Green Energy and Technology

Ashwani K. Gupta · Ashoke De ·

Suresh K. Aggarwal · Abhijit Kushari ·
Akshai K. Runchal Editors

Advances in
Energy and
Combustion
Safety and sustainability
Green Energy and Technology
Climate change, environmental impact and the limited natural resources urge
scientific research and novel technical solutions. The monograph series Green Energy
and Technology serves as a publishing platform for scientific and technological
approaches to “green”—i.e. environmentally friendly and sustainable—technologies.
While a focus lies on energy and power supply, it also covers "green" solutions in
industrial engineering and engineering design. Green Energy and Technology
addresses researchers, advanced students, technical consultants as well as decision
makers in industries and politics. Hence, the level of presentation spans from
instructional to highly technical.
**Indexed in Scopus**.
**Indexed in Ei Compendex**.

More information about this series at http://www.springer.com/series/8059

Ashwani K. Gupta · Ashoke De ·
Suresh K. Aggarwal · Abhijit Kushari ·
Akshai K. Runchal
Editors

Advances in Energy
and Combustion
Safety and sustainability
Editors
Ashwani K. Gupta Ashoke De
Department of Mechanical Engineering Department of Aerospace Engineering
University of Maryland Indian Institute of Technology Kanpur
College Park, MD, USA Kanpur, Uttar Pradesh, India

Suresh K. Aggarwal Abhijit Kushari

Department of Mechanical and Industrial Department of Aerospace Engineering
Engineering Indian Institute of Technology
University of Illinois at Chicago Kanpur, Uttar Pradesh, India
Chicago, IL, USA

Akshai K. Runchal
Analytic and Computational Research, Inc.
Los Angeles, CA, USA

ISSN 1865-3529 ISSN 1865-3537 (electronic)

Green Energy and Technology
ISBN 978-981-16-2647-0 ISBN 978-981-16-2648-7 (eBook)
https://doi.org/10.1007/978-981-16-2648-7

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Singapore Pte Ltd. 2022
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface

The amount of energy used worldwide is significant, with more than 500 quads of
energy per year. Most of the propulsion and power systems use liquid fuels repre-
senting 19 million barrels of liquid fuel per day alone in the USA and about 5
times more this amount worldwide. The development of environmentally benign
and sustainable energy sources that drive the society and industry continues to be
an important fundamental challenge for engineers and scientists. The known avail-
able fossil fuel resources are rapidly depleting, and as a result, there is an increased
emphasis on developing renewable and alternative energy sources in a sustainable
manner. In the near-term, fossil fuels are expected to remain the main driver of our
economy and infrastructure as they offer dispatchability and high energy density. We
have seen continued improvements in fossil fuel utilization for different technologies
over the past century. Current technologies using fossil fuels for energy are also recog-
nized as contributing to global warming, degrading of the environment, and a burden
on the economy with extensive cleanup costs. Greenhouse gas emissions are known to
be significant contributor to climate change, rising global temperatures, and lasting
impact on the protective ozone layer. The consequent increase in global average
temperatures, melting of glaciers, and sea-level rise are now serious threats to the
human society’s very core of existence. Simultaneously, our requirements for energy
needs continue to grow due to the considerable growth of the human population
in developing countries and increasing living standards in the emerging economies.
This means that the average per capita energy use will only increase substantially in
the coming decades, with fossil fuels to remain important major sources of energy
for all industrial and power applications through direct combustion of fuels. Society
urgently needs safe, sustainable, transportable, and more efficient energy sources for
various industrial and domestic needs at affordable costs.
The world population has reached over 7.8 billion and is projected to increase by
more than 25% to 9.9 billion by 2050. The worldwide energy supply in 2019 was
about 170,000 TWh, or 580 quads (quadrillions BTU), and is expected to increase
to 815 quads by 2040. Per IEA, if we continue along the current path, without any
additional policy changes, the energy demand will rise by 1.3% per year up to 2040. If
the policy intentions and targets stated by major energy user nations are implemented,

v
vi Preface

this energy demand rate could decrease to 1% per year. While considerable progress
is being made to shift to renewable energy sources, fossil fuels are expected to provide
75% of the world’s energy in 2040. While the environment is already under stress,
either of these scenarios places a significant additional burden on the environment
with a somewhat slower rise in emissions but still far short of sustainability and
safety. Therefore, we urgently need a sustainable development scenario that maps
out a way to meet energy demand goals, requiring rapid and widespread changes
across all energy sectors. The current and foreseeable clean energy technologies are
not likely to provide the answer. We need new and innovative technologies with
improved efficiency to meet the Paris Agreement’s goals by controlling the rise in
global temperatures to well below 2 °C and pursuing efforts to limit it to 1.5 °C and
meet global objectives cost-effective energy access and cleaner air. The spectrum
of the world’s energy needs requires multiple solutions. Reduction in emissions (in
particular CO2 and particulates) must be achieved from all kinds of gas, liquid, and
solid fuel use. The dispatchable technologies provide efficient and cost-effective
energy solutions in all energy-using sectors.
An international forum to discuss the challenges outlined above was started in
2004, focusing on combustion, power, and propulsion technologies using different
kinds of fuels available to us. Over the years, it has evolved into annual international
workshops on Sustainable Energy, Power and Propulsion (ISEPP), which provide a
broad framework of international collaborations and share ideas about research and
developments related to cleaner and efficient utilization of energy sources and to
develop sustainable alternative and renewable energy technologies. The 11th work-
shop in this series was held from February 24 to 26, 2020, at the Indian Institute
of Technology, Kanpur, India. The workshop’s objectives were to promote inter-
national collaborations and disseminate knowledge related to safe and sustainable
energy initiatives and technology development. The workshop attracted over 100
participants from different academic and research organizations and representa-
tives from governments, industry, and academia. Many world-renowned scientists
and researchers from Japan, Malaysia, Thailand, Saudi Arabia, the UK, USA, and
India presented their latest research and development findings and shared ideas,
informal talks and informal technical exchanges. The TEQIP (Ministry of Human
Resource Development, India), the Defense Research and Development Organiza-
tion (DRDO), India, The National Science Foundation (NSF), USA, and the Office
of Naval Research (ONR) Global sponsored the workshop. It was co-sponsored by
industries, several leading research organizations, and universities, including Begell
House Publishers, Springer/Springer Nature Publishers, Siemens, TSI, Inc., ACRi
Los Angeles, NIT Kurukshetra, IIT Kanpur, University of Maryland, College Park,
and the University of Illinois at Chicago. The editors wish to express their sincere
thanks to the faculty, staff, and students at IIT Kanpur for their outstanding support
and warm hospitality to all participants at this international workshop.
This research monograph provides a wealth of knowledge on the latest research
and development presented by the eminent scientists and engineers at the work-
shop. The broad theme of the monograph is energy, combustion, power, safety, and
sustainability. The chapters are grouped into six sections: Propulsion Systems, Fuels,
Preface vii

Flames, and Reacting Systems, Combustion Systems, Fires, and Sustainable Energy
Technologies. These chapters deal with prototype systems using state-of-the-art diag-
nostic and analytic techniques. Each chapter is written by a renowned international
authority in the field of study. Several chapters deal with specific design aspects.
In contrast, the others deal with novel concepts using alternative energy sources for
improved performance of next generation of advanced power and propulsion systems,
with due considerations to the emission of various trace pollutants, such as polycyclic
aromatic hydrocarbons (PAH), dioxins, furans, NOx, unburnt hydrocarbons, carbon
monoxide, carbon, and particulates. The contributions of cutting-edge research and
development technologies should be of paramount interest to researchers and engi-
neers working on safe and sustainable energy technologies in academia and industry.
We hope the book proves to be an asset to those working in combustion and clean
energy.

College Park, USA Ashwani K. Gupta

Kanpur, India Ashoke De
Chicago, USA Suresh K. Aggarwal
Kanpur, India Abhijit Kushari
Los Angeles, USA Akshai K. Runchal
Introduction

This book is evolved from the International Workshop on Energy, Environment and
Multiphase Flows held at IIT Kanpur during February 24–26, 2020, with the sponsor-
ship from the Office of Naval Research Global (ONR Global), US National Science
Foundation (NSF), TEQUIP (India), Defense Research and Development Organiza-
tion (India), Begell House Publishers, Springer/Springer Nature Publishers, Siemens,
and TSI Inc. The workshop was co-sponsored by IIT Kanpur, University of Mary-
land College Park, USA, the University of Illinois at Chicago, USA, and Analytic
and Computational Research Inc. (ACRI) Los Angeles, USA. This workshop repre-
sented the 11th International Workshop on the Broad Theme of Energy, Power,
and Propulsion, with the first workshop in 2004. Over the years, the workshop has
grown into a world-class forum for sustainable fuels, energy, power, and propulsion,
focusing on clean energy production and utilization. This volume is focused on fuels,
combustion, energy, safety, and sustainability representing many areas of engineering
importance in energy, power, and propulsion. The Chief Guest at the workshop
was Dr. Satish Kumar, Director of National Institute of Technology, Kurukshetra,
Haryana, India, who provided a wealth of information from his rich, extensive prior
experiences in various leadership positions at Defense Research and Development
Organization (DRDO), India. Renowned international authorities delivered some
28 invited lectures in their specific expertise on the various topics covered at the
workshop. Twenty-four topics on advances in energy and combustion, emphasizing
safety and sustainability from the selected authorities, were delivered. Subsequently,
they contributed a chapter on their topic, which are included in this book. This
book is aimed at researchers working in energy conversion, waste, pollution control,
modeling, diagnostics, fluid mechanics, heat transfer, biofuels production, and energy
sustainability and safety. This book should also benefit the design and development
engineers of power plant equipment and managers from industry and government

ix
x Introduction

with the latest scientific developments reported on safe and sustainable use of fossil
fuels, biofuels, and low-grade fuels in specific applications.

Ashwani K. Gupta
Ashoke De
Suresh K. Aggarwal
Abhijit Kushari
Akshai K. Runchal
Contents

Propulsion Systems
Injector Dynamics and Pressure Gain in Rotating Detonation
Engines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
K. Kailasanath
Low Emissions Propulsion Engine Characterization Process . . . . . . . . . . . 13
Hukam C. Mongia
Aerodynamic and Aero-Acoustic Performance of an Adjustable
Pitch Axial Flow Fan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Rajat Arora, Ramraj H. Sundararaj, T. Chandrasekar, Sharad Saxena,
and Abhijit Kushari
Proposed Thrust Profile Design of Pulse Detonation Engine (PDE)
for Aerospace Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Subhash Chander and T. K. Jindal

Fuels
The Formation of PAH Compounds from the Combustion
of Biofuels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
A. Williams, A. R. Lea-Langton, and K. D. Bartle
Review of Biomass Energy Resources with Livestock Manure . . . . . . . . . . 125
Osama M. Selim, Juan Espindola, and Ryoichi S. Amano
Higher Alcohols as Diesel Engine Fuel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Naveen Kumar, Harveer S. Pali, Ankit Sonthalia, and Sidharth
Photocatalytic Hydrogen from Water Over Semiconductors . . . . . . . . . . . 175
Siow Hwa Teo, Aminul Islam, and Yun Hin Taufiq-Yap

xi
xii Contents

Flames and Reacting Systems

Evaluation of Hazard Correlations for Hydrogen-Rich Fuels Using
Stretched Transient Flames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Michael Peter Meyer and Rune Peter Lindstedt
Experimental Investigation of Turbulent Flow/Flame Structure
of Double Swirler Burner . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Dhanalakshmi Sellan and Saravanan Balusamy
Flame Root Dynamics and Their Role in the Stabilisation of Lifted
Flames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
James C. Massey, Zhi X. Chen, and Nedunchezhian Swaminathan
Effect of Oxygenation on the Liftoff, Stabilization, and Blowout
Characteristics of Laminar Co-flow Jet Flames . . . . . . . . . . . . . . . . . . . . . . . 273
Rajesh Kumar, Krishna C. Kalvakala, and Suresh K. Aggarwal
Laminar Burning Velocity Measurements at Elevated Pressure
and Temperatures and the Challenges in Kinetic Scheme
Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
Robin John Varghese and Sudarshan Kumar

Combustion Systems
Direct Numerical Simulation of Preignition and Knock in Engine
Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
Minh Bau Luong and Hong G. Im
Numerical Investigation of Flow-Acoustics Coupling
in a Half-Dump Combustor Using Hybrid CFD-CAA Methodology . . . . 337
Bharat Bhatia and Ashoke De
Emissions from FEHA Fueled RQL Gas Turbine Combustor
Under High-Pressure and High-Temperature Simulated in-Flight
Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
Hitoshi Fujiwara and Keiichi Okai
Performance Evaluation of Upward Swirl Combustor with Reverse
Fuel Injector and Hydrogen Blending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
Parag Rajpara, Rupesh Shah, and Jyotirmay Banerjee

Fires
Quantification of Enhanced Fire Severity in Modern Buildings . . . . . . . . 413
Dharmit Nakrani and Gaurav Srivastava
The Tea Light Candle and the Global Waste Problem . . . . . . . . . . . . . . . . . 429
Ali S. Rangwala and Kemal S. Arsava
Contents xiii

Sustainable Energy Technologies

Approach to Solve Renewable Energy Problems . . . . . . . . . . . . . . . . . . . . . . 441
Sourabh Kumar and Ryoichi S. Amano
Energy Recovery from Waste Tires Via Thermochemical Pathways . . . . 477
Kiran R. G. Burra, Zhiwei Wang, Matteo Policella, and Ashwani K. Gupta
Biomass Conversion to Power in Southeast Asian Countries:
Current Situation and Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523
Somrat Kerdsuwan and Krongkaew Laohalidanond
Role of Artificial Roughness in the Performance Improvement
of Solar Air Heaters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555
Navneet Arya, Rajneesh Kumar, and Varun Goel
Estimation of Mechanical Behavior of Cementitious Material
Using Microstructural Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 579
Praveen Verma and Ayushman Bajpai
About the Editors

Dr. Ashwani K. Gupta is a Distinguished University Professor at the University

of Maryland (UMD), USA. He obtained his Ph.D. and D.Sc. from the University
of Sheffield, UK. He was awarded D.Sc. from Southampton University, UK, and
Honorary Doctorates from the University of Wisconsin–Milwaukee, USA, King
Mongkut’s University of Technology North Bangkok, Thailand, and University of
Derby, UK. He received the President Kirwan Research Award and College of Engi-
neering Research Award at UMD. He has received several honors and awards from
AIAA and ASME, and has co-authored over 800 papers, three books, five patents,
and edited 18 books in the areas of combustion, swirl flows, high temperature air
combustion (HiTAC), distributed combustion, wastes to energy, acid gas treatment,
fuel reforming, and air pollution. He is Honorary Fellow of ASME and RAeS (UK),
and Fellow of AIAA, SAE, and AAAS, and Member of European Academy of
Sciences and Arts (EASA).

Dr. Ashoke De is a Professor in the Department of Aerospace Engineering, Indian

Institute of Technology (IIT) Kanpur. He received his MS from IIT Kanpur, and Ph.D.
from Louisiana State University, USA. Before joining IIT Kanpur, he has worked as
post-doctoral scholar at Technical University of Delft (TU-Delft), Netherlands and
as Research Engineer in GE Global research at Bangalore, India. He is the recipient
of DAAD Fellowship (2016), IEI-Young Engineer’s Award-2014 (Aerospace Engi-
neering), Early Career Research award-2015 (DST, India), Expertise Award-2010
(GE Global research), and P. K. Kelkar Young Research Fellowship (2015–2018,
IITK). He is AvHumboldt Fellow, Fellow of West Bengal Academy of Science and
Technology, Associate Editor of International Journal of Energy for a Clean Environ-
ment (2019–present), and currently member of APS, AIAA (senior member), ASME,
SIAM, FMFP, ISHMT and Combustion Institute. He leads large scale initiatives in
the modelling of turbulent flows at IIT Kanpur and authored over 130 papers, and
four edited books. His current research interests include multiphase modeling, hybrid
RANS/LES model development, supersonic flows and Fluid-Structure interactions
(FSI).

xv
xvi About the Editors

Dr. Suresh K. Aggarwal received his Ph.D. from Georgia Institute of Technology,
Atlanta, USA. He was a member of the research staff at Princeton University, New
Jersey, USA and a Senior Research Engineer at CMU. He then joined the Univer-
sity of Illinois at Chicago, USA and was promoted to Professor in 1995. He served
as Director of Graduate Studies, and was also a Visiting Scientist at ANL, Visiting
Professor at Ecole Centrale-Paris, France and Guest Professor at Jiangsu Univer-
sity, China. His research interests include Combustion, Multiphase Reacting Flows,
Emissions, Clean Energy, and Fire Suppression. He has authored over 340 publica-
tions and graduated 17 Ph.D. and 46 MS students. He is Fellow of ASME, AAAS
and Associate Fellow of AIAA.

Dr. Abhijit Kushari is Professor in the Department of Aerospace Engineering,

Indian Institute of Technology (IIT) Kanpur, India. Dr. Kushari received his Ph.D.
from Georgia Institute of Technology, Atlanta, Georgia, in 2000 and joined IIT
Kanpur as an Assistant Professor in 2001. He became a Professor in 2014. His
research interests are Aerospace Propulsion, Gas Turbine engines, Turbo-machinery,
Liquid Atomization Flow Control and Combustion Dynamics. He has authored more
than 130 technical papers. He has supervised six Ph.D. and 50 M.Tech. theses in
Aerospace Propulsion fields.

Dr. Akshai K. Runchal is an expert in Computational Fluid Dynamics (CFD) and

numerical simulation of flow, heat and mass transport processes in engineering and
environmental sciences. In late 1960s under the guidance of Prof. D. B. Spalding
at Imperial College, London, Dr. Runchal, and Dr. Wolfshtein, developed the Finite
Volume Method (FVM) for CFD applications. His experience spans a range of prob-
lems including those related to design, production, operation and environmental
impact of industrial and urban projects. One of his specific areas of expertise has
been analysis and management of natural and man-made disaster scenarios. Dr.
Runchal received his Ph.D. from Imperial College, London (UK) in 1969 and a
Bachelor of Engineering with Honors from Punjab Engineering College (Chandi-
garh, India) in 1964. He has taught and guided research students at UCLA, Cal Tech,
CSU (Northridge), Imperial College (London), and IIT (Kanpur). He is the author or
co-author of six books and over 200 technical publications. He has delivered keynote
and invited talks at more than 100 conferences and seminars in the USA, the UK,
France, Germany, India, Australia, Belgium, Sweden, Croatia, Czechoslovakia and
Slovenia.
Propulsion Systems
Injector Dynamics and Pressure Gain
in Rotating Detonation Engines

K. Kailasanath

Abstract Rotating detonation engines are an example of a pressure gain combustion

device. By definition, detonation is a pressure gain combustion process, and hence,
a combustor based on detonations will always produce pressure gain. However, this
does not always translate into pressure gain for an engine based on detonations
since there are other components involved in an engine. Numerical simulations of
an idealized “engine,” consisting of a simple inlet (with injectors), combustor and
exhaust nozzle is used to illustrate this issue. The inlet injector area ratio is shown to
be a key parameter. A poor choice of this parameter could result in inlet losses that
are larger than the pressure gain in the combustor, resulting in an overall system that
does not produce pressure gain. Some other choices will produce overall pressure
gain but may induce unwanted pressure oscillations. Hence, a suitable design of the
inlet-combustor interface is shown to be a critical factor in the overall development
of rotating detonation engines.

Keywords Detonation engines · Pressure gain · Injectors

1 Background

Typical combustion engines in current use for propulsion or power use flames to
convert the fuel into products and release the chemical energy. In theory, the conver-
sion process takes place under constant pressure conditions, while in practice, there
is some pressure loss. Detonations provide an alternate form of energy conversion.
Detonations are a pressure gain combustion process and hence result in higher
thermodynamic efficiency, which can result in better fuel efficiency and reduced
emissions.
There is world-wide interest in exploring and developing detonation engines [1].
An overview of detonation engines as well as recent developments has been discussed
elsewhere [2]. For the sake of completeness and to motivate the work presented in
this paper, some of the key observations from the past are repeated here. The initial

K. Kailasanath (B)
Lorton, VA, USA

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 3
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_1
4 K. Kailasanath

emphasis on the development of detonation engines was focussed on “intermittent” or

“pulsed” detonation engines, called PDE. The basic concept and early work has been
reviewed by Eidelman et al. [3]. There was uncertainty even in the estimates of the
ideal performance of these engines [4]. Advances in laser diagnostics and numer-
ical simulations enabled better performance estimates as well as development of
laboratory-scale devices that produced close to the estimated performance. Research
in the 1990s and early 2000s also identified key factors that control the performance
and operation of pulse detonation engines. The inherent intermittency of the PDE,
as well as the need to repeatedly initiate detonations rapidly and with the use of very
little energy, are major obstacles in the development of PDEs. The attention then
shifted to continuous detonation engines with a single initiation event to overcome
the above obstacles. Rotating detonation engines (RDE) are a form of continuous
detonation engine, where a detonation moves continuously in a annular, cylindrical
chamber, as long as it is supplied continuously with fresh detonable mixture. Deto-
nation needs to be initiated only once at the beginning of the operation. Hence, this
concept overcomes two of the key difficulties in the development of PDEs. However,
this does introduce new issues that need to be researched and resolved. One such
issue is pressure gain in these engines.
In this paper, key factors that control the pressure gain in these devices are explored
in more detail using numerical simulations of a generic rotating detonation engine
(RDE). A generic configuration has been chosen so that the results can have wider
applicability than choosing a specific engine under current development. The focus is
also on identifying key factors that may be useful in further development of this type
of engine. The fuel considered in the simulations is hydrogen, because its chemistry
is well understood and can be represented in a tractable manner in complex three-
dimensional simulations. However, the observations presented are more general and
not dependent on the specific fuel.

2 A Rotating Detonation Engine

The generic configuration of a rotating detonation engine used in the simulations is

shown in Fig. 1. It consists of an annular cylinder as the combustion chamber with
an injector plate at one end (bottom of the figure) and an exhaust nozzle at the other
end (top of the figure).
The fresh fuel–air mixture is injected from the bottom and the products of deto-
nation flow out through the nozzle at the top. Once a detonation wave is initiated in
the combustion chamber, it keeps moving in a circumferential direction (rotating)
as long as the inflow of detonable fuel–air mixture is sustained. Further details of
the flow field are not presented here since they have been discussed previously in a
number of papers [5, 6].
Injector Dynamics and Pressure Gain … 5

Fig. 1 Simulation of the

flow field in a generic
rotating detonation engine
(RDE)

2.1 Modeling Detonation Engines

The emphasis of this paper is on the impact of the injector geometry on the flow field
and dynamics within the engine. Hence, a full three-dimensional simulation that
can adequately represent the geometrical details is required. However, the standard
unsteady Euler equations can be used since viscous effects have very little influence
over the very short time scales involved in detonation propagation. This also allows
us to keep the computational cost reasonable and conduct a representative set of
simulations to explore various parametric changes.

3 Effect of Area Ratio

Before considering specific types of inflow injectors, let us first consider the generic
problem. A key characteristic defining an inflow injector is the open area as a fraction
of the total area available. We call this the area ratio for the purpose of our discussion
here. A series of numerical simulations was conducted in which this area ratio was
varied from 0.1 (small opening) to 0.75 (large opening). The pressure in the upstream
plenum chamber was held constant at 10 atmospheres for all the cases. The pressure
in the combustor chamber just at the exit of the injector and the pressure at the exit
of the combustor after combustion were noted in each of the cases. These results are
presented in Table 1, along with those holding the mass flow constant. Of course in all
cases, the pressure at the exit plane of the combustor is higher than at the inlet, because
6 K. Kailasanath

Table 1 Effect of inlet area-ratio on pressure gain

Set 1. Constant plenum pressure
Area ratio Pplenum (atm) Pt,inlet * Pt,exit (atm) % inlet choked
0.1 10 1.79 3.92 87.8%
0.2 10 2.99 6.66 90.7%
0.3 10 4.25 9.39 94.4%
0.4 10 5.35 11.91 94.6%
0.5 10 6.27 14.08 94.9%
0.75 10 8.69 18.38 95.0%
Set 2. Approximately hold mass flow constant
0.1 16.47 3.13 6.59 82.4
0.2 10.00 2.99 6.66 90.7
0.3 6.94 3.15 6.72 72.0
0.4 5.53 3.21 6.55 64.9
0.5 4.71 3.21 6.78 60.3
0.75 3.40 3.39 6.21 26.7
*Inlet pressure only integrated over choked area

of the pressure gained during the detonative combustion process. However, when we
also take into account the initial plenum pressure, we observe that the pressure at the
exit is lower than the plenum pressure for all cases where the area ratio is smaller
than 0.4. This indicated that for these cases, the pressure drop across the injector is
larger than the pressure gain in the combustor. Thus, a key injector design parameter
to achieve overall pressure gain is the inlet area ratio, which needs to be about 0.4 or
larger. The second set of data in Table 1, where the mass flow is held constant, also
suggests the same conclusion. These results provide us with some guidance on the
overall design of the injector. However, further consideration must be given because
with the larger opening there is the possibility of back flow into the plenum chamber.
In the case of a flight device, this could result in back flow or pressure oscillations in
the inlet. Therefore, a better understanding of how to isolate the pre-mixture plenum
from the combustion chamber is essential to practically achiever pressure gain in
these types of devices. Let us look at some specific type of injectors to understand
this problem better.

4 Impact of Slot Injectors

The basic RDE geometry considered has an inner diameter of 80 mm and an outer
diameter of 100 mm, giving a wall thickness of 20 mm for the injectors. First a simple
slot type of injector (as shown in Fig. 2) was simulated. A total of 50 injectors were
Injector Dynamics and Pressure Gain … 7

Fig. 2 Schematic of the slot

injectors used in the
simulations

used with spacing, w of 5.457 mm and a height T of 10 mm. The throat area, At , was
1.131 for the area ratio of 0.2 and 2.262 for the area ratio of 0.4.
With these slot injectors, both the temperature and pressure fields within the
combustion chamber show the clear presence of relatively cool jets at the injectors
and dead zones of higher temperature gas existing between the jets. No back flow
into the mixture plenum was observed in the cases simulated. However, pressure
waves are seen to propagate into the mixture plenum through the injector opening,
as the detonation wave moves past the injectors, as shown in Fig. 3.

5 Impact of Cylindrical Injectors

Next cylindrical injectors were considered. Again 50 injectors were used resulting
in the same average spacing as in the previous case. A representation of the installed
injectors is shown in a cut-out of the RDE geometry in Fig. 4.
The simulations with cylindrical injectors show the same basic pattern as with the
slot injectors. However, the presence of the dead zones between injectors (discussed
above) is smaller because of the shape of the injector. As before, there is pressure
feedback into the plenum, as seen in Fig. 5. The pressure gradient shows much
more structure than with the slot injectors. Multiple shock waves emanate from the
injectors into the high-pressure region behind the detonation wave.
8 K. Kailasanath

Fig. 3 Instantaneous snapshot of the pressure gradient field showing the presence of pressure waves
within the plenum chamber

Fig. 4 Cut open view of the RDE geometry showing the cylindrical injectors
Injector Dynamics and Pressure Gain … 9

Fig. 5 Instantaneous snapshot of the pressure gradient field for the simulation with cylindrical
injectors

6 Impact of Pintel Injectors

The next type of injector considered was of the pintel type, as shown in Fig. 6.

Fig. 6 Schematic of the

pintel type of injector
10 K. Kailasanath

Fig. 7 Instantaneous snapshot of the pressure gradient field for the simulation with pintel injectors

The pintel type of injector resulted in a much smoother temperature and pressure
profile within the combustion chamber but it did produce pressure waves propagating
into the plenum chamber, as in the previous cases discussed above. As before, the
pressure gradient distribution within the plenum chamber is shown in Fig. 7.

7 Performance and Stability

The stability and overall performance with the various types of injectors were
compared to simulations where inflow boundary conditions (ideal case, for reference)
were used rather than physical injectors of the various types discussed above. There
were only small changes in the performance as measured by the specific impulse.
With an area ratio of 0.2, the performance for all cases including those with the ideal
reference case was within 1%. With the larger area ratio of 0.4, there was a decrease
in performance with the physical injectors ranging from 2.4 to 7.7%. The reduction
in performance was most noticeable as low (3 atm) and high pressures (15 atm).
Similarly, the stability of the waves in the system was also impacted for the extremes
in operating pressure. Several different modes of failure including propagation in
the reverse direction were observed. For the lower pressure case, detonation wave
failure and re-ignition continually happens near the injection plate. Sometimes, the
re-ignition event can be strong enough to initiate a detonation wave propagating in
the opposite direction to the original direction of propagation.
To visually convey the level of pressure oscillations, the pressure traces in the
center of the plenum chamber are compared for the simulations with the different
aspect ratios of 0.2 and 0.4 for the slot injector case in Fig. 8.
In general, for any type of injector, the amplitude of pressure oscillations was
larger for the larger area ratios.
Injector Dynamics and Pressure Gain … 11

Fig. 8 Comparison of pressure traces in the center of the plenum for the slot injector system with
an area ratio (AR) of 0.2 and 0.4. The mean pressure is 4 atm

8 Concluding Remarks

Numerical simulations have been used to explore the impact of various types of
injectors on the flow field and performance of rotating detonation engines. It is
shown that some choices for the fuel–air injectors could result in a large pressure
drop that could negate the pressure gained during the detonative combustion process.
The inlet injector area ratio is shown to be a key parameter. Relatively large area ratios
(greater than 0.4) will provide better pressure gain. However, with the larger area
ratios, the pressure feedback and oscillations in the injector plenum chamber are
shown to be larger. In some cases, instability and even failure of the detonation could
result. Hence, the choice of the particular type of injector, especially the area ratio,
is critical for the successful development of these engines. Further injector choices,
including isolators to limit the upstream pressure propagation, are suggested for
further research and development.
12 K. Kailasanath

References

1. Kailasanath K (2000) Review of propulsion applications of detonation waves. AIAA J

38(9):1698–1708
2. Kailasanath K (2020) Recent developments in the research on pressure-gain combustion devices.
In: Gupta AK, De A, Aggarwal SK, Kushari A, Runchal A (eds) Innovations in sustainable energy
and cleaner environment. Springer, pp 3–21
3. Eidelman S, Grossmann W, Lottati I (1991) Review of propulsion applications and numerical
simulations of the pulse detonation engine concept. J Prop Power 7(6):857–865
4. Kailasanath K (2001) A review of PDE research—performance estimates. AIAA paper 2001-
0474. AIAA, Reston, VA
5. Schwer DA, Kailasanath K Numerical investigation of rotating detonation engines. AIAA 2010-
6880. AIAA, Reston, VA
6. Schwer DA, Kailasanath K Numerical investigation of the physics of rotating-detonation-
engines. In: Proceedings of the combustion institute, vol 33. The Combustion Institute,
Pittsburgh, PA, pp 2195–2202
Low Emissions Propulsion Engine
Characterization Process

Hukam C. Mongia

Abstract Most of the propulsion engines use Low-Emissions rich-quench lean

combustors LEC except recently introduced GEnx (2009) and LEAP-X (2016) based
on the application of a second-generation lean dome combustion technology TAPS
demonstration (2003). These two technologies are targeted for replacing pre-LEC
rich domes or the 1st generation lean-dome technology (known popularly as Dual-
Annular Combustors DAC) combustors. Sixteen combustors have been selected from
three engine manufacturers (CFM International, GE Aircraft Engines, and Rolls-
Royce Plc.) involving these four technologies covering a broad range of ICAO
takeoff pressure ratios (21–49) and thrust rating (77–514 kN) with engine emis-
sions certification testing conducted during 1985 and 2017. Even though the ICAO
emissions regulatory criteria given by engines’ landing takeoff emissions of NOx ,
CO and hydrocarbons normalized by takeoff thrust, g/kN (LTO NOx , CO and HC)
should be acceptable, they hardly provide any guidance during combustion design and
development process. This paper summarizes a low-emissions technology evalua-
tion criterion that originally started in the middle 1970s under the NASA-sponsored
Pollution Reduction Technology Program (PRTP). Application of this criteria to
the selected 16 propulsion engine combustion products shows that the LEC takeoff
NOxEI correlation (NOxEI = 0.0563 × PR1.1796 with R 2 = 0.986) along with idle
COEI and HC correlation (COEI = 8.231 + 7.196 × HCEI with R 2 = 0.9949)
and idle HCEI entitlement of zero can be used for comparing all rich-dome
combustion technologies and products. The corresponding takeoff NOxEI corre-
lations for the 1st and 2nd generation lean dome combustion products (DAC and
TAPS) NOxEI = 1.110E − 03 × PR2.950 with R 2 = 0.9772 and NOxEI =
7.480E − 05 × PR3.4534 with R 2 = 0.9985, respectively, should be used for
guiding future lean-dome combustion products in addition to matching or exceeding
idle COEI versus takeoff NOxEI tradeoff correlation established by GEnx product,
namely Idle COEI = 32.183 Takeoff NOxEI−0.223 with R 2 = 0.8967 while
simultaneously its idle COEI and HCEI is close to that of the LEC correlation
COEI = 8.231 + 7.196 × HCEI.

H. C. Mongia (B)
CSTI Associates, LLC, Yardley, PA, USA

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 13
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_2
14 H. C. Mongia

Keywords Gas turbine combustors · Combustion technology · CFD · Design

development process · Low emissions combustors

1 Introduction

All the major aero propulsion engine manufacturers have been intensively involved
in reducing pollutant emissions starting in the 1970s leading to the 1st generation of
low-emissions combustors in several modern engines and enforcement of the Inter-
national Civil Aviation Organization ICAO 1st set of emissions regulatory standards
known popularly as Committee on Aviation Environmental Protection (CAEP) CAEP
1 https://www.icao.int/environmental-protection/Pages/Caep.aspx where from Fig. 1
has been reproduced. The first CAEE committee and subsequent CAEP committee
members have maintained a highly admirable professional environment that helped
formulate emissions regulatory levels equitably for all Original Equipment Manu-
facturers OEM’s, small, medium, and large size propulsion engines as shown in
Fig. 2.
The original CAEP (or CAEP1) regulatory standards for engine exhaust emissions
during landing takeoff are given by the following expressions:
 
83.6
SAE Smoke Number = Minimum 50, 0.274
F00

Hydrocarbons HC = 19.6 gm CH4 /kN.

Carbon monoxide CO = 118 gm/kN.

Fig. 1 Major milestones of CAEP since 1986; reproduced from https://www.icao.int/environme

ntal-protection/Pages/Caep.aspx
Low Emissions Propulsion Engine Characterization Process 15

Fig. 2 LTO NOx stringencies from the original CAEP (adopted in 1981 through CAEP2, ––
CAEP8 (adopted respectively in 1993, 1999, 2005 and 2011) compared to Mid-term and Long-
Term goals; reproduced from https://www.icao.int/Meetings/EnvironmentalWorkshops/Docume
nts/2014-Kenya/4-1_LAQ-Technology_notes.pdf

Oxides of nitrogen NOx = 40 + 2xPR for all engines manufactured before January
1, 2000 with rated thrust F00 > 26.7 kN. Here, PR is the ICAO-specified engine takeoff
pressure ratio.
At 30 PR, the CAEP1 regulatory standard requires that engines with LTO NOx
more than100 gm of NO2 /kN will not be allowed to operate. Therefore, if possible,
OEM’s would like to have a 30% margin from the standards shown by dotted blue
lines in Fig. 3 except for smoke where the preference is generally for 50% margin.
It is educational to pay attention to how low-emissions combustion product
evolution took place starting with the engine emissions databank available
from the link: https://www.easa.europa.eu/easa-and-you/environment/icao-aircraft-
engine-emissions-databank.
We plan to cover engines manufactured by the three OEM (CFM, GE, and RR)
listed in this database. For the GE engines, the first engine to incorporate rich-
dome Lean-Combustion or Rich-Quench-Lean Combustion LEC/RQL technology
was the CF6-80C2 (with emissions test start date of January 13, 1995) followed by
quick (but not optimized) transition of LEC technology to GE90 (April 24, 1997).
Only minor LEC optimization was undertaken for the GE90-110/115B (November
23, 2003) because of the long-term objective of replacing lean-dome Dual-Annular
Combustion system (1st generation) by the 2nd generation lean-dome Twin-Annular
partially Premixing Swirlers based concentric flames TAPS [1, 2]. A new centerline
CF34-10E was the 4th LEC combustion product (July 24, 2004) followed by drop-in
16 H. C. Mongia

1000 250
RB211/84 RB211/84
CF6/83 200

% CAEP 1 CO
CF6/83
% CAEP 1 HC

RB211/91 RB211/91
CF6-80C2PLEC 150 CF6-80C2PLEC
100 OEM's preference
100

50
10 0
20 25 30 35 20 25 30 35
Takeoff Pressure Rao Takeoff Pressure Rao
a. LTO hydrocarbons (HC) emissions b. LTO CO emissions
% CAEP 1 SAE Smoke Number

140 110
RB211/84 RB211/84
120 100

% CAEP 1 NOx
CF6/83 CF6/83
100 RB211/91 90 RB211/91
80 CF6-80C2PLEC CF6-80C2PLEC
80
60 70
40 60
20 50
0 40
20 25 30 35 20 25 30 35
Takeoff Pressure Rao Takeoff Pressure Rao
c. Characteristic smoke emissions d. LTO NOx emissions

Fig. 3 Remarkable propulsion engine combustion product emissions reduction trends established
by the primary leaders of combustor designers community before 1992

LEC combustor for the CFM56-5B and CFM56-7B emissions tested on September
29, 2003.
In summary, the author believes that GE and CFM LEC combustion product devel-
opment activities were not followed as intensively [4] as for the 1st and 2nd generation
lean dome products [1–3], namely DAC and TAPS. The DAC combustion products
development during the early 1990s started from an extensive technology know-
how and empirical correlations [5–8]. The LEC product development effort relied
heavily on the anchored CFD approach [9] since the most reputed semi-analytical
correlations [10, 11] didn’t produce right trends for NOx as function of fuel/air ratio
FAR [12]. However, the anchored modeling approach needed to be modified for
future application in development of lean combustion product [13, 14] which led to
formulation and validation of comprehensive models overviewed in [15]. Improved
semi-empirical and semi-analytical correlations developed later [16–18], if used, can
provide more insight needed for future LEC combustion products.
The RR combustion community is very proud of its RQL combustion products
which provided the author motivation to assess emissions characteristics of their
rich-domes relative to LEC. It should be pointed out that the author doesn’t have
first-hand knowledge of the RR RQL combustion products. Therefore, his process of
selection and grouping of their engine families was driven mostly by the RR engines’
emissions test start date. The legend RB211/84 in Fig. 3 includes all the RB211 engine
models with engine start dates falling between March 1979 and July 1984 whereas
Low Emissions Propulsion Engine Characterization Process 17

the legend RB211/91 groups together engine tested between September 1986 and
April 1991.
Similarly, the legend CF6-80C2PLEC in Fig. 3 are the engine models tested on
May 29, 1983 whereas CF6/83 includes all CF6 engine models tested prior to May
29, 1983. The summary of these engine groupings given in Table 1 highlights the
minimum and maximum extent of HC, CO, NOx and smoke emissions.
It’s remarkable for the RR RB211/84 combustion team to step up to reduce HC
from its maximum value of 915 to take the challenge of reducing it enough to meet the
CAEP1 requirement. Its maximum value was 67.7 as shown in Table 1 for RB211/92.
Similarly, good progress was made on meeting CO which came down from 192.9 to
24.4. However, the RR combustion team struggled in bringing down NOx and smoke
emissions.
The GE engines group CF6/83 didn’t have as much challenging tasks. Here, they
had to reduce its maximum HC value of 225 to less than 100, which came down to
99.5 in the CF6-80C2 which achieved or exceeded OEM’s target values shown in
Fig. 3.
Subsequently, CAEP NOx regulatory standards have gone through more stringent
requirements with introduction of CAEP2, CAEP4, CAEP6, and CAEP8 for engines
produced after January 1 of the years 2000, 2004, 2008 and 2014, respectively. The
corresponding LTO NOx regulatory values at 30 PR for all engines with takeoff thrust
≥89.1 kN are 80, 67, 59, and 50. In other words, relative to CAEP1, the percentage
reductions in the regulatory NOx standard are 20%, 33%, 41%, and 50%. Or, the
successive NOx reduction steps going from CAEP1 to CAEP8 have been 20%, 16%,
12%, and 15%, respectively.
What started out as a simple expression for CAEP1 NOx standard (namely, 40 +
2PR) has become quite complicated CAEP8 standard as illustrated below. This was
driven mostly by one objective: The regulatory standards should be fair to all OEM’s
whether manufacturing small, medium or large engines in regard to rated static engine
takeoff thrust F00 and the attendant overall operating takeoff pressure ratio PR. Some
engines static takeoff pressure ratio may be medium, high and very high. These
groups are currently categorized into the following groups:

≤ 26.7 kN
F00 = 26.7 − 89 kN
> 89 kN

and

≤ 30
PR = 30 − 104.7
> 104.7

The CAEP8 regulatory standards for engines that fall into more than 26.7 kN but
not more than 89 kN are quite complex as summarized below:
 18

Table 1 Minimum and maximum range of LTO emissions normalized by regulatory levels of the original CAEP (CAEP1) proposed in 1981 for the selected
four engine families
Legends Hydrocarbons Carbon monoxide Oxides of nitrogen SAE Smoke number
Minimum Maximum Minimum Maximum Minimum Maximum Minimum Maximum
RB211/84 13.3 915.4 27.6 192.9 63.9 94.4 9.8 119.7
RB211/92 3.6 67.7 18.8 24.4 83 99.9 11.4 84.2
CF6/83 59.2 225.3 35.3 91.3 61.6 78.9 20.9 81.6
CF6-80C2PLEC 69.4 99.5 47.5 62.4 41.8 53.1 35.5 53.3
H. C. Mongia
Low Emissions Propulsion Engine Characterization Process 19

LTO NOx = 40.052 + 1.5681PR − 0.3615F00 − 0.0018PR

× F00 when PR ≤ 3041.9435 + 1.505PR
− 0.5823F00 − 0.005562PR
× F00 when 30 < PR < 104.7

Fortunately, we will not discuss this class of engines in this chapter. Focus of
this chapter is on propulsion engines with takeoff thrust more than 89 kN for which
CAEP8 standards are quite simple:

7.88 + 1.408PR when PR ≤ 30

LTO NOx = −9.88 + 2PR when 30 < PR < 104.7
32 + 1.6PR when PR > 104.7

As mentioned earlier, CAEP1 NOx requirement for overall pressure ratio OPR of
30 is 100 g/kN compared to 50 g/kN as per the CAEP8 NOx requirement. It means
that the CF6-80C2PreLEC’s LTO NOx emission levels 41.8 ≤ LTO NOx ≤ 53.1
emissions tested in 1985 essentially met CAEP8 requirement applicable for engines
manufactured after January 1, 2014. For this reason, in addition to the fact most of
this chapter is about TAPS engines, we will present figures when appropriate in term
of CAEP8 NOx requirement, as summarized in Fig. 4. It should be emphasized that
applicable NOx standards for the CF6-80C2 engine families tested in 1985 and 1995
(identified respectively as CF6-80C2PreLEC and CF6-80C2LEC) are CAEP1 and
CAEP2, not CAEP8 as shown in Fig. 4. We expressed these in terms of CAEP8 so
that it can be compared directly with the emission characteristics of the GEnx and
LEAP-X engine families summarized in Fig. 5.
When normalized by CAEP8 regulatory levels, the CF6-80C2PreLEC gives LTO
NOx ranging between 91 and 102% which was reduced only slightly to fall between
93 and 97 for the LEC product. However, the resulting low-emission combustion
LEC product reduced significantly the unburned hydrocarbons and CO emissions.
We will have more discussion on this topic in Sect. 3 which provides the list of 16
engines selected and their emissions normalized by CAEP8 standards.
The author, very proud of his association with the TAPS team that developed the
technology, conducted an extensive CFM TAPS engine demonstration program [1,
2] leading to prestigious team recognition award and the 1st GE Aviation’s Edison
award in 2003, followed by highly successful combustion product development as
part of the GEnx combustion, was very disappointed to see the results summarized in
Fig. 5a. How can we have 70% increase in LTO NOx of the two models of LEAP-X
supposedly having been scaled from GEnx?
The author asked one of his former combustion colleagues, “Is there a typograph-
ical error in the ICAO engine emissions databank in regard to LEAP-X with the
attendant increase in LTO NOx?” His response - “I wish it was”—is what motivated
the author to pull together material for this chapter.
Since the author started paying little attention to the development of rich-dome
LEC combustion products after the GE90-110/115B tested in 2003, he out of curiosity
20 H. C. Mongia

120 CF6-80C2 PreLEC Tested in 1985

100 NOx

% CAEP 8 Standards
HC
80
CO OEM
60 Smoke Preference

40

20

0
25 27 29 31 33 35
Takeoff Pressure Rao

a. CF6-80C2 engine family tested in 1985 identified as CF6-80C2PreLEC

120 CF6-80C2LEC Tested in 1995

NOx
100 Smoke
% CAEP 8 Standards

CO
y = -0.3415x2 + 20.801x - 220.61
80 HC R² = 0.6289 OEM
Preference
60
y = 0.2595x1.4946
R² = 0.8355
40 y = 6858.3x-1.652
R² = 0.9492
20
y = 59080x-2.454
R² = 0.9687
0
25 27 29 31 33 35
Takeoff Pressure Rao

b. CF6-80C2 engine family tested in 1995 identified as CF6-80C2LEC showing that its
maximum LTO NOx value of 97.3 doesn’t exceed CAEP 8 regulatory level of 100.
However, when expressed in terms of its relevant regulatory level, namely CAEP 2, the
NOx emission range falls between 57-63% CAEP 2, falling below the desirable 70%
level. Smoke emissions given by has more than 50% margin, CO
given by more than 70% margin and HC given by has
more than 80% margin from the CAEP 8 standards.

Fig. 4 Back to back comparison of the LTO emissions characteristics of the CF6-80C2 engine
families tested in 1985 and 1995 identified as CF6-80C2PreLEC and CF6-80C2LEC, respectively

wanted to compare its LTO NOx with the RR most recent Trent Extra Wide Body
emission tested in 2012 leading him to ask a question. Is RR Trent XWB’s 20% lower
LTO NOx compared to GE90-110/115B indicative of RR’s continuing improvement
in its RQL technology while GE’s focus was on developing the 2nd generation
lean-dome TAPS technology and product?
Is the LEAP-X setback indicative of innate limitations of lean-domes or simply
an anomaly caused by several reasons? How will it impact future TAPS-based lean
combustion products including GE9X supposedly a TAPS combustion system? These
questions and many more provided the motivation for the author to undertake a
Low Emissions Propulsion Engine Characterization Process 21

100
GEnx
LEAP-1A
90 LEAP-1B
70% increase
in LTO NOx
80

% CAEP8 NOx
70

60

50

40
30 35 40 45 50
ICAO Takeoff Pressure Rao

a. LEAP-X (LEAP-1A and Leap-1B) supposedly scaled from GEnx gave 70% higher LTO
NOx; why?
75
70 CFM56 TI
y = 1.7462x - 8.2888
65 CF34-10E R² = 0.9799
CF6-80C2 20%
60
LTO NOx, g/kN

GE90
55
LEC
50
Trent XWB y = 1.8202x - 22.69
45 R² = 0.9892
40
35
30
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao

b. Has GEAE’s rich-dome LEC combustion products fallen behind RR rich-dome RQL,
here Trent Extra Wide Body Trent XWB tested in April 1, 2012? GE90-110/115B (leg-
end GE90) was tested on November 23, 2003. LTO NOx of LEC products represented
well by compared to
.

Fig. 5 Back to back comparison between the LTO NOx emissions characteristics of GEnx versus
LEAP-X and GE rich-dome LEC versus Trent Extra Wide Body engine models

systematic assessment of the gaseous emissions characteristics of modern propulsion

engines.
The low-emissions combustion technology process characterization originally
developed as part of [19–21] has worked very well for all the programs the author
has been involved since the middle 1970s, as summarized in [25]. Section 2 summa-
rizes the application of this approach for evaluating the CF6-80C PreLEC and LEC
combustion products.
A list of 16 modern propulsion engines (Table 3) and their LTO characteristic
values expressed in terms of CAEP8 requirement are summarized in Sect. 3 followed
by an extensive discussion on their LTO NOx characteristics in Sect. 4.
22 H. C. Mongia

Application of the high-power NOx methodology of Sect. 2 is demonstrated to all

rich-dome combustors of Table 3 in Sect. 5 clearly demonstrates that recent LTO NOx
reductions can be attributed to these engines’ reduction in their LTO fuel burn. Their
takeoff and climbout NOxEI versus pressure ratio characteristics have not changed
since the middle 1990s.
High- and low-power emissions tradeoffs are summarized in Sect. 6 for all 16
rich- and lean-dome combustors followed by deep dive into lean domes in Sect. 7.
If TAPS combustion system design is optimized properly, its mission NOx emis-
sions potentials are significantly superior than rich-domes’ as summarized in Sect. 9
preceded by Sect. 8. Here a Data Reduction and Analysis DRA is explored for rich-
dome combustor emissions over its sea-level emissions from idle to maximum power
followed by its extension over the entire flight envelop. Finally, a brief summary is
provided in Sect. 10.

2 Characterization of CF6-80C2 Combustors’ Emissions

The ICAO landing takeoff gaseous emissions (pollutants, Pi ) is calculated from:

Charcateristic factor 
LTO Pi = 0.7 × Ẇ f Takeoff × TakeoffPi
F00
+ 2.2 × Ẇ f Climbout × ClimboutPi + 4 × Ẇ f Approach

× ApproachPi + 26 × Ẇ f Idle × IdlePi

Here, Pi denotes gaseous emissions hydrocarbons gm CH4 /kg fuel), CO gm/kg

fuel and oxides of nitrogen NOx gm NO2 /kg fuel, respectively. Ẇ f i denotes the fuel
flow rates (kg/min) during the takeoff, climbout, approach, and taxi idle operational
modes, respectively, with the corresponding times of 0.7, 2.2, 4, and 26 min. F00
(kN) is engine’s static takeoff thrust at the standard atmospheric conditions.
Clearly, LTO emissions are determined from engine performance (here, fuel flow
rates) and emissions indices of HC, CO, and NOx , represented in this chapter as
HCEI, COEI, and NOxEI, respectively.
Characteristic factors, as explained on p. 106 of [22] take into consideration effect
of sample size, namely, the number of engines tested and the repeatability of emis-
sions data of HC, CO, NOx and smoke number. From [22], we have created Table 2
as an illustration to show effect of number of engines tested, here 1, 2, 3 and in the
limiting case of 10 engines.
Most of the ICAO engine emissions data comprise of single engine with three test
runs in order to calculate average LTO values of HC, CO, NOx and Smoke number.
However, its characteristic LTO values are expected not to exceed the ICAO regula-
tory levels. This means that average LTO values have to be multiplied, respectively,
by 1.2274, 1.5401, 1.1592, and 1.2872. Sometimes, OEM’s might test three engines;
and for this case, the multiplication factors reduce to 1.0815, 1.1666, 1.0593, and
Low Emissions Propulsion Engine Characterization Process 23

Table 2 ICAO recommended characteristic factors for HC, CO, NOx, and smoke number when
number of engines tested are 1, 2, 3, and 10
Engines Characteristic factors for
Tested HC CO NOx Smoke #
1 1.2274 1.5401 1.1592 1.2872
2 1.1393 1.3012 1.0996 1.1727
3 1.0815 1.1666 1.0593 1.1000
10 1.0431 1.0848 1.0316 1.0524

Table 3 List of 16 propulsion engines manufactured by the three engine manufacturers covering
a broad range of ICAO takeoff pressure ratios (21–49) and thrust rating (77–514 k N) with engine
emissions certification testing conducted during 1985 and 2017
Number Legend Takeoff pressure ratio Takeoff thrust (kN) Engine test dates
Minimum Maximum Minimum Maximum From To
Engines by CFM International and GE Aircraft Engines
1 LEAP-1A 30.0 38.6 106.8 143.1 11/21/2016 12/2/2016
2 LEAP-1B 36.8 42.0 111.3 130.4 3/8/2016 3/9/2016
3 CFM56-5B TI 22.7 32.6 96.1 142.3 9/29/2005 3/23/2006
4 CFM56-7B TI 21.4 29.0 86.7 121.4 9/29/2005 3/23/2006
5 CFM56-7B 22.8 28.8 91.6 121.4 8/30/1997 9/12/1997
DAC
6 CFM56-5B 24.4 32.8 102.2 142.4 8/1/1996 9/1/1996
DAC
7 GEnx 35.5 47.5 255.3 349.2 6/19/2012 6/21/2012
8 CF34-10E 25.5 27.2 77.4 83.7 6/5/2009 6/6/2009
9 GE90 39.7 42.2 492.6 513.9 11/23/2003 11/23/2003
10 GE90-DAC1 35.2 40.4 360.6 426.7 2/24/1995 7/28/1995
11 CF6-80C2 27.5 33.1 231.1 287.0 1/13/1995 1/17/1995
12 CF6-80C2PLEC 27.5 32.9 231.1 270.0 5/29/1985 6/3/1985
Engines by Rolls-Royce Plc.
13 Trent XWB 36.8 48.6 334.0 436.7 4/1/2012 4/20/2016
14 T1000 36.2 49.4 268.0 363.9 9/14/2009 11/30/2017
15 TPh5 34.5 45.4 251.9 345.9 3/16/1995 11/30/2017
16 T1994 34.0 41.5 300.3 413.1 9/1/1994
Range 21.4 49.4 77.4 513.9 5/29/1985 11/30/2017

1.1. Rarely, the OEM’s run10 engines for collecting emissions data; but if required,
then these multiplication factors reduce to 1.0431, 1.0848, 1.0316, and 1.0524.
Figure 6 gives a schematic of the CF6-80C2 combustion system cross-section
adopted from [4] in order to identify schematically the major changes between the
PreLEC and LEC combustors, here, specifically, swirl cup (SC), splash plate, dome,
24 H. C. Mongia

Cooling airflow
distribution
SC

Fig. 6 Schematic layout of the CF6-80C combustion system showing modifications in swirl cup
design (for reducing smoke) and cooling airflow distribution in order to optimize wall temperature
levels and gradients of the swirl cup splash plate, dome, and primary panels. Modified from [4]

and primary panels’ cooling air modifications. These modifications were not intended
for changing the spray characteristics or the primary zone flow dynamics. Conse-
quently, LEC takeoff NOxEI was expected to change only slightly from that of
PreLEC. The quality of the LEC NOx emission data is considerably better than the
data scatter band from the PreLEC as summarized in Fig. 7. Takeoff NOxEI data can
be analyzed by several empirical, semi-empirical, or semianalytical correlations, see
[10–12, 16–18]. However, here we plot takeoff NOxEI as a function of the ICAO

60 CF6-80C2
CF6-80C2PLEC GE LEC Technology
55
LEC
50 CF6PLEC
Takeoff NOxEI, g/kg

45
40
35
y = 0.0541x1.809
30
R² = 0.99
25
20
15
10
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao

Fig. 7 Takeoff NOxEI versus pressure ratio of the CF6-80C2 with PreLEC and LEC combustion
systems compared to GE LEC technology trendline expressed as LEC Takeoff NOxEI = 0.0541 ×
PR1.809 with R 2 = 0.99 shown by dotted black curve. Red curve is given by 1.12xLEC Takeoff
NOxEI which passes through data scatter of the CF6-80C2 PreLEC
Low Emissions Propulsion Engine Characterization Process 25

takeoff pressure ratio in Fig. 7 because by doing so does not require any proprietary
details about the combustion system.
Details on the development of the GE LEC NOxEI technology will be provided
later. Since there is wider data scatter for the PreLEC, we have arbitrarily plotted
1.12 times LEC Takeoff NOxEI trendline for now in order to illustrate that one may
be inclined to suggest that PreLEC takeoff NOxEI is only 12% higher than that of
the LEC technology. Therefore, at least for this engine model, the emphasis was not
on reducing NOx unlike for the RR combustors moving forward after 1991, refer to
Table 1.
It is well known that design modifications for reducing smoke and high-power
NOx may be accompanied with challenges for low-power emissions, operability
that includes ground and altitude ignition, and lean blowout. Other challenges may
include combustor exit temperature quality and combustion system durability. Here,
our focus will be on idle CO and unburned hydrocarbon emissions which in addition
to generally accepted tradeoff with reducing high-power NOx can be impacted by
atomization characteristics, and entrainment of the cooler film cooling air, and atten-
dant quenching of HC and CO. Alternatively, lack of cooling air for protecting the
combustor dome structure can lead to higher liner or dome wall temperature levels
and gradients and resulting impact on durability and shop visits. Based on lessons
learned from [19–21], we have continued to use since the middle 1970s illustrations
similar to Figs. 8 and 9.
If the target takeoff NOxEI goal were 25, the CF6-80C2 LEC combustion product
reduced simultaneously NOx , idle COEI and HCEI emissions by 10%, 58%, and 85%
compared to the PreLEC, and hence one can claim success in demonstrating lower
emissions technology combustor without increasing smoke emissions as summarized
previously in Fig. 4.

60
55 CF6-80C2 CF6-80C2PLEC
50
45
Idle COEI, g/kg

40
y = 175.36x-0.408
35 R² = 0.7689
30
25
20 y = 235.84x-0.768
15 R² = 0.7787
10
20 25 30 35 40
Takeoff NOxEI, g/kg

Fig. 8 Idle COEI plotted as function of takeoff NOxEI for the CF6-80C2 PreLEC and LEC
combustors are represented well by PreLEC Idle COEI = Takeoff 175
NOxEI0.408
with R 2 = 0.7689
and LEC Idle COEI = Takeoff 236
NOxEI0.768
with R 2 = 0.7787 showing that compared to PreLEC,
LEC reduces NOx by approximately 10% while simultaneously at takeoff NOxEI of 25 it reduces
idle COEI by 58%
26 H. C. Mongia

60
55 CF6-80C2 CF6-80C2PLEC
50

Idle COEI, g/kg

45
40
y = 2.894x + 15.669
35 R² = 0.9992
30
25
20
y = 7.196x + 8.2309
15 R² = 0.9949
10
0 2 4 6 8 10 12 14
Idle HCEI, g/kg

a. Idle COEI plotted as function of idle HCEI giving good quality correlations
and

14
CF6-80C2 CF6-80C2PLEC
12
Idle HCEI, g/kg

10

8 y = 0.3453x - 5.4017
R² = 0.9992
6

4 y = 0.1383x - 1.1301
R² = 0.9949
2

0
10 20 30 40 50 60
Idle COEI, g/kg

b. Idle HCEI plotted as function of idle COEI giving good quality correlations
and
.

Fig. 9 Idle COEI plotted as function of idle HCEI of the CF6-80C2 PreLEC and LEC combustor
(Part a) is preferred over the bottom where idle HCEI has been plotted as function of idle COEI
wherefrom we can calculate that for 25 takeoff NOxEI, the LEC combustor reduced NOx , idle
COEI and HCEI by 10%, 58%, and 85%

From Fig. 4b, it is remarkable to see that LEC combustion system emission certi-
fied in January 1995 can meet the most stringent CAEP8 NOx standard in a propulsion
engine designed in the late 1970’s through early 1980s in spite of its higher LTO fuel
burn. It also has 50%, 75%, and 85% margin from the regulatory requirements of
smoke, CO, and HC emissions. This raises one potential question, (Are the CAEP8
standards not stringent enough?) which is partially answered by Fig. 4c.
A word of caution about the right interpretation of Fig. 9 which comprises of
several aspects of the combustion processes taking place simultaneously within a
typical propulsion combustor. This includes air atomization characteristics, aero
tradeoff between NOx formation and CO oxidation, quenching of partially burned
Low Emissions Propulsion Engine Characterization Process 27

hydrocarbons and CO by cooling air and or escape of part of spray’s excessively

large size droplets.
The use of Figs. 7, 8 and 9 combined with the methodology [23, 24] and applicable
CFD tools [9, 15] has led successfully to evolving generation of low emissions gas
turbine combustors as summarized in [25].

3 LTO Emissions of Selected Propulsion Engines

Focus of this section is to compare characteristic LTO emissions data normalized

by the CAEP8 regulatory levels. However, in Sect. 4 we plan to use average LTO
emissions data followed by average indices of emissions in Sect. 5. The reason for
using CAEP8 NOx standards is simple: it is applicable “for engines of a type or
model of which the date of manufacture of the first individual production model was
after 1st January 2014.” Therefore, this section is intended for young combustion
technologist to notice that unless there is further reduction in climbout and takeoff
NOx of rich-dome combustion technology and or LTO fuel burn, it will be difficult
to have 30% margin from the regulatory NOx standard.
Twelve modern propulsion engines manufactured by the CFM International and
GE Aircraft Engines were selected in order to compare swirl cup [3] based Low-
Emissions Combustors LEC (engines numbered 3, 4, 8, 9, and 11 listed in Table 3),
pre-LEC (#12), lean-dome Dual-Annular Combustors DAC (5, 6, and 10), and TAPS
#1, 2, and 7. Details on these four combustion technologies are given in [1–4].
All the selected four engine families from RR (#13–16) belong to Rich-quench-
Lean RQL category which is essentially similar to LEC technology practiced by GE
and the CFM.
As summarized in Table 3, the selected 16 engines cover a very broad area in regard
to takeoff pressure ratio (21.4–49.4) and rated takeoff thrust that ranges between 77
and 514 kN. The emissions certification dates of these engines cover a long period
of 32 years, 5/29/1985 through 11/30/2017.
Figures 10, 11, 12 and 13 summarize landing takeoff emissions of NOx , unburned
hydrocarbons HC, CO and characteristics SAE smoke numbers of the selected 11
GE and CFM engines showing performance of the three combustion technologies
(low-emissions rich-domes LEC, dual annular combustors DAC and lean-dome twin
annular partially mixing swirlers TAPS), expressed as % of CAEP8 regulatory stan-
dards. Essentially all three technologies meet these requirements, some with consid-
erable margins whereas others approaching or slightly exceeding the limits. But
these figures don’t shed much light in regard to underlying reasons for explaining
the observed wide scatter band. For that we will have to go to Sects. 4–6.
If OEM’s had a choice for their future combustion products, they would like to
have 30% margin from the regulatory standards for NOx , 50% for HC, CO, and
smoke shown as dotted blue lines in these figures. However, when the products
turn out higher than these goals, some of them are willing to contribute toward
improving technology. For example, some of the CFM DAC products have margins
28 H. C. Mongia

100 120
LEC
110 DAC
90

% CAEP8 NOx
% CAEP8 NOx

DAC 100
LEC
80 90
70 80
GEnx OEM's preference
LEAP-1A
TAPS 70 CF34-10E
60 LEAP-1B
GE90
CFM56-5B_TI 60
CFM56-7B_TI Only LEAP-X subject GE90-DAC1
50 CFM56-7B DAC to CAEP8 Requirements 50 y = 0.4963x1.2669
CF6-80C2
CFM56-5B DAC R² = 0.943
40 40
20 25 30 35 40 45 20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao ICAO Takeoff Pressure Rao

Fig. 10 Characteristic landing takeoff LTO NOx, expressed as % CAEP8 NOx, for the CFM (L)
and GEAE (R) engines

120 80
LEAP-1A
70
100 LEAP-1B GEnx
CFM56-5B_TI 60 CF34-10E
% CAEP HC
% CAEP HC

CFM56-7B_TI GE90
80 CFM56-7B DAC 50 GE90-DAC1
CFM56-5B DAC
CF6-80C2
60 OEM's
40
40
Preference 30
20
20 10
0 0
20 25 30 35 40 45 20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao ICAO Takeoff Pressure Rao

Fig. 11 Characteristic LTO unburned hydrocarbons HC, expressed as % CAEP HC, for the CFM
(L) and GEAE (R) engines

120 100
LEAP-1A 90
100 LEAP-1B
80
% CAEP CO

CFM56-5B_TI
OEM's Preference
70
% CAEP CO

CFM56-7B_TI
80 CFM56-7B DAC
CFM56-5B DAC 60
60 50
GEnx
40 CF34-10E
40 30 GE90
GE90-DAC1
20 20 CF6-80C2
10
0 0
20 25 30 35 40 45 20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao ICAO Takeoff Pressure Rao

Fig. 12 Characteristic LTO CO, expressed as % CAEP CO, for the CFM (L) and GEAE (R) engines

of 10%, 3%, and 0.4% from NOx , HC and CO standards, respectively. On the other
hand, some of the GE90 DAC models have margins of −9% and 33% for NOx
and smoke. Environmentally conscious passengers don’t like to fly in airplanes with
visible exhaust plume. During Boeing 777 with the GE90 DAC I flyby over GEAE
main building, the author received a phone call from his VP of Engineering asking
explanation for his visual observation of exhaust smoke. For the time being, he was
satisfied by my explanation that it must be flying outside the LTO cycle.
Low Emissions Propulsion Engine Characterization Process 29

90 80
80 LEAP-1A
LEAP-1B
70
% CAEP Smoke

% CAEP Smoke
70 CFM56-5B_TI 60
CFM56-7B_TI
60 CFM56-7B DAC
OEM's Preference
50
50 CFM56-5B DAC
40 GEnx
40
30 CF34-10E
30 GE90
20 20 GE90-DAC1
CF6-80C2
10 10
0 0
20 25 30 35 40 45 20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao ICAO Takeoff Pressure Rao

Fig. 13 Characteristic SAE Smoke number, expressed as % CAEP Smoke, for the CFM (L) and
GEAE (R) engines

Fearing that 33% margin from smoke standard may not be enough, we increased
our goal to more than 50% which we exceeded in the follow-on rich-dome GE90
product with 65% margin, as shown in Fig. 13. But truly lean-dome products are
expected to exhibit much higher margins from smoke standards and one may get
decent margin from optimized rich domes also. For example, the lean-dome TAPS
engines GEnx and LEAP-X have more than 95% margin from smoke standards. The
CFM DAC’s smoke margin is more than 88% except at 25 overall pressure ratio OPR
it reduces to 66%. The rich dome CF6-80C2 has more than 50% margin. But rich
domes can have more variability in their smoke emissions as shown by the CFM56
Tech Insertion engines.
Some OEM’s are more willing to pursue lean-dome technology and product devel-
opment activities including the author who prefers concentric lean-flames over radi-
ally [5–8] or axially [19–21] staged concepts realizing well that its more complex
flow characteristics and unravelling surprises to be discussed later in Figs. 15, 16 and
43. In spite of less than stellar outcome of the DACs, the CFM and GEAE contributed
more than $100 million toward development of the TAPS combustion technology,
CFM TAPS engine demonstration [1, 2], the GEnx combustion system design, fabri-
cation, limited full annular combustor rig test followed by first engine to test FETT
in March 2006 culminating with toll gate TG 6.
In summary, the CFM family of engines with TAPS II technology have only 6%
margin for NOx compared to 32% with GE’s GEnx. The author is glad that the
LEAP-X combustion used nomenclature TAPS II and not TAPS 2 which according
to [1] was intended to get 50% reduction in NOxEI from the 1st generation TAPS
technology (TAPS 1) used in GEnx. We will have more discussion on LEAP-X,
GEnx, TAPS1, TAPS2, and TAPS3 later.
30 H. C. Mongia

4 Landing Takeoff NOx Characteristics

Figure 14 presents average landing takeoff NOx (not characteristic values) of the
selected eleven GEAE/CFM engine models (Table 3) as function of takeoff pressure
ratio along with several lines identified as % of CAEP1 NOx regulatory standard,
namely LTO NOx = 40 + 2 × PR. Here, 50% CAEP1 implies that its value at 30PR
is 50 g/kN; and likewise 35 and 20 for 35% and 20%CAEP1 legends, respectively.
It should be noted that the slope of these lines is 2.0. It should be recalled that for all
practical purposes, 50% CAEP1 is equivalent to CAEP8 standard. Even though the
CFM56-5B and CFM56-7B Tech Insertion combustion systems are identical, their
LTO values are different; the former has slightly lower LTO NOx values. We will
discuss this apparent anamoly later in Sect. 5. Due to some reasons, there is more data
scatter for the CFM56-5B_TI than for the CFM56-7B_TI which lies slightly below
50% CAEP1 line. We observed similar behavior in HC (see Fig. 11), CO (Fig. 12)
and smoke emissions as shown in Fig. 13.
In spite of some data scatter of the dual-annular combustion systems for the engine
models CFM56-5BDAC and CFM56-7BDAC, it shows slightly higher values for the
latter compared to the former, consistent with the rich-dome SACs. There is some
tendency for losing advantage of DAC over SAC; but it becomes clearly evident
when we combine DAC data from both the CFM56 and GE90 engine families. The
CFM engine models lie between 35 and 50% of CAEP1, DAC’s giving slightly lower
values. But the lean dome dual annular combustion system of the nominal 40 pressure
ratio GE90 increases to 50%CAEP1 level a loss of 15 points which is attributed to
its lower share of combustion air. As per the conventional wisdom, its primary zone
equivalence ratio is closer to 0.9 instead of being closer to the desirable range of less
than 0.7 or 0.75.
In summary, DAC technology didn’t live up to its expectations inferred somewhat
wrongly from its technology development and two engine demonstrator programs
summarized in [5–8]. These programs didn’t even try to improve fuel/air mixed-
ness quality unlike the small engine pollution reduction technology program PRTP
described in [19–21] where more than 70% NOx reduction was achieved after
first modification. DAC technology programs didn’t spend enough resources in
addressing their low-power challenges unlike PRTP where 75% of the effort was
spent in addressing low-power challenges. Moreover, the DAC activities didn’t put
enough emphasis on addressing significant variation in the combustor exit tempera-
ture radial profiles caused by radial staging of the fuel nozzles. Poor fuel/air mixed-
ness, radial staging of the fuel nozzles and stubborn refusal to adopt twin-wall liner
cooling technology from [8] were the main reasons for the failure of DAC tech-
nology in its transitioning to market competitive low emissions propulsion engine
combustion products.
In spite of the great freedom the author enjoyed during his 15 years at GE Aviation,
he was unable to get the senior management to buy into developing more cooling air
efficient metalic twin-wall technology which would have made more air available
Low Emissions Propulsion Engine Characterization Process 31

for combustion with attendant increased potential for NOx reduction for GE90 DAC
I.
Therefore, the TAPS1 technology development activities proceeded despite the
adverse consequences of using state-of-the-art film and multihole cooling techniques
used routinely in GE propulsion engine combustion products unlike the competitors
with supposedly more efficient twin-wall cooling technologies, Floatall and Tiled.
For example, P&W reported first use of Floatwall in its low-emissions engine model
PW4168 with emissions test start date of February 14, 1993, two years before the
TAPS technology initiative was kicked off. Ten years later, RR reported the use
of Tiled constructed combustion system in its engine model Trent900 with engine
emissions start date of October 17, 2003.
But lack of twin-wall’s use in LEAP-X combustor doesn’t explain the failure
of the LEAP-X TAPS combustion products summarized in Fig. 14. It starts with
20%CAEP1 at 30 overall pressure ratio OPR, as expected, but increases somewhat
erratically and sometimes consistently to 35%CAEP1 by 42 OPR. One cannot miss
similarity with the DAC’s where we thought we understood losing 15 points going
from 30 to 40 OPR engines. Did the scaling from GEnx to LEAP-X process fail
for lack of proper hypotheses formulation [23, 24] or unintended consequences of
making wrong selection between the two different internal flowfield characteristics

Fig. 14 Average LTO NOx 100

characteristics for the CFM 90
LEAP-1A
LEAP-1B
(T) and GEAE (B) engines CFM56-5B_TI
80
LTO NOx, g/kN

CFM56-7B_TI
70 CFM56-7B DAC
CFM56-5B DAC
60
50
40
30
20
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao

100
GEnx
90 CF34-10E
80 GE90
LTO NOx, g/kN

GE90-DAC1
70 CF6-80C2

60

50

40

30 y = 2.0973x - 51.464
R² = 0.971
20
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao
32 H. C. Mongia

predicted analytically in [27], Figs. 15 and 16? We will come back to this topic later
in Sect. 7.
We are inclined to come to very different tentative conclusions from the bottom
part of Fig. 14 which summarizes the LTO NOx characteristics of the three technolo-
gies of GEAE, namely lean-dome GE90-DAC1, rich-dome LEC products (CF34-
10E, CF6-80C2, and GE90 (rated for the ICAO takeoff thrust between 493 and
514 kN) and TAP1 technology-based GEnx for rated takeoff thrust between 255
and 349 kN. It is true that DAC technology product gave LTO NOx falling along
50% CAEP1 compared to the LEC along 43% CAEP1 and TAP1 significantly lower
value of 14%CAEP1. But that doesn’t mean that combustion technologist can claim
that his TAP1 combustion technology produced approximately 1/3rd NOx emissions
compared to the LEC-based rich-dome GE engines for the same reason the Rolls
Royce combustion technology engineer can claim that its Trent X wide-body (Trent
XWB) has 20% lower NOx emissions compared to the GE90-115B engine model,

Combustor 1

RCZ
RCZ

Combustor 2
Fig. 15 Reacting flow mean axial velocity contours (m/s) predicted by comprehensive combustion
system analytical modeling approach [27]

Combustor 1

Combustor 2
Fig. 16 Mean gas temperature contours (°F) predicted by comprehensive combustion system
analytical modeling approach [27]
Low Emissions Propulsion Engine Characterization Process 33

Fig. 17 Average LTO NOx 80

characteristics of the CFM 75 CFM56 TI
and GEAE engines’ rich 70 CF34-10E
CF6-80C2

LTO NOx, g/kN

domes compared with RR 65
GE90 20%
Trent X Wide Body (Trent 60 Trent XWB
XWB) showing 20% LTO 55 GE90-DAC1
NOx reduction compared to 50 CF6-80C2PLEC
GE90-115B engine model 45 y = 1.8202x - 22.69
R² = 0.9892
40
35
30
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao

as summarized in Fig. 17. Why? Because the GE90-115B’s landing takeoff cycle-
specific fuel burn of 3 kg/kN is 6% higher than that of Trent XWB as summarized
in Fig. 18 for several engines as function of takeoff pressure ratio and thrust.
There is no simple honest way of filtering out the effect of engine fuel burn out of
the LTO NOx emissions to claim how much contribution combustion technologist has
made in reducing emissions. Even more important area of concern for the combustion
technologist should be about answering wide scatter band in LTO NOx of combustors
based on supposedly the same technology base, for example, LEC for GE and RQL.
And then how much benefit RR got because of its use of more cooling air-efficient
double-wall Tiled compared to single wall cooling technology products of GE? These
questions will be answered in rest of this chapter starting with the material presented
in Sect. 5.

5 Rich-Dome Combustors’ NOx Assessment

Consistent performance in developing combustion technology and products meeting

customers’ expectations has been of paramount importance to the author since the
early 1970s because of what generally happened to Jack Haasis who was his best on-
the-job combustion teacher. His superb quality work’s productivity was continuously
sabotaged by self-proclaimed burned-out combustion experts driven mostly by self-
promotional misplaced egos. In order to keep these saboteurs at arm’s length, it was
very important for the author to reduce data similar to Fig. 17 to some definite conclu-
sions in order to help him for providing guidance in achieving continuous systematic
advances in both combustion technology and products. Again, to emphasize what
has been mentioned before, the author is not familiar with the Rolls Royce combus-
tion products other than what can be obtained from reliable COMPLETE-STORY
publications which hardly exist.
Figure 19b presents a cluster of T1994 that represents LTO NOx data of the Trent
engine models 768, 772, 875, 877, 884, 892, and 895 emission certified on September
1, 1994. This remarkable data set became the benchmark for RQL combustors’ NOx
34 H. C. Mongia

4.1
CFM56 TI
3.9 CF34-10E
CF6-80C2
LTO Fuel Burn/F, kg/kN
3.7
GE90
3.5 Trent XWB
T1000
3.3
TPh5
3.1 T1994

2.9

2.7

2.5
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao
4.1 CFM56 TI
3.9 CF34-10E
CF6-80C2
LTO Fuel Burn/F, kg/kN

3.7 GE90
3.5 Trent XWB
T1000
3.3 TPh5
3.1 T1994

2.9
2.7
2.5
0 100 200 300 400 500 600
ICAO Takeoff Thrust F, kN

Fig. 18 LTO fuel burn divided by takeoff thrust F versus takeoff pressure ratio (top) and takeoff
thrust shown in the bottom

for other OEM’s to compete with. As far as the author knows, these combustors
didn’t use advanced liner cooling ‘Tiled’. Tiled were used in the Trent models 7000-
72, 7000-72C, 970-84, 972-84, 553-61, 556-61, 970-84, 972-84 and 972E-84 in
addition to using ‘Phase V’ RQL technology. The Trent engine models 768 and 772
also belong to Phase V RQL technology group; but there is no mention of “Tiled’ in
these two models. All these 11 models have been lumped together under the legend
TPh5 in Figs. 19b.
Figures 17 and 19 raises at least four questions:
(1) Is Trent’s 20% lower LTO NOx compared to GE90-115B due to its superior
RQL combustion technology or Trent’s lower LTO fuel burn, Fig. 18?
(2) GEAE/CFM engines cover takeoff thrust range between 77 and 514kN
compared to the Trent family of engines with narrower range of 252 to 437kN.
The corresponding range of takeoff pressure of the Trent is 34–49 compared to
Low Emissions Propulsion Engine Characterization Process 35

70
CFM56 TI
65
CF34-10E

LTO NOx, g/kN

60 CF6-80C2
55 GE90
y = 1.7462x - 8.2888
50 LEC R² = 0.9799

45 Linear (LEC)

40
35
30
20 25 30 35 40 45
ICAO Takeoff Pressure Rao
a. Average LTO NOx versus takeoff pressure ratio characteristic described reasonably

75
CFM56 TI
70 CF34-10E
65 CF6-80C2
LTO NOx, g/kN

GE90
60
Trent XWB
55 T1000
50 TPh5
T1994
45
40
35
30
20 25 30 35 40 45 50 55
ICAO Takeoff Pressure Rao

b. Cluster of Trent family of engines that use RR RQL combustion technology compared
to the GEAE/CFM LEC engines

Fig. 19 Average LTO NOx versus takeoff pressure ratio characteristic of five GEAE/ CFM LEC
combustors fall along a linear trendline (top) while all RR Trent family engines emission certified
between 1994 and 2017 divided into four groups form a wide scatter band. Why?

GE/CFM’s of 21 to 42. Why more scatter in LTO NOx data of RR compared

to GE/CFM, compare Fig. 19b with Fig. 19a?
(3) Is “Phase V” RQL technology the best RR technology or it’s simply used as a
marketing tool?
(4) How good was RR in transitioning its RQL technology to the pre-LEC
combustor of Allison AE3007 combustion product?
The 4th question can be answered at least partially by the author since he was
involved in converting a high smoke, industrial engine 570 combustion system with
aerodynamic diffuser, dual-orifice pressure atomizer, and double-wall convection-
film cooled liner to a dump diffuser, piloted airblast nozzle with converging passage
cup to reduce smoke, and effusion cooled liner. No attempt was made to reduce NOx ,
and, therefore, the baseline combustor (identified Type 1 in Fig. 20) falls into the
pre-LEC category; this engine emissions test start date was March 31, 1994. Since
36 H. C. Mongia

50
23

Takeoff NOxEI, g/kg

LTO NOx, g/kN
45 22 Type 1

21 Type 3 Reduced Emissions

40 Type 2
20
19
35 Type 1
18
Type 3 Reduced Emissions
30 17
Type 2
16
25 15
15 16 17 18 19 20 21 15 16 17 18 19 20 21

Takeoff Pressure Rao Takeoff Pressure Rao

a. Landing takeoff LTO NOx versus takeoff b. Takeoff NOxEI versus takeoff pressure ratio.
pressure ratio.
50 45
Type 1
45 Type 3 Reduced Emissions 40

Idle COEI, g/kg

Idle COEI, g/kg

40 Type 2 35

35 30

30 25

25 20 Type 1
Type 3 Reduced Emissions
20 15
Type 2
15 10
15 17 19 21 23 0 2 4 6 8

Takeoff NOxEI, g/kg Idle HCEI, g/kg

c. Idle COEI versus takeoff NOxEI. d. Idle COEI versus idle HCEI.

Fig. 20 On checking RR design process consistency in transitioning RQL technology into pre-
RQL Type 1 AE3007 (engine emissions test start date March 31, 1994) through Type 3 Reduced
Emissions (June 6, 2001) and Type 2 engine emissions tested on August 28, 2002

the author left Allison during January, 1994, he doesn’t know the details about Type
3 Reduced Emissions (with emission test start date June 6, 2001) and Type 2 engine
emissions tested on August 28, 2002. As summarized in Fig. 20a, Type 3 Reduced
Emissions configuration did reduce LTO NOx at 17.9PR from 41.53 to 38.49 g/kN,
mere 7%. But the baseline combustor has some LTO data repeatability issues varying
between 41.53 and 45, approximately 8%. Type 3 engine model’s maximum takeoff
pressure ratio is 20.2 compared to the baseline’s being 17.9. The author can’t tell if
Type 2 combustor tested in 2002 is a predecessor of Type 3 tested in 2001. But, if
we do, then the intended reduction in LTO NOx happened first by increasing through
Type 2 mod followed successfully through Type 3 configuration giving only 8%
reduction in LTO NOx .
As explained next, the preferred output of reduced emissions combustor program
should be simultaneous reduction of takeoff NOx , idle CO and HC or as a minimum
reduction in takeoff NOx without adversely impacting idle CO and HC emissions.
Here, for the AE3007 emissions reduction effort, 8% reduction in LTO NOx was
followed by increased level of idle CO and HC emissions as summarized in Fig. 20;
a clear contrast to Figs. 7, 8 and 9. The author would be hesitant to categorize this
piece of emission reduction effort as spectacular.
A casual look at the NOx emissions data at the four operating points of the landing
takeoff cycle shows that the takeoff NOxEI is generally higher than that of the
Low Emissions Propulsion Engine Characterization Process 37

Fig. 21 Contributions of 31%

Takeoff NOx contribuon, %

NOx emissions during 29%
takeoff (top) and climbout
27%
(bottom) expressed as % of
LTO NOx 25% CFM56 TI
CF34-10E
23%
CF6-80C2
21% GE90
Trent XWB
19%
T1000
17% TPh5
T1994
15%
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao
53%
Climbout NOx contribuon, %

52%
51%
50%
CFM56 TI
49%
CF34-10E
48% CF6-80C2
47% GE90
46% Trent XWB
T1000
45%
TPh5
44%
T1994
43%
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao

climbout phase of the airplane. But the ICAO LTO cycle assumes that the takeoff
phase lasts for 0.7 min compared to 2.2 min for the climbout. It is therefore important
to know how much contributions to the LTO NOx emissions come during the takeoff
and climbout operations. Figure 21 answers this question for all the engines listed
in Table 3 wherefrom we see that the takeoff NOx emission contribution ranges
between approximately 19% and 29%. However, the climbout phase contributes
higher percentage of NOx emissions ranging between approximately 44% and 53%.
The remaining 17% to 27% contribution of NOx comes from the combined taxi-idle
and approach operation.
It is very hard to make any conclusions from the RR LTO NOx data scatter of
Fig. 19b. Therefore, we will first look at the GEAE/CFM engine data-driven mostly
by the fact that the author is very familiar with these engines. From Fig. 19a, we
see that the LTO NOx data of the five LEC engine models (CF34-10E, CFM56-5B
and CFM56-7B Tech Insertion (identified as CFM56 TI), CF6-80C2, and GE90-
110/115B identified as GE90 can be represented well by a linear equation fit, namely
LEC LTO NOx = 1.746 × PR − 8.289 with R 2 = 0.9799.
As discussed earlier in regard to Fig. 14, the slope of the NOx regulatory standard
CAEP1 and subsequent more stringent requirements for the medium and large high-
pressure ratio engines slope is 2.0. Therefore, if the engine combustors could have
38 H. C. Mongia

slope of ≤2.0, the NOx margin from the regulatory standards will not reduce with
increasing pressure ratios. Here, the LEC combustors have slope of 1.75.
Equally good is to have best-fit power law is obtained for the takeoff
NOxEI of these five LEC engine models, as summarized in Fig. 22a given by
LEC Takeoff NOxEI = 0.0544 × PR1.806 with R 2 = 0.9788.
As discussed earlier, the climbout operation contributes more NOx emissions
than the takeoff phase. Therefore, it will be desirable to check whether the NOxEI
correlation during climbout operation can be derived similar to that of the takeoff
NOxEI. Unfortunately, the engines’ climbout overall pressure ratio values unlike
takeoff’s are not available publicly. Additionally, the ratio of the climbout to takeoff
pressure ratios varies from engine to engine and varies as a function pressure ratio.

55
CFM56 TI
50
CF34-10E
Takeoff NOxEI, g/kg

45
CF6-80C2
40
GE90
35 y = 0.0544x1.806
LEC
30 R² = 0.9788
Power (LEC)
25
20
15
10
20 25 30 35 40 45
ICAO Takeoff Pressure Rao

a. Takeoff NOxEI versus takeoff pressure ratio described well by

.

40
CFM56 TI
35 CF34-10E
Climbout NOxEI, g/kg

CF6-80C2
30
GE90

25 LEC
y = 0.1256x1.4864
Power (LEC) R² = 0.9588
20

15

10
20 25 30 35 40 45
ICAO Takeoff Pressure Rao

b. Climbout NOxEI versus takeoff pressure

described well byratio
. X-axis value should
have been climbout pressure ratio (which is not available publicly), not takeoff pressure
ratio explaining qualitatively inferior performance of the best-fit power law.

Fig. 22 Average LTO NOx (Part a), takeoff NOxEI (Part b), and climbout NOxEI (Part c) versus
takeoff pressure ratio characteristic of the five GEAE and CFM LEC combustors
Low Emissions Propulsion Engine Characterization Process 39

Consequently, one cannot get the desired correlation for climbout unless we assume
that the ratio Takeoff PR/Climbout PR is essentially constant for these five LEC
engines. In that case, we can assume that climbout NOxEI is a function of takeoff
pressure ratio. As summarized in Fig. 22b, we see that the climbout NOxEI is also
reasonably good, namely, LEC Climbout NOxEI = 0.1256 × PR1.4864 with R 2 =
0.9588 (Fig. 23).
One obvious weakness in the LEC correlations for NOxEI is that these are heavily
weighted in favor of nominal 20–30 pressure ratio engines. Since Trent XWB’s
takeoff and climbout NOxEI are very close to that of the GE90 for its range of
takeoff pressure ratio. We, therefore, took the liberty of adding Trent XWB’s NOxEI
data to the LEC database. The resulting LEC correlations are weighted essentially
well for application in engines with PR ranging between 20 and 50. The following
NOxEI correlations will be used for the rest of this chapter:

LEC Takeoff NOxEI = 0.0563 × PR1.796 with R 2 = 0.9876

LEC Climbout NOxEI = 0.0977 × PR1.563 with R 2 = 0.9741

It is encouraging to see that all the Trent engine models’ rich-dome combustors
which has wide LTO NOx data (shown previously in Fig. 19b) when expressed in
takeoff NOxEI, as summarized in top part of Fig. 25 scatter very nicely around
the LEC takeoff technology curve (given LEC ± 2) except for the two engine
models shown within a rectangle. A similar behavior is observed for the Trent’s
climbout NOxEI except all but one model of Trent XWB falls on uniformly biased
slightly higher values than the LEC technology given approximately by LEC + 4.
Here, again the same two engine models stand out as outliers. However, the Trent
engine models show more scatter in their takeoff NOxEI data than the GE/CFM LEC
engine models. But one can’t help in concluding that in regard to both the takeoff
and climbout NOxEI both OEM’s rich-domes’ NOx emissions characteristics are
essentially similar. It appears that Trent engines’ takeoff NOxEI technology has not
improved since 1994 in spite of changing single wall film cooling configuration to
double-wall tiled construction (Fig. 24).
In summary, even though the landing takeoff emission characteristics are the right
metrics for evaluating impact of propulsion engines on the local air quality, high-
power NOx emissions indices provide better guidance in developing low-emissions
combustion technology and products. The rich-dome low NOx products known popu-
larly as LEC or RQL for all the modern engines produced by the three OEM’s produce
similar levels of takeoff and climbout NOxEI given by the following two correlations
which are applicable over a wide range of takeoff thrust and overall pressure ratios,
namely, 21 ≤ Takeoff Pressure Ratio ≤ 49, 77kN ≤ Takeoff Thrust ≤ 514 kN.

LEC Takeoff NOxEI = 0.0563 × PR1.796 with R 2 = 0.9876

LEC Climbout NOxEI = 0.0977 × PR1.563 with R 2 = 0.9741

40 H. C. Mongia

70
CFM56 TI y = 1.7462x - 8.2888
65 R² = 0.9799
CF34-10E
60 CF6-80C2

LTO NOx, g/kN

55 GE90

50 LEC
Trent XWB
45
Linear (LEC)
40

35

30
20 25 30 35 40 45
ICAO Takeoff Pressure Rao

a. LTO NOx showing clearly that LTO NOx of Trent X Wide Body is considerably smaller
than the trendline for the GE LEC combustion products.
60
CFM56 TI CF34-10E
55
CF6-80C2 GE90
50
LEC Trent XWB
Takeoff NOxEI, g/kg

45 Power (LEC)
40
35
y = 0.0544x1.806
30 R² = 0.9788
25
20
15
10
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao

b. Takeoff NOxEI data for Trent X Wide Body Trent XWB is only slightly lower than that
of GE90-110/115B.
45
CFM56 TI
40 CF34-10E
Climbout NOxEI, g/kg

35 CF6-80C2
GE90
30
LEC
25 Trent XWB
y = 0.1256x1.4864
20 Power (LEC)
R² = 0.9588

15

10
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao

c. Climbout NOxEI data for Trent X Wide Body, Trent XWB is literally sitting on the top
of GE90-110/115B leading to updated LEC correlations summarized in Fig. 24.

Fig. 23 Average LTO NOx (Part a), takeoff NOxEI (Part b) and climbout NOxEI (Part c) versus
takeoff pressure ratio characteristic of the five GEAE and CFM LEC combustors plus Trent X Wide
Body, Trent XWB
Low Emissions Propulsion Engine Characterization Process 41

Fig. 24 Takeoff NOxEI 70

CFM56 TI CF34-10E
(top) and climbout NOxEI
60 CF6-80C2 GE90
(bottom) versus takeoff

Takeoff NOxEI, g/kg

LEC Trent XWB
pressure ratio characteristic
50 Power (LEC)
of the five GEAE and CFM
LEC combustors plus Trent y = 0.0563x1.796
40 R² = 0.9876
X Wide Body, Trent XWB
30

20

10
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao
45
CFM56 TI
40 CF34-10E
Climbout NOxEI, g/kg

35 CF6-80C2
GE90
30 y = 0.0977x1.563
LEC
R² = 0.9741
25 Trent XWB

20 Power (LEC)

15

10
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao

6 High- and Low-Power Tradeoffs for Rich and Lean

Domes

When the author started working on an axially-staged partially premixing combustion

system in November 1974 [19], he went through three major design iterations before
settling on the one, followed by its fabrication and testing. The baseline design
achieved program goals for idle unburned hydrocarbons and simulated takeoff NOx .
The author’s low emissions combustion technology development and test experience
gained from the remaining five modifications described in [19] in addition to 5
modifications in [20] and another set of 3 refinements in [21] gave him first-hand
appreciation on how hard it is to have a low-emission combustion technology system
that meets simultaneously all the design requirements. That is when he learned that
getting low NOx at takeoff is easy but not simultaneously low idle COEI for preferable
limit of 30 COEI but not exceeding 45 COEI, as shown in Fig. 26 for the selected
eleven GE/CFM engine combustion products we have been discussing. Perhaps less
challenging is getting low idle HCEI limits of approximately 3 and 5, as summarized
in Fig. 27.
It’s very unusual to discover a low-NOx lean dome combustion product that is
comparable with rich-dome combustion products in regard to idle CO-takeoff NOx
42 H. C. Mongia

Fig. 25 Takeoff NOxEI 60

Trent XWB
(top) and climbout NOxEI
55 T1000
(bottom) versus takeoff

Takeoff NOxEI, g/kg

pressure ratio characteristic TPh5
50
of the Rolls Royce RQL T1994
LEC+2
combustion products LEC
45
compared with the updated
GE Low Emissions LEC-2
40
Technology trendline
35

30
30 35 40 45 50
ICAO Takeoff Pressure Rao
50
Trent XWB
45 T1000
Climbout NOxEI, g/kg

TPh5 LEC+4
40
T1994
LEC
35

30

25

20
30 35 40 45 50
ICAO Takeoff Pressure Rao

and idle CO and HC characteristics. But here, we have GEnx, the first lean-dome
to exceed or meet rich-dome’s idle CO and HC emissions. Our benchmark here is
the CF6-80C2LEC, and GEnx is even better. Even though CO and HC emissions
of LEAP-X are equally low, we don’t group all of their engine models with GEnx
and CF6-80C2 for the reasons obvious from the data comparison summarized in
Figs. 26 and 27. These figures show that three of the LEAP-1A and five of LEAP-
1B engine models fall outside of the two boxes, namely Box 1 and Box 2; the
former is preferred over the latter. Hypothesis proposed for conducting fundamental
investigations in order to improve LEAP-X emissions characteristic will be discussed
in the next section.
In summary, GEnx combustor makes the benchmark for the lowest idle CO-HC
and idle CO-takeoff NOx given by the following correlations:

Idle COEI = 32.183/Takeoff NOxEI0.223

Idle COEI = 10 + 7 × Idle HCEI

The last expression is slightly modified later in Sect. 7.

Low Emissions Propulsion Engine Characterization Process 43

7 Comparative Evaluation of Lean Domes’ Products

In this section, we start with Fig. 28 that provide emissions characteristics of

the two combustors, namely the LEC trendline for takeoff NOxEI, and GEnx for
showing tradeoff between high-power NOx and idle CO and HC emissions with the
corresponding expressions being:

55
LEAP-1A
50
LEAP-1B
45 CFM56-5B_TI
CFM56-7B_TI
LEC
Idle COEI, g/kg

40 CFM56-7B DAC
CFM56-5B DAC
35
Box 2
30
25
20
15
Box 1

10
10 20 30 40 50 60 70
Takeoff NOxEI, g/kg

a. CFM combustion products showing right NOxEI limit of 30.9 set by the CFM56-5B
Tech Insertion CFM565B_TI. Three engine models of the LEAP-1A and only one en-
gine model of the LEAP-1B fall under Box 1. Two engine models of CFM56-5B_TI
and one model of CFM56-7B_TI fall under Box 1; the rest of the models fall under Box
2 where all but three engine models of the CFM56 DAC belong too. The remaining
three models of the CFM56 DAC fall outside of Box 2.
50 GEnx
CF34-10E
45 GE90
40 GE90-DAC1
CF6-80C2
Idle COxEI, g/kg

Box 2
35

30

25 LEC
20
Box 1
15

10
10 20 30 40 50 60 70
Takeoff NOxEI, g/kg

b. GEAE combustion products. Right hand limit value of NOxEI set by the maximum
takeoff NOxEI value of 39.1 from the GE90-115B. The three models of GE90 DAC I
fall outside of Box 2. Box 2 contains all three engine models of GE90-110/115B, two
models of GE90 DAC I and essentially all models of CF34-10E. All engine models of
the CF6-80C2 and GEnx fall within Box 1.

Fig. 26 Tradeoff characteristics for idle COEI and takeoff NOxEI for the CFM (Part a) and GEAE
(Part b) engines followed by hypotheses-based LEAP-X emissions reduction in Part c; see [24, 25]
44 H. C. Mongia

50
GEnx
45
CF6-80C2
40 1A 1B GEnx
2 CFM56-7B_TI

Idle COEI, g/kg

35 LEAP-1A
30 LEAP-1B
CFM56-5B_TI
25

20
1
15

10
10 20 30 40 50 60 70
Takeoff NOxEI, g/kg

c. High-end NOxEI limits established for GEnx’s 46 NOxEI at 47.5PR (for 20%
lower NOxEI than LEC), 31 NOxEI at 31 PR (when LEAP-1A has same value of NOxEI as
the LEC), and 38 NOxEI at 38PR when NOxEI of LEAP-1B is comparable with that of LEC.
This means that only three engine models each of the LEAP-1A and LEAP-1B fall withing
the box 1 while others don’t exceed takeoff NOxEI of the LEC. On the other hand, all but
one of the GEnx engine models fall within box 1.

Fig. 26 (continued)

Fig. 27 Spray quality and 55

quenching effects expressed 50
through idle COEI and HCEI 45
for the CFM (L) and GEAE
Idle COEI, g/kg

40
(R) engines showing that
several engines fall in the 35
neighborhood of the lowest 30 LEAP-1A
LEAP-1B
idle HCEI-COEI curve given 25 2 CFM56-5B_TI
by Idle COEI = 20 CFM56-7B_TI
10 + 7 × Idle HCEI. Box 1 1
CFM56-7B DAC
15 CFM56-5B DAC
is preferred over Box 2
10
0 2 4 6 8 10
Idle HCEI, g/kg
50
GEnx
45 CF34-10E
GE90
40
Idle COxEI, g/kg

GE90-DAC1
35 CF6-80C2

30
25 2
20
1
15
10
0 1 2 3 4 5 6
Idle HCEI, g/kg
Low Emissions Propulsion Engine Characterization Process 45

70
LEC
60

Takeoff NOxEI, g/kg

50
y = 0.0563x1.796
40 R² = 0.9876

30

20

10
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao
a. Takeoff NOxEI versus takeoff pressure ratio of LEC combustion products given by

24

22 GEnx
Idle COxEI, g/kg

20 CF6-80C2

18

16

14
y = 32.183x-0.223
R² = 0.8967
12

10
10 20 30 40 50
Takeoff NOxEI, g/kg
b. Lowest idle COEI versus takeoff NOxEI characteristics of the lowest GEAE combustion
products are the rich-dome CF6-80C2 and lean-dome TAPS GEnx; the latter given by

60
y = 7.1701x + 15.87
50 R² = 0.9995
Idle COxEI, g/kg

40

30
GEnx
GE90-DAC1
20
CF6-80C2
y = 7.196x + 8.2309 LEAP-1A
10 R² = 0.9949
LEAP-1B
0
0 1 2 3 4 5
Idle HCEI, g/kg
c. Lowest idle COEI versus idle HCEI group comprises of the rich-dome CF6-80C2, lean-
domes TAPS GEnx and LEAP-1B given by

Fig. 28 A group of low-emissions (benchmark) combustion products selected for comparison with
lean-dome DAC and TAPS combustion products
46 H. C. Mongia

LEC Takeoff NOxEI = 0.0563 × PR1.796 with R 2 = 0.9876

32.183
GEnx Idle COEI = with R 2 = 0.897
Takeoff NOxEI0.223

CF6 − 80C2 Idle COEI = 8.2 + 7.2 × Idle HCEI with R 2 = 0.9949

One can’t miss similarity with GE90 DAC I Idle COEI = 15.9 + 7.2 ×
Idle HCEI with R 2 = 0.9995. This implies empirically that spray quality of the
swirl cups used in both of these engines are essentially similar. However, seeing
almost twice the value of intercept, 15.9 versus 8.7 implies empirically that there is
more CO quenching in the GE90 DAC I in spite of:
(1) Both engine dome splash plates are convection cooled, see Figs. 6 and 29.
(2) Idle operating conditions T3 and P3 of the GE90 are higher than those of the
CF6-80C2.
(3) The centerbody is undercooled.

Fig. 29 A schematic representation of the GE90 DAC I combustion system; reproduced from [4]
Low Emissions Propulsion Engine Characterization Process 47

These three factors lead empirically to suggest that the higher level of CO
quenching in GE90 DAC I might have been caused by the mixing of the main combus-
tion air of the inner annular passage and the outer annulus pilot air downstream of
the centerbody.
The main reason for selecting the LEC as the benchmark is that its correlations
for both the takeoff and climbout NOxEI give good agreement with all rich-dome
combustion products of the three major OEM’s, namely GE, CFM, and RR; refer to
Figs. 24 and 25.
The main reason for selecting GEnx as benchmark for tradeoff between idle COEI
and takeoff NOxEI, and idle CO-HC is that it has demonstrated lower values than the
best of GE’s low emissions combustion products in this regard, namely CF6-80C2,
refer to Figs. 26 and 27. And finally, GEnx is currently the lowest LTO NOx and
takeoff NOxEI combustion product.
We start with the DAC combustion products emissions performance keeping in
mind that these two concurrently running engine programs, one for the CFM56, a
nominal 30 pressure ratio engine, and another one for a new centerline 40 pressure
ratio engine were based on the outstanding technology programs conducted by the
best of combustion technologists of their era [5–8]. Both of these engine programs
were initiated before January 1994 when the author joined GE Aviation to help guide
a “world-class combustion section.” The first CFM56-5B DAC’s engine emissions
certification test started on June 15, 1994, followed by the CF6-80C2 LEC engine
start date of January 13, 1995, the GE90-DAC I on February 24, 1995, the CFM56-
5B DACII on August 1, 1996. The author who regarded only one out of the four-
engine programs success was furious. Subsequently, he insisted for more freedom
for conducting aerothermal design activities which he was given for the CFM56-
7B DAC and the resulting engine emissions certification start date of August 30,
1997. Back to back comparison between the two (CFM56-5B and CFM56-7B) DAC
engines’ CO, NOx, and HC emissions summarized in Figs. 26 and 27 clearly shows
that the author’s direct involvement didn’t have any impact on the outcome.
It should be pointed out that the turbomachinery technology activities of [8]
led to the layout and successful execution of the GE90 turbomachinery product
development while simultaneously the GE90 DACI combustion system turned out
to be a challenging tradeoff as discussed in the following paragraphs.
We will start with the CFM DAC first because it didn’t have to extend beyond
the engine operating pressure ratios of the two engine demonstrators reported
in [7, 8]. Obviously, comparison of any lean-dome combustion products needs
to be made with a benchmark rich-dome combustion system’s emissions which
in our case is summarized in Fig. 28a for the LEC takeoff NOxEI is given by
LEC Takeoff NOxEI = 0.0563 × PR1.796 with R 2 = 0.9876, idle COEI and
HCEI given by CF6 - 80C2 Idle COEI = 8.2 = 7.2 × Idle HCEI with R 2 = 0.9949
in Fig. 28c.
Logically speaking, one should select the CF6-80C2 LEC combustor’ data on
idle COEI versus takeoff NOxEI summarized in Fig. 28b. But we have chosen the
GEnx engine instead, the engine that was tested in 2009; because, if someone’s
rich-dome can’t compete with lean-dome’s idle COEI versus takeoff NOxEI, his
48 H. C. Mongia

grandmother may not like it. Therefore, GEnx’ expression GEnx Idle COEI =
32.183
Takeoff NOxEI0.223
with R 2 = 0.897 will be used as benchmark.
The three CFM DAC engine products gave only slight reduction in takeoff NOxEI
emissions compared to LEC, as summarized in Fig. 30a. The empirical correlation
combined for the both 30 and 40 overall pressure ratio OPR engines (CFM56 and
GE90) shows that DAC technology gave 8% higher NOx than the LEC technology
at 32 OPR. Even with the empirical correlation based the three CFM DAC engines
data, it gave us 4% higher NOx value than the LEC technology line. Compared to
the CF6-80C2 LEC data, DAC gave us 6% higher NOx . We have 38% higher NOx
with short GE90 DAC1 combustion system compared to LEC technology in addition
to inferior performance in regard to idle COEI versus takeoff NOxEI summarized
previously in Fig. 27 and 28. Back to back comparison with the lower-cost the Trent
engine emissions certified in September 1, 1994, the GE90 DACI gave 35% higher
NOx .
Poor emissions performance of the lean DAC products in spite of the best effort
put by the teams regarded very capable by the author who also thought very high of
the teams that developed the technology [5–8] should have normally dampened the
zeal for pursuing lean-dome technology development for propulsion engines. The
driving force for pursuing lean-dome technology was the author’s commitment to his
customer for developing dual annular TAPS combustion system for achieving 50%
reduction from LEC [1, 2]. A similar commitment was made to another customer for
transitioning the TAPS technology into the single annular TAPS GEnx combustion
product whose takeoff NOxEI is presented in Figs. 30b and 31 that include TAPS2
and TAPS3 technologies. Results are summarized in Tables 4, 5, 6 and 7.
Since the takeoff NOxEI results for the LEAP-1A and LEAP-1B engines are
significantly different from those of the GEnx (compare Fig. 30b, c), it’s perhaps
important to summarize the key aspects of the GEnx design process used by the
author who functioned as the team leader from conceptual design through first engine
to test (FETT) of GEnx completed by March 2006.
It should be mentioned that all of the GE and CFM LEC propulsion engines can
trace to its origin to the CF6-80C2 engine tested on January 13, 1995; similarly, for the
RR RQL engines to its origin the Trent engine family tested on September 1, 1994. In
summary, all of these LEC/RQL engines emissions certified between 1994 and 2017
are represented well by LEC Takeoff NOxEI = 0.0563 × PR1.796 with R 2 = 0.9876
implying that LEC/RQL should be considered a matured technology.
But that is not the case with the lean-dome propulsion engine products. The
development of the lean-dome DAC combustion products stopped after GE90-DACI
tested on February 24, 1995 and the CFM56-7B on August 30, 1997 because of
its less than stellar performance in regard to low-power HC and CO emissions in
addition to takeoff NOx emissions as summarized in Fig. 30a. It gave slightly lower
NOx in nominal 30 OPR (overall pressure ratio) engines than LEC, but significantly
higher in 40 OPR engines. When DAC’s correlation is extended to 49 OPR, it is
projected to have 76% higher takeoff NOx than its counterpart LEC engines.
TAPS technology initiated in 1995 was expected to address several deficien-
cies in DAC’s products including durability, shifting combustor exit radial profile
Low Emissions Propulsion Engine Characterization Process 49

70
LEC y = 1.110E-03x2.950E+00
R² = 9.772E-01
60 CFM56-5B DAC 38% increase

Takeoff NOxEI, g/kg

CFM56-5B DACII
50
CFM56-7B DAC

40 GE90-DAC I
y = 0.0563x1.796
Lean DAC R² = 0.9876
30

20

10
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao
st
a. Disappointing NOx performance of the 1 generation lean-dome DAC for 40 OPR en-
gine GE90-DAC I giving 38% increase in takeoff NOxEI emissions compared to LEC.

LEC
70
GEnx-1B
Takeoff NOxEI, g/kg

60 GE90-DAC I

50 19%

40 56%

30 y = 0.0563x1.796 63%
R² = 0.9876
20 y = 7.480E-05x3.453E+00
R² = 9.985E-01
10
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao
st
b. TAPS based GEnx combustion system gives 56-63% reduction from the 1 generation
lean-dome GE90-DAC I, but only 19% lower NOx than LEC at 47.5 OPR. Refer to Fig.
31 for TAPS2 and TAPS3 technology potentials.

70 LEC
LEAP-1A
Takeoff NOxEI, g/kg

60 LEAP-1B
39%
DAC
50

40

30 y = 0.0563x1.796
R² = 0.9876 *
20

10
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao
c. What went wrong with TAPS II combustion products of LEAP-1A and LEAP-1B? We
are back to NOx performance of the 1st generation DAC technology getting 39% higher
NOx than the LEC technology given by
.

Fig. 30 Takeoff NOxEI of DAC’s based engines (CFM56-5B DAC I, CFM56-5B DAC II, CFM56-
7B DAC and GE90-DAC I), LEC engines’ best-fit curve, TAPS based engines GEnx, LEAP-1A,
and LEAP-1B
50 H. C. Mongia

Fig. 31 Takeoff NOxEI 120

LEC
characteristics of five y=0.00111x2.95
technologies, DAC based on 100 GEnx-1B y = 0.0563x1.796

Takeoff NOxEI, g/kN

DAC y = 7.48E-05x3.4534
four engines data covering
80
24.6 ≤ P R ≤ 40.4, six LEC TAPS2 76%
engines data covering 60 TAPS3
21.4 ≤ P R ≤ 48.6 and
GEnx engine data covering 40
35.5 ≤ P R ≤ 47.5
20 79%

0
20 25 30 35 40 45 50
ICAO Takeoff Pressure Rao

Table 4 GEnx technology

Technology or Goal (%) Achieved or expected (%)
exceeded its goals for
Product
reducing takeoff NOxEI from
GE90-92B DAC I at 40.4 Rich-dome LEC 25 27
overall pressure ratio technology
GEnx Technology 50 56

Table 5 Reduction from

Product or TAPS Demonstrated or Takeoff pressure
LEC technology and the
technology expected (%) ratio
corresponding overall takeoff
pressure ratio GEnx-1B64/P2 39 40.9
TAPS2 Goal 70 40.9
TAPS3 Goal 85 40.9

Table 6 Reduction from

Product or TAPS Demonstrated or Takeoff pressure
LEC technology and the
technology expected (%) ratio
corresponding overall takeoff
pressure ratio GEnx-1B76/P2 19 47.5
TAPS2 Goal 59 47.5
TAPS3 Goal 80 47.5

Table 7 Predicted takeoff

Technology Takeoff NOxEI % Reduction from LEC
NOxEI at 49 takeoff pressure
ratio and % reduction from LEC 61.11 0
LEC technology DAC 107.50 −76%
GEnx 51.38 16%
TAPS2 25.69 58%
TAPS3 12.85 79%
LEC technology curve based on engines data covering 21.4 ≤
P R ≤ 48.6 and GEnx engine data covering 35.5 ≤ P R ≤ 47.5
Low Emissions Propulsion Engine Characterization Process 51

from idle to maximum power and attendant adverse impact on fuel burn, higher
idle CO/HC emissions, and higher high-power NOx for the next-generation 40 OPR
engines. Summary of TAPS technology development of the two full-scale combus-
tors, Fig. 32, for potential applications in the next-generation medium size and large
propulsion engines is provided in [1, 2] including two next-generation TAPS tech-
nologies colloquially identified as TAPS2 and TAPS3 implying TAPS2Takeoff NOx =
TAPS1Takeoff NOxEI /2 and TAPS3Takeoff NOx = TAPS2Takeoff2
NOxEI
. The two TAPS2 and
TAPS3 curves shown in Fig. 31 are simply takeoff NOxEI expression of GEnx
(namely, GEnx Takeoff NOxEI = 7.480E − 05 × PR3.453 ) divided by 2 and 4,
respectively.
Two of the curves (namely, SAC TAPS engine data, and DAC TAPS full annular rig
FAR) shown in Fig. 33 are for the TAPS configurations shown in Fig. 32. Discussion
on Fig. 33 will follow after discussing Fig. 34 through xx.
When we started with conceptual design of GEnx, the author had data on:
• The Tech56/CFM TAPS combustion system that successfully completed engine
demonstration including emissions, performance, operability, and endurance.
• The DAC TAPS combustion system from its full annular rig test rig at T3,
corrected flow rates, fuel/air ratio, and P3 for all the GE90 engine operating condi-
tions except P3 at takeoff was slightly lower. This data set included emissions,
performance, operability, and liner wall temperature levels and gradients.
During the conceptual design of GEnx that started in 2004, the team had at its
disposal stark reminder that the geometric details of the TAPS’ pilot mattered for
nominal 30 OPR engines, see Fig. 34. Hypotheses for the CFM TAPS’s higher NOx
emissions compared to DAC TAPS (see Fig. 33) while simultaneously lower idle CO
and HC of the latter, see [1], were:
1. The CFM TAPS flame was anchored off the pilot’s lip recirculation zone LRZ
that was discovered later through CFD simulation in 2007, see Figs. 15 and 16,
and experimentally in 2008, see Fig. 35.

Fig. 32 Schematic layout of TAPS combustion systems; (L) Tech56/CFM TAPS combustion
system, (R) GE90 DAC TAPS combustion system. Reproduced from [1]
52 H. C. Mongia

10
LEC Engine Data
SAC TAPS Engine Data
DAC-TAPS FAR Data
TAPS FAR Data
1
TAPS Engine Data
TAPS3 FT Data
1
NOxEI/P30.5

0.1

0.01
600 700 800 900 1000 1100 1200 1300
T3 (F)

Fig. 33 Typical comparison between a rich-dome low-emission √ sea-level operation compared with
three TAPS1 configurations expressed in terms of N O x E I / P3 versus combustor inlet temperature
T3 compared to TAPS1 and TAPS3 results. There are two TAPS engines data identified as SAC
TAPS and TAPS, the latter has data from a full annular rig (FAR) that gives good agreement
with engine data. Refer to discussion on Fig. 32 TAPS3 takeoff NOxEI curve regarding projected
operational range of TAPS3 based engine which allows variable pilot fuel flow split capability

Fig. 34 Pilot exit outer-wall details US Patent 6,865,889, May 19, 2005 (left), “Good Karma, Good
Coke” appropriately phrased by Al Mancini, improved idle efficiency from mid-70% to 97%+ on
the engine test stand. Reproduced from [2]
Low Emissions Propulsion Engine Characterization Process 53

Fig. 35 Schematic representation of TAPS mixer and flow characteristics; here RZ means recir-
culation zones caused by pilot P, pilot lip L, and sudden expansion corner C formed between the
TAPS mixer and combustor dome and liner wall, identified, respectively, as PRZ, LRZ, and CRZ.
Reproduced from [17]

2. Mainly because of the smaller size of the DAC TAPS mixer compared to the
CFM TAPS, its flame might be a lifted flame with attendant reduction in NOx
emissions. It was verified numerically in 2007, see Figs. 15 and 16.
3. Lower idle CO and HC emissions of DAC TAPS compared to the CFM TAPS
could be attributed to the former’s higher idle T3 and P3 conditions.
4. But wouldn’t GEnx TAPS mixer that is bigger than the CFM TAPS lead to its
flame anchoring off LRZ? Options:
(a) Go without the pilot exit lip features of the CFM TAPS, Fig. 34.
(b) Put more emphasis on TAPS2 mixers.

Therefore, the first pizza-pie full-scale annular combustor rig testing for GEnx
included three TAPS2 concepts and only one TAPS1 concept intended for GEnx.
Thanks to good secular Karma deeds of the GEnx TAPS team, all three TAPS2
concepts literally fell along the intended TAPS2 characteristics as shown in Fig. 31.
TAPS2 configurations had some durability challenges which didn’t need to be
addressed because the TAPS1 NOx characteristics fell on the top of DAC TAPS.
How did that happen? The author has no idea.
TAPS3 conceptual concept screening, selection of a particular design, single-cup
test rig and combustor design, fabrication, and instrumentation completed by the
GEnx team was followed by testing at the NASA Glenn Research Center covered by
Raghavan Pandalai. The parametric testing at simulated cruise condition, P3 = 400
psi, T3 = 1000 °F and FAR = 0.0273 included variations in the pilot fuel flow split
for the reasons explained in Sect. 9. According to Raghavan Pandalai, the author told
54 H. C. Mongia

him, “If you don’t get test results in line with my TAPS3 expectations, you better
hide yourself in a cave.”
As hoped for and summarized in Fig. 3 of [15], the pilot fuel flow split had huge
impact on NOxEI emissions that ranged between 0.2 g/kg and 7.5 g/kg. It should be
obvious that TAPS3 NOxEI less than 1.0 at 400 psi, 1000 °F T3 and 0.027 FAR makes
it highly competitive with dry low NOx mixers used in the LM2500 and LM6000
aeroderivative engine, see [32].
It is, therefore, very important to know for any TAPS mixer configuration the
actual fraction of the pilot fuel flow rate and that of the main participating in its
so-called flame stabilizing region(s) and their residence time(s), see Fig. 35. TAPS’
emissions and performance are impacted strongly by its degree of mixedness, frac-
tion of pilot and the main injectors’ fuel flow rates, characteristics of the flowfield
including the recirculation zones (PRZ, LRZ, and CRZ) and the mixing layers. CFD
simulations are not accurate enough to sort through the complexity of the TAPS
flowfield characteristics. Al Mancini, the primary contributor on the TAPS injector
technology development, appropriately cast the right-hand side of Fig. 34 as “good
Karma and good Coke”.
A second set of data for verification of TAPS3 curve is from the sector rig testing
results reported in [31]. This sector test program included testing of LTO cycle points
of a next-generation engine cycle. The rig test results of LTO NOx value of 20.7 g/kN
and 18.9% CAEP6 standard were used to calculate its ICAO takeoff pressure ratio of
55. It’s interesting to notice that our TAPS3 expression recklessly extrapolated from
46.9 to 55 OPR gives takeoff NOxEI of 19.1 g/kg compared to the data, 17.6 g/kg,
reported in [31].
In summary, for any future TAPS-based combustion products when properly
designed, its takeoff NOxEI should fall between data of the GEnx, TAPS2, and
TAPS3 curves shown in Fig. 31. The experience gained from the GEnx and LEAP-X
TAPS combustion products indicates that idle COEI and HCEI should fall between
the data bands presented in Fig. 28c. In cases of high takeoff NOx values (Fig. 30c),
the LEAP-X team should:
• Investigate fundamental flow characteristics of LEAP-X TAPS mixer, Fig. 35. But
the team should pay attention to the lessons learned by the author as summarized
in [26].
• Extend the DRA methodology for rich domes summarized in Sect. 8 to TAPS
combustions systems including CFM TAPS, DAC TAPS, GEnx, sector test results
reported in [31], LEAP-1A and LEAP-1B.
Hopefully, this process will get the LEAP-X back to the right path for producing
low emissions TAPS design.
Low Emissions Propulsion Engine Characterization Process 55

8 Engine Emissions Data Reduction and Analysis DRA

The author has relied on qualitative guidance from CFD simulations but subject to
two verification criteria, namely hypotheses’ expectations and empirical data correla-
tions; the latter is discussed first because it has direct relevance to the curve identified
as 1 in Fig. 33. √
Combustion engineers use empirical correlations, like NOxEI/ P3 versus T3
from a typical LEC engine operating at sea-level from idle to maximum power
shown in Fig. 33. Moreover, instead of using best-fit expressions commonly used in
research papers (namely, power law, exponential, logarithmic, or linear functions),
industry uses 4th or 6th order polynomials in order to get the highest possible values
of R 2 . For example, the LEC engine data in Fig. 33 when fitted with a 4th-order
polynomial gave R 2 = 0.9991.√ On the other hand, the power-law used for the TAPS
engine data gave NOxEI/ P3 ∝ T38.497 with R 2 = 0.9887. The latter approach may
not provide good σ value; but it surely sends a strong warning against extrapolating
empirical correlations. That was the root cause for failure of seeing through the fog
for estimating a priori NOx emissions of the GE90-DAC I.
The author had the same fear for relying on the Tech56/CFM TAPS NOxEI
extrapolated curve 1 in Fig. 33 for scaling up to the higher operating pressure ratio
GEnx. On the other hand, DAC TAPS NOxEI data (with 60 TAPS mixers) was
significantly lower than the Tech56/CFM TAPS with only 20 mixers. But we wanted
to use only 22 mixers in GEnx in order to keep its acquisition cost-competitive with
counterpart rich-domes.
Since most of the semi-empirical and semi-analytical correlation works by the
research community are mainly focused on rich-dome emissions and operability, we
will focus on this topic while fully recognizing that it may not have much relevance
to TAPS flames.
Prof. Lefebvre has done outstanding service for the gas turbine engine combustion
community by offering several semi-analytical correlations for NOx , CO, HC, LBO
and ignition [10] followed by further refinements by Rizk and Mongia [11]. We will
focus on NOx correlation works first.
According to Lefebvre [10], NOx emission index from propulsion engine
combustors can be correlated by:

NOxEI = 9E − 8P30.25 VNOx exp0.01Tst /(W a36 TP Z ) (1)

Here T PZ is calculated from the primary zone fuel/air ratio, FARPZ , which is
defined as FAR36 /f . f is the fraction of combustor air participating in the primary
zone. This expression is very simple (which requires best fitting the engine data with
two empirical constants, V NOx and f ) and elegant for qualitative assessing different
combustors in regard to NOx emissions.
It, however, does not have the accuracy level desired by the designers, as pointed
out by Danis, Pritchard, and Mongia [10]. Moreover, it can predict trends very
56 H. C. Mongia

Fig. 36 Typical comparison

between test rig NOxEI data Matched with Eq. 1
at a given P3, T3, and Wa3
Data
conditions with parametric
variation in combustor’s
overall fuel/air ratio and

NOxEI
correlation given by Eq. 1.
Source [12]

Combustor Fuel/Air Ratio

different from data, as shown in Fig. 36 which is very relevant for developing
rich-dome combustion technology and products.
Rizk and Mongia [11] recast the Lefebvre’s expression into the following
equation:

NOxEI = 0.459E − 8P30.25 τNOx exp0.01Tst (2)

, which implies that with a NOx characteristic time τNOx and the primary zone
fraction air f one should be able to fit the NOxEI data.
Rizk and Mongia [11] gave an additionally improved correlation, namely
 0.5
0.15E16 τNOx − 0.5τev exp− Tst
71000

NOxEI = (3)
P30.05 (P/P3 )0.5

The author likes this correlation because his data from the early 1980s could be
interpreted only when he assumed the pressure exponent equal to zero.
The author also felt that the root cause for poor agreement with engine data could
be the overly simplified assumption of a constant value of the NOx characteristic
time τNOx .
This was addressed in [16] by modeling five production engines CFM56-2/3,
CFM56-5/7, CF6-80C2 PreLEC and LEC, and GE90 DAC2 by using a 3-reactor
model shown in Fig. 37 which allows independent variation of three characteristic
times τ1 , τ2 and τ3 . Embedded in these characteristic times are the spray evaporation
times which takes into consideration the spray characteristics of the selected product
combustors and their engine operating conditions from idle to maximum power.
As per the standard practice unique to each OEM, the airflow distribution around
the combustion system is known. There can be several ways of estimating the axial
variation of the equivalence ratios φ 1 , φ 2 and φ 3 as function of power setting. Perhaps,
Low Emissions Propulsion Engine Characterization Process 57

Primary NOx Secondary Dilution

Reactor (1) NOx Reactor CO Reactor

Fig. 37 Three perfectly stirred reactor model used for conducting Data Reduction and Analysis
DRA of engine sea-level emissions data from idle to maximum power. Here, we assumed τ1 = τ2 .
The equivalence ratios φ1, φ2, and φ3 of the selected five production engines were estimated from
the combustor airflow distributions and a finite set of axial stations

one-dimensional heat transfer analysis can be used as guide. But the process must be
consistent across all the combustors being analyzed. Similarly, a standard practice
must be used regarding the location and number of axial stations selected; Fig. 38
can be used as an illustration. It should be made clear that effort described in [16]
was called Data Reduction and Analysis, DRA with the main objective being can
any model patterned after Equations 1 or 2 get good agreement with HC, CO, and
NOx emissions data of rich-dome propulsion engines from minimum to maximum
power?
Readers should refer to [16] for details; here, we summarize the key findings.
Figure 39 shows that the calculated τ1 and τ2 from idle to maximum power will have
to change in order to match the data as summarized in Fig. 39 for NOxEI and COEI.

Fig. 38 Typical cumulative axial distribution of combustion zones with power setting
58 H. C. Mongia

Fig. 39 In order to match engine emissions data for specified variation in φ 1 , φ 2 and φ 3 , τ 1 and
τ 3 vary from idle to maximum power for the five selected production engines

When we went with the average but distinct values of τ1 and τ2 for each of the five
combustors, the agreement with the data was very poor.
In summary, semi-analytical emissions models with one constant value of
τNOx , VNOx and or fraction of combustion air participating in the primary is expected
to give only qualitative agreement with engine emission data from idle to maximum
power during its sea-level operation. Therefore, state of the art semi-analytical models
are not good enough to reduce sea-level engine emissions data to calculate LTO emis-
sions characteristics. That is the main reason why OEM’s continue to use an approach
similar to Fig. 33 for emissions certification. But, this approach should not be used
for calculating mission NOx emission or for extrapolating to predict emissions from
next-generation higher pressure ratio engines.
Low Emissions Propulsion Engine Characterization Process 59

Fig. 40 The data reduction and analysis DRA proposed in [29] matches NOx and CO emissions
data of five production engines

If one were to compare rich-dome with lean-dome combustion product emissions,

the criteria used by the author since the middle 1970 [1–4] is not enough; namely
takeoff NOx versus P3, idle CO versus takeoff NOx , and idle CO versus idle HC.
Therefore, an attempt was made to check the validity of the following assumptions
for the seven rich-dome combustors out of the twelve overviewed in [15].
a. For the entire engine operation from idle to maximum thrust at sea-level, NOxEI
is a function of stoichiometric flame temperature, COEI and HCEI are functions
of NOxEI.
b. Assumption (a) can be extended to the propulsion engines’ entire flight envelope
from sea-level to maximum altitude covering their minimum to maximum thrust
capabilities. Altitude emissions data reported by Diehl and Biaglow [33] was
used for this exercise.
60 H. C. Mongia

Readers should refer to [17, 18] for details; here, we have reproduced summary
results as presented in Figs. 41 and 42. Agreement between correlations and data
was good for all the seven rich-dome combustion products; but the quality of engine
emissions was considered to be better for Combustors 1, 5, 6, and 7 which are covered
in Fig. 41.
NOxEI correlation versus data, along x-axis, fall between error band of ±4% and
the resulting standard deviation σ of 2% which is considered comparable with the

Fig. 41 3rd generation of 80 Comb 1

semiempirical correlations
for the entire engine Comb 5
operating points from idle to 60 Comb 6
σ = 2%
maximum thrust. Comb 7
Reproduced from [17]
40 4%
-4%

20
NOxEI Correlaon vs. Data
0
0 20 40 60 80
100 Comb 1
Comb 5
σ =1.0 EI
80 Comb 6
Comb 7
60 2
-2
40

20
COEI Correlaon vs. Data
0
0 20 40 60 80 100
10 Comb 1
Comb 5 σ =0.2 EI
8 Comb 6
Comb 7
6 0.4
-0.4
4

2
HCEI Correlaon vs. Data
0
0 2 4 6 8 10
Low Emissions Propulsion Engine Characterization Process 61

Fig. 42 3rd generation of 20

semiempirical correlations PerfectFit NOxEI
for the TFE731-2 engine
operating points from 1.076
minimum to maximum thrust 15
0.924
range at simulated altitude
from 640 m to 13,200 m.
The root-mean-square errors 10
for NOx , CO and HC are
3.8%, 1.43EI and 1.04EI,
respectively. Reproduced
5
from [18]

NOxEI correlaons vs. data

0
0 5 10 15 20

COEI
60
PerfectFit

50 2.85
-2.85
40

30

20

10
COEI correlaons vs. data
0
0 10 20 30 40 50 60

20
HCEI
PerfectFit
2.09
15
-2.09

10

HCEI correlaons vs. data

0
0 5 10 15 20
62 H. C. Mongia

TFE731-2 Turbofan Engine Cross-secon

Fig. 43 The TFE731-2 engine cross-section and its reverse-flow annular combustion system.
Reproduced from [19]

experimental σ values of these four combustors. When it comes to assessing the

quality of correlation for CO and HC, the author prefers to express the error bands in
terms of ±δCOEI and ±δHCEI along with the corresponding σ values, as illustrated
in the middle and bottom parts of Fig. 41.
In summary, the approach recommended in [17] can reproduce data very well for
all the seven combustors. For the four-engine combustors with good quality emissions
data, the approach achieved standard deviation σ values of 2%, 1.0 EI and 0.2 EI for
NOx, CO, and HC, respectively, for the entire engine operating points from idle to
maximum thrust.
It should be emphasized that the emissions data quality for all engine emissions
certification programs is expected to be much superior than generic altitude engine
test programs similar to the one reported in [33]. Moreover, this engine, small turbofan
TFE731-2, Fig. 43, designed in the late 1960s and production start in the early 1970s
was going through its growing pains. It is small engine, 15.6 kN rated takeoff thrust
with a reverse-flow annular combustor fueled by radially inserted fuel nozzle. The
root-mean-square errors for NOx , CO and HC are 3.8%, 1.43EI, and 1.04EI, respec-
tively, through somewhat higher than for the four medium and large size engines
covered in Fig. 41 (namely, σ values of 2%, 1.0 EI, and 0.2 EI) should be regarded
quite good considering that it covers engine operation from sea-level to 13,200 m.
In summary, the exploratory investigations summarized in [17, 18] offers an
approach based on fundamentals that can be defended significantly better than
the currently practiced OEM’s approach. Once this approach is refined to meet
OEM’s expectations, it can be used to extrapolated for predicting emissions from
the next-generation rich-dome combustion products. Obviously, a lot needs to be
done for extending this approach to lean-dome combustion product, present, and
next-generation.
Low Emissions Propulsion Engine Characterization Process 63

9 TAPS Technology NOx Emissions Potentials

Lean-dome combustion products for propulsion engines (CFM56-5B, CFM56-7B,

GE90, GEnx, and LEAP-X) with varying degree of partial premixing of fuel/air
within the mixers comprised of two or more co- or counter-rotating swirlers and
nearly perfectly premixed aero-derivative industrial engines (LM2500 and LM6000)
starting in the late 1990s have performed safely while simultaneously meeting
combustion systems’ design requirements including cost of ownership, performance,
emissions and operability throughout the operating range of the engines in use without
encountering flashback and/or autoignition within the mixers, a generally mentioned
area of concern in fundamental research publications. Practical, affordable, and
durable solutions for mitigating combustion dynamics when or if encountered during
the last phases of product development have also been performing reliably [28–30].
Picking back from Fig. 30a where we see setback (for the GE90-DAC I) and
less than stellar reduction in NOx for the CFM56-5/7B with the DAC technology,
and in Fig. 30c for LEAP-X where a similar setback is encountered with the TAPS
technology. While simultaneously we see from Fig. 30b superb emissions reduction
with GEnx TAPS combustion system and 31 the TAPS technology luring poten-
tials for reducing takeoff NOxEI up to 80% compared to the rich-dome LEC tech-
nology. Is the LEAP-X setback due to the team’s inability to “see through the fog”,
micromanagement by burned-out combustion specialists or lack of tools?
Here, we pick up the most boring part, a commonly accepted OEM’s process for
LTO NOx calculation from sea-level engine emissions data represented in Fig. 33
Here, engine NOxEI data has been corrected to standard humidity via an agreed to
empirical correlation with some nagging questions still unresolved. The assump-
tion NOxEI/P3n=0.5 ∝ T3 over the entire engine operating line is not valid.
Even for the takeoff or climbout P3 and T 3 conditions, it is not unusual to see
0.3 ≤ n ≤ 0.6 for different rich-dome combustors. Applicability of this approach
for estimating mission NOx emissions which is an order of magnitude more than
LTO NOx emissions are fraught with several questions.
The alternate approach [16–18] is consistent with fundamentals of NOx formation
[34, 35] allowing linear interpolation between engine data points for calculating
NOx emissions at each of the four LTO operating conditions of a reference engine
running in standard ambient conditions with reference fuel, namely, stoichiometric
flame temperatures at takeoff, climbout, approach, and idle. No need to know NOx
pressure exponent n which can vary with different rich- and lean-domes. No humidity
correction is needed. Effect of fuel properties that can vary even for the same test
facility [36] on NOx emissions can be corrected for the reference fuel agreed to by
all OEM’s. The sea-level engine emissions data from minimum to maximum power
can be used for calculating NOx emissions for the airplane’s entire flight envelope.
Tremendous amount of insight is provided when engine NOx emissions data from
the seven rich-dome product combustors are plotted as function of 1000/Tst , Fig. 44
from [7].
64 H. C. Mongia

Fig. 44 NOxEI versus 70

1 2 3 4
1000/T st characteristics of 60 5 6 7
the seven rich-dome 50
combustors. Source [15, 17].

NOxEI
40
Here, T st is the
stoichiometric flame 30
temperature which is a 20
function of P3 , T 3 and 10
thermodynamic properties of 0
the fuel 0.205 0.215 0.225 0.235
1000/Tst
a. For the entire sea-level engine operating lines.
65
1 2 3 4
60 5 6 7
55
50
NOxEI

45
40
35
30
25
20
0.207 0.209 0.211 0.213 0.215
1000/Tst
b. Zoomed-in high-power operation.
28
1 2 3 4
26
5 6 7
24
22
20
NOxEI

18
16
14
12
10
8
0.215 0.217 0.219 0.221 0.223 0.225
1000/Tst
c. Zoomed-in mid-power operation.
10
1 2 3 4
9 5 6 7
8
7
NOxEI

6
5
4
3
2
0.225 0.227 0.229 0.231 0.233 0.235 0.237
1000/Tst
d. Zoomed-in low-power operation.
Low Emissions Propulsion Engine Characterization Process 65

Figures similar to Fig. 44 for GE90-DAC I, GEnx, LEAP-X, TAPS2, and TAPS3
with the pilot fuel flow split as a variable will provide the insight needed to sort
through the root causes leading eventually back to continuing NOx reduction by
using cleverly TAPS technology.
The author felt that by controlling pilot fuel flow split with airblast fueled pilot,
Fig. 45a is better than dual-orifice pressure atomizers shown in Fig. 45b especially for
the next-generation TAPS combustion products. But if the details of air atomizing
devices lead to adversely impacting the intents of the low NOx performing flow
characteristics, Fig. 35, the author wouldn’t be surprised by their consequences.
TAPS figure similar to Fig. 44 for all the TAPS configurations (CFM TAPS,
DAC TAPS, LEAP-1A, LEAP-1B, 55 pressure ratio engine TAPS technology sector
[31], flame tube TAPS3 [15], GEnx and its three FAR1 TAPS2 configurations) will
provide the insight required to “see through the fog” created by the results of LEAP-
1A and LEAP-1B. The TAPS figure can also be used to check the relevance of
fundamental investigations [26] and resulting phenomenological description of the
critical processes identified in Fig. 35.
In summary, the author feels that TAPS potentials as expressed in Figs. 31 and 33
and summarized in Table 7 are achievable, and regards the LEAP-X setback simply
a stark reminder for the team that lean-dome product development is a continuing
evolution process.

(a) Airblast fueled TAPS pilot (b) Dual-orifice fueled TAPS pilot

Fig. 45 Airblast or dual-orifice fueled TAPS pilot (reproduced respectively from US patents
8,171,735 and 8,387,391) with details adversely influencing TAPS flowfield characteristics (see
Fig. 35 can lead to setbacks)
66 H. C. Mongia

10 Summary

All original equipment manufacturers OEMs of the propulsion engines have been
working hard since the early 1970s for reducing undesirable exhaust gaseous emis-
sions, oxides of nitrogen NOx , unburned hydrocarbons HC and carbon monoxide
and smoke. The regulatory agency in conjunction with OEMs have established good
metrics, the landing takeoff cycle LTO emissions for quantifying impact of propulsion
engines’ pollutants on the local air quality. OEM’s made rapid progress in reducing
HC, CO, and smoke emissions by the middle 1980s and have successfully applied
their rich-dome low-emission combustion LEC technology including NOx devel-
oped during the middle 1990s across their entire product line while simultaneously
improving safety, cost of ownership, and operability range.
While LTO emissions are the right engine-level metrics, the tradeoffs between
low NOx at high-power versus low-power CO and HC emissions can be quantita-
tively assessed by comparing characteristics of takeoff NOxEI versus engine overall
pressure ratio, idle COEI versus takeoff NOxEI and idle COEI versus idle COEI. 16
engines of the three OEM’s, CFM, GE, and RR investigated using publicly available
ICAO engine emissions data have shown that rich-dome low NOx products known
popularly as LEC or RQL for all the modern engines produced by the three OEM’s
produce similar levels of takeoff and climbout NOxEI given by

LEC Takeoff NOxEI, LEC TO = 0.0563 × PR1.796 with R 2 = 0.9876 (4)

LEC Climbout NOxEI, LEC CO = 0.0977 × PR1.563 with R 2 = 0.9741 (5)

The RR Trent engines takeoff and climbout NOxEI data fall within the following
data bounds:ss

LEC TO − 2 ≤ Trent engine Takeoff NOxEI ≤ LEC TO + 2

LEC CO ≤ Trent engine Climbout NOxEI ≤ LEC CO + 4

These NOx correlations are applicable over a wide range of takeoff thrust and
overall pressure ratios, namely, 21 ≤ Takeoff Pressure Ratio ≤ 49, 77 kN ≤ Takeoff
Thrust ≤ 514 kN.
The following correlations for idle COEI and HCEI should be used as entitlements
for assessing both rich-dome and TAPS emissions capability:

32.183
Idle COEI = with R 2 = 0.8967 (6)
Takeoff NOxEI0.223

Idle COEI = 7.199 × Idle HCEI + 8.231 with R 2 = 0.9949 (7)

Low Emissions Propulsion Engine Characterization Process 67

The use of these correlations combined with the methodology [23, 24] and appli-
cable CFD tools [9, 15] has led successfully to evolving generation of low emissions
rich-dome combustors with insignificant reduction from the takeoff NOxEI expres-
sion (4) since the early 2000s. Therefore, any claim in LTO reduction in recently
certified propulsion engine should be attributable to the reduction in LTO fuel burn
and not advances in rich-dome combustion technology. Spray quality technology
should be considered matured in addition to well-practiced rich dome combustors
scale-up. Then, COEI values higher than Eq. 5 may be due to overcooled and/or
unoptimized dome and/or primary zone combustion structure.
TAPS combustion system scale-up should likewise be considered straightfor-
ward for teams that make proper use of the extensive database collected over the
last 20 years complimented by comprehensive combustion system analysis and
relevant fundamental investigation guided by Data Reduction and Analysis DRA
approach [17, 18], a significant advance over the current “black box” OEM’s preferred
approach. Moreover, this approach is applicable for estimating mission NOx emission
that is generally an order of magnitude higher than LTO NOx emission.
Passively controllable air atomizing fuel insertion system for the TAPS’ pilot is
preferred over hybrid or dual-orifice for obvious reasons. Variable pilot fuel flow split
(% of total fuel flow rate) with pilot swirlers designed in conformance with TAPS
low-emissions flow characteristic should be able to address operability challenges
while simultaneously meeting takeoff NOxEI given by Eq. 7 within ±2 EI meeting
or exceeding idle COEI vs. takeoff tradeoff given by Eq. 5 and idle emissions by
Eq. 6.

GEnx Takeoff NOxEI = 7.480E − 05 × PR3.453 with R 2 = 0.9985 (8)

In order to ensure that TAPS engine takeoff NOxEI is at least 25% lower than
its counterpart LEC combustion system, TAPS1 based GEnx technology should be
limited to takeoff pressure not exceeding 45.
The TAPS2 based combustion products, Eq. 9, should be used for 45 ≤ PR ≤ 55
with resultant reduction of 50% from LEC at 55 PR.
When takeoff pressures are higher than 55, TAPS3 based combustion product,
Eq. 10 development should be pursued in order to guarantee at least 50% reduction
from LEC.

TAPS2 Takeoff NOxEI = GEnx Takeoff NOxEI/2 45 ≤ PR ≤ 55 (9)

TAPS3 Takeoff NOxEI = GEnx Takeoff NOxEI/4 PR > 55 (10)

68 H. C. Mongia

References

1. Mongia HC (2003) TAPS—a fourth generation propulsion combustor technology for low
emissions. AIAA2003–2657
2. Mongia HC (2011) Engineering aspects of complex gas turbine combustion mixers Part V: 40
OPR. AIAA2011–5527
3. Mongia HC (2011) Engineering aspects of complex gas turbine combustion mixers Part IV:
Swirl cup. AIAA2011–5526
4. Mongia HC (1997) Recent progress in low-emissions gas turbine combustors. ISABE
5. Bahr DW, Gleason CC (1975) Experimental clean combustor program Phase I—final report.
NASA CR-134737
6. Gleason CC, Roger DW, Bahr DW (1976) Experimental clean combustor program Phase II—
final report, NASA CR-134971, August 1976
7. Gleason CC, Bahr DW (1979) Experimental clean combustor program Phase III—final report.
NASA CR-135384, June 1979
8. Stearns EW et al (1982) Energy efficient engine core design and performance report. NASA
CR-168069, December 1982
9. Danis AM, Burrus DL, Mongia HC (1996) Anchored CCD for gas turbine combustor design
and data correlation. ASME1996-GT-143
10. Lefebvre AH (1984) Fuel effects on gas turbine combustion liner temperature, pattern factor
and pollutant emissions. AIAA J Aircraft, v 21, no 11, pp 887–898
11. Rizk NK, Mongia HC (1993) Semianalytical correlation for NOx , CO, and UHC emissions. J
Eng Gas Turbines Power 115:612–619
12. Danis AM, Pritchard BP, Mongia HC (1996). Empirical and semi-empirical correlation of
emissions data from modern turbopropulsion gas turbine engines. ASME96-GT-86
13. Hura HS, Joshi ND, Mongia HC (1998) Dry low emissions premixer CCD modeling and
validation. ASME1998-GT-444
14. Hura HS, Mongia HC (1998) Prediction of NO emissions from a lean dome gas turbine
combustor. AIAA1998-3375
15. Mongia HC (2008) Recent progress in comprehensive modeling of gas turbine combustion.
AIAA2008-1445
16. Mongia HC, Vermeersch M, Held TJ (2001) A DRA modeling approach for correlating gaseous
emissions of turbo-propulsion engine combustors. AIAA2001–3419
17. Mongia HC (2010) On initiating 3rd generation of correlations for gaseous emissions of aero-
propulsion engines. AIAA2010-1529
18. Mongia HC (2010) Correlations for gaseous emissions of aero-propulsion engines from sea-
level to cruise operation. AIAA2010-1530
19. Bruce TW, Davis FG, Kuhn TE, Mongia HC (1977) Pollution reduction technology program
small jet aircraft engines Phase 1. Final report. NASA CR-135214
20. Bruce TW, Davis FG, Kuhn TE, Mongia HC (1978). Pollution reduction technology program
small jet aircraft engines Phase 2. Final Report. NASA CR-159415
21. Bruce TW, Davis FG, Kuhn TE, Mongia HC (1981) Pollution reduction technology program
small jet aircraft engines Phase 3. Final Report. NASA CR-165386
22. Annex 16 to the Convention on International Civil Aviation Environmental Protection Volume
II Aircraft Engine Emissions, International Civil Aviation Organization
23. Mongia HC (2016) An art of developing aviation engine combustion technologies and products.
Energy, combustion and propulsion new perspectives, pp 239–268, ISBN: 978-19-1039-029-0.
Eds. Agarwal etc. Athena Academic Ltd
24. Mongia HC, Ajmani K, Sung CJ (2020) Hypotheses driven combustion technology and design
development approach pursued since early 1970’s. in 50 Years of CFD in engineering sciences:
a commemorative volume in memory of D. Brian Spalding, pp 439–484, Ed. A. Runchal, ISBN
978-981-15-2669-5. Springer Nature Singapore Pte Ltd
Low Emissions Propulsion Engine Characterization Process 69

25. Mongia HC, Ajmani, K, Sung CJ (2020) Fundamental combustion research challenged to meet
designers’ expectations. In: Chapter 11 in a Springer book, Ashoke De et al (eds) Sustainable
development for energy, power, and propulsion, expected to be published in 2020
26. Mongia, HC (2020) Synopsis of propulsion engines combustion technology/product develop-
ment and analysis substantiated during last 47 years. In: Gupta et al (eds) Advances in IC
engines and combustion technology, Select Proceedings of NCICEC 2019. Springer Nature
Singapore Pte Ltd
27. Mongia H, Krishnaswami S, Sreedhar PSVS (2007) Comprehensive gas turbine combustion
modeling methodology. Fluent’s Int Aerospace CFD Conference, June 18, 2007, Paris
28. Mongia HC, Held TJ, Hsiao GC, Pandalai RP (2003) Challenges and progress in controlling
dynamics in gas turbine combustors. J Propulsion Power 19: 822–829
29. Lieuwen TC, Yang V (2005) Progress in astronautics and aeronautics, vol 210
30. Anderson WE, Lucht RP, Mongia H (2015) Integrated physics-based modeling and experiments
for improved prediction of combustion dynamics in low-emission systems. NASA/CR-2015-
NNX11AI62A
31. Lee CM et al (2013) NASA project develops next generation low-emissions combustor
technologies. AIAA2013-0540
32. Joshi ND, Mongia HC, Leonard G, Steggmaier JW, Vickers EC (1998) Dry low emissions
combustor development. ASME 98-GT-310
33. Diehl LA, Biaglow JA (1974) Measurement of gaseous emissions from a turbofan engine at
simulated altitude conditions. NASA TM X-3046
34. Turns SR (2000) An introduction to combustion. The McGraw-Hill Companies, Inc
35. Law CK (2006) Combustion physics. Cambridge University Press, Cambridge
36. World Fuel Sampling Program, CRC Report No. 647, June 2006
Aerodynamic and Aero-Acoustic
Performance of an Adjustable Pitch
Axial Flow Fan

Rajat Arora, Ramraj H. Sundararaj, T. Chandrasekar, Sharad Saxena,

and Abhijit Kushari

Abstract To increase the operational regime of an axial flow fan has been of
immense focus lately. Variable-pitch rotor blades provide a crucial means of adapting
to various inlet flow conditions. The operating envelope of an axial flow fan is char-
acterized by flow instabilities. Experiments were conducted on a six-bladed scaled
down version of an inducer-type industrial fan. The performance was evaluated at
three different blade pitch configurations of 8, 16, and 22° with a speed variation of
460, 560, 660, 760, 860, and 960 RPM. Static pressure measurement was carried out
at the inlet of fan and exit of fan at sampling rate of thirty times the rotor order of
maximum RPM. Sound pressure was measured at the inlet of test facility. Experi-
ments conducted revealed that with increase in the pitch angle, flow rate, and static
pressure rise increases but the trend remains nonlinear, exception being the low flow
rates operating regime. Total efficiency increased and then decreased with increase
in pitch angle. Noise levels were highest at high pitch angles and low flow rates indi-
cating the existence of instabilities in the system. The presence of high amplitude
pressure fluctuations was confirmed by power spectral density plots of static pressure
and microphones data. The dominant frequency of 66 Hz appeared at RPM of 860
at near to fully closed operation of exit throttle. Other frequencies 22 Hz and 45 Hz
were found to be dominant at lower speeds.

Keywords Axial flow fan · Aero-acoustics · Variable-pitch fan · Aerodynamic

performance

Nomenclature

B Number of blades

R. Arora · R. H. Sundararaj · T. Chandrasekar · S. Saxena · A. Kushari (B)

Indian Institute of Technology Kanpur, Department of Aerospace Engineering, Kanpur, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 71
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_3
72 R. Arora et al.

BPF Blade passage frequency

Ca Axial flow velocity (m/s)
Cd Discharge coefficient of bell mouth 0.98 assumed
D Diameter of the casing
FSP Fan static pressure
N Rotor speed in revolutions per minute
Nd Non dimensional speed
Nref Reference or designed 960 RPM
OASPL Overall sound pressure level in decibel (dB)
P Static pressure in Pascal
P Static pressure rise in Pascal (SPR)
Ṗ Power in Watts
RO Rotor order N/60
RPM Revolutions per minute
U Blade velocity at mean (m/s)
V Voltage in Volts
i Current in Ampere
α Blade pitch angle in degree
ηm Mechanical efficiency of motor
ηt Total to total efficiency
∀ Volumetric flow rate in m3 /s
∅ Flow coefficient at mean
 Work done coefficient at mean

Subscript

1 Inlet of the fan

2 Outlet of the fan
ref Reference Pressure for air (20 µ Pa)
rms Root mean square

1 Introduction

Axial flow fan in an industry application is desired to operate efficiently at different

operating regimes, e.g., full load and part load operation. The operation is intrin-
sically limited by aerodynamic instabilities incurred during distortions in the inlet
flow, mechanical deterioration, and transient operations [1]. The instabilities can be
characterized as rotating stall or surge [2], where surge is associated with the global
oscillations [3], and rotating stall is the local flow adaptation technique leading to
non-uniformity in the flow around the annulus. A disturbance hence generated can be
Aerodynamic and Aero-Acoustic Performance of an Adjustable … 73

further amplified [4, 5], which will be manifested by overall drop in overall perfor-
mance, increased noise and vibration, lower efficiencies, and catastrophic mechanical
failures.
In contrast to a fixed-pitch fan, a variable-pitch axial flow fan has several benefits
such as high operating efficiency, a large flow rate, and a larger operating range [6].
Variable-pitch axial flow fans find applications in the thermal power stations where
fresh air is needed (forced draft) or process involving high temperatures such as in
boilers (induced draft). These types of fans are proven to have a negligible loss in
power at the output, and hence, they are also used in pulverized air fans and flue
gas desulphurization (FGD) applications. Even though the variable-pitch axial fan
provides and an efficient alternative, the increase in pitch angle leads to reduction
in stall-free operating margin [1]. The limiting factor that imposes an upper limit on
the pitch angle is not reported systematically.
The work presented here is directed at providing a depiction in the variation
of aerodynamic and acoustic performance when pitch angle is increased beyond
the designed values. The results of detailed experimentation at two measurement
planes are provided. The diagnostic for detecting the instabilities is discussed on
the basis of unsteady pressure measurement and non-intrusive sound measurements
technique of sound signals [7, 8]. The details pertaining to experimental facility,
instrumentation implemented, methodology adapted and experimental results are
presented in subsequent sections.

2 Experimental Setup and Instrumentation

The experimental setup shown in Fig. 1 is a six-bladed axial flow fan. The setup
is equipped with a bell mouth to establish clean flow at inlet of the fan without an
appreciable loss in total pressure. The flow enters through the bell mouth and exits
toward the throttle cone side. The pitch angle is altered manually; pre-calibrated slots
for different pitch angle are provisioned in the hub of the rotor. The diameter of the
fan casing is 1 m, and the length of the duct is 6D. Total length if the test including
bell mouth is 7D. Tip diameter and hub diameter for the current configuration are

Fig. 1 Photograph of the experimental test facility

74 R. Arora et al.

0.98D and 0.12D, respectively. The designed speed and designed pitch angle for
the current fan are N ref = 960 RPM and α = 8 degree, respectively. The setup has
a provision for flow rate variation via a throttle cone of length 0.5D. The throttle
cone is mounted on a linear traverse, operated with a stepper motor that facilitates
the accurate control of valve location from fully open to fully closed condition. The
facility is equipped with instrumentation allowing the measurement of quantities to
derive the aerodynamic and acoustic performance of the fan.
The inlet measurement plane is positioned at 0.9D ahead of the fan, whereas the
outlet plane measurement is located at 1.2D aft of the fan. Total pressure at the inlet is
measured by Pitot static probe aligned with the axis of the setup. Total pressure at the
exit is measured by 4 Kiel probes also aligned with the axis of the test facility. Static
pressure measurement at the inlet and exit is measured by six equally spaced pressure
ports along the circumference of the casing. The circumferentially averaged value
of the pressure is used for estimation of flow properties. All the pressure ports are
connected to electronically scanned pressure (ESP) scanner with a range of 1000 Pa
through flexible tubes. These scanners consist of an array of 32 silicon piezo-resistive
pressure sensors, one for each pressure port, whose analog output is electronically
multiplexed at the rate of 16,000 Hz through a built-in instrumentation amplifier.
The data is acquired at a rate of 500 samples per second for duration of 10 s. The
Nyquist criterion for the above sampling rate is 250 Hz which is 15 times the rotor
order at designed speed (16 Hz) of the fan.
Sound measurements were ascertained with the assistance of a PCB free field
130A24 microphones having a sensitivity of 10 mV/Pa. Microphone was placed at
a distance of 1D ahead of the bell mouth aligned with the axis of the tunnel. The
voltage was acquired through NI-6211 DAQ at a sampling rate of 50,000 samples
per second for a duration of 4 s. The fan is operated with a calibrated DC shunt motor
with maximum capacity of 3000 W. The RPM of the fan is regulated using a variable
frequency drive (VFD). The input voltage to motor is measured using multimeter, and
the input current is measured using clamp meter. All the measurements are interfaced
with LabVIEW application software. The errors in performance variables of fan map
are estimated to be less 0.1% in static pressure and flow rates and within ± 1 dB in
overall sound pressure level.

3 Methodology

The fan was operated and maintained constant at a selected speed, and the throttle
cone was inserted in the casing to vary the exit area and hence the flow rate. The values
static pressure and sound were recorded for several exit area blockages from fully
open to fully close condition. The experiments were repeated for various rotational
speeds as listed in Table 1. The experiments were conducted for designed pitch
angle α = 8° and two other settings of α = 16° and α = 22°. Table 1 summarizes the
operating speed and characteristic frequencies of the system at these conditions. It is
to be noted that pitch angle was not varied continuously during the experiments. The
Aerodynamic and Aero-Acoustic Performance of an Adjustable … 75

Table 1 Test conditions and

Non-dimensional Rotor order RO BPF (N × B/60 in
characteristic frequencies
speed Nd (N/60 in Hz) Hz)
0.48 7.6 46
0.58 9.3 56
0.68 11 66
0.79 12.6 76
0.89 14.3 86
1 16 96

two set of experiments were repeated for a given pitch setting, followed by change
in pitch setting, and the procedure was repeated for the next case.
Plots of static pressure rise versus flow rate, total efficiency versus flow rate, work
done coefficient versus flow coefficient, and overall sound pressure level constitutes
fan aerodynamic and acoustic fan characteristics. These plots were generated using
the standard formulas listed in Eqs. 1–10. Fan static pressure is the capability of
the fan to push air against a resistance (defined as fan total pressure rise—velocity
pressure), which in current set of experiments is represented by P2 since the inlet
stagnation pressure was always at ambient condition. Frequency domain analysis
of times series data obtained from pressure scanner and microphone was separately
performed using Welch’s power spectral density estimate with a frequency resolu-
tion close to 1 Hz. This analysis would aid in identifying the frequency bands of
instabilities if present.

2(P0 − P1 )
∀ = Cd A (1)
ρ

P = P2 − P1 (2)

Ṗin = V × i × ηm (3)

Ṗout = ∀(PO2 − PO1 ) (4)

Ṗout
ηt = (5)
Ṗin

2(PO1 − P1 )
Ca = (6)
ρ
π NDmean
U= (7)
60
76 R. Arora et al.

Ca
∅= (8)
U
(PO2 − PO1 )
= (9)
ρU 2
 2
Prms
OASPL = 10 log10 2
(10)
Pref

4 Experimental Results

In the current section, aerodynamic and acoustic results for designed blade pitch angle
of 8° are discussed, and effect of change in pitch angle on performance parameter
is further discussed for non-dimensional speed of N d = 0.89. Figure 1 shows the
variation of static pressure rise at different off design conditions. For a given rotor
speed, SPR increases from fully open to fully closed throttle valve position. With
increase in rotational speed the flow rate as well as SPR increases. The trend obtained
from experimental result is consistent with a typical compressor or fan map available
in the literature [9, 10]. Figure 2 shows total efficiency increases then decreases with
reduction in flow rates with slight variation in the trend at very low speed of N d
= 0.48. This nature of curve is also in agreement with the characteristic curves
obtained in a fan experiments. It is to be realized at this juncture that maximum
efficiency points do not correspond to maximum static pressure rise plot. Operating
requirements at higher SPR values necessitate a bargain between static pressure rise
and efficiency. The fan operates more efficiently at values close to the designed speed
N d = 1 (Figs. 3, 4 and 5).

Fig. 2 Static pressure rise 100

versus flow rate at α = 8° Nd = 0.48
Nd = 0.58
80
Static Pressure Rise (Pa)

N = 0.68
d
Nd = 0.79
60
Nd = 0.89
N =1
40 d

20

0
1 2 3 4 5 6
Flow rate (m3/s)
Aerodynamic and Aero-Acoustic Performance of an Adjustable … 77

Fig. 3 Total efficiency 0.65

versus flow rate at α = 8° N = 0.48
d
0.6 N = 0.58
d
0.55 N = 0.68
d

Total effeciency
Nd = 0.79
0.5
Nd = 0.89
0.45
N =1
d
0.4

0.35

0.3

0.25
1 2 3 4 5 6
Flow rate (m3/s)

Fig. 4 Workdone coefficient

Nd = 0.48
versus flow coefficient at α
= 8° Nd = 0.58
Nd = 0.68
0.1 Nd = 0.79
Nd = 0.89
ψ

Nd = 1

0.05

0
0.05 0.1 0.15 0.2 0.25 0.3
φ

Fig. 5 Overall sound 95

pressure level versus flow N = 0.48
d
rate at α = 8° Nd = 0.58
90
Nd = 0.68
Nd = 0.79
OASPL (dB)

85
Nd = 0.89
N =1
d
80

75

70
1 2 3 4 5 6
Flow rate (m3/s)
78 R. Arora et al.

To investigate the effect of pitch angle variation on performance parameters,

comparative plots are shown from Figs. 6, 7, 8, and 9 at non-dimensional speed of
0.89. It can be seen from Fig. 6 that for a given value of SPR, flow rate increases
from α = 8° to α = 22° . This increase in flow rate is attributed to decrease in the
inlet flow blockage at higher values of blade pitch angle; however, the trend in not a
linear function of α. Increase in the values of static pressure rise with increasing α is
contributed partially by increase in the FSP and decrease in the value of P1 at higher
α. The SPR values at α = 22° tend to rise gradually as compared to α = 16° finally
collapsing at flow rate of 4 m3 /s. The trend of Fig. 8 was found similar to the trend
obtained in Fig. 6. At this stage, flow separation seems to be a factor contributing to
the above nature of the flow physics. Along the same lines, lower value of efficiency
and higher values of noise are observed in Figs. 7 and 9.

Fig. 6 Plot of static pressure 100

rise versus flow rate at 8 Degree 16 Degree 22 Degree
various pitch angles for N d 90
= 0.89
Static Pressure Rise (Pa)

80

70

60

50

40

30

20
1 2 3 4 5 6
Flow rate (m3/s)

Fig. 7 Plot of total 0.65

efficiency versus flow rate at 8 Degree 16 Degree 22 Degree
various pitch angles for N d 0.6
= 0.89
0.55
Total effeciency

0.5

0.45

0.4

0.35

0.3

0.25
1 2 3 4 5 6
Flow rate (m3/s)
Aerodynamic and Aero-Acoustic Performance of an Adjustable … 79

Fig. 8 Plot of work done 0.16

coefficient versus flow 8 Degree 16 Degree 22 Degree
coefficient at various pitch 0.14
angles for N d = 0.89
0.12

0.1

ψ
0.08

0.06

0.04

0.02
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
φ

Fig. 9 Plot of overall sound 100

pressure level versus flow 8 Degree 16 Degree 22 Degree
rate at various pitch angles
for N d = 0.89 95
OASPL (dB)

90

85

80

75
1 2 3 4 5 6
Flow rate (m3/s)

To further investigate the cause of reduced efficiency and higher noise level, plots
of pressure amplitude variation with frequencies are plotted for α = 16° and α = 22°
at N d 0.89. The graphs are plotted up to frequencies of 150 Hz for better distinction.
Legend in Figs. 10, 11, 12, 13, 14, and 15 refers to the values of flow coefficient
φ. Higher values of φ correspond to fully open, whereas lower values correspond
to almost fully closed condition of exit valve. The subsequent graphs can be read
in conjunction with Figs. 6, 7, 8, and 9 for better interpretation. Figure 10 shows
the fluctuations in static pressure P1 at the inlet plane measurement at α = 16° . The
sharp jump in fluctuations at frequency of 67 Hz was observed at low flow rates
corresponding to φ = 0.15. The other small peaks existed at the values of operating
range are around 22 and 45 Hz. This point corresponds to very high noise level, lower
efficiency and low work done capacity as can be seen as observed in Figs. 7, 8, and
9. Similar frequency was observed in static pressure fluctuations at the exit plane P2
80 R. Arora et al.

Fig. 10 Plot of pressure 6

fluctuations in P1 at pitch 0.3
angle of 16° and N d = 0.89 0.29
5
0.28

Pressure Amplitude (Pa)

0.27
4 0.26
0.25
3 0.24
0.23
0.22
2
0.2
0.19
1 0.16
0.15
0
0 50 100 150
Frequency in (Hz)

Fig. 11 Plot of pressure 6

fluctuations in P2 at pitch 0.3
angle of 16° and N d = 0.89 0.29
5
Pressure Amplitude (Pa)

0.28
0.27
4 0.26
0.25
3 0.24
0.23
0.22
2
0.2
0.19
1 0.16
0.15
0
0 50 100 150
Frequency in (Hz)

with an additional frequency of 3.9 Hz. The fact that this frequency exits for all the
exit blockages, it manifests the presence of rotating stall for full operating regime.
The rotor order for the current rotational speed is 14.3 Hz as listed in Table 1.
Figures 12 and 13 show the fluctuations at α = 22° and N d = 0.89. It depicts
very high fluctuations in the static pressure P1 at 8.8 Hz, corresponding to φ = 0.1.
Other frequencies peaks were in the range of 64–66 Hz for values of φ < 0.19. The
frequencies (22 and 45 Hz) observed in Figs. 10 and 12 are similar but the magnitude
of fluctuations is slightly on the higher side in the case of α = 22° . The trend of higher
amplitude of fluctuations and similar frequencies remains similar for Figs. 11 and
13. The authors conducted similar analysis for other non-dimensional speed, and it
was found that for lower speeds the spectrum consisted of similar frequencies 22 and
45 Hz. The frequencies appearing in the spectrum seem to be following a pattern,
Aerodynamic and Aero-Acoustic Performance of an Adjustable … 81

Fig. 12 Plot of pressure 6

fluctuations in P1 at pitch 0.34
angle of 22° and N d = 0.89 0.32
5 0.32

Pressure Amplitude (Pa)

0.3
4 0.29
0.28
0.26
3
0.24
0.23
2 0.21
0.19
1 0.17
0.15
0.1
0
0 50 100 150
Frequency in (Hz)

Fig. 13 Plot of pressure 6

fluctuations in P2 at pitch 0.34
angle of 22° and N d = 0.89 0.32
5 0.32
Pressure Amplitude (Pa)

0.3
4 0.29
0.28
0.26
3
0.24
0.23
2 0.21
0.19
1 0.17
0.15
0.1
0
0 50 100 150
Frequency in (Hz)

where 22 Hz appears to be fundamental and 45 and 66 Hz as first and second overtone.

The reason for a particular mode getting excited is still unclear at the present stage.
The authors conducted similar analysis to generate power spectral density plots on
the data obtained from microphone placed in front of the test facility for the above-
mentioned conditions of α = 16° and α = 22° at N d = 0.89. It can be seen that
similar frequencies are manifested in sound measurement. The frequencies 22, 45,
and 66 Hz getting excited are confirmed by Figs. 14 and 15. The authors conducted a
modal analysis to obtain the natural mode shapes and frequencies of a structure. The
natural frequency of the bending mode was around 42 Hz, and the second bending
mode was 181 Hz eliminating the possibility of structural vibration exciting the
frequencies shown by power spectrum.
82 R. Arora et al.

Fig. 14 Plot of acoustic 0.5

pressure fluctuations in 0.3
microphone at pitch angle of 0.29
16° and N d = 0.89 0.4 0.28

Acoustic Pressure (Pa)

0.27
0.26
0.3 0.25
0.24
0.23
0.2
0.22
0.2
0.1 0.19
0.16
0.15
0
0 50 100 150
Frequency in (Hz)

Fig. 15 Plot of acoustic 0.5

pressure fluctuations in 0.34
microphone at pitch angle of 0.32
22° and N d = 0.89 0.4 0.32
Acoustic Pressure (Pa)

0.3
0.29
0.3 0.28
0.26
0.24
0.2 0.23
0.21
0.19
0.1 0.17
0.15
0.1
0
0 50 100 150
Frequency in (Hz)

The possibility of air column contained by the duct vibrating at a forced frequency
by the rotor is discussed by Wallis [11]. If the applied frequency is found to be similar
to that of natural frequency of the air column as in case of organ pipes, resonance
occur leading to intensification in noise levels. Further, investigation at the current
stage is required to make any claims regarding the above-mentioned possibility. The
main result of the study conducted is the onset of instabilities at high pitch angles
fan operating at high speed and reduced flow rates. The detailed study is intended to
push the operating envelope further with safe operation.
Aerodynamic and Aero-Acoustic Performance of an Adjustable … 83

5 Conclusions

The goal of this investigation was to identify the change in performance map at pitch
angles values higher than the designed values. Initial phase of analysis suggests
that the fan operation at high pitch angles and lower flow rates is susceptible to
flow instabilities leading to inefficient operation. Overall noise levels were found
around to be the higher side. The frequencies appearing seemed to be dependent
on the exit blockage. Further, improvements in measurement locations with addi-
tional measurement technique like vibration measurement using accelerometer and
turbulence measurement using hotwire are required at the current stage to better
understand the cause associated with instabilities.

Acknowledgements The work was carried out using the axial flow fan facility developed and
located within National Wind Tunnel Facility (NWTF) premises at IIT Kanpur. The authors
gratefully acknowledge the support of Head NWTF for facilitating the current research.

References

1. Bianchi S, Corsini A, Sheard AG (2013) Experiments on the use of signal visualization tech-
nique for in-service stall detection in industrial fans. Adv Acoust Vibr 2013:10. Article ID
610407
2. Lawless PB, Fleeter S (1993) Rotating stall acoustic signature in a low speed centrifugal
compressor: Part 1-vaneless diffuser. ASME Paper no 93-GT-297
3. Greitzer EM (1976) Surge and rotating stall in axial flow compressors—Part I: theoretical
compression system model. J Eng Power 98(2):190–198
4. Velarde-Suarez S, Ballestesteros-Tajadura R, Santolaria-Morros C, Gonzalez-Perez J (2001)
Unsteady flow pattern characteristics downstream of a forward-curved blades centrifugal fan.
J Fluid Eng 123/265
5. Cumpsty NA (1989) Part-circumference casing treatment and the effect on compressor stall.
ASME Paper No 89-GT-312
6. Ye X, Fan F, Zhang R, Li C (2019) Prediction of performance of a variable-pitch axial fan with
forward-skewed blades. Energies 12(12):2353
7. Bently D, Goldman P, Yuan J (2001) Rotor dynamics of centrifugal compressor in rotating
stall. Orbit 2Q01, pp 40–50
8. Bently D, Goldman P (2000) Vibrational diagnostics of rotating stall in centrifugal compressors.
Orbit 1Q01, pp 32–40
9. Dixon S, Hall C (2014) Fluid mechanics and thermodynamics of turbomachinery. Butterworth-
Heinemann
10. Cohen H, Rogers GFC, Saravanamuttoo HIH (1996) Gas turbine theory. Longman group
Limited
11. Wallis RA (1961) Axial flow fans design and practices. Academic Press, New York, London
Proposed Thrust Profile Design of Pulse
Detonation Engine (PDE) for Aerospace
Applications

Subhash Chander and T. K. Jindal

Abstract The paper deals with challenging assignment of development and opti-
mization of pulse detonation engine (PDE) to test it in aerospace system configura-
tions. Our researchers have successfully developed and deployed static test bed for
PDE firing earlier. This has been trial proven in dynamic PDE firings in last one
decade. In the present case of design of propulsion system for aerospace vehicle,
which includes trial setup for prolonged testing in autonomous mode, pneumatic
air/oxygen supply (compressed), fuel supply, ignition system, command genera-
tion system (wired/remote), recorders/telemetry, safety systems including thrust
stand hardware need to be system engineered meticulously. Extensive testing of
subsystem/subassembly levels need to incorporated before to get confidence to func-
tion in dynamic environment. This included 150 bar compressed air delivery system,
which needs to operate in highly pulsating environment at test bed. A small recorder
system is also installed to record four channels of data acquisition system (DAQ) to
validate the performance of PDE; full function trial run in autonomous configuration
resulted in successful test runs, repeatedly. It generated a lot of data to predict system
performance to make it function better. All these exercises are executed simultane-
ously and efforts are still on in this regard. Careful data analysis has resulted in
validation of predicted mission parameters in actual trials at various test conditions.
The same data is utilized to inject in experimental aerospace vehicle (EAV) six DOF
model and tailor it to meet these applications. The performance of the vehicle in
terms of speeds, range, intercept dynamics, weight advantage and compact form
factor achievement, etc. need to be explored in this paper.

Keywords Autonomous PDE rig · Detonation · Ignition system · Thrust

generation · Measurements

S. Chander (B)
DRDO, Terminal Ballistics Research Lab, Chandigarh, India
e-mail: [email protected]
T. K. Jindal
Punjab Engineering College (Deemed To Be University), Sector 12, Chandigarh, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 85
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_4
86 S. Chander and T. K. Jindal

1 Introduction

Pulse detonation engine is a modern propulsion system that utilizes repetitive deto-
nations to produce thrust. PDE offers the potential to provide increased perfor-
mance, while simultaneously reducing the engine weight, miniaturization, cost and
complexity relative to conventional propulsion systems currently in service. Due to
its obvious advantages, attention has been paid to the scientific and technical issues
concerning pulse detonation engine worldwide. Despite extensive research in PDE
over the past several decades, it is not yet ready to be used for practical propulsion
applications [18–20]. Many reputed institutions like NASA, DAPRA, DRDO, GE,
P&W, EADS and many others are seriously engaged in active research in this area
and are in a mature stage of development of such technologies and systems. Many
academic organizations like University of Taxas, Arlington, USA, Nogaya Univer-
sity, Japan, N.N. Semenov Institute of Chemical Physics, Russia, NUS, Singapore
and Punjab Engg. College, Chandigarh, etc. are engaged in active research in this
area.
One of the key barriers to the realization of an operational PDE lies in the diffi-
culty to initiate the detonation wave in the engine, in a reliable and controllable
manner. Detonations are difficult to achieve within a practical length in hydro-
carbon/air mixtures, which are less sensitive to detonation initiation taking atom-
ization, vaporization and detonation into account. There are challenges in reducing
noise and dampening the vibrations due pulsating power coupled by handling thermal
management for prolonged runs.

2 The Setup

The dedicated test setup for experimental trials is used to investigate, address the
issues and observe the performance improvements due these modifications as shown
in Fig. 1 [20, 23, 31].
For this purpose, detonation process details are being explored extensively. Deto-
nation can be initiated either directly or by deflagration followed by deflagration to
detonation transition (DDT). Direct detonation initiation energy and power require-
ments are function of cell size or cell width (λ) of a detonable mixture, which is the
measure of the combustion reaction rates of the fuel mixture. Minimum energies are
of the order of 250 kJ for several hydrocarbon/air mixtures, which are prohibitive for
practical pulse detonation engines [31]. It is desirable to use low initiation energy
system to produce fully developed detonations with in an acceptable distance [21]. An
important diagnostic tool for such exercise is sensor networking, which is explained
below.
Title Suppressed Due to Excessive Length 87

Fig. 1 Schematic of demonstrator test configuration

3 Sensor Networking

A sensor network is designed and powered using various rating of SMPS to meet the
diagnostics requirements of the setup. Main sensors used for this purpose are [22,
37].
1. MEAS Pressure sensors (4/10 bar) for line pressure measurements
2. Analogue gauges (20/100 bar) for indications and manual recording
3. PCB/Zuma pressure gauge (10/20 bar)
4. Max load cells (10 kN) for force measurement
5. K/J type thermocouple network for hot mixture and tube surface temperature
measurements.
Some other devices are also used:
1. Solenoid valves for gas filling to settling chambers/PDE tube
2. Flow controllers for auto filling of gases to settling chambers.
88 S. Chander and T. K. Jindal

Fig. 2 PDE structured control plan

A special network is created to establish the connectivity is shown in Fig. 2

[31]. Organizing the sensors and control components has helped team to achieve
a structured architecture of the plan. It has created a modular approach to highly
reliable electronics and instrumentation setup to ease the rapid recording/processing
of data in an organized pattern. These modifications have also helped team to easy
detection of problems and resolving the major issues in a simplified diagnostics
approaches.

4 The Engine

The PDE developed is based on C2 H2 as fuel and O2 as oxidizer. A typical data set
obtained is plotted (Fig. 3) and is used as reference thrust input for our problem.
Same data is assumed for all four tubes. The thrust of total system is based on
this content of PDE-based sustainer. This study will be able to make researchers
understand the acceleration growth, velocity and down range extension trends. This
preliminary work will become a vital study for identifying the suitability of PDE
engine as sustainer as compared to conventional system. Further improvements can be
reinforced after seeing the performance enhancements observed and its verification
in detailed complex 6° of freedom model. Its effect on other systems and handling
techniques is to be explored further [22, 45, 30].
Title Suppressed Due to Excessive Length 89

Fig. 3 Multiple tube PDE thrust profile

5 The Ignition System Hardware

The ignition control system (ICS) used for this purpose is shown in Fig. 4 [23].
The objective of this research is to study the thrust generation algorithms to obtain
solutions of pulsating thrust produced by PDE.
The cyclic PDE engine system is centrally controlled by a custom built software
program developed and installed on a computer controller-based data acquisitions and

Fig. 4 Multitube multicycle PDE AFCS—electrical connectivity line diagram multicycle PDE
control
90 S. Chander and T. K. Jindal

control hardware installed in robust environment PDE test bed with structured termi-
nations. The controller computer is linked to a data processing, analysis, modeling
and simulation systems in the rig through a network connection to the especially
built PDE domain.
The software program controlled engine operation by managing engine supply
acetylene gas operation via solenoid valves using opening time programmed located
in the rig. A D-sub μ-connector-based pulse generator also located in the control room
controlled timing of trigger signals sent to both the mixture igniters via electrical
relays (12 VDC) of the system. 24 VDC and 220 VAC master power relay modules
are also installed in control room. Further, an emergency shut-off button coupled with
emergency stop lever too. These electrical relay switches controlled power within the
rig for engine control and instrumentation. These instrumentations include load cells,
temperature and pressure transducers connected through termination boxes. The flow
of gases inside pulse detonation engine is controlled by solenoid-based flow control
valves with non-return valve for safety purposes coupled by flash arrestors. The
emergency shut-off button would disable rig by closing all supply gas ball valves, it
will also interrupt fuel injection coupled with stoppage of ignition trigger signals.
The typical PDE cycle with details is shown in Fig. 5.

PDE THRUST Profile

PDE WEIGHT BURN_OUT Profile

PDE PHASES

Fig. 5 EAV embedded PDE phases in each cycle

Title Suppressed Due to Excessive Length 91

6 Automated Multicycle ICS Software

Software is especially designed to give the commands for firing of PDE and record
the outputs in correct sequences and timings [31]. The software is written in NI
Labview working in robust test environment with NiDAQ and DOP libraries. The
detailed analysis and plotting is carried out in MATLAB.
The software is divided into three parts as below:
• Firing control software
• DAQ analog input/output software module
• MATLAB-based analysis utility.
These are explained as below:

6.1 Firing Control Software

The flow chart of software is shown in Fig. 6 and software screen of National Instru-
ments Labview Firing Control Software is shown in Fig. 7, hich are self expan-
tory. The software allow user to select the gas release time in GUI Environment.
Then Gas release button knob is pressed.

Fig. 6 Multicycle ICS software flowchart

92 S. Chander and T. K. Jindal

Fig. 7 Firing software GUI

6.2 DAQ Analog Input/output Software Module

NI DAQ cards will generate analog input commands in form of voltages, which
will generate commands to open solenoid valves, firing spark initiations through
command signals. NI DAQ cards will also monitor pressures through pressure
sensors converted voltages, thrust through load cells converted voltages, temper-
atures through thermocouple converted voltages sensed by DAQ and fed back to VI
for display and storage to.xls file (Fig. 8). It is visually explained through flowchart
[31].

6.3 MATLAB-Based Analysis Utility

This explains the development of MATLAB-based software AFCS for PDE analyzing
detonation, signal conditioning, display and .XLS file save for further analysis.
Title Suppressed Due to Excessive Length 93

a. S ensor online display b. Flow chart

Fig. 8 Sensor output online plots generation

MATLAB program was written to analyze, filter, plot, compare and conclude
the test achievements for further collaborations (Fig. 9). Since it is involving highly
flammable fuels at closed vicinity of PDE system so appropriate check lists were
developed for further enhancing g safeties.

7 Conversion of Single Shot ICS to Multitube

and Multicycle ICS

IFCS for multicycle multitube operations is a mission critical hardware. The derived
scheme for multitube multicycle operations [23].
The design is similar to single shot, but large no. of components makes it impos-
sible to ensure high reliability and safety. The system has been trial proven after
repetitive testings.
94 S. Chander and T. K. Jindal

PDE Simulation (Pr.) with 50 Hz filter

p1 Bar
p2 Bar
p3 Bar
p4 Bar

20

15
bar

10

0
24 24.05 24.1 24.15 24.2 24.25
time in sec.

Fig. 9 Depicting P1 , P2 , P3 and P4 cross plot of one typical run in MATLAB

8 EAV Configurations with Pulse Detonation Engine

Option

The area of research was identified as use of PDE as secondary propulsion system
(sustainer) for tactical EAV application. The technical concepts of design, operations
and integration are as follows [31].

Fig. 10 Proposed tactical missile with PDE sustainer

Title Suppressed Due to Excessive Length 95

Fig. 11 Proposed EAV configuration with PDE sustainer

A typical tactical EAV (Fig. 10) is being utilized to experiment the trajectory vari-
ations due PDE incorporations. The basic existing booster sustainer configuration is
studied and their forces, accelerations, velocities and down range are being analyzed.
The experimental aerospace vehicle (EAV) launch weight is <130 kg. The booster
is having <4 s run and sustainer has got about <7 s run. It is launched from ground
against incoming missiles, UAVs and aircraft etc. It is a supersonic 12 km+ range
weapon. It is a quick reaction seeker/INS guided missile. It has got classical warhead.
It can be also fired in many other modes too. Finalization of PDE replacement
of rocket sustainer, no. of tubes and frequency of firing including its associated
modifications/corrections required to be carried out in various subsystems (Fig. 11):
• Since the thrust is sufficient for the current mission, so the feasibility part is
concluded
• Since the system has to fit in EAV, so, sizing and shaping has been designed
analytically and through modeling and simulation
• The need of design of fuel and oxidizer tanks also was considered and hence
implemented
• The piping for transport of fuel and oxidizer from tank to PDE injector was also
forecasted and realized
• Mounting mechanism and gas path design was also carried out.

9 Thrust Analysis of EAV Propulsion

The thrust of EAV with existing configuration and PDE modified configuration is
examined in detail. The thrust profile of both the configurations is cross plotted for
comparison (Fig. 12).
It is clear that PDE is not able to take parallel performance to rocket sustainer in
the present form, so it is planned to apply some changes to suit the requirements,
which are covered below.
96 S. Chander and T. K. Jindal

THRUST DEVELOPMENT - PDE based (Green)

Fig. 12 Pulsating thrust handling plan

10 Staggered Firing Techniques

It was tried to fire PDE in staggered mode to reduce the dead thrust band, which
did improve the performance; however, it was still not meeting the performance
criteria in totality (Fig. 13). Now, the frequency increase is injected gradually, at
about 27 Hz per tube in staggered mode did give promising results (Figs. 14, 15);
however, there is a 10% pulsating effect, but no dead thrust band. Additionally, it
had super-imposition of thrust cycles, resulting in higher thrust than before. This is
simply exceptional achievement in performance of the EAV as it did improve the
range, intercept speeds and kill mechanisms, which is a game changer for PDE in
future [32].
• Modified EAV is being utilized to experiment the trajectory variations due PDE
incorporations in MATLAB simulation as follows:
• Basic existing booster sustainer configuration

Fig. 13 Staggered firing sequence timing events in each PDE cycle

Title Suppressed Due to Excessive Length 97

PDE PULSATING THRUST SUPERMIPOSITION TECHNIQUES

Fig. 14 Superimposed staggered firing sequence timing events in each PDE cycle

PULSATING THRUST HANDLING – RESULTANT THRUST

Fig. 15 EAV configuration with PDE-based sustainer augmentation

• PDE sustainer firing configuration—forces, accelerations, velocities and down

range.
This research promise better latex, intercept velocities and down range etc. Multi-
tube and multicycle operations hold key to the solution. This novel approach with
98 S. Chander and T. K. Jindal

change in fire control system will help to optimize the performance and handling
characteristics too.
The effects of PDE-based EAV system performance is deteriorated after PDE
introduction due pulsating power. I explored techniques to improve the performance
through configurations and physical parameter variations and reduction of pulsating
power effects to open up new option available for propulsion, which is having better
power/weight ratio, form factor and easier to maintain/operate in future.
Consequent to our work on PDE in single shot/multicycle operations in ground
demonstration and experimentation rigs, we have optimized the performance of PDE.
Our team too explored the use of PDE as sustainer for single shot EAV. Existing
AV system is being adapted for this purpose and PDE developed data is used for
this highly ambitious iteration. The removal of rocket-based sustainer propulsion
and insertion of PDE alternative is being experimented in theoretical work and
3D modeling. The PDE-related hardware is also being inserted inside the custom
design of EAV. The space constraints, routing of components, CG/MI distribution
and resulting re-orientation of various components are being finalized. This work is
being limited by various other aspects due pulsating flows options, longer duration
run challenges, etc. need to be tried. General system engineering (SE) aspects and
its effect on the system performance also force designers to study into change of key
parameters very meticulously. The current paper deals with these aspects and lead
to finalization of weapon system propulsion by using advanced features of modeling
and simulation tools coupled with state-of-the-art integration/assembly techniques
with analytical backing. The improvements carried out thereon, after meticulous
selection of vital parameters in the engine, ground rig and instrumentation setup, etc.
It also yielded in robust optimal design coupled by better consistency of dynamic
performance of PDE powered AV and optimization of main subsystems of PDE.
Further, team will also reduce the development cycle and trial complexity also, in
addition to generate data to look further diversified use.
The PDE related hardware is also being inserted inside the custom design of
EAV. The space constraints, routing of key piping, wiring of ignition system,
fuel/oxidizer tanks, reorienting exhaust of exit nozzle, CG/MI distribution and
resulting re-orientation of various components are being finalized. This work is
being limited by various other aspects including channelization of hot gases, vibra-
tion/shock mounting due pulsating flows options, longer duration run challenges
etc. need to be tried. General system engineering (SE) aspects and its effect on the
system performance also force designers to study into change of key parameters
very meticulously. The current paper deals with these aspects and lead to final-
ization of weapon system propulsion by using advanced features of modeling and
simulation tools coupled with state-of-the-art integration/assembly techniques with
analytical backing. Researchers are actively engaged in PDE research to optimize
the thrust and suppress sound. Handling power inconsistency, thermal management
DDT are the other areas of focused research due issues. The improvements carried
out thereon, after meticulous selection of vital parameters in the engine, ground rig
and instrumentation setup, etc.
Title Suppressed Due to Excessive Length 99

By using this setup, trials have been carried out in detail and results are examined.
This propulsion data is inserted in 6° of freedom model of aerospace vehicle to assess
the performance of EAV [20, 21].
With the above modifications in EAV to incorporate PDE, overall system perfor-
mance has improved, as key problem areas have been addressed by improving fin
deflection limits δ from 15 to 18°. Effectivity of control is also improved by tweaking
deflection rates δ’ from 300 to 350°/s. Robustness and response of system can be
improved through deflection accelerations δ” from 3.75 × 106 to 4.0 × 106 °/s. Finally,
after computing other parameters for EAV, 6° of freedom model is run to estimate
the performance evaluation of the EAV and suggest corrective actions to this effect
[30].

11 Conclusion and Scope for Future Work

• PDE-based secondary propulsion successfully designed for EAV mission. It is

proved to be superior to current sustainer in term of launch weight, intercept zone
enhancement, better interception, controllability and agility.
• It is using simpler PDE technology, which has paved a new concept of pulsating
power usage in aerospace propulsion.
• EAV configuration with PDE-based sustainer augmentation—Modified EAV
is being utilized to experiment the trajectory variations due PDE incorporations
in MATLAB simulation as follows:
• Basic existing booster sustainer configuration.
• PDE sustainer firing configuration—forces, accelerations, velocities and down
range.
• This research promises better latex, intercept velocities and down range, etc.
Multitube and multicycle operations hold key to the solution. This novel approach
with change in fire control system will help to optimize the performance and
handling characteristics too.
• The effects of PDE-based EAV system performance is deteriorated after PDE
introduction due pulsating power. I explored techniques to improve the perfor-
mance through configurations and physical parameter variations and reduction of
pulsating power effects to open up new option available for propulsion, which is
having better power/weight ratio, form factor and easier to maintain/operate
in future.
• Cost of weapon system got reduced extensively due classical hardware, especially
for single shot devices.
• Fire control system safeties are now embedded in design due remote operations
of stand-offs.
• Due multiuse remote operations, launch operations become affordable.
• Operations of rocket mode PDE do not get affected by height or other environ-
mental changes.
100 S. Chander and T. K. Jindal

12 Future Works Includes

• Creating working SAM model to prove concepts

• Sensor networking
• Trying similar concepts on UAV and satellite launch vehicle
• Trying liquid-based/air breathing PDE/RDE
• Thermal management for longer runs
• Optimizing PDE power for better applications
• Optimizing ignition energies and fuel/oxidizer flows.

Acknowledgements Authors are thankful to the Director, PEC University of Technology and the
Director, TBRL, Chandigarh, for their support and guidance.

References

1. Zhang F-Y, Fujiwara T, Miyasaka T, Nakayama E-I, Hattori T, Azuma N, Yoshida S, Tamugi A
(2003) Experimental study of key issues on pulse detonation engine development. Trans Japan
Soc Aero Space Sci 45(150):243–248. https://doi.org/10.2322/tjsass.45.243
2. Schauer F, Stutrud J, Bradley R (2001) Detonation initiation studies and performance results for
pulsed detonation engine applications. In: 39th AIAA aerospace sciences meeting and exhibit.
AIAA, pp 2001–1129
3. Chander S, Jindal TK (2012) Integration challenges in design and development of pulse deto-
nation test rig. Int J Adv Res Electr Electron Instrum Eng 1(4), ISSN 2278-8875. https://doi.
org/10.15662/ijareeie
4. Lu FK (2016) Progress and challenges in the development of detonation engines for propulsion
and power production. Appl Mech Mater 819:3–10. ISSN 1662-7482
5. Chander S, Jindal TK (2014) Performance enhancement of surface to air missile by application
of pulse detonation engine based secondary propulsion system. Int J Mech Engg (IJME). Recent
Science Publications
6. Chander S, Jindal TK (2013) Design of automated fire control system for C2 H2 /O2 pulse
detonation rig. Int J Adv Res Electr Electron Instrum Eng 2(2). ISSN 2278-8875. https://doi.
org/10.15662/ijareeie
7. Jindal TK (2012) Pulse detonation engine—a next gen propulsion. Int J Modern Eng Res
(IJMER) 2(6). ISSN 2249-6645
8. Jindal TK (2012) Single cycle pulse detonation engine testing. Int J Mech Eng 40(10). ISSN
2051-3232
9. McManus K, Furlong E, Leyva I, Sanderson S (2001) MEMS-based pulse detonation engine
for small scale propulsion applications. AIAA paper, pp 2001–3469
10. Valli DM, Jindal TK (2013) Thrust measurement of single tube valveless pulse detonation
engine. Int J Sci Eng Res (IJSER) 4(3). ISSN 2229-5518
11. Soni SK, Singh A, Sandhu M, Goel A, Sharma RK (2013) Numerical simulation to investigate
the effect of obstacle on detonation wave propagation in a pulse detonation engine combustor.
Int J Emerg Technol Adv Eng 3(3): 458–464. ICERTSD-2013. ISSN 2250-2459
12. Ostertagová E, Ostertag O (2013) Methodology and application of oneway ANOVA. Am J
Mech Eng I(7): 256—261. ISSN (Print) 2328-4102, ISSN (Online) 2328-4110, 10.12691
13. Chander S, Jindal TK (2014) A study of possible technical aspects needed to be addressed
in respect of system engineering of surface to air missile by application of pulse detonation
engine. Int J Mech Eng (IJME) 1(1): 40–43. ISSN 1694-2302
Title Suppressed Due to Excessive Length 101

14. Chander S, Jindal TK (2018) Performance analysis of surface to air missile by application of
pulse detonation engine based secondary propulsion system. Ph.D. thesis submitted to Punjab
Engineering College (Deemed to be University), Chandigarh
15. Robbins TJ, Sinibaldi JO (2006) Fuel injection strategy for a next generation pulse detonation
engine. Naval Postgraduate School, Monterey, California
16. Rodriguez J (205) Investigation of transient plasma ignition for a pulse detonation engine.
Master’s thesis, Naval Postgraduate School, Monterey, California
17. Kailasanath K (2000) Review of propulsion applications of detonation waves. AIAA J 38(9)

Proceedings/Transactions Articles

18. Smirnov N, Nikitin VF, Biochinto AP, TyurniKov MV, ShevtSova VM, VBaskaKov V (1999)
In: Proceedings of the 12th ONR propulsion meeting
19. Roy GD, Frolov SM, Borisov AA, Netzer DW (2014) Pulse detonation propulsion: challenges,
current status, and future perspective. Prog Energ Combust Sci 30. www.sciencedirect.com
20. Chander S, Valli DM, Jindal TK (2013) Multiple option levels based full function PEC aero
simulation (PAeroSim) development for pulse detonation engines and systems. In: Symposium
on applied aerodynamics and design of aerospace vehicle (SAROD 2013), Hyderabad, India
21. Schauer F, Stutrud J (1999) AFRL’s in house research pulse detonation engine. In: 11th annual
symposium on propulsion
22. Falempin F, Bouchaud D, Forrat B, Desbordes D, Daniau E (2001) Pulsed detonation engine
possible application to low cost tactical missile and to space launcher. In: AIAA-2001-3815,
37th AIAA/ASME/SAE/ASEE joint propulsion conference, Salt Lake City, Utah
23. Cloutier JR, Evers JH, Feeley JJ (1988)Assessment of air-to-air missile guidance and control
technology. In: American control conference, Atlanta, Georgia
24. Panicker PK, Lu FK , Wilson DR (2009) Practical methods for reducing the deflagration-to-
detonation transition length for pulse detonation engines. In: Proceedings of the 9th interna-
tional symposium on experimental and computational aerothermodynamics of internal flows
(ISAIF9), Gyeongju, Korea
25. Falempin F, Bouchaud D, Daniau E (2000) Pulsed detonation engine towards tactical missile
application. In: AIAA-2002-3473, 36th AIAA/ASME/SAE/ASEE joint propulsion conference,
Huntsville, Alabama

Company or NASA Report and Patents

26. Chao TW, Winterberger E, Shepherd JE (2001) On the design of pulse detonation engine. Galcit
Report FM00–7, Graduate Aeronautics Laboratories, California Institute of Technology
27. Jindal TK, Chauhan YS (2012) Development of pulse detonation rig. A Report on sponsored
research by TBRL
28. Chander S, Singh S, Ray SK et al (2008) System engineering for Antiship missile—a report.
RCI, Hyderabad
29. Lam M, Tillie D, Leaver T, McFadden B (2004) Pulse detonation engine technology: an
overview. Submitted to: Michael Schoen, Applied Science 201, The University of British
Columbia
30. Chander S (2007) Pointmass study of guided bomb—a report. RCI, Hyderabad
31. Chander S, Chaudhuri SK et al (2008) Missile launched PGMs—a report. RCI, Hyderabad
32. Chander S, Jindal TK (2019) Pulse detonation engine—pulsating power handling schemes. A
petent filed with Department of Patents, Govt. Of India, 2019 (Patent Pending)
102 S. Chander and T. K. Jindal

AIAA and Other Book Series

33. Travis J, Kring J (2009) LabVIEW for everyone. Pearson Education

34. Thomson WT, Dahleh MD (1998)Theory of Vibration with applications, 5th ed. Prentice Hall,
Inc
35. Fleeman E (2006) Tactical missile design. AIAA Edu. ISBN-10 1563477823
36. Operations and maintenance manual, six component, 500 pound pulse detonation measurement
system. Pacific Press Co., N00244-02-C-0060, Manual No. 110028, September 2002
37. Chin SS (1961)Missile configuration design. McGraw-Hill Book Company, Inc., New York.
(OCoLC) 602683910
38. Garnell P (1980) Guided weapon control systems. Brassley’s Defence Publishers, London
39. SK Ray (2007) Missile control system. DESIDOC, DRDO, Delhi
40. Ramamurti K (2010) Explosions and explosion safety. Tata Mcgraw Hills Publishing Co. Ltd.
ISBN 9780070704473
41. Kothari CR, Garg G (2014) Research methodology: methods and techniques. New Age
Publishers. ISBN-13 978–8122436235
42. Montgomery DC, Runger GC (2003) Applied statistics and probability for engineers. Wiley
43. Royal Military College of Science Guided weapons handbook, 14th ed., pp 1–1
44. JF Rouse Guided weapons. Royal Military College of Science, Shrinavenham, UK, Brassey’s,
London
45. Mullins SS, Penner BP (1959) Explosions, detonations, flammability and ignition. Pergamon
Press
Fuels
The Formation of PAH Compounds
from the Combustion of Biofuels

A. Williams, A. R. Lea-Langton, and K. D. Bartle

Abstract Polycyclic aromatic hydrocarbons (PAH) are byproducts resulting from

the incomplete combustion of hydrocarbon-based fuels such as fuel oils, coal and
biofuels. PAH are implicated in the formation of soot, another undesirable byproduct
formed during combustion. Such compounds are harmful to human health, and soot
plays an important role in climate change. Black carbon is known to absorb visible
solar radiation, and the associated organic carbon fraction, including PAH, that is
co-emitted can absorb radiation at shorter wavelengths leading to climate forcing.
As a result of climate change considerations and the need to reduce greenhouse gas
emissions, fossil fuels are slowly being replaced by lower carbon renewable biofuels.
However, these biofuels are often carbon based, and so, their incomplete combustion
produces a similar range of pollutant species to fossil fuels. These substitute fuels
can form PAH compounds, although often to a lesser degree due to differences in the
fuel chemistry. The EPA 16 list is widely used to determine the extent of pollution
due to PAH but does not include other types of polycyclic aromatic compounds such
as oxygenated species (O-PAC) or nitrogenated species (N-PAC) so does not reflect
many of the priority pollutants.

Keywords Biomass · PAH · O-PAH · Renewable fuels

1 Introduction

There has been considerable interest in the emission of PAH species from the combus-
tion of fossil fuels such as coal, diesel and heavy fuel oils because many species
have been shown to be harmful to human health [1–6]. Soot is an associated co-
product with adverse health and climate impacts. Because of climate change concerns

A. Williams (B) · K. D. Bartle

School of Chemical and Process Engineering, University of Leeds, Leeds LS2 9JT, UK
e-mail: [email protected]
A. R. Lea-Langton
Department of Mechanical Engineering, Aerospace and Civil Engineering, School of
Engineering, University of Manchester, Manchester M13 9PL, UK

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 105
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_5
106 A. Williams et al.

fossil fuels are being replaced by biofuels in some applications, which includes
mainly solid, liquid and gaseous forms. The solid biofuels involve wood, agricul-
tural residues, charcoal and biochar; the liquids are mainly ethanol and biodiesel;
and the gaseous fuels are biogas or products from gasification of biomass. Biomass
accounts for about 10% of the global energy supply and 80% of the renewable energy
[7]. Biomass fuels have the advantage that they share many of the properties of fossil
fuels and can easily substitute for them for heating and electricity generation, road
vehicles and as aircraft fuels.
The use of solid renewable fuels has increased markedly in recent years [7, 8].
Solid fuels, mainly wood logs, pellets from waste wood or agricultural residues have
been widely used for many years, but there has been a considerable increase because
of climate change concerns; however, there are consequential pollution issues partic-
ularly with regard to smoke pollution [4, 5, 8]. There has also been an increase in
use of liquid biofuels where the fast pyrolysis of biomass feedstock is potentially an
economic method of producing renewable liquid fuels.
Road vehicles have proved to be a suitable case for the substitution of fossil fuels
by the biofuel equivalent either in total or by blending. This is the case for petrol or
diesel, and the substitutes may be bioethanol/biobutanol or biodiesels derived from
transesterification of bio-oils, e.g. from rape seed oil [9, 10].
Aviation bio-jet fuels present different challenges although the same types of fuels
may be used [11, 12]—Alcohols are too volatile to be used alone but could be used
in blends with Jet A1. This is also the case with bio-oils which can be blended with
petroleum fuels although currently the fuel specifications are very strict.
Much depends on the fuel refining operations which can result in a range of fuels
types [11, 13] including heavy fuels which can be burned as fuel sprays in furnaces
or marine engines as substitutes for heavy fuel oils [3, 14, 15]. Another source of
these heavy oils could be via the processing of algae [16].
Oxygenated PAH (O-PAH) species can be formed from the combustion of solid
and liquid biofuels [e.g. 17]. There is considerable interest in this because of the
importance in relation to the health aspects [10–16] and due to the role that they play
in climate chemistry [4, 18, 19]. It is known that black carbon absorbs visible solar
radiation; however, the organic fraction has also been found to absorb at the shorter
visible wavelengths. Some of these species are thought to arise by secondary reaction
in the atmosphere, which comes about because of the formation of oxygenated PAH
species that are able to react to give larger molecular structures.
In all these applications, the bio-replacements, although renewable and sustain-
able, can be carbon containing, and consequently, there is a route available leading
to the formation of PAH species. The key to the understanding of this is a knowledge
of the mechanism of formation of PAH from the hydrocarbon component, although
some petroleum-based fuels already contain a small amount of PAH species. Thus,
fuels from bio-oils are more likely to produce PAH than an alcohol such as ethanol;
butanol containing a higher carbon content than ethanol would produce slightly more
PAH.
The Formation of PAH Compounds from the Combustion of Biofuels 107

2 The Formation of Polyaromatic Hydrocarbons (PAH)

from Hydrocarbons

The pyrolysis and oxidation reactions that take place during the combustion of fossil
fuels result in the generation of a number of species which can react and form into
benzene. This then reacts to produce small PAH molecules which grow to produce
larger PAH molecules and subsequently into soot particles. This only occurs if there
are sufficient building block materials available; that is, if the fuel–air ratio is rich
enough for soot to be formed. The routes to PAH formation and to soot are thus
intertwined; PAH in combustion products are essentially byproducts of the soot
forming process and consequently also linked in steady-state combustion equipment
to CO emissions.
Historically, a number of mechanisms have been postulated based on acetylene
(ethyne) as the main precursor. The most widely used chemical model is the Hydrogen
Abstraction C2 H2 Addition (HACA) set of reactions [20], and together with the
intermediate CPDyl (cyclopentadienyl) gives the soot precursors, some of which
have five-membered rings and the PAH species. It should be noted that the PAH
species act as the precursor to soot formation and may be produced and emitted even
below the critical limit for soot formation.
In the HACA mechanism [20], the initial pyrolysis product of fossil fuels is
acetylene which polymerizes to give benzene, which go on to form larger PAH
species [20–23]. The structures of the PAH present in the combustion products are
determined by the thermodynamic stabilities of the molecules produced.
The HACA mechanism is set out below in a simplified form:

Ai + H = Ai− + H2
Ai− + C2 H2 = AiC2 H2
AiC2 H2 + C2 H2 = Ai+1 + H·

where Ai is an aromatic molecule with i-peri-condensed rings and Ai− is its radical.
So, overall:

Ai− + C2 H2 = products = PAH growth

where A1 would be benzene and H are the hydrogen atoms produced during the
combustion and pyrolysis processes. A2 would be naphthalene and so on.
A fuller mechanism leading to the formation of PAH species is shown in Fig. 2
including the steps up to the formation of soot. Five-membered rings in the precursors
can then lead to pentagonal curved network regions and hence to bowl-shaped struc-
tures (so-called protofullerenes) and curved carbon atom sheets consistent with the
observed morphology of the soot of the ‘protographene’ PAH such as anthanthrene
which act as ‘dead-end’ PAH and terminate HACA reaction sequences.
108 A. Williams et al.

Fig. 1 IEA estimates of world energy use until 2035 (based of ref 7)

Fig. 2 Schematic of the formation of PAH and the pathway to the point of soot from [20, 22, 23]

There are a number of PAH forming/soot models which include the original HACA
model [20], but have additional growth chemistry or invoke the role of aromatic
molecules present in the fuel such as naphthalene and indene which react further
with CPD or CPDyl to form larger PAH. These may react with five-membered ring
PAH yielding open structures or undergo ring closure to yield large PAH. There
is also a more recent model [24] which involves a different growth mechanism
incorporating resonance-stabilized hydrocarbon radical chain reactions (CHRCH
mechanism), and these routes result in the formation of clusters. Some evidence for
this route also comes from ATOF-MS studies [25]. Another recent model for soot
formation includes the formation of reactive clusters and may involve ions [26].
The more general model used for the formation of soot involves stacked PAH
layers as precursors which bind together through pi-pi interactions between graphene
sheets to give crystallites in flame gases of size approximating to that of circum-
coronene (C54 H18 ). This approach is briefly outlined Fig. 3 and may involve the
icospiral mechanism described in [23] to build up the soot structure.
Most of the early understanding of the mechanism of PAH formation comes
from studies associated with diesel fuels, coal or fuel oil combustion and industrial
operations involving the pyrolysis of hydrocarbons. Investigations of soot formation
The Formation of PAH Compounds from the Combustion of Biofuels 109

Fig. 3 Schematic of soot formation through icospiral mechanism, based on [23]

during the combustion of a wide variety of fuels, gaseous and liquid hydrocarbons,
coal and biomass have allowed us to classify the properties of a number of soot forms.
These range from the nature of the structure of black carbon (BC) and the associated
organic material (OC) which arises from the partial combustion of diesel fuels and
wood.
Although hundreds of PAH exist in the environment, only 16 PAH, i.e. naph-
thalene, acenaphthylene, acenaphthene, fluorene, phenanthrene, anthracene,
fluoranthene, pyrene, benzo[a]anthracene, chrysene, benzo[b]fluoranthene,
benzo[k]fluoranthene, benzo[a]pyrene, dibenzo[a,h]anthracene,
benzo[ghi]perylene, and indeno[1,2,3-cd]pyrene, have been selected by the US Envi-
ronmental Protection Agency (US EPA) as priority pollutants, which are monitored
routinely for regulatory purposes. The toxicity of these varies from species to species.
There is still uncertainty about the process of early growth relating to PAH forma-
tion although most research has been concerned with the process of soot inception
[23, 24]. Soot inception include the dimerization of PAH species [20], clustering of
resonance-stabilized radicals [24], or ion-induced nucleation of curved PAH species
[26]. This latter route would involve other flame ions such as K+ which would mainly
be present during the combustion of solid biomass materials such as wood.
The well-developed sequence of PAH then leads to planar carbon sheets through
the addition of C2 and C4 to the carbon network in the HACA mechanism. Recently,
however, Johansson et al. [24] have pointed out that the HACA mechanism incor-
porates hydrogen abstraction with high energy barriers, whereas a scheme in which
highly reactive resonance stabilized radicals (RSR) are clustered with PAH molecules
results in the regeneration and growth of the RSR and the effective dimerization of
the PAH and extension of the protographene structures.
The early stages of the clustering of hydrocarbons by radical chain reactions
(CHRCR) mechanism are shown in Fig. 4. The first stage involves molecular growth
of the pool of RSR that acts as seeds for initiating hydrocarbon clustering. Elec-
tronic structure calculations predict a sequence of RSR which are readily generated
through radical chain reactions. Each reaction adds acetylene (C2 H2 ) or vinyl (C2 H3 )
to an RSR and generates a new RSR. Thus, these radicals are not consumed and
grow to larger species. For instance, propargyl (C3 H3 ) is an RSR that can generate
cyclopentadienyl by reaction with acetylene.
110 A. Williams et al.

Fig. 4 CHRCH (clustering of hydrocarbons by radical chain reactions) Mechanism (from [24])

The second stage comprises the clustering of hydrocarbon caused by radical chain
reactions. This mechanism provides a satisfactory way of describing the formation
of PAH species, but the clustering step has also to explain how the well-defined onion
structures occur which are seen by electron microscopy (see Fig. 3).
The formation of larger PAH species from the small precursors via the different
routes can have different outcomes as summarized in Fig. 5.
In addition, the PAH species may be methylated depending on the combustion
condition and the temperature of the cooling process of the exhaust gases, and this
process is significantly different between vehicles and stoves [27–30]. The nature of
different types of PAH, whether methylated or demethylated species from specific
sources-marine, heavy diesel and gasoline car engines have been identified [28] and
are given in Table 1.
Several studies have been made of emission factors for a number of individual PAH
species for different fuel types and combustion applications [31–36]. For practical
purposes, the total PAH emission are given, and this includes both particle bound and

Fig. 5 Different types of PAH species that can occur

The Formation of PAH Compounds from the Combustion of Biofuels 111

Fig. 6 Mechanism of biomass pyrolysis leading to PAH and soot formation [39, 40]

Fig. 7 Lignin units, from

left to right:
p-hydroxyphenol (H),
guaiacyl (G) and syringyl (S)

Table 1 Types of PAH compound emitted by different types of engines together with emissions
from wood stoves, based on [28]
Diesel cars and wood stoves Marine engines, heavy-duty Gasoline vehicles
diesel vehicles
2–5 rings not methylated 2–3 rings, substituted with 1 or 2 ring methylated species:
naphthathalenes, methyl or ethyl groups, e.g. toluene,
phenanthrene, 7-ethyl-1-methyl phenanthrene xylene,
pyrene methyl naphthalene
e.g. pyrene e.g. toluene

gaseous PAH; the amount of particle bound PAH varies from system to system but is
usually about 10% mass. This includes the volatile naphthalene and acenaphthalene,
and the results are greater than some other methods.
A selection of emission factors is shown in Table 2. It should be noted that emis-
sions from engines and boilers are not measured using the same sampling technique
and it is not appropriate to directly compare the emission factors.
112 A. Williams et al.

Table 2 Selection of total

System PAH emission factor, References
PAH emission factors from
µg/MJ
different combustion systems
Coal, domestic 40 [31]
burning
Oil boiler, No. 2 fuel 1.4 [32]
oil
Oil boiler soy 1.1 [32]
biodiesel
Oil boiler, animal 0.6 [32]
biodiesel
HFO boiler, older 60 [31]
design
Diesel truck, heavy 26 [33]
duty, DI diesel fuel
Same engine, with 17 [33]
rapeseed fuel
Wood stove, pellet 2 [36]
stove
Wood stove, modern, 36 [36]
logwood
Wood stove, older 1000 [35]
design logwood

From the data given here, it is seen that for the same type of equipment and on
the same thermal basis that the biofuels emit less that the fossil fuel.

3 The Formation of PAH from Biomass

PAH species can be formed during the combustion of all types of biofuel. From the
early development of PAH species to the inception of soot, there are also opportunities
to form O-PAH species, which are discussed in the next section. Studies have been
concerned with the interpretation of the identity and sources of PAH in emissions
from burning biomass together with evidence from other sources [5, 17, 28, 35–
38]. Particular attention is directed to those species which play an important role a
determining the toxicity of combustion products.
The pyrolysis of biofuels during combustion leads to a number of PAH species
resulting from the carbon content of the biofuel. Thus, biogas which will be very low
carbon containing will produce very low amounts of PAH, whilst solid biofuels such
as wood produce greater amounts. They also produce oxygenated PAH as a result of
being highly oxygen-containing compounds. The two main classes of O-compounds:
firstly, cellulose products form small PAH via HACA reactions and together with
CPDyl, etc., give the soot precursors, some of which have five-membered rings [17,
The Formation of PAH Compounds from the Combustion of Biofuels 113

35, 39, 40]. The initial pyrolysis products of the lignin are phenols, e.g. 4-methyl-2-
methoxyphenol from eugenol which goes on to form precursors of larger PAH species
[20, 22]. The five-membered rings in these PAH compounds formed by both routes
can lose an H by H-abstraction and react with an oxygen atom forming compounds
such as 1-indanone, 9-fluorenone and peri-naphthalenone. Five-membered rings in
the precursors can then lead to pentagonal-curved network regions and hence to
bowl-shaped structures (so-called protofullerenes) and curved carbon atom sheets
consistent with the observed morphology of the soot, as previously described. Some
of these O-PAH are the analogues of the ‘protographene’ PAH such as anthanthrene
which act as ‘dead-end’ PAH and terminate HACA reaction sequences.
The smaller O-PAH also plays a role in the reactions leading to the formation
of the larger molecules. It is suggested that the formation of soot involves stacked
PAH layers as precursors which bind together through pi-pi interactions between
graphene sheets to give crystallites in flame gases of size approximating to that of
circumcoronene (C54 H18 ). As well as growing in size they can, under certain circum-
stances, they can be oxidized by reaction with O or O2 to give ketones. From the early
development of PAH species to the inception of soot, there are many opportunities
to form larger O-PAH and some of these become trapped in the soot matrix leading
to a disordered soot structure, which in turn are more reactive to oxidation. This may
be advantageous in the case of regenerative diesel soot traps.
The early stages involve the liberation of small aromatics, especially indene and
naphthalene, formed from cyclopentadienyl radical which is the product of elimina-
tion of CO from phenols [41–43]. This sequence is particularly relevant to biomass
combustion in the light of the contribution of lignin. The reactive hydrocarbon
acenaphthylene has also been implicated as a product from further reactions here.

Lignin coumaryl, coniferyl, sinapyl alcohols phenols CpDyl

This sequence produces aromatic ketones in the ratio of O-PAH to PAH given by
the thermodynamic stabilities. Such reaction in the precursors can lead to pentagonal
regions and hence to bowl-shaped structures, protofullerenes and curved C atom
sheets consistent with the observed morphology of the soot. Some of these O-PAH
are structural analogues of protographenes such as coronene and anthanthrene which
act as dead ends and terminate HACA reaction sequences.
However, extension to smaller PAH along with CPDyl accounted for the soot
precursors, some of which had five-membered rings and could lose H by abstraction
through reaction with O or O2 .
The mechanism is based on two main studies biomass studies [39, 40], and a
more detailed biomass pyrolysis model is now available [42]. Routes leading to the
reactions of the lignin decomposition model have also been studied [42, 43].
114 A. Williams et al.

4 OXY-PAH (Designated Here as O-PAH)

O-PAH species can be formed during the combustion of all types of biofuel as well as
fossil fuels. From the early development of PAH species to the inception of soot, there
are also opportunities to form O-PAH species which are discussed in the next section.
Studies have been concerned with the interpretation of the identity and sources of
PAH in emissions from burning biomass together with evidence from other sources
[5, 17, 28, 35, 39, 40] Particular attention is directed to those species which play a
determining the toxicity of combustion products [44–47]. The ratio of the O-PAH
to the PAH present in the combustion products is determined by the thermodynamic
stabilities of these molecules; these can act as precursors of the oxygenated involatile
atmospheric PAH species observed. Here, we particularly consider the interpretation
of the identity and sources of oxygen-containing PAH in emissions from burning
biomass, which is a major source, together with evidence from other sources.
The experimental data used to draw conclusions about the formation of O-PAH
are taken from the following sources:
1. Samples of soot were taken from a wood burning stove. Soot formation
was determined firstly by collecting and filtering all particulate pine wood
combustion products [39, 40].
2. Fitzpatrick et al. [17] studied the emission of oxygenated species from the
combustion of pine wood and its relation to soot formation using a firebox.
Samples were analysed by GC–MS.
3. Liquid biomass model compounds were burned in a wick burner [41] using a
2 mm wick diameter, a height of 7 mm and surrounding uniform flow.
4. Shen et al. [37, 38] measured the emission factors for 9-fluorenone, anthracene-
9, 10-dione, benzanthrone and benzo[a]anthracene-7,10 dione from wood stoves
in rural China. They also measured the emission of N-PAH.
5. Measurements have also been made of O-PAH in the atmosphere, and some of
this work is used here [48, 49]. The data are complicated because samples come
from many sources, and some have undergone oxidation in the atmosphere.
Typically, samples of urban air particulates were collected on quartz filters and
analysed by a variety of methods: Py-GC-MS, Py-GC-AED for selective detec-
tion of oxygen compounds; or samples were separated by extraction with super-
critical carbon dioxide, and multi-dimensional chromatography and detection
by high-resolution MS to give molecular formulae of individual compounds.

4.1 O-PAH Sources

Broadly, classified experimental data from analysis of biomass emissions revealed

the information shown in Tables 3 and 4. There are interesting differences in the
combustion of wood in a stove, which are shown in Table 3, between the emissions
from the lump wood and the smaller particles of wood in briquettes due to differences
The Formation of PAH Compounds from the Combustion of Biofuels 115

Table 3 Relative masses of

Emission Stove lump Stove sawdust Pine pyrolysis
typical emissions from a pine
mg/MJ wood briquette products
wood-fired unit [35]. Pine
pyrolysis products are given PAH 14.0 12.0 3200
in the RH column Alkyl PAH 7.6 4.7 1900
Phenols 13.0 2.5 29,100
O-PAC 1.7 1.2

Table 4 Emission factors for

O-PAH species B.Pt °C Lump wood
O-PAH in flue gas from a
wood stove, (µg/MJ) [35] 1-naphthaldehyde 160 310

2-Me-1,4-naphthoquinone 212 100

1-indanone 243 550

Perinaphthenone (phenalen-1-one) 273 160

9-fluorenone 342 350

Anthraquinone (anthracene-9,10-dione) 380 120

Pyrene-1-carbaldehyde 160

in the way in which the biomass decomposition products are released and burn from
the reacting solid source.
Wood that burns in an open crib or in wild fires do so with a considerable amount
of excess air at least at the flame extremity because they are not enclosed. The
temperatures in a wood burning stove are much less. In terms of mass, PAH makes
the most important contribution to emissions, but there are also significant amounts
of phenols, phenoxy compounds and alkyl PAH.
In Table 3, the ‘phenols’ were found to consist of 33% phenol, 39% cresols (o,
m and p) and 10% dimethyl phenols. The amounts of the emitted PAH and alkyl
PAH combustion products follow that of the pyrolysis products, but at lower concen-
trations. This suggests either that they result from ‘slippage’ or that the combus-
tion products are in partial equilibrium. The term O-PAC includes the polyaromatic
116 A. Williams et al.

Table 5 O-PAH in urban air particulates (ng/m3 ) and EFs for wood stove particulates (mg/kg) [38]
Species Urban air Afghanistan [46] EFs [13]
particulates [48] mg/kg
9-Fluorenone 0.05 0.27 0.84
Phenanthrene-9-carboxaldehyde 0.06
10-ethylanthracene-9-carboxaldehyde 0.05
Benz[a]anthracene-7,12-dione 0.19 0.07
Anthraquinone 0.21 0.46 0.61
2-methyl 0.20 0.36
anthraquinone
Benzanthrone 1.01 6.3 0.67
Naphthacenequinone 0.14

compounds as well as the O-PAH. Table 4 gives emission factors for O-PAH in flue
gas from the same wood stove as in Fig. 3, (µg/MJ) [38].
The O-PAH found in the flue gases from this unit forms about 5% of the total
PAH. Table 5 gives data obtained for O-PAH on particulates in urban air in a UK city
[48, 49] and in rural Afghanistan [46] are shown in Table 5, together with emission
factors (EF) from a wood burning stove in China (Shen et al. [38]). Some of the
pollutants are clearly from wood burning, and others (aldehydes) are from diesel
fuelled vehicles.
It is interesting to note that fast pyrolysis and the analysis of bio-oils produced by
Stas et al. [50] give O-PAH compounds, but they are different to those emitted in the
flue gases. The pollutants found in the environment (Table 5) are also different to those
found in bio-oil. There are differences also between the total O-PAH emitted from
a wood stove and those present on air particulates, presumably due to the volatility
of the lower O-PAH species. The O-PAH compounds observed in the combustion
products of biomass, especially those on the particulates cover only a small fraction of
the range produced-small volatile molecules are not collected and the larger species
exist as organic aerosols. The O-PAH species formed by combustion can be classified
into four broad groups (but with considerable overlap) which are set out below.
Group 1. Phenolic Products
Lignin is a major polymeric component of biomass which decomposes in a complex
manner as a result of differences in the thermal stability of its oxygen functional
groups. Early decomposition products are guaiacol and syringol derivatives followed
by highly substituted phenols which lose alkyl substituents [51].
A number of phenols, 30 monohydric, 14 dihydric and 44 methoxy and dimethoxy
phenols have been identified in bio-oil [50, 51] and can be considered as the primary
products input to the biomass combustion process. Lignin decomposes in the temper-
ature range 240–500 °C, and the process is complex [51–53]. These early products
are shown in Fig. 8 and can react further giving small furans and PAH; some of these
have five-membered rings. The products will depend on the pyrolysis time [52, 53].
The Formation of PAH Compounds from the Combustion of Biofuels 117

Fig. 8 Pyrolysis of pine as a function of pyrolysis time at 500 °C (increasing from left to right):
hydroxyfurfural; levoglucosan; methoxyphenol; 4-vinyl-2-methoxyphenol

Fig. 9 Outline of reaction paths from phenol via CPD to naphthalene or to dibenzofuran

Fig. 10
Methylbenzo[b]naphtho[1,2-
d]furan,

Fig. 11 Oxidation of a 5-ring PAH species to a ketone

118 A. Williams et al.

Detailed analyses have been made of the fission of the bridges in lignin leading
both to small molecules and to heavier condensates the tars [52]. Many of these
products are found in the emission products from the combustion of biomass, for
example [5, 17, 39].
Group 2. Compounds derived from phenols
The phenols produced from the initial decomposition step of lignin pyrolysis can
themselves react further with other phenolic products or can react further producing
small furans and PAH compounds; some of these have five-membered rings [54]. The
most important biomass products are then a series of PAH some of which become
oxidized forming compounds such as 1-indanone, 9-fluorenone, benzofluorene-ones
and larger O-PAH formed by the decomposition of lignin. In addition, there are
a range of cyclic compounds, but these are treated as a special case and mainly
discussed in Group 3.
An example of a product from lignin pyrolysis is eugenol [41, 43]. This in
turn undergoes thermal decomposition giving a range of oxygenates, some with
a single ring but also some are O-PAH compounds [43] which in turn decom-
pose reaching a maximum at about 700 °C. The main phenolic product of thermal
decomposition is 2-methoxy-4-methylphenol [41, 43]. Progressive degradation of
the side chain of eugenol produces simpler phenols which are sources of substituted
CPDyl compounds (C5 H5 ). The most important biomass products are then a series
of PAH some of which become oxidized forming compounds such as 1-indanone,
9-fluorenone (see Tables 2 and 3), benzofluorenones and larger O-PAH formed by
the decomposition of lignin.
A number of studies have been made of the rapid pyrolysis of a variety of the
phenols produced by the pyrolysis of biomass [54–56]. It has been found that
phenol produces, amongst other PAH species, dibenzofuran, m-cresol produces
1,2-dimethoxydibenzofuran, guaiacol produces benzofuran; pyrocatechol produces
1-dihydroindanone; and syringol produces related polymethoxyphenols and ethyl
phenol-5-methylbenzofuran and analogues.
Phenols initially lose H atoms mainly by abstraction processes, and the phenoxy
groups so formed rearrange to form cyclic-O compounds (such as furans) by conden-
sation reactions or phenols decompose with elimination of CO to form cyclopenta-
diene (CPDyl) with subsequent reaction [54]. The is cyclohexadiene from which are
derived CPDyl structures; these in turn isomerise to naphthalenes and indenes which
are the source of a number of O-PAH compounds such as dibenzofuran, (see Tables 4
and 5), alkylated dibenzofurans and also benzonaphthofurans. But this route also
leads to other larger PAH species and the formation of soot precursors [20, 22, 23].
Larger cyclic-O compounds are formed by condensation reactions of phenols,
e.g. methylbenzo[b]naphtho[1,2-d]furan, which can be formed by the interaction of
phenolic products [55].
Kojima et al. suggested [55] that the ortho position of phenol is important in the
formation of O-PAH and that such phenols are abundant in biomass emissions.
The Formation of PAH Compounds from the Combustion of Biofuels 119

Group 3. O-PAH
As discussed in the earlier section, a group of oxygen-containing PAH represented by
benzofuran, dibenzofuran and benzonaphthofuran and their substituted derivatives
can be formed by reactions of phenoxy in which CO is not eliminated. Thus, the
model substance for lignin, guaiacol [53, 56] yields benzofuran. These can also arise
from the formation of PAH compounds formed by HACA [20, 24]; (ii) through
cyclopentadiene (CPD) generated by decomposition of phenoxy radical; but with
further reaction with naphthalene and indene to form larger PAH which may be
further condensed with CPD, cyclodehydrogenation and ring closure to yield large
PAH. After early growth, these give flat sheets of PAH. The routes can also lead to
curved PAH with cyclopenta groups on the periphery.
Group 4. PAH aldehydes and ketones
It is clear the aldehydes are formed during the combustion of wood (Table 2) but is
also abundant in the atmosphere (Table 5) because of the contribution of emissions
from vehicles, particularly diesel vehicles. Diesel vehicles fuelled by conventional
or biodiesel fuels emit many lower aldehydes as well. In the case of diesel emissions,
these are derived either from unburned fuel (alkyl derivatives) or PAH originating as
combustion products in the aromatic fraction of diesel fuels which may contain two to
three PAH with short alkyl side chains [33]. Not all such alkyl compounds are burned,
see for example Table 3 which lists some of the alkyl anthracene/phenanthrene, alkyl
fluoranthene, pyrene carboxyaldehydes and benzofluorenones of urban air particu-
lates collected during a pollution episode. The routes leading to their formation are
complex because they can involve atmospheric oxidation.
The ketones are formed by the oxidation of PAH which contain 5-membered
rings, e.g. PAH ketones such as 9-fluorenone, naphthalenone and cyclo-
pentaphenanthrenone, or alkyl PAH. Taken with the presence of ketone derivatives
of 5-membered ring PAHs, such as the parent [Shen et al. 37, 38] and numerous 9-
fluorenone derivatives found in urban air, examples such as those in Table 5, provide
abundant evidence for the participation of oxidized hydrocarbons in biomass combus-
tion emission chemistry, including soot formation. Table 5 compares concentrations
of some representative O-PAH levels of the order of 0.01 ngm−3 are accessible to
quantitation with these methods. A typical route leading to formation is by oxidation.
The ratio of the O-PAH to the PAH present in the combustion products is deter-
mined by the thermodynamic stabilities of these compounds; these can act as precur-
sors of the oxygenated involatile PAH species observed in the atmosphere. Five-
membered rings in the precursors can then lead to pentagonal-curved network regions
and hence to bowl-shaped structures (so-called protofullerenes) and curved carbon
atom sheets consistent with the observed morphology of the soot. Some of these
O-PAH are the analogues of the ‘protographene’ PAH such as benzo[ghi]perylene
which act as ‘dead-end’ PAH and terminate HACA reaction sequences.
120 A. Williams et al.

5 Toxicity of PAH and O-PAH Compounds

The toxicity of the combustion products from biomass burning is associated with the
gas phase contribution and from the fine particles (<PM2.5) and the PAH compounds
associated with it. The toxicity of the combustion products can be determined
by the B[a]P-eq-based PAH concentration (see Polycyclic Aromatic Hydrocarbons
Factsheet https://ec.europa.eu/jrc/sites/jrcsh/files/Factsheet%20PAH_0.pdf).
The presence of ketone and quinone derivatives of PAH may pose significant
health hazards in addition to the carcinogenesis properties of some PAH [45, 57].
Indeed, O-PAH may be the key compounds in the toxicity of particulate matter; it has
been suggested by Walgraeve et al. [45] and others that PAH quinones are more toxic
than the corresponding PAH. Moreover, measurements on NIST Standard Reference
Samples (SRMs) revealed [44] that levels of O-PAH similar to or greater than those
of PAH, with emission factors for residential stoves 1–2 orders of magnitude greater
than for open burning, Shen et al. [37, 38].
There are seasonal effects which are indicated by variations with time of the ratio
O-PAC/parent PAH, and of vapour/particle phase equilibria [27]. The mode of action
of quinones is thought to differ from that of PAC in general in terms of mutagenicity
and carcinogenicity.
It follows that candidates must be sought for extending the EPA16 list to make it
more relevant to the products of biomass combustion. Andersson et al. [6] have
proposed that benzofurans and quinones are representative O-PAH and the 10
proposed [6] are widespread in the environment. Other oxygen compounds in emis-
sions are the aldehyde and keto (quinone) derivatives of PAH and O-PAH resulting
from the oxidation of alkyl 2–4 ring aromatics as previously discussed.
The most toxic PAH compounds are the nitro-PAH [58], but they are present
in only small quantities in the atmosphere. It was shown [58] that in a loca-
tion in China biomass burning was the main source of cresols, 2,6-dimethyl-4-
nitrophenol, 4-nitrocatechol and methyl-nitro-catechol, but other major sources were
coal combustion and vehicles.

5.1 Control of PAH and O-PAH Emissions

It is seen from the previous discussion that the routes leading to both smoke formation
and to PAH are linked in the combustion processes, and to a lesser extent to O-PAH
since this may also be formed post-combustion. Not only are there toxicity effects
from the PAH family as discussed above, but these molecules can have an effect on
the climate [4, 18, 19, 58]. The total PAH emitted together with O-PAH and the soot
can be minimized by better combustion control. Thus, in the case of biomass burning
in a pellet stove the amount of PAH formed is 0.1% of the soot formed [5] and it is
likely, but there is a lack of evidence that the PAH emission is proportional to the
soot emitted. The O-PAH is about 5% of the total PAH formed (see Table 4) but
The Formation of PAH Compounds from the Combustion of Biofuels 121

can increase by reactions in the atmosphere. The use of alcohol fuels in vehicles is
advantageous because of the lower carbon content and lower soot forming tendency
though PAH can be formed from ethanol [59].
In all cases, they can be further reduced by catalytic combustion, for example as
described in reference [60].

6 Conclusions

An analysis has been made of the extent of PAH and the oxygenates (O-PAH and
phenols) present in biomass combustion emissions in exhaust gases and in urban air
by a number of techniques. It is difficult to compare the toxic impact from different
sources because measurement techniques are different; products from engines are
measured at a specific filter temperature whilst different filter temperatures are used
for stationary units such as stoves or furnaces. Nevertheless, it is seen that PAH is
linked to the formation of smoke (and CO) and is best controlled by optimizing
the overall combustion process, and this would involve any staged combustion and
the cooling of the combustion products. Oxygenated PAH products such as O-PAH
could be decreased in this way in the exhaust gases, but additional reactions in the
atmosphere may influence the atmospheric concentrations.
The choice of fuel can be advantageous depending on the application. In the case
of bio-oils, there could be significant differences between raw and processed bio-
oils, the final refined products of flash pyrolysis and alcohols, much depending on the
carbon and oxygen content and the aromatic content. In the case of the combustion of
solid fuels, the principal contributors are the numerous phenols liberated by pyrolysis
of lignin and their secondary pyrolysis products, especially if they are only partially
combusted.

Acknowledgements We acknowledge helpful discussions with our colleagues.

References

1. IARC Diesel engine exhaust carcinogens. The Lancet Oncology (online). 15 June 2012. https://
doi.org/10.1016/S1470-2045(12)70280-2
2. Abdel-Shafy HI, Mansour MSM (2016) A review on polycyclic aromatic hydrocarbons: source,
environmental impact, effect on human health and remediation. Egypt J Pet 25:107–123
3. Wu D, Li Q, Ding X, Sun J, Li D, Fu H et al (2018) Primary particulate matter emitted from
heavy fuel and diesel oil combustion in a typical container ship: characteristics and toxicity.
Environ Sci Technol 52:12943–12951
4. Samburova V, Connolly J, Gyawali M, Yatavelli RLN, Watts AC et al (2016) Polycyclic
aromatic hydrocarbons in biomass-burning emissions and their contribution to light absorption
and aerosol toxicity. Sci Total Environ 568:391–401
122 A. Williams et al.

5. Orasche J, Seidel T, Hartmann H, Schnelle-Kreis J, Chow JC, Ruppert H, Zimmermann R

(2012) Comparison of emissions from wood combustion. Part 1: emission factors and charac-
teristics from different small-scale residential heating appliances considering particulate matter
and polycyclic aromatic hydrocarbon (PAH)-related toxicological potential of particle-bound
organic species. Energ Fuels 26:6695−6704
6. Andersson J, Achten TC (2015) A critical look at the 16 EPA PAHs. Polycycl Aromat Comp
35:143–146
7. World Energy Outlook (2013) © OECD/IEA, 2013, fig. 2.5, p 63
8. Williams A, Jones JM, Ma L, Pourkashanian M (2012) Pollutants from the combustion of solid
biomass fuels. Prog Energy Combust Sci 38:113–137
9. Knothe G, Razon LF (2017) Biodiesel fuels. Prog Energ Combust Sci 58:36–59
10. Zabed H, Sahu JN, Suely A, Boyce AN, Faruq G (2017) Bioethanol production from renewable
sources: current perspectives and technological progress. Renew Sustain Energ Rev 71:475–501
11. Wang M, Dewil R, Maniatis K, Wheeldon J, Tan T, Baeyens J, Fang Y (2019) Biomass-derived
aviation fuels: challenges and perspective. Prog Energ Combust Sci 74:31–49
12. International Air Transport Association (2019) IATA Fact Sheet: Sustainable Aviation Fuels,
May 2019
13. Sharifzadeh M, Sadeqzadeh M, Guo M, Borhani TN, Murthy Konda NVSN, Garcia MC, Wang
L, Jason Hallett J, Shah NP (2019) The multi-scale challenges of biomass fast pyrolysis and
bio-oil upgrading: review of the state of art and future research directions. Energ Combust Sci
71:1–80
14. Bartle KD, Fitzpatrick EM, Jones JM, Kubacki ML, Plant R, Pourkashanian M, Ross AB,
Williams A (2011) The combustion of droplets of liquid fuels and biomass particles. Fuel
90:1113–1119
15. Tótha P, Ögrena Y, Sepmana A, Vikströma T, Grenc P, Wiinikkaa H (2019) Spray combustion
of biomass fast pyrolysis oil: experiments and modeling. Fuel 237:589–591
16. Lea-Langton AR, Biller P, Ross AB (2013) Investigation of HTL derived micro-algal bio-
crudes as substitutes for marine shipping fuels. In: The 3rd International conference on algal
biomass, biofuels and bioproducts, Toronto, June 2013
17. Fitzpatrick EM, Ross AB, Bates J, Andrews GE, Jones JM, Phylaktou H, Pourkashanian M,
Williams A (2007) Emission of oxygenated species from the combustion of Pine wood and its
relation to soot formation. Proc Saf Environ Protect 85:430–440
18. Trostl J, Chuang WK, Gordon H, Heinritzi M, Yan C, Molteni U, Ahlm L, Frege C, Bianchi
F et al (2016) The role of low-volatility organic compounds in initial particle growth in the
atmosphere. Nature 533(7604):527–531
19. Barsanti KC, Kroll JH, Thornton JA (2017) Formation of low-volatility organic compounds in
the atmosphere. Recent advancements and insights. J Phys Chem Lett 8:1503−1511
20. Frenklach M, Wang H (1994) In: Bockhorn H (ed) Soot formation in combustion: mechanisms
and models. Springer, Berlin, pp 162−192
21. Colket MB, Hall RJ (1994) In: Bockhorn H (ed) Soot formation in combustion: mechanisms
and models. Springer, Berlin, pp 442−470
22. Smedley JM, Williams A, Bartle KD (1992) A Mechanism for the formation of soot particles
and soot deposits. Combust Flame 91:71–82
23. Wang Y, Chung SH (2019) Soot formation in laminar counterflow flames. Prog Energ Combust
Sci 74:152–238
24. Johannsson KO, Head-Gordon MP, Schrader PE, Wilson, KR, Michelsen HA (2018)
Resonance-stabilized hydrocarbon radical chain reactions may explain soot inception and
growth. Science 361:997–1000
25. Wilson JM, Baeza-Romero MT, Jones JM, Pourkashanian M, Williams A, Lea-Langton AR,
Ross AB, Bartle KD (2013) Soot formation from the combustion of biomass pyrolysis products
and a hydrocarbon fuel, n-decane: an ATOFMS study. Energ Fuels 27:1668–1678
26. Bowal K, Martin JW, Misquitta AJ, Kraft M (2019) Ion induced soot nucleation using a new
potential for curved aromatics. Combust Sci Tech 191:747–765
The Formation of PAH Compounds from the Combustion of Biofuels 123

27. Muñoz M, Haag R, Honegger P, Zeyer K, Mohn J, Comte P, Czerwinski J, Heeb NV (2018) Co-
formation and co-release of genotoxic PAHs, alkyl-PAHs and soot nanoparticles from gasoline
direct injection vehicles. Atmos Environ 178:242–254
28. Czech H, Stengel B, Adam T, Sklorz M, Streibel T, Zimmermann R (2017) A chemometric
investigation of aromatic emission profiles from a marine engine in comparison with residential
wood combustion and road traffic: implications for source apportionment inside and outside
sulphur emission control areas Atmos. Environ 167:212–222
29. Lea-Langton AR, Ross AB, Bartle KD, Andrews GE, Jones JM, Pourkashanian M, Williams
A (2012) Low temperature PAH formation in diesel combustion. JAAP 103:119–125
30. Lea-Langton AR, Andrews GE, Bartle KD, Jones JM, Williams A (2015) Combustion and
pyrolysis reactions of alkylated polycyclic aromatic compounds: the decomposition of 13C
methylarenes in relation to diesel engine emissions. Fuel 158:719–724
31. Ravindra K, Sokhi R, Van Grieken R (2008) Atmospheric polycyclic aromatic hydrocarbons:
source attribution, emission factors and regulation. Atmos Environ 42:2895–2921
32. Miller CA Characterizing Emissions from the Combustion of Biofuels U.S. Environmental
Protection Agency Office of Research and Development National Risk Management Research
Laboratory Air Pollution Prevention and Control Division Research Triangle Park, NC
33. Lea-Langton AR, Li H, Andrews GE (2008) Comparison of particulate PAH emissions for
diesel, biodiesel and cooking oil using a heavy-duty DI diesel engine. SAE Technical Paper
Series 2008-01-1811
34. Dandajeh HA, Talibi M, Ladommatos N, Hellier P (2019) Influence of combustion charac-
teristics and fuel composition on exhaust PAHs in a compression ignition engine. Energies
12:2575
35. Ross AB, Jones JM, Chaiklangmuang S, Pourkashanian M, Williams A, Kubica K, Andersson
JT, Kerst M, Danilhelka P, Bartle KD (2002) Measurements and prediction of the emission of
pollutants from the combustion of coal and biomass in a fixed bed. Fuel 81:571–582
36. Czech H, Miersch T, Orasche J, Abbaszade G, Olli Sippula O, Tissari et al (2018) Chemical
composition and speciation of particulate organic matter from modern residential small-scale
wood combustion appliances. Sci Total Environ 612:636–648
37. Shen G, Tao S, Wang W, Yang Y, Ding J, Xue M, Min Y et al (2011) Emission of oxygenated
polycyclic aromatic hydrocarbons from indoor solid fuel combustion. Environ Sci Technol
45:3459–3465
38. Shen G, Tao S, Wei S, Zhang Y, Wang R et al (2012) Emissions of parent, nitro, and oxygenated
polycyclic aromatic hydrocarbons from residential wood combustion in rural China. Environ
Sci Technol 46:8123–8130
39. Fitzpatrick EM, Jones JM, Pourkashanian M, Ross AB, Williams A, Bartle KD (2008)
Mechanistic aspects of soot formation from the combustion of pine wood. Energ Fuels
22:3771–3718
40. Fitzpatrick EM, Jones JM, Pourkashanian M, Ross AB, Williams A, Bartle KD, Kubica K
(2009) The mechanism of the formation of soot and other pollutants during the co-firing of
coal and pine wood in a fixed-bed combustor. Fuel 88:2409–2417
41. Atiku FA, Lea-Langton AR, Bartle KD, Jones JM, Williams A, Burns I, Humphries G (2017)
Some aspects of the mechanism of formation of smoke from the combustion of wood. Energ
Fuels 31:1935–1944
42. Ranzi E, Debiagi PEA, Frassoldati AA (2017) Mathematical modeling of fast biomass pyrol-
ysis and bio-oil formation: kinetic mechanism of biomass pyrolysis. ACS Sustain Chem Eng
5:2867–2881
43. Ledesma EB, Campos C, Cranmer DJ, Foytik BL, Ton MN et al (2013) Vapor-phase cracking
of eugenol: distribution of tar products as functions of temperature and residence time. Energ
Fuels 27:868–878
44. Layshock JA, Wilson G, Anderson KA (2010) Ketone and quinone-substituted polycyclic
aromatic hydrocarbons in mussel tissue, sediment, urban dust, and diesel particulate matrices.
Environ Toxicol Chem 29:2450–2460
124 A. Williams et al.

45. Walgraeve C, Demeestere K, Dewulf J, Zimmermann R, Van Langenhovewingbows H (2010)

Oxygenated polycyclic aromatic hydrocarbons in atmospheric particulate matter: molecular
characterization and occurrence. Environ 44:1831–1846
46. Wingfors H, Hägglund L, Magnusson R (2011) Characterization of the size-distribution of
aerosols and particle-bound content of oxygenated PAHs, PAHs, and n-alkanes in urban
environments in Afghanistan. Atmos Environ 45:4360–4369
47. Zhou J, Zotter P, Bruns EA, Stefenelli G, Bhattu D, Brown S et al (2018) Particle-bound reactive
oxygen species (PB-ROS) emissions and formation pathways in residential wood smoke under
different combustion and aging conditions. Atmos Chem Phys 18:6985–7000
48. Lewis AC, Askey SA, Holden KM, Bartle KD, Pilling MJ (1997) On-line. LC-GC-and LC-
GC-MS—multidimensional methods for the analysis of PAC in fuels, combustion emissions
and atmospheric samples. J High Res Chromatogr 20:109–114
49. Lewis AC, Kupiszewska D, Bartle KD, Pilling MJ (1995) City centre concentrations
of polycyclic aromatic hydrocarbons using supercritical fluid extraction. Atmos Environ
29:1531–1542
50. Staš M, Kubička D, Chudoba J, PospíšI M (2014) Overview of analytical methods used for
chemical characterization of pyrolysis bio-oil. Energ Fuels 28:385–402
51. Pelucchi M, Cavallotti C, Cuoci A, Faravelli T, Frassoldati A, Ranzi E (2019) Detailed kinetics
of substituted phenolic species in pyrolysis bio-oils. React Chem Eng 4:490–506
52. Zhang T, Li X, Qiao X, Zheng M, Guo L, Song W, Lin W (2016) Initial mechanisms
for an overall behavior of lignin pyrolysis through large-scale ReaxFF molecular dynamics
simulations. Energ Fuels 30:3140–3150
53. Maduskar S, Facas GC, Papageorgiou C, Williams CL (2018) Five rules for measuring biomass
pyrolysis rates: pulse heated analysis of solid reaction kinetics of ignocellulosic biomass. ACS
Sustain Chem. Eng. 6:1387–1399
54. Akazawa M, Kojima Y, Kato Y (2015) Formation of polycyclic compounds from phenols by
fast pyrolysis. EC Agric 1:67–85
55. Kojima Y, Inazu K, Hisamatsu Y, Okochi H, Baba T, Nagoya T (2010) Comparison of PAHs,
nitro-PAHs and oxy-PAHs associated with airborne particulate matter at roadside and urban
background sites in downtown Tokyo, Japan. Polycycl Aromat Comp 30:321–333
56. Liu C, Ye L, Yuan W, Zhang Y, Zou J, Yang J et al (2018) Investigation on pyrolysis mechanism
of guaiacol as lignin model compound at atmospheric pressure. Fuel 232:632–638
57. Barradas-Gimate A, Murillo-Tovar MA, Díaz-Torres JJ, Hernández-Mena L, Saldarriaga-
Noreña H et al (2017) Occurrence and potential sources of quinones associated with PM2.5 in
Guadalajara, Mexico. Atmosphere 8:140
58. Li M, Wang X, Lu C, Li R, Zhang J, Dong S, Yang L, Xue L, Chen J, Wang W (2020) Nitrated
phenols and the phenolic precursors in the atmosphere in urban Jinan, China. Sci Total Environ
714:13670
59. Viteri F, López A, Millera Á et al (2019) Influence of temperature and gas residence time on
the formation of polycyclic aromatic hydrocarbons (PAH) during the pyrolysis of ethanol. Fuel
236:820–828
60. Soufi J, Pitault I, Gelin P, Meille V, Vanoye L et al (2016) Catalytic combustion of polycyclic
aromatic hydrocarbons (PAH) over zeolite type catalysts: effect of Si/Al ratio, structure and
acidity. In: French conference on catalysis (FCCat), May 2016, Frejus, France
Review of Biomass Energy Resources
with Livestock Manure

Osama M. Selim, Juan Espindola, and Ryoichi S. Amano

Abstract This paper presents different thermochemical conversion processes on

different biomass species. The results show it is recommended to use the lowest
heating rate to allow a quasi-equilibrium state through slow heating, hence avoiding
measurement errors. Thermal degradation of the three main components of the
chicken manure was obtained. The initial results show that for the slow heating rates,
5 °C/min, the thermal degradation of the cow manure is different compared to that
one obtained from chicken manure. The Hemicellulose decomposition took place at
250 and 300 °C for the chicken manure and cow manure, respectively. The Cellulose
decomposition was started at 300 °C for chicken manure and 470 °C for cow manure.
Rice husk, the peak of the Pyrolysis of rice husk, is less distributed than the Pyrol-
ysis of chicken manure due to the absence of Hemicellulose and less Lignin content.
For CO2 gasification, the chemical reactions, for all different heating rates tested,
were endothermic. Consequently, the energy must be supplied in terms of heating to
sustain the reaction, while air gasification was exothermic, which means that the reac-
tion can be sustained without external heating where the self-ignition was observed
between 450 and 600 °C. In addition, it was observed that carbon dioxide had the
most complicated mechanism with four stages. CO-Pyrolysis, results show the 40%
RH-60%CH decreasing the activation energy by 12% compared to Chicken manure.
In addition, an increase in the mass conversion by more than 3% was achieved. The
40% CM-60% CH shows a positive result in terms of keeping an exothermic reaction
over the co-Pyrolysis process.

Keywords Thermochemical · Pyrolysis · Gasification · Co-Pyrolysis · Chicken

manure · Cow manure · Rice husk

O. M. Selim · J. Espindola · R. S. Amano (B)

University of Wisconsin Milwaukee, Milwaukee, WI 53211, USA
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 125
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_6
126 O. M. Selim et al.

1 Introduction

Energy consumption in the world is predominantly from fossil fuels for transporta-
tion and electricity generation [1]. Fossil fuels cause problems such as air pollution,
greenhouse gases, and environmental issues from the extraction. Also, high price
fluctuation and possible fuel depletion are threats to the fossil-fuel-based economy.
Furthermore, society is turning into energy consumption monitoring and low carbon
practices [2]. Consequently, interest in alternative and renewable energy sources has
been heightened. In 2020, the generation of electricity was 19% from renewable
sources, and it is expected to take 36% of the share by 2050 [1]. Biomass raises as
one of the alternatives to meet the increasing energy demand because it is abundant,
carbon-neutral, and clean. Biomass is defined an organic material that comes from
living organisms and can be converted to energy. Biomass can be classified into
first-, second-, and third-generation biomass. The first-generation biomass converts
carbohydrates and starches into biofuel (e.g., corn). The second generation includes
agricultural, animal waste, and wood, while the third generation refers to algae [3].
To convert biomass into usable bioproducts, three main technologies can be used:
(1) Mechanical conversion, (2) biochemical conversion and, (3) thermochemical
conversion [4, 5]. In mechanical conversion, mechanical forces reduce the size of
the biomass. This step usually prepares the fuel for combustion or another thermo-
chemical process. The biochemical process uses microorganisms to degrade biomass.
Among the technologies used for biological decomposition are anaerobic digestion
and hydrolytic fermentation. Anaerobic digestion produces mainly methane, while
fermentation produces ethanol [6, 7]. Thermochemical process uses heat to promote
chemical reactions that decompose the organic matter in the biomass, yielding bio-
oils and/or Syngas (depending on the temperature and atmosphere) [8]. Pyrolysis
uses an inert atmosphere such as (nitrogen or argon) to decompose the biomass
in a reactor at elevated temperatures, the products of Pyrolysis are bio-oils, non-
condensable gases, and biochar [9], while gasification uses controlled amounts of
oxygen to partially oxidize the fuel, producing Syngas and ashes [9].
There are three main technologies for Pyrolysis reactors, batch reactors, moving
bed, reactors with movement caused by mechanical forces, and reactors when the
movement I caused by fluid flow (fluidized bed) [10]. The gas yield depends on several
factors: (1) type of biomass, (2) particle size, (3) heating rate, and (4) temperature
and residence time [11]. This study aims to review recent advances in the Pyrolysis
of feedstock from the perspective of the kinetics and product yields.

1.1 Pyrolysis Process

Pyrolysis is a thermochemical process that decomposes the livestock in the absence

of oxygen. Pyrolysis happens at medium temperatures (300–800 °C) or high tempera-
tures (800–1300 °C) [5]. Pyrolysis products are (1) bio-oil (tar), (2) non-condensable
Review of Biomass Energy Resources with Livestock Manure 127

gases (or Syngas) that is a mix of different gases such as H2 , CH4 , CO, C2 HX , and
C3 Hx, and (3) char. The yield of each of these products depends on several factors,
such as the kind of feedstock, temperatures, heating rates, and residence time [11,
12]. According to the operating conditions, Pyrolysis can be classified as slow, fast,
or flash Pyrolysis [13].
Slow Pyrolysis uses low heating rates (usually less than 10 °C/s) and long resi-
dence times (from 5 to 30 min) [14]. Hussein showed that low hearing rate and
high residence time promote the completion of chemical reactions [15], giving as a
result higher yields of Syngas and biochar. Vieira et al. [16] optimized slow Pyrol-
ysis parameters for a fixed bed reactor. Results showed that the optimal conditions
were given for the longest residence time and the slowest heating rate (5 °C/min and
120 min).
Fast Pyrolysis uses heating rates higher than 100 °C/s and residence times from
0.5 to 2 s. Fast Pyrolysis has the advantage of low residence time, reducing the overall
cost of the process [14, 17]. Besides, biochar yields are much lower in fast Pyrolysis
compared with slow Pyrolysis reducing the solid handling. The main product of fast
Pyrolysis is bio-oils.
Flash Pyrolysis uses heating rates greater than 500 °C/s and residence times of
less than one second. This technology requires special reactor configuration and very
small particle size to ensure temperature uniformity [18].
Abundant research compares the yields of each technology. Yuan [17] compares
the biochar yields of fast and slow Pyrolysis of walnut shells, showing that the
mechanism of biochar formations is similar in slow and fast Pyrolysis, but the yields
of biochar in slow Pyrolysis are much higher than the yields of biochar in fast
Pyrolysis.
Al Arni [19] studied the conversion of bio-oil from sugarcane bagasse, finding
that slow Pyrolysis produces more Syngas and less bio-oil compared to fast Pyrol-
ysis. Pyrolysis follows three main steps: drying, primary Pyrolysis, and secondary
Pyrolysis [11, 20].
Drying
In the drying step, the moisture on the biomass is removed, and moisture in livestock
varies between 5 and 20% [21]. Biomass drying is mainly done to reduce the volume
of the biomass and increase the efficiency of the thermochemical process. Reed
et al. [22] showed that high moisture reduces the value of the fuel. The removal of
moisture for biomass is usually done by thermal drying. However, thermal drying
can be an expensive process, such as fuel’s value, and energy cost from drying must
be a trade-off. Researchers have focused on modeling drying processes to optimize
drying conditions for biomass [23, 24]. Studies conclude that small particle size and
low heating rates are the most effective ways to dry.
Primary Pyrolysis
Primary Pyrolysis happens at temperatures between 250 and 600 °C [11]; the chem-
ical reactions are associated with the thermal cracking of Hemicellulose, Cellulose,
and Lignin. Gaur et al. [25] compiled the degradation for different kinds of biomass,
128 O. M. Selim et al.

including the independent three components (Hemicellulose, Cellulose, and Lignin)

using thermal gravimetrical analysis (TGA). Gaur showed that Hemicellulose and
Cellulose decompose at a faster rate than Lignin. Also, the peaks of degradation
occur at temperatures between 270, 340, and 380 °C, respectively. Pyrolysis of
biomass occurs in different sets of competitive reactions (such as char formation,
depolymerization, or methanation) [26]. Yang [27] showed that the degradation of
Hemicellulose and Lignin is highly exothermic for temperatures below 500 °C; after
that temperature, the endothermic reactions are dominant, and the degradation of
Hemicellulose and Lignin becomes highly endothermic. The opposite behavior was
observed for Cellulose. For temperatures below 400 °C, the degradation of Cellulose
was highly endothermic. However, after 400 °C the exothermic reactions dominate
the decomposition of Cellulose.
Secondary Pyrolysis
Secondary Pyrolysis occurs at temperatures from 600 °C to 1,400 °C. This stage
further decomposes the products (tars) from the primary Pyrolysis, yielding Syngas
and char. Several publications [11, 15, 28] showed that hydrogen formation is
promoted at higher temperatures, which increases the heating value of the Syngas. In
addition, the char formed in the primary Pyrolysis can work as a catalyzer increasing
the total degradation at higher temperatures.

1.2 Gasification

Gasification is an efficient thermochemical decomposition of solid carbonaceous

material to produce gases of higher heating value compared to the solid precursor
in the presence of gasifying agents such as steam, air, or CO2 . The major compo-
nents of product gas are H2 , CO, CO2 , CH4 , and other higher hydrocarbons. Unde-
sirable by-products containing a mixture of heavy aromatic hydrocarbon residues,
referred to as tar, are also formed, which must be further pyrolyzed or reformed to
enhance the Syngas yield. The factors that affect the composition of Syngas evolved
from gasification are gasifying agent, temperature, and feedstock composition, and
other reactor operational parameters affect the composition of Syngas evolved from
gasification. Production of homogeneous fuel (Syngas as intermediate) from hetero-
geneous biomass is a combination of sources including chicken manure, poultry
litter, biomass, and solid wastes, and coal makes it an attractive solution over direct
combustion of these low-value heterogeneous feedstocks providing feed flexibility
along with versatility. Abundant literature is available on the effects of using different
catalysts and gasifiers, and favorable temperatures on biomass gasification [29–44].
Equations (1–5) show the major governing reactions (considering C to represent the
biomass) that occur during gasification [45].

Char Oxidation : C + O2 ⇒ CO2 (1)

Review of Biomass Energy Resources with Livestock Manure 129

Boudouard Reaction : C O2 + C  2C O (2)

Water − Gasshift Reaction : C O + H2 O  C O2 + H2 (3)

Water − Gas Reaction : +H2 O  C O + H2 (4)

Thermal Cracking : Cn Hm ⇒ C + C x Hy + H2 (5)

Chicken droppings, waste beddings, waste food, and feathers from the coops form
the major source from poultry litter. This biomass has high nitrogen content due to
proteins and amino acids, high phosphorus, and ash content, making poultry litter
a low-grade feedstock that is different from other conventional biomasses such as
wood. Such differences endeavor the motivation for an investigation into the pyrolytic
behavior of chicken litter [46]. One of these differences investigated from thermal
gravimetric analysis (TGA) was the start of decomposition at lower temperatures
compared to coal, wherein the maximum weight loss rate was observed at 643 K
[47]. The TGA results on the Pyrolysis of chicken litter showed three stages of weight
loss, unlike woodchip Pyrolysis, which occurs in two stages because of lignocellulose
content. Three main peaks of chicken litter decomposition were observed during
pyrolysis process which are corresponding to the decomposition of Cellulose and
Hemicellulose with a peak at a temperature less than that of wood chips, followed by
manure and Lignin decomposition and the last stage of residual char devolatilization.
This is unlike TGA of woodchip Pyrolysis, which showed only two stages [46].
Higher manure content is assisted in the faster decomposition of waste litter. An
increase in temperature increased Syngas yields and decreased liquid and char yields
[48]. CO2 yields increased due to decarboxylation of mineral carbonates in the ash
at temperatures above 973 K until a stable value above 1273 K [49]. Highly viscous
bio-oils produced from fast Pyrolysis of chicken litter had less acidity and higher
heating value compared to bio-oils from hardwood due to the decomposition of
protein content in a chicken litter into hydrocarbons [50].
Although the high inorganic content in the chicken litter can provide fertilizing
capabilities and catalytic activity in Syngas production, it causes handling difficul-
ties due to the low fusion temperature of inorganics present since the chicken litter
ash contains high amounts of Ca, Na, and K. Such high amounts of ash content
(>20%) in a chicken litter with phosphorous and potassium oxides make it an effi-
cient fertilizer even after gasification [51]. The gas evolution from inorganic matter
in manure, such as decarboxylation of carbonates and the low melting points of its
constituents, induces error into the modeling of product gases from high-temperature
gasification and Pyrolysis if ash is considered inert as in conventional models [52].
Chicken litter gasification was demonstrated by blending with coal and found no
change in the heating value of product gas [53]. The ultimate analysis showed
enough oxygen present in a chicken litter to gasify its carbon contents; such a method
130 O. M. Selim et al.

was called “auto-gasification” [54]. This investigation also found ineffective avail-
ability of carbon to react with oxygen due to reactions with other hydrocarbons and
water–gas reaction, causing oxidation of CO evolved to form CO2 . Use of catalysts
in steam gasification of chicken manure in a fluidized bed at temperatures around
873 K produced Syngas rich in H2 content and total gas yields with the addition of
Ni-Al2 O3 catalyst [55]. Supercritical gasification of chicken manure was also inves-
tigated [56, 57]. The abundant literature on biomass, although helpful in assessing
chicken litter, concrete understanding of the differences with biomass, and effects
of other gasifying parameters, is still lacking and much desired. Using steam as the
gasifying agent is known to generate a higher-quality Syngas with high amounts of
hydrogen content [58–60]. The high latent heat required for evaporation makes steam
the most expensive gasifying agent, and thus long reaction time is undesirable. On
the other hand, air, which is the cheapest agent, generates the lowest quality Syngas
and fastest reaction rates. Using a mixture of air (oxygen) and steam as the gasifying
agent provides the benefit of improved Syngas quality, while the exothermic reaction
of air boosts the reaction rates. Steam can be injected at later stages in gasifiers [61]
to reform the Syngas and improve its quality, or it can be mixed with air at the early
stages of the gasification [62], to give even improved concentrations of hydrogen.
The concentration of oxygen in the gasifying agent plays an important role in the
quality of Syngas evolved and the rates of reaction. Thus, the oxygen concentration
needs to be optimized to seek for the most feasible reaction conditions [63].

1.3 Co-pyrolysis and Co-gasification

Co-Pyrolysis and co-gasification are viable options to enhance the performance of

the thermochemical processes. Depending on the biomass mixed (and the ratio),
synergetic effects can promote the Syngas yield, the energy of activation, or degra-
dation rate. Guo et al. [64] demonstrated that the extractives in singular biomass
could improve the thermal degradation of the biomass components (Hemicellulose
and Cellulose). Mallick [65] showed that the binary conversion of the biomass,
depending on the ratio, can be considered as an additive equation.
Research has focused on mixing lignocellulosic biomass with coal [66, 67].
Results show that some biomasses can produce positive synergetic effects increasing
the yield of volatiles without affecting the process. He [68] mixed cotton and coal,
and results showed that the rate of decomposition increased for the co-Pyrolysis
blends compared to the biomasses stand-alone.
Mixtures of lignocellulosic biomasses have been largely studied. Another
common path is to mix lignocellulosic biomass with plastics. Burra et al. [69] studied
the co-Pyrolysis of pinewood with plastic waste. Results show non-synergetic effects
between the two biomasses. The best performance was obtained by mixing pinewood
with BPC. Hossain et al. [70] studied the co-Pyrolysis of a solid tire with rice husk
to show the possibility of getting liquid products that are comparable with petroleum
fuels. However, these liquid products are only produced if the Pyrolysis conditions
Review of Biomass Energy Resources with Livestock Manure 131

are correctly selected. Costa et al. [71] investigated mixing the rice husk with plastic
waste using different pressures and residence time; an enhancement in the biomass
conversion was observed. Also, the best performance was obtained for lower pres-
sure, high temperatures, and short residence times, or low pressure, low temperature,
and high residence time.
Co-gasification targets to promote reactions that yield hydrogen to increase the
quality of the produced Syngas. Similarly, to Pyrolysis, mixtures of lignocellu-
losic biomasses have been largely studied. Ng et al. [72] studied air gasification
of chicken manure and woody biomass, and an increase on the lower heating value
(LHV) of the Syngas was accounted to the properties of chicken manure (softer,
smaller size, and loosely packaged). Dayananda [73] experimentally studied the
co-gasification of chicken manure and rice husk. Results show that the mixing
ratio of 70% chicken manure and 30% rice husk produced the Syngas with the
highest energy value. Seçer [74] studied the efficiency of co-gasification of biomass
hydrolysates (products from hydrolysis), lignocellulosic biomass, and coal. Results
conclude that biomass hydrolysates produce more hydrogen than the other biomasses
because hydrolysis breaks down the structure of the biomass in smaller monomers,
leading to higher Syngas yields. Co-gasification yields higher CO2 due to the higher
oxygen on biomass. However, this CO2 is consumed on further reactions such as
Boudouard reaction, which is later converted into hydrogen in other consecutive
reactions (water–gas shift reaction).
The main objective of this paper is to explain the chemical kinetics of the thermo-
chemical conversion process for different biomass species. Consequently, it helps
in explaining and gives a better understanding of some of the complicated ther-
mochemical processes, such as co-Pyrolysis. Due to the mass production of these
biomass species (chicken manure, cow manure, and rice husk), the chemical kinetics
of different thermochemical processes were studied.

2 Experimental Setup and Procedures

Figure 1 shows the differential thermal gravimetry (DTG) apparatus, which is used
simultaneously to perform the thermal gravimetric analysis (TGA) and differential
thermal analysis (DTA). The device is consisting of three main parts. The first part is
the electric furnace, which supplies the required heat of reaction and can be controlled
to generate different heating rates and then maintain the reaction at a constant desired
temperature. The second part is the measurement system, which consists of rods that
are each fitted with thermocouples; then, both rods are connected to a sensitive
balance mechanism. One of the detector rods is used as a reference/control, and it
carries an empty sample cell, while the other rod is used to measure the changes of
mass and temperature of the active sample cell relative to the control cell. The third
part of the device is the autosampler, which is a robotic system capable of loading
and unloading the samples on the detectors automatically. The weight difference
between the two detectors, obtained by the delicate balance, indicates the sample
132 O. M. Selim et al.

Fig. 1 Shimadzu DTG-60AH; a main device components and b gas flow through the device

weight. The difference between the voltage readings of the two sensors is measured
for the differential thermal analysis. Any dry non-corrosive gas agent can be used
with the device. The measurable range of the mass is ± 500 mg, with a resolution
of 1 μg and ± 1% accuracy. The device can provide up to 1500 °C with ± 1 °C
uncertainty. The thermocouples used here are Pt-10% pt/Rh thermocouples. The
measurable range for the differential thermal analysis is 1000 μV with ± 1% μV
accuracy and noise level less than 1 μV.(Table 1)
To minimize the effect of gas turbulence on the readings, a regular calibration
was performed on the device all over the temperature range and operating gas flow
rate. Alumina pans were used for all case studies. Equation () is used to calculate
the extent of reacti6on, while its rate is calculated by using Eq. (7). The Arrhenius
rate expression was used as the first step of any kinetics model [75]. To calculate

Table 1 Proximate and

Chicken manure Cow manure
ultimate analysis of chicken
manure and cow manure [76, Proximate analysis (wt.% dry)
77] Volatile content 65.56 62
Ash content at 550 °C 21.65 23
Fixed carbon 12.8 15
Ultimate analysis (wt.% dry)
Carbon 35.59 35.2
Hydrogen 4.57 3.1
Nitrogen 4.98 2.2
Sulfur 1.45 0.7
Oxygen 35.52 33.3
HHV (in MJ/kg) 13.15 16
Review of Biomass Energy Resources with Livestock Manure 133

Table 2 Kinetic parameters

β ( °C /min) n Log(A/β) Ea (kJ/mole)
for nitrogen Pyrolysis with
different heating rates [80] 5 (250–360 °C) 5 14.9 84
5 (>360 °C) 5 10.9 63.1
10 (250–360 °C) 5 15 88.1
10 (>360 °C) 5 11 65.1
15 (250–360 °C) 5 15.3 89.6
15 (>360 °C) 5 11.2 69.0
20 (250–360 °C) 5 15.6 93.2
20 (>360 °C) 5 11.4 72.3
25 (250–360 °C) 5 16 95.8
25 (>360 °C) 5 11.7 76.4
30 (250–360 °C) 5 16.5 96.5
30 (>360 °C) 5 11.8 80.9
35 (250–360 °C) 5 16.8 98.3
35 (>360 °C) 5 11.9 84.5
40 (250–360 °C) 5 17 87.8
40 (>360 °C) 5 11.9 99

the activation energy, the following procedure was followed; Eqs. (9) and (10) were
combined to get Eq. (11); then by taking the natural log, Eqs. (9) through (12) are
used to calculate the activation energy where Eqs. (9) and (10) are combined to
get Eq. (11). Equation (12) is obtained by taking the natural log of Eq. (11), where
the activation energy E a and the exponent constant A can be obtained as shown in
Tables 2, 3, 4, 5, and 6.

w0 − W t νt
α= = (6)
w0 − w f νf

Table 3 Kinetic parameters

β (o C/min) n Log(A/β) Ea (kJ/mole)
for air gasification with
different heating rates [80] 5 3 11.2 70.3
10 3 10.1 66.7
15 3 10.1 64.5
20 3 10.9 70.8
25 3 10.3 67.6
30 3 10.9 70.3
35 3 10.3 67.9
40 3 10.2 69.0
134 O. M. Selim et al.

Table 4 Kinetic parameters

β ( °C /min) n Log(A/β) Ea (kJ/mole)
for carbon dioxide
gasification with different 5 (250–360 °C) 5 9.8 63.2
heating rates [80] 5 (360–630 °C) 5 4.3 39
5 (>630 °C) 5 72.5 575.5
10 (250–360 °C) 5 8.9 60.2
10 (360–630 °C) 5 5.0 38.7
10 (>630 °C) 5 68.1 523
15 (250–360 °C) 5 9.1 61.5
15 (360–630 °C) 5 3.5 32.9
15 (>630 °C) 5 68 556
20 (250–360 °C) 5 8.9 61.3
20 (360–630 °C) 5 3.8 34.8
20 (>630 °C) 5 67.3 551
25 (250–360 °C) 5 8.0 57.1
25 (360–630 °C) 5 3.0 305
25 (>630 °C) 5 66.5 549
30 (250–360 °C) 5 8.7 60.3
30 (360–630 °C) 5 3.0 33.3
30 (>630 °C) 5 68.1 561.5
35 (250–360 °C) 5 8.0 56.4
35 (360–630 °C) 5 2.9 32.6
35 (>630 °C) 5 63.6 544.7
40 (250–360 °C) 5 7.5 56.4
40 (360–630 °C) 5 3.7 37.6
40 (>630 °C) 5 62 519.7

Table 5 Biochar percentage

%CM % Biochar
for different heating rates and
mix ratios [70] 5 °C/min 10 °C/min 15 °C/min
100 34 39.3 36
90 39 35 44
80 37 37 38
70 30 33 35
60 26 36 37
50 33 32 32
Review of Biomass Energy Resources with Livestock Manure 135

dα (αt − αt−1 )
= (7)
dt t
 
−E a
k(T ) = A exp (8)
RT
 
dα k(T ) A −E a
= f (α) = exp f (α) (9)
dT β β RT
dα
= k(T )(1 − α)n (10)
dT
 
dα k(T ) A −E a
= f (α) = exp (1 − α)n (11)
dT β β RT
     
dα
A Ea
ln dT
= ln − (12)
(1 − α)n β RT

3 Operation Conditions

Chicken manure used in this study was a dried chicken manure fertilizer of irregular
form, with sizes ranging from 1 to 3 mm. The apparent density of particles measured
from fluid immersion was 1.5 g.ml−1 . The ultimate and proximate analysis, along
with the higher heating value (HHV) [76] of the sample, is given in Table 1.

4 Results and Discussion

Three thermochemical processes were investigated in this paper: Pyrolysis, gasifica-

tion, and co-Pyrolysis.

4.1 Pyrolysis of Chicken Manure

The thermal gravimetric and differential thermal analyses were conducted for the
chicken manure using nitrogen as gas agent with eight different heating rates from
room temperature up to 1000 °C. The three main components in the chicken manure
are Hemicelluloses, Cellulose, and Lignin. The thermal degradation of Hemicellu-
loses is known to peak at 240 °C and Cellulose at 380 °C, while Lignin has more
of a steady degradation with a small peak at high temperature 600–800 °C [78].
136 O. M. Selim et al.

1
0.9
0.8

Extent of reaction
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
150 250 350 450 550 650 750 850 950
Temperature (oC)

5 C/mi n 10 C/mi n 15 C/mi n 20 C/mi n

25 C/mi n 30 C/mi n 35 C/mi n 40 C/mi n

Fig. 2 Extent of the reaction of chicken manure at different heating rates [80]

Figure 2 shows the extent of reaction of chicken manure at different heating rates
when nitrogen is used. It can be observed that all heating rates had the same trend
with the increase of the temperature, and the extent of reaction increases. In addition,
when the heating rate increased, the progress of the extent of reaction seems to be
delayed to a higher temperature. Furthermore, the figure shows that the faster the
heating rate, the more delayed is the progress of each reaction with respect to temper-
ature. The main reason is that increasing the heating rate does not allow enough time
for each reaction to be completed before increasing the temperature. Another plau-
sible reason can be the response of the measuring device. This delay can be observed
as well when using different gas agents. This kind of behavior was also observed for
wood [79]. The Pyrolysis reaction was best represented by dividing the decomposi-
tion into three main stages. The first stage begins with the start of thermal cracking
up to 250 °C, the second stage with a faster reaction between 250 and 350 °C, and
finally, the third stage with a steady reaction from 350 °C to the end of the reaction.
These three distinct stages were used to find the kinetics of the reaction. Starting from
350 °C and up to 700 °C, the thermal degradation is a complicated overlap between
the three main components (Hemicellulose, Cellulose, and Lignin) decomposition,
and the effect of heating rate on the extent of reaction is more significant compared to
the other temperature ranges. At the end of each reaction, a residual mass that failed
to react because of either insufficient activation energy or the absence of suitable
reacting media was left over.
The rate of change of the extent of reaction of chicken manure with the temperature
at different heating rates is shown in Fig. 3. Similar to Fig. 2, it can be observed that
the trend is the same for all tested heating rates. At 150 °C for the low heating rates,
all the moisture content has evaporated from the sample. For the highest two heating
rates, 35 and 40 °C, two peaks take place near 150 °C due to the evaporation of any
moisture in the sample. The main reason is that the heating rate is faster than the
evaporation rate of the moisture content.
Review of Biomass Energy Resources with Livestock Manure 137

0.0045
0.004
0.0035

da/dt, 1/sec
0.003
0.0025
0.002
0.0015
0.001
0.0005
0
150 250 350 450 550 650 750 850 950
Temperature (oC)

5 C/mi n 10 C/mi n 15 C/mi n 20 C/mi n

25 C/mi n 30 C/mi n 35 C/mi n 40 C/mi n

Fig. 3 Rate of Change of the extent of reaction with the temperature at different heating rates [80]

For all heating rates, the first two peaks are within the temperature range of 250–
350 °C, and then the third peak started to occur at 670 °C for the lowest heating rates
(5 °C /min).
It can be observed that the peak value is increased with the increase in the heating
rate, as well as the temperature at which the peak occurs. Starting with 5 °C/min, the
first peak took place at 255 °C, and after increasing the heating rate, the temperature
at which peak occurs is increased as well to reach 298 °C for the heating rate of
40 °C/min. The same scenario for the second peak is 320, 325, 335, 345, and 343 °C
for the heating rates 5, 15, 20, 30, and 40 °C /min, respectively. The three peaks are
indications for the three main components: Hemicellulose, Cellulose, and Lignin,
respectively. The values obtained in this study are matching the values obtained by
Yang et al. for determining the degradation temperature range of these main three
components [81].
Kinetics of Reaction: Table 2 shows the activation energy and the Arrhenius constant
at different heating rates. For the temperature range from 250 to 360 °C, it can be
observed that the average activation energy is 91 kJ/mole and the average value for
log A = 14.5 s−1 . For the range greater than 360 °C, the average activation energy
is 75.6 kJ/mole while log A = 10 s−1 .
Differential Thermal Analysis (DTA): Figure 4 shows the DTA variation with the
temperature at different heating rates. For all the heating rates, it can be observed
that the trend is the same for all the heating rates. The reaction is relatively steady
except for a large peak downward for the faster heating rate cases by activating the
endothermic reaction due to the quick breakdown of Cellulose and Hemicelluloses
below 350 °C. With the increase in the heating rate, the magnitude of the peaks
increases. The more massive peak is a result of a fast increase in the temperature of
the control (empty) pan, while the active pan is enduring high rates of endothermic
reactions hindering the sample temperature increase. Consequently, for 30 °C/min
and higher, the trend looks different from the other cases.
138 O. M. Selim et al.

Fig. 4 Change of the DTA 150

with the temperature at
different heating rates [80] 100

DTA, μV
50

0
150 350 550 750 950
-50

-100
Temperature (oC)
5 C/mi n 10 C/mi n 15 C/mi n 20 C/mi n
25 C/mi n 30 C/mi n 35 C/mi n 40 C/mi n

4.2 Pyrolysis of Cow Manure

Figure 5 shows the extent of the reaction for the cow manure at eight different
heating rates. It can be observed that all the heating rates have the same trend and
increasing the heating rates responsible for delaying the extent of the reaction to
a higher temperature. The Pyrolysis reaction was best presented by dividing the
decomposition into four main stages. First stage is up to 255 °C, which represents
the thermal cracking. A second faster reaction is from 255 to 315 °C. Then, from
315 to 745 °C represents the thermal degradation of the three main components:
Hemicellulose, Cellulose, and Lignin.
The degradation temperature of the three main components can be obtained from
Fig. 6. The first peak represents the Hemicellulose degradation. For all the heating
rates, the first peak occurs at 321 °C. At 475 °C, the thermal degradation of the
Cellulose takes place for the lower heating rates and 500 °C for the 40 °C/min. A
plausible reason for the small fluctuation between 350 and 500 °C is the degradation
of impurities such as bedding. Relatively, more minor bumps appear at 760 °C for 35

1
0.9
0.8
Extent of reaction

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
150 250 350 450 550 650 750 850 950
Temperature oC
5 C/min 15 C/min 25 C/min 35 C/min
10 C/min 20 C/min 30 C/min 40 C/min

Fig. 5 Extent of reaction variation with temperature for cow manure at different heating rates [80]
Review of Biomass Energy Resources with Livestock Manure 139

0.01

0.008

da/dt, 1/sec
0.006

0.004

0.002

0
150 250 350 450 550 650 750 850 950
o
Temperature ( C)

5 C/min 15 C/min 25 C/min 35 C/min

10 C/min 20 C/min 30 C/min 40 C/min

Fig. 6 Rate of change of the extent of reaction with the temperature for cow manure at different
heating rates [80]

and 40 °C/min, which is due to the degradation of the low residuals of Lignin after
combustion.
Figure 7 shows the DTA variation with the temperature at different heating rates.
All heating rates are following the same trend. All the heating rates are exothermic,
which means that the reaction is self-sustainable. For the higher heating rates, the
sharp decline after the first peak implies that the reaction turns into endothermic.
The main reason is that the faster heating rate does not allow enough time to thermal
equilibrium besides the heat transfer limitation. Based on this conclusion and for
figure clarity, the 5 °C/min is used to compare between the chicken manure and cow
manure.

200

150

100
DTA, μV

50

0
150 250 350 450 550 650 750 850 950
- 50
Temperature (oC)

5 C/min 10 C/min 15 C/min 20 C/min

25 C/min 30 C/min 35 C/min 40 C/min

Fig. 7 Change of the DTA with the temperature for cow manure at different heating rates [80]
140 O. M. Selim et al.

4.3 Pyrolysis of Rice Husk

Rice husk-chemical content consists of 50% of Cellulose, 25–30% Lignin, and 15–
20% silica [25]. No peak below 270 °C is expected because the degradation peak of
Hemicellulose is at temperatures between 240 and 270 °C [69]. The mass degrada-
tion of rice husk can be divided into three stages. The first stage is the same as that
for chicken manure—the water is evaporated from 100 to 225 °C. The second stage
starts around 250 °C and peaks at a temperature between 323 and 340 °C, thus corre-
sponding to the degradation of the Cellulose. Approximately, 50% of the biomass is
degraded in this stage. The peak of the Pyrolysis of rice husk is less distributed than
the Pyrolysis of chicken manure due to the absence of Hemicellulose and less Lignin
content. The third stage occurs within the 400–650 °C range. In this stage, the Lignin
degrades. The percentage of Lignin in the studied rice husk prior to degradation is
small because there is no peak at this stage. Also, the Lignin is more thermally stable
than the Cellulose. Thus, degradation occurs at a decreased rate. The total conversion
for this biomass is 60% wt, signifying that 40% of the biomass remained as biochar
(see Fig. 8).
When the heating rate is increased, the peak of the degradation rate and the
temperature (at which the peak occurs) increases (Fig. 9). This result is the same
as obtained for the chicken manure case in Fig. 3. It can also be observed that the
DTG curve has only one dominant peak, which takes place at 330, 340, and 350 °C,
for the 5, 10, and 15 °C/min, respectively. A plausible reason for that result is that
the rice husk has more Cellulose content than Hemicellulose. This explanation is
not applicable in the case of chicken manure, where Hemicellulose and Cellulose
percentages are closer to each other. When the heating rate was increased, the TG
curves shifted slightly to the higher temperature side in the second stage but merged in
the third stage. Other than that, no significant changes in degradation were observed.

Fig. 8 TG curves for rice husk at different heating rates [70]

Review of Biomass Energy Resources with Livestock Manure 141

Fig. 9 DTG curves for rice husk at different heating rates [70]

4.4 Gasification of Chicken Manure

Air Gasification
In this case, the air was used as a gas agent. As a result, the chemical reactions were
not only limited to the Pyrolysis process but also gasification to take place because of
the incomplete combustion of the gases and fixed carbon in the presence of O2 . With
the same as nitrogen Pyrolysis, the same parameters were studied at the same heating
rates. The same remarkable results are obtained with case A, as seen in Fig. 10, where
all different heating rates had the same trend. As the temperature increases, the extent
of reaction increases. When the heating rate increased, the progress of the extent of
reaction seems to be delayed to a higher temperature. For the gasification part, the
reaction can be divided into five main stages. The first stage takes place, starting from
the thermal cracking up to 250 °C. A faster second reaction takes place between 250
and 350 °C and then from 350 to 450 °C another fast stage with a mild slope compared

1
0.9
0.8
Extent of reaction

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
150 350 550 750 950
Temperature (oC)
5 C/min 10 C/min 15 C/min 20 C/min
25 C/min 30 C/min 35 C/min 40 C/min

Fig. 10 Extent of the reaction of chicken manure at different heating rates [80]
142 O. M. Selim et al.

0.006
0.005

dα/dt, 1/sec
0.004
0.003
0.002
0.001
0
150 250 350 450 550 650 750 850 950
Temperature (oC)
5 C/min 10 C/min 15 C/min 20 C/min
25 C/min 30 C/min 35 C/min 40 C/min

Fig. 11 Rate of change of the extent of reaction with the temperature at different heating rates [80]

to the previous region. The fluctuation in temperature between 450 and 600 °C is
because of the self-ignition. Right after 600 °C, the extent of reaction is more than
95%, and the reaction tends to be very slow.
Again, three significant peaks can be shown in Fig. 11: 255, 340, and 500 °C. As
expected, the magnitude of the peak is increased as the heating rate increased as well
as the temperature at which the peak occurs. As shown in the figure, for 5 °C/min,
the first peak occurs at 255 °C. For 10 °C/min, one can see that the first peak is
shifted to the right (higher temperature) as well, and the peak value is higher. For
40 °C/min, the peak value is 0.005 s−1 , which is the highest peak, and it happened at
293 °C, and so on. The increase in the peak value is resulting from the faster change
in temperature. The characteristics of hemicellulose and Cellulose can be seen on the
first two peaks. Another indication of self-ignition can be seen from the temperature
fluctuation between 350 and 450 °C, which was already observed in case A. A tiny
bump appears at 680 °C for 35 and 40 °C/min, which is due to the degradation of
the low residuals of Lignin after combustion.
Kinetics of reaction (TGA). The values for the Arrhenius equation constant and
the activation energy are shown in Table 3. Even though the reaction appeared
more complicated than the N2 Pyrolysis case, single kinetic reaction constants were
calculated for the whole conversion reaction where the average activation energy is
68.4 kJ/mole and the average log A = 9.3 s−1 .
Differential thermal analysis (DTA). In Fig. 12, the change of the DTA with the
temperature at different heating rates is shown; the same peaks are formed at the
same temperatures obtained from Figs. 10 and 11, and these peaks indicate the
thermal degradation of the Hemicellulose, Cellulose, and self-ignition. The fact that
these peaks are positive (upward) is due to the oxidation of some of the evolving
gases rendering the reaction as exothermic. After 700 °C, no significant reaction was
recorded, and the weight of the sample was stable.
Review of Biomass Energy Resources with Livestock Manure 143

500

400

DTA, μV 300

200

100

0
150 250 350 450 550 650 750 850 950
-100 o
Temperature ( C)
5 C/min 10 C/min 15 C/min 20 C/min
25 C/min 30 C/min 35 C/min 40 C/min

Fig. 12 Change of the DTA with the temperature at different heating rates [80]

Carbon Dioxide Gasification

The extent of reaction. Another gasification case was studied when CO2 is used.
In this case, as shown in Fig.13, the four stages of the gasification reaction are as
follows: start of thermal cracking up to 250 °C, the faster reaction between 250 and
360 °C, steady reaction from 360 to 700 °C, and finally, a quick reaction at 700 °C
to the end of the reaction. These four distinct stages were used to find the kinetics of
the reaction.
In Fig.14, three significant peaks can be observed close to the following tempera-
ture: 250, 360, and 700–800 °C. The first two peaks are characteristics for Hemicel-
lulose and Cellulose, while the third peak (700–800 °C) is an overlap between the
peak for Lignin and the peak for the Boudouard reaction between fixed carbon and
CO2 , respectively. The peak value increases with the heating rate due to the faster
increase of the temperature with time.

1
Extent of reaction

0.8

0.6

0.4

0.2

0
150 350 550 750 950
Temperature (oC)
5 C/min 10 C/min 15 C/min 20 C/min
25 C/min 30 C/min 35 C/min 40 C/min

Fig. 13 Extent of the reaction of chicken manure at different heating rates [80]
144 O. M. Selim et al.

0.06

0.05

dm/dt, mg/sec 0.04

0.03

0.02

0.01

0
150 250 350 450 550 650 750 850 950
Temperature (oC)
5 C/min 10 C/min 15 C/min 20 C/min
25 C/min 30 C/min 35 C/min 40 C/min

Fig. 14 Rate of change of the extent of reaction with the temperature at different heating rates [80]

Kinetics of reaction (TGA). Table 4 shows the values of the Arrhenius equation
constant and the activation energy. For the range 250–360 °C, the average value for
Ea = 59.6 kJ/mole and the average value for log A = 9.8 s−1 ; for the range 360–
630 °C, the average value for Ea = 34.9 kJ/mole and the average value for log A =
4.8 s−1 ; and for > 630 °C, the average value for Ea = 547.6 kJ/mole and the average
value for log A = 68 s−1 .
As shown in Fig.15, the reaction is relatively steady from 350 to 700 °C with an
exothermic reaction where the DTA values are positive. The endothermic reactions
took place below 350 °C due to the fast breakdown of Cellulose and Hemicelluloses,
and another endothermic reaction of CO2 with the fixed carbon in the chicken manure
at a temperature greater than 700 °C. After 700 °C, when the heating rate increases,
the curve started to drop sharply, which means that the reaction is endothermic.
Also, the magnitude of the peak increases. Comparing the different heating rates,

60
40
20
DTA, μV

0
150 250 350 450 550 650 750 850 950
-20
-40
-60
-80
Temperature (oC)

5 C/min 10 C/min 15 C/min 20 C/min

25 C/min 30 C/min 35 C/min 40 C/min

Fig. 15 Change of the DTA with the temperature at different heating rates [80]
Review of Biomass Energy Resources with Livestock Manure 145

100%

80%

Mass Residual 60%

N2
40%
AIR
20% CO2

0%
150 250 350 450 550 650 750 850 950
o
Temperature ( C)

Fig. 16 Mass residual change versus temperature for the three tested gas agents [80]

after 700 °C, one can observe that, for 5 °C/min, the reaction is exothermic, where
the curve goes up, while for the 40 °C/min, it sharply goes down. The faster rate of
change in temperature does not allow enough time for thermal equilibrium between
the reference and sample, and thus the peak value appears larger.
In terms of mass residual, Fig.16 shows the average mass residual for the three
gas agents. The average residual mass in the case of N2 Pyrolysis is 25% of the total
mass, and it is independent of the heating rate, while for the air gasification and CO2
gasification, the average residual mass is 23% and 19%, respectively.

4.5 Co-pyrolysis

Co-pyrolysis of chicken and cow manure

The two livestock manures have been mixed at four different ratios: 20, 40, 60, and
80% cow manure of the total mixture weight. The same three parameters are used
to give a better idea of the chemical kinetics for the mixture of thermal degradation.
Starting with the extent of reaction, Fig. 17 shows that the 40 and 60% cow manures
give a slower weight consumption compared to 20 and 80% cow manures. Figure 18
shows that the extent of reaction rate is increasing with increase in the cow manure
percentage, which was expected as it approaches the cow manure characteristic to
stand alone.
Regarding the DTA, from Fig. 19 we see that all the tested samples are exothermic
except for the 60% cow manure, whose reaction turns into an endothermic reaction
after 620 °C, which means further energy is needed to complete the Pyrolysis process.
There are several reasons for the inconsistency in the trend with 60% cow manure
in comparison with other cases. The first one could be the bedding of cow manure
(some wood straws). Another reason, as you can see, increasing the cow manure
percentage makes the reaction to be endothermic, and this is mainly because of the
protein and amino acids that need a large amount of energy to be lost. In general, the
thermal degradation of the cow manure is hard to predict because of the variation of
146 O. M. Selim et al.

0.8
Extent of Reaction
0.6

0.4

0.2

0
150 250 350 450 550 650 750 850 950
o
Temperature ( C)
20% COW MANURE 40% COW MANURE
60% COW MANURE 80% COW MANURE

Fig. 17 Extent of reaction variation with the temperature at different cow manure percentages [25]

0.002

0.0015
da/dt (1/sec)

0.001

0.0005

0
150 250 350 450 550 650 750 850 950
Temperature (oC)
20% COW 40% COW 60% COW 80% COW

Fig. 18 Rate of change of the extent of reaction with the temperature at different cow manure
percentage [25]

its components. All of these reasons could participate in this discrepancy. Overall, it
can be observed that the 40% cow manure gives a stable exothermic reaction to the
end of the process.
Co-pyrolysis of Chicken Manure and Rice Husk
For the biomass blends, the TG curves for the mixtures are shown in Fig. 20. For
clarity, only the heating rate with the most noticeable effect (5 °C/min) is shown.
However, it is worth mentioning that all the heating rates had the same effect on
biomass degradation. The first stage occurs at temperatures between 100 and 220 °C,
which corresponds to the humidity extraction. The humidity content of the blends was
10%wt. The second stage is given at temperatures between 250 and 360 °C. In this
Review of Biomass Energy Resources with Livestock Manure 147

100

80

60
DTA (μV)

40

20

0
150 250 350 450 550 650 750 850 950
-20
Temperature (oC)
20% COW 40% COW 60% COW 80% COW

Fig. 19 Change of the DTA with the temperature at different cow manure percentages [25]

Fig. 20 TG curves for different mixing ratios at 5 °C/min [70]; you can use the color version

stage, the Hemicellulose and Cellulose from both biomasses decompose. This stage
has the most massive mass degradation with around 40%wt on average. Therefore,
the kinetic parameters are estimated at only this single stage. The third stage occurs at
temperatures higher than 360 °C, during which the crystalline Cellulose and Lignin
are slowly degraded. This stage decomposes about 20% of the total volatiles.
Figure 21 shows that the degradation curve follows a straight line during the
second stage. In order to evaluate the decomposition, the mass degradation and
temperature between 250 and 340 °C are approximated to a straight line (Fig. 21).
The degradation increased when more rice was added. At 250 °C, the lost mass is
almost identical within all blends; an acceleration occurs in the degradation at 340 °C,
causing increased degradation from 0.7 to 0.51 (marked in Fig. 21). However, the
lines are almost parallel—the decomposition of the different blends takes place at
the same temperatures. This result is because both the chicken manure and rice husk
have similar volatile matter [25, 70].
148 O. M. Selim et al.

Fig. 21 TG curves between 250 and 340 °C at 5 °C/min [70]

The first and third stages of degradation remained relatively unchanged. For
the second stage, the degradation rate increases when the percentage of rice husk
increases. The chicken manure’s degradation rate is increased because there is addi-
tional heating provided by the exothermic reaction that comes from rice husk’s char
gas formation. However, the degradation peak is slightly delayed, causing a need for
higher temperatures to reach the degradation peak (Fig. 22).
Table 5 shows the remaining biomass for different heating rates. The increase in
the heating rate reduced the conversion percentage for all the blends except for a
few cases where the remaining mass was higher than the mass of singular biomass.
The effect of the synergetic effect is also seen in the energy of activation, which is
discussed in the next sections.
Differential Thermal Analysis
The differential thermal analysis curves for different mixing ratios are shown.
The curves are consistent with the DTG curves, the maximum degradation rate,
and the peaks in the thermal energy, which all occur at the same temperature. Three
peaks can be seen. The first peak represents water degradation at a temperature of

Fig. 22 DTG curves for different blends [70]

Review of Biomass Energy Resources with Livestock Manure 149

120 °C. This stage corresponds to water extraction degradation of Hemicellulose

and Cellulose and additionally peaks at a temperature of about 350 °C. The energy
extracted remains stable up to 380 °C. Combining the two samples increases the
amount of energy released during the reaction than it would be if each biomass sample
stood alone because the two samples each provide the same exothermic reaction over
the same temperature range. At the early stages of Pyrolysis, chicken manure has
the least energy extraction mainly because all the energy supplied by the furnace is
being used to remove moisture content and breaking up Hemicellulose. However, for
0% chicken manure (100% rice husk), it can be observed that the existing energy is
higher due to lesser moisture content and the absence of Hemicellulose. By adding
more rice husk (10 and 20%), the DTA curve is then located between 0 and 100%
chicken manure, which emphasizes this conclusion. When more rice husk is added,
the exothermic reaction is dominant (Fig. 23).
Starting from 260 °C (which is corresponding to thermal degradation of Hemi-
cellulose in chicken manure) to 480 °C (which is the start of Lignin degradation),
it is observed that the amount of energy released from the reaction for all mixture
blends is higher than each sample alone. The main reason behind this result is the
overlap between two exothermic reactions taking place at the same time. Among all
cases, the 60% chicken manure mixture gives the highest exothermic reaction, which
makes the reaction self-sustaining to the end of Pyrolysis.
Pyrolysis Kinetics
The kinetic parameters for the reaction between 250 °C and 360 °C were calculated
for each blend at different heating rates. The average values for each combination are
shown in Table 6. The average energy of activation of chicken manure and rice husk
is 108 and 95.5 kJ mol−1 , respectively. The pre-exponential factor is 2.05 ×1010 min
−1
and 1.17 ×108 min−1 for chicken manure and rice husk, respectively. The order
of reaction of rice husk was 2.0, while the order of reaction for chicken manure was

Fig. 23 DTA curves for different blends [70]; you can use the color version
150 O. M. Selim et al.

Table 6 Averaged kinetic parameters [70]

% cm n ln(A/b) Ea (Kj/mol) A (min-1)
100 5 17.7 108.1 2.05E + 10
90 4.7 21.1 112.7 2.44E + 10
80 4.4 21.3 109.7 5.40E + 09
70 4.1 21.3 106.8 3.78E + 09
60 3.8 13.3 95.0 5.87E + 08
50 3.5 17.3 101.1 6.97E + 08
0 2 16.8 95.5 1.17E + 08

5.0. For the blends, the order of the reaction was found to vary linearly with the blend
ratio.
The addition of rice husk increased the energy of activation compared with indi-
vidual types of biomass for the first two mixing ratios. This increase could be due to
the difference in particle sizes of rice husk compared to chicken manure. That size
difference negatively affects the Pyrolysis reactions due to the temperature gradi-
ents, also leading to lower conversion. However, after adding more than 20% RH,
the energy of activation and pre-exponential factor starts decreasing and reaches its
minimum energy of activation—95.0 kJ mol−1 for the blend (60%CM + 40%RH).
This minimum value signifies a decrease of 12.1% of the energy of activation when
compared with chicken manure’s energy of activation. This decrease means that the
synergetic effects overcome the heat and mass transfer effects of the mixture (Table
6). The comparable behavior was observed by Dayananda [74] for the gasification
of chicken manure and rice husk under a fluidized bed.
The increase in the heating rate, in turn, increases the energy of activation, which
is in agreement with the literature for a different biomass Pyrolysis [35, 36]. Several
factors affect the kinetics of the biomass. First, the particle size of the biomass plays
a substantial role because the rapid change in temperature prevents the particles from
having an even temperature distribution. Also, because less time is given before the
temperature increases, the reactions in the Pyrolysis cannot be complete, increasing
the energy of activation (Fig. 24). Also, the pyrolytic reactions cannot be perfect
because less time is given before the temperature decreases, thus increasing the
energy of activation.

5 Conclusions

The thermal gravimetric and differential thermal analyses were conducted for chicken
manure using three different gasifying media (nitrogen, air, and carbon dioxide) with
eight different heating rates (5, 10, 15, 20, 25, 30, 35, and 40 °C/min.) from room
temperature to 1000 °C. The main difference between the nitrogen Pyrolysis and
carbon dioxide gasification occurred after the temperature regime (>700 °C). At this
Review of Biomass Energy Resources with Livestock Manure 151

Fig. 24 Energy of activation for different cases [70]; you can use the color version

regime, the reaction turned into the endothermic reaction for the CO2 as it is governed
by the Boudouard reaction, where the reaction turned into the endothermic reaction
for the CO2. Air gasification is exothermic, and ignition was observed between 450
and 600 °C. Therefore, with air gasification, the reaction has the potentials to be
self-sustainable with no external heating. The chicken manure thermal degradation
implied the presence of the three components: Hemicellulose, Cellulose, and Lignin.
The highest reaction rates were observed at temperatures corresponding to known
peak characteristics of the three components. Gasification with air had an extra peak
for ignition at 600 °C. The reaction kinetic parameters for the conversion reaction
were calculated for the different media. The nitrogen Pyrolysis was divided into two
regions at 360 °C with the order of reaction of five for both regions. Kinetic parameters
for air gasification were calculated using a third-order single region reaction. CO2
had the most complicated mechanism of the three cases and was divided into three
regions at 360 and 630 °C. The kinetic parameters varied with the heating rate. It
is recommended to utilize values generated by the lowest heating rate because the
slow heating rate allows a quasi-equilibrium state, thus decreasing the effects of
measurement errors due to delay in response or any transient condition error. Four
different case studies were examined where the cow manure was added to chicken
manure. The 40% cow manure of the total sample gives a positive result where
the reaction is steady exothermic reaction all over the Pyrolysis process. Further
research is required to obtain the amount of the produced Syngas and cumulative
energy yield. Rice husk is compatible with chicken manure, causing an observable
positive synergetic effect and increasing the conversion of the biomass. This effect is
observed in the energy of activation. The energy of activation decreases when the rice
husk ratio increases, thus decreasing the energy needed to begin Pyrolysis. The energy
of activation decreases due to the synergetic effect of the rice husk on the blends.
Moreover, adding rice husk increases the degradation rate, thus reducing the length
of time that the biomass has to be in the reactor, resulting in an overall decrease in the
energy consumed during Pyrolysis. The blend with the best performance was 60%CM
+ 40%RH. This blend decreased the energy of activation by 12.1% with respect to
chicken manure. Additionally, a maximum increase in the conversion of 28% was
152 O. M. Selim et al.

achieved when the heating rate was 5 °C/min. These optimistic results need to apply
more gas analysis to give a detailed explanation of the Syngas reactions taking place.
A gas analysis of these results needs to be studied to give a detailed explanation of
the Syngas reactions taking place. Co-Pyrolysis of chicken manure and rice husk
demonstrated improved Pyrolysis of chicken manure. However, more research must
be conducted in evaluating other parameters that can affect the co-Pyrolysis between
these two types of biomass, such as particle size.

References

1. Outlook AE (2020). Energy information administration. Department of Energy, Washington,

DC 20585, pp 1–161
2. Lv Y, Bi J, Yan J (2018) State-of-the-art in low carbon community. Int J Energy Clean Environ
19(3–4)
3. Ryu HW, Kim DH, Jae J, Lam SS, Park ED, Park Y-K (2020) Recent advances in catalytic
co-Pyrolysis of biomass and plastic waste for the production of petroleum-like hydrocarbons.
Bioresource Technol 123473
4. Ong HC, Chen W-H, Farooq A, Gan YY, Lee KT, Ashokkumar V (2019) Catalytic thermo-
chemical conversion of biomass for biofuel production: a comprehensive review. Renew Sustain
Energy Rev 113:109266
5. Pandey A (ed) (2011) Biofuels: alternative feedstocks and conversion processes. Academic
Press
6. Chen H, Wang L (2016) Technologies for biochemical conversion of biomass. Academic Press
7. Pandey A, Bhaskar T, Stöcker M, Sukumaran R (eds) Recent advances in thermochemical
conversion of biomass. Elsevier
8. Rosendahl L (ed) In: Biomass combustion science, technology and engineering. Elsevier
9. Verma M, Godbout S, Brar SK, Solomatnikova O, Lemay SP, Larouche JP (2012) Biofuels
production from biomass by thermochemical conversion technologies. Int J Chem Eng
10. Molino A, Chianese S, Musmarra D (2016) Biomass gasification technology: the state-of-the-
art overview. J Energy Chem 25(1):10–25
11. Neves D, Thunman H, Matos A, Tarelho L, Gómez-Barea A (2011) Characterization and
prediction of biomass Pyrolysis products. Prog Energy Combust Sci 37(5):611–630
12. Guedes RE, Luna AS, Torres AR (2018) Operating parameters for bio-oil production in biomass
Pyrolysis: a review. J. Anal Appl Pyrolysis 129:134–149
13. Pandey A (2008) Handbook of plant-based biofuels. CRC Press
14. Heidari A, Eshagh K, Habibollah Y, Lu HR (2019) Evaluation of fast and slow Pyrolysis
methods for bio-oil and activated carbon production from eucalyptus wastes using a life cycle
assessment approach. J Cleaner Prod 241:118394
15. Hussein MH (2016) Experimental investigation of chicken manure pyrolysis and gasification.
Ph.D. Thesis, University of Wisconsin-Milwaukee
16. Vieira FR, Romero Luna CM, Arce GLAF, Ávila I (2020) Optimization of slow Pyrolysis
process parameters using a fixed bed reactor for biochar yield from rice husk. Biomass
Bioenergy 132:105412
17. Yuan T, He W, Yin G, Shiai Xu (2020) Comparison of bio-chars formation derived from fast
and slow Pyrolysis of walnut shell. Fuel 261:116450
18. Al Arni S (2018) Comparison of slow and fast Pyrolysis for converting biomass into fuel.
Renew Energy 124:197–201
19. Goyal HB, Seal D, Saxena RC (2008) Bio-fuels from thermochemical conversion of renewable
resources: a review. Renew Sustain Energy Rev 12(2:504–517
Review of Biomass Energy Resources with Livestock Manure 153

20. Tanoue K-I, Hinauchi T, Oo T, Nishimura T, Taniguchi M, Sasauchi K-I (2007) Modeling
of heterogeneous chemical reactions caused in Pyrolysis of biomass particles. Adv Powder
Technol 18(6):825–840
21. Dry ME (1996) Practical and theoretical aspects of the catalytic Fischer-Tropsch process. Appl
Catal A 138(2):319–344
22. Reed TB (2002) Encyclopedia of biomass thermal conversion: the principles and technology
of Pyrolysis, gasification and combustion. Biomass Energy Foundation Press
23. Gebreegziabher T, Oyedun AO, Hui CW (2013) Optimum biomass drying for combustion–a
modeling approach. Energy 53:67–73
24. Han J, Choi Y, Kim J (2020) Development of the process model and optimal drying conditions
of biomass power plants. ACS Omega 5(6):2811–2818
25. Gaur S, Reed TB (1995) An atlas of thermal data for biomass and other fuels. No. NREL/TP-
433–7965. National Renewable Energy Lab., Golden, CO (United States)
26. Balat M (2008) Mechanisms of thermochemical biomass conversion processes. Part 1: Reac-
tions of Pyrolysis Energy Sources Part A: Recovery Utilization Environ Effects 30(7):620–635
27. Yang H, Yan R, Chen H, Lee DH, Zheng C (2007) Characteristics of hemicellulose, cellulose
and lignin Pyrolysis. Fuel 86(12–13):1781–1788
28. Uddin MN, Daud WMAW, Abbas HF (2014) Effects of pyrolysis parameters on hydrogen
formations from biomass: a review. RSC Adv 4(21):10467. https://doi.org/10.1039/c3ra43
972k
29. Kumar A, Jones DD, Hanna MA (2009) Thermochemical biomass gasification: a review of the
current status of the technology. Energies 2:556–581. https://doi.org/10.3390/en20300556
30. Bahng M-K, Mukarakate C, Robichaud DJ, Nimlos MR (2009) Current technologies for anal-
ysis of biomass thermochemical processing: a review. Anal Chim Acta 651:117–138. https://
doi.org/10.1016/j.aca.2009.08.016
31. Kalinci Y, Hepbasli A, Dincer I (2009) Biomass-based hydrogen production: a review
and analysis. Int J Hydrogen Energy 34:8799–8817. https://doi.org/10.1016/j.ijhydene.2009.
08.078
32. Woolcock PJ, Brown RC (2013) A review of cleaning technologies for biomass-derived Syngas.
Biomass Bioenergy 52:54–84. https://doi.org/10.1016/j.biombioe.2013.02.036
33. Sutton D, Kelleher B, Ross JRH (2001) Review of literature on catalysts for biomass
gasification. Fuel Process Technol 73:155–173. https://doi.org/10.1016/S0378-3820(01)002
08-9
34. Bulushev DA, Ross JRH (2011) Catalysis for conversion of biomass to fuels via Pyrolysis and
gasification: a review. Catal Today 171:1–13. https://doi.org/10.1016/j.cattod.2011.02.005
35. Tanksale A, Beltramini JN, Lu GM (2010) A review of catalytic hydrogen production processes
from biomass. Renew Sustain Energy Rev 14:166–182. https://doi.org/10.1016/j.rser.2009.
08.010
36. Saxena RC, Seal D, Kumar S, Goyal HB (2008) Thermo-chemical routes for hydrogen rich gas
from biomass: a review. Renew Sustain Energy Rev 12:1909–1927. https://doi.org/10.1016/j.
rser.2007.03.005
37. Puig-Arnavat M, Bruno JC, Coronas A (2010) Review and analysis of biomass gasification
models. Renew Sustain Energy Rev 14:2841–2851. https://doi.org/10.1016/j.rser.2010.07.030
38. Chhiti Y, Kemiha M (2013) Thermal conversion of biomass pyrolysis and gasification: a review.
Int J Eng Silences 2:75–85
39. Pereira EG, da Silva JN, de Oliveira JL, Machado CS (2012) Sustainable energy: a review of
gasification technologies. Renew Sustain Energy Rev 16:4753–4762. https://doi.org/10.1016/
j.rser.2012.04.023
40. Parthasarathy P, Narayanan KS (2014) Hydrogen production from steam gasification of
biomass: influence of process parameters on hydrogen yield—a review. Renew Energy
66:570–579. https://doi.org/10.1016/j.renene.2013.12.025
41. Mohan D, Pittman CU, Steele PH (2006) Pyrolysis of wood/biomass for bio-oil: a critical
review. Energy Fuel 20:848–889. https://doi.org/10.1021/ef0502397
154 O. M. Selim et al.

42. McKendry P (2002) Energy production from biomass (part 1): overview of biomass. Bioresour
Technol 83:37–46. https://doi.org/10.1016/S0960-8524(01)001183
43. Wang L, Weller CL, Jones DD, Hanna MA (2008) Contemporary issues in thermal gasification
of biomass and its application to electricity and fuel production. Biomass Bioenergy 32:573–
581. https://doi.org/10.1016/j.biombioe.2007.12.007
44. Kamin´ ska-Pietrzak N, Smolin´ ski A (2013) Selected environmental aspects of gasification
and co-gasification of various types of waste. J Sustain Min 12:6–13.https://doi.org/10.7424/
jsm130402
45. Higman C, van der Burgt M, Higman C, van der Burgt M (2008) Chapter 1—introduction.
Gasification 1–9. https://doi.org/10.1016/B978-07506-8528-3.00001
46. Kim S-S, Agblevor FA (2007) Pyrolysis characteristics and kinetics of chicken litter. Waste
Manage 27:135–140. https://doi.org/10.1016/j.wasman.2006.01.012
47. Dejong W, Dinola G, Venneker B, Spliethoff H, Wojtowicz M (2007) TG-FTIR Pyrolysis
of coal and secondary biomass fuels: determination of Pyrolysis kinetic parameters for main
species and NOx precursors. Fuel 86:2367–2376. https://doi.org/10.1016/j.fuel.2007.01.032
48. Kim S-S, Agblevor FA, Lim J (2009) Fast Pyrolysis of chicken litter and turkey litter in a
fluidized bed reactor. J Ind Eng Chem 15:247–252. https://doi.org/10.1016/j.jiec.2008.10.004
49. Agblevor FA, Beis S, Kim SS, Tarrant R, Mante NO (2010) Biocrude oils from the fast Pyrolysis
of poultry litter and hardwood. Waste Manage 30:298–307. https://doi.org/10.1016/j.wasman.
2009.09.042
50. Joseph P, Tretsiakova-McNally S, McKenna S (2012) Characterization of cellulosic wastes and
gasification products from chicken farms. Waste Manage 32:701–709. https://doi.org/10.1016/
j.wasman.2011.09.024
51. Font-Palma C (2012) Characterisation, kinetics and modelling of gasification of poultry manure
and litter: an overview. Energy Convers Manage 53:92–98. https://doi.org/10.1016/j.enconman.
2011.08.017
52. Priyadarsan S, Annamalai K, Sweeten JM, Holtzapple MT, Mukhtar S (2005) Co-gasification
of blended coal with feedlot and chicken litter biomass. Proc Combust Inst 30:2973–2980.
https://doi.org/10.1016/j.proci.2004.08.137
53. Kirubakaran V, Sivaramakrishnan V, Premalatha M, Subramanian P (2007) Kinetics of auto-
gasification of poultry litter. Int J Green Energy 4:519–534. https://doi.org/10.1080/154350707
01583102
54. Xiao X, Le DD, Li L, Meng X, Cao J, Morishita K et al (2010) Catalytic steam gasification
of biomass in fluidized bed at low temperature: conversion from livestock manure compost to
hydrogen-rich Syngas. Biomass Bioenergy 34:1505–1512. https://doi.org/10.1016/j.biombioe.
2010.05.001
55. Yanagida T, Minowa T, Nakamura A, Matsumura Y, Noda Y (2008) Behavior of inorganic
elements in poultry manure during supercritical water gasification. Nihon Enerugi Gakkaishi/J
Jpn Inst Energy. https://doi.org/10.3775/jie.87.731
56. Yanagida T, Minowa T, Shimizu Y, Matsumura Y, Noda Y (2009) Recovery of activated carbon
catalyst calcium nitrogen and phosphate from effluent following supercritical water gasification
of poultry manure. Bioresour Technol 100:4884–4886. https://doi.org/10.1016/j.biortech.2009.
05.042
57. Pinto F, André R, Miranda M, Neves D, Varela F, Santos J (2016) Effect of gasification agent
on co-gasification of rice production wastes mixtures. Fuel 180:407–416. https://doi.org/10.
1016/j.fuel.2016.04.048
58. Mahinpey N, Gomez A (2016) Review of gasification fundamentals and new findings: reactors,
feedstock, and kinetic studies. Chem Eng Sci 148:14–31. https://doi.org/10.1016/j.ces.2016.
03.037
59. Karatas H, Olgun H, Akgun F (2012) Experimental results of gasification of waste tire with
air and CO2 air and steam and steam in a bubbling fluidized bed gasifier. Fuel Process Technol
102:166–174. https://doi.org/10.1016/j.fuproc.2012.04.013
60. Sharma S, Sheth PN (2016) Air–steam biomass gasification: experiments, modeling and
simulation. Energy Convers Manage 110:307–318. https://doi.org/10.1016/j.enconman.2015.
12.030
Review of Biomass Energy Resources with Livestock Manure 155

61. Broer KM, Woolcock PJ, Johnston PA, Brown RC (2015) Steam/oxygen gasification system
for the production of clean syngas from switchgrass. Fuel 140:282–292. https://doi.org/10.
1016/j.fuel.2014.09.078
62. Arabloo M, Bahadori A, Ghiasi MM, Lee M, Abbas A, Zendehboudi S (2015) A novel modeling
approach to optimize oxygen–steam ratios in coal gasification process. Fuel 153:1–5. https://
doi.org/10.1016/j.fuel.2015.02.083
63. Yuan H, Lu T, Zhao D, Wang Y, Kobayashi N (2015) Influence of oxygen concentration and
equivalence ratio on MSW oxygen-enriched gasification syngas compositions. In: Progress in
clean energy, vol 2. Springer, Cham, pp 165–176
64. Guo X, Wang S, Wang K, Qian LIU, Luo Z (2010) Influence of extractives on mechanism of
biomass Pyrolysis. J Fuel Chem Technol 38(1):42–46
65. Mallick D, Poddar MK, Mahanta P, Vijayan, Moholkar S (2018) Discernment of synergism in
Pyrolysis of biomass blends using thermogravimetric analysis. Bioresource Technol 261:294–
305
66. Collot A-G, Zhuo Y, Dugwell DR, Kandiyoti R (1999) Co-Pyrolysis and co-gasification of
coal and biomass in bench-scale fixed-bed and fluidised bed reactors. Fuel 78(6):667–679
67. Fan Y, Li Y, Zhiqiang Wu, Sun Z, Yang B (2018) Kinetic analysis on gaseous products during
Co-Pyrolysis of low-rank coal with lignocellulosic biomass model compound: effect of Lignin.
Energy Procedia 152:916–921
68. He Q, Qinghua Guo Lu, Ding JW, Guangsuo Yu (2019) Rapid Co-Pyrolysis of lignite and
biomass blends: analysis of synergy and gasification reactivity of residue char. J Anal Appl
Pyrol 143:104688
69. Burra KG, Gupta AK (2018) Kinetics of synergistic effects in Co-Pyrolysis of biomass with
plastic wastes. Appl Energy 220:408–418
70. Hossain MS, Islam MR, Rahman MS, Kader MA, Haniu H (2017) Biofuel from Co-Pyrolysis
of solid tire waste and rice husk. In Energy Procedia, vol 110, pp 453–458. Elsevier Ltd. https://
doi.org/10.1016/j.egypro.2017.03.168
71. Costa P, Pinto F, Miranda M, André R, Rodrigues M (2014) Study of the experimental conditions
of the co-Pyrolysis of rice husk and plastic wastes
72. Ng WC, You S, Ling R, Gin KY-H, Dai Y, Wang C-H (2017) Co-gasification of woody
biomass and chicken manure: syngas production biochar reutilization, and cost-benefit analysis.
Energy 139:732–742
73. Dayananda BS, Manjunath SH, Girish KB, Sreepathi LK (2013) An experimental approach on
gasification of the chicken litter with rice husk. Int J Innov Res Sci Eng Technol 2(7):2837–2842
74. Seçer A, Küçet N, Fakı E, Hasanoğlu A (2018) Comparison of co–gasification efficiencies of
coal lignocellulosic biomass and biomass hydrolysate for high yield hydrogen production. Int
J Hydrogen Energy 43(46):21269–21278
75. White JE, Catallo WJ, Legendre BJ (2011) Biomass Pyrolysis kinetics: a comparative critical
review with relevant. J Anal Appl Pyrolysis
76. Phyllis2 (2015) database for biomass and waste [Online]. Available at https://www.ecn.nl/phy
llis2/Biomass/View/3501
77. Ruiz-Gómez N, Quispe V, Ábrego J, Atienza-Martínez M, Murillo MB, Gea G (2017) Co-
Pyrolysis of sewage sludge and manure. Waste Manage 59:211. https://doi.org/10.1016/j.was
man.2016.11.013
78. Selim OM, Hussein MS, Amano RS (2020) Effect of heating rate on chemical kinetics of
chicken manure with different gas agents. J Energy Resourc Technol 142(10)
79. Yang H (2007) Characteristics of hemicellulose cellulose and lignin Pyrolysis. Fuel 1781–1788
80. Selim OM, Amano RS (2020) Co-pyrolysis of chicken and cow manure. ASME J Energy
Resour Technol 143(1):011301. https://doi.org/10.1115/1.4047597
81. Espindola J, Selim OM, Amano RS (2020) Co-pyrolysis of rice husk and chicken manure.
ASME J Energy Resour Technol 143(2):022101. https://doi.org/10.1115/1.4047678
Higher Alcohols as Diesel Engine Fuel

Naveen Kumar , Harveer S. Pali , Ankit Sonthalia , and Sidharth

Abstract India is one of the fastest-growing economies in the world and requires a
vast amount of energy which is mainly met by fossil fuels. India, however, spends
a huge amount of foreign exchange for importing crude petroleum due to limited
indigenous production. The diesel consumption in India is nearly four times higher
than gasoline. It is widely used in locomotives, city transport buses, heavy trucks,
electric generators, agricultural equipment, and underground mining equipment due
to its versatility and higher efficiency. Diesel engines have the main environmental
disadvantage of producing more soot particles and nitrogen oxides. They must also
meet the stringent Euro VI emission standards, which are difficult to attain just
by modifying the engine design. In this situation, higher alcohols such as butanol,
pentanol, hexanol, and octanol have vast potential. Previous work on the use of
lower alcohols as a blend with diesel in compression-ignition engines has suggested
a reduction in emissions. However, problems such as phase separation and increased
fuel consumption have also been encountered during the use of lower alcohols, e.g.,
ethanol in diesel engines. Higher alcohols have the potential to be used along with
diesel to make a homogenous blend without any phase separation. They also have
molecular oxygen present which improves the combustion and reduces emissions.
Higher alcohols are sustainable and have the potential to address the issues of climate
change and energy security. However, the use of these fuels in India is still in the
early stage and large-scale production and adaptation will take its own time.

N. Kumar (B) · H. S. Pali · A. Sonthalia · Sidharth

Centre for Advanced Studies and Research in Automotive Engineering, Delhi Technological
University, Bawana Road, Delhi 10042, India
H. S. Pali
Department of Mechanical Engineering, National Institute of Technology Srinagar, Srinagar,
Jammu & Kashmir, India
A. Sonthalia
Department of Automobile Engineering, SRM Institute of Science and Technology, NCR
Campus, Uttar Pradesh, Ghaziabad 01204, India
Sidharth
Department of Mechanical and Automation Engineering, Maharaja Agrasen Institute of
Technology, Rohini, Delhi 110086, India

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 157
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_7
158 N. Kumar et al.

Keywords Butanol · Pentanol · Hexanol · Octanol · Higher alcohols ·

Compression-ignition engine

Abbreviations

ABE Acetone–Butanol–Ethanol
ATF Aviation turbine fuel
BTE Brake thermal efficiency
BSFC Brake specific fuel consumption
CI Compression ignition
CO Carbon monoxide emission
CRDI Common Rail Direct Injection engine
CuO Copper oxide
EGR Exhaust gas recirculation
HC Unburned hydrocarbon emission
HD Heavy duty diesel engine
HRR Heat release rate
LD Light duty diesel engine
NiO Nickel oxide
NOx Oxides of nitrogen emission
PM Particulate matter emission
ZnO Zinc oxide

1 Introduction

Humans from primitive times have required energy for their survival. This led to
a process of searching and developing energy alternatives, like the employment
of animals for waste, biomass, and wind energy along with others. It allowed the
exploration of the whole world by different civilizations over many millennia.
Fossil fuels are the main provider of energy in the world. They are formed when
the remains of large amount of plants and animals buried deep inside the earth and
at the bottom of the seas and lakes are decomposed for millions of years. Fossil fuels
are chiefly classified into coal, natural gas, and oil.
Oil was formed mainly from old plants and microorganisms that lived in the
ocean and seas of salty water. Oil has been a main source of energy in the developed
countries, which will be exhausting very soon. The use of fossil fuels generates a
large amount of greenhouse gases into the atmosphere, in addition to nitrogen oxides
and sulfur oxides as combustion exhaust. The extraction of fossil fuels also causes
oil and fuel spills in the open sea harming the environment.
Higher Alcohols as Diesel Engine Fuel 159

Fig. 1 Global energy consumption [1]

For these reasons, the exploration of alternative fuels has regained interest due to
economic, social, and environmental considerations worldwide.
From Fig. 1, it is seen that the world’s energy consumption has increased by
nearly 1.3% in the year 2019. Fossil fuels are fulfilling most of the energy demand.
Oil consumption has increased by 0.9 million barrels per day in the year 2019 when
compared to 2018. India accounted for an increase in oil consumption in the year
2019 by 159,000 barrels per day [1].
Indian economy is heavily dependent upon the usage of diesel fuel. It is the
livelihood of many people in India. Several applications like tractors, transport buses,
trucks, and industries employ diesel engines. Diesel consumption during 2018–19
was 39.46%, whereas petrol consumption was only 13.3% followed by 11.7% for
LPG, 10% for petroleum coke, 6.6% for Naphtha, and 3.9% for ATF.
Diesel engines are specifically preferred where the power output and higher effi-
ciency are highly desired. The main applications of diesel engines are the transporta-
tion of public and goods, power generation, heavy industries, and the agriculture
sector. Diesel despite its better thermal efficiency is not preferred owing to its harmful
emissions.
Due to this, alternative sources of energy were found to be capable of replacing
or minimizing the consumption of diesel. It is seen from Fig. 2 that the consumption
of diesel is many folds to that of petrol in India. Therefore, focus on diesel fuel is
necessary for a developing country like India [2].
160 N. Kumar et al.

Fig. 2 Diesel versus petrol consumption in India for the years 2012–2018 [2]

It is for this reason that alternative fuels such as alcohol obtained from biomass
can be a suitable substitute. They also have a positive economic impact, giving energy
independence, and improving competitiveness. Environmentally, they contribute to
reducing air pollution, generating biodegradable by-products. Socially, they help the
growth and diversification of the rural economy and quality of life.

2 Comparison of Higher and Lower Alcohols

Alcohols are having a general formula as Cn H2n + 1 OH. The most popular ones are
methanol and ethanol having numerous applications in the food and industrial sector.
Alcohols can also be used in both spark-ignition and compression-ignition engines.
However, their suitability is to be checked before usage in the engines. The simplest
method for checking the suitability is the comparison of its property with the fuel
used in the engine. Several researchers have studied the suitability of lower chain
alcohols like methanol, ethanol in a gasoline engine, whereas higher chain alcohols,
namely butanol, pentanol, hexanol, and octanol were primarily studied in a diesel
engine.
It is observed from Table 1 that the properties of lower alcohols (methanol, ethanol,
and propanol) like cetane number and calorific value are quite lower than diesel.
Moreover, methanol and ethanol are highly miscible in water [6] making it a non-
preferable candidate for the diesel engine.
Several studies carried out in the past with methanol, ethanol, and propanol
concluded that at some point in time, there will be phase separation of diesel/biodiesel
with these lower alcohols. Therefore, the obvious choice comes out to be higher alco-
hols as a substitute for diesel. The preferred higher alcohols include butanol, pentanol,
hexanol, and octanol.
Higher Alcohols as Diesel Engine Fuel 161

Table 1 Comparison of several chain alcohols [3–9]

Fuel Methanol Ethanol Propanol Butanol Pentanol Hexanol Octanol
properties
General CH3 OH C2 H5 OH C3 H7 OH C4 H9 OH C5 H11 OH C6 H13 OH C8 H17 OH
formula
Density (at 791.3 789.4 803.7 809.7 814.8 821,8 830
15 °C)
Kinematic 0.58 1.13 2.96 3.17 3.44 5.32 5.5
Viscosity (at
40 °C)
Calorific 20.08 26.83 29.82 32.01 32.16 37.1 38.4
value
Flash point 12 13 22 35 48 59 81
Cetane 3.8 8 12 17 20 23 38.4
number/
index
Oxygen % 49.93 34.73 26.62 21.59 18.15 15.7 12.31

With the increase in the number of carbon atoms in the chain of the higher alco-
hols, it is observed that the calorific value, flash point, and cetane number of alcohol
improves. This makes hexanol and octanol a suitable candidate as a diesel substitute.
On the contrary, properties like kinematic viscosity and density also increase with
the increase in carbon atoms. It is worthwhile to mention that based on the proper-
ties discussed in Table 1, all higher alcohols shall work suitably without any other
alteration in a diesel engine.

3 Production of Higher Alcohols

The commercial method for the production of 1-butanol is hydrogenation of n-

butyraldehyde. The aldehyde is formed by the Oxo reaction of propylene. In this
reaction, a mixture of iso and n-butyraldehyde is obtained which can initially be
separated, and the individual aldehydes are then hydrogenated to form butanol. It
is also possible that the mixture of butyraldehydes is hydrogenated, and then, the
mixture of n- and isobutanol is separated through distillation. The hydrogenation is
carried out over a heterogeneous catalyst in the vapor phase. For example, 60 to 40
percent of hydrogen to nitrogen is passed over a mixture of iso and n-butyraldehyde
in the presence of CuO–ZnO–NiO catalyst at 25–196 °C and 0.7 MPa, resulting in
98.6% conversion efficiency of the corresponding alcohols [10]. In an old method,
ethanol was dehydrogenated to acetaldehyde followed by its condensation to croton-
aldehyde and lastly hydrogenation of crotonaldehyde resulting in the formation of
1-butanol. Alternatively, the acetaldehyde can be obtained from ethylene using the
Wacker process [11]. In the US, the Zeigler-Natta chain growth reaction is also
162 N. Kumar et al.

used for producing small amounts of n-butanol from ethylene [12]. The Acetone–
Butanol–Ethanol (ABE) production process is the earliest commercial process for
producing butanol [13, 14]. In this process, corn products or molasses are fermented
with Clostridium acetobutylicum resulting in the formation of acetone, ethanol and
butanol which is later separated. Another fermentation route for producing 1-butanol
is the use of anaerobic bacteria, namely Butyribacteriummethylotrophicum that is
based on synthesis gas. As compared to the catalytic processes used for converting
syngas to alcohol, this new method is insensitive to sulfur presence. However, only
1 g/L of butanol could be produced from this process which is very low for it to be
commercially viable.
A patent for a novel route has been filed in which butadiene is used to produce
1-butanol. In this process, butadiene is first catalytically converted to a mixture of
branched and linear butenyl ethers. The branched ethers are converted to linear ethers
and then hydrogenated to 1-butanol. Isobutyl alcohol is commercially produced by
hydrogenating isobutyraldehyde which in turn is obtained from the propylene hydro-
formylation. The alcohol is also a co-product of methanol synthesis produced from
the Fischer Tropsch process [15]. 2-butanol is produced from Raffinate II type feed-
stocks, that is, C-4 refinery streams that contain saturated C-4 s and n-butenes after
removal of isobutylene and butadiene. In the indirect hydration process, the n-butenes
are esterified in sulphuric acid, and intermediate butyl sulfate is hydrogenated. For
producing t-butyl alcohol, isobutane is taken as the starting material, and arco propy-
lene oxidation process is used, wherein propylene oxide is the main product, and
t-butyl alcohol is the co-product. The process can also be modified to produce only
t-butyl alcohol. In this process, the isobutane is fixed oxidized, and the produced t-
butyl hydroperoxide can be decomposed under catalytic conditions to give gasoline
grade t-butyl alcohol (GTBA) [16–19]. The oxygen released from the decomposition
process can be reutilized for the oxidation of isobutane. Another industrial route for
producing t-butyl alcohol is acid-catalyzed hydration of isobutylene.
Hydroformylation of butenes is used for the commercial production of pentanol.
The syngas reacts with C4 olefins in the presence of rhodium or cobalt-containing
catalyst to yield C5 aldehydes. The C5 aldehyde is then hydrogenated to yield corre-
sponding alcohol. Figure 3 shows the schematic of the process. From various cracking
processes raffinate I is used as a raw material. 1,3 butadiene is separated and Raffi-
nate I contains 44–49 vol.% isobutene, 24–28 vol.% 1-butene, 19 –21 vol.% 2-
butene, and 8–11 vol.% C4-hydrocarbons. The isobutene present in the raffinate
is reacted with methanol and converted to methyl tert-butyl ether. The remaining
olefin mixture is used for producing 1-pentanol and 2-methyl-1-butanol [20]. If the
hydroformylation of n-butenes is carried out in the presence of cobalt catalysts, then
70% 1-pentanol and 30% 2-methyl-1-butanol is obtained, whereas with rhodium-
triphenyl phosphane catalyst, the produced mixture contains 90% 1-pentanol and
10% 2-methyl-1- butanol [21]. At 130 °C, rhodium phosphite catalyst can convert
2-butene to a mixture of 1-pentanol and 2-methylbutanal in the ratio 1.13:1 [22]. At
110 °C, rhodium tricyclohexyl phosphane catalyst converts 2-butene to a mixture of
1-pentanol and 2-methylbutanal in the ratio 0.2:1 [23]. Using rhodium or cobalt cata-
lyst, isobutene is converted to 3-methyl-1-butanol [24]. With cobalt catalyst, small
Higher Alcohols as Diesel Engine Fuel 163

Vent Gas Vent Gas

b b
Recycle Gas Recycle Gas

Liquid c d Pentanol
a
product, Fractionation
Largely
and refining
aldehyde
Synthesis Gas
Hydrogen
Butenes
e
Cobalt Catalyst
a) Oxo reactor (100 – 200°C, 10 – 45 MPa); b) Separator; c) Catalyst
removal(decobalting); d) Hydrogenation unit (50 –250°C, 0.3 – 24 MPa); e)
Catalyst recovery

Fig. 3 Production process of pentanol

amount of 2,2-dimethyl-1-propanal is also produced [21]. Apart from pure olefins

and raffinate, other olefin mixtures containing four carbon atoms can be used for
hydroformylation. At 100 °C, with Co(CO)4 as a catalyst, a mixture of 1-butene,
2-butene, and isobutene in the ratio of 1:1:2 can be converted to a mixture of 1-
pentanol, 2-methyl-1-butanol, and 3-methyl-1-butanol in a ratio of 15:5.2:11 using
the hydroformylation technique [25]. The catalyst used for the hydrogenation of C5
aldehydes is CuO-ZnO, Cu-TiO2 , nickel, and zirconia [26–28].
Pentenes can also be hydrated to produce 2-pentanol, 3-pentanol, 2-methyl-2-
butanol, and 3-methyl-2-butanol, as shown in Table 2. Apart from the hydration
process, modification in the hydroboration reaction can be applied for stereoselec-
tive reactions resulting in the formation of pentanol [29]. 2-methyl-2-butene can be

Table 2 Hydration of pentenes

Olefin Alcohol Process catalyst References
1-pentene 2-pentanol zeolites, [37]
aluminosilicates
2-methyl-2-butene 2-methyl-2-butanol acid ion [38]
exchangers
2-methyl-1-butene/2-methyl-2-butene 2-methyl-2-butanol acid ion [39]
exchangers
2-methyl-2-butene 2-methyl-2-butanol Zeolites [40]
2-methyl-2-butene 2-methyl-2-butanol layer silicates [41]
2-methyl-1-butene/2-methyl-2-butene 2-methyl-2-butanol acetic acid [42]
164 N. Kumar et al.

reacted with mono bromoborane–dimethyl sulfide complex resulting in the forma-

tion of 3-methyl-2-butanol [30]. Dimesityl borane reacts with 2-pentene at 65 °C
to form a mixture of 76% 2-pentanol and 24% 3-pentanol. 2-pentanol, 3-pentanol,
and 3-methyl-2-butanol can also be produced by hydrogenating their corresponding
ketones [31]. Catalysts for these reactions are nickel, copper, palladium, and special-
ized catalyst systems such as ruthenium-polysterene, cationic rhodium complexes,
and ClMgAlH4 [32]. An acid-stable catalyst such as rhenium (VII) oxide (Re2 O7 ) or
osmium tetroxide (OsO4 ) was developed to produce 2,2-dimethyl-1-propanol from
pivalic acid at 90 °C and 100 bar. The yield of the alcohol was 94.3%. Other cata-
lysts used for the process are HTcO4 or KReO4 –HClO4 , zirconia [33], and CuO–
ZnO–Al2 O3 [34]. Isobutene can be reacted with formaldehyde (Prins reaction) to
form 3-methyl-1-butanol. Depending on conditions, 3-methyl-3-buten-1-ol or 4,4-
dimethyl-1,3- dioxane intermediates are formed by reacting isobutene with formalde-
hyde; then, the hydrogenation of the intermediates in the presence of a catalyst such
as L2 Cu(OAc)2 /Cr2 O3 –ZnO–BaO results in the formation of 3-methyl-1-butanol.
Pentanol can also be produced by fermentation using engineered microorganisms
[35] or glucose biosynthesis [36].
1-Hexanol can also be produced by the Ziegler process from ethylene. It is sold
under the commercial name of Alfol – 1 or Sasol. It is used as a solvent, an ingredient
for the perfume industry, and also a plasticizer. Nitrates of 1-hexanol are also used as
a cetane improver. 2-Methyl-1-pentanol is produced by aldol condensation of propi-
onaldehyde followed by hydrogenation of the intermediate 2-methyl-2-pentenal. It
is also used as a solvent. 4-Methyl-2-pentanol is produced as a by-product during the
synthesis of methyl isobutyl ketone. Dow chemicals sell this alcohol under the name
methylamyl alcohol [43]. The paint industry uses this alcohol as a solvent, and it is
also used for cleaning semiconductors. Its other uses are as brake fluid, floatation
aid, and as a fungicide [44]. 2-Ethyl-1-butanol is prepared by the aldol condensation
of 1-butanal and acetaldehyde followed by the hydrogenation of the intermediate.
It is a colorless, pleasant-smelling liquid used as a flow improver and solvent for
paints and varnishes. It is used as raw material for making corrosion inhibitors and
penetrating oils. It is also used as a cleaning agent for the printed circuit boards.
1-Hexanol can also be prepared by engineering microbes. However, Table 3 shows
that the yield of alcohol is less with these microbes. 1-Hexanol was synthesized

Table 3 Production of
Engineered host organisms Source Yield Reference
1-hexanol from microbes
Escherichia coli Glucose 18.5 mg/L [36]
Escherichia coli Glucose 47 mg/L [45]
Escherichia coli Glucose 302 mg/L [50]
Clostridium acetobutylicum Glucose 30 mg/L [46]
Ralstoniaeutropha Glucose 280 mg/L [46]
Clostridium carboxidivorans Syngas 1 g/L [49]
Higher Alcohols as Diesel Engine Fuel 165

from glucose by using a non-natural metabolic pathway in which the chain elon-
gation was carried out by an engineered E. Coli [36]. The authors optimized the
process and observed the highest yield of 18.5 mg/L. Dekishima et al. [45] modified
the 1-butanol pathway that utilizes the E. Coli strain Clostridium co-enzyme A and
used it for synthesizing 1-hexanol from glucose. For producing n-hexanol, Machado
et al. replaced the enzyme Clostridium acetobutylicum Hbd with Ralstoniaeutropha
homologue PaaH1 [46]. The authors used an anaerobic growth platform that resulted
in a ten-fold yield of 1-hexanol. The research group also showcased a 67% increase in
1-hexanol production by applying a selection platform for the PaaH1 evolution. For
producing hexanol and butanol from biomass, an indirect method was devised that
chemically converted 90% of carbon into carbohydrates [47]. A one-step method for
simultaneous production of 1-hexanol and 1-butanol from bioethanol at its super-
critical state was developed by Ghaziaskar et al. [48]. The authors achieved 35%
ethanol conversion efficiency with the highest selectivity of 62% of 1-butanol and
21% of 1-hexanol. Another method is the gasification of the biomass to syngas (CO
+ H2 ) and passing this gas through the bacteria Clostridium carboxidivorans which
autotrophically grows using this gas and produces alcohols like butanol and hexanol
[49]. Although the yield using this method is low, it is viable from a commercial
viewpoint as the medium used is low cost.
The predominant method for producing n-octanol from petrochemical processes is
Ziegler alcohol synthesis in which ethylene is oligomerized [43] or the oxo synthesis
process [51]. 2-ethylhexanol is also alcohol with eight carbon atoms, and it holds
significant use in the manufacture of ester plasticizers for soft polyvinyl chloride.
The alcohol is produced in four steps: First, butyraldehyde is first aldolized and then
dehydrated. The butyraldehyde is itself produced by hydroformylation of propene.
In the reactor carrying out aldolization, butyraldehyde in the presence of aqueous
sodium hydroxide reacts to form 2-ethyl-2-hexanal [52]. The reactants are contin-
uously stirred so that secondary reactions taking place due to overheating can be
stopped. The primary aldol addition must take place rapidly, and the dehydration of
hydroxyaldehyde must be carried out quickly as the aldol is unstable resulting in a
decrease in product quality. The reaction takes place mostly between 80–150 °C at
a pressure below 0.5 MPa. The ratio of aldehyde to aqueous sodium hydroxide is
maintained between 1:10 and 1:20 resulting in 99% conversion rates [53, 54]. The
heat produced from aldolization can be used for steam generation. In the second step,
in a phase separator, the mixture is separated into a lower aqueous phase containing
the aldolized solution and an upper organic phase. Some parts of the aldolized solu-
tion can be recycled, and the rest can be removed via a side stream. To the leftover
solution, water is added and sodium hydroxide is added to maintain the required
concentration. The side stream is then purified by acid treatment, oxidation, and
extraction resulting in the partial recycling of valuable products [52]. In the third
step, the organic product is hydrogenated on a nickel catalyst in single stage [55].
The hydrogenation can be carried out in multiple stages either in the combination
of the gas–liquid phase or liquid–liquid phase [56]. The heat release by breakage
of C = C is very high resulting in temperature control problems thereby decreasing
yield. In single-step hydrogenation, the product is remixed to dissipate heat, and
166 N. Kumar et al.

medium pressure is adequate for conversion [57]. Modern plants use two stages for
removing residual carbonyl compounds thus ensuring that high-grade 2-ethylhexanol
is obtained [58]. For the hydrogenation, heterogeneous catalysts such as copper [59],
nickel [57, 60, 61], or mixed systems [62] are used that give 100% conversion and
greater than 99% selectivity. In the final step, the hydrogenated product is distilled.
First, light ends are separated that can be used for manufacturing 1-butanol. Then,
pure 2-ethylhexanol is collected, and lastly, recyclable intermediate fractions are
separated. Octanol can also be produced from biomass. Recently, a multifunctional
catalyst was developed by Julius et al. [51] producing C8 linear alcohols and di-octyl
ether from a lignocellulosic biomass platform. A yield of up to 93% of linear C8
alcohol products was obtained using this route. Xia et al. [63] achieved a yield of
67% from lignocellulosic biomass using an efficient catalyst system. Even microor-
ganisms have been used for producing octanol from biomass. 1-Butanol pathway was
extended by Machado et al. [46] for increasing the carbon chain length of the alcohol
and reported a yield of 70 mg/L of 1-octanol. Akhtar et al. [64] used Escherichia
coli for engineering a biosynthetic pathway for producing 1-octanol.

4 Use of Higher Alcohols as a CI Engine Fuel

Most of lower carbon alcohols are used as fuel for spark-ignition engines due to
their high octane number. The cetane number of lower carbon alcohols is very low as
compared to petroleum diesel. Therefore, this restricts researchers to use these alco-
hols for compression-ignition engines. In this scenario, butanol, pentanol, hexanol,
and octanol got more attention from researchers for diesel engine fuel. The cetane
number and heating value of these alcohols are much higher than methanol and
ethanol. Various researches have been published by the number of researchers in
which they used higher alcohols as fuel for compression-ignition engines [65–70].

4.1 Application of Butanol in CI Engine

Exhaustive research has been carried out on butanol as a potential CI engine fuel.
Armas et al. studied emission of cold and hot start condition of diesel engine fuel
blended with butanol with having same oxygen contained in test fuel (~3.3% in mass)
and observed that the butanol diesel blends have a positive effect in reducing smoke
and PM concentration during hot engine start with similar nitrogen oxide (NOx)
emissions. On the cold start, the blends produced combustion instability [71].
Similarly, Ibrahim et al. investigated the performance and emission of a blend, of
diesel, biodiesel, and butanol in different ratio and observed that the optimal alter-
native fuel among all the fuels tested was the D50B50 fuel as its use increased the
maximum thermal efficiency of the engine by 6.5% and reduced the specific fuel
Higher Alcohols as Diesel Engine Fuel 167

consumption of the engine brake by 5% less compared to diesel fuel. NO emis-

sions increased significantly with increased engine load and increased slightly with
the use of oxygenated fuels, and all fuels tested did not adversely affect engine
stability [72]. Karabektas and Hosoz tested isobutanol diesel fuel blends on the
single-cylinder diesel engine found that the brake power slightly decreased with
the volumetric increase in the isobutanol subsequently BSFC partially increased
for similar conditions. Emission results revealed that CO and NOx emissions were
decreased, and HC was increased [73]. In another study, Ileri et al. investigated the
effect of a diesel-rapeseed oil-n-butanol on a four-stroke, four-cylinder turbocharged
direct injection diesel engine. Emissions and engine performance was conducted at
full load and variable RPM. Exhaust temperature and brake power got decreased
and BSFC increased. Moreover, HC emissions were reduced, and NOx emissions
were partially increased. Huang et al. investigated the effect of the pilot on low-
temperature combustion in the high-speed diesel engine fueled with n-butanol- diesel
blends. The addition of n-butanol to diesel leads pressure rise rate and BSFC. As well
as an increase in pilot injection timings retards the HRR [65]. Wang et al. reviewed
n-butanol/biodiesel mechanism and predicted exhaust emissions like CO, HC, NOx,
soot, and heat release rate under different conditions agreed well with the exper-
imental data. During the ignition, process biodiesel interacts with n-butanol, and
biodiesel produced OH radicals in low in-cylinder temperature. These OH radicals
are consumed by n-butanol to produce HO2 radicals. In the successive reaction, it
is transformed into H2 O2 , which separates into OH radicals at elevated in-cylinder
temperature [74]. In another similar study, Sahin et al. found n-butanol fumigation
decrease the NOx [67]. Jamrozik et al. rewired co-combustion of alcohol and found
a positive effect on mean effective pressure, thermal efficiency, and stability of how
exhaust emissions were significantly reduced [75]. Zheng et al. worked on a dual fuel
mode RCCI engine. Investigation showed the difference between two combustion
modes at different EGR rates and n-butanol ratios. The optimum EGR rate was 30%.
Yilmaz et al. trialed on a four-stroke, single-cylinder indirect injection diesel engine
and found lower exhaust gas temperature NOx however a little change in CO and HC.
Singh et al. used the n-butanol as an additive eucalyptus biodiesel biodiesel-diesel
blends for performance and emission of CI engine [76]. Similarly, other researchers
examined the performance of diesel engines at 20% of butanol with petroleum diesel.
However, a further increase in the percentage of butanol put some adverse effect on
the performance of the engine [66, 68, 77, 78]. This improvement is due to the oxygen
content in the butanol which improves the combustion characteristics of blended fuel
at higher load. The blend of butanol with diesel fuel also leads to a decrease in exhaust
emissions of the diesel engine. However, some emissions like hydrocarbon increase
slightly [79].
168 N. Kumar et al.

4.2 Application of Pentanol in CI Engine

Pentanol has good fuel characteristics with good corrosion impediment. As well as it
has diesel-like viscosity, calorific value, and stoichiometric properties used in the CI
engine. Apart from this, it has good ignition property, better blend stability, and higher
oxygen content [68]. Wi et al. explored the pentanol-diesel blends up to 30% in the
four cylinders’ direct injection diesel engine and found increased ignition delay and
similar other combustion characteristics. Moreover, HRR curves of the blends shift
closer to the TDC with increasing load [80]. A higher percentage of pentanol-diesel
blends is also utilized by few researchers [81, 82]. Kumar et al. tested the CI engine
at high load with different settings of RGR and injection timings. The study exhib-
ited combustion characteristics like prolonged ignition delay, lower peak in-cylinder
pressure, and more HRR against neat diesel. Some other researchers investigated the
effects of pentanol in vegetable oil in the CI engine [83]. Pentanol-diesel blended fuel
offers greater atomization and better combustion due to lower viscosity and more
oxygen content [84]. Generally, higher emissions of HC and CO was observed by
many researchers; moreover, drop in smoke emission was also observed. NOx emis-
sion showed mixed trends in the working conditions and specific engine type [85, 86].
Higher pentanol content minimizes particulate mass concentration and total particle
number concentration [87]. Radheshyam et al. investigated the effects of pentanol
blends and EGR on the CRDI diesel engine and found a higher delay period for all
blends. BSFC increases and BTE decreases with an added concentration of pentanol.
CO and HC emissions were increased, and NOx emissions were reduced with the
addition pentanol blends and EGR [88]. In another study, Yesilyurt et al. scrutinized
binary and ternary blend of pentanol of a diesel engine. After the study, pentanol
proved its potential as a promising higher alcohol fuel to fulfill the future energy
burden of the automobiles [7, 89].

4.3 Application of Hexanol in CI Engine

Hexanol is a high-carbon bio-alcohol, and it has high energy density and cetane
number. So it is partially used as diesel engine fuel for the last decade. Duncan
et al. explored diesel hexanol blends and suggested potential fuel for CI engine [90].
Sathiyagnanam et al. used the hexanol as a co-solvent to prevent the separation of
ethanol from diesel and improved performance and lower emissions [91]. De poures
et al. investigated the influence of injection timing and exhaust gas recirculation on
hexanol as a CI engine fuel. It has resulted in longer ignition delay, higher peak pres-
sure, and HRR. Authors suggested that 30% hexanol blend was advantageous in terms
of smoke density up to 35.9%; however, it had partially increased in NOx emission
up to 3% [9]. In another assessment, Ashok et al. examined calophyllum inophyllum
biodiesel with hexanol and decanol. All ternary blends diesel, biodiesel, and higher
alcohol revealed higher BTE and lower BSFC. Moreover CO, HC, and smoke were
Higher Alcohols as Diesel Engine Fuel 169

lower. Furthermore, peaks in-cylinder pressure and HRR were significantly lower
than neat diesel [92]. Pandian et al. dopped hexanol into the cashew nut shell oil
biodiesel and found a reduction in emissions. Along with significant improvement
in brake thermal efficiency and reduction in brake-specific fuel consumption [93]. A
similar study was carried out by Raju et al. on the mahua oil biodiesel with hexanol
and concluded that BSFC, HC, and smoke were lower to pure diesel, and BTE and
NOx were slightly higher [94]. In another study, Suhaimi et al. analyzed perfor-
mance, emission, and combustion of CI engine using hexanol blend up to 20%. The
peak pressure of hexanol blends was more which improved the fuel–air mixing ratio.
Moreover, it increased the HRR due to enriched oxygen content [95].

4.4 Application of Octanol in CI Engine

Octanol is higher carbon chain alcohol which is explored extensively in the present
days. Now, researchers started using octanol as fuel for diesel engines. The results
quoted by the researchers are pretty impressive and attractive as well [81, 96–98].
In the initial stage of octanol production, octanol is used to produce in industries
through chemicals and other alcohols and is mostly used in perfumes and cosmetic
products [64]. Once the properties of octanol were tested and found impressive as
fuel and the researchers start working on it as an additive and as the main fuel, the
producer starts working on alternative methods of octanol production. Deep et. al used
diesel blended with octanol and observed that increase in the volumetric blending
percentage of octanol in mineral diesel, improved the BTE, diminished the HC and
CO emissions [98]. Rajesh et al. analyzed the performance of the engine using the
blend of diesel and n-octanol up to 30% and compared with diesel under both natu-
rally aspirated and exhaust gas recirculation mode and found significant improvement
in the engine performance. Due to prolong ignition delay, it developed higher peaks
of in-cylinder pressure and higher heat release rates during the premixed combus-
tion phase. The thermal efficiency is increased while the specific fuel consumption
decreased with an increase in the n-octanol fraction. The emissions of smoke, NOx,
hydrocarbons, and CO decreased with the addition of n-octanol. NOx and smoke
emissions also remained low at all EGR rates. Both BTE and BSFC suffered higher
EGR rates. HC and CO emissions increased as EGR rates increased. N-octanol has
shown great promise for the replacement of fossil diesel by up to 30% [81]. Zhang
et. al conducted several experiments for engine performance and emissions on both
a light engine (LD) and a heavy engine (HD). Besides, cold start tests of mixtures
were performed on a multi-cylinder LD diesel engine. Blends of four different alco-
hols were used with diesel, n-butanol, isobutanol, 2-ethyl hexanol, and n-octanol.
To keep the cetane number close to diesel, the blend is completely mixed with a
cetane index improver, di-tertiary butyl peroxide, or hydrogen treated vegetable oil.
The results indicated that with a similar CN and the same engine configuration,
the alcohol/diesel blends and diesel fuel showed the same onset of combustion and
identical heat release processes. The blends generated slightly faster combustion and
170 N. Kumar et al.

higher indicated thermal efficiency than diesel fuel under most conditions tested and
in both engines. The diesel blends of n-butanol and 2-ethyl hexanol showed good
cold-start performance in the multi-cylinder LD engine. Emissions showed that the
total amount of particulate matter (PM) decreased when alcohol/diesel blends were
used, and the diameter of the PM appears to decrease as the amount of oxygen in the
fuel increases in the LD engine. It was concluded that alcohol/diesel blends produce
much less soot than diesel fuel in both types of engines owing to the higher oxygen
content in the blends but cause slightly increased NO formation [6]. Bharti et al.
analyzed the combustion behavior of n-octanol by reactive force field simulation
[97]. Ashok et al. investigated the effect of n-octanol with calophyllum inophyllum
biodiesel on compression-ignition engine characteristics. Five different fuel blends of
n-octanol from 10 to 50% with calophyllum inophyllum biodiesel were prepared on
volume basis. The experiments showed better results compared to diesel [99].

5 Conclusion

Higher alcohols produced from microorganisms are considered third-generation

biofuels. They have the capability of addressing energy security and environmental
problems. However, huge capital is required presently for producing them but higher
benefits in terms of animal life, vegetation, and human health can be reaped from
their use. In compression-ignition engines, higher alcohols can be easily utilized
which would reduce the reliance on food crops and simultaneously increase the frac-
tion of renewables in the fuel. The properties of alcohol play a major role in engine
performance and emission. Due to the low viscosity and density of the alcohols,
the spray characteristics were improved. Due to the low cetane number, the ignition
delay increases that result in higher pressure rise. The lower calorific value of alcohol
tends to increase fuel consumption. As the alcohols contain oxygen, it improves the
local equivalence ratio resulting in lower soot emissions. It is seen that 20% blending
of butanol, pentanol, and hexanol results in improved engine performance and lower
emissions, whereas 30% blend of octanol is well suited for engine applications.

References

1. BP Stats BP Statistical Review of World Energy (2019)

2. Government of India Indian Petroleum and Natural Gas Statistics 2018–19 (2020)
3. Jeevahan J, Lakshmi Sankar S, Karthikeyan P, Sriram V, Britto Joseph G (2020) Compar-
ative investigation of the effects of lower and higher alcohols/bio-diesel blends on engine
performance and emissions characteristics of a diesel engine. Int J Ambient Energy 41:652–658
4. Fernández JC, Arnal JM, Gómez J, Dorado MP (2012) A comparison of performance of higher
alcohols/diesel fuel blends in a diesel engine. Appl Energy 95:267–275
5. Gautam M, Martin DW (2000) Combustion characteristics of higher-alcohol/gasoline blends.
Proc Inst Mech Eng Part A J Power Energy 214:497–511
Higher Alcohols as Diesel Engine Fuel 171

6. Zhang T, Nilsson LJ, Björkholtz C, Munch K, Denbratt I (2016) Effect of using butanol and
octanol isomers on engine performance of steady state and cold start ability in different types
of diesel engines. Fuel 184:708–717
7. Babu D, Anand R (2017) Effect of biodiesel-diesel-n-pentanol and biodiesel-diesel-n-hexanol
blends on diesel engine emission and combustion characteristics. Energy 133:761–776
8. Atmanli A (2016) Comparative analyses of diesel–waste oil biodiesel and propanol, n-butanol
or 1-pentanol blends in a diesel engine. Fuel 176:209–215
9. De Poures MV, Sathiyagnanam AP, Rana D, Rajesh Kumar B, Saravanan S (2017) 1-Hexanol as
a sustainable biofuel in DI diesel engines and its effect on combustion and emissions under the
influence of injection timing and exhaust gas recirculation (EGR). Appl Therm Eng 113:1505–
1513
10. Cropley JB, Burgess LM, Loke RA (1984) Butyraldehyde hydrogenation: a case study in
process design. Chemtech (United States) 14:374–380
11. Crabtree RH (1988) Butyl alcohols. In: Organometallic chemistry transition metal first. Wiley
Inc, New York, NY, pp 173–176
12. Consulting S (1950) In: Chemical economics handbook, California
13. Donmez S, Ozcelik F, Pamir HH (1990) Optimal conditions for acetone-butanol production
from molasses. Doga Turk Tarim Orman Derg 14:71–81
14. Kwon GS, Kim BH, Ong ASH (1989) Studies on the utilization of palm oil wastes as the
substrates for butanol fermentation. Elaeis 1:91–102
15. Keim W (1989) Carbon monoxide: feestock for chemicals, present and future. J Organomet
Chem 372:15–23
16. Worrell G (1980) Tertiary butyl alcohol production, US4296263A
17. Grane HR, Jr. Jubin JC, Worrell GR (1979) Removing water from tertiary butyl alcohol,
US4239926A
18. Grane HR, Jr. Jubin JC, Worrell GR (1979) Preparing oxygen-containing fuel wherein tertiary
butyl alcohol is the major product of the process, US4294999A
19. Grane HR, Jr. Jubin JC, Worrell GR (1980) Manufacture of tertiary butyl alcohol, US4296262A
20. Arpe H-J, Hawkins S (2010) Industrial organic chemistry, 5th edn. Wiley-VCH Verlagsge-
sellschaft, Weinheim
21. Scheidmeir WJ (1972) Cheminform abstract: hydroformylation of butenen and pentenen, synth,
products and possibilities of their use. Chem Inform sd. 3:383–387
22. Billig E, Abatjoglou AG, Bryant DR, Murray RE, Maher JM (1986) Transition metal complex
catalyzed reactions, US4717775A
23. Young DA (1986) Improved rhodium catalyzed hydroformylation of internal olefins,
EP0195656B1
24. Imai T (1982) Preparation of alcohols, US4438287A
25. Barker GE, Forster D (1983) Preparation of alcohols from olefins having from 3 to 7 carbon
atoms, EP0094456A1
26. Bonnier J-M, Damon J-P, Masson J (1987) Raney nickel as a selective catalyst for aldehyde
reduction in the presence of ketones. Appl Catal 30:181–184
27. Delk FS, Vāvere A (1984) Anomalous metal-support interactions in CuTiO2 catalysts. J Catal
85:380–388
28. Pai C-C (1979) A heterogeneous vapor phase process for the catalytic hydrogenation of
aldehydes to alcohols, EP0008767A1
29. Aoyama Y, Tanaka Y, Fujisawa T, Watanabe T, Toi H, Ogoshi H (1987) Catalytic reac-
tions of metalloporphyrins 3 Catalytic modification of hydroboration-oxidation of olefins with
rhodium(III) porphyrin as catalyst. J Org Chem 52:2555–2559
30. Brown HC, Ravindran N (1977) Monochloroborane-methyl sulfide, H2BClS(CH3)2, and
dichloroborane-methyl sulfide, HBCl2S(CH3)2, as new stable hydroborating agents with high
regiospecificity. J Org Chem 42:2533–2534
31. Pelter A, Singaram S, Brown H (1983) The dimesitylboron group in organic chemistry 6
Hydroborations with dimesitylborane. Tetrahedron Lett 24:1433–1436
172 N. Kumar et al.

32. Fujitsu H, Matsumura E, Takeshita K, Mochida I (1981) Reactivity of ketones in homogeneous

catalytic hydrogenation with cationic rhodium complexes. J Chem Soc Perkin Trans 1:2650–
2655
33. Matsuhita H, Shibagaki M, Takahashi K (1988) Verfahren zur Herstellung von Alkoholen,
EP0285786B1
34. Butter SA, Stoll I (1986) Synthese von Neoalkoholen, EP0180210A2
35. Cann AF, Liao JC (2010) Pentanol isomer synthesis in engineered microorganisms. Appl
Microbiol Biotechnol 85:893–899
36. Zhang K, Sawaya MR, Eisenberg DS, Liao JC (2008) Expanding metabolism for biosynthesis
of nonnatural alcohols. Proc Natl Acad Sci 105:20653 LP–20658
37. Okumura Y, Kamiyama S, Furukawa H, Kaneko K (1984) Process for hydration of olefins,
EP0127486A1
38. Malessa R, Schleppinghoff B (1989) Process for the preparation of tert-amylic alcohol,
EP0325144A3
39. Giles JH, Stultz JH, Jones SW (1980) Process for hydration of olefins to produce alcohols,
US4182920A
40. Maksimov AI, Levitskii II, Minachev KM (1981) Dehydration of 2-methyl-2-butanol on
zeolites. Bull Acad Sci USSR Div Chem Sci 30:2275–2278
41. Ballantine JA, Davies M, O’Neil RM, Patel I, Purnell JH, Rayanakorn M, Williams KJ, Thomas
JM (1984) Organic reactions catalysed by sheet silicates: ester production by the direct addition
of carboxylic acids to alkenes. J Mol Catal 26:57–77
42. Henke AM, Odioso RC, Schmid BK (1962) Hydration of olefins in the presence of a solvent,
US3285977A
43. Falbe J, Bahrmann H, Lipps W, Mayer D, Frey GD (2013) Alcohols Aliphatic. Ullmann’s
Encycl Ind Chem
44. Lewis HL (1973) Control of fungi on cotton plants, US3778509A
45. Dekishima Y, Lan EI, Shen CR, Cho KM, Liao JC (2011) Extending carbon chain length of
1-butanol pathway for 1-Hexanol synthesis from glucose by engineered Escherichia coli. J Am
Chem Soc 133:11399–11401
46. Machado HB, Dekishima Y, Luo H, Lan EI, Liao JC (2012) A selection platform for carbon
chain elongation using the CoA-dependent pathway to produce linear higher alcohols. Metab
Eng 14:504–511
47. Verser DW (2012) Method for the indirect production of butanol and hexanol, US8252567B2
48. Ghaziaskar HS, Xu C (2013) (Charles) One-step continuous process for the production of
1-butanol and 1-hexanol by catalytic conversion of bio-ethanol at its sub-/supercritical state.
RSC Adv 3:4271–4280
49. Phillips JR, Atiyeh HK, Tanner RS, Torres JR, Saxena J, Wilkins MR, Huhnke RL (2015)
Butanol and hexanol production in clostridium carboxidivorans syngas fermentation: medium
development and culture techniques. Bioresour Technol 190:114–121
50. Marcheschi RJ, Li H, Zhang K, Noey EL, Kim S, Chaubey A, Houk KN, Liao JC (2012) A
synthetic recursive “+1” pathway for carbon chain elongation. ACS Chem Biol 7:689–697
51. Julis J, Leitner W (2012) Synthesis of 1-Octanol and 1,1-Dioctyl ether from biomass-derived
platform chemicals. Angew Chemie Int Ed 51:8615–8619
52. Bahrmann H, Hahn H-D, Mayer D (2011) 2-Ethylhexanol. Ullmann’s Encycl Ind Chem
53. Lemke H, Duval R (1972) A process for the preparation of 2-ethylhexanol, DE1966388A1
54. Nienburg HJ, Nicolai E, Hagen W (1955) A process for preparing ethyl—isopropylacrolein.
DE927626C
55. Matthey J (1977) Low-pressure oxo process yields a better product mix. Chem Eng J 84:110
56. Egerev, Pritsker, Tkachenko, Shapiro, Kustov, Gurevich, Sedova (1984) Ledovskikh The
method of obtaining 2-ethylhexanol, SU 1084268
57. Dümbgen G, Neubauer D (1969) Large-scale production of oxo alcohols from propylene at
BASF. Chemie Ing Tech 41:974–980
58. Cornils B, Mullen A (1980) 2-EH: what you should know. Hydrocarb Process 59:93
59. Horn G (1978) C.D.F. A process for producing copper traegerkatalysatoren, DE2538253B1
Higher Alcohols as Diesel Engine Fuel 173

60. Corr H, Haarer E, Hoffmann H (1968) S.W. A process for the preparation of saturated aliphatic
alcohols, DE1277232B
61. Kubicka R, Veprek J, Hum M, Macek Lv, Jasansky Z (1981) Method of making the
2-ethylhexanole, CS207933B1
62. Egerev, Pritsker, Tkachenko, Shapiro, Kustov, Gurevich, Sedova (1979) Ledovskikh Method
of preparing 2-ethylhexanol, SU692824A1
63. Xia Q, Xia Y, Xi J, Liu X, Wang Y (2015) Energy-efficient production of 1-octanol from
biomass-derived furfural-acetone in water. Green Chem 17:4411–4417
64. Akhtar MK, Dandapani H, Thiel K, Jones PR (2015) Microbial production of 1-octanol: a
naturally excreted biofuel with diesel-like properties. Metab. Eng. Commun. 2:1–5
65. Huang H, Liu Q, Yang R, Zhu T, Zhao R, Wang Y (2015) Investigation on the effects of pilot
injection on low temperature combustion in high-speed diesel engine fueled with n-butanol-
diesel blends. Energy Convers Manag 106:748–758
66. Sinha SK, Kumar N (2019) Utilization of blends of biodiesel and higher alcohols in a small
capacity diesel engine. SAE Tech Pap 2019-April (2019), pp 1–8
67. Şahin Z, Durgun O, Aksu ON (2015) Experimental investigation of n-butanol/diesel fuel blends
and n-butanol fumigation—evaluation of engine performance exhaust emissions heat release
and flammability analysis. Energy Convers Manag 103:778–789
68. Vinod Babu VBM, Madhu Murthy MMK, Amba Prasad Rao G (2017) Butanol and pentanol:
the promising biofuels for CI engines—a review. Renew Sustain Energy Rev 78:1068–1088
69. Mack JH, Flowers DL, Buchholz BA, Dibble RW (2005) Investigation of HCCI combustion
of diethyl ether and ethanol mixtures using carbon 14 tracing and numerical simulations. Proc
Combust Inst 30:2693–2700
70. Zhang Q, Yao M, Zheng Z, Liu H, Xu J (2012) Experimental study of n-butanol addition on
performance and emissions with diesel low temperature combustion. Energy 47:515–521
71. Armas O, García-Contreras R, Ramos Á (2012) Pollutant emissions from engine starting with
ethanol and butanol diesel blends. Fuel Process Technol 100:63–73
72. Ibrahim A (2016) Performance and combustion characteristics of a diesel engine fuelled by
butanol-biodiesel-diesel blends. Appl Therm Eng 103:651–659
73. Karabektas M, Hosoz M (2009) Performance and emission characteristics of a diesel engine
using isobutanol-diesel fuel blends. Renew Energy 34:1554–1559
74. Wang X, Liu H, Zheng Z, Yao M (2015) Development of a reduced n-butanol/biodiesel
mechanism for a dual fuel engine. Fuel 157:87–96
75. Jamrozik A, Tutak W, Pyrc M, Gruca M, Kočiško M (2018) Study on co-combustion of diesel
fuel with oxygenated alcohols in a compression ignition dual-fuel engine. Fuel 221:329–345
76. Singh R, Singh S, Kumar M (2020) Impact of n-butanol as an additive with eucalyptus biodiesel-
diesel blends on the performance and emission parameters of the diesel engine. Fuel 277:118178
77. Sidharth, Kumar N (2019) Performance and emission studies of ternary fuel blends of diesel,
biodiesel and octanol. Energy Sources Part A Recover Util Environ Eff 0:1–20
78. Kumar P, Kumar N (2018) Study of ignition delay period of n-Butanol blends with JOME
and diesel under static loading conditions. Energy Sources Part A Recover Util Environ Eff
40:1729–1736
79. Goga G, Chauhan BS, Mahla SK, Cho HM (2019) Performance and emission characteristics
of diesel engine fueled with rice bran biodiesel and n-butanol. Energy Rep 5:78–83
80. Wei L, Cheung CS, Huang Z (2014) Effect of n-pentanol addition on the combustion, perfor-
mance and emission characteristics of a direct-injection diesel engine. Energy 70:172–180
81. Rajesh Kumar B, Saravanan S, Rana D, Anish V, Nagendran A (2016) Effect of a sustainable
biofuel - N-octanol - on the combustion, performance and emissions of a di diesel engine
under naturally aspirated and exhaust gas recirculation (EGR) modes. Energy Convers Manag
118:275–286
82. Rajesh Kumar B, Saravanan S (2016) Use of higher alcohol biofuels in diesel engines: a review.
Renew Sustain Energy Rev 60:84–115
83. Sivalakshmi S, Balusamy T (2011) Performance and emission characteristics of a diesel engine
fuelled by neem oil blended with alcohols. Int J Ambient Energy 32:170–178
174 N. Kumar et al.

84. Li L, Jianxin W, Zhi W, Jianhua X (2015) Combustion and emission characteristics of diesel
engine fueled with diesel / biodiesel / pentanol fuel blends. Fuel 156:211–218
85. Santhosh K, Kumar GN, Radheshyam, Sanjay PV (2020) Experimental analysis of performance
and emission characteristics of CRDI diesel engine fueled with 1-pentanol/diesel blends with
EGR technique. Fuel 267:117187
86. Yilmaz N, Atmanli A (2017) Experimental assessment of a diesel engine fueled with diesel-
biodiesel-1-pentanol blends. Fuel 191:190–197
87. Huang H, Guo X, Huang R, Lei H, Chen Y, Wang T, Wang S, Pan M (2020) Effect of n-
pentanol additive on compression-ignition engine performance and particulate emission laws.
Fuel 267:117201
88. Radheshyam, Santhosh K, Kumar GN (2020) Effect of 1-pentanol addition and EGR on the
combustion, performance and emission characteristic of a CRDI diesel engine. Renew Energy
145:925–936
89. Yesilyurt MK, Yilbasi Z, Aydin M. (2020) The performance, emissions, and combustion char-
acteristics of an unmodified diesel engine running on the ternary blends of pentanol/safflower
oil biodiesel/diesel fuel, Springer International Publishing
90. Aloko D, Adebayo GA, Oke OE (2007) Evaluation of diesel-hexanol blend as diesel fuel.
Leonardo Electron J Pract Technol 10:151–156
91. Sathiyagnanam AP, Saravanan CG, Gopalakrishnan M (2010) Hexanol-ethanol diesel blends
on DI-diesel engine to study the combustion and emission. In: WCE 2010-World Congress
Engineering, vol 2. pp 1602–1606
92. Ashok B, Nanthagopal K, Darla S, Chyuan OH, Ramesh A, Jacob A, Sahil G, Thiyagarajan S,
Geo VE (2019) Comparative assessment of hexanol and decanol as oxygenated additives with
calophyllum inophyllum biodiesel. Energy 173:494–510
93. Pandian AK, Munuswamy DB, Radhakrishanan S, Devarajan Y, Ramakrishnan RBB,
Nagappan B (2018) Emission and performance analysis of a diesel engine burning cashew
nut shell oil bio diesel mixed with hexanol. Pet Sci 15:176–184
94. Dhana Raju V, Kirankumar K, PSK Engine performance and emission characteristics of a direct
injection diesel engine fuelled with 1-Hexanol as a fuel additive in mahua seed oil biodiesel
blends. Int J Therm Environ Eng 13:121–127
95. Syuhaida N, Asri M (2018) Analysis of combustion characteristics. Engine Perform Emissions
of Diesel-Organ Germanium Fuel Blends 220:1–17
96. Kumar N (2018) Sidharth some studies on use of ternary blends of diesel, biodiesel and n-
octanol. Energy Sources Part A Recover Util Environ Eff 40:1721–1728
97. Bharti A, Banerjee T (2016) Reactive force field simulation studies on the combustion behavior
of n-octanol. Fuel Process Technol 152:132–139
98. Deep A, Kumar N (2014) Assessment of the performance and emission characteristics of
1-Octanol / diesel fuel blends in a water cooled compression ignition engine
99. Ashok B, Nanthagopal K, Anand V, Aravind KM, Jeevanantham AK, Balusamy S (2019)
Effects of n-octanol as a fuel blend with biodiesel on diesel engine characteristics. Fuel
235:363–373
Photocatalytic Hydrogen from Water
Over Semiconductors

Siow Hwa Teo, Aminul Islam, and Yun Hin Taufiq-Yap

Abstract Water splitting to produce hydrogen (H2 ) over a semiconductor photo-

catalyst using solar energy is a promising process for the large-scale production
of clean chemical energy in the form of H2 , a clean and renewable energy carrier.
Many efforts have been implemented to discover photocatalysts that workable under
visible-light irradiation to efficiently utilize solar energy. Generally, a suitable cocata-
lyst to provide an active redox site, is a recognize manner to modify the water-splitting
photocatalysts. Apart from that, the constructed two-step photoexcitation photocat-
alysts, mimicking the natural photosynthesis system, using two different semicon-
ductor powder, have many merits, including amplified light harvesting, spatially
separated reductive and oxidative active sites, and perfectly preserved powerful redox
ability, which advantage the photocatalytic performance. Especially, a photocatalytic
framework called Z-scheme, which is fundamentally comprises of a H2 -photocatalyst
and an O2 -photocatalyst to perform water reduction and oxidation, individually.
This chapter describe the fundamental development of water-splitting photocatalytic
systems.

Keywords Semiconductor · Water-splitting photocatalyst · Cocatalyst ·

Z-scheme · H2 evolution

S. H. Teo (B) · Y. H. Taufiq-Yap (B)

Faculty of Science and Natural Resources, Universiti Malaysia Sabah, Jalan UMS, 88400 Kota
Kinabalu, Sabah, Malaysia
e-mail: [email protected]
Y. H. Taufiq-Yap
e-mail: [email protected]
A. Islam
Department of Petroleum and Mining Engineering, Jashore University of Science and
Technology, Jashore 7408, Bangladesh

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 175
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_8
176 S. H. Teo et al.

Nomenclature

AGeometric Area of reactor

Ec Conductive band minimum
EF Fermi level
EH2 /H2 O Hydrogen evolution potential
EO2 /H2 O Oxygen evolution potential
Ev Valence band maximum
Ev Planck’s constant
I light Irradiance intensity at λ
Io Power density of incident light
jp Photocurrent density at the measured potential
J Photocurrent density at 0 V at a certain wavelength
Psun Energy free of sunlight
v Frequency
V Voltage
vac Vacuum level
rH2 Hydrogen production rate
V Bias potential
W Work function
G Gibbs free energy
λ Wavelength
λ Wavelength

Abbreviations

Ag Silver
Au Gold
Br Bromide
Br2 Bromine
C Carbon
CB Conductive band
CH4 Methane
CNTs Carbon nanotubes
CO2 Carbon dioxide
Cu Copper
Cu2 O Copper oxide
DMF Dimethyl fluoride
EQN Equation
g-C3N4 Graphitic carbon nitride
Photocatalytic Hydrogen from Water Over Semiconductors 177

GO Graphene oxide
HER Hydrogen evolution reaction
HI Hydrogen iodine
H2 Hydrogen
H2 O WaterWater
IPCE Incident photon to current conversion efficiency
MoS2 Molybdenum disulfide
NHE Normal hydrogen electrode
Nd2 O5 Neodymium oxide
NiO Nickel oxide
NPs Nanoparticles
OER Oxygen evolution band
O2 Oxygen
pH Potential of hydrogen or power of hydrogen
PC Photocatalyst
PCE Photo conversion efficiency
Pt Platinum
QE Quantum efficiency
RuO2 Ruthenium oxide
SHT Solar to hydrogen
Si Silicon
SrTiO3 Strontium titanium oxide
THF Tetrahydrofuran
TiO2 Titanium oxide
UV Ultraviolet
VB Valence band
WO3 Tungsten oxide
vs. Versus
Zno Zinc oxide
2D Two dimensional

1 Introduction

Solar energy is considered as most promising candidate to substitute the fossil fuels.
It also leads us to achieve clean, sustainable and environment friendly fuel and it is
predicted that only 0.01% solar energy conversion can be enough to solve the world’s
energy crisis [1, 2]. The most abundant component across the universe, hydrogen (H2 )
is burned to generated only water (H2 O) and hence it is reflected as a perspective clean
energy source for the future once it can be made inexpensively and expeditiously [3,
4]. One of the best way to keep and manage the solar energy is photocatalysis which
stores energy in the form of chemical fuel [5]. Therefore, solar H2 production can
attain directly from H2 O splitting process using solar energy as driving force to split
178 S. H. Teo et al.

water molecules to H2 and oxygen (O2 ) on the surface of catalyst. Additionally, the
method is also as source for O2 , carbon dioxide (CO2 ) and methane (CH4 ), which
are the products for other fuel production routes [6].
In 1972, photocatalytic water splitting was emerged, which the first report is autho-
rized by Fujishima and Honda [9] using a photoelectrochemical system. Through the
report published in Nature, suggesting that the H2 O molecule is oxidized via photoin-
duced holes on TiO2 surface with the assistant of small electrical voltage, until near
recent twentieth century, it witnessed the nonstop progression of photocatalysis [2,
5, 8, 10] including photocatalytic water splitting on powder photocatalyst particles.
An overall photocatalytic water splitting process (based on semiconductor) typically
includes three essential steps [10], (1) absorption of light followed by electron–hole
separation inside the semiconductor, (2) migration of the charge carries toward the
surface, and (3) chemical reactions at the surface of the semiconductor making H2 and
O2 evolution. Those processes demonstrated both the oxidizing sites and the reduc-
tion sites on the same catalyst (Fig. 1a). The photons with appropriate energy are
harvested by semiconductor to excite electrons from valence band (VB) to conduc-
tion band (CB), leaving holes in the VB. Consequently, both electrons and holes are
separated and then move towards catalyst’s surface, unless they recombine in the
bulk or on the surface. At the end, the electrons and holes on surface will allocate to
the adsorbed species to initiate corresponding water redox processes, respectively.
Water splitting is thermodynamically not favourable process and an absorption
of light having an energy more than (or equal to) the band gap of the photocatalysts
that consist of the semiconductor materials in order to drive the reaction forward
[6, 8]. Conceptually, the band gap energy should be at least 1.23 eV for overall
water oxidation process [9]. Practically, an activation barrier is created when the
electron moving between molecules photocatalysts and water. Therefore, the photon
energy requires should be greater than the minimum band gap energy in order to
perform photocatalytic water splitting [11]. For the hydrogen evolution reaction
(HER) and the oxygen evolution reaction (OER) to occur on the catalyst’s active
sites, the minimum/ bottom of the CB energy level should be located at a more

Fig. 1 a Schematic illustration for a typical photocatalytic process based on semiconductor [7];
b Band structure of semiconductor and redox potential of water splitting [8]
Photocatalytic Hydrogen from Water Over Semiconductors 179

negative than the H2 evolution potential (E H2 /H2O , 0 V versus NHE, pH = 0) (Eq.

(1)), whereas the maximum/ top of the VB must be more positive than O2 evolution
potential (E 2/H2O , 1.23 eV vs. NHE, pH = 0) [9] (Eq. (2)).

4H+ + 4e− → 2H2 (1)

2H2 O + 4h+ → 4H+ + O2 (2)

In Fig. 1b, the band positions of various materials are shown. Band engineering
is one of the approaches used to design the visible light sensitivity of water split-
ting materials. Low band gap photocatalytic materials are regularly more sensitive
to photon-corrosion as compared to larger band gap materials, but higher band bap
are not adept of harvesting visible light [12]. To achieve the target, many researchers
are concentrated in band gap modification, including coupled with narrow band
gap semiconductors, surface plasmonic effect, dye sensitization in order to improve
visible light activity of the semiconductors [13–17]. Meanwhile, heterojunction
composite-based photocatalysts can be boosted to obtained apposite band gap align-
ment for visible light susceptible water splitting [5, 11]. The non-single compound-
based photocatalysts are expedient over single material-based photocatalysts as they
capable to harvest more visible light than the individual materials. Other than that,
heterojunction architecture with appropriate band edge also facilitates an enhanced
charge separation [2, 6, 8]. Additionally, the efficient photon absorption, charge
carrier transportation, and separation may be attained through physical improvements
such as growth of nanostructured photocatalytic materials with various morpholo-
gies for high crystallinity and stability, lesser defects, and minimum charge diffusion
length [13, 18, 19].
The photocatalytic activity of a substantial is evaluated by solar to hydrogen (STH)
energy conversion for overall redox reaction. The STH efficiency is identified as:

Output energy rH2 × G

SHT = = (3)
Incident solar light energy PSun × AGeometric

where
rH2 H2 production rate
G Gibbs free energy
PSun Energy flux of sunlight
AGeometric Area of reactor

Theoretical rH2 can be counted from the photon’s number in the solar spectrum at a
different quantum efficiency (QE).

2 × rH2
QE(hv) = (4)
Io (hv)
180 S. H. Teo et al.

Hence, hypothetical STH at different QE can be predicted by Eqs. (3) and (4),
respectively. For sacrificial agent-assisted water splitting, photo-conversion effi-
ciency (PCE) and incident photon to current conversion efficiency (IPCE) are used
which are calculated by Eqs. (5) and (6).
 
|V |
PCE = jp × 1.23 − (5)
Io

where
jp Photocurrent density at the measured potential
V bias potential versus NHE
Io power density of incident light

1240 × J
IPCE = (6)
λ × Ilight

where
J Photocurrent density at 0 V at a certain wavelength (λ)
Ilight Irradiance intensity at λ.

The multiple combines different auspicious properties from various materials, which
may also improve physiochemical properties and electronic configuration, hence
slow recombination of photo-generated charges, sustain to photo-corrosion, and an
appropriate band gap can be achieved, which is possible to reach a targeted STH
efficiency. Furthermore, the suitable band alignment and high stability of the photo-
catalytic material in an aqueous mixture are also tremendously important for the
water splitting reaction.

2 Different Type of Junction Used in Photocatalysis

Designation of a suitable semiconductor photocatalyst is the key factor to solve

the current energy and environmental issues owing to its ability to maximize solar
energy to stimulate various photocatalytic reactions. Hence, it is essential for the
beginners/researchers in this field to understand the different type of heterojunctions
used in photocatalysis, which include Type-II heterojunction and Z-schemes. Next,
we will discuss the difference of each heterojunction.
For Type-II heterojunction, this type of junction is composed of two semicon-
ductors with staggered band structure configuration, as shown in Fig. 2a. Under
the incident light with sufficient energy, the electron in photocatalyst I (PC I) (with
Photocatalytic Hydrogen from Water Over Semiconductors 181

Fig. 2 a Schematic illustration of Type-II heterojunction. CB, VB, PC I, and PC II stand for the
conduction band, valence band, photocatalyst I and photocatalysts II, respectively, b schematic
illustration of charge carrier transfer in traditional Z-scheme photocatalysts; A and D stand for
electron acceptor and donor, respectively, c schematic illustration of charge carrier transfer in all-
solid-state Z-scheme photocatalysts, d schematic illustration of charge carrier transfer in direct
Z-scheme photocatalysts. E means the electric field [20]

higher CB position) will be migrating to photocatalyst II (PC II) (lower CB posi-

tion). Whilst the holes move in a reverse way, gave rise to an accumulation of holes
at PCI for an oxidation reaction and the accumulation of electrons at PC II for a
reduction reaction. Although the mentioned effect promotes the spatial separation
and suppressed charge recombination, however, this kind of heterojunction may fail
to drive a specific photocatalytic reaction [21].
A Z-scheme photocatalyst, in contrast could perform excellent photocatalytic
activity owing to (1) simultaneous preservation of strong reduction and oxidation
capabilities; (2) spatial separation of oxidative and reductive active sites; (3) high
charge separation efficiency; (4) wide spectrum of photocatalysts, and (5) extended
light harvesting range. It should be notable that a semiconductor with narrow bandgap
should be selected in order to achieve points (4) and (5).
Depending on the type of charge carrier mediator introduced, the Z-scheme photo-
catalysts are classified into three, one of them is traditional Z-scheme photocata-
lyst, which is usually composed of two different photocatalysts coupled through
a reversible redox ion pair (Fig. 2b), taking examples: Fe3+ /Fe2+ and IO3− /I− are
182 S. H. Teo et al.

used as the charge carriers transfer medium. When the photocatalysts are exposed
to light, the electrons generated in CB of PC II are consumed by electron acceptor
species while the holes in VB of PC I are consumed by electron donor species. The
photogenerated holes in VB of PC II participated in the oxidation reaction, while the
electrons in the CB of PCI performed a reductive reaction, hence gave rise to enhance
photocatalytic performance. Nevertheless, there are some shortenings of this type of
photocatayst, as listed below [22–24]:
i. Redox mediator-induced back reactions are thermodynamically feasible due
to the generated electrons and holes with strong redox power are consumable
by shuttle redox ion pairs.
ii. Light shielding effect in solution systems.
iii. Slow charge carrier transfer and solution pH sensitivity
iv. Reduced reaction rate due to unstable redox mediators.
While, for all-solid-state Z-scheme photocatalyst, an electron solid conductor
(Au, Ag, and Cu NPs) is utilized as a charge transfer bridging in all-solid-state Z-
scheme photocatalyst (Fig. 2c). Typically, noble metals are commonly employed as
excellent electron mediators, in addition, graphene, CNTs, etc. are also good for
photocatalytic performance and stability performance. The all-solid-state Z-scheme
photocatalysts are viable to work in both liquid and gas phases, ascribed to the
solid conductor prompted for a fast charge carrier transfer. However, the problems
of this photocatalyst would be the expensive noble metals and its shielding effect
[23, 25–28].
Lastly, a direct-Z-scheme photocatalyst is meant for a direct contact between two
semiconductors without the need of charge carrier transfer mediator [29–33]. As
shown in Fig. 2d, both the semiconductors are in close contact without any mediator,
which resulted in suppressed backward reactions, shielding effects, and resistant to
corrosion. In this case, a work function difference between both photocatalysts is
important to induce charge redistribution and to form internal electric field, which is
significantly affect the charge carrier separation and transfer process [34, 35].
To fully understand the mechanisms, there are two types of photocatalytic systems
to be discussed, which are (i) PC I has a higher CB and VB positions and smaller
work function (higher Fermi level) than PC II, and (ii) typical p–n junction. For the
first case (Fig. 3a), free electrons of PC I can be easily transfer to PC II when PC I
and PC II are in contact until their Fermi levels are equilibrated (Fig. 3b). Hence, a
built-in electric field and a band edge bending are formed, attributed to the PC I side
is positively charged, while negatively charged at the PC II interface. The Z-scheme
charge transfer mode is much in favour for this case. The existence of internal electric
field favors the recombination between the photogenerated electrons in the CB of
PC II and photogenerated holes in the VB of PC I [36] (Fig. 3c). The factors such
as internal electric field, the extra potential barrier induced from band bending, and
Coulomb repulsion hinder the transferred of photogenerated electrons from the CB
of PC I to PC II CB, as well as the transfer of photogenerated holes (Fig. 3d).
Meanwhile, the electrons from the CB of PC I and holes from the VB of PC II
are maintained and spatially separated, in addition, they can participate in specific
Photocatalytic Hydrogen from Water Over Semiconductors 183

Fig. 3 Schematic illustration of semiconductor–semiconductor junction with staggered band

configurations (W 1 < W 2 ) [18]: a before contact, b in contact, c photogenerated charge carrier
transfer process in direct Z-scheme mode, d photogenerated charge carrier transfer process in
Type-II mode, and e photogenerated charge carrier recombination. W 1 and W 2 denote the work
function of PC I and PC II, respectively. V ac , E c , E v , and E F stand for the vacuum level, conduction
band minimum, valence band maximum, and Fermi level, respectively

reductive and oxidative photocatalytic reactions, respectively. It is worth noting that

the induced electric field can suppress the recombination between the photogenerated
electrons in the CB of PC I and the photogenerated holes in the VB of PC II, as
shown in Fig. 3e. In contrast, the second type of the system, p–n junction (Fig. 4a),
is significantly different from the first one (Fig. 3a). When p-type semiconductor
is contacted with the n-type semiconductor, free electrons of n-type semiconductor
can be transferred to p-type semiconductor to form a built-in electric field (Fig. 4b).
Under this circumstance, a p–n junction charge carrier transfer mode (Fig. 5c) is
formed rather than direct Z-scheme charge carrier transfer mode (Fig. 4d) owing to
the existed built-in electric field [37, 38]. Thus, the photogenerated charge carrier
184 S. H. Teo et al.

Fig. 4 Schematic illustration of p–n junction [18]: a before contact, b in contact, c transfer of
photogenerated charge carriers in p–n junction mode, and d not-allowed transfer of photogenerated
charge carriers in direct Z-scheme mode

transfer mode in p–n junction is unflavored for direct Z-scheme mode (Fig. 4c and
d).
Similar to direct Z-scheme photocatalysts, all-solid-state photocatalysts
exhibiting similar VB and CB edge bending. At the metal–semiconductor interface
of all-solid-state photocatalysts, the free electrons will be migrated from a semi-
conductor to a metal when metal has larger work function than the semiconductor,
resulted in an upward bent semiconductor band induced by an electric field [23, 25].
Nevertheless, when metal has smaller work function than semiconductor, the free
electrons will then be migrated from a metal to a semiconductor, which led to a
downward bending of semiconductor band edge. While for the traditional Z-scheme
photocatalysts, free electron redistribution would not be occurring (no internal elec-
tric field between them) because there isn’t direct contact between two different
photocatalysts in the presence of an appropriate shuttle redox ion mediator [31].
Hence, the VB and CB edges of each component are not affected when traditional
Z-scheme photocatalysts are formed.

3 Why Semiconductor in Photocatalytic Hydrogen

Production

Why semiconductor for photocatalysis? Semiconductor-based photocatalysis such

as TiO2 (the most commonly reported) has emerged as an effective photocatalyst
for various applications owing to its renewable, clean, and safe technology where
photocatalyst requires only solar energy as the prime energy source [39, 40]. The
pioneering work of TiO2 semiconductor photocatalysis was initiated by Fujishima
and Honda for photochemical water splitting [9]. TiO2 based two dimensional (2D)
nanosheets are derived from the exfoliation of titanate layer. Although it exhibits
Photocatalytic Hydrogen from Water Over Semiconductors 185

Fig. 5 a Schematic illustration of the crystal structure of a typical lepidocrocite-type titanate and
its exfoliation into 2D titania nanosheets [43], b Schematic illustration for the formation of 3D
ultrathin porous N-doped g-C3 N4 [44]

similar properties as rutile and anatase TiO2 , however, it has larger band gap due to
the size quantization. Taking an example, Ti0.91 O2 0.36− nanosheets has a band gap of
3.8 eV, as compared to anatase TiO2 with a band gap of 3.2 eV [41]. Usually layered
titanates are prepared through high temperature solid state reaction of a mixture of
TiO2 and alkali metal carbonates, subsequently processed with acidic solution to form
protonated intermediate via ion-exchange process (Fig. 5a). In addition, other layered
metal oxides nanosheets such as WO3 , titanoniobate, and perovskite oxides are also
186 S. H. Teo et al.

prepared through high temperature solid state reactions and wet-chemical exfolia-
tion processes, where high temperature synthesis process is unfavouring industrial
purpose.
Doping of TiO2 has been showing effective approach to extend light absorption
to visible region, however, low photocatalytic efficiency due to rapid charge recom-
bination could be the main challenge of the existing photocatalysts [42]. Thus, to
overcome the recombination issue and to improve charge separation, developing a
suitable semiconductor composites or engineering a 2D material such as graphene or
graphitic carbon nitride (g-C3 N4 ) is one of the best options. The g-C3 N4 is denoted
as the most stable allotrope among the carbon nitride and g-C3 N4 exists in the form
of 2D sheets consisting of tri-s-triazines interconnected with tertiary amines. What
is the difference between TiO2 and g–C3 N4 ? Unlike TiO2 (active in UV region),
g-C3 N4 is a visible light active photocatalyst due to its bandgap of ca. 2.7 eV with
a suitable conduction and valence band positions. In addition, g-C3 N4 is thermally
and chemically stable in most solvents such as water, alcohols, DMF, THF, diethyl
ether, and toluene.
Recently, ultrathin 2D g-C3 N4 has attracted consideration attentions due to its
large specific surface area, and low carrier recombination properties, which thus
demonstrated superior potential for electronics, energy conversion and storage appli-
cations. A highly anisotropic 2D g-C3 N4 nanosheets prepared through a top-down
strategy (thermal oxidation etching of bulk g-C3 N4 in air) has demonstrated a high
specific surface area of 306 m2 g−1 , accompanied with larger band gap, thus improved
electron transport ability along the in-plane direction, increased photoexcited charge
carrier lifetime, and excellent photocatalytic activities than bulk g-C3 N4 [45]. Never-
theless, the 2D g-C3 N4 nanosheets usually suffer from severe restacking problem,
which is highly detrimental to the photocatalytic activity. Thus, owing to its tunable
morphologies, g-C3 N4 is highly favored in producing an enlarged specific surface
area, improved charge transfer efficiency and surface catalytic reaction. In accord
to the advantages of three dimensional (3D) porous structure, superior photocat-
alytic performance has been reported for a 3D mesoporous g-C3 N4 . This photocat-
alyst, in short denoted as UM3 has shown remarkable photocatalytic H2 evolution
performance and stability, as compared to the pristine (Fig. 5b). Hence, the research
on developing more potential semiconductors which is cost-effective, efficient in
photocatalytic activity and charge transport is still ongoing.

4 Facile Developments of Photocatalyst for Water Splitting

Despite the TiO2 has demonstrated improved photocatalytic activity towards water-
splitting, its efficiency is still far underneath the prerequisite for commercializa-
tion due to the inherent constraint of TiO2 . Hence, the researchers have been
concentrated to create other potential cocatalysts to improve the productivity of
water-splitting reaction. The photocatalysts combined with at least two elements
for H2 evolution through photocatalytic water splitting that are more active than
Photocatalytic Hydrogen from Water Over Semiconductors 187

TiO2 . For instant, Liu et al. [46] reported efficient Au–Cu/TiO2 /MoS2 composite
photocatalyst for H2 production through water splitting. Firstly, TiO2 nanosheets
were blended, at that point hybridized them with MoS2 nanosheets by a simple
aqueous solution. Subsequently, the Au-Cu nanoparticles were effectively inserted
into the TiO2 /MoS2 composite by a refluxing technique followed by calcination,
lastly obtained TiO2 /MoS2 /Au-Cu ternary composite catalyst. The balanced mixture
of the ternary composite improves the photocatalytic efficiency and leds the sepa-
ration of photo-induced electron–hole separation. Zhao et al. [47] synthesized NiO-
TiO2−x /C photocatalysts for H2 production through water splitting. Because of the
cooperative energy of carbon nanosheets, NiO, and TiO2 , the optimum (1 wt.%) NiO-
TiO2-x /C-T650 nanocomposites was demonstrated efficient photocatalytic hydrogen
evolution from water, beating 18-fold in efficiency higher than TiO2 /C. However,
NiO/TiO2 nanoparticles generates agglomerate into enormous bunches, obstructing
effective photo-induced electron transfer, which reduces a space of NiO/TiO2 p-n
heterojunctions for the photocatalytic reaction.
Stabilization of Cu2 O onto TiO2 was reported by Wei et al. [48] by adjustment
of crystal facets and defects of TiO2 (Fig. 6). Both experimental and hypothet-
ical conception demonstrated that 101-faceted TiO2 could make electron importing
channel into the Z-scheme system. The Cu2 O/TiO2 heterostructures with {101}
facet was reported to be 251-fold activity for photo-induced water splitting reac-
tion compared with other facets, along with high stability. However, the impact of
defect in Z-scheme system on the stability of Cu2 O should be investigate in detail
for H2 evolution.
Some different semiconductors, for example, Nb2 O5 , NiO–GO, NiO–TiO2 ,
Pt/RuO2 and SrTiO3 were demonstrated to show photocatalytic water splitting [49–
51]. However, the proficiency of these mass photocatalysts is restricted by active sites
of reaction and low change transfer efficiency because of the electron/hole recom-
bination. In this manner, discovering steady and productive photocatalyst materials
actually stays challenging. Two dimensional materials which have been an inter-
esting issue in a decade ago, presenting tunable electronic and optical properties that
make them promising candidates for water splitting [52]. A few exploratory examina-
tions have demonstrated that 2D material-based photocatalysts display exceptionally

Fig. 6 Schematic illustration of synthetic routes for Cu2 O/TiO2 heterostructures [48]
188 S. H. Teo et al.

improved photocatalytic activities contrasted with the conventional bulk materials

[53, 54].
To overcome the photocorrosion issue of composite photocatalyst while keeping
up its high effectiveness, a photocatalytic framework called Z-scheme, which imitates
the Z-scheme system in the characteristic photosynthesis of green plants, has been
created to produce H2 and O2 at the same time. The Z-scheme is a double photo-
catalyst framework that fundamentally comprises of a H2 -photocatalyst and an O2 -
photocatalyst to perform water reduction and oxidation, individually. Wen et al. [55]
synthesized ZnO coupled with g-C3 N4 to create direct Z-scheme, photocatalytic
systems attributable to their well-matched band positions. The outcomes affirmed
that the inserted Au NPs act as an electron mediator to advance vectorial electron
move in direct Z-scheme ZnO/g-C3 N4 heterojunction. Recently, some works reported
that the ZnO/TiO2 heterojunctions demonstrated increased photocatalytic activity
due to their matched energy band alignments and high electron mobility of ZnO that
efficiently promote the carrier separation and transfer [56]. However, the reported
ZnO/TiO2 heterostructures were usually sealed structures with lower surface areas,
such as the core–shell structured ZnO/TiO2 or branched ZnO/TiO2 and disordered
ZnO rods due to the aggregation of ZnO seeds on the TiO2 nanorods, which result
in the obstruction of light absorption for TiO2 , usually demonstrating unsatisfactory
water splitting performance. Besides, ZnO also suffers from the low photocorrosion
resistance in most electrolytic media and the morphologies and structures of ZnO
seriously affect the photostabilities of the ZnO/TiO2 heterojunctions.
Niu et al. [57] evaluated the photogenerated electrons/holes recombination prop-
erties of C3 N/C3 B catalyst and compared the performance for H2 generation through
water splitting. The 2D C3 N/C3 B p-n heterostructure was found to be promising
metal free Z-scheme photocatalyst for water splitting reaction. The strong redox
ability of C3 B and C3 N layers are driving force to split water without addition
of overpotential or co-catalyst. Bismuth-based oxides have attracted attention in
photocatalytic water splitting because of their high dielectric permittivity, reason-
able band structure (2.8 eV) and position of valence band [58]. The Z-scheme photo-
catalyst offer successful charge seperation and move without adding any sacrifi-
cial reagent. Nitrogen/fluorine codoping decreases the defect concentration, which
reduced the recombination of photoexcited electrons, contrasted with nitrogen-doped
photocatalyst. It has been suggested that fluorine doping into TiO2 improved its
photocatalytic water splitting under UV illumination by supressing recombination
of electron/hole. Miyoshi et al. [59] compared the activity of nano-TiO2 :N,F and
TiO2 :N,F in Z-scheme system under visible light irradiated Z-scheme water-splitting.
It was concluded that Z-scheme system of TiO2 :N,F was found to increase four-
fold compared to the bulk TiO2 :N,F. While TiO2 :N,F in Z-scheme system has been
reported more effective than bulk TiO2 :N,F, the impact of TiO2 :N,F molecule size
remain unknown.
Utilization of Nanostructured Si essentially improves the solar H2 conversion
proficiency of an ideal photoelectrochemical system. Oh et al. [60] demonstrated that
the water splitting efficiency of nanoporous ‘black’ Si was more than the polished Si
electrode. In similar, n-type nano-Si appeared about 20% improvement in quantum
Photocatalytic Hydrogen from Water Over Semiconductors 189

efficiency because of the decreased reflectance and high photocorrosion in concen-

trated HI. Recently, n-type Si nanowire demonstrated the high water splitting activity,
contrasted with the planar Si, in the presence of Br2 /Br solution. Zhang et al. [61]
investigated the activity of water splitting by running Si photocatalyst. It has been
reported that the photocatalytic hydrogen development execution of silicon catalyst
incredibly impacts on a surface functional group of catalyst. This outcome is consis-
tent with the information proclaimed by Rye et al. [62] who endeavored to synthesized
Pt/Si photocatalyst for H2 production through water splitting. Nonetheless, the costly
platinum co-catalyst remains challenge for real application.
There is a basic postulation that the H atoms on Si surface are electron-deficient
and may fill in as an electron sink to improve charge separation. Similar phenomenon
has been discussed by other researchers and reported that oxidation occurs on Si–OH
and H2 produces on Si–H surfaces by reduction process, as shown the Eqs. (7–9):

Oxidation : h+ + Si + H2 O → Si − OH + H+ (7)

Reduction :e− + Si − H + H+ → Si + H2 (8)

Reduction : e− + Si − OH + H+ → Si + H2 O (9)

The function of photogenerated electron and holes on the photocatalyst is schemat-

ically shown in Fig. 7. The function of photogenerated electrons on Si photocatalyst
could be elucidated based on the dynamic charge separation [63, 64]. As recently
revealed the ultrathin layer of catalyst led the detachment of photogenerated electrons

Fig. 7 Water splitting mechanism on Si photocatalyst [63]

190 S. H. Teo et al.

because of intrinsic effect of Si [65]. While conceding to this conceivable explana-

tion, there is still need to investigate the molecular interaction and kinetic of carriers
for photocatalytic water splitting reaction.

5 Conclusion

In the course of recent years, a few semiconductor materials and photocatalytic

frameworks have been produced for the water-splitting irradiation under UV and
visible light. It has been seen that photo-induced charge separation, avoidance of
water-splitting in reverse reaction, and use of a wide range of light energy are the
basic prerequisites for accomplishing high proficiency of water splitting. The high
evolution of H2 energy through water splitting has been evident by the addition of
sacrificial reagent that could supress the openings to repress charge recombination.
In any case, to accomplish feasible hydrogen evolution, sacrificial reagent must be
persistently added during reaction. Besides, photocatalytic reactor design to accom-
plish separate H2 and O2 development, for example, the H-type reactor and the novel
Z-scheme, should be required for efficient hydrogen production. Different strategies,
for example, potentially doping of metal or metal oxide particles on the photocatalyst
have been effectively utilized to improve the performance of photocatalytic reaction.
Photocatalytic water splitting is a cross-discipline innovation that requires the contri-
bution of specialists from various fields (i.e., scientific experts, electrical architects,
material researchers, and physicists). The advancement of new innovations requires
cooperation with a solid hypothetical foundation for a superior comprehension of
the hydrogen production catalyst so as to think of a minimal effort for hydrogen
evolution.

Acknowledgements The authors acknowledge the financial support of the Impact Research
Scheme (SPB, Project Number: SP0005-2020) from UMS Research Grant, Universiti Malaysia
Sabah.

References

1. Razek SA, Popeil MR, Wangoh L et al (2020) Designing catalysts for water splitting based
on electronic structure considerations. Electron Struct 2:023001. https://doi.org/10.1088/2516-
1075/ab7d86
2. Lin L, Hisatomi T, Chen S et al (2020) Visible-light-driven photocatalytic water splitting:
recent progress and challenges. Trends Chem 2:813–824. https://doi.org/10.1016/j.trechm.
2020.06.006
3. Safari F, Dincer I (2020) A review and comparative evaluation of thermochemical water splitting
cycles for hydrogen production. Energy Convers Manag 205:112182. https://doi.org/10.1016/
j.enconman.2019.112182
Photocatalytic Hydrogen from Water Over Semiconductors 191

4. Chen Y, Ji S, Sun W et al (2020) Engineering the atomic interface with single platinum atoms for
enhanced photocatalytic hydrogen production. Angew Chemie—Int Ed 59:1295–1301. https://
doi.org/10.1002/anie.201912439
5. Miseki Y, Sayama K (2019) Photocatalytic water splitting for solar hydrogen production using
the carbonate effect and the Z-Scheme reaction. Adv Energy Mater 9:1–15. https://doi.org/10.
1002/aenm.201801294
6. Kumaravel V, Mathew S, Bartlett J, Pillai SC (2019) Photocatalytic hydrogen production using
metal doped TiO2 : a review of recent advances. Appl Catal B Environ 244:1021–1064. https://
doi.org/10.1016/j.apcatb.2018.11.080
7. Zhang N, Gao C, Xiong Y (2019) Defect engineering: a versatile tool for tuning the activation of
key molecules in photocatalytic reactions. J Energy Chem 37:43–57. https://doi.org/10.1016/
j.jechem.2018.09.010
8. Kudo A, Miseki Y (2009) Heterogeneous photocatalyst materials for water splitting. Chem
Soc Rev 38:253–278. https://doi.org/10.1039/b800489g
9. Fujishima A, Honda K (1972) Electrochemical photolysis of water at a semiconductor
electrode. Nature 238:37–38. https://doi.org/10.1038/238037a0
10. Wang Y, Vogel A, Sachs M et al (2019) Current understanding and challenges of solar-driven
hydrogen generation using polymeric photocatalysts. Nat Energy 4:746–760. https://doi.org/
10.1038/s41560-019-0456-5
11. Maeda K (2011) Photocatalytic water splitting using semiconductor particles: history and recent
developments. J Photochem Photobiol C Photochem Rev 12:237–268. https://doi.org/10.1016/
j.jphotochemrev.2011.07.001
12. Su J, Guo L, Bao N, Grimes CA (2011) Nanostructured WO3/BiVO4 heterojunction films
for efficient photoelectrochemical water splitting. Nano Lett 11:1928–1933. https://doi.org/10.
1021/nl2000743
13. Kudo A (2013) New materials for photocatalytic and photoelectrochemical water splitting. AIP
Conf Proc 1568:7–10. https://doi.org/10.1063/1.4848079
14. Domen K, Kondo JN, Hara M, Takata T (2000) Photo-and mechano-catalytic overall water
splitting reactions to form hydrogen and oxygen on heterogeneous catalysts. Bull Chem Soc
Jpn 73:1307–1331
15. Domen K, Hara M, Kondo JN et al (2001) New aspects of heterogeneous photocatalysts for
water decomposition. Korean J Chem Eng 18:862–866. https://doi.org/10.1007/BF02705609
16. Kudo A (2006) Development of photocatalyst materials for water splitting. Int J Hydrogen
Energy 197–202
17. Maeda K, Domen K (2007) New non-oxide photocatalysts designed for overall water splitting
under visible light. J Phys Chem C 111:7851–7861. https://doi.org/10.1021/jp070911w
18. Xu Q, Zhang L, Yu J et al (2018) Direct Z-scheme photocatalysts: principles, synthesis, and
applications. Mater Today 21:1042–1063. https://doi.org/10.1016/j.mattod.2018.04.008
19. Yuan Y, Lv J, Jiang X et al (2007) Large impact of strontium substitution on photocatalytic
water splitting activity of BaSnO3 . Appl Phys Lett 91:27–30. https://doi.org/10.1063/1.277
8631
20. Low J, Yu J, Jaroniec M et al (2017) Heterojunction photocatalysts. Adv Mater 29:1–20. https://
doi.org/10.1002/adma.201601694
21. Bai S, Jiang J, Zhang Q, Xiong Y (2015) Steering charge kinetics in photocatalysis: Intersection
of materials syntheses, characterization techniques and theoretical simulations. Chem Soc Rev
44:2893–2939. https://doi.org/10.1039/c5cs00064e
22. Maeda K (2013) Z-scheme water splitting using two different semiconductor photocatalysts.
ACS Catal 3:1486–1503. https://doi.org/10.1021/cs4002089
23. Zhou P, Yu J, Jaroniec M (2014) All-solid-state Z-scheme photocatalytic systems. Adv Mater
26:4920–4935. https://doi.org/10.1002/adma.201400288
24. Bala Chandran R, Breen S, Shao Y et al (2018) Evaluating particle-suspension reactor designs
for Z-scheme solar water splitting via transport and kinetic modeling. Energy Environ Sci
11:115–135. https://doi.org/10.1039/c7ee01360d
192 S. H. Teo et al.

25. Li H, Tu W, Zhou Y, Zou Z (2016) Z-Scheme Photocatalytic systems for promoting photocat-
alytic performance: recent progress and future challenges. Adv Sci 3. https://doi.org/10.1002/
advs.201500389
26. Iwase A, Ng YH, Ishiguro Y, et al (2011) Reduced graphene oxide as a solid-state electron
mediator in Z- scheme photocatalytic water splitting under visible light. Am Chem Soc. pp
11054–11057
27. Li H, Yu H, Quan X et al (2016) Uncovering the key role of the fermi level of the electron
mediator in a Z-Scheme photocatalyst by detecting the charge transfer process of WO3-metal-
gC3N4 (Metal = Cu, Ag, Au). ACS Appl Mater Interfaces 8:2111–2119
28. Li X, Yan X, Lu X et al (2018) Photo-assisted selective catalytic reduction of NO by Z-scheme
natural clay based photocatalyst: insight into the effect of graphene coupling. J Catal 357:59–68.
https://doi.org/10.1016/j.jcat.2017.10.024
29. Jia Q, Iwase A, Kudo A (2014) BiVO4-Ru/SrTiO3: Rh composite Z-scheme photocatalyst for
solar water splitting. Chem Sci 5:1513–1519. https://doi.org/10.1039/c3sc52810c
30. Song H, Yu YT, Norby P (2009) Efficient complete oxidation of acetaldehyde into CO2 over
Au/TiO 2 core-shell nano catalyst under UV and visible light irradiation. J Nanosci Nanotechnol
9:5891–5897. https://doi.org/10.1166/jnn.2009.1263
31. Liu X, Chen N, Li Y et al (2016) A general nonaqueous sol-gel route to g-C3N4 -coupling
photocatalysts: the case of Z-scheme g-C3N4 /TiO2 with enhanced photodegradation toward
RhB under visible-light. Sci Rep 6:1–16. https://doi.org/10.1038/srep39531
32. Zhou D, Chen Z, Yang Q et al (2016) Facile construction of g-C3N4 nanosheets/TiO2 Nanotube
Arrays as Z-scheme photocatalyst with enhanced visible-light performance. ChemCatChem
8:3064–3073. https://doi.org/10.1002/cctc.201600828
33. Luo J, Zhou X, Ma L et al (2016) Enhancing visible light photocatalytic activity of direct
Z-scheme SnS2 /Ag3 PO4 heterojunction photocatalysts. Mater Res Bull 81:16–26. https://doi.
org/10.1016/j.materresbull.2016.04.028
34. Cui J, Li Y, Liu L et al (2015) Near-infrared plasmonic-enhanced solar energy harvest for
highly efficient photocatalytic reactions. Nano Lett 15:6295–6301. https://doi.org/10.1021/
acs.nanolett.5b00950
35. Liu J, Cheng B, Yu J (2016) A new understanding of the photocatalytic mechanism of the direct
Z-scheme g-C3 N4 /TiO2 heterostructure. Phys Chem Chem Phys 18:31175–31183. https://doi.
org/10.1039/c6cp06147h
36. Zhang L, Li S, Liu B, et al (2014) Highly efficient CdS / WO 3 photocatalysts : Z- scheme
photocatalytic mechanism for their enhanced photocatalytic H 2 evolution under visible light
37. Zheng L, Zheng Y, Chen C et al (2009) Network structured SnO2 /ZnO heterojunction nanocat-
alyst with high photocatalytic activity. Inorg Chem 48:1819–1825. https://doi.org/10.1021/ic8
02293p
38. Son DY, Im JH, Kim HS, Park NG (2014) 11% efficient perovskite solar cell based on ZnO
nanorods: an effective charge collection system. J Phys Chem C 118:16567–16573. https://doi.
org/10.1021/jp412407j
39. Bard AJ (1980) Photoelectrochemistry. 207:139–144
40. Hoffmann MR, Martin ST, Choi W, Bahnemann DW (1995) Environmental applications of
semiconductor photocatalysis. Chem Rev 95:69–96. https://doi.org/10.1021/cr00033a004
41. Sakai N, Ebina Y, Takada K, Sasaki T (2004) Electronic band structure of titania semiconductor
nanosheets revealed by electrochemical and photoelectrochemical studies. J Am Chem Soc
126:5851–5858
42. Cao S, Low J, Yu J, Jaroniec M (2015) Polymeric photocatalysts based on graphitic carbon
nitride. Adv Mater 27:2150–2176. https://doi.org/10.1002/adma.201500033
43. Luo B, Liu G, Wang L (2016) Recent advances in 2D materials for photocatalysis. Nanoscale
8:6904–6920. https://doi.org/10.1039/c6nr00546b
44. Tian N, Zhang Y, Li X et al (2017) Precursor-reforming protocol to 3D mesoporous g-C3N4
established by ultrathin self-doped nanosheets for superior hydrogen evolution. Nano Energy
38:72–81. https://doi.org/10.1016/j.nanoen.2017.05.038
Photocatalytic Hydrogen from Water Over Semiconductors 193

45. Niu P, Zhang L, Liu G, Cheng HM (2012) Graphene-like carbon nitride nanosheets for improved
photocatalytic activities. Adv Funct Mater 22:4763–4770. https://doi.org/10.1002/adfm.201
200922
46. Liu Y, Xu C, Xie Y, et al (2020) Au–Cu nanoalloy/TiO2/MoS2 ternary hybrid with enhanced
photocatalytic hydrogen production. J Alloys Compd 820.https://doi.org/10.1016/j.jallcom.
2019.153440
47. Zhao X, Xie W, Deng Z et al (2020) Salt templated synthesis of NiO/TiO2 supported carbon
nanosheets for photocatalytic hydrogen production. Colloids Surfaces A Physicochem Eng
Asp 587:124365. https://doi.org/10.1016/j.colsurfa.2019.124365
48. Wei T, Zhu YN, An X et al (2019) Defect modulation of Z-Scheme TiO2 /Cu2 O photocatalysts
for durable water splitting. ACS Catal 9:8346–8354. https://doi.org/10.1021/acscatal.9b01786
49. Ghouri MI, Ahmed E, Ali A et al (2020) The synergistic role of ternary combined semiconduc-
tors (Nb2 O5 –MoS2 -MO) hetrojunctions in photocatalytic water splitting. Mater Chem Phys
255:123320. https://doi.org/10.1016/j.matchemphys.2020.123320
50. Nogueira AE, Silva GTST, Oliveira JA et al (2020) CuO decoration controls Nb2 O5 photocat-
alyst selectivity in CO2 reduction. ACS Appl Energy Mater 3:7629–7636. https://doi.org/10.
1021/acsaem.0c01047
51. Fo Y, Wang M, Ma Y et al (2020) Origin of highly efficient photocatalyst NiO/SrTiO3 for
overall water splitting: insights from density functional theory calculations. J Solid State Chem
292:121683. https://doi.org/10.1016/j.jssc.2020.121683
52. Faraji M, Yousefi M, Yousefzadeh S et al (2019) Two-dimensional materials in semiconductor
photoelectrocatalytic systems for water splitting. Energy Environ Sci 12:59–95. https://doi.org/
10.1039/c8ee00886h
53. Fu CF, Sun J, Luo Q et al (2018) Intrinsic electric fields in two-dimensional materials boost
the solar-to-hydrogen efficiency for photocatalytic water splitting. Nano Lett 18:6312–6317.
https://doi.org/10.1021/acs.nanolett.8b02561
54. Lv X, Wei W, Sun Q et al (2017) Two-dimensional germanium monochalcogenides for photo-
catalytic water splitting with high carrier mobility. Appl Catal B Environ 217:275–284. https://
doi.org/10.1016/j.apcatb.2017.05.087
55. Wen P, Sun Y, Li H et al (2020) A highly active three-dimensional Z-scheme ZnO/Au/g-
C3N4 photocathode for efficient photoelectrochemical water splitting. Appl Catal B Environ
263:118180. https://doi.org/10.1016/j.apcatb.2019.118180
56. Zhou T, Wang J, Chen S et al (2020) Bird-nest structured ZnO/TiO2 as a direct Z-scheme
photoanode with enhanced light harvesting and carriers kinetics for highly efficient and stable
photoelectrochemical water splitting. Appl Catal B Environ 267:118599. https://doi.org/10.
1016/j.apcatb.2020.118599
57. Niu X, Bai X, Zhou Z, Wang J (2020) Rational design and characterization of direct Z-Scheme
photocatalyst for overall water splitting from excited state dynamics simulations. ACS Catal
10:1976–1983. https://doi.org/10.1021/acscatal.9b04753
58. Ma Y, Jiang X, Sun R, et al (2020) Z-scheme Bi2O2.33/Bi2S3 heterojunction nanostructures
for photocatalytic overall water splitting. Chem Eng J 382:123020. https://doi.org/10.1016/j.
cej.2019.123020
59. Miyoshi A, Miyoshi A, Kato K et al (2020) Nano: Versus bulk rutile TiO2:N, F in Z-scheme
overall water splitting under visible light. J Mater Chem A 8:11996–12002. https://doi.org/10.
1039/d0ta04450d
60. Oh J, Deutsch TG, Yuan HC, Branz HM (2011) Nanoporous black silicon photocathode for H2
production by photoelectrochemical water splitting. Energy Environ Sci 4:1690–1694. https://
doi.org/10.1039/c1ee01124c
61. Zhang RQ, Liu XM, Wen Z, Jiang Q (2011) Prediction of silicon nanowires as photocatalysts
for water splitting: band structures calculated using density functional theory. J Phys Chem C
115:3425–3428. https://doi.org/10.1021/jp111182c
62. Ryu J, Jang YJ, Choi S et al (2016) All-in-one synthesis of mesoporous silicon nanosheets
from natural clay and their applicability to hydrogen evolution. NPG Asia Mater 8:e248–e249.
https://doi.org/10.1038/am.2016.35
194 S. H. Teo et al.

63. Islam A, Teo SH, Awual MR, Taufiq-Yap YH (2020) Assessment of clean H2 energy production
from water using novel silicon photocatalyst. J Clean Prod 244:1–12. https://doi.org/10.1016/
j.jclepro.2019.118805
64. Islam A, Teo SH, Awual MR, Taufiq-Yap YH (2019) Improving the hydrogen production from
water over MgO promoted Ni–Si/CNTs photocatalyst. J Clean Prod 238. https://doi.org/10.
1016/j.jclepro.2019.117887
65. Islam A, Teo SH, Awual MR, Taufiq-Yap YH (2020) Ultrathin assembles of porous array for
enhanced H2 evolution. Sci Rep 10:1–14. https://doi.org/10.1038/s41598-020-59325-4
Flames and Reacting Systems
Evaluation of Hazard Correlations for
Hydrogen-Rich Fuels Using Stretched
Transient Flames

Michael Peter Meyer and Rune Peter Lindstedt

Abstract The use of sustainably produced hydrogen as part of fuel blends provides
a promising route towards reducing the impact of combustion on climate change. The
use of such blends also increases the reactivity of the fuel leading to increased fire
and explosion risks and can necessitate revised device designs. Applicable models,
e.g. for flames in the flamelet regime of combustion, typically rely upon data derived
using steady (strained) laminar flames. Parameters include laminar burning veloc-
ities and Markstein lengths. However, interactions between turbulence and flame
structures are transient in nature. In the current work, physico–chemical properties
are extracted from transient spherical laminar flame computations to investigate the
impact of strain. It is shown (i) that the response of transient flames to rates of strain is
qualitatively similar, but quantitatively different, to strained steady flames and (ii) that
the response of flame structures is correlated with the leading edge. The computations
have also been used to (iii) identify the impact of the Lewis number on the transient
flame response. (iv) Difficulties in relation to the selection of computational flame
iso–contours and the non-uniqueness of Markstein lengths are highlighted and it is
shown that agreement with experimental data demands that properties are extracted
in a consistent manner. Finally, (v) as an application example, the methodology is
used to estimate the minimum ignition spark energies and quenching distances for
methane/hydrogen/air and propane/hydrogen/air mixtures.

Keywords Hydrogen · Hydrogen - Hydrocarbon blends · Ignition energies ·

Quenching distances · Hazard classifications · Risk assessment

1 Introduction

The impact of climate change and the need for environmental protection have pro-
moted the development of low carbon footprint and clean energy technologies [1].

M. P. Meyer · R. P. Lindstedt (B)

Department of Mechanical Engineering, Imperial College, London SW7 2AZ, UK
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 197
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_9
198 M. P. Meyer and R. P. Lindstedt

State-of-the-art gas turbines operate under fuel-lean conditions and are typically opti-
mized for natural gas [2]. A partial substitution with hydrogen leads to the expansion
of the lean operating limit [3] and provides a route towards reduced environmental
impact provided the hydrogen is produced using renewable sources. However, the
change in fuel composition can lead to fuel flexibility problems for manufacturers [4,
5] and to increased safety concerns [6]. The mixture reactivity is strongly dependent
upon the hydrogen content with direct implications for flame propagation speeds,
explosion over-pressures, auto-ignition points, and turbulence–chemistry interac-
tions. The latter can lead to differences in flame surface areas [7] and combustion
instabilities [1, 5, 8]. Li et al. [6] measured flame speeds and over-pressures gener-
ated in an obstructed flame tube for a wide range of binary and ternary H2 , CO, and
CH4 mixtures, with methane showing a stronger inhibiting effect on the mixture reac-
tivity. It was shown that a scaling based on the amount of air required to fully oxidize
the mixture correlated the fuel composition impact on explosion over-pressures [6]
and turbulent flow fields [9]. Simatos et al. [10] further investigated the impact of
hydrogen content for lean (φ = 0.80) binary CH4 /H2 and CO/H2 fuel blends on auto-
ignition in turbulent shear layers and showed that the stronger inhibiting effect of
methane prevailed.
The turbulent burning velocity (ST ) is a key parameter required, for example, to
evaluate the strength of explosions [6] and the flashback propensity in gas turbine
engines [11]. Investigations seeking to correlate ST with the turbulence intensity
are quite common e.g. [12]. Wang et al. [13] investigated the effect of hydrogen
content and explored scaling relations under gas turbine relevant conditions, and
found that hydrogen addition alters the correlation due to preferential diffusive-
thermal effects [14]. The impact of hydrogen blending on methane was found to be
particularly prominent under ultra-lean conditions—of particular relevance to gas
turbines operating in premixed or piloted premixed modes under such conditions.
The Lewis number effect was also found to be significant for such fuel blends [15]
with the impact of pressure also investigated [16, 17]. Multiple turbulent burning
velocity definitions (e.g. Bray [18]) have been advanced and Driscoll [19] has shown
that different definitions [20, 21] are not equivalent. Yet, the definitions of Bray [18],
Lawn and Schefer [21] and a variant proposed by Driscoll [19] can agree within
±20% for premixed twin opposed jet flames [22] with the matter further explored by
Hampp et al. [23] for a wide range of hydrogen blends. Models and correlations for
ST inevitably feature the laminar burning velocity with the impact of stretch effects
also frequently included.
A large number of computational studies of one–dimensional steady–state flames
using detailed chemistry have been performed following the seminal investigations
by Dixon–Lewis [24], Miller and Bowman [25] and Warnatz [26], Such studies are
typically aimed at understanding pollutant emissions and the thermo–chemical struc-
ture of flames. However, practical turbulent flames are inevitably subject to unsteady
interactions with the flow field. The transient behavior of curved premixed flame
structures is of particular practical relevance due to the coupling with chemistry
and transport processes [27]. The non–unique nature of such interactions may be
exemplified by steady–state computations of premixed laminar flames in back–to–
Evaluation of Hazard Correlations for Hydrogen … 199

burnt geometries (e.g. [28]) wherein the flame is stabilized against burnt gases [29,
30]. Comparatively few detailed computational studies of transient laminar flames
have been performed despite the increasing awareness of the importance of time–
dependent effects on flame structures. Past studies include the use of asymptotic
methods, aimed to clarify the physics (e.g. Buckmaster [31] and Clavin [32]), and
numerical studies featuring simplified chemistry e.g. Bradley et al. [33] and Ghoniem
et al. [34]. Stahl and Warnatz [35] studied the effects of temporally varying strain
rates on premixed one–dimensional flames, while Yoo and Im [36] studied the tran-
sient dynamics of edge flames in a non-premixed hydrogen-air counterflow. Malik
and Lindstedt [37, 38] studied the transient response of inhomogeneous hydrogen
and methane flames to pressure fluctuations and stretch. Ciardiello et al. [39] recently
studied the controlling parameters for ignition transients in premixed annular com-
bustors and concluded that the laminar burning velocity is a key parameter.
The advantage of spherically expanding/contracting flames is that the rate of strain
is uniformly distributed over the flame surface and the geometry is hence amenable
to comprehensive computational studies due to symmetry. A computational method
using a mass-based Lagrangian frame of reference is here combined with detailed
transport and detailed chemistry featuring 300 reactions and 48 species to study the
response of transient flames to rates of strain which is shown (i) to be qualitatively
similar, but quantitatively different, to strained steady flames. It is further shown
(ii) that the response of flame structures is correlated with the leading edge. The
computations have also been used to (iii) identify the impact of the Lewis num-
ber on the transient flame response. (iv) Difficulties in relation to the selection of
computational flame iso–contours and the non-uniqueness of Markstein lengths are
highlighted and it is shown that agreement with experimental data demands that
properties are extracted in a consistent manner. Finally, (v) as an application exam-
ple, the methodology is used to estimate the minimum ignition spark energies and
quenching distances for methane/hydrogen/air and propane/hydrogen/air mixtures.

2 Extraction of Stretch Related Parameters

Karlovitz et al. [40] described the influence of a velocity gradient on a flame by

introducing the concept of stretch and this well-known concept is illustrated in Fig. 1.
Stretch can be either positive or negative. An outwardly propagating spherical flame
is an example of the former, whereas a contracting spherical flame experiences a
negative stretch. Transient flames experience variable velocity gradients, though
stretch–free steady (e.g. one–dimensional flat flames) or stretched stationary cylin-
drical and spherical flames, maintained by a line or point source, are frequently used
to extract data. In order to define stretch, a surface s (e.g. an iso-contour), located
within the flame structure moves with a velocity u L and the surrounding flow field
has a velocity v as is shown in Fig. 1. The flame stretch K is then the rate of change
of a small element A of that surface normalized by A:
200 M. P. Meyer and R. P. Lindstedt

Fig. 1 Sketch of a surface

element s showing the
velocity u L of the surface
element, velocity v of the
fluid and the normal unity
vector n

1 dA
K = (1)
A dt
Buckmaster [31] and Matalon [41] have shown that K can be expressed as,

K = ∇t · vt + (u L · n)(∇ · n) (2)

where n is the unit normal vector pointing in the direction of the unburnt gas with
∇t and vt the tangential components. The first term in Eq. (2) can be written as

∇t · vt = −n · ∇ × (v × n). (3)

The stretch rate can be recast and expressed in two components, where K c is due to
flame curvature (curvature stretch) [42] and K s is due to strain (flowfield strain),

K c = (u L · n)(∇ · n) (4)
K s = −n · v(∇ · n) + n · n(∇v) (5)

where u L is the laminar burning velocity of the flame normal to the surface s. For
spherical flames K c and K s can be expressed as,
ug
Ks = 2 r
(6)
Kc = 2n r urL (7)

where n r is the normal unit vector in radial direction and is equal to 1 for outwardly
propagating flames and equal to −1 for contracting flames. The gas velocity ahead of
the flame is denoted u g and the radius r is typically defined by the leading edge. The
diffusivity of the mixture is usually discussed in terms of the Lewis number (Le).

λ
Le = (8)
ρc p D
Evaluation of Hazard Correlations for Hydrogen … 201

For Le = 1 the mixture is diffusion–transport neutral and thus pure stretch has only a
small impact. Frankel and Sivashinsky [43] show, via an asymptotic analysis assum-
ing the flame radius to be much larger than the thermal–diffusion thickness, that the
upstream properties (e.g. the mass burning rate m b and flame thickness δb ) are not
affected by stretch when Le = 1 with only the downstream burning rate changed
from its unstretched value to m b (Au /Ab ) following the area change caused by ther-
mal expansion and flow divergence. If Le = 1 the response of a flame towards stretch
can change markedly (e.g. Law [44], Sun and Law [45]) as the temperature depends
on the relative rates of mass and heat transport. For Le > 1 the mass gain is lower
than the heat loss and consequently the burnt gas temperature (Tb ) is reduced from
the adiabatic value (Tad ) so that Tb < Tad and for Le < 1, conversely, Tb > Tad . The
effect of stretch for non–unity Lewis number can be observed, for example, with
lean CH4 –air mixtures. Due to the higher diffusivity of the “lighter” CH4 , it diffuses
more readily towards the reaction zone and thus changes the local equivalence ratio
towards stoichiometric conditions. The burning rate of these mixtures is accordingly
enhanced by strain. The opposite would be true for propane. Such effects are explored
below under transient conditions.

3 A Lagrangian Calculation Method for Transient

Laminar Flames

The current calculation method is based on the simulation of transient laminar flames
in a spherically symmetric co–ordinate system. Flames of this type are predominantly
influenced by tangential scalar gradients that affect the local balance between chem-
istry and transport processes. A mass-based (M) Lagrangian frame of reference is
adopted and the governing equations are shown below [46, 47].
Continuity (mass conservation)

1 ∂M dR
ρ= and u = (9)
R2 ∂ R dt
Enthalpy (assuming constant pressure across the flame)
⎧ ⎫
∂ ⎨ 2  ⎬
Nsp
∂h ∂  2 
ρ = −ρ R Jh + ρ R h k −Jk + Jn (10)
∂t ∂M ∂ M ⎩ k=1 ⎭

Species mass fractions

∂Yk ∂  2 
ρ = −ρ R Jk + Rk Mk (11)
∂t ∂M
Scalar fluxes
202 M. P. Meyer and R. P. Lindstedt

∂Yk 1 ∂n
Jk = −ρ 2 Dk R 2 − Yk − ρWk Yk − ρVc Yk (12)
∂M n ∂M

λ ∂h λ ∂Yk
Jh = −ρ R 2 and Jn = −ρ R 2 (13)
Cp ∂ M Cp ∂ M

Thermal diffusion
D k k 1 ∂ T
Wk = −ρ R 2 (14)
Xk T ∂ M

Correction velocity (to ensure mass conservation)


Nsp 
1 ∂Yk 1 ∂n
Vc = −  Nsp ρ Dk R
2 2
− Yk + ρWk Yk (15)
ρYk k=1
∂M n ∂M
k=1

In the above equations, ρ is the density, Dk is the diffusion coefficient for species
k, Mk its molecular weight, n the mole number (the inverse of the mean molecular
weight), λ the thermal conductivity, C p the heat capacity at constant pressure, X k the
mole fraction of species k and k its thermal diffusion ratio. The solution algorithm
features second–order temporal and spatial accuracy along with adaptive dynamic
grid refinement and a block implicit structure for all solved variables. Flame struc-
tures are resolved using a minimum of 100 locally refined computational cells. The
transport coefficients for viscosities and binary diffusivities are evaluated as specified
by Lindstedt and Maurice [48]. Thermal diffusion (Soret effect) is included using
the formula of Chapman and Cowling [49].
The governing equations are discretised using a finite volume approach featuring
a staggered grid formulation. The resulting set of equations is solved using a block
implicit method. A Newton linearisation technique is utilised for the source terms in
the species equations:
⎡ ⎤

Nreac 
Nsp f
ξ jk 
Nsp
ξ rjk
 jk ⎣k j ⎦
f
Rk = l − k rj l (16)
j=1 l=1 l=1


Nsp
∂ Rk ∂ l  ν+1  ∂ Rk  ν+1 
Rkν+1 = Rkν + Yl − Ylν + T − Tν (17)
l=1
∂ l ∂Yl ∂ T

Dependent variables, such as temperature and density, are included in the block
matrix structure to fully account for dependencies in the transport equations. The
forward integration time step ( t) is controlled by the Courant (Co) number. Second–
Evaluation of Hazard Correlations for Hydrogen … 203

order convergence of the iterative process during the solution of the block-matrix
structure is maintained for Co ≤ 0.5.
Flames were ignited by defining a radical depleted and isobaric ignition kernel of a
radius approximately equal to the flame thickness. The initial temperature and species
concentrations in the fully burnt gas were deduced by assuming chemical equilibrium.
Initial profiles within the ignition region were set using linear interpolation from
reactants to the fully burnt state. The procedure has the benefit of providing weak
ignition kernels.

4 Results for Methane–Air and Propane–Air Mixtures

In order to evaluate the accuracy of the current simulation technique, computa-

tions of expanding (outwardly propagating) spherical methane–air and propane–
air flames were performed. The flame speeds are compared to experimental data
obtained by Taylor [50]. The same initial conditions (i.e. P = 101,325 Pa and To
= 296 K) and equivalence ratio ranges of 0.58 ≤ φ ≤ 1.41 for CH4 –air flames and
0.69 ≤ φ ≤ 1.64 for C3 H8 –air flames were used. Taylor [50] used a schlieren pho-
tographic technique to locate the flame front and the evolution of the 305 K isotherm
was extracted from the computations (c.f. Bradley et al. [33]).
Figures 2 and 3 show comparisons between simulations and experimental data.
The agreement appears satisfactory both qualitatively and quantitatively. Towards
the rich and lean flammability limits some discrepancies occur. This may arguably
be attributed to a stronger ignition source required in the experimental study.

2.5
φ = 0.90
2.0
1.5
Flame Speed [m/s]

φ = 0.79
1.0
0.5 φ = 0.68
0.0
φ = 1.00
2.5
2.0
φ = 1.23
1.5
1.0
φ = 1.35
0.5
0.0
0.000 0.005 0.010 0.015 0.020
Radius [m]

Fig. 2 The computed flame speed (u f ) variation with radius compared with experimental data
(symbols) by Taylor [50] for methane–air flames. Equivalence ratios are listed along the curves
204 M. P. Meyer and R. P. Lindstedt

3.0
2.5 φ = 1.00
2.0
Flame Speed [m/s] 1.5 φ = 0.79
1.0
0.5 φ = 0.69
0.0
3.0 φ = 1.32
2.5
2.0 φ = 1.53
1.5
1.0
0.5 φ = 1.64
0.0
0.000 0.005 0.010 0.015 0.020
Radius [m]

Fig. 3 The computed flame speed (u f ) variation with radius compared with experimental data
(symbols) by Taylor [50] for propane–air flames. Equivalence ratios are listed along the curves

5 Impact of the Rate of Strain on Transient Flame

Structures

The influence of strain on laminar burning velocities is illustrated in Figs. 4 and 5

for CH4 /H2 and C3 H8 /H2 mixtures with air. It can be noted that the effects are
proportional to the hydrocarbon fuel content Hence, pure methane and propane flames
are more strongly affected by the rate of strain than pure hydrogen flames.
The thermo–chemical structure offers a starting point for the evaluation of the
response of flames to transient strain. In particular, the net rate of radical production
provides an opportunity for a detailed evaluation. The radical pool considered in
the present work features H + O + OH + HO2 + CH3 + CH + CHO + 1 CH2 +
3
CH2 + CH2 O + CH3 + CH3 O + CH2 OH + C2 O + C2 H + C2 HO + C2 H3 + C2 H5
+ C3 H + C3 H2 + C3 H2 O + C3 H3 + a-C3 H5 + s-C3 H5 + t-C3 H5 + n-C3 H7 + i-
C3 H7 . Figures 6 and 7 show the net rate of radical production versus temperature
(rather than physical distance) for lean and rich methane–air flames and for lean and
rich propane–air flames, respectively. Distinct points can be identified with Ti the
temperature where the net rate of radical production becomes positive (i.e. the ignition
temperature). For example, Dixon–Lewis [51] correlated extinction limits with the
temperature difference between the maximum flame (Tb ) and ignition temperatures
( Tbi = Tb − Ti ). Other reference planes, e.g. defined by the temperature where
the maximum radical production occurs (Tm ) or the temperature where the radical
consumption rate exceeds production at the hot end of the flame (To ), can also be
defined along with arbitrary iso–contours, e.g. at 400 K, 500 K, 600 K and 1000 K.
As shown in Figs. 6 and 7, flame structures respond differently to increasing rates
of strain depending on the type of fuel and stoichiometry. The results confirm that
the Lewis number of the deficient reactant plays an important role in the response.
Evaluation of Hazard Correlations for Hydrogen … 205

0.6 3.0

0.5 2.5

Burning Velocity [m/s]

0.4 60% CH4/40% H2 2.0
40% CH4/60% H2
20% CH4/80% H2
0.3 100% H2 1.5

0.2 100% CH4 1.0

90% CH4/10% H2
80% CH4/20% H2
0.1 0.5
70% CH4/30% H2

0 0
0.0 500 1000 1500 0 2000 4000 6000 8000
−1
Strain Rate [s ]

Fig. 4 Laminar burning velocities (u L ) of stoichiometric CH4 /H2 mixtures based on the propaga-
tion of the 305 K isotherm as a function of the rate strain

0.6 3.0

0.5 2.5
Burning Velocity [m/s]

0.4 30% C3H8/70% H2 2.0

20% C3H8/80% H2
10% C3H8/90% H2
0.3 100% H2 1.5

0.2 100% C3H8 1.0

90% C3H8/10% H2
70% C3H8/30% H2
0.1 0.5
50% C3H8/50% H2

0 0
0 500 1000 1500 0 2000 4000 6000 8000
−1
Strain Rate [s ]

Fig. 5 Laminar burning velocities (u L ) of stoichiometric C3 H8 /H2 mixtures based on the propa-
gation of the 305 K isotherm as a function of the rate of strain

Due to the higher diffusivity of the comparable “lighter” species, the burning rate
increases for lean CH4 –air mixtures and rich propane–air mixtures (where oxygen
is the deficient reactant).
Inwardly propagating flames can be used to gain additional information about
flame response to curvature. The advantage is that the unburnt gas is not affected
by the flame front as it remains quiescent. Therefore, the straining of the flame
front is only due to curvature effects. However, in the limit of small radii, the flame
thickness becomes the same order of magnitude as the radius and the flame can not
be assumed to be asymptotically thin. Figures 8 and 9 show the rate of radical pool
production for expanding and contracting CH4 –air and C3 H8 –air flames. The radical
206 M. P. Meyer and R. P. Lindstedt

0.8
CH4: φ = 1.41
0.4
Ti Tb
0.0
Rate [kmol /m /s] Tu To
3

−0.4

−0.8
CH4: φ = 0.58 Increasing
0.2 Strain
Ti
0.0
Tu To Tb
−0.2

−0.4
300 600 900 1200 1500 1800 2100
Temperature [K]

Fig. 6 The variation of net rates of radical production in lean and rich methane–air flames with
increasing rates of strain. The radical pool considered here consists of (H + O + OH + HO2 + CH3
+ CH + CHO + 1 CH2 + 3 CH2 + CH2 O + CH3 + CH3 O + CH2 OH + C2 O + C2 H + C2 HO + C2 H3
+ C2 H5 + C3 H + C3 H2 + C3 H2 O + C3 H3 + a-C3 H5 + s-C3 H5 + t-C3 H5 + n-C3 H7 + i-C3 H7 )

3.0
C3H8: φ = 1.64 Ti
1.5
Ti Tb
0.0
Rate [kmol/m /s]

Tu To
3

Increasing
−1.5 Strain
−3.0
C3H8: φ = 0.69
0.8
Ti Tb
0.0
Tu To
−0.8

−1.6
300 600 900 1200 1500 1800 2100
Temperature [K]

Fig. 7 The variation of net rates of radical production in lean and rich propane–air flames with
increasing rate of strain. See Fig. 6 for the radical pool species

consumption in the pre–heat zone (i.e. between Tu and Ti ) and radical production in
the main reaction zone (Ti – To ) for the expanding flame agrees qualitatively with the
influence of strain shown in Figs. 6 and 7. Increasing strain has an opposite effect on
the pre–heat zones of expanding and contracting flames. In the main reaction zone,
no simple explanation can be given with respect to the response of the rate of radical
production due to the narrowing of the reaction zone for contracting flames.
Markstein [52, 53] proposed a relationship, based on small perturbations that
permits linearised relations resulting in the following relationship between flame
front curvature and the laminar burning velocity,
Evaluation of Hazard Correlations for Hydrogen … 207

3.0
Tm Increasing
2.0 CH4 : φ = 0.79
Strain
1.0
Tu Ti To Tb
0.0
Rate [kmol/m /s]
3

−1.0
Outward
−2.0

2.0

0.0

−2.0
Inward
−4.0
300 600 900 1200 1500 1800 2100
Temperature [K]

Fig. 8 The variation of net rates of radical production for inwardly and outwardly propagating lean
methane–air flames. See Fig. 6 for the radical pool species. The variation with increasing rate of
strain for the outwardly and inwardly propagating flames is also indicated

8.0
Tm
C3H8 : Φ = 1.32
4.0
Tu Ti To Tb
0.0
Rate [kmol/m /s]

−4.0 Increasing
3

Outward Strain
−8.0
4.0

0.0

−4.0
Inward
−8.0
300 600 900 1200 1500 1800 2100
Temperature [K]

Fig. 9 The variation of net rates of radical production in inwardly and outwardly propagating rich
propane–air flames. See Fig. 6 for the radical pool species. The variation with increasing rate of
strain for the outwardly and inwardly propagating flames is also indicated
208 M. P. Meyer and R. P. Lindstedt
 
LM
u L = u∞
L 1+ (18)
r

where u L is the strained burning velocity, u ∞

L the unstrained burning velocity and L M
is the characteristic Markstein length, which is of the order of the flame thickness [53].
The Markstein length implicitly includes the effects of heat conduction, diffusion of
the reaction chain carriers and non–unity Lewis numbers. However, in more general
terms, L M describes not only the response of the burning velocity to the curvature
of the flame front, but the influence of stretch on the flame (e.g. Clavin [32], Clavin
and Williams [54]). Therefore, the above relationship can be reformulated as

1 dA
u∞
L − u L = LM = LM K . (19)
A dt

The Markstein length is a measure of the change in the burning velocity per unit
stretch [50]. More than one Markstein length can be defined and these do not vary
linearly with strain. Therefore, a “global” Markstein length L M can be expressed as
a linear combination of the separate lengths due to the rate of strain or stretch due to
curvature (Clavin and Joulin [55]),

u∞
L − u L = Ls K s + Lc K c (20)

where K s denotes the impact of strain and K c that of curvature (see Sect. 2).
For an outwardly propagating spherical flame the two Markstein lengths Ls and
Lc differ quite significantly. According to the definitions of K s and K c , the stretch
caused the flow field usually yields higher rates of strain than the curvature as the
gas velocity is, after the initial phase, much larger than the laminar burning velocity
due to the expansion of the burnt gases where u f = u g + u l at the leading edge. As
a first–order approximation it is, therefore, possible to ignore the contribution due to
curvature in Eq. (20) and to define a single Markstein length as shown in Eq. (21).

u∞
L − uL
LM = (21)
2u f (r )/r

Examples of the evolution of the flame speed and the laminar burning velocity for
methane/hydrogen/air mixtures are shown in Figs. 10 and 11. Lindstedt et al. [56]
used this approach where u f (r ) is the flame speed with the value for the unstretched
laminar burning velocity u ∞ L obtained by using the asymptotic value extrapolated to
zero rate of strain (cf. Vagelopoulos et al. [57]). The evolution of the corresponding
stoichiometric flame speeds and laminar burning velocities for propane/hydrogen/air
flames are shown in Figs. 12 and 13.
The selection of a reference surface, (e.g., iso-contour) for the determination of
the Markstein length is a key issue that has received relatively limited attention.
Fristrom [58, 59] used the upstream edge of the luminous zone (i.e. at the begin-
ning of the reaction zone). Bradley et al. [33] selected the 305 K isotherm, while
Evaluation of Hazard Correlations for Hydrogen … 209

4.0 20

Flame Speed [m/s] 3.0 15

60% CH4 /40% H2
40% CH4 /60% H2
20% CH4 /80% H2
2.0 100% H2 10

100% CH4
1.0 90% CH4 /10% H2 5
80% CH4 /20% H2
70% CH4 /30% H2

0 0
0 0.005 0.010 0.015 0.02 0.005 0.010 0.015 0.02
Radius [m]

Fig. 10 The flame speed (u f ) as a function of radius for stoichiometric methane/hydrogen/air

flames obtained by tracking the 305 K isotherm

0.6 3.2
Burning Velocity [m/s]

0.5 2.4
60% CH4 /40% H2
40% CH4 /60% H2
20% CH4 /80% H2
0.4 100% H2 1.6

100% CH4
0.3 90% CH4 /10% H2 0.8
80% CH4 /20% H2
70% CH4 /30% H2

0.2 0
0 0.005 0.010 0.015 0.02 0.005 0.010 0.015 0.02
Radius [m]

Fig. 11 The laminar burning velocity (u l ) as a function of radius for stoichiometric

methane/hydrogen/air flames obtained by tracking the 305 K isotherm

Dowdy et al. [60] used the location where the mass flux in a spherical flame corre-
sponds to that of a steady, one–dimensional planar flame. Figure 14 shows Markstein
lengths based on different isotherms as a function of the radius for a stoichiometric
methane–air flame. The values for the current first–order approximation depend on
the tracked iso-contour to the extent that the qualitative response to strain changes
sign. Hence, L M is not a unique physico–chemical parameter (Lipatnikov [61]).
Tables 1 and 2 show the Markstein lengths for different stoichiometries and isotherms
for methane and propane–air flames. The observation of increased burning rates at
higher strain rates for lean CH4 –air and rich C3 H8 –air flames are reflected by the
Markstein lengths which tend towards negative for these conditions. Qualitatively
210 M. P. Meyer and R. P. Lindstedt

4.0 20

Flame Speed [m/s] 3.0 15

30% C3H8 /70% H2

2.0 20% C3H8 /80% H2 10
10% C3H8 /90% H2
100% H2
100% C3H8
1.0 90% C3H8 /10% H2 5
70% C3H8 /30% H2
50% C3H8 /50% H2

0.0 0
0 0.005 0.010 0.015 0.02 0.005 0.010 0.015 0.02
Radius [m]

Fig. 12 The flame speed (u f ) as a function of radius for stoichiometric propane/hydrogen/air

flames obtained by tracking the 305 K isotherm

0.6 4
Burning Velocity [m/s]

0.5 3

30% C3H8 /70% H2

0.4 20% C3H8 /80% H2 2
10% C3H8 /90% H2
100% H2
100% C3H8
0.3 90% C3H8 /10% H2 1
70% C3H8 /30% H2
50% C3H8 /50% H2

0.2 0
0 0.005 0.010 0.015 0.02 0.005 0.010 0.015 0.02
Radius [m]

Fig. 13 The laminar burning velocity (u l ) as a function of radius for stoichiometric

propane/hydrogen/air flames obtained by tracking the 305 K isotherm

Fig. 14 Markstein lengths 0.8

(L M ) based on the tracking Ti
0.6 1000 K
of different isotherms shown
as a function of radius for a 500 K
0.4
LM [mm]

305 K
stoichiometric methane–air
flame at different positions 0.2
within the flame
0.0

−0.2

−0.4
0.00 0.01 0.02 0.03
Radius [m]
Evaluation of Hazard Correlations for Hydrogen … 211

Table 1 Computed Markstein lengths L M for ambient ( p = 101,325 Pa and T = 296 K) methane–air
flames at different equivalence ratios (φ) and isotherms
Mixture φ 305 K 500 K 600 K 1000 K Ti Tb
CH4 0.63 −0.46 −0.10 0.04 0.56 0.62 0.66
0.68 −0.47 −0.04 0.06 0.53 0.62 0.64
0.79 −0.30 0.01 0.08 0.41 0.53 0.56
1.00 −0.17 0.05 0.10 0.40 0.55 0.56
1.23 −0.14 0.05 0.12 0.53 0.85 0.85
1.41 0.06 0.06 0.17 1.38 2.01 2.23

Table 2 Computed Markstein lengths L M for ambient ( p = 101,325 Pa and T = 296 K) propane–air
flames at different equivalence ratios (φ) and isotherms
Mixture φ 305 K 500 K 1000 K Ti Tb
C3 H8 0.69 0.26 0.58 1.08 1.74 1.84
0.79 0.08 0.34 0.72 1.16 1.21
1.00 0.01 0.17 0.45 0.74 0.81
1.21 −0.03 0.11 0.30 0.59 0.71
1.32 −0.07 0.05 0.21 0.50 0.50
1.42 −0.14 −0.03 0.09 0.23 0.28
1.53 −0.21 −0.10 −0.08 0.01 0.01
1.64 −0.29 −0.20 −0.23 −0.28 −0.32

the trend in L M obtained from the different isotherms is similar when proceeding
from fuel rich to fuel lean mixtures for each fuel. Furthermore, Markstein lengths
based on the thermo–chemical properties of the flame, such as Ti and Tm , are quali-
tatively and quantitatively similar. However, such properties can not be determined
experimentally.
The calculated Markstein lengths for methane–air and propane–air flames are
compared in Figs. 15 and 16 with experimental data from Taylor [50], the cor-
rected values from Tseng et al. [62] and the indirect determination by Searby and
Quinard [63]. Taylor [50] used a schlieren technique to determine the flame front. As
can be seen in Figs. 15 and 16, the trends in the computed and measured Markstein
lengths are similar. Weinberg [64] pointed out that the location of the schlieren image
in a flame is about 200 K higher than previously thought. Bradley et al. [33] comment
that the isotherm obtained by schlieren corresponds better to a 450 K iso-surface. The
current computations confirm these observations. Given that flame properties derived
either on the basis of the internal structure (e.g. Ti ) or integral properties are not acces-
sible experimentally, the leading edge isotherm remains a natural choice. Significant
caution is, however, required when comparing experimental and computational data.
212 M. P. Meyer and R. P. Lindstedt

0.50

L [mm] 0.25

0.00

Taylor (1991)
−0.25 Tseng et al. (1993)
305 K Isotherm
500 K Isotherm
600 K Isotherm
−0.50
0.50 0.75 1.00 1.25 1.50 1.75
Equivalence Ratio [ − ]

Fig. 15 The computed Markstein lengths L M (lines), based on different isotherms, as a function
of equivalence ratio in comparison to experimental data (symbols) for methane–air flames

0.6

0.4
L [mm]

0.2

0.0 Taylor (1991)

Tseng et al. (1993)
Searby & Quinard (1990)
−0.2
305 K Isotherm
500 K Isotherm
−0.4
0.50 0.75 1.00 1.25 1.50 1.75
Equivalence Ratio [ − ]

Fig. 16 The computed Markstein lengths L M (lines), based on different isotherms, as a function
of equivalence ratio in comparison to experimental data (symbols) for propane–air flames

6 Minimum Ignition Energies and Quenching Distances

The response of incipient flame kernels to stretch is critical in the context of mini-
mum ignition energies and quenching distances. As shown above, there is a strong
dependency on the reference surface selected when the response is defined in terms
of the Markstein length. However, transient computations of the current kind can,
irrespective of such concerns, be used directly to estimate values when combined
with well-established theory. Transient laminar flames propagating in methane and
propane fuel blends with hydrogen were accordingly used to investigate the response
to strain and to estimate the minimum ignition energies and quenching distances using
Evaluation of Hazard Correlations for Hydrogen … 213

the approach of Williams [65] as outlined below. Fuels are categorized into different
hazard classes (e.g. [66]) with methane and propane Class IIA fuels with hydrogen
a Class IIC fuel.
The Energy Institute applies a hazard criterion that features a 30% hydrogen con-
centration limit for all types of hydrocarbon/hydrogen mixtures. Mixtures featuring
lower hydrogen concentrations are treated as ordinary fuels [67]. Hydrocarbon mix-
tures with hydrogen from an increasingly important class of process-related fuels
and their inherent flammability characteristics comprise a source of concern. The
increased explosion hazards associated with syngas-related fuels blends have been
explored by Hampp et al. [23] for a wide range of fuel compositions. Li et al.
[6, 9] have further shown that methane has a significant mitigating impact on the
strength of explosions in CH4 /H2 fuel blends. However, these studies do not address
aspects such as the minimum ignition energy or the quenching distance. The direct
determination of the latter through computational means is possible though excep-
tionally expensive. To enable screening of a large number of mixtures, an alternative
approach based on the extraction of relevant properties from transient spherical flame
simulations is explored below.
Expressions for the minimum spark energy and quenching distance have been
derived by Williams [65] on the basis that the amount of energy provided should be
sufficient for a temperature increase that permits the initiation of combustion and
that the ignited gas–filament thickness should exceed a critical limit so that flame
propagation becomes self–sustaining. Williams [65] showed that the initial spark
energy H can be expressed as,

H = (Aδb ) ρo c p (T∞ − Tu ) (22)

where A denotes the filament cross sectional area, Tu and T∞ denote the ambient
and maximum flame temperature respectively, and δb is the corresponding flame
thickness. The latter quantity may be expressed as,

μu 1 c p μu
δb = Pr = (23)
ρu u L Pr λu

where the flame front values of the properties c p , μu , ρu are used. The quenching
distance dmin can be assumed to be proportional to the flame thickness δb as suggested
by Mullins and Penner [69] and Williams [65],

dmin = aδb (24)

where a is a proportionality constant. Given the above definitions and that the cross
sectional area A must be at least equal to the square of the quenching distance (i.e.
A ≥ dmin
2
) the minimum initial spark energy H can be expressed as [65],

H = a 2 δb3 ρu c p (T∞ − Tu ) (25)

214 M. P. Meyer and R. P. Lindstedt

where the proportionality constant a is here assigned the midrange consensus value of
40, see Mullins and Penner [69] and Williams [65]. The maximum flame temperature
(T∞ ) is extracted from the simulations and the laminar burning velocity (u L ) can be
obtained either by tracing an iso–surface (e.g. T = 305 K) or via an integration
over the entire flame structure. As strain effects become large for small radii, the
tracking of the leading edge iso-contour presents a choice that provides an opportunity
for experimental verification. The current simulations permit δb to be evaluated via
Eq. (23) as a function of the rate of strain by determining the laminar burning velocity.
Minimum ignition energies and quenching distances are strongly dependent on the
fuel type, equivalence ratio and ambient conditions (e.g. Calcote et al. [68]), as shown
in Fig. 17. The cases considered here are stoichiometric methane/hydrogen mixtures
with the following compositions (in volume %) : 0/100, 10/90, 20/80, 30/70, 40/60,
50/50, 60/40, 70/30, 80/20, 90/10, and 100/0. Stoichiometric propane/hydrogen mix-
tures with the following compositions (in volume %): 0/100, 5/95, 10/90, 20/80,
30/70, 50/50, 70/30, 90/10, and 100/0 where also considered.
The initial conditions are set to standard atmospheric conditions (T = 298 K and
P = 101 kPa). Minimum ignition energies and quenching distances for the stoichio-
metric methane/hydrogen mixtures are shown in Figs. 18 and 19. The agreement
between the simulations and experimental data for cases featuring pure methane and
hydrogen mixtures is good. Calculated values, including uncertainties, are within
25% of the experimental determinations. It may also be noted that the distribution of
minimum spark ignition energies as a function of mixture composition shows a high
degree of similarity to the corresponding distribution of quenching distances. This
is not surprising, given the formulation of the minimum ignition energy through the
quenching distance. The trend is graphically illustrated in Fig. 20.
Figures 18, 19 and 20 permit a comparison of quenching distances and minimum
ignition energies obtained for the range of mixtures of CH4 /H2 with pure ethylene
as a typical Class IIB fuel. A similarity can thus be observed between the ignition
behavior of pure ethylene and a 20/80% methane/hydrogen mixture.
Computed minimum ignition energies and quenching distances for stoichiometric
propane/hydrogen mixtures are shown in Figs. 21 and 22. The computed minimum
ignition energy and quenching distance for the pure propane case are in excellent
agreement with experimental data.
A comparison with the stoichiometric methane/hydrogen mixtures (Figs. 18 and
19) reveals distinct differences. The quenching distance for propane/hydrogen
remains essentially constant over a wide range of propane content (30 < C3 H8 [%] <
100). For mixtures containing less than 30% propane the gradient of the quenching
distance and minimum ignition energy curves changes quite abruptly in contrast to
the more gradual change for the methane–containing mixtures. The observed differ-
ences imply a dependence on the type of hydrocarbon fuel considered.
The corresponding quenching distances and minimum ignition energies, for the
range of mixtures considered here, may also be seen in Fig. 23. Comparisons with sto-
ichiometric methane/hydrogen mixtures further highlight the dependency of quench-
ing distances and minimum ignition energies on the specific type of the hydrocarbon
component of the mixture.
Evaluation of Hazard Correlations for Hydrogen … 215

Fig. 17 Correlation of
minimum spark ignition Cyclohexane Isooctane
energy with quenching Di−isopropylether
distance adopted from 10 Di−isobutylene
Calcote et al. [68] n−Heptane
Vinyl acetate
1−Heptyne
Cyclohexene

Minimum Spark Ignition Energy [10 J]

Benzene

−4
2−Pentene Isobutane
n−Pentane
Diethylether
Methane
Methyl formate
Propionaldehyde
Propylene oxide Propane
1,3−Butadiene
Methyl−acetylene

Ethylene oxide
1.0
Ethylene
Hydrogen sulphide

Hydrogen Acetylene

Carbon disulfide

0.1
1.0 10
Minimal Quenching Distance [mm]

Fig. 18 Computed 10
minimum spark ignition
energies as a function for
stoichiometric CH4 /H2
H [10 J]

mixtures. The dash-dotted

−4

H(C2H4)
line indicates the value for 1.0
ethylene and the error bars
the variation obtained by
Exp. (H2, CH4)
extracting data at different
Calc. H2(φ=1.0),CH4(φ=1.0)
radii (see Sect. 7)
0.1
0 20 40 60 80 100
Hydrocarbon in Fuel [%]

7 Evaluation of Uncertainties

The computation of minimum ignition energies and quenching distances through the
extraction of relevant flame properties from simulations of transient spherical flames
developing from weak ignition kernels requires an assessment of intrinsic uncertain-
ties. Given the significant influence of strain on the evolution of the flame structure at
small radii, a source of uncertainty is related to the choice of a specific radius for the
extraction of properties of interest. The influence of the chosen radius on computed
216 M. P. Meyer and R. P. Lindstedt

Fig. 19 Computed 2.5

quenching distances as a
function for stoichiometric 2.0
CH4 /H2 mixtures. The

dmin [mm]
dash-dotted line indicates the 1.5 dmin(C2H4)
value for ethylene and the
error bars the variation 1.0
obtained by extracting data
at different radii (see Sect. 7) 0.5 Exp. (H2, CH4)
Calc. H2(φ=1.0),CH4(φ=1.0)
0.0
0 20 40 60 80 100
Hydrocarbon in Fuel [%]

Fig. 20 Correlation of
minimum spark ignition
energy with quenching
distance for CH4 /H2 10 100%/0%
mixtures (boxes). Abridged 90%/10%
experimental data (triangles)
Minimum Spark Ignition Energy [10 J]

80%/20%
−4

after Calcote et al. [68] CH4

70%/30%
60%/40%
50%/50%
40%/60%
30%/70%

1.0 20%/80%
C2H4

10%/90%

Calcote et al.(1952)
0%/100% CH4/H2
H2

0.1
1.0 10
Minimal Quenching Distance [mm]

laminar buring velocities for stoichiometric methane/hydrogen mixtures is shown

in Fig. 24. Associated uncertainty factors are shown in Table 3 in the form of nor-
malised differences (in %) between values obtained at r = 5 mm and r = 10 mm and
those obtained at r = 20 mm. The differences are non–monotonic with respect to the
mixture composition and a maximum deviation of about 45% for minimum ignition
energies and 18% for quenching distances is deduced for the 30/70% mixture. Similar
remarks can be made for the stoichiometric propane/hydrogen mixtures with respect
both to the trend and magnitude. Table 4 shows that the maximum observed differ-
ences were found for a 20/80% propane/hydrogen mixture and are around 10% for
Evaluation of Hazard Correlations for Hydrogen … 217

Fig. 21 Computed 10
minimum spark ignition
energies as a function for
stoichiometric C3 H8 /H2

H [10 J]
mixtures. The dash-dotted

−4
H(C2H4)
line indicates the value for 1.0
ethylene and the error bars
the variation obtained by
extracting data at different Exp.(H2, C3H8)
radii (see Sect. 7) Calc.H2(φ=1.0),C3H8(φ=1.0)
0.1
0 20 40 60 80 100
Hydrocarbon in Fuel [%]

Fig. 22 Computed 2.2

quenching distances as a
function for stoichiometric 1.9
C3 H8 /H2 mixtures. The
dmin [mm]
1.6
dash-dotted line indicates the
dmin(C2H4)
value for ethylene and the 1.3
error bars the variation
1.0
obtained by extracting data
at different radii (see Sect. 7) Exp.(H2, C3H8)
0.7
Calc.H2(φ=1.0),C3H8(φ=1.0)
0.4
0 20 40 60 80 100
Hydrocarbon in Fuel [%]

the quenching distance and 27% for the minimum ignition energy. This indicates that
the current approach is consistent. Furthermore, the maximum observed differences
are moderate and arguably within the uncertainties of experimental measurements,
which typically feature discrepancies of about a factor of two (e.g. Calcote et al.
[68]).
A further issue relates to the use of properties derived using data obtained by
tracking the 305 K isotherm. As noted above for Markstein lengths, properties can
be obtained at different isotherms or as an integral property of the flame. However,
it can be shown that the influence of the chosen radius on computed values based on
the integral burning velocity is significantly higher with differences up to 100%. This
can be attributed to the fact that the integral burning velocity at small radii is more
strongly affected by transient strain effects as the timescale associated with estab-
lishing a flame structure is more significant. Accordingly, integral properties deviate
more strongly due to the transient behavior at high rates of strain. The observed
discrepancies therefore arguably highlight the advantages of a flame front based
approach that can also be experimentally verified.
218 M. P. Meyer and R. P. Lindstedt

Fig. 23 Correlation of
minimum spark ignition
energy with quenching
distance for C3 H8 /H2 10
mixtures (boxes). Abridged
experimental data (triangles)

Minimum Spark Ignition Energy [10 J]

90%/10%

−4
after Calcote et al. [68] 70%/30%
100%/0%
50%/50%
30%/70% C3H8

20%/80%

10%/90%
1.0
5%/95% C2H4

Calcote et al.(1952)
0%/100% C3H8/H2
H2

0.1
1.0 10
Minimal Quenching Distance [mm]

Fig. 24 Extracted flame 3.0

front burning velocities for At 20 mm Radius
Burning Velocity [m/s]

stoichiometric CH4 /H2 At 10 mm Radius

mixtures at radii of 5, 10 and At 5 mm Radius
2.0
20 mm

1.0

0.0
0 20 40 60 80 100
CH4 in Fuel [%]
Evaluation of Hazard Correlations for Hydrogen … 219

Table 3 Normalised differences (in %) for quenching distance and minimum spark energy values,
based on the flame front burning velocity for stoichiometric CH4 /H2 mixtures, at radii R = 5 mm
and R = 10 mm against values based on R = 20 mm
Mixture Radius = 5 mm Radius = 10 mm
CH4 /H2 (dmin ) (H) (dmin ) (H)
100–0% 5.21 15.0 3.87 11.3
90–10% 6.40 18.0 4.90 14.0
80–20% 8.15 22.5 5.38 15.3
70–30% 9.40 25.6 5.49 15.6
60–40% 10.9 29.3 5.55 15.8
50–50% 11.7 31.2 5.87 16.6
40–60% 13.3 34.9 7.45 20.7
30–70% 18.0 44.8 12.56 33.2
20–80% 10.3 27.9 4.55 13.1
10–90% 6.48 18.2 3.25 9.44
0–100% 9.66 26.3 3.95 11.4

Table 4 Normalised differences (in %) of quenching distances and minimum spark energies, based
on flame front burning velocities for stoichiometric C3 H8 /H2 mixtures, at radii R = 5 mm and R =
10 mm against values based on R = 20 mm
Mixture Radius = 5 mm Radius = 10 mm
C3 H8 / H2 (dmin ) (H) (dmin ) (H)
100–0% 5.76 18.3 0.86 2.56
90–10% 3.93 12.3 1.73 5.10
70–30% 0.42 1.24 2.85 8.32
50–50% 4.99 14.2 3.41 9.90
30–70% 9.18 23.2 4.88 13.3
20–80% 9.96 27.0 5.18 14.75
10–90% 0.57 28.5 5.11 14.56
5–95% 9.13 25.0 4.06 11.67
0–100% 9.66 3.95 26.3 11.4

8 Conclusions

Computational simulations of spherical transient flames for various fuels and fuel
mixtures have been performed using comprehensive detailed chemistry and transport.
Changes in the flame speed and burning velocities induced by variations of the
hydrogen content and the hydrocarbon fuel type have been extensively investigated.
Results featuring flames without hydrogen enrichment are in good agreement with
experimental data. Similar comparisons for the whole range and types of mixtures
were not possible due to a lack of experimental data.
220 M. P. Meyer and R. P. Lindstedt

Given the importance of flow field strain on the development of ignition ker-
nels, the simulation of transient spherical methane and propane flames was also
used to examine the uniqueness of Markstein lengths. These lengths, defined on the
basis of flame iso–contours (i.e. reference surfaces), were investigated using detailed
thermo–chemical flame structures with properties extracted by tracking different
iso–contours. The latter included the ignition temperature where the rate of radical
production becomes positive when proceeding from the cold front and the temper-
ature at the back end of the flame where the rate of production of the radical pool
becomes negative. It was shown that the Lewis number of the deficient reactant at
the leading edge of the flame correlates the overall response of the flame towards
stretch and that the Markstein length can serve as a qualitative indicator in relation
to flame behavior. However, due to the complex behavior of flames such simplified
parameters do not permit a quantitative characterization. Moreover, transient flames
respond differently to strain than their steady-state counterparts due to the temporal
evolution of the flame structure. This is especially the case for ignition events due to
the time scale required to establish the radical pool.
As an application example, quenching distances and minimum ignition energies
were computed for stoichiometric methane/hydrogen/air and propane/hydrogen/air
fuel blends with the current calculation method combined with the theoretical anal-
ysis by Williams [65]. The approach provides uncertainty quantification and an effi-
cient methodology with the accuracy expected to be similar to experiments. However,
caution is required as in practise issues such as stratification as well as dilution may
become crucial.
The advantage of the current approach is that it is applicable under conditions (e.g.
high pressure and temperature) where experimental quantification is not feasible.
Furthermore, the increasing accuracy of detailed chemical mechanisms for complex
fuels creates an opportunity for the accurate theoretical determination of hazard-
related properties for emerging fuels and fuel blends. Finally, while ignition energies
and quenching distances form an important part of hazard assessments, it must be
recognized that the influence of laminar transport properties diminishes in turbulent
combusting flows. Hence, the data presented in the current work is complementary
to such investigations.

Acknowledgements The authors wish to gratefully acknowledge the financial support of the
Energy Institute’s Technical Partners for part of the current study along with Members of its Area
Classification Working Group that originally technically reviewed that work and the Members that
subsequently revalidated it. The technical advice and encouragement of Dr I.A.B Reid and Dr V.
Sakthitharan is also greatly appreciated.
Evaluation of Hazard Correlations for Hydrogen … 221

References

1. Lieuwen TC, Yang V, Yetter R (eds) (2009) Synthesis gas combustion: fundamentals and
applications CRC Press, Boca Raton ISBN 978-1-4200-8535-8
2. Milosavljevic VD, Persson M, Lindstedt RP, Váos EM (2003) Evaluation of the thermoacoustic
behaviour of an industrial gas turbine combustion chamber with a second moment method.
Cranfield University Press, pp 72–77
3. Korb B, Kawauchi S, Wachtmeister G (2016) Fuel 164:410
4. Milosavljevic VD, Lindstedt RP, Cornwell MD, Gutmark EJ, Váos EM (2006) In Roy G, Yu
KH, Whitelaw JH, Witton JJ (2006) Advances in combustion and noise control. Cranfield
University Press, pp 149–165
5. Lieuwen TC, McDonell V, Santavicca D, Sattelmayer T (2008) Combust Sci Technol
180(6):1169
6. Li T, Hampp F, Lindstedt RP (2018) Process Safety Environmental Protection 116:663
7. Law CK, Jomaas G, Bechtold JK (2005) Proc Combust Inst 30(1):159
8. Milosavljevic VD, Cornwell MD, Gutmark EJ, Lindstedt RP, Váos EM (2003) Methods and
mechanisms for the stabilization of lean partially premixed low-temperature flames. Cranfield
University Press, pp 78–83
9. Li T, Hampp F, Lindstedt RP (2019) Comb Flame 203:105
10. Simatos P, Hampp F, Lindstedt RP (2020) in Innovations. In: Gupta A, De A, Aggarwal S,
Kushari A, Runchal A (eds) Sustainable energy and cleaner environment. Springer Verlag,
Singapore, Green energy and technology, pp 333–356
11. Lin YC, Daniele S, Jansohn P, Boulouchos K (2013) J Eng Gas Turbines Power 135
12. Burke EM, Güthe F, Monaghan RFD (2016) Proceeding ASME Turbo Expo, Seoul, GT2016-
57804
13. Wang J, Zhang M, Xie Y, Huang Z, Kudo T, Kobayashi H (2013) Exp Thermal Fluid Sci 50:90
14. Daniele S, Jansohn P, Mantzaras J, Boulouchos K (2011) Proc Comb Inst 33(2):2937
15. Muppala SPR, Nakahara M, Aluri NK, Kido H, Wen JX, Papalexandris MV (2009) Int J
Hydrogen Energy 34:9258
16. Venkateswaran P, Marshall AD, Seitzman JM, Lieuwen TC (2013) J Eng Gas Turb Power 136
17. Liu CC, Shy SS, Chiu CW, Peng MW, Chung HJ (2011) Int J Hydrogen Energy 36:8595
18. Bray KNC (1990) Proc Roy Soc A - Math Phy 431(1882):315
19. Driscoll JF (2008) Prog Energy Combust 34(1):91
20. Shepherd IG, Cheng RK (2001) Combust Flame 127(3):2066
21. Lawn CJ, Schefer RW (2006) Combust Flame 146(1):180
22. Goh KHH, Geipel P, Lindstedt RP (2014) Combust Flame 161:2419
23. Hampp F, Goh KHH, Lindstedt RP (2020) Process Safety Environ Protect 143:66
24. Dixon-Lewis G (1990) Proc Combust Inst 23:305
25. Miller JA, Bowman CT (1989) Prog Energy Combust Sci 15:287
26. Warnatz J (1980) Proc Combust Inst 18:369
27. Poinsot T (1996) Proc Combust Inst 26:219
28. Lindstedt RP (1993) Proc Anglo - German Combustion Symposium. The British Section of
the Combustion Institute, p 372
29. Hampp F, Lindstedt RP (2017) Combust Flame 182:248
30. Hampp F, Lindstedt RP (2017) Proc Comb Inst 36(2):1911
31. Buckmaster J (1979) Acta Astronautica 6:741
32. Clavin P (1985) Prog Energy Combust Sci 11:1
33. Bradley D, Gaskell P, Gu X (1996) Combust Flame 104:176
34. Ghoniem AF, Soteriou MC, Knio OM, Cetegen B (1992) Proc Combust Inst 24:223
35. Stahl G, Warnatz J (1991) Combust Flame 85:285
36. Yoo CS, Im H (2005) Proc Combust Inst 30:349
37. Malik NA, Lindstedt RP (2010) Combust Sci Technol 182:1171
38. Malik NA, Lindstedt RP (2012) Combust Sci Technol 184:1799
222 M. P. Meyer and R. P. Lindstedt

39. Ciardiello R, de Oliveira P, Skiba A, Mastorakos E, Allison P (2020) Combust Flame 221:296
40. Karlovitz B, Dennington D, Knappschafe D, Wells F (1953) Proc Combust Inst 4:613
41. Matalon M (1983) Combust Sci Technol 31:169
42. Candel S, Poinsot T (1990) Combust Sci Technol 70:1
43. Frankel M, Sivashinsky G (1983) Combust Sci Technol 31:131
44. Law CK (1988) Proc Combust Inst 22:1381
45. Sun CJ, Law CK (2000) Combust Flame 121:236
46. Lindstedt RP, Sakthitharan V (1999) Parallel processing and direct simulation of transient
premixed laminar flames with detailed chemical kinetics. In: Allen RJ, Guest MF, Simpson
AD, Henty AS, Nicole DA (eds) High-performance computing. Kluwer Academic / Plenum
Publishers, New York, pp 417–428
47. Meyer MP (2001) The application of detailed and systematically reduced chemistry to transient
laminar flames. Ph.D. thesis, Imperial College of Science, Technology and Medicine
48. Lindstedt RP, Maurice LQ (1995) Combust Sci Technol 107:317
49. Chapman S, Cowling TG (1970) The mathematical theory of non-uniform gases. Cambridge
University Press, Cambridge
50. Taylor S (1991) Burning velocity and the influence of flame stretch. Ph.D. thesis, University
of Leeds
51. Dixon-Lewis G (1996) Proc Roy Soc London, Part A 452:1857
52. Markstein G (1951) J Aero Sci 18:199
53. Markstein G (1964) Non-steady flame propagation. Pergamon Press
54. Clavin P, Williams FA (1982) J Fluid Mech 116:251
55. Clavin P, Joulin G (1989) In: Borghi R, Murthy SNB (eds) Lecture notes in engineering:
turbulent reactive flows. Springer-Verlag, Berlin, p 213
56. Lindstedt RP, Meyer MP, Sakthitharan V (1998) EUROTHERM seminar No. 61 on detailed
studies of combustion phenomena, p D52
57. Vagelopoulos CM, Egolfopoulos FN, Law CK (1994) Proc Combust Inst 25:1341
58. Fristrom R (1965) Phys Fluids 8:273
59. Fristrom R, Westenberg A (1965) Flame structure. McGraw Hill, New York
60. Dowdy DR, Smith DB, Taylor SC (1990) Proc Combust Inst 23:325
61. Lipatnikov AN (1996) Combust Sci Technol 119:131
62. Tseng L, Ismail M, Faeth G (1993) Combust Flame 95:410
63. Searby G, Quinard J (1990) Combust Flame 82:298
64. Weinberg FJ (1955) Fuel 34:S84
65. Williams FA (1985) Combustion Theory. The Benjamin/Cummings Publ, Company Inc
66. British Standards Institution, Code of Practice for the Selection, Installation and Maintenance of
Electrical Apparatus for use in Potentially Explosive Atmospheres. Part 1: Basic Requirements
for all Parts of the Code. Technical Report BS 5345: Part 1 (1976)
67. Energy Institute, EI model code for safe practice Part 15: Area classification for installations
handling flammable fluids. Tech. Rep. 4th Edition, 61 New Cavendish Street, London W1G
7AR, UK (2015)
68. Calcote H, Gregory C, Barnett C, Gilmer R (1952) Ind Eng Chem 44:2656
69. Mullins BP, Penner SS (1959) Explosions, detonations, flammability and ignition. Pergamon
Press, New York
Experimental Investigation of Turbulent
Flow/Flame Structure of Double Swirler
Burner

Dhanalakshmi Sellan and Saravanan Balusamy

Abstract To develop efficient combustors under various operating conditions, a

deeper understanding of the effects of swirl flow mechanisms on gas turbine combus-
tion is essential. The double swirl flow arrangement has an advantage over the
single swirl flow by proving better mixing and flame anchoring due to the increased
shearing action between the two swirl flows. The flow field of the open-type, double
swirler burner at various Reynolds numbers is analyzed using the 2D particle image
velocimetry (PIV) technique. The spatial distribution of OH* and CH* chemilu-
minescence is analyzed for one premixed (φ= 0.75) and three stratified LPG/air
flames. Medium swirlers are used for velocity measurements, while two different
axial swirlers, defined as low and medium swirlers with geometric swirl numbers
0.5 and 0.8, respectively, are used for chemiluminescence studies. The analysis of
non-reacting flow field results shows that the increase in the Reynolds number of outer
swirl increases the radial component of the velocity; thus, the swirl flow spreads in
the radial direction. As a result, turbulence parameters such as Reynolds stress and
turbulence intensity are also dominant in the radial direction. As a consequence of
this phenomenon, in the worst case, the spread of the flame front and further local
quenching could occur. In comparison, the increase in internal swirl flow increases
the axial components of the velocity and the flow field becomes axially elongated.
The inner recirculation zone shifts downstream of the flow field, suggesting that
the reaction zone could be moved to the downstream under the reacting condition.
The acquired OH*/CH* chemiluminescence signals indicate that the premixed LPG
flames are elongated while the upstream stratified mixture conditions lead to more
compact flames that are appropriate for gas turbine applications. Nonetheless, for
medium swirler cases, OH*/CH* chemiluminescence is much more compact in form
than low swirl. Intensity variation is independent of the flame strain of propane;
thus, the equivalence ratio calculated from the OH*/CH* intensity ratio has helped
to understand the interaction between flame and turbulence at various operating
conditions.

D. Sellan · S. Balusamy (B)

Department of Mechanical and Aerospace Engineering, Indian Institute of Technology
Hyderabad, Kandi, Sangareddy, Telangana 502285, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 223
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_10
224 D. Sellan andS. Balusamy

Keywords Double swirler burner · Reynolds stress · Turbulent intensity ·

Stratified flame · Chemiluminescence · OH*/CH* · PIV · LPG/air flames

Nomenclature

φ Equivalence ratio
φ1 Equivalence ratio at an inner annulus
φ2 Equivalence ratio at an outer annulus
Rei Reynolds number at an inner annulus
Reo Reynolds number at an outer annulus
ui Bulk velocity at an inner annulus
uo Bulk velocity at an outer annulus
uc Co-flow bulk velocity
τR Reynolds stress
TI Turbulent intensity
I Chemiluminescence intensity

Abbreviations

LPG Liquefied petroleum gas

PIV Particle image velocimetry
CCD Charge-coupled device
ICCD Intensified charge-coupled device
SR Stratification ratio

1 Introduction

In practical combustion devices, stabilization of flame in the lean premixed condi-

tion is of critical and complex due to the low heat release rate. The burner featured
with a swirler is capable of stabilizing the flame by enhancing the mixing between
incoming fresh fuel/air mixture and hot reactants [1, 2]. This is more pronounced
in the double swirl flow arrangement [3–5]. The study of turbulent swirl flow under
isothermal conditions is important before understanding the case of reaction flow and
helps to optimize burner design and validate numerical data [6]. Many combustion
devices, such as gasoline and gas turbine engines, operate in a stratified condition to
reduce pollution and improve combustion performance [7–10]. Stratified combustion
is a special type of partially premixed mode of combustion where the flame prop-
agates through the non-uniform equivalence ratio gradient field [11–14]. Past few
Experimental Investigation of Turbulent Flow/Flame Structure … 225

decades the stratified combustion research is of interest due to its practical relevance
[11, 13, 15–17].
During the combustion process, many reactions take place at the flame front where
highly localized heat release occurs. Identifying these reaction zones at the flame is
essential for understanding the flame structure and behavior at various operating
conditions. The chemiluminescence of the OH* and CH* radicals formed during
the chemical reaction provides important information on the structure and nature of
the reaction zone of the flame. Therefore, they are used as markers for identifying
the reaction zone. Chemiluminescence imaging is one of the reliable methods of
non-invasive optical diagnostics techniques used to capture the intermediate free
radicals in the reaction zone. Chemically exited OH and CH radicals emit photons at
a specific wavelength when returning to their ground level, and this light is captured
by intensified charge-coupled device (ICCD). For a perfectly premixed flame, the
OH* and CH* chemiluminescence intensity depends linearly on the mass flow rate
of the air/fuel mixture and exponentially on the global equivalence ratio in case of
globally uniform equivalence ratio distribution [18]. Therefore, the local equivalence
ratio of the flow field can be measured from OH*/CH* chemiluminescence ratio [19].
As the chemiluminescence emission gives the reaction zone location and intensity
is directly proportional to the species concentration, it can be used as an indicator
for lean blow-off [20]. The flame stability was understood from the flame structure,
and heat release rate measurements were obtained from chemiluminescence images
of premixed and stratified propane flames [21, 22]. Because premixed flames have
uniformly distributed species concentration, there are direct methods to measure
local equivalence ratio and heat release rate. There are many studies conducted to
investigate the properties of premixed flames using chemiluminescence techniques
on various fuels [23–27].
In the current study, the effect of Reynolds number on the turbulent structure under
non-reacting conditions is being investigated using 2D particle image velocimetry
(PIV). Also, the spatial distributions of OH* and CH* are captured by the chemilu-
minescence technique for turbulent premixed and stratified liquefied petroleum gas
(LPG)/air flames. An open-type double swriler burner with bluff body configuration
is used in this study. The current study uses LPG due to its availability, combustion
performance, and low emission levels [28–32]. Many research groups are focusing
on introducing LPG into gas turbine applications [30, 33, 34]. The non-reacting flow
PIV data shows the influence of Reynolds number at inner and outer annuli alter
the flow structure which helps to predict the flow structure and flame stability. The
overall flame structure and stability is understood by OH* and CH* chemilumines-
cence imaging. The stratified flames are short and compact than the premixed flames
and insensitive to stratification ratio (SR).
226 D. Sellan andS. Balusamy

2 Experimental Methods

This section describes the swirl-stabilized burner unit and diagnostics techniques
used in this study.

2.1 Swirl-Stabilized Burner Unit

A swirl-stabilized axisymmetric burner is designed and fabricated to study the

premixed and stratified flames. Figure 1a shows the main components of the burner
in different colors. The burner mainly consists of three annuli: inner (aqua), outer
(orange), and co-flow (green) and two mechanical swirlers, one (red) located in the
inner annulus (23 mm diameter) and the other (purple) in the outer annulus (43.5 mm
diameter) and a bluff body. The inner annulus provides a premixed fuel–air mixture
flow at an equivalence ratio φ1 , which is equal or different than the equivalence ratio
at an outer annulus φ2 as shown in the cross-sectional view of the burner (Fig. 1b).
The mixing of fuel–air is taking place before entering the inner and outer annuli,
inside the feed line to the burner.
The premixed fuel–air was made to pass through the bottom of the plenum (gray)
to the inner annulus. Similarly, the premixed flow through the outer annulus enters
four sideways of the plenum by 90° with reference to the burner axis. The height of the
annuli (inner-333 mm, outer-297, co-flow-250) is chosen in such a way that it provides
fully developed flow to the burner exit. The entrainment of ambient air is prevented by
using a cylindrical co-flow (161 mm diameter) where only filtered air is supplied. The
air filtering at the co-flow is achieved by using two 3D printed honeycomb structures

Fig. 1 Swirl-stabilized axisymmetric burner. a cut-sectional view b cross-sectional view of the

burner exit
Experimental Investigation of Turbulent Flow/Flame Structure … 227

and two metallic perforated plates. The airflow at the cylindrical co-flow is initially
allowed to pass through the two honeycomb structures then through the perforated
plates. A collar (metallic silver) type of structure is used to hold the co-flow cylinder
where the air is allowed to pass through four sideways, ensuring symmetrical inflow
of air.
To achieve different swirling environments, two different swirlers are fabricated,
such as the low swirler (L) and the medium swirler (M). The mechanical swirlers are
fabricated using a 3D laser metal printing process. Also, this configuration allows
us to independently investigate the effect of swirl number (S) and the Reynolds
number (Re). Along with the swirlers, flame stabilization is established by bluff
body configuration which is shown in Fig. 1b. A bluff body with 6 mm radius placed
on inner swirler such a way that it does not affect the flow of inner swirler.
The current burner design concept is similar to the Cambridge/Sandia stratified
swirl burner (CSB), and it has been demonstrated that flames can be stabilized over
a range of flow rates, stratification ratios, and swirl numbers [35, 36]. The swirl
introduced by CSB is basically an aerodynamic swirl where both the swirl number
and Reynolds number are dependents on each other. Whereas in this burner both inner
and outer annulus flows can be run with the axial flow or on various swirl intensities
(low to strong) by using different sets of axial swirlers in both annuli which allows
independently to study the effect of swirl and Reynolds number.

2.2 2D-Particle Image Velocimetry (PIV)

PIV is a non-intrusive technique used to measure the instantaneous velocity field

of flow. 2D-PIV measures the two components (radial and axial) instantaneous
velocity field of flow. Figure 2 shows the schematic diagram of the two-dimensional
(2D) PIV setup along with flow configuration. PIV setup consists of a dual-cavity
laser (Litron, 532 nm), laser sheet arrangement, and a charged-couple device (CCD)
camera equipped with a Nikkor lens and an optical filter (532 ± 10 nm). The laser
system emits two laser pulses with a given delay time between them, and the optical
unit expands the beam into a laser sheet, which illuminates the flow field. The micron-
level particles are illuminated by a laser sheet, and the laser sheet generated using
532 nm wavelength, 135 mJ Litron double pulsed Nd: YAG laser system. A CCD
camera, which is synchronized with the laser system, acquires the image pairs at
a given time interval t. The captured images are post-processed using 4G Insight
software (TSI).
The overall flow setup along with the burner is shown in Fig. 2. Mass flow
controllers (MFC-Alicat MCR-D series) are used to regulate the flow rates of the
inner and outer annulus air with an accuracy of ±1% of full scale. The airflow through
three annuli is seeded with micrometric vegetable oil particles for non-reacting flow
PIV measurements using particle seeders. The oil particle generators are fabricated
in-house. In these particle generators, Laskin nozzle design is used to efficiently
produce oil particles at the micron level. Seeding of these particles is obtained by
228 D. Sellan andS. Balusamy

Fig. 2 Schematic of a 2D-PIV experimental setup for non-reacting flow study

using compressed air through a seed particle generator and uniform seed density is
achieved by adjusting the needle valve that controls the airflow.

2.3 Chemiluminescence Imaging

The experimental setup of measurement of simultaneous OH*/CH* chemilumines-

cence is shown in Fig. 3. The chemiluminescence imaging is a line-of-sight technique,
which captures the signal coming from the chemical excitation. The light emitted
from the chemically excited OH* and CH* radicals is recorded by an intensified
charge-coupled device (ICCD). Two ICCD (PI-MAX4) cameras equipped with UV
lens of focal length 105 mm, providing a 16-bit image of 1024 × 1024 pixels and
band-pass filters of 431 nm (CH*) and 308 nm (OH*), are used to capture the simul-
taneous CH* and OH* chemiluminescence images. Images were captured using the
light-field software (Princeton Instruments). Proper calibration was performed, and
background corrections were done during the post-processing of images.
Experimental Investigation of Turbulent Flow/Flame Structure … 229

Fig. 3 Schematic of chemiluminescence imaging of OH* and CH* for reacting flow study

2.4 Operating Conditions

The 2D PIV for non-reacting flow study was carried out for six experimental cases,
which are given in Table 1. The outer Reynolds number (Reo ) and co-flow Reynolds
number (Rec ) are maintained constant for three different inner Reynolds numbers
(Rei ). Similarly, the inner Reynolds number (Rei ) and co-flow Reynolds number
(Rec ) are maintained constant for three different outer Reynolds number (Reo ).
Experiments were performed to investigate the one premixed and three stratified
LPG/air flames. To understand the flame structure and stability, four test cases were
studied using low and medium swirlers with geometric swirl number of 0.5 and 0.8,
respectively. The test matrix is shown in Table 2. The outer swirl bulk velocity is
maintained to be twice that of the inner swirl bulk velocity to promote mixing while
the co-flow bulk velocity is kept constant at 0.4 m/s and thermal input power is also
kept constant. The stratification ratio (SR) is defined as the ratio of inner equivalence
ratio (φ1 ) to the outer equivalence ratio (φ2 ).

Table 1 Test matrix for PIV study

Case No Bulk velocity (m/s) Reynolds number
Inner (ui ) Outer (uo ) Co-flow (uc ) Inner (Rei ) Outer (Reo ) Co-flow (Rec )
N1 5 6 0.4 3577 6669 2826
N2 10 6 0.4 7154 6669 2826
N3 15 6 0.4 10,731 6669 2826
N4 7.5 4 0.4 5309 4402 2826
N5 7.5 6 0.4 5309 6581 2826
N6 7.5 8 0.4 5309 8804 2826
230 D. Sellan andS. Balusamy

Table 2 Test matrix for chemiluminescence study

Case No Inner swirl flow Outer swirl flow Stratification
Equivalence Bulk velocity Equivalence Bulk velocity ratio SR = φφ21
ratio (φ1 ) m/s ratio (φ2 ) m/s
R1 0.75 7.40 0.75 15.27 1.0
R2 1.00 9.17 0.50 18.84 2.0
R3 1.20 9.20 0.40 18.81 3.0
R4 1.40 9.21 0.35 18.89 4.0

3 Results and Discussion

3.1 Velocity Measurements of Non-reacting Turbulent Flow

Using 2D-PIV

To understand the isothermal flow structures, an unconfined axial swirl burner with
the various Reynolds numbers (Re) (Table 1) of inner and outer flow is used under
isothermal condition (298 K). Two components velocity field and turbulent parame-
ters such as Reynolds stress and turbulent intensity are determined using 2D particle
image velocimetry (PIV). For constant outer swirl and co-flow Reynolds number, the
effect of changing inner swirl Reynolds number is studied. Similarly, for constant
inner and co-flow Reynolds numbers, the effect of changing the outer Reynolds
number also studied. For each case, 200 images are captured and post-processing
was performed using 4G Insight software (TSI). The mean of 200 images is plotted
to represent the velocity field and turbulence parameters such as Reynolds stress and
turbulent intensity. ‘x’ indicates the axial co-ordinate, and ‘r’ indicates the radial
co-ordinates for all the results provided in this section.
Radial velocity contour for six cases is shown in Fig. 4. The top row shows that
increasing the inner swirl Reynolds number, the radial component of the velocity field
is diminishing because the adverse pressure gradient dominates in axial direction.
The radial spread of the flow is prevented by the outer swirl and co-flow when the
inner swirl Reynolds number is increased. Therefore, in the case of reacting flow, the
flame may elongate axially. The effect of outer swirl flow plays an essential role in
the shear action between the inner swirl flow and co-flow conditions, which shows
the increase in the radial flow field than the axial flow field. From case N4 to N6,
as increasing outer swirl Reynolds number the radial velocity increases, also strong
vortex break down was observed which is shown in the bottom row of Fig. 4.
Figure 5 shows the axial velocity vector field for various inner Reynolds number
and the center recirculation zone (CRZ) in the velocity field which is shown clearly.
It is observed that for the first case the axial component of velocity is radially spread,
whereas, in the third case the radial spread is decreased and axial velocity is increased.
The CRZ becomes very strong when the inner swirl Reynolds number is increased.
The axial velocity component is increased while increasing the inner swirl Reynolds
Experimental Investigation of Turbulent Flow/Flame Structure … 231

Fig. 4 Mean radial velocity components. Top row: Variation of inner flow Reynolds number.
Bottom row: Variation of outer flow Reynolds number

number. As stated before the adverse pressure gradient becomes non-favorable for
radial component; therefore, axial components are increasing. In the case of N4, the
inner axial Reynolds number dominates the outer flow; therefore, the outer spreading
is less, and the magnitude of axial velocity is less due to less overall bulk velocity
(inner bulk velocity + outer bulk velocity). But the flow structure becomes wider
for the case N5 than N4 and N6. In the case of N6, due to an increase in outer axial
bulk velocity, the mean axial velocity starts dominating the outward pull. Therefore
in the case of N6, the flow structure has a lesser radial spread compared to case N5
but more than case N4. Due to these flow structures, the flame length will be affected
in reacting conditions. When considering the same flow condition (case N5 and case
N6) for the reacting case, the flame length may reduce, because there may be an
effect of thermal expansion radially.
The streamline plots are very useful in identifying the recirculation zones in the
flow field. Figure 6 shows the streamline superimposed with axial velocity compo-
nents and the inner recirculation zones (IRZ) is clearly identified. From cases N1
to N3, the adverse pressure gradient at the center of the swirl flow generates IRZ
and when an increase in inner swirl Reynolds number, the IRZ is moving close to
the centerline of the burner exit while shifting toward downstream. When consid-
ering the same flow conditions with the reacting case, the reaction zone may shift
downstream. Due to an increase in inner Reynolds number, the fuel/air may not mix
232 D. Sellan andS. Balusamy

Fig. 5 Mean axial velocity component. Top row: Variation of inner flow Reynolds number. Bottom
row: Variation of outer flow Reynolds number

properly at the exit of the burner and flame front may elongate axially. The IRZ
becomes strong in the third case compared to the first case. When the inner Reynolds
number dominates over the outer (N4), the IRZ is very strong and close to the centre-
line of the burner exit. Ìn case N5, when increasing the outer axial bulk velocity, the
inner recirculation zones are moved outward and axially upward. On the other hand,
when outer swirl Reynolds number dominates on the inner swirl Reynolds number,
IRZ becomes weak and moves radially outwards. Also, case N5 shows compact inner
recirculation zones, whereas N3 has elongated IRZ.
Figure 7 shows the mean Reynolds stress (τR ) contour for each case. Reynolds
stress is the momentum induced by the turbulent flux that acts as a medium to
transfer energy from mean flow to turbulence. From the case, N1 to N3 the increase
in Reynolds stress in axial direction indicates that the energy transfer to the turbu-
lence more axially; therefore, the shear between inner and outer swirl flow is more
at upstream where combustion is intensive due to velocity gradients. The mean
Reynolds stress (τR ) contour for various outer swirl flow conditions with constant
inner swirl flow and co-flow is shown in the bottom row of Fig. 7. An increase in outer
axial bulk velocity is contributing to Reynolds stress increment at radially outward as
well as axially upward. When the Reynolds stress is high, the energy transfer to the
turbulence will also be high, and thus, it promotes the fuel/air mixing for complete
combustion.
Experimental Investigation of Turbulent Flow/Flame Structure … 233

Fig. 6 Streamlines are superimposed with mean axial velocity. Top row: Variation of inner flow
Reynolds number. Bottom row: Variation of outer flow Reynolds number

The turbulent intensity is one of the major parameters to characterize the swirl flow
under both non-reacting and reacting conditions. Figure 8 shows the total turbulent
intensity plot of the flow filed. The top row shows the increase in inner swirl Reynolds
number, increases the turbulent intensity axially. The increase in inner swirl Reynolds
number increases all the flow parameters axially; therefore, the flame may become
elongated axially and the reaction zone is like to happen at downstream of the flow.
The turbulent intensity contour for various outer swirl flow conditions with constant
inner Reynolds number and co-flow Reynolds number is shown in the bottom row
of Fig. 8. With an increase in outer flow, turbulent intensity also increases in both
the axial and radial direction. As the Reynolds stress transfers the energy to the
turbulence, the increase in Reynolds stress leads to an increase in the turbulence
level. However, increasing the swirl Reynolds number at the flow field is not always
favorable to enhance flame stabilization like an increase in turbulence level at lean
fuel conditions may lead to blow-off. Therefore, it is essential to study the flow
characteristics under various operating conditions such as different swirl number,
mass flow rate, and equivalence ratio in order to identify the optimized operating
conditions.
234 D. Sellan andS. Balusamy

Fig. 7 Mean Reynolds stress for different operating conditions. Top row: Variation of inner flow
Reynolds number. Bottom row: Variation of outer flow Reynolds number

3.2 OH*/CH* Chemiluminescence Study of LPG Flames

Flame structures and behavior can be investigated by observing intermediate radicals

in the reaction zone, and there is a higher local heat release rate in the reaction
zone that occurs at the flame front. Intermediate radicals such as OH* and CH*
are good indicators for identifying the flame front. Chemiluminescence imaging
is one of the reliable methods of non-intrusive optical diagnostics used to study
the concentration of intermediate free radicals in the reaction zone. In this work,
the spatial distribution of OH*/CH* using chemiluminescence is studied for the
premixed and stratified liquefied petroleum gas (LPG) flames using an open-type
double swirl-stabilized burner. Two different axial swirler, named as low swirler and
medium swirler with geometric swirl numbers of 0.5 and 0.8, respectively, are used in
this study. The spatial distribution of OH* and CH* chemiluminescence images are
captured simultaneously for one premixed and three stratification conditions, which
is shown in Table 2.
To understand the flame–turbulence interaction at premixed and stratification
conditions, instantaneous CH* chemiluminescence images are presented in Fig. 9. It
is observed that in premixed condition, the reacting wake is spread axially upward.
As we approach the lean stratification condition, the reacting wake structure is more
compact and the flame front is closer to the burner exit. The premixed LPG flames
Experimental Investigation of Turbulent Flow/Flame Structure … 235

Fig. 8 Mean turbulent intensity plots for various cases. Top row: Variation of inner flow Reynolds
number. Bottom row: Variation of outer flow Reynolds number

Fig. 9 Instantaneous CH* chemiluminescence images for various stratification ratios at low swirl
flow condition (Swirl number 0.5)

are elongated, whereas the stratified conditions flame becomes more compact at the
low swirling condition, as shown in Fig. 10. Indeed, the outer equivalence ratio in
stratified conditions is below the lower flammable limits, but the combustion is still
happening due to the back support from the rich inner flow region. For the medium
swirl configuration, OH* and CH* contour plots show much more compact flame
shapes than low swirl configuration as we approach the lean stratified condition.
The increased swirl strength promotes the mixing of burned gases with incoming
LPG/air, leading to faster combustion thus compact in shape. As we look closer, the
236 D. Sellan andS. Balusamy

Fig. 10 Mean OH * and CH * chemiluminescence contours for different stratification conditions.

Top row: low swirl. Bottom row: medium swirl. The left side is CH* plot, and the right side is the
corresponding OH*

spatial distribution of OH* and CH* is very similar between all three stratified cases
due to the enhanced mixing, which makes the flame becomes more robust to the local
stratifications while the reaction zone stays closer to the exit of the burner.
Previous researchers have shown that intensity variation is independent of flame
strain for propane flame [37], and similar assumptions can be made for the LPG
flame as well. The flame equivalence ratio is calculated based on Eq. (1).
  
OH∗
− 0.527
φ = 0.6 − ln CH
∗ 0.171 (1)
1.135

where OH*/CH* ratio is calculated from the instantaneous images captured simul-
taneously using two ICCD cameras. The mean contour plots of the equivalence ratio
for different stratification conditions are shown in Fig. 11.
From the equivalence ratio contour plots shown in Fig. 11, we can see the
combined effect of mixture stratification and swirl strength on flame shapes. As
expected, the premixed case (R1) for both swirl configurations shows the more spatial
distribution of combustion zones as compared to that of stratified cases (R2-R4).
Among the stratified cases, the primary reaction zone is more compact and closer
to the exit of the burner for the medium swirl condition because of faster mixing
of streams, which balances the slow down of chemical kinetics with the increase of
mixture stratification. The corresponding probability distribution of the equivalence
ratio for various test conditions is shown in Fig. 12. The premixed cases show the
lower distribution as expected, whereas the low swirl is wider than the medium swirl.
A similar trend is observed between the stratified cases between low and medium
Experimental Investigation of Turbulent Flow/Flame Structure … 237

Fig. 11 Equivalence ratio contour plot based on the ratio of OH*/CH* for different stratification
conditions. Top row: low swirl. Bottom row: medium swirl

Fig. 12 Probability density function of the equivalence ratio is calculated for different stratification
ratios. Top row: Low swirl. Bottom row: Medium swirl

swirl conditions. Indeed, the PDF pattern in almost the same for all three medium
swirls stratified cases as compared to the low swirl cases. This experimental investi-
gation on turbulent stratified flames using the chemiluminescence technique shows
the potential of using the equivalence ratio derived from the OH*/CH* ratio could
help us to understand the flame–turbulence interaction for LPG flames under various
operating conditions.
238 D. Sellan andS. Balusamy

4 Conclusion

Experiments are performed to understand the effect of the Reynolds number on the
non-reacting flow structure and the effect of the swirl on the premixed/stratified flame
structures. An open-type swirl-stabilized burner with double swirler arrangement is
designed, and necessary flow systems are developed. 2D PIV technique is used
to measure the flow field, and the chemiluminescence imaging technique is used to
capture the flame structures. A series of experiments are performed for premixed and
stratified flames. The analysis of non-reacting flow fields showed that the operating
parameters play a critical role in the flow structure, such as varying the inner and
outer flow rate leads to different recirculation zone size and strength. The increase
in the inner flow rate elongates the IRZ while the increase in the outer flow rate
widens the IRZ for the MM swirler combination. Similar behavior is observed for
other swirler combinations.
The spatial distribution of OH* and CH* is observed from simultaneously
acquired OH* and CH* chemiluminescence images under low swirler and medium
swirler combinations. Premixed and stratified turbulent flame structures are observed
from the instantaneous CH* images. From the intensity of the OH* images, the flame
front is located. In the stratified case, the flame front appears compact and closer to
the exit of the burner than to the premixed case. Enhanced swirl effect facilitates
the mixing of LPG and air; thus, the structure of turbulent LPG/air flames under
premixed and stratified conditions is more compact for medium swirler combina-
tions than low swirler combinations, and the lean blow-off limit is delayed. The
equivalence ratio plots derived from the OH*/CH* ratio help to understand the effect
of stratification and turbulence mixing on the flame characteristics of the LPG under
various operating conditions.

Acknowledgements This research is supported by the Science and Engineering Research Board
(SERB) of India through grant No. ECR/2015/000343. The optical diagnostic setup is supported by
the Fund for Improvement of S&T Infrastructure (FIST) of India through grant No.: SR/FST/ETI-
397/2015/(C). The author is grateful to the Ministry of Human Resource Development (MHRD) of
India for providing the Ph.D. fellowship.

References

1. Lilley DG (1977) Swirl flows in combustion: a review. AIAA J 15(8):1063–1078. https://doi.

org/10.2514/3.60756
2. Ahmed SA, Nejad AS (1992) Swirl effects on confined flows in axisymmetric geometries. J
Propuls Power 8(2):339–345. https://doi.org/10.2514/3.23483
3. Chong CT, Hochgreb S (2015) Measurements of non-reacting and reacting flow fields of a
liquid swirl flame burner. Chinese J Mech Eng 28(2):394–401. https://doi.org/10.3901/CJME.
2015.0109.011
4. Elbaz AM, Moneib HA, Shebil KM, Roberts WL (2019) Low NOX–LPG staged combustion
double swirl flames. Renew Energy 138(X):303–315. https://doi.org/10.1016/j.renene.2019.
01.070
Experimental Investigation of Turbulent Flow/Flame Structure … 239

5. Sellan D, Murugan R, Balusamy S (2019) Experimental and numerical analysis of turbulent

swirl flow structure in double swirler burner. In: ASME 2019 gas turbine india conference,
GTINDIA 2019, vol 2. https://doi.org/10.1115/GTINDIA2019-2739
6. Vanierschot M, Van den Bulck E (2008) Influence of swirl on the initial merging zone of a
turbulent annular jet. Phys Fluids 20(10). https://doi.org/10.1063/1.2992191
7. Zhao F, Lai MC, Harrington DL (1999) Automotive spark-ignited direct-injection gasoline
engines. Prog Energy Combust Sci 25(5):437–562
8. Candel S (2002) Combustion dynamics and control: progress and challenges. Proc Combust
Inst 29(1):1–28
9. Huang Y, Yang V (2009) Dynamics and stability of lean-premixed swirl-stabilized combustion.
Prog Energy Combust Sci 35(4):293–364
10. Karagiannaki C et al (2009) Performance evaluation of a model swirl burner under premixed
or stratified inlet mixture conditions. Combust Flame 28(1):1–8
11. Balusamy S, Cessou A, Lecordier B (2009) Direct experimental determination of laminar
burning velocity using a new PIV approach pp 8–11
12. Balusamy S, CessouA,Lecordier B (2009) Measurement of laminar burning velocity–A new
PIV approach. 4th Eur Combust Meet Vienna 3:1–6
13. Balusamy S, Cessou A, Lecordier B (2014) Laminar propagation of lean premixed flames
ignited in stratified mixture. Combust Flame 161(2):427–437
14. Masri AR (2015) Partial premixing and stratification in turbulent flames. Proc Combust Inst
35(2):1115–1136
15. Lauvergne R, Egolfopoulos FN (2000) Unsteady response of C3H8/Air laminar premixed
flames submitted to mixture composition oscillations. Proc Combust Inst 28(2):1841–1850
16. Pasquier N, Lecordier B, Trinité M, Cessou A (2007) An experimental investigation of flame
propagation through a turbulent stratified mixture. Proc Combust Inst 31(1):1567–1574
17. Balusamy S, Cessou A, Lecordier B (2011) Direct measurement of local instantaneous laminar
burning velocity by a new PIV algorithm. Exp Fluids 50(4):1109–1121
18. Guyot D, Guethe F, Schuermans B, Lacarelle A, Paschereit CO (2010) CH*/OH* chemilumi-
nescence response of an atmospheric premixed flame under varying operating conditions. Proc
ASME Turbo Expo 2(PARTS A AND B):933–944
19. Kojima J, Ikeda Y, Nakajima T (2000) Spatially resolved measurement of OH*, CH*, and C2*
chemiluminescence in the reaction zone of laminar methane/air premixed flames. Combust Inst
28(2):1757–1764
20. Muruganandam TM, Kim B, Olsen R,Patel M, Romig B, Seitzman JM (2003) Chemilumines-
cence based sensors for turbine engines. In: 39th AIAA/ASME/SAE/ASEE Jt Propuls Conf
Exhib July 2003
21. Aleiferis PG, Hardalupas Y, Taylor AMKP, Ishii K, Urata Y (2004) Flame chemiluminescence
studies of cyclic combustion variations and air-to-fuel ratio of the reacting mixture in a lean-burn
stratified-charge spark-ignition engine. Combust Flame 136(1–2):72–90
22. Karagiannaki C, Paterakis G, Souflas K, Dogkas E, Koutmos P (2015) Performance evaluation
of a model swirl burner under premixed or stratified inlet mixture conditions. J Energy Eng
141(2):1–9
23. Taamallah S, Shanbhogue SJ, Ghoniem AF (2016) Turbulent flame stabilization modes
in premixed swirl combustion: Physical mechanism and Karlovitz number-based criterion.
Combust Flame 166:19–33
24. Jourdaine P, Mirat C, Caudal J, Lo A, Schuller T (2017) A comparison between the stabilization
of premixed swirling CO2 -diluted methane oxy-flames and methane/air flames. Fuel 201:156–
164
25. Nori VN, Scitzman JM (2009) CH* chemiluminescence modeling for combustion diagnostics.
Proc Combust Inst 32(1):895–903
26. Foley C et al (2009) Performance evaluation of a model swirl burner under premixed or stratified
inlet mixture conditions. Combust Flame 28(1):1–8
27. Chaudhari RR, Sahu RP, Ghosh S, Mukhopadhyay A, Sen S (2013) Flame color as a lean
blowout predictor. Int J Spray Combust Dyn 5(1):49–65
240 D. Sellan andS. Balusamy

28. Miao J, Leung CW, Cheung CS, Huang Z, Jin W (2016) Science direct effect of H2 addition
on OH distribution of LPG/air circumferential inverse diffusion flame. Int J Hydrogen Energy
41(22):9653–9663
29. Mishra DP (2004) Emission studies of impinging premixed flames, vol 83, pp 1743–1748
30. Shaddix CR, Molina A (2009) Particle imaging of ignition and devolatilization of pulverized
coal during oxy-fuel combustion. In: 32nd international symposium on combustion, vol 32 II,
pp 2091–2098
31. Makmool U, Jugjai S, Tia S (2013) Structures and performances of laminar impinging multiple
premixed LPG-air flames. Fuel 112:254–262
32. Choi GH, Chung YJ, Bin Han S, Performance and emissions characteristics of a hydrogen
enriched LPG internal combustion engine at 1400 rpm. Int J Hydrogen Energy, 30(1):77–82
33. Enagi II, Al- KA, Zainal ZA (2017) Combustion chamber design and performance for micro
gas turbine application. Fuel Process Technol 166:258–268
34. Elbaz AM, Roberts WL (2019) Stability and structure of inverse swirl diffusion flames with
weak to strong swirl. Exp Therm Fluid Sci 112(November 2019):109989
35. Zhou R, Balusamy S, Sweeney MS, Barlow RS, Hochgreb S, Flow field measurements of a
series of turbulent premixed and stratified methane/air flames. Combust Flame 160(10):2017–
2028
36. Sweeney MS, Hochgreb S, Dunn MJ, Barlow RS (2012) The structure of turbulent stratified
and premixed methane/air flames I: Non-swirling flows. Combust Flame 159(9):2896–2911
37. Orain M, Hardalupas Y (2010) Influence du combustible sur la mesure de richesse par
chimiluminescence dans les flammes prémélangées. Comptes Rendus Mec 338(5):241–254
Flame Root Dynamics and Their Role
in the Stabilisation of Lifted Flames

James C. Massey , Zhi X. Chen , and Nedunchezhian Swaminathan

Abstract This chapter aims to provide a broad overview on the importance of lifted
flames in turbulent flows with and without swirl in practical applications for energy
production and propulsion. The stabilisation of lifted flames is governed by numerous
physical processes that control the behaviour of the flame root or the leading edge.
The flame lift-off height, which is the physical distance of the flame root above the
burner, is a quantity of practical importance and obtaining accurate predictions using
computational fluid dynamics (CFD) is challenging. The large eddy simulation (LES)
paradigm has proven to be successful in accurately capturing the flame lift-off height
in turbulent reacting flows in simple and complex geometries of practical relevance.
The objective of this chapter is to present an overview of the stabilisation mechanisms
that have been observed in simulations with relevance to modern applications. An
overview of the LES framework for turbulent reacting flows and different sub-grid
combustion models are briefly discussed with a focus on the unstrained flamelet
combustion model that is used in the case studies presented here. The first part of
the simulation results focuses on the canonical jet flame configuration, where it is
seen that the lift-off height is sensitive to the jet velocity and the fuel used. The
second part of the results focuses on a more complex configuration, which is a gas
turbine model combustor with two radial swirlers. The flame root in swirling flows
is typically more robust, but failed ignition and local extinction cause the flame root
to oscillate and leads to flame lift-off. The amplitude of this oscillation can lead to
the occurrence of thermo-acoustic oscillations and flame blow-off under appropriate
conditions. The discussion is presented on a physical basis and the observations are
compared with measurements. The chapter concludes by summarising the role and
importance of modelling flame stabilisation.

Keywords Computational fluid dynamics · Gas turbiness · Lifted flames ·

Partially premixed flames · Swirling flows · Turbulent reacting flows

J. C. Massey (B) · Z. X. Chen · N. Swaminathan

Department of Engineering, University of Cambridge, Trumpington Street,
Cambridge CB2 1PZ, UK
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 241
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_11
242 J. C. Massey et al.

1 Introduction

Chemical reactions in combustion processes are accompanied by a flame, but the type
of flame that exists is dependent on the prescribed boundary conditions. Premixed
combustion is present when the fuel and oxidiser are mixed prior to ignition, whereas
non-premixed combustion is present when the fuel and oxidiser enter through sep-
arate inlets, and mixing and burning occur simultaneously [1]. Both combustion
processes can exist in laminar and turbulent flows. In practical combustion systems,
such as industrial gas turbines and aero engines, turbulent flows are expected and
this causes a degree of unavoidable mixing between the fuel and oxidiser before
ignition. Therefore, compositionally inhomogeneous mixtures are present and this
leads to partially premixed combustion [2]. Flames in large industrial boilers and
gas turbines are stabilised using a non-premixed jet configuration and swirling flows
are often used, since swirling flows can extend the range of flame stability in the
combustion system [3]. Due to the presence of partially premixed combustion and
ubiquitous high Reynolds number turbulent flows in practical systems, understanding
the underlying mechanisms that lead to flame stabilisation is important.
The jet flame is a simple configuration where partially premixed combustion is
present. Fuel enters through a nozzle into quiescent air at atmospheric conditions
and hence, compositionally inhomogeneous mixtures that are within and beyond the
flammability limits are present. For a low fuel jet exit velocity, the flame is able to
stabilise by remaining attached to the nozzle rim. When the jet velocity is increased,
the flame is locally quenched due to turbulence induced strain and the flame stabilises
at a downstream location where reactants mix with hot combustion products [2].
These are referred to as lifted flames and these are typically preferred to protect the
burner nozzle, such as in commercial boilers [4]. Combustion of stratified mixtures
in gas turbine engines and diesel engines has elements of lifted flame attributes and
hence, laboratory-scale lifted flames remain of high interest to the community [5].
A typical lifted flame configuration is shown in Fig. 1, and the mean flame base
and stoichiometric mixture fraction contour are shown in red and grey respectively.
The flame base is the stabilisation location of the flame and its position downstream
of the fuel nozzle exit is referred to as the flame lift-off height H . The mean flame
base is usually marked with a specified temperature value in numerical simulations
and OH-PLIF measurements in experimental studies. The stabilisation mechanisms
of lifted flames have been hypothesised in detail in the study by Wohl et al. [6] and
the review by Pitts [7] addressed the main flame stabilisation mechanisms that have
been identified, followed by more recent reviews by Lyons [5] and Lawn [8]. In the
study by Vanquickenborne and van Tiggelen [9], it was suggested that the local gas
mixture at the flame base is premixed and hence, the flame stabilisation is due to the
balance between the local flow velocity and the local burning velocity of the mixture
upstream of the flame base along the stoichiometric mixture fraction contour. The
burning velocity is known as the turbulent flame speed sT , which is a function of
unstrained laminar flame speed s L0 and the local turbulence statistical characteristics,
such as the integral length scale and velocity fluctuations. The study by Kalghatgi
Flame Root Dynamics and Their Role … 243

Fig. 1 Schematic of a lifted

flame, which is adapted from
the schematic shown by
Lawn [8]

[10] investigated the turbulent flame speed by using different fuels and jet velocities.
An equation for sT was deduced using simple theoretical arguments and experimental
data, and it was suggested that the ratio sT /s L0 is proportional to the square root of
the local turbulent Reynolds number ReT . It was also shown that the lift-off height
was linearly dependent on the fuel nozzle jet velocity U j .
Peters and Williams [11] argued that the major flame stabilisation is related to
flamelet quenching at the flame base region due to high scalar dissipation rates and
flame stretching. However, the lack of experimental data in the flame base region
could not ascertain this theory and the study by Pitts [7] highlights that this theory does
not consider the partial premixing of the fuel and oxidiser upstream of the flame. The
study by Müller et al. [12] introduced a model for partially premixed combustion that
includes both premixed flame propagation and flame quenching by using premixed
and non-premixed flamelets respectively. It has since been proposed that high levels
of partial premixing lead to the presence of triple flames (or sometimes referred to as
edge flames), where the flame base consists of a lean and rich premixed flame and a
trailing non-premixed flame between the two premixed flame branches, which is due
to excess oxidiser and fuel from the lean and rich branches respectively. These flame
structures have been studied and reviewed in significant detail in earlier studies [2,
12–15], but the flame stabilisation mechanisms have been widely hypothesised.
244 J. C. Massey et al.

Another explanation for the flame stabilisation mechanism in lifted flames is due
to mixing of fuel with hot combustion products in the large-scale eddies formed in the
jet flow. The review by Pitts [7] observed using species concentration measurements
that the fuel–air mixture composition at the flame base is highly intermittent. For short
time periods, mixtures within the flammability limit are present that allows the flame
base to propagate upstream, otherwise only fuel or air is present, which prevents the
flame base from propagating upstream. Therefore, large-scale neighbouring eddies
need to supply hot combustion products to the edges of the jet, along with fresh
air and fuel–air mixtures within the flammability range. These large-scale structures
are important, since there is no large recirculation zone present that can supply
hot combustion products to the edges of the jet [16]. Correlations have also been
developed and proposed using experimental observations to determine the blow-off
velocity for lifted flames [16, 17]. Lifted flames are highly prone to blow-off, which
is treated as the complete extinction of the flame and must be avoided to ensure
efficient and smoothly operating combustion systems.
Due to the inherent instability of lifted flames, the risks of blow-off and their
influences on system operation, swirling flows are used to improve flame stability
because the flame root becomes more robust and improves system stability over
a broad range of operating conditions [18]. A schematic of a non-premixed swirl
flame configuration is shown in Fig. 2. An inner shear layer (ISL) containing high
vorticity is formed from the trailing edge of the burner nozzle. The vortex structures
are convected further downstream with the incoming jet stream until vortex break-
down occurs, which leads to the formation of an inner recirculation zone (IRZ). The
incoming fuel is entrained into the IRZ and a flame will form along the ISL once
the fuel–air mixture is ignited. The IRZ acts as a continuous ignition source since
hot products are entrained into the recirculation zone and this continuous supply
of heat and radical species makes the flame root more robust [19]. In addition, a
hydrodynamic instability that is referred to as a precessing vortex core (PVC) can
be present in flows with larger swirl numbers (S > 0.6–0.7) [3, 20]. The instanta-
neous structure of the PVC is shown in Fig. 8 for a gas turbine model combustor
simulation, along with the IRZ and the flame. The PVC influences the stability of
the flame and can lead to significant changes in the combustion chamber, since the
PVC can enhance fuel–air mixing in the flame root region or lead to the presence
of thermo-acoustic instabilities and flame lift-off, as shown in a gas turbine model
combustor experiment [21, 22].
Swirling flow configurations are beneficial, since they allow quicker mixing of
fuel and air, which is conducive for partially premixed combustion and this leads to
compact combustion systems [3]. This helps achieving fuel-lean combustion in order
to meet the ever-increasing norms on emissions of carbon dioxide (CO2 ) and nitro-
gen oxides (NOx ). Lean combustion is also beneficial as shorter flames are present
and the size of combustion chambers can be reduced. However, operating under lean
conditions can make combustion systems prone to combustion instabilities, which
arise due to close coupling among acoustics, combustion and swirling flow dynam-
ics [23–25]. These instabilities can become detrimental to lean systems through failed
ignition, flashback, autoignition and flame blow-off. Moreover, it is near impossible
Flame Root Dynamics and Their Role … 245

Fig. 2 Schematic of a swirl-stabilised flame, which is adapted from the schematic shown by
Feikema et al. [18]

to achieve perfect premixing because the mixing time scales are usually larger than
the residence time scales for the reactants and hence, combustion is typically in the
partially premixed regime [26]. Therefore, understanding the stabilisation mecha-
nisms for partially premixed flames is imperative for the design of such combustion
systems.
The design of combustion systems is increasingly reliant on computational fluid
dynamics (CFD), which is now on par with experimental tools for research and devel-
opment. There are three commonly used CFD approaches, namely direct numerical
simulation (DNS), large eddy simulation (LES) and Reynolds-averaged Navier–
Stokes (RANS) simulation, in the order of descending computational cost. With
the state-of-the-art computing technology, the use of DNS for practical combustion
systems is not yet possible. This approach is typically used in simplified configura-
tions (often referred to as “flame-in-a-box”) for combustion model development and
validation. On the other hand, steady RANS and unsteady RANS (URANS) tech-
niques have been developed as computationally inexpensive alternatives for complex
geometries in practical devices. However, these approaches consider ensemble- or
time-averaged conservation equations for mass, momentum, energy and scalars. In
addition, they resolve only one length scale and one time scale, which are typically
246 J. C. Massey et al.

(integral scales) larger than the scales of interest to investigate flame behaviour. This
means that RANS simulations cannot capture highly unsteady phenomena, such as
the small-scale flame stabilisation mechanisms. Hence, there has been an emergence
of LES approaches for modelling turbulent combustion over the last two decades,
since such approaches can capture the transient phenomena exhibited by turbulent
flows and flames.
The device-dependent energy containing large scales and its cascade to smaller
scales, down to a cut-off scale , are resolved and captured in LES by using fil-
tered conservation equations. The contributions of device-independent or universal
small scales, typically known as the sub-grid scale (SGS) range, are modelled using
closure models for the unresolved residual stresses and scalar fluxes [27]. There-
fore, the resulting computational cost is largely reduced compared to that for DNS.
The LES approach is particularly suitable for physical problems where the tran-
sient behaviours are predominantly driven by the resolvable large scales that often
vary from one case to another (e.g. geometry, operating conditions, etc.), whereas
SGS models accounting for the effect of non-resolvable small scales can be derived
prior to the CFD simulations. The turbulent combustion processes occurring in many
practical devices fall into this category, especially for those exhibiting highly tran-
sient characteristics, such as the lifted flame dynamics and stabilisation, which is
of interest here. This makes LES the ideal candidate for the present study. Since
combustion is mostly at the SGS level, where the typical chemical length scales
(< O(10−4 ) m) are by orders of magnitude smaller than the device-dependent turbu-
lent flow scales (> O(10−2 ) m), accurate and robust modelling is required to capture
the SGS turbulence–combustion interactions.
The last two decades has seen numerous SGS models developed for turbulent
combustion and a detailed review of these models is beyond the scope of this work
(see previous studies for extensive reviews [28–30]). A presumed joint probabil-
ity density function (PDF) approach belonging to the statistical category is used to
model the SGS turbulent combustion here. This approach has been tested and val-
idated for various laboratory and practical configurations with stable and unstable
conditions using RANS and LES methodologies (see the review by Chen et al. [31]).
The aim of this chapter is to present an overview of simulations using this mod-
elling approach for turbulent partially premixed combustion with a focus on flame
stabilisation mechanism in various burner configurations.
This chapter is organised as follows. The combustion modelling framework is
described in Sect. 2 along with a brief summary of the other models that have been
used for the particular target cases presented in later sections. The canonical non-
swirling jet flame is first presented in Sect. 3, which focuses on the sensitivity of the
lift-off height to the fuel jet velocity for different fuels. A more complex configuration
featuring a gas turbine model combustor with swirling flows is investigated in Sect. 4
with detailed analyses on flame root dynamics and its stabilisation mechanisms under
various conditions. Concluding remarks are presented in Sect. 5.
Flame Root Dynamics and Their Role … 247

2 Turbulent Reacting Flow Modelling

The modelling methodology is introduced in two parts in this section. The first part
presents the general LES governing equations for turbulent reacting flow simulations
and the SGS modelling for unclosed terms in those equations is summarised. In the
second part, the chosen combustion modelling approach involving additional scalar
transport equations and closure models is described in appropriate detail, in order to
facilitate a better interpretation of the simulation results presented in Sects. 3 and 4.

2.1 Large Eddy Simulation

Simulations with the LES methodology are becoming increasingly popular, due to the
continuous advancement of computational resources. However, the computational
cost depends on the amount of scales that are inherently resolved in the LES, i.e. the
more scales that are resolved, the higher the cost. This is related to the chosen filter
size, which is proportional to the local cell volume and is estimated as  ≈ V 1/3 . If
 is larger, i.e. fewer scales are resolved by the numerical grid, then the SGS model
needs to account for the effects of more unresolved scales. As an overall guidance,
Pope [27] suggested that 80% of the turbulent kinetic energy should be resolved for
an acceptable LES and this is commonly practised.
The transport equations required for a turbulent flow field include the conversa-
tion of mass, momentum and energy. When applying the LES filtering to these equa-
tions, several unclosed terms appear and these require SGS modelling. To reduce the
number of unclosed terms in variable density flows, a density-weighting procedure
is used, which is known as Favre filtering. For a physical variable ϕ, its density-
weighted form is written as  ϕ = ρϕ/ρ [32]. Applying this procedure to the filtered
transport equations for mass, momentum and the thermochemical enthalpy (sensible
plus formation) leads to

∂ρ  
 = 0,
+ ∇ · ρU (1)
∂t
DU
ρ = −∇ p + ∇ · τ − ∇ · τ R , (2)
Dt
Dh Dp  
ρ = + ∇ · ρα∇h − ∇ · τ hR , (3)
Dt Dt

where D/Dt = ∂/∂t + U  · ∇ is the material derivative and τ is the molecular

viscous stress tensor. The filtered density is obtained using the equation of state
 0T
ρ = p M/  with M and 0 being the mixture molecular weight and the univer-
sal gas constant respectively. The molecular transport terms, which are the second
terms on the right-hand side of Eqs. (2) and (3), are expected to be small for high
Reynolds number flows, but these terms cannot be neglected in the LES framework,
248 J. C. Massey et al.

as they are relevant at the SGS level and for numerical stability in the simulation. The
pressure-heating source term D p/Dt in Eq. (3) is included if compressibility effects
are expected to be significant in certain cases, e.g. thermo-acoustic instabilities, and
if there is bulk compression, e.g. internal combustion engines.
The residual stress tensor τ R ≡ (ρ U U − ρU  U)
 in Eq. (2) is decomposed into its
isotropic and anisotropic parts respectively as τ R = 23 ρ ksgs I + τ r , where the SGS
kinetic energy  ksgs is equal to half of the isotropic part of the residual stress tensor
and is typically absorbed into the filtered pressure term [27]. The anisotropic resid-
ual stress tensor is typically modelled using the Boussinesq eddy viscosity hypoth-
esis [33, 34]:  
1 
τ r = −2 ρνT  S− ∇·U  I , (4)
3

where   + (∇ U)
S = 21 [∇ U   ] is the symmetric strain rate tensor and νT is the SGS
eddy viscosity [27].
The residual scalar flux terms are commonly modelled using a gradient transport
hypothesis, such as the last term in Eq. (3). This is analogous to Fick’s law of diffusion,
where the transport of the scalar flux is in the direction of the scalar gradient −∇ϕ
and is written as τ ϕR = ρ Uϕ  − ρ U  ϕ = −ρ(νT /σϕ )∇ ϕ , where the dimensionless
number σϕ is a parameter that is physically consistent with the transported variable;
for Eq. (3), this dimensionless number is the turbulent Prandtl number σϕ ≡ Pr T ,
which is taken to be 0.7 in common practice.
The eddy viscosity νT requires closure and its model should be carefully chosen.
Numerous models for the eddy viscosity have been developed, which include alge-
braic (zero equation) models or solving additional transport equations to obtain SGS
turbulence information. The classical Smagorinsky model [35] is a simple algebraic
formulation that is still widely used, and it has an empirical model constant Cs , which
can take different values for different cases. Proposed modifications to dynamically
compute Cs using resolved quantities have been developed [36–38] to overcome this
issue. However, this approach (or any dynamic procedures) should be used with care,
as it is highly susceptible to numerical errors leading to unphysical values of Cs , e.g.
negative effective (molecular plus turbulent) viscosity. Other popular algebraic mod-
els include the eddy viscosity model proposed by Vreman [39] and the wall-adapting
local eddy (WALE) model [40], and both approaches give improved predictions of νT
naturally approaching zero in near-wall regions. The σ -model proposed by Nicoud
et al. [41] uses a single-value decomposition on the velocity gradient tensor as a basis
for determining νT , which also allows for νT to scale with the wall distance cubed.
The k-equation model [42–44] involves directly obtaining the SGS turbulent kinetic
energy  ksgs by using its transport equation. This approach is commonly referred to
as a one equation model, since it is assumed that the length scale for  ksgs is  and
therefore, no transport equation is required for its dissipation. For the problems inves-
tigated here, the non-dynamic Smagorinsky model is chosen, due to its adequacy and
simplicity and also that the region of interest is sufficiently far from the walls.
Flame Root Dynamics and Their Role … 249

The framework presented thus far is sufficient for variable density non-reacting
turbulent flows, while additional transport equations and closure models are required
to capture combustion phenomena, which are described next.

2.2 Combustion Modelling

For the LES of turbulent combustion, the typical grid size is usually larger the laminar
flame thickness δ L0 and thus, combustion occurs entirely at the SGS level. Develop-
ing SGS models to accurately capture the small-scale turbulence–combustion inter-
actions has been the focal point of turbulent combustion modelling research. The
state-of-the-art SGS combustion models can be classified into three categories: phe-
nomenological, geometric and statistical models [28]. Phenomenological models
evaluate the full thermochemistry on the fly and compute the reaction rate using a
chemical mechanism. Such models include the artificially thickened flame model [45,
46] and the linear eddy model [47, 48]. These models often require a fine grid res-
olution, which lead to high computational costs. Geometric models are inspired and
derived from the morphology of turbulent flames, such as the flame surface den-
sity [49–51] and the G-equation (or level-set) [32, 52] models. Statistical models
typically involve a multi-variant probability density function (PDF), where ideally
its shape can be obtained from solving the PDF transport equation [53–56]. This
approach is sophisticated as the reaction rate source term naturally appears in a
closed form, but requires very high computational costs and a model for scalar mix-
ing physics at the small scales, which are known as micro-mixing models. Other sta-
tistical models have been proposed to reduce the computational burden by assigning
the shape of the PDF and these models are referred to as presumed PDF approaches.
These include the conditional moment closure [57–59] and multiple mapping con-
ditioning [60, 61] models, which are still computationally demanding due to the on
the fly chemistry evaluation for the conditional reaction rates. The presumed PDF
approach further reduces the cost by assuming that the thermochemical states of
laminar flames do not change in a turbulent environment, i.e. the turbulent flame
is seen as an ensemble of fluctuating laminar flames. As such, the chemistry can
be tabulated prior to the CFD simulation as a function of a number of control vari-
ables [62, 63]. This approach is attractive for practical applications because of its
high computational efficiency and yet adequate accuracy. Derived from the early
attempts by Bradley et al. [64] using a mixedness-reactedness concept, many model
variants have been developed including the BML [65, 66], FGM [67], FPI [68] and
FPV [69] models.
In this work, a presumed joint PDF approach with unstrained premixed flamelets
is used for SGS combustion modelling. This approach revisits flamelet based models
and the development of these models has demonstrated that the variance of a reactive
scalar should be transported, and its dissipation rate should include the influences
from turbulent–combustion interactions. This modelling framework is hence referred
to as a flamelets revised for physical consistencies (FlaRe) approach. The LES–FlaRe
250 J. C. Massey et al.

modelling framework described here has been tested and validated for laboratory-
scale premixed [70–72] and partially premixed flames [73, 74], and also for gas
turbine combustors [75–80]. In this approach, a mixture fraction ξ is used to describe
the turbulent mixing process and a reaction progress variable c is used to describe the
combustion progress. The mixture fraction is calculated using the definition proposed
by Bilger [81]. The normalised progress variable is defined as c = ψ/ψ eq , where
ψ = YCO + YCO2 and the superscript ‘eq’ denotes the chemical equilibrium value
for a given mixture fraction. Transporting the progress variable requires care, since
unclosed terms arise when deriving the progress variable transport equation from first
principles [82]. Scaled and unscaled progress variable approaches have been tested
by Chen et al. [74] , where it is observed that the scaled progress variable approach
performs better in capturing local extinction. The transport equations for the Favre-
filtered mixture fraction and progress variable, as well as their SGS variances, are
written as
 
Dξ νT
ρ  ∇ · ρD + ρ ∇ξ , (5)
Dt ScT
2  
Dσξ,sgs νT νT
∇
2
ρ  ∇ · ρD + ρ ∇σξ,sgs − 2 ρ
2
χξ,sgs + 2 ρ ξ , (6)
Dt ScT ScT
 
D
c νT
ρ  ∇ · ρD + ρ ∇c + ω̇∗ , (7)
Dt ScT
2  
Dσc,sgs νT νT 2
ρ  ∇ · ρD + ρ ∇σc,sgs
2
− 2 ρχc,sgs + 2 ρ ∇c (8)
Dt ScT ScT
 
+ 2 c ω̇∗ − 
c ω̇∗ ,

where D is the scalar molecular diffusivity and ScT = 0.7 is the turbulent Schmidt
number. The sub-grid scalar dissipation rate (SDR) terms χ ξ,sgs and χ
c,sgs are unclosed
and require modelling. The sub-grid SDR for ξ is modelled using a linear relaxation
model χξ,sgs = Cξ (νT /2 )σξ,sgs
2
[52]. This algebraic expression is suitable for pas-
sive scalars, since their gradients are generated through turbulence. However, the
effects of chemical reactions, thermal expansion and the multi-scale turbulence–
combustion interactions are important for the SDR of reactive scalars [83]. Hence,
the term χ
c,sgs requires a more sophisticated closure model. The algebraic expression
proposed by Dunstan et al. [84], which is part of the LES–FlaRe model development
for premixed and partially premixed combustion [72, 73], is adopted here.
The filtered reaction rate term ω̇∗ in Eq. (7) takes the form of [82, 85]

ω̇ψ c d2 ψ eq 1 dψ eq
ω̇∗ = + ρχξ + 2 ρχ ξ c . (9)
ψ eq ψ eq dξ 2 ψ eq dξ
    
Premixed ω̇fp Non-premixed ω̇np Cross dissipation term ω̇cdr
Flame Root Dynamics and Their Role … 251

The three terms on the right-hand side represent the contributions from premixed
and non-premixed combustion modes, and their interactions resulting from the cross
dissipation rate. These are denoted as ω̇fp , ω̇np and ω̇cdr respectively. The cross dis-
sipation term is neglected following previous studies [82, 86].
The first term of Eq. (9) signifies the contribution of premixed combustion mode,
where the unscaled reaction rate is ω̇ψ = ω̇CO + ω̇CO2 . This term is modelled as [86]

1 1
ω̇fp (η, ζ )
ω̇fp = ρ  ζ ) dη dζ ,
P(η, (10)
ρ(η, ζ )
0 0

where ω̇fp (η, ζ ) and ρ(η, ζ ) are the flamelet reaction rate and density respectively.
They are obtained from one-dimensional unstrained planar laminar premixed flame
calculations over the flammability range. The density-weighted joint PDF is approx-
imated using statistically independent beta PDFs as P(η,  ζ) ≈ P β (η ; 
ξ , σξ,sgs
2
)×

Pβ (ζ ; 
c, σc,sgs ), where η and ζ are the sample space variables for the mixture fraction
2

and progress variable respectively. The DNS study by Chen et al. [87] demonstrated
that the sub-grid correlation is not influential on the time-averaged statistics because
the contribution related to the large-scale fluctuations is resolved in LES. There-
fore, the sub-grid correlation is not considered here and the statistical independence
assumption is made for simplicity.
The non-premixed contribution of Eq. (9) is modelled using [86]

1
1 d2 ψ eq (η) 
ω̇np = ρ 
cχ
ξ P (η) dη . (11)
ψ eq (η) dη2
0

The non-premixed contribution does not come from counterflow diffusion flamelets
and is instead a correction term for the premixed contribution. This term contains
the filtered mixture fraction scalar dissipation rate, which is the sum of the resolved
and SGS contributions χ  
ξ = D(∇ ξ · ∇ ξ) + χξ,sgs . The non-premixed contribution
is typically only significant in the stoichiometric regions, since d2 ψ eq /dη2 tends to
zero rapidly when moving away from stoichiometry [86].
The final term that requires closure is the reaction source term in Eq. (8) and is writ-
ten as (c ω̇∗ − c ω̇∗ ) = (c ω̇fp − 
c ω̇fp ) + (c ω̇np − 
c ω̇np ), where (c ω̇np − 
c ω̇np ) = 0
following Eq. (11). The term c ω̇fp is evaluated in a similar manner to Eq. (10) [86].
The Favre-filtered temperature T  is obtained using the filtered enthalpy obtained
from its transport equation, the mixture-averaged enthalpy of formation h  0f and
an effective specific heat capacity  c p through the approximation T  = T0 + ( h−
 f )/
h
0
c p , where ‘0’ refers to a reference temperature, which is 298.15 K in this
work. The three thermochemical quantities of the mixture, h  0
,c and M,  are
f p
calculated in a manner similar to Eq. (10), as described in detail by Ruan et al. [86].
252 J. C. Massey et al.

These three terms are included in a look-up table along with the two reaction rate
source terms in Eqs. (7) and (8).
Simulations of non-swirling jet flames [73, 86, 88, 89] and swirl-stabilised flames
in a gas turbine model combustor [76–80] using the LES–FlaRe modelling framework
described here are presented next to shed light on the physical mechanisms for the
dynamics and stabilisation of the lifted flame root.

3 Non-swirling Jet Flames

The first configuration to be analysed is the canonical round jet flame configuration
(see the schematic in Fig. 1), where several experiments of lifted flames have been
undertaken using hydrogen [90–92] and hydrocarbon fuels [10, 17, 93, 94], and the
parameters that have been varied include the Reynolds number (fuel jet velocity or
nozzle diameter) or the air dilution. These experiments have been used as initial cases
for developing the FlaRe modelling approach described in the previous section. The
model is assessed on its ability to capture the average lift-off height when the inlet
conditions are changed over a wide range. The flame stabilisation mechanisms are
explored with the aid of simulation data and the findings are discussed in this section.

3.1 Effects of Inlet Conditions

Hydrogen jet flames [90–92] are investigated first, since the chemical time scales
(τc = δ L0 /s L0 ) are significantly shorter than those for hydrocarbon flames, and fewer
species and elementary steps are involved in hydrogen–air combustion. The com-
bustion model was tested a priori using DNS data [96, 97] and a posteriori using
2D axisymmetric steady-state RANS simulations [86]. The experiment by Cheng et
al. [90] was used for model assessment and validation. It was shown that including
the non-premixed contribution ω̇np in Eq. (9) improved the predictions of the lift-off
height. Neglecting the non-premixed term causes the flame to stabilise closer to the
fuel nozzle exit, as observed in the study by Ruan et al. [86]. For RANS simulations,
the fluctuations of c and ξ are more significant and the use of statistically independent
PDFs may not be suitable because these fluctuations are entirely modelled. It was
found that including this correlation within the joint PDF using a copula method, as
described by Darbyshire and Swaminathan [98], improved the lift-off height predic-
tion. This is shown in Fig. 3a with the comparison against the experiment by Cheng et
al. [90]. Additional RANS simulations were undertaken to capture the lift-off height
for different jet velocities and compared to the experiment by Brockhinke et al. [92].
The jet velocity was varied between 500 and 850 m/s, which corresponds to pipe
Reynolds numbers that are approximately within the range 9000 < Re D < 15500.
It is shown in Fig. 3a that the average lift-off heights are well captured by the sim-
ulations using the FlaRe approach over the range of jet velocities used and that the
Flame Root Dynamics and Their Role … 253

Fig. 3 Comparisons between experimental studies and 2D axisymmetric RANS simulations of the
lift-off height for jet flames with different jet velocities for a hydrogen [86], b propane [95] and
c methane [95]

lift-off height increases with the jet velocity. The lift-off heights obtained using the
CMC model are also good [99] but this approach requires a substantially higher
computational cost.
The combustion modelling framework was then applied to jet flame configura-
tions using hydrocarbon fuels [95]. The results of 2D axisymmetric RANS sim-
254 J. C. Massey et al.

Fig. 4 Comparisons between experimental measurements [93] and 2D axisymmetric RANS simu-
lations [88] of the lift-off height for jet flames with different jet velocities and methane with varying
air dilution

ulations for propane and methane jet flames are shown in Fig. 3b, c respectively,
where two different nozzle diameters have been used for the methane cases. Previ-
ous simulation results using other combustion models are also shown wherever avail-
able for comparison. The velocities for the cases with propane are varied between
32 and 74 m/s, which correspond to a Reynolds number range of approximately
41, 500 < Re D < 96, 000, while the velocities for the cases with methane are varied
between 16 and 100 m/s, which correspond to a Reynolds number range of approx-
imately 3600 < Re D < 45, 000. It is shown that the lift-off height is well captured
for both hydrocarbon fuels over the wide range of jet velocities used. In addition, the
lift-off height is more sensitive to the jet velocity when methane is used in comparison
with the propane cases. To study the effect of air dilution on the lift-off height, the
FlaRe modelling framework was then applied to the jet flame configuration studied
experimentally by Ahmed and Mastorakos [93]. The inlet velocity and the amount of
air dilution by volume into the fuel jet were varied. These results are shown in Fig. 4
and it is shown that the comparisons between the experimental data and simulation
results obtained using 2D RANS [88] are very good. For a given jet velocity, the
lift-off height increases with the dilution level, which is due to the stoichiometric
mixture fraction contour being shifted towards the centre of the jet, where the local
velocities are larger, and the flame stabilises at a location further downstream. For
case F4 (40% air dilution), using a velocity that was higher than 22 m/s leads to
blow-off in the simulation, which was consistent with the observations in the exper-
imental study [93]. Therefore, it is demonstrated that the FlaRe combustion model
is capable of accurately capturing the variations in the average lift-off height over
a wide range of velocities and different fuels by using computationally inexpensive
2D axisymmetric steady RANS simulations.
Although the comparisons between the results obtained using the RANS frame-
work and the measured lift-off heights are good, it is also of interest to analyse the
transient evolution of the flame leading edge from its ignition location. Experimental
Flame Root Dynamics and Their Role … 255

data are available in the study by Ahmed and Mastorakos [93] for methane with 30%
air dilution for two different jet velocities and ignition locations. Simulations were
performed for the transient evolution of the flame leading edge using 2D and 3D
URANS simulations [89]. It was demonstrated that the final lift-off height was the
same in 2D and 3D simulations for both jet velocity cases, which further justifies the
axisymmetric assumption made for the steady RANS calculations presented earlier.
However, the interactions between the leading flame point and large-scale structures
cannot be captured using the URANS methodology, as the flow field and flame are
averaged. Therefore, simulations with the LES framework are required to capture the
instantaneous behaviour of the leading flame point and the propagation dynamics.
The findings using the LES–FlaRe approach are discussed next.

3.2 Flame Stabilisation

In the previous study using the 2D URANS framework [88], various stages of the
flame evolution were identified and these compared well with the experimental obser-
vations using high-speed movies [93]. The flame kernel is first convected downstream
quickly by the incoming flow and the kernel then grows in a spherical shape due to
both molecular and turbulent diffusion. The flame then expands in the radial direction
and the flame brush upstream edge remains at a location near the ignition location
until the flame is able to position itself between the lean and rich flammability mixture
fraction contours, where the most upstream point is located along the stoichiometric
mixture fraction contour. The flame then propagates upstream until it reaches its
final stabilisation height. The previous study by Chen et al. [89] was useful to gain
insights into the general flame stabilisation stages, but it did not capture the transient
interactions with the large-scale eddies, as the relevant scales associated with tur-
bulence and mixing are modelled using the URANS framework. It was also shown
in the study by Chen et al. [89] that compared to the measured position, the com-
puted flame leading edge was always closer to the fuel nozzle during the propagation
stage, which suggests that the propagation speed of the flame base was over predicted.
Hence, simulations with the LES framework then performed to obtain improvements
in the flame leading edge propagation stage and observe the influences on the flame
stabilisation due to large-scale eddies.
The LES–FlaRe framework described in Sect. 2 was applied to the same jet
burner [93] to study the transient evolution of the flame leading edge by Chen et
al. [73]. Two cases were studied using jet velocities of U j = 12.5 m/s and U j =
25.5 m/s of methane with 30% air dilution by volume; these are referred to here
respectively as F3-1 and F3-2. The ignition location is 30D j downstream of the fuel
nozzle exit for both cases, but measurements have also been taken for the ignition
location of 40D j downstream of the fuel nozzle exit. Studies with the LES framework
have been undertaken on the evolution of the leading flame edge with a jet velocity
of U j = 12.5 m/s using the artificially thickened flame model [100], and with a
256 J. C. Massey et al.

Fig. 5 Comparison of the time series of the flame leading edge location using experimental mea-
surements [93] and from numerical studies [76, 100–102]

jet velocity of U j = 25.5 m/s using the transported PDF approach with Eulerian
stochastic fields [101] and the CMC model [102].
The experimental study recorded a time series of the axial position of the most
upstream flame point using high-speed line-of-sight imaging and OH-PLIF measure-
ments. In the aforementioned LES studies, the flame leading point was tracked by
using T = 1200 K to 1500 K. These simulation results are compared to the experi-
mental measurements for the most upstream flame point in Fig. 5. For the low-velocity
case (F3-1), the results are available for the present simulations with the LES–FlaRe
approach [73] and the previous transported PDF [101] simulations. It can be seen
that the flame leading edge position is under predicted using LES–FlaRe, suggesting
that the propagation speed of the flame base is over predicted, which is attributed
to over predictions in the local reaction rates. In addition, the flame reaches a final
stabilisation height approximately 100 ms earlier than as seen in the experiment. On
the other hand, the flame leading edge location is consistently over predicted using
the transported PDF model, suggesting the propagation speed of the flame base is
under predicted. The final lift-off height for both approaches is approximately 5D j
above the fuel nozzle exit, which is 2D j higher than the experimentally observed
value of approximately 3D j above the fuel nozzle exit. For the high-velocity case, the
LES results obtained using LES–FlaRe [76], thickened flame [100] and CMC [103]
approaches are compared. It is shown in Fig. 5 that the flame leading edge position
is under predicted using the LES–FlaRe approach during the first 200 ms period,
which is similar to the behaviour for case F3-1. The final lift-off height for case F3-2
was approximately 14D j , which was approximately 2D j higher than the experi-
mentally observed value of approximately 12D j above the fuel nozzle exit. It had
been postulated that the statistical independence assumption for the presumed PDF
leads to the over predictions in the final lift-off height for both simulations using the
Flame Root Dynamics and Their Role … 257

Fig. 6 Evolution of the flame leading edge trajectory for case a F3-1 and b F3-2 [73]

LES–FlaRe approach [73]. However, it was later shown in the DNS study by Chen
et al. [87] that the correlation that represents the interactions between the mixture
fraction and progress variables fields was seen to be small for lifted flames. The
LES study using the artificially thickened flame approach over predicts the flame
leading point position for the first 150 ms, while the final lift-off height was similar
to the value obtained using LES–FlaRe. The CMC model consistently over predicts
the location of the flame leading edge throughout the entire 300 ms of simulation
runtime. Therefore, it is demonstrated that the LES–FlaRe approach performs well
in comparison with the other approaches but at a much lower computational cost.
The use of CFD is beneficial in comparison with experimental techniques, since
it is possible to extract further information from the simulation results, in order to
obtain further insights into the physical mechanisms. However, it should be noted that
CFD cannot replace experiments, but both are required to improve the understanding
and knowledge of the flame stabilisation mechanisms. An example of an advantage
for using CFD is the analysis of the leading flame point trajectory in 3D space for
cases F3-1 and F3-2 [73], as shown in Fig. 6a, b respectively, which are obtained
using the results from the simulations with LES–FlaRe approach. The ignition plane
for both cases is located at x/D j = 30, and it is shown that the flame leading point
initially moves up (i.e. downstream) along the central jet axis (solid black line). This
downstream movement from the ignition plane is more evident in case F3-2, since
258 J. C. Massey et al.

the fuel jet velocity is greater. The radial expansion that follows is different between
the two cases and also the flame leading edge path, as the local mixing conditions are
different for the two LES realisations. After reaching a certain radial distance from
the jet axis, which is observed to be approximately 5D j , the flame leading point then
moves upstream for both cases. The flame edge continues to propagate upstream
and follows a path where near-stoichiometric mixtures are present. These mixtures
are created due to the presence of large-scale eddies and the induced flame–flow
interactions. For case F3-1, it is shown in Fig. 6a at x/D j = 20 that the flame leading
edge follows a spiral path towards the stabilisation plane located at x/D j = 5. This
motion is similar for case F3-2 from x/D j = 24, but the radius of the spiral remains
larger than in case F3-1, as shown in Fig. 6b. It is also seen in Fig. 6b that the trajectory
in the region 14 ≤ x/D j ≤ 16 fluctuates more. This is due to the higher incoming
velocity, which causes more local extinction to be present, as the mixing time scales
as shorter and lead to the creation of mixtures out of the flammability range. When
the flame is near the stabilisation plane for both cases, it is shown in Fig. 6 that the
flame leading point moves around the jet axis with a circular trajectory. The spread
of the scattered points is wider for case F3-2 because the lift-off height is greater
and hence, the radial position for a given mixture fraction is larger. It was shown by
Chen et al. [73] that the radial location of the leading point is found to be within
lean mixtures. However, the presence of mixtures is not the only reason for the flame
stabilisation location and its presence may be attributed to other physical processes,
such as turbulence–combustion interactions and the presence of large-scale eddies.
It has been discussed thus far that the flame stabilisation processes are con-
trolled by numerous processes, which include local mixing, large flow structures
and turbulence–combustion interactions. This causes the flame root’s location to
change intermittently. These processes are different in swirling flows and the flame
stabilisation processes can lead to significant changes in the flame behaviour within
a confined flow configuration. These aspects are investigated next for gas turbine
combustors.

4 Gas Turbine Combustors

For practical combustion systems, it is possible that unexpected flame behaviour can
be present, which can lead to serious consequences. An example is the presence of
pressure fluctuations and if their amplitudes are large enough, this leads to increased
flow/acoustic oscillations within the engine, flashback of the flame and burning part
of the fuel injection system [25, 104]. This behaviour is referred to as thermo-acoustic
instabilities, which are the combination of the influences from turbulent flows, heat
release from the flame, hydrodynamic stability and acoustics. If the heat release
rate is in phase with the pressure oscillations and of similar amplitudes, thermo-
acoustic instabilities arise, which can significantly influence the flame stabilisation
in practical combustors [105]. Another safety hazard that must be considered for gas
turbine combustors is the risk of complete flame blow-off. This issue becomes critical
Flame Root Dynamics and Their Role … 259

as modern gas turbines operate under very lean conditions and understanding flame
stabilisation near the lean blow-off (LBO) limit is of paramount practical importance.

4.1 Model Combustor

The gas turbine model combustor [21, 22] developed by the German Aerospace
Centre (DLR) featuring a dual-swirler configuration with detailed measurements for
the flow/flame structures is of interest here. Three benchmark flames operating under
thermo-acoustically stable (flame A) and unstable (flame B) conditions, and close to
the LBO limit (flame C) are listed in Table 1 with their operation details. A schematic
of this burner is shown in Fig. 7. Dry air at atmospheric pressure and room temperature
enters a single plenum and the flow is split through two radial swirlers. The two co-
swirling flows enter the combustion chamber through central and annular nozzles,
in order to achieve quick fuel–air mixing. Non-swirling methane is fed through a
nozzle ring that contains 72 channels and is located between the two air nozzles.
The exit planes of the central air and fuel nozzle ring are 4.5 mm below the exit of
the annular air nozzle and the entrance to the combustion chamber. State-of-the-art
laser diagnostics techniques have been used to investigate the physical processes in
the three flames using stereoscopic particle image velocimetry (stereo-PIV), Raman
spectroscopy, laser Doppler velocimetry (LDV), OH/CH/CH2 O/acetone planar laser-
induced fluorescence (PLIF) and OH∗ /CH∗ chemiluminescence with high repetition
rates [21, 22, 106–108].
The extensive experimental datasets [21, 22] have served for the further validation
of the LES–FlaRe combustion modelling framework, reported in Refs. [76, 77] for
flames A and B, and Refs. [79, 80] for flame C. Flame A has also been attempted
with the LES framework using other combustion models including FPV and F-
TACLES [109], EDC [110] and FGM [111] approaches, and flame C was simulated
using the CMC model [103]. To date, the LES–FlaRe approach is the only model
to capture the correct behaviours of the unstable flame B undergoing pronounced
thermo-acoustic oscillations. Since the statistics for the flow and flame structures
have been well captured by the LES–FlaRe approach in previous studies, the focus
here is on the transient behaviours of the flame root and its stabilisation mechanisms
under stable, unstable and close to LBO conditions. The Raman measurements were

Table 1 Summary of the operating conditions for the three flames in the gas turbine model com-
bustor [21, 22]
Flame φglob ṁ a (m/s) ṁ f (m/s) S Pth (kW)
A 0.65 18.25 0.697 0.9 34.9
B 0.75 4.68 0.205 0.55 10.3
C 0.55 4.68 0.15 0.55 7.6
260 J. C. Massey et al.

Fig. 7 Schematic of the gas turbine model combustor [21, 22]

used to indicate the mean structure of the flame (temperature and species mass
fractions) and identify the mean lift-off height of the flame. The lift-off height H
was tracked using the most upstream point of T  = 1200 K above the fuel nozzle, as
shown in Fig. 8 using H , which is also illustrated in the schematic in Fig. 2.
The measured lift-off heights for flames A, B and C are H ≈ 5 mm, 4 mm and
6 mm respectively, but the behaviour in the flame root region for the three cases
is different. The experimental studies have observed the main flame stabilisation
mechanisms, as well as a wide range of transient phenomena. A PVC is observed in
isothermal and reacting flows for this configuration, which is a form of hydrodynamic
instability that appears in swirling flows usually with high swirl numbers [3]. It is a
helical vortex structure residing in the shear layer between the swirling jet and the
inner recirculation zone, creating a low pressure region as marked in Fig. 8. It is pos-
sible that the PVC can be dampened or disappear when combustion is present [112],
but the flame stabilisation mechanisms for all three flames studied here are highly
influenced by the PVC, since it is located in the ISL where the flame root is present.
The ISL originates from the fuel nozzle (as illustrated in Fig. 2) and is close to the
flame and outside of the IRZ, as shown by the pink lines in Fig. 8. The PVC gen-
erates regions of a high vorticity magnitude and since the PVC passes through the
unburnt reactants, this induces strong mixing processes between the fuel and air.
These large-scale structures are convected along the ISL and this leads to flame–
vortex interactions, where hot combustion products are entrained into the vortices.
Flame Root Dynamics and Their Role … 261

This has been observed in the experiments [107, 113] to be a dominant mechanism
for flame stabilisation. To visualise this mechanism, a mid-plane snapshot of the
filtered reaction rate and the stoichiometric mixture fraction contour are shown in
Fig. 8. This is taken from the flame C simulation when the flame root was stable.
In the flame root region where H is marked, it is shown on the right-hand side that
the PVC passes through an area of unburnt reactants near the flame root. The stoi-
chiometric mixture fraction is in the vicinity of the PVC, indicating that the fuel–air
mixture will become well-mixed and ignition should occur, as the velocity vectors
in the inset of Fig. 8 indicate a vortex. The left-hand side of Fig. 8 indicates the next
stage, which is PVC induced flame roll-up. The vortex is close to the stoichiomet-
ric mixture fraction and the flame root, which is also suggesting that the fuel–air
mixture within the vortex is ignited. On the right-hand side of Fig. 8, a large-scale
vortex with a high reaction rate has been convected further downstream. The velocity
vectors shown in the inset of Fig. 8 suggest that the vorticity is strong, and hot com-
bustion products and radical species are entrained into the vortex region to sustain
combustion. The reaction zone is then quenched when vortex breakdown occurs, as
shown in the downstream regions of Fig. 8.
The aforementioned PVC induced sequence can be observed continuously as the
PVC spins about the centre axis in flame A, which stably retains a ‘V’ shape flame.
Although such a flame stabilisation process is still present in flame B, the flame shape
changes significantly with the presence of pressure oscillations, which causes a flat
flame to be close to the dump plane, and the PVC cannot be always identified. As
observed in the LES studies by Chen et al. [76, 77], the pressure fluctuations caused
variations in the air mass flow split between the central and annular air passages
during one revolution of the PVC. The different acoustic impedances due to the
swirlers cause fuel jet flapping behaviour when the PVC is out of plane, i.e. not
near the fuel nozzle. This causes enhanced radial mixing of the fuel and leads to a
shorter and flattened flame shape, which was also observed in the experiments by
Boxx et al. [114] and Steinberg et al. [115]. It has also been observed [77] that the
structure of the PVC is influenced by the pressure fluctuations within the combustion
chamber. The PVC induced flame roll-up and enhanced mixing effects are present, as
also seen in flame A, at the maximum and minimum pressure inside the combustion
chamber. The axial distance between the vortices is observed to be higher when
the pressure is higher, which suggests that the PVC extends for higher pressures
and retracts for lower pressures. This behaviour was also seen in the experimental
study by Steinberg et al. [115] and occurred in numerous thermo-acoustic cycles. In
addition, there was frequent disappearance of the flame root during large pressure
fluctuations, and the root of the PVC contracted and expanded radially for high and
low pressures respectively. The flame root was able to reform when the pressure
fluctuations become smaller, which was also observed in the experimental study by
Boxx et al. [114]. For flame C, the flame stabilisation sequence described for Fig. 8
is also different, since the global equivalence ratio is leaner than in flame B (see
Table 1) and both cases have similar inlet velocities. As a result, the stoichiometric
mixture fraction contour is further away from the PVC and mixing enhancement is
weakened, which occasionally leads to failed ignition during a later flame roll-up
262 J. C. Massey et al.

Fig. 8 Snapshot of the filtered reaction rate, the stoichiometric mixture fraction (white lines) and
the zero axial velocity contour (pink lines) in the x–y mid-plane for flame C. The inset shows the
filtered reaction rate and the velocity vectors

process. Given the different flame base behaviours taking place in the three flames, it
is of interest to compare the flame leading edge behaviour, which is discussed next.
As previously mentioned, the flame root and its azimuthal position are highly
influenced by the PVC, which has also been seen when analysing time sequences
of the LES snapshots for the three flames. The flame leading point, i.e. the most
upstream location of the T  = 1200 K iso-surface in the IRZ, and the fluctuation
of its azimuthal position θ on the transverse plane are analysed. The fast Fourier
transform (FFT) is applied to time series of this fluctuation to obtain the spectral
behaviour for the three flames in Fig. 9. For flames A and B, it is shown in Fig. 9a,
Flame Root Dynamics and Their Role … 263

Fig. 9 Spectral behaviour of

the azimuthal movement for
the flame root using the
azimuthal position
fluctuations for a flame A,
b flame B and c flame C
264 J. C. Massey et al.

b respectively that the FFT signals have a single pronounced peak corresponding
to the PVC frequency. This suggests that the PVC is the dominant mechanism for
the stabilisation of this flame and the azimuthal position of the flame follows the
rotation of the PVC. For flame C, the FFT looks significantly different as there is a
peak that is at the PVC frequency, but it is much less prominent. Flame C is close
to the LBO limit and is highly unstable. This is typical in modern jet engines, since
the aim is to operate at fuel-lean conditions in order to reduce pollutant emissions
and therefore, there is a risk of flame blow-off. These risks were captured for this
configuration in the study by Stöhr et al.[107] using simultaneous stereo-PIV and
OH-PLIF techniques. It was observed in the study by Stöhr et al. [107] that flame
lift-off occurred 1–2 times per second and lasted between 0.1 and 0.15 s. The lift-off
height was observed to reach a height of 30–40 mm. This lift-off event was triggered
due to failed ignition at the flame root and within the vortices. This suggests that the
PVC induced fuel–air mixing and flame roll-up is weaker and this leads to the flame
becoming unstable. This is reflected in Fig. 9c as there is some low-frequency noise
below the PVC frequency.
It was also reported by Stöhr et al. [107] that during an unstable blow-off event,
the flame root was extinguished and this led to complete flame lift-off, i.e. the flame
stabilised at the combustor exit. The ultimate blow-off event was only observed when
further reducing the global equivalence ratio. Highly transient flame root dynamics
were observed, as also seen for flame B [114], but failed ignition within the convected
large-scale vortex structures was also observed before the blow-off event. This occurs
if re-ignition does not occur after more than 2 ms, which is the duration for one
rotation of the PVC. During the lift-off event, the flame stabilised downstream and
the flame then moved back upstream and returned to the location of H ≈ 6mm. In the
LES study by Massey et al. [79], a lift-off event was captured using the LES–FlaRe
framework. This was not captured in the CMC study by [103], since the model
used was a non-premixed model with single conditioning on the mixture fraction
and strong partial premixing is present at the flame root region and vortices. When
a lift-off event occurs, different flame shapes were observed leading to significant
temporal variations in the the heat release rate [79]. This phenomenon captured in
the LES is demonstrated in Fig. 10, which shows two time series of the integrated
heat release rate and the lift-off height. The lift-off height is also moving-averaged
over a 4.5 ms window and the averaged values are shown using horizontal grey lines.
The dash-dotted line in the bottom of Fig. 10 shows the temporal variation of the
heat release rate integrated over the entire combustion chamber, denoted as Q̇ glob ,
for 45 ms, which is an arbitrarily chosen interval that included a lift-off event in the
simulation. The volume integrated heat release rate varies with time, but it is close
to the thermal power of Pth = 7.6 kW for the experiment. However, it is difficult to
identify the lift-off event from this quantity. Therefore, the temporal variation of the
heat release rate is integrated over a smaller volume around the fuel nozzle region
to gather information near the flame root. This heat release rate, denoted as Q̇ nozzle ,
is also shown in Fig. 10, using the solid line, but multiplied by 100 to show on the
same scale as the global values. It is evident that the heat release rate in this region
Flame Root Dynamics and Their Role … 265

Fig. 10 Time series of the lift-off height and volume integrated heat release rate for a lift-off event
in flame C

changes significantly over the time interval shown and it is clear that the flame has
lifted off in the range 124 < t < 141 ms, as the flame is out of the near fuel nozzle
region. The trend seen for the lift-off height is directly linked to Q̇ nozzle . After the first
window of 4.5 ms, the lift-off height fluctuates up to t = 131 ms, since the unstable
behaviour and fluctuating heat release rate are caused by the flame root trying to
establish itself. Beyond this time, the averaged lift-off height significantly increases,
suggesting that the flame recedes downstream and does not stabilise near the fuel
nozzle, which is shown by the low heat release rate seen during this time window.
The last window of 4.5 ms shows that the average height is very similar to the first
window and hence, the flame leading edge is re-established at its typical location
with a more stable flame in its typical ‘V’ shape.
266 J. C. Massey et al.

Two instants are marked with red points in Fig. 10 at t = 108.05 ms and 111.05 ms,
which represent precursors to the lift-off event. As seen in Fig. 10, the lift-off height
increases and the heat release rate in the flame root region decreases after both of
these instances. On the right-hand of the first frame, there is no reaction along the
stoichiometric mixture fraction line and the reaction rate is very weak in comparison
with the reaction rate field in Fig. 8. This is caused by the air entrainment between
the stoichiometric reactant mixture and hot products, and this prevents ignition at the
flame root region. This causes the flame to move away in the radial direction in the
right half of the domain and leads to the sudden drop in the values of Q̇ nozzle that is
seen in the time series in Fig. 10. Consequently, the flame root recedes downstream,
leading to an increase in the flame lift-off height around t = 108 ms, as shown in the
H time series. As seen in the second frame, a weak vortex is shown on the right-hand
side, and a pocket of rich mixture is present along the centreline near the bottom of
the combustion chamber in the region where the flame root typically stabilises. This
prevented ignition of the vortex on the right-hand side and there was no flame roll-
up. The flame then moved to a downstream position and therefore, the flame root is
shifted from its typical location leading to an lift-off event.
The analyses presented here of the simulations using the LES–FlaRe framework
have demonstrated that the underlying mechanisms that influence the flame root
dynamics for flames under thermo-acoustically unstable conditions and near the
lean blow-off limit compare well with the experimental observations. This suggests
that the modelling framework is suitable for simulating full-scale gas turbine com-
bustors, since the approach is also economical from a computational cost perspective.
Combustion models need to have a low computational cost to simulate such config-
urations, since the grid sizes are typically large and the geometries often include
multiple swirlers and complex injection systems. A brief overview of some progress
with simulations of industrial burners using the LES–FlaRe framework is presented
next.

4.2 Industrial Burners

Simulations of practical configurations are challenging, since experiments which

can be used to validate combustion models are scarce. This is because such prac-
tical configurations operate at elevated pressures and the reactants are preheated.
Therefore, non-intrusive techniques are required, which are very expensive for con-
figurations operating at high pressures. Furthermore, the gases/combusting mixtures
can become increasingly translucent with optical depth as pressure increases and
thus, laser diagnostics can become challenging. Measurements have been conducted
for one configuration, which is the DLR-AT Big Optical Single Sector (BOSS) rig,
where OH and spray measurements have been obtained [116–119]. The combustor
rig is fitted with a Rolls-Royce developmental lean burn injector having a central
pilot surrounded by a main jet. The operating conditions are chosen to represent
Flame Root Dynamics and Their Role … 267

take-off conditions, and air with a temperature above 700 K and a pressure above
10 bar passes through the swirlers in the burner. Liquid kerosene is injected through
fuel atomisers and this produces a partially premixed flame that stabilises in the shear
layers formed along the inner and outer recirculation zones. It has been observed that
the flame is sensitive to changes in its shape due to turbulence–combustion interac-
tions. The flame is mainly attached in the shear layer of central and outer recirculation
zones, and the flame has an ‘M’ shape. However, it has been observed that the opening
angle of the pilot can cause the flame shape to change to a ‘V’ shape.
The study by Langella et al. [120] used the LES–FlaRe formulation to gain insights
into the flame switching behaviour, and it was demonstrated that the flame configu-
rations and spray characteristics were well captured. Analysis of the flames showed
that a PVC was not present, but a small vortex formed intermittently close to the pilot
and this occurred with a flipping movement of the pilot jet. The change to the ‘V’
shape occurs when the vorticity is not strong enough, and this caused richer fuel–air
mixture pockets, leading to the flame moving upstream and the flame root stabilising
closer to the injectors. It was also observed that when the flame has established its
‘V’ shape, the flame influences the vorticity field and local richer fuel–air mixture
pockets were formed. This also causes the flame to change to a ‘V’ shape. The ‘V’
shape flame is undesirable, since it is close to the fuel injector system and poses risks
of damaging the injector nozzle. Therefore, increasing the vorticity magnitude can
create more homogeneous mixtures to prevent ‘V’ shape flames occurring. Capturing
these underlying mechanisms behind flame stabilisation in real-engine combustors
with flamelet models demonstrates that such models are promising for modelling
full-scale gas turbine combustors and serve as a useful tool for industry.

5 Conclusions

This chapter has presented a broad overview of lifted flames in turbulent flows with
and without swirl in laboratory and practical applications used for energy production
and propulsion. The stabilisation mechanisms of lifted partially premixed flames
are complex and understanding these mechanisms is important for the design of
combustion systems. Numerical modelling techniques with the LES paradigm are
becoming more widely used as a CFD tool, due to their capabilities of modelling
small-scale mixing and turbulence–combustion interactions, and resolving the flow
field at the larger scales of turbulence. Different sub-grid combustion models have
been briefly discussed with a focus on the LES–FlaRe modelling framework, which
offers a reduced computational cost while maintaining physical consistencies for the
sub-grid processes and their interactions. The two lifted flame configurations have
been investigated: the canonical jet flame and a gas turbine model combustor with
co-axial swirlers. The RANS formulation of the FlaRe framework is able to accu-
rately capture the lift-off height for non-swirling jet flames with different fuels and
jet velocities. However, the framework lacks the capability of modelling the transient
evolution of the flame root and therefore, the LES framework is required. The flame
268 J. C. Massey et al.

leading edge evolution for two jet velocities obtained using the LES–FlaRe approach
compared well against results obtained using more computationally expensive mod-
els. In the gas turbine model combustor, the average flame stabilisation height is less
sensitive to changes in the inlet velocity and the global equivalence ratio. However,
the flame root behaviour and its stabilisation are highly sensitive to changes to these
inlet conditions. The simulations undertaken with the LES–FlaRe approach demon-
strated that the underlying mechanisms for flames under thermo-acoustically stable
and unstable conditions, and close to lean blow-off were well captured in compar-
ison with the experimental observations. The PVC is the dominant mechanism for
flame stabilisation even in the presence of self-induced thermo-acoustic instabilities
and under lean blow-off conditions. The LES–FlaRe framework has been applied to
the industrial burner BOSS rig and it is briefly discussed that the modelling frame-
work is able to produce the switch in flame shape. The underlying flame stabilisation
mechanisms for lifted flames can therefore be captured using an LES flamelet based
closure in a variety of configurations. The modelling approach can serve as a use-
ful approach for industry, due to the reduced computational cost while maintaining
physical consistencies within the framework.

Acknowledgements The authors acknowledge the financial support from Mitsubishi Heavy
Industries, Ltd., Japan. This work used the ARCHER UK National Supercomputing Service
(https://www.archer.ac.uk) with the resources provided by the UKCTRF (e305).

References

1. Spalding DB (1979) Combustion and mass transfer. Pergamon Press, Oxford, UK

2. Peters N (2000) Turbulent combustion. Cambridge University Press, Cambridge, UK
3. Syred N (2006) Prog Energy Combust Sci 32(2), 93–161
4. Mansour MS (2003) Combust Flame 133(3), 263–274
5. Lyons KM (2007) Prog Energy Combust Sci 33(2), 211–231
6. Wohl K, Kapp NM, Gazley C (1948) Symp Combust Flame Explos Phenom 3(1):3–21
7. Pitts WM (1988) Symp (Int) Combust 22(1), 809–816
8. Lawn CJ (2009) Prog Energy Combust Sci 35(1), 1–30
9. Vanquickenborne L, van Tiggelen A (1966) Combust Flame 10(1), 59–69
10. Kalghatgi GT (1984) Combust Sci Technol 41(1–2), 17–29
11. Peters N, Williams FA (1983) AIAA J 21(3), 423–429
12. Müller CM, Breitbach H, Peters N (1994) Symp (Int) Combust 25(1):1099–1106
13. Buckmaster J, Matalon M (1988) Symp (Int) Combust 22(1):1527–1535
14. Buckmaster J (2002) Prog Energy Combust Sci 28(5), 435–475
15. Domingo P, Vervisch L (1996) Symp (Int) Combust 26(1):233–240
16. Broadwell JE, Dahm WJA, Mungal MG (1984) Symp (Int) Combust 20(1):303–310
17. Miake-Lye RC, Hammer JA (1988) Symp (Int) Combust 22(1):817–824
18. Feikema D, Chen RH, Driscoll JF (1990) Combust Flame 80(2), 183–195
19. Gupta AL, Lilley DG, Syred N (1984) Swirl flows. Abacus Press, Royal Tunbridge Wells,
UK
20. Syred N, Beér JM (1974) Combust Flame 23(2), 143–201
21. Weigand P, Meier W, Duan XR, Stricker W, Aigner M (2006) Combust Flame 144(1–2),
205–224
Flame Root Dynamics and Their Role … 269

22. Meier W, Duan XR, Weigand P (2006) Combust Flame 144(1–2), 225–236
23. Matalon M (2009) Proc Combust Inst 32(1), 57–82
24. Lieuwen TC (2012) Unsteady Combustor Physics. Cambridge University Press, Cambridge,
UK
25. Dowling AP, Mahmoudi Y (2015) Proc Combust Inst 35(1), 65–100
26. Masri AR (2015) Proc Combust Inst 35(2), 1115–1136
27. Pope SB (2000) Turbulent flows. Cambridge University Press, Cambridge, UK
28. Gicquel LYM, Staffelbach G, Poinsot T (2012) Prog Energy Combust Sci 38(6), 782–817
29. Pope SB (2013) Proc Combust Inst 34(1), 1–31
30. Swaminathan N, Bray KNC (eds) (2011) Turbulent premixed flames. Cambridge University
Press, Cambridge, UK
31. Chen ZX, Langella I, Swaminathan N (2019) The role of CFD in modern jet engine combustor
design. Environ Impact Aviat Sustain Solut IntechOpen
32. Williams FA (1985) Combustion theory, 2nd edn. Benjamin Cummings, Menlo Park, CA,
USA
33. Hinze JO (1959) Turbulence. McGraw-Hill, New York, NY, USA
34. Tennekes H, Lumley JL (1972) A first course in turbulence. MIT Press
35. Smagorinsky J (1963) Mon. Weather Rev. 91(3), 99–164
36. Germano M, Piomelli U, Moin P, Cabot WH (1991) Phys Fluids A Fluid Dyn 3(7), 1760–1765
37. Moin P, Squires K, Cabot W, Lee S (1991) Phys Fluids A Fluid Dyn 3(11), 2746–2757
38. Lilly DK (1992) Phys. Fluids A Fluid Dyn 4(3), 633–635
39. Vreman AW (2004) Phys Fluids 16(10), 3670–3681
40. Nicoud F, Ducros F (1999) Flow Turbul Combust 62(3), 183–200
41. Nicoud F, Toda HB, Cabrit O, Bose S, Lee J (2011) Phys Fluids 23(8):085,106
42. Yoshizawa A, Horiuti K (1985) J Phys Soc Japan 54(8), 2834–2839
43. Ghosal S, Lund TS, Moin P, Akselvoll K (1995) J Fluid Mech 286:229–255
44. Kim WW, Menon S (1995) 33rd AIAA aerospace sciences meeting and exhibit. AIAA 95-
0356, Reno, NV, USA
45. Butler TD, O’Rourke PJ (1997) Symp (Int) Combust 16(1):1503–1515
46. Colin O, Ducros F, Veynante D, Poinsot T (2000) Phys Fluids 12(7), 1843–1863
47. Kerstein AR, Ashurst WT, Williams FA (1988) Phys Rev A 37(7), 2728–2731
48. McMurtry PA, Menon S, Kerstein AR (1992) Symp (Int) Combust 24(1):271–278
49. Marble FE, Broadwell JE (1977) Project squid technical report TRW-9-PU. Purdue University,
West Lafayette, IN, USA
50. Bray KNC, Cant RS (1991) Proc R Soc A 434(1890), 217–240
51. Hawkes ER, Cant RS (2001) Combust Flame 126(3), 1617–1629
52. Pitsch H (2006) Annu Rev Fluid Mech 38(1), 453–482
53. Dopazo C, O’Brien EE (1974) Phys Fluids 17(11), 1968–1975
54. Pope SB (1985) Prog Energy Combust Sci 11(2), 119–192
55. Valiño L (1998) Flow Turbul Combust 60(2), 157–172
56. Rembold B, Jenny P (2006) J Comput Phys 220(1), 59–87
57. Klimenko AY (1990) Fluid Dyn 25:327–331
58. Bilger RW (1993) Phys Fluids A Fluid Dyn 5(2), 436–444
59. Navarro-Martinez S, Kronenburg A, Di Mare F (2005) Flow Turbul Combust 75(1–4), 245–
274
60. Klimenko AY, Pope SB (2003) Phys Fluids 15(7), 1907–1925
61. Cleary MJ, Klimenko AY (2009) Flow Turbul Combust 82(4), 477–491
62. Libby PA, Bray KNC (1980) Combust Flame 39(1), 33–41
63. Liew SK, Bray KNC, Moss JB (1981) Combust Sci Technol 27(1–2), 69–73
64. Bradley D, Gaskell PH, Lau AKC (1990) Symp (Int) Combust 23(1):685–692
65. Bray KNC, Libby PA, Masuya G, Moss JB (1981) Combust Sci Technol 25(3–4), 127–140
66. Libby PA, Bray KNC (1981) AIAA J 19(2), 205–213
67. van Oijen JA, de Goey LPH (2000) Combust Sci Technol 161(1), 113–137
68. Gicquel O, Darabiha N, Thévenin D (2000) Proc Combust Inst 28(2), 1901–1908
270 J. C. Massey et al.

69. Pierce CD, Moin P (2004) J Fluid Mech 504:73–97

70. Langella I, Swaminathan N, Pitz RW (2016) Combust Flame 173:161–178
71. Langella I, Swaminathan N, Williams FA, Furukawa J (2016) Combust Sci Technol 188(9),
1565–1591
72. Langella I, Swaminathan N (2016) Combust Theory Model 20(3), 410–440
73. Chen Z, Ruan S, Swaminathan N (2017) Proc Combust Inst 36(2), 1645–1652
74. Chen ZX, Langella I, Barlow RS, Swaminathan N (2020) Combust Flame 212:415–432
75. Langella I, Chen ZX, Swaminathan N, Sadasivuni SK (2018) J Propuls Power 34(5), 1269–
1284
76. Chen ZX, Swaminathan N, Stöhr M, Meier W (2019) Proc Combust Inst 37(2), 2325–2333
77. Chen ZX, Langella I, Swaminathan N, Stöhr M, Meier W, Kolla H (2019) Combust Flame
203:279–300
78. Chen ZX, Swaminathan N (2020) Fuel 275:117,868
79. Massey JC, Chen ZX, Swaminathan N (2019) Combust Sci Technol 191(5–6), 1019–1042
80. Massey JC, Chen ZX, Swaminathan N (2021) Flow Turbul Combust 106(4):1355–1378
81. Bilger RW (1988) Symp (Int) Combust 22(1):475–488
82. Bray K, Domingo P, Vervisch L (2005) Combust Flame 141(4), 431–437
83. Swaminathan N, Bray KNC (2005) Combust Flame 143(4), 549–565
84. Dunstan TD, Minamoto Y, Chakraborty N, Swaminathan N (2013) Proc Combust Inst 34(1),
1193–1201
85. Domingo P, Vervisch L, Bray K (2002) Combust Theory Model 6(4), 529–551
86. Ruan S, Swaminathan N, Darbyshire O (2014) Combust Theory Model 18(2), 295–329
87. Chen ZX, Doan NAK, Ruan S, Langella I, Swaminathan N (2018) Combust Theory Model
22(5), 862–882
88. Chen Z, Ruan S, Swaminathan N (2015) Combust Flame 162(3), 703–716
89. Chen Z, Ruan S, Swaminathan N (2016) Combust Theory Model 20(4), 592–612
90. Cheng TS, Wehrmeyer JA, Pitz RW (1992) Combust Flame 91(3–4), 323–345
91. Brockhinke A, Andresen P, Kohse-Höinghaus K (1996) Symp (Int) Combust 26(1):153–159
92. Brockhinke A, Haufe S, Kohse-Höinghaus K (2000) Combust Flame 121(1–2), 367–377
93. Ahmed SF, Mastorakos E (2006) Combust Flame 146(1–2), 215–231
94. Røkke NA, Hustad JE, Sønju OK (1994) Combust Flame 97(1), 88–106
95. Ruan S, Chen Z, Swaminathan N (2015) 25th international colloquium on the dynamics of
explosions and reactive systems. Leeds, UK
96. Mizobuchi Y, Tachibana S, Shinio J, Ogawa S, Takeno T (2002) Proc Combust Inst 29(2),
2009–2015
97. Mizobuchi Y, Shinjo J, Ogawa S, Takeno T (2005) Proc Combust Inst 30(1), 611–619
98. Darbyshire OR, Swaminathan N (2012) Combust Sci Technol 184(12), 2036–2067
99. Kim IS, Mastorakos E (2005) Proc Combust Inst 30(1), 911–918
100. Lacaze G, Richardson E, Poinsot T (2009) Combust Flame 156(10), 1993–2009
101. Jones WP, Prasad VN (2011) Proc Combust Inst 33(1), 1355–1363
102. Zhang H, Giusti A, Mastorakos E (2019) Proc Combust Inst 37(2), 2125–2132
103. Zhang H, Mastorakos E (2019) Flow Turbul Combust 102(4), 909–930
104. Poinsot T (2017) Proc Combust Inst 36(1), 1–28
105. Huang Y, Yang V (2009) Prog Energy Combust Sci 35(4), 293–364
106. Stöhr M, Sadanandan R, Meier W (2009) Proc Combust Inst 32(2), 2925–2932
107. Stöhr M, Boxx I, Carter C, Meier W (2011) Proc Combust Inst 33(2), 2953–2960
108. Stöhr M, Arndt CM, Meier W (2015) Proc Combust Inst 35(3), 3327–3335
109. See YC, Ihme M (2015) Proc Combust Inst 35(2), 1225–1234
110. Benim AC, Iqbal S, Meier W, Joos F, Wiedermann A (2017) Appl Therm Eng 110:202–212
111. Donini A, Bastiaans RJM, van Oijen JA, de Goey LPH (2017) Flow Turbul Combust 98(3),
887–922
112. Selle L, Lartigue G, Poinsot T, Koch R, Schildmacher KU, Krebs W, Prade B, Kaufmann P,
Veynante D (2004) Combust Flame 137(4), 489–505
113. Stöhr M, Boxx I, Carter CD, Meier W (2012) Combust Flame 159(8), 2636–2649
Flame Root Dynamics and Their Role … 271

114. Boxx I, Stöhr M, Carter C, Meier W (2010) Combust Flame 157(8), 1510–1525
115. Steinberg AM, Boxx I, Stöhr M, Carter CD, Meier W (2010) Combust Flame 157(12), 2250–
2266
116. Schneider D, Meier U, Quade W, Koopman J, Aumeier T, Langfeld A, Behrendt T, Hassa C,
Rackwitz L (2010) 27th congress of international council of the aeronautical sciences. Nice,
France
117. Freitag S, Behrendt T, Heinze J, Lange L, Meier U, Hassa C, Rackwitz L (2011) 24th european
conference on liquid atomization and spray systems. Estoril, Portugal
118. Meier U, Heinze J, Lange L, Hassa C, Rackwitz L, Doerr T (2012) CEAS Aeronaut J 3(1),
45–53
119. Meier U, Lange L, Heinze J, Hassa C, Sadig S, Luff D (2015) J Eng Gas Turbines Power
137:7
120. Langella I, Heinze J, Behrendt T, Voigt L, Swaminathan N, Zedda M (2019) Proceedings of
the ASME Turbo Expo 2019: Turbomachinery Technical Conference and Exposition. Volume
4B: Combustion, Fuels, and Emissions, V04BT04A047. ASME GT2019-91879, Phoenix,
AZ, USA
Effect of Oxygenation on the Liftoff,
Stabilization, and Blowout
Characteristics of Laminar Co-flow Jet
Flames

Rajesh Kumar, Krishna C. Kalvakala, and Suresh K. Aggarwal

Abstract This chapter summarizes the results of a computational study that exam-
ined the effects of a new oxygenation strategy on the flame structure, liftoff
height, stabilization, and blowout characteristics in non-premixed laminar co-flow
jets. Oxygenated ethylene–air flames were established by employing an innovative
strategy of using N2 -diluted fuel jet along with O2 -enriched oxidizer jet such that
the stoichiometric mixture fraction (f st ) is varied but the adiabatic flame temperature
is not materially changed. For each case, the flame liftoff and blowout processes
were examined by using CO2 dilution in the fuel jet. A time accurate, implicit,
axisymmetric CFD code that uses a detailed description of transport and chemistry
(111 species and 784 reactions) and that includes radiation effects was adopted.
The computational model and chemical mechanism were validated using published
measurements of the flame length, temperature, and species profiles of ethylene/air
jet flames. Simulations indicated burner attached flames for all levels of oxygen
enrichment, without CO2 dilution. The flame stabilization mechanism in the near-
field region can be described by a balance between the reaction rate and local scalar
dissipation rate. Further, as the oxygen enrichment is increased, it becomes more diffi-
cult to get lifted flames. However, once a flame is lifted, its liftoff height increases
rapidly with the increase in CO2 dilution, i.e., a lifted oxygenated flame exhibited a
greater sensitivity to CO2 dilution.

Keywords Oxygenation · Dilution · Liftoff · Blowout

1 Introduction

Flame extinction is one of the fundamental and classical phenomena in combustion

science, where both chemistry and transport play a vital role. This phenomenon has
high potential to dictate the efficiency and emissions from combustion systems used in
transportation and power generation applications. In addition, it has direct relevance

R. Kumar · K. C. Kalvakala · S. K. Aggarwal (B)

University of Illinois at Chicago, Chicago, IL 60607, USA
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 273
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_12
274 R. Kumar et al.

in safe, stable, and reliable operation of these combustion systems. For instance,
flame extinction closely correlates with the flame flashback in micro-combustors,
flame quenching due to flame-wall interaction in furnaces, and with lean burning
limit in gas turbines. This further emphasizes the importance of flame extinction in
the design of combustion devices, including furnaces and gas turbine combustors.
In many systems, flame extinction is often described in terms of flammability
limits, burning velocity, liftoff, and blowout characteristics. In order to gain funda-
mental understanding of this mechanism, laminar lifted flames have been commonly
investigated [1–7]. In a jet flame configuration, lifted flames can be established in
near field or far field by adjusting the flow conditions, dilution of fuel, premixing,
or fuel composition. For instance, increase in jet velocity for laminar premixed jets
makes the flame to liftoff from the burner [8, 9], which changes the flame structure to a
large extent. Increase in liftoff height is accompanied by flame structure transitioning
from a non-premixed flame to a double flame and then to a triple flame containing
three reaction zones—a lean premixed, a rich premixed, and a non-premixed zone
[10, 11]. In addition to jet velocity, several other factors including fuel and air-stream
dilution [9, 12–15], temperature of air stream [16–18], and partial premixing [9, 12,
14] have been identified to have a significant impact on stabilization of lifted flames.
Lee et al. [9] examined the effects of nozzle diameter and jet velocity on the liftoff
characteristics of N2 -diluted, non-premixed propane jet flames. The nozzle diameter
and amount of dilution were found to have prominent effect on the liftoff and blowout
processes, and three regimes for liftoff and blowout were identified depending on the
jet diameter and velocity. Lock et al. [13] examined the effects of fuel- and air-stream
dilution on the liftoff and blowout of partially premixed (PPFs) and non-premixed
(NPF) methane flames. It was noticed that dilution of either of air or fuel stream with
CO2 tends to liftoff the flame, eventually leading to blowout. However, with fuel-
stream dilution, PPFs stabilized at a higher liftoff height and blew out at a lower CO2
dilution than NPFs. In contrast, with air-stream dilution, NPFs were lifted higher, and
blew out at a lower CO2 dilution than PPFs. In addition, results showed remarkable
similarity in the blowout or extinction behavior of co-flow and counterflow flames
with respect to the level of partial premixing and air- and fuel-stream dilutions.
In a subsequent study, Lock et al. [14] elaborated on the role of air- and fuel-
stream dilution on the flame liftoff and blowout characteristics for three different
fuels—methane, acetylene, and ethylene–under PPF and NPF conditions. It was
observed that ethylene and acetylene flames exhibit stronger resistance to blowout
or in other words are more difficult to extinguish than methane flames. Kim et al.
[17] investigated the liftoff characteristics of N2 -diluted propane diffusion flames at
elevated temperatures. It was observed that the lifted flames maintain a tribrachial
or triple flame structure up to a temperature of 900 K. In addition, while liftoff
height was found to decrease with temperature, the lifted flames were stabilized by
a balance between local flow velocity and the tribrachial flame propagation speed
regardless of temperature. Further, Choi and Chung [18] experimentally examined
the liftoff characteristics of methane, ethylene, ethane, and n-butane fuels at elevated
temperatures. The study highlighted the importance of auto-ignition temperature on
the flame liftoff and blowout behavior. It was observed that at temperatures lower
Effect of Oxygenation on the Liftoff, Stabilization, and Blowout … 275

than auto-ignition temperature, the liftoff and blowout characteristics varied with
Schmidt number (Sc) of the fuel, while at higher temperatures, regardless of Sc
number, ignition delay time of fuel/air mixture had a significant effect on flame
liftoff and blowout. The results were consistent with previous studies by Chung and
Lee [1, 2], which emphasized the importance of Schmidt number (Sc) in the liftoff
process and observed that lifted flames can be stabilized for Sc < 0.5 or Sc > 1.
In recent times, oxygen-enhanced combustion (OEC) and the use of oxyfuels
have gained significant amount of interest for their potential to reduce soot emis-
sions and easier CO2 capture [19–22]. On the other hand, there are very few studies
that highlighted the significance of OEC strategy on liftoff and blowout characteris-
tics [23–26]. Pickett et al. [23] in diesel spray flames observed that the liftoff length
was inversely proportional to the ambient oxygen concentration. Xue and Ju [26]
through experiments examined the influence of fuel-bound oxygen on the liftoff
phenomenon using di-methyl ether (DME), oxygen-diluted propane, and n-butane
jet diffusion flames. It was observed that DME flames required higher flow veloci-
ties for blowout. Further, an increase in oxygen addition, both propane and n-butane
flames changed from direct liftoff regime to direct blowout regime. Chung et al. [27]
examined auto-ignited laminar DME flames with high-temperature co-flow air to
observe a unique non-monotonic behavior in liftoff height with jet velocity. Chen
and Axelbaum [24] investigated the extinction phenomena under counterflow flames
through a unique strategy of using oxygenated combustion. The strategy involved
simultaneous O2 -enrichment of oxidizer stream and N2 dilution of fuel steam, such
that the stoichiometric mixture fraction was increased without altering the adia-
batic flame temperature. It was observed that under counterflow configuration, the
oxygenated flames were significantly more resistant to extinction, as their extinction
strain rates were an order of magnitude higher than those without oxygenation. In
subsequent studies, this oxygenation strategy has been examined under counterflow
and co-flow conditions [19, 21, 28–30] for its potential to reduce polycyclic hydro-
carbons (PAHs) and soot emissions. However, the effect of such strategy on liftoff
and blowout phenomena in jet flames has not been examined in previous studies.
Motivated by this, the present study aims to numerically investigate the effect of
this oxygenation strategy along with CO2 dilution on the flame liftoff and blowout
behavior. Simulations are performed using a time accurate, implicit algorithm with
detailed chemistry and transport models. The oxygenated ethylene flames are estab-
lished in co-flow jets by removing a specified amount of N2 from the oxidizer jet
and introducing it into the fuel jet. For each oxygenated flame, its liftoff and blowout
behavior is characterized by using an increasing amount of CO2 dilution in the fuel
jet. Thus, at a fixed oxygenation level or stoichiometric mixture fraction, the diluent
concentration in the fuel jet is increased until the blowout of lifted flame occurs.
The detailed flame structure and its propagation characteristics during the liftoff and
blowout process are investigated for different oxygenation levels. It is important to
note that most of the results in this chapter have been previously reported in an AIAA
conference paper [30].
276 R. Kumar et al.

2 Computational Model

Ethylene laminar jet flames in a co-flow configuration are simulated using a time-
dependent, axisymmetric algorithm, the UNICORN (Unsteady Ignition and Combus-
tion using Reactions) [31, 32]. The algorithm numerically solves the continuity, axial-
and radial-momentum, energy, and species conservation equations on a staggered
grid system. The finite difference forms of the momentum equations are obtained
using an implicit QUICKEST (Quadratic Upstream Interpolation for Convective
Kinematics with Estimated Streaming Terms) scheme and those of the species and
energy equations using a hybrid scheme of upwind and central differencing [33].
The pressure field is computed by solving the discrete form of the pressure Poisson
equation at all the grid points simultaneously by using the LU (Lower and Upper
diagonal) matrix-decomposition technique.
The general form of the conservation equations in cylindrical coordinates can be
represented as:
   
∂(ρ) ∂(ρu) ∂(ρv) ∂ ∂ ∂ ∂
+ + =  + 
∂t ∂r ∂z ∂r ∂r ∂z ∂z

ρu  ∂
− + + S (1)
r r ∂r

where  is the variable,ρ is the density, t the time, u and v represent the radial (r )
and axial (z) velocity components, respectively. The transport coefficients   and
source term S  in the above equation are provided in Table 1 of Ref. [34], and further
details are discussed in Ref. [35]. Further, a sink term based on an optically thin gas
assumption is included in the energy equation to account for thermal radiation from
the flame [36]. This is expressed as qrad = −4σ K p (T 4 − T04 ) [37], where T denotes
the local flame temperature, and K p accounts for the absorption and emission from
participating species (CO2 , H2 O, CO, CH4 and soot particles), and is expressed
as K p = P X i K p,i , where K p,i denotes the mean absorption coefficient of kth
species, σ the Stefen–Boltzmann coefficient constant, and T0 the ambient temper-
ature. The values of K p,i are calculated using a polynomial approximation to the

Table 1 Input values of species mole fractions for various oxygenated flames
CASES—percentage Fuel stream Air stream Stoichiometric
of N2 removed from air C H mole N2 mole O2 mole N2 mole mixture fraction
2 4
(%) fraction fraction fraction fraction (f st )

0 1 0.0 0.21 0.79 0.064

25 0.262 0.738 0.262 0.738 0.243
50 0.151 0.849 0.347 0.653 0.423
100 0.081 0.919 1.0 0.0 0.782
Effect of Oxygenation on the Liftoff, Stabilization, and Blowout … 277

Fig. 1 Schematic of the

computational domain and
boundary conditions.
Adopted from [30]

experimental data provided in Ref. [37]. The thermodynamics and transport proper-
ties in these equations are considered as temperature and species dependent. First,
the species enthalpies are calculated using the polynomial curve fits developed for
the temperature range 300–5000 K. The methodology for calculating the molecular
viscosity, thermal conductivity, and the binary diffusion coefficients of the species,
and then of the mixture is described in Ref. [15].
Figure 1 illustrates the computational domain simulated in the present study. The
domain extends 200 and 50 mm in axial (Z) and radial (r) directions, respectively.
In addition, an isothermal insert of 2 × 0.38 mm is used to simulate the fuel nozzle
wall. A grid-independent study has been performed to understand the dependency of
grid on computed results. Figure 2 presents a comparison of computed temperature
profiles along the center line for three different grid systems and with respect to
experimental data. It is observed that the computed temperature profiles are nearly
independent of the grid system; thus, a 401 (Z) × 101(r) grid system will be adopted
for rest of the simulations. The outflow boundaries are located sufficiently far from
the flame region to minimize the effect of enclosure on the predictions. The inner jet
contains fuel and diluent (N2 and CO2 ) mixture, while the outer jet issues oxidizer.
The velocity of the inner and outer jets is 35 cm/s and is fixed for all the cases
investigated. As mentioned, oxygenated ethylene flames are established by transfer-
ring a specified amount of N2 from the outer jet into the fuel jet. In other words,
the oxidizer stream is made richer in oxygen while simultaneously diluting the fuel
278 R. Kumar et al.

Fig. 2 Comparison of
experimental (symbols) and
predicted (solid lines)
temperature profiles along
the central line at different
grid sizes. For all the cases,
the flame is computed using
the same initial conditions.
Adopted from [30]

stream with N2 . Consequently, the stoichiometric mixture fraction (f st ), (given by

Eq. (2)) changes. In addition, for a given f st , the diluent (CO2 ) concentration in
the fuel jet is slowly increased in order to analyze the flame liftoff, stabilization, and
blowout processes. The ethylene flame chemistry is modeled using the USC II mech-
anism [38] with 111 species and 784 reactions. Table 1 provides details including
species mole fraction in the fuel and oxidizer streams for base (or CO2 undiluted)
oxygenation levels, along with the f st values. Table 2 provides the N2 and CO2 mole
fractions in the fuel stream for all cases at various diluent concentrations simulated.
  −1
Y F MW O ϑ O
f st = 1 + (2)
Y O MW F ϑ F

Table 2 Mole fractions of N2 and CO2 in the fuel stream for various diluted flames
Cases 0% Oxygen 25% Oxygen 50% Oxygen 100% Oxygen
enrichment enrichment enrichment enrichment
X N2 X CO2 X N2 X CO2 X N2 X CO2 X N2 X CO2
D1 0 0 0.738 0 0.849 0 0.919 0
D2 0 0.5 0.585 0.207 0.738 0.131 0.849 0.075
D3 0 0.67 0.485 0.344 0.653 0.231 0.790 0.140
D4 0 0.75 0.385 0.478 0.556 0.345 0.738 0.196
D5 0 0.80 0.339 0.541 0.506 0.404 0.693 0.246
D6 0 0.83 0.303 0.590 0.485 0.430 0.653 0.289
D7 0 0.85 0.295 0.601 0.465 0.453 0.634 0.309
Effect of Oxygenation on the Liftoff, Stabilization, and Blowout … 279

where Y f and Y O are mass fractions of fuel and oxygen at the respective boundaries,
ϑ F and ϑ O are the stoichiometric coefficients for fuel and oxygen, and MW stands
for molecular weight.

3 Results and Discussion

3.1 Validation of Computational Model

The computational algorithm and the reaction mechanism have been extensively
validated previously for a variety of steady and unsteady flames including opposed-
jet diffusion flames [39], burner-stabilized flames [8, 40], and lifted jet flames [41].
As discussed in cited studies, there is generally good agreement between predic-
tions and measurements in terms of flame topology including the locations of the
various reaction zones (i.e., the rich premixed (RP), lean premixed (LP), and non-
premixed (NP) zones), and the liftoff heights. We herein provide another validation
against the experimental data of nitrogen-diluted ethylene diffusion flames reported
by Smooke et al. [42]. Figure 3 compares the measured and predicted temperature
and acetylene profiles along the centerline for two N2 -diluted ethylene jet flames. It
is observed that the computational model captures the experimental data, both quali-
tatively and quantitatively to a good extent. The computational model reproduces the
rapid rise and subsequent decrease in temperature. However, there are some differ-
ences between measured and predicted peak temperatures, which may be attributed
to the over prediction of radiation losses by the optically thin model. Figure 3b
compares the measured and predicted acetylene profiles along the centerline for the
two flames. Again, the computational model reasonably captures the main features
of the flame and the changes in the acetylene concentration with N2 dilution. Thus,
overall, the computational model is able to reproduce the experimental flame data in
a satisfactory manner.

3.2 Effects of Oxygenation and Dilution on the Flame

Structure

Figure 4 presents the computed flame structures for various levels of oxygenation (see
Table 1). For each case, the flame structure is shown in terms of heat release contours
(q) and velocity vectors (v). The stoichiometric mixture fraction (f st ) line for each
flame is shown by a black line. As the oxygenation level or f st is increased, the flame
height decreases considerably which can be attributed to the increased availability
of oxygen combined with the reduced availability of fuel. In other words, at higher
values of f st , the required stoichiometric condition for combustion is achieved at a
shorter height. In addition, with increase in f st , the heat release rate at the flame
280 R. Kumar et al.

Fig. 3 Comparison of
experimental (symbols) [42]
and predicted (solid lines)
(a) temperature and (b)
acetylene mole fraction
profiles along the central line
in N2 -diluted ethylene/air
co-flow jet flames with 40%
and 60% ethylene in the fuel
jet. Adopted from [30]

base increases, while for all f st conditions considered, the flame remains attached
to the burner. These changes associated with increase in oxygenation on the burner-
stabilized flame height and structure, especially near the base region, have significant
implications on the flame liftoff behavior, as elaborated in the following sections.
Figure 5 presents the topologies/structures of various oxygenated flames for
different CO2 dilution levels. Again, the flame structure is presented in terms of
heat release rate contour, velocity vectors, and stoichiometric mixture fraction line.
Figures 5a–d correspond to cases with f st = 0.064, 0.243, 0.423, and 0.782, respec-
tively. For any given f st , increase in CO2 dilution makes the flame to liftoff from the
burner and stabilize at a certain height. The liftoff height is observed to be in the low
velocity region that extends from the wake of the burner rim. It is also observed that
the flame remains attached to the burner until the dilution levels (sum of N2 and CO2
mole .fractions) of 66%, 81%, 88%, and 93% for the four oxygenation cases: f st =
0.064, 0.243, 0.423, and 0.782, respectively. Further, any additional increase in CO2
Effect of Oxygenation on the Liftoff, Stabilization, and Blowout … 281

Fig. 4 Predicted heat release rate contours and velocity vectors for various oxygenated flames with
f st = 0.064, 0.243, 0.423, and 0.782 (Table 1). The dark black line represents the stoichiometric
mixture fraction (f st ). Adopted from [30]

dilution, results in a sudden increase in liftoff height, followed by flame blowout.

For instance, for the flame (a) with f st = 0.064, 75% dilution lifts the flame base to a
height of 12 mm which then moves rapidly to a height of 41 mm as the dilution level
increases to 83%. The flame blows out for a dilution level of 85%. In addition, for
all cases with different f st , the lifted flame at lower heights exhibits a double flame,
containing rich premixed (RP) and non-premixed (NP) reaction zones. With increase
in the dilution level (which results in higher flame heights), the mixing in the wake
region above the burner rim is enhanced, which allows for increased entrainment of
oxidizer into the fuel stream. As a consequence, the lifted flame structure changes
from double to triple flames. For f st = 0.782 flame, the rich premixed zone is weaker
and appears as a merged flame due to high dilution rate in the fuel stream. In addition,
the lifted flames become increasingly more sensitive to CO2 dilution as the oxygen
enrichment level is increased, which is discussed in the next section.
In order to characterize the effect of oxygenation on the flame structure, we present
in Fig. 6 the state relations for two extreme cases (f st = 0.064 and 0.782, or 0% and
100% oxygenation) in terms of the profiles of major reactant and product species
(C2 H4 , O2 , H2 O, and CO2 ) and intermediate fuel species (H2 and CO) with respect
to the mixture fraction (f ) at an axial location 2 mm above the burner. Here f = 1
and 0 indicate the fuel and oxidizer sides, respectively. Following Bilger [43], the
mixture fraction Z(F) is defined as:
282 R. Kumar et al.

Fig. 5 Predicted heat release rate contours and velocity vectors for various oxygenation
 levels,
 f st
= 0.064 (a), 0.243 (b), 0.423 (c), and 0.782 (d), and various dilution levels X CO2 and X N2 . The
dark black line represents the stoichiometric mixture fraction (f st ). Adopted from [30]

 MC  Mh
Z (F) = NCi Yi + Nhi Yi
Mi Mi

Here, N Ci and N hi are the number of carbon and hydrogen atoms, respectively, in
species i, Y i the mass fraction of species i, M the molecular weight, and summation
is over all species. Since the fuel or oxidizer stream may also contain diluent, Z can
be normalized as follows:
[Z (F) − Z (F, l)]
f =
[Z (F, r ) − Z (F, l)]

where subscripts r and l correspond to fuel and oxidizer streams, respectively.

Effect of Oxygenation on the Liftoff, Stabilization, and Blowout … 283

Fig. 6 State relationships in terms of scalar profiles with respect to mixture fraction (f ) for two
flames with f st = 0.064 (a) and f st = 0.782 (b), but without CO2 dilution. The vertical dashed line
represents the stoichiometric mixture fraction (f st ) value. Adopted from [30]

Figure 6 depicts a typical non-premixed flame structure in the mixture fraction

coordinate for both flames. The reaction zone is located by the peak values of heat
release rate (HRR), CO2 , and H2 O mole fractions near f = f st . In addition, it is
indicated that as the level of oxygenation is increased, the stoichiometric mixture
fraction (f st ) increases from f st = 0.0637 to 0.7817 for the 100% O2 -enriched case,
and the flame location moves from the oxidizer to fuel side. The increased f st value
also implies that mixing required to establish the flame is reduced compared to the
case with lower f st value.

3.3 Effect of Dilution on Flame Liftoff and Blowout

Figure 7 presents the variation in flame liftoff height with respect to diluent mole frac-
tion for all four f st cases. It is observed that, with addition of diluent, the flame liftoff
height for 0% oxygenated flame initially increases in a gradual, nonlinear fashion,
and then followed by a rapid increase until the blowout occurs (at X CO2 = 0.85).
Further, these results for the flame without oxygenation (f st = 0.064) are consistent
with previous experimental data [14] in terms of CO2 mole fraction required for
blowout of ethylene/air flames. In short, this again underlines the capability of the
computational algorithm and the chemical mechanism used in this study. In contrast
to 0% oxygenated flames, the liftoff height for oxygenated flames (25, 50 and 100%)
increases very rapidly after flame liftoff, indicating greater sensitivity
 of the flame
 to
dilution level. Further as shown in Fig. 7, the critical dilution X CO2 + X N2 levels
for extinction are 5%, 89%, 92%, and 94% for 0%, 25%, 50%, and 100% oxygen-
enriched flames, respectively. Critical values of CO2 and N2 dilutions for blowout
284 R. Kumar et al.

Fig. 7 Liftoff height plotted

as a function of fuel-stream
dilution (CO2 + N2 ) for
various oxygenated flames;
0%, 25%, 50%, and 100%
correspond to oxygenation
level by removing N2 from
air stream and adding it to
fuel stream as indicated in
Tables 1 and 2. Adopted
from [30]

are provided in Table 2 (Case: D7). In addition, with the increase in the level of
oxygenation, flames tend to blowout at a lower height and the range of dilution from
liftoff to blowout reduces drastically.
The stabilization of lifted non-premixed flames has generally been explained
based on the dynamic balance between a triple flame speed and local flow velocity.
This stabilization is based on the existence of a triple flame structure at the flame
base, which is more appropriate for flames that are stabilized in the far field. For
lifted flames stabilized in the near field, Briones and Aggarwal [15] and Takahashi
and Katta [44] hypothesized a stabilization mechanism based on a dynamic balance
between the characteristic reaction rate and scalar dissipation rate. To examine this
hypothesis in the context of present study, we have computed the scalar dissipation
rate for various dilution and oxygenation levels. The scalar dissipation rate is defined
as χ = 2D F−mix (∇ f )2 , where f is the mixture fraction and D f −mix is the fuel
diffusivity with respect to the mixture.
Figure 8 presents the variation of local χ with respect to axial position along the
fst line for various oxygenated flames (D6 case in Table 2) near blowout (extinction)
conditions. For each oxygenated flame, the flame base is located at the point of
minimum χ value. The enhanced blowout characteristics of oxygenated flames can be
explained by an increasingly lower value of the scalar dissipation rate (χ ), implying
stronger reactivity as the amount of oxygenation is increased.
Effect of Oxygenation on the Liftoff, Stabilization, and Blowout … 285

Fig. 8 Scalar dissipation

rate (χ) plotted versus the
axial position (z) along the
stoichiometric mixture
fraction line for flames near
blowout. 0, 25, 50, and 100%
correspond to oxygenation
level by removing specified
amount of N2 from oxidizer
stream and adding it to fuel
stream. Adopted from [30]

3.4 Blowout (Extinction) Characteristics of Oxygenated

Co-flow Flames

Several studies have highlighted the importance of using the peak flame temperature
(T max ) as an indicator of flame strength or its resistance to extinction [11, 14]. Figure 9

Fig. 9 Peak flame

temperature plotted as a
function of (CO2 + N2 )
dilution for various
oxygenated flames. Adopted
from [30]
286 R. Kumar et al.

plots T max as a function of fuel-stream dilution for the four oxygenation cases. For
each case, T max decreases as the amount of dilution is increased, and the minimum
T max reached indicates flame condition close to extinction. For the base case (f st =
0.064), the peak flame temperature decreases more gradually as the amount of dilution
is increased. In contrast, at higher oxygenation levels, T max decreases more rapidly
with dilution, with the amount of dilution required for flame extinction or blowout
increasing with f st . However, for all the four cases, the predicted temperatures near
blowout are within a narrow range of 1680–1730 K. Further, for the base case, T max
exhibits a nonlinear behavior because the diluent has two effects: the dilution effect
and the thermal effect that cools the flame. On the other hand, for oxygenated flames,
T max decreases almost linearly, with the thermal effect of CO2 dilution being more
prominent. In oxygenated flames, thermal effect of CO2 is more prominent than the
dilution effect of either CO2 or N2 . For instance, for the flame with 100% enrichment,
as CO2 dilution is increased from X CO2 = 0.246 to 0.289, T max decreases from 1763
to 1682 K. Any further increase in CO2 causes the flame blowout or extinction.

4 Conclusions

A computational investigation on the flame liftoff, stabilization, and blowout char-

acteristics of oxygenated non-premixed ethylene/air jet flames has been carried out.
A time accurate, implicit algorithm that uses a detailed description of the chem-
istry and includes radiation effects is used for this study. The oxygenated flames are
established by simultaneously using an O2 enriched oxidizer jet along with a N2
diluted fuel jet in such a way that the adiabatic flame temperature is not materially
changed. The flame liftoff and blowout processes are then examined by using CO2
dilution in the fuel jet. The computational algorithm and chemical mechanism have
been validated using the experimental data on flame structure reported in terms of
temperature and species profiles for diluted ethylene/air jet flames [42]. Important
observations are as follows,
1. Results indicate burner-attached flames for all levels of oxygen enrichment,
but without CO2 dilution. As CO2 dilution is increased to a critical value, the
flame lifts off from the base and stabilizes in the near-field region. With further
increase in dilution, the lifted flame gets stabilized further downstream of burner
rim in a low velocity region. With still further increase in CO2 dilution, the liftoff
height rapidly increases until the blowout occurs. The amount of critical dilution
required for extinction (blowout) increases with the increase in oxygenation

level. For instance, the predicted critical dilution X N2 + X CO2 values for
oxygenation levels of 0%, 25%, 50%, and 100% are 0.85, 0.89, 0.92, and 0.94,
respectively, highlighting the increased flame resistance to extinction. Also, for
the 0% enrichment case, there is a good agreement between the measured and
predicted values of CO2 X CO2 = 0.85 required for flame blowout.
Effect of Oxygenation on the Liftoff, Stabilization, and Blowout … 287

2. As the oxygen enrichment is increased, it becomes more difficult to establish

lifted flames. However, once the flame is lifted, its liftoff height increases rapidly
with the increase in CO2 dilution, i.e., a lifted oxygenated flame exhibits greater
sensitivity to CO2 dilution.
3. The predicted range of dilution for stable lifted flames reduces monotonically
with the increase in oxygen enrichment. For the flame with 100% oxygen enrich-
ment, the flame lifts off at dilution level of 93% dilution and blows out at 94%
dilution, indicating a stable range of only 1%, as compared to a value of 19%
for the non-oxygenated flame.
4. The stabilization mechanism for lifted flames in the near-field region can be
described by a balance between the reaction rate and local scalar dissipation
rate.

Acknowledgements Computations were performed using the UNICORN code provided by Dr.
Viswanath R. Katta. His assistance throughout this study is gratefully acknowledged.

References

1. Chung SH, Lee BJ (1991) On the characteristics of laminar lifted flames in a nonpremixed jet.
Combust Flame 86:62–72
2. Lee BJ, Chung SH (1997) Stabilization of lifted flames tribrachial flames in a laminar
nonpremixed jet. Combust Flame 109:163–172
3. Chen YC, Bilger RW (2000) Stabilization mechanism of lifted laminar flames in axisymmetric
jet flows. Combust Flame 122:377–399
4. Ghosal S, Vervisch L (2001) Stability diagram for lift-off and blowout of a round jet laminar
diffusion flame. Combust Flame 123:646–655
5. Takahashi F, Schmoll WJ, Katta VR (1998) Attachment mechanism of diffusion flames. Proc
Combust Inst 27:675–684
6. Takahashi F, Katta VR (2000) A reaction kernel hypothesis for the stability limit of methane
jet diffusion flames. Proc Combust Inst 28:2071–2078
7. Ruetsch GR, Vervisch L, Linan A (1995) Effects of heat release on triple flames. Physics of
Fluid 7:1447–1454
8. Azzoni R, Ratti S, Puri IK, Aggarwal SK (1999) Gravity effects on triple flames: flame structure
and flow instability. Phys Fluids 11:3449–3464
9. Lee BJ, Kim JS, Chung SH (1994) Effect of dilution on the liftoff of non-premixed jet flames.
Proc Combust Inst 25:1175–1181
10. Chung SH (2007) Stabilization, propagation and instability of tribrachial triple flames. Proc
Combust Inst 31:877–892
11. Aggarwal SK (2009) Extinction of laminar partially premixed flames. Prog Energy Combust
35:528–570
12. Qin X, Puri IK, Aggarwal SK (2002) Characteristics of lifted triple flames stabilized in the
near field of a partially premixed axisymmetric jet. Proc Combust Inst 29:1565–1572
13. Lock A, Briones AM, Aggarwal SK, Puri IK, Hegde U (2007) Liftoff and extinction
characteristics of fuel- and air-stream-diluted methane-air flames. Combust Flame 149:340–352
14. Lock A, Aggarwal SK, Puri IK (2009) Effect of fuel type on the extinctions of fuel and air
stream diluted partially premixed flames. Proc Combust Inst 32:2583–2590
288 R. Kumar et al.

15. Briones AM, Aggarwal SK (2006) A numerical investigation of flame liftoff, stabilization, and
blowout. Phys Fluids 18:043603–1–11
16. Choi BC, Kim KN, Chung SH (2009) Autoignited laminar lifted flames of propane in coflow
jets with tibrachial/edge and mild combustion. Combust Flame 156:396–404
17. Kim KN, Won SH, Chung SH (2007) Characteristics of laminar lifted flames in coflow jets
with initial temperature variation. Proc Combust Inst 31:947–954
18. Choi BC, Chung SH (2010) Autoignited laminar lifted flames of methane, ethylene, ethane
and n-butane jets in coflow air with elevated temperature. Combust Flame 157:2348–2356
19. Skeen SA, Yablonsky G, Axelbaum RL (2010) Characteristics of non-premixed oxygen-
enhanced combustion: II Flame structure effects on soot precursor kinetics resulting in soot-free
flames. Combust Flame 157:1745–1752
20. Kalvakala KC, Katta VR, Aggarwal SK (2018) Effects of oxygen-enrichment and fuel unsat-
uration on soot and NOx emissions in ethylene, propane and propene flames. Combust Flame
187:217–229
21. Kalvakala KC, Aggarwal SK (2021) Soot emission reduction in oxygenated co-flow jet flames.
Proc Combust Inst 38
22. Bhatia P, Katta VR, Krishnan SS, Zheng Y, Sunderland PB, Gor JP (2012) Simulations of
normal and inverse laminar diffusion flames under oxygen enhancement and gravity variation,
Combut Theory Model 16(5):774–798
23. Siebers D, Higgins B, Pickett L (2002) Flame lift-off on direct injection diesel fuel jets: Oxygen
concentration effects, SAE Transactions. J Engines 111(3):1490–1509
24. Chen R, Axelbaum RL (2005) Scalar dissipation rate at extinction and the effect of oxygen-
enriched combustion. Combust Flame 142:62–71
25. Zhou J, Ai Y, Kong W (2013) The liftoff properties of dimethyl ether jet diffusion flames with
preheating, Proc. of ASME Turbo Expo 2013: Turbine Technical Conference and Exposition,
1B, 2013 V01BT04A020
26. Xue Y, Ju Y (2006) Studies on the liftoff properties of dimethyl ether jet diffusion flames.
Combust Sci Tech 178(12):2219–2247
27. Al-Noman SM, Chul CB, Chung SH (2018) Autoignited lifted flames of dimethyl ether in
heated coflow air. Combust Flame 195:75–83
28. Kalvakala KC, Aggarwal SK (2019) Polycyclic aromatic hydrocarbons and soot emissions in
oxygenated ethylene diffusion flames at elevated pressures. J Eng Gas Turb Power 141:Article
071022
29. Kumfer BM, Skeen SA, Chen R, Axelbaum RL (2006) Measurement and analysis of soot
inception limits of oxygen enriched co-flow flames. Combust Flame 147:233–242
30. Kumar R, Katta VR, Aggarwal SK (2018) On the effects of oxygen-enrichment on flame liftoff,
stabilization and blowout in ethylene flames. In: 2018 AIAA aerospace sciences meeting,
AIAA-2018–1478
31. Katta VR, Goss LP, Roquemore WM (1994) Effect of nonunity Lewis number and finite rate
chemistry on the dynamics of a hydrogen air jet diffusion flame. Combust Flame 96:60–74
32. Roquemore WM, Katta VR (2000) Role of flow visualization in the development of UNICORN.
J Visualization 2:257–272
33. Leonard BP (1979) A stable and accurate convective modeling procedure based on quadratic
upstream interpolation. Comput Methods Appl Mech Engg 19:59–98
34. Shu Z, Aggarwal SK, Puri IK, Katta VR (1997) A numerical investigation of the flame structure
of an unsteady inverse partially premixed flame. Combust Flame 111:295–311
35. Shu Z, Aggarwal SK, Puri IK, Katta VR (1997) Flame-Vortex Dynamics In An Inverse Partially
Premixed Combustor; The Froude Number Effects. Combust Flame 111:276–295
36. Qin X, Puri IK, Aggarwal SK, Katta VR (2004) Gravity, radiation, and coflow effects on
partially premixed flames. Phys Fluids 16:2963–2974
37. Siegel R, Howell JR (1981) Thermal radiation heat transfer. Hemisphere Publishing Corpora-
tion, New York
38. Wang H, You X, Joshi AV, Davis SG, Laskin A, Egolfopoulos F, Law CK (2007) USC Mech.
version II. High-temperature combustion reaction model of H2 /CO/C1 -C4 compounds. http://
ignis.usc.edu/USC_Mech_II.htm
Effect of Oxygenation on the Liftoff, Stabilization, and Blowout … 289

39. Katta VR, Meyer TR, Brown MS, Gord JR, Roquemore WM (2004) Extinction criterion for
unsteady, opposing-jet diffusion flames. Combust Flame 137:198–221
40. Shu Z, Krass BJ, Choi CW, Aggarwal SK, Katta VR, Puri IK (1998) An Experimental and
numerical investigation of the structure of steady two-dimensional partially premixed methane-
air flames. Proc Combust Inst 27:625–632
41. Katta VR, Takahashi F, Linteris GT (2004) Suppression of cup-burner flames using carbon
dioxide in microgravity. Combust Flame 137:506–522
42. Smooke MD, Hall RJ, Colket MB, Fielding J, Long MB, McEnally CS, Pfefferle LD (2004)
Investigation of the transition from lightly sooting towards heavily sooting co-flow ethylene
diffusion flames. Combust. Theory Model. 8:593–606
43. Bilger RW (1989) The structure of turbulent nonpremixed flames. Proc Combust Inst 22:475–
488
44. Takahashi F, Katta VR (2009) A reaction kernel hypothesis for the stability limit of methane
jet diffusion flames. Proc Combust Inst 28:2071–2078
Laminar Burning Velocity Measurements
at Elevated Pressure and Temperatures
and the Challenges in Kinetic Scheme
Optimization

Robin John Varghese and Sudarshan Kumar

Abstract Reliable data on fundamental combustion parameters are essential to vali-

date kinetic schemes under a wide range of experimental conditions. Laminar burning
velocity is one such fundamental intrinsic property of a flammable mixture. The chal-
lenges in determining the effect of simultaneous change in temperature and pressure
on the burning velocity of low-calorific value syngas–air mixtures are presented.
Comparing the present experiments with available kinetic schemes at elevated
temperature and pressure reveals poor prediction capabilities. The major contributing
factor to this discrepancy is the high sensitivity of key reactions at high temperatures
and pressure conditions. The challenges of direct use of the experimental data to
optimize kinetic schemes for the syngas–air mixtures are discussed.

Keywords Burning velocity · Optimization · Elevated temperatures and

pressures · Diverging channel

Nomenclature

α Temperature exponent
αt Thermal diffusivity
β Pressure exponent
δ Flame thickness
φ Equivalence ratio
ρ Mixture density
Su Laminar burning velocity
χH2 Hydrogen content

R. J. Varghese (B) · S. Kumar

Department of Aerospace Engineering, Indian Institute of Technology, Bombay, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 291
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_13
292 R. J. Varghese andS. Kumar

Abbreviations

FFCM Foundational fuel chemistry model

GRI Gas research institute
IGCC Integrated gasification and combined cycle
LBV Laminar burning velocity
LCV Low-calorific value

1 Introduction

The use of fossil fuels has been on the rise since the industrial revolution. Preser-
vation of resources demands efficient combustion of fossil fuels, and the protec-
tion of the environment stresses a restraint on the pollutant emissions. Environment
(shrinking oil reserves) and ecology (climate change, air quality) concern over the
past decades has improved awareness to shift the dependence on these fuels to alter-
native sources for various power generation needs and transportation. However, the
alternative energy source should primarily (a) decrease fossil energy use, (b) reduce
pollutant emissions, (c) sustainable, (d) scalable. Synthesis gas derived from coal
or biomass is considered a viable alternative for stationary power generation, and
biomass gasification results in a mixture of H2 and CO gas with a significant quan-
tity of inert diluents. The gasification product is characterized by large compositional
variability and low-calorific value (LCV). The primary combustion parameters like
burning velocity and ignition delay of these LCV syngas mixtures are not available in
the literature at high temperature and pressure conditions. However, they are essential
for the integrated gasification and combined cycle (IGCC) power plants to efficiently
use the resources.
Laminar-burning velocity (LBV) is a fundamental combustion entity that char-
acterizes combustible mixtures’ overall reactivity. LBV impacts the stabilization of
flames which directly influences the limits of flashback and helps validate kinetic
schemes. LBV is also one of the critical parameters that model the combustion of
fuels in the spark-ignition engines.
Various stationary and dynamic flame methods are available for the determination
of LBV. LBV data on low-calorific value syngas–air mixtures are limited, and exper-
imental research on simultaneous variation of temperature and pressure for LBV
determination is non-existent. Most syngas–air mixtures’ investigations focus on the
effect of single diluents [1–4] or different concentrations of carbon monoxide [5].
LBV is available for the syngas–air system with a considerable variation in H2 /CO
ratios at different equivalence ratios for various N2 / CO2 dilution [2, 3, 6–9]. Yan
et al. [10] reported LBV of three variants of syngas–air mixtures derived from a
gasification plant using the heat-flux method at 298–358 K.
Laminar Burning Velocity Measurements at Elevated Pressure … 293

Shang et al. [11] investigated the effect of inert diluents (CO2 /N2 ) on LBV of
producer gas mixtures using the Bunsen flame method. The effect of N2 was deter-
mined to be milder in comparison with the same volume of CO2 . Similarly, Natarajan
et al. [2] reported an LBV of syngas mixtures at various temperatures and pressures.
The performance of available reaction mechanisms at high-temperature conditions
(>450 K) was comparatively inaccurate. Han et al. [12] reported LBV of syngas
mixtures at elevated pressures and temperatures for different hydrogen content (χH2
= 0.25, 0.5, 0.75), with spherically expanding flames. Natarajan et al. [2] and Han
et al. [12] report the only available literature for LBV of syngas–air mixtures, where
both the change in temperature and pressure were considered individually. The effect
of the simultaneous change in the mixture temperature and pressure on the laminar
burning velocities of syngas–air mixtures was not considered [2, 12]. This chapter
presents LBV for a simultaneous variation of temperature and pressure for syngas–air
mixtures and the significance of this consideration for optimizing reaction chemistry.
The synthesis gas compositions considered in the present study are similar to
the output of the biomass gasifiers and are described by large quantity (60–80%)
of diluents (CO2 and N2 ) [13], and the remaining volume (40–20%) constitutes of
combustible gases. Productive use of these compositions in IGCC combustors merits
an accurate knowledge of LBV at high temperature and pressure conditions. The
limited research available on LBV of syngas–air mixtures does not account for the
compositional variability. The validation and optimization of kinetic schemes based
on LBV are limited at high temperatures and non-existent at elevated pressures for
the present study’s compositions.
The experimental method presented in the chapter is the preheated diverging
channel method. The in-house developed and perfected method in the Combustion
Laboratory of the Indian Institute of Technology Bombay has undergone consider-
able improvements. The most recent was the extension of the method to include a
simultaneous change in the thermodynamic conditions (Pressure and Temperature),
and the facility was validated using the LBV measurements of methane–air mixtures.
In this chapter, LBV of syngas–air mixtures at elevated temperatures and pressures are
presented, along with a modified power–law correlation for representing LBV at high
temperatures and pressures. The composition of the syngas–air mixture presented in
the chapter is referred to as SG from hereon, is H2 -15%, CO-15%, CO2 -15%, N2 -
55%. The kinetic scheme optimization challenges for the LCV syngas mixtures with
LBV at high temperature and pressures as the target variable for genetic algorithm
(GA)-based optimization is also discussed briefly.
294 R. J. Varghese andS. Kumar

2 Methods

2.1 Experimental Method

The experimental technique employed for the measurement of LBV in the present
work is the preheated diverging channel. The experimental section consists of a quartz
diverging channel of rectangular cross section, an infrared heater, and a gas supply
system controlled using a computer. The diverging quartz channel stabilizes planar
flames at elevated temperatures, and the infrared heater placed inside the cylindrical
pressure vessel heats the diverging channel to achieve a positive temperature gradient
along the flow direction. The uniform distribution of temperature and velocity profiles
along the transverse direction at various axial positions are presented in [5]. The
infrared heater also aids in compensating the heat loss from the flame to the channel
walls [5].
Figure 1 shows the relevant details of the diverging channel experimental section
along with the cylindrical pressure vessel. The pressure vessel is initially filled at a
specified pressure, and the relief valve mechanism used for regulating the pressure
inside the chamber is accurate up to ±0.1 atm. The infrared heater is placed below
the rectangular quartz channel and is switched on at the required heating rate. The
temperature inside the channel at various axial locations are measured using a k-
type thermocouple for various heating conditions. The fuel–air mixture is fed into
the channel at a specified inlet velocity and ignited at the channel’s exit using an
ignition device. The flame gradually moves upstream and stabilizes where the local
flow velocity is equal to the laminar burning velocity. LBV is determined from the
flame area and unburnt gas temperature at the flame stabilization location using a
rearranged mass conservation equation. The Peclet number and Reynolds number of
the flow inside the channel are low, and hence, the assumption that the unburnt gas
temperature is the same as the temperature of the quartz wall is reasonable [14, 15].

2.2 Computational Method

LBV and sensitivity of various pre-exponential factors of different elementary reac-

tions were determined using the freely propagating flame code (PREMIX). The
conservation equations and the calculation of species and temperature profiles of
PREMIX are discussed in detail in [16]. LBV prediction capabilities of Aramco 3
(581 species and 3037 reactions) [17–19], Davis mechanism (14 species and 28
reactions) [20] GRI-Mech 3.0 (53 species and 325 reactions) [21], and founda-
tional fuel chemistry model (FFCM-1) [22] (38 species and 291 reactions) were
analyzed initially. Comparing the four kinetic schemes’ predictions with the present
LBV measurements of syngas–air mixtures revealed FFCM-1 to be in closer agree-
ment until elevated temperatures of 500 K and 1 atm. This observation is consistent
Laminar Burning Velocity Measurements at Elevated Pressure … 295

Connected to AC/
Mains

Power
Connected to PC/Command Controller
Module
IgniƟon
Electrode

Channel T
Heater (1200 W)
High Pressure
vessel

Diverging Manual Signal

Mass Flow Valve Lines
Fuel

Channel
Air

Controller
T
Thermocouple Pressure Pressure Gas
Gauge Regulator Lines

Fig. 1 a Schematic of the experimental facility b High-pressure vessel 3D model—Sectional

view, A-Pressure relief valve, B-Glass window, C-Pressure gauge, D-Diverging channel, E-
Thermocouple, F-Infrared heater

with the conclusions of Natarajan et al. [2] for GRI-Mech 3.0 and Davis mecha-
nisms. However, for a simultaneous variation of temperature and pressure, the LBV
prediction competency of various kinetic schemes deteriorate. Therefore, a genetic
algorithm-based optimization for the best-performing kinetic scheme (FFCM-1)
was attempted. The target variables for the optimization and various constraints are
discussed in detail in a latter section. The base model for optimization consists of 38
296 R. J. Varghese andS. Kumar

species and 310 reactions. 19 reactions from Aramco 3 were merged to the FFCM-1
model. The Arrhenius parameters for four reactions were specified as logarithmic
functions of pressure, and the rate constants were linearly interpolated with respect
to the logarithm of pressure in between the specified pressures. This merged kinetic
scheme was used as the base model for optimization.

3 Results and Discussion

The SG–air mixture’s laminar burning velocities were measured using the diverging
channel technique at elevated temperatures (300–650 K) and pressures (1–5 atm).
The unique composition of the mixture and unavailability of literature at the ther-
modynamic conditions presented makes direct comparison challenging. Therefore,
the comparison with the best performing kinetic scheme (FFCM-1) and optimized
reaction scheme is presented in the following sections. However, the measured LBV
for various fuels such as CH4 [23], LPG [24], C3 H8 [24], H2 /CO/CO2 /N2 [5, 25–27],
C1–C5 alcohols [28, 29], DME [30], and methyl formate [31] using the diverging
channel technique were reported to be accurate at various conditions in comparison
with other methods. The method’s extension to elevated pressures and subsequent
validation of the technique using LBV of CH4 -air mixtures were reported recently
[32].

3.1 LBV at High Temperature and Pressure

The diverging channel facility for the measurement of LBV at elevated temperatures
and pressures was used for the syngas–air mixture (SG). Figure 2 shows the measured
LBV of stoichiometric SG–air at different temperatures (T u = 300–650 K) and
mixture pressures (Pu = 1, 2, 3, 5 atm).
The measured LBV at various temperatures increases with an increase in the
mixture enthalpy at a specified pressure. The increase in LBV with an increase in
temperature at various pressures is similar. For instance, the comparisons at T u = 450,
600 K indicates LBV is doubled for both Pu = 1, 5 atm. However, the extrapolated
values of S u,o, and power exponents of temperature (α) are distinct. S u,o decreases
with an increase in pressure, whereas temperature exponents (α) increase slightly.
The decrease in LBV with an increase in pressure may be attributed to the increased
density, higher third-body reaction effects, and the reduction in thermal diffusivity.
LBV predictions using the FFCM-1 base model and optimized model are also
compared with the measured values in Fig. 2. LBV predictions deviate from the
measured values at high temperature and pressure conditions for the base model.
The discrepancy in the predictions using the optimized model reduced considerably.
The uncertainty in the reaction rate parameters is significantly high, and optimization
Laminar Burning Velocity Measurements at Elevated Pressure … 297

Fig. 2 LBV of
1 atm
stoichiometric SG–air 1.25
2 atm
mixtures at elevated 3 atm
temperatures and pressures 5 atm
1.00 Power-law fit
Base Model

Su (m/s)
Optimized
0.75

0.50

-- 1 atm 3 atm 5 atm

0.25
Su,o (m/s) 0.17 ± 0.01 0.12 ± 0.01 0.09 ± 0.01
α 2.67 ± 0.06 2.74 ± 0.11 2.78 ± 0.08

1.5 1.6 1.7 1.8 1.9 2.0 2.1

Tu / Tu,o

would be substantially improved if quantum chemistry calculations are considered

to develop a base model for the reaction mechanism.

3.2 Optimization Approach

A genetic algorithm-based approach to optimize the rate parameters of the mech-

anism was employed. The target data for optimization was the measured burning
velocities of syngas–air mixtures at different equivalence ratios and various temper-
ature and pressure conditions. The overall relative squared error describes the objec-
tive function for minimization between the LBV obtained from experiments and
simulation. A complete set of the target parameters is provided in the supplemen-
tary material. The optimization algorithm’s objective is to match the target data with
the freely propagating flame speed predictions using each modified rate parameter.
The parameters’ uncertainty range was defined as the pre-exponential factor (A)
(ratio) 1.5, temperature dependence factor, n = ±0.2, and activation energy Ea =
±21 kJ/mol. The RMS error of the target data and flame speed predictions using the
back-end freely propagating flame code is initially calculated.
The procedure of optimization can be summarized as (1) 15 datasets are generated
randomly considering the specified uncertainties of reaction rate parameters as the
initial population; (2) The freely propagating flame code is used to calculate the flame
speeds using the initial mechanisms; (3) The RMS error of the function evaluation
and target data is defined as the objective function and is calculated from (2, 3). (4)
The convergence distance and percentage (0.1, 80%) of the objective function are
evaluated; (5) New population is generated, and mutation may occur during this step,
and flame speeds are calculated for the new population. (6) The objective function is
re-evaluated, the populations are merged, and best-fit mechanisms are selected as the
298 R. J. Varghese andS. Kumar

Table 1 Optimized reaction rate constants (Units: cal/mol)

Reaction number Reaction A n Ea
R4 H2 + OH = H + H2 O 4.4E+13 0 6990
R272 CH3 + H(+M) = CH4 (+M) 1.3E+16 −0.6 383
R280 CH3 + OH = CH2 (S) + H2 O 4.9E+14 −0.7 −446
R281 CH3 + OH = CH2 O + H2 3.5E+05 1.4 −3244
R282 CH3 + OH = CH2 OH + H 1.6E+10 1.0 3210
R283 CH3 + OH = H + CH3 O 1.2E+09 1.0 11,940
R284 CH3 OH(+M) = CH3 + OH(+M) 2.1E+18 −0.6 92,541
R285 CH3 OH(+M) = CH2 OH + H(+M) 7.9E−03 5.0 84,467
R301 CO + H2 (+M) = CH2 O(+M) 4.3E+07 1.5 79,600

next generation. (7) The steps are repeated until the predefined maximum generation
(2000) is reached. Table 1 shows the optimized rate constants of sensitive reactions.

3.3 Variation of Power Exponents of Temperature

and Pressure

The conventional power–law correlation available for laminar burning velocities can
be expressed as:
 α  β
Tu Pu
Su = Su,o (1)
Tu,o Pu,o

where the exponents of temperature and pressure are considered to be constants.

Figure 3a shows the power exponents of temperature determined from the
measured burning velocities at various pressures, along with a comparison with
a linear curve fit. The comparisons with the available kinetic schemes show a similar
increase in the power exponents with increased pressure. However, the values are
slightly underpredicted for both the optimized and base FFCM-1 models. This differ-
ence in the predictions after optimization indicates the need for quantum chemistry
calculations at high temperature–pressure conditions and expressing the reaction
factors in terms of the pressure range. The temperature exponents were considered
constant and independent of temperature range and the pressure they were measured.
Varghese et al. [32] proposed a modified power–law correlation representing temper-
ature exponents’ variation with a change in the mixture pressure. Here, in syngas–air
mixtures, a similar power–law correlation is reasonable for an accurate representation
of temperature exponents at various pressures.
Laminar Burning Velocity Measurements at Elevated Pressure … 299

2.9

α 2.8

2.7 SG1 - Present

φ = 0.9
φ = 1.0
φ = 1.1
φ = 1.25
Optimized
2.6 FFCM-1
Curve Fit

1 2 3 4 5
a)
Pu (atm)

-0.35

-0.40
β

-0.45 SG1 - Present

φ = 0.9
φ = 1.0
φ = 1.1
φ = 1.25
Curve Fit
-0.50
300 350 400 450 500 550 600
b)
Tu (K)

Fig. 3 a Power exponents of temperature (α) as a function of pressure and b pressure exponents
(β) at different temperatures

Similarly, the comparison of power exponents of pressure determined from the

measured LBV shows an increasing trend. The predictions of both the kinetic
schemes significantly underpredict the pressure exponents. Perhaps measurements at
higher pressures (5–20 atm) would be necessary to accurately calculate the pressure
exponents’ variation.
Figure 3 shows the temperature and pressure exponents at various pressures
and temperatures. Power exponents (α) increase linearly with an increase in pres-
sure, and pressure exponent (β) increases with temperature. This clearly shows that
300 R. J. Varghese andS. Kumar

the assumption of constant exponents is inaccurate when considering the simulta-

neous variation of thermodynamic conditions. Therefore, the pressure exponents are
expressed in terms of temperature ratios and, similarly, temperature exponents as a
linear function of pressure.
   
Pu Tu
α = αo + α1 1 − ; β = βo + β1 1 − (2)
Pu,o Tu,o

The conventional power–law correlation is modified considering the variations of

power exponents of temperature and pressure (Eq. 2), and Eq. 1 can be modified as:
   
 αo +α1 1− PPu,o
u  βo +β1 1− TTu,o
u
Tu Pu
Su = Su,o (3)
Tu,o Pu,o

3.4 Comparison of Optimized Model with Available Kinetic

Schemes

The diverging channel was used to measure LBV of syngas–air mixtures at elevated
temperatures (300–650 K) and pressures (1–5 atm). The LBV determined from the
experiments were used to evaluate the performance of available kinetic schemes and
the optimized reaction mechanism.
Figure 4 shows the predictions of five different kinetic schemes (GRI-Mech3.0,
Aramco3.0, FFCM-1 base model and optimized model) at T u = 450 K and elevated
pressures (Pu = 1 − 5 atm).

Fig. 4 Comparison of SG1-air

measured burning velocities 60
Present Tu = 450 K
at T u = 450 K and elevated
1 atm
pressures with various 50 2 atm
kinetic schemes 3 atm
5 atm
40
Su (cm/s)

30

20
Kinetic Schemes
Optimized
10 Base Model
GRI-Mech 3.0
Davis

0
0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
φ
Laminar Burning Velocity Measurements at Elevated Pressure … 301

LBV decreases with an increase in pressure at T u = 450 K. The mixture density

increases with an increase in pressure, reducing the mean free path for molecular
collisions and leading to enhanced third-body recombination reactions. The oxida-
tion rate also depends on the change in thermal diffusivity at higher pressures and
contributes to the slower flame propagation at elevated pressures. The mixture pres-
sure increase indicates a peak shift of LBV from rich mixtures to near stoichiometric
mixtures. For instance, at Pu = 1 atm, the peak burning velocity is present for very
rich mixtures (φ = 1.4), whereas, at 5 atm, the peak burning velocity occurs for stoi-
chiometric mixtures. The flame structure of stoichiometric mixtures at two different
pressures (Pu = 1, 5 atm) is compared to provide further insight into the peak shifting
of syngas–air mixtures at elevated pressures.
The prediction accuracy of various kinetic schemes has improved from subse-
quent optimizations. FFCM-1 based model may be considered the best-performing
reaction mechanism available for methane–air mixtures when a simultaneous varia-
tion of temperature and pressure are considered. Optimization of the base FFCM-1
model has considerably improved the LBV predictions at elevated pressures and
temperatures.

3.5 Flame Structure

The PREMIX calculations for freely propagating the SG–air mixture’s flames at
T u = 450 K, and Pu = 1, 5 atm for various equivalence ratios were carried out.
Figure 5 shows the comparison of the minor species’ mole fractions for different
equivalence ratios at 1 atm and 5 atm. For a direct comparison of different equivalence
ratios, the abscissa was defined as dimensionless temperature (τ ). τ = (T − T u )/
(T ad – T u ), where T u is the unburnt gas temperature and T ad is the adiabatic flame
temperature. The mole fractions of H radicals increase with an increase in equivalence

0.0020
0.007 1 atm, φ = 1.0 H H
5 atm, φ = 1.0
1 atm, φ = 1.4 5 atm, φ = 1.4
0.006
1 atm, φ = 1.6 0.0015 5 atm, φ = 1.6
0.005
Mole fraction

Mole fraction

0.004 OH
0.0010
0.003
OH
O O
0.002
0.0005

0.001

0.000 0.0000

-0.001
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Dimensionless temperature (τad) Dimensionless temperature (τad)

Fig. 5 Flame structure comparison of syngas–air mixtures at T u = 450 K and different pressures
(1, 5 atm)
302 R. J. Varghese andS. Kumar

R33 : CO+OH=H+CO2

R32 : CO+OH=H+CO2
5 atm
R17 : HO2+H=2OH 1 atm

R16 : HO2+H=H2+O2

R15 : H+O2(+M)=HO2(+M)

R13 : H2O+M=H+OH+M

R4 : OH+H2=H+H2O

R3 : O+H2=H+OH

R1 : H+O2=O+OH

-0.15 -0.10 -0.05 0.00 0.05 0.10 0.15

Normalized Sensitivity Coefficients

Fig. 6 Normalized sensitivity coefficients of SG–air mixtures at T u = 450 K and different pressures
(1, 5 atm)

ratio from stoichiometric until φ = 1.4 and then decreases for Pu = 1 atm and T u
= 450 K. The values of O and OH radicals decrease continuously as the fuel–air
mixture transitions to the rich side. This continuous reduction in mole fractions of
O and OH mole fractions is followed for 5 atm case. However, the flame thickness
reduces for higher pressures, which can be identified by the peak values closer to
dimensionless temperature τ = 1. Turns [33] define flame thickness (δ = 2α t /S u )
for binary diffusion, unity Lewis number, and linear temperature profile. LBV can
be considered to be indirectly proportional to the flame thickness in the case of
multi-component diffusion and non-unity Lewis number.
Sensitivity coefficients for gas temperature with respect to the reaction A-factors
were calculated for stoichiometric SG–air mixtures at T u = 450 K and Pu = 1, 5 atm.
Figure 6 shows the normalized sensitivity coefficients for gas temperature with
respect to the reaction pre-exponential factors for stoichiometric SG–air mixtures at
an unburnt gas temperature of 450 K and different mixture pressures (1, 5 atm). The
sensitive reactions for both the pressures for stoichiometric mixtures remain the same.
However, the sensitivity coefficients of the high-pressure mixture are enhanced. The
third-body reaction (R15: H + O2 (+M) = HO2 (+M)) has positive sensitivity for Pu
= 5 atm, whereas the 1 atm case indicates a negative sensitivity. The third-body
reactions’ enhanced sensitivity is expected at denser mixture conditions due to the
enhanced third-body collisions. The amplified negative sensitivity of the reactions
at elevated pressures, and the shift in third-body collisions’ sensitivity indicates a
decrease in burning velocities at elevated pressures may be favored.
Laminar Burning Velocity Measurements at Elevated Pressure … 303

4 Conclusions

LBV of practical syngas compositions, which are an output of biomass gasifiers,

was measured for various equivalence ratios at elevated temperatures and pressures.
The power exponents of temperature and pressure for the measured burning veloci-
ties were determined. The temperature exponents were demonstrated to be a strong
function of the temperature range at which they are calculated and the mixture pres-
sure. The temperature and pressure exponents’ analysis indicated that considering the
power exponents to be constant leads to inaccurate extrapolations. The power expo-
nents of temperature and pressure are expressed as pressure and temperature ratios.
The dependences were suggested individually earlier by Babkin and Kozachenko
[33], Han et al. [34] and a combined modified power–law correlation was proposed
by Varghese et al. [32] for methane–air mixtures. The modified power–law correla-
tion and the variation of power exponents of temperature and pressure proposed are
accurate for syngas–air mixtures.
The best-performing kinetic scheme (FFCM-1) for the syngas mixtures at ambient
conditions performed poorly at high pressure–temperature conditions. The major
contributing factor to this discrepancy is the high sensitivity of key reactions at
high temperatures and pressure conditions. Genetic algorithm-based optimization
of the base model with the current study’s experimental results as the target data
was attempted. A base kinetic scheme was constructed with FFCM-1 (38 species
and 291 reactions) and supplemented by 23 reactions from Aramco 3. The pressure-
dependent Arrhenius factors for rate constants were incorporated in the base model.
The prediction capabilities for burning velocities at high temperature and pressure
conditions have significantly improved with the optimization. The validation datasets
of the modified mechanism at the syngas–air mixtures are promising.

Acknowledgements Financial assistance by the Government of India in the form of a fellowship

for Senior Research Fellow (SRF) is sincerely acknowledged.

Appendix: Experimental Data

The following tables provide the experimental results obtained using the diverging
channel method for various compositions. Table 2 shows the detailed composition
of the syngas considered. Tables 3, and 4 show the laminar burning velocity and
temperature exponents obtained from the experiments, respectively. Table 5 shows
the target parameters used for the optimization.
304 R. J. Varghese andS. Kumar

Table 2 Detailed composition of the syngas used in the present work

Sr. no % Volume H2 /CO ratio
CO CO2 H2 N2
SG1 15 15 15 55 1
SG2 20 15 15 50 0.75
SG3 25 15 15 45 0.6

Table 3 Laminar burning velocity of SG1-SG3 with uncertainties at various equivalence ratio at
ambient conditions (300 K, 1 atm)
φ SG1 Su SG2 Su SG3 Su
Unit cm/s cm/s cm/s cm/s cm/s cm/s
0.71 10.21 ±0.49 12.50 ±0.58 15.97 ±0.80
0.90 16.14 ±0.77 20.00 ±0.99 23.67 ±1.17
1.00 18.23 ±0.91 23.00 ±1.14 26.10 ±1.26
1.10 20.65 ±1.00 24.20 ±1.45 28.95 ±1.45
1.25 21.62 ±1.03 26.27 ±1.34 30.59 ±1.53

Table 4 Temperature exponent (α) of different compositions at various equivalence ratios (φ) and
Pu = 1 atm
φ SG1 α SG2 α SG3 α
0.70 3.06 ±0.09 2.62 ±0.10 2.50 ±0.05
0.90 2.77 ±0.08 2.56 ±0.09 2.37 ±0.05
1.00 2.67 ±0.08 2.51 ±0.09 2.35 ±0.05
1.10 2.60 ±0.08 2.46 ±0.09 2.33 ±0.05
1.25 2.63 ±0.07 2.43 ±0.09 2.30 ±0.05
Table 5 Target data for optimization
(φ) Pu (atm) T u (K) LBV (cm/s) Pu (atm) LBV (cm/s) (φ) Pu (atm) T u (K) LBV (cm/s) Pu (atm) LBV (cm/s)
0.7 1 300 9.7 3 4.5 1.3 2 300 16.8 5 9.3
0.9 1 300 15.5 3 10.0 1.4 2 300 16.4 5 8.6
1.0 1 300 17.4 3 12.0 0.7 2 450 24.6 5 10.5
1.1 1 300 20.1 3 13.0 0.9 2 450 39.4 5 26.1
1.3 1 300 20.5 3 14.0 1.0 2 450 43.7 5 28.7
1.4 1 300 20.2 3 12.5 1.1 2 450 48.2 5 31.9
0.7 1 450 33.6 3 16.1 1.3 2 450 49.8 5 29.2
0.9 1 450 47.8 3 32.0 1.4 2 450 51.5 5 30.4
1.0 1 450 51.4 3 36.5 0.7 2 600 60.0 5 26.6
1.1 1 450 57.6 3 39.0 0.9 2 600 88.9 5 60.6
1.3 1 450 59.6 3 42.0 1.0 2 600 95.5 5 63.9
1.4 1 450 60.4 3 40.8 1.1 2 600 104.5 5 70.5
0.7 1 600 81.0 3 39.9 1.3 2 600 107.7 5 65.7
0.9 1 600 106.0 3 73.1 1.4 2 600 115.8 5 74.3
Laminar Burning Velocity Measurements at Elevated Pressure …

1.0 1 600 110.7 3 80.2

1.1 1 600 121.6 3 85.1
1.3 1 600 127.0 3 91.6
1.4 1 600 131.3 3 94.6
0.7 2 300 7.0 5 2.8
0.9 2 300 12.5 5 8.0
1.0 2 300 14.5 5 9.3
1.1 2 300 16.2 5 10.4
305
306 R. J. Varghese andS. Kumar

References

1. Shih H-Y (2009) Computed extinction limits and flame structures of H2 /O2 counterflow
diffusion flames with CO2 dilution. Int J Hydrogen Energy 34(9):4005–4013
2. Natarajan J, Lieuwen T, Seitzman J (2007) Laminar flame speeds of H2 /CO mixtures: Effect
of CO2 dilution, preheat temperature, and pressure. Combust Flame 151(1–2):104–119
3. Burke MP, Dryer FL, Ju Y (2011) Assessment of kinetic modeling for lean H2 /CH4 /O2 /diluent
flames at high pressures. Proc Combust Inst 33(1):905–912
4. Wang J, Huang Z, Kobayashi H, Ogami Y (2012) Laminar burning velocities and flame
characteristics of CO–H2 –CO2 –O2 mixtures. Int J Hydrogen Energy 37(24):19158–19167
5. Varghese RJ, Kolekar H, Hariharan V, Kumar S (2018) Effect of CO content on laminar burning
velocities of syngas-air premixed flames at elevated temperatures. Fuel 214:144–153
6. Bouvet N, Chauveau C, Gökalp I, Halter F (2011) Experimental studies of the fundamental
flame speeds of syngas (H2 /CO)/air mixtures. Proc Combust Inst 33(1):913–920
7. Burbano HJ, Pareja J, Amell AA (2011) Laminar burning velocities and flame stability analysis
of H2 /CO/air mixtures with dilution of N2 and CO2 . Int J Hydrogen Energy 36(4):3232–3242
8. Lee JH, Kim JH, Park JH, Kwon OC (2010) Studies on properties of laminar premixed
hydrogen-added ammonia/air flames for hydrogen production. Int J Hydrogen Energy
35(3):1054–1064
9. Prathap C, Ray A, Ravi MR (2012) Effects of dilution with carbon dioxide on the laminar
burning velocity and flame stability of H2 –CO mixtures at atmospheric condition. Combust
Flame 159(2):482–492
10. Yan B, Wu Y, Liu C, Yu JF, Li B, Li ZS et al (2011) Experimental and modeling study
of laminar burning velocity of biomass derived gases/air mixtures. Int J Hydrogen Energy
36(5):3769–3777
11. Shang R, Zhang Y, Zhu M, Zhang Z, Zhang D, Li G (2016) Laminar flame speed of CO2 and
N2 diluted H2 /CO/air flames. Int J Hydrogen Energy 41:15056–15067
12. Han M, Ai Y, Chen Z, Kong W (2015) Laminar flame speeds of H2 /CO with CO2 dilution at
normal and elevated pressures and temperatures. Fuel 148:32–38
13. He Y, Wang Z, Weng W, Zhu Y, Zhou J, Cen K (2014) Effects of CO content on laminar burning
velocity of typical syngas by heat flux method and kinetic modeling. Int J Hydrogen Energy
39(17):9534–9544
14. Maruta K, Kataoka T, Kim NI, Minaev S, Fursenko R (2005) Characteristics of combustion in
a narrow channel with a temperature gradient. Proc Combust Inst 30(2):2429–2436
15. Kumar S, Maruta K, Minaev S, Fursenko R (2008) Appearance of target pattern and spiral
flames in radial microchannels with CH4 -air mixtures. Physics of Fluids. 2008;20.
16. Smooke DM, Miller JA, Kee RJ (1983) Determination of adiabatic flame speeds by boundary
value methods. Combust Sci Technol 34:79–90
17. Zhou C-W, Li Y, O’Connor E, Somers KP, Thion S, Keesee C et al (2016) A comprehensive
experimental and modeling study of isobutene oxidation. Combust Flame 167:353–379
18. Metcalfe WK, Burke SM, Ahmed SS, Curran HJ (2013) A hierarchical and comparative kinetic
modeling study of C1–C2 hydrocarbon and oxygenated fuels. Int J Chem Kinet 45(10):638–675
19. Li Y, Zhou C-W, Somers KP, Zhang K, Curran HJ (2017) The oxidation of 2-butene: a high
pressure ignition delay, kinetic modeling study and reactivity comparison with isobutene and
1-butene. Proc Combust Inst 36(1):403–411
20. Davis SG, Joshi AV, Wang H, Egolfopoulos F (2005) An optimized kinetic model of H2 /CO
combustion. Proc Combust Inst 30:1283–1291
21. Smith GP, Golden DM, Frenklach M, Moriarty NW, Eiteneer B, Goldenberg M et al (1999)
GRI-Mech 3.0. p. http://www.me.berkeley.edu/gri_mech.
22. Smith G, Tao Y, Wang H (2016) Foundational fuel chemistry model version 1.0 (FFCM-1).
https://web.stanford.edu/group/haiwanglab/FFCM1
23. Akram M, Saxena P, Kumar S (2013) Laminar burning velocity of methane–air mixtures at
elevated temperatures. Energy Fuels 27(6):3460–3466
Laminar Burning Velocity Measurements at Elevated Pressure … 307

24. Akram M, Kumar S (2012) Measurement of laminar burning velocity of liquified petroleum
gas air mixtures at elevated temperatures. Energy Fuels 26(6):3267–3274
25. Varghese RJ, Kolekar H, Kumar S (2019) Laminar burning velocities of H2 /CO/CH4 /CO2 /N2
-air mixtures at elevated temperatures. Int J Hydrogen Energy 44(23):12188–12199
26. Varghese RJ, Kolekar H, Kumar S (2019) Demarcation of reaction effects on laminar burning
velocities of diluted syngas–air mixtures at elevated temperatures. Int J Chem Kinet 51(2):95–
104
27. Varghese RJ, Kumar S (2020) Machine learning model to predict the laminar burning veloc-
ities of H2 /CO/CH4 /CO2 /N2 /air mixtures at high pressure and temperature conditions. Int J
Hydrogen Energy 45(4):3216–3232
28. Katoch A, Asad M, Minaev S, Kumar S (2016) Measurement of laminar burning velocities of
methanol–air mixtures at elevated temperatures. Fuel 182:57–63
29. Katoch A, Chauhan A, Kumar S (2018) Laminar burning velocity of n-propanol and air mixtures
at elevated mixture temperatures. Energy Fuels 32(5):6363–6370
30. Varghese RJ, Kishore VR, Akram M, Yoon Y, Kumar S (2017) Burning velocities of
DME(dimethyl ether)-air premixed flames at elevated temperatures. Energy 126
31. Kumar R, Katoch A, Singhal A, Kumar S (2018) Experimental investigations on laminar
burning velocity variation of methyl formate-air mixtures at elevated temperatures. Energy
Fuels 32(12):12936–12948
32. Varghese RJ, Kolekar H, Kishore VR, Kumar S (2019) Measurement of laminar burning veloc-
ities of methane-air mixtures simultaneously at elevated pressures and elevated temperatures.
Fuel 257:116120
33. Turns SR (1996) Introduction to combustion: McGraw-Hill Companies
34. Han P, Checkel MD, Fleck BA, Nowicki NL (2007) Burning velocity of methane/diluent
mixture with reformer gas addition. Fuel 86(4):585–596
Combustion Systems
Direct Numerical Simulation
of Preignition and Knock in Engine
Conditions

Minh Bau Luong and Hong G. Im

Abstract Theoretical framework and high-fidelity direct numerical simulations

(DNS) on preignition and knock encountered in internal combustion (IC) engines
are briefly reviewed in this chapter. The theoretical framework is presented first, and
systematically followed by one-dimensional and multi-dimensional simulations. A
number of high-fidelity simulations under realistic IC-engine conditions were used to
demonstrate the super-knock mechanism in the presence of the complex chemistry-
turbulence interaction. The ratio of ignition delay time to eddy-turnover time, τig /τt ,
and the ratio of the most energetic length scale of turbulence and temperature fluctu-
ations, lt /l T , affect the knock intensities are discussed. Different statistical metrics
extracted from the multi-dimensional simulations to predict knock intensity are also
presented. The quantitative determination of strong and weak ignition modes using
the predicted Sa-based metrics is also cover in this chapter.

Keywords Direct numerical simulation (DNS) · Ignition criteria · Strong

ignition · Weak ignition

1 Introduction

Modern downsized and boosted internal combustion (IC) engines provide higher ther-
modynamic efficiencies and cleaner combustion. However, these engines operated
under extreme conditions of high load suffer from a higher possibility of abnormal
preignition process, a premature auto-ignition induced by hot spots leading to ignition
front propagation [1–7]. The mechanism and the source of such irregular preignition
events are still not well-understood [8]. The early flames developed from preigni-
tion kernels increase the temperature and pressure of the end-gas that may induce
a stronger end-gas auto-ignition, leading to extremely high-knock intensities, and
even super-knock under high-load and/or high-temperature conditions. As represen-

M. B. Luong (B) · H. G. Im
Clean Combustion Research Center, King Abdullah University of Science and Technology,
Thuwal, Saudi Arabia
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 311
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_14
312 M. B. Luong and H. G. Im

Start of Knock
220 @128 bar/ -
2°CA/ MFB 65% Start of Knock @
90 bar/ 20°CA/
MFB 62%
170
p [bar]

120 Start of Pre-

Ignition
@-19/-6 ° bTDC S2

70
K2
N2
20
-20 -10 0 10 20 30
KLSA= -2.3° Crank Angle degree

Fig. 1 Examples of in-cylinder pressure traces show preignition and superknock in a turbocharged
SI engine: super-knock cycle (S2) with preignition at 19 CAD BTD and KI ∼ 14.5 MPa, moderate
knock cycle (K2) with preignition at 6 CAD BTDC and KI ∼ 2.4 MPa, and normal (N2) cycle
without preignition (taken from Refs. [6, 7])

tatively shown in Fig. 1 [6], super-knock cycle (S2) is characterized by high-pressure

amplitudes and intensities due to localized detonation development that may lead to
severe structural damage to the engine components. Therefore, a reliable criterion to
predict irregular combustion phenomena is of critical importance [1, 3, 4, 6, 9–12].
A large spectrum of turbulence length and time scales combined with extremely
short time scales of complex combustion process in detonation simulations pose a
significant challenge in reproducing the phenomena using computational simula-
tions, which require extremely fine spatial and temporal resolutions to capture the
highly intermittent localized characteristics. This chapter reviews our recent efforts
to investigate the fundamental issues of the knock events using high-fidelity direct
numerical simulations (DNS) with the capability of fully resolving all temporal and
spatial scales and the complex interaction of thermochemistry and turbulence [3, 4,
13, 14]. First-principle DNS is capable of unraveling the complex interplay between
turbulence and chemical reactions to provide a better understanding of the mecha-
nism of detonation development encountered in modern combustion devices under
extreme high-load operating conditions, and to develop a reliable predictive model
for real-world industrial applications [15–31].
In the following, a theoretical framework on super-knock development encoun-
tered in internal combustion (IC) engines is briefly reviewed, followed by high-
fidelity DNSs in the subsequent sections. A systematically parametric set of two-
dimensional (2-D) and three-dimensional (3D) DNS simulations are conducted at
engine-relevant conditions [1–4, 32]. Effects of various parameters such as length
and time scales of turbulence and scalar fluctuations on the occurrence and intensity
of detonation are examined and a scaling relation for knock prediction is proposed
[33–43].
Direct Numerical Simulation of Preignition … 313

2 Developing Detonation Regime Diagram

2.1 Theoretical Background

2.2 Zeldovich’s Theory on Reactivity Gradient

According to Zeldovich’s theory [44], a spontaneous ignition front is defined as a suc-

cessive ignition events in the presence of spatial reactivity gradients. Its propagation
speed, Ssp , is determined as
Ssp = |∇τig |−1 , (1)

where τig is the ignition delay time of the local reactant mixture as schematically
shown in Fig. 2, defined as the time to reach the maximum heat release as illustrated
in Fig. 3.
Based on the relative ratio of Ssp to the laminar flame speed, Ssp /SL , and to the
sound speed, Ssp /a, of the local mixture as illustrated in Figs. 2 and 3, Zeldovich [44]
classified four different ignition regimes of a reaction front propagation in a com-
bustible gaseous mixture:
1. weak ignition: In the presence of large spatial variations in τig , Ssp ∼ SL ,
the local ignition establishes a slow-speed thin-front deflagration (see Figs. 2c
and 3).
2. spontaneous ignition: For a mixture with a smaller level of τig variations,
Ssp  a, the mixture ignites spontaneously and Ssp exhibits a rapid ignition
front propagation (see Figs. 2a and 3).
3. detonation: If Ssp is sufficiently large and comparable to the sound speed, Ssp ∼ a
as visually shown in Fig. 2b, the combustion wave couple with the acoustic
wave may lead to detonation development characterized by extremely high-peak
amplitudes of pressure oscillations.
4. homogeneous ignition: In the asymptotic limit of zero gradient, Ssp becomes
infinity, resulting in the purely homogeneous ignition mode.

2.3 ε−ξ Regime Diagram

Gu and Bradley [45] further mapped different propagation modes into a regime
diagram characterized by two key non-dimensional parameters, ξ and ε, which are
defined as:
ξ = a/Ssp , (2)

ε = (rhs /a)/τe , (3)

314 M. B. Luong and H. G. Im

Fig. 2 Schematic of 1-D spatially monotonic distribution of ignition delay times, τig1 < τig2 · · · <
τig6 , (first row), and the relative speed between the reaction front and the acoustic wave (second row)
with a spontaneous ignition, b detonation with a perfect coupling between the reaction front and
acoustic wave, and c deflagration

Fig. 3 Representative 2D contours of HRR show deflagrations (first row) and spontaneous ignition
fronts (second row) and their temporal evolutions of HRR (right column). The definition of ignition
delay time, τig , and excitation time, τe (the time duration from 5% of the peak HRR to the peak
HRR), are shown in the right column
Direct Numerical Simulation of Preignition … 315

Fig. 4 Conditions for the 45

occurrence of developing
40
detonations in terms of Xi ξ u
and epsilon for a 35
stoichiometric
30
50%H2 –50%CO/air mixture
at 5 atm [45, 46]. Three 25
N2

ξ
points (S2, K2, and N2) in
20 DEVELOPING
the red line correspond to the
three pressure traces shown 15 DETONATION
in Fig. 1 (taken from
Refs. [6, 7])
10 K2
5
S2 ξ l
0
0 5 10 15 20 25
ε

where ε is defined as the ratio of the residence time, rhs /a, of the acoustic wave
within the hot spot with a radius, rhs , to the excitation time, τe , in which most of the
chemical energy is released. τe is typically defined as the time interval between 5%
of the peak heat release rate (HRR) and the peak HRR as schematically illustrated
in Fig. 3.
Based on the range of ξ , five different modes of ignition front propagation within
a hot spot were identified [45, 46]:
1. ξ = 0 for instant thermal explosion,
2. 0 < ξ < ξl for spontaneous (supersonic) ignition where the autoignitive wave is
ahead of acoustic wave,
3. ξl < ξ < ξu developing and developed detonation, through the thermoacoustic
coupling process referred to as the shock wave amplification by coherent energy
release (SWACER) mechanism [47, 48]),
4. ξl ≥ ξ < a/SL subsonic autoignitive deflagration where SL is the laminar burn-
ing velocity,
5. ξl ≥ ξ < a/SL laminar burning deflagration propagating at the speed compara-
ble to the laminar flame speed.
Figure 4 shows the ξ −ε peninsula with the upper and lower limit, ξu and ξl ,
respectively, between which the developing detonation regime is identified for a
stoichiometric 50%H2 –50%CO/air mixture at 5 atm [45, 46]. Note that the upper and
lower ξ limits of the C-shaped curve strongly depend on the initial conditions of T0 ,
P0 , and φ0 , fuel types, and the kinetic model used in simulations [49, 50]. In general,
the developing detonation regime is expected when ξ ranges between 1 and O(10)
quantity [49, 50]. As representatively shown in Fig. 5, due to the NTC behavior
of two-stage ignition fuels (e.g., DME), they are much more prone to detonation
development (gray region with roughly 1 < ξ < 10) than single-stage ignition fuels
(e.g., ethanol) for the same initial conditions of temperature, pressure, equivalence
ratio, temperature gradient.
316 M. B. Luong and H. G. Im

Fig. 5 The ξ values as a 10

2

function of temperature for

ethanol/air mixtures (red) DME
Ethanol, φ = 1
and DME/air mixtures (blue) Ethanol, φ = 0.5
at φ of 0.5 and 1.0, and 1
10

ξ
pressure of 35 atm, and
∂ T /∂r = −2 K/mm. The
shaded region of 1 ≤ ξ ≤ 10
is prone to the developing
detonation for a small hot 100
spot, i.e., ε < 10 600 800 1000 1200
T (K)

2.4 Quantitative Analysis

Luong et al. [3] proposed three a posteriori statistical metrics to quantitatively assess
the knock intensity, defined as follows:

 q̇| P > 1.1Pe 
FH =  , (4)
q̇

 q̇| P > PCJ 
FCJ =  , (5)
q̇

 q̇| P > PVN 
FVN =  , (6)
q̇

where q̇ is the HRR; Pe , PCJ and PVN correspond to the homogeneous constant-
volume equilibrium pressure, the Chapman–Jouguet pressure, and the von Neumann
pressure, respectively. The summation is operated over all the cells of the computa-
tional domain and integrated over time. As such, Fi (i = e, C J, V N ) measures the
fraction of HRR associated with higher pressures exceeding the corresponding Pi ,
evaluated at the initial conditions of T0 , P0 , and φ0 .
According to Fig. 4 [45], the developing detonation regime is bounded by the
upper and lower ξ limits of the C-shaped curve, ξu and ξl , respectively, suggesting
that the detonation intensity depends strongly on the statistical quantities extracted
from the spatial distribution of ξ such as its volume-average ξ , probability density
function. To this end, FD , defined as the volume fraction of ξu < ξ < ξl , is proposed
to predict knock intensity:
ξu
FD = P(ξ )dξ , (7)
ξl

which serves as an a priori predictive indicator of the subsequent detonation devel-

opment while FH is an a posteriori statistical assessment of knock intensity. Note
Direct Numerical Simulation of Preignition … 317

that both FD and FH range from zero to unity. FD was found to have a strong corre-
lation with the fraction of the total heat release associated with detonation extracted
from the full DNS data, FH [3]. The mixture is found to be more prone to strong
detonation as FD approaches unity if ε p  1. For each initial condition, FD together
with the volume-average, ξ , and probability density function, P(ξ ), are computed to
quantify the knock intensity [3].
In the next sections, the above predictive criteria are discussed using 1D simula-
tions first, followed by selected 2D and 3D simulations with and without turbulence
in order to characterize the spontaneous ignition and detonation development.

2.5 One-Dimensional Detonation Development

The developing detonation ε−ξ diagram [45, 46] has been used in many 1-D studies
to investigate the interaction between flame propagation and end-gas auto-ignition,
and the effects of boundary conditions, domain size, hot-spot location, hot-spot inter-
action, and cool flames on subsequent detonation formulation [6, 49–71]. The effects
of the bulk-mixture gradient on developing detonation were also studied by Sow et
al. [60] and Nogawa and Terashima [72]. In the following, two representative 1-D
cases with distinct detonation development characteristics are discussed.
The first case is shown in Fig. 6a features a deflagration to detonation transition
(DDT) event where a hot spot is developed into deflagration, gradually consuming
the unburned mixture, and finally accelerating to make a transition to detonation.
Specifically, the initial temperature profile is designed such that ξ ranges from 0 to
1000 within the hot spot (see Fig. 6). Figure 7 shows the temporal evolution of the
temperature and pressure profiles. For such a strong ignition source, autoignition
immediately forms a detonation front within the hot spot length, which is subse-
quently attenuated quickly upon leaving the hot spot (lines 1–4 in Fig. 7). Due to the
low reactivity of the unburned mixture at T0 = 1000 K, the heat release rate within
the reaction zone is not fast enough to sustain the detonation front such that it decays
quickly and transits back into deflagration at x ∼ 10 mm. As more fuel/air mixture is
gradually consumed by deflagration, the temperature and pressure of the remaining
unburned mixture ahead of the flame are increased due to the compression heating
effect (lines 5–7 in Fig. 7). Due to the increased reactivity of the unburned mixture
together with the flame acceleration, the deflagration front transits into detonation
at x ∼ 20 mm and rapidly consumes the rest of the unburned mixture (lines 8–12 in
Fig. 7). Much higher pressure peaks are observed during this period. These processes
are typically referred as to deflagration to detonation transition (DDT).
The second case (i.e., T0 of 1200 K with Tpeak of 1264 K in Fig. 6) features a direct
detonation initiation in which the reaction front and the pressure wave are perfectly
coupled to form an extremely high shock wave. The initial hot spot is designed for
ξ to range between 3 and 6 (see Fig. 6). Due to the high reactivity of the unburned
mixture with T0 of 1200 K, the hot spot immediately gives rise to a detonation front
within the hot spot length, thereby rapidly consuming the unburned mixture (see
318 M. B. Luong and H. G. Im

Fig. 8). Once the detonation is established, the rapid heat release rate in the reaction
zone is able to sustain the steady detonation front that propagates and consumes the
rest of the unburned mixture.
The above cases show that the two parameters, ξ and ε, within the hot spot and
the bulk mixture determines the nature of the early detonation development. Figure 9
shows the compiled results of a series of 1-D simulations for various combinations
of these parameters. For a fixed T0 of 1200 K and φ of 1, the 1-D simulations were
performed by varying Ths and rhs such that both ξ and ε are varied accordingly. As
seen in Fig. 9, the maximum pressure above the equilibrium pressure (i.e., > 1.1Pe )
for 0.2 < ξ < 20. The peak Pmax occurs at ξ ∼ 6 where the perfect coupling between
the reaction front and the acoustic wave evolving into a stable detonation shock wave.
The range of ξ was used to determine the volume fraction of ξ and FD to predict the
amplitude of knock intensity in multi-dimensional cases.

a 2000
103
b 1280 8

1260

. ξ
102

.
6
T (K)

T (K)

1500 1240
ξ

ξ
. 101 1220 ξ 4

1000 1200
100
2
0 5 10 0 5 10
x (mm) x (mm)

Fig. 6 Two representative initial conditions of a hot spot with a 1-D linear-temperature profile: a
T0 = 1000 K and Tmax = 2000 K featuring deflagration to detonation transition (DDT) (see Fig. 7),
and b T0 = 1200 K and Tmax = 1264 K featuring direct detonation development (perfect coupling
between the acoustic wave and ignition front, see Fig. 8). The speed of sound, a, and ξ are evaluated
at the center of hot spots (marked as dots). The spatial distribution of ξ within the hot spots is also
shown. Note that τig and τe are computed by using a 0-D homogeneous adiabatic constant-volume
reactor using the localized temperature and pressure

a 3500 b 60
10 12
3000 40

DDT
P (MPa)

2500 20
T (K)

1 2 4 5 6 7 8 10 12 1 2
2000 increasing time
7
1500 6
4
5
1000 3
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
x (mm) x (mm)

Fig. 7 The temporal evolutions of a temperature and b pressure at different time. The initial
conditions of T0 of 1000 K and P0 of 35 atm (see Fig. 6), a stoichiometric ethanol/air mixture
Direct Numerical Simulation of Preignition … 319

a 3500
b 40
30
3000
20

P (MPa)
2500
T (K)

increasing time 10
2000
increasing time
1500

1000 3
0 1 0 1
0.5 10 10 50 0.5 10 10 50
x (mm) x (mm)

Fig. 8 The temporal evolutions of a temperature and b pressure at different time.The initial con-
ditions of T0 of 1000 K and P0 of 35 atm (see Fig. 6), a stoichiometric ethanol/air mixture

a 3
b 15
ε ≈ 2.8, rhs = 2 mm 0.2 0.5 1 rhs= 2 mm Pmax/PVN
ε ≈ 1.4, rhs = 1 mm 2.5

2 ε ≈ 0.7, rhs = 0.5 mm 10

2.0
Pmax/PVN

ε ≈ 0.3, rhs = 0.2 mm PZND 1.0

ξ

PCJ 0.6
1.1Pe
1 5 0.3

1.1Pe
0 -1 0
10 10
0
10
1
20 0 1 2 3 4 5
ξ ε
Fig. 9 a The normalized maximum pressure as a function of ξ , and b the ξ − ε diagram for different
hot spot sizes (taken from Luong et al. [3])

2.6 2-D and 3-D Simulations of Super-Knock Under Realistic

IC-engine Conditions

The basic characterization in 1D problem is now extended to multi-dimensional

conditions in the presence of spatial fluctuations in temperature and velocity fields.
Parametric 2-D and 3-D simulations were systematically performed to validate the
criteria proposed in Fig. 2.4. The distribution of ξ and ε were varied by changing
T over a wide range from 0.5 to 20 K, and l T of 1, 2, 5, and 10 mm. For each
initial condition, the full detailed information of spatial distribution of ξ and the
corresponding statistical quantities of ξ can be computed. Figure 10 shows the spatial
distribution of ξ of four cases with different T and l T that result in no-knock, mild-
knock and super-knock [3].
Figure 11 shows a representative 2-D contour of temperature field with T0 of
1200 K, T of 15 K, and l T of 5 mm. According to Peters et al. [53, 73], the temperature
field is subdivided into the dissipation elements over which the temperature profile
320 M. B. Luong and H. G. Im

(a) lT = 1 mm, T  = 15 K, no knock (b) lT = 2 mm, T  = 15 K, mild knock

(c) lT = 5 mm, T  = 15 K, super-knock (d) lT = 5 mm, T  = 0.6 K, mild knock

Fig. 10 Representative 2-D contours of ξ distribution computed from four different initial condi-
tions for three different l T of 1 mm, 2 mm, 5 mm, with T of 15 K (a–c) and 0.6 K (d)

Fig. 11 Representative 2-D

contour of temperature field
with T0 of 1200 K, T of
15 K, and l T of 5 mm
overlaid by the dissipation
elements. The white lines
mark the boundary between
the dissipation elements
Direct Numerical Simulation of Preignition … 321

is monotonic, ranging from the peak to the trough of each dissipation element. For
isotropic temperature field fluctuations, Peters et al. [53] found rhs comparable with
the mean distance of the dissipation elements, lDE , and approximated as lDE ≈ 2λT ,
where λT is the Taylor mixing scale [53]. In the multi-dimensional problems, lDE is
believed to be a minimum size of a hot spot of unburned reactants required so that it
allows a sufficient run-up distance of detonation development [12, 53].
Based on the Taylor mixing scale, Towery et al. [12] proposed a statistical model to
evaluate the predictive ε p for highly compressible homogeneous isotropic turbulence
fluctuations. ε p is defined as

ε p = 2λT /(arms τexo ) (8)

where a and τexo are the mass-weighted sound speed and exothermicity time, respec-
tively (see Ref. [12] for details). In Refs. [3, 4], at the the initial conditions of
T0 = 1200 K, P0 = 35 K, and φ0 = 1, ε p = 1 corresponds to the initial temperature
field of l T = 3.3 mm. The cases with ε p < 1 (i.e., l T of 1 and 2 mm with ε p of 0.3
and 0.6) are less likely to develop detonation while the cases with ε p of 1.2, 1.4, and
3.8 for l T of 4 mm, 5 mm, and 10 mm, respectively, are more likely to develop strong
detonation if 1 < ξ < 10.
For a uniform mixture with only thermal fluctuations, the predictive ξ p is scaled
as [11, 12]  
dτig  dτig 
ξ p = a| ∇T | ∼
= a   |∇T
| (9)
dT dT 

| ∼
where the statistical mean temperature gradient is further approximated by |∇T =
∼
T /λT , with the Taylor mixing length scale, λT = (l T α/T ) [11]. These predictive
0.5

parameters, ξ p , and ε p , will be validated in the subsequent sections.

2.6.1 Baseline Cases with No Turbulence

In this subsection, three baseline (no turbulence) cases with l T of 1 mm, 2 mm, and
5 mm denoted as BL1, BL2, and BL5, respectively, are chosen for discussion. The
probability density functions (PDF) of ξ , the temporal evolution of the normalized
Pmax , and the contours of pressure at the time of the peak of FH for three baseline cases
are shown in Figs. 12 and 13. According to the scaling of ξ p above, with a fixed T
of 15 K, increasing l T from 1 to 5 mm decreases ξ p which is consistent with the PDF
of ξ shifting toward the developing detonation regime (see Fig. 12). Consequently,
BL5 is predicted to be more prone to developing detonation as compared to BL1 and
BL2.
Consistent with the PDFs of ξ and ε p = 0.3, Pmax ≤ PVN and the isocontours of
pressure in Figs. 12 and 13, respectively, confirm that BL1 features a spontaneous
ignition process. BL2 with FVN ≈ 0.07 is characterized as a mild detonation process
as also quantitatively predicted by ε p = 0.6. Unlike BL1 and BL2, a strong detonation
322 M. B. Luong and H. G. Im

a 0.1 b 10
lT = 1 mm
0.08 8 lT = 2 mm
lT = 5 mm

Pmax/PVN
0.06 6
PDF

5 mm
0.04 4

2 mm 2
0.02
lT = 1 mm
0 0
0 20 40 60 80 100 0.7 0.8 0.9 1
ξ t/τ0ig

Fig. 12 a The probability density functions (PDF) of ξ for three baseline cases (without turbulence)
with three different most energetic length scale of temperature, l T of 1 mm, 2 mm, and 5 mm,
featuring no knock, mild knock, and super-knock, respectively, and b their temporal evolution of
the normalized maximum pressure, Pmax /PVN , [4]. The gray area in the PDF figure depicts the
region of 1 < ξ < 10 determined by a series of 1-D simulations under the same initial conditions,
which is prone to developing detonation (see Fig. 9a) (adapted from Luong et al. [3, 4])

Fig. 13 Representative 2-D contours of pressure for three cases with the same T of 15 K, and l T of
1 mm, 2 mm, and 5 mm with no knock, mild knock, and super-knock, from left to right, respectively
(adapted from Luong et al. [3])

process is observed for BL5, quantified by FVN ≈ 0.2 and Pmax  PVN , which is
accurately predicted by ε p = 1.6 for BL5.
The statistical quantities, FH , FD and ξ , of all the baseline cases are plotted in
Fig. 14. Three main points are noted from Fig. 14. First, FD has a better agreement
with FH for ε p > 1 (i.e., cases with l T > 4 mm in Fig. 14b). Second, the magnitude of
FH for 3-D cases is comparable with the corresponding 2-D cases as seen in Fig. 14a
despite the peak FVN of the 3-D cases slightly lower than that of the 2-D cases
[3]. Third, Fig. 14 also reveals a good correlation between ξ and the knock intensity
characterized by the FH magnitude. The correlation indicates that the conditions with
ξ ranging between 0.5 and 30 experience developing detonation and the strongest
detonation intensity (FH approaches unity) occurs at ξ ∼ 6 with ε p > 1. Specifically,
the maximum values of FH at ξ ∼ 6 are approximately 0.18, 0.22, 0.55, and 0.75
for l T of 1 mm, 2 mm, 5 mm, and 10 mm, respectively. These results verify that the
prediction of ε p for knock propensity is consistent with FH , the detonation intensity
is greatly alleviated with decreasing l T (mild-to-no detonation for l T < 3.3 mm with
ε p < 1).
Direct Numerical Simulation of Preignition … 323

a 1 b 1
1 mm 1 mm
0.8 2 mm 0.8 2 mm
4 mm 4 mm
5 mm 0.6 5 mm
0.6
FH, FD

10 mm 10 mm

FH
0.4 0.4

0.2 0.2

0 -1 0
10 100 101 102 0 0.2 0.4 0.6 0.8 1
ξ FD

Fig. 14 a FH (color symbols) and FD (black symbols) as a function of ξ , with T ranging from
0.5 K to 20 K for different l T ranging from 1 mm to 10 mm, b the correlation between FH and FD .
Filled (open) symbols are for 3-D (2-D) cases (taken from Luong et al. [3])

2.6.2 Effect of Turbulence

The previous subsection demonstrated that no detonation was observed for BL1 while
BL5 exhibited a strong detonation process because a minimum run-up distance is
required for developing detonation as accurately predicted by εt . The results suggest
that decreasing (increasing) the length scale of temperature fluctuations (hence the
dissipation element length) is a key parameter to alleviate (promote) knock intensity.
In this subsection, the additional effect of turbulent velocity fields on knock intensity
is also investigated [4].
The isocontours of pressure and HRR of two representative 2-D simulations and
a 3-D simulation in Figs. 15, 16 and 17 are chosen for discussion. With the same tur-
bulent velocity fluctuation, u of 83.3 m/s, and root-mean-square temperature fluctu-
ation, T of 15 K, l T /le and τig /τt were found to be two key parameters to determine
knock propensity.
First, increasing turbulent intensity by increasing u and fixing le = 5 mm is not
effective in reducing knock intensity since with τig /τt of 1.0, turbulence is not fast
enough to modify the autoignition and the subsequent detonation process as visually
seen in Fig. 16. Rather, the main effect of turbulent eddies is to weakly wrinkle and
distort the detonation fronts. Second, a finer turbulence structure, le of 1 mm l T =
5 mm, and τig /τt of 5.0, effectively attenuates the detonation intensity as visually
seen in Figs. 16 and 17. High turbulence intensity together with a short mixing-
time scale significantly increases the amount of reactants burned by corrugating
turbulent flames rather than by the developing detonation phase. Fast and small
turbulence structures also induce smaller and finer localized detonation waves during
the developing detonation process. Third, for the 3-D cases with a slightly lower FH ,
a consistent agreement between the 2-D and 3-D cases was observed in terms of
Pmax , the duration of detonation process, and the level of pressure oscillations, P
[4].
324 M. B. Luong and H. G. Im

Fig. 15 Two-dimensional contours of pressure and heat release rate (HRR) at the times of onset of
detonation development (left) and the peak detonation intensity (right) for a 2-D case with the most
energetic length scale of temperature and turbulent field, l T of 5 mm and le of 5 mm, respectively,
the turbulent velocity fluctuation, u of 83.3 m/s, and the ratio of the ignition delay time to turbulent
time scale τig /τt of 1.0 [4] (taken from Desai et al. [14])

Fig. 16 Two-dimensional contours of pressure and heat release rate (HRR) at the times of onset of
detonation development (left) and the peak detonation intensity (right) for a 2-D case with the most
energetic length scale of temperature and turbulent field, l T of 5 mm and le of 1 mm, respectively,
the turbulent velocity fluctuation, u of 83.3 m/s, and the ratio of the ignition delay time to turbulent
time scale τig /τt of 5.0 [4] (taken from Desai et al. [14])

Contrary to the case with le = 1 mm l T = 5 mm being able to attenuate detona-

tion intensity, a fast turbulent mixing, τig /τt of 5, together with le = 5 mm  l T = 1
mm, promotes detonation development as seen in Fig. 18. Turbulence with a larger
eddy size effectively spreads the temperature field over a wider area, stretching out
the run-up distance, and thus facilitates the developing detonation [4].
Direct Numerical Simulation of Preignition … 325

Fig. 17 The volume rendering of HRR and the iso-surface of pressure (green) with P = PCJ at
successive times prior to the onset of the detonation formation (top row) until the detonation intensity
reaches its peak (bottom row) for a 3-D case with the most energetic length scale of temperature
and turbulent field, l T of 5 mm and le of 1 mm, respectively, the turbulent velocity fluctuation, u
of 83.3 m/s, and the ratio of the ignition delay time to turbulent time scale τig /τt of 5.0 [4] (taken
from Desai et al. [14])

Fig. 18 Temporal evolution 3

of the normalized maximum Baseline, no turbulence lT = 1 mm
pressure, Pmax /PVN for the le=1 mm, u =16.6 m/s, ig/ t
=1
cases with l T of 1 mm with le=1 mm, u =83.3 m/s, ig/ t
=5
different le and u (adapted 2
Pmax /PZND

le=5 mm, u =16.6 m/s, ig/ t

= 0.2
from Luong et al. [4]) le=5 mm, u =83.3 m/s, ig/ =1
t

PCJ

0
0.7 0.8 0.9 1
0
t/ ig
326 M. B. Luong and H. G. Im

3 Weak and Strong Ignition Modes

3.1 Theoretical Background

Based on Zeldovich’s theory and validated by high-fidelity DNS data, Sankaran et

al. [15] proposed a criterion to delineate the weak or strong ignition modes, referred
to as the Sankaran number (Sa) herein, which is based on the ratio of laminar flame
speed, SL , to the ignition front speed, Ssp = |∇τig |−1 [44].

SL
Sa = β , (10)
Ssp

where β is a weighting factor of 0.5 to account for some delay in developing a

deflagration from the ignition kernel due to the interactions of thermal and radial
diffusion between the neighboring regions. Sa = 1 serves as the boundary between
the strong (Sa < 1) and the weak (Sa > 1) ignition.
In practical combustion devices, the detailed information of the spatially resolved
scalar fluctuations is not available, and thus the prediction for in a real engine should
be based on the statistical quantities such as the mean and variance of fluctuations in
RANS or LES simulations. Im et al. [11] conducted a theoretical scaling analysis by
incorporating the turbulent effect into the original Sankaran criterion to derive the
turbulent ignition regime diagram as shown in Fig. 19.
For compositionally homogeneous reactant mixtures with turbulent velocity and
temperature fluctuations, the diagram classifies ignition modes into three cate-
gories: weak exhibiting a dominant slow-speed deflagration mode of fuel consump-
tion, reaction-dominant strong, and mixing-dominant strong exhibiting a dominant
volumetric-ignition mode of fuel consumption. Pal et al. [74] and Luong et al. [1]
conducted a series of DNS simulations and found consistent results with the pre-

Fig. 19 Regime diagram

proposed by Im et al. [11] for
strong and weak ignition
modes for nearly
homogeneous reactant
Reaction intensity

mixture with velocity and

temperature fluctuations
Direct Numerical Simulation of Preignition … 327

dictions by the turbulent ignition regime diagram. The ignition criterion [11] was
further extended to accommodate both temperature and concentration fluctuations
and validated for the mixtures with NTC and non-NTC behaviors [1].
The gradient of ignition delay time, ∇τig , for a mixture with a uniform composition
and pressure is solely due to temperature variations expanded by the chain rule as
[11]:  
 dτig 
|∇τig | = ∇T  , (11)
dT

where dτig /dT represents the ignition delay sensitivity to temperature as represen-
tatively shown in Fig. 20b, and ∇T is the temperature gradient of the bulk mixture
[11, 74]. In turbulent conditions, the temperature gradient is approximated by the
|, which is estimated based on the variance
statistical mean temperature gradient, |∇T
of the temperature, T , fluctuations and the Taylor mixing scale, λT , as

| ≈ T ,
|∇T (12)
λT

where the Taylor mixing scale is approximated as

λT = lRel−0.5 , (13)

with Rel = u l/ν being the turbulent Reynolds number, ν the kinematic viscosity of
the bulk mixture gas, u the root mean square (RMS) turbulent velocity fluctuation,
and l the integral scale of velocity field.
Further scaling analysis yields the expression for the predicted Sankaran number,
Sa p , [11] as follows:
 
1  dτig 
Sa p = K Dal
−1/2
, K =β  ,
(τ f τig )1/2  dT T  (14)

where Dal = τt /τig is the ignition Damköhler number, defined as the ratio of the
turbulence integral time scale, l/u , to the ignition delay time, and τ f = α/SL2 is
the characteristic flame time, where α is the thermal diffusivity of the initial bulk
mixture. K is referred to as the normalized thermal ignition sensitivity.
In the presence of both temperature and equivalence ratio (concentration) fluctu-
ations, τig is a function of T and φ such that |∇τig | is expanded by using the chain
rule as flows:  
 ∂τig ∂τig 

|∇τig | = ∇T + ∇φ , (15)
∂T ∂φ

where ∇φ and ∂τig /∂φ are the equivalence ratio gradients, and the ignition delay
sensitivity to equivalence ratio, respectively [1]. Assuming that the temperature and
equivalence ratio fields are correlated with the turbulent flow field, it results in a
328 M. B. Luong and H. G. Im

 ∼ φ /λT as approximated in
comparable integral length scale, l T ∼ lφ ∼ le , and ∇φ
the temperature case.
As such K in Fig. 14 is extended as:
   
 ∂τig   
K =β
1   + s  ∂τig φ  ,
(τ f τig ) 1/2  ∂T T   ∂φ  (16)

where the sign of s is to account for the NTC behavior and the correlation of T -φ
distribution. s is positive if T and φ fluctuations are uncorrelated, thereby being
additive to |∇τig | that synergistically promotes the deflagration mode regardless of
the NTC regime. s adopts a negative (positive) sign for a negative T − φ correlation
if the product of ∂τig /∂ T and ∂τig /∂φ is positive (negative) that the combined effects
of T and φ on K become subtractive (additive).

3.2 Prediction of Weak/strong Combustion Modes

Many fuels with and without the NTC behavior were chosen to investigate the effect of
temperature and equivalence ratio fluctuations on the combustion modes by varying
the initial mean temperatures and different fluctuation levels [1, 20, 27, 28, 30,
75–77]. The initial conditions and physical parameters are chosen to be relevant to
IC engine conditions [1, 33–43]. The DNS solution were processed to verify the
predictive accuracy of the ignition criteria.
Figure 20 shows the homogeneous ignition delay time of DME/air mixtures and
ethanol/air mixtures and the sensitivity of ignition delay time to temperature that is
needed to compute Sa p . DME exhibits the negative-temperature coefficient (NTC)
regime in which τig increases with increasing temperature such that its τig varies
non-monotonically with temperature. As such, within/near the NTC regime, the
variation of dτig /dT for DME is small compared to that of single-stage fuels (e.g.,
ethanol), which in turn leads to a small |∇τig | for the same temperature fluctuation
level. It suggests that within/near the NTC regime, the mixture with T only is more
susceptible to spontaneous ignition as quantitatively predicted by Sa p in Fig. 21 [2].
Figure 22 shows a typical initial condition used in the simulations. These DNS
cases were adopted to perform the statistical analysis of Sa such that the volume-
average (mean) Sa, Sa, the probability density function of Sa, P(Sa), and the volume
fraction of the regions with Sa <1, FSa,S , extracted from the spatial Sa distribution
are obtained. In addition, Sap for each simulation was also computed based on the
information of the initial mean bulk mixture condition such as T0 , p0 , and φ0 , and
the RMS values, T and φ as representatively shown in Fig. 21 as a function of
temperature.
The isocontours of HRR for some representative DME cases are shown in Fig. 23.
Spontaneous ignition characterized by volumetric ignition is dominant even with T
of 60 K at T0 of 900 K due to a small variation of τig with temperature within/near the
Direct Numerical Simulation of Preignition … 329

a 10
1
b 10
-1

DME Ethanol DME Ethanol

|dτig/dT | (ms/K)
τig (ms)

0
10-2
10

Ethanol, φ = 1
Ethanol, φ = 0.5 10-3
-1
10
600 800 1000 1200 600 800 1000 1200
T (K) T (K)

Fig. 20 a The 0-D homogeneous ignition delay time, τig , and b the absolute value of the sensitivity
of τig to temperature, |dτig /dT |, as a function of temperature for ethanol/air mixtures (red) and
DME/air mixtures (blue) at φ of 0.5 and 1.0, and pressure of 35 atm

Fig. 21 The predicted Sa 10

2

number, Sa p , as a function of
Ethanol, φ = 1
temperature for ethanol/air
mixtures (red) and DME/air DME, φ = 1
mixtures (blue) at φ of 0.5 1 Ethanol, φ = 0.5
10
and 1.0, and pressure of 35 DME, φ = 0.5
Sap

atm, and assuming the

identical integral length scale
of velocities and temperature 100
field, lt = l T = 1 mm,
τt /τig = 1, and T = 15 K.
The shaded region of
Sa p < 1 is for strong 10-1
ignition, while the region of 600 800 1000 1200
Sa p > 1 is for weak ignition T (K)

NTC regime as shown in Fig. 20). On the contrary, deflagration propagation becomes
dominant for the cases with T0 of 680 and 1045 K outside the NTC regime. Similar
results were found for n-heptane [20, 29–31, 75, 76, 78, 79].
Contrary to the T -only cases, in the presence of both T and φ, especially when
they are negatively correlated, T0 of 900 K results in a much lower HRR (due to the
synergistic effect of T and φ ) than those of T0 outside the NTC regime (due to the
subtractive effect of T and φ for the cases with T0 of 770 and 1045 K), as seen in
the bottom row of Fig. 23.
Quantitative assessment of combustion modes for the previous cases was also
conducted. For each initial field of a DNS case, the exact distribution of Sa within the
domain and its associated statistical quantities can be computed. For direct validations
with the predictive quantities such as FS and Sa, the fractional contributions of weak
and strong ignition to the total heat release were extracted from the detailed DNS
solution.
Figure 22a-b shows representative initial fields of temperature and equivalence
ratio distribution. Based on these detailed DNS information, the spatial distribution
330 M. B. Luong and H. G. Im

Fig. 22 a–b initial fields of temperature and equivalence ratio with a negative T –φ distribution at
the initial mean temperature and equivalence ratio of 900 K–0.5 with their fluctuation of T and φ
of 30 K and 0.15, respectively, and c–d the corresponding ignition delay time and Sa distribution.
The black iso-lines of Sa = 1 delineates two distinct combustion modes of strong (Sa < 1) and weak
(Sa > 1) ignition (taken from Luong et al. [2])

of τig and Sa, and the corresponding probability density function (PDF) of Sa, P(Sa)
are computed in Figs. 22c–d and 24, respectively. The volume fraction of the regions
attributed to strong ignition was also estimated as

1
FSa,S = P(Sa)dSa , (17)
0

which was proposed as a quantitative metric to predict the fractional cumulative heat
release rate associated with strong ignition directly computed from DNS solution
fields, FDa,S as [20, 27, 28, 30, 75, 76]

 q̇| Da > Da0 
FDa,S =  , (18)
q̇

where the summation is operated over the total number of computational cells in the
DNS domain, q̇ is the HRR, and Da is Damköhler number defined as the ratio of the
Direct Numerical Simulation of Preignition … 331

Fig. 23 2-D contours of the normalized HRR at the maximum spatial-averaged HRR occurrence
for the DME/air mixtures at the initial pressure of 30 atm. The first (second) row corresponds to the
cases with temperature fluctuations only, T of 60 K, (with both temperature and equivalence ratio
fluctuations, T of 30 K and φ of 0.15 [2]) (adapted from Luong et al. [1])

local reaction and diffusion terms for a selected scalar variable [16, 17, 20, 28, 30,
71, 75, 76, 80]. Strong ignition is dominant if FSa,S ∼ 1. Strong and weak ignition
is delineated by a threshold value, Da0 , of order unity that depends on the chosen
initial condition. The significance of the Sa-based metric is that the prediction of
combustion modes is based purely on the initial conditions.
Consistent with the qualitative observations in the previous section, Sa and FSa,S
for the DME cases of T0 at 680 K, 900 K, and 1045 K are found to be, respectively,
4.2 & 0.20, 0.5 & 0.92, and 1.3 & 0.4. The cases with T0 of 680 K and 1045 K have
Sa greater than one, suggesting a weak-ignition dominance, while for the case with
T0 of 900 K, a strong ignition is expected. The predictive accuracy of Sa and FSa,S
is further verified by the magnitude of FDa,S as shown in Fig. 25.
Figure 25 shows a strong correlation between San = Sa/(1+ Sa) and FSa,S regardless
of T0 , T , φ , and fuel types. A good correlation between Sap and FSa,S was also
demonstrated, which is attributed to the agreement between the predicted Sa p and
the volume-average Sa. Based on this universal correlation, the magnitude of either
Sa or Sap can be used to quantitatively infer FSa,S , which characterize the combustion
intensity. In general, a higher Sa corresponds to a lower FSa,S , which translates into a
lower combustion intensity. In particular, San = 0.5 serves as a transition point from
332 M. B. Luong and H. G. Im

Fig. 24 A representative 0.8

probability density function
of Sa distribution, PDF(Sa).
SI and WI denote strong
ignition and weak ignition,
0.6
respectively with a threshold
of Sa = 1 (taken from Luong

PDF
et al. [2]) 0.4

0.2
SI WI
0
0 1 2 3 4 5
Sa

Fig. 25 The volumetric 1

fraction of Sa < 1, FSa,S , as
a function of the normalized
0.8
Sa, San =Sa/(1+Sa) to fit
within a range between 0 and
1, with 0.5 corresponding to 0.6
FSa,S

Sa = 1 (taken from Luong et

al. [1]) 0.4 Yoo et al., 2011
Ethanol
0.2 Luong et al., 2015
DME
0
0 0.2 0.4
_ _0.6 0.8 1
Sa/(1+Sa)

SI to WI dominance with 60% heat released by spontaneous ignition mode (FSa,S

= 0.6). Spontaneous ignition dominance (FSa,S = 1.0) occurs at Sa < 0.3. On the
contrary, weak ignition is expected to be dominant at Sa > 1.0. In summary, either
Sa or Sap can be used to quantitatively predict the combustion modes over a wide
range of initial conditions and different fuel types.

4 Summary

Recent developments in the theoretical study to predict detonation development in

IC engine conditions were summarized. A brief review was given on the theoretical
framework, starting from the original Zeldovich theory, and the regime diagrams per
Bradley [46] and Im [11], and various proposed predictive criteria by Luong et al. [3,
4]. A large collection of DNS simulation data were used for a priori and a posteriori
Direct Numerical Simulation of Preignition … 333

test for the validation of different predictive criteria based on the initial solution field,
and its extension to large turbulent conditions with statistical average and variances
has been suggested.
In real combustion engines, additional complexities arise from the front-to-wall
and front-to-front interactions which may further enhance or suppress the occurrence
and intensity of the detonation event. This is of strong practical significance and is
currently being investigated by large-scale simulations [3, 4].

Acknowledgements This work was sponsored by King Abdullah University of Science and Tech-
nology and used the resources of the KAUST Supercomputing Laboratory.

References

1. Luong MB, Hernández Pérez FE, Im HG (2020) Prediction of ignition modes of NTC-fuel/air
mixtures with temperature and concentration fluctuations. Combust Flame 213:382–393
2. Luong MB, Hernández Pérez FE, Sow A, Im HG (2019) Prediction of ignition regimes in
DME/air mixtures with temperature and concentration fluctuations. AIAA SciTech 2019 Forum
https://doi.org/10.2514/6.2019-2241
3. Luong MB, Desai S, Hernández Pérez FE, Sankaran R, Johansson B, Im HG (2020) A statistical
analysis of developing knock intensity in a mixture with temperature inhomogeneities. Proc
Combust Inst 37
4. Luong MB, Desai S, Hernández Pérez FE, Sankaran R, Johansson B, Im HG (2006) Effects of
turbulence and temperature fluctuations on knock development in an ethanol/air mixture. Flow
Turbul Combust
5. Wang Z, Liu H, Reitz RD (2017) Knocking combustion in spark-ignition engines. Prog Energy
Combust Sci 61:78–112
6. Kalghatgi GT, Bradley D (2012) Pre-ignition and ‘super-knock’ in turbo-charged spark-ignition
engines. Int J Engine Res 13:399–414
7. Kalghatgi GT (2015) Developments in internal combustion engines and implications for com-
bustion science and future transport fuels. Proc Combust Inst 35:101–115
8. Figueroa-Labastida M, Badra J, Elbaz AM, Farooq A (2018) Shock tube studies of ethanol
preignition. Combust. Flame 198:176–185
9. Meyer JW, Oppenheim AK (1971) On the shock-induced ignition of explosive gases. Symp
(Int) Combust 13:1153–1164
10. Lutz AE, Kee RJ, Miller JA, Dwyer HA, Oppenheim AK (1989) Dynamic effects of autoignition
centers for hydrogen and c1, 2-hydrocarbon fuels. Symp (Int) Combust 22:1683–1693
11. Im HG, Pal P, Wooldridge MS, Mansfield AB (2015) A regime diagram for autoignition of
homogeneous reactant mixtures with turbulent velocity and temperature fluctuations. Combust
Sci Technol 187:1263–1275
12. Towery CAZ, Poludnenko AY, Hamlington PE (2020) Detonation initiation by compressible
turbulence thermodynamic fluctuations. Combust Flame 213:172–183
13. Hernández Pérez FE, Mukhadiyev N, Xu X, Sow A, Lee BJ, Sankaran R, Im HG (2018)
Direct numerical simulations of reacting flows with detailed chemistry using many-core/GPU
acceleration. Comput Fluids 173:73–79
14. Desai S, Yu JK, Song W, Luong MB, Hernández Pérez FE, Sankaran R, Im HG (2020) Direct
numerical simulations of reacting flows with shock waves and stiff chemistry using many-
core/GPU acceleration. Comput Fluids, In revision
15. Sankaran R, Im HG, Hawkes ER, Chen JH (2005) The effects of non-uniform temperature
distribution on the ignition of a lean homogeneous hydrogen-air mixture. Proc Combust Inst
30:875–882
334 M. B. Luong and H. G. Im

16. Chen JH, Hawkes ER, Sankaran R, Mason SD, Im HG (2006) Direct numerical simulation of
ignition front propagation in a constant volume with temperature inhomogeneities: I. funda-
mental analysis and diagnostics. Combust Flame 145:128–144
17. Hawkes ER, Sankaran R, Pébay P, Chen JH (2006) Direct numerical simulation of ignition
front propagation in a constant volume with temperature inhomogeneities: II. parametric study.
Combust Flame 145:145–159
18. Bansal G, Im HG (2011) Autoignition and front propagation in low temperature combustion
engine environments. Combust Flame 158:2105–2112
19. Gupta S, Im HG, Valorani M (2011) Classification of ignition regimes in HCCI combustion
using computational singular perturbation. Proc Combust Inst 33:2991–2999
20. Yoo CS, Lu T, Chen JH, Law CK (2011) Direct numerical simulations of ignition of a lean
n-heptane/air mixture with temperature inhomogeneities at constant volume: parametric study.
Combust Flame 158:1727–1741
21. Yu R, Bai X-S (2013) Direct numerical simulation of lean hydrogen/air auto-ignition in a
constant volume enclosure. Combust Flame 160:1706–1716
22. El-Asrag HA, Ju Y (2013) Direct numerical simulations of exhaust gas recirculation effect
on multistage autoignition in the negative temperature combustion regime for stratified HCCI
flow conditions by using H2 O2 addition. Combust Theory Model 17:316–334
23. El-Asrag HA, Ju Y (2014) Direct numerical simulations of NOx effect on multistage autoigni-
tion of DME/air mixture in the negative temperature coefficient regime for stratified HCCI
engine conditions. Combust Flame 161:256–269
24. Bhagatwala A, Lu T, Chen JH (2014) Direct numerical simulations of HCCI/SACI with ethanol.
Combust Flame 161:1826–1841
25. Bhagatwala A, Sankaran R, Kokjohn S, Chen JH (2015) Numerical investigation of spontaneous
flame propagation under RCCI conditions. Combust Flame 162:3412–3426
26. Bansal G, Mascarenhas A, Chen JH (2015) Direct numerical simulations of autoignition in
stratified dimethyl-ether (DME)/air turbulent mixtures. Combust Flame 162:688–702
27. Yoo CS, Luo Z, Lu T, Kim H, Chen JH (2013) A DNS study of ignition characteristics of a
lean iso-octane/air mixture under and SACI conditions. Proc Combust Inst 34:2985–2993
28. Luong MB, Luo Z, Lu T, Chung SH, Yoo CS (2013) Direct numerical simulations of the ignition
of lean primary reference fuel/air mixtures with temperature inhomogeneities. Combust Flame
160:2038–2047
29. Kim SO, Luong MB, Chen JH, Yoo CS (2015) A DNS study of the ignition of lean PRF/air
mixtures with temperature inhomogeneities under high pressure and intermediate temperature.
Combust Flame 162:717–726
30. Luong MB, Yu GH, Chung SH, Yoo CS (2017) Ignition of a lean PRF/air mixture under
RCCI/SCCI conditions: a comparative DNS study. Proc Combust Inst 36:3623–3631
31. Yu GH, Luong MB, Chung SH, Yoo CS (2019) Ignition characteristics of a temporally evolving
n-heptane jet in an iso-octane/air stream under RCCI combustion-relevant conditions. Combust
Flame 208:299–312
32. An Y, Jaasim M, Vallinayagam R, Vedharaj S, Im HG, Johansson B (2018) Numerical simula-
tion of combustion and soot under partially premixed combustion of low-octane gasoline. Fuel
211:420–431
33. Mittal G, Sung C-J (2006) Aerodynamics inside a rapid compression machine. Combust Flame
145:160–180
34. Wang Y, Rutland CJ (2005) Effects of temperature and equivalence ratio on the ignition of
n-heptane fuel spray in turbulent flow. Proc Combust Inst 30:893–900
35. Kokjohn SL, Musculus MPB, Reitz RD (2015) Evaluating temperature and fuel stratification
for heat-release rate control in a reactivity-controlled compression-ignition engine using optical
diagnostics and chemical kinetics modeling. Combust Flame 162:2729–2742
36. Tang Q, Liu H, Yao M (2017) Simultaneous measurement of natural flame luminosity and
emission spectra in a RCCI engine under different fuel stratification degrees. SAE Int J Engines
10:2017–01–0714
Direct Numerical Simulation of Preignition … 335

37. Schießl R, Maas U (2003) Analysis of endgas temperature fluctuations in an SI engine by

laser-induced fluorescence. Combust Flame 133:19–27
38. Kaiser SA, Schild M, Schulz C (2013) Thermal stratification in an internal combustion engine
due to wall heat transfer measured by laser-induced fluorescence. Proc Combust Inst 34:2911–
2919
39. Söderberg F, Johansson B, Lindoff B (1998) Wavelet analysis of in-cylinder ldv measurements
and correlation against heat-release. SAE paper 107:444–456
40. Funk C, Sick V, Reuss DL, Dahm WJA, Turbulence properties of high and low swirl in-cylinder
flows, SAE paper (2002) 2002–01–2841
41. Miles PC (2008) Turbulent flow structure in direct-injection, swirl-supported diesel engines.
In: Flow and combustion in reciprocating engines, experimental fluid mechanics, Springer, pp
173–256
42. Akkerman V, Ivanov M, Bychkov V (2009) Turbulent flow produced by piston motion in a
spark-ignition engine. Flow Turbul Combust 82:317–337
43. Petersen BR, Ghandhi JB (2010) High resolution scalar dissipation and turbulence length scale
measurements in an internal combustion engine. SAE Int J Engines 3:65–83
44. Zeldovich YB (1980) Regime classification of an exothermic reaction with nonuniform initial
conditions. Combust Flame 39:211–214
45. Gu X, Emerson D, Bradley D (2003) Modes of reaction front propagation from hot spots.
Combust Flame 133:63–74
46. Bradley D, Morley C, Gu XJ, Emerson DR (2002) Amplified pressure waves during autoigni-
tion: Relevance to CAI engines. SAE Technical Paper (2002) 2002–01–2868
47. Lee JH, Knystautas R, Yoshikawa N (1978) Photochemical initiation of gaseous detonations.
Acta Astronautica 5:971–982
48. Lee JHS (2008) The Detonation Phenomenon. Cambridge University Press
49. Su J, Dai P, Chen Z (2020) Detonation development from a hot spot in methane/air mixtures:
effects of kinetic models. Int J Engine Res 146808742094461
50. Gao Y, Dai P, Chen Z (2020) Numerical studies on autoignition and detonation development
from a hot spot in hydrogen/air mixtures. Combust Theo Modell 24:245–261
51. Bates L, Bradley D, Paczko G, Peters N (2016) Engine hot spots: modes of auto-ignition and
reaction propagation. Combust Flame 166:80–85
52. Bates L, Bradley D (2017) Deflagrative, auto-ignitive, and detonative propagation regimes in
engines. Combust Flame 175:118–122
53. Peters N, Kerschgens B, Paczko G (2013) Super-knock prediction using a refined theory of
turbulence. SAE Int J Engines 6:953–967
54. Dai P, Chen Z, Chen S, Ju Y (2015) Numerical experiments on reaction front propagation in
n-heptane/air mixture with temperature gradient. Proc Combust Inst 35:3045–3052
55. Yu H, Chen Z (2015) End-gas autoignition and detonation development in a closed chamber.
Combust Flame 162:4102–4111
56. Dai P, Qi C, Chen Z (2017) Effects of initial temperature on autoignition and detonation
development in dimethyl ether/air mixtures with temperature gradient. Proc Combust Inst
36:3643–3650
57. Terashima H, Matsugi A, Koshi M (2017) Origin and reactivity of hot-spots in end-gas autoigni-
tion with effects of negative temperature coefficients: Relevance to pressure wave develop-
ments. Combust Flame 184:324–334
58. Pan J, Wei H, Shu G, Chen R (2017) Effect of pressure wave disturbance on auto-ignition mode
transition and knocking intensity under enclosed conditions. Combust Flame 185:63–74
59. Wei H, Chen C, Shu G, Liang X, Zhou L (2018) Pressure wave evolution during two hotspots
autoignition within end-gas region under internal combustion engine-relevant conditions. Com-
bust Flame 189:142–154
60. Sow A, Lee BJ, Hernández Pérez FE, Im HG (2019) Detonation onset in a thermally stratified
constant volume reactor. Proc Combust Inst 37:3529–3536
61. Pan J, Wei H, Shu G, Chen Z, Zhao P (2016) The role of low temperature chemistry in
combustion mode development under elevated pressures. Combust Flame 174:179–193
336 M. B. Luong and H. G. Im

62. Pan J, Dong S, Wei H, Li T, Shu G, Zhou L (2019) Temperature gradient induced detonation
development inside and outside a hotspot for different fuels. Combust Flame 205:269–277
63. Desai S, Sankaran R, Im HG (2019) Unsteady deflagration speed of an auto-ignitive dimethyl-
ether (DME)/air mixture at stratified conditions. Proc Combust Inst 37:4717–4727
64. Desai S, Sankaran R, Im HG (2020) Auto-ignitive deflagration speed of methane (CH4 ) blended
dimethyl-ether (DME)/air mixtures at stratified conditions. Combust Flame 211:377–391
65. Robert A, Richard S, Colin O, Martinez L, De Francqueville L (2015) LES prediction and
analysis of knocking combustion in a spark ignition engine. Proc Combust Inst 35:2941–2948
66. Robert A, Richard S, Colin O, Poinsot T (2015) LES study of deflagration to detonation
mechanisms in a downsized spark ignition engine. Combust Flame 162:2788–2807
67. Ali MJM, Luong MB, Sow A, Hernández Pérez FE, Im HG (2018) Probabilistic approach to
predict abnormal combustion in spark ignition engines. SAE paper 2018–01–1722
68. Wei H, Chen C, Zhou H, Zhao W, Ren Z (2016) Effect of turbulent mixing on the end gas
auto-ignition of n-heptane/air mixtures under IC engine-relevant conditions. Combust Flame
174:25–36
69. Chen L, Wei H, Chen C, Feng D, Zhou L, Pan J (2019) Numerical investigations on the effects of
turbulence intensity on knocking combustion in a downsized gasoline engine. Energy 166:318–
325
70. Zhong L, Liu C (2019) Numerical analysis of end-gas autoignition and pressure oscillation in
a downsized SI engine using large eddy simulation. Energies 12:3909
71. Zhang T, Sun W, Wang L, Ju Y (2019) Effects of low-temperature chemistry and turbulent
transport on knocking formation for stratified dimethyl ether/air mixtures. Combust Flame
200:342–353
72. Nogawa T, Terashima H (2020) Effects of globally stratified temperature distributions and NTC
characteristics on end-gas combustion modes. Combust Sci Technol 1–25
73. Wang L, Peters N (2006) The length-scale distribution function of the distance between extremal
points in passive scalar turbulence. J Fluid Mech 554:457–475
74. Pal P, Valorani M, Arias PG, Im HG, Wooldridge MS, Ciottoli PP, Galassi RM (2017) Computa-
tional characterization of ignition regimes in a syngas/air mixture with temperature fluctuations.
Proc Combust Inst 36:3705–3716
75. Luong MB, Lu T, Chung SH, Yoo CS (2014) Direct numerical simulations of the ignition of a
lean biodiesel/air mixture with temperature and composition inhomogeneities at high pressure
and intermediate temperature. Combust Flame 161:2878–2889
76. Luong MB, Yu GH, Lu T, Chung SH, Yoo CS (2015) Direct numerical simulations of ignition
of a lean n-heptane/air mixture with temperature and composition inhomogeneities relevant to
HCCI and SCCI combustion. Combust Flame 162:4566–4585
77. Zhang J, Luong MB, Hernández Pérez FE, Han D, Im HG, Huang Z (2019) Exergy loss of
dme/air mixtures and ethanol/air mixtures with temperature and concentration fluctuations
under HCCI/SCCI conditions: a DNS study, submitted to proceedings of the combustion insti-
tute (2019)
78. Luong MB, Yu GH, Chung SH, Yoo CS (2017) Ignition of a lean PRF/air mixture under
RCCI/SCCI conditions: chemical aspects. Proc Combust Inst 36:3587–3596
79. Luong MB, Sankaran R, Yu GH, Chung SH, Yoo CS (2017) On the effect of injection timing
on the ignition of lean PRF/air/EGR mixtures under direct dual fuel stratification conditions.
Combust Flame 183:309–321
80. Echekki T, Chen JH (2003) Direct numerical simulation of autoignition in non-homogeneous
hydrogen-air mixtures. Combust Flame 134:169–191
Numerical Investigation
of Flow-Acoustics Coupling
in a Half-Dump Combustor Using
Hybrid CFD-CAA Methodology

Bharat Bhatia and Ashoke De

Abstract The present work investigates the flow-acoustics coupling in a half-dump

combustor emulating methane-air non-premixed combustion for varying Reynolds
numbers (Re) and fuel flow rates. The study uses hybrid computational aero-acoustics
(CAA) simulation methodology, which is based on Lighthill’s acoustic analogy, to
compute the acoustic flow field. The underlying turbulent flow is modeled using
a hybrid Reynolds Averaged Navier Stokes/Large Eddy Simulation (RANS/LES)
approach, while the turbulence-chemistry interaction is handled using steady laminar
flamelet model (SFLM). Chemical kinetics for methane-air combustion is repre-
sented using the GRI-3.0 mechanism. In simulations, time-resolved data is analyzed
to understand the flow physics in greater detail, while the comparison is made with the
measurements to benchmark the numerical code used herein. In particular, the infor-
mation of non-dimensional numbers, i.e., Helmholtz and Strouhal numbers, allows to
characterize the instability modes and distinguishes between hydrodynamic (vortex
shedding modes) and acoustics instabilities (acoustic modes). The observation yields
that unsteady pressure fluctuations either come from the natural acoustics mode of
the channel or vortex shedding process and thereby exhibiting multiple frequencies in
the domain for lower Re case, while a single dominant frequency for higher Re case.
Moreover, the computed Rayleigh index using the unsteady heat release and pressure
oscillations from reacting conditions confirms the mode of instability, whether it is
driving (positive net value) or damping (negative net value), at different frequency
levels.

Keywords Hybrid RANS/LES · CFD-CAA · Flow-acoustics coupling

B. Bhatia · A. De (B)
Department of Aerospace Engineering, Indian Institute of Technology Kanpur, Kanpur 208016,
India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 337
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_15
338 B. Bhatia and A. De

1 Introduction

Flow-acoustics couplings or combustion instabilities in real-life combustion systems

such as gas turbines, rocket engines, burners, and furnaces are of paramount impor-
tance in the last couple of decades [1–11]. The reason is the generated noise and
exciting acoustic instabilities, which have a significant impact on the structural
loading in confined reacting systems. Vortex shedding reported being one of the
essential driving mechanisms of flow-acoustics coupling [5], where the interactions
between large-scale and small-scale turbulence structures lead to the generation of
periodic heat lease. The shear layer vorticity in dump combustor is primarily formed
due to the interaction of high and low-speed streams at the dump plane. Many studies,
including both experimental and computational approaches, have already reported
the importance of vortex structures on flow-acoustics coupling [1–11]. Also, some
studies reported combustion instability in practical combustor using a high fidelity
turbulence model, i.e., large eddy simulation (LES) [2, 3, 11].
In this work, we utilize a hybrid computational aero-acoustics (CAA) method-
ology, based on Lighthill’s analogy, for predictions of the acoustic flow field. Mainly,
this approach uses two-step procedures to compute the acoustic flow field such as
(a) firstly, the scale resolved turbulence calculations are done using high fidelity
turbulence model, and then, (b) the acoustic flow field is computed using the CAA
formulation invoking the results of the first step. In general, this is the most afford-
able approach in the current scenario as the direct computation of both the sources
and resulting acoustic fields is not a viable solution for practical systems [9, 12].
Moreover, we have invoked hybrid RANS/LES [9, 13] approach to model the under-
lying turbulent flow field. The hybrid approach utilizes the advantages of both RANS
and LES at reasonable computational cost, where the LES model resolves the sepa-
rated flow regions, and the RANS model becomes active close to the boundary layer
(attached to a solid surface). Hybrid RANS/LES has gained a lot of attention or an
excellent alternative option to LES for simulating flows with large-scale separation
where one can obtain better results with a reasonable computational cost.
Therefore, the primary focus of the present work is two-fold: (a) to access the
predictive capability of hybrid RANS/LES model in conjunction with CAA method-
ology and (ii) to analyze and shed light on the physics of flow-acoustics coupling
and the associated unsteady vortex dynamics in a half-dump combustor. The primary
aim is to differentiate between hydrodynamic mode of instability (low Re) with the
acoustic mode of instability (high Re), which is quite often mixed-up in the open
literature and termed as acoustic instability [9]. Also, the emphasis is put to establish
the theory behind low Re hydrodynamic instability (primarily due to unsteady vortex
dynamics) and high Re acoustic instability (mostly due to strong coupling between
heat release and pressure fluctuations). Another goal is to access the applicability
of a reduced-order model for practical applications. To achieve the said objective,
we have investigated a laboratory scale dump combustor for which measurements
are available for comparison and assessment [7]. Furthermore, we have analyzed
the unsteady flow features (heat release and pressure fluctuations) to estimate the
Numerical Investigation of Flow-Acoustics Coupling … 339

Rayleigh index, which confirms the onset of instabilities. Also, the detailed anal-
ysis of non-dimensional numbers (Helmholtz and Strouhal numbers) for varying
Reynolds number and fuel flow rates differentiates between two instability modes,
i.e., natural acoustic modes and hydrodynamics modes (vortex shedding modes) in
the combustor.

2 Numerical Details

In the hybrid approach, the hybrid RANS/LES model employs to compute the turbu-
lent flow field, while the sound pressure level is calculated using CAA formulation
at the next step. CFD simulations provide the required acoustics source information
to the CAA model to compute this far-field sound pressure level. The general form
of the Favre-filtered governing equations for the transport of mass, momentum, and
turbulence can be expressed as:
 
∂  ∂   ∂ ∂ φ̃
ρ̄ φ̃ + u j φ̃ =
ρ̄  ρ̄ D + Sφ  (1)
∂t ∂x j ∂x j ∂x j

where φ̃ represents the flow field variable solved for, and Sφ  represents the source
of φ̃. The scalar diffusion coefficient and mean density are represented by D and
ρ, respectively. The turbulence model provides turbulence scale information. In the
context of the present work, we have invoked the SST k-ω-based hybrid RANS/LES
model in conjunction with a steady laminar flamelet model (SLFM) [13]. The inherent
advantage of this hybrid method lies in the calculation of eddy viscosity using zonal
information. It automatically switches from RANS region to the LES region through
the following modification of dissipation term of the turbulent kinetic energy:

Yk = ρβ ∗ kωFhybrid (2)

where F hybrid is defined as:

 
lt
Fhybrid = max ,1 (3)
Cd max

where C d is the model constant with a value of 0.61, max is the maximum
 local
√ 
grid spacing (x , y , z ), and the turbulent length scale definition lt = 0.09∗ω
k

comes from RANS model. However, the grid-induced separation in some portion of
attached boundary layer (where the local surface grid spacing < the boundary layer
thickness, i.e., s < dδ, with d of the order one) remains a significant problem in
hybrid formulation, and this requires some special treatment to avoid such undesir-
able separation, or instead, the boundary layer needs to be protected. In the present
340 B. Bhatia and A. De

scope of the work, this is achieved through a blending function of the SST model as:
 
lt
Fhybrid = max (1 − FS ), 1 (4)
Cd max

With F S = 0, F 1 , F 2 , where F 1 , F 2 are the blending functions of the SST model,

while F S = F 2 provides the maximum level of protection against grid-induced
separation [13].

2.1 Turbulence-Chemistry Interaction Model

The local flame structure of the turbulent flames can be described as an ensemble of
laminar and one-dimensional local structures. With unitary Lewis number, the steady
equations can be expressed concerning mixture fraction Z and scalar dissipation rate
χ as follows:

χ ∂ 2 Yi ω̇i
0= + (5)
2 ∂Z 2 ρ
N N
χ 1 ∂ 2h χ 1 ∂ 2 Yk 1
0= − hk − h k ω̇k (6)
2 Cp ∂ Z2 2 Cp k=1
∂ Z 2 ρC p k=1

The connection between the physical space and mixture fraction space is achieved
through the scalar dissipation rate, which quantifies the deviation from equilibrium
and is defined as:
2
∂Z
χ = 2D (7)
∂x j

The scalar dissipation rate varies along the flamelet and is modeled as Eq. (7) as
depicted below:

2 2

χ (Z ) = χst exp 2 erfc−1 (2Z st ) − 2 erfc−1 (2Z ) (8)

where erfc−1 stands for the inverse of the complementary error function. The turbu-
lent flame brush is represented as the ensemble of diffusion flamelets, where the
Favre averaged temperature and species mass fraction of a turbulent flame can be
determined as:
¨
φ̃ = φ(Z , χst ) p(Z , χst )d Z dχst (9)
Numerical Investigation of Flow-Acoustics Coupling … 341

where a presumed β-PDF is used to define the probability of the mixture fraction.
The temperature and mean density have an extra dimension of mean enthalpy H 
to consider the non-adiabatic steady diffusion flamelets. The species mass fraction
is assumed to have an eligible effect by the heat loss or gain by the system. The
evolution of mixture fraction in the physical space is represented by the transport
equation of Z and Z  is given as:
 
D μeff
(ρ Z̃ ) = ∇. ∇ Z̃ (10)
Dt σt
   2
D  μeff  ε 2
2
(ρ Z ) = ∇. ∇Z 2 + C g μt . ∇ Z̃ − Cd ρ Z (11)
Dt σt k

where Z  = Z − Z̄ , μeff = μl +μt σt C g , and Cd are 0.85, 2.86, and 2.0, respectively.
More details regarding flamelet modeling can be found in the literature [13–16].

2.2 Acoustics Modeling Using Lighthill’s Analogy

(CFD-CAA)

Lighthill’s [17] analogy separates the analysis of acoustics into two steps. The first
step is sound generation induced by fluid flow in any real continuous medium. The
second step is sound propagation in an acoustic medium at rest, exerted by external
fluctuating sources, which are calculated from the first step. In Lighthill’s analogy,
the momentum equation is rewritten in the conservative form while leaving out the
viscous stresses as:
 
∂u i ∂u i ∂P
ρ + uj =− (12)
∂t ∂x j ∂ xi
 
∂ 2ρ
Eliminating ρu i and subtracting co2 ∂ xi ∂ x j
c02 (δ 2 ρ  /δxi δx j ) gives the Lighthill’s
equation

∂ 2ρ 2 
2 ∂ ρ ∂ 2 Ti j
− c = − (13)
∂t 2 o
∂ xi ∂ xi ∂ xi ∂ x j

where Ti j = ρu i u j + (P − co2 ρ)δi j . More details about the Lighthill’s analogy can
be found in the literature [13, 17].
342 B. Bhatia and A. De

Fig. 1 Computational domain: a 3D view with dimensions, b BCs and probing locations

2.3 Setup of the CFD Calculation

2.3.1 Computational Domain and Grid

Because of the nature of the combustor geometry, the simulations are performed
in a 3D configuration of the given combustor geometry [7], as depicted in Fig. 1a.
Figure 1b projects the 2D sectional view where different lengths are identified as
streamwise length L = L x = 48 h (includes the entrance/inlet section of L e = 14.67 h),
transverse length L y = 2 h, and spanwise length of L z = 2 h. Here, h represents
the step height, and the value of 30 mm has been used in the present calculations.
The computational domain consists of a structured mesh of hexahedral cells, where
the streamwise and spanwise directions use uniform grid spacing, and the transverse
direction invokes non-uniform distribution along with near-wall grid refinement. The
used grid (fine mesh) consists of 936 × 91 × 14 grid points in all three directions.
Also, two other meshes, having 624 × 61 × 9 grid points (medium mesh) and 312
× 31 × 5 grid points (coarse mesh) , have been considered for grid independence
analysis.

2.3.2 Boundary Conditions

The boundary of the computational domain at the highest axial distance is set to the
pressure outlet (Fig. 1b). For the current configuration, the Reynolds number based on
step height is defined as Re = Uo h/νRe = U0 h/ν, and the expansion ratio as ER =
L y / (L y − h) = 2. At the left inlet boundary (X = −14.6 h), the turbulent inlet velocity
profile is set using the information of turbulent boundary layer thickness at the step
(δ99/ h = 1.2). All the walls are no-slip walls, and the periodic boundary condition
is utilized along the spanwise directions. For stability, the Courant–Freidrich–Lewy
(CFL) condition determines the time step chosen for the simulations.
The time-resolved unsteady computations provide statistical data. To achieve
useful statistical information, we have collected large numbers of samples (~20
Numerical Investigation of Flow-Acoustics Coupling … 343

flow-through times) after completion of the initial transient phase (~10 flow-through
times). Figure 1b shows the fuel injection location, which is located 1 mm below
the step corner. The desired fuel flow (ṁ f ) rate is obtained through a square hole
of 0.13 h, which is centered along the spanwise direction. The unsteady signals of
pressure and heat release fluctuations are also monitored in consistency with the
measurement’s locations [7]. Figure 1b depicts three such points, which are located
at X p /L = 0.35, 0.49, and 0.83, where X p is the distance from the inlet of the compu-
tational domain, and L is the length of the combustor to maintain consistency with
the measurements in reference [7].

2.3.3 Computational Algorithm

The flow simulations are performed ANSYS FLUENT [13], where the pressure
and velocity coupling is achieved through the Semi-Implicit Method for Pressure
Linked Equations (SIMPLE) algorithm. The second-order discretization scheme is
consistently used for all the terms. Particularly, the convective fluxes of all the equa-
tions excluding momentum are discretized using the second-order upwind scheme,
while the bounded second-order central difference scheme is invoked for momentum
equations; while the diffusion term discretization involves second-order central
differencing. Time discretization employs a second-order implicit scheme, and the
methane-air combustion is represented using detailed chemistry, i.e., GRI 3.0 mech-
anism containing 52 species and 325 reactions [18]. The second-order discretization
scheme is consistently used for all the terms.

3 Results

In this section, we initially present non-reacting simulation results along with grid
independence and compare them with the measurements to have confidence in the
grid and turbulence model. Finally, we analyze the coupled reacting flow simulations
for varying fuel flow rates and Reynolds numbers to examine flow-acoustics coupling
in the chosen half-dump combustor.

3.1 Grid Independence and Non-reacting Flow Results

Here, we report two different cases for validation of the solver.

(a) Re = 5100

The backward facing step (BFS) geometry is schematically shown in Fig. 1. This
case was experimentally investigated by Jovic and Driver [19]. The Re based on step
344 B. Bhatia and A. De

Table 1 Recirculation
Re Experimental Coarse mesh Fine mesh (0.62 M)
bubble length (X r /h)
(0.22 M)
comparison for Re = 5100
5100 6.1 6.32 6.15

height is 5100. We have considered two different sets of grids: One is with coarser
gird of 0.22 M (million) cells and another one with fine resolution of 0.6 M (million)
cells. The computational domain has streamwise length L = L x = 30 h (includes the
entrance/inlet section of L e = 10 h), transverse length L y = 6 h, and spanwise length of
L z = 4 h. The expansion ratio of this case is ER = 1.2. At the left-most boundary, the
turbulent inlet velocity profile is imposed (δ99/ h = 1.2), while the rightmost boundary
is set to pressure outlet. No-slip boundary condition is used for the walls, the periodic
boundary condition along the spanwise directions, and the free-stream condition is
utilized at the top-most boundary along the transverse direction. The statistics are
collected over ~30 flow-through times after completion of the initial transient phase
(Table 1).
Figure 2 shows the predicted results of mean and rms of axial velocity at two
different axial locations using a fine grid only. Mean and rms velocity profiles are in
good agreement with the measurements, and the size of the recirculation bubble is
well predicted.
(b) Re = 18,000

The considered configuration is already illustrated in Fig. 1. The domain dimension

and grid details are already reported in Sect. 2.3. Here, we have analyzed three sets
of grids to achieve grid independence. Figure 3 depicts the mean and rms profiles of
axial velocity at different locations. The mean and rms profiles using medium and
fine grids are in excellent agreement with each other alongside the measurements
of [7]. However, the rms profiles using the coarse grid show some discrepancies,

Fig. 2 Mean and rms of axial velocity for Re = 5100. Symbols are measurements [19], and lines
are predictions
Numerical Investigation of Flow-Acoustics Coupling … 345

Fig. 3 Mean and rms of axial velocity for Re = 180,00. Symbols are measurements [7], and lines
are predictions

Fig. 4 Recirculation bubble

length (X r /h) for Re =
18,000. Lines represent the
streamwise velocity at first
cell above the wall

and these discrepancies are more visible in Fig. 4 which illustrates the recirculation
bubble predictions.
The velocity profiles are well captured in both the recirculation and recovery
regions. Necessarily, the turbulent activities are dominant in the recirculation regions
and decay gradually in the downstream locations. At downstream, the inner layer of
Reynolds stresses behaves like an ordinary turbulent boundary layer, whereas in the
outer layer, it decays slowly due to the presence of large eddies that are generated in
the upstream shear layer. Coarse mesh shows some over predictions in the recovery
region. Moreover, the length of the recirculation bubble using coarse mesh is X r /h
= 6.16, while the same using the other two meshes is X r /h = 6.03 and is in good
agreement with the experiments (X r /h = 6.01). Although the overall data predictions
using medium and fine meshes are in god agreement with measurements, the fine
346 B. Bhatia and A. De

mesh has been considered for details CAA calculations due to its better grid resolution
+
in wall units. The grid resolution of the fine mesh in wall units is x + = 2, ymin = 0.14,
+
ymax = 3, and z = 6, where ‘+’ superscript denotes near-wall units in terms of
+

friction velocity uτ and kinematic viscosity ν.

3.2 Reacting Flow Results Using CFD-CAA

The study on the reacting cases is conducted on various fuel and air mass flow
rates with detailed chemistry. In the detailed discussion, the results of reacting flow
simulations are compared with the results of non-reacting flow simulations (that is
cold flow simulations), which helps to segregate the frequency contribution due to
hydrodynamic factors. This result along with the instantaneous pressure and heat
release signal helps in the determination of the driving mechanism of combustion
instability. Moreover, the Rayleigh index is also plotted to support the data. The
simulation processes, for both reacting flow and non-reacting flow, are done for two
extreme cases with Reynolds number Re = 47,000 and Re = 18,000 while for the
intermediate Reynolds number (i.e., Re = 10,000, 26,000, 30,000, 32,000, 36,000,
and 45,000), only reacting flow simulations are carried out.
The numerical simulations are carried for eight different Reynolds numbers and
fuel flow, as shown in Table 2. First, the analysis of all the three fuel mass flow rate
cases is shown, which highlights the mass flow effect on the combustor character-
istics. Later, the detailed investigation is carried out for the fuel mass flow rate of
142 mg/s. For analyzing the acoustic signal at various locations across the domain,
three receiver probe points are located in a similar position as in the experiments. The
dominant frequency is noted at each probe. This frequency is used in the calculations
and analysis carried out for the simulations, including the calculation of Helmholtz

Table 2 Various cases for flow-acoustic analysis. Here, quantities marked with (*) are calculated
according to the given equation
Reynolds no. x 104 Fuel flow rate
142 mg/s 229 mg/s 316 mg/s
φ∗ MR* φ∗ MR* φ∗ MR*
1.0 0.04762 0.275882
1.8 0.027603 0.085149 0.044515 0.137318 0.061426 0.189486
2.6 0.019308 0.040811 0.031138 0.065815 0.042968 0.090819
3.0 0.017438 0.030654
3.2 0.015142 0.026942
3.6 0.013297 0.021287 0.021443 0.034329 0.02959 0.047372
4.5 0.011114 0.013624
4.7 0.010661 0.012489 0.017192 0.020141 0.023724 0.027792
Numerical Investigation of Flow-Acoustics Coupling … 347

(He) number and Strouhal (St) number. The definitions of He and St are recasted as:

H e = f ∗ L/c
St = f ∗ h/U (14)

where f is observed frequency, L is the length of the combustor, c is the speed of the
sound at the reference condition (inlet in this case), h is the height of the step at dump
plane, and U is the average airflow velocity at the inlet. The advantage of using these
non-dimensional numbers lies in the fact that the acoustic and the hydrodynamic
contributions to the frequency become discernible. Helmholtz number takes care of
the acoustics contribution while the Strouhal number looks into the hydrodynamic
factors.
(*)—The calculation of the equivalence ratio and the momentum ratio is carried
out by the following formula:

φ = s ∗ (F/A)
(ρu 2 ) F
MR = (15)
(ρu 2 )ox

3.2.1 Fuel Flow Rate of 142 Mg/S (ṁ f = 142 mg/s)

Figure 5 illustrates two non-dimensional numbers, i.e., Helmholtz number (He)

and Strouhal number (St), at different locations for varying Reynolds number (Re).
Notably, when the He corresponding to the dominant frequency at the probing point
is constant with the changing Reynolds number (Re), the frequency f also remains
constant, while the change of He with respect to Re indicates the role of hydrody-
namics as a contributor to the frequency f . The results indicate that for the lower
Reynolds number range, if the observed frequency of the flow is equivalent to the
vortex shedding frequency in the recirculation zone, then St decreases hyperbolically,
and He linearly increases, while St approaches a constant value at higher Reynolds
number. The results here show four data clusters in Helmholtz number plot in Fig. 5a.
The corresponding data can also be observed showing Strouhal number variation
against the Reynolds number range (Fig. 5b). The data matches the experimental
data quite well.
Figure 5c illustrates the schematic of the whole process of lock-on (represented
from [7]). The named clusters—I, II, III, IV represent the pattern of variation of the
dominant frequency with the increasing Re. Here, f a denotes the acoustic frequency
while f v denotes the vortex shedding frequency. Cluster I denotes the complete
hydrodynamic dominance, while Cluster II shows acoustic dominance. Cluster III
and IV exhibit the regime when the flow field gets lock-on the combustor acoustic
modes. Hence, the data points (Fig. 5a) in the nearly constant He (Cluster II) appear to
be transition region dominated by the acoustic modes, where the transitioning takes
348 B. Bhatia and A. De

II
III

IV

(a) (b)
Dominant frequency

fa3
fv1 fv2
fa2
fa1
amplitude

Reynolds number
(c) (d)

Fig. 5 a Helmholtz number, b Strouhal number at three points for ṁ f = 142 mg/s, c schematic
of the dominant frequency measured at the probe with varying Re showing flow-acoustics lock-on
during combustion, and d amplitude versus Re. Experimental data are taken from [7]

place from linearly increasing trend at lower Re (cluster I) to a linearly increasing

trend at higher Re (cluster III & IV). The 1st transition (Cluster I→II) is primarily
attributed to the vortex shedding at the dump plane, and it is similar to the case
of typical vortex shedding phenomena in the wake of oscillating bodies. With the
increase of Reynolds number in the lock-on regime, this vortex shedding frequency
shift into the natural frequency of the acoustic mode of the duct, as reported earlier [7].
The corresponding data points in St plot exhibit an asymptotic behavior to a constant
value (after decreasing hyperbolically) with increasing Re that signifies the nonlinear
rise in amplitude due to flow-acoustics lock-on. The dominant frequency for this floc-
acoustics lock-on happens to be the vortex shedding frequency at the step/dump plane.
Mostly, the data points in the linearly increasing portion of the plot (cluster I, III, IV)
are between successive vortex shedding modes, and the linear variation with different
slopes indicates different modes of vortex shedding, suggesting the possibility of
Numerical Investigation of Flow-Acoustics Coupling … 349

vortex pairing or merging in a lower shedding frequency (lower Strouhal number)

and consequently a lower slope. The observed dominant frequency corresponds to
the vortex shedding, as the vortex shedding frequency at the step gives rise to the St
= 0.2, which happens to be the frequency at which flow-acoustic lock-on takes place
[7, 20]. The decrease of Strouhal number (Cluster IV) to half of its value in cluster
III is indicative of the dominance of the vortex merging process in high Reynolds
number cases.
This interesting phenomenon is represented by the cluster (III & IV), which after
the initial linear growth of He (shown as cluster III), suddenly drops to cluster IV. This
is a feature of flow-acoustic lock-on (Fig. 5a). This difference of lock-on phenomena
is visible for just the clusters III and IV, and not clusters I. This means that although
the linearly increasing clusters III and IV show hydrodynamic behavior (just like the
cluster I) but they are controlled by the combustor acoustics (through the lock-on
phenomena) and accompanied by a steep rise in the amplitude (Fig. 5d).
The abrupt increase in the amplitude of noise corresponds to the flow-acoustics
lock-on at higher Re. Along with this amplitude plot (Fig. 5c), the schematic of flow-
acoustics lock-on (Fig. 5c) suggests the actual process of shifting the natural acoustic
mode of the duct into vortex shedding variation with increasing Re. The present
predictions using hybrid RANS/LES approach seem to be much better compared
to previous URANS predictions [21], where authors reported all the modes to be
acoustics mode at all Re. However, the present simulations capture two distinct
modes at lower Re due to hydrodynamic instability, and a single mode at higher
Re when flow-acoustics lock-on takes place/hydrodynamic instability merges with
acoustics instability as reported in the literature [20].

3.2.2 Fuel Flow Rate of 229 Mg/S (ṁ f = 229 mg/s)

A similar trend is observed in the higher fuel flow rate of 229 mg/s, where the four
Reynolds number cases are simulated (Re = 18,000, 26,000, 36,000, and 47,000).
Again, in this case, the results here indicate four data clusters in Helmholtz number
plot in Fig. 6a, like the previous case. The corresponding data can also be observed in
Fig. 6b showing Strouhal number variation against the Reynolds number range. The
results match the experimental data quite well. A notable difference is that the slope
of the linearly increasing cluster I, III, and IV has increased on the Helmholtz plot,
and the transition difference has decreased. This shows the higher sensitivity of the
vortex shedding phenomenon to the increasing Reynolds number. Further, a delay
of 3rd transition is also evident from both the figures, especially while looking at the
measurements [7]. However, the present simulations with 4 Re do not completely
cover the 3rd transition, as we have done for the previous case (Fig. 5). The frequency
of the vortex shedding can be known from the Strouhal number plot in Fig. 6b,
whereas the cluster II corresponding to the constant portion of the plot indicates the
dominance of acoustic modes in Fig. 6a.
350 B. Bhatia and A. De

II
III

IV

(a) (b)

Fig. 6 a Helmholtz number and b Strouhal number at three points for ṁ f = 229 mg/s.
Experimental data are taken from [7]

3.2.3 Fuel Flow Rate of 316 Mg/S (ṁ f = 316 mg/s)

For this case also, four Reynolds number cases are simulated as specified in Table 2.
The data agree with the experimental results quite well. In this case, the second
transition almost diminishes. Moreover, the third transition is delayed significantly,
also evident from both the figures. The frequency of the vortex shedding can be
known from the Strouhal number plot in Fig. 7b, while the largely flat region in
Fig. 7a indicates the domination of acoustic modes.

Fig. 7 a Helmholtz number and b Strouhal number at three points for ṁ f = 316 mg/s.
Experimental data are taken from [7]
Numerical Investigation of Flow-Acoustics Coupling … 351

3.2.4 Detailed Discussion for ṁ f = 142 mg/s

The literature strongly suggests that the movement of the recirculation bubble at
the dump plane is primarily the driving mechanism for flow-acoustics coupling in
the present case [8]. Primarily, the shear layer oscillations at the downstream of the
dump plane also modulate the heat release fluctuations and lead to the generation of
acoustics oscillations. Also, the frequency of this shear layer oscillations matches
with the natural acoustics modes of the combustor. These two processes are in close
resonance with each other through the symptomatic rise and fall of the amplitudes.
When the periodic pressure and heat release fluctuations are in phase, it excites the
high-amplitude pressure oscillations. The presence of such high-amplitude pressure
oscillations leads to the generation of acoustics oscillations, and sustenance becomes
stronger in the presence of heat generation, as reported by Rayleigh [22]. Thus, this
acoustic excitation becomes stronger when the unsteady pressure and heat release
fluctuations get into the same phase. The most widely used criterion for such acoustics
excitation is known as Rayleigh’s criterion and is expressed using local Rayleigh
Index (RI) over one cycle of instability as [23]:


t+(2π/ω)
1
RI = p  q  dt (16)
T
t

where p  and q  are the pressure and heat release oscillations, T is the time for one
cycle of instability, and ω is the frequency in rad/s. In general, the RI is a good marker
to identify the growth or damping of instabilities under reacting conditions. The
positive value of RI indicates the coupling of pressure and heat release fluctuations to
give rise to amplified pressure fluctuations, while the negative value of RI represents
the damping of the same.
Before moving the detailed identification of the frequencies, let us examine how
the natural acoustic modes of the present combustor can be estimated. Here, one can
use the simple relationship to evaluate the natural acoustic modes of the combustor as
   2  2 1/2
2
n
f = c2 nlxx + l yy + nlzz , where nx , ny , nz are the order of modes, c is the
speed of sound, and lx , l y , l z are the length of the domain in three directions. Due to the
size of the combustor geometry, there exists no dominant modes along the transverse
or spanwise directions as the first mode in ‘y/z’ direction is ~2900 Hz under cold
condition, i.e., T = 300 K. Hence, it is understood that the only longitudinal modes  are
dominant in this combustor and can approximately be estimated as f = 2 lx . The c nx

first few natural acoustics modes along the longitudinal direction are approximately
~120.5, 241, 361.5, 602.75, 723.31, 843.8, 964.4, 1085, 1205.51 Hz, and so on.
Here, we restrict our discussion two Re cases, i.e., Re = 47,000 and 18,000. For
both cases, we will present reacting flow results followed by the non-reacting results
to establish the resemblance between the instability mode and driving mechanism.
352 B. Bhatia and A. De

All these cases, the sampling rate is 50 kHz (t = 2 × 10−5 s), while a large number
of sampling size and block size are considered for calculations.
Figure 8 illustrates the predictions for Re = 47,000 at two extreme locations in
the domain. As observed, multiple dominant frequencies (58, 148, 293, 567, 713,
860, 1007, 1152, 1302, 1573, 1702, 1866 Hz, and so on in the range of 2000 Hz)
are observed from the PSD plots while the observed frequencies are in similar range
at two locations. Since there exists feedback between pressure and heat release fluc-
tuations at the natural acoustic mode of the combustor, the spectra exhibit multiple
peaks at these locations, which are either harmonics or superposition of two different

Fig. 8 Reacting CFD-CAA results for Re = 47,000 (ṁ f = 142 mg/s) at two locations: PSD
versus frequency, HRR and pressure fluctuations versus time, PSD of pressure and HRR versus
frequency
Numerical Investigation of Flow-Acoustics Coupling … 353

Fig. 9 Non-reacting CFD-CAA results for Re = 47,000 at two locations: PSD of pressure versus
frequency

modes. To identify the driving force, one has to look at the complete spectra of both
pressure and heat release, as depicted in Fig. 8 at both locations. Two points, i.e.,
X p /L = 0.35 belongs to the vortex shedding region, and X p /L = 0.83 is far away from
the vortex shedding region, show similar behavior. While looking at pressure and
heat release fluctuation together, one can notice the most correlated frequency (i.e.,
beat frequency) is ~58 Hz. In particular, this frequency gets generated primarily due
to the upstream shear layer oscillations, i.e., movement of the recirculation bubble.
Analyzing the other observed peaks in spectra, one can notice that they either belong
to the natural acoustic mode of the combustor or to a superposition of natural acoustic
mode and lock-on mode (58 Hz). For example, 293 Hz is the superposition of second
natural mode + lock-on mode (~241 + 58 Hz), 567 Hz is the superposition of 4th
natural mode + lock-on mode (~482 + 58 Hz), 713 Hz and 860 Hz are approximately
sixth and seventh natural mode, 1007 Hz is the superposition of 8th natural mode +
lock-on mode (~964 + 58 Hz), and similarly, other modes can also be identified.
While correlating these spectra with the cold flow results in Fig. 9, noticeably, the
reacting flow results strongly resemble the cold flow predictions. As usual, multiple
peaks, including some low amplitude peaks in the vortex shedding region, are also
present in cold flow results (Fig. 9), but those mostly include the ones which are
dominant in reacting flows (Fig. 8). Interesting to note here is that the cold flow
predictions also exhibit the lock-on frequency of 58 Hz which confirms the exci-
tation of natural acoustic mode with lock-on to vortex shedding mode in the shear
layer. The presence of low amplitude peaks supports the cyclic behavior of shear
layer vortex (at step) due to recirculation and flow reversal.
To further support the above observation, the net value of the local Rayleigh
index is estimated to confirm that heat release and pressure oscillations are in phase.
Figure 10 illustrates the local Rayleigh index (RI), which appears to be positive for
most of the dominant frequencies obtained for this Reynolds number, which in turn
indicates the feedback between pressure and heat release fluctuations. The positive
value of RI at the most correlated frequency (58 Hz) strongly suggests the excitation
of acoustic mode with lock-on to the vortex shedding mode in the shear layer region
[7, 8]. Thus, this correlated frequency is termed as acoustic locked-on frequency.
354 B. Bhatia and A. De

Fig. 10 Reacting CFD-CAA

results: rayleigh index for Re
= 47,000 for X p /L = 0.83

Moreover, the positive value of RI indicates that instability is driving (i.e., positive
net value) at most dominant frequencies, and the presence of acoustic locked-on
frequency and acoustic modes shows that acoustics is the driving mechanism of
combustion instability at this Reynolds number. This finding indicates that the insta-
bility or flow-acoustics coupling in this combustor primary takes place due to shear
layer oscillations (vortex shedding) coupled with unsteady heat release fluctuations
at the dump plane.
While analyzing the data for Re = 18,000, one can notice completely different
behavior, as illustrated in Fig. 11. The observed dominant frequencies are also
different at different locations, i.e., 42 Hz and 136 Hz, at X p /L = 0.35 and 0.83,
respectively. The different dominant frequencies at different locations confirm the
excitation of hydrodynamic instabilities without locking on to the acoustics mode.
The hydrodynamic mode appears primarily due to shear layer oscillations at the dump
plane (motion of recirculation bubble). The first frequency (42 Hz) arises due to the
motion of the recirculation bubble, while the second one (136 Hz) is approximately
the first bulk mode frequency of the combustor duct. Hence, unlike the higher Re
case, the pressure and heat release fluctuations are not in phase and do not attribute to
the resonant acoustic mode (Fig. 11). The plot local RI supports the argument also, as
illustrated in Fig. 12, as no feedback exists between pressure and heat release rate at
this Re. It means the combustion at this Re is entirely stable, and the instabilities are
hydrodynamic in nature. Moreover, the non-reacting spectra (Fig. 13) further bolsters
this argument as the observed dominant frequencies are the same as of reacting cases
(Fig. 11). Hence, the mode of instability is purely hydrodynamic in nature at lower
Re, while flow-acoustics lock-on takes place at higher Re.
Numerical Investigation of Flow-Acoustics Coupling … 355

Fig. 11 Reacting CFD-CAA results for Re = 18,000 (ṁ f = 142 mg/s) at two locations: PSD of
pressure and HRR versus frequency, HRR and pressure fluctuations versus time

Fig. 12 Reacting
CFD-CAA results: Rayleigh
Index for Re = 18,000 for
X p /L = 0.83
356 B. Bhatia and A. De

Fig. 13 Non-reacting CFD-CAA results for Re = 18,000 at two locations: PSD of pressure versus
frequency

3.2.5 Instability Map

Further, the ratio of flow timescale to the acoustic timescale is calculated for this
combustor. This ratio is equal to χ = St/H e = Lh M1 , where M is the Mach
number of the airflow at the combustor inlet (Table 2). The characteristic length for
the acoustic oscillations is the length of the combustor, while the characteristic length
for the shear flow or heat release process is the step height. The data are taken from
all the three receiver probe points in a numerical domain for the three mass flow rates
at different Reynolds number. The plot χ against the equivalence ratio φ indicates a
map of the combustor’s performance with respect to its acoustic stability behavior,
as shown in Fig. 14. The flow-acoustics interaction is found to occur when the order
of χ is equal to unity or 1/n, where n is the integer multiple of natural acoustic mode.
This suggests the ratio of the hydrodynamic timescale to the acoustics timescale of
the chamber.
The map explains the dominance of acoustics in higher Re or lower φ flows. It
is similar to what has been discussed earlier on St and He plots. For all the three
fuel flow rates, St is found to depend hyperbolically on flow velocity, in other words,
directly proportional to 1/u. This observation for all the fuel flow rates is clearly visible
together in this single instability map (Fig. 14), because as φ (depends hyperbolically
on flow velocity) decreases, χ (= St/He) also decreases.
As discussed earlier, the data corresponding to two lower most Re cases (Re =
10,000 and Re = 18,000) lying in cluster I of Fig. 5a show the complete dominance
of the hydrodynamics. These two cases, hence, have a value of χ nearer to or greater
than 1.0 and are far apart from the rest of the higher Re data points. The predicted
results clearly indicate that the shear layer oscillation is the primary reason behind the
instabilities associated with this half-dump combustor. At this lower Re, the acoustic
Numerical Investigation of Flow-Acoustics Coupling … 357

Fig. 14 Map for acoustic

instability for three locations
for different fuel flow rates

pressure does not get coupled with the shear layer oscillation to create a feedback
loop between flow and acoustics. Hence, the hydrodynamic mode of instability is
the dominant one without flow-acoustics coupling at lower Re, whereas at higher
Re cases (of cluster III and IV in Fig. 5a), the value of χ decreases to less than 0.5,
and the acoustics pressure couples with the shear layer oscillation and provides a
necessary feedback loop between flow and acoustics. Hence, flow-acoustic lock-on
happens and acoustic instability becomes the dominant mode of instability at higher
Re.
As expected, the curve is found to shift rightwards with an increase in the fuel
rate or equivalence ratio, but with a gentler slope. It indicates a lesser differences
in the domination of the acoustics or the hydrodynamic factors for their respective
cases of Re. This is evident in the previous plots of He (Fig. 5a, 6a and 7a), where
the 2nd transition second difference vanishes as the fuel flow rate is increased, and
at the highest case of fuel flow rate in this study, it is only acoustics dominated.
It is understood that under the turbulent reacting condition, the heat release (shear
layer burning) becomes periodic due to the energy cascading process (vortex breaks
down to small scales) and becomes higher with increasing turbulent conditions
(higher Re cases). Henceforth, this periodic heat release, in particular, feeds back
the necessary energy to acoustics pressure to have flow-acoustics lock-on situa-
tion at higher Re cases as compared to lower Re cases. This also stands for the
acoustics-dominated cases of high-fuel flow rate where the heat release is also high.

4 Conclusions

The present work reports the flame-acoustics coupling behavior in a half-dump

combustor using numerical simulations. Hybrid RANS/LES methodology is used
358 B. Bhatia and A. De

for the simulation of the turbulent flow to increase computational efficiency. Further,
the steady laminar flamelet model (SLFM) is used for turbulence-chemistry interac-
tion for the reacting flow case. Using CAA methodology, the unsteady pressure and
heat oscillations are monitored at specific locations to get the information of the flow
features and the calculation of Helmholtz and Strouhal numbers, Rayleigh Index,
at different frequencies. From the analysis of Helmholtz and Strouhal numbers,
it is concluded that the hydrodynamic frequency dominates at smaller Reynolds
numbers, while the flow-acoustics lock-on takes place at higher Reynolds numbers.
The increase in fuel flow rate delays the transition and decreases the abrupt difference
in each transition. Further, the detailed analysis confirms the existence of a correla-
tion between pressure fluctuations and heat release at different Reynolds numbers.
The observation suggests that acoustics is the driving mechanism for the combustion
instability at high Reynolds number, while hydrodynamics is the driving mechanism
for combustion instability at low Reynolds.

Acknowledgements Financial support for this research is provided through Science and Engi-
neering Research Board (SERB), India. Simulations are carried out on the computers provided by
the Indian Institute of Technology Kanpur (IITK) (www.iitk.ac.in/cc), and the manuscript prepa-
rations, as well as data analysis, have been carried out using the resources available at IITK. This
support is gratefully acknowledged.

References

1. Candel S (2002) Combustion dynamics and control: progress and challenges. Proc Combust
Inst 29(1):1–28
2. Yu KH, Trouve AC, Daily JW (1991) Low frequency pressure oscillations in a model ramjet
combustor. J Fluid Mech 232:47–72
3. Menon S (1992) Active combustion control in a ramjet using large-eddy simulations. Combust
Sci Technol 84:51–79
4. Poinsot T, Veynante D (2005) Theoretical and Numerical Combustion. R. T. Edwards,
Philadelphia
5. Schadow SC, Gutmark EJ (1992) Combustion instability related to vortex shedding in dump
combustors and their passive control. Prog Energy Combust Sci 18:117–132
6. Sivasegaram S, Whitelaw JH (1987) Oscillations in axi-symmteric dump combustors. Combust
Sci Technol 52:413–426
7. Shreenivasan OJ (2009) Onset of acoustic instability by vortex shedding in a non-premixed
combustor, Ph.D. dissertation, Indian Institute of Technology, Chennai, India
8. De A, Kushari A, Sudharsan K, Kondreddy G (2016) CFD-CAA simulation of flow acoustic
coupling in a half-dump combustor. J Propul Power 32(1):95–107
9. David P, An overview of computational aero acoustic modeling at NASA Langley, Aerody-
namics, Aerothermo-dynamics, and Acoustics Competency, NASA Langley Research Center,
Hampton, VA 23681-2199, U.S.A
10. Kailasanath K, Gardner JH, Boris JP, Oran ES (1987) Numerical simulations of acoustic vortex
interactions in a central dump ramjet combustor. J Propul Power 3:525–533
11. Angelberger C, Egolfopoulos F, Veynante D (2000) Large eddy simulations of chemical and
acoustic effects on combustion instabilities. Flow Turbul Combust 65(2):205–220
12. Christopher KW, Tam (2006) Recent advances in computational aeroacoustics. Fluid Dyn Res
38(9):591–615
Numerical Investigation of Flow-Acoustics Coupling … 359

13. ANSYS Fluent 18.0 User’s guide, Canonsburg, PA, USA

14. Saini R, De A (2017) Assessment of soot formation models in lifted ethylene/air turbulent
diffusion flame. Therm Sci Eng Prog 3:49–61
15. Saini R, Prakash S, De A, Yadav R (2018) Investigation of NOx in piloted stabilized methane-
air diffusion flames using finite-rate and infinitely-fast chemistry based combustion models.
Therm Sci Eng Prog 5:144–157
16. Reddy M, De A, Yadav R (2015) Effect of precursors and radiation on soot formation in
turbulent diffusion flame. Fuel 148:58–72
17. Lighthill J (1978) Acoustic streaming. J Sound Vib 61(3):391–418
18. Smith GP, Golden DM, Frenklach M, Moriarty NW, Eiteneer B, Goldenberg M, Bowman CT,
Hanson RK, Song S, Gardiner WJ, Lissianski VV (2000) GRI 3.0, Gas Research Institute,
Chicago, IL. http://www.me.berkeley.edu/gri_mech
19. Jovic S, Driver D (1994) Backward-facing step measurements at low Reynolds number, Reh
= 5100, NASA Ames Research Center
20. Chakravarthy SR, Obla J (2007) Experimental characterization of onset of acoustic instability
in a non-premixed half dump combustor. J Acoust Soc Am 122(1):120–127
21. Nath A (2012) Unsteady reynolds-averaged navier-stokes simulation of flow-acoustic coupling
in a half-dump combustor. Master’s dissertation, Indian Institute of Technology, Chennai, India
22. Rayleigh JWS (1945) The theory of sound. Dover, Newyork
23. Putnam AA (1971) Combustion driven oscillations in industry. Elsevier Publishing Co, New
York
Emissions from FEHA Fueled RQL Gas
Turbine Combustor Under
High-Pressure and High-Temperature
Simulated in-Flight Conditions

Hitoshi Fujiwara and Keiichi Okai

Abstract Combustion rig tests of a rich burn–quick quench–lean burn (RQL) gas
turbine combustor and of a single concentric lean-burn combustor were performed
using kerosene (Jet A1)100% hydro-treated ester and fatty acid (HEFA) alternative
jet fuel. The results showed a clear tendency on the suppression of non-volatile PM
(nvPM) emission with increase in blending ratio of HEFA. A flight mission analysis
utilizing the present experimental data gave emission profiles over the flight path.

Keywords Hydro-treated ester and fatty acid (HEFA) · Fuel property ·

Combustion test · Particulate matter · Flight mission analysis

Nomenclature

ATM Air Traffic Management

ASTM American Society for Testing and Materials
CO Carbon monoxide
CO2 Carbon dioxide
EI Emission Index
FAME Fatty and methyl-ester fuel
FT-SPK Fischer–Tropsch Synthetic Paraffin Kerosene
HEFA Hydro-treated ester and fatty acid
JAXA Japan Aerospace Exploration Agency
LTO Landing and takeoff
NEDO New Energy Industrial Technology Development Organization
NOx Nitrogen oxides
nvPM Non-volatile particulate matter
ROC Rate of climb
RQL Rich burn Quick quench and Lean burn

H. Fujiwara (B) · K. Okai

Propulsion Research Unit, Aeronautical Research Directorate, Japan Aerospace Exploration
Agency, 7-44-1 Jindaiji-Higashi, Chofu 182-8522, Tokyo, Japan
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 361
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_16
362 H. Fujiwara and K. Okai

SPK Synthetic paraffin kerosene

TEC Toyo Engineering Cooperation
THC Total hydrocarbons
TRA Technology Reference Aircraft
UOP Universal oil products

1 Introduction

Global climate change due to the rapid increase in CO2 emissions especially caused
by aviation is one of the critical issues that can be solved through international
collaborations. Although the amount of CO2 emissions from aviation consists of
only around 2% of the total CO2 emissions, it is of crucial and important to start
suppressing emissions caused by air traffic as early as possible. Recent rapid growth
of aviation transportation has caused many activities that could curtail CO2 reduction
form aviation. In other fields, much efforts are in progress, such as electric and hybrid
vehicles in the automotive sector, and carbon dioxide capture and storage (CCS)
installations in power generation plants. It should be noted that more than 90% of
CO2 emissions from commercial aircraft operations are generated by large aircraft,
indicating areas of pursuing future research to reduce commercial aircraft emissions
with a focus on technology applicable to large commercial aircrafts [1].
In aviation, alternative fuels are considered one of the important options to
suppress CO2 emission. Their specifications are strictly defined in ASTM D7566
“Standard Specification for Aviation Turbine Fuel Containing Synthesized Hydro-
carbons” [2]. Its Annexes define not only the chemical and physical properties but
also the manufacturing process of those fuels, which is of prime importance from
the aviation safety point of view.
The JAXA investigations started in 2010 on the bio-derived aviation alternative
fuels. At JAXA, the impact of introducing bio-fuels in turbofan jet engines was
investigated [3] and combustion testing initiated in 2011 with an available bio-fuel
(that was not certified) called FAME fuel (Fatty acid methyl-ester) [4]. As expected,
the fuel had higher CO emission at low load conditions due to low flame temperatures
[5].
In 2017, Mitsubishi Power along with their partners (Toyo Engineering Coop-
eration (TEC), JERA and JAXA) started a project to conduct a pilot-scale plant
testing on a Fischer–Tropsch synthetic paraffin kerosene (FT-SPK) fuel produced
from lignocellulosic biomass. This project was supported by NEDO [6]. For this
project, JAXA participated in conducting the final combustion tests of the (bio-fuel)
product.
This chapter presents the results of our investigations on the aviation-certified
bio-fuels to understand the limitations and potentials of bio-derived aviation fuels as
a contribution toward more environmentally friendly aviation. It further focuses on
emission characteristics with Jet A1 and HEFA fuels using same high-temperature
Emissions from FEHA Fueled RQL Gas Turbine Combustor … 363

and high-pressure that simulated in-flight test conditions. A simplified flight mission
analysis was conducted along with data provided on the emission profiles over the
flight path.
The alternative turbine fuel used in this study was hydro-treated ester and fatty
acid (HEFA) made from tallow fat. The fuel was provided by Honeywell UOP/Nikki
Universal. The manufacturing process for this fuel is specified in Annex 2 of ASTM
D7566. The HEFA fuel was available for use in commercial flights, if the blending
ratio did not exceed 50 vol.% as per ASTM D7566.
The specifications for HEFA fuel are provided in ASTM D7566 Annex 2 [7]. To
understand the similarities and differences between HEFA fuel and the baseline Jet
A1 fuel, the properties of HEFA fuel were analyzed. Table 1 summarizes the results.
Several considerations on the analysis are summarized below:
1. HEFA showed a very low freezing point, which is considered to be caused by
branched paraffins.
2. Dynamic viscosity and surface tension of HEFA and baseline jet fuel differed
only a little from each other. The slight differences might affect the atomization
behavior of HEFA fuels as compared to the baseline fuel.
3. The density of HEFA is lower than that of baseline jet fuel.
4. The lubricity of HEFA is lower than that of baseline jet fuel due to lack of sulfur
content.
5. The most remarkable difference between the two fuels lie in its aromatics
content. HEFA contains fewer aromatic compounds, while the baseline jet fuel
contains about 20% of aromatic compounds on volume basis, which consists of
many benzene structure compounds and its derivatives that are low polycyclic
aromatics hydrocarbons (PAHs).

2 Experimental Apparatus and Conditions

A schematic diagram of the single burner combustion rig test is shown in Fig. 1.
Photograph of the JAXA AP7 high-pressure test facility is shown in Fig. 2. The
facility allows combustion testing at maximum inlet temperature, pressure, and air
mass flow of 1000 K, 40 bar, and 4 kg/s, respectively. This test rig is usually used
for the demonstration of a new combustor concept and for the development of some
innovative measurement technologies. An exhaust gas sampling probe having eight
(ϕ = 0.8) sampling holes was located at the exit of the combustor liner.
Exhaust sample gas was introduced to the measurement instruments using a
stainless-steel pipe that had a valve to control the mass flow rate and temperature
of the sample gas prior to introducing the gas sample to the instrument. NOx was
measured using a chemiluminescence detector (CLD), CO and CO2 were measured
using nondispersive infrared (NDIR) detectors, and THC was measured using a
flame ionization detector (FID). They were incorporated in the Horiba MEXA ONE
analyzer. Particulate matter (PM) mass was measured using a photo-acoustic soot
sensor (PASS) in AVL MSS 483. PM number was measured through condensation
particle counter (CPC).
364 H. Fujiwara and K. Okai

Table 1 Chemical and physical analysis of the fuels

Properties HEFA D7566 Table Jet A1 Jet A1: HEFA D7566 Test
A2.1-2 = 50:50 Table 1 method
Freezing point −58.5 <−40 −48.0 −51.5 <−47 JIS K
(°C) 2276
Dynamic 1.963 N/A 1.655 1.786 <8 (−20 °C) JIS K2283
viscosity
[mm2 /s] @15
°C
Surface 22.5 N/A 23.6 23.1 N/A JIS K
tension 2241
[mN/m]
Density 0.7554 0.73-0.77 0.7886 0.7732 0.775-0.840 JIS K
[g/cm3 ] @15 2249-1
°C
Net heat of 44.14 N/A 43.38 43.73 >42.8 JIS K
combustion 2279
[MJ/kg]
Physical JIS K
distillation 2254
Initial boiling 146.5 N/A 148.5 148.5 N/A
point [°C]
10% 165.0 <205 164.5 165.0 <205
recovered
temp. [°C]
50% 208.0 Report 193.5 198.5 Report
recovered
temp. [°C]
90% 253.5 Report 237.5 247.0 Report
recovered
temp. [°C]
Final boiling 269.0 <300 259.0 262.0 <300
point [°C]
Lubricity 1.04 N/A 0.83 0.87 <0.85 ASTM
[mm] D5001
Carbon 84.7 N/A 86.1 85.4 N/A JPI-5S-65
[mass%]
Hydrogen 15.2 N/A 13.8 14.5 N/A
[mass%]
Sulfur <0.0001 <0.0015 0.0006 0.0003 <0.3 JIS K
[mass%] 2541-6
Composition ASTM
Paraffin [vol. 98.4 Report 79.1 89.1 N/A D1319
%]
(continued)
Emissions from FEHA Fueled RQL Gas Turbine Combustor … 365

Table 1 (continued)
Properties HEFA D7566 Table Jet A1 Jet A1: HEFA D7566 Test
A2.1-2 = 50:50 Table 1 method
Olefin [vol. 0.9 N/A 3.0 1.6 N/A
%]
Total 0.7 N/A 17.9 9.3 8-25
aromatics
[vol. %]
Aromatics ASTM
Benzene [vol. <0.1 N/A 19.1 10.2 N/A D6379
%]
[mass%] <0.1 N/A 21.3 11.6 N/A
Naphthalene <0.1 N/A 0.1 0.1 N/A
[vol. %]
[mass%] <0.1 N/A 0.2 0.1 N/A
Total <0.1 N/A 19.2 10.3 8.4-26.5
aromatics
[vol. %]
[mass%] 0.1 N/A 21.5 11.8 N/A

Fig. 1 A schematic diagram of the combustion test rig

366 H. Fujiwara and K. Okai

Fig. 2 Photograph of the JAXA AP7 high-pressure combustion test rig

Fig. 3 RQL combustor

Figure 3 shows the RQL combustor [8] examined in this study. A Parker-Hannifin
type air blast fuel nozzle, as shown in Fig. 4, was used for the injection of fuel into
the combustor. In tests, 10% of total air flow entered through the upstream while the
remaining 90% of total air entered through the air holes on the liner (as shown in
Emissions from FEHA Fueled RQL Gas Turbine Combustor … 367

Fig. 4 Parker-Hannifin type air blast fuel nozzle

Fig. 3) that served as combustion/dilution air. The fuel nozzle had only one fuel inlet
and no fuel staging were performed.
Inlet air conditions, shown in Table 2, were tested, wherein the total air/fuel ratio
(AFR) ranged from approximately 38–120. Inlet air conditions were determined, so
that the emissions close to in-flight conditions can be obtained. The present efforts
at JAXA had been in force since 2013 with focus on engine technology development
and is called “Research and Development for Green Engine Technology.” Figure 5
shows the present condition in reference to the target of “Research and Development
for Green Engine Technology.” The target engine is a 20,000 lbf class turbofan engine
with bypass ratio of 13.

Table 2 Experimental condition

Case # Representation Inlet temperature Inlet pressure (kPa) Pressure loss ratio
(K) (%)
Case 1 Idle (7% MTO) 524 548 2.5
Case 2 Approach 644 1350 4.6
(30%MTO)
Case 3 Cruise 739 1350 4.1
Case 4 Climb 803 1350 4.2
temp. 13.5 bar
Case 5 Climb 803 2000 4.2
temp. 20 bar
Case 6 MTO 837 2000 4.1
temp. 20 bar
368 H. Fujiwara and K. Okai

Fig. 5 Experimental
conditions and referenced
R&D target

Fig. 6 Emission index (EI)

of CO emission
characteristics for Jet A1 and
HEFA fuels

3 Experimental Results1

Combustion tests with fuel supply of 100% HEFA or 100% jet fuel were performed to
investigate the effects of fuel change from jet fuel to HEFA. Figure 6 shows emission
index (EI) for CO as a function of the air/fuel ratio. The EI CO decreased with
increase in of thermal load intensity for both the fuels. A remarkable difference can
be seen for the idle condition as compared to the other conditions. The results show
that for each condition, the emission characteristics are similar from the two fuels.

1From AIAA 2020–1220 [9]; reprinted by permission of the American Institute of Aeronautics
and Astronautics, Inc.
Emissions from FEHA Fueled RQL Gas Turbine Combustor … 369

Fig. 7 Emission index (EI)

of NOX emission
characteristics for Jet A1 and
HEFA fuels

Figure 7 shows the emission index (EI) of NOX as a function of air/fuel ratio. The
EI NOX increased with the increase of pressure and temperature for both the fuels.
For each condition, the emission characteristics were observed to be similar for the
two fuel cases.
Figure 8 shows the emission index (EI) of THC as a function of air/fuel ratio. The
EI THC decreased with increase of thermal load intensity for both the fuels, similar
to that observed for EI CO. For each condition, the emission characteristics were
similar for the two fuel cases.
Figures 9 and 10 show mass (mg/m3 ) and number of nvPM (particles/cm3 ) emis-
sion, respectively. NvPM mass for each fuel generally increased with the increase in
thermal load intensity, but this tendency is rather complex. The reduction of nvPM
mass with HEFA was more critical than the number of nvPM. These results are in
good agreement with the previous study by the authors [10].

Fig. 8 Emission index (EI)

of THC emission
characteristics for Jet A1 and
HEFA fuels
370 H. Fujiwara and K. Okai

Fig. 9 PM mass emission

(mg/m3 ) characteristics for
Jet A1 and HEFA fuels

Fig. 10 PM number
emission (particles/cm3 )
characteristics for Jet A1 and
HEFA fuels

In order to have a better understanding of the PM emission characteristics, pressure

and temperature effects were investigated. Figure 11 shows the effect of pressure
on PM emission characteristics at a constant temperature condition of 803 K. These
figures show that as the pressure increased both mass and number of particle emission
decreased. However, the pressure effect is less eminent in the PM number case.
Figure 12 shows the effect of temperature on PM emission characteristics at
constant pressure condition of 1350 kPa. In the case, temperature effect on PM mass
was not as clear as that for the constant temperature case. However, the results clearly
showed that the PM number decreased with increase in temperature.
There are a number of reports available on the soot-emission characteristics for
both PM mass and number of particles in combustion of conventional jet fuel and
certified bio-fuel (blend). However, not many deals with the actual in-flight and/or
rating conditions. Moore and co-workers [11] demonstrated reduction of soot emis-
sion from alternative fuels compared to conventional jet fuel at cruise condition,
Emissions from FEHA Fueled RQL Gas Turbine Combustor … 371

(a) PM mass density (b) PM number concentration

Fig. 11 Effect of pressure on PM mass density and particle number concentration emission
characteristics for Jet A1 and HEFA at constant temperature condition

(a) PM mass density (b) PM number concentration

Fig. 12 Effect of temperature on PM mass density and particle number concentration emission
characteristics for Jet A1 and HEFA at constant pressure condition

suggesting a reduction of climate effects (contrail formation). The general tendency

seen in the present study is in consistent with the reference. Anderson and co-workers
[12] demonstrated suppression of soot number and mass for various part load condi-
tions. They showed that the mass reduction remained unchanged with the change of
thermal load and that particle number density decreased with the increase of engine
power. This tendency is consistent with the findings reported in the present study.
Brem and co-workers [13] reported on their measurement of both particulate mass
and number and showed that effect of mass and number is comparable and that the
372 H. Fujiwara and K. Okai

improvement of emission characteristics by blending HEFA fuel is marginal when

investigated at the defined landing–takeoff (LTO) cycle condition. This is slightly
different from the present results, but it should be noted that the reference work is
HEFA/Jet A blend that had a relatively smaller fraction of HEFA content in the blend.
If the ratio of HEFA in the fuel blend increases, the discrepancy in tendency for mass
and number might become apparent. Liati and co-workers examined in-detail the
emission of soot under both idle and climb-out condition with due attention paid
to soot reactivity [14]. They suggested that at climb-out (high-load) condition, the
HEFA blend showed higher reactivity potential to bear higher health risk compared
to soot produced from Jet A1 fuel. Their study points out the importance of evaluation
for the HEFA fuel in a number of different projects.
A correlation for soot formation is given by Döpelhheuer [15]. From their equa-
tion, the soot density (in mg/cm3 ) is reported to be proportional to p1.35 . The data
from Fig. 11a is re-plotted in Fig. 13a with correlated values for pressure as given
in ref. [32]. The results showed a good general agreement. Also, previous report by
the authors [10] showed that at relatively low pressure and temperature condition
(350 kPa and 450 K), the variation of soot density (in mg/m3 ) with HEFA concen-
tration (%) in the mixture fuel showed quasi-linear rapid decay, although variation
of number concentration with fuel mixture ratio was linear, as seen in Fig. 13b.
Assuming this as a valid tendency at elevated pressures, the correlation above was
replotted and the results are shown in Fig. 14. The results obtained showed that, with
the fact that the mass density curve with respect to fuel mixture ratio was a rapid
decay, while the value of soot concentration kept to a low value with increase of
HEFA in the fuel mixture due to the rapidly decaying tendency.
The effect of fuel change from baseline jet fuel to HEFA is remarkable, especially
on soot formation and soot emission. The reduction of emission from HEFA to

(a) Experimental data and correlation (b) Tendency of change of PM mass

of PM mass concentration profile. concentration with fuel mixture ratio.

Fig. 13 Data for correlation of soot concentration in the fuel mixture under elevated pressure
condition
Emissions from FEHA Fueled RQL Gas Turbine Combustor … 373

Fig. 14 Correlation of soot

concentration in the fuel
mixture under elevated
pressure condition

baseline jet fuel is attributed largely to the reduced aromatic concentration in the
HEFA fuel. Other combustion properties tested included the emission of CO, NOx,
and THC. They all showed almost no effect from the change of fuel. In the following
sections, based on the emission data analyzed above, a simplified flight mission
analysis was conducted and is discussed below.

4 Simpliefied in-Flight Emission Analysis

The present experimental condition was determined based on an engine operation

condition for the conceptual engine configuration studied in “Research and Develop-
ment for Green Engine Technology.” The concept engine was defined under reference
JAXA’s concept technology reference aircraft TRA 2022A described by Nomura
[16]. In the following, investigation on emission profiles of a hypothetical flight
mission is presented.
The TRA2022A is one of the options defined in the study of conceptual design
study of 120-seat class future passenger aircraft, that is anticipated to enter service in
early 2020s. One of the goals of TRA2022s is to reduce fuel consumption by more
than 30% compared to baseline reference aircraft. Figure 15 shows a conceptual
drawing of the vehicle and its mission profile. Table 3 shows specifications of TRA
2022A. The description of mission flight profile in Nomura [16] provides in-detail
aerodynamic characteristics with the empirical data. In the present chapter, simplified
version of vehicle definition was used and the lift-to-drag coefficient relation was
depicted from an empirical expression of the aerodynamic characteristics given by
Sun and co-workers [17], see Fig. 16. The engine performance data is given as maps
of lapse rate and other expressions as seen in Fig. 17. Basic performance description
of the turbofan engine was determined from the previous work by Fukuyama and
co-workers [18, 19]. For the descent performance, description by Jenkinson and
co-workers [20] was referenced.
374 H. Fujiwara and K. Okai

(a) TRA 2022A three-view drawing (b) Mission profile of TRA 2022A

Fig. 15 Reference Aircraft model TRA-2022A [16]

Table 3 Main characteristics

Cruise speed M0.78
Range 360 nm
Number of seats 126
Main wing width 98.1 ft
Main wing area 1014.0 ft2
Main wings aspect ratio 9.5
Fuselage length 123.3 ft
Fuselage diameter 12.2 ft
Takeoff thrust 20,348 × 2 lbf
Maximum takeoff weight 127,980 lb

Fig. 16 Lift–Drag polar for flight mission analysis

Emissions from FEHA Fueled RQL Gas Turbine Combustor … 375

(a) Thrust lapse at take off

(b) Thrust lapse at max climb (c) Thrust lapse at max cruise

(d) Thrust lapse at descent (e) Fuel flow [lb/hr/FN*] at descent

Fig. 17 Information for engine characteristics calculation

376 H. Fujiwara and K. Okai

5 Results of in-Flight Emission Analysis

Based on the mission definition shown in Fig. 15, flight mission analysis was
conducted. The methods used were based on the method described by Fukuyama
and co-workers [18, 19] as well as in the other literature [20]. Figure 18 shows
calculated 2D-flight path. The results on variation in the corresponding fuel flow rate
and mission fuel burn are shown in Figs. 19 and 20, respectively. As seen previously,
emission data remained unchanged other than nvPM emissions. In the following, fuel
flow and emissions are representing either of the fuels (Jet A1, HEFA, and mixture
of the two). For soot (nvPM) emissions, differences between the fuels are examined.
By referencing to the flight profile given above, several emission profiles were
investigated by utilizing the experimental data shown in this chapter.
There are many empirical formulas proposed for describing EINOx, and the
method described by Chandrasekaran and Guha was utilized here [21].

Fig. 18 Flight path

Fig. 19 Fuel flow variation

Emissions from FEHA Fueled RQL Gas Turbine Combustor … 377

Fig. 20 History of fuel burn


Ttc
EINOx = 0.00036 ptc e 200

where ptc and T tc are combustor inlet pressure and temperature, respectively. The
expression was adjusted to the experimental data and correlated as shown in Fig. 21a.
The NOx emission history in relation to flight distance is presented in Fig. 21b. The
figure shows that the NOx outflow rate to be initially large, then decreases and then
kept to a relatively low value over the cruise phase. This suggests that emissions
under takeoff and climb phase conditions are quite large, and that the total amount of
emission should be relatively small but stand out in summation under cruise altitude
conditions. In each flight altitude and path, these data can be utilized to investigate
the effect of emission on the atmosphere and climate.

(a) NOx emission from Correlation (b) NO x emission rate over flight

Fig. 21 NOx emission consideration

378 H. Fujiwara and K. Okai

Similarly, soot-emission characteristics are described by utilizing the expression

provided in [22].
The expression for conventional jet fuel is written as:
 
− 20,000
csoot ∝ p3 e TFL

where p3 is combustor inlet pressure and T FL is defined as the flame temperature. The
results obtained from this correlation are shown in Fig. 22a. Using the expression,
the results on soot-emission profile are shown in Fig. 22b. The emission profile is
similar to the one for NOX , as seen from Fig. 21b, but one remarkable point is
that the ratio of the emission value of cruise phase to take off for the jet fuel case
(0% HEFA) is relatively large compared to the NOx case. Soot emission is now
recognized as increasingly important since it serves as kernel to generate contrails,
which is considered as a potential of source on global warming. Based on the previous
investigation as seen in Fig. 14, HEFA and mixture fuel cases are also plotted, see
Fig. 22b. Increase in HEFA fuel in the mixture resulted in decreased soot emission
over the flight path. Inclusion of relatively small fraction (20%) of HEFA decreased
large amount of soot emission due to the tendency seen in Fig. 14.
Finally, emission of water vapor and CO2 were investigated. The expression used
was as given in [23].
When fuels consumption is given as:

f = t · sfc · Th

where, t, sfc, and Th represent the time, specific fuel consumption, and thrust,
respectively. The emission of CO2 an H2 O can be written as:

(a) Soot emission from Correlation (b) Soot outflow rate over flight

Fig. 22 nvPM emission consideration

Emissions from FEHA Fueled RQL Gas Turbine Combustor … 379

(a) Emission rate over the flight profile. (b) Emission per height at climb phase.

Fig. 23 CO2 and H2 O emission consideration

E CO2 = E ICO2 · f = 3.155 · f

E H2 O = E IH2 O · f = 1.237 · f

Figure 23a shows the emission profiles of fuel, CO2 , and H2 O with respect to flight
altitude. As the expression implies, the emission of CO2 and H2 O are simply related to
a constant factor (i.e., fuel consumption). Therefore, the reduction of these emissions
is realized by decreasing the fuel consumption. However, more importantly, emission
amount at each altitude may affect differently to the atmosphere. Contrail’s formation
mechanism is complex and related to the amount of water (supercooled) vapor, soot
or aerosol kernels, and other factors. These emission profiles with respect to the
altitude would be important. Figure 23b shows the emission scale indicating amount
per height, which is emission outflow rate in kg/s divided by rate of climb (ROC).
This implies that in the middle of climb phase, one can have a relatively large amount
of emission.
To examine the effect of those emissions on climate change, many flight histories
should be referred and integrated over many years. But as a first attempt, these profiles
provide insights into the reason and effects of each emission with respect to a specified
flight mission. This chapter shows an example to interpret the experimental data to
understand the direct effect of emission characteristics to the atmosphere.

6 Conclusions

Combustion tests with jet fuel and HEFA up to 20 bar and 837 K were performed.
The results showed that mass and number density of PM emission was suppressed
with HEFA, while the emissions of NOx, CO, and THC with HEFA fuel were similar
380 H. Fujiwara and K. Okai

to those with Jet A1 fuel. At higher pressure and temperature, the suppression effect
of PM emission with HEFA fuel was apparent. However, the hypothetical diameter
determined from the PM mass and number density data suggested that the particle
diameter becomes closer between the Jet A1 and HEFA fuels. This study clearly
showed that pressure and temperature effect on the PM emission to be large and that
further investigations are necessary with actual conditions in order to closely evaluate
the soot-emission characteristics under flight conditions including the impact of
climate.
The present study demonstrated a way to utilize emission data at ground test
facility for the simulations on in-flight evaluation, thus to be used for the investi-
gation of direct effect on emissions to the atmosphere. The study clearly defined
a hypothetical turbofan engine and aircraft that flies. The corresponding combus-
tion and emission data from experiments with simulated combustor entry condition
(pressure and temperature) can easily provide prediction map data for the emis-
sions during flight. By calculating a mission profile, some emission characteristics
were obtained as histories of time and/or flight path (location). To further investi-
gate the effect of emissions on climate change, assumptions and information on Air
Traffic Management (ATM) should be incorporated. For those purpose, the present
table-style emission data would be beneficial.

Acknowledgements This work was in part supported by JSPS KAKENHI Grant Number
19H02341.
The authors thank Dr. Yoshitaka Fukuyama of JAXA for his guidance and comments on engine
systems analysis.
The authors also thank Professor Ashwani K. Gupta of the University of Maryland for his
comments and extensive revisions to the chapter.

References

1. National Academies of Sciences, Engineering, and Medicine (2016) Commercial aircraft

propulsion and energy systems research: reducing global carbon emissions. The National
Academies Press, Washington DC. 10.17226/23490
2. ASTM D7566-18 (2018) Standard specification for aviation turbine fuels containing synthe-
sized hydrocarbons, ASTM International, West Conshohocken, PA
3. Fukuyama Y, Fujiwara H, Okai K (2009) Alternative fuels and their impact on the turbofan
engine design and performance under the realistic flight conditions, ACGT paper No. 51, asian
joint congress on gas turbines 2009 (ACGT 2009), Tokyo, Japan (ACGT2009-051)
4. Okai K, Fujiwara H, Hongoh M, Shimodaira K (2012) Application of a Bio-fuel to a single
sector combustor for an experimental small aero-engine, asian joint congress on gas turbines
2012 (ACGT 2012), Shanghai, China, (ACGT 2012-2121)
5. Lefebvre AH, Ballal DR (2010) Gas Turbine Combustion: alternative fuels and emissions, 3rd
edn, CRC press, Taylor & Francis
6. Okai K, Mizuno T, Shinoda K, Fujii A, Kojima Y, Sakai K, Shibahara Y (2020) Development
and testing of integrated process of woody biomass gasification and fischer-tropsch synthesis
for bio-derived aviation fuel, AIAA 2020-3670, AIAA Propulsion and Energy Forum
Emissions from FEHA Fueled RQL Gas Turbine Combustor … 381

7. Abe Y, Toba M, Mochizuki T, Yoshimura Y, Oxidative degradation behaviour of fatty

acid methyl ester in fish oil biodiesel and improvement of oxidation stability by partial
hydrogeneration, J Japan Petroleum Inst 52(6):307–315 (in Japanese)
8. Makida M, Yamada H, Shimodaira K (2014) Detailed research on rich-lean type single sector
and full annular combustor for small aircraft engine. In: 29th Congress of the international
council of the aeronautical sciences (ICAS 2014), St. Petersburg, Russia, 2014 (ICAS2014-
0628)
9. Okai K, Fujiwara H, Makida M, Shimodaira K (2020) The effect of the fuel change from
petroleum kerosene to HEFA alternative jet fuel on the emission of an RQL gas turbine
combustor under high pressure and high temperature conditions, AIAA 2020-1220, AIAA
Sci-Tech 2020
10. Okai K, Fujiwara H, Makida M, Shimodaira K (2019) The effect of the fuel change from
petroleum kerosene to HEFA alternative jet fuel on the number of nvPM emission of an RQL
gas turbine combustor, AIAA 2019-1772, AIAA Sci-Tech 2019
11. Moore RH, Thornhill KL, Weinzierl B, Sauer D, D’Ascoli E, Kim J, Lichtenstern M, Scheibe
M, Beaton B, Beyersdorf AJ, Barrick J, Bulzan D, Corr CA, Crosbie E, Jurkat T, Martin R,
Riddick D, Shook M, Slover G, Voigt C, White R, Winsead E, Yasky R, Ziemba LD, Brown
A, Schlager H, Anderson BE (2017) Biofuel blending reduces particle emissions from aircraft
engines at cruise conditions, nature. https://doi.org/10.1038/nature21420
12. Anderson B (2020) The ACCESS-II science and implementation teams, alternative-fuel effects
on contrails and cruise emission (ACCESS-2) flight experiment.https://science.larc.nasa.gov/
large/data/ACCESS-2/presentations/4%20RichMoore_FuelEffectsOnBC.pdf. Accessed 20
Dec. 2020
13. Brem B, Durdina L, Elser M, Schoenenberger D, Setyan A, Wyss S, Munoz M, Schreiber D,
Liati A, Haag R, Rentsch D, Fischer A, Mohn J, Heeb NV (2018) Non-volatile particle matter
mass and number emissions of an aero gas turbine fuelled with alternative fuel blends. In: 22nd
ETH-Conference on combustion generated nanoparticles, June 18th 2018
14. Liati A, Schreiber D, Alpert PA, Liao Y, Brem BT, Corral Arroyo P, Hu J, Jonsdottir HR,
Ammann M, Dimopoulos P, Eggenschwiler PD (2019) Aircraft soot from conventional fuels
and biofuels during ground idle and climb-out conditions: Electron microscopy and X-ray
micro-spectroscopy. Environ Pollut 247:658–667
15. Döpelhheuer A (2001) Quantities, characteristics and reduction potential of aircraft engine
emissions. SAE 2001-01-0308
16. Nomura T (2013) Conceptual design of future passenger aircraft aimed at reducing fuel
consumption, JAXA-RR-13-007, Dec 2013
17. Sun J, Hoekstra JM, Elebroek J (2018) Aircraft drag polar estimation based on a stochastic
hierarchical model, presented n 8th SEAR innovation days, Dec 2018
18. Fukuyama Y, Yamane T (2018) Flight mission fuel burn evaluation for aero-engine related
technology variation, ACGT 2018-TS45. In: Proceedings of the asian congress on gas turbines,
Aug 2018
19. Fukuyama Y, Yamane T, Nishizawa T (2019) Overview of aero-engine, airframe and flight char-
acteristics influence on the flight mission energy consumptions, IGC-2019-234. In: Proceedings
of the international gas turbine congress 2019 Tokyo, Nov 2019
20. Jenkinson LR, Simpkin P, Rhodes D (1999) Cibil Jet aircraft design. AIAA Education Series,
American Institute of Aeronautics and Astronautics, Inc.
21. Vera-Morales M, Hall CA (2009) Modelling performance and emissions from aircraft for the
aviation integrated modelling project. In: AIAA 2009-1263, 47th AIAA aerospace sciences
meeting and exhibit, Jan 2009
22. Chandrasekaran N, Guha A (2012) Study of prediction methods for NOx emission from
turbofan engine. J Propul Power 28(1)
23. Kugele A (2005) Aircraft particulate matter emission estimation through all phases of flight.
EEC/SEE/2005/0014, EUROCONTROL Experimental Centre
24. Srichar B, Ng HK, Chen NY (2012) Integration of linear dynamic emission and climate models
with air traffic simulations. In: AIAA 2012-4756, AIAA guidance, navigation and control
conference, Aug 2012
Performance Evaluation of Upward
Swirl Combustor with Reverse Fuel
Injector and Hydrogen Blending

Parag Rajpara, Rupesh Shah, and Jyotirmay Banerjee

Abstract Modern gas turbine combustors are required to provide high combustion
efficiency, stability over wide range of operating conditions and fuel flexibility while
meeting strict low emission norms, with other requirements such as low heat loads
on combustor liner wall. One of the promising ways to meet such requirements is
the provision of swirler. In conventional Can combustor, the swirler is positioned at
upstream end of the primary region. This arrangement is termed as backward swirl
arrangement. When swirler is placed between primary and secondary region, it is
termed as upward swirl arrangement. Influence of key parameters on the perfor-
mance of a Can-type upward swirl gas turbine combustor is reported in this chapter.
First, a relative assessment of thermal and emission characteristics of conventional
Can-type combustor and upward swirl Can-type combustor is presented. Test results
of exit temperature profile, liner wall temperature and emission characteristics at
exit of the combustor are compared for both the combustors. Results of large eddy
simulation (LES) are compared to identify and understand the flow and combus-
tion regimes using methane as a fuel. Temperature distribution obtained in conven-
tional combustor indicates that backward swirl flow arrangement provided for flame
stability and air–fuel mixing has drawback that it produces localized hot spots of
high temperature gradients in the primary zone resulting in the formation of thermal
NOx emissions. In addition, the thermal pattern created due to swirling of combus-
tion products exerts high heat loads on the liner wall of combustor. When air is
allowed to enter in the primary zone through upward swirl arrangement, it helps to
confine the flame near the combustor axis. This peculiar flow regime set the liner
wall at a low temperature compared to conventional combustion chamber. Compar-
ison of emission characteristics demonstrates that NOx emissions are significantly
low in upward swirl combustor compared to conventional combustor. However, low
NOx levels are achieved by incurring unacceptable penalties in combustion chamber
performance in terms of low combustion efficiency and high CO emission levels.

P. Rajpara
Department of Mechanical Engineering, Marwadi University, Rajkot, Gujarat, India
R. Shah (B) · J. Banerjee
Department of Mechanical Engineering, Sardar Vallabhbhai National Institute of Technology,
Surat, Gujarat, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 383
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_17
384 P. Rajpara et al.

To address the difficulties of upward swirl combustor, a modification to the fuel

injection strategy is proposed. A new concept of reverse fuel injection is introduced.
The existing upward swirl combustor has conical-shaped injector where fuel jets are
arranged on the periphery of 90° cone. In reverse fuel injection strategy, the fuel
exits the fuelling device in reverse axial direction towards the wall of hemispherical
dome. The dome wall decelerates the fuel jet first, and then fuel flows in all direc-
tions. The injector length is chosen as variable for parametric investigation. Research
indicates that the burned gas recirculation and highly turbulent shear flow caused by
reverse fuel flow pattern improves fuel–air mixing and produces larger energy density
during combustion. This improves the combustion efficiency and reduces CO emis-
sion levels remarkably. The length of injector has significant effect on increase in
NOx levels and wall temperature. The reverse fuel injector with 5 mm injector length
is identified as optimum configuration for achieving noticeable reduction in CO
emission without any adverse effect on NOx levels and wall temperature. Chem-
istry between fuel and air is further augmented by blending methane with hydrogen.
Results of experiments conducted for constant energy input condition are presented
and discussed. For the same energy input, hydrogen addition to methane increases
the mole fractions of free radicals H, O and OH, which promotes chemical reactions
and produces intensive combustion. From flame visualization, it can be said that
flames are relatively broader and shorter at higher hydrogen concentrations. Shorter
flame obtained at higher hydrogen percentage indicates more rapid combustion and
improved flame stability. Addition of hydrogen up to 5% in fuel blend contributes to
additional reduction in CO emissions with marginal increase in NOx emissions.

Keywords Upward swirl combustor · Reverse fuel injector · Hydrogen blending ·

CO and NOx emissions · Combustion efficiency · Flame visualization

1 Introduction

Combustion chambers are required to burn fuel completely, create little pressure
drop, produce an acceptable temperature profile, maintain stable operation over a
wide range of conditions and satisfy stringent emission norms. Another important
requirement is to maintain low heat loads on liner wall of gas turbine combustor even
at higher engine pressure ratio. All these requirements are difficult to be satisfied
simultaneously. Enrichment of mixing between fuel and air by means of swirl is one
of the promising way to meet such performance requirements. Swirling of inlet air in
combustion enhances the rate of mixing by creating strong turbulence at the mixing
layers. This helps in achieving higher fuel burning rate, flame stability and controlled
emission. Several researchers have studied the relation between swirl flow and NOx
emissions in conventional gas turbine combustors [1–8]. In general, conventional
combustors have swirler positioned at the upstream end of the combustor dome
around the fuel injector (as shown in Fig. 1). This arrangement is termed as backward
swirl arrangement. Studies have shown that backward swirl arrangement provides
Performance Evaluation of Upward Swirl Combustor … 385

Primary jets Dilution jets

Wall cooling
Air Air

Air Air cts

odu Exhaus
t pr t
Fuel Ho

Primary Secondary Dilution

Swirler zone zone zone

Fig. 1 Conventional gas turbine combustor features

flame stability and air–fuel mixing in conventional combustor. This arrangement has
drawback of localized hot spots of high temperature gradients in the primary zone.
This facilitates formation of thermal NOx emissions. In addition, the thermal pattern
created due to swirling of combustion products exerts high heat loads on the liner
wall of combustor. There are many practical methods and approaches available to
alleviate liner wall temperature and NOx levels in gas turbine combustion chambers.
Traditional mechanisms available for wall-cooling and emission control in the field
of gas turbine combustion are discussed first followed by philosophy of upward swirl.

2 Wall-Cooling Techniques

Extensive efforts have been put by the researchers towards reducing wall-cooling
requirements from conventional gas turbine combustors. Wassell and Bhangu [9]
suggested that the composite liner wall structure made out of Transply1 is capable
of reducing cooling air requirement significantly as compared to the traditional film
cooling in combustor. This large saving in cooling air benefits in terms of better
improvement in pattern factor and control of emissions. Wadia [10] recommended the
use of composite matrix liner, as it demands least amount of cooling air if cost is not an
issue. Bailey et al. [11] studied the cooling characteristics of combustor liner using jet
impingement method along with rib turbulators2 for heat transfer augmentation. They
reported that rib turbulators enhance the heat transfer and keep the liner relatively
cool. However ribbed structure results in increase of overall engine weight. Various

1 Transply is one kind of composite liner structure developed by Rolls-Royce Company. It is

produced by brazing of two or more metal alloys having high temperature strength.
2 Turbulators are designed to develop turbulent flow and help to augment heat transfer on tube side

of fire tube boilers, heat exchangers and other heat transfer devices.
386 P. Rajpara et al.

other methods such as film cooling, jet impingement cooling, effusion cooling3 ,
transpiration cooling4 , combined slot injection and effusion cooling and many other
are also preferred by researchers [12–18] to keep low liner wall temperature in gas
turbine combustor. However, at higher engine pressure ratio, the cooling effectiveness
of air used for liner cooling reduces due to higher inlet air temperature. Increase in the
pressure ratio also leads to an increase of turbine inlet temperature. This is beneficial
in improving the pattern factor in order to maintain uniform temperature distribution
at the exit of combustor. However, higher turbine inlet temperature has adverse
effect on liner wall temperature, specifically at the tail end of combustor. Supply
of excess air for combustor liner cooling means lesser air for combustion. This
in turn results into incomplete combustion and formation of carbon monoxide and
unburned hydrocarbons. Therefore, the only option left with combustor engineers and
researchers is to use the available cooling air most efficiently to minimize the wall-
cooling requirements in gas turbine combustor with optimum emission of pollutants.

2.1 Emission Control Mechanisms

The principal pollutants emitted by gas turbines are carbon dioxide (CO2 ), carbon
monoxide (CO), unburned hydrocarbons (UHC), water vapour (H2 O), NOx , SOx , O2 ,
N2 and particulates. When emission control methods were originally introduced, NOx
was the primary regulated gas turbine emission. Various methods such as water/steam
injection [19], lean premixed prevaporized combustion [20], catalytic combustion
[21], rich-burn/quick-mix/lean-burn combustor [22], staged combustion [23], selec-
tive catalytic reduction [24] and exhaust gas recirculation [25] worked well in alle-
viating NOx emissions from gas turbine combustor. However, the fulfilment of these
techniques depends on reducing the temperature of flame, which tends to increase
the formation of carbon monoxide (CO) and unburned hydrocarbons (UHC). Other
concerns with these techniques are those of durability, maintainability and safety.

3 Effusion cooling is the simplest approach of wall cooling defined by large number of holes
drilled on liner wall normal to the wall. It can be applied to specific portion or entire liner wall.
The advanced effusion cooling technique is “Angled Effusion Cooling (AEC)” in which the cooling
holes are drilled at an angle to the liner wall. AEC improves cooling effectiveness of the cool air
film but increases the liner weight.
4 Transpiration cooling technique is one in which the cooling air passes through pores uniformly

dispersed over the liner surface. The cooling air jets emerging from each pore quickly settle to form
a protective layer of cool air over the entire surface of liner wall. This technique is most efficient
one, but its practical application is hampered due to limitation of available porous material.
Performance Evaluation of Upward Swirl Combustor … 387

2.2 Philosophy of Upward Swirl

Philosophy of upward swirl is new to the field of gas turbine combustion technology.
It has potential to meet conventional combustor requirements. However, this tech-
nology has not received much attention. For the very first time, Arai et al. [26] and
Furuhata et al. [27] have coined the concept of upward swirl to achieve low NOx
emissions in micro gas turbine combustors. The concept is presented in Fig. 2.
The combustor consists of primary zone and the secondary zone, and both are
connected by the throat zone. In order to enhance burned gas recirculation in the
primary zone, swirler was set at the end of the primary zone. The combustion air
was introduced through swirler and forced to flow upward towards the centrally
positioned fuel injector. It was reported that burned gas recirculation and highly
turbulent shear flow caused by upward swirl promote mixing and provide a premixed
like combustion resulting in low NOx emissions from combustor. Combustion tests
showed that emissions of CO and NOx were very low in the wide range of flow
conditions.
In another study, Shah and Banerjee [6] made the prototype of Can-type upward
swirl combustor. Combustion experiments were carried out using conventional Can-
type combustor, backward swirl combustor and upward swirl Can-type combustor.
Obtained results showed that the annular entry of primary air in upward swirl configu-
ration helps to confine the flame near the combustor axis. Thus, the liner wall remains
at a very low temperature in upward swirl combustor compared to backward swirl
Can combustor. Experimental results demonstrated that upward swirl combustor has
substantially low NOx emissions compared to backward swirl combustor. Unfortu-
nately, the poor combustion efficiency and increase in formation of CO emissions are

Swirler
Recirculation region Air
Fuel injector

Flame

Upward swirl air flow Air

Primary zone Throat Secondary zone

Fig. 2 Schematic diagram of upward swirl combustor [27]

388 P. Rajpara et al.

formidable challenges that must be overcome to secure implementation of upward

swirl combustion in operational gas turbine.
Motivated by the desire to improve the performance of a Can-type upward swirl
combustor and to address above-mentioned challenges, parametric investigations
are performed. First, the performance of backward swirl Can-type combustor and
upward swirl Can-type combustor is assessed. Thermal and emission characteristics
are compared. Strategies adopted to improve the performance of upward swirl Can
combustor are tested, and results are discussed.

3 Backward Swirl and Upward Swirl Can-Type Combustor

The conventional combustor under consideration is shown in Fig. 3. It is a roughly

three/fourth scale model of a Can-type gas turbine combustor. It is representative
of the Rolls-Royce Tay gas turbine combustor [6]. The conventional combustor
presented in Fig. 3 comprises of fuel injector, swirler followed by hemispherical
head (dome), one row of primary holes and one row of secondary holes on cylin-
drical barrel and circular to rectangular nozzle. Fuelling device is a conical-shaped
fuel injector where fuel jets are arranged on 90° cone. It injects gaseous fuel through
ten 1.7 mm diameter holes circumferentially placed on pitch circle diameter of a 90°
cone located centrally at the upstream end of hemispherical head. A swirler with 10
flat vanes at 45° angle to the flow is placed circumferentially around the fuel injector
on the head of the combustor. In conventional combustor, one row of primary holes
and one row of secondary holes are located in the intermediate barrel section. There
are 6 primary holes and 12 dilution holes having diameter of 10 mm pierced through
the liner surface in conventional combustor.

Primary Secondary Dilution

zone zone zone

5
25
R37
Ø40

Ø15

Swirler
Fuel injector Primary hole axis Dilution hole axis
6 holes, Ø=10mm 12 holes, Ø=10mm
46
126
206
All dimensions are in mm

Fig. 3 Schematic diagram of backward swirl Can combustor

Performance Evaluation of Upward Swirl Combustor … 389

Swirler
8 vanes, 45° angle
Dilution holes axis
Fuel injector 12 holes, Ø=10mm

25
R37
R39

R49
Ø40

Ø15

65
175
240

All dimensions are in mm

Fig. 4 Schematic diagram of upward swirl Can combustor

The Can combustor geometry developed with upward swirl arrangement is shown
in Fig. 4. In upward swirl configuration, swirler is set between primary and secondary
zones. The primary air required for combustion enters into the combustion chamber
through swirler and flows in upstream direction towards the centrally located fuel
injector. Fuel is injected into combustion chamber through a conical-shaped fuel
injector having 10 number of fuel holes of diameter 1.7 mm arranged on 90° cone
located at the centre and at the upstream end of the dome region. One row of dilution
air holes is provided on the liner wall of the combustor barrel in the dilution zone.

4 Experimental Procedure and Test Conditions

Figure 5 represents the schematic diagram of experimental test setup developed

for reactive experimentation on Can-type gas turbine combustor. The test setup is
equipped with fuel line, airlines and the measurement system. Measurements of
temperature and emission at the exit of combustor are carried out. The temperature
distribution at the exit of combustor is measured by L-shaped K-type thermocouples.
Emissions are measured at the exit of combustor using “Testo” made flue gas analyser
(Testo 340). Thermal imaging of outer wall of combustor is performed using “Testo”
made thermal imaging camera (Testo 875-1i). Details of various equipments and
measurement accuracy have already been explained elsewhere [28].
All experiments are conducted at atmospheric conditions. In present study, the
total air flow rate to the combustor is kept constant at 0.07 kg/s. Normally, the gas
turbine engines operate at an air to fuel ratio (AFR) in the range of 40–70. Experiments
are performed for different values of air–fuel ratio (AFR) ranging between 50 and
70.
390 P. Rajpara et al.

Fig. 5 Schematic diagram of upward swirl combustor test rig

5 Numerical Methodology

Computations are performed to understand and analyse flow and combustion physics
inside the combustor. Computational fluid dynamics (CFD) tool Ansys Fluent is
used to model flow, turbulence, heat transfer and reactions for the combustor. Large
eddy simulation (LES) is used as the main approach to describe the turbulent flow.
The effect of bulk radiation is included using discrete ordinates (DO) model. The
interaction between turbulence and chemistry in a non-premixed flow regime is
discussed by means of the well-known PDF model. The detailed description about
governing equations and physical models may be found elsewhere [28].
In present study, computations are carried out with initial air and fuel temperature
as 300 K. Combustor inlet (air and fuel) surfaces are defined as mass flow inlet.
Pressure boundary condition is assigned at combustor outlet. Total mass flow rate of
air at combustor inlet is kept constant at 0.07 kg/s. The fuel mass flow rate is varied
to achieve desired value of air-fuel ratio. In present work, computational results
obtained for AFR equal to 50 are compared with the test results using methane as
fuel. Further details about solution methods and convergence criteria for different
governing equations are available in [28].
Performance Evaluation of Upward Swirl Combustor … 391

6 Conventional Can Combustor Versus Upward Swirl Can

Combustor

Assessment of thermal and emission characteristics of conventional and upward swirl

Can combustor is carried out through evaluation of exit temperature variation, flame
characteristics, emission level and thermal imaging of combustor outer wall.

6.1 Exit Temperature Variation

To measure temperature variation at the combustor exit, thermocouples are placed

at the mid of the rectangular exit section as shown in Fig. 6. Test results obtained at
AFR equal to 50 for the conventional and the upward swirl combustor are compared
as illustrated in Fig. 7.

Fig. 6 Sketch diagram of temperature measurement arrangement at combustor exit

Fig. 7 Exit temperature -0.05 -0.025 0 0.025 0.05

variation for conventional ( 1 1
) and upward swirl ( 0.9 0.9
)
0.8 0.8
0.7 0.7
0.6 0.6
T/Tad

0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
-0.05 -0.025 0 0.025 0.05
Position along mid horizontal line (m)
392 P. Rajpara et al.

In Fig. 7, the ordinate represents the position along the mid-horizontal line while
the abscissa represents the temperature value normalized by the adiabatic flame
temperature value of methane with air as oxidizer. In Fig. 7, zero position represents
the face centre location of the exit section and ±0.05 represents the sidewalls of
combustor on mid horizontal line at combustor exit. The comparison reveals that
temperature variation at the exit is opposite in both combustors. In conventional
combustor, the lowest temperature is in the vicinity of face centre of exit plane while
the highest temperature is observed at the sidewalls of exit section. On the other
hand, the sidewalls of upward swirl combustor are at relatively low temperature
compared to core region of the exit section. This depicts that alteration in swirl
flow arrangement significantly influences the thermal flow patterns inside the Can
combustor. In the following section, the flame characteristics of two combustors are
studied and discussed.

6.2 Flame Characteristics

For AFR 50, simulation results for both the swirl flow configurations are compared
in terms of temperature contours on vertical plane of combustor and is shown in
Fig. 8. It implies from Fig. 8 that the swirling flow and primary jet impingement
utilized for flame stability and air–fuel mixing in conventional combustor forces the
combustion products to flow near the combustor liner wall. This sets the combustor
liner wall to a very high temperature. When air is allowed to enter in the primary
zone through upward swirl arrangement, it helps flame to confine near the combustor
axis. This peculiar flow regime set the liner wall at a low temperature compared to
the conventional Can combustor. Same can be observed from temperature contours
shown in Fig. 8.

Fig. 8 Temperature contours on vertical plane of backward swirl combustor (left) and upward swirl
combustor (right) (Unit: K)
Performance Evaluation of Upward Swirl Combustor … 393

Fig. 9 Thermal images of outer wall surface for conventional combustor (left) and upward swirl
combustor (right) (Unit: K)

6.3 Outer Wall Temperature

The change in flue gas flow regime within the combustion chamber caused by
different swirl flow arrangements result into significant change in the wall tempera-
ture distribution. To estimate the effect of swirl flow arrangement on wall temperature,
the combustor outer wall temperature is measured with the help of infrared ther-
mography. Figure 9 compares the thermal images of outer cylindrical wall surface
for conventional and upward swirl combustors. Test results show that outer wall
surface is relatively cool in case of upward swirl combustor compared to conven-
tional combustor. This is mainly attributed to different thermal flow patterns within
the combustion chamber of two configurations. Since high temperature gases domi-
nate the liner wall region of backward swirl combustor, the radiative intensity in this
region is maximum. Therefore, the radiative heat transfer from liner to outer wall is
larger resulting in higher temperature at the outer wall surface. On the other hand, in
upward swirl combustor, the main flow resulting from burning of fuel–air mixture
flows in the vicinity of combustor axis. Therefore, the liner wall region remains at
relatively low temperature. Since the relative distance between the liner wall and the
outer wall and the nature of wall surfaces are same in both combustors, the radiative
heat transfer from liner to outer wall is relatively small in upward swirl combustor
compared to backward swirl.

6.4 Emission Characteristics

The emission performance of two combustors is evaluated by carrying out experi-

ments to measure pollutant species such as NOx and CO at different values of AFR
394 P. Rajpara et al.

Conventional combustor 7000

25
Upward swirl combustor
6000
20

CO emission (ppm)
NOx emission (ppm)

5000
15
4000
10
3000

5 2000

0 1000
50 55 60 65 70 50 55 60 65 70
Air-Fuel Ratio (AFR) Air-Fuel Ratio (AFR)

Fig. 10 Measured emission values (at 15% O2 )

as illustrated in Fig. 10. Test results show that NOx levels are very low in upward
swirl compared to conventional backward swirl at all AFRs. The mechanism of NOx
formation in combustor is correlated to its flame characteristics. In backward swirl
combustor, the fuel injector and the dome wall are exposed to high temperature gases
in the primary region (as seen in Fig. 8). The high temperature wall and fuel injector
act as high-energy inert reactants, which accelerates the formation of two active
molecules of atomic oxygen (O). This situation favours the generation of nitrogen
oxides (NOx ) in conventional combustor through reactions O + N2 → NO + N and
N + O2 → NO + O mainly governed by Zeldovich mechanism [28]. On the other
hand, in upward swirl configuration, fuel burns in the vicinity of combustor centreline
and the combustor core remains at higher temperature compared to combustor wall.
Hot spots of high temperature gradients observed in the primary zone of conventional
combustor are eliminated in upward swirl combustor. This result into significant
reduction in NOx levels from upward swirl combustor compared to backward swirl
combustor.
Although the upward swirl method is very effective in reducing wall temperature
and NOx levels, but CO levels increase significantly with AFR as can be seen from
Fig. 10. The increase in CO emission is an indication of poor fuel–air mixing and
slow fuel burning rate. Inefficient fuel burning in upward swirl is reflected as lower
exit gas temperature and poor combustion efficiency as shown in Table 1. To meet
the goals of low CO emission and high combustion efficiency while maintaining

Table 1 Performance
Parameters Conventional Upward swirl
parameters at the exit (AFR =
combustor combustor
50)
Average 1183 991
temperature (K)
Combustion 92.45 72.3
efficiency (%)
Performance Evaluation of Upward Swirl Combustor … 395

NOx levels and wall temperature within an acceptable range, the modification in
fuel injection strategy of upward swirl is required. The performance of upward swirl
combustor with modified fuel injection system is evaluated.

7 Assessment of Upward Swirl Combustor with Reverse

Fuel Injector

A new concept of reverse fuel injection is introduced to improve combustion effi-

ciency and reduce CO emission level for upward swirl combustor. The concept is
shown in Fig. 11. In reverse fuel injection, fuel is supplied into combustor through
injection holes in reverse axial direction towards the wall of hemispherical dome in
primary zone. The wall of hemispherical dome decelerates the fuel jet first, and then,
the fuel spreads in all directions in the primary zone.
To arrive at optimum dimensions of the reverse fuel injector geometry, the injector
length (distance between injection holes and the dome wall) is considered as a vari-
able. Parametric study is carried out with three variants of reverse fuel injector having
injector length equal to 5 mm, 10 mm and 15 mm. Figure 12 presents schematic
diagram of three variants of reverse fuel injector geometry. Results obtained for
methane fuel are presented here, which shows the influence of fuel injection strategy
on flow field, combustion and emission characteristics of upward swirl combustor.

Fig. 11 Schematic flow diagram of upward swirl combustor with reverse fuel injector
396 P. Rajpara et al.

Fig. 12 Schematic diagram of different reverse fuel injector variants

7.1 Influence of Reverse Fuel Injection on Flow Structure

Effect of fuel injection strategy on flow structure is carried out by means of plotting
streamlines along the longitudinal plane of combustor. Figure 13a–d demonstrates
streamline plots on longitudinal plane of combustor for existing conical fuel injector
and three variants of reverse fuel injector. For upward swirl combustor with conical
fuel injector, streamline pattern depicts that fuel jets are not able to penetrate deep
into swirling primary air and mixing take place close to the injector in the vicinity of
combustor axis (refer Fig. 13a). Streamlines show that fluid expands and accelerates
in the secondary zone. Streamlines are straight and parallel to the combustor axis
in the secondary zone. The nature of streamlines in dilution zone depicts that flow
converges as it passes through a circular to rectangular nozzle section.
Flow structure of upward swirl combustor appears to be significantly altered in
primary and secondary regions when reverse fuel injection strategy is used. For the

Fig. 13 Streamline on vertical plane for a conical fuel injector, b 5 mm reverse fuel injector,
c 10 mm reverse fuel injector, d 15 mm reverse fuel injector
Performance Evaluation of Upward Swirl Combustor … 397

first model of reverse fuel injector with 5 mm injector length, streamline pattern show
that fuel jet strikes at the dome wall and then spreads in all directions in the primary
zone (refer Fig. 13b). The primary air from swirler flows towards the fuel injector
and interacts with the fuel jet near the injector wall. The air–fuel mixture prepared
in this region propagates in the downstream direction. The downstream propagation
of gaseous mixture is retarded by the flow of swirling primary air as can be seen
from a bunch of air flow streamlines near the injector wall in Fig. 13b. This reduces
the fluid velocity in the centre of the flow resulting in the formation of wake in the
downstream of the primary zone as demonstrated in Fig. 13b. The formation of wake
is beneficial in terms of achieving better mixing.
With an increase in injector length from 5 mm to 10 mm, the fuel jet strikes at the
dome wall from a longer distance. The increased size of horseshoe vortex formed
due to fuel–air interaction near the injector wall reveals that fuel jet spreads more in
radial direction with higher strength (as seen from Fig. 13c). The strength of central
recirculating region decreases, which reveals that combustion gases produced from
fuel–air burning moves faster.
With further increase in injector length from 10 mm to 15 mm, the size of horse-
shoe vortex formed due to fuel–air interaction increases revealing increased fuel jet
penetration into the primary air (refer Fig. 13d). However, the size and strength of
the central recirculating region decreases, which indicates that overall residence time
of combustion gases reduces in combustor due to faster mixing between fuel and air
in primary region.

7.2 Influence of Fuel Injection on Combustion

Influence of fuel injection on combustion is determined in the form of energy density

and fuel mass fraction variation for different fuel injector geometries. The energy
density quantifies the amount of heat released during combustion. The energy density
of flow is defined as

E = ρ × Cp × T (1)

where ρ is density, C p is specific heat and T is temperature.

Figure 14a–d illustrates the LES results of energy density contours for different
types of fuel injection system at AFR 50. The energy density is higher in regions
where fuel–air mixture burns. In upward swirl combustor, fuel–air mixture burns in
the vicinity of combustor centreline, and as a consequence, the energy density is seen
higher in this region.
With reverse fuel injector of 5 mm injector length, the rapid near-field mixing
and formation of central recirculating region causes better mixing of combustion
products, which produces more amount of energy density in combustor. This is
revealed from Fig. 14b where the region of higher energy density is wider in 5 mm
reverse fuel injector combustor. As a result of intensified reactions, fuel burning rate
398 P. Rajpara et al.

Fig. 14 Energy density contours for a conical fuel injector, b 5 mm reverse fuel injector, c 10 mm
reverse fuel injector, d 15 mm reverse fuel injector (unit: kJ/m3 )

is enhanced, which in turn decreases the amount of unburned fuel in combustor as

can be seen from Fig. 15a.
Injection of fuel from 10 mm distance from the dome wall produces wider reaction
zone in the primary region. This can be observed from increased volume of high-
energy density region in the primary zone as shown in Fig. 14c. The high-energy

Fig. 15 Fuel mass fraction contours for a conical fuel injector, b 5 mm reverse fuel injector, c 10 mm
reverse fuel injector, d 15 mm reverse fuel injector
Performance Evaluation of Upward Swirl Combustor … 399

density region expands more when fuel is injected from 15 mm distance from the
dome wall. Wider reaction zone produced at higher injector length is associated
with the rapid fuel–air mixing caused by increased fuel jet penetration. However,
an increase of reverse fuel injector length produces little effect on augmentation in
energy density and reduction in unburned fuel in combustor.

7.3 Influence of Fuel Injection on Wall Temperature

Figure 16a–d illustrates thermal images of combustor outer wall for all type of fuel
injector configurations considered here. It is found that increasing the reverse fuel
injector length (distance between fuel injection holes and the dome wall) elevated the
combustor outer wall temperature. As discusses earlier, the combustion gases expand
radially towards the liner wall by increasing the length of the injector, resulting in an
increase in the rate of heat transfer from the inner to the outer wall, thereby increasing
the surface temperature of the outer wall. For 5 mm injector length, the elevation in
wall temperature is negligible, whereas the maximum wall temperature is obtained
when injector length is increased to 15 mm.

Fig. 16 Thermal images of outer wall for a conical fuel injector, b 5 mm reverse fuel injector,
c 10 mm reverse fuel injector, d 15 mm reverse fuel injector (unit: K)
400 P. Rajpara et al.

7.4 Influence of Fuel Injection on Emission Characteristics

Emission measurement in terms of NOx and CO is obtained for different fuel injector
geometry. Measured values of NOx emission for all cases at different AFR values
are illustrated in Fig. 17. The upward swirl combustor with conical fuel injector
emits very low amount of NOx emission at all AFRs. NOx emission increases when
reverse fuel distribution method is used. The mechanism of NOx emission formation
is correlated with the flame characteristics, which is presented in Fig. 18a–d.

15 7000

12 6000
NOx emission (ppm)

CO emission (ppm)
5000
9
4000
6
3000
3
2000

0 1000
50 55 60 65 70 50 55 60 65 70
Air-Fuel Ratio (AFR) Air-Fuel Ratio (AFR)

Fig. 17 Emission characteristics for different fuel injection systems conical fuel injector (
), 5 mm ( ), 10 mm ( ) and 15 mm reverse fuel injector ( )

Fig. 18 Temperature on vertical plane for a conical fuel injector, b 5 mm reverse fuel injector,
c 10 mm reverse fuel injector, d 15 mm reverse fuel injector (unit: K)
Performance Evaluation of Upward Swirl Combustor … 401

For upward swirl combustor with conical fuel injector, slower burning rate of fuel
results into less amount of energy produced during combustion, which in turn result
into lower primary zone temperature and corresponding value of NOx emission is
low. Fuel delivery through reverse fuel injector improves fuel–air mixing, which
leads to better combustion and larger energy release. This leads to an increase in
gas temperature, which accelerates the formation of thermal NOx according to the
Zeldovich mechanism [28]. It is found that at low AFR values the concentration of
NOx emission remains same for conical fuel injector and the first model of 5 mm
reverse fuel injector. Reverse fuel injector with 5 mm length produces little effect on
NOx emission elevation at higher AFR value. Increasing the distance of fuel injection
holes from the dome wall (injector length) promotes the formation of NOx emission
from combustor. This is because as the length of injector increases, reaction zone
gets wider in the primary zone, which in turn result into higher flame temperature
in primary zone and thereby produces large amount of NOx emission. Among three
variants of reverse fuel injector, lowest NOx emission is achieved in 5 mm injector,
whereas highest value of NOx emission is produced in 15 mm injector.
The effect of fuel injection on CO levels is illustrated in Fig. 17. Using reverse fuel
injection strategy, efficient mixing is achieved between fuel and air. This results in
enhanced rate of fuel burning and better combustion of CO into CO2 . When compared
for the three variants of reverse fuel injector, it is found that increasing the injector
length produces little effect on CO reduction and performance improvement. This is
because as the injector length increases, the depth of fuel jet penetration in primary
zone increases while the size and strength of central recirculation zone in combustor
decreases (refer Fig. 13b–d). Larger fuel jet penetration in radial direction helps in
achieving better fuel–air mixing and intensive combustion in primary zone, while
reduced size and strength of central recirculation zone decreases the residence time
of combustion products. This slows down the rate of combustion of CO into CO2 .
The performance parameters such as average exit gas temperature and combustion
efficiency are improved due to better combustion of fuel in reverse fuel injection
system as shown in Table 2.
From the measured results, it is observed that reverse fuel injection system works
well in terms of achieving reduction in CO and maintaining low NOx and low wall
temperature when injector length is kept at 5 mm. Increasing the length of fuel

Table 2 Performance of
Fuel injector models Average exit Combustion
combustor for various fuel
temperature (K) efficiency (%)
injection systems (AFR = 50)
Conical fuel injector 991 72.35
5 mm reverse fuel 1099 83.65
injector
10 mm reverse fuel 1102 83.97
injector
15 mm reverse fuel 1112 85.02
injector
402 P. Rajpara et al.

injector produces little effect on CO reduction but has remarkable effect on increase
of NOx emission and wall temperature. There is scope for further reduction in CO
emission level using 5 mm reverse fuel injector geometry. The CO emission can
further be alleviated by blending hydrogen with methane, which augments combus-
tion chemistry between fuel and air. Analysis of chemical reaction, flame structure
and emission characteristic is reported for different methane–hydrogen fuels used in
upward swirl combustor for the optimum geometry of 5 mm reverse fuel injector.
Results are presented and discussed in the following sections.

8 Evaluation of Upward Swirl Combustor with Reverse

Fuel Injector and Hydrogen Blending

Hydrogen is a carbon-free, high-energy-density substance that has been widely used

as a supplement with methane/natural gas in gas turbines. The strategy of adding
hydrogen fraction to methane has advantage that it makes the system partially carbon-
free, preserves a clean environment, extends flammability limits and reduces the total
cost of fuel mixture [29–34].
The upward swirl combustor with reverse fuel injector is evaluated for hydrogen
blended methane fuel under constant energy input condition. The methane–air
combustion case (0% H2 ) is considered as a reference case. A variable fraction
of hydrogen, between 0 and 10% in mass, is added at constant energy input condi-
tion. Various parameters such as methane and hydrogen flow rates, total fuel mixture
flow rate, total airflow rate and equivalence ratio for same energy input condition are
listed in Table 3.

Table 3 Parameters for constant energy input condition

Cases H2 % (by Energy Fuel mixture CH4 H2 Total air Equivalence
mass) input (kW) flow rate Volume flow flow rate ratio (Ø)
(kg/s) rate (SLPM) (kg/s)

1 0 70 0.0014 129 0 0.07 0.345

2 1 70 0.00138 126 10 0.07 0.342
3 2 70 0.00136 123 20 0.07 0.340
4 3 70 0.00134 120 31 0.07 0.338
5 4 70 0.00132 117 41 0.07 0.337
6 5 70 0.001301 114 51 0.07 0.336
7 7 70 0.001261 107 72 0.07 0.333
8 10 70 0.001201 98 102 0.07 0.328
Performance Evaluation of Upward Swirl Combustor … 403

8.1 Influence of Hydrogen Addition on Chemical Kinetics

The chemical kinetic effect of hydrogen addition is analysed by means of mole

fraction profiles of various flame species derived from LES results. Some chemical
species including major species CH4 and CO2 and the free radicals O, H and OH are
selected for analysis. Figure 19a–e illustrates the mole fraction profiles of various
chemical species for pure methane–air combustion (reference case) and the fuel
mixture containing 10% hydrogen.
As illustrated in Fig. 19a–b, with hydrogen addition, the mole fraction of CH4
reduces while the mole fraction of CO2 increases across the combustor as compared to
the reference case of pure methane–air combustion. The decrease/increase in carbon-
related species in the flame is due to decrease in the molar fraction of CH4 reactant
and the chemical kinetic effect. The dominant chemical reactions contributing to
CH4 are identified and presented as follows [35],

OH + CH4 ↔ CH3 + H2 O (2)

H + CH4 ↔ CH3 + H2 (3)

O + CH4 ↔ OH + CH3 (4)

OH + CO ↔ H + CO2 (5)

From Eqs. (2–4), it is implied that main CH4 consumption reactions are the reac-
tions of reactant CH4 with intermediate species H, O and OH. These free radicals H,
O and OH are highly active during combustion and play major role in the chemical
reactions [35, 36]. The main reactions forming free radicals H, O and OH are

OH + H2 ↔ H + H2 O (6)

H + O2 ↔ O + OH (7)

As illustrated in Fig. 19c–e, the mole fraction of intermediate radicals H, O and OH

increases as hydrogen is added. Increase of H2 mol fraction due to hydrogen addition
increases the rate of reaction of Eq. (6) and thus forms more H and promotes the
reaction rate of Eq. (7). The rate of consumption of CH4 , therefore, increases due to
promotion of Eq. (7) as hydrogen is added. This implies that combustion reactions are
intensified and combustion of methane–air flame is promoted. The increase of CO2
mol fraction due to hydrogen addition is attributed to the increase of OH radicals,
which promotes the oxidation of CO into CO2 as per Eq. (5).
404 P. Rajpara et al.

Fig. 19 Mole fraction profiles of various species a CH4 , b CO2 , c H, d O and e OH

Performance Evaluation of Upward Swirl Combustor … 405

8.2 Influence of Hydrogen Addition on Flame Structure

Flame visualization study is carried out to understand the behaviour of different

hydrogen–methane flames. Luminous photographs of hydrogen–methane flames
captured in a quartz chamber are presented in Fig. 20a–h. Flame visualization reveals
that when methane alone is burned in combustor, a much wider and longer flame
is produced, which is demonstrated by wider blue region in Fig. 20a for reference
case (0% H2 ). Methane–air flame is observed to extend beyond the downstream end
of combustor as reaction continues to occur at the downstream. This is due to poor
mixing in the reaction zone when methane alone is burned.
A little change is observed in the wider and longer blue flame region produced for
methane–air combustion when the hydrogen mass concentration is increased from
0 to 2% in fuel (as shown in Fig. 20b–c). This might be because of little change
in volumetric flow rate of fuel (with fixed air flow rate) at combustor inlet at same
energy input. Significant change in structure and appearance of methane–air flame

(a) (b) (c)

(d) (e) (f)

(g) (h)

Fig. 20 Luminous photographs of different methane–hydrogen flames a 0% H2 , b 1% H2 , c 2%

H2 , d 3% H2 , e 4% H2 , f 5% H2 , g 7% H2 , h 10% H2
406 P. Rajpara et al.

is observed when hydrogen fuel mass concentration is increased from 2 to 10% in

fuel. Figure 20d–h shows that the flames are wider but shorter at higher hydrogen
percentages. In addition, the brightness of visible blue flame region is decreased and
flame appearance is changed from typical blue flame to pale orange-blue. Shorter
flames at higher hydrogen percentages are mainly attributed to high reactivity of
hydrogen, which increases the burning rates and shifts the reaction zone upstream
towards the burner head [37, 38].

8.3 Influence of Hydrogen Addition on Emission

Characteristics

Figure 21 shows the influence of hydrogen addition over the emission levels of CO
and NOx for the same energy input. As illustrated in Fig. 21, hydrogen addition to
methane produces remarkable reduction in CO emission level. Reaction described in
Eq. (5) is the main reaction involved in consumption of CO. This reaction is identified
as the most important reaction for hydrocarbon combustion and nearly all of the heat
released during combustion is expected to occur in this reaction [35]. As discussed
earlier, the addition of hydrogen at same energy input increases the mole fraction of
free radicals H, O and OH in the flame. Increase of OH concentration in the flame
promotes the reaction rate and thereby converts CO into CO2 .
The NOx concentration level, on the other hand, show opposite trends with
hydrogen enrichment. The dominant reactions contributing to NO formation in
methane–hydrogen–air flames are [35, 39],

N + OH ↔ NO + H (8)

N + O2 ↔ NO + O (9)

Fig. 21 Emission levels at 3000 16

different hydrogen
2700 14
percentages
NOx emission (ppm)
CO emission (ppm)

12
2400
10
2100
8
1800 6
1500 4

1200 2
0
0 1 2 3 4 5 6 7 8 9 10
Hydrogen mass (%)
Performance Evaluation of Upward Swirl Combustor … 407

Fig. 22 Flame temperature in primary zone a reference case (0% H2 ), b 10% H2

N2 + O ↔ NO + N (10)

HNO + H ↔ H2 + NO (11)

The reactions described in Eqs. (8–11) occur in high temperature regions [35].
The NOx formation mechanism is correlated with flame temperature in the primary
zone. Figure 22 presents LES temperature contours on vertical plane of combustor
obtained at AFR 50 for reference case and combustion with 10% hydrogen fuel blend.
It is observed that the flame temperature in primary zone increases with hydrogen
blending. This is because hydrogen has higher combustibility than methane, which
results into intensified reactions and thus higher reaction zone temperature. This
favours the formation of thermal NOx at higher hydrogen concentration.
It is interpreted that injection of hydrogen by 5% in fuel blend produces another
40% reduction in CO emission level with marginal increase in level of NOx emission
from 5 ppm to 8 ppm. Further increase in hydrogen concentration from 5% to up to
10% in fuel mixture reduces the CO level, but results in higher combustion temper-
ature and higher levels of NOx . Therefore, limiting the hydrogen concentration to
5% in the fuel mixture can contribute to a noticeable reduction in CO emission level
with marginal increase in the NOx emission level in upward swirl combustor.

9 Conclusions and Future Endeavours

Low NOx levels and low wall-cooling requirements are achieved in a Can-type
combustor by adopting upward swirl philosophy. Upward swirl however suffers from
408 P. Rajpara et al.

poor combustion efficiency and large CO emissions. The development of reverse fuel
injection strategy appears to be promising for improvement in combustion efficiency
and reduction in CO emission levels from upward swirl combustor. The length of
injector has significant influence on increase in wall temperature and the NOx level.
The length of reverse fuel injector can be kept around 5 mm to achieve better perfor-
mance in terms of reduced CO level with negligible effect on the wall temperature
and the NOx level. The CO emission can further be alleviated by blending methane
with hydrogen. For the same energy contribution at the combustor inlet, injection of
hydrogen up to 5% in the fuel mixture reduce CO emissions by another 40% with
little rise in NOx emission value from 5 to 8 ppm.
In future, the performance evaluation of upward swirl combustor with reverse
fuel injector needs to be carried out under different operating conditions. To achieve
high level of combustion efficiency with low level of CO and NOx emissions, the air
supply into upward swirl combustor may be redistributed by suitable mechanism.

Acknowledgements The authors wish to acknowledge the funding of this current research by the
Aeronautics Research and Development Board (ARDB), New Delhi, Government of India, under
the Grant-In-Aid Scheme (Sanction letter no. ARDB/01/1041742/M/I dated 7 October 2014).

References

1. Goldstein RJ, Lau KY, Leung CC (1983) Velocity and turbulence measurements in combustion
systems. Experiments in Fluids 1. https://doi.org/10.1007/bf00266261
2. Bicen AF, Jones WP (1986) Velocity characteristics of isothermal and combusting flows in a
model combustor. Combust Sci Technol 49. https://doi.org/10.1080/00102208608923900
3. Bicen AF, McGuirk JJ, Palma JMLM (1989) Modelling gas turbine combustor flow fields in
isothermal flow experiments. Proc Inst Mech Eng Part A: J Power Eng 203. https://doi.org/10.
1243%2FPIME_PROC_1989_203_015_02
4. Jones WP, Wilhelmi J (1989) Velocity, temperature and composition measurements in a
confined swirl driven recirculating flow. Combust Sci Technol 63. https://doi.org/10.1080/
00102208908947115
5. Koutmos P, McGuirk JJ (1989) Investigation of swirler/dilution jet flow split on primary zone
flow patterns in a water model can-type combustor. J Eng Gas Turbines Power 111. https://doi.
org/10.1115/1.3240253
6. Shah RD, Banerjee J (2014) Development of an upward swirl CAN type combustor. Ph. D.
thesis. National Institute of Technology Surat; 2014
7. Shah RD, Banerjee J (2016) Thermal and emission characteristics of a CAN combustor. Heat
Mass Transfer 52. https://doi.org/10.1007/s00231-015-1572-9
8. Zhang K, Ghobadian A, Nouri JM (2017) Comparative study of non-premixed and partially-
premixed combustion simulations in a realistic Tay model combustor. Appl Therm Eng 110.
https://doi.org/10.1016/j.applthermaleng.2016.08.223
9. Wassell AB, Bhangu JK (1980) The development and application of improved combustor wall
cooling techniques.In: ASME international gas turbine conference and products show, Mar
1980
10. Wadia AR (1988) Advanced combustor liner cooling technology for gas turbines. Defence Sci
J 38
Performance Evaluation of Upward Swirl Combustor … 409

11. Bailey JC, Intile J, Fric TF, Tolpadi AK, Nirmalan NV, Bunker RS (2003) Experimental and
numerical study of heat transfer on turbine combustor liner. J Eng Gas Turbine Power 125.
https://doi.org/10.1115/1.1615256
12. Gao L, Ekkad SV, Bunker RS (2003) Heat transfer under stretched arrays of impinging jets.
In: ASME IGTI conference, Atlanta, GA 2003
13. Hebert R, Gao L, Ekkad SV, Bunker RS (2004) Effect of streamwise pressure gradient on jet
impingement heat transfer. In: 6th ASME ISHMT joint conference, India, Jan 2004
14. Scrittore JJ, Thole KA, Burd SW (2005) Experimental characterization of film-cooling effec-
tiveness near combustor dilution holes. In: ASME turbo expo 2005: power for land, sea, and
air, Jan 2005
15. Li L, Liu T, Peng XF (2005) Flow characteristics in an annular burner with fully film cooling.
Appl Therm Eng 25. https://doi.org/10.1016/j.applthermaleng.2005.03.009
16. Li L, Peng XF, Liu T (2006) Combustion and cooling performance in an aero-engine annular
combustor. Appl Therm Eng 26. https://doi.org/10.1016/j.applthermaleng.2005.11.023
17. Para SR, Li X, Subbuswamy G (2009) Numerical study of mist film cooling in combustor at
operating conditions. ASME International Mechanical Engineering Congress and Exposition,
Jan 2009
18. Ali ABS, Kriaa W, Mhiri H, Bournot P (2012) Numerical investigations of cooling holes system
role in the protection of the walls of a gas turbine combustion chamber. Heat Mass Transfer
48. https://doi.org/10.1007/s00231-011-0932-3
19. Touchton GL, September (1984) Influence of gas turbine combustor design and operating
parameters on effectiveness of NOx suppression by injected steam or water. In: ASME joint
power generation conference
20. Leonard G, Stegmaier J (1994) Development of an aeroderivative gas turbine dry low emissions
combustion system. J Eng Gas Turbines Power 116. https://doi.org/10.1115/1.2906853
21. Enga BE, Thompson DT (1979) Catalytic combustion applied to gas turbine technology.
Platinum Metals Rev 23
22. Mosier SA, Pierce RM (1980) Advanced combustion systems for stationary gas turbine engines.
EPA Contract 68
23. Adachi S, Iwamoto A, Hayashi S, Yamada H, Kaneko S (2007) Emissions in combustion of lean
methane-air and biomass-air mixtures supported by primary hot burned gas in a multi-stage
gas turbine combustor. Proc Combust Inst 31. https://doi.org/10.1016/j.proci.2006.07.239
24. Davis LB, Washam RM (1989) Development of a dry low NOx combustor. In: ASME
international gas turbine and aeroengine congress and exposition, June 1989
25. Wilkes C, Gerhold B (1980) NOx reduction from a gas turbine using exhaust gas recirculation.
ASME Paper No. 80-JPGC/GT-5
26. Arai M, Amano S, Furuhata T (2007) Combustion characteristics in a micro gas turbine
combustor with a recirculation zone induced by an upward swirl. J Environ Eng 2. https://
doi.org/10.1299/jee.2.124
27. Furuhata T, Amano S, Yotoriyama K, Arai M (2007) Development of can-type low NOx
combustor for micro gas turbine (fundamental characteristics in a primary combustion zone
with upward swirl). Fuel 86. https://doi.org/10.1016/j.fuel.2007.01.005
28. Rajpara P, Dekhatawala A, Shah R, Banerjee J (2018) Influence of fuel injection method on
performance of upward swirl can-type combustor. Appl Therm Eng 130. https://doi.org/10.
1016/j.applthermaleng.2017.11.017
29. Phillips JN, Roby RJ (1999) Enhanced gas turbine combustor performance using H2 -enriched
natural gas. In: ASME international gas turbine and aeroengine congress and exhibition, June
1999
30. Schefer RW, Oefelein J (2002) Reduced turbine emissions using hydrogen-enriched fuels. In:
Proceedings of the 2002 US DOE hydrogen program review-NREL/CP-610-32405. June 2002
31. Schefer RW, Wicksall DM, Agrawal AK (2002) Combustion of hydrogen-enriched methane in
a lean premixed swirl-stabilized burner. Proc Combust Inst 29. https://doi.org/10.1016/s1540-
7489(02)80108-0
410 P. Rajpara et al.

32. Juste GL (2006) Hydrogen injection as additional fuel in gas turbine combustor. Evaluation of
effects. Int J Hydrogen Energy 31. https://doi.org/10.1016/j.ijhydene.2006.02.006
33. Cozzi F, Coghe ALDO (2006) Behavior of hydrogen-enriched non-premixed swirled natural
gas flames. Int J Hydrogen Energy 31. https://doi.org/10.1016/j.ijhydene.2005.05.013
34. Emadi M, Karkow D, Salameh T, Gohil A, Ratner A (2012) Flame structure changes resulting
from hydrogen-enrichment and pressurization for low-swirl premixed methane–air flames. Int
J Hydrogen Energy 37. https://doi.org/10.1016/j.ijhydene.2012.04.017
35. Wang J, Huang Z, Tang C, Miao H, Wang X (2009) Numerical study of the effect of hydrogen
addition on methane–air mixtures combustion. Int J Hydrogen Energy 34. https://doi.org/10.
1016/j.ijhydene.2008.11.010
36. Burbano HJ, Amell AA, Garcia JM (2008) Effects of hydrogen addition to methane on the
flame structure and CO emissions in atmospheric burners. Int J Hydrogen Energy 33. https://
doi.org/10.1016/j.ijhydene.2008.04.020
37. Kim HS, Arghode VK, Linck MB, Gupta AK (2009) Hydrogen addition effects in a confined
swirl-stabilized methane-air flame. Int J Hydrogen Energy 34. https://doi.org/10.1016/j.ijh
ydene.2008.10.034
38. Shih HY, Liu CR (2014) A computational study on the combustion of hydrogen/methane
blended fuels for a micro gas turbines. Int J Hydrogen Energy 39. https://doi.org/10.1016/j.ijh
ydene.2014.07.046
39. Kahraman N, Tangöz S, Akansu SO (2018) Numerical analysis of a gas turbine combustor
fueled by hydrogen in comparison with jet-A fuel. Fuel 217. https://doi.org/10.1016/j.fuel.
2017.12.071
Fires
Quantification of Enhanced Fire Severity
in Modern Buildings

Dharmit Nakrani and Gaurav Srivastava

Abstract Modern materials used in the interior as well as exterior of a building pose
new fire safety challenges. From an interior perspective, the traditional materials such
as natural wood and fabrics are being increasingly replaced by modern synthetic
alternatives, which are typically petroleum-based and are hence, more susceptible
to fire. Similarly, from an exterior perspective, current building façade systems are
predominantly based on combustible materials such as aluminum composite panels as
opposed to the traditional non-combustible materials such as terracotta, concrete, and
brick masonry. Modern materials offer significantly improved performance in certain
aspects compared to their traditional alternatives—for instance, most petroleum-
based polymers are much better thermal insulators compared to concrete or masonry
and hence provide significantly better energy efficiency to a building. However, these
modern materials are also significantly more susceptible to being vehicles of fire
spread and hence pose challenges to the designers and engineers, where they have to
find a balance in this pareto optimal choice of modern versus traditional. The current
work examines this issue and presents methods to quantify fire severity of modern
buildings both from interior and exterior perspectives. It is clear that the downside of
modern materials is too prominent to be ignored, and some of the estimation methods
must be factored in the modern building design philosophy.

Keywords Fire risk · Building fire · Façade · ACP · MDF · Furnishing

Nomenclature

P Probability
T Temperature

D. Nakrani · G. Srivastava (B)

Department of Civil Engineering, Indian Institute of Technology Gandhinagar, Palaj,
Gandhinagar, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 413
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_18
414 D. Nakrani and G. Srivastava

1 Introduction

In recent decades, interior and exterior of buildings have witnessed major changes
in the types of materials being used. In most modern habitats, there has been a
noticeable enhancement in the use of petroleum-based polymeric materials such
as plastics, foams, and adhesives. Consequently, modern building fires have also
become more devastating than earlier. From 1993 to 2017, a total of 92.6 million fire
incidents were reported, which caused more than one million deaths and significant
economic losses; the total annual financial loss due to fires is estimated to be 1% of
the world GDP [1, 2]. Developing countries such as India and Pakistan recorded the
highest number of fire-related casualties and the second-highest number of fires [2].
The modern built environment possesses higher internal fuel load (room furnish-
ings that can catch fire) primarily due to the use of synthetic materials in contemporary
furnishings as compared to traditionally available alternatives. Traditional furnish-
ings were typically made from natural materials such as solid wood, cotton, and wool.
In modern furnishings, engineered wood products such as plywood, particle boards
(made from wood chips, sawdust, glue, formaldehyde-based resins, and wax) have
replaced natural solid wood, and polyurethane, nylon, polyester, polyethylene (PE),
acrylic, and other petroleum-based polymers have replaced cotton and wool. These
synthetic materials usually have greater calorific values, lower ignition temperatures,
faster burning times, and produce much more toxic gases during combustion when
compared to natural materials.
Another recent change in the building landscape, especially in developing nations,
is the increase in the number of high rise buildings in cities for both commer-
cial and residential purposes as a result of the increasing population density. Rapid
urbanization, expanding cities, and rising heights of buildings intensified discussions
on sustainability of the modern built environment [3]. Consequently, sustainability
and energy performance became the main parameters in building design and led to
changes in the building envelope materials and their installation techniques. These
building envelopes, typically known as façade or curtain wall systems, render the
building sustainable, make them more energy efficient, and enhance the esthetics.
However, many materials used in building envelops (such as insulation boards and
aluminum composite panels) are also petroleum-based and significantly enhance the
susceptibility of such buildings to fire. In the 2017 fire at the Grenfell Tower, London,
its façade system was found to be the main contributor toward the immense severity
of the incident.
The present work intends to provide a bird’s eye view toward quantification of the
enhanced fire severity of modern buildings due to changes in their internal furnishings
and external façades.
Quantification of Enhanced Fire Severity in Modern Buildings 415

2 Fire Severity Due to Internal Furnishings

Changes in the interior of buildings primarily include (a) the increased use of more
flammable synthetic materials such as plastics and textiles, (b) the increased quantity
of combustible materials, and (c) the use of goods with the unknown composition and
uncertain flammability behavior. Over time, home and office contents have transi-
tioned from being comprised of natural materials to dominated by synthetic materials
[4].
Modern office and residential buildings use corrugated plastic roofs, partition
walls, plastic doors, false ceiling panels, core panels for walls, interior finish mate-
rials, etc. Table 1 lists the time to ignition and peak heat release rates (HRR) observed
through cone calorimeter tests (at incident heat flux level of 50 kW/m2 ) for some
materials which were used traditionally in residential and office occupancies and
have been widely replaced by their modern alternatives.
It can be observed from Table 1 that time to ignition is considerably shorter for
most of the polymeric materials being used in modern occupancies as compared
to the traditional materials. Also, peak HRR is greater for modern materials made
from plastics, acrylic fabric as compared to traditional materials. Further, polymeric
materials have greater calorific values compared to traditional materials [5]. A study
led by Babrauskas [6] compared specimens of upholstered chairs made with different
materials. The cotton-padded chair covered in a cotton fabric produced a peak HRR
of 370 kW at 910 s after ignition. The foam-padded chair covered in polyolefin fabric
produced a peak HRR of 1990 kW at 260 s after ignition. While both chairs exhibited
similar total heat released of about 420 MJ, the synthetic chair synthetic chair yielded
almost five times greater peak HRR than that of the natural chair and this peak was
attained four times faster.

Table 1 Combustion characteristics of various interior materials used in residential and office
buildings
Material Time to ignition (s) Peak HRR Calorific value
(kW/m2 ) (MJ/kg)
Traditional Ghana teak wood 25 298 19
materials Pine wood 30 160 20
Heavy cotton 8 189 15
fabric
Modern Particle board 24 249 21
alternatives Plywood 23 328 20
Polyester 6 210 22
Acrylic fabric 12 386 29
Polyurethane 5 227 24
Polyethylene 31 725 40
(PE)
416 D. Nakrani and G. Srivastava

The survey by Zalok [7] for office buildings showed a significant decrease in
the contribution of cellulosic material (70%) and a greater proportion of plastics
(22%) over the past decades. Further, the survey conducted by Issen [8] for single-
family residential homes in the US found the contribution of cellulosic materials
to be 65% and plastics around (25%). The decrease in the proportion of cellulosic
materials is compensated with an increase in the proportion of polymeric materials.
The introduction of appliances such as computers, printers, air-conditioners, and
refrigerators in modern occupancies has primarily led to an increase in plastic-based
combustible contents. This high intensity of fuel load in modern occupancies plays
a key role in faster fire propagation, shorter flashover times, and rapid changes in
fire dynamics. A full-scale experimental study conducted by Kerber [4] aimed at
characterizing fire development within modern and legacy rooms, concluded that
flashover can occur within 5 min of ignition in modern rooms, whereas legacy rooms
took almost 30 min to flashover (flashover is the point when a local fire suddenly
engulfs an entire room; any evacuation or first-responder firefighting operations are
effective only before the flashover stage). The development of room temperatures in
legacy and modern rooms is shown in Fig. 1 [1, 4].
The overall increase in the total amount of combustible materials within a
typical room, measured through fire load energy density (FLED)—the amount of
combustibles per unit area of a compartment—has been clearly observed through
several fire load surveys across the world. Table 2 summarizes the FLED values in
office-type occupancies (of about 30 m2 floor area) over the last five decades reported
by various surveys and studies. It can be observed that the FLED has significantly
increased due to the use of more flammable materials.
A study by Khan and Srivastava [15] carried out for office and dormitory buildings
observed that, on an average, fire load present in modern buildings is about three times
greater than what is prescribed by building codes (used for design). Based on the
FLED and the compartment characteristics (area, size, windows, doors, etc.), one
can predict the time–temperature curve to be expected during a potential fire through

Fig. 1 Development of
room temperature in legacy
and modern building rooms.
[1, 4]
Quantification of Enhanced Fire Severity in Modern Buildings 417

Table 2 FLED of office

Timeframe Location Mean FLED References
buildings over the past five
(MJ/m2 )
decades around various
geographical locations 1970–1980 USA and UK 365 [9, 10]
1980–1990 New Zealand 496 [11]
1990–2000 New Zealand and 590 [12–14]
India
2000–2010 Canada 850 [7]
2010–2020 India 1205 [15]

the parametric relations available in the Eurocodes [16]. Figure 2 shows some such
parametric fire curves for a typical modern compartment considering the increased
FLED along with the standard fire curves (the ISO 834 and the ASTM E119 fire
curves). The standard fire curves are the ones which are used to prescribe design
requirements for building components. Many building components are required to
have a one-hour fire rating, indicating that the components should not fail when
subjected to a standard fire for a period of one hour. However, if one observes the
parametric fire curves of Fig. 2, it is evident that a component tested with the standard
fire will not sustain one hour of the parametric fire (since the temperatures attained
by the parametric fire are greater than that of the standard fire).
The enhancement in the fire risk can be characterized probabilistically through
Monte Carlo simulations by considering appropriate probability distribution of FLED
(typically, generalized extreme value or gamma distributions [15]) based on the
observed data. This allows the consideration of the uncertainties associated with
the actual distribution of FLED and a robust quantification of the risk. Given that
standard fire temperatures play a key role in defining fire ratings of different building

Fig. 2 Parametric fire

curves compared with
standard fire curves [15]
418 D. Nakrani and G. Srivastava

Fig. 3 P30 and P30 versus

opening factor for dormitory
buildings [15]

components, one may define the fire risk as the probability of the real fire temperature
exceeding the standard fire temperature during different times of a fire. Specifically,
it can be defined as

Pm = P(Tmactual > Tmstdandard )

where m denotes time in minutes, Tmactual is the actual fire temperature after m minutes,
and Tmstdandard is the standard fire temperature after m minutes. Figure 3 shows P30 and
P60 computed for dormitory building configurations using Monte Carlo simulations
with a sample size of 1000 [15]. The opening factor quantifies the extent of openings
in the room in the form of doors and windows (that will allow fresh air ingress and
aid in combustion). It is interesting to note that the opening factor of 0.10 m0.5 comes
out as the ‘optimal’ opening factor from a P60 point of view. Many such insights can
be obtained through such fire risk quantification studies.
Modern polymeric materials also impact the amount of smoke and hazardous
gases generated during their combustion which ultimately reduces the egress time
for occupants and increases health and life risks as compared to traditional mate-
rials. It can be seen from Fig. 4 that CO and CO2 yield obtained through cone
calorimeter results of cellulosic materials (wood, cotton fabric, etc.) is significantly
lower as compared to modern polymeric materials (polyester fabric, nylon fabric,
PE, polyurethane foam, etc.).

3 Fire Severity Due to External Façades

Façade is an important component of modern buildings which is typically non-

load bearing, and its purpose is to separate the indoor environment from external
Quantification of Enhanced Fire Severity in Modern Buildings 419

Fig. 4 CO and CO2 yield of different materials used in residential and office buildings [17]

factors. It also caters to aesthetic, economic, energy, and security considerations.

Earlier cladding systems comprised of materials such as terracotta stone, cement
tiles, masonry veneer, pre-cast concrete panels, and non-structural glass. These
cladding materials were non-combustible and hence possessed negligible fire risk
to the building occupants.
Modern commercial and residential buildings increasingly utilize a façade system
based on glass or a combination of glass and other materials such as aluminum
composite panels (ACP), medium/high-density fiberboards (MDF/HDF), and high-
pressure laminates (HPL). Further, refurbishing of tall concrete buildings often
involves covering the exterior with a rain-screen façade system, consisting of an
outer screen, a cavity, and an inner layer of insulation. This outer screen is usually
ACP, HPL, or mineral fiberboard. The increased use of ACP in the construction sector
is confirmed by the global sales data. Construction application accounted for around
52% of the global ACP market in 2018. Global ACP market was valued at USD
5.33 billion in 2018 and is projected to reach USD 8.71 billion in 2026 [18]. The
newer materials usually possess superior insulating properties as compared to tradi-
tional concrete- or masonry-based systems and hence positively impact the ambient
thermal insulation of the buildings, which leads to lower costs for maintaining the
inside temperatures during regular operations.
While such façade systems render building energy-efficient, reduce air and water
infiltration, and enhance esthetics, they can become easy vehicles for movement of
420 D. Nakrani and G. Srivastava

fire along the building [19, 20]. Bonner and Rein [21] found that the number of
worldwide fires in tall buildings with spread via façade is currently, on average, 4.8
per year, and the total has increased by seven times in the last 30 years. Nyugen et.al.
[22] summarized recent major fire accidents from around the world. Investigations
carried out after these fire accidents have shown that the presence of combustible
materials in the cladding, typically a polymer core sandwiched between aluminum
panels, has been a major contributor in the rapid spread of fire.
Before the installation in actual buildings, façade systems are usually required
to be certified for their fire performance using applicable testing standards ranging
from small (bench) to large-scale tests. Results from one such bench-scale test, using
the cone calorimeter, which provides HRR for several façade materials (at incident
heat flux level of 50 kW/m2 ) are shown in Fig. 5.
Table 3 lists the ignition time and peak HRR (at incident heat flux of 50 kW/m2 )
of commonly used façade materials. The table highlights the short time to ignition
and high HRR values for modern façade materials, whereas the traditional façade

Fig. 5 Heat release rate of various façade materials

Table 3 Combustion characteristics of different façade materials

Material Thickness (mm) Time to ignition (s) Peak heat release rate References
(kW/m2 )
ACP 4 207 350.96 [23]
MDF 12 19 263.52 [24]
HDF 8 44 685.96 [24]
HPL-PF 10 48 572.12 [25]
Quantification of Enhanced Fire Severity in Modern Buildings 421

Fire stop Fire stop

Deck slab Deck slab
Façade Façade

3: Fire 3: Fire
spreads to spreads to
next floor next floor

2. Façade breaks
2: Fire stop
1: Fire starts 1: Fire starts material fails

(a) External spread (b) Internal spread

Fig. 6 Mechanisms of fire spread in the façade systems

systems are non-ignitable. The time to ignition of ACP and HPL is higher than that of
MDF and HDF as the former comprise of a combustible core of polymeric material
sandwiched between metal sheets thereby reducing the direct exposure, whereas the
latter are wood-based panels.
White and Delichatsios [26] studied various small-scale tests and observed that
these tests are applied only to individual material components and represented
specialized exposure conditions. Moreover, bench-scale tests also did not help to
ascertain the fire spread mechanism over the façade systems in real-fire scenarios.
Fire spread in façade systems typically takes place through two mechanisms: external
spread through the leap-frog mechanism, and internal spread [27], as shown in Fig. 6.
It is therefore essential to understand how the whole façade system will perform
and which fire spread mechanism will contribute to fire spread over the cladding
systems. Srivastava et al. [23] highlighted the need for considering the system in its
entirety rather than performing component-level testing. Despite many years of use
of façade systems, full-scale testing remains the only robust option to gain knowledge
about the performance of these systems under real fire scenarios.
A summary of full-scale façade tests of various building standards is given in
Table 4. It can be observed from Table 4 that the geometry, fire source, specimen
support details, test duration, measurements, and acceptance criteria vary greatly
for different testing standards. Also, all full-scale façade tests simulate an internal
post-flashover fire with flames ejecting from windows. However, these tests do not
give an idea about the performance of these systems under the impact of exterior fire
sources or the interaction of façade fire with room furnishings of upper floors.
The full-scale standard façade tests typically consider a single compartment setup,
i.e., the fire compartment is at the ground floor with a standard wood crib as the
fire source and no compartment (and hence, no fire load) at upper floors of the
experimental setup. Thus, the interaction between the furniture of the compartment
and the façade system cannot be studied for upper floors.
422 D. Nakrani and G. Srivastava

Table 4 Summary of different full-scale façade fire tests

Test Arrangement Façade Standard fire Test Failure criteria
dimensions source duration
(min)
ISO Re-entrant H: 4 m above 5.5 MW 23–27 None specified
13,785:2002 corner “L” window lintel gas burner in
arrangement W: 3 m Wing enclosure
(wing wall) width: 1.2 m
BS 8414 part 1 Re-entrant H: 6 m above 3 MW 30 Exterior or
and Part 2 corner “L” window soffit timber crib interior fire
arrangement W:2.6 m Wing in opening spread 5 m above
(wing wall) Width: 1.5 m window within
15 min
DIN 4102—20 Re-entrant H: 5.5 m 320 kW 20 Exterior or
corner “L” W: 2 m gas burner in interior fire
arrangement wing width: opening spread 3.5 m
(wing wall) 1.4 m above the
window
NFPA 285 Single wall Two-story test 900 kW 30 Exterior fire
surface frame; H: 5.3 m gas burner in spread > 3.05 m
W: 4.1 m bottom above the
enclosure window
and 400 kW
gas burner in
window
(ignited
5 min after
room burner)
SP FIRE 105 Single wall H: 6 m W: 4 m 2.5 MW 15 Exterior fire
surface includes a Heptane fuel spread > 3.2 m
500 mm eave at tray in an above the
top enclosure window
CAN/ULC Single wall H:7.25 m 5.5 MW 25 Exterior fire
S134 surface above window propane spread > 5 m
W: 6 m burner or above the
timber crib in window
an enclosure
GB/T 29,416 Re-entrant H:9 m Timber crib – –
corner “L” W: 2.6 m Wing or gas burner
arrangement width: 1.5 m in an
(wing wall) enclosure

Srivastava et al. [27] developed a full-scale façade testing facility at Indian Institute
of Technology Gandhinagar, India, where full-scale façade systems can be studied
under real fire scenarios. The testing facility is a three-story steel structure with three
independent compartments at each level, as shown in Fig. 7. The plan dimensions of
the compartments at each story are 10 × 20 (approximately 3 m × 6 m) with a story
height of 10 . The facility has been designed to simulate both fire spread mechanisms
Quantification of Enhanced Fire Severity in Modern Buildings 423

Face 3

Face 1

Fig. 7 Full-scale façade fire experimental facility at IIT Gandhinagar [27]

(external and internal). Two faces (Faces 1 and 2) of the test structure can simulate
external spread mechanism, while one face (Face 3) has been designed to simulate the
internal spread mechanism. This facility allows (i) to test façade systems including
curtain wall and perimeter firestop as per actual installation practices, (ii) use of
real furniture to simulate real fire scenario, and (iii) provision of fully furnished
compartments at all floor levels to enable the study of interaction between façade
fire and compartment components.
Several full-scale tests with different combinations of façade assemblies such as
glass-ACP and glass-MDF have been performed at this facility. Before the onset
of fire in experiments, fire scenarios are developed using a combination of real
office/residential furniture and rated wooden cribs with the desired FLED based on
the considered design parameters. The test structure is instrumented with K-type ther-
mocouples for temperature measurements, internal video cameras to monitor smoke
movement, an elaborate smoke collection system, and external infrared cameras to
measure surface temperatures.
Through numerous experiments, it has been observed that façade fire in assemblies
with glass and ACP can reach from one story to the next through the leap-frog
mechanism in about 3 min and hence can severely limit the response and egress
time for higher floors of a building. Once the fire breaches the compartment of
origin and combustible façade catches fire, the vertical spread over the external
surface between story is less than 3 min. Further, the experiments showed that the
façade system can be exposed to heat flux levels in excess of 100 kW/m2 with
maximum temperatures reaching 1000 °C. The existing testing standards consider
lower incident flux levels, and the gas burners used in most existing test methods can
reach maximum temperatures of around 700 °C. These findings indicate a significant
gap between current testing standards and real fire performance of façade systems
[23].
Glass-MDF façade assembly tested under real fire scenarios has been found to be
better than glass-ACP systems as no flame spread was observed over the surface once
424 D. Nakrani and G. Srivastava

(a) ACP with exposed PE core residue (b) MDF after getting charred

Fig. 8 Behavior of different façade materials after being exposed to real fire scenarios

the fire breached the compartment of origin. External vertical spread over the MDF
surface was not observed as opposed to rapid flame spread over the ACP surface.
Failure of MDF panels wherein they fall out from their positions occurred in the
compartment of origin within 10 min of the ignition, whereas ACP typically falls
out within 4 min of the ignition.
ACP comprises of a PE infill placed between two aluminum sheets. Aluminum
melts around 660 °C, while PE can catch fire in a matter of seconds (see Table 1).
During a fire, flame temperatures typically range from 600 to 800 °C, and thus,
aluminum begins to melt early. The combustible core of PE aids the melting of
aluminum as the polymer at the periphery catches fire and begins to drip and vaporize
thereby increasing the local fuel concentration. However, MDF is a wood-based panel
and hence does not melt. During combustion, it usually chars and retains its strength
for a relatively longer time when compared to ACP. The combustion of PE leaves
residue in the aluminum sheets, while MDF fails after considerable charring, as
shown in Fig. 8. Due to these inherent material-level differences in ACP and MDF,
an ACP-based façade tends to fail much faster than that of an MDF-based façade
system.
When façade panels begin to fail (and fall), they create more and more openings
on the façade side of the building. This allows more access for fresh air inside the
fire compartment and makes it easier for the fire to sustain. The extent of openings,
measured through the opening factor (or ventilation factor), can be used as an indi-
cator of the spread of fire within a compartment as well as the inter-story spread.
Figure 9 shows the cumulative ventilation factors for two full-scale experiments:
one with glass-ACP façade system and other with glass-MDF system with the same
initial conditions and fire scenario. It can be observed that in the ACP-based system,
the ground and first floors become fully open from the façade-side within 10 min
of ignition, whereas in the MDF-based system, it takes about 20 min for the ground
floor to become fully open and the fire does not propagate to first or second floors
(they remain completely closed from the façade-side).
Overall, the degree to which a façade system might become involved in a fire
is dependent on its overall flammability. By flammability, it is referred to the ease
Quantification of Enhanced Fire Severity in Modern Buildings 425

(a) Glass-ACP façade assembly (b) Glass-MDF façade assembly

Fig. 9 Ventilation factor w.r.t time plots for different façade assemblies

of fire to ignite and spread through the façade. Flammability is a broader concept
that depends on how different materials are combined, oriented, and connected
in the façade system, and how they interact during a fire. While the traditional
non-combustible façade systems cannot become vehicles of fire like their modern
combustible counterparts, they are also less energy-efficient and increase the overall
heating/cooling costs of the building during day to day operations.

4 Conclusions

This disposition aimed to highlight the dilemma of current designers and engineers
dealing with modern buildings. The advantages related to esthetics and enhanced
functionality of modern materials is on one side, and the disadvantage of their greater
fire susceptibility is on the other. The engineers (and perhaps occupants) need to be
cognizant of the need to balance the two sides in order to achieve a reasonably safe
design. Quantification of the enhanced fire severity due to modern materials is an
essential exercise for anyone to understand the flip side of modern materials.
Cone calorimeter results of various traditional materials and modern engineered
alternatives have shown that modern materials are easier to ignite and have higher
HRR values. Various fuel load surveys from across the globe have shown the consis-
tent rise in the overall fire load within buildings. Specifically, for India, the fire loads
have become three times greater than those reported in previous studies conducted
in the 1990s. The proportion of natural cellulosic materials has decreased consid-
erably in recent times, whereas the proportion of plastics and synthetic materials,
which have much higher calorific values than cellulose, has increased. Parametric
design fire curves provide further quantification of the fire severity by predicting
time–temperature curves of potential fires within building compartments. Parametric
curves computed through actual fire loads observed in recent surveys tend to be more
426 D. Nakrani and G. Srivastava

severe than the standard fire curves which are typically employed for characterizing
fire ratings of different components of a building. A probabilistic characterization of
the enhanced fire risk due to the changes in the materials used for room furnishings
can also be done through Monte Carlo simulations based on the developed probability
models for fire loads to assess optimal design parameters for building compartments.
Modern exterior façade materials such as ACP, HDF, MDF, and HPL are also
highly susceptible to ignition and flame spread, while the traditional façade materials
such as terracotta stone, cement tiles, masonry veneer, pre-cast concrete panels, and
non-structural glass are all non-combustible. The inability of bench-scale tests in
assessing the performance of modern combustible façade systems under real fire
scenarios and full-scale standardized facade tests available have been discussed along
with their limitations. This highlights the need for full-scale testing in actual site
conditions under real scenarios. A comparison of the performance of dominantly
used glass-ACP façade system and a glass-MDF system under real fire scenarios
showed significant insights. It was observed in glass-ACP assembly that the time
of vertical spread of fire between stories is less than 3 min indicating an extremely
short window of time available for egress and rescue operations. The tests also showed
that during real fires, these combustibles façade systems can be exposed to heat flux
levels above 100 kW/m2 which the standardized full-scale tests do not account for.
Overall, this work provided insights toward the quantification of enhanced fire
severity of modern buildings due to changes in internal furnishings and external
façade systems. It is expected that more research in this area with some of the
discussed factors will lead to the improvement of building standards and eventually
produce safer designs in the future.

Acknowledgements The support from the Centre for Safety Engineering, IIT Gandhinagar, is
gratefully acknowledged. In the development of the full-scale fire testing facility, the support
from Underwriters Laboratories LLC is acknowledged. The support from Ministry of Human
Resource Development (Government of India) under the Uchchatar Avishkar Yojana (UAY) is
also acknowledged.

References

1. Kodur V, Kumar P, Rafi MM (2019) Fire hazard in buildings: review, assessment and strategies
for improving fire safety. PSU Res Rev 4(1):1–23
2. Brushlinsky NN, Ahrens M, Sokolpv S, Wagner P (2019) World fire statistics
3. Zeng C, Deng X, Dong J, Hu P (2016) Urbanization and sustainability: comparison of the
processes in ‘BIC’ countries. Sustain 8(4):1–18
4. Kerber S (2012) Analysis of changing residential fire dynamics and its implications on
firefighter operational timeframes. Fire Technol 48(4):865–891
5. Ljungberg LY (2007) Materials selection and design for development of sustainable products.
Mater Des 28(2):466–479
6. Babrauskas V, Lawson RJ, Walton DW, Twilley HW (1982) Upholstered furniture heat release
rates measured with a furniture calorimeter
7. Zalok E (2011) Validation of methodologies to determine fire load for use in structural fire
protection. Fire Prot Res Found, 1–65
Quantification of Enhanced Fire Severity in Modern Buildings 427

8. Issen LA (1980) Single-family residential fire and live loads survey. NIST Interagency/Internal
Rep
9. Baldwin R, Law M, Allen G (1970) Survey of fire loads in modern office buildings—some
preliminary results. Fire Saf Sci 808:1–1
10. Culver CG (1976) Survey results for fire loads and live loads in office buildings. Natl Bur Stand
Build Sci Ser 85
11. Barnett CR (1984) Pilot fire load survey carried out for the New Zealand Fire Protection
Association. MacDonald Barnett Partners, Auckl
12. Buchanan AH, Abu AK (2017) Structural design for fire safety, 2nd edn. Wiley, West Sussex
13. Kumar S, Rao CVSK (1997) Fire loads in office buildings. J Struct Eng 123(March):365–368
14. Kumar S, Rao CVSK (1995) Fire load in residential buildings. Build Environ 30(2):299–305
15. Khan NA, Srivastava G (2018) Enhanced fire severity in modern Indian dwellings. Curr Sci
115(2):320–325
16. Eurocode 1 (2004) Actions on structures—part 1–2: general actions -actions on structures
exposed to fire, London
17. Wg Cdr GA Sachdev VS, Kotresh TM, Vyawahare MK (2004) Heat release characteristics of
the basic materials used for flying clothing. Ind J Aerosp Med 48(1):53–58
18. “Aluminium Composite Panels (2019) Global market analysis, insights and forecast, 2019–
2026
19. Gandhi P et al (2017) Performance of glass-ACP façade system in a full-scale real fire test in
a G+2 structure. Procedia Eng. 210:512–519
20. Peng L, Ni Z, Huang X (2013) Review on the fire safety of exterior wall claddings in high-rise
buildings in China. Procedia Eng. 62:663–670
21. Bonner M, Rein G (2018) Flammability and multi-objective performance of building façades:
towards optimum design. Int. J. High-Rise Build. 7(4):363–374
22. Nguyen K, Weerasinghe P, Mendis P, Ngo T, Barnett J (2016) Performance of modern building
façades in fire: a comprehensive review. Electron J Struct Eng 16(1):69–86
23. Srivastava G, Nakrani D, Ghoroi C (2020) Performance of combustible facade systems with
glass, ACP and firestops in full-scale, real fire experiments. Fire Technol 56(4):1575–1598
24. Lee BH et al (2011) Evaluating the flammability of wood-based panels and gypsum
particleboard using a cone calorimeter. Constr Build Mater 25(7):3044–3050
25. McKenna et al (ST) Fire behaviour of modern façade materials—understanding the Grenfell
Tower fire. J Hazard Mater 368(2018):115–123 (2019)
26. White N, Delichatsios M (2015) Fire hazards of exterior wall assemblies containing combustible
components, 1st edn. Springer, New York
27. Srivastava G et al (2018) Development of a unique full-scale real-fire façade testing facility at
IIT Gandhinagar. Curr Sci 115(9):1782–1787
The Tea Light Candle and the Global
Waste Problem

Ali S. Rangwala and Kemal S. Arsava

Abstract In 2012, the World Bank estimated that each person living on planet Earth
produces approximately 1.2 kg of waste per day, amounting to 1.2 billion tons per
year. By 2025, this number is expected to reach a staggering 2.2 billion tons, which
raises the fundamental question: What is and what happens to waste? Domestic
burning of waste is a frequent disposal technique in many parts of the world. Further,
even when waste is moved to dumpsites in many developing countries, it is not
uncommon for the material to be burned by open, uncontrolled fires. In short, fire
is the primary mode of waste disposal. Fire provides volume reduction and prevents
disease, but inefficient combustion brings its hazards in the form of toxic/noxious
gasses. This paper will discuss the work we have been doing at WPI related to burning
hazardous waste cleanly and efficiently by observing a tea light candle’s behavior.

Keywords Fire · Flame refluxer · Candle · Waste burning

1 Background

The world population generates 1–2 billion metric tons of municipal solid waste per
year globally1 . In the last decade, this waste’s nature has shifted to include increasing
amounts of plastics and electronic waste as shown in Table 1. The “waste” can be
considered a reasonably good combustible with an approximate heat of combustion
slightly higher than wood and equal to 20 kJ/g.
Fire is the most efficient means of reducing the volume of solid waste. This has
led developed areas to invest in waste incineration facilities, which require a large
amount of infrastructure and capital investment. In less-developed areas, fires for
waste management are typically open-air and present both health and environmental
hazards. The purpose of this paper is to demonstrate new techniques in burning
waste outdoors, which can reduce the environmental impact by increased combustion
temperatures and added capability to burn waste streams that are moist and have lower

A. S. Rangwala (B) · K. S. Arsava

Worcester Polytechnic Institute, Worcester, MA 01609, USA
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 429
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_19
430 A. S. Rangwala and K. S. Arsava

Table 1 Typical composition

% kJ/g
of municipal waste
Paper 41 10.4
Yard waste 15.2 9.3
Plastics 10 46
Food waste 16.8 7.6
Wood 3.8 16
Textiles 2.2 15
Rubber and leather 2.4 19.5
Glass 7.6
Metals 9
Miscellaneous 2
kJ/g
Coal 29
Natural gas 45
Propane 50
Methanol 22
Wood 16.5
Waste 20

heat of combustion. A simple example of a tea light candle is used to demonstrate

the problem and the solution.
When the waste is burned in the open, the fire takes the form of a diffusion flame
as shown in Fig. 1. Typically, fires of this type burn at a regression rate of 0.1 mm/s.
This rate is slow, mainly because fuel vaporization is slow because of the highly
endothermic nature of the pyrolysis process. If we imagine 1 cubic foot (0.31 ×
0.31 × 0.31 m) of solid waste burning as a pool in one plane, it will burn out in
(0.31 m)/(0.1 mm/s) ~ 52 min. However, if the same solid waste can be burned in a
premixed mode, like in a Bunsen burner, for example, then its regression rate increases

Fig. 1 Comparison between

rates of burning between a
turbulent pool fire and a
Bunsen burner flame. The
fuel consumption rate is
prolonged when burning as a
pool, which results in
incomplete combustion and
an overall slow-burning
process. For waste burning,
this translates to a more
significant strain on the
environment
The Tea Light Candle and the Global Waste Problem 431

to ~ 1 m/s and will burn out in (0.31 m)/(1 m/s) = 0.3 s! This drastic reduction in burn
time has advantages. The increased temperatures cause the flame to be less sooty as
shown in Fig. 1, which compares a hydrocarbon pool fire (diffusion) and a Bunsen
burner premixed flame. The question then is, can we increase the burning rate or
regression rate of the free diffusion flame? We will proceed to answer this question
by using a common tea light candle and show how we can increase its regression
rate by a factor of 10 times.

2 The Tea Light Candle

2.1 How Long Does It Take for a Tea Light Candle to Burn?

Let us consider the example of a candle flame shown in Fig. 2. If paraffin is ignited
with a match, it is challenging to burn. The candle’s paraffin melts because of heat
from the flame; it travels by capillary forces through the wick, where it then evaporates
to become paraffin vapor, a gaseous fuel. The process is illustrated in Fig. 2.
The slow-burning candle flame takes around 4 h, 30 min to be fully consumed.
The long burning time makes it amenable for decorations at home. The wick ensures
that burning is complete, resulting in a bright yellow flame with no visible soot. Solid
paraffin in a tea light represents an ideal solid fuel because it is homogeneous and
melts at low temperatures to form a low viscosity liquid. Polyethylene and PVC,
typical plastics in solid waste, will melt at low temperatures, but they have different
flow properties from paraffin in the molten state. It will burn slowly by first using the
flame’s heat to form a melt pool, gasifying to create fuel vapor. The entire process is
inherently slow because of the conversion of condensed fuel to vaporized fuel. An
obvious way of burning the candle faster is to increase the number of wicks discussed
in the next section.

Fig. 2 Candle flame [1]

432 A. S. Rangwala and K. S. Arsava

Fig. 3 Addition of wicks to

increase the burning rate of a
tea light candle

2.2 How Can the Burning Rate Be Increased?

The burning rate of the tea light candle can be increased by adding more wicks as
shown in Fig. 3. With two wicks, the candle is consumed in 2 h, with three wicks in
1 h 20 min, and four wicks in 1 h as shown in Fig. 3. Adding more wicks does not
improve the burning further because the wicks increase the paraffin’s melting rate,
and at some point, when the paraffin is completely melted in the container, it burns
like a small pool fire. At this stage of burning, the purpose of wicks is lost.

2.3 How Can the Burning Rate Be Increased Even Further?

Figure 4 shows the tea light candle, with the addition of four copper rods of 1 mm
in diameter. The height of the copper rods was approximately two times the base’s
height as shown in Fig. 4. As the candle burns, it heats the copper rods, which
then capture and conduct (“reflux”) the flame heat back to the paraffin. The wax is
observed to melt in 2 min, and in 22 min, a pool fire is observed and the wicks drop
off. The burning rate continues to accelerate because of the copper. At ~30 min, the
formation of bubbles is observed on the copper rod’s surface in the molten pool, and
consequently, the burning rate increases because of the improvement in heat transfer
rates because of nucleate boiling. The flame height at around 33 min is ~10 times
a normal candle flame, and the candle is consumed in 37 min. The enhanced flame
height after 30 min is shown in Fig. 4. It would have been impossible to generate an
increase in flame height of this magnitude without the onset of nucleate boiling that
occurs because of the rods’ presence.
Figure 5 shows a plot of the burning time of the tea light candle. Adding wicks
improves the burning rate to some extent. However, the addition of the copper rods
improves the burning rate much more. Further, the molten pool is wholly consumed
The Tea Light Candle and the Global Waste Problem 433

Fig. 4 Increase in burning rate by a factor of 10 by addition of four copper rods—the Flame
Refluxer™ concept

Fig. 5 Burning of a tea light candle. The copper rods can burn out the tea light candle in ~35 min
by creating an additional heat transfer pathway via conduction through the copper

from 30 to 37 min. Thus, if additional wax were to be supplied at this stage of burning,
then it would be consumed at a significantly higher rate (~10 times).
The slower burning during normal usage is the resistance to heat transfer from
the flame to the surface. Torero et al. [2] characterized the theoretical efficiency of
heat transfer for burning condensed fuels showing it could be as low as 0.3%. With
434 A. S. Rangwala and K. S. Arsava

such a small fraction of the total heat released actually returned to the fuel surface,
the vaporization and consequently combustion is slow as most of the heat is lost to
the environment through the buoyant convection plume and thermal radiation [2–
6]. The addition of copper rods to the tea light candle creates a secondary pathway
for heat transport to the paraffin wax. This pathway is coupled to the flame, and as
heat is transferred, the burning rate improves, which enables further heat transport
and so on. Once nucleate boiling starts on the copper rod’s surface, the burning rate
increases by a factor of 10, and the paraffin candle is consumed in a few minutes!
The baseline (unenhanced) burning rate of a tea light-sized pool (or diameter) of
paraffin wax was measured to be ~2 g/m2 s. What is the cause of the enhancement,
especially the role of the bubble dynamics during the last 7 min of the burn when
the flame height is ten times higher than the baseline? The controlling parameters’
role can be further explained, assuming that the heated copper rods will produce
bubbles because of subcooled nucleate boiling. Recent experiments by the research
group have demonstrated that the bubbles generation rate per unit area ḃg ~ 4 × 105
bubbles/m2 -s [7, 8]. Note that the generation rate will vary with several parameters,
like the thermofluid properties of molten wax and the surface roughness of the copper.
However, the ballpark number allows us to get an order of magnitude analysis on
bubbles’ influence. Assuming that the bubbles are spherical and the average bubble
size is r ~ 1 mm [9],

4 3
ṁ bubble−vapor = πr ρg ḃg , (1)
3

where ṁ bubble−vapor is the mass flux of fuel vapor generated by bubbles reaching the
surface and consequently releasing the vapor of density ρg . Substituting r = 1 mm,
ρg = 7300 g/m3 , and ḃg = 4 × 105 bubbles/m2 -s leads to ṁ bubble−vapor = 1.5 g/m2 -s.
Further, the mass flux of fuel vapor because of a micron droplet spray by bubble
bursts on the surface, ṁ bubble−spray , is given by

4 3
ṁ bubble−spray = πr ρ n drop ḃg , (2)
3 drop
where n drop is the number of droplets produced per bubble burst and rdrop equals
the average drop radius. Spiel [10] experimentally observed that a 1 mm bubble
bursting on the water surface creates anywhere from 50 to 200 drops of 10–100 µm
size. Assuming that each bubble of paraffin vapor upon bursting produces ~100 drops
of 100 µm size and substituting ρ = 749,000 g/m3 , rdrop = 10–4 m, n drop = 100, and
ḃg = 4 × 105 bubbles/m2 -s into Eq. (2) give ṁ bubble−spray = 15.65 g/m2 -s.
The above example shows that a significant enhancement in the burning rate over
the baseline, increasing from 2 to 17.2 g/m2 -s (~9×), is achieved mainly because of
bubble bursts on the surface. With an increase in the rod surface area, for example,
by changing surface conditions [11, 12], the burning rate will increase further. Also,
by optimizing the rod geometry and the number of rods, the time to melt the pool
The Tea Light Candle and the Global Waste Problem 435

and create the nucleate boiling can be reduced even further. In short, the tea light
candle can be burned in an even shorter time. One can hypothesize that the upper
limit will be controlled by the critical heat flux (CHF) condition during subcooled
nucleate boiling. Besides, it will depend on the film thickness of the bubble at the time
of bursting. The latter is coupled with flame heat flux by evaporation, a parameter
not explored in the literature [9]. Elucidation of these mechanisms has significant
implications and has never been addressed in fire research.

3 The Flame Refluxer Concept and Applications

We have seen that conductive nonflammable objects, such as metal rods, can be placed
into a condensed fuel to enhance the burning rate. This idea led to the development
of a patented technology named the Flame Refluxer™ [14–16]. The radiative and
convective heat generated by the combustion is directed back to condensed fuel
to create a feedback loop that effectively enhances the burning rate. The increased
burning rate also results in a reduction of smoke [16]. The simple but effective design
of the Flame Refluxer™ requires no moving parts, atomizing nozzles, or compressed
air, which makes it more reliable than existing burner technologies. The fact that it
uses heat generated by combustion itself to improve burning makes it amenable to
rapid waste burning. Figure 6 shows a timeline of the project’s evolution from concept
to a tested technology at both laboratory and field. The idea has been successfully
applied to burn hydrocarbon liquids, emulsions, waste liquids trapped in the soil,
human/animal waste, and solid fuel slurries mixed in water. Details of the Flame
Refluxer™ in different applications are available in reports [17, 18] and papers [13,
16, 19–22].

Fig. 6 Timeline of the Flame Refluxer™ from concept to application in different areas, a trans-
former fires [13], b arctic oil spill burning [5], c waste burner (solid fuel), d waste oil burner (liquid)
[14], e offshore in situ burning [15], f biowaste burning
436 A. S. Rangwala and K. S. Arsava

4 Conclusions

The paper demonstrates how a tea light candle’s burning rate can be increased by
a factor of 10 by using four copper rods immersed in the wax. Nucleate boiling
(formation of bubbles on the metal’s surface immersed in the liquid) is the main
reason for significant enhancement of mass burning rate. The metal object immersed
in the flame zone creates a high-efficiency pathway for heat transfer from the hot
flame zone (~800–1000 °C) to the cooler fuel layer (~200–400 °C). This is because
the portion of the object immersed in the fuel can easily reach temperatures above
the fuel boiling point resulting in subcooled nucleate boiling, which increases the
heat transfer rates by order of magnitude (×100). This is the main reason for the
fuel’s faster evaporation and the corresponding increase in the burning rate.
The idea of immersing a metal object in fuel to enhance its burning rate led to the
development of a patented technology named Flame Refluxer ™. This technique can
be extended to burn a variety of wastes quickly and efficiently, creating a pathway
for efficient waste disposal in specific applications related to waste burning.

Acknowledgements The Flame Refluxer is funded by the Bureau of Safety and Environ-
mental Enforcement, U.S. Department of the Interior, Washington, DC, under contract number
E14PC00043. The contents do not necessarily reflect the BSEE’s views and policies, nor does
mention of the trade names, or commercial products constitute endorsement or recommendation
for use. Many thanks to people, who made contributions to this study, including Karen Stone, Kristi
McKinney, Glenn Mahnkenn, Professor Raghavan Vasudevan (IIT Madras), Professor Jose Torero
(University College, London), and students at WPI (in alphabetical order) Apoorv Walavalkar,
Aravind Krishnan, Dave Petrow, Drew Egger, Francisco Kang, Hamed Farmahini, Hayri Sezer,
Honggang Wang, John Ramos, Karl T. Sundberg, Kathryn Hall, KiJana M. Haney, Li Chang,
Lydia Shi, Mahesh S. Kottalgi, Nathaniel G. Sauer, Nathan S. Ghion, Nicolas Tobar, Panyawat
Tukaew, Raymond Ranellone, Shijin Kozhumal, Shivaprasad Arava, Steve Stephansky, Trevor
Borth, Veronica Kimmerly, Xiaoyue Pi, Zhaozhao Meng.

References

1. SFPE (2008) SFPE handbook of fire protection engineering, 4th edn. Society of Fire Protection
Engineering, Bethesda
2. Torero JL, Olenick SM, Garo JP, Vantelon JP (2003) Determination of the burning character-
istics of a slick of oil on water. Spill Sci Technol Bull 8(4):379–390
3. Brosmer M, Tien C (1987) Radiative energy blockage in large pool fires. Combust Sci Technol
51(1–3):21–37
4. Gritzo L, Nicolette V, Tieszen S, Moya J, Holen J (1995) Heat transfer to the fuel surface in
large pool fires. In: Chan SH (ed) Transport phenomenon in combustion. Taylor and Francis
5. Shi X, Bellino P, Simeoni A, Rangwala A (2016) Experimental study of burning behavior of
large-scale crude oil fires in ice cavities. Fire Saf J 79:91–99
6. Jiang F, De Ris J, Qi H, Khan M (2011) Radiation blockage in small scale PMMA combustion.
Proc Combust Inst 33(2):2657–2664
7. Pi X, Rangwala AS (2019) The influence of wire orientation during nucleate pool boiling in
subcooled dodecane. Int J Heat Mass Transf 137:1247–1257
The Tea Light Candle and the Global Waste Problem 437

8. Sauer N, Pi X, Arsava KS, Rangwala AS (2020) Controlling mechanisms of burning rate

enhancement while using Flame RefluxerTM technology during in situ burning of crude oil
spills. In: Proceedings of international oil spill conference
9. Poulain S, Villermaux E, Bourouiba L (2018) Ageing and burst of surface bubbles. J Fluid
Mech 851:636–671
10. Spiel DE (1995) On the births of jet drops from bubbles bursting on water surfaces. J Geophys
Res Oceans 100(C3):4995–5006
11. Jones BJ, McHale JP, Garimella SV (2009) The influence of surface roughness on nucleate
pool boiling heat transfer. J Heat Trans 131(12)
12. Chung SH, Law CK (1984) Importance of dissociation equilibrium and variable transport
properties on estimation of flame temperature and droplet burning rate. Combust Flame 55:225–
235
13. Shi X, Rangwala AS (2013) Influence of immersed objects on liquid pool boilover. AICHE
Annual Meeting, San Francisco
14. Rangwala AS, Shi X, Arsava KS, Mahnken G (2017) Methods and systems for clean-up of
hazardous spills. Google Patents
15. Arsava KS, Rangwala AS, Mahnken G (2019) Systems and methods for in-situ clean up of
burnable materials. Google Patents
16. Tukaew P (2017) Outdoor gas emission sampling system: a novel method for quantification of
fires in outdoor conditions
17. Rangwala AS, Arsava KS, Mahnken G, Shi X (2015) A novel experimental approach to enhance
burning of oil-water emulsions by immersed objects. In: B.o.S.a.E. Enforcement (ed) Worcester
Polytechnic Institute
18. Rangwala AS (2017) An offshore oil burn enhanced by floating immersed objects. In: B.o.S.a.E.
Enforcement (ed) Worcester Polytechnic Institute
19. Arsava KS, Raghavan V, Rangwala AS (2018) Enhanced oil spill clean-up using immersed
thermally conductive objects. Fire Technol 54(6):1745–1758
20. Arsava KS, Borth TJ, Fields SL, Hansen K, Kozhumal S, Mahnken G, Petrow D, Rangwala
AS, Sezer H, Stone K (2019) Application of flame RefluxerTM concept to ISB–experimental
results of 5 field trials in mobile, alabama. In: Proceedings-42nd AMOP technical seminar on
environmental contamination and response. Emergencies Science and Technology Section
21. Sezer H, Arsava KS, Kozhumal SP, Rangwala AS (2017) The effect of embedded objects on
pool fire burning behavior. Int J Heat Mass Transf 108:537–548
22. Sezer H, Arsava KS, Rangwala AS (2017) Oil spill clean-up using immersed metal wool. J
Environ Chem Eng 5(5):5196–5206
Sustainable Energy Technologies
Approach to Solve Renewable Energy
Problems

Sourabh Kumar and Ryoichi S. Amano

Abstract Computational fluid dynamics had shown significant development in soft-

ware capability and hardware during the last two decades. It has become an essen-
tial tool in various product development stages and is used actively in advanced
research technology. Renewable energy varies from traditional to new resources such
as biomass, fuel cells, water, wind, and solar. These renewable sources are playing a
significant role in electric generation, and both energy producers and consumers are
demanding more deployment of these sources. There is a great advantage in using
numerical simulation methods for innovative renewable technologies such as wind
energy, biomass, solar power, hydropower, gas and steam turbines, fuel cells, tidal,
wave, and many others. CFD analysis provides an additional resource for these inno-
vative renewable energy sources to predict results that can save time and money and
reduce physical prototyping and testing resources. This paper intends to outline and
discuss all CFD applications in renewable energy sources, helping understand these
methods’ advantages and usage.

Keywords Computational fluid dynamics · Renewable energy · Wind turbine ·

Hydropower · Biomass · Tidal · Solar and geothermal

Abbreviations

A Area
Cp Pressure coefficient
CFD Computational fluid dynamics
DES Detached Eddy simulation
k Velocity

S. Kumar
Jacobs Engineering Inc, Bingham Farm, USA
R. S. Amano (B)
University of Wisconsin, Milwaukee, WI, USA
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 441
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_20
442 S. Kumar and R. S. Amano

K Kelvin
LES Large eddy simulation
m Meter
OWSC Oscillating wave surge converter
OWC Oscillating wave converter
P Power output
PA Point absorber
Pave Power output
Q Volume of water flow
RANS Reynolds averaged Navier–Stokes
SST Shear stress transport
T Time
U Velocity
VOF Volume of fluid
WEC Wave energy converter
W Watt

Nomenclature

ï Efficiency
ρ Density
g Acceleration due to gravity
h Acceleration due to gravity
ω Specific rate of dissipation

1 Introduction

Fluid dynamics involves a partial differential equation for physical laws. CFD solvers
transform these partial differential equations into algebraic equations and solve using
computers numerically. Computational fluid dynamics is a method of solving numer-
ically the laws that govern the movement of fluid. The complete system can be trans-
formed into a virtual product and can be visualized virtually for analysis. One of
the advantages of using CFD over the experimental method is that the flow analysis
can be achieved without interfering with the flow and can access the flow variable’s
details where conventional experimental techniques fail. CFD can be implemented
to save time and resources during design iterations for faster and economical than
tests. CFD analysis can be used for measuring all quantities with higher resolution
in space and time. CFD process can be a handy tool for minimizing the test if the
methodology has been verified and validated with tests. With the advancement in the
Approach to Solve Renewable Energy Problems 443

numerical algorithm and computing power, there is the widespread use of CFD tech-
niques in various fields such as gas turbine, heat transfer, multiphase, combustion,
and several other areas.
Overall, the energy demand increases over the years, and dependence on fossil
fuels causes environmental problems. Increase the use of renewable energy sources
will solve the environmental concern along with solving power demand. Renewable
energy is the energy source that replenishes naturally, and these resources are inex-
haustible but are limited by the amount of energy per unit of time [1]. Over the last
century, a significant energy generation source has been fossil fuel, which results
in greenhouse gas emissions. Reduction in emissions and air pollution can be made
by increasing renewable energy sources such as water, wind, bioenergy, solar, tidal,
hydro, geothermal, and ocean energy.
This chapter is mainly explaining the CFD application and methods in the field of
renewable energy sources. CFD simulations are appropriate and of great importance
for virtual investigation and prediction of renewable energy systems. These CFD
simulations can provide several design iterations for flow around wind turbine, waves,
hydro, solar, and biomass. These CFD simulations can reduce the prototype test and
increase trust in the new renewable sources. The main objective of this chapter is to
describe the various renewable energy sources briefly. CFD simulation methods are
discussed to provide insight into the process used for modeling. There is an attempt to
introduce and review the methods used to address renewable energy sources problem
using CFD simulations.

2 Renewable Energy

Renewable energy is a natural source that can be replenished over time. Renewable
energy sources are more uniformly distributed geographically than fossil fuels. These
systems are independent of fuel supply with clean output. As environmental control
is becoming more stringent, there is an increase in demand for renewable sources.
Renewable energy generation over the years has been shown in Fig. 1 [2]. Over the
years, the major obstacle to the increase in renewable energy use was higher cost than
conventional sources. Due to the increase in innovative methods, the cost is coming
down, shown in Fig. 2 [3].
Some of the primary renewable energy sources currently being used are:
1 Hydropower
2 Biomass power
3 Geothermal power
4 Ocean power (tidal/wave/thermal)
5 Wind power
6 Solar power (solar heating and thermal electric/ photovoltaic).
444 S. Kumar and R. S. Amano

Other renewables refer to geothermal, biomass, waste and tidal. Traditional

biomass is not included

Fig. 1 Renewable energy generation worldwide [2]

Fig. 2 Global electricity costs [3]

2.1 Hydropower

Hydropower is the oldest source of energy used for electricity generation. The
hydropower plant’s mechanism works on flowing water. Change in water eleva-
tion called head determines the available energy of water flows through the penstock
and pushes the turbine for electricity generation. Due to water being the primary
source of hydroelectric power, the plants are made near the source location. Figure 3
[4] shows a traditional hydropower plant setup. Hydropower is a well-established
technology that accounts for global electricity generation. This water energy is either
Approach to Solve Renewable Energy Problems 445

Fig. 3 Hydroelectric power generator [4]

in the form of potential energy (reservoirs) or kinetic energy (rivers). In both cases,
electricity is generated by passing the water through large water turbines.
Most of the conventional hydroelectric facilities include:
1 Run of the river system—the water current’s force applies pressure on the turbine,
including a weir to divert water flow to hydro turbines.
2 Storage system—water is accumulated in the reservoirs created by dams on
streams and rivers and is released through hydro turbines to generate electricity
[4].
Power output from a dam is a product of efficiency ï, the potential energy per
unit volume, and the volume of water flowing per second Q

P = nρgh Q (1)

Power output is dependent on h Q.

Hydro turbines are rotating equipment that converts high-pressure water to
mechanical power. For low head and high flow rate, Kaplan turbines, which are axial
flow, are used. For higher heads, the Francis turbines, which are radial inflow reaction
turbines, are used. Pelton wheels are used for small flow rates where a nozzle is used
to accelerate the high-pressure water stream to the high-velocity jet, impinges on the
wheel turbine’s blades. The overall efficiency of converting waterpower to electricity
is being 80% [5].
The significant advantage of hydropower is low operating costs, a low impact on
the atmosphere, long plant life, and quick response to electricity demands [6]. Some
446 S. Kumar and R. S. Amano

additional benefits are water for irrigation and flood control. The return of investment
for these plants is extended. Some severe social and environmental issues need to be
considered, such as population displacement, sedimentation, water quantity changes,
flooding, and aquatic life. Overall developing hydropower sites can be economically
competitive with fossil fuel plants and being renewable and pollution-free.

2.2 Biomass Power

Biomass is a plant- and animal-derived material such as straw, logs, dung, and crop
residue, which are used directly or indirectly as fuels, and these fuels are often
called biofuels. Biomass is generally used by combustion or by converting it to
biofuels such as bioethanol and biodiesel. Biomass is a carbon-neutral source as the
amount of carbon released during combustion has been previously removed from the
atmosphere during photosynthesis into making plant material.
These are agricultural crop, or they are byproduct which can replace some fossil
fuel consumption. Another form of biomass is animal waste or human waste, where
digesters can produce methane, and the residue of this process can be used as farm
fertilizers.
The essential processes that use the energy of biomass are as follows.
1. Combustion: Woody plants and grasses can be burnt directly to provide process
heat or electric power.
2. Gasification: The process that converts biomass into gaseous fuel with hydrogen,
carbon dioxide, and carbon monoxide is known as gasification. Gasification
is done at high temperatures without combustion with controlled amounts of
oxygen or steam, which conserves the heating value with fuel combustible in
the boiler furnace.
3. Pyrolysis: Thermal decomposition of biomass produces combustible products
in solid, liquid, and gaseous forms.
4. Fermentation: Distillation of carbohydrates produces ethanol liquid fuel used
as a blender in gasoline for motor fuel.
5. Anaerobic digestion: Produces a gaseous mixture of carbon dioxide and
methane.
The heating value of biomass is lost due to easy use, which increases the production
cost. The overall price to the consumer is higher for biofuels, and hence, biomass
energy supply must use unutilized byproduct and low capitalization cost for biomass
conversion to higher-value fuel.
For biomass energy crops except for wood, a year-round steady supply of feed-
stock for energy needs storage post-harvesting, which increases the cost of biomass
fuels. Kumar and Reisel [7] presented a mathematical model for calculating energy
usage and corn ethanol production cost. The model investigated solar and wind energy
as potential energy sources to reduce carbon footprints during ethanol production.
Approach to Solve Renewable Energy Problems 447

2.3 Geothermal Power

Geothermal energy is the earth’s natural heat, tapping hot temperature via steam
at great depth or moderate temperature at shallow depth. The different temperature
ranges of water or steam at different locations can be effectively used for various
applications such as
(a) Hot water or steam from high-temperature geothermal reservoirs can be used
for electric generation
(b) Moderate and low can be used for direct applications such as heating buildings.
There is a high temperature below the earth level. This hot interior has a high
magnitude of energy to be exploited by drilling wells to the depth of 10 km and
pumping high-temperature steam and water effectively for geothermal energy plants.
The most practical and economic sites for geothermal energy are surface temper-
ature, higher and underground water, and steam close to the surface. Figure 4 has
a typical geothermal power plant. Production well supplies warm water and steam
mixture and fed to the separator for separating to steam from liquid—the liquid being
pumped back through the injection well. Steam is provided to a turbine for electricity
generation. The condenser cools the exhaust steam and reinjects it to the ground.
Some of the challenges of geothermal resources can be locating a reservoir of
enough high temperature, which are assessable by drilling. In addition to this, supply
should sustain for a considerable duration, and the resistance to the pump should
be less to make the plant economical. Lately, geothermal heat pumps are utilizing

Fig. 4 Flow of fluid in geothermal power plant

448 S. Kumar and R. S. Amano

the heat storage capacity of underground water or steam. Water from the reservoir
supplies heat to the heat pump and then releases a high amount of heat during the
cold weather.

2.4 Ocean Power (Tidal/wave/thermal)

Tidal
The regular tidal rise and fall of the ocean level can be used to generate electricity.
The tidal rise and fall are due to the gravitational force on the opposite side of the
earth by moon and sun, which produces higher elevation toward and away from the
moon and lower elevation in the middle. Spring tides are due to the full moon, which
is maximum elevation, and the time when solar effects are minimum is called neap
waves.
Tidal power system uses ocean current to drive turbines to generate electricity.
Tidal barrage for energy generation uses a low head turbine and can work most of
the day between high and low tides by using barrage. Sluice gates located along the
barrage control the level of water across the turbine. Sluice gates are closed when the
tides turn in or start to come in, and barrage holds back the rising sea level, creating
a water level difference, which can be used for power generation. A similar reverse
phenomenon is used during low tides to generate electricity called tidal barrage ebb
generation—Fig. 5 [8] shows the tidal barrage power system.
Power from the tidal barrage

ρg Ah 2
Pave = (2)
2T
The advantage of this system is carbon-free energy, and it improves flood protec-
tion. Large tidal barrage construction has some drawbacks, including enormous
capital cost, long construction times and intermittent operations, and long-term
ecological impact.

Fig. 5 Tidal barrage energy

generation [8]
Approach to Solve Renewable Energy Problems 449

Wave: The effect of wind mainly causes the waves on the surface of the sea. The
wave crest experiences air closer and moves faster. From Bernoulli’s theorem, air
pressure is reduced; hence amplitude increases, and waves are generated. As wave
crest collapses, fluid is displaced and forced to rise above the equilibrium level. The
motion of fluid below the water level decays exponentially with depth. The wave
energy moves with the wave, and by using suitable mechanical devices, the energy
in a wave can be converted to mechanical power. The wavelength of waves is the
superposition of different wavelengths; hence, the tool should be robust enough to
withstand random fluctuations of waves.
Incident power per unit width of the wavefront is [9]

gλ
P = 1/4 ρga 2
(3)
2π

Most of the wave power is available when the wind exceeds its average value.
Wave energy at the location is available for a small fraction of time with 15–20%
capacity. There are two types of devices primarily used for extracting wave energy.
(a) Floating bodies anchored in place
(b) structure fixed to the seabed.
By the flow of the waves, parts of the structure will have relative displacement
producing power. These are the power produced by a device that intercepts a fraction
of waves. Some of the wave power devices being used are oscillating water column,
which uses air turbine in a duct above the water surface. The base is being open to
the sea so that water oscillates in a vertical direction. That results in the air above
the water surface to be in phase with the water column-free surface and drives the
air turbine. The turbine used is called the Wells turbine. Figure 6 [10] shows an
oscillating water column.
Submerged devices use the change in pressure below the surface when waves
pass, with wave crest having high pressure and low pressure for wave trough.

Fig. 6 Oscillating water column [10]

450 S. Kumar and R. S. Amano

Floating devices float on the water surface with incident waves. The shapes are
designed such that most wave energy is absorbed with minimum reflection and
transmission.
Thermal
This is the concept of generating mechanical power using the differential temperature
of warm surface water and cold deep water. The warm water is for the heat source,
and cold deep water for cooling to condense working fluid. These are termed ocean
thermal energy conversion (OTEC) plants with small thermodynamics efficiency as
the temperature difference is low. There are mainly two types of plants utilized.
(A) Open cycle: This utilizes water evaporated from the warm stream under low
pressure to supply a turbine discharging to a lower pressure condenser cooled
by cold water.
(B) Closed cycle: This cycle uses working fluid as ammonia for more effective
turbine output use. Evaporator and condenser are more efficient and expensive
for this cycle.
Figure 7 shows the fluid flow in both cycles. A large volume of warm and cold
water is needed with large diameter ducts. Some of the issues with these power
systems are fouling water supply pipe and heat exchanger by a marine organism
and a heat exchanger’s corrosion. The power generated should be higher than the
pumping power. The whole system cost should be kept economical for making this
renewable energy to be competitive.

2.5 Wind Power

Since ancient times, wind energy has been in use for mechanical power. The present-
day wind turbine has the latest technology in design and manufacturing and is mainly
used for electrical power generation. The wind generation source is the sun, which is
absorbed by water and land and heats the surrounding air. A high amount of power
is in the wind, which can solve the power issue, but as the wind source is diffuse, it
is tough to harness this. Wind availability is variable with time and location.
Wind Turbines
(A) Horizontal axis wind turbine (HAWT): A typical HAWT is shown in Fig. 8.
The main components consist of bearing the turbine shaft, gearbox, and gener-
ator enclosed inside the nacelle. The wind turbine blades are mounted to the
shaft and nacelle and are oriented by yaw control. The rotor is upwind of the
tower to avoid blades not getting enough wind—this kind of wind turbine is
predominantly used for power generation.

(B) Vertical axis wind turbine (VAWT): This type of wind turbine’s main rotor
shaft is a traverse of wind with all significant components at the base. One
Approach to Solve Renewable Energy Problems 451

Fig. 7 a Open cycle and,

b closed cycle ocean thermal
power plant

popular design is the Darrieus design, driven by lift forces generated by wind
flow over the aerofoil shaped blades. VAWT is easier to maintain since it does
not need a yaw mechanism with a gearbox and generator at the ground level.
Overall, these are less cost-effective compared to HAWT.

Power of HAWT: The kinetic energy of the wind provides wind power flowing
over a cross-sectional area. An average direction of the wind is

P = 1/2 AρU 3 (4)

452 S. Kumar and R. S. Amano

Fig. 8 Horizontal axis wind turbine

However, not all power can be extracted by a wind turbine because some fraction of
the kinetic energy moves downstream to maintain airflow. This effect of a theoretical
maximum of 59% for extracting power from wind is known as the Betz limit.
Hence, power extracted by the wind turbine is

P = 1/2 AρU 3 C p (5)

where Cp is power coefficient.

The blade’s design is such that the thrust on the turbine is generated and translated
to rotational energy by shaping the wind blades as the aerofoil. The turbine blades are
designed with a twist, which increases going toward the root to maintain the angle
of attack optimum.
Blade width also increases going toward the root of the blade. For turbine blades,
several factors are considered, such as blade strength, material, and fatigue. For
higher power efficiency, a large tip speed ratio of 6–10 is needed.
Wind site investigation for wind turbine installation plays a significant role. Over
a period, excellent and regular measurements are needed to investigate a turbine or
number of turbines, wind farm.
Approach to Solve Renewable Energy Problems 453

Lately, the acceptance of wind warmth is positive because they are a renewable
energy source with zero carbon emission. There are some concerns, such as a visual
threat to birds and space. Overall, the wind power return of investment depends on
capital cost, an ongoing cost, maintenance, operational cost, and the revenue from
the power generated.

2.6 Solar Power

Sun emits electromagnetic radiation, and energy flux per unit area is called irradiation.
Solar irradiance is of two types of beam irradiance, which retains solar direction
and diffuses irradiance, which does not control the direction. The solar radiation
reaching the earth has a wide band of wavelength. The fraction of incoming beam
that is scattered or absorbed depends on the wavelength of incoming light. Scattering
is mostly for low wavelengths, and absorption is for long wavelengths. The average
solar power incident on earth is 1000 Wm−2 , far greater than the current world power
consumption.
Solar energy can be used as a source of power in the form of electricity and
thermal. Photovoltaic solar panels can do the conversion to electricity, and other
ways can be used as thermal applications such as indoor heating of spaces and fluids.
Solar energy is a clean and renewable power source that can be harnessed
anywhere with proper conversion methods. There are several efficient conversion
methods available and will briefly be discussed.
A. Flat plate collector: The most economical form of solar thermal energy collector
is a flat plate collector. It has a thermally conducting collector plate with a
passage that consists of fluid for heat transfer. There is a need for optimum
design to minimize the ratio of cost to heat collection.
B. Focusing collector: The main reason to use focusing solar collector systems
is to increase solar radiation’s intensity on the collector and get solar energy
at high temperatures with a small collector area. Solar irradiation increases
by concentration ratio, which is mirror dimension to image dimension. The
focusing system collects solar energy at high temperatures and can be used in
the power generation’s heat engine cycle.
C. Photovoltaic cells: These are solid-state devices used to generate electricity
from solar energy. The photovoltaic cells are made of n-type and p-type
semiconductor materials doped with another element. This creates donor and
receptor electrons and at the junction of two layers creates an electric poten-
tial difference. Photovoltaic cell efficiencies are rated for a solar irradiance
of 1 KW/m2 under maximum power conditions. Photovoltaic power is direct
current (DC) power and needs to be converted to alternating current. The
maximum efficiency of photovoltaic cells falls in the range of 25–30%.
Some of the other developing technologies based on solar energy are electrochem-
ical cells, solar-driven Stirling engines, and solar chimney.
454 S. Kumar and R. S. Amano

3 Computational Fluid Dynamics

Computational fluid dynamics are a set of methods given to the computer to perform
fluid flow simulations. It involves analyzing various systems, including fluid flow,
heat transfer, multiphase flow, and chemical reaction. High-fidelity CFD models
can be visualized in a computer by using a virtual environment. CFD is presently
used as simulation tools in various industries to influence design at an earlier stage,
saving a lot of time and resources. For any CFD analysis, the most important thing
is to identify and formulate the flow problem in terms of the physics such as 2D or
3D, symmetrical or not, turbulent or laminar, compressible or incompressible, and
transient or steady. This section is a brief outline of the CFD analysis process, and
the attempt is to address the CFD analysis outline.
For all significant CFD solvers, three major elements are
1. Preprocessor
2. Solver
3. Post-processor.
The activities involved in preprocessor ranges from definition of the geometry of
the computational domain, grid generation, selection of physical systems to simu-
lated, fluid property definition, and specifying the boundary conditions. The solution
of all physical quantities is based on the grid generated to capture the physics accu-
rately. The accuracy can be captured by having a finer grid in a high gradient area
compared to a low gradient area. The flow problems are defined at nodes inside each
grid, and hence, the accuracy is highly dependent on grids.
Material properties of the fluid and selecting a proper turbulence model for physics
are essential parameters for analysis.
Solvers provide a solution to the underlying equations by approximating flow
variables by simple functions, discretization by substituting the approximate function
into governing flow equation and solving the algebraic equations. The major CFD
solvers use a finite volume approach for solving the problem. There is a relationship
between the numerical algorithm and physical conservation in the finite volume
approach.
Post-processing is understanding and visualizing the physical properties of the
simulation results obtained from numerical data. Post-processing of solution involves
data visualization of various flow fields such as vector plots, domain geometry, grid
visualization, surface plots, and streamlines. There are options for creating animation
for dynamic results.
Approach to Solve Renewable Energy Problems 455

4 Application of CFD in Renewable Energy Source

4.1 Hydropower CFD

Hydropower generation consists of a turbine, intake valve, spillway, dam, and the
entire unit. CFD tools can be applied to an individual component of a power gener-
ation unit or a whole unit for rapid and cost-effective improvements. This simula-
tion can help understand the complex working mechanism to design new facilities,
upgrade and maintain the existing one.
Water turbine applications are mainly divided into two.
1. Impulse turbines which are for low flow rate and high heads such as Pelton
turbine. These turbines convert the head of water to kinetic energy, which is
discharged to a runner’s curved bucket to get useful work.
2. Reaction turbines such as Francis, propeller, Kaplan, and Deriaz, which are
medium to high flow rates which are below 450 m head. These get energy by
converting accelerating water flow in a runner with pressure drop. These turbines
can be arranged horizontally or vertically.
CFD is being implemented in predicting complex flow in hydraulic machines for
continuous improvement in all aspects of design such as efficiency, power output,
and head. It helps to understand the detailed flow pattern for customized projects,
reducing the cavitation and improving efficiency. The main reason for discussing
different types of water turbines is to identify the CFD method. The first step is to
identify the type of turbine being used for power generation.
Steady-State Reynolds Averaged Navier–Stokes (RANS) model
The steady flow RANS model application is one of the dominant tools for water
turbine flow design and analysis. This analysis takes into account for loss effect,
boundary layer effects, and overall component flows effect. The analysis of the draft
tube, diffuser, and pump turbine gives a more realistic flow analysis.
Keck et al. [11] discussed the basic principle of hydraulic turbines with particular
emphasis on using CFD as a tool. They used 3D Euler and 3D Navier–Stokes for
Francis runner simulation, and the CFD domain included the whole Francis turbine.
Figure 9 shows the Francis turbine grid used for simulation. They compared measured
and numerical results for circumferential and meridional velocity components at the
runner outlet and demonstrated computational results’ reliability. They concluded
that details of flow separation, loss sources, and loss distribution in components,
matching components at design and off the system, and low-pressure levels with
risk of cavitation can be modeled with CFD. The significant challenges were grid
generation and grid modification. Simulation of vertical Kaplan turbine was done,
which had optimized guide vane and runner blade design for maximum efficiency
[12]. Based on these simulations, flow and pressure coefficient efficiency shell have
been calculated with similar specific speeds (Fig. 10).
456 S. Kumar and R. S. Amano

Fig. 9 Grid used in Francis turbine [11]

Fig. 10 Vertical Kaplan turbine [12]

Manitoba Hydro used CFD solver to get water flow behavior around construction
dams and permanent structures. They used it to guide spillway design, intake and
tailrace channel design, and river management during construction. The results’ main
output is discharge capacity, water levels, velocities, and flow patterns at various
points on the construction timeline and different spillway gates [13].
Approach to Solve Renewable Energy Problems 457

Moving reference frame

The rotating domain’s inclusion and development enhanced the hydraulic machines’
CFD prediction capability. In this technique, an interface between the rotating and
stationary domain was implemented, effectively analyzing all turbine components.
This method can be used with steady-state simulations as well. Keck et al. [14, 15]
used the full-stage simulation method to calculate the first numerically predicted hill
chart, which used rotating and non-rotating components. These simulation results
are validated with experiment and have a good agreement. They concluded that an
unsteady rotor–stator interface has a more realistic flow and disturbance transported.
Additional development in simulation technique: The advancement in simulation
computing resources has given the luxury of implementing other physics to accurately
predict the flow. The simulation physics includes multiphase, cavitation, turbulence,
fluid–structure coupling, and unsteady flow. Tilahun et al. [16] discussed in detail
the design, modeling, and performance analysis of a Pelton turbine. They studied
how the size of the turbine affects the efficiency of the turbine. Some of the physical
assumptions were considered for simulation for reducing computational expense,
such as no casing, symmetrical case, single jet, no hub, periodic torque, and scaling
down. For simulation, the multiphysics model of liquid and air, domain interface
for rotating and stationary as fluid-fluid and K-ω SST turbulence model was used.
The flow visualization study provides insights into the reasons for efficiency and low
efficiency in reduced size of the turbine are caused by a large amount of water leaving
bucket through the lip. Akinyemi et al. [17] studied electrical generation capacity by
evaluating paddle wheel with different speed, and the effect of side and the bottom
fin was evaluated. Curved paddle wheels had higher efficiency than straight paddle
wheels.
Some of the critical physical phenomenon to be modeled with the unsteady flow
is dynamic loading on structural components, which produces.
1. Draft tube vortex: In the draft tube, a rotating pressure vortex is created with
low pressure in the center, which can create damage, and hence some additional
measures such as aeration through hub and fins are used. High-fidelity CFD
simulations can capture these complex phenomena of rolling up vortices at the
shear layer between the outer and inner regions. Cavitation is caused when the
static pressure of the vortex core drops below vapor pressure. For these types
of simulations, DES and LES turbulence model approach needs to be adopted.
2. Von Karman vortex at trailing edge of stay vane/guide vane and runner blade:
Lockey et al. [18] investigated CFD analysis of von Karman vortex shedding.
In the first step, the steady-state analysis was carried out to understand velocity
distribution at the stay vane’s (stationary) trailing edge to confirm shredding
frequency. In the second step, advanced time-dependent CFD analysis of vortex
shedding was analyzed. Von Karman shedding can be used as an additional
numerical parameter for accuracy.
3. Guide vane and runner blade interactions-Fluid structure Interaction: One of the
essential aspects of turbine design is to analyze the structure’s response to the
dynamic load of the flow field. Pressure amplitude is determined from the CFD
458 S. Kumar and R. S. Amano

flow field for the structure’s natural frequency, and then FEA analysis is done
to determine the structure’s response. Sick et al. [19] investigated the pressure
pulsation from CFD to the structure of the turbine. They presented the natural
frequency of a Francis runner, including added mass effect due to surrounding
water. This coupling of fluid mechanics with structure mechanics is becoming
more relevant for robust and efficient design.
Application of CFD in the hydraulic turbine is becoming more critical, instru-
mental in designing and understanding complex flow patterns. The use of CFD will
enable us to study highly complex multiphysics and accurately predict the turbine’s
flow. Overall, CFD usage improves and reduces the time for designing turbines.

4.2 Biomass Power

Computational fluid dynamics is becoming an essential tool for developing and

designing new processes and equipment to convert biofuel into fuels and improve
component performance. Biomass is becoming cost-effective and is increasingly
becoming an important renewable source with improved performance using CFD.
Biomass thermochemical conversion refers to gasification of fuel, pyrolysis,
carbonization for heat energy. The main difference among these methods is feeding
properties, oxidizer, temperature, and time. In this section, the CFD application
for biomass thermochemical process is discussed to optimize design and operation.
Wang et al. [20] examined the fundamentals involved in developing a CFD solution.
This section briefly outlines the approach taken by Wang et al.
Gasification and pyrolysis
This process is the thermal process of producing solid (charcoal), liquid (bio-oil),
and gas in the absence of air. The CFD process could provide flow and temperature
patterns and concentration details to improve the process.
Fletcher et al. [21] discussed a CFD model to simulate an entrained flow biomass
gasifier’s flow and reaction. Biomass in this model is simulated using the Lagrangian
approach as it enters the gasifier, and it releases it is volatile and undergoes gasifi-
cation. Figure 11 shows the geometry used for the CFD solution. Lower inlets were
used for the inlet of biomass and air and upper inlet for the steam inlet. The initial
condition lower chamber was with 1000 K air to start the devolatilization for parti-
cles. The upper chamber has 30% steam at 400 K. The biomass was assumed to
be dry when it enters the gasifier, with water initially present in the biomass being
injected as steam with heated air. Equations for the concentration of CH4 , H2 , CO,
CO2 , H2 O, O2 , and reaction with CO, CO2 , and H2 O are modeled. Gas composition
details with temperature and velocity at different scenarios can be evaluated. They
suggested that simulation can examine the effect of gasifier height and steam flux in
the upper inlet.
Approach to Solve Renewable Energy Problems 459

Fig. 11 Geometry of
gasifier [21]

Sharma [22] developed a model for the reactive, porous bed of biomass gasifier
to simulate pressure drop, the temperature profile in the bed, and flow rates. He
concluded temperature profile being susceptible to gas flow rate and heat generation
in the oxidation zone. In contrast, high bed temperature, gas flow rate, and reduction in
feedstock particle size increase in pressure drop through the gasifier. The downstream
zones’ temperature is more sensitive to any change in heat generation in the bed
compared to the upstream zone. Figure 12 shows the schematic and sectional view
of the gasifier used for modeling. The packed bed in the gasifier was modeled based on
a saturated porous medium in which fluid flow increases inflow due to the particles’
gasification constituting the beds. This simulation predicted temperature profile and
pressure drop for varying operating parameters such as gas flow rate, heat generation
in oxidation zone, the moisture content in the feedstock, particle size, and input air
temperature. Luo et al. [23] modeled wood pyrolysis in a fluidized bed reactor, and
they simulated the main operation parameters for product distribution.
Biomass combustion
One of the most extensive applications of CFD models has been to power station
boiler and furnaces. These simulations can be used for qualitative and quantitative
analysis of the biomass system to improve the current process.
Kaer et al. [24] discussed CFD analysis of straw-fired grate boiler. Based on
CFD analysis, they suggested poor mixing in the furnace is the main reason for high
amounts of unburnt carbon in the fly ash. Initial boundary conditions were incorpo-
rated using stand-alone in-house code. The boiler is comprised of 395 air nozzles
460 S. Kumar and R. S. Amano

Flow Direction

Sectional view of gasifier (22)

Fig. 12 Schematic diagram of a gasifier system [22]

that influence flow and combustion pattern. The platen secondary and tertiary super-
heater in the first and second passes were modeled as slabs, and primary superheater
tube banks in the third pass were accounted for using source terms in the third pass
were accounted for using source terms in momentum and energy equations to simu-
late heat transfer and pressure drop. Figure 13 shows the boiler section used for
simulation. In these kinds of boiler, thermochemical fuel simulation takes place on
the grate. The heat released by heterogeneous combustion of entrained particles is
insignificant, and only gas combustion is considered. Gas combustion was modeled
using the eddy breakup model and two-step reaction mechanism with CO as the
intermediate species.
Kaer et al. [25] discussed the biomass-fired ash problem. They investigated
deposits on furnace walls and superheater tubes, which includes impacts on heat
transfer rates. The model was validated with straw-fired boiler and predicted the
Approach to Solve Renewable Energy Problems 461

Flow Direction

Schematic of fuel bed used for numerical model (24)

Fig. 13 Schematic of lower section of grate-fired boiler with operating conditions [24]
462 S. Kumar and R. S. Amano

Fig. 14 Particle trajectories of charcoal burnout [26]

correct operational trends. In this analysis deposit, formation models were included
got a comprehensive investigation.
Syred et al. [26] worked on fragmentation models for solid fuel combustion using
CFD. They suggested that existing models for solid fuel combustion are robust for
initial flow field aerodynamics, temperature but are not accurate when followed
through the complete system. Using enhanced retention of burning particles in
centrifugal force fields can be resolved by changes in particle size distribution and
fragmentation as particle gasify or burn. In this study, single-step fragmentation was
used, which needs to be refined for more accurate analysis. A proper understanding
of the path of a burning particle in coal-burning plants is required. Typical temper-
atures are 1300 K with a 10-step reaction scheme using finite rate eddy dissipation
models. Figure 14 shows the fragmentation application and particle transport.
Biogas Fermentation
By using CFD, improving fermentation material fluid mixing for biogas in anaerobic
digesters can provide distribution and visualization of flow patterns. This virtual
method is gaining importance because it allows for an optimal design of digester
shape, mixing pattern, and different fluid parameters and methods.
The biogas simulation can be having low Reynolds number turbulence. Multiphase
flow simulations are commonly used for biogas, and mostly Eulerian–Lagrangian,
and Euler–Euler are employed. The biogas reforming reactor was modeled in CFD,
which included reaction kinetics for exhaust gas reforming biogas to synthetic
gas [27]. The biogas reactor was modeled as an isothermal heterogeneous reactor
consisting of two zones connected at the interface, heat zone, and monolithic zone.
Inside monolithic zone reforming surface reaction occurs with the presence of
rhodium catalyst.
Approach to Solve Renewable Energy Problems 463

Dapelo et al. [28] developed the Euler–Lagrangian model to simulate gas mixing
of sludge for anaerobic digestion. Momentum transfer from bubbles to liquid drives
the fluid motion, and they are modeled together. The diameter of vertical columns is
small, which reduces the bubble rise. The focus was to resolve flow patterns away
from bubble plume.
Wang et al. [29] discussed in detail the CFD applications in biogas fermentation to
concentrate mainly on analyzing flow patterns during mixing. The mixing schemes
of biogas are mixing medium, mixing parameters, and digester shape. The mixing
parameters are kinetic energy diffusion, phase, and temperature distribution. Mixing
mediums are solid, liquid, and gas divided into mechanical mixing, hydraulic mixing,
and pneumatic mixing. The shape of the digester has a critical influence on the
flow pattern. Many CFD shows that these parameters can change the flow pattern
significantly.
Biofuels: CFD has been utilized for the modeling and improvement of equipment
for the biofuel production process. Biodiesel is produced through the transesteri-
fication reaction, where triglyceride feedstock reacts with alcohol to produce fatty
acid methyl esters (FAME) with a catalyst that generates biodiesel glycerol. For
homogeneous catalysis, stirred tank reactor along with tubular reactors is used. For
heterogeneous catalysis, the trickle bed reactor is used. [30].
Adeyemi et al. [31] developed numerical modeling of alkali catalyzed waste
cooking oil transesterification in a stirred tank reactor. They implemented the
Reynolds average Navier–Stokes equation, a 3D model with rate constant and eddy
dissipation model for turbulence. Multiple reference frame model was chosen to
describe the rotational motion of impeller.
Boer et al. [32] used CFD to encapsulate the liquid–liquid flow phenomenon in
reaction. They investigated the visualization results and studied different simulation
settings. Figure 15 shows the simulation setup along with the boundary condition
for the CFD of biodiesel. CFD has been applied to modeling and improvement
of equipment for the biofuels production process. For current CFD models, flow,
turbulence, energy, mass transfer, and chemical reactions can be implemented.

4.3 Geothermal Power

Geothermal wells’ potential is being explored all around for new techniques and
systems. The most preferred means of extracting the thermal energy contained
in geothermal wells are electrical and mechanical. Experiments are costly for
the geothermal power production process. CFD modeling can be used effectively
to provide an initial preliminary understanding of the whole process. Predicting
behavior and production performance of geothermal wells can be done for opti-
mized power production. This section outlines some of the research CFD methods
for various aspects of the geothermal system.
Huang et al. [33] studied the wellbore phenomenon by analyzing flow and heat
transfer conditions in the well. Wellbores in enhanced geothermal systems extend
464 S. Kumar and R. S. Amano

Fig. 15 Simulation setting and boundary conditions [32]

from the ground surface, providing large heat transfer areas between the flowing fluid
and the surrounding formations. They simulated unsteady fluid flow within a vertical
injection and production well with heat transfer. Figure 16 shows the domain used
for simulation. The fractured reservoir was located at some distance from the ground
surface, with some distance kept between two wellbores with a specified diameter.
Details of the CFD setting include rock formation enclosing the reservoir set to be
the impermeable solid phase for heat transfer conduction. The fractured reservoir is
a fluid region, and velocity inlet and pressure outlet boundary conditions are applied
to the inlet of injection and production well. Initial temperature increases with depth
with a constant gradient. The time-evolving simulation shows that almost the entire
pipe is hydro fully developed but not thermally fully developed in the initial phase.
The upper section’s heat transfer direction is opposite to that in the lower section due
to the opposite gradient vector.
Renaud et al. [34] studied 2D axisymmetric, single-phase, and steady state of the
reservoir and wellbore heat extraction from a geothermal well near a magmatic
chamber using a simplified geological assumption. The model included many
assumptions, and with current computing power 3D, multiphysics simulation can
be done. They concluded that a decrease in mass flow rate production drives more
heat from the reservoir but cools down faster when flowing up to the surface.
Jiang et al. [35] studied the heat exchange process in a geothermal reservoir.
They considered heat reservoirs as porous medium and domain, including an open
channel for injection and production wells. Figure 17 shows the three different region
schematic diagram. In the simulation, a single domain with multiple subregions is
Approach to Solve Renewable Energy Problems 465

Fig. 16 Schematic of simulated domain [33]

Fig. 17 Schematic of
multiple regions; region 1:
porous heat reservoir; region
2: doublet wall boreholes;
region 3: solid rock
enclosing heat reservoir [35]

considered. Region 1 is viewed as the heat reservoir, region 2 the rock enclosing the
heat reservoir, and region 3 the injection and production wells.
In addition to CFD for the geothermal system, some component level CFD is also
essential for increasing the plant’s efficiency. Mubarok et al. [36] studied numerical
simulation of two-phase geothermal fluid flows in the production pipeline using
a different flow meter. These CFD models were instrumental in developing the
predicting capability of fluid flow behavior, including net pressure drop, velocity,
turbulent kinetic energy, mass flow rate, temperature, enthalpy, and entropy. These
466 S. Kumar and R. S. Amano

results are useful for selecting the appropriate pressure flow meter to measure
geothermal wells’ output and power potential.
Yousefi et al. [37] studied a geothermal reservoir that was modeled as a bottom
heated square box filled with the water-CO2 mixture. Different parameter effect
on overall performance was investigated. They concluded that if CO2 percentage
increases, then there is a substantial increase in convective heat transfer. They
suggested including multiphase analysis for the accurate prediction of heat transfer.
Overall, the increased use of CFD in the geothermal plant for system and compo-
nent level can increase plant efficiency. More research in this field is needed to
enhance CFD usage.

4.4 Ocean Power (Tidal/wave/thermal)

Ocean power can be a significant contributor to a satisfying increase in energy

demands over the years. As for other sources, CFD can be a useful tool for early stage
optimization and fine-tuning the device design before the final physical prototype.
Windt et al. [38] discussed in detail wave energy converters (WECs) and the use
of high-fidelity numerical modeling, and the best practices guidelines for applying
CFD in the field of wave energy.
For numerical simulation of wave energy converters, device type, device motion,
and the number of devices are important aspects. Based on the operating principle,
wave energy converters are classified as point absorber (PA), absorbs power from the
wavefield, terminators, aligned parallel to incident wavefronts, being geometrically
wide to intercept and stop the incoming wave. Attenuators are placed perpendicular
to the incoming tide and derive power from the wave’s phase difference. Oscil-
lating wave surge converters extract energy from horizontal wave motion. Pressure
differential devices are placed below the ocean surface and utilize the change in pres-
sure caused by passing waves to pump fluid through the system. Figure 18 shows
the primary wave energy converters being used. Device motion also needs to be
considered, which is an essential factor for model preparation.

Fig. 18 Different WEC categories, a PA, b terminator, c attenuator, d OWSC, e OWC, and f pressure
differential [35]
Approach to Solve Renewable Energy Problems 467

Tatum et al. [39] modeled tidal steam turbine, which included profiled flow and
gravity waves. A full-scale tidal stream turbine was transiently modeled in and oper-
ating at maximum power point with both in and out phase waves with its period of
rotation. They concluded that a naturally varying sea state might yield a soothing
effect in this turbine response. For this analysis, a rectangular cross section was
modeled with a turbine situated below the free surface. The velocity profile and
the wave characteristic were included in the CFD model. In this analysis, k-ω SST
Reynolds averaged Navier–Stokes (RANS) equation was used with transient and free
surface modeling.
Karthikeyan et al. [40] compared marine current turbines with different pitch
angles. They suggested 200 hub pitch angles for simulation due to the wider operating
and peak value of Cp and CFD for various free stream velocities. Jafaria et al. [41]
worked on a wave energy generator CFD analysis. In this research wave, an energy
converter was simulated inside a wave tank, including water. This analysis included
RANS equations coupled with the volume of fluid (VOF) model to generate three-
dimensional numerical linear propagating waves and solve the fluid flow. In this
simulation, the converter’s upper buoy was designed to move upward by ocean waves
developed.
Li et al. [42] studied a wave energy generating system. A 6 DOF analysis technique
is applied to CFD analysis on a floating body with incoming waves. A spring constant
is adopted to control multifloating bodies’ motion and calculate the total average
power absorption. The results showed average power absorption varies with both
positions of the floating body and wavelength.
Some of the major points to consider during the ocean wave CFD analysis are
its definition and applications, including domain definition, load, and performance
studies for results. Fidelity of the subsystem needs to define for accuracy of the results.
The free surface interface needs to be defined as the most accurate and efficient way.
Depending on the computational cost and applications, a turbulence model needs
to be chosen for 3D fluctuation of flow qualities. During these simulations, one of
the important approaches is to consider wave generation and absorption, such as
relaxation method, static boundary method, dynamic method, mass source, impulse
source, and numerical beach method. The body representation method, which uses
boundary fitted mesh with mesh distortion, can be implemented. For more large
amplitude displacement, multi-DOF body motion methods such as overset mesh can
be utilized. Validation of such CFD methods is vital for confidence in the data.
Schmitt et al. [43] discussed challenges and advantages in the application of RANS
CFD methods in the design process of wave energy converters. They concluded that
the accuracy and repeatability of CFD simulations are compared to experimental
testing. An essential aspect of the CFD method is mesh adaptation of surface motion.
468 S. Kumar and R. S. Amano

4.5 Wind Power

Wind energy is one of the most important renewable sources of energy. The compu-
tational method has been extensively used over the years to improve wind turbine
performance. To reduce the uncertain behavior of forces and to reduce the design
risk application of CFD methods and its validation is crucial. Extensive research
and effort have been made to understand the aerodynamics of wind turbine and the
pressure dominated flow in the surrounding. The present wind energy technology
has been advanced with a significant contribution of aerodynamic research.
To accurately numerical prediction of the wind turbine, there is a need to under-
stand wind turbines’ aerodynamics. Vermeer et al. [44] discussed the aerodynamics
of wind turbines in detail and regarding wakes made a distinct division into the near
and far region. Near the trail, regions are area one rotor diameter behind the rotor.
In this region, the rotor’s presence is apparent by many blades, blade aerodynamics,
stalled flow, tip vortices, and 3-D effects. Vortices generated at tip and root have sharp
gradients in the velocity and turbulence intensity. Momentum and energy from flow
provide power to the turbine, which has an axial pressure gradient and a decrease in
axial velocity.
For this region, performance and physical processes of power extraction are
crucial. This region is considered uniform, steady, and parallel flow. The important
aerodynamics aspect being wind shear, rotor-tower interaction, yawed conditions,
and stall.
For the far wake region, the focus is more on wind farm situation and wake
models, wake interference and turbulence models. These kinds of analyses are more
focused on mutual influence when placed in clusters. The following wind turbine
will experience lower velocity with high turbulence intensity. This unsteady effect
and lower velocity decrease power.
Sanderse et al. [45] discussed six different methods used for wind turbine analysis.
These methods are
1. Kinematic method: This method is based on an analytical approach that uses
the far wake’s self-similar nature to obtain expressions for velocity deficit and
turbulence intensity.
2. Blade element momentum theory (BEM): This method uses momentum balance
on rotating stream tube through turbine and forces generated by aerofoil lift
and drags at various sections along the blade. This can be used to get the
aerodynamics blade characteristic.
3. The vortex lattice method is a numerical method used for modeling the thin
lifting surface of discrete vortices and is used at the preliminary stage of design.
The influence of thickness and viscosity is neglected.
4. The panel method is a numerical scheme for solving inviscid, incompressible,
and irrotational flow. Blade geometry and viscous effects can be included with
boundary layer code.
5. Generalized actuator disk method: In this method, rotor is represented as an
actuator disk or actuator line. The disk is represented by a porous disk that
Approach to Solve Renewable Energy Problems 469

allows the flow to pass through the rotor. The actuator disk exerts a force on
the flow, which is added to the momentum equations, and turbulence sources
are also included, which are actuator disk, actuator line, and actuator surface
model.
6. Direct method: This method includes all the CFD methods used for the full
modeling of the rotor. With the development of computational resources, a new
approach to predict wind turbine wake using high-fidelity turbulent model is
made.
The fully resolved rotor or direct method includes the rotor’s exact geometry
and the sliding mesh method for the rotor region. For accurate prediction of the
boundary layer, fine resolution of mesh is required near the blade surface and wake
region. Initially, steady-state simulation can be run using a moving reference frame
model, which uses motion sources from the rotating part, but the mesh is static. This
can be used for accurate transient solutions.
Inlet boundary conditions need to be accurate for wind turbine wakes, and hence
sheared velocity profile, the anisotropy of turbulence, and the inflow’s stationary
nature are essential. Some of the methods are for RANS simulations prescribing
velocity components and turbulence quantity. For LES, unsteady inflow data are
necessary, two different types of methods exist: synthesized inlet methods and
precursor simulation methods. A turbulent inflow profile is used close to the rotor to
prevent turbulent fluctuations decay. The standard method for the LES and RANS
approach is to use a wall function to use wall friction [45]. For upper boundary
symmetry condition can be used, the streamwise direction can use periodic condition.
Kalvig et al. [46] studied a wind turbine’s performance with an actuator disk,
actuator line, and a fully resolved rotor. The CFD results were validated with wind
turbine experimental results and concluded that a fully resolved rotor mesh for direct
method rotor produces the most accurate results. Figure 18 is the wind tunnel test
section. Figure 19 shows the mesh used in the simulation with a turbine rotor in focus
(Fig. 20).
Sumner et al. [47] reviewed RANS-based models and LES-based models for
complex flow. They studied by dividing the simulation based on a length scale smaller
than the rotor diameter and the other at a larger one. Tachos et al. [48] discussed
aerodynamics characteristic of the NERL phase II rotor, a horizontal axis downwind
wind turbine rotor, used the RANS turbulence model, and concluded that the CFD
could predict the aerodynamics of wind turbine rotor. Figure 21 shows the periodic
domain considered for simulation.
For wind turbine aerodynamics, CFD simulations can reproduce results compa-
rable to test effects and are currently being actively used for aerodynamic analysis.
The major challenge is the modeling of turbulence and flows separation. For all prac-
tical purposes, RANS-based CFD methods are performing better than other analytical
methods. LES-based CFD methods are eventually going to be actively used for wind
aerodynamical use.
With the current success and more confidence with simulation tools, there has
been an increase in using it in the design stage. Siemens’s general-purpose analysis
470 S. Kumar and R. S. Amano

Fig. 19 Model wind turbine in the wind tunnel [46]

Fig. 20 Mesh around the wind turbine and blade [46]

tools provide airflow around the turbine blade and give a fluid and structural link. This
kind of analysis can be effective for a quick change in geometry with automatic CFD.
German company Windgiant used CFD to design a multibladed fan for a residential
area. Green Energy Technologies used CFD for the development of the Wind Cube
turbine system for an urban area. Vestas wind turbine manufacturers used automated
methodology to predict the aerodynamics of the blade [49].

4.6 Solar Power

Solar energy is the cleanest and accessible electricity sources in the area with high
solar radiation. There is a need for developing new processes and methods to get
innovative and cost-effective solar panels. Simulation methods can be used effectively
Approach to Solve Renewable Energy Problems 471

Fig. 21 Periodic domain

used for simulation [48]

throughout the solar energy industry. In this section, some of the CFD applications
in the solar industry are discussed.
Solar still and desalination: Desalination of water is done with thermal and
membrane methods. One of the thermal desalination methods uses solar energy as
a heat source, and evaporated water, after colliding with cold glass, is converted to
pure water.
Kharea et al. [50] studied a multiphase 3D CFD model of simple solar stills.
Solar stills by Kharea et al. It was a transient simulation with results validated with
experimental data. Figure 22 shows a solar still schematic diagram, and a similar
geometry was used for simulation. They concluded that high solar intensity results in
quick evaporation–condensation rate and hence higher distillate, and the productivity
decreases with an increase in water depth.
In this simulation, the volume of fraction (VOF) model can be used for air and
water with long-run hours with higher time steps. The boundary condition for glass,
bottom, and collecting surface can be assumed constant temperature. Solar intensity
can be based on absorption factor and emissivity of glass, water, and bottom. The
sidewall’s heat transfer coefficient needs to maintain constant and no-slip wall condi-
tion for the liquid phase and free slip for the gas phase—some simulation assumptions
like constant thermophysical properties of aluminum, glass, and air (Fig. 22).
This simulation with modified physics to capture physical phenomenon can be
used as a powerful tool for designing parametric analysis of solar stills.
Solar collector: Solar electric heating systems use heat collector, which uses solar
energy to generate electricity. In this kind of system, a parabolic linear collector is
used to heat and increase the working fluid temperature. A working fluid pile lines
pass and directly receive heat from solar radiation and are reflected from parabolic
plates. Figure 23 shows a solar collector.
472 S. Kumar and R. S. Amano

Fig. 22 Energy flow in solar still [50]

Fig. 23 Energy flow in solar still [51]

Manjunath et al. [51] studied solar air heater CFD analysis. These air heaters have
a lower thermal performance level due to high heat losses and poor air heat transfer
characteristics. They reviewed thermal performance enhancer by using turbulence
enhancers such as corrugations, vortex generators. Gawande et al. [52] studied solar
air heaters used different approaches to investigate a solar air heater’s thermal perfor-
mance. They reviewed different articles for the type of computational domain (2d/3d),
turbulence model used, type of commercial software, and range of operating and
geometrical parameters used. Figure 24 shows a conventional solar air heater.
For CFD analysis, defining material properties, selecting an appropriate physical
model and turbulence model, and proper boundary conditions at all boundary zones
are essential for this kind of analysis.
Approach to Solve Renewable Energy Problems 473

Fig. 24 Conventional solar air heater [53]

Avila-Marin et al. [53] studied modeling strategies for porous structure as a solar
receiver in a central receiver system. They discussed central receiver systems, volu-
metric absorbers, which work at a higher temperature and concentration factor. Volu-
metric absorbers are porous structures where the air is heated directly by a porous
matrix. They concluded that the porous absorber simulation’s most accurate strategy
is direct simulation or discrete ordinate or Monte Carlo methods for the radiative
heat transfer.
Once the solar panel is mounted on a roof, it will have to withstand wind loading
from all directions. To understand the sunroof panel’s force to withstand the wind
load, a safe, reliable CFD simulation can help. When CFD results are available, then
solar panel mounts can be designed to withstand those loads.

5 Conclusions

This chapter’s main objective was to discuss the importance of renewable energy
sources and CFD application in those fields. All renewable sources are briefly
discussed to provide background information. CFD is becoming an essential tool
in all aspects of all energy sectors’ processes and equipment. With the increase in
computing resources and CFD algorithms, complex phenomena are achieved by
simulation. Over the years, the larger spatial discretization and increased resolution
in high physics gradient can be achieved. Turbulence modeling is the center of most
CFD simulations; with the current trend RANS modeling, it is used predominantly,
but LES is also significantly implemented.
This chapter briefly discussed the CFD methodology for accurate flow and physics
prediction to reduce the design cycle and prototype costs. Global warming has put
renewable energy sources in the forefront, and a sooner energy system transition is on
the horizon. CFD-supported research in this field will continue to grow and expand
for an innovative solution. The critical research and reviews were discussed in this
474 S. Kumar and R. S. Amano

chapter to enhance the understanding of CFD application. Different methodologies

were discussed, which will be helpful for CFD applications.
Fuel cells are one of the renewable sources and were not discussed in this chapter.
In fuel cells, fluid flow, heat transfer, electrochemistry, reactants oxidation, and
reduction rates are computed. CFD will be instrumental in future developments and
analysis of fuel cells and all other renewable energy sources.

References

1. https://www.eia.gov/energyexplained/renewable-sources
2. https://ourworldindata.org/renewable-energy
3. https://www.utilitydive.com/news/renewable-energy-prices-keep-falling-when-do-they-bot
tom-out/555822/
4. https://www.eia.gov/energyexplained/hydropower/
5. Bay JA, Golomb DA, Energy and the environment, 1st Edition, Oxford University Press, New
York, 2002
6. https://www.mr-cfd.com/services/cfd-consulting/renewable-energy-mr-cfd-cfd-analysis-con
sultation-training-simulation/
7. Kumar S, Reisel JR (2010) Modelling of energy usage for the refining of ethanol from corn.
Int J Sustain Energy
8. https://www.alternative-energy-tutorials.com/tidal-energy/tidal-barrage.html
9. Andrew J, Jelley N (2007) Energy science: principle, technology, and impacts 1st Edition,
Oxford University Press, New York, 2007
10. Vertechy R, Fontana M, Rosati Papini GP, Bergamasco M (2016) Oscillating-water-column
wave-energy-converter based on dielectric elastomer generator. In: Proceedings of SPIE-the
international society of optical engineering
11. Drtina P, Sallaberger M (1999) Hydraulic turbine- basic principles and state-of-the-art
computational fluid dynamics applications. J Mech Eng Sci 213:85–102
12. http://www.hfm.tugraz.at/en/references/turbine/development-of-hydraulics-fors-a-vertical-
kaplan-turbine.html
13. Sydor KM, Pamela J (2010) Waterman engineering and design: the value of CFD modeling in
designing a hydro plant. Renew Energy World
14. Keck H, Drtina P, Sick M (1996) Numerical hill chart prediction by means of CFD stage
simulation for a complete Francis turbine. In: Proceedings of the 18 IAHR symposium, Valencia
15. Keck H, Sick M (2008) Thirty years of numerical flow simulation in hydraulic turbomachines.
Acta Mech 201:211–229
16. Nigussie T, Engeda A, Dribssa E (2017) Design, modeling, and CFD analysis of a micro hydro
pelton turbine runner: for the case of selected site in Ethiopia. Int J Rotating Mach 2017:17.
(Article ID 3030217)
17. Akinyemi OS, Liu Y (2015) CFD modeling and simulation of a hydropower system in
generating clean electricity from water flow. Int J Energy Environ Eng 6:357–366
18. Lockey K, Keller M, Sick M, Gehrer A (2006) Flow induced vibrations at stay vanes: experience
at site and CFD simulation of von Kármán vortex shedding. Hydro2006, Porto Carras, Greece
19. Sick M, Lais S, Weiss TH (2007) Numerical prediction of flow induced dynamic load in water
turbines: recent developments and results. In: Proceedings of Hydro2007, Granada, Spain
20. Wang Y, Yan L (2008) CFD studies on biomass thermochemical conversion. Int J Mol Sci
9:1108–1130
21. Fletcher DF, Haynes BS, Christo FC, Joseph SD (2000) A CFD based combustion model of
an entrained flow biomass gasifier. Appl Math Model 24(3):165–182
Approach to Solve Renewable Energy Problems 475

22. Sharma AK (2007) Modeling fluid and heat transport in the reactive, porous bed of downdraft
(biomass) gasifier. Int J Heat Fluid Flow 28:1518–1530
23. Luo ZY, Wang SR, Cen KF (2005) A model of wood flash pyrolysis in fluidized bed reactor.
Renew Energy 30(3):377–392
24. Kaer SK (2004) Numerical modelling of a straw-fired grate boiler. Fuel 83(9):1183–1190
25. Kaer SK, Rosendahl LA, Baxter LL (2004) Extending the capability of CFD codes to assess
ash related problems in biomass fired boilers. In: Division of fuel chemistry, no 12, Proceedings
of 227th ACS annual meeting¸ Anaheim California, Mar 28–Apr 1, 2004
26. Syred N, Kurniawan K, Grifths T, Gralton T, Ray R (2007) Development of fragmentation
models for solid fuel combustion and gasification as subroutines for inclusion in CFD codes.
Fuel 86(14):2221–2231
27. Hamedi MR, Tsolakis A, Lau CS (2014) Biogas upgrading for on-board hydrogen production:
reforming process CFD modelling. Int J Hydrog Energy 39(24):12532–12540
28. Dapelo D, Alberini F Bridgeman J (2015) Euler-Lagrange CFD modelling of unconfined gas
mixing in anaerobic digestion. Water Res 85(15):497–511
29. Wang J, Xue Q, Guo T, Mei Z, Long E, Wen Q, Huang W, Luo T, Huang R (2018) A review
on CFD simulating method for biogas fermentation material fluid. Renew Sustain Energy Rev
97:64–73
30. Quiroz-Pérez E, Gutiérrez-Antonio C, Vázquez-Román R (2019) Modelling of produc-
tion processes for liquid biofuels through CFD: a review of conventional and intensified
technologies. Chem Eng Process Intensif 143:107629
31. Adeyemi NA, Mohiuddin A, Nor RM (2013) CFD modeling of waste cooking oil transesteri-
fication in a stirred tank reactor. World Appl Sci J 21:151–158. https://doi.org/10.4028/www.
scientific.net/AMM.390.86
32. de Boer K, Bahri PA (2009) Investigation of liquid- liquid two phase flow in biodiesel produc-
tion. In: Proceedings of 7th international conference on CFD in the minerals and process
industries
33. Huang X, Zhu J, Li J (2015a) Analysis of wellbore heat transfer in enhanced geothermal system
using CFD modeling. In: World geothermal congress, April 2015, p 6
34. Renaud T, Stebel M, Verdin PG, Falcone G (2018) CFD modeling of a high enthalpy geothermal
context geo-energy engineering centre. In: 43rd workshop on geothermal reservoir engineering,
Stanford University
35. Jiang F, Luo L, Chen J (2013) A novel three-dimensional transient model for subsurface heat
exchange in enhanced geothermal systems. Int Commun Heat Mass Transfer 41:57–62
36. Mubarok MH, Cater JE, Zarrouk SJ (2020) Comparative CFD modelling of pressure differential
flow meters for measuring two-phase geothermal fluid flow. Geothermics 86
37. Yousefi S, Atrens AD, Sauret E, Dahari M, Hooman K (2015) CFD convective flow simulation
of the varying properties of CO2 –H2 O mixtures in geothermal systems. Sci World J 2015:8.
https://doi.org/10.1155/2015/843068
38. Windt C, Davidson J, Ringwood JV (2018) High-fidelity numerical modelling of ocean wave
energy systems: a review of computational fluid dynamics-based numerical wave tanks. Renew
Sustain Energy Rev 93:610–630
39. Tatum S, Allmark M, Frost C, O’Doherty D, Mason-Jones A, O’Doherty T (2016) CFD
modelling of a tidal stream turbine subjected to profiled flow and surface gravity waves. Int J
Marine Energy 15:156–174
40. Karthikeyan T, Ezhilsabareesh K, Samad A (2016) Parametric analysis of a tidal current turbine
using CFD techniques. Progress Renew Energies Offshore
41. Jafari M, Babajani A, Hafezisefat P, Mirhosseini M, Rezania A, Rosendahl L (2018) Numerical
simulation of a novel ocean wave energy converter. In: International scientific conference
“environmental and climate technologies”, CONECT
42. Li KM, Parthasarathy N, Choi YH, Lee YW (2013) Study on wave energy generation of multi
floating bodies for energy absorption by CFD. J Korean Soc Power Syst Eng 17(5)
43. Schmitt P, Doherty K, Clabby D, Whittaker T (2012) The opportunities and limitations of
using CFD in the development of wave energy converters. In: Proceedings of the international
conference on marine and offshore renewable energy
476 S. Kumar and R. S. Amano

44. Vermeer LJ, Sorensen JN, Crespo A (2003) Wind turbine wake aerodynamics. Prog Aerosp
Sci 39:467–510
45. Sanderse B, Van Der Pijl SP, Koren B (2011) Review of computational fluid dynamics for wind
turbine wake aerodynamics. Wind Energy 14:799–819
46. Kalvig S, Manger E, Hjertager B (2014) Comparing different CFD wind turbine modeling
approaches with wind tunnel measurements. J Phys Conf Ser 555:012056
47. Sumner J, Watters CS, Masson C (2010) CFD in wind energy: the virtual, multiscale wind
tunnel. Energies 3:989–1013
48. Tachos NS, Filios AE, Margaris DP, Kaldellis JK (2009) A computational aerodynamics
simulation of the NREL phase II rotor. Open Mech Eng J 3:9–16
49. https://www.digitalengineering247.com/article/renewable-energy-options-multiply-with-cfd-
solutions/
50. Khare VR, Singh AP, Kumar H, Khatri R (2016) modelling and performance enhancement of
single slope solar still using CFD. In: International conference on recent advancement in air
conditioning and refrigeration, RAAR
51. Manjunath MS, Madhwesh N (2019) A review of CFD investigations on heat transfer augmen-
tation of forced convection solar air heater through enhanced fluid turbulence levels. Int J Mech
Prod Eng Res Dev 9:1309–1322
52. Gawande VB, Dhoble AS, Zodpe DB, Chamoli S (2016) A review of CFD methodology used in
literature for predicting thermo-hydraulic performance of a roughened solar air heater. Renew
Sustain Energy Rev 54:550–605
53. Avila-Marin AL, Fernandez-Reche J, Martinez-Tarifa A (2019) Modelling strategies for porous
structures as solar receivers in central receiver systems: a review. Renew Sustain Energy Rev
111:15–33
Energy Recovery from Waste Tires Via
Thermochemical Pathways

Kiran R. G. Burra, Zhiwei Wang, Matteo Policella, and Ashwani K. Gupta

Abstract Demand for automotive tires has been increasing at a significant rate
to reach production rate of 2.9 billion in 2017 that contributed to the generation of
almost 1 billion waste tires. The pathways for handling of these waste tires have been
limited to aggregate construction applications and direct incineration as fuel. The rate
of recycling feasibility was limited by the availability of waste tires with structure
intact for their upgrading and reuse in the economy. In the USA, although 40% of
the waste tires are disposed by combustion for energy recovery in cement kilns, pulp
and paper mills, and electricity generation, the presence of sulfur from vulcanized
rubber, the production of hazardous pollutants, and low temperatures caused due
to slow heterogenous combustion leads to significant energy and resource losses.
Additionally, this pathway is limited to thermal energy production and thus lacks
versatility. Alternative thermochemical pathways such as pyrolysis and gasification
offer better pathways for the utilization of these wastes as they provide uniform
products such as synthetic gas, bio-oil and char. Versatility is also achieved via
syngas production as it is a precursor to liquid fuels and various other essential
petrochemicals. These pathways provide improved energy efficiency, feasibility, and
scalability for increased waste tire utilization and value outcome compared to the
current application pathways. Results are reported from the investigations on high-
temperature pyrolysis and CO2 -assisted gasification of waste tires with focus on the
evolutionary behavior of syngas production, its constituents, and energy yield in lab-
scale fixed-bed reactor. The impact of CO2 addition, temperature, and the addition
of biomass feedstock to the waste tire was investigated to understand the feasibility
of waste tire disposal via this pathway while also utilizing the CO2 pollutant and
maintaining high energy efficiency. Lack of inhibitive effects observed when tire
was co-processed with biomass, makes the disposal of waste tires along with other

K. R. G. Burra · Z. Wang · M. Policella · A. K. Gupta (B)

Department of Mechanical Engineering, The Combustion Laboratory, University of Maryland,
College Park, MD 20742, USA
e-mail: [email protected]
Z. Wang
College of Environmental Engineering, Henan University of Technology, Zhengzhou 450001, PR,
China

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 477
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_21
478 K. R. G. Burra et al.

wastes easier along with established feasibility regime for efficient and economical
disposal of waste tires while recovering energy and utilizing CO2 .

Keywords Waste tires · Syngas · Gasification · Pyrolysis · Recycling · Energy

recovery

Nomenclature

α Extent of mass-loss conversion

BET Brunauer–Emmett–Teller
BR Butadiene rubber
Energyinput Electrical energy consumed by furnaces
GC-TCD Gas chromatography-thermal conductivity detector
LHV Low-heating value
Mi Mass flow rate of species ‘i’
mfeedstock Initial mass of the feedstock in the lab-scale reactor
msyngas Cumulated mass of syngas
M(T ) Mass of sample left in the TGA at temperature ‘T ’
NR Natural rubber
OEE Overall energy efficiency
PET Polyethylene terephthalate
ρi Density of species ‘i’ at GC inlet
SBR Styrene–butadiene rubber
T Temperature
TDF Tire-derived fuel
TGA Thermogravimetric analysis
Vi Volumetric flow rate
Xi Mole fraction of species ‘i’ recorded by the GC

1 Introduction

With the growing penetration of automobiles in the marketplace, their sustainability

is dependent on the availability and management of the fuel, and its materials. The
increased utilization of fossil fuels in transportation sector and industry is leading to
lack of fuel availability and increase in global warming from the imbalance between
carbon emissions and its sequestration. This can be seen by the trend of CO2 concen-
tration in the atmosphere which reached 415 ppm in 2020 from ~280 ppm in the
pre-industrial era [1]. Several efforts are being made here to support CO2 manage-
ment and control in the transportation sector with increased penetration of electric
vehicles and light-duty trucks. California has promulgated 100% electric vehicle
Energy Recovery from Waste Tires Via Thermochemical Pathways 479

sales by 2035 and phase out of gasoline powered vehicles by 2045. While renew-
able alternatives and carbon-free transportation are still in developing phase, their
scalability and penetration to completely replace the carbon-based infrastructure are
questionable. These resources need to be supplemented with carbonaceous fossil
fuel resources for the foreseeable future to achieve sustainable energy while seeking
carbon neutral footprint. All electric vehicles are justifiable only with the electricity
produced from renewable sources.
In addition to the fuel crisis, the handling and disposal of waste tires created by
increases in the automotive sector poses significant challenge to sustainably continue
producing or recovering these tires. An estimated 2.9 billion automobile tires were
produced in 2017, while almost 1 billion waste tires were generated [3]. Predictions
based on the current demand growth in automotive sector of 4.1% are expected to
result in production of almost 3.2 billion tires by 2022 [4]. Such a rapid growth
with the current trends of recovery will be unsustainable. In the USA alone, 255
million waste tires were disposed in 2017, wherein 15% were landfilled, and 41%
were combusted for energy recovery [2]. Figure 1 reveals the statistics in 2017 and
the current modes of utilization of waste tires which account for 81.4% going to the
market while 15% landfilled and the rest remained unaccounted [2]. Tire-derived fuel
(TDF) accounts for 41% of the utilization. Other sectors include cement kilns (19%),
pulp and paper mills (12%), and electric utilities (10%) for the energy recovered.
Almost 30% of this waste tires are ground for use in applications that included asphalt
for roads, sport surfaces and mulch, and extrusion/molded products. Utilization in
construction offers lower-value applications compared to other applications. The

Current Pathways of Waste Tire Utilization

Tire derived Fuel Ground rubber

Civil Engineering Exported
Electric Arc Furnaces Reclamation Projects
Agricultural Baled Tires/Market
Punched/Stamped Other
Land Disposed Unaccounted

Fig. 1 Fate of waste tires in the USA in 2017 [2]. [Color codes are does not account for all the
colors used]
480 K. R. G. Burra et al.

infrastructure that accounted for in TDF includes fossil fuel equivalent combustion
units where the waste tire is possibly mixed with other fuels such as coal, and biomass
for energy recovery. While these pathways operate with solid fuel efficiencies which
is lower than gas-phase combustion due to lower temperatures attained, they also pose
environmental hazard in terms of emissions of polychlorinated dibenzodioxins, Zn,
SOx from high S content in vulcanized rubber, polyaromatic hydrocarbons, partic-
ulates, odors and other toxic emissions [5]. Thus, to operate in low-environmental
impact mode regulated by the government, it requires multi-stage flue-gas cleaning,
which significantly lowers the net efficiency of the waste tire utility and recuperated
energy.
These statistics reveal the need for alternative pathways for waste tire disposal,
especially for energy purposes with the consideration of the imminent fuel crisis and
rapidly increasing CO2 concentration in the environment. Thermochemical path-
ways such as pyrolysis and gasification for oil and syngas yields provide significant
advantages in this aspect of feedstock utilization. Pyrolysis involves thermal decom-
position of carbonaceous feedstocks at high temperature to form syngas, oil and char
products [6]. Depending on the desired product phase, the operating conditions are
modified. High heating rates and low vapor residence times are used when oil is
desired, while low heating rates and high residence times are used for increased char
residue, and high temperatures above 700 °C when syngas is desired. In the case of
gasification, the sample is thermally decomposed at temperatures above 750 °C in
the presence of a mild oxidizing/gasifying agent such as H2 O, CO2 , diluted O2 /air
to yield significantly high syngas compared to pyrolysis [6]. While temperature has
a crucial role in the yield of syngas, its composition is significantly controlled by
the gasifying agent where steam (H2 O) yields H2 -rich syngas, CO2 yields CO-rich
syngas, and air/O2 is used for autothermal operation. Although these processes are
endothermic, the reformed uniform syngas, which is the focus of this chapter, is of
high quality in terms of its compositional uniformity, heating value, yield, energy
recovery and capability to modify composition with minimal downstream resources,
via water–gas shift reaction. Additionally, the versatility in applications offered by
syngas is unparalleled, because various pathways have been extensively studied and
implemented industrially such as Fischer–Tropsch synthesis for liquid fuels from
syngas [6]. Significant literature is available on the utilization of these pathways for
waste tire conversion, and some of it will be discussed in the context of evolutionary
behavior and energetic feasibility later in this chapter to establish the knowledge
and its gap compared to the available studies [3–5, 7–29]. This will be preceded
by understanding of waste tire constituents, composition and their behavior in the
context of thermochemical conversion.

2 Characterization of the Waste Tires

To characterize waste tires’ feedstock, we need to initially investigate the composition

and its variability in the waste tires collected. Typically, tires consist of a blend of
Energy Recovery from Waste Tires Via Thermochemical Pathways 481

natural (NR) and synthetic rubber (SR) wherein polybutadiene rubber (BR) and
styrene–butadiene rubber (SBR) are the most common examples of SR [30]. Carbon
black is added to this blend for enhanced resistance against abrasion, in addition
to extracting heat from the tread and the belt. For reinforcement, steel wires/textile
fabrics (polyester, rayon and nylon cords) are incorporated into this rubber mix for
enhancing strength and durability of tires [23]. During the vulcanization of the rubber
constituents, ZnO is added to control and improve its physical properties, while
sulfur is added as cross-linking agent between the elastomer chains while hardening
the final product and improving its resistance against thermal deformation at high
temperatures. While these are the major components, additional trace compounds
are also added to improve their quality and manufacturing, such as a mix of aromatic
hydrocarbons called extender oils and resins as plasticizers for improved shaping and
molding, antioxidants and antiozonants to combat oxidization and ozone influence
[5]. Different sources, i.e., different automobiles classes, have their specific needs
and thus, the characteristics of the tires utilized are also modified in terms of the
elastomers’ composition, reinforcement and its material, treading, geometry, carbon
black and other additives. For example, while passenger car tires are composed of
35% natural rubber (NR) and 65% polybutadiene rubber (BR), truck tires contain
51% NR, 10% BR and 39% styrene–butadiene rubber (SBR) [2]. Table 1 provides
further differentiation on the composition of typical passenger car and truck tires.
Literature also provides proximate and ultimate analysis of waste tires varying in their
sources and properties, and they are summarized in Table 2 [5]. This table reveals
the significance of waste tires in terms of the calorific value of this feedstock that is
seen more than 30–35 MJ/kg with some as high as 40 MJ/kg. These heating values

Table 1 Typical constituents and composition of waste tire and their utility [14, 30]
Component Passenger tire (wt. %) Truck tire (wt. %) Comments/examples
Rubber 47 45 Synthetic and natural rubbers,
examples: styrene–butadiene rubber,
natural rubber (polyisoprene), nitrile
rubber, chloroprene rubber,
polybutadiene rubber
Carbon black 21.5 22 Added to strengthen the rubber and
help in abrasive resistance
Metal 16.5 21.5 Steel belts and chords for strength
Textile 5.5 – Reinforcement
Zinc oxide 1 2 Used with stearic acid to control
vulcanization and enhance physical
properties of rubber
Sulfur 1 1 Used to control rubber polymer
chains, harden and prevent excessive
deformation at high temperatures
Additives 7.5 5 Example: Clay/silica as a partial
replacement to carbon black
Table 2 Proximate and ultimate analysis of various waste tires and their calorific value [5]
482

Elemental analysis (wt. % dry basis) Proximate analysis (wt. % as received basis) Calorific value (MJ/kg) References
C H N S O Ash Ash Volatiles Fixed carbon Moisture
81.72 6.54 0.55 1.87 2.68 6.64 6.64 62.58 30.07 0.71 nr [31]
83.8 6.9 0.6 2 2.3 4.4 4.3 63.4 30.4 1.9 nr [32]
83.8 7.6 0.4 1.4 3.1 3.7 3.7 67.3 28.5 0.5 36.45 [33]
82.36 6.92 2.3 1.4 2.03 5 4.95 73.74 20.22 1.09 37.06 [25]
75.5 6.75 0.81 1.44 15.5 + 20.1 57.5 20.85 1.53 29.18 [21]
85.9 8 0.4 1 2.3 2.4 2.4 66.5 30.3 0.8 40 [19]
75.4 7.03 0.21 1.62 5.09 10.21 10.21 62.32 26.26 1.31 33.29 [34]
84.33 7.81 0.49 1.66 3.32 2.4 7.1 62.2 29.4 1.3 40 [12]
82.8 7.6 0.5 1.3 4.5 3.3 3.3 68.7 27.2 0.8 36.46 [35]
86.7 8.1 0.4 1.4 1.3 2.1 8 61.9 29.5 0.7 36.2 [8]
80.29 7.25 0.31 1.84 4.9 5.41 5.3 67.5 25.2 2.1 37.3 [36]
85.05 6.79 0.5 1.53 1.75 4.4 4.35 62.24 32.28 1.14 34.9 [37]
81.5 7.1 0.5 1.4 3.4 6.1 6.07 64.87 28.56 0.5 36.8 [38]
86.09 6.74 0.19 1.93 1.35 3.7 3.7 65.5 29.4 0.9 nr [26]
86.7 6.9 0.3 1.9 0.9 3.3 4.4 64 30.7 0.9 31.8 [39]
83.92 6.83 0.78 0.92 3.39 4.16 4.16 64.97 30.08 0.75 38.6 [40]
85.25 7.94 0.41 1.38 1.19 3.83 3.83 64.09 31.14 0.94 nr [41]
83 6.79 0.32 1.37 3.46 5.06 5 64.1 29.7 1.2 35 [42]
67.08 6.12 0.17 2.05 24.58 + 19.1 59.69 19.45 1.72 27.37 [11]
81.79 7.99 0.48 1.81 3.04 4.9 4.88 65.74 28.98 0.4 38.3 [43]
(continued)
K. R. G. Burra et al.
Table 2 (continued)
Elemental analysis (wt. % dry basis) Proximate analysis (wt. % as received basis) Calorific value (MJ/kg) References
C H N S O Ash Ash Volatiles Fixed carbon Moisture
84 7.19 0.49 1.42 3.3 3.6 3.6 65.6 30 0.8 38.8 [44]
83.15 6.78 0.28 1.77 0.84 7.1 7.1 61.9 29.9 1.1 37.35 [45]
74.3 7.2 0.9 1.71 15.89 + 18.9 58.2 21.3 1.6 30.5 [17]
Energy Recovery from Waste Tires Via Thermochemical Pathways
483
484 K. R. G. Burra et al.

are comparable to some of the high-quality fuels such as anthracite and bituminous
coals, and biofuels such as alcohol fuels, but lower than domestic and transportation
fuels such as gasoline, kerosene and methane which are higher than 42–45 MJ/kg.
Table 2 also reveals significantly high carbon content and fixed carbon in the waste
tires which reveals high quality of the feedstock and emphasizes the importance of
carbon black’s presence in the context of its conversion as this means high levels
of char residue from pyrolysis. By extension, it also means that if gasified, major
reactions responsible for mass conversion are heterogenous reaction of the fixed
carbon components with the gasifying agent. This reveals the potential of waste tires
for use as an ideal source of energy and for coal replacement applications.
So, for the context of understanding the feasibility of waste tires, we restrict our
discussions in this chapter to specific types of tires in terms of the vehicle class.
From the statistics of 2017 given in Table 3, significant portions (almost 73%) of
waste tires are from passenger car, while commercial truck and bus tires account
for up to 11.4% of the total waste tires collected [2]. With this table as the basis,
we restricted our experimental investigation to passenger vehicle waste tire. Further
details of our materials and methods and observations will be preceded by various
results available in the literature regarding pyrolyzing and gasifying waste tires for
syngas production.

Table 3 Types of tires disposed in the USA in 2017 [2]

Tire class Millions of Tires Market % Average weight (kg) Weight (1000 ×
tons)
Light-duty tires 254.6 88.6 10.2 2864.3
Passenger tire 209.7 73.0
replacements
Light truck tire 31.3 10.9
replacements
Tires from scrapped 13.6 4.7
vehicles
Commercial tires 32.7 11.4 54.4 1962.0
Medium, wide-base, 19.7 6.9
heavy truck
replacement tires
Tires from scrapped 13.0 4.5
trucks and buses
Total tires hauled 287.3 100.0 15.24 4826.3
Used tires culled 37.9 13.2 15.24 637.1
Net scrap tires 249.4 4189.2
Energy Recovery from Waste Tires Via Thermochemical Pathways 485

3 Thermochemical Conversion of Waste Tires

Various types of processes were proposed and investigated to convert waste tires for
energy, fuels or chemicals recovery. Significant portion of the waste tire pyrolysis
had been vested in the production of oil, possibly via fast pyrolysis. Literature also
provides us with a compiled comparison of the state of the art of pyrolysis of waste
tires from different studies in several review articles [5, 14, 16]. Thus, we limit the
scope of this chapter, in the context of pyrolysis results reported in the literature, to
the results compiled from these review articles and their findings. Table 4 provides
us with a comparison of various pyrolysis studies reported and reveals the relative
yield of oil, char and gas products [14]. From this table, one can see that the char
yield was around 30–50% (by wt.), depending on the reactor conditions used. Such
a yield of char is significantly high that is limited to the processes designed for oil
production. This is a result of the high carbon black and thus fixed carbon content,
which is a characteristic of the waste tire feedstock. Additionally, Table 5 provides
information on the quality of oil produced in comparison with petrochemical fuels.
This reveals the requirement of downstream catalytic refining of the oil, possibly
novel hydrogenation, to make it compatible for fuel applications, especially in terms
modifying the flash point, calorific value enhancement, particulate and carbon residue
removal, and improving the boiling point characteristics [14]. The requirement of
significant refining adds to the issue of high char yields which leads to an end-result
of relatively low yields of the final fuel products. The high char yield, unwanted
high gas yield and catalytic refining requirements currently limit the capability of
pyrolysis as a pathway for the purposes of oil/liquid fuel production.
To alleviate these issues, various strategies have been proposed. As the gas yield
from this pyrolysis for liquid process is in the range of around 2–25% (by wt.), valu-
able utilization of this by-product is essential to support the overall process. The high
char yield from pyrolysis still makes this process less attractive and calls for another
strategy to target both the issues on high gas and less char yields. Gasification is
proposed to avoid direct formation of liquid yield and reform all the intermediates
from waste tires into syngas rich in H2 , CO, CH4 , impurities of C2 and C3 hydrocar-
bons and CO2 . This not only improves the effective yield of the desired product to
above 70% (by wt.), but also provides uniform and versatile syngas, which can be
used for various applications. Even modifying pyrolysis to focus on gas yield, such
as high temperature mesh reactor tests shown in Table 4 reveals the gas yield to be
higher than 70% (by wt.) and minimized char yield to fixed carbon content levels.
Significant knowledge is available on gasification of waste tires involving various
reactor systems, operating conditions and gasifying agents, and they are summa-
rized in some of the well-compiled review articles in the literature [16]. To obtain a
complete picture on the gasification of waste tires, one needs to first investigate pyrol-
ysis under similar conditions as this is the first inherent step in gasification. Conesa
et al. found from pyrolysis of waste with high vapor residence times (>10 min) that
the increase in temperature from 450 to 1000 °C, changed the char yield from 35 to
37% due to the high residence time, while the gas yield increased from 27 to 62%
486 K. R. G. Burra et al.

Table 4 Summary of different pyrolysis tests on waste tires and their relative product yields [14]
Reactor Experimental conditions Maximum oil yield References
T (°C) Oil Char Gas
(wt.%) (wt.%) (wt.%)
Fixed bed, 400–700 °C 500 40.26 47.88 11.86 [46]
batch
Fixed bed, 500–1000 °C, 500 58 37 5 [36]
batch 1200 °C/min
Fixed bed, 300–720 °C, 720 58.8 26.4 14.8 [19]
batch 5–80 °C/min, 50 g
Fixed bed, 450–600 °C, 5 °C/min, 475 58.2 37.3 4.5 [12]
batch 3 kg
Fixed bed, 950 °C, ~2 °C/min, 950 20.9 40.7 23.9 [47]
batch 1000 kg
Fixed bed, 350–600 °C, 5 & 400 38.8 34 27.2 [10]
batch 35 °C/min,
Fixed bed, 300–700 °C, 15 °C/min, 700 38.5 43.7 17.8 [48]
batch 175 g
Fixed bed, 375–500 °C, 10 °C/min, 425 60 ~30 ~10 [11]
batch 10 g
Fixed bed, 375–575 °C, 750 g 475 55 36 9 [49]
batch,
internal fire
tubes
Fixed, wire 390–890 °C, 70–90 °C/s 860 ~5 ~22 ~73 [29]
mesh, fast
reactor
Moving 600–800 °C, 3.5–8 kg/h 600 48.4 39.9 11.7 [31]
screw bed throughput
Rotary kiln 550–680 °C, 4.8 kg/h 550 38.12 49.09 2.39 [50]
Rotary kiln 450–650 °C, 12–15 kg/h 500 45.1 41.3 13.6 [37]
Fluidized 740 °C, 1 kg/h 740 30.2 48.5 20.9 [51]
bed
Fluidized 750–780 °C, 30 kg/h 750 31.9 38 28.5 [51]
bed
Fluidized 700 °C, 200 kg/h 700 26.8 35.8 19 [51]
bed
Fluidized 450–600 °C, ~220 kg/h 450 55 42.5 2.5 [52]
bed
Circulating 360–810 °C, 5 kg/h 450 ~52 ~28 ~15 [35]
fluidized bed
Conical 425 and 500 °C 500 ~62 ~35 ~3 [53]
spouted bed
(continued)
Energy Recovery from Waste Tires Via Thermochemical Pathways 487

Table 4 (continued)
Reactor Experimental conditions Maximum oil yield References
T (°C) Oil Char Gas
(wt.%) (wt.%) (wt.%)
Vacuum, 425 and 500 °C, 25 & 500 ~60 ~34 ~4 [54]
conical 50 kPa
spouted bed
Vacuum 485–550 °C, batch 520 45 36 6 [55]
(80–180 kg) and
continuous
Vacuum 500 °C, pilot-scale 500 56.5 33.4 10.1 [56]
semi-continuous
Vacuum 450–600 °C, 100 g batch 550 47.1 36.9 16 [40]
Closed batch 350–450 °C, 30 °C/min 450 ~63 ~30 ~7 [57]
reactor
Drop tube 450–1000 °C, 30 g/h 450 37.8 35.3 26.9 [26]
reactor

at the expense of oil yield that decreased from 38% to negligible levels [26]. This is
because of high-temperature operation leads to enhanced secondary reactions such as
thermal cracking of the hydrocarbon intermediates to form light hydrocarbons, H2 ,
CO and CO2 at significant vapor residence times. These results are also evident from
the data shown in Tables 6 [14]. From Table 4, one can see that high heating rates
(70–90 K/s), high temperature and vapor residence times (15 min) in a wire-mesh
microreactor led to significantly high gas yield of 73% [29]. But at temperatures of
450–600 °C with short residence times (~2 min), heating rate of 5 °C/min led to
maximum oil yield of 58%. These results also showed that temperature had minimal
effect on the relative yields. Although significant literature is available on pyrolytic
breakdown of waste tires, focus on syngas yield and evolutionary behavior during
gasification is still lacking. Our studies into high-temperature pyrolysis of waste
tires and its mixtures with biomass provide comprehensive information (given in
the following sections) that establishes the feasibility of syngas recovery from waste
tires.
Various reactor systems have been utilized for the gasification of waste tires, which
included fixed bed, bubbling and circulating fluidized bed reactors of both lab-scale
and pilot-scale, rotary-kiln reactors and plasma gasifiers [5, 16]. Knowledge on the
gasification at pilot scale has been limited to using air and steam as the gasifying
agent. The addition of air oxidized some of the inflammable components evolved
from waste tire decomposition while the exothermic reaction from this oxidation
supports the endothermic decomposition reactions responsible for volatile evolution.
This leads to syngas with relatively high CO2 compared to other gasifying agents but
provided with the advantage of autothermal operation, i.e., alleviation of the need
of external heat/energy source for continuous operation and maintaining the desired
temperature. Due to the ease of energy transfer and its similarity to incineration,
488 K. R. G. Burra et al.

Table 5 Properties of oil products obtained from waste tire pyrolysis in the literature [14]
Property Ref. [47] [37] [10] [58] Kerosene [59] Gas oil Light fuel oil
Flash point 20 17 65 43 40 75 79
(o C)
Carbon 2.2 1.78 – – – < 0.35 –
residue (%)
Density 0.91 0.962 0.833 0.924 0.84 0.78 0.89
(kg.l−1 )
Viscosity (cSt 6.3 – – 3.77 1.2 3.3 21
at 40 °C)
Viscosity (cSt – 2.44 1.01 0.924 – – –
at 50 °C)
Viscosity (cSt 2.38 – – – – – –
at 60 °C)
Carbon 88.0 84.26 79.61 – – 87.1 85.5
(wt.%)
Hydrogen 9.4 10.39 10.04 – 13.6 12.8 12.4
(wt.%)
Nitrogen 0.45 0.42 0.94 – – 0.05 0.15
(wt.%)
Sulfur (wt.%) 1.5 1.54 0.11 0.72 0.1 0.9 1.4
Oxygen 0.5 3.39 9.3 – – – –
(wt.%)
Initial B.Pt 100 – 38.5 70 140 180 200
(o C)
10% B.Pt 140 – 58.2 114.5 – – –
(o C)
50% B.Pt 264 – 174.8 296.1 200 300 347
(o C)
90% B.Pt 355 – – 386.4 315 – –
(o C)
Calorific 42 41.7 42.66 38 46.6 46 44.8
value (MJ/kg)
Ash (wt.%) 0.002 Trace – 0.31 – 0.01 0.02
Moisture 4.6 0.88 (wt.%) – – – 0.05 0.1
(vol.%)

autothermal mode is the most preferred way of gasification currently used at large
scale at the expense of syngas quality [3]. A novel pathway of pilot-scale solar air
gasification was reported in Wieckert et al. concerning the conversion of waste tires
in comparison with various other feedstocks including coal that reported sustainable
energy efficiency of 27% in waste tire conversion [60]. While solar concentrator
supported gasification has significant potential in waste tire conversion, the lack
of knowledge and the challenges associated with energy losses in the current solar
Table 6 Proximate and ultimate analysis, and calorific value of the tested feedstocks [4]
Sample Volatile Fixed Ash (wt.%) Moisture LHV Carbon Hydrogen Nitrogen Oxygen Sulfur
(wt.%) carbon (wt.%) (MJ kg−1 ) (wt.%) (wt.%) (wt.%) (wt.%) (wt.%)
(wt.%)
Waste tire 62.5 27.9 8.9 0.7 33.3 81.85 6.66 1.7 9.8 1.37
Pine bark 63.9 26.2 4.8 5.2 18.8 50.37 4.20 1.61 43.81 0.03
Energy Recovery from Waste Tires Via Thermochemical Pathways
489
490 K. R. G. Burra et al.

concentrator reactors limit this pathway but is a pursuable challenge for carbon
neutral production of energy/chemicals from waste tires and other wastes.
Increase in temperature was found to increase the syngas yield and thus the energy
recovery ratio (syngas energy/waste tire energy) as found by Raman et al. who
reported the energy recovery ratio improved linearly from 0.1 to 0.4 with increase in
temperature from 900 to 1060 K [61]. Other studies have found 0.4 for the highest
energy recovery obtained under these conditions of air gasification. In the case
of steam gasification, Donatelli et al. found that increase in steam/feedstock ratio
decreased the calorific content of the syngas (possibly by reforming all the hydrocar-
bons into H2 and CO) to result in decreased energy recovery ratio [20, 62]. Various
gasifying agents were investigated to convert waste tires including air, steam, and
CO2 but studies on CO2 gasification are visibly limited [16]. While air gasification
provides autothermal operation and steam gasification provides high H2 yield, the
advantages of CO2 gasification are multi-fold and more eco-friendly. Incorporation
of CO2 into waste tire gasification leads to utilization of CO2 for char gasification
and volatiles reforming which results in syngas enhancement and improved carbon
conversion. It also assists in net conversion of CO2 into valuable syngas which leads
toward making this process more carbon neutral. High-temperature CO2 is avail-
able as flue gas in some industries, such as coal combustion plants, which can be
synergistically used for waste tire conversion for effective carbon utilization. Further
studies into the scope of CO2 gasification with respect to waste tire conversion are
necessary, and lab-scale studies have been carried out by our group in this direction
that support effective and efficient conversion of waste tires using CO2 . Following
sections provide detailed experimental setup and result that established feasibility
of CO2 -assisted gasification of waste tires as a sustainable pathway for waste tire
conversion.
The influence of catalyst in waste tire conversion was conducted using dolomite
addition that resulted in lowered calorific value of syngas, and thus the energy
recovery ratio. However, the utilization of transition metal (Ni, Co, Fe, Cu) supported
on Al2 O3 carried out by Zhang et al. reported the formation of carbon nanotubes in
the char yield that constituted to 8–12 wt.% of the char [63]. Studies on the compar-
ison of commercial Ni-based catalysts, with mineral catalysts such as dolomite and
olivine, were also reported in the literature using bubbling fluidized bed gasifier.
These studies showed improved H2 and syngas yield via reforming/cracking reac-
tions at the expense of calorific output. The issues of catalyst deactivation from
coke deposition were as expected due to high levels of char formation tendency and
hydrocarbon presence with Ni/Al2 O3 leading to H-abstraction and enhanced H2 yield
from 24 to 57%. Utilization of this carbon deposition for the production of carbon
nanotubes was also investigated [63]. Further studies into such a pathway of H2 and
carbon nanotubes are essential to establish its feasibility and operating conditions for
optimal products as control over the quality of carbon nanotubes, and their extrac-
tion is critical to establish their production. Syngas with H2 content as high as 99%
was also produced from the gasification of waste tires in the presence of CaO [64].
Although the review article claims CaO to be catalyst, H2 enhancement was carried
out via well-known sorption enhanced reforming to remove CO2 via carbonization of
Energy Recovery from Waste Tires Via Thermochemical Pathways 491

CaO into CaCO3 which can drive the product equilibrium to high H2 partial pressures
[16]. But the reusability of CaO was low, and thus the state-of-the-art feasibility is
questionable as reported in gasification and sorption-enhanced reforming studies in
the literature.
The gasification of waste tires with other feedstocks including coal products
such as lignite, biomass/biowaste such as pinewood, palm kernel shells, municipal
sludge, olive husk, almond shells, palm fruit and plastic wastes such as polyethylene
terephthalate (PET) from plastic bottles, and their mixtures were also carried out for
various selected goals [16]. Almond shells and palm fruits were added to improve the
char reactivity from an otherwise stable low-reactive tire char and found significant
improvement in carbon conversion levels via co-conversion. Even in the cases of co-
processing, although the values of syngas yield, quality and energy with respect to
temperature, catalyst and feedstock ratio were carried out, systematic studies into the
impact of feedstock composition and the feedstock ratio on the syngas components
and conversion behavior, especially the understanding of possible nonlinear synergy
is still lacking in the literature. Enhanced syngas evolution via nonlinear interaction
between polymer wastes and biomass/coal were reported with selected plastics but
such studies involving waste tires are still lacking [65–70]. Studies to understand the
impact of co-processing waste tire with biomass via pyrolysis and gasification have
also been carried out and reported in the following sections [4, 15, 18].

4 Methods and Materials

4.1 Feedstock Materials

For studies to establish the feasibility of waste tire conversion via thermochemical
pathways, we utilized used-passenger car tire as such waste constitutes to almost
90% of the total waste tires collected in the USA [2]. The waste tire used was a
Goodyear winter radial tire reinforced with textile fibers instead of metal wires. For
lab-scale pyrolysis and gasification studies, this tire was cut into small cubes having
each side of approx. 1.5–2 cm. Thermogravimetric studies were also carried out to
characterize this feedstock, and for such microscale studies, the waste tire was frozen
using liquid N2 and then grounded with rubber particles of approx. ~100 µmin size.
For each case, the total tested feedstock of waste tire included a mix of 60 wt.%
from the tire tread and 40 wt.% from the side wall. Due to reinforcement and tire
construction, the composition of the waste tire varied from the side wall to tread.
This proportional mixture was utilized to accurately represent a real waste tire.
For co-pyrolysis and co-gasification studies, pine bark was chosen as a represen-
tative forestry residue biomass feedstock that was added to the waste tire. The pine
bark feedstock was obtained from a local nursery, and the acquired feedstock was
dried at 105 °C for 24 h to minimize moisture content before blending with the waste
492 K. R. G. Burra et al.

tire. The samples acquired were characterized using proximate and ultimate anal-
ysis to establish the fixed carbon and volatile content along with carbon, hydrogen,
nitrogen and sulfur (CHNS) content of the chosen feedstocks. Thermogravimetric
analysis (TGA) was utilized following modified ASTM 7582–15 standard for prox-
imate analysis, while low heating value (LHV) of the samples and the pyrolysis of
char blends was obtained using rapid screening device (5E-KCIII, China); see Table
6 for these results.

4.2 Thermogravimetric Analysis (TGA)

Thermogravimetric analysis was carried out using TA Instruments SDT Q600 with
each sample weighing approximately 1–3 mg. The TGA was equipped with hori-
zontal beam balance in an electric furnace and was capable of operating at temper-
atures of up to 1500 °C. The samples of waste tire and pine bark were ground to
140 mesh particle size separately and as blends for these analyses. The samples were
heated from room temperature to 100 °C and maintained for 10 min to remove mois-
ture, followed by heating the sample at 10 or 20 °C/min to reach to temperatures
of 950 °C in the presence of high-purity Ar at 100 ml/min. During this heating, the
sample mass was recorded to understand the kinetics of their thermal decomposition.
At 950 °C, the sample was maintained for 10 min, while the purge gas was changed
to dry air. This step was carried out to combust the solid-char residue thus leaving
behind only the ash contents. The mass-loss data from these tests allowed us to not
only obtain the pyrolysis kinetics but also proximate analysis.

4.3 Lab-Scale Fixed Bed Reactor

Pyrolysis and CO2 -assisted gasification experiments were conducted using a lab-
scale fixed-bed reactor system that was operated in semi-batch mode at different set
temperatures of up to 1000 °C. The reactant gases flowed through the reactor contin-
uously along with the product gases while for any given test, the sample feedstock
was introduced as a batch of fixed mass. This reactor system was powered by two
electric furnaces—Lindberg/Blue M Mini-Mite for preheating the incoming gases
and Lindberg/Blue M 1200 °C Split-hinge tube furnace for sample heating at any
defined temperature. The input gases included N2 as the tracer gas, CO2 as gasifying
agent, and Ar to purge residual gases from the system. The flow rates of these gases
were controlled separately using orifice flow meters. The evolved product gases
were collected in sampling bottles (for short sample interval collection) or trans-
ported directly to a gas chromatograph for detailed analysis of the gases evolved.
The feedstock sample holder was made of stainless-steel wire mesh, and this was
inserted into the uniform heated zone of the reactor system at appropriate time via
a quick-disconnect coupling at back end of the reactor. A schematic diagram of the
Energy Recovery from Waste Tires Via Thermochemical Pathways 493

Fig. 2 Schematic of the experimental setup used for pyrolysis and gasification [15]

experimental setup details is provided in Fig. 2. The setup was equipped with ice-bath
condenser and particulate filters to cool the product gases and remove moisture, tar
and other particulate content in the product gases prior to their transport for gas anal-
ysis. Online product gas analysis was carried out using gas chromatography-thermal
conductivity detector (GC-TCD) (Agilent Micro GC 3000A) which was calibrated
for the quantification of mole fractions of H2 , CO, CO2 , CH4 , C2 H4 , C2 H2 , C2 H6 , O2
and N2 (using Restek refinery gas standard #1). The time required for each sample
for gas analysis was almost 2.6 min. Additionally, the online gas sampling system
was also equipped with gas sampling bottles in case of requirement of sampling
times faster than 2.6 min. The electrical furnaces were monitored for their energy
consumption (EML-2000, Canada) to estimate the energy requirements of each of
the pyrolysis and gasification tests to obtain accurate values of overall energy effi-
ciency beyond typically reported on energy recovery data. The char samples were
characterized for their morphology using scanning electron microscopy (JSM-6510,
Japan) at beam voltage of 20 kV after these samples were sputtered with Pt for better
scanning characteristics. The char samples were also characterized for their surface
area and pore volume using surface area and porosity analyzer (Tristar II 3020, USA).
Each test began by first allowing the input gases to go through the reactor system
while the two furnaces reached the desired set point temperatures. The temperatures
of interest were from 700 to 1000 °C. The input gases were chosen depending on
the test conducted. For pyrolysis and co-pyrolysis tests, 2.1 sccm of N2 flowed while
avoiding entry of any other gas to maintain inert conditions during purging and
analyzing the evolved gases. In the case of CO2 gasification, a gas mixture of 75
494 K. R. G. Burra et al.

vol.% CO2 and 25 vol.% N2 with a total flow rate of 2.1 sccm was used to allow
for CO2 -assisted gasification. Here the vapor residence times were comparable to
that in pyrolysis tests. Once the setpoint temperature was attained and the gas flow
rates fixed, the sample holder containing 35 g of feedstock sample was inserted into
the main reactor via quick-disconnect coupling. Tested feedstock samples included
waste tire, pine bark and their mixtures at varying relative mass fractions of 1:3,
1:1 and 3:1 to not only analyze the influence of the mass ratio but also compare
the results with the weighted results from separate processing of pure feedstocks to
examine any possible interaction. While most of the product gases evolved from the
main reactor were vented out, a fraction of it passed through the ice-bath condenser
system that removed moisture and tar prior to analyzing the product gases. This dry
and tar-free syngas were collected at 0.5, 1, 2, 3, 4 min from the start of the reaction
using sampling bottles as such short sampling times of 1 min were not possible
for online gas analysis using micro-GC and that our previous tests using different
feedstocks had revealed that during this time, most of the syngas was released. From
5 min, this syngas was sent directly to the micro-GC for direct online gas analysis
where the gas was collected at time intervals of 2.6 min and analyzed for the mole
fractions of different gases against the calibrated gases.

Xi
Mi = ∗ VN2 ∗ ρi (1)
X N2

Using the mole fraction of each gas species with respect to mole fraction of N2 ,
the gas flow rate of each of the gas species was calculated as the flow rate of N2
was constant and it did not change during the reaction. From the known flow rate
of N2 , it can be used as internal calibration reference to measure the flow rate of
each of the gas species using the Eq. (1). In this equation, X i is the measured mole
fraction of species ‘i’ at given time, VN2 is the volumetric flow rate of N2 at the
reactor inlet, ρi is the standard density of species ‘i’ and Mi is the mass flow rate of
species ‘i’ at given time. This method of obtaining gas flow rates assumes ideal gas
behavior of the species, the inertness of N2 toward these reactions, and that the syngas
sample composition is representative of the composition of the evolved syngas prior
to venting. Additionally, since each gas sample was collected over span of 10 s, this
data represented an averaged value over that timespan at each time duration. This
was carried out until the syngas at the GC was negligible. Ar was used to flush the
reactor system and sampling lines of residual gases between each test. After this,
the gas samples from sampling bottles were analyzed individually using micro-GC
while Ar flushing was carried between the analyses for each of these bottles. The
operating conditions of this experimental setup are summarized in Table 7.
Energy Recovery from Waste Tires Via Thermochemical Pathways 495

Table 7 Summary of the operating reactor conditions for pyrolysis and gasification tests [15]
Pyrolysis
Reactor temperature 673, 773, 873, 973, 1073, and 1173 K
Operating pressure Atmospheric
Tracer gas and inert medium 2.1 sccm of N2 (at standard 294 K and 1 atm.)
Sample 35 g of waste tire 2 × 2 cm pieces (60% tread and 40%
sidewall)
Apparent vapor residence time ~21 s
Isothermal reaction time 2.5 h
Gasification
Reactor temperature 973, 1073, 1173, and 1273 K
Operating pressure Atmospheric
Tracer gas and inert medium 2.1 sccm of 75% vol. CO2 and 25% vol. N2 gas mixture (at
standard 293 K and 1 atm.)
Sample 35 g of waste tire 2 × 2 cm pieces (60% tread and 40%
sidewall)
Apparent vapor residence time ~21 s
Isothermal reaction time 53 min

5 Results and Discussion

5.1 Thermogravimetric Analysis

5.1.1 Waste Tire Decomposition

This analysis was carried out for gaining insights into the pyrolytic kinetics and its
associated mass-loss behavior during the thermal decomposition of waste tire. This is
essential as it provides us with the characteristic temperatures of decomposition and
possibly also the compositional details of the rubbers used in the examined waste tire
sample. With inert operation of TGA tests, one could quantify the volatile content
sourced from the rubbers, accelerators and other additives present in the tire. The
combustion step using air at high temperature was carried out to quantify the inorganic
content and the fixed carbon content (mostly carbon black) in the case of waste tire.
TGA tests in CO2 environment were avoided since the decomposition here is slow and
at their characteristic temperatures (due to low heating rate) which were significantly
low for CO2 to be active. Additionally, to normalize the mass-loss data obtained from
TGA into non-dimensional format, conversion (α) was obtained, and the results are
reported using this data and its derivative with respective to instantaneous temperature
(DTG = dα/dT ). Normalization was carried out according to Eq. (2), where M(T )
is the mass of the remaining sample in the TGA at temperature T, T o is the initial
temperature (100 °C) and T f is the final temperature (950 °C).
496 K. R. G. Burra et al.

M(To ) − M(T )
α(T ) = (2)
M(To ) − M(Tf )

This conversion data for thermal decomposition of waste tire is presented in

Fig. 3 along with DTG (derivative with respect to temperature). These results show
similarity compared to those reported in the literature. The DTG data is useful to
represent the behavior of waste tire decomposition and reveals the volatile evolution
to be a combination of three stages (i.e., showing three peaks) at temperatures of
450–600 K, 600–680 K, and 680–800 K. Each of these regions could be a combi-
nation of decomposition patterns from different constituents, such as decomposition
of unstable additives such as extender oils and other low-boiling point additives that
contribute to the first stage of decomposition while the second stage could be from
natural rubber (NR) decomposition and initiation of decomposition of polybutadiene
rubber (BR), and styrene-butadiene rubber (SBR). The third stage could be signifi-
cantly from the decomposition of BR and SBR based on the straight elastomer and
is based on the TGA studies reported in the literature [41]. Most of the literature on
such analysis of tires has reported the latter two stages corresponding to elastomers
decomposition and in some cases, the second stage contributed more to the mass-
loss than the third due to different proportions of the rubber mix. The high mass-loss
from the third peak revealed that the specific waste tire examined here contained
significantly high BR and SBR proportions compared to NR and that the ratio of BR
to SBR could be predicted from the position and contribution of the third peak based
on the literature reported on varying ratios.

DTGwaste tire = wNR ∗ DTGNR + wBR ∗ DTGBR

+ wSBR ∗ DTGSBR (3)

To quantify the relative composition of individual rubbers in the waste tire exam-
ined, we employed a three-component curve-fit to the DTG data based on the straight

Fig. 3 Extent of mass-loss (α) and its derivative with temperature during waste tire pyrolysis [15]
Energy Recovery from Waste Tires Via Thermochemical Pathways 497

Table 8 Estimated
Polymers Composition (wt.%)
composition of elastomers in
the tested waste tire [15] Natural rubber (NR) 5
Butadiene rubber (BR) 40
Styrene–butadiene rubber (BR) 55

elastomer DTG data from the literature, neglecting the presence of any interaction’s
contribution to the mass distribution and additives contribution to the mass-loss in
the latter two stages. Equation (3) revealed that the equation used to represent such
a model was favorable, where wi represent the mass fraction of the respective elas-
tomer in the waste tire, and the DTGNR , DTGBR and DTGSBR were DTG of the
respective straight-chain elastomers obtained from the literature. The DTGwaste tire
obtained from our data and the fit obtained had a coefficient of correlation of R2 =
0.996. The estimated compositional results from this analysis are given in Table 8
that reveals significant presence of BR and SBR.

5.1.2 Decomposition of Waste Tire-Pine Bark Blend

Investigations involving TGA on blends of waste tire with other feedstocks are essen-
tial to understand the presence of any interaction and establish the feasibility of a
feed-flexible gasifier where waste tires can be converted in the presence of any other
feedstocks for sustainable operation and avoid over dependence on any individual
feedstock. For this, blends of waste tire and pine bark were mixed in defined frac-
tion and ground into similar size powder form. The relative fractions of each in the
blends were 1:1, 1:3 and 3:1. These samples will be represented by WxPy where
x represents waste tire mass fraction and y represents pine bark mass fraction; for
example, W1P3 represents a sample with 25 wt.% waste tire and 75 wt.% pine bark.
The results from these blends will be compared with the calculated weighted results
from the individual components. Figure 4 reveals the conversion and DTG behavior

Fig. 4 Effect of waste tire-pine bark mixture ratio on a the extent of mass-loss, α, and b its derivative
[18]
498 K. R. G. Burra et al.

of the blends in comparison with pure components—waste tire (W1P0) and pine
bark (W0P1). Pine bark also exhibited the presence of multiple peaks corresponding
to the decomposition of lignocellulosic components such as hemicellulose, cellu-
lose and lignin, respectively, with increase in temperature. Note that lignin partially
contributed to all the stages of decomposition. Pine bark decomposition started early
from 220 °C while waste tire pyrolysis was concentrated between 250 and 550 °C.
Comparison of blended DTG with weighted results (not shown here), i.e., comparing
DTGWxPy with (x*DTGwaste tire + y*DTGpine bark )/(x + y), revealed the lack of any
difference in behavior. Although this reveals the lack of any synergistic interaction
between waste tire and pine bark, it is important to note the lack of any inhibitive
behavior either. This means that the decomposition of either components was possibly
occurring independent of each other. This also means that conversion of waste tire
in the presence of biomass feedstocks such as pine bark can be carried out in the
same reactor without the loss of any efficiency and easily predict the decomposition
behavior without any issues of interaction. Although any interaction seems to be
missing according to the DTG behavior, further lab-scale studies were also carried
out via pyrolysis and CO2 gasification to confirm such behavior and observe for
any high-temperature, heating rate effects and volatile–volatile interaction behaviors
that cannot be observed from TGA results. Such a behavior on the lack of interaction
from TGA results but presence of significant synergistic interaction during lab-scale
high temperature gasification and pyrolysis was observed in various biomass–plastic
mixtures such as pinewood–polyethylene. The lab-scale studies described here in the
later sections revealed the observation of such a synergistic behavior.

5.2 Lab-Scale Pyrolysis and Gasification

5.2.1 Product Gas and Char Yield

Waste Tires

Online gas chromatography analysis was used to analyze and quantify dry, tar-free
and particulate-free product gases for H2 , CO, CO2 , CH4 , C2 H4 , C2 H6 and C2 H2 as
they represent the major compounds evolved from pyrolysis and gasification of tires.
In this chapter, we refer to the combination of H2 , CO, CH4 , C2 H4 , C2 H2 and C2 H6
as syngas since these components are important due to their calorific value, while
the combination of this syngas with CO2 will be referred to as product gas since
one can consider CO2 as a by-product. From the mass flow rates obtained with time
for different temperatures, cumulative yield of these components was calculated
over a 50-min time duration for both pyrolysis and gasification to understand the
influence of temperature and gasifying agent on the syngas evolution and thus the
carbon conversion. Char yields were also measured at the end of reaction for these
conditions to gain insights into the tendency of charring and the loss of yield by
forming char as by-product.
Energy Recovery from Waste Tires Via Thermochemical Pathways 499

Fig. 5 Effect of temperature on a syngas and b char yield during pyrolysis and CO2 -assisted
gasification of waste tires [15]

Figure 5 reveals the variation of syngas and char yield (as wt. % of waste tire
mass after 2.5 h) with temperature and gasifying agent. At lower temperatures of
700–800 K, insignificant amounts of syngas evolved due to the inability of conver-
sion of the volatiles evolved from waste tire to further crack or interact with CO2
to form any of the evolved syngas components examined. But, at these tempera-
tures char yield was more than 50% that decreased to ~38% as the temperature was
increased to beyond the maximum temperature of complete conversion (~770 K) as
seen from the DTG results presented in Fig. 3. From 800 K, syngas yield increased
strongly with temperature because at high operational temperatures of the lab-scale
reactor, the sample experienced high heating rates immediately after placing it in the
reactor, leading to increased volatiles evolution. These volatiles undergo secondary
reactions such as thermal cracking, and CO2 -reforming in the gas phase leading to
the formation of syngas components. These gas-phase reactions leading to the forma-
tion of H2 , CO and light hydrocarbons are endothermic and thus with increase in
temperature, the syngas yield is enhanced. In the case of gasification, temperatures
above 973 K are required for CO2 to actively react with the intermediates from the
decomposing waste tire. Syngas yield as high as 50% from pyrolysis and 100% from
gasification was obtained from waste tire conversion. The literature also reported
high gaseous yields and were dependent on the reactor used to obtain high tempera-
tures and heating rates, high vapor residence times and slow quenching that provided
the necessary conditions for enhanced secondary reactions of cracking and reforming
in the gas phase. Comparing the yields from pyrolysis and gasification, increase in
temperature increased the differences in the gas yield as the char reforming reactions
also contributed leading to improved CO yields.
In the case of char yield, beyond the temperature of complete devolatilization from
TGA results (770 K), the char yields did not change significantly with temperature
during pyrolysis as the char obtained was stable, similar to the char residue from
TGA results. But in the case of gasification, beyond 1173 K, the char yield decreased
significantly by about 50% due to Boudouard reaction of the char with CO2 which
was only active at high temperatures beyond 1100 K based on our previous studies
on CO2 -assisted gasification of various other kinds of feedstocks.
500 K. R. G. Burra et al.

Fig. 6 Effect of waste tire-pine bark blend ratio on the evolution behavior and yield of syngas
during co-pyrolysis at 900 °C (solid bars: experimental; hollow bars: weighted yields) [4]

Waste Tires with Pine Bark

Investigations into co-pyrolysis and co-gasification of waste tires with pine bark
at 900 °C were conducted using lab-scale reactor system to examine any possible
interaction. Figure 6 shows the influence of feedstock mixture fraction on the
syngas yield during co-pyrolysis in comparison with estimated yields calculated
by weighted aggregate of the individual feedstocks processed separately. The syngas
yields revealed no significant interaction between waste tire and pine bark pyrolysis
when processed together as the blend co-pyrolysis yields were similar to the weighted
yields. At the examined temperature, the syngas yield was found to be ~40 wt.% for
all the blend samples and pure samples. This was probably due to the high thermal
stability of fixed carbon in waste tire leading to its low reactivity toward pine bark
pyrolysis. Variation of char yield after 48 min with blend ratio during co-pyrolysis
was also measured, and the results are shown in Fig. 7. While no significant inter-
action was observable, as the differences between the yields from co-pyrolysis and
the weighted yields were negligible, the yield decreased with increase in pine bark
content due to relatively high charring tendency from waste tire from its fixed carbon
compared to pine bark.
Co-gasification was carried out at both 800 and 900 °C under similar conditions
as in our other gasification studies but at different blend ratios. Figure 8 reports
the influence of feedstock mixture on the syngas yield at these temperatures. The
results obtained showed a decrease in syngas yield from co-gasification at 800 °C
compared to separate feedstock gasification, but no such inhibition was observed at
900 °C. Further investigation into the composition of the syngas can provide us with
information about any such interaction.
Energy Recovery from Waste Tires Via Thermochemical Pathways 501

Fig. 7 Effect of feedstock

blend ratio on the char yield
from co-pyrolysis of waste
tire and pine bark at 900 °C
[4]

5.2.2 Syngas Components’ Evolution and Yield

Waste Tires

From online product gas analysis capabilities using our lab-scale reactor system, we
were able to obtain the temporal evolutionary behavior of flow rates of syngas and its
components. Figure 9 reveals the impact of temperature on the evolutionary behavior
of syngas during pyrolysis and gasification from the waste tire sample. The flow rate
curves showed peak, and with increase in temperature, the peak value increased,
but its position shifted toward earlier (shorter) times. This is because increase in
temperature not only drives the equilibrium toward syngas product due to the global
endothermicity of both pyrolysis and CO2 -assisted gasification, but also improves
the kinetics of these reactions. This behavior was observed in the evolution of all
the major components of syngas; see Fig. 10. Hydrogen (H2 ) and CH4 were the
dominant components impacted by the temperature change during pyrolysis while
CO and H2 during CO2 -assisted gasification. While the evolution of syngas flowrate
completed in 25 min during pyrolysis, it decreased to a constant nonzero value during
CO2 -assisted gasification and then stayed at that flow rate for extended periods of
time; see Fig. 10. CO evolution showed similar behavior that extended for a long
period of time during CO2 -assisted gasification. During this process, the contribution
of CO to syngas yield was very high. This was due to low reactivity of tire-char
residual with CO2 after 25 min, possibly due to high crystallinity, low imperfections
and O, H content, and low surface area compared to other carbonaceous materials
such as biomass. Long gasification times were reported to be required even in the
case of steam gasification despite relatively higher reactivity of tire char with steam
compared to CO2 .
Figure 11 reveals the cumulative yields of the individual syngas components and
their variation with temperature during pyrolysis and CO2 -assisted gasification, while
502 K. R. G. Burra et al.

Fig. 8 Effect of feedstock blend ratio on the syngas yield from co-gasification of waste tire and
pine bark at a 800 °C and b 900 °C [18]
Energy Recovery from Waste Tires Via Thermochemical Pathways 503

Fig. 9 Effect of temperature on the evolution of syngas from waste tires under a pyrolysis and
b gasification [15]

Fig. 12 shows their mole fraction in cumulative product gas yield (including CO2 ).
In pyrolysis, increase in temperature increased the yields of H2 , CO, CH4 and C2
hydrocarbons. During pyrolysis, decarboxylation and decarbonylation occur of the
oxygenates in the waste tire such as extender oils, secondary reactions with char,
along with decarboxylation of inorganic content such as CaCO3 , CaSiO4 and other
metal carbonates. Low molecular mass hydrocarbons (CH4 , and C2 ) were generated
by gas-phase cracking of the volatiles released from the decomposition of BR and
SBR polymers in addition to H2 . Hydrogen is also released from aromatization and
cyclization of the volatiles in the gas phase, while these reactions are significantly
enhanced with temperature. Comparing the H atomic content in waste tire (from
ultimate analysis shown in Table 6) with the yields of H2 and CH4 revealed that
significant portion of H content in the solid sample got converted to syngas.
Figure 12 reveals that while the net CO2 yield enhanced with temperature, its mole
fraction in product gas decreased. At low-temperature pyrolysis, while decarboxyla-
tion reaction is favored to release CO2 , insignificant conversion of heavy hydrocarbon
intermediate volatiles into syngas components resulted in high CO2 mole fraction.
As the temperature increased, other components enhanced due to the contribution
of enhanced secondary gas-phase reactions at a higher rate leading to the observed
net reduction in CO2 mole fraction. This is also because of enhanced equilibrium
drive toward higher CO content at high temperature compared to CO2 . This enhanced
thermal cracking of heavy intermediates from rubber components can be seen from
enhanced CH4 mole fraction in product gas; see Fig. 12. However, in the case of CO2 -
assisted gasification, this increase in mole fractions of CH4 and C2 with temperature
is met with opposing reactions such as dry hydrocarbon reforming to form H2 and
CO which results a decrease in these hydrocarbon mole fractions beyond 1173 K.
Equations (4)–(11) represent the major reactions during pyrolysis and gasification
of waste tires and biomass. Improved rate of increase in CO yield due to Boudouard
reaction also contributed to the lowered hydrocarbon yields at high temperature; see
Eq. (5).
504 K. R. G. Burra et al.

Fig. 10 Effect of temperature on the evolution of H2 , a, b, CO, c, d and CH4 , e, f during pyrolysis
(left) and CO2 -assisted gasification (right) of waste tires [15]

Waste tire + pine bark + CO2

→ H2 + CO + CO2 + CH4 + Cm Hn
(C2 H2 + C2 H4 + C2 H6 + C3 H8 + . . .)
+ H2 O + char + tar H > 0 (4)

C + CO2  2CO H = +172 kJ/mol (5)

Energy Recovery from Waste Tires Via Thermochemical Pathways 505

Fig. 11 Effect of temperature on the cumulative yields of a H2 , b CO and c CH4 from pyrolysis
and CO2 -assisted gasification of waste tires [15]

Fig. 12 Effect of temperature on the mole fractions of individual species from cumulative syngas
yield from a pyrolysis and b CO2 -assisted gasification of waste tires [15]
506 K. R. G. Burra et al.

C + 2H2 → CH4 H = −75 kJ/mol (6)

C + H2 O → CO + H2 H = +131 kJ/mol (7)

CO + 3H2  CH4 + H2 O H = −206 kJ/mol (8)

CO2 + H2  CO + H2 O H = 41 kJ/mol (9)

 
Tar Cx H y + xCO2 → 2xCO + (y/2)H2 H
= 0.9 to 0.3 MJ/mol (10)

Tar → CH4 + H2 + H2 O + Cm Hn H > 0 (11)

Equations (5), (7), (9) and (10) contribute to provide significant enhancement of
CO yield with increase in temperature, and the rate of increase was higher at high
temperature. At temperatures beyond 1073 K, CO yield increased with temperature
by 1.5 times/100 K to provide yields as high as 25.6 mmol/g which accounted for
0.72 g per gram of waste tire feedstock. Below this temperature, the yield of all the
syngas components from CO2 -assisted gasification was almost the same as that from
pyrolysis due to low activity of CO2 at these temperatures. TGA results and literature
also reveal this to be the case while this also supports CO2 -assisted gasification to
be an extension of pyrolysis to include CO2 reforming of the evolving volatiles and
char into CO and H2 ; see Figs. 5 and 10 that reveal lowered hydrocarbon and char
yields with increase in CO yield compared to pyrolysis at high temperatures.

Waste Tires with Pine Bark

To investigate further into any possible synergistic or inhibitive interaction between

waste tire and pine bark during their co-processing via pyrolysis and CO2 -assisted
gasification, we examined the evolution of syngas components and yield and
compared the results with the weighted data calculated from separate conversion
of the feedstock. The weighted yields were calculated with the initial mass fraction
of the feedstock components to be their respective weights in the aggregate.
Figure 13 reveals the influence of feedstock mass fraction on the variation of
major syngas components’ mass flow rate with time along with their cumulative
yield compared with weighted yields during co-pyrolysis. The results reveal the
yields from blended feedstock to be almost same as the weighted yields from mono-
conversion. This means that in pyrolysis to obtain gaseous yields, waste tire and pine
bark feedstock show no interaction with each other and the behavior of the reaction
products was a superimposition of their behavior when pyrolyzed separately. This
Energy Recovery from Waste Tires Via Thermochemical Pathways 507

Fig. 13 Effect of feedstock blend fraction on the evolution and yield of a H2 , b CO and c Cn Hm
from co-pyrolysis of waste tire and pine bark [4]
508 K. R. G. Burra et al.

simplifies the modeling and design of feed-flexible reactors as it alleviates the need
for any further studies into nonlinear behavior. Additionally, this also provides the
design option to modify the syngas composition and their respective component yield
by modifying the mixture composition of the feedstock. While the syngas yield did
not change significantly with the feedstock in the case of co-pyrolysis, hydrocarbon
yield was favored with high waste tire content, while CO yield was favored with
high pine bark content, due to oxygenates present in the biomass. H2 yield also
increased with increase in pine bark content but to a lesser extent compared to other
species. As the net syngas yield was almost the same for all the feedstock mass
fractions, it suggests that the decreased mass of Cn Hm with increase in pine bark was
appropriately compensated by the increase in CO mass yield. The variation in CO
yield varied from less than 5 wt.% from pure waste tire pyrolysis to almost 30 wt.%
from pure pine bark, while Cn Hm yield varied between 40 wt.% from pure waste tire
to ~10 wt.% from pure pine bark.
Similar data showing the evolution of syngas components and their yield is
reported in Fig. 14 for both 800 and 900 °C that not only provide information
about the variability on the evolutionary behavior with feedstock variation, but also
provide insights into possible interaction that could lead to non-additive yields.
Increase in temperature led to significant increase in the peak H2 mass flow rate
from ~0.035 g/min to as much as ~0.075 g/min due to enhanced cracking and CO2 -
reforming. Increase in pine bark content also increased the H2 flow rate and yield due
to the higher reactivity of the biomass to react with CO2 to form H2 and CO. At 800 °C,
some variation between the H2 yields from the blends and the weighted results was
found. However, at 900 °C, these differences disappeared to finally provide blended
yields same as the weighted yields. CO flow rate and yield also increased with
increase in pine bark content not only due to its higher reactivity compared to waste
tire but also due to inherently higher oxygenate content in the biomass. The CO
yield also increased with temperature by almost twice from 800 to 900 °C due to
enhancement of CO from pyrolysis along with char gasification and CO2 -reforming.
The behavior of CO flow rate with time showed not only a peak but also a steady CO
yield from ~10-min time due to Boudouard reaction involving char-CO2 gasification
to provide enhanced CO yields. Such a constant flow rate was due to slow reaction
rate of the char with CO2 , possibly due to low porosity. Synergistic reduction of
Cn Hm was also found at 800 °C, which also led to lower syngas yield, as seen from
Figs. 8 and 14. But this was not observed at 900 °C, possibly due to higher influence
of gas-phase reactions with CO2 leading to similar yields as the weighted yields.
From these results, we can see that at operational temperatures of 900 °C required
for CO2 -assisted gasification, the syngas component yields from co-gasification were
superimposed sum of the yields from individual components to prove the compat-
ibility of waste tire with biomass gasifiers. So, subject to the net product output
required, the feedstock can be modified to include waste tires without the loss of
conversion efficiency.
Energy Recovery from Waste Tires Via Thermochemical Pathways 509

Fig. 14 Effect of temperature and feedstock blend fraction on the evolution and yield of a, b H2 ,
c, d CO and e, f Cn Hm from co-gasification of waste tire and pine bark at 800 °C and 900 °C,
respectively [18]
510 K. R. G. Burra et al.

Fig. 14 (continued)
Energy Recovery from Waste Tires Via Thermochemical Pathways 511

Fig. 15 Effect of temperature on a energy yield and b LHV of the syngas yield from pyrolysis and
CO2 -assisted gasification of waste tire [15]

Fig. 16 Effect of
temperature on the overall
energy efficiency of
pyrolysis and CO2 -assisted
gasification of waste tire [15]

5.2.3 Syngas Energy Recovery and Overall Energy Efficiency

Waste Tires

The energy yield from pyrolysis and gasification in the form of calorific content
of syngas was obtained by aggregating the yield of individual syngas components’
yield multiplied with their respective lower heating value (LHV). Similarly, LHV
of the CO2 -free syngas was also obtained by dividing the respective energy yield
with the syngas yield to provide a perspective of the quality of flammable gases
that can be obtained. From this energy yield, the overall energy efficiency was also
calculated using the Eq. (12) which also considers the electrical energy consumed
by the furnaces.

Overall energy efficiency

m syngas ∗ LHVsyngas
= (12)
Energyinput + m feedstock ∗ LHVfeedstock

Figure 15 reveals the energy yield and LHV of the syngas, while Fig. 16 reveals
the overall energy efficiency of both pyrolysis and CO2 -assisted gasification of waste
tire. The energy yield increased with increase in temperature for both gasification and
512 K. R. G. Burra et al.

pyrolysis although higher content of hydrocarbons in the syngas from pyrolysis led to
higher energy yield from pyrolysis compared to gasification, which was dominated
by CO yield due to higher LHV of hydrocarbons compared to CO. So, the energy
yield and efficiency reported here are from the perspective of utilizing the products for
energy recovery purposes. Pyrolysis yielded overall energy efficiency of almost 38%,
while gasification yielded 30% to support the feasibility of waste tire conversion for
energy recovery. Although this may seem relatively low, it only accounts for syngas
energy and significant portions of energy were also transformed into char residue
that offers its own value.

Waste Tires with Pine Bark

To further understand the energy yield and overall energy efficiency, variations
obtained from co-pyrolysis and co-gasification of waste tire with pine bark were
also examined. Additionally, to obtain accurate representation of the overall energy
recovery feasibility, heating value of char residue was also measured and the calorific
content of this char was also incorporated into the overall energy efficiency as the
product energy output to observe the capability of our lab-scale reactor experiments
in establishing the feasibility of this process. Additionally, studies incorporating
pine bark also revealed the pathways of modifying energy input and output for better
design of conversion reactors.
Figure 17 reveals the overall energy efficiency variation with time along with inset
cumulative results of efficiency at different times that included consideration on char
energy from co-pyrolysis. The variation with time reveals the efficiency to reach
maximum at 10 min while beyond that temperature resulted in a slow decrease in the
efficiency. At 15 min, the efficiency reached as high as 45% while considering only
syngas energy, and by 47.7 min the efficiency reached to 35%. Note that reactors need
to be designed by considering both the extent of conversion and the process efficiency.
So, these results will provide with the required understanding to design reactors for
high yield and efficiency along with optimal syngas composition. Incorporation of
the char energy into efficiency after 47.7 min led to almost constant efficiency of 48%
for all the waste tire-pine bark blends examined. This makes the overall process to be
feasible and sustainable. Comparison of the energy efficiency from the blends with
the weighted results, we can also see a slight non-additive synergistic enhancement in
the efficiency when the feedstocks used were blends compared to separate pyrolysis.
Such synergistic enhancement could form cumulative enhancement of all the syngas
yields to a lesser extent contributing to an overall significance.
From these results, we can see that significant portions of the feedstock were
converted into char leading to the limitations of efficiency. While the efficiency is
acceptable, the char needs to be either converted for energy purposes or some other
utility for better economic output from waste tire conversion. A combination of
syngas production with extraction of char including carbon black for utilization in
material applications is necessary for sustainable operation of this pathway. Figure 18
shows surface area of the char produced from co-pyrolysis of waste tire and pine bark.
Energy Recovery from Waste Tires Via Thermochemical Pathways 513

Fig. 17 Effect of feedstock blend fraction on the overall energy efficiency (OEE) from co-pyrolysis
of waste tire and pine bark [4]

Fig. 18 Effect of feedstock

blend fraction on the surface
area of char residue from
co-pyrolysis of waste tire
and pine bark [4]

The results show relatively low-surface area from waste tire compared to pine bark,
and their co-pyrolysis led to inhibitive lowered char surface area compared to separate
pyrolysis. Figure 19 shows the difference in the morphology of chars from waste
char and pine bark and their variation with co-pyrolysis. Further studies are essential
to explore future direction toward the applicability of these chars for high-value
purposes, such as carbon catalysts, activated carbon and other valuable carbon solid
products to make thermochemical conversion of waste tires economically attractive.
514 K. R. G. Burra et al.

(a) W:P=100:0 (1000×) waste tire (b) W:P=0:100 (1000×) pine bark

(c) Waste tire solid in W:P=75:25 (d) Pine bark solid in W:P=75:25
(1000×) (1000×)

(e) Waste tire solid in W:P=50:50 (f) Pine bark solid in W:P=50:50
(1000×) (1000×)

(g) Waste tire solid in W:P=25:75 (h) Pine bark solid in W:P=25:75
(1000×) (1000×)

Fig. 19 Morphology of char residue from waste tire and pine bark during their co-pyrolysis at
different blend ratios [4]
Energy Recovery from Waste Tires Via Thermochemical Pathways 515

5.2.4 CO2 Consumption During CO2 Gasification

In addition to providing waste tire disposal and energy recovery, CO2 -assisted gasi-
fication also provided with positive utility to high-temperature CO2 utilization and
thus reduced net carbon emissions from this disposal process. In our studies, we
characterized the capability of CO2 gasification in CO2 utilization by examining the
net CO2 consumption during the overall gasification process. This was carried out for
both waste tire gasification and its co-gasification with different ratios of pine bark.
Figures 20 and 21 reveal the influence of temperature and feedstock composition
on the cumulative CO2 consumption. Increase in temperature led to increase in CO2
consumption with values as high as 0.7 g of CO2 consumed per each gram of waste
tire at 1000 °C. This consumption rises from the Boudouard reaction and gas-phase
CO2 reforming reactions which enhanced with temperature. Comparison revealed
CO2 consumption during pine bark conversion was higher than waste tire conversion.

Fig. 20 Effect of
temperature on CO2
consumption during
gasification of waste tires
[15]

Fig. 21 Effect of feedstock

blend ratio on CO2
consumption during
co-gasification of waste tires
with pine bark [18]
516 K. R. G. Burra et al.

Additionally, co-gasification also seemed to not provide any non-additive behavior

and was similarly superimposable as the syngas yields from co-gasification.
These studies provide important pathways that establish the capability of ther-
mochemical pathways of pyrolysis and CO2 -assisted gasification for efficient,
sustainable and economic recovery of waste tires to energy.

6 Conclusions

This chapter provides improved understanding for efficient disposal of real waste
tires via thermochemical pathways such as pyrolysis and CO2 -assisted gasification
in return for syngas which provides with versatility in utilization and uniformity
in handling. Syngas obtained from this pathway is useful for various applications
including use as fuel for energy recovery and value-added chemicals production such
as methanol, and Fischer–Tropsch transportation fuels. Thermogravimetric analysis
revealed the waste tire samples to thermally decompose between temperatures of
120–550 °C and included mass-loss rate peaks of different elastomer content and the
additives such as extender oils. Curve-fitting this data with TGA results from straight-
chain elastomers of similar nature to those present in the examined waste tire revealed
the composition to be dominated by butadiene and styrene–butadiene rubbers, while
natural rubber’s contribution was only about 5%. Pyrolysis and gasification were
carried out using a lab-scale fixed bed reactor to understand the evolution, yield and
composition of the product gases extracted. Temperature had significant influence on
the syngas yield and composition, where high temperatures led to increased syngas.
Endothermic reactions such as thermal cracking of volatile intermediates released
from waste tire pyrolysis, CO2 -reforming of the hydrocarbons into H2 and CO, and
Boudouard reaction of CO2 with char residue were enhanced on both equilibrium
and kinetic front to increase the yield and lead to faster release of syngas. Waste tire’s
char poses a difficulty in gasification as the CO yield from its conversion was low
and continued for extended periods of time that concluded low reactivity of its chars.
BET surface area value of char from waste tire was only about a half of that from pine
bark char. Co-pyrolysis and co-gasification of waste tire and pine bark biomass were
also conducted in various proportions of their blends. The results revealed almost
no synergistic or inhibitive interaction between waste tire conversion and pine bark
conversion, and the results obtained were found to be a weighted superimposition
of the results from their respective conversions when conducted separately. These
results are a step closer to developing feed-flexible gasifiers that can provide with
sustained syngas output and efficiency from various feedstocks with minimal changes
to its operating conditions. The lack of any inhibitive interactions means that waste
tire can be co-processed with pine bark or vice versa, and the predictable behavior
allows one for easier designing of the reactor, while the feedstock can be adjusted to
maintain the energy input and other design parameters.
Energy Recovery from Waste Tires Via Thermochemical Pathways 517

Overall energy efficiency was also calculated for these processes based on the
syngas’s calorific content, the measured electrical energy requirements for the lab-
scale reactors and the calorific content of the feedstock. The results revealed efficiency
to improve with increase in temperature that reached as high as 38% in pyrolysis
and 30% in CO2 -assisted gasification wherein higher unreformed hydrocarbons in
pyrolysis products and their high heating value compared to reformed CO from gasi-
fication led to this differences. Incorporating char energy from the measured heating
values of the chars to the overall energy efficiency resulted in efficiency to reach
almost 50%. Additionally, the co-processing of waste tire with pine bark was able
to provide a cumulative synergistic enhancement in efficiency beyond the weighted
efficiency that could be achieved by separate conversion of the two components.
These results establish the feasibility of pyrolysis and CO2 -assisted gasification of
waste tire and its mixtures with biomass as sustainable pathway for their disposal
while providing usable syngas.
A major challenge for future studies into these pathways is the waste tire char that
was found to be of low reactivity and surface area while being a significant by-product
that accounted for almost 30% of the sample mass. Developing pathway for conver-
sion and utilization of this char is essential to not only make these pathways sustain-
able and efficient in material management, but also to provide economic support.
Future research needs to focus on producing high-value carbon products from these
char residues, such as carbon catalysts, activated carbon for sorption, electrochem-
ical and energy storage applications to replace expensive materials (such as synthetic
graphite and fossil fuel-derived activated carbon), high-quality solid fuel, and carbon
dots for biomedical and other imaging applications. These activities will help provide
improved value to economically support the thermochemical pathways so that waste
tires can be disposed in an eco-friendly fashion while sustainably recovering energy
and producing value-added chemicals.

Acknowledgements This work was supported by Office of Naval Research (ONR), and it is grate-
fully acknowledged. K. G. Burra received Ann. G Wylie fellowship, and this support is gratefully
acknowledged.

References

1. Keeling R (2020) Scripps institute of oceanography 2020. https://scrippsco2.ucsd.edu/.

Accessed 26 Sep 2020
2. Association U tire rubber. 2017 U. S. scrap tire management summary about the U. S. Tire
Manufacturers Association 2018
3. Machin EB, Pedroso DT, de Carvalho JA (2017) Energetic valorization of waste tires. Renew
Sustain Energy Rev 68:306–315. https://doi.org/10.1016/j.rser.2016.09.110
4. Wang Z, Burra KG, Zhang M, Li X, Policella M, Lei T et al (2020) Co-pyrolysis of waste tire
and pine bark for syngas and char production. Fuel 274:117878. https://doi.org/10.1016/j.fuel.
2020.117878
518 K. R. G. Burra et al.

5. Martínez JD, Puy N, Murillo R, García T, Navarro MV, Mastral AM (2013) Waste tyre pyrol-
ysis—a review. Renew Sustain Energy Rev 23:179–213. https://doi.org/10.1016/j.rser.2013.
02.038
6. Burra KG, Gupta AK (2018) Thermochemical reforming of wastes to renewable fuels. In:
Runchal AK, Gupta AK, Kushari A, De A, Aggarwal SK (eds) Energy propuls. A sustain.
Technol. Approach, Springer, Singapore, pp 395–428. https://doi.org/10.1007/978-981-10-
7473-8_17
7. Hasan A, Dincer I (2019) Comparative assessment of various gasification fuels with waste tires
for hydrogen production. Int J Hydrogen Energy 44:18818–18826. https://doi.org/10.1016/j.
ijhydene.2018.11.150
8. González JF, Encinar JM, Canito JL, Rodrı́guez JJ (2001) Pyrolysis of automobile tyre waste.
Influence of operating variables and kinetics study. J Anal Appl Pyrolysis 58–59:667–83.
https://doi.org/10.1016/S0165-2370(00)00201-1
9. Colom X, Faliq A, Formela K, Cañavate J (2016) FTIR spectroscopic and thermogravimetric
characterization of ground tyre rubber devulcanized by microwave treatment. Polym Test
52:200–208. https://doi.org/10.1016/j.polymertesting.2016.04.020
10. Banar M, Akyıldız V, Özkan A, Çokaygil Z, Onay Ö (2012) Characterization of pyrolytic oil
obtained from pyrolysis of TDF (Tire Derived Fuel). Energy Convers Manag 62:22–30. https://
doi.org/10.1016/j.enconman.2012.03.019
11. Kar Y (2011) Catalytic pyrolysis of car tire waste using expanded perlite. Waste Manag
31:1772–1782. https://doi.org/10.1016/j.wasman.2011.04.005
12. Cunliffe AM, Williams PT (1998) Composition of oils derived from the batch pyrolysis of
tyres. J Anal Appl Pyrolysis 44:131–152. https://doi.org/10.1016/S0165-2370(97)00085-5
13. Karatas H, Olgun H, Engin B, Akgun F (2013) Experimental results of gasification of waste
tire with air in a bubbling fluidized bed gasifier. Fuel 105:566–571. https://doi.org/10.1016/j.
fuel.2012.08.038
14. Williams PT (2013) Pyrolysis of waste tyres: a review. Waste Manag 33:1714–1728. https://
doi.org/10.1016/j.wasman.2013.05.003
15. Policella M, Wang Z, Burra KG, Gupta AK (2019) Characteristics of syngas from pyrolysis
and CO2 -assisted gasification of waste tires. Appl Energy 254:113678. https://doi.org/10.1016/
j.apenergy.2019.113678
16. Oboirien BO, North BC (2017) A review of waste tyre gasification. J Environ Chem Eng
5:5169–5178. https://doi.org/10.1016/j.jece.2017.09.057
17. Ucar S, Karagoz S, Ozkan AR, Yanik J (2005) Evaluation of two different scrap tires
as hydrocarbon source by pyrolysis. Fuel 84:1884–1892. https://doi.org/10.1016/j.fuel.2005.
04.002
18. Wang Z, Burra KG, Lei T, Gupta AK (2019) Co-gasification characteristics of waste tire and
pine bark mixtures in CO2 atmosphere. Fuel 257:116025. https://doi.org/10.1016/j.fuel.2019.
116025
19. Williams PT, Besler S, Taylor DT (1990) The pyrolysis of scrap automotive tyres. Fuel 69:1474–
1482. https://doi.org/10.1016/0016-2361(90)90193-T
20. Portofino S, Donatelli A, Iovane P, Innella C, Civita R, Martino M et al (2013) Steam gasification
of waste tyre: influence of process temperature on yield and product composition. Waste Manag
33:672–678. https://doi.org/10.1016/j.wasman.2012.05.041
21. Rofiqulislam M, Haniu H, Rafiqulalambeg M (2008) Liquid fuels and chemicals from pyrolysis
of motorcycle tire waste: product yields, compositions and related properties. Fuel 87:3112–
3122. https://doi.org/10.1016/j.fuel.2008.04.036
22. Chen DT, Perman CA, Riechert ME, Hoven J (1995) Depolymerization of tire and natural
rubber using supercritical fluids. J Hazard Mater 44:53–60. https://doi.org/10.1016/0304-389
4(95)00047-X
23. Lah B, Klinar D, Likozar B (2013) Pyrolysis of natural, butadiene, styrene-butadiene rubber
and tyre components: modelling kinetics and transport phenomena at different heating rates
and formulations. Chem Eng Sci 87:1–13. https://doi.org/10.1016/j.ces.2012.10.003
Energy Recovery from Waste Tires Via Thermochemical Pathways 519

24. Berrueco C, Esperanza E, Mastral FJ, Ceamanos J, García-Bacaicoa P (2005) Pyrolysis of

waste tyres in an atmospheric static-bed batch reactor: analysis of the gases obtained. J Anal
Appl Pyrolysis 74:245–253. https://doi.org/10.1016/j.jaap.2004.10.007
25. Xiao G, Ni M-J, Chi Y, Cen K-F (2008) Low-temperature gasification of waste tire in a fluidized
bed. Energy Convers Manag 49:2078–2082. https://doi.org/10.1016/j.enconman.2008.02.016
26. Conesa JA, Martín-Gullón I, Font R, Jauhiainen J (2004) Complete study of the pyrolysis and
gasification of scrap tires in a pilot plant reactor. Environ Sci Technol 38:3189–3194. https://
doi.org/10.1021/es034608u
27. de Marco RI, Laresgoiti M, Cabrero M, Torres A, Chomón M, Caballero B (2001) Pyrolysis of
scrap tyres. Fuel Process Technol 72:9–22. https://doi.org/10.1016/S0378-3820(01)00174-6
28. Kim S, Park JK, Chun H-D (1995) Pyrolysis kinetics of scrap tire rubbers. I: Using DTG
and TGA. J Environ Eng 121:507–14. https://doi.org/10.1061/(ASCE)0733-9372(1995)121:
7(507)
29. Zabaniotou A, Stavropoulos G (2003) Pyrolysis of used automobile tires and residual char
utilization. J Anal Appl Pyrolysis 70:711–722. https://doi.org/10.1016/S0165-2370(03)000
42-1
30. Evans A, Evans R (2006) The composition of a tyre: typical components. Waste Resour
Action Program 2006. http://www.wrap.org.uk/sites/files/wrap/2 - Composition of a Tyre -
May 2006.pdf=https://doi.org/10.3233/HAB-1993-4306
31. Aylón E, Fernández-Colino A, Murillo R, Navarro MV, García T, Mastral AM (2010) Valori-
sation of waste tyre by pyrolysis in a moving bed reactor. Waste Manag 30:1220–1224. https://
doi.org/10.1016/j.wasman.2009.10.001
32. Scala F, Chirone R, Salatino P (2003) Fluidized bed combustion of tyre derived fuel. Exp
Therm Fluid Sci 27:465–471. https://doi.org/10.1016/S0894-1777(02)00249-2
33. Lee JM, Lee JS, Kim JR, Kim SD (1995) Pyrolysis of waste tires with partial oxidation in a
fluidized-bed reactor. Energy 20:969–976. https://doi.org/10.1016/0360-5442(95)00049-M
34. Chang YM (1996) On pyrolysis of waste tire: degradation rate and product yields. Resour
Conserv Recycl 17:125–139. https://doi.org/10.1016/0921-3449(96)01059-2
35. Dai X, Yin X, Wu C, Zhang W, Chen Y (2001) Pyrolysis of waste tires in a circulating fluidized-
bed reactor. Energy 26:385–399. https://doi.org/10.1016/S0360-5442(01)00003-2
36. Leung DYC, Yin XL, Zhao ZL, Xu BY, Chen Y (2002) Pyrolysis of tire powder: influence of
operation variables on the composition and yields of gaseous product. Fuel Process Technol
79:141–155. https://doi.org/10.1016/S0378-3820(02)00109-1
37. Li S-Q, Yao Q, Chi Y, Yan J-H, Cen K-F (2004) Pilot-scale pyrolysis of scrap tires in a
continuous rotary kiln reactor. Ind Eng Chem Res 43:5133–5145. https://doi.org/10.1021/ie0
30115m
38. Roy C, Labrecque B, de Caumia B (1990) Recycling of scrap tires to oil and carbon black
by vacuum pyrolysis. Resour Conserv Recycl 4:203–213. https://doi.org/10.1016/0921-344
9(90)90002-L
39. Senneca O, Salatino P, Chirone R (2000) A fast heating-rate thermogravimetric study of the
pyrolysis of scrap tyres. Fuel Energy Abstr 41:104. https://doi.org/10.1016/S0140-6701(00)909
20-2
40. Zhang X, Wang T, Ma L, Chang J (2008) Vacuum pyrolysis of waste tires with basic additives.
Waste Manag 28:2301–2310. https://doi.org/10.1016/j.wasman.2007.10.009
41. Murillo R, Aylón E, Navarro M, Callén M, Aranda A, Mastral A (2006) The application of
thermal processes to valorise waste tyre. Fuel Process Technol 87:143–147. https://doi.org/10.
1016/j.fuproc.2005.07.005
42. Larsen MB, Schultz L, Glarborg P, Skaarup-Jensen L, Dam-Johansen K, Frandsen F et al (2006)
Devolatilization characteristics of large particles of tyre rubber under combustion conditions.
Fuel 85:1335–1345. https://doi.org/10.1016/j.fuel.2005.12.014
43. López FA, Centeno TA, Alguacil FJ, Lobato B (2011) Distillation of granulated scrap tires in
a pilot plant. J Hazard Mater 190:285–292. https://doi.org/10.1016/j.jhazmat.2011.03.039
44. Betancur M, Daniel Martínez J, Murillo R (2009) Production of activated carbon by waste tire
thermochemical degradation with CO2 . J Hazard Mater 168:882–887. https://doi.org/10.1016/
j.jhazmat.2009.02.167
520 K. R. G. Burra et al.

45. Dı́ez C, Martı́nez O, Calvo LF, Cara J, Morán A (2004) Pyrolysis of tyres. Influence of the
final temperature of the process on emissions and the calorific value of the products recovered.
Waste Manag 24:463–9. https://doi.org/10.1016/j.wasman.2003.11.006
46. Aydın H, İlkılıç C (2012) Optimization of fuel production from waste vehicle tires by pyrolysis
and resembling to diesel fuel by various desulfurization methods. Fuel 102:605–612. https://
doi.org/10.1016/j.fuel.2012.06.067
47. Williams PT, Bottrill RP, Cunliffe AM (1998) Combustion of tyre pyrolysis oil. Process Saf
Environ Prot 76:291–301. https://doi.org/10.1205/095758298529650
48. Laresgoiti M, Caballero B, de Marco I, Torres A, Cabrero M, Chomón M (2004) Character-
ization of the liquid products obtained in tyre pyrolysis. J Anal Appl Pyrolysis 71:917–934.
https://doi.org/10.1016/j.jaap.2003.12.003
49. Islam RM, Hossain Joardder MU, Kader MA, Islam Sarker MR (2011) Valorization of solid
tire wastes available in Bangladesh by thermal treatment. In: Proceedings of the WasteSafe
2011—2nd international conference on solid waste management in the developing countries,
Khulna, Bangladesh, pp 1–9
50. Galvagno S, Casu S, Casabianca T, Calabrese A, Cornacchia G (2002) Pyrolysis process for the
treatment of scrap tyres: preliminary experimental results. Waste Manag 22:917–923. https://
doi.org/10.1016/S0956-053X(02)00083-1
51. Kaminsky W, Mennerich C, Zhang Z (2009) Feedstock recycling of synthetic and natural
rubber by pyrolysis in a fluidized bed. J Anal Appl Pyrolysis 85:334–337. https://doi.org/10.
1016/j.jaap.2008.11.012
52. Williams PT, Brindle AJ (2003) Fluidised bed pyrolysis and catalytic pyrolysis of scrap tyres.
Environ Technol 24:921–929. https://doi.org/10.1080/09593330309385629
53. Olazar M, Aguado R, Arabiourrutia M, Lopez G, Barona A, Bilbao J (2008) Catalyst effect
on the composition of tire pyrolysis products. Energy Fuels 22:2909–2916. https://doi.org/10.
1021/ef8002153
54. Lopez G, Olazar M, Aguado R, Elordi G, Amutio M, Artetxe M et al (2010) Vacuum pyrolysis
of waste tires by continuously feeding into a conical spouted bed reactor. Ind Eng Chem Res
49:8990–8997. https://doi.org/10.1021/ie1000604
55. Roy C, Chaala A, Darmstadt H (1999) The vacuum pyrolysis of used tires. J Anal Appl Pyrolysis
51:201–221. https://doi.org/10.1016/S0165-2370(99)00017-0
56. Pakdel H, Pantea DM, Roy C (2001) Production of dl-limonene by vacuum pyrolysis of used
tires. J Anal Appl Pyrolysis 57:91–107. https://doi.org/10.1016/S0165-2370(00)00136-4
57. Miranda M, Pinto F, Gulyurtlu I, Cabrita I (2013) Pyrolysis of rubber tyre wastes: a kinetic
study. Fuel 103:542–552. https://doi.org/10.1016/j.fuel.2012.06.114
58. Murugan S, Ramaswamy MC, Nagarajan G (2008) The use of tyre pyrolysis oil in diesel
engines. Waste Manag 28:2743–2749. https://doi.org/10.1016/j.wasman.2008.03.007
59. Harker JH, Backhurst JR (1981) Fuel and energy
60. Wieckert C, Obrist A, Von ZP, Maag G, Steinfeld A (2013) Syngas production by thermo-
chemical gasification of carbonaceous waste materials in a 150 kW th packed-bed solar reactor.
Energy Fuels 27:4770–4776. https://doi.org/10.1021/ef4008399
61. Pattabhi Raman K, Walawender WP, Fan LT (1981) Gasification of waste tires in a fluid bed
reactor. Conserv Recycl 4:79–88. https://doi.org/10.1016/0361-3658(81)90036-9
62. Donatelli A, Iovane P, Molino A (2010) High energy syngas production by waste tyres
steam gasification in a rotary kiln pilot plant. Experimental and numerical investigations. Fuel
89:2721–2728. https://doi.org/10.1016/j.fuel.2010.03.040
63. Zhang Y, Wu C, Nahil MA, Williams P (2015) Pyrolysis-catalytic reforming/gasification of
waste tires for production of carbon nanotubes and hydrogen. Energy Fuels 29:3328–3334.
https://doi.org/10.1021/acs.energyfuels.5b00408
64. Lerner AS, Bratsev AN, Popov VE, Kuznetsov VA, Ufimtsev AA, Shengel’ SV et al (2012)
Production of hydrogen-containing gas using the process of steam-plasma gasification of used
tires. Glas Phys Chem 38:511–6. https://doi.org/10.1134/S1087659612060041
65. Burra KG, Gupta AK (2018) Synergistic effects in steam gasification of combined biomass and
plastic waste mixtures. Appl Energy 211:230–236. https://doi.org/10.1016/j.apenergy.2017.
10.130
Energy Recovery from Waste Tires Via Thermochemical Pathways 521

66. Wu Z, Yang W, Tian X, Yang B (2017) Synergistic effects from co-pyrolysis of low-rank coal
and model components of microalgae biomass. doi:https://doi.org/10.1016/j.enconman.2016.
12.060
67. Zhang Y, Geng P, Liu R (2018) Synergistic combination of biomass torrefaction and co-
gasification: Reactivity studies. Bioresour Technol 245:225–233. https://doi.org/10.1016/j.bio
rtech.2017.08.197
68. Chen W, Shi S, Zhang J, Chen M, Zhou X (2016) Co-pyrolysis of waste newspaper with
high-density polyethylene: synergistic effect and oil characterization. Energy Convers Manag
112:41–48. https://doi.org/10.1016/j.enconman.2016.01.005
69. Déparrois N, Singh P, Burra KG, Gupta AK (2019) Syngas production from co-pyrolysis and
co-gasification of polystyrene and paper with CO2 . Appl Energy 246:1–10. https://doi.org/10.
1016/j.apenergy.2019.04.013
70. Ahmed II, Nipattummakul N, Gupta AK (2011) Characteristics of syngas from co-gasification
of polyethylene and woodchips. Appl Energy 88:165–174. https://doi.org/10.1016/j.apenergy.
2010.07.007
Biomass Conversion to Power
in Southeast Asian Countries: Current
Situation and Perspectives

Somrat Kerdsuwan and Krongkaew Laohalidanond

Abstract As fossil fuels are still the main primary energy source used among the
Association of Southeast Asian Nation countries (ASEAN), ASEAN member states
have the ambition to reduce their dependence on fossil fuel and to encourage the
utilization of renewable energy, which can be integrated into the energy mix. Hence,
alternative energy efficiency and renewable energy plans have been issued in ASEAN.
As almost all countries are agricultural-based economies, there are abundant agri-
cultural and forestry by-products or residues that can be used as lingo-cellulosic
biomass feedstock for heat and power generation, e.g., rice husks and straw, cassava
rhizome, palm oil shell, fronds and empty fruit bunches, corn stalks and empty cobs,
as well as sugarcanes tops, leaves, and bagasse. This lingo-cellulosic biomass can be
converted to power through direct combustion and indirect combustion technology.
The boilers generally used in the direct combustion of biomass power plants are
stoker-fired, fluidized bed and pulverized boilers, and equipped with cyclone, elec-
trostatic precipitator, or bag filters to control emissions. In contrast, indirect combus-
tion using gasification technology, e.g., a downdraft, updraft, or cross-draft gasifier,
is used to generate syngas which can run a gas or diesel engine to generate power
of less than 1 MWe . A suitable business model of a power plant is a small power
producer (SPP), where the installed power capacities are lower than 90 MWe . There
are many key drivers to achieve renewable energy (RE) goals, e.g., tax incentives and
feed-in-tariff power buying schemes. It is to believed that the increase in the portion
of RE in ASEAN can contribute to the depletion of fossil fuel and the emission of
greenhouse gas, finally to achieve sustainable development goals (SDGs).

Keywords Biomass · Power plant · Renewable energy · ASEAN countries

S. Kerdsuwan (B) · K. Laohalidanond

Department of Mechanical and Aerospace Engineering, Faculty of Engineering, The Waste
Incineration Research Center, Science and Technology Research Institute, King Mongkut’s
University of Technology North Bangkok, Bangkok, Thailand

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 523
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_22
524 S. Kerdsuwan and K. Laohalidanond

1 Introduction

According to the increase in the world population, more comfortable living standards,
and industrialization and globalization, it is inevitable that a great amount of energy
is consumed [1, 2]. There is a strong link between the global population and global
energy consumption, as graphically shown in Fig. 1, where it is evident that the
growth in the global population leads to an increase in global energy consumption.
In 2020, the world population is showing an increase of 46.15% from 1990, and the
world energy consumption has risen to up to 84.18% [3, 4]. It is predicted that the
global population and global energy consumption will increase in the future [3, 5].
The population growth and the final energy consumption in the ASEAN countries,
comprising Brunei Darussalam, Cambodia, Indonesia, the Lao People’s Democratic
Republic (Lao PDR, Laos), Malaysia, Myanmar, the Philippines, Singapore, Thai-
land, and Vietnam (Fig. 2) behave as other countries in the world, as mentioned
above. ASEAN’s population has been estimated to be 667.3 million in 2020, which
corresponds to a share of 8.56% of the world population in 2020. Regarding the
economical point of view, the GDP of ASEAN member states was 3 trillion US
dollars, with a share of 3.5% in the global GDP, in 2018 [6, 7]. With the total area of
4,522,945 m2 , the population density accounts for 147 inhabitants per square kilo-
meter, as listed in Table 1 [3, 8]. The United Nations [8] forecasted that ASEAN’s
population will continuously increase to 791.98 million by 2050, as shown in Fig. 3.
According to economic expansion, urbanization, and industrialization in ASEAN,
the final energy consumption has sharply increased over the past few decades, from
178.33 metric tons of oil equivalent (Mtoe) in 1990 to 625.07 Mtoe in 2020, and it is
projected based on the business-as-usual model (BAU) to be 1,152.71 Mtoe in 2040
[9], as illustrated in Fig. 4.

Fig. 1 Relationship between population and energy consumption from 1990 to 2050 [3–5]
Biomass Conversion to Power in Southeast Asian Countries… 525

Fig. 2 ASEAN members

Table 1 Area, population, and density of each country in ASEAN in 2020 [6, 10]
Country Area (sq km) Population Density GDPa (Billion US,
(Million, %) (Capita/sq km) %)
Brunei 5765 0.44 (0.07) 75.80 13.56 (0.45)
Cambodia 181,035 16.72 (2.51) 92.36 24.63 (0.82)
Indonesia 1,904,569 273.52 (40.99) 143.61 1.04 (34.88)
Laos 236,800 7.28 (1.09) 30.74 18.10 (0.61)
Malaysia 330,803 32.37 (4.85) 97.85 358.41 (12.00)
Myanmar 676,578 54.41 (8.15) 80.42 77.26 (2.59)
The Philippines 342,353 109.58 (16.42) 320.08 342.69 (11.48)
Singapore 710 5.85 (0.88) 8,239.44 364.08 (12.19)
Thailand 513,120 69.80 (10.46) 136.03 505.06 (16.91)
Vietnam 331,212 97.34 (14.39) 293.89 241.04 (8.07)
Total 4,522,945 667.31 147.54 2.99 (100.00)
(100.00)
a GDP in 2018

The main energy sources in ASEAN have changed from time to time. For example,
biomass and waste were a dominant energy source during the 1990s; however, due to
the rapid development in terms of industrialization, urbanization, and motorization,
the share of fossil fuel, e.g., oil, coal, and natural gas in the ASEAN energy mix
increased from 56.4% in 1990 to 75% in 2015, while the share of biomass and waste
decreased from 40% in 1990 to 19% in 2015. It is predicted that the share of fossil
526 S. Kerdsuwan and K. Laohalidanond

Fig. 3 ASEAN’s population and its projection to 2050 [8]

Fig. 4 ASEAN’s final energy consumption and its projection to 2040 [9]

fuel in primary energy supply will rise to 80% in 2050 and the share of biomass and
waste will decline to 9.2% by 2050 [11]. Figure 5 presents the primary energy supply
in ASEAN from 1990 to 2050 [11]. Among the energy sources, coal, oil, and natural
gas are the major energy sources in ASEAN.
According to the aforementioned, ASEAN countries will face a challenge in
energy security, as they rely on fossil fuel energy sources, which have limited world
reserves. In order to tackle this problem, the increase in the share of renewable
Biomass Conversion to Power in Southeast Asian Countries… 527

Fig. 5 ASEAN’s primary energy supply and its projection to 2040 [11]

energy, e.g., solar, wind, geothermal, hydro, biomass, etc., in the energy mix has been
encouraged. This chapter will deal with the using of biomass as a part of renewable
energy in ASEAN to reduce the fossil fuel by setting up the goal to achieve. The
technologies to convert biomass to power as well as business model used will be
discussed, together with the key drivers and perspectives to approach the setting up
target.

2 Renewable Energy in ASEAN

Not only the limitation of fossil fuel reserves but their use as the main energy source
leads to unfavorable impacts on the environment, particularly regarding greenhouse
gas (GHG) emissions, which is the key contributor to the global warming effect.
ASEAN has recognized the problem of both energy security and problems with the
environment, and hence, the ASEAN plan of action for energy cooperation (APAEC)
2016–2025 has been issued, by which the share of renewable energy of 23% in the
power sector, the transportation sector, and the industrial sector should be achieved
by 2025 [12]. Although some ASEAN countries produce electricity from renewable
energy sources, e.g., the countries along the Mekong River (a trans-boundary river
in East Asia and Southeast Asia and runs through China, Myanmar, Laos, Thai-
land, Cambodia, and Vietnam) have a 40–80% share of hydro energy for electricity
production and some countries produce electricity from geothermal, solar PV, and
528 S. Kerdsuwan and K. Laohalidanond

wind; almost all countries have strongly relied on fossil fuel for electricity produc-
tion, as seen in Fig. 6 [13, 14]. The current energy situation is far behind the target of
reaching a 23% share of renewable energy in energy mix by 2025. In order to reach
this target, renewable energy has been promoted at the national level though specific
key drivers. The renewable energy target compared to the achievement in 2017 and
the policy of each country are tabulated in Table 2.
In Table 2, the national policy is differentiated from country to country. The
countries on the Mekong riverside promote the utilization of hydropower, while the
use of geothermal energy is partly encouraged in Indonesia and the Philippines,
where the countries sit on the “ring of fire.” Solar, wind, and biomass, including
waste, and biogas are mostly fostered in Thailand and Vietnam.
The key drivers for the achievement of renewable energy targets can be cate-
gorized according to national policy, fiscal incentives, regulatory instruments, and
grid access [16, 17]. National policy is announced by the government in order to
reach renewable energy targets and related laws. Fiscal incentives encourage plant
developers to invest in renewable energy through the reduction of upfront costs, e.g.,
income tax exemptions, VAT exemptions, import/export fiscal benefits, accelerated

Fig. 6 Share of energy source for electricity production in 2017, exception for Laos, where the
information is based on the year 2015 [13, 14]
Biomass Conversion to Power in Southeast Asian Countries… 529

Table 2 National policy for promoting renewable energy in power generation [15–18]
Country Target Current achievement Key drivers
Brunei • 10% RE share in 0.05% Need to be developed
electricity production
in 2035
Cambodia • 2.2 GW (80%) n.a. Tax incentives
hydropower in 2020
Indonesia • 16.7 GW RE planed • 9.5 GW RE in 2018 Feed-in-tariff
new capacity during (hydro 5.6 GW,
2019–2028 (hydro geothermal 1.9 GW,
9.5 GW, geothermal solar 0.1 GW, wind
4.6 GW, solar 0.9 GW, 0.1 GW and biomass
wind 0.9 GW and 1.8 GW
biomass 0.8 GW
• 23% RE in 2025
• 31% RE in 2030
Laos • 0.7 GW RE in 2025 n.a. Tax incentives
(small hydro 0.4 GW,
biomass 0.1 GW, solar
0.1 GW and wind
0.1 GW
• 30% RE in total energy
consumption in 2025
• >15 MW hydropower
in 2015
Malaysia • 2.1 GW RE in 2020a 4% in 2017 Feed-in-tariff and
• 20% RE in 2025a capital subsidies
Myanmar • Hydro 8.9 GW, other • Hydro 3.3 GW, other Need to be developed
RE 2 GW (total 46% RE 0.1 GW in 2018
RE) by 2030–2031
The Philippines • 9.9 GW RE planed • 1.7 GW RE in 2018 Tax incentives
new capacity during (hydro 0.1 GW, feed-in-tariff and capital
2010–2030 (hydro geothermal 0.1 GW, subsidies
5.4 GW, geothermal solar 0.9 GW, wind
1.5 GW, solar 0.3 GW, 0.4 GW, and biomass
wind 2.3 GW, biomass 0.2 GW)
0.3 GW, and marine
0.1 GW
Singapore • Solar 0.35 GW in 2020 • Solar 0.2 GW in 2018 Tax incentives and
and 1 GW beyond feed-in-tariff
2020
Thailand • 30% RE in 2036 (solar • 10.4 GW RE in 2018 Tax incentives and
6 GW, hydro 3.3 GW, feed-in-tariff
wind 3 GW, biomass
6.78 GW, biogas
0.6 GW)
• 20.8 GW RE planed
new capacity during
2018–2037a
(continued)
530 S. Kerdsuwan and K. Laohalidanond

Table 2 (continued)
Country Target Current achievement Key drivers
Vietnam • 21% RE of 60 GW • 18.5 GW RE in 2018 Tax incentives and
installed in 2020 (hydro 18 GW, solar feed-in-tariff
• 13% RE of 96 GW in 0.1 GW, wind 0.2 GW,
2025 and biomass 0.2 GW)
• 23% RE of 130 GW in
2030 (small hydro
20.2 GW, Solar
4.3 GW, wind 2.7 GW,
and biomass 2.7 GW)
a Excluding large hydro

depreciation, etc. Regulatory instruments are the incentives provided to developers

in terms of feed-in-tariffs (FiT), grid access quotas, net metering, certificates, etc.
The grid access quota is to ensure that developers obtain access to the power distri-
bution line for renewable-based electricity [16, 17]. Additionally, the facilitation of
permission and license procedures can also motivate the investment in renewable
energy [16, 17, 19]. Table 2 shows the key drivers for the achievement of renewable
energy targets at the nation level.
Although there are many types of renewable energy sources to be executed in the
future, only the cost of electricity generation from biomass and hydro is competitive
against conventional fuel since energy conversion technologies have fully been devel-
oped and commercialized [20]. Besides cost-effective power production, biomass
plays an important role in global renewable energy and it is environmental friendly.
Since 2000, the biomass energy supply has continuously increased from 42.8 EJ in
2000 to 55.6 EJ in 2017, which corresponds to the share of 69–79% in total renewable
energy supply [21], as presented in Fig. 7. Considering the biomass energy supply in
Asia, it can be found that the share of bioenergy in renewable energy supply in 2017
accounted for 65% of 33.2 EJ, Fig. 8 [21]. A similar trend of bioenergy supply in
total renewable energy supply in ASEAN was reported with its share of 87% [22]. It
can be implied that ASEAN owns an abundant amount of bioenergy, both in terms
of agricultural residue and forestry residue, since agriculture products are the core
driver in economic growth in most ASEAN countries [6, 22], whose agriculture land
has a share ranging from 10% in Laos to more than 40% in Thailand and the Philip-
pines [23, 24]. According to the high bioenergy supply and cost-competitive power
generation, this chapter discusses biomass energy in ASEAN and its conversion
technology, especially regarding power generation.
Biomass Conversion to Power in Southeast Asian Countries… 531

Fig. 7 Global renewable energy supply in EJ [21]

Fig. 8 Renewable energy supply in EJ by continents and share of renewable energy in Asia [21]
532 S. Kerdsuwan and K. Laohalidanond

3 Biomass as Feedstock for Energy Conversion

3.1 Biomass Residue

Since most ASEAN countries are agro-economic, there are enormous agricultural
land and forestry land, as can be noticeably seen in Table 3, for the cultivation of
agriculture crops and forestry products.
The top-six ASEAN agricultural products are rice, cassava, palm oil, coconut,
corn, and sugarcane, with the annual production capacity of 115.4, 71.2, 60.1, 35.8,
33.7, and 18.5 million tons, respectively. Table 4 shows the average production

Table 3 Land use in each ASEAN country in 2011 (%) [24]

Country Agricultural land Forest land Other landa
Brunei 2.5 71.8 23.7
Cambodia 32.1 56.5 11.4
Indonesia 31.2 51.7 17.1
Laos 10.6 67.9 21.5
Malaysia 23.2 62 14.8
Myanmar 19.2 48.2 32.6
The Philippines 41.0 25.9 33.1
Singapore 1.0 3.3 95.7
Thailand 41.2 37.2 21.6
Vietnam 34.8 45.0 20.2
a Built-up areas, roads, and other transportation features, barren land, or wasteland

Table 4 Annual amount of agricultural crops in each ASEAN country during 2015–2019 (1000 t)
[25–28]
Country Rice Cassava Palm oil Coconut Corn Sugar cane
Brunei 1 3 0 0 0 0
Cambodia 5,445 7,646 0 68 915 223
Indonesia 35,552 16,119 38,300 17,970 11,440 2,095
Laos 1,901 2,279 0 0 1149 214
Malaysia 1,818 53 19,108 512 58 4
Myanmar 12,782 416 0 538 2,228 380
The Philippines 11,612 2,749 88 14,334 7,769 2,217
Singapore 0 0 0 0 0 0
Thailand 18,783 31,509 2577 896 5,145 11,974
Vietnam 27,471 10,441 0 1,495 4,951 1,440
Total 115,365 71,215 60,073 35,813 33,655 18,547
Biomass Conversion to Power in Southeast Asian Countries… 533

capacity of each agricultural crop by country during 2015–2019. It can be remark-

ably noticed that Brunei and Singapore have no agricultural products because their
economies do not rely on agriculture but on crude oil and natural gas for Brunei and
on the business and service sectors for Singapore [24].
As agricultural countries, there are substantial amounts of residue derived from
the above-mentioned crops, which will be discussed in greater detail.
Rice straw and rice husks are the residue from rice planting. Rice straw is the
stems and leaves left over in rice fields after the harvesting process. Although rice
straw can be used as fodder, fertilizer, mushroom growth carriers, etc., the utilization
of rice straw in ASEAN can hardly be found, with only 20%, 5%, and 1% of total rice
straw in Thailand, the Philippines, and Malaysia [29], respectively. In practical terms,
many countries in ASEAN directly return rice straw to the field as a fertilizer after
harvesting in order to clear the land for new planting [29, 30], e.g., 29.7–40.2% of rice
straw is burnt in the Philippines [31]. Rice husks are the outer rice shell that occurs
during the rice milling process. Rice husks have been considered as an energy source
for decades [32], since they are easily collected in rice mills and have a uniform and
homogeneous shape, which can be conveniently handled. Rice husks have served as
solid fuel for power generation in most ASEAN, such as Thailand, Malaysia, and
Cambodia [33–35]. Examples of rice straw and rice husks are illustrated in Fig. 9.
Cassava is the second-ranked agricultural crop mostly planted in ASEAN. Its
root is the raw material for the starch-derived industry, e.g., human food, animal
feed, and chemicals [36]. Additionally, the cassava root is recognized as an energy-
containing plant for the production of liquid biofuel via fermentation [36, 37]. The
residue from cassava is cassava rhizome (Fig. 10), which is unused and discharged
in the field. Some research has focused on the utilization of cassava rhizome for the
production of liquid biofuel through the pyrolysis process and solid biofuel through
the carbonization or torrefaction process [37, 38]. The use of cassava rhizome as fuel
in co-firing power plants can be found in Thailand [39–41].

Fig. 9 Paddy rice residue (a) straw and (b) husks

534 S. Kerdsuwan and K. Laohalidanond

Fig. 10 Cassava rhizome

Palm oil trees are widely cultivated in Indonesia, Malaysia, and Thailand. Palm
oil fruit bunches are harvested and delivered to the palm oil industry for crude palm
oil production. Not only used as raw material for the food industry, palm oil can
be used as raw material in the transesterification process for the production of fatty
acid methyl esters, which can be blended in diesel oil and served as bio-diesel for
compression ignition engines [42, 43]. The by-products from palm oil processing are
trunks, fronds, empty fruit bunches (EFBs), and shells [44, 45]. Palm oil shells are
completely used as fuel for steam and power generation in the palm oil manufacturing
process [46], while other by-products are fibrous materials, which are generally
utilized as raw material for pulp, paper, and particle board production, as well as for
energy production [43, 46]. Figure 11 shows the residue from palm oil tree and palm
oil mill.
Coconut is a multipurpose plant with a variety of uses, from high-grade food
ingredients, cosmetics, and chemicals to furniture and home decor [47, 48]. Coconut
fruit is harvested and processed for food production, e.g., coconut milk, coconut
oil, and coconut meat, as well as for some chemical manufacturing. Coconut wood
is used as tables, chairs, parts of houses, etc. The coconut residue comprises shell,
fiber, and frond. Shell and fiber can occur during the manufacturing process and
can be used for many purposes, e.g., as aggregate in concrete and road construction
material, as agricultural substrate, etc. On the other hand, coconut fronds are left
over in the field after harvesting. Therefore, there is an abundant amount of coconut
fronds available [49, 50]. A picture of coconut fronds is presented in Fig. 12.
Corn or maize is mostly planted for use as a primary source of animal fodder
production and bioethanol production [51]. The residue from corn is the stem, peel,
and empty corn cob [51, 52], as shown in Fig. 13. Similar to rice straw, corn stalks,
including stems and leaves, are left over in the fields and directly return to the field as
a fertilizer for the next planting [31]. Corn peels and empty corn cobs are by-products
Biomass Conversion to Power in Southeast Asian Countries… 535

Fig. 11 Palm oil residues (a) shells, (b) EFBs, (c) fronds, and (d) trunks

Fig. 12 Coconut fronds

536 S. Kerdsuwan and K. Laohalidanond

Fig. 13 Corn residue (a) stems and leaves, (b) peels, and (c) empty corn cobs

from the corn seed removal process. Corn stems, peels, and empty corn cobs can be
utilized as feedstock for energy production [51].
Sugarcane is a sugar-containing crop that is used in the sugar mill industry. The
residue from sugarcane is the tops, leaves, and bagasse (Fig. 14). Like rice straw and
corn stems, sugarcane tops and leaves (TLs) are burnt and directly returned to the
field [31]. Bagasse is a dry effluent from the sugar mill process after extracting the
juice. Bagasse originating in sugar mills is mainly used as fuel in situ for steam and
power generation [53, 54].
Table 5 summarizes the agricultural crop residue considered as biomass available
highlighted in this chapter. Palm oil shells and bagasse are not included in this chapter,
since almost all of them are completely used in situ in palm oil extraction plants and
sugar mills.
Forestry products also play a significant role as solid woody biomass, since they
can be used as fuel for traditional cooking and heating in the rural areas where
electricity cannot be accessed. The main forestry products considered in this chapter

Fig. 14 Sugarcane residue (a) tops and leaves and (b) bagasse
Biomass Conversion to Power in Southeast Asian Countries… 537

Table 5 Main agricultural

Agricultural crop Residue
residue considered as biomass
available in ASEAN Paddy rice Rice straw, rice husks
Cassava Cassava rhizome (CR)
Palm oil Empty fruit bunches (EFBs), fronds, stems
Coconut Fronds
Corn Stems, peels, empty cobs (ECs)
Sugarcane Tops and leaves (TLs)

Fig. 15 Forestry biomass (a) rubber trees and (b) eucalyptus trees

are rubber trees, eucalyptus trees, and white popinac trees [55], as shown in Fig. 15.
Apart from agricultural crops and woody biomass, there are some fast-growing,
cellulose-rich crops, e.g., cattail and napier grass, which can be used for energy
production [55].

3.2 Crop-to-Residual Ratio (CRR)

In order to estimate the amount of available biomass, it is essential to investigate the

crop-to-residual ratio (CRR), which is defined as the amount of unused residue per
one-unit mass of agricultural crop. The CRR has been investigated, based on a case
study of Thailand, by harvesting the agricultural crop from the planting area of 3
× 3 square meters and by weighing the amount of agricultural crop and its residue.
Finally, the CRR is calculated based on one kilogram of agricultural crop. Figure 16
shows the example of planting area for the investigation of CRR and its values are
listed in Table 6.
The amount of biomass can be estimated using Eq. (1) and the amount of biomass
from the top-six agricultural crops in ASEAN is shown in Table 7.
538 S. Kerdsuwan and K. Laohalidanond

Fig. 16 Example planting area for the investigation of CRR

Table 6 CRR: a case study

Agricultural crop Residue CRR
of Thailand
Paddy rice Straw 0.38
Husks 0.22
Cassava CR 0.12
Palm oil EFBs 0.22
Fronds 0.38
Stems 0.72
Coconut Fronds 1.00
Corn Stems 1.10
Peels 0.24
ECs 0.13
Sugarcane TLs 0.23

Table 7 Estimated annual

Agricultural crop Residue Amount (1000 t/year)
amount of biomass from the
top-six agricultural crops in Paddy rice Straw 43,838.70
ASEAN (average from 2015 Husks 25,380.30
to 2019) (1000 t)
Cassava CR 8,545.80
Palm oil EFBs 13,216.06
Fronds 22,827.74
Stems 43,252.56
Coconut Fronds 35,813.00
Corn Stems 37,020.50
Peels 8,077.20
ECs 4,375.15
Sugarcane TLs 4,266.04
Biomass Conversion to Power in Southeast Asian Countries… 539

Table 8 Annual amount of forestry products in each ASEAN country during 2015–2019 (1000 t)
[28]
Country Wood charcoal Wood pellets Wood chips and particles Wood residue
Brunei 0 0 0 0
Cambodia 37 0 0 0
Indonesia 664 134 1,788 388
Laos 93 0 0 0
Malaysia 29 396 788 1966
Myanmar 332 0 0 0
The Philippines 407 0 0 0
Singapore 5 0 0 0
Thailand 1,466 128 3,142 6,060
Vietnam 414 1,459 13,900 560

Amount of biomass = CRR × amount of agricultural product (1)

Considering woody biomass from forestry, the amount of some forestry products
by country was statistically recorded by the Food and Agriculture Organization of
the United Nations [28]. The average amount of some forestry products that can be
utilized as fuel is given in Table 8.

3.3 Biomass Properties

The Waste Incineration Research Center (WIRC) has investigated biomass properties
in Thailand [56]. Biomass has been collected and ground to the size of less than 1 mm
in diameter for fuel property determination. The fuel properties to be investigated in
this study are proximate analysis, ultimate analysis, and heating value and density,
which are shown in Tables 9, 10 and 11.
Proximate analysis and ultimate analysis as well as heating value were investigated
according to ASTM [57]. Bulk density was calculated using Eq. (2).

Bulk density = mass of biomass in container/volume of container (2)

540 S. Kerdsuwan and K. Laohalidanond

Table 9 Proximate analysis (% wt. dry basis)

Residue Moisturea VM FC Ash
Paddy rice Straw 44.06 67.31 16.23 16.46
Husks 14.99 63.52 18.28 18.20
Cassava CR 60.66 73.87 19.04 7.09
Palm oil EFBs 22.34 80.38 16.70 2.92
Fronds 56.65 72.04 20.13 7.83
Stems 65.54 76.36 18.22 5.42
Shellsb 12.00 77.50 18.52 3.98
Coconut Fronds 68.77 73.17 19.09 7.74
Corn Stems 53.64 72.65 16.62 10.73
Peels 37.28 75.82 21.17 3.01
ECs 47.79 77.40 20.23 2.37
Sugarcane TLs 48.91 73.58 19.86 6.56
Bagasseb 50.73 85.21 11.89 2.90
Rubber timber 22.99 77.57 21.11 1.32
Eucalyptus timber 38.32 76.48 22.71 0.81
White popinac timber 35.57 76.54 21.43 2.03
Cattail 72.39 74.97 16.19 8.84
Napier grass 86.92 64.94 20.82 14.24
a As received basis, VM Volatile matter, FC Fixed carbon
b From literature [58]

4 Biomass Conversion Technology

4.1 Technology

Biomass can be converted to heat and power through either the thermo-chemical,
biological, or mechanical process. The selection of a suitable process depends on the
biomass type and end product required, as shown in Fig. 17.
Since biomass available in ASEAN, listed in Table 7 is lingo-cellulosic material,
which mostly contains cellulose, hemicellulose, and lignin [59, 60]. The conver-
sion of lingo-cellulosic material to power is a thermo-chemical process, including
carbonization, gasification, and pyrolysis for intermediate fuel production and
combustion for heat and power production [22, 59, 61–63].
Rice husks, palm oil shells, and bagasse have successfully been utilized in
commercial-scale power plants in ASEAN for decades [64]. According to the expe-
rience of commercial-scale rice husk power plants [65], 1 ton of paddy rice can
produce approximately 650–700 kg of white rice and 220 kg of rice husks as by-
product. White rice is final product for sale as food and rice husks are used as fuel for
heat and power generation. With respect to energy consumption, a rice mill consumes
Biomass Conversion to Power in Southeast Asian Countries… 541

Table 10 Ultimate analysis (% wt. dry basis)

Residue C H N S O Ash
Paddy rice Straw 39.34 5.52 1.05 0.05 37.58 16.46
Husks 39.62 5.25 1.27 0.01 35.65 18.20
Cassava CR 45.11 5.98 1.38 0.02 40.42 7.09
Palm oil EFBs 49.29 6.97 0.48 0.00 40.34 2.92
Fronds 47.42 5.85 0.18 0.00 38.72 7.83
Stems 45.77 5.80 0.21 0.24 42.56 5.42
Shellsa 50.57 5.70 0.32 0.02 39.50 3.98
Coconut Fronds 43.75 6.05 0.00 0.00 42.46 7.74
Corn Stems 42.86 5.85 1.56 0.05 38.95 10.73
Peels 46.02 6.23 0.99 0.01 43.74 3.01
ECs 47.82 6.27 1.13 0.01 42.40 2.37
Sugarcane TLs 45.77 6.06 0.83 0.08 40.70 6.56
Bagassea 43.33 6.16 0.24 0.06 47.31 2.90
Rubber timber 49.64 6.22 0.11 0.00 42.71 1.32
Eucalyptus timber 50.35 6.34 0.40 0.00 42.10 0.81
White popinac timber 49.31 6.12 0.46 0.02 42.06 2.03
Cattail 45.78 5.96 0.63 0.14 38.65 8.84
Napier grass 42.64 4.46 0.78 0.03 37.85 14.24
a From literature [58]

30–60 kWh/t of paddy rice for the milling and drying process, where rice husks of
220 kg can produce approximately 90–125 kWh of electricity, as shown in Fig. 18.
Figure 19 illustrates the amount of palm oil and residue from the palm oil industry,
as well as the energy required for palm oil production. Approximately, 140–200 kg
of palm oil can be extracted from one ton of fresh fruit bunches. The by-products
from the processing are 600–700 kg of palm oil mill effluent, which can be fermented
to produce 20 cubic meters of biogas, 190 kg of fiber and shells, as well as 230 kg of
empty fruit bunches, which can be used as fuel to generate approximately 120 kWh
of electricity.
Likewise, bagasse as a by-product from sugar mills has proven to be capable of
being used as feedstock for commercial power plants. One ton of sugarcane can
generate approximately 100–121 kg of sugar and 290 kg of bagasse, which can be
utilized as fuel for the production of 100 kWh of electricity, while a sugar mill plant
needs 25–30 kWh of electricity and 0.4 t of steam for the production process, as
presented in Fig. 20.
The biomass power plant technology can be categorized according to direct
combustion and indirect combustion (gasification), depending on the plant scale.
A large-scale power plant with a power generation capacity greater than 1 MWe uses
direct combustion technology, while indirect combustion technology (gasification)
542 S. Kerdsuwan and K. Laohalidanond

Table 11 Bulk density and heating value

Residue Bulk density (kg/m3 ) HHV (kcal/kg) LHV (kcal/kg)
Paddy rice Straw 63.97 3,696.36 3,395.46
Husks 90.00 3,696.67 3,393.33
Cassava CR 193.00 4,090.00 3,764.17
Palm oil EFBs 80.00 4,910.00 4,550.00
Fronds 213.90 4,250.00 3,940.00
Stems 530.00 4,120.00 3,810.00
Shellsa 400.00 4,572.51 4,942.37
Coconut Fronds 288.40 3,900.00 3,570.00
Corn Stems 46.25 3,975.00 3,647.50
Peels 20.13 4,146.00 3,776.00
ECs 120.32 4,315.00 3,975.00
Sugarcane TLs 295.00 4,209.29 3,879.29
Bagassea 120.00 3,560.56 4,466.64
Rubber timber 306.37 4,390.00 4,060.00
Eucalyptus timber 357.00 4,460.00 4,113.33
White popinac timber 253.00 4,383.33 4,053.33
Cattail 169.00 4,280.00 3,970.00
Napier grass 166.00 4,240.00 3,440.00
a From literature [58]

Fig. 17 Biomass conversion pathways for power production. Adapted from [59]
Biomass Conversion to Power in Southeast Asian Countries… 543

Fig. 18 Power production capacity from rice industry. Adapted from [65]

Fig. 19 Power production capacity from palm oil industry. Adapted from [65]

is suitable for a small-scale biomass power plant with the capacity of less than 1 MW
of electricity.
In direct combustion technology, biomass is completely burnt with an excess
amount of air to generate heat, which can further be used for steam generation in the
boiler. The steam from the boiler drives the steam turbine, coupled with the generator,
to produce electricity. The flue gas from the process must be cleaned for the removal of
544 S. Kerdsuwan and K. Laohalidanond

Fig. 20 Power production capacity from sugar industry. Adapted from [65]

particulates, oxide of nitrogen, and oxide of sulfur using an electrostatic precipitator

(ESP) or bag filter in order to comply with emission regulations, which are required
to control only the particulates, SO2 and NOx [66, 67]. Boilers generally used in
biomass power plants are stoker-fired boilers (<50 MW of electricity), fluidized bed
boilers (30–300 MW of electricity), and pulverized boilers (>600 MW of electricity)
[68, 69]. The stoker-fired boiler is a conventional boiler for biomass and is still in use
today because it is reliable, convenient, and suitable for different types of feedstock
[68]. Compared to a stoker-fired boiler, the fluidized bed boiler provides higher
combustion efficiency because the heat-carrier medium, e.g., sand, inside the boiler
contributes to the continuous heat transfer; however, a fluidized bed boiler is suitable
for small-size and homogenous feedstock, and therefore the fuel preparation process
is required [68, 69]. The pulverized boiler is used in practical terms for coal-fired
power plants. When using biomass as feedstock in a pulverized boiler, the biomass
must be dried and finely ground [69]. Figure 21 is a schematic diagram of direct
combustion technology.
In contrast to the direct combustion process, biomass is incompletely burnt in the
sub-stoichiometric condition (inadequate amount of air) in the gasifier to produce

Fig. 21 Direct combustion technology

Biomass Conversion to Power in Southeast Asian Countries… 545

Fig. 22 Indirect combustion (gasification) technology

combustible gas (syngas) in the indirect combustion technology. The syngas must
be cleaned to meet the requirements of the engine before being used as fuel in an
internal combustion engine connected with a generator for electricity production.
There are normally three types of gasifiers used for power production from biomass:
an updraft gasifier, a downdraft gasifier, and a four-stage gasifier with a separate
reaction zone [70]. Compared to the updraft gasifier, the downdraft gasifier produces
lower tar- and particulate-containing syngas [58, 70]. Both the downdraft gasifier and
the updraft gasifier can more easily be operated and have lower investment costs than
the four-stage gasifier, in which four reaction zones—drying, pyrolysis, oxidation,
and reduction—take place in separate reactors. Resulting from the separation of the
reaction zones in different reactors, tar and volatile matter are oxidized in the gas
phase without contacting the solid material, and consequently, the tar can efficiently
be decreased [70]. The syngas cleaning process includes cyclone, cooling tower, wet
scrubber, and bag filter, as illustrated in Fig. 22.

4.2 Examples of Direct Combustion Technology

An example of a direct combustion biomass power plant is illustrated in Fig. 23.

This biomass power plant is located in Petchaboon province, northern Thailand, and
requires 954,025 t/year of bagasse as feedstock in a stoker-fired boiler to produce
27 MW of electricity. 15 and 4 MWe thereof are used in a sugar mill and in a power
plant, respectively, and the remaining 8 MWe is connected to a power distribution
grid and sold to the Provincial Electricity Authority (PEA). Particulates and any
pollutants are removed from the flue gas by an electrostatic precipitator before being
discharged to the atmosphere [71].
The next example is a biomass power plant that uses multi-fuel, including rice
husks and wood chips as feedstock. This biomass power plant is located in Nakhon
Ratchasima province in northeastern Thailand and purposes to produce 55 MW of
electricity (as designed). Rice husks are used as the main feedstock with the amount
of 310,100 t/year and wood chips serve as supplementary fuel with the amount of
47,800 t annually. Feedstock is introduced into the circulating fluidized bed (CFB)
boiler using sand as a heat carrier in order to produce 45 MW of electricity, 25 MWe
of which is distributed in a grid and sold to the Electricity Generating Authority of
Thailand (EGAT), 15 MWe of which is delivered to a rice mill and the rest, 5 MWe ,
546 S. Kerdsuwan and K. Laohalidanond

Fig. 23 Direct combustion bagasse power plant. Adapted from [71]

is used in a power plant. The flue gas from the boiler is cleaned in 2 steps: a hot
cyclone separator and ESP. A hot cyclone separator is connected next to the boiler,
where large particulates are trapped and re-circulated in the boiler for re-combustion.
ESP is used to remove fine particulates and any pollutants from the flue gas before
the flue gas is released through the stack. Figure 24 is a process flow diagram of a
multi-fuel power plant [71].

Fig. 24 Rice husk and wood chip in a multi-fuel power plant. Adapted from [71]
Biomass Conversion to Power in Southeast Asian Countries… 547

4.3 Examples of Indirect Combustion Technology

Unlike direct combustion technology, the indirect combustion (gasification) of

biomass has not been widely used on a commercial scale yet. Twelve pilot biomass
gasification plants for heat and power production in Thailand have been introduced
to demonstrate the development and the feasibility of this technology. Among the
12 gasification plants, 7 plants use a downdraft gasifier, 4 plants apply an updraft
gasifier, and only 1 plant uses 4-stage gasification: drying, pyrolysis, oxidation, and
reduction, with separate reactors. Regarding the biomass feedstock, forestry residue,
e.g., wood chips and wood waste, is used as feedstock in 7 of the 12 plants, 4 of
them utilize multi-fuel from forestry residue and agricultural residue, and only one
plant uses empty corn cobs as single feedstock. The end-use of 8 pilot plants is heat
as a substitute to liquefied petroleum gas (LPG) for application in the production
process and only 4 pilot plants aim to produce electricity with the small capacity of
250–350 kWe [70]. The details of the 12 biomass gasification pilot plants are given
in Table 12.

5 Biomass for Heat and Power in ASEAN: The Perspective

As already discussed, nowadays fossil fuel, e.g., coal, oil, and natural gas, is still the
main primary energy source in ASEAN. In order to reduce the dependence on fossil
fuel and to encourage the utilization of renewable energy, alternative energy efficiency
and renewable energy plans have been issued in ASEAN member states. In Sect. 1,
the final energy consumption was presented in a business-as-usual (BUA) model. In
this section, the final energy consumption is shown based on the alternative-policy-
scenario (APS) model, by which the total energy consumption is predicted using
available alternative energy efficiency and renewable energy plans as an assumption.
The final energy consumption BAU model in 2025, 2030, 2035, and 2040 is compared
to APS, as illustrated in Fig. 25. It can be seen that the final energy consumption
based on the APS model can decrease by 12–15% compared to the BAU model.
Considering the final energy consumption by source, the share of renewable
energy in final energy consumption will increase from 11.6% for the BAU model to
13.8% for the APS model by 2040. This means that after implementing alternative
energy efficiency and renewable energy plans, the utilization of renewable energy is
predicted to increase, as shown in Fig. 26.
As almost all countries in ASEAN, excluding Brunei and Singapore, are
agricultural-based economies, there are abundant agricultural and forestry by-
products or residue that can be used as lingo-cellulosic biomass feedstock for power
generation, e.g., rice husks and straw, cassava rhizome, palm oil shells, fronds and
empty fruit bunches, corn stalks and empty cobs, as well as sugarcane tops, leaves,
and bagasse. This lingo-cellulosic biomass can be converted to power through direct
combustion and indirect combustion technology. The installed capacity and projected
548 S. Kerdsuwan and K. Laohalidanond

Table 12 Twelve biomass gasification pilot plants in Thailand [70]

Feedstock Preparation Gasification Gas cleaning Application
Mixed wood chips Size reduction Downdraft with Cyclone 250 kWe power
air/steam Cooling tower production by
Venturi scrubber gas engine
Bio packed bed
filter
Empty corn cobs Size reduction Downdraft with air Cyclone 350 kWe power
Spray tower production by
Wet scrubber gas engine
Fabric filter
Wood chips Size reduction 4-stage Hot gas filtration 250 kWe power
gasification: Gas cooler production by
drying, pyrolysis, Scrubber combined heat
oxidation, and and power
reduction
Eucalyptus wood Size reduction Downdraft with air Cyclone 300 kWe power
Filter production by
gas engine
Wood waste from Size reduction Downdraft with air Cyclone Heat used in
furniture Scrubber, filter hotel
(25 toe/year)
Wood waste Size reduction Downdraft gasifier Cyclone Heat for
with air Wet scrubber fertilizer drying
Using plasma (84 toe/year)
torch
Wood chips Size reduction Updraft gasifier Gas cooler Heat for palm oil
with air Cyclone industry
Fabric filter (60 toe/year)
Wood chips, Size reduction Updraft gasifier Gas cooler Heat for rubber
empty corn cobs, with air Cyclone drying
bagasse, straw Fabric filter (60 toe/year)
Palm empty fruit Size reduction Updraft gasifier Gas cooler Heat for drying
bunches, husks, with air Cyclone palm fruit
chips Fabric filter (132 toe/year)
Wood waste from Size reduction Downdraft gasifier Cyclone Heat for ceramic
furniture with air Filter process
Double throat (205 toe/year)
Corncob, wood Size reduction Cross-draft and Heat for corn
chips updraft with air mill industry
(115 toe/year)
Empty corn cobs, Size reduction Downdraft with air Cooling tower Heat for steam
shells, coconut Scrubber, filter generation
fiber (192 toe/year)
Biomass Conversion to Power in Southeast Asian Countries… 549

Fig. 25 Final energy consumption BAU and APS. Adapted from [9]

Fig. 26 Share of renewable energy consumption BAU and APS. Adapted from [9]

target of power generation from biomass according to country policy are listed in
Table 13.
Brunei and Singapore have no biomass in the country, and therefore there is
no target on biomass power plant. However, other countries have the ambitious
target to replace their energy from fossil fuel with biomass, especially Thailand,
Indonesia, Malaysia, and the Philippines. Almost of biomass power plants use residue
550 S. Kerdsuwan and K. Laohalidanond

Table 13 Installed capacity and projected target of power generation from biomass according to
the country’s policy [9, 17, 22, 71–78]
Country Installed capacity Projected target
Brunei 0 No target
Cambodia 23 MW 74 MW by 2040
Indonesia 1856 MW in 2018 5500 MW by 2025
Laos 39.7 MW in 2016 58 MW by 2025
Malaysia 330 MW in 2015 800 MW by 2020
1340 MW by 2030
Myanmar Very few 147 MW in 2035
17% share in electricity production in 2050
The Philippines 220 MW in 2015 192.31 MW additional capacity in 2018
313.42 MW additional capacity in 2019-2022
500 MW additional capacity in 2023-2040
Singapore 0 No target
Thailand 2451.82 MW in 2014 5570 MW in 2036
Vietnam 1% share in electricity production in 2020
~1.2% share in electricity production in 2025 ~2.1%
share in electricity production in 2030

from the production line as feedstock in order to produce heat and power for in-
house utilization, and the excess power is sold to the national grid line. These are
considered as small power producers (SPP), where the installed capacities are lower
than 90 MWe . With key drivers such as tax incentives and the feed-in-tariff power
buying scheme, it is believed that ASEAN member states can achieve the target of
power generation from biomass in order to increase the portion of RE in ASEAN,
as well as to reduce the utilization of fossil fuel and the emission of greenhouse gas
to achieve the sustainable development goals (SDGs) as suggested by the United
Nations.

References

1. Ram A, Saurabh T, Ashwani KA, Pankaj K (2019) Population-urbanization-energy Nexus: a

review. Resources 136(8)
2. Yoshihisa S (2007) Energy consumption: an environmental problem. IEEJ Trans 1:12–16
3. Worldometer (2020) World population by year. https://www.worldometers.info/world-popula
tion/world-population-by-year/
4. Enerdata (2020) Global energy statistical yearbook 2019. https://yearbook.enerdata.net/
5. US Energy Information Administration (2020) EIA projects nearly 50% increase in world
energy usage by 2050, led by growth in Asia. https://www.eia.gov/todayinenergy/detail.php?
id=41433
6. ASEAN Secretariat (2019) ASEAN Key Figures 2019
7. The World Bank GDP (Current US). https://data.worldbank.org/indicator/NY.GDP.MKTP.CD
Biomass Conversion to Power in Southeast Asian Countries… 551

8. United Nation (2020) World population prospects 2019. https://population.un.org/wpp/Dow

nload/Standard/CSV/
9. ERIA Energy outlook and energy saving potential in East Asia 2019
10. ASEANStatsDataPortal Gross domestic product (GDP) in ASEAN, at current prices (nominal),
in US dollars (Annually; Quarterly). https://data.aseanstats.org/
11. Endang JMS, Tan CS, Goh CH (2019) Energy security in ASEAN region: its challenges. In:
IOP conference series: earth and environment science, vol 268
12. Visal V, Beni S, Alysius DP, Nadhilah S (2020) A review of renewable energy development
and its policy under nationally determined contrinutions in ASEAN. Int J Smart Grid Clean
Energy 9(1):149–161
13. International Energy Agency (2020) Data and statistics, data table. https://www.iea.org/data-
and-statistics/data-tables?country=VIETNAM&energy=Electricity&year=2017
14. ERIA (2018) Lao PDR energy statistics
15. Renewable Energy Institutes (2019) Renewable energy to replace coal power in Southeast Asia,
Pragmatism to deliver a sustainable bright future
16. IRENA (2018) Renewable energy market analysis, Southeast Asia
17. ERIA (2018) Distributed energy system in Southeast Asia
18. Ministry of Energy (2020) Thailand power development plan 2015–1036 (PDP 2015). https://
www.egat.co.th/en/images/about-egat/PDP2015_Eng.pdf
19. ASEAN Center for Energy (2016) ASEAN renewable energy policies
20. ASEAN Center for Energy (2019) Levelised cost of electricity (LCOE) for selected renewable
energy technologies in the ASEAN member state II
21. World Energy Association (2019) Global bioenergy statistics
22. Maw M, Dagmar J, Myo MW, Aung MT, Tomáš P (2019) Biomass energy: an overview of
biomass sources, energy potential, and management in Southeast Asian Countries. Resource
81(8)
23. OECD/FAO (2017) OECD-FAO agricultural outlook 2017–2026
24. Central Intelligence Agency (2010) The world factbook. https://www.cia.gov/library/publicati
ons/the-world-factbook/fields/288.html
25. United States Department of Agriculture (2020) Foreign agricultural service. https://ipad.fas.
usda.gov/countrysummary/Default.aspx?id=VMSjdd
26. World Data Atlas (2020) Agriculture, crop production. https://knoema.com/atlas/topics/Agr
iculture
27. Statista (2020) Farming. https://www.statista.com/statistics/658221/asia-pacific-cassava-pro
duction-by-country/
28. (2020) Food and Agriculture Organization of the United Nation, FAOSTAT. http://www.fao.
org/faostat/en/#data/FO
29. Rosmiza MZ, Davies WP, Rosniza ACR, Mazdi M, Jabil MJ (2014) Farmers’ knowledge on
potential uses of rice straw: an assessment in MADA and Sekinchan, Malaysia. Malays J Soc
Space 10(5):30–43
30. Danutawat T, Nguyen TKO (2007) Effects from open rice straw burning emission on air quality
in the Bangkok Metropolitan region. ScienceAsia 33:339–345
31. ESCAP-CSAM (2018) Status of straw management in Asia-Pacific and options for integrated
straw management. http://www.un-csam.org/Publication/StatusOfStrawMgrAP_final_31July
2018.pdf
32. Kuen SL, Paul HW, Lin CJ, Ching IJ A process development for gasification of rice husk. Fuel
Process Technol
33. Jittima P, Shabbir HG (2017) Sustainable utilization of rice husk ash from power plants: a
review. J Clean Prod 167:1020–1028
34. N.a. (2020) Resource efficient and cleaner production for sustainable rice milling
in Cambodia. https://prospernet.ias.unu.edu/wp-content/uploads/2012/09/SPC-learning-case-
10_teaching-note_final.pdf
35. Shafie SM, Mahlia TMI, Masjuki HH, Rismanchi B (2012) Life cycle assessment (LCA) of
electricity generation from rice husk in Malaysia. Energy Procedia 14:499–504
552 S. Kerdsuwan and K. Laohalidanond

36. Reinhardt H, Keith F (2008) Cassava production and utilization in Asia and its potential as a
bio-fuel. In: NexGen bio-ethanol conference, Bangkok, Thailand, April 22–24
37. Nakorn T, Prasert R, Pruk A, Adisak P (2017) Biochar production from cassava rhizome in a
semi-continuous carbonization system. In: 4th International Conference on Power and Energy
Systems Engineering, CPESE 2017, Berlin, Germany, September 25–29
38. Suntorn S, Keartisak S, Adisak P (2012) Production of bio-oil via fast pyrolysis of cassava
rhizome in a fluidised-bed reactor. Energy Procedia 14:668–673
39. The Consulting Engineer Association of Thailand (2020) Study and design of power plant
using cassava rhizome fired. http://www.ceat.or.th/2010/index.php/energy/288-study-and-des
ign-of-power-plant-using-cassava-rhizome-fired-.html
40. Power Solution Technologies Public Company Limited (2020). http://www.pst.co.th/produc
tGroup-78-view.html
41. Consultants of Technology Company Limited (2020). https://www.cot.co.th/th/project/study-
and-design-of-cassava-rhizome-fired-power-plant-eastern-and-northeastern-regions-of-tha
iland/
42. Ogan IM, Marie-Josée D, Michael N (2015) Palm oil: processing, characterization and
utilization in the food industry—a review. Food Bioscience 10:26–41
43. Sheil D, Casson A, Meijaard E, Van Nordwijk M, Gaskell J, Sunderland-Groves J, Wertz K,
Kanninen M (2009) The impacts and opportunities of oil palm in Southeast Asia: what do we
know and what do we need to know? https://www.cifor.org/publications/pdf_files/OccPapers/
OP-51.pdf
44. Muhammad JI, Azizah B, Farah HA, Rizafizah O (2018) Palm oil industry in South East Asia
and the effluent treatment technology—a review. Environ Technol Innov 9:169–185
45. Kerdsuwan S, Laohalidanond K (2011) Renewable energy from palm oil empty fruit bunch.
In: Renewable energy—trends and applications, IntechOpen Publisher, pp 123–150. ISBN:
978-953-307-939-4
46. Shuit SH, Tan KT, Lee KT, Kamaruddin AH (2009) Oil palm biomass as a sustainable energy
source: a Malaysian case study. Energy 34:1225–1235
47. Ahuja SC, Siddharth A, Uma A (2014) Coconut—history, uses, and folklore. Asian Agri-
History 18(3):221–248
48. Romulo NA (2009) The situation and prospects for the utilization of coconut wood in Asia and
Pacific. http://www.fao.org/3/a-am610e.pdf
49. Neetesh K, Dilip K (2014) Utilization of coconut shell in different forms in concrete. Int J Sci
Res Dev 2(7):158–160
50. Tay LT, Ramadhansyah PJ, Norhidayah AH, Haryati Y, Dewi SJ (2015) A review of utilization
o coconut shell and coconut fiber in road construction. Jurnal Teknologi (Sci Eng) 76(14):121–
125
51. Elpiniki S, Alexandra S, Georgios C, Nicholaos D (2020) Maize as energy crop. https://www.
intechopen.com/books/maize-production-and-use/maize-as-energy-crop
52. Otchere AG, Hagan EB (2014) Potential for electricity generation from maize residues in rural
Ghana: a case study of Brong Ahafo Region. Int J Renew Energy Technol Res 3(5):1–10
53. Yasujiro W (2003) Utilization of bagasse energy in Thailand. Mitig Adapt Strat Glob Change
8:253–260
54. Mitr Phol Bio Power (2020) Case study: bagasse cogeneration development in Thailand’s
sugar industry. http://www.fao.org/fileadmin/templates/rap/files/meetings/2014/140723-d1s3.
Bagasse.pdf
55. Saranya Y (2004) Bioconversion of alpha cellulose from weeds to ethanol. https://cuir.car.
chula.ac.th/xmlui/handle/123456789/27065 (in Thai)
56. Kerdsuwan S, Loahalidanond K (2016) Availability and characteristic studies of using 23
biomass residues for developing of distributed community power plants. In: European biomass
conference and exhibition proceedings, pp 185–188
57. ASTM International
58. Loahalidanond K (2008) Theoretische Untersuchungen und thermodynamische Model-
lierungen der Biomasse vergasung und der Fischer-Tropsch-Synthese zur Herstellung von
Biomass Conversion to Power in Southeast Asian Countries… 553

Dieselkohlenwasserstoffen aus thailändischen Biomassen. Dissertation. RWTH Aachen

Univerity
59. Daniela T, Martin D, Volker L, Jan L, Andreas O (2015) Flexible bioenergy supply for balancing
fluctuating renewables in the heat and power sector—a review of technologies and concepts.
Energy Sustain Soc 5(35)
60. Department of Alternative Energy Development and Efficiency, Ministry of Energy Thailand
(2020) Feasibility study on cellulosic ethanol production in commercial scale. (in Thai)
61. Kusch S, Maria VM (2009) Integration of lignocellulosic biomass into renewable energy
generation concepts. ProEnvironment 2:32–37
62. Kerdsuwan S, Laohalidanond K (2015) Advancements in biomass and waste gasification for
power generation. In: Handbook of clean energy systems. Wiley
63. Gupta AK, Kerdsuwan S (2014) Efficient energy conversion of wastes and fuels in power
systems. Appl Sci Eng Prog
64. (2014) The working group for bioenergy science technology and innovation policy for Thailand
in the context of AEC, ASEAN bioenergy technology status report 2014
65. EC-ASEAN COGEN Programme (COGEN 3) (2004) Applicability of European Technologies
in ASEAN market
66. (2020) Emission standard, Stationary engines-World Bank guidelines. https://dieselnet.com/
standards/inter/wbank.php#emstd
67. International Finance Cooperation, World Bank Group (2008) Environmental, health, and
safety guidelines-Thermal power plants. (2020) https://www.ifc.org/wps/wcm/connect/f82
a5f06-f3f7-4033-8ea6-b767523cda8e/FINAL_Thermal%2BPower.pdf?MOD=AJPERES&
CVID=jqeD9Eg&id=1323162579734
68. Arnie I (2020) Biomass equipment options for steam and power. http://biomassmagazine.com/
articles/2470/biomass-equipment-options-for-steam-and-power
69. Wolf JP, Dong N (2013) Biomass combustion for power generation: an introduction: biomass
combustion science, technology and engineering. Elsevier Inc.
70. Krongkaew L, Palita C, Somrat K (2015) Status of using biomass gasification for heat and
power in Thailand. Energy Procedia 79:385–390
71. (2020) Report on environmental impact assessment (EIA) and monitoring. (in Thai) http://eia.
onep.go.th/
72. Trois D (2019) Bioenergy power plant in Indonesia: rules and implementation. German-
Indonesia Business Forum, Jakarta
73. Phonepasong S (2017) Bioenergy use in Lao PDR. In: Regional workshop on sustainable
bioenergy production and use in Southeast Asia: lessons learnt and future opportunities, 16–17
November 2017, Hanoi, Vietnam
74. (2013) National biomass strategy 2020: new wealth creation for Malaysia’s biomass industry.
https://www.cmtevents.com/MediaLibrary/BStgy2013RptAIM.pdf
75. Asian Development Bank (2016) Myanmar energy sector, assessment, strategy and
road map. https://www.adb.org/sites/default/files/institutional-document/218286/mya-energy-
sector-assessment.pdf
76. Ruby BDG Biomass sector roadmap. https://www.doe.gov.ph/sites/default/files/pdf/announ
cements/iloilo-a-bs3-02-biomass-sector-roadmap.pdf
77. Ministry of Energy Thailand (2015) Alternative energy development plan: AEDP2015. http://
www.eppo.go.th/images/POLICY/ENG/AEDP2015ENG.pdf
78. Approval of the revised national power development master plan for the 2011–2020 period
with the vision to 2030. Hanoi, 18 March 2016. https://policy.asiapacificenergy.org/sites/def
ault/files/PDP%207%20revised%20Decision%20428-QD-TTg%20dated%2018%20March%
202016-ENG.pdf
Role of Artificial Roughness
in the Performance Improvement
of Solar Air Heaters

Navneet Arya, Rajneesh Kumar, and Varun Goel

Abstract Solar radiation is a free, abundant, and clean source of thermal energy. A
solar air heater (SAH) is a simple device which acts as a heat exchanger in a way that
it extracts thermal energy from the solar radiation. There are various applications in
which SAH is used such as space heating and crop drying. As the performance of
SAH is low due to less heat transfer coefficient of air which acts as a working fluid in
SAHs. Among different techniques, artificial roughness is considered to be the best
technique for improving the performance of SAH. Providing roughness element in
the direction of flow will help in creating local turbulence which helps in augmenting
the heat transfer. Due to increase in turbulence, friction also increases in the flow
channel. So, roughness may be provided in such a manner that no significant rise in
friction takes place. A lot of roughness geometries were investigated experimentally,
numerically to see the impact of roughness elements on heat transfer and friction. The
impact of roughness elements on the performance of SAH has been studied for rough-
ness geometry using computational fluid dynamics (CFD). It is a powerful numerical
tool which can be utilized to study the flow structure near roughness elements in solar
air heaters. This chapter provides the comprehensive information about the different
cross-sectional roughness elements, and the effect of these roughness elements on
the performance of SAH has also been discussed.

Keywords Roughness elements · Solar air heater · Flow structure · CFD · Heat
transfer

Nomenclature

havg Convective heat transfer coefficient

N. Arya · V. Goel
Mechanical Engineering Department, National Institute of Technology
Hamirpur, Hamirpur 177005, India
R. Kumar (B)
Mechanical Engineering Department, CGC Chandigarh Engineering College, Landran (Mohali),
Ajitgarh, Punjab, India

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 555
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_23
556 N. Arya et al.

k Thermal conductivity
W Width of absorber plate
H Duct height
v Velocity

Abbreviations

CFD Computational Fluid Dynamics

f penalty Frictional penalty
Nuenh Nusselt number enhancement
SAH Solar air heater
TPP Thermohydraulic performance parameter

1 Introduction

1.1 Energy Scenario

Energy is the one of the essential component for living comfortable life. Most of our
day-to-day activities are accomplished with the use of energy. Population growth is
one of the main reason which significantly effects the amount of energy consumption
[1].
The International Energy Agency (IEA) published a report tilted, ‘World Energy
Outlook’ in 2012 and in which the total energy supply was estimated to 532.5 EJ, and
among this, 53.8% is harvested from fossil fuels. Prediction shows that the global
annual energy demand going to be increased to 778 EJ by 2035 [2]. It is also being
expected that over the next two decades, global energy demand is increased by 50%
[3], and for meeting the expected energy demand, rate of consumption of conventional
fuels (i.e., fossil) will increase drastically. Beside this, the bigger challenge is faced
by the energy sector, specially oil and gas companies. It is estimated that these
companies consume one-tenth of the produced energy [2].
The biggest problem that is associated with the fossil fuels is the degradation of
environment because of the emission of greenhouse gases (GHG) which leads to
global warming. In the next two decades, it is expected that there is 40% increase
in the CO2 emissions, and it would increase to 44.1 Gt by 2035 from 31.2 Gt in
2011. The CO2 emission is also accompanied with the other greenhouse gases, and
therefore, it is estimated that other greenhouse gases increased up to 1000 ppm
equivalent of CO2 with the increase of CO2 emission. With this, the average global
temperature is expected to be increased by 6 °C [2].
Role of Artificial Roughness in the Performance Improvement … 557

1.2 Importance of Solar Energy

The sun is the power source of the energy in our solar system, and the radiation
comes from the sun supports the life on the earth. The solar radiations formed due
to 5.5 × 103 K temperature on the sun surface which is resulted due to fusion of
hydrogen [4, 5].
Solar energy is a clean and sustainable source of energy which is freely available
in on the Earth during sunshine. Approximately, 1.6 × 105 terawatts (TW) of energy
are received from the sun in terms of solar radiation. Among them, 5 × 104 TW are
reflected back to the atmosphere and rest is utilized for performing various activities
on Earth such as photosynthesis, wind and wave conversion, and solar heating or
ocean thermal energy [6] as shown in Fig. 1.
As per the report published by IEA in 2011, it is claimed that India receives 5 ×
103 trillion kWh of solar energy, and it is interesting to see that it is much higher
than the annual global consumption of energy [7, 8]. Understanding the importance
of solar energy, Government of India had taken an initiative and in 1992 established
an exclusive ministry, i.e., Ministry of New and Renewable Energy (MNRE), which
devoted to promote use of renewable energy and technology associated with it [9].
Beside this, a non-banking financial institution has been established under MNRE
to provide financial assistance for projects which are dedicated for development of
sustainable energy-efficient components.

Fig. 1 Occurrence of
various activities from the
solar energy (units 1012 W)
558 N. Arya et al.

1.3 Introduction to Solar Air Heater

Solar air heater (SAH) is a type of heat exchanger which helps in extracting heat from
the solar radiations and utilized same to heat the air. A SAH consists of a glass plate,
absorber plate (usually painted black to increase heat absorbing capacity), and both
of them are encased in a wooden box in such a way that the glass plate is exposed to
the direct sun light, while the absorber plate faces the glass plate. The air is guided
to flow either between the glass and absorber plate or at the back of absorber plate
depending on the design. Figure 2 shows the schematic of SAH. There is various
application in which SAH can be used such as drying corps or dry fruits or herbs,
heating living space, and drying timbers. [10–15].
There are many factor on which it can be concluded that the SAH is advantageous
over the other solar energy harvesting devices such as solar water heater (SWH), and
those factors are:
1. The fabrication of system is simple and compact.
2. Only one working medium is required, i.e., air.
3. As working fluid is air, there is no corrosion.
4. Leakage of working fluid does not cause serious effect on the environment, apart
from low performance of SAH.
5. There is marginal variation in the pressure in the system and the surrounding.
6. In comparison with the SWH, maintenance of SAH is cheaper.
Among these advantages, there are some demerits are also associated with the
SAH, and they are highlighted below:
1. The main problem observed in the SAH is associated with its performance
because of low heat transfer coefficient of air.

Fig. 2 Schematic of conventional SAH

Role of Artificial Roughness in the Performance Improvement … 559

2. As there is marginal change in the air density, the large amount of working fluid
needs to be handled.
3. Due to working medium, the use of SAH is limited to temperature values.

2 Artificial Roughness and Its Importance in SAH

The working medium used in SAH is air, and the heat transfer coefficient obtained
between the absorber plate and air is low which leads its poor performance. There
are different techniques available in the literature which can be used to increase heat
transfer coefficient in SAH as presented in Table 1. All these techniques improves
heat transfer either by increasing heat transfer area or by increasing heat transfer
coefficient of air.
The artificial roughness is one of the simplest and easiest ways of improving heat
transfer through the SAH. In this technique, the roughness elements are either pasted
or directly fabricated over absorber plate, and it is assumed that the area of roughness
element is negligible in comparison with the area of absorber plate. Therefore, it can
be concluded that the heat transfer is improved by providing roughness over the
absorber plate, and this improvement in heat transfer is achieved because of creating
local turbulence which results in local heat transfer coefficient increment, and hence,
heat transfer through the SAH increases. Initially, this technique was used in cooling
of turbine blade passages in which roughness is placed on both sides of the passage.
But, in case of SAH, the roughness is provided only on single side where absorber
plate is placed. With the time, this technique emerged as most effective method for
increasing the performance of SAH.
The formation of viscous sub-layer is one of the main reasons that causes low heat
transfer in SAH. These layers oppose transfer of heat from air to plate (absorber),
and it leads to higher temperature zone near the absorber plate. For improving the
heat transfer through the SAH, it becomes important to break these layers. The main

Table 1 Different techniques for improving performance of SAH

Method Principle
Multi-glazing By reducing losses from glass cover
Multi-passing By providing more time to air for the heat transfer
Recycle Re-heating of air by re-directing the conventional air outlet again into
SAH and forced to flow at the back of absorber plate
Jet impingement Making air jet to cause higher local turbulence in the SAH for
improving SAH performance
Winglets Creates vortices in the flow domain for increasing interaction with the
absorber plate
Fins By increasing the area for heat transfer
Artificial roughness Break-up of viscous sub-layers which resulted flow separation along
with secondary flow generation
560 N. Arya et al.

purpose of providing artificial roughness over absorber plate is to break viscous

sub-layer in such a way that the effective heat transfer takes place within the SAH.

2.1 General Roughness Parameters

The size and arrangement of the roughness over the absorber plate are decided on the
basis of roughness parameters. As effective design of SAH is concerned, selection of
roughness parameter value is very important. Generally, relative rib height (e/D) and
relative rib pitch (P/e) are two important roughness parameters, and these roughness
parameters can be non-dimensionalized as relative rib height (defined as the ratio
between rib height and hydraulic diameter; e/D) and relative rib pitch (defined as ratio
between rib pitch and rib height; P/e). However, there are other roughness parameters
as well such as rib aspect ratio, rib attach angle, relative chamfered height, etc. But,
the impact of roughness also depends upon the orientation/arrangement of the rib
over the absorber plate.
With the variation of roughness parameters, there is noticeable changes which
can be seen in heat transfer and flow characteristics. Actually, there is a formation of
viscous sub-layer over the absorber plate under turbulent flow which acts as barrier in
heat transfer from absorber plate to air flowing through the SAH. The e/D parameter
helps in breaking of viscous sub-layers, and for this, it is provided to have rib height
greater than the viscous sub-layer thickness. The formation of viscous sub-layers
under turbulent flow condition on the absorber plate can be significantly affected
with the change in relative rib height (e/D). While the flow separation and occurrence
of reattachment points vary with the variation of P/e value for the rib element,
there is tremendous variation in heat transfer which is observed along with the flow
characteristics in SAH. Therefore, it can be concluded that both these parameters are
very important for designing of energy-efficient SAH.

3 Literature Review

In recent years, a lot of research has been carried out for improving the thermal and
hydraulic performance of SAH. Though, the air has low heat transfer coefficient, but
beside this, there is development of laminar sub-layers over the absorber plate which
acts as a barrier in transfer of heat from absorber plate to the air flowing through
the SAH [16–18], and therefore, it poses poor performance. There are different
techniques which can be used for improving SAH performance, and these techniques
are presented in Table 1. Among all the techniques, artificial roughness is more
promising for heat transfer improvement [16].
In present section, the literature review is carried out on artificial roughness
depending on the shaped rib roughness geometries which can be employed over
the absorber plate for improving the performance of (SAH). The schematic view of
Role of Artificial Roughness in the Performance Improvement … 561

Fig. 3 Schematic view of the SAH with rib elements

SAH with rib roughness elements is shown in Fig. 3. The roughness element can be of
various shapes, and these rib elements can be arranged in repeated or periodic pattern
over the absorber plate. Figure 4 shows the schematic view of basic geometries of
the roughness elements.
In this current chapter, the literature review has been carried out using different
cross-sectional roughness elements at a time, and different possible patterns consid-
ered by the researchers have been presented in Figs. 5, 6, 7, and 8 for circular, square,
rectangular, and chamfered ribs, respectively. It is seen that transverse rib pattern is
very popular in which the direction of air is perpendicular to the rib element, while
some complex patterns such as arc shape, V-shape, and S-shape were also inves-
tigated by the researchers. Under the following headings, the literature review has
been performed:
1. Circular cross-sectional ribs and their different patterns
2. Square and rectangular cross-sectional ribs and their different patterns
3. Chamfered al ribs and their different patterns.
In conventional SAH, there is occurrence of low temperature gradient over the
absorber plate which leads to low heat transfer coefficient or heat transfer. The heat
transfer coefficient can be increased by creating local turbulence in the flow field,
and therefore, the repeated rib roughness is provided over the absorber plate. These
repeated rib roughness helps in developing secondary flow (caused due to flow sepa-
ration, vortices and reattachment of viscous sub-layers) around the rib elements along
with the main flow, and these associated flows play vital role in the flow dynamics and
heat transfer within the SAH and thus the performance of SAH. Furthermore, these
flows around the roughness has strong dependency on the shape and pattern of rib
elements which is provided over the absorber plate, and it resonates in augmentation
of heat transfer along with pressure drop.
562 N. Arya et al.

Fig. 4 Basic shape of rib roughness

3.1 Circular Cross-Sectional Ribs and Their Different

Patterns

The circular cross-sectional rib element is widely used by the researchers in SAH
for improving performance because it can be easily fabricated. Different roughness
patterns were proposed and studied by the researcher over the absorber plate as it is
shown in Fig. 5. The transverse arrangement of circular cross-sectional rib element
is most common [19–27], while in some studies rib element are arranged in complex
patterns as well such as single or multiple V-pattern with or without gaps [28, 29],
single or multiple arc-shaped with or without gaps [30–32], innovative arc-shaped
[33], etc. The details of parameters considered by different researcher for various rib
patterns is provided in Table 2.
Role of Artificial Roughness in the Performance Improvement … 563

Fig. 5 Different patterns which are studied by the researchers using circular cross-sectional
roughness
564 N. Arya et al.

Fig. 6 Different patterns which are studied by the researchers using square cross-sectional
roughness

The rib top surface plays an important role in the separation of free-shear layer
which occurred by placing rib elements in the flow direction. The separated free-shear
layer glides over the rib-top surface and results improvement in performance of SAH.
In case of circular cross-sectional rib elements, the rib-top surface area is negligible,
and thus, the augmentation of heat transfer is poor. Furthermore, these ribs provide
low pressure drop as well which leads to low friction factor (f ) value. In comparison
with smooth SAH, the highest improvement in heat transfer and friction penalty
(f penalty ) is claimed by Prasad and Saini [20], i.e., 2.38 and 3.39, respectively, for
e/D = 0.033, P/e = 10. While, the highest thermohydraulic performance parameter
(TPP) value is claimed by Verma and Prasad [19], i.e., 1.71.
As there is no effective separation of shear layer takes place due to circular rib
element, the researchers have adopted different rib patterns over the absorber plate
to enhance flow separation. Bhardwaj et al. [27] had used the inclined rib element
Role of Artificial Roughness in the Performance Improvement … 565

Fig. 7 Different patterns which are studied by the researchers using rectangular cross-sectional
roughness

Fig. 8 Different patterns which are studied by the researchers using chamfered roughness
566 N. Arya et al.

Table 2 Different studies published on circular cross-sectional roughness elements

Authors Year Arrangement Cross section Range of key parameters
of rib element of flow Re (×10−3 ) e/D P/e
passage
Verma and 2000 Transverse Rectangular 5–20 0.01–0.03 10–40
Prasad [19] continuous
Prasad and 1988 Transverse Rectangular 5–50 0.02–0.033 10–20
Saini [20] continuous
Yadav and 2013 Transverse Rectangular 3.8–18 0.021–0.042 7.14–35.71
Bhagoria [21] continuous
Chaube et al. 2006 Transverse Rectangular 3–20 0.09 13.3
[22] continuous
Aharwal et al. 2008 Transverse Rectangular 3–18 0.037 10
[23] continuous
Kumar et al. 2017 Semi-circular Triangular 4–24 0.0167–0.067 7.5–100
[24]
Kumar et al. 2018 Transverse Triangular 4–18 0.015–0.06 4–20
[25] continuous
Aharwal et al. 2009 Transverse Rectangular 3–18 0.018–0.037 4–10
[26] with gap
Bhardwaj 2017 Transverse Rectangular 4.5–28 0.21–0.043 4–16
et al. [27] inclined
Singh et al. 2011 Down Rectangular 3–15 0.015–0.043 4–12
[28] V-pattern
Singh et al. 2015 Down Rectangular 3–15 0.043 4–12
[29] V-pattern
Sahu and 2016 Arc-shaped Rectangular 2–39 0.023–0.042 10
Prasad [30] pattern
Pandey et al. 2016 Multi-arc Rectangular 2.1–21 0.016–0.044 4–16
[31] shape with
gap pattern
Kumar et al. 2019 Multi-arc Rectangular 11–19 0.043 10
[32] shape with
gap pattern
Kumar et al. 2020 Innovative Rectangular 2.1–23 0.018–0.045 4–16
[33] arc-shaped
pattern

pattern over the absorber plate, and using this pattern, Nu and f increased by 2.96
and 3.13 times the smooth duct. But, the inclined rib element provides easy path to
the secondary flow and guide it toward the side walls of the SAH passage. During
this process, the increased temperature lowers the heat carrying capacity of the air
and leads to reduction in heat transfer. This reason motivated the researcher toward
V-rib pattern in which the secondary flow is directed toward the centerline of the
flow passage, and beside this, the distance covered by the secondary flow is reduced
Role of Artificial Roughness in the Performance Improvement … 567

which increases heat carrying capacity [28, 29]. The TPP obtained in case of V-rib
pattern is 2.06 [29]. Another pattern, i.e., arc-shaped rib, is also investigated by the
researchers [30–32]. In case of arc-shaped rib element, the gradual slop is provided
to the secondary flow (in comparison with V-pattern) which leads comparatively
higher TPP, i.e., 3.45 [31]. Another innovative arc-shaped rib pattern was introduced
by Kumar et al. [33] in which the arc-shaped rib elements are arranged in the form of S
shape along the width of SAH. The secondary flow associated from these types of rib
pattern results in higher local turbulence in between the two successive rib elements
and gives best THP value, i.e., 3.61, in comparison with transverse, inclined, V-
shaped and arc-shaped rib elements. Therefore, it can be concluded that the pattern
of rib element over plate (absorber) has significant effect on the performance of SAH.

3.2 Square and Rectangular Cross-Sectional Ribs and Their

Different Patterns

The square and rectangular cross-sectional roughness provides more area over the
rib top, and thus, the separation of free-shear takes place much effectively than the
circular cross-sectional roughness. Beside this, the local eddies formed at the front of
the roughness while flow strikes the rib element is comparatively low in rectangular
rib element due flat surface of the roughness element. There are different patterns of
rib elements which have been studied by the researcher for square and rectangular
cross-sectional roughness, and they are presented in Figs. 6 and 7, respectively.
Different studies are performed using different ranges of roughness parameters which
are shown in Tables 3 and 4 for square and rectangular cross-sectional rib elements,
respectively.
In literature, different patterns are studied using square rib element such as trans-
verse [22, 34–37], perforated transverse [38], and V-shaped [39, 40]. In case of trans-
verse square rib elements, the maximum THP is claimed of the order of 1.88 which is
estimated by Singh and Bhagoria [35] by investigating the rectangular duct SAH with
the help of commercial ANSYS software using finite volume method. Similar study
is also performed by Kumar et al. [37] using triangular cross-sectional duct SAH,
and it is seen that under similar range of roughness parameter the maximum THP
obtained is 1.44 which is comparatively lower as that obtained for rectangular duct
SAH. Thus, an important conclusion can be made that the SAH performance is not
only affected by the shape or pattern of the rib roughness, but it varies significantly
with the cross section of flow passage through which air flows.
The effect of perforated rib element has been investigated by the Singh and Singh
[38]. It is concluded from the study that the for low Reynolds number, i.e., Re < 3000,
the flow reattachment zone is limited to small area and which restrict the increment
in Nu. But, for higher Re (i.e., Re> 9000), the flow reattachment zone is stretched to
successive rib element, which helps better augmentation of heat from the absorber
plate, and due to this, the thermal efficiency of the SAH is found to be 68.55%. In
568 N. Arya et al.

Table 3 Different studies published on square cross-sectional roughness elements

Authors Year Arrangement Cross section Range of key parameters
of rib element of flow Re (×10−3 ) e/D P/e
passage
Taslim et al. 1997 Transverse Square 2–18 0.133–0.25 5–10
[34] continuous
Chaube et al. 2006 Transverse Rectangular 3–20 0.09 13.3
[22] continuous
Singh and 2014 Transverse Rectangular 3.8–18 0.042 10.71
Bhagoria [35] continuous
Abhay et al. 2018 Transverse Rectangular 3.8–18 0.042 7.14–17.86
[36] continuous
Kumar et al. 2019 Transverse Triangular 4–18 0.02–0.04 5–15
[37] continuous
Singh and 2018 Transverse Rectangular 3–15 0.042 4–30
Singh [38] wave profiled
Karwa and 2009 Discrete Rectangular 1.07–26.35 0.07 6.6–53.3
Chauhan [39] V-down
Kumar et al. 2012 Discrete Rectangular 2–20 0.043 10
[40] V-down

Table 4 Different studies published on rectangular cross-sectional roughness elements

Authors Year Arrangement Cross section Range of key parameters
of rib element of flow Re (× e/D P/e
passage 10−3 )
Jaurker 2006 Rectangular Rectangular 3–21 0.0181–0.0363 4.5–10
et al. [41]
Aghie et al. 2015 Rectangular Rectangular 10 0.018–0.037 10
[42]
Kumar 2017 Rectangular Triangular 4–18 0.02–0.04 5–15
et al. [16]
Layek et al. 2007 Rectangular Rectangular 3–21 0.022–0. 04 4.5–10
[43] with
alternative
grooved
Alfarawl 2017 Rectangular Rectangular 12.5–86.5 0.0468 6.6–53.3
et al. [44] with
alternative
semi-circular
Gupta and 2017 Tapered Rectangular 3.8–18 0.042–0.084 3.6–17.9
Varshney rectangular
[45]
Role of Artificial Roughness in the Performance Improvement … 569

another pattern, the rib elements are arranged in discrete V-shaped, and the effect
of providing gaps has been investigated [40]. By discretion of the rib elements, the
local velocity of secondary flow increased drastically at the location of gap which
produces highly intense local turbulent in the flow field and gives considerably higher
augmentation in heat transfer. Thus, the TPP found in discrete V-shaped pattern is
3.7.
In case of rectangular cross-sectional rib elements, three different patterns have
been investigated by the researcher, i.e., transverse [16, 41, 42], combination of
rectangular rib with alternative groove [43, 44], and transverse tapper rib [45]. As
the rib-top area is higher in case of rectangular rib element in comparison with
circular and square rib, the highest augmentation of heat is observed. Two different
studies were found in the literature in which rectangular [41, 42] and triangular [16]
flow passage has been used by the researchers. The highest TPP obtained is 1.85
[41] and 1.89 [16] in rectangular [41] and triangular [16] duct SAH, respectively.
It is found that the THP obtained in the triangular duct is comparatively higher as
that in rectangular duct SAH. In other two studies, Layek et al. [43] and Alfarawl
et al. [44] used the rectangular rib element in combination with alternative triangular
and circular groove, respectively. It is claimed that with this combination (i.e., rib
with triangular groove), the significantly higher local turbulence occurred in the
flow passage due to vortices developed around and in the groove with its break-
up with the alternative rib element. Due to this reason, the heat transfer increased
by 170% in comparison with smooth SAH. While in case of rectangular rib with
successive circular groove, the TPP found is 1.39. In comparison with the above-
discussed rectangular rib patterns, the best TPP is found in case of transverse tapered
rectangular rib elements, i.e., 1.91. Due to rib tapered rib element, the turbulent
kinetic energy increased, and it is guided toward the centerline (where the rib height
is minimum), and thus, the vortex formation takes place which results higher thermal
performance in comparison with other discussed patterns.

3.3 Chamfered Ribs and Their Different Patterns

The different configuration of chamfered rib roughness is presented in Fig. 8, and

the range of parameter considered is presented in Table 5. The transverse rib rough-
ness is commonly used by the researcher [46–51], and most of the studies were
carried out using rectangular flow passage [46–49]. The highest THP is claimed by
Gawande et al. [49] with value of 2.07 in comparison with other similar studies,
while few studied are available on triangular flow passage SAH. In these studies, the
chamfered rib roughness is classified into two basic geometries based on the location
of provided chamfer, i.e., forward-facing and backward-facing chamfer [50, 51]. It
is observed that the location of providing chamfer plays an important role in the
SAH performance. It is noticed that the highest TPP (with value of 2.08) obtained
in case of forward-facing chamfered rib roughness is approximately 4% higher than
the backward-facing chamfered ribs, and this resulted due to slope occurred due to
570 N. Arya et al.

Table 5 Different studies published on chamfered rectangular cross-sectional roughness elements

Authors Year Arrangement Cross section Range of key parameters
of rib element of flow Re (×10−3 ) e/D P/e
passage
Han and 1988 Transverse Rectangular 8–80 0.021–0.078 10–20
Park [46] trapezoidal
Karwa 1999 Transverse Rectangular 3–20 0.0141–0.0328 4.5–8.5
et al. [47] chamfered
rectangular
Karwa 2001 Transverse Rectangular 3.75–16.35 0.0197–0.0441 4.58–7.09
et al. [48] chamfered
Gawande 2016 Chamfered rib Rectangular 3.8–18 0.042 7.4–17.86
et al. [49]
Kumar 2018 Forward Triangular 4–17 0.018–0.043 12
et al. [50] chamfered rib
Kumar 2019 Backward Triangular 4–17.1 0.018–0.043 12
et al. [51] chamfered
Layek 2009 Chamfered rib Rectangular 3–21 0.03 10
et al. [52] with
alternative
groove
Bhagoria 2002 Wedge Rectangular 12.5–86.5 0.018–0.037 6.6–53.3
et al. [53]

the chamfering over the rib-top surface. In this case, the rib-top surface allows the
smooth passage of free-shear layer and reduces the eddy formation at the start of the
rib element which leads to better augmentation of heat.
In the literature, two different rib patterns were studied by the researchers, i.e.,
chamfered rib with alternative triangular groove [52] and wedge pattern [53]. It is
found that the TPP obtained in case of wedge pattern is 2.0 which is approximately
13.63% higher as that estimated in case chamfered rib with alternative triangular
groove pattern.

4 Effect of Duct Cross-Section

The performance of SAH is not only affected by shape or pattern of the rib roughness
over the absorber plate, but it is also shows the significant influence of cross-sectional
shape of the flow passage which is considered in design of SAH. Table 6 shows the
comparison of THP values for triangular and rectangular duct SAH. The comparison
has been carried out using different cross-sectional-shaped rib roughnesses, and it
can be seen from the table that TPP value changes noticeably with the change of flow
passage of SAH. This variation in THP might be resulted due to number of available
corners in the duct. There is coexistence of laminar and turbulent flow near the corner
Role of Artificial Roughness in the Performance Improvement … 571

Table 6 Comparison of THP between triangular and rectangular duct SAH

Cross-sectional of rib element SAH with triangular SAH with rectangular passage
passage
Authors TPP Authors TPP
Circular Kumar et al. [25] 1.82 Verma and Prasad [19] 1.71
Square Kumar et al. [37] 1.44 Chaube et al. [22] 1.94
Rectangular Kumar et al. [16] 1.89 Chaube et al. [22] 2.17
Chamfered Kumar et al. [50] 2.15 Chaube et al. [22] 1.95

region which leads to flow stagnation, and there is formation of hot spots [11]. These
hot spots indicate the trapped heat which not carried away the main or primary flow,
and thus, the heat transfer reduced in the duct. Beside this, the velocity profile is
also an important factor which affects the heat transfer in the duct. In equilateral
triangular duct, the main flow concentrated at one-third of duct height from the base
surface, while it is at the center of the duct in case of rectangular duct. Due to this,
the main flow is near to the base surface, and higher heat trapped may take place at
the corner opposite to the base surface which leads poor thermal performance of the
triangular duct in comparison with the rectangular duct.
With the use of rib roughness, the thermal performance of triangular duct can be
increased to the significant level. As the main flow is near to the base surface, the
turbulence created in the main flow due to secondary flow resulted better mixing
and thus increases the heat transfer augmentation in SAH. It can be concluded from
Table 6 that the it is better option to consider triangular duct in case of rib with
circular and chamfered cross section because it resulted higher TPP, i.e., 1.82 and
2.15, in comparison with rectangular duct SAH, while rectangular duct SAH is better
for square and rectangular cross-sectional rib elements.

5 Parameters Used for Predicting SAH Performance

For measuring the desired output parameters, all the properties measured during
experimentation were calculated at bulk temperature, and collected data is utilized
to determine thermal (as Nusselt number; Nuavg ) and hydraulic (as friction factor;
f ) performance of SAH. For the calculation of Nu and f , following equations were
used:
The heat transfer through the SAH is predicted using Nusselt number (Nu) which
is a non-dimensional number, and mathematically, it is calculated as:

h avg
Nuavg = ×D (1)
kair

where
572 N. Arya et al.

area
D =4× (2)
perimeter
 
W×H
=4× (for rectangular duct) (2.1)
2(W + H )
 
1
×WH
=4× 2
(for triangular duct) (2.2)
2(W + H )

There is unavoidable effect which appeared in terms of pressure drop in the SAH,
and it is calculated as:
 −1
1 P 1
f = × × ρair × vair
2
×D (3)
4 L 2

where

Pt = ρg × h (4)

It is seen that there is one desirable affect that is heat transfer which is occurred
along with the unwanted affect, i.e., friction factor (f ). Therefore, one more param-
eter needs to be considered to estimate the effectiveness of the SAH which is
thermohydraulic performance parameter (TPP), and it is expressed as:

(Nu/Nu2 )
TPP = (5)
( f / f s )1/3

6 Comparative Performance Assessment of Various

Roughness Used in SAH

The different proposed design of rib roughness has shown strong influence on the
heat and flow characteristics of SAH, and these different designs can be compared on
the basis of thermal and hydraulic performances in order the find out the best possible
cross section of rib element. With the use of roughness, thermal performance (i.e.,
Nu) changes drastically in comparison with the smooth SAH, but there also appears
an unavoidable effect, i.e., pressure drop or friction factor. In other words, it can be
said that the heat transfer increases at the expense of friction factor. On the basis
of cost effectives is concerned, it is important to design a SAH in such a way that
the Nu should be as higher as possible with f as small as possible, such that the
heat transfer dominates the f and opposite to this leads to the faulty design in which
the desired outcome; i.e., heat transfer is comparatively lower than f . The external
Role of Artificial Roughness in the Performance Improvement … 573

power consumed by the system is comparatively higher than the corresponding heat
transfer which results in wastage of energy.

6.1 Based on Nusselt Number and Friction Factor

Figure 9 presents the comparison of highest Nuenh value which is claimed by the
different researchers using various cross-sectional rib elements with different patterns
over the absorber plate. It is observed from the figure that the Nuenh changes dras-
tically with the variation of rib cross section and with the slight change in the
cross-sectional area or pattern of the rib element resulted noticeable change in the
thermal performance of SAH. By providing the rib element in the path of air flow, the
dynamics of flow change considerable inside the SAH due to flow separation and its
reattachment over the absorber plate. The secondary flow induced due to rib element
develops local turbulence at or near the rib element. The best Nuenh is observed in
case of innovative arc-shaped rib pattern with value of 3.6 for circular cross-sectional
rib element [25], whereas the lowest Nuenh is obtained in case of transverse circular
cross-sectional rib element, and it is due to the minimum rib-top area which does
not effectively support the secondary flow in the SAH and resulted in poor local
turbulence thus lower Nuenh .
The heat transfer improvement in the ribbed SAH is also accompanied with the
pressure drop or friction factor (f ). For effective design of SAH, the f is required
to be as small as possible. Figure 10 presents the comparison of f penalty obtained

3.5

3
Nuenh

2.5

1.5

Singh and Bhagoria [35] Kumar et al [37]

Verma and Prasad [19] Prashad and Saini [20]
Kumar et al. [25] Yadav and Bhagoria [21]
Layek et al. [44] Alfarawl et al. [45]

Fig. 9 Comparison of optimum Nuenh claimed in various studies

574 N. Arya et al.

4.5

3.5
fpenalty

2.5

1.5

Singh and Bhagoria [35] Kumar et al [37] Verma and Prasad [19]
Prashad and Saini [20] Kumar et al. [25] Layek et al. [44]
Alfarawl et al. [45] Kumar et al. [43] Layek et al. [53]
Kumar et al. [51] Kumar et al. [52]

Fig. 10 Comparison of maximum f penalty claimed in various studies

in SAH with different cross-sectional rib element under different rib patterns. The
minimum value of f penalty is claimed by Singh and Bhagoria [35], i.e., 1.44, and it is
obtained in transverse circular cross-sectional rib element, while the highest f penalty is
obtained in innovative arc-shaped rib pattern of circular cross-sectional rib element
[25]. For other combination of rib element patterns, the f penalty lies in between these
two values. Most importantly, it is also found that the higher f penalty corresponds to
lower Re, and as with the increase of Re, the f penalty starts decreasing because the
viscous forced in the flow is dominated by the inertia forces.

6.2 Based on Thermohydraulic Performance Parameter

It is noticed from Figs. 9 and 10 that both Nuenh and f penalty varied noticeably with
the shape of rib element and it orientation over the absorber plate. Furthermore,
both these values obtained maximum for forward chamfered rib pattern [50], while
minimum for transverse circular rib element [37]. Thus, it is quite difficult to come
to a conclusion on the basis of this observation. It is very important to introduce
a parameter which includes effect of both heat transfer and flow characteristics in
Role of Artificial Roughness in the Performance Improvement … 575

2.2

1.8
TPP

1.6

1.4

1.2

Singh and Bhagoria [35] Kumar et al1 [37] Verma and Prasad [19]
Prashad and Saini [20] Kumar et al. [25] Layek et al. [44]
Alfarawl et al. [45] Kumar et al. [43] Layek et al. [53]
Kumar et al. [51] Kumar et al. [52]

Fig. 11 Comparison of optimum TPP claimed in various studies

order to estimate the effectiveness of the rib element. Keeping this in mind, the
thermohydraulic performance parameter (TPP) is calculated for the various available
rib elements. According to TPP, the effectiveness of roughness can only be ensured
if its value is greater than one. In other words, it can be said that if the TPP is greater
than one, the heat transfer which takes place in the system is comparatively higher
in comparison with the effort made to maintain the flow through the system. The
comparison of TPP for various rib roughness configurations is shown in Fig. 11. The
best TPP obtained is 2.14 in case of triangular duct SAH having forward chamfered
rib elements of rectangular cross-sectional area [50]. In case of rectangular duct
SAH, the TPP found is 2.1 for rectangular cross-sectional rib element [35].

7 Conclusions

In this chapter, the comprehensive review of literature has been carried out in order
to analyze the best configuration of rib element which resulted in higher thermo-
hydraulic performance parameter. From the current report, it is concluded that the
cross-sectional area of the duct used in SAH plays an important role in the deciding
thermohydraulic performance parameter in addition to rib cross-sectional area and
its orientation on the absorber plate.
The heat transfer in ribbed SAH is influenced by the rib-top surface. Higher the top-
surface area, more is the heat transfer through the SAH. On this fact, it is concluded
that the transverse rib pattern of the highest thermohydraulic performance parameter
is obtained in case of rectangular cross-sectional rib element and relatively worst
576 N. Arya et al.

thermohydraulic performance parameter is found in case of circular cross-sectional

rib element.
As per rib pattern is concerned, the best thermohydraulic performance parameter
corresponds to arc-shaped rib elements in comparison with other studied rib patterns
such as transverse, inclined, and V-shaped. Beside this, by providing the gap in the
rib element, there is further enhancement in the performance of SAH.

References

1. Halabi MA, Al-Qattan A, Al-Qtaibi A (2005) Application of solar energy in the oil industry-
current status and future prospects. Renew Sustain Energy Rev 43:296–314
2. IEA (2012) World energy outlook. International Energy Agency, Head of Communication and
Information Office: Paris, France. Available at www.iea.org/publications/freepublications/pub
lication/name-49561-en.html
3. OPEC (2013) World oil outlook. Secretariat, Organisation of the Petroleum Exporting
Countries: Vienna, Austria. Available at www.opec.org
4. Kleissl J (2013) Solar energy forecasting and resource assessment
5. Tiwari GN, Tiwari Shyam A (2016) Handbook of solar energy: theory, analysis and
applications. Springer Nature, Berlin
6. Tiwari GN, Mishra RK (2012) Advance renewable energy sources. RSC publishing, UK
7. The Potential for Solar Energy in India Is Huge: IEA. http://www.eai.in/club/users/krupali/blo
gs/627
8. Solar Energy in India-Pros, Cons and the Future. http://www.mapsofindia.com/my-india/india/
scope-of-solar-energy-in-india-pros-cons-and-the-future
9. Manju S, Sagar N (2017) Progressing towards the development of sustainable energy: a crit-
ical review on the current status, applications, developmental barriers and prospects of solar
photovoltaic system in India. Renew Sustain Energy Rev 70:293–313
10. Lingayat AB, Chandramohan VP, Raju VRK, Meda V (2020) A review on indirect type
solar dryers for agriculture crops-dryer setup, its performance, energy storage and important
highlights. Appl Energy 28:11400
11. Kumar R, Varun Kumar A (2016) Thermal and fluid dynamic characteristics of flow through
triangular cross-section duct: a review. Renew Sustain Energy Rev 61:123–140
12. Bharadwaj G, Varun Kumar R, Sharma A (2017) Heat transfer augmentation and flow charac-
teristics in ribbed triangular duct solar air heater: an experimental analysis. Int J Green Energy
14(7):587–598
13. Kumar R, Goel V, Kumar A (2017) A parametric study of the 2D model of solar air heater with
elliptical rib roughness using CFD. J Mech Sci Technol 31(2):959–964
14. Prasad K, Mullick SC (1983) Heat transfer characteristics of a solar air heater used for drying
purposes. Appl Energy 13(2):83–93
15. Hu J, Zhang G (2019) Performance improvement of solar air collector based on airflow
reorganization: a review. Appl Therm Eng 155:592–611
16. Kumar R, Kumar A, Goel V (2017) A parametric analysis of rectangular rib roughened
triangular duct solar air heater using computational fluid dynamics. Sol Energy 157:1095–1107
17. Kumar R, Kumar A, Goel V (2018) Effect of rounded corners on the heat transfer and fluid
flow through triangular duct. J Heat Transfer 140(12):
18. Incropera F, Dewitt PD, Bergman TL, Lavine AS, Setharamu KN, Setharama TR (2013)
Fundamentals of heat and mass transfer, 5th edn. Wiley, United States
19. Verma SK, Prasad BN (2000) Investigation for the optimal thermohydraulic performance of
artificially roughened solar air heaters. Renewable Energy 20(1):19–36
Role of Artificial Roughness in the Performance Improvement … 577

20. Prasad BN, Saini JS (1988) Effect of artificial roughness on heat transfer and friction factor in
a solar air heater. Sol Energy 41(6):555–560
21. Yadav AS, Bhagoria JL (2013) A CFD (computational fluid dynamics) based heat transfer and
fluid flow analysis of a solar air heater provided with circular transverse wire rib roughness on
the absorber plate. Energy 55:1127–1142
22. Chaube A, Sahoo PK, Solanki SC (2006) Analysis of heat transfer augmentation and flow
characteristics due to rib roughness over absorber plate of a solar air heater. Renewable Energy
31(3):317–331
23. Aharwal KR, Gandhi BK, Saini JS (2008) Experimental investigation on heat-transfer enhance-
ment due to a gap in an inclined continuous rib arrangement in a rectangular duct of solar air
heater. Renew Energy 33:585–596
24. Kumar R, Kumar A, Varun (2017) Computational fluid dynamics based study for analysing
heat transfer and friction factor in semi-circular rib- roughened equilateral triangular duct. Int
J Numer Meth Heat Fluid Flow 27(4):941–957
25. Kumar R, Goel V, Kumar A, Khurana S, Singh P, Bopche S (2018) Numerical investigation of
heat transfer and friction factor in ribbed triangular duct solar air heater using computational
fluid dynamics (CFD). J Mech Sci Technol 32(1):399–404
26. Aharwal KR, Gandhi BK, Saini JS (2009) Heat transfer and friction characteristics of solar air
heater ducts having integral inclined discrete ribs on absorber plate. Int J Heat Mass Transf
52(25–26):5970–5977
27. Bhardwaj G, Varun, Kumar R, Sharma A (2017) Heat transfer augmentation and flow charac-
teristics in ribbed triangular duct solar air heater: an experimental analysis. Int J Green Energy
14(7):587–598
28. Singh S, Chander S, Saini JS (2011) Heat transfer and friction factor correlations of solar air
heater ducts artificially roughened with discrete V-down ribs. Energy 36(8):5053–5064
29. Singh S, Chander S, Saini JS (2015) Thermo-hydraulic performance due to relative roughness
pitch in V-down rib with gap in solar air heater duct—comparison with similar rib roughness
geometries. Renew Sustain Energy Rev 43:1159–1166
30. Sahu MK, Prasad RK (2016) Exergy based performance evaluation of solar air heater with
arc-shaped wire roughened absorber plate. Renew Energy 96:233–243
31. Pandey NK, Bajpai VK, Varun (2016) Experimental investigation of heat transfer augmentation
using multiple arcs with gap on absorber plate of solar air heater. Sol Energy 134:314–326
32. Kumar R, Goel V, Singh P, Saxena A, Kashyap AS, Rai A (2019) Performance evaluation and
optimization of solar assisted air heater with discrete multiple arc shaped ribs. J Energy Storage
26:
33. Kumar R, Kashyap AS, Singh P, Goel V, Kumar K (2020) Innovatively arranged curved
ribbed solar assisted air heater: performance and correlation development for heat and flow
characteristics. J Solar Energy Eng 142(3)
34. Taslim D, Li ME, Kercher T (1996) Experimental heat transfer and opposite walls. J Turbomach
118:20–28
35. Singh A, Bhagoria JL (2014) A numerical investigation of square sectioned transverse rib
roughened solar air heater. Int J Therm Sci 79:111–131
36. Abhay L, Chandramohan VP, Raju VRK (2018) Numerical analysis on solar air collector
provided with artificial square shaped roughness for indirect type solar dryer. J Clean Prod
190:353–367
37. Kumar R, Kumar A, Goel V (2019) Performance improvement and development of correlation
for friction factor and heat transfer using computational fluid dynamics for ribbed triangular
duct solar air heater. Renewable Energy 131:788–799
38. Singh I, Singh S (2018) CFD analysis of solar air heater duct having square wave profiled
transverse ribs as roughness elements. Sol Energy 162:442–453
39. Karwa R, Chauhan K (2009) Performance evaluation of solar air heaters having v-down discrete
rib roughness on the absorber plate. Energy 35(1):398–409
40. Kumar A, Saini RP, Saini JS (2012) Experimental investigation on heat transfer and fluid flow
characteristics of air flow in a rectangular duct with multi v-shaped rib with gap roughness on
the heated plate. Sol Energy 86(6):1733–1749
578 N. Arya et al.

41. Jaurker AR, Saini JS, Gandhi BK (2006) Heat transfer and friction characteristics of rectangular
solar air heater duct using rib-grooved artificial roughness. Sol Energy 80(8):895–907
42. Aghaie AZ, Rahimi AB, Akbarzadeh A (2015) A general optimized geometry of angled ribs for
enhancing the thermo-hydraulic behaviour of a solar air heater channel—a Taguchi approach.
Renewable Energy 83:47–54
43. Layek A, Saini JS, Solanki SC (2007) Heat transfer and friction characteristics for artificially
roughened ducts with compound turbulators. Int J Heat Mass Transf 50:4845–4854
44. Alfarawi S, Abdel-Moneim SA, Bodalal A (2017) Experimental investigations of heat transfer
enhancement from rectangular duct roughened by hybrid ribs. Int J Therm Sci 118:123–138
45. Gupta AD, Varshney L (2017) Performance prediction for solar air heater having rectangular
sectioned tapered rib roughness using CFD. Therm Sci Eng Prog 4:122–132
46. Han JC, Park JS (1988) Developing heat transfer in rectangular channels with rib tabulators.
Int J Heat Mass Transf 31(1):183–195
47. Karwa R, Solanki SC, Saini JS (1999) Heat transfer coefficient and friction factor correlations
for the transitional flow regime in rib-roughened rectangular ducts. Int J Heat Mass Transf
42(9):1597–1615
48. Karwa R, Solanki SC, Saini JS (2001) Thermo-hydraulic performance of solar air heaters
having integral chamfered rib roughness on absorber plates. Energy 26(2):161–176
49. Gawande VB, Dhoble AS, Zodpe DB, Chamoli S (2016) Analytical approach for evaluation
of thermohydraulic performance of roughened solar air heater. Case Stud Therm Eng 8:19–31
50. Kumar R, Kumar A, Goel V (2019) Simulation of flow and heat transfer in triangular cross-
sectional solar-assisted air heater. J SolEnergy Eng 141(1):
51. Kumar R, Goel V, Kumar A (2018) Investigation of heat transfer augmentation and friction
factor in triangular duct solar air heater due to forward facing chamfered rectangular ribs: a
CFD based analysis. Renewable Energy 115:824–835
52. Layek A, Saini JS, Solanki SC (2009) Effect of chamfering on heat transfer and friction
characteristics of solar air heater having absorber plate roughened with compound turbulators.
Renewable Energy 34(5):1292–1298
53. Bhagoria JL, Saini JS, Solanki SC (2002) Heat transfer coefficient and friction factor correla-
tions for rectangular solar air heater duct having transverse wedge shaped rib roughness on the
absorber plate. Renew Energy 25(3):341–369
Estimation of Mechanical Behavior
of Cementitious Material Using
Microstructural Information

Praveen Verma and Ayushman Bajpai

Abstract Concrete is a widely used material, so estimating its mechanical prop-

erties is essential for the safety and durability of structures. It has a complex
multi-component and multiscale structure that forms through a chemical reac-
tion and then continue to change with time. Generally, cement paste is the most
porous and also most chemically active part of concrete. Hence, understanding the
cement microstructure and its variation over time is essential for understanding the
mechanical properties. The microstructural information can be extracted computa-
tionally using modeling platform-tools. The generated microstructural information
is modeled as a Representative volume element in finite element analysis software
to generate some cubes of several tens of microns in dimensions subjected through
prescribed deformation mode to generate effective elastic tensor of the material. The
current work focusses on the estimation of effective properties of cementitious mate-
rial through finite element analysis. The calculated elastic properties, such as Elastic
moduli and poison’s ratio, are validated through previous studies.

Keywords Microstructure · Hydration · Elastic modulus · RVE ·

Homogenization · Calcium silicate hydrate

Abbreviations

C2 S Dicalcium silicate
C3 S Tricalcium silicate
C3 A Tricalcium aluminate
C4 AF Tetra calcium aluminoferrite
CSH2 Gypsum

P. Verma (B)
Department of Civil Engineering, IIT Roorkee, Roorkee, India
A. Bajpai
Department of Civil Engineering, Ram Manohar Lohia Avadh University Ayodhya,
Faizabad, India

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2022 579
A. Gupta et al. (eds.), Advances in Energy and Combustion, Green Energy and Technology,
https://doi.org/10.1007/978-981-16-2648-7_24
580 P. Verma and A. Bajpai

CH Calcium hydroxide
CSH Calcium silicate hydrate gel
C3 AH6 Hydrogarnet
C6 AS3 H32 Ettringite
C4 ASH12 Monosulfate AFM
FH3 Iron hydroxide
RVE Representative volume element
ITZ Interfacial transition zone
MM Multiscale modelling
FEA Finite element analysis
VCCTL Virtual cement and concrete testing laboratory
NIST National Institute of Standards and Technology
E Elastic modulus
G Shear modulus
K Bulk modulus
K∗ Effective bulk modulus
G∗ Effective shear modulus
μ Poisson’s ratio
ρ Mass density
E1 Tensile deformation (Uniaxial)
G12 Shear deformation in x-y plane
Ei Homogeneous modulus corresponding to applied axial deformation
mode (i = 1, 2, 3)
Gi j Homogeneous modulus corresponding to shear deformation mode (i, j
= 1, 2, 3)
CV Computational volume
α Degree of hydration
Vf Volume fraction of fibre material
Vm Volume fraction of matrix material
σi j Average stress tensor over the entire RVE
εi j Average strain tensor over the entire RVE

1 Introduction

As we all know that the cementitious materials are highly heterogeneous and complex
structure because their microstructures have random features at different length scale
level such as CSH, cement paste, mortar, and concrete level [1, 2]. Every scale
represents a random composite. At the nanoscale, the CSH is a random composite
which exists in two different forms such as high-density CSH and low-density CSH
because of different packing densities. Cement paste has a complex structure at
microscale with a random arrangement of coarse and fine aggregate in concrete.
However, at the meter scale generally concrete is considered to be uniform with bulk
Estimation of Mechanical Behavior of Cementitious Material … 581

Fig. 1 Different levels of cement-based composites [6]

characteristic properties such as compressive strength, elastic modulus, and others

[3]. The multiscale level for cement-based composites can be illustrated in Fig. 1. It
can be seen from the figure that there are different levels of cement-based materials
that effect the loading behavior and properties of concrete [4, 5]. The modeling at the
level of concrete is difficult because here concrete is modeled by considering three-
phase material such as cement paste, interfacial transition zone (ITZ), and aggregate
phase. Therefore, finding effective properties of cementitious materials are required
to help engineers during design process.
Multiscale modeling (MM) of materials is the most prominent priority of the
research community for achieving material properties. Now, it is possible to model
cementitious materials by selecting RVE size which governs the overall properties
of material [7]. The elastic properties of individual phase and their densities such
as clinkers, hydrated and unhydrated products can be estimated using molecular
dynamics (MD) simulations. These properties are employed during modeling of
RVE of different sizes. However, it is very difficult to analyze the behavior of the
material for predicting its property using different realistic models of its microstruc-
ture However, the properties of the individual components of concrete, knowledge
of its microstructure, and their relationship to each other is useful for controlling the
properties of concrete [8–11].
In this paper, an attempt is made for estimating the effective properties of hetero-
geneous CSH-based multiphase cementitious composite, based on knowledge of
individual phases. The focus of this paper is two-fold: (i) to develop a microstructural
information of cement paste during hydration process. (ii) to model the cement paste
582 P. Verma and A. Bajpai

Fig. 2 Particle size distribution for simulation corresponding to Blaine Fineness 370 m2 /kg used
in the hydration model

in FEA software to get homogenized elastic properties of cementitious materials. The

main aim of this study is to determine the homogenized properties at macrolevel using
micromechanics-based model when small strain quasi-static loading conditions are
applied.

2 Methodology

2.1 Model for Cement Paste

In order to investigate the microstructure development with time and the change of its
properties, in terms of its mechanical properties quantitative topological and geomet-
rical description is needed. The most used digital microstructure model Virtual
Cement and Concrete Testing Laboratory (VCCTL) modeling platform developed by
National Institute of Standards and Technology (NIST) researchers which were based
on CEMHYD3D developed by Bentz and Bentz [12]. There is quickly generation of
microstructure using this model which is based on discretization of computational
volumes into cubic voxel, each occupies by a single phase of cement constituents
[13].
The modeling platform VCCTL was used to model hydration of cement paste and
3CaO·SiO or C3 S in chemistry notation are the main phase of Portland cement and
resulting three-dimensional microstructures. The advantage of using microstructures
from is that rather generating random spatial distribution of phases, it simulates
the processes that lead to microstructural development, generating more realistic
representations of microstructures.
Estimation of Mechanical Behavior of Cementitious Material … 583

2.2 Simulation

The composition of Portland cement that is simulated is shown in Fig. 3 and their
Particle Size Distribution is shown in Fig. 2. There are three simulations is done at
a different water-cement ratio i.e. 0.3, 0.35, and 0.4. The simulation is carried out
using the VCCTL modeling tools in which user may input or choose the material
characterization and the computational cubic volume is associated with one million
(µm) in size and this size is sufficient for simulating hydration and porosity.
The density and elastic properties of each phases are assumed as given in Tables 1
and 2. Each reaction of cement clinkers is calculated in volumetric terms, and the
hydrated products e.g. ettringite and C–S–H (Calcium Silicate Hydroxide) their depo-
sition starts around the cement particles and the other products, i.e. Portlandite nucle-
ates in the pores, and the particles are produced throughout the first 24 h of hydration
to give a final number around 20% of the number of C3 S particles in the system.

3 Homogenization of Cementitious Composites

Homogenization simply means to find out the effective properties of heterogeneous

materials. Homogenization is a method for finding the microscopic fields and prop-
erties based on the microstructure parameters and local properties of heterogeneous
media and the effective properties represent the overall behavior of heterogeneous
media which depends upon the phase properties and microstructure information of
heterogeneous materials.

Fig. 3 Compositions of
Portland cement
584 P. Verma and A. Bajpai

Table 1 Elastic properties of chemical phases measured by the mechanical tests [4, 11, 12, 14–17]
Phase Young modulus (GPa) Poisson’s ratio (µ)
Tricalcium silicate (C3 S) 117.6 0.314
Dicalcium silicate (C2 S) 117.6 0.314
Tricalcium aluminate (C3 A) 117.6 0.314
Tetracalcium aluminoferrite (C4 AF) 117.6 0.314
 
Gypsum CSH2 45.7 0.33
Calcium hydroxide (CH) 42.3507 0.324
Calcium silicate hydrate gel (CSH) 22.4 0.25
Hydrogarnet (C3 AH6 ) 22.4 0.25
Ettringite (C6 AS3 H32 ) 22.4 0.25
Monosulfate AFM (C4 ASH12 ) 0.0423 0.324
Iron hydroxide (FH3 ) 22.4 0.25
Empty porosity 0.001 0.00
Water filled porosity 0.001 0.499924

Table 2 Bulk density of each phase of cement used in the simulation [18]
Phase Chemical formula Notation Density (g/cm3 )
Air – – 0.00
Alite (CaO)3 (SiO2 ) C3 S 3.21
Aluminate (CaO)3 (Al2 O3 ) C3 A 3.03
Anhydrite CaSO4 CS 3.28
Belite (CaO)2 (SiO2 ) C2 S 3.28
Calcium silicate (CaO)3 (SiO2 ) 3H2 O C–S–H 2.12
hydrate
Calcium silicate (CaO)1.7 (SiO2 ) 1.8H2 O C–S–H 2.60
hydrate
Ettringite (CaO)3 (Al2 O3 )(CaSO4 )3 32H2 O C6 A$3 H32 1.78
Ferrite (CaO)4 (Al2 O3 )(Fe2 O3 ) C4 AF 3.73
Free lime CaO C 2.52
Gypsum CaSO4 2H2 O C$H2 2.32
Monosulfate (CaO)3 (Al2 O3 )(CaSO4 ) 12H2 O C4 A$H12 2.00
Periclase MgO M 3.58
Portlandite Ca(OH)2 CH 2.24
Quartz SiO2 S 2.65
Water H2 O H 1.00
Hydrogarnet (CaO)3 Al2 O3 6(H2 O) C3 AH6 2.52
Estimation of Mechanical Behavior of Cementitious Material … 585

3.1 Representative Volume Element (RVE)

During analysis, the selection of RVE size must be necessary for homogenization.
The concept of representative volume element (RVE) was developed by Hill [19]. For
the composite materials, the representative elementary volume (REV) (also called
the representative volume element (RVE) or the unit cell) is the smallest volume
over which a measurement can be made that will yield a value representative of the
whole. The Representative volume element is defined as a small volume taken that is
less than the actual volume of material that capable to capture the overall properties
of the composite. Alternatively, Representative volume element also be defined as
a Volume that shows the effective properties of chosen RVE size is same as the
effective properties of overall behavior.

3.2 RVE Generation

The software package virtual cement and concrete testing laboratory (VCCTL) based
on CEMHYD3D V.3 which was developed by the National Institute of Standards and
technology. Researchers also use this modeling platform to simulate the hydration
of cement paste and concrete and the formation of digital generated microstructures
for type-1 Portland cement [12]. This software creates a microstructure informa-
tion which depends particle size distribution, surface area fraction, and the volume
fraction of the several constituents of cement phases. Microstructural information
stored at a different degree of hydration. For obtaining the result, it is important to
choose a cubic size of 4 different domains of “10 × 10 × 10” µm (1 K), “20 ×
20 × 20” µm (8 K) and “25 × 25 × 25” µm (15.625 K) are considered in this
study. A 3-D periodic microstructure is generated and which contains typically four
cement clinkers minerals, gypsum, and other hydrated products, in the form of spher-
ical particles [12] and these spherical particles are placed into a 3-D Computational
Volume having individual voxels size is 1 µm and the particle size ranges varies
from 1 to 35 µm in Diameter of material properties with respect to the “boundary
condition” applied [20] (K Sab—Eur. J. Mech & 1992).

3.3 Finite Element Analysis (FEA) Based Microstructure

Modeling

The microstructure voxel-based information generated by hydration model which

was imported into finite element software using a python scripting, MATLAB coding,
and geometry modeled using continuum C3D8 brick elements of dimension “10 ×
10 × 10” µm (1 K), “20 × 20 × 20” µm (8 K) and “25 × 25 × 25” µm (15.625 K)
as in Fig. 4.
586 P. Verma and A. Bajpai

a) b)

c)

Fig. 4 a 1 K, b 8 K, c 15.625 K models of hydrated cement paste

Mechanical properties can find out by creating a micromechanical model based

on the microstructural information available during simulation, and the appropriate
constitutive equation has been created. Microstructure modeling can be divided into
three parts: geometry, boundary condition, and constitutive equations. The modeled
microstructure of cement paste as in Fig. 4.
Estimation of Mechanical Behavior of Cementitious Material … 587

3.4 Boundary Conditions

Boundary conditions are applied to the modeled geometry of various sizes which was
shown in the literature [11, 12, 14, 21]. For the estimation of effective properties,
of the cementitious material the modeled geometry of sizes 1, 8, and 15.625 K was
subjected to various deformation modes which were shown in Fig. 4.
For the estimation of effective properties of a composite model, it may be neces-
sary to impose uniform strain to compute elastic properties in which the pure tensile
deformation and pure shear deformation are applied as shown in Fig. 5a and b. The
homogenized elastic properties are calculated by assigning the elastic properties of
each hydrated or unhydrated phase of cement and subjected to the RVE for imposing
deformation modes. The elastic moduli are the ratio of average stress to the average
strain The elastic moduli are the ratio of average stress and strain corresponding to
deformation mode.
avg avg
σi σi
Ei = avg , G i j = avg (1)
∈i ∈j

where E i is a homogeneous modulus corresponding to applied deformation mode (i,

j = 1, 2, 3) And in case of pure shear deformation the tensorial shear strain can be
computed as (Fig. 6)
 
1 1 x y
∈x y = (θ1 + θ2 ) = + (2)
2 2 y x

Two examples related to the uniaxial deformation (E 1 ) along the x-axis and shear
deformation (G12 ) in the x-y plane are explained in details
Case 1 Applied strain: ∈11 = 0 and (∈22 , ∈33 , ∈12 , ∈13 , ∈23 = 0) for uniaxial
deformation along the x-axis.

Nodes on faces: Nodes on edges: Nodes on vertices:

∈11 = 0, i = 1 ∈11 = 0, i = 1 ∈11 = 0, i = 1
u 1F2 − u 1F1 − L x ∈11 = 0 u 1E2 − u 1E4 − L x ∈11 = 0 u 1V 3 − u 1V 5 − L x ∈11 = 0
u 1F4 − u 1F3 = 0 u 1E1 − u 1E3 − L x ∈11 = 0 u 1V 2 − u 1V 8 − L x ∈11 = 0
u 1F6 − u 1F5 = 0 u 1E6 − u 1E8 − L x ∈11 = 0 u 1V 7 − u 1V 1 − L x ∈11 = 0
u 1E5 − u 1E7 − L x ∈11 = 0 u 1V 4 − u 1V 6 − L x ∈11 = 0
u 1E11 − u 1E9 = 0
u 1E10 − u 1E12 = 0

Case 2 Applied strain: ∈12 and ∈21 = 0 and (∈11 , ∈22 , ∈33 , ∈12 , ∈13 , ∈23 = 0) for
shear deformation in x-y plane.
588 P. Verma and A. Bajpai

Fig. 5 Model subjected to displacements a tensile deformation (E 1 ) and b Pure shear (G12 )
boundary conditions
Estimation of Mechanical Behavior of Cementitious Material … 589

Fig. 6 Prescribed boundary condition

Nodes on faces: Nodes on edges: Nodes on vertices:

∈12 = 0, i = 1 ∈12 = 0, i = 1 ∈12 = 0, i = 1
u 1F2 − u 1F1 = 0 u 1E2 − u 1E4 − L y ∈12 = 0 u 1V 3 − u 1V 5 − L y ∈12 = 0
u 1F4 − u 1F3 − L y ∈12 = 0 u 1E1 − u 1E3 + L y ∈12 = 0 u 1V 2 − u 1V 8 − L y ∈12 = 0
u 1F6 − u 1F5 = 0 u 1E6 − u 1E8 = 0 u 1V 7 − u 1V 1 − L y ∈12 = 0
u 1E5 − u 1E7 = 0 u 1V 4 − u 1V 6 + L y ∈12 = 0
u 1E11 − u 1E9 − L y ∈12 = 0
u 1E10 − u 1E12 − L y ∈12 = 0
590 P. Verma and A. Bajpai

Nodes on faces: Nodes on edges: Nodes on vertices:

∈21 = 0, i = 2 ∈21 = 0, i = 2 ∈21 = 0, i = 2
u 2F2 − u 2F1 − L x ∈21 = 0 u 2E2 − u 2E4 − L x ∈21 = 0 u 2V 3 − u 2V 5 − L x ∈21 = 0
u 2F4 − u 2F3 = 0 u 2E1 − u 2E3 − L x ∈21 = 0 u 2V 2 − u 2V 8 − L x ∈21 = 0
u 2F6 − u 2F5 = 0 u 2E6 − u 2E8 − L x ∈21 = 0 u 2V 7 − u 2V 1 + L x ∈21 = 0
u 2E5 − u 2E7 − L x ∈21 = 0 u 2V 4 − u 2V 6 + L x ∈21 = 0
u 2E11 − u 2E9 = 0
u 2E10 − u 2E12 = 0

3.5 Homogenization

There are different homogenization approaches. Analytical homogenization and

homogenization based on volume averages of field quantities such as stress, strain,
and the effective properties can be found according to the definition of effective
properties of the composites. Finite element analysis can perform the average and
the calculation of local field quantities. The simplest approach of the estimation of
mechanical properties, under the loading condition such as uniaxial tension test [11].
There are several studies conducted for finding the effective properties by imposing
boundary conditions are briefly discussed in [14, 21].

3.5.1 Homogenization Based on FEM

FEM is a tool for micromechanical analysis that can be conveniently applied even
to highly complex microstructures where a discrete mesh can be generated. Similar
to the traditional FEM used in structural mechanics, here the stress and strain fields
at each integration point in the CV are approximated after imposing kinematic or
uniform boundary conditions. In the kinematic boundary conditions, a uniform defor-
mation is applied to two opposite faces of the CV. In the static boundary condition, a
uniform force is applied to one of the faces, and the opposite face is held fixed. The
Elastic modulus and Poisson’s ratio can be extracted using the averages of stress and
strain fields over the whole CV. It is assumed that the average properties of a repre-
sentative volume element are equal to the average properties of the heterogeneous
material.

3.5.2 Average Fields

If the exact micro field σi j and ∈i j in the RVE are Known under the applied load, the
averaged stresses and strain over the RVE are given by Kurukuri and Eckardt [22].
Estimation of Mechanical Behavior of Cementitious Material … 591

1
σi j = σi j dV (3)
V

1
εi j = ∈i j dV (4)
V

where V = Volume of RVE.

The averages are then treated as the effective stress and strain fields in the
homogenized RVE. When linear elasticity is assumed, the generalized Hook’s law
applies

σi j = Ci jkl εi j (5)

Ci jkl is a fourth-order tensor. It is also known as the constitutive matrix.

4 Results and Discussion

4.1 FEM-Based Homogenization

The microstructure of each periodic microstructure domain is captured at various

degrees of hydration (DOH (α) = 0.3, 0.5, or 0.8) in order to study the evolution
of elastic properties of cement paste. In case of small domain size i.e. less than
8 K, the volume fraction of various constituent’s phases cannot be maintained due
to low number of particles present in the computational volume. In the case of low
domain size 1 K the volume fraction of individual constituents becomes unfeasible.
A significant increment is observed in the volume fraction of C–S–H as domain size
is increased from 1 to 8 K. This is because of more active surface of C–S–H is formed.
The elastic tensor (Voigt notation) for an isotropic material is given by
⎡ ⎤ ⎡ ⎤⎡ ⎤
σ11 C11 C12 C13 0 0 0 ∈11
⎢σ ⎥ ⎢C 0 ⎥ ⎢ ⎥
⎢ 22 ⎥ ⎢ 21 C22 C23 0 0 ⎥⎢ ∈22 ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎢ σ33 ⎥ ⎢ C31 C32 C33 0 0 0 ⎥⎢ ∈33 ⎥
⎢ ⎥=⎢ ⎥⎢ ⎥ (6)
⎢ τ12 ⎥ ⎢ 0 0 0 C44 0 0 ⎥⎢ γ12 ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎣ τ23 ⎦ ⎣ 0 0 0 0 C55 0 ⎦⎣ γ23 ⎦
τ13 0 0 0 0 0 C66 γ13

where the individual components of the stiffness tensor are given by

−E 1 (1 − ν23 ν32 )
C11 = (7)

−E 1 (ν21 + ν23 ν31 )
C12 = (8)

592 P. Verma and A. Bajpai

−E 1 (ν31 + ν21 ν32 )

C13 = (9)

−E 2 (ν12 + ν13 ν32 )
C21 = (10)

−E 2 (1 − ν13 ν31 )
C22 = (11)

−E 2 (ν32 + ν12 ν31 )
C23 = (12)

−E 3 (ν13 + ν12 ν23 )
C31 = (13)

−E 3 (ν23 + ν13 ν21 )
C32 = (14)

−E 3 (1 − ν12 ν21 )
C33 = (15)


C44 = G 12 , C55 = G 13 , C66 = G 23 (16)

 = ν12 ν21 + ν13 ν31 ν23 ν32 + ν12 ν23 ν31 + ν13 ν21 ν32 − 1 (17)

To determine the anisotropy more precisely, the following parameters were

described by Kanit et al. [23].

2Y44
a= (18)
Y11 − Y12

where
C44 + C55 + C66 C12 + C23 + C13 C11 + C22 + C23
Y44 = , Y12 = , Y11 =
3 3 3
The coefficient a is equal to 1 if the elastic behavior is perfectly isotropic. So, there-
fore the effective engineering constants for an isotropic material can be determined
by

Y11 = C1 ; Y12 = C2
2
λ = C2 , G = Y11 , K = λ + G
3
λ G(3λ + 2G)
ν= ,E = (19)
2(λ + G) λ+G
Estimation of Mechanical Behavior of Cementitious Material … 593

Table 3 Effective properties obtained from elastic matrix assuming orthotropic symmetry
Domain size (µm) DOH E (GPa) G (GPa) λ (GPa) K (GPa) ν a
20 × 20 × 20 0.8 23.097 9.348567 8.312469 14.54485 0.235334 0.874179
20 × 20 × 20 0.5 18.7496 7.3485 9.034908 13.93395 0.275732 1.031
20 × 20 × 20 0.3 15.4262 5.885233 9.650537 13.57403 0.310591 0.940254
25 × 25 × 25 0.8 23.955 9.801 7.834262 14.36826 0.222119 0.854855

After applying the boundary condition, the total of six pure axial (E i ) and shear
deformation (G i j ) cases are analyzed.

n n n
σ11i
Vi σ22i
Vi σ33i
Vi
E 11 = , E 22 = , E 33 = (20)
i=1
∈11 Vi
i
i=1
∈22 Vi
i
i=1
∈33 Vi
i

n n n
τ12
i
Vi τ23
i
Vi τ31
i
Vi
G 12 = , G 23 = , G 31 = (21)
i=1
γ12
i
Vi i=1
γ23
i
Vi i=1
γ33
i
Vi
n n n
∈i22 Vi ∈i33 Vi ∈i11 Vi
ν12 = , ν23 = , ν31 =
i=1
∈i11 Vi i=1
∈i22 Vi i=1
∈i33 Vi
n n n
∈i11 Vi ∈i22 Vi ∈i33 Vi
ν21 = , ν32 = , ν13 = (22)
∈i
i=1 22
Vi ∈i
i=1 33
Vi i=1
∈i11 Vi

Therefore all E 11 , E 22 , E 33 , G 12 , G 23 , G 31 , ν12 , ν21 , ν23 , ν32 , ν31 , ν13 are calcu-
lated for finding the elastic matrix obtained by applying boundary conditions to
different size of RVE domain at DOH 0.8, 0.5, 0.3 using finite element software
(Table 3).
⎡ ⎤ ⎡ ⎤⎡ ⎤
σ11 29.47 9.03522 7.6743 0 0 0 ∈11
⎢ σ ⎥ ⎢ 9.03513 ⎥⎢ ∈ ⎥
⎢ 22 ⎥ ⎢ 33.30632 8.22781 0 0 0 ⎥⎢ 22 ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎢ σ33 ⎥ ⎢ 7.67692 8.2293 26.3221 0 0 0 ⎥⎢ ∈33 ⎥
⎢ ⎥=⎢ ⎥⎢ ⎥
⎢ τ12 ⎥ ⎢ 0 0 0 10.28 0 0 ⎥⎢ γ12 ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎣ τ23 ⎦ ⎣ 0 0 0 0 8.56 0 ⎦⎣ γ23 ⎦
τ13 0 0 0 0 0 9.2057 γ13

Elastic matrix for RVE size domain 8K at DOH = 0.8

The strength of cement paste is controlled by the formation of C–S–H and the
refinement of pore space. As the hydration process of cement paste proceeds the
formation of C–S–H results in the reduction of free water from the pores which
reduces the Poisson ratio and increment in the elastic modulus as in Figs. 7, 8, and
9 respectively.
594 P. Verma and A. Bajpai

Elastic Modulus M. Shahzamanian et al.E(Gpa)

30
25
20
E (GPa)

15
10
5
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
DOH

Fig. 7 Effect of degree of hydration on elastic modulus (E)

Poisson's Ratio M. Shahzamanian et al ν

0.4
Poisson's ratio (ν)

0.3

0.2

0.1

0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
DOH

Fig. 8 Effect of degree of hydration on Poisson’s ratio

Shear Modulus (Gpa) M. Shahzamanian et al G(Pa)

12
10
8
G (GPa)

6
4
2
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
DOH

Fig. 9 Effect of degree of hydration on shear modulus

Estimation of Mechanical Behavior of Cementitious Material … 595

The elastic modulus (E) and shear modulus (G) of cement paste obtained by the
homogenization based on FEM their variation with respect to DOH is shown in
Fig. 9. The out coming results validated with the previous study by Shahzamanian
et al. [24].

4.2 Analytical Homogenization

Typical representatives of homogenization methods are rule of mixtures [1], Hashin-

Shtrikman bound [25], Mori-Tanaka [26] or self-consistent [19] method. This
approach is also termed as analytical homogenization approach.
The assumption is taken that the analytical homogenization with the real
microstructure of cementitious materials leads to significant refinements of results.
The validity of results checked and compared with the homogenization based on
FEM. The key term is the representative volume element (RVE) that will able to
analyze and capture limited piece of the heterogeneity in the material. Then the
numerical methods must be used for calculation of fluctuation of stress or starin field
to bridge the length scale of the materials. This approach is referred to as a numerical
approach based on FEM. Here we have to discuss the method of prediction of models
and bounds for elastic predictions of composites are reviewed. The method comprise
the following.

4.2.1 Rule of Mixtures Based Homogenization

A rule of mixtures approach independent of the microstructure of the material is used

to compute the effective bulk properties of the cementitious material.
The theoretical extreme upper and lower bounds effective material properties of
multiphase materials are the Voigt and Reuss [27] bounds. The rule of mixtures-based
Voigt upper Bound on the effective bulk (K*) and shear moduli (G) of a mixture of
n phase material is given by
n
K∗ = fi K i (23)
i=1
n
G∗ = fi G i (24)
i=1

The inverse rule of mixtures based Reuss lower bound on effective bulk (K*) and
shear modulus (G*) is given by
n
1 fi
= (25)
K∗ i=1
Ki
596 P. Verma and A. Bajpai

Table 4 Volume fraction of various constituent of cement phases at w/c-0.3 and (α) = 0.8
Phase Phase ID Volume fraction % E (GPa) Poisson’s ratio (µ) K (GPa)
Alite (C3 S) 1 5.1 117.6 0.314 105.3763
Belite (C2 S) 2 5.78 117.6 0.314 105.3763
Aluminate 3 0.33 117.6 0.314 105.3763
(C3 A)
Ferrite (C4 AF) 4 3.03 117.6 0.314 105.3763
Gypsum 5 0.27 45.7 0.33 44.80392
Water-filled 6 12.08 0.001 0.499924 2.192982
porosity
Calcium silicate 7 42.4 22.4 0.25 14.93333
hydrate gel
(C–S–H)
Calcium 8 15.3 42.3507 0.324 40.10483
hydroxide
Ettringite 9 4 22.4 0.25 14.93333
(C6 AS3 H32 )
Monosulfate 10 4.7 0.0423 0.324 0.040057
Iron Hydroxide 11 0.624 22.4 0.25 14.93333
Hydrogarnet 12 4.9 22.4 0.25 14.93333

n
1 fi
= (26)
G∗ i=1
Gi

K ∗ = effective bulk moduli, G ∗ = effective shear moduli

The analytical homogenization is done at water-cement ratio 0.3, 0.35, and 0.4
corresponding to each water-cement ratio volume fraction of each hydrated and
anhydrate phases are taken at the degree of hydration (α = 0.8) as in Tables 4, 5, and
6.
The effective properties of cement paste calculated by the homogenization based
on Finite element method and analytical homogenization using Voigt’s and Reuss’s
bound which gives the upper and lower bounds of composite materials. The effective
elastic properties has been calculated using voigt and reuse model as in Table 7.
So, it is estimated from the above results that the analytical homogeniza-
tion overestimates the effective properties of cement paste when compared with
microstructure-based homogenization.
Above Fig. 10, illustrates the variation of elastic modulus at different water-cement
ratio from the graph it may be concluded that the water-cement ratio increases there
is reduction of elastic properties due to more water content leads to decrease in the
strength of cement paste.
Estimation of Mechanical Behavior of Cementitious Material … 597

Table 5 Volume fraction of various constituent of cement phases at w/c-0.35 and (α) = 0.8
Phase Phase ID Volume fraction % E (GPa) Poisson’s ratio (µ) K (GPa)
Alite (C3 S) 1 5.12 117.6 0.314 105.3763
Belite (C2 S) 2 5.81 117.6 0.314 105.3763
Aluminate 3 0.28 117.6 0.314 105.3763
(C3 A)
Ferrite (C4 AF) 4 3.025 117.6 0.314 105.3763
Gypsum 5 0.25 45.7 0.33 44.80392
Water-filled 6 16 0.001 0.499924 2.192982
porosity
Calcium silicate 7 42.5 22.4 0.25 14.93333
hydrate gel
(C–S–H)
Calcium 8 15.25 42.3507 0.324 40.10483
hydroxide
Ettringite 9 4.75 22.4 0.25 14.93333
(C6 AS3 H32 )
Monosulfate 10 4.12 0.0423 0.324 0.040057
Iron Hydroxide 11 0.66 22.4 0.25 14.93333
Hydrogarnet 12 5.52 22.4 0.25 14.93333

Table 6 Volume fraction of various constituent of cement phases at w/c-0.4 and (α) = 0.8
Phase Phase ID Volume fraction % E (GPa) Poisson’s ratio (µ) K (GPa)
Alite (C3 S) 1 3.18 117.6 0.314 I05.3763
Belite (C2 S) 2 4.08 117.6 0.314 I05.3763
Aluminate 3 0.125 117.6 0.314 I05.3763
(C3 A)
Ferrite (C4 AF) 4 2.23 117.6 0.314 I05.3763
Gypsum 5 0.21059 45.7 0.33 44.80392
Water-filled 6 21.4863 0001 0.49 2.192982
porosity
Calcium silicate 7 37.5 22.4 0.25 14.93333
hydrate gel
(C–S–H)
Calcium 8 13.2 42.3507 0.324 40.10483
hydroxide
Ettringite 9 5.229 22.4 0.25 14.93333
(C6 AS3 H32 )
Monosulfate 10 2.35 0.0423 0.324 0.040057
Iron Hydroxide 11 0.623 22.4 0.25 14.93333
Hydrogarnet 12 5.5 22.4 0.25 14.93333
598 P. Verma and A. Bajpai

Table 7 Comparison of
Effective properties Voigt (V) Reuss (R)
theoretical bounds on
homogenized elastic modulus E 28.40 5.12 × 10-4
for 1 M RVE (DOH = 0.8) ν 0.3145 0.29
K 25.52 0.6235
G 11.83141 2.00 × 10-4

Analytical Homogenization using Voigt Model At DOH 0.8

40
Elastic Modulus, E (GPa)

35
30
25
20
15
10
0.25 0.3 0.35 0.4 0.45
w/c ratio

Fig. 10 Analytical homogenization using Voigt model at a different water-cement ratio at degree
of hydration (α) = 0.8

5 Conclusions

Estimation of elastic properties is an important step in analysis and design of

composite materials. The work aimed at the simulation of the hydration of cement
paste and the overall goal is to determine the effective elastic properties of cement
paste. The microstructure was produced using the modeling platforms. The study
is done to determine the effective properties of type-1, Portland cement using
homogenization based on FEA approach and analytical homogenization.
The elastic properties of the 3-D microstructure of hydrating cement paste were
calculated at the early ages by applying finite element method. And it is found that
low values of elastic modulus are found at a low degree of hydration due to not
much gain of strength of cement paste at early ages. The calcium silicate hydroxide
(C–S–H) gel formed between the particle and creates a percolated network that can
increase the mechanical properties of the microstructure as hydration process take
place. This corresponds to the increase in the elastic moduli. Due to an increase in
the volume fractions of C–S–H and simultaneously decrease in volume fraction of
other major constituents, which leads microstructure to become more isotropic. As
the hydration of the cement paste proceeds, there is the formation of C–S–H that
results in the depletion of free water content in the pores due to pore space is taken
by the portlandite or calcium hydroxide which reduces the Poisson’s ratio. Analytical
homogenization overestimates the effective properties of cement paste as compared
to homogenization based on FEM solution.
Estimation of Mechanical Behavior of Cementitious Material … 599

Acknowledgements We would like to thank Mr. Amardeep Dahiya, Assistant professor at National
institute of technology, Uttarakhand and Dr. Siddharth Shukla, professor at Dr. Rammanohar Lohia
Avadh University for providing their valuable guidance and constant encouragement throughout the
work. This work is simply the reflection of their thoughts, ideas, and concepts and all their efforts.
I am highly obliged to them for his kind and valuable suggestions and of course their valuable time
during the period of the work. I would acknowledge my gratefulness to my parents for their support,
love, patience and for being a constant source of inspiration. The financial support from Ministry
of Human Resource Development (MHRD), Government of India is gratefully acknowledged.

References

1. Mehta PK, Monteiro PJM (2017) Concrete microstructure, properties and materials
2. Neville AM (1995) Properties of concrete, vol 4. Longman, London
3. Garboczi EJ, Bentz DP (1996) Multi-scale picture of concrete and its transport properties:
introduction for non-cement researchers. US Department of Commerce, National Institute of
Standards and Technology, Gaithersburg
4. Bernard O, Ulm F-J, Lemarchand E (2003) A multiscale micromechanics-hydration model for
the early-age elastic properties of cement-based materials. Cem Concr Res 33(9):1293–1309
5. Lee J et al (2009) A multiscale model for modulus of elasticity of concrete at high temperatures.
Cem Concr Res 39(9):754–762
6. Yang H et al (2017) A critical review on research progress of graphene/cement based
composites. Compos A Appl Sci Manuf 102:273–296
7. Zaoui A (2002) Continuum micromechanics: survey. J Eng Mech 128(8):808–816
8. Torquato S (2002) Random heterogeneous materials. Microstructure and macroscopic proper-
ties. Interdisc Appl Math 16
9. Milton GW (2002) The theory of composites (Cambridge Monographs on Applied and
Computational Mathematics). Cambridge University Press, Cambridge
10. Christensen RM (2012) Mechanics of composite materials. Courier Corporation
11. Haecker C-J et al (2005) Modeling the linear elastic properties of Portland cement paste. Cem
Concr Res 35(10):1948–1960
12. Bentz DP, Bentz DP (1995) A three-dimensional cement hydration and microstructure program:
I. hydration rate, heat of hydration, and chemical shrinkage. US Department of Commerce,
National Institute of Standards and Technology
13. Bullard JW (2014) Virtual cement and concrete testing laboratory: version 9.5 user guide
14. Bernard F, Kamali-Bernard S, Prince W (2008) 3D multi-scale modelling of mechanical
behaviour of sound and leached mortar. Cem Concr Res 38(4):449–458
15. Velez K et al (2001) Determination by nanoindentation of elastic modulus and hardness of pure
constituents of Portland cement clinker. Cem Concr Res 31(4):555–561
16. Monteiro PJM, Chang CT (1995) The elastic moduli of calcium hydroxide. Cem Concr Res
25(8):1605–1609
17. Boumiz A et al (2000) Modelling the development of the elastic moduli as a function of the
hydration degree of cement pastes and mortars. In: 2nd international RILEM workshop on
hydration and setting: why does cement set? An interdisciplinary approach (Dijon, 11–13 June
1997)
18. Taylor HF (1997) Cement chemistry. Thomas Telford
19. Hill R (1965) Theory of mechanical properties of fibre-strengthened materials—III. Self-
consistent model. J Mech Phys Solids 13(4):189–198
20. Nemat-Nasser S, Hori M (2013) Micromechanics: overall properties of heterogeneous
materials, vol 37. Elsevier
600 P. Verma and A. Bajpai

21. Ren Z-Y, Zheng Q-S (2002) A quantitative study of minimum sizes of representative volume
elements of cubic polycrystals—numerical experiments. J Mech Phys Solids 50(4):881–893
22. Kurukuri S, Eckardt S (2004) A review of homogenization techniques for heterogeneous
materials. Weimar, Germany
23. Kanit T et al (2006) Apparent and effective physical properties of heterogeneous materials:
representativity of samples of two materials from food industry. Comput Methods Appl Mech
Eng 195(33–36):3960–3982
24. Shahzamanian MM et al (2014) Representative volume element based modeling of cementitious
materials. J Eng Mater Technol 136(1)
25. Hashin Z, Shtrikman S (1963) A variational approach to the theory of the elastic behaviour of
multiphase materials. J Mech Phys Solids 11(2):127–140
26. Mori T, Tanaka K (1973) Average stress in matrix and average elastic energy of materials with
misfitting inclusions. Acta Metall 21(5):571–574
27. Zuo L, Humbert M, Esling C (1992) Elastic properties of polycrystals in the Voigt-Reuss-Hill
approximation. J Appl Crystallogr 25(6):751–755