Downloaded from orbit.dtu.

dk on: Jan 13, 2021

Dynamic Performance of a PEM Fuel Cell System

Rabbani, Raja Abid

Publication date:
2013

Document Version
Publisher's PDF, also known as Version of record

Link back to DTU Orbit

Citation (APA):
Rabbani, R. A. (2013). Dynamic Performance of a PEM Fuel Cell System. DTU Mechanical Engineering.
DCAMM Special Report, No. S154

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright
owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

 Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
 You may not further distribute the material or use it for any profit-making activity or commercial gain
 You may freely distribute the URL identifying the publication in the public portal

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Dynamic Performance of a PEM Fuel
Cell System

Abid Rabbani

PhD Dissertation

Department of Mechanical Engineering

Technical University of Denmark
Kongens Lyngby 2013
Dynamic Performance of a PEM Fuel Cell system

Copyright © 2013 by Abid Rabbani. All rights reserved

PhD Thesis

DTU Mechanical Engineering

Thermal Energy Section
Technical University of Denmark

Nils Koppels Allé, Bldg. 403,

DK–2800 Kongens Lyngby,
Denmark.
Phone: (+45) 4525 4107
Fax: (+45) 4588 4325
www.mek.dtu.dk
ISBN: 978–87–7475–359–9
DCAMM Special Report: S154

ii
 Prefaace

This thesis is submitted

s in partial fulfilment of the req quirementss for the Ph.D.
P
degree at the Technicaal University of Deenmark (D DTU). Thee research was
condducted und der the suppervision oof Assoc. Prof.
P Masouud Rokni at the Theermal
Enerrgy Section, Departtment of Mechaniccal Engineeering andd represen nts a
culm
mination off work and learning tthat has takken place from
f April 2010 to March
M
20133. The project was reaalized withiin the fram
mework of LINK2009
L wwhich is mainly
m
finannced by th he Danish
h Energy A Agency EU UDP progrram. Presennted work k was
carriied out in direct colllaborationn with H2L Logic A/S, a compaany involveed in
desiggn and devvelopment of
o automottive fuel celll systems.

This thesis maainly coverrs dynamicc aspects of a PEM M fuel cell system an nd is
preseented as a monograp ph, but it aalso contain
ns a number of paperrs based on
n the
workk in this ressearch.

Abbid Rabbanni
Lyngbby, March 2013

iii
Preface

iv
 Acknowledgements

A PhD dissertation is apparently a work of one, but only with the support of many
helping hands behind the curtain. Fortunately for my case, I was accompanied by
an extended team of professionals who contributed in their different ways to help
me through the completion of this thesis. For this, I would like to extend my
appreciation to them.

First and foremost, I offer my sincerest gratitude to my supervisor and mentor,

Assoc. Prof. Masoud Rokni, whose patience and kindness, as well as academic
experience have been invaluable to me. I appreciate his vast knowledge and
expertise in many areas, his assistance in writing scientific papers, and allowing me
the freedom and opportunity to select the course of my research. I must also
acknowledge Henrik Mortensen from H2Logic Company, for his suggestions and
collaboration throughout this period. Special thanks to my colleague Elham
Hosseinzadeh for engaging in many productive discussions and her valuable
suggestions.

The support and encouragement of my colleagues at TES section has been

indispensable, and I would like to extend my deepest gratitude to all of them.
During my stay at DTU, I got to share my office with fellow PhD students who
truly broadened the value of work: Tuong–Van Nguyen, Jorrit Wronski and
Leonardo Pierobon. I very much enjoyed exchange of ideas and thoughts, and all
the fun moments together. Furthermore, I am grateful to the Head of Section
Assoc. Prof. Brian Elmegaard, for always taking time to discuss problems and his
encouragement. Special thanks go to Arne Egelund for his valuable comments and
assistance in translation of thesis abstract. I owe heartfelt appreciation to the
section secretary, Eva Stitz Larsen, for her motherly love and exceptional care
during the course of my studies.

v
Acknowledgements

I would take this opportunity to express due indebtedness to my beloved parents,

who have been a constant source of emotional, moral and financial support
throughout my life. I must not forget to thank my brothers, each of them with
unique traits: Amjad’s integrity, Majid’s foresight, Sajid’s perseverance and Atif’s
wisdom, all of which I admire and seek for my personal and professional
development. Undoubtedly, my wife Saima deserves a special word of gratitude for
her understanding, moral support, patience and love through the duration of my
research.

vi
 Abstract

Concerns about climate change and projections of fossil fuel depletion within the
next century have prompted our interest towards energy efficient and therefore
more sustainable alternatives. Transport sector is one of the major consumers of
oil and predictions imply a further surge with an increasing trend of more vehicles.
This necessitates focus for alternative fuel which is not only sustainable but also
environmentally benign well into the future. Hydrogen is expected to play a major
role in future energy economy in this context and fuel cells have shown a
significant potential as an efficient solution for harnessing energy of hydrogen.
The polymer electrolyte membrane fuel cells (PEMFCs) currently appear to be the
preferred fuel cell for the automotive industry due to their inherent virtue of being
compact with quick start–up and higher efficiencies with very low emissions.
Nonetheless, there are many issues related to the fuel cell durability, cost and
performance which hinder their commercialization and competence with present
automotive engines.

The overall aim of this research is to investigate the dynamic behaviour of a

PEMFC system in order to improve the efficiency of the system. Such systems
when employed in automobiles are subjected to continuous changes in load and
external operating conditions, making it important to investigate the dynamic
performance of the system. Current research was focused towards addressing key
problems related to fuel and air supply, species transportation within the fuel cell,
water and thermal management of the system. These objectives were achieved by
firstly developing a PEMFC stack and a humidifier model in a component based
simulation tool (Aspen Dynamics) and then simulations of the entire system with
integrated controls were carried to emulate an actual fuel cell system operating
under varying operating conditions with the influence of external parameters.

vii
Abstract

Effects of these controls and other components on overall performance of the

system were also investigated in this work.

System analyses were conducted for start–up sequence and transitory load changes.
Issues related to thermal and water management were addressed with considerable
focus on species crossover mechanisms across the fuel cell. Finally, study on
nitrogen crossover was conducted and associated purging strategies are discussed.
Different purging techniques were simulated to further gain insight into hydrogen
dilution issues caused by nitrogen accumulation in the fuel cell anode. The
research conducted here could be helpful in understanding intricate transient
processes within the fuel cell and system model could be used as a thorough tool
for predicting PEMFC dynamics and to contribute as a basis for the design and
optimization of system controls and purging strategies.

viii
 Resumé

Bekymringer om klimaændringer og forudsigelser om udtømning af de fossile

brændstofressourcer inden for det næste århundrede har resulteret i et ønske om
energieffektive og dermed mere bæredygtige alternativer. Transportsektoren er en
af de største forbrugere af olie og med den forventede stigning i antal køretøjer vil
dette forbrug yderligere stige. Dette nødvendiggør at der sættes fokus på alternative
brændstoffer, som ikke kun er ressourcemæssigt bæredygtige, men også
miljøvenlige langt ind i fremtiden. Brint forventes at spille en stor rolle i
fremtidens energiøkonomi og i denne sammenhæng har brændselsceller vist et
betydeligt potentiale som en effektiv måde at udnytte energien i brint.
Polymerelektrolytmembran brændselsceller (PEMFCs) synes i øjeblikket at være
den foretrukne brændselscelle til bilindustrien i kraft af at være kompakt med
hurtig opstart, højere effektivitet og meget lave emissioner. Ikke desto mindre er
der mange forhold i forbindelse med brændselscellers holdbarhed, deres
omkostninger og ydeevne, som vanskeliggør deres anvendelse i eksisterende
bilmotorer– og dermed deres kommercialisering.

Det overordnede formål med denne forskning er at undersøge den dynamiske

opførsel af et PEMFC system med henblik på at forbedre effektiviteten af systemet.
Sådanne systemer er, når de anvendes i biler, udsat for stadige ændringer i
belastning og eksterne driftsbetingelser, hvilket gør det vigtigt at undersøge
systemets dynamiske ydeevne. Den udførte forskning var fokuseret på at løse
centrale problemer i forbindelse med brændstof–og lufttilførsel, stoftransport
inden for brændselscellen samt styring af vand–og varmestrømme i systemet. Disse
mål blev opnået ved først at udvikle modeller af en PEMFC stak og en
luftfbefugter i et komponentbaseret simuleringsværktøj (Aspen Dynamics) og
derefter simulere hele systemet med integreret styring med det formål at efterligne

ix
Resumé

en virkelig brændselscelle, der arbejder under forskellige driftsbetingelser og under

indflydelse af ydre parametre. Virkningen af styringesenheder og andre
komponenter på den samlede ydelse af systemet blev også undersøgt i dette
arbejde.

Systemanalyser blev udført for opstartsekvenser og forbigående

belastningsændringer. Spørgsmål vedrørende styring af vand– og varmestrømme
blev behandlet med særligt fokus på stofovergang på tværs af brændselscellen.
Endelig blev undersøgelse af kvælstofovergang gennemført og relevante
rensningsstrategier diskuteret. Forskellige rensningsteknikker blev simuleret for
yderligere at få indsigt i problemet med brintfortynding som følge af
nitrogenakkumulering i brændselscellens anode. Den forskning, som er udført
her, vil kunne være en hjælp til at forstå indviklede forbigående processer i
brændselscellen, og systemmodellen vil kunne anvendes som et nyttigt redskab til
at forudsige dynamikken af en PEMEC og danne grundlag for design og
optimering af styrings– og rensningsstrategier for systemet.

x
 List of Publications

The main body of this thesis is based on the following papers:

Rabbani, A & Rokni, M. “Dynamic Simulation of a Proton Exchange Membrane

Fuel Cell System for Automotive Applications”, Proceedings of Sustainable
Energy & Environmental Protection (SEEP) Conference, Dublin, 2012; 311–
316.

Rabbani A, Rokni M, Hosseinzadeh E, Mortensen H. “Start–up Analysis of a PEM

Fuel Cell System in Vehicles”, International Journal of Green Energy, 2013; 11:
1–21.

Rabbani, A & Rokni, M. “Dynamic characteristics of an automotive fuel cell

system for transitory load changes”, Journal of Sustainable Energy Technologies
& Assessments, 2013; 1: 34–43.

Rabbani, A & Rokni, M. “Effect of nitrogen crossover on purging strategy in PEM

fuel cell systems”, accepted for Journal of Applied Energy, 2013.

Other related publications

Hosseinzadeh E, Rokni M, Rabbani A, Mortensen H. “Thermal and water
management of low temperature PEM fuel cell in forklift truck power system”,
Journal of Applied Energy, 2013; 104: 434–444.

xi
List of Publications

xii
 Contents

Preface iii 
Acknowledgements v 
Abstract vii 
Resumé ix 
List of Publications xi 
Contents xiii 
List of Figures xvii 
List of Tables xxi 
Nomenclature xxiii 
1  Introduction 1 
1.1  Motivation ................................................................................................ 1 
1.2  Scope of the research project .................................................................... 2 
1.3  Objectives of the thesis ............................................................................. 3 
1.4  Methodology ............................................................................................. 3 
1.5  Limitations and Validation....................................................................... 4 
1.6  Overview of Thesis .................................................................................... 5 
2  Background 7 
2.1  Hydrogen economy ................................................................................... 7 
2.2  PEM fuel cells for automobiles ................................................................. 9 

xiii
Contents

2.3  PEM Fuel cell system .............................................................................. 12 

2.3.1  Fuel cell theory ...................................................................................... 12 
2.3.2  Reactant supply system........................................................................ 14 
2.3.3  Water Management............................................................................... 14 
2.3.4  Heat Management ................................................................................. 14 
2.3.5  Control system ...................................................................................... 15 
2.4  Literature review ..................................................................................... 15 
3  System Configuration 19 
3.1  PEM Fuel Cell stack................................................................................ 21 
3.1.1  Membrane ............................................................................................. 22 
3.1.2  Catalyst Electrode layer ......................................................................... 22 
3.1.3  Gas Diffusion Layer .............................................................................. 23 
3.1.4  Bipolar plates......................................................................................... 23 
3.2  Balance of Plant (BoP) ............................................................................ 23 
3.2.1  Humidifier ............................................................................................ 24 
3.2.2  Air Blower and pumps .......................................................................... 25 
3.2.3  Heat Exchangers.................................................................................... 26 
3.2.4  Flow valves............................................................................................. 26 
3.2.5  Water trap ............................................................................................. 27 
4  System controls 29 
4.1  Operating conditions .............................................................................. 30 
4.2  Air flow control....................................................................................... 33 
4.3  Fuel flow control ..................................................................................... 35 
4.4  Temperature control ............................................................................... 36 
5  Modelling of PEM Fuel Cell Stack 39 
5.1  Fuel cell Potential ................................................................................... 39 
5.1.1  Thermodynamic Potential .................................................................... 39 
5.1.2  Theoretical fuel cell potential ............................................................... 40 
5.2  Electrochemical Model ........................................................................... 41 
5.2.1  Activation loss ....................................................................................... 42 
5.2.2  Ohmic Loss ........................................................................................... 44 
5.2.3  Concentration Loss ............................................................................... 45 
5.3  Membrane Transportation Sub–models ................................................ 46 
5.3.1  Water crossover ..................................................................................... 47 
5.3.2  Nitrogen crossover ................................................................................ 50 
5.4  Flow Modelling ....................................................................................... 51 
5.5  Heat Transfer .......................................................................................... 52 
5.6  System Efficiency .................................................................................... 53 
5.7  Model Validation .................................................................................... 54 

xiv
 Contents

6  Exergy analysis of Fuel Cell Stack 57 

6.1  Overview ................................................................................................. 57 
6.2  Laws of thermodynamics and PEM fuel cells ......................................... 58 
6.3  Exergy balance in PEM fuel cell.............................................................. 60 
6.4  Simulation results ................................................................................... 62 
6.4.1  Exergy efficiency of PEM Fuel Cell ....................................................... 63 
6.4.2  Effect of stack current ........................................................................... 65 
6.4.3  Influence of operating temperature ...................................................... 66 
6.4.4  Impact of ambient temperature ............................................................ 68 
6.4.5  Effect of liquid water ............................................................................. 69 
6.5  Summary ................................................................................................. 70 
7  System Analysis on Start–up 71 
7.1  Premise .................................................................................................... 71 
7.2  System start–up ....................................................................................... 72 
7.2.1  Stack warm–up periods ......................................................................... 73 
7.2.2  Power generation and system efficiency ................................................ 75 
7.2.3  Cell voltage at start–up sequence .......................................................... 77 
7.2.4  Fuel and oxidant flows .......................................................................... 79 
7.2.5  Air humidification effects ..................................................................... 81 
7.2.6  Water saturation at cathode............................................................... 83 
7.3  Summary ................................................................................................. 84 
8  Transitory load changes –System Analysis 85 
8.1  Overview ................................................................................................. 85 
8.2  Results and discussion ............................................................................ 86 
8.2.1  Voltage loss during load change............................................................ 86 
8.2.2  Thermal management of the system ..................................................... 87 
8.2.3  System efficiency and power consumption ........................................... 89 
8.2.4  Water management during load changes ......................................... 91 
8.2.5  Stream properties at fuel cell ports ....................................................... 94 
8.3  Summary ................................................................................................. 96 
9  Nitrogen crossover & anode purging methodologies 97 
9.1  Premise .................................................................................................... 97 
9.2  Simulation results and discussion .......................................................... 98 
9.2.1  Gas Buildup at anode ......................................................................... 99 
9.2.2  Anode outlet bleed .............................................................................. 101 
9.2.3  Anode outlet purge ............................................................................. 102 
9.2.4  Automatic Anode purge...................................................................... 104 
9.2.5  Anode purge at load changes .............................................................. 106 
9.3  Summary ............................................................................................... 109 

xv
Contents

10   Concluding remarks 111 

10.1  General summary .................................................................................. 111 
10.2  Summary of analysis.............................................................................. 112 
10.2.1  Modelling of the PEM fuel cell system................................................ 112 
10.2.2  Start–up analysis of fuel cell system .................................................... 113 
10.2.3  System response during transitory load changes ................................. 114 
10.2.4  Effect of nitrogen crossover on purging strategy ................................. 114 
10.3  Improvements for future work ............................................................. 115 
Bibliography 119 
Appendix A 129 
A.1.  PEM Fuel Cell component model ........................................................ 129 
A.2.  Humidifier component model ............................................................. 151 
A.3.  Control system codes ............................................................................ 159 
A.4.  Aspen Tasks, feed streams & efficiency ................................................ 165 
Appendix B 171 
Paper I............................................................................................................. 171 
Appendix C 199 
Paper II ........................................................................................................... 199 
Appendix D 223 
Paper III .......................................................................................................... 223 
Appendix E 249 
Paper IV .......................................................................................................... 249 
 
 
 

xvi
 List of Figures

Figure 2.1: Energy consumption and corresponding GHG emissions for

different types of hydrogen production processes . .......................................... 8
Figure 2.2: Predictions for GHG pollution for different vehicles . ..................................... 9
Figure 2.3: Vehicle mass for different types of battery electric vehicles in
comparison with fuel cell vehicles . ................................................................ 10
Figure 2.4: Comparison of storage volumes for battery operated vehicles and fuel
cell electric vehicles . ....................................................................................... 11
Figure 2.5: Working principle of a single PEM fuel cell . .................................................. 13
Figure 3.1: Schematic of a PEM fuel cell system with auxiliary components and
control layout. ................................................................................................. 19
Figure 3.2: Structure of a single fuel cell ........................................................................... 21
Figure 3.3: Exploded view of PEM fuel cell stack . ............................................................ 22
Figure 3.4: (a) Efficiency profile of air compressor versus air mass flow, (b) Power
consumption of air compressor at different stack currents. ........................... 25
Figure 4.1: Control blocks for the PEMFC system. ........................................................... 30
Figure 4.2: Stack data provided by Ballard (a) Nominal pressure drop within the
fuel cell, (b) Inlet pressures for reactants. ....................................................... 32
Figure 4.3: (a) Reactant stoichiometry used in the fuel cell control system, (b)
Stack inlet and outlet temperature difference maintained by coolant
mass flow. ........................................................................................................ 33
Figure 4.4: Air flow controls (a) Air compressor power versus stack current, (b)
Controller output variation with change in process variable. ........................ 34
Figure 4.5: Change in O2 stoichiometry with current drawn from stack. ......................... 35
Figure 4.6: Fuel flow controls (a) Anode inlet pressure versus stack current, (b)
Control input and output variables. ............................................................... 35
Figure 4.7: (a) Fuel stoichiometry as a function of current, (b) Power
consumption of recirculation pump against stack current. ............................ 36

xvii
List of Figures

Figure 4.8: Temperature controls (a) PID controller process and output
variables, (b) Stack operating temperature as a function of coolant
mass flow. ........................................................................................................ 37
Figure 4.9: Thermal management (a) Coolant inlet temperature versus external
coolant flow, (b) Power consumption graph of air radiator. .......................... 38
Figure 5.1: Representation of a typical Voltage–Current curve denoting different
polarization dominated regions . .................................................................... 42
Figure 5.2: Schematic diagram of gas transport phenomena across PEM fuel cell
MEA. ............................................................................................................... 47
Figure 5.3: (a) Comparison of stack operational data and calculated polarization
curves for PEM fuel cell with relative error, (b) System efficiency
profile at operating current range. .................................................................. 54
Figure 5.4: (a) Calculated H2 consumption versus Ballard data with relative
error, (b) Heat and power generation with stack efficiency at
corresponding current densities. .................................................................... 55
Figure 6.1: Schematic of exergy balance for a PEM fuel cell stack. ................................... 60
Figure 6.2: (a) Energy and exergy efficiency of PEMFC at different current loads,
(b) Exergy destruction and entropy generation at increasing current
densities. ......................................................................................................... 63
Figure 6.3: Total exergy entering and leaving the fuel cell stack at different
operating currents with the amount of fuel cell power produced and
exergy destroyed. ............................................................................................. 64
Figure 6.4: Percentage of exergy destruction at corresponding FC power
production. ..................................................................................................... 64
Figure 6.5: Energy and exergy efficiency of stack at system start–up: (a) at current
of 40A, (b) at current of 100A. ....................................................................... 65
Figure 6.6: Cell voltages and stack temperature at system start–up: (a) at current
of 40A, (b) at current of 100A. ....................................................................... 66
Figure 6.7: Effect of increase in stack operating temperature by 10°C (a)
Exergetic efficiency and liquid water fraction, (b) heat produced by
the FC stack. ................................................................................................... 67
Figure 6.8: Entropy generation in the stack with change in stack operating
temperature. .................................................................................................... 67
Figure 6.9: Effect of ambient temperature on entropy generation and exergetic
efficiency; (a) ambient temperature of 25°C and (b) 15°C
respectively. ..................................................................................................... 68
Figure 6.10: Heat produced by liquid water and its effect on exergetic efficiency
at system start–up: (a) at current of 40A, (b) at current of 100A. .................. 70
Figure 7.1: Start–up at 40A; (a) Stack heat–up times and corresponding coolant
flow rate, (b) Flow rates in external cooling circuit and air radiator. ............. 73
Figure 7.2: Start–up at 60A; (a) heat–up time and coolant flow, (b) External
cooler and radiator flow rates. ........................................................................ 74
Figure 7.3: Start–up at 80A; (a) heat–up time and coolant flow, (b) External
cooler and radiator flow rates. ........................................................................ 74

xviii
 List of Figures

Figure 7.4: Start–up at 100A; (a) Stack heat–up times and corresponding coolant
flow rate, (b) Flow rates in external cooling circuit and air radiator. ............. 75
Figure 7.5: System efficiency and stack power; (a) at start–up current of 60A (b)
at start–up current of 100A, auxiliary power consumption at start–
up of 60A (c) and 100A (d) respectively. ........................................................ 76
Figure 7.6: (a) Cell voltage as a function of operating temperature with start–up
current of 80A, (b) Stack power versus cell current density. .......................... 77
Figure 7.7: Effect of sudden temperature drop on voltage and power. ............................. 78
Figure 7.8: Contribution of different losses in cell voltage reduction. .............................. 79
Figure 7.9: Fuel and air flows at inlet and outlet of the stack: (a) Fuel flow at
anode inlet, (b) Fuel flow at anode outlet, (c) Air flow at cathode
inlet and (d) Air flow at cathode outlet. ......................................................... 80
Figure 7.10: Effect of air inlet relative humidity on fuel inlet and outlet: (a) Air
inlet humidity of 95%, (b) 90%, (c) 85% and (d) 80% respectively. .............. 81
Figure 7.11: Relative humidity of fuel at inlet and outlet: (a) with 1–2% liquid at
anode inlet (b) with no liquid at anode inlet.................................................. 82
Figure 7.12: Percentage of liquid water at cathode outlet; (a) at start–up current
of 40A and (b) at start–up current of 100A.................................................... 83
Figure 8.1: (a) Voltage change with stack current, (b) Variation in overpotentials
at transitory change of 60–100–60A. ............................................................. 87
Figure 8.2: Current ramp–up from 60–100A and vice versa: (a) Changes in cell
voltages and operating temperature, (b) Changes in coolant flow rate
to maintain the fuel cell operating temperature, (c) variation in
external cooling flows and air radiator. .......................................................... 88
Figure 8.3: (a) System efficiency variation with load change, (b) stack power
output and auxiliary power consumption. ..................................................... 90
Figure 8.4: Auxiliary power consumption versus current; (a) Air blower power,
(b) Air radiator and recirculation consumption. ............................................ 90
Figure 8.5: Current ramp–up from 60–100A and vice versa: (a) Water crossover
through PEMFC membrane, (b) Effect of water transportation on
anode RH........................................................................................................ 91
Figure 8.6: Current ramp–up from 60–100A and vice versa: (a) Water activity in
fuel cell, (b) Back diffusion coefficient for water in membrane. .................... 92
Figure 8.7: Liquid water at cathode outlet; (a) Current ramp–up from 60A to
100A and vice versa, (b) Current ramp–up from 60A to 120A and
back. ................................................................................................................ 93
Figure 8.8: Current ramp–up from 60–100A and vice versa:, (a) Stream
properties at anode inlet and (b) outlet respectively, (c) Effect on fuel
stoichiometry and recirculation pump, (d) Air flow into the cathode. .......... 95
Figure 9.1: Calculated nitrogen crossover rates against provided experimental
data. ................................................................................................................ 99
Figure 9.2: N2 permeance and crossover rate at (a) fuel cell start–up and (b)
Steady–state operations................................................................................. 100

xix
List of Figures

Figure 9.3: (a) Cell voltage decay due to N2 buildup and (b) Increment in liquid
water at anode exhaust.................................................................................. 101
Figure 9.4: (a) N2 levels for different bleed fractions at anode inlet and (b)
Composition of bleed stream from system start–up..................................... 102
Figure 9.5: Performance of Fuel cell for fixed purge interval of 60 s; (a) Voltage
rejuvenation with semi–stable profile and (b) system efficiency and
power during purge sequences...................................................................... 103
Figure 9.6: Molar fractions at purge routines and H2 vent out rate. ............................... 104
Figure 9.7: Automatic purge intervals and corresponding molar fractions of
purges stream; (a) at drawn current of 120 A, and (b) 60 A. ........................ 105
Figure 9.8: N2 crossover through membrane for fixed purge interval routine
during load changes from 120A to 60A and vice versa. ............................... 106
Figure 9.9: Effect of load changes of 120–60–120 A on fixed purge interval; (a)
H2 purge rate (b) N2 concentration at anode inlet and cell voltage. ............. 107
Figure 9.10: N2 crossover through membrane for automatic purge interval
sequence during load changes from 120 A to 60 A and vice versa. ............. 108
Figure 9.11: Effect of load changes of 120–60–120 A on automatic purge
interval; (a) H2 purge rate (b) N2 concentration at anode inlet and
cell voltage. .................................................................................................... 109

xx
 List of Tables

Table 3.1: Nominal operating conditions for PEMFC stack ..................................... 20 
Table 5.1: Key parameters for Nafion 117 membrane ............................................... 49 
Table 6.1: Maximum achievable voltage and efficiency for H2–O2 fuel cells
at different temperatures ............................................................................. 59 
Table 6.2. Standard molar chemical exergy, of substances at 298.15 K and 1
bar................................................................................................................... 62 

xxi
List of Tables

xxii
 Nomenclature

Roman symbols
A area (cm2)
a activity (–)
aw water activity (–)
C constant (–)
Ct stack thermal capacitance (kW)
c concentration (mol cm–3)
Dw water diffusion (cm2s–1)
D water diffusion coefficient (–)
E theoretical voltage (V)
e error (%)
E N2 activation energy (kJ mol–1)
E exergy rate (kW)
E x energy rate (kW)
E x D rate of exergy destruction (kW)
F Faraday’s constant (C mol–1K–1)
fv volumetric ratio of water in membrane (–)
g 0f Gibbs free energy (J mol–1)
H enthalpy (J mol–1)
I current (A)
i current density (A cm–2)
i0 exchange current density (A cm–2)
ilim limiting current density (A cm–2)
iloss internal current density (A cm–2)

xxiii
Nomenclature

imax maximum current density (A cm–2)

J H 2O water molar flux (mol s–1cm–2)
J N2 molar flux of nitrogen (mol s–1cm–2)
k reaction speed constant (mol s–1cm–2)
KP proportional gain (–)
K N2 nitrogen permeance (kmol s–1cm–1bar–1)
Mm molecular weight of membrane (Kg mol–1)
n molar flow rate (mol s–1)
Ncell number of cells (–)
ne electrons transferred (mole molfuel–1)
nel number of electrons in rate step reaction (–)
ndrag electro osmotic drag (–)
OP controller output (–)
P pressure (bar)
P average pressure (bar)
PH 2 hydrogen partial pressure (–)
PO2 oxygen partial pressure (–)
P net stack power (kW)
Pstack power produced by stack (kW)
Q heat transfer rate (kW)
r resistance (Ω cm2)
R universal gas constant (J mol–1K–1)
RH relative humidity (–)
S stoichiometric ratio (–)
S entropy (J mol–1K–1)
T temperature (K)
t instantaneous time (s)
 integration variable (s)
tm membrane thickness (cm)
V voltage (V)
W rate of work delivered (kW)
yi molar fraction of species (–)

Greek symbols
 transfer coefficient (–)
N 2
perm scale factor (–)

xxiv
 Nomenclature

 symmetry factor (–)

 efficiency (–)
 water content (–)
 dry membrane density (g cm–3)
m membrane activity (S cm–1)
 constant (–)

Subscripts
a anode
act activation
air air
Ar argon
c cathode
cell single fuel cell
conc concentration
d derivative
diffusion back diffusion
drag electro–osmotic drag
dry dry membrane
el electrical
exh fuel cell cathode exhaust
f hydrogen
FC fuel cell
i integral
in inlet
inv inverter
int rev internally reversible
ion ionic
loss loss to the ambient
mass mass flow
max maximum
mem membrane
ohmic ohmic
out outlet
ref reference state
sat saturated
st stored
stack fuel cell stack
sys system

xxv
Nomenclature

Superscripts
0 standard conditions
a anode
c cathode
ch chemical
ph physical
sat saturation
t total

Abbreviations
BEV Battery Electric Vehicle
BoP Balance of Plant
DOE Department of Energy, USA
DTU Technical University of Denmark
EMF Electro–Motive Force
FCV Fuel Cell Vehicle
FCEV Fuel Cell Electric Vehicle
FCHEV Fuel Cell Hybrid Electric Vehicle
GDL Gas Diffusion Layer
GHG Greenhouse Gas
HHV Higher Heating Value
ICE Internal Combustion Engine
LHV Lower Heating Value
MEA Membrane Electrode Assembly
OCV Open Circuit Voltage
PEMFC Polymer Electrolyte Membrane Fuel Cell
PFSA Per–Fluoro Sulfonic Acid
PID Proportional Integral Derivative
SOFC Solid Oxide Fuel Cell

xxvi
Chapter 1

1 Introduction

1.1 Motivation

For decades, fossil fuels have dominated the energy market of the world. However,
it is becoming increasingly obvious that they ought to diminish sooner or later.
Regardless of it, our growing reliance on these fuels is another concern. It is
reported that fossil fuels provided 85% of the total primary energy demand in
2008, which raises a number of concerns. Combustion of these fossil fuels leads to
local and global emissions of pollutants such as NOx, SOx and greenhouse gases
(GHG). It is believed that GHG emissions associated with the provision of energy
services is a major cause of climate change. Increase in global average temperature
since the mid of 20th century is undoubtedly attributed to the increase in GHG
levels. Furthermore, the combustion of fossil fuels accounted for 56.6% of all
anthropogenic GHG emissions and these concentrations have continued to grow
39% above pre–industrial levels by the end of 2010. In the last two centuries,
global use of fossil fuels (coal, oil and gas) has substantially increased to lead the
energy supply market, resulting in rapid surge of CO2 emissions. The amount of
carbon in fossil fuel reserves and resources not yet burned has the potential to add
more quantities of CO2 to the atmosphere and the overwhelming majority of the
projections show even higher emissions in 2100 compared with those in 2000.
Rise in GHG concentrations imply an increase in global mean temperatures with
potentially irreversible climate changes, contributing to adverse impacts on water
resources, ecosystems, human health, food security and aquatic life near coasts. In
order to be confident of deterring the temperature, atmospheric GHG
concentrations would need to be stabilized; i.e. that global emissions of CO2 have
to decrease by 50 to 85% below the levels of year 2000 by 2050. At the moment,
CO2 concentrations are continuing to increase, necessitating an immediate goal of
inverting this trend towards a decline till 2015 [1].

1
Introduction 1

Transport sector’s utilization of world primary energy consumption of oil has

increased from 45.4 % (1973) to 61.5 % (2010), which is predicted to surge with
an increasing trend of more vehicles [2]. Also, world mobility is bound to escalate
with growing population in developing countries like China and India.
Consequently, individual mobility in the future is directly associated with the
availability of fuel for transportation. Though oil production is expected to peak in
the near future, oil reserves are finite and will deplete soon. Scarcities of energy
sources are another reason for price fluctuations and speculations in the energy
market, which raise the issue of energy security for many countries which have a
strong dependence on imported oil.

All these factors necessitate focus for alternative fuels which are not only
sustainable but also environmental friendly. Hydrogen is suggested to be one of
the options and is argued to be an attractive substitute for petroleum as the main
fuel for transportation in the future. Fuel cell systems, being the most capable
utilizers of hydrogen, have received substantial attention in recent years and
research on these systems has drastically increased mainly due to their
characteristic advantages of clean and efficient energy conversion. Existing fuel cell
systems are categorized based on the type of electrolyte and preferred operating
conditions. Among various types of fuel cells, the Proton Exchange Membrane
Fuel Cells (PEMFC) is currently the best choice for portable power generation due
to its relatively low operating temperature, quick start–up, high power density and
efficiency, system robustness and low degradation due to corrosion. This thesis
focuses on PEMFC system for automotive applications such as forklifts.

1.2 Scope of the research project

The work presented here is a part of a larger project (LINK–2009), which is a

consortium of academia and industrial partners. This project encompasses
different aspects of hydrogen based infrastructure with specific focus on
implementation of fuel cell technologies including a demonstration of a new PEM
Fuel Cell Vehicle. The project consists of ten work packages and DTU–
Mechanical Engineering is leading WP7 (as well as WP8), which deals with the
design and development of the fuel cell system. The current PhD project is aimed
at investigation of fuel cell transients and the research is directly collaborated with
H2Logic A/S, which is one of the industrial partners involved in integration of
fuel cell systems in material handling vehicles.

2
1 Introduction

Other project collaborators include:

 Vestforsyning A/S
 Rotrex A/S
 Dana–Tank A/S
 Hytor A/S
 Teknologisk Institute
 Parker Hannifin Danmark A/S
 Hydrogen Link Association
 Holstebro Kommune

1.3 Objectives of the thesis

The aim of this dissertation is to investigate the dynamic behaviour of a PEMFC

system in order to improve system efficiency when it is operating at different loads
and operating conditions. One of the objectives was to identify components that
are not responding properly at different operating conditions, and finally,
recommending new operational strategies for different operating scenarios.

The abovementioned objectives were achieved by indulging in the following

aspects of the current research work:

i. Development of a dynamic PEMFC stack model.

ii. Implementation of controls for the fuel cell system.
iii. Exergy analysis of the fuel cell stack.
iv. Start–up and transitory load change analysis.
v. Water and thermal management of the fuel cell stack.
vi. Investigation of species crossover across the fuel cell membrane.
vii. Analysis of anode–purge strategies at different operating conditions.

1.4 Methodology

The main portion of the work is based on computer simulations. The

characteristics of the PEMFC system described above are implemented in Aspen
Plus Dynamics TM [3] which is a simulation tool for process modelling and energy
system analysis. It is a component based simulation tool with a built–in unit
operation modules library. The component library includes models of heat
exchangers, reactors, turbo machinery, decanters and separators, pressure relief
valves, controllers along with utility components and manipulators. These
components are built upon by a standard set of equations representing their

3
Introduction 1

physical properties, such as isentropic and mechanical efficiencies in blowers and

turbines. Also, user–based custom models can be imported into the program.

In Aspen plus Dynamics, the physical model of the complete system is devised by
connecting the relevant component models through work, heat or material streams
and by defining operating conditions for the system. The numerical solvers then
convert these inputs into set of dynamic mathematical equations which are either
solved by sequential modulation or simultaneously with equation oriented
methods. These equations include mass and energy conservation for all units and
connecting streams, as well as relations for the states and thermodynamic
properties of the working fluids. In addition, the program has the ability to run
steady–state, initialization and dynamic simulations for the same model by
changing model or module specifications. Occurrences of disturbances and
operation constraints can be enforced by adding control modules to the
constructed system. An implicit integrator ensures stable solution of the dynamic
simulation and varies the integration step to ensure simulation accuracy.

Model for PEMFC stack and humidifier is built using Aspen Custom Modeller.
The fuel cell stack model constitutes equations for fuel cell electrochemistry, mass
and energy balances, water crossover in membrane and fuel cell thermodynamics.
The model is implemented into ASPEN Dynamics and system controls are
implemented in order to emulate real system behaviour characterizing the
transient nature of operations.

1.5 Limitations and Validation

This project is only focused on dynamics of the PEMFC system and only a few
potential issues related to fuel cell performance were addressed to keep the focus
well within the scope of research. Though steady–state models of the fuel cell stack
were developed for validation purposes, most of the emphasis has been set on
dynamic operations of the system. Moreover, only a specific fuel cell stack of 21.2
kW nominal power is selected for modelling and power conditioning modules
such as inverters and batteries have been neglected.

Veracity of the developed models is validated by previous studies and design data
provided by the fuel cell manufacturers. Assumptions relating to component
models and operating conditions are based on open literature and personal
communications. Simulation results and associated analysis should therefore be
considered in view of all the assumptions presented in the dissertation and
appended papers.

4
1 Introduction

1.6 Overview of Thesis

The thesis is based on appended papers and is divided into 10 chapters and 2
appendices:

Chapter 1 presents the statement of purpose for this research project. The
objectives of the study are outlined with a brief description of the primary
methodologies used, and limitations curtailing the work are outlined.

Chapter 2 provides the background of fuel cells with the challenges faced, key
constituents of the fuel cell system and literature review of the underlying studies.

Chapter 3 presents the layout of the studied system with breakdown of PEM fuel
cell and description of ancillary components.

Chapter 4 elaborates on system control design and associated strategies for fuel cell
stack management.

Chapter 5 elucidates on the detailed modelling of the PEM fuel cell stack.

Chapter 6 covers the exergy analysis of the fuel cell stack.

Chapter 7 encompasses the simulation of start–up sequence of the fuel cell system
and discusses on its preliminary findings.

Chapter 8 presents the system response and stack behaviour during transitory load
changes.

Chapter 9 describes the effect of nitrogen crossover in the fuel cell and its impact
on purging strategies of anode recirculation.

Chapter 10 summarizes the work with concluding remarks and suggestions for
improvements in proceeding studies.

Appendix A contains the programming codes of modelled components in Aspen

dynamics.

Appendices B–E comprise of journals and conference papers published during this
research.

5
Introduction 1

6
Chapter 2

2 Background

2.1 Hydrogen economy

Hydrogen economy refers to the vision of using hydrogen as a clean energy

source and is anticipated to compete with coal, natural gas and gasoline as a fuel
for power generation and transportation applications in the future. Nonetheless,
hydrogen is an energy carrier and not an energy source i.e. it is not found in pure
form and it must be produced from other means of energy sources; such as natural
gas, biomass, alcohols and water. Currently, most of the hydrogen is produced in
the chemical industry by steam reforming of natural gas. As we are gradually
becoming dependent on natural gas and its extensive use will progressively make it
more expensive, reserves for natural gas will eventually deplete in the long run.
Additionally, production of hydrogen from fossil fuel reformation still generates
CO2, which would further require its capture and sequestration.

Due to the reasons above, the climate change impact of using hydrogen depends
on the carbon footprint of the energy used to produce it. Other alternatives of
producing hydrogen with very low CO2 formation or no CO2 at all need be
explored. Electrolysis of water by electricity produces no CO2, though electric
power used has to be supplied from the grid raising questions on carbon dioxide
emissions associated to the electrical power used. However, if electricity generated
through renewable resources such as wind power, hydropower and solar
photovoltaic cells is used in water electrolysis, hydrogen produced in this process is
accounted to be virtually carbon–free. Biomass feed stocks such as wood can be
chemically or thermally reformed to produce hydrogen as well. This method
releases relatively less carbon dioxide and argued to be almost neutral since all of
the produced CO2 is recycled by the growth cycle of biomass. The biological
reformation of biomass using micro–organisms is yet another mode of hydrogen
production with a relatively lower carbon footprint.

7
Background 2

Future prospects of hydrogen advocate it to be an important energy carrier and

deliver energy in a usable form to end consumers. Hydrogen may be able to store
useful energy from intermittent renewable sources, such as solar and wind power
plants when there is low demand for electricity or during part load conditions.
Electricity from these sources could be used to electrolyze water and produce
hydrogen, which can be easily stored onsite. Figure 2.1 depicts the amount of
energy required per kg of H2 produced and associated GHG emissions for three
main processes. Production of H2 by using Photovoltaic (PV) could be considered
the best choice in terms of both energy requirements and emissions. Hydrogen is
already used extensively in the process industry so issues related to its storage,
handling and large–scale distribution can be addressed quite easily. Hydrogen is
therefore considered by many a key proponent for the future energy systems with
low carbon footprints and potential cost benefits as well.

Figure 2.1: Energy consumption and corresponding GHG emissions for different types of
hydrogen production processes [4].

One of the most promising ways of utilizing hydrogen is its use as a fuel for
automotive applications. Hydrogen when used in conjunction with a fuel cell
directly generates electricity which transfers the power to vehicle through electric
motor. Fuel cells are a promising technology and are deemed attractive because
they are far more efficient than the internal combustion engines and hold several
other advantages. Hydrogen is ubiquitous to fuel cells which operate best on pure
hydrogen. PEM fuel cells use hydrogen and oxygen to produce electricity, heat and

8
2 Background

water as the only by–product. In comparison to batteries, fuel cells do not

discharge and provide power as long as fuel is supplied.

2.2 PEM fuel cells for automobiles

PEM fuel cells have shown the greatest potential for automotive applications. The
PEM system allows compact design and achieves a high energy to weight ratio. The
principal advantage of the PEMFC is its ability to operate at moderate
temperatures of 80°C and a relatively quick start–up when hydrogen is used as the
fuel. System efficiencies for PEMFCs falls above 50% in comparison to internal
combustion engines (ICE), which are around 30% (gasoline)–40% (Diesel)
efficient[5]. Also, the fuel cell technology enables low noise and vibration
operation, even during rapid accelerations and in some cases is found to be 50%
lower than the corresponding ICEs [6]. In respect, fuel cells also contribute the
least to the environmental emissions; direct hydrogen fuel cell vehicles emit only
water vapour. Figure 2.2 displays the projected GHG pollutions for different types
of light duty vehicles in the coming years.

Figure 2.2: Predictions for GHG pollution for different vehicles [7].

Major limitations in commercialization of the PEM fuel cell systems are high
manufacturing costs and complex water management systems. Varying seasonal
ambient temperatures are a further challenge. During winter, the fuel cell stack is
more susceptible to damage caused by ice formation within the stack. Heating of

9
Background 2

the stack for cold start–ups is therefore usually recommended by the

manufacturers to prevent freeze damage. Furthermore at colder ambient
conditions, the fuel cell takes more time to reach the operating temperatures and
the stack performs inefficiently during this period.

Thermal management is another issue which affects the stack life. Lower
temperatures hinder the performance of the fuel cell whereas on the other hand,
excessive heat could cause damage to the stack. Effective management requires
control of operating temperatures through cooling mechanisms, as well as
adequate supply of reactants require blowers and pumps that consume around
15% of the power produced by the fuel cell. Another challenge is the cost of these
fuel cells. At present, the estimated service life of PEMFCs operating in a vehicle is
in the range of 2000–4000 hours. The DOE has set the target 5,000 hours of
service life for 2015. Reaching such goals would require advanced manufacturing
techniques for more durable stacks, which further increase production costs. There
are also plans to reduce this cost to half of that at present by 2015 [8]. Currently,
the fuel cells cost significantly higher than internal combustion engines and until
mass–produced, they will not be competing with the engines used currently.

Figure 2.3: Vehicle mass for different types of battery electric vehicles in comparison with
fuel cell vehicles [7].

Partly for these reasons, and partly due to uncertainties of practical and
commercial advent of fuel cell vehicles (FCVs), market attention has been diverted

10
2 Background

towards battery electrical vehicles (BEVs). Even with these limitations, the fuel cells
have quite a many advantages over batteries. A practical concern for BEVs is the
weight and volume occupied by the on–board battery in the vehicle. On the other
hand, FCV would only require storage for the fuel. Fig. 2.3 exemplifies the travel
distance for vehicles with different energy storages and corresponding weight of
the vehicles. It can be observed that in order to have larger driving range, the mass
of the vehicle with batteries increases substantially. In this respect, the FCV can
afford to conquer longer distances and can be regarded similar to IC engine cars.

Hydrogen is a high energy content fuel. Still, its low density presents many
technical challenges in on–board storage design. One approach of avoiding such a
problem is on–board reforming of hydrogen–rich fuels. In case of PEMFC systems,
pure hydrogen is the preferable fuel of choice as other fuels can cause
decomposition of the polymer membrane. PEM fuel cells being very sensitive to
carbon monoxide require additional purification making it an expensive option
[9], [10]. Out of all prospective options, on–board compressed hydrogen storage in
cylinders seems to be a feasible storage option. At the moment, the price of
hydrogen is about twice to that of gasoline and there is also a need to develop the
hydrogen infrastructure for availability and supply of fuel to dispensing outlets.
Refuelling stations for hydrogen to support the increasing number of vehicles are
very expensive to build presently.

Figure 2.4: Comparison of storage volumes for battery operated vehicles and fuel cell electric
vehicles [7].

11
Background 2

In comparison to FCVs, the charging unit for the plugin BEVs is cheaper at the
moment; however the charging times are very high compared to the refilling of
hydrogen for fuel cells. Moreover, storage volume occupied by the batteries is
many folds more than required for the PEMFCs. The logarithmical curves of
storage volumes for batteries place limitations in terms of size of the vehicle. In
comparison, the hydrogen tank pressurized at 700 bar takes the least of the volume
(see Fig. 2.4), and is more appropriate for light duty vehicles where size of the
automobile is one of the main concerns.

In spite of analogous outcomes of the fuel cells and IC engines in automotive

applications, FCVs require an on–board battery coupled to the fuel cell, mainly
due to the poor response time and narrow power ranges of the PEM fuel cells. A
hybrid FCEV would be a better choice for near future, mainly due to reduction in
fuel cell stack costs by employing smaller FC modules and a corresponding plugin
battery to supplement the fuel cell at start–ups and during high load operations. It
is also likely that tougher environmental regulations are implemented by some
governments in the future, which would mandate the use of FCEVs/FCHEVs or
BEVs. Also, fossil fuels will eventually get depleted and fuel cells could become the
power source in the future. Meanwhile, research and development of the fuel cell
could continue to make it economically and practically viable for providing
alternative options against the IC engines.

2.3 PEM Fuel cell system

2.3.1 Fuel cell theory

A fuel cell is a device that electrochemically converts fuel into electricity through a
chemical reaction, with energy loss in the form of heat. A single PEMFC consists
of a polymer electrolyte membrane between two electrodes and two gas diffusion
layers that facilitate oxygen and hydrogen reactions at cathode and anode
respectively. These cells are connected in series to construct a fuel cell stack
capable of developing large voltage potentials and consequently generate high
powers. Figure 2.4 shows the basic operations of a single PEMFC. Hydrogen gas is
supplied at the anode and oxygen or air is fed at the cathode. Hydrogen molecules
are dissociated into ions and electrons at the cell anode. The negative electrode
provides a path for the electrons to flow out into the external load and reach the
positive electrode. H+ ions, at the same time cross through the membrane and
combine with the electrons and oxygen in the cathode to produce water, which is

12
2 Background

expelled out of the fuel cell. If air is used instead of pure oxygen, then cathode
exhaust also constitutes unreacted air along with the produced water.

Figure 2.5: Working principle of a single PEM fuel cell [11].

The two chemical reactions occurring at the anode and the cathode are described
as:

H 2  2H   2e 

1
2 O2  2H   2e   H 2O

H2  12 O2  H2O  heat

For hydrogen fuelled PEMFC, the reactions at cathode and anode only produce
water and heat in addition to the electrical power. Such a power system seems
quite simple; nonetheless it requires well–suited conditions to operate successfully.
These conditions are controlled by ancillary components which define the overall
system performance and effect the lifetime of the fuel cell stack. A complete fuel
cell system would comprise the following components:

13
Background 2

 Fuel cell stack

 Fuel and oxidant supply system
 Water management
 Heat management
 Electrical power conditioning subsystem
 Controls

2.3.2 Reactant supply system

PEMFC systems are comparatively compact due to allowance of higher operating

pressures. Air entering the stack is pressurized by a compressor or blower
depending on the anode pressure of the fuel. Although this increases the power
density of the stack, it comes at the expense of additional cost of compressor and
power consumed by it which leads to a reduction in overall system efficiency.
Moreover, the fuel cell stack usually operates at higher fuel and air stoichiometry
in order to avoid catalyst degradation and related operational problems [12],[13].
Removal of generated water is also attributed to the higher stoichiometry in the
fuel cell cathode. Similarly for the anode side, hydrogen is fed by a control valve at
low utilization factors. Hydrogen starvations within cell sites are another issue to
be handled to have a stable operation and prolonged stack life. It is therefore
imperative to have an efficient and reliable feed system for the PEMFC stack.

2.3.3 Water Management

The proton conductivity of the fuel cell depends heavily on the water content of
the membrane. To ensure sufficient conductivity and prevent damage to the
membrane, it must be sufficiently humidified. On the other hand, a lot of effort
has to be put into expelling the generated water to avoid the flooding, or blockage
at the reaction sites on the electrode. For systems with recirculation loops,
additional equipment is required to condense, trap and reuse the water collected
at the exhausts of the stack

2.3.4 Heat Management

Most of the reactions occurring in the fuel cell are exothermic in nature and
produce heat in addition to electrical power. In case of PEMFCs, the electrical
efficiency can reach up to 60% and rest is converted to heat energy. For physical
stability of the stack, a constant operating temperature needs to be maintained by
effective rejection of the produced heat. A cooling circuit comprising of heat

14
2 Background

exchangers and air radiators is responsible for thermal management of the system
with liquid coolant as the medium. Proper sizing of heat exchangers and radiators
is critical in system design as they would occupy a larger volume which is
undesirable in light duty vehicles.

2.3.5 Control system

Fuel cell operating requirements in vehicles are more inflexible than stationary
applications. These systems have to operate at varying conditions related to
temperatures, pressures, power load and humidity. All the auxiliary components,
such as air and fuel supply system which include air blower (compressor) and
control valves, and the thermal control system which comprises of heat exchangers,
coolant pumps and air radiators need to control the input for fuel cell stack.
Therefore, system level control of key parameters becomes useful in stable and
efficient operation of the PEMFC system.

2.4 Literature review

Many PEM fuel cell models have been developed in recent years. Whereas, steady–
state models of these systems are present in abundance. A few others have studied
these system components individually, where most of the research focuses on the
fuel cell or the stack itself. Ceraolo et al. [14] developed a simplistic dynamic
model based on cathode kinetics. Amphlett et al. [15] extended their previous
steady–state model and presented a generalized transient model. Another bulk
dynamic model catering to inverter load effects is presented by Yerramalla et al.
[16]. Pathapati et al. [17] developed a fuel cell model that included effects of
double charge layers at cell sites along with flow dynamics and pressure gradient in
flow channels. Jia et al. [18] also followed suit and developed a dynamic model in
MATLAB/Simulink to investigate fuel cell transient electrical responses under
various operating conditions.

Multi–dimensional models on the other hand, disseminate complex characteristics

of reactant flows and charge transportation within fuel cells. Hu et al. [19]
represented a three–dimensional computational PEMFC model with conventional
and inter–digitized flow fields. Another three–dimensional mixed–domain PEM
fuel cell model of Kim et al. [20], which integrates intrinsic transport mechanisms,
has been applied to investigate effects of the fully coupled transport phenomena.
These models might be useful in predicting cell or stack performance, however due

15
Background 2

to high computational times and limitations to integrate with BoP components,

these models are not suitable for system–level modelling.

Start–up behaviour of PEMFC stacks at sub–zero temperatures has been studied

and reported by few researchers. Li et al. [21] conducted experiments and validated
their model based on the results. Yan et al. [22] investigated effects of sub–freezing
temperatures on fuel cell performance and start–up. A model for freeze start is also
developed by Mangold et al. [23] and compared to experimental data. Meng [24]
and Sundaresan and Moore [25] both proposed thermal models for cold start of
PEMFC. However, start–up behaviour of a stack above the freezing temperatures
which accounts for multiple varying operating conditions and changing auxiliary
components outputs has not been reported to a greater extent.

Heat management in PEMFCs being a critical factor in its operations and

performance is accounted for in open literature as well. Issues related to
temperature dynamics are dealt and studied by Vasu and Tangirala [26], which
could predict the effects of temperature and feed flows on system transient
behaviour. Khan and Iqbal [27] proposed a transient model to predict voltage
output and cell efficiency, and a thermal model including heat transfer coefficients
and energy balance for the stack. Shan and Choe [28] analysed the temperature
distribution on fuel cells by developing a two–dimensional model. Another
control–oriented thermodynamic model is also proposed by del Real et al.
[29].Coolant control strategies were suggested by Ahn and Choe [30] after
investigation of temperature effects on the system. Jung and Ahmed [31] developed
a stack model based on real–time simulator in MATLAB/ Simulink environment
and validated it with experimental setup of Ballard Nexa fuel cell. A thermal
management system for a PEMFC was designed by Asghari et al. [32]. Influence of
temperature on fuel cell’s characteristics is also reported by Beicha [33]. While
some researchers such as Park and Choe [34] presented a transient stack model
and analysed the temperature distribution on fuel cells, management of stack
temperature by coolant flow controls becomes vital in efficient operations and
reliable performance of the fuel cell system.

A very few studies have been conducted on exergy analysis of dynamic fuel cell
systems, although most of studies are carried out on individual fuel cells or stacks
at steady–state. Blomen [35] was one of the first to introduce the concept of exergy
analysis in fuel cells. Study on the energy and exergy analysis of a PEMFC–assisted
combined heat and power system is published by Saidi et al. [36]. Barbir and
Gomez [37] investigated the efficiency and economics of PEM fuel cells. Kazim [38]
reported an exergy analysis of a PEM fuel cell with respect to operating parameters,
and performed exergo–economic (thermo–economic) analysis of a PEM fuel cell at

16
2 Background

various operating conditions [39]. A similar study was conducted by Mert et al.
[40]. Recently Miansari et al. [41], reported a performance analysis of PEMFC
based on exergetic parameters. Leo et al. [42] and Uyanga et al. [43] studied
exergies for PEM fuel cells in marine applications and experimental work on Nexa
module, respectively.

Literature available on water and gas crossover through PEMFC membranes is not
uncommon. Kocha et al. [44] characterized gas crossovers by applying an in–situ
electrochemical technique to determine hydrogen crossover rates and devised a
model to predict the amount of nitrogen accumulation in the anode channels.
Ahluwalia and Wang [45] modelled and analysed the buildup of N2 in the anode
recirculation loop and the impact of this buildup on the performance of 90 kW
PEMFC stack. Catalano et al. [46] reported the effects of relative humidity on gas
permeability and swelling in Nafion membranes used in fuel cells. Baik and Kim
[47] calculated nitrogen permeability coefficients (NPC) by employing a mass
spectrometer and analyzed N2 crossover under open circuit voltage and power
generation conditions. Weber [48] reported the effects of membrane pinholes and
increase in gas–permeation on PEMFC’s performance. It was also shown that
water and thermal management is also affected where there are large pinholes in
the membrane.

From a control perspective, it is important to develop a complete model that

emulates the real behaviour of PEMFC stack when subject to varying operating
conditions. Therefore, a need to develop a system–level model is identified that
responds to dynamic effects of all the necessary BoP components which is lacking
in the aforementioned studies. Of all publications in open literature, relatively
limited studies account for control–oriented PEMFC models. Pukrushpan et al.
[49] presented a transient study on PEM fuel cell, which included compressor flow
controls and incorporated inertia dynamics, membrane humidification and
pressure effects into the model. A control–oriented thermal dynamic model is also
proposed by Del Real et al. [29]. Ahn and Choe [30] investigated temperature
effects on the overall system performance and suggested strategies for coolant
controls.

There are a few published papers investigating the purge process modelling in
general or specifically pertaining to individual aspects of the process. Zhu et al. [50]
experimentally determined the critical flow rate in anode exhaust stream by
manual purging of Nexa Power module. Tang et al. [51] also investigated the
transient response of the aforementioned module and identified factors
influencing its performance. Zhai et al. [52] also presented their study on anode
water flooding and simulated the gas purge effect for medium current densities.

17
Background 2

Gou et al. [53] presented a one–dimensional computational model to study the

dynamic behaviour of pressure in anode flow field during the purge process. It was
shown that different current densities had minimal influence on the pressure drop
and the pressure swing during this transient process. A dynamic three–phase
transport model was developed by Wang et al. [54], which lead in identification of
the optimum water uptake parameters for purge cycles during startup from
subfreezing temperatures and subsequent shutdowns. Karimaki et al. [55] setup a
test bench to study inert gas buildup effects on fuel cell stack and introduced the
use of online hydrogen sensor to measure gas crossover across the cell. Promislow
et al. [56] built a simple analytical model to describe the steady state profile of
anode N2 concentration in PEMFCs with straight gas channels and identified an
optimum bleed rate for the specified stack. Some others have published papers
addressing the anodic dead–end mode of operations in the fuel cell anode. Hou et
al. [57] developed a dynamic voltage model for a ‘dead–ended’ FC to simulate
hydrogen purging in order to prevent water accumulation in the anode. Muller et
al. [58] also correlated nitrogen accumulation with a 20–cell temporal fuel cell
performance operating in anodic ‘dead–end’ mode by estimating permeability
characteristics of typical Nafion membranes. Siegel et al. [59] modelled a PEM fuel
cell operating with a ‘dead–end’ mode and experimentally verified the evolution of
liquid water and nitrogen fronts along the length of anode channels. They further
developed the 1D model with periodically–purged anode channels and
incorporate simple resistance model for predicting nitrogen permeance and water
transportation through the membrane and presented their findings in Yesilyurt et
al. [60]. Another experimental study was conducted by Choi et al. [61] in which
purge characteristics of a cathodic dead–end mode PEMFC for a submarine were
analysed.

18
Ch
hapter 3

3 Systeem Configguratio
on

The fuel cell stack

s is basically susttained by supporting
s equipmennt and toggether
consstitute a fu
uel cell systtem. Figuree 3.1 show ws the scheematics of fuel cell syystem
analyysed in thiss study. Th
he PEM fueel cell stack [62], has a nominal ppower outp put of
2
21.2 kW and constitutees 110 cellls with thee cell area of 285.8 cm . Auxxiliary
equipment, also referred to as balan nce of plan
nt (BOP) in ncludes coomponents such
as aiir blower an
nd humidifier, pumpps, heat excchangers an nd radiatorr for the co
ooling
uit, flow valves
circu v and controllerss. Compreessed air iss cooled aand humid dified
before enteringg the cath hode side of the stack. On th he other siide, pressuurized
hydrrogen fromm storage tan nk is fed too the anodee of fuel celll.

Figgure 3.1: Scchematic of a PEM fuel cell system with

w auxiliaary componennts and conttrol
layout.

19
System Configuration 3

Since a higher fuel stoichiometry is maintained, anode exhaust is recirculated back

to the feed stream via a recirculation pump. Liquid water present in the anode
exhaust is collected in a water trap, whereas the remaining water present in the
stream is purged with the other gases by a solenoid valve before mixing with the
inlet stream.

Heat produced in the stack is absorbed by the coolant which circulates in a circuit
associated with the stack and a heat exchanger. An external cooling loop,
connected to the interior heat exchanger, in turn cools the water in the internal
circuit. This circuit also comprises of a heat exchanger to precool the blown air
and an air radiator for heat rejection from the system. Flow of water is regulated by
pumps in the respective circuits. This system layout is implemented in Aspen
Dynamics simulation tool [3] along with the developed models of Fuel cell stack
and the humidifier.

Table 3.1: Nominal operating conditions for PEMFC stack

Current (A)
Reactant Parameter
15 30 60 120 240 300
Fuel (Pure Hydrogen)
Inlet stoichiometry Min. 6.3 3.4 2.2 1.9 1.6 1.6
Nominal Inlet Pressure (kPa) 115 116 131 155 200 220
Pressure Drop (kPa) 9 13 13 14 16 18
Oxidant (Ambient Air)
Inlet stoichiometry Min. 5.1 2.4 1.8 1.8 1.8 1.8
Nominal Inlet Pressure (kPa) 108 110 117 138 180 200
Pressure Drop (kPa) 8 13 12 16 40 53
Relative Humidity (%) 95 95 95 95 95 95
Coolant
Inlet Temperature Max. (°C) 60 60 60 60 60 60
Outlet Temperature Max. (°C) 61 63 66 67 68 70
Minimum Coolant flow (lpm/cell) 0.05

In order to ensure a stable operation of the system, a control system is needed to

regulate flows, temperatures and pressures in the feed streams, cooling circuits and
plant components. Here, PID controllers are deployed to emulate the system in
real time and thereby analyse response characteristics under varying operating
conditions. Aspen Dynamics comprises of built–in PID controllers and
manipulators. Combinations of these components are used to devise controls for

20
3 System Configuration

entire system. These controllers collect data from various component inlet and
outlets which are regarded as pressure, temperature and flow transmitters, and
manipulate the corresponding components to reach the desired state. Due to this
fact, current system responses and its behaviour are attributed to the formulated
control strategy which is based on fuel cell stack limitations and recommendations
by the manufacturer. Component properties related to these limitations are
discussed in the following subsections whereas intricate details of system controls
are dealt in the relevant section later.

3.1 PEM Fuel Cell stack

In a PEMFC, the electrolyte membrane is pressed between two catalyst layers

followed by gas diffusion layers. These layers form the electrochemical part of the
cell and are often merged to the term MEA (membrane electrode assembly).
Bipolar plates on each side of MEA complete a single cell formation. The main
components used in a single cell are briefly described in terms of their
functionality and structure.

Figure 3.2: Structure of a single fuel cell [63].

21
System Configuration 3

Figure 3.3: Exploded view of PEM fuel cell stack [64].

3.1.1 Membrane

The prime function of the membrane is to transport the protons from dissociated
hydrogen molecules at electrode sites of the anode. Other functions include
separating the anode and cathode reactants by acting as electron insulator while
providing high ionic current conductivity. The most commonly used electrolyte in
PEMFC is a polymer based material (Semi crystalline Teflon with perfluoro
sulfonic acid [PFSA] groups) branded as Nafion®. These membranes are mostly
selected due to their good availability at low cost, lesser water drag, low gas
permeability, strength and flexibility, and chemical as well as thermal stability.

The ionic conductivity of PFSA membranes could reach to that of aqueous

electrolytes; however they are strongly dependent on the membrane water content.
Therefore the uptake of water and its distribution within the membrane becomes a
defining factor for ionic conductivity of the fuel cell.

3.1.2 Catalyst Electrode layer

The catalyst layers on the anode and cathode side of the membrane promote and
speed up the electrochemical reactions. The catalyst present in the layer helps in
dissociating reactant molecules to generate external electronic current. Currently,
platinum based catalysts are usually used, in spite of being very expensive. To
ensure close contact of the three phases (electric, ionic, gas), catalyst particles are
finely dispersed on an electrically conducting carbon paper. In this manner the

22
3 System Configuration

amount of active platinum exposed to the reactants is maximized, greatly

increasing power density of the cell. The catalyst electrodes are attached to the
adjacent components by spraying or coating the catalyst ink either on gas diffusion
layer or the membrane and hot pressing the MEA together.

3.1.3 Gas Diffusion Layer

The gas diffusion layer (GDL), also known as the backing layer is a critical
component in fuel cell assembly. The main function of a GDL is to distribute the
reactant gases to and from the reaction sites while providing electronic
conductivity. Additionally, its role is to minimize water flooding and maximize the
electronic contact at catalyst interfaces. Since a conducting substrate is required to
distribute electrons from anode to cathode, a chemically inert carbon layer is used
in the manufacture of GDL layers. An ideal GDL should possess adequate
hydrophobic characteristics and enhance electronic contact with low resistance,
and effectively transport reactant gases to the catalyst layers. Lastly, the GDL
structure should be strong enough to withstand the pressure of the tightly clamped
stack cells, which is necessary to prevent leakages and in reduction of contact
resistances.

3.1.4 Bipolar plates

The bipolar plate is also known as the flow field plate due to the integral design for
reactant flows. The bipolar plates connect the MEA to the stack and are attached
to anode on one side and cathode to the other, thus the name bipolar. Their
primary role is the supply, distribution and removal of reactants and products
from the stack. They also collect current from the cell terminals. In order to ensure
heat removal, the bipolar plates are generally equipped with cooling channels for
circulation of coolants through the stack. The bipolar plates therefore need to be
high electronic conductors, impermeable with high chemical and mechanical
stability, low weight and low cost with appropriate manufacturability.

3.2 Balance of Plant (BoP)

PEM fuel cells entail supplementary equipment to manage the characteristics of

gases entering and leaving the system for maximum power output and to prolong
the stack life. Key objectives of these BoP components include regulating fuel cell
temperature, regulation of fuel and oxidant stoichiometry at anode and cathode,
humidification of fuel cell and removal of liquid water from the flow fields.

23
System Configuration 3

Suitable regulation of these characteristics is necessary for efficient and

uninterrupted fuel cell operation. Brief description of these components is as
follows:

3.2.1 Humidifier

One of the key issues in PEMFC is the dehydration in the membrane. A fully
hydrated membrane supports the ionic crossover between the electrodes, as well as
extends its life. Therefore, reactants in the PEMFC need to be humidified before
entering the stack. In the present model, a humidifier utilizes the water produced
by chemical reaction inside the fuel cells to humidify the inlet air. A simple model
based on mass and energy balance is implemented here. Although some empirical
models could be considered, these vary over a wide range depending upon the
types of humidifiers used. Due to the lack of sufficient data, it is assumed that the
inlet air is optimally humidified before entering the stack. The relative humidity of
air entering the cathode is set to 95% in the calculations; although other values
can be chosen. This assumption could be justified as it is very close to the real
operational conditions. On the anode side, there is no humidifier and the fuel is
humidified by means of water crossover and its recirculation. It is observed that
water crossover from cathode to anode through the membrane is adequate enough
to raise the relative humidity in the anode exhaust to 100%. This recirculation,
when mixed with inlet stream delivers the desired humidity into the anode of the
fuel cell.

A static model of humidifier is used to add the desired amount of water into the
cathode inlet stream. This model calculates the amount of water required to reach
the set relative humidity value based on the air temperature at the outlet of the
humidifier. The relative humidity (RH) can be defined as the ratio of partial
pressures of vapour, to the partial pressure of that in the saturated mixture, at the
same temperature:

Pv
RH  (3.1)
Psat

Aspen custom modeller has built–in functions to calculate saturation pressures at

different temperatures. The partial pressure of vapour in the air is calculated based
upon the molar fraction of water in the air.

Pv  xv  Pair (3.2)

24
3 System Configuration

where Pair is the total pressure of air. For a set RH value, this gives the molar
fraction of water vapour needed at the outlet of humidifier, which in turn
calculates the desired molar flow of water. Simple mass balance of all inlet and
outlet streams determines the amount of water transferred from cathode exhaust
to the air entering the cathode:

m exh,in  m exh,out  m H 2O (3.3)

m air ,out  m air ,in  m H 2O (3.4)

In a similar manner, a simplistic heat transfer model is used to define air

temperatures at the cathode inlet. Here the assumption has been made that the
cathode exhaust, which is at higher temperature than the air entering the
humidifier, transfers heat at an effectiveness of 0.85. The temperature of air
leaving the humidifier is determined by the following energy balance:

m air C p ,air Tair ,out  Tair ,in   0.85  m exhC p ,exh Texh,in  Texh,out  (3.5)

Where subscripts air and exh denote the air stream directed towards fuel cell inlet
and exhaust stream from the cathode respectively.

3.2.2 Air Blower and pumps

In the above proposed system, air blower, anode recirculation pump and water
pumps are one of the BoP components which are also regulated by the control
system. Aspen Dynamics TM contains models of these units in its library as well.

Isentropic efficiency Power consumption

1200
0,5
1000
0,4
800
Power (W)

0,3
600
η

0,2 400
0,1 200
0 0
0 20 40 60 80 0 30 60 90 120 150 180
Mass flow (kg/hr) Current (A)

Figure 3.4: (a) Efficiency profile of air compressor versus air mass flow, (b) Power
consumption of air compressor at different stack currents.

25
System Configuration 3

Since the nominal power of the PEMFC is only 21 kW, mass flow rates of fuel and
air are very low. For example, at an average load of 10 kW, fuel and air flows are
around 0.00014 kg/s and 0.0088 kg/s respectively. Therefore, very low values of
isentropic efficiencies are suggested in this study, while the mechanical efficiency is
set at 90%.

The efficiency of a blower ranges from 15% to 48% in the calculations based on
experimental measurements. The mass flow dependent variance of isentropic
efficiency can be seen in Fig. 3.4a, whereas power consumption data for different
stack currents is shown in Fig. 3.4b. Similarly, calculated pump efficiencies are also
very low for the cooling water circuits and are determined to be around 70%.

3.2.3 Heat Exchangers

As can be seen in Fig. 3.1, a network of heat exchangers constitutes the thermal
management of the prescribed system. They extract the heat produced within the
fuel cell stack and maintain the desired operating temperature which is critical for
fuel cell durability and performance. Although heat exchanger models used here
are predefined in Aspen Dynamics, some of the parameters have been assumed on
the basis of media entering the hot and cold sides of these heat exchangers. The
heat exchanger, which is connected to the internal cooling loop, has liquid water
on both its hot and cold side. Therefore, a UA value of 1.0 kW/K is assumed.
Whereas, UA values for air pre–cooler and radiator are approximated to be 0.05
kW/K and 0.3 kW/K respectively. In performing the simulations, the pressure
drop was assumed to be 0.05 on both sides of heat exchangers. The corrected
LMTD is calculated in addition to the corresponding inlet and outlet temperatures
of hot and cold streams.

3.2.4 Flow valves

A valve is placed between the hydrogen tank and inlet manifold of anode which
enables or disables the hydrogen supply. This regulatory valve adjusts high
hydrogen pressure of the tank to the desired operating pressure of the fuel cell.
Since the system does not operate on dead–end mode, the amount of hydrogen
regulated by this valve equals the stoichiometric hydrogen required of the fuel cell.
Stoichiometric ratio is defined as the amount of reactant supplied to the amount
which is consumed in the reaction.

Flow and pressure of oxidant into the cathode is regulated by the blower. The
amount of stoichiometric oxygen for the fuel cell reaction is manipulated by a

26
3 System Configuration

controller which regulates the electrical power of the blower, and hence
compression and air flow.

3.2.5 Water trap

A knockout drum for liquid water is placed before the purge valve on the anode
recirculation loop of the system. This container is simulated using a separator
block in Aspen plus, which after determining phase properties of the inlet stream
separates the liquid and vapour species at the outlet. Liquid water is collected at
the bottom of the tank and the vapour stream is recirculated back into the anode
feed. Though in simulations, all of the liquid is expelled from the stream, it is
more likely to experience few liquid droplets in the recycled loop.

27
System Configuration 3

28
Chapter 4

4 System controls

Current research aims to build up a system which meets the requirements of actual
stack running under varying load and operating conditions. As mentioned earlier
that BoP components are critical for proper functioning of the stack. These
components are able to regulate three major control loops in the fuel cell system:
reactant supply, the water management, and the heat management. This section
presents the methodology of controlling system parameters and operating
conditions for the system to achieve the prescribed goals through stable
operations. Classic proportional–integral (PI) controllers, which are widely used in
industrial control systems, are employed to regulate different components and flow
streams. Principally, these controllers calculate “error” value as the difference
between a measured process variable and a desired set–point, and attempt to
minimize this error by adjusting the process control inputs. Aspen Plus Dynamics
has built–in PID controllers with options of specifying process and output variable
ranges, tuning and filtering of controls and selection of ideal, series or parallel
algorithms. Classical notation for output of the ideal PID controller employed in
this study is specified as:

 1 t d 
OP  K p  e(t ) 
 Ti  e( )d  T
0
d
dt
e(t ) 

(4.1)

where, OP is the controller output, KP is the proportional gain which is set

different value for different components. Ti and Td are integral and derivative
times respectively and e is the calculated error between set point and processed
variable at instantaneous time t.  is the variable of integration taking values from
0 to present time. Key parameters to be controlled in the proposed system are
reactant inlet stoichiometries, inlet pressures, coolant inlet and operating
temperatures of the stack. Control strategies incorporating these PIDs and
regulatory mechanisms for different operating parameters are discussed

29
System controls 4

accordingly. However, optimization of these controllers is not performed in the

present study. Figure 4.1 elucidates on processed and manipulated variable for
different control blocks implemented in the proposed system.

Figure 4.1: Control blocks for the PEMFC system.

4.1 Operating conditions

The principle functionality of the PEMFC stack is to provide stable electrical

power over a range of operating and environmental conditions. However, fuel cell
functionality is constrained within specific limitations pertaining to operational
integrity and stack lifetime. Optimum fuel cell performance depends on a number
of operating conditions including current density, reactant stoichiometry, relative
humidity, inlet pressures, and cell temperature. Each of these conditions impacts
the performance of the fuel cell individually or with combined effect.

Operating the stack below its minimum theoretical stoichiometry leads to

premature durability issues. Stoichiometry is defined as the ratio of actual flow rate
to the flow rate required to support the reaction. The minimum theoretical
stoichiometry required to support the reaction is 1.0 for both fuel and oxidant. In

30
4 System controls

practice, higher stoichiometries, or flows, are required to ensure adequate reactant

concentrations and to remove water. The fuel cell performance is not sensitive to
stoichiometries at low currents because the minimum flow rates required for water
removal are more than adequate to provide the necessary concentrations. At high
currents, the reactant concentration at the catalyst reaction sites is the critical
factor.

For high current loads, excess flows are also needed to manage the water
production in the cathode and to dilute the higher nitrogen crossover to the
anode. In the case of fuel starvation, damage can be done to a cell operated much
below the average cell voltage. The extent of the damage increases with decreasing
current density due to a higher average cell voltage. The cathode, however, is much
more sensitive to air stoichiometry. As air stoichiometry is decreased at high load,
cell voltage drops and the cell voltage standard deviation increases significantly. Air
starvation might not cause damage to the cells, though it could lead to the
appearance of individual low cell voltages. Air flow requirement increases with
current and is dependent on the number of cells in the stack.

Low voltage cells can occur within the stack because of insufficient reactant
concentration at catalyst reaction sites. This could be caused due to inadequate
reactant stoichiometry, N2 build–up in the anode, water blockage, and large
internal leaks between anode and cathode. Over the stack lifetime, greater leaks
can develop and cause operational problems. The development of internal and
external leaks with lifetime operation is highly dependent on stack operating
conditions. Large internal leaks have a negative impact on stack operation and
system efficiency. Since the anode inlet pressure is higher than the cathode side,
hydrogen will transfer to the cathode side, will not be used in the fuel cell reaction
and thus will be wasted fuel. The onset of an internal leak will be accelerated by
high temperature and low humidity in the cells. Temperature is also one of the key
factors for dry cell operations, since at high temperature the water is in vapour
state and easier to remove. Nonetheless, there still is some liquid water in the flow
channels.

Over the period of stack life, there are three main ways in which external leaks can
occur: gas diffusion through the external seals, overheating and seal damage due to
high reactant pressures. Elevated temperatures may also trigger external leaks if
operated over longer durations. Further, high pressure cycling also leads to a
reduction interface pressure, resulting in an external leak.

For a system with anode recirculation, build–up of inert gases will decrease the
hydrogen concentration to the point where fuel starvation spots will occur. The

31
System controls 4

crossover rate of nitrogen from the cathode to the anode is highly dependent on
the operating conditions of the stack. Generally, N2 crossover will increases with
the increase in cathode pressure and cell temperatures. Ensuring the cathode
pressure is lower than the anode pressure will minimize nitrogen crossover and
improve cell stability. Additionally, from a safety perspective, it is preferred to
operate the anode at a higher pressure than the cathode. In the event of an
internal leak, hydrogen would preferentially leak to the cathode and be diluted by
the air, or combine with oxygen to produce water. However, there is a maximum
cross pressure limit beyond which mechanical stresses could be produced in the
stack resulting in damage [62].

For a maximum service life and efficiency of the stack, Ballard recommends
operating conditions within which the stack should operate, which are followed in
this study. Figures 4.2 and 4.3 portray the graphical representation of the suggested
operating parameters for the specified fuel cell stack.

Anode Pressure drop H2 Inlet pressure

0,6 2,5 Air Inlet pressure
Cathode Pressure drop
0,5
2,0
0,4
∆P (bar)

P (bar)

0,3 1,5

0,2
1,0
0,1

0 0,5
0 50 100 150 200 250 300 0 50 100 150 200 250 300
I (A) I (A)

Figure 4.2: Stack data provided by Ballard (a) Nominal pressure drop within the fuel cell,
(b) Inlet pressures for reactants.

During normal steady state operation, reactant pressure should be above coolant
pressure whereas for short transients and during start–up, coolant pressure may
exceed reactant pressure. In a system with fuel recirculation, the consumption is
slightly above 1.0 stoichiometry as 1% to 2% of the flow will be required for
purging.

32
4 System controls

7 H2 Stoichiometry Coolant Delta T

12
Air Stoichiometry
6
10
5
Stoichiometric Ratio

8
4
6

∆T (◦C)
3
2 4

1 2

0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
I (A) I (A)

Figure 4.3: (a) Reactant stoichiometry used in the fuel cell control system, (b) Stack inlet and
outlet temperature difference maintained by coolant mass flow.

4.2 Air flow control

The control loops of hydrogen and air supply mechanism should maintain the
optimal reactant ratio and prevent shortages that occur during abrupt and
amplified changes in the external load. It is reported in [65] that the proposed
PEMFC system is not sensitive to reduction in fuel flow during load fluctuations;
however if the hydrogen stoichiometry is below 1.0, then fuel starvation can cause
irreversible damage to the stack. On the other hand, oxygen starvation affects the
fuel cell to a greater degree and has a much larger effect on stack performance.
Oxygen starvation is a complex phenomenon that occurs when oxygen partial
pressure falls below the critical level outlined by the reaction stoichiometry. It
greatly affects the voltage potential of the cell and can cause short circuit in severe
cases. Hence, an adequate air supply is required.

Figure 4.3a shows the fuel and air stoichiometries for the selected fuel cell stack. It
can be seen that at low current loads, high amounts of excess reactant flows are
desired. For stack operating at low power and low pressures, water formed by the
reaction in the cathode side of the cells needs to be ejected out of the stack, which
is done by supplying high amounts of air. While the oxidant consumption
depends on the current load, the amount of oxygen delivered to a fuel cell is
directly related to the blower power. Here, data from pressure and flow sensors is
transmitted to the controller where an algorithm based on data from Fig. 4.3a,
translates it to be the process variable and set point for the PI controller, which
regulates the blower power in order to maintain the desired oxygen ratio.

33
System controls 4

Air flow into the fuel cell is dependent on the current drawn from the stack.
Figure 4.4a displays the current ramp tests against the power consumption of the
air blower. Oxygen required by the fuel cell is instantly supplied by the blower
which is regulated by the controls at the backend. In the PID controllers
implemented here, the set–point for air flow controller is defined as the difference
between the recommended air stoichiometric curves and the measured air
stoichiometry from the flows at the inlet and outlet of the fuel cell stack. As can be
seen in Fig. 4.4a, that the set–point is always set to 0 and the controller
manipulates the blower power to coincide the process variable with the set–point.
With each current change, the process variable distances from the set–point and
the controller is actuated.

     
120.0
1.5

1.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2

Blower Power (kW)

80.0

0.8

Controller Output
Process Variable
Current (A)
1.0

Set Point
40.0

0.4
0.5

0.0
0.0

-0.4
0.0

0.0 50.0 100.0 150.0 200.0 250.0 0.0 50.0 100.0 150.0 200.0 250.0
Time (Seconds) Time (Seconds)

Figure 4.4: Air flow controls (a) Air compressor power versus stack current, (b) Controller
output variation with change in process variable.

A similar change in the measured air stoichiometry is observed. For an abrupt

current rise, air stoichiometry is reduced until the activation of control which
increases power supplied to the blower and consequently delivers the required
amount of air to the fuel cell stack.

34
4 System controls

80.0 120.0
2.5
Air stoichiometry
2.25

Currentl (A)
40.0
2.0

0.0
1.75

-40.0
1.5

0.0 50.0 100.0 150.0 200.0 250.0

Time (Seconds)

Figure 4.5: Change in O2 stoichiometry with current drawn from stack.

4.3 Fuel flow control

Fuel starvation also causes stack damage in the form of performance degradation.
When the fuel stoichiometry is lower than required there is a non–uniform fuel
concentration within the stack, which leads to localized fuel starvation. Similar to
the air flow control, an algorithm for controlling hydrogen flow is devised along
with a PI controller, which regulates the control valve opening for optimal fuel
supply at the desired pressures.

     
120.0

1.5

0.5
2.0

0.4
1.75

80.0

Controller Output
1.0
Process Variable
Pressure (bar)

Current (A)

Set Point

0.3
40.0

0.5
1.5

0.2
1.25

0.0
0.0

0.1
-0.5
1.0

0.0 50.0 100.0 150.0 200.0 250.0 0.0 50.0 100.0 150.0 200.0 250.0
Time (Seconds) Time (Seconds)

Figure 4.6: Fuel flow controls (a) Anode inlet pressure versus stack current, (b) Control
input and output variables.

35
System controls 4

Figure 4.6a expresses the variation in fuel inlet pressure to the ramping of stack
current. The proportional valve opens or closes to change the hydrogen flow into
the stack. This control process is elucidated in Fig. 4.6b, where the output variable
(valve opening position) is controlled to reach the desired fuel stoichiometry.

     

120.0
120.0

0.16
4.0

Recirculation pump power (kW)

3.5
H2 Stoichiometry

80.0

0.12
80.0
Current (A)

Current (A)
3.0

0.08
40.0
40.0
2.5
2.0

0.04
0.0

0.0
1.5

-40.0
1.0

0.0 50.0 100.0 150.0 200.0 250.0 0.0 50.0 100.0 150.0 200.0 250.0
Time (Seconds) Time (Seconds)

Figure 4.7: (a) Fuel stoichiometry as a function of current, (b) Power consumption of
recirculation pump against stack current.

Variations in fuel stoichiometry against the drawn current can be noticed in Fig.
4.7a. Excess fuel ratio is reduced from 2.2 to 1.6 at current ramp–up from 60A to
100A. Larger current changes could induce even lower fuel stoichiometry; it is
therefore always preferred to operate the fuel cell well above its stoichiometric
requirements. In a similar manner, the pressures in anode recirculation loop are
governed by PI controllers which collect data from anode upstream and
manipulate recirculation pump power accordingly. Figure 4.7b portrays the power
consumed by the recirculation pump in the same event of current change.

4.4 Temperature control

Thermal management in PEMFC systems is of vital importance, basically due to

the fact that heat produced in the selected fuel cell cannot be dissipated by
convection and radiation through the stack surface. A consistent and stable
operation of around 70°C thus requires a liquid cooling system. Since the
operating temperature of the fuel cell is not very high, a low temperature
difference with the ambient requires having a large heat transfer surface.

36
4 System controls

Therefore, an efficient thermal control system becomes of substantial importance

to ensure optimum system performance.

As shown in Fig. 4.1, the cooling system for the proposed fuel cell consists of
internal and external cooling circuits. As mentioned earlier, the coolant mass flow
rate defines the variance in the stack temperature or simply maintains the fuel cell
operating temperature. In this case, temperature in the stack can be controlled by
the coolant flow rate which acts as an input signal and is adjusted by the PI
controller. Based on data from Fig. 4.3b, equations defining stack temperature as a
set–point for controller are developed. In addition, the controller simultaneously
collects data from temperature transmitter at the coolant outlet stream, which then
changes the coolant flow accordingly by sending output signals to the driving
pump. Employment of such algorithm in the system controller ensures a stable
operation under normal steady–state conditions, however for system start–up
scenario, a different approach is required. This approach is selected because at
start–up sequence, temperature of the stack is equal to the ambient and the
above–mentioned control strategy would not allow the stack temperature
difference to increase more than 10°C and hence increase the time to reach
steady–state operations, which is not desirable. In this regard, at start–up
sequence, the coolant flow is restricted to a minimum value until the desired stack
operating temperature is reached which is when the steady–state controls begin
functioning. Though, a lower coolant flow is needed to increase the stack heat–up
time at start, a very high temperature difference may cause damage to the fuel cells
in the stack. Henceforth, the maximum temperature difference is limited to 17°C
for this study.
1000.0

     
75.0
80.0
2.5

Internal coolant flow (kg/hr)

Stack temperature (C)

0.0

Controller Output
60.0
Process Variable

750.0

50.0
Set Point
-7.5 -5.0 -2.5

40.0

500.0

25.0
20.0

250.0
-10.0

0.0

0.0 200.0 400.0 600.0 0.0 200.0 400.0 600.0

Time Seconds Time (Seconds)

Figure 4.8: Temperature controls (a) PID controller process and output variables, (b) Stack
operating temperature as a function of coolant mass flow.

37
System controls 4

Thermal controls are not actuated by stack current changes, but with temperatures
of the coolant entering and leaving the system. The set–point for stack operating
temperature control is set by difference between the desired coolant outlet
temperatures and the measured ones. In Fig. 4.8a, it can be seen that the
controller output takes longer time to reach the set–point values once it is
activated, which is a common trait of thermal management in fuel cells. Coolant
flow rates increase or decrease slowly in comparison to other controls, as can be
observed in Fig. 4.8b.

Temperature of the coolant entering the stack can similarly be controlled by the
flow of water in the external circuit. Control signal to the associated pump
regulates the electrical power of the pump and hence the coolant inlet temperature
into the stack. Figure 4.9a demonstrates the actuation of inlet temperature
controls. Once the temperature of the coolant entering the stack reaches 61°C,
flow in the external circuit is increased to maintain it. In a similar fashion,
temperature of water in the external circuit is dependent on radiator fan speed.
Radiator power consumption is also not directly affected by the PEMFC stack
current, and is activated to cool down the external coolant to the desired
temperatures (see Fig. 4.9b). For the present case, the radiator control is set to
maintain the external coolant temperature to 50°C.

     
50.0 75.0 100.0 125.0

     
300.0
70.0

0.1
External mass flow (kg/hr)

Air radiator power (kW)

Coolant Inlet temp (C)

0.075
60.0

200.0

Current (A)
50.0

0.05
100.0

0.025
40.0

25.0
30.0

0.0

0.0 200.0 400.0 600.0 0.0 200.0 400.0 600.0

Time (Seconds) Time (Seconds)

Figure 4.9: Thermal management (a) Coolant inlet temperature versus external coolant
flow, (b) Power consumption graph of air radiator.

38
Chapter 5

5 Modelling of PEM Fuel Cell Stack

The PEMFC system proposed in Fig. 3.1 is constructed by using Aspen Dynamics
simulation tool. As mentioned in the 1st chapter, that Aspen Dynamics contains a
vast library of component models which can be configured to design and simulate
various process engineering systems. However, in order to have a detailed study of
intricate fuel cell phenomena, a comprehensive fuel cell model is devised. This
model constitutes equations for fuel cell electrochemistry, mass and energy
balances, water and nitrogen crossovers in membrane and fuel cell
thermodynamics in addition to the flow models of reactants and products at fuel
cell interface. Governing equations for these sub–models are discussed below:

5.1 Fuel cell Potential

5.1.1 Thermodynamic Potential

Since the fuel cell directly converts chemical energy into electrical energy, it can be
described by the Gibbs free energy, which is a thermodynamic potential measuring
the maximum available work from a process. For a chemical reaction, the net
change in the Gibbs free energy is equal to the difference between the Gibbs free
energy of products and the reactants. In case of PEMFCs, where oxygen and
hydrogen are the only reactants and water as the product, change in Gibbs free
energy g 0f , is calculated at standard conditions.

 
g 0f  g 0f H 2O
 
 g 0f H2

2
 
1 0
gf O2
(5.1)

Here, g 0f is the specific Gibbs free energy of formation and is a function of

temperature, pressure and gas phase [66].

39
Modelling of PEM Fuel Cell Stack 5

5.1.2 Theoretical fuel cell potential

For an ideal reversible fuel cell, the maximum electrical work at a constant
temperature and pressure is equal to the Gibbs free energy:

Wel  G 0   n  F  E 0 (5.2)

where E is the ideal potential of the cell, n is the number of electrons and F is the
Faraday’s constant. Number of electrons for a H2–O2 cell reaction is 2, hence the
formation for ideal fuel cell potential becomes:

 G 0
E0  (5.3)
2F

This potential is valid for standard temperature and pressure. However, PEM fuel
cells usually operate at higher temperatures and because Gibbs free energy is a
function of temperature, it decreases the cell potential to a certain extent. The
actual theoretical cell potential can be accounted in the Nernst equation:

RT  a H 2 aO2 
1/ 2

EE  0
ln (5.4)
2 F  a H 2O 

here activity, a, is defined as:

Pi
a (5.5)
P0

Pi is the partial pressure of the gases and P0 is the standard pressure. By assuming
the gases are ideal, their activity is proportional to their partial pressure and the
activity of liquid water is equal to 1, the Nernst equation takes the form:

E  E0 
RT
2F

ln PH 2 PO2
1/ 2
 (5.6)

Theoretical cell voltage is decreased when partial pressure of water vapour

increases. For evaluation of temperature, T, the stack temperature is selected, as
the temperature gradient of gases at inlet and outlet is reasonably low. PH2 and PO2
are average partial pressures of the hydrogen and oxygen and are given as:

 y H ,out  y H 2 ,in 
PH 2   2  Pa (5.7)
 2 

40
5 Modelling of PEM Fuel Cell Stack

 yO ,out  yO2 ,in 

PO2   2  Pc (5.8)
 2 

Here, y is the molar fraction of the species at their respective terminals. Assuming
that all of the available energy is converted into electrical power, the efficiency of a
fuel cell can be stated as:

G 0
 rev  (5.9)
H

The Change in enthalpy of formation, ∆H, is also temperature dependent and the
ideal efficiency decreases with the temperature increase. The cell efficiency could
be calculated based on higher or lower heating values, HHV and LHV respectively.
For LHV, the water is assumed to be in vapour form, while liquid form of water is
assumed for HHV calculations. HHV is usually higher than LHV because the
latent heat of vaporization in the reactants and the products is accounted for as
well.

Vcell
 cell  for HHV (5.10)
1.48

Vcell
 cell  for LHV (5.11)
1.25

Where Vcell is the average cell voltage and cell is the cell efficiency at standard
temperature and pressure.

5.2 Electrochemical Model

An actual fuel cell has a lower efficiency than the ideal one because of the
irreversible losses in the cell reaction. The fuel cell terminal voltage drops from the
open circuit voltage and this drop in voltage is proportional to the current drawn
by the electric circuit. This phenomenon is known as polarization. There are three
main types of polarization: activation polarization, ohmic polarization and
concentration polarization. In a practical fuel cell, the voltage drop is significantly
more at low and high current densities. At low current densities, the activation
polarization is predominant and at high current densities, the concentration
polarization is a major cause of losses. The ohmic polarization is nothing but cell
resistance and hence with increase in cell current the voltage drop increases
proportionately. Low temperature fuel cells exhibit high activation polarization at

41
Modelling of PEM Fuel Cell Stack 5

low current densities. On the other hand, high temperature fuel cells have a low
activation losses and ohmic polarization is the most important cause for losses.

The thermodynamic efficiency and net power of the system are determined by the
current drawn and voltage produced by the stack. The average cell voltage of a fuel
cell is defined by an analytical expression:

Vcell  E  Vact  Vohmic  Vconc (5.12)

where E is the theoretical voltage, Vact the activation overpotential, Vohmic ohmic
overpotential and Vconc denotes concentration loss.

Figure 5.1: Representation of a typical Voltage–Current curve denoting different

polarization dominated regions [67].

It can be seen in Fig. 5.1 that the open circuit voltage (OCV) starts far below from
the Nernst potential of the cell. This is due to internal current losses as well as fuel
crossover through the membrane. These aspects are further discussed in the
following section and their impact on cell performance is incorporated into
activation overpotential equations.

5.2.1 Activation loss

Activation overpotential is the voltage required to overcome the activation energy

of the cell reactions. This occurs due to slow charge transfer reaction at the surface

42
5 Modelling of PEM Fuel Cell Stack

of the electrodes and a proportion of the total voltage is needed in the reaction
kinetics. Also, a certain amount of voltage is utilized in transportation of electrons
to and from the electrodes, thereby further reducing the potential at the fuel cell
terminals. The electron transfer, also referred as exchange current density is
dependent on temperature and pressure of reactants. The reactions at the anode
are very fast in comparison to the ones at cathode, therefore anode overpotential is
sometimes neglected [68]. Reduction in activation losses can be achieved by raising
the temperature, increasing the active surface of the catalysts, and increasing the
amount of reactants.

The total activation losses in the cell are equal to sum of anode and cathode
contributions. Knowing this and assuming equal transfer coefficients in both
electrodes, the Butler–Volmer equation is simplified to Tafel slope and used in the
current model:

RT  i  iloss  RT  i  iloss
 


Vact  Vact ,c  Vact ,a  ln   F ln  i
(5.13)
 c F  i0 ,c  a  0 ,a



Here, R is the universal gas constant and α denotes the transfer coefficients on the
anode and cathode sides. The term iloss accounts for the losses due to the internal
current generation and fuel crossover. Although a fuel cell electrolyte is designed
to conduct positive ions, some electrons cross the membrane internally. Since
these electrons do not follow the external path, they give rise to internal currents.
Some fuel may also diffuse through the membrane and react directly with the
oxygen without producing any external current. These internal losses affect by
reducing the current density by a few mA when at higher temperatures, however
the voltage loss could be at least 0.2V less than the reversible voltage with PEMFC
operating at normal air and pressure [66]. Here, the value for internal current
density, iloss is assumed to be equal to 0.002 A/cm2 [69].

The equations below are valid for evaluating the transfer coefficients on the anode
and cathode side respectively:

a    nel (5.14)

 c  1     nel (5.15)

The symmetry factor,   0.5 is chosen, nel is the number of electrons in the rate
step of the reaction and equal to 4 for anode and 1 for cathode. The exchange
current density, i0, is also a function of temperature and catalyst activity. An

43
Modelling of PEM Fuel Cell Stack 5

analytical expression is chosen here, which predicts the value of exchange current
density at the anode and cathode [69].

 1     n  F  E 
i0,a  n  F  k a  exp  (5.16)
 RT 

   n F E 
i0,c  n  F  k c  exp  (5.17)
 RT 

Where, k is the reaction rate coefficient and found by calibration of the model.

The equations for activation losses mentioned above have been used in the current
study, though there are alternative methods which have been considered as well.
According to the method proposed by Berger [70], the activation overpotential can
be determined by the following equation:

  
Vact  1   2   3  T ln cO 2   4  T ln i  (5.18)

The values for the constants (ξ), can be found by solving the set of equations
defined below, where c* are the concentrations of the species.

G 0f G 0f
1   (5.19)
nF  c nF

2 
R
 c nF
   c 
ln nFAcell k c c H 
1 c 
H 2O
c
 nFR ln4FA
cell  
k a c H 2 (5.20)

R 1   c 
3  (5.21)
 c nF

 R R 
 4     (5.22)
 nF  c nF 

5.2.2 Ohmic Loss

Ohmic overpotential is generated by the resistance to the ionic flow in the

electrolyte, resistance to the flow of electrons through the electrodes, and contact
resistance at the cell terminals. Ohmic losses can be reduced by using electrodes
with high electronic conductivity and thinner membranes which are conducive to
high ion transfer. However, thinner membranes would increase fuel and gas
crossovers, provide less structural stability and in some cases might not insulate the

44
5 Modelling of PEM Fuel Cell Stack

electrodes, resulting in short circuit. These limitations would demand a trade–off

of between reduction in ohmic losses and membrane thickness. The electrolyte
and electrodes both obey Ohm’s law and the formulation for the calculation of
ohmic losses used in this model is presented below.

Vohmic  rel  rion   i (5.23)

The ionic resistance of the membrane is denoted by rion and rel represents the
electronic resistance in the electrodes and the contact resistance at the terminals of
the bipolar plates. The electronic resistance is reported to be very low when
compared to the ionic resistance in the membrane; therefore it is usually ignored
in the summation of overall ohmic resistance [71]. The ionic resistance increases
with the current drawn from the stack and on the contrary decreases with the cell
temperature. A correlation for Nafion 117 membrane was formulated by Mann et
al. [72], which is used here:

  T  2.5 
2

C1  1  0.03i  0.062    i 
  303  
rion   tm (5.24)
  T  303 
mem  0.634  3i   expC2   
  T 

where λmem is the average water content of the membrane and tm is the membrane
thickness. Other terms are correction factors for operating temperatures other
than 303 K; C1 and C2 are fitted to be 180 and 16.4 respectively. There are
however another simpler correlations reported in [67], [73]:

tm
rion  (5.25)
m

The membrane activity, σm is a function of water content and the temperature, and
is determined by the following equation [67]:

  1 1 
 m  0.005139mem  0.00326  exp1268    (5.26)
  303 T  

5.2.3 Concentration Loss

The concentration overpotential arises due to a reduction in the reactant

concentrations at the electrode–membrane interface. At high current densities, the
reactions occurring within the cell are faster. The fuel and reactants are consumed
faster than they are reach the cell sites, which can be seen as a characteristic voltage

45
Modelling of PEM Fuel Cell Stack 5

drop in the polarization curve. This drop is even more pronounced if the reactants
are not pure and mixed gases are fed to the fuel cell e.g. air is used instead of
oxygen [66]. One of the main reasons for such losses is product accumulation
which dilutes the oxygen concentration at the catalytic layer, resulting in
concentration gradients. Such an overpotential could be related to the limiting
current density of the cell, which is the maximum current drawn from the cell. At
these high currents, the transport of species is comparatively slower than the
needed concentrations, at which case the partial pressures of these reactants reach
near zero at reaction sites, consequently dropping the voltage potential to very low
levels. This effect can be expressed mathematically by correlating to the limiting
current densities [74], [75], and used in the present model.

RT  i  RT 
  i 

Vconc  Vconc ,c  Vconc ,a  ln1   n F ln1  i 
(5.27)
ne F  ilim,c  e  lim,a 

Limiting current densities at cathode and anode are calculated as the product of
maximum current density and the stoichiometry of the reactants:

ilim,c  S O2  imax (5.28)

ilim,a  S H 2  imax (5.29)

The concentration overpotential is usually prominent at high currents, which are

seldom drawn from the cell for longer stack life and higher efficiencies. At average
loads, these losses do not play a significant role and can be neglected in fuel cell
modelling.

5.3 Membrane Transportation Sub–models

In addition to electronic and ionic transfers in PEMFCs, a number of complex

transport phenomena occur in the form of gas and vapour transportation across
the fuel cells. Species crossover in the MEA components is well recognized, yet
detailed mechanisms of the intricate processes are not fully understood in some
cases due to their complex nature. Nevertheless, most of these processes affect the
performance and life cycle of the fuel cell and therefore need further diligence.
Figure 5.2 provides schematic of gas and vapour transport phenomena in a typical
MEA of a hydrogen fuel cell. As it can be seen in the figure, there are six basic
processes of species transport across the membrane.

46
5 Modellling of PEM
M Fuel Cell Stack

Figgure 5.2: Scchematic diaagram of gas transport phenomena across

a PEM ffuel cell ME
EA.

Connduction off protons is the prim mary modee of transfference, whhich react with
oxyggen molecu ules on thee cathode sside to gen nerate electtrical poweer and watter as
reacttion produ uct. Some fuel and ooxygen also o crosses over
o and reeact to pro oduce
wateer without contributin
c ng to the sttack power production n. The lossses associatted to
thesee internal crossovers have beeen discussed in the electrochhemical model m
sections of thiss chapter. Some
S moleecules of water
w are alsso draggedd along with h the
hydrrogen ions to the cath hode, whilee higher waater concen ntrations att cathode result
r
in diiffusion tow
wards the anode
a of th
he fuel cell. Nitrogen being
b an innert gas rem
mains
inacttive in the reactions occurring w within these cells. Hoowever, som me proporrtions
of n
nitrogen peermeate tow wards the anode as well. Mod del formulaation for water
w
transsportation and nitroggen perme ation throu ugh the fuuel cell are presented here
accordingly.

5.3.1 Water crossover

c

Prop per manageement of water

w in PPEMFCs pllays a vitall role in eenhancing stack
lifetiime and in achieving higher perfformance goals.
g Dryin
ng of the m
membrane could
c
resullt in permmanent meembrane daamage, wh hereas on the contraary, high water
w
flood ding could
d also lead to cell revversal also resulting in permaneent cell damage

47
Modelling of PEM Fuel Cell Stack 5

[66], [76]. Drying of membrane at the anode side occurs if the fuel is not
adequately hydrated and hampers the ionic flow causing additional heat
generation within the cell. This in turn leads to increased dehydrating of the
membrane and subsequently decreasing cell voltage output. On the other hand,
PEMFCs operating at higher current densities and lower temperatures experience
water flooding. When higher currents are drawn from the stack, higher amount of
water is produced which saturates and condenses in the cathode, leading to
clogging in flow channels restricting the air flow to the catalyst electrode layer.
These cells with limited air flow channels would have a significantly less voltage
potential across them, which would reduce the output stack voltage.

Nafion membranes are characterized in terms of their thickness and water uptake
isotherms. In the present study, water content profiles are assumed to be linear
across the membrane thickness, as suggested by [77], [78]:

c  a
 mem  (5.30)
2

The membrane water content, λ, specifies the amount of water uptake per sulfonic
acid groups in the membrane and is a function of water activity at respective anode
and cathode electrode interfaces. The following expressions have been used here
[79]:

c&a  0.043  17.18a w  39.85a w2  36a w3 0 < aw < 1 (5.31)

c&a  14  1.4aw  1 0 < aw ≤ 3 (5.32)

The value of λ is equal to 14 for a perfectly hydrated membrane and in some cases
could reach 22 e.g. under super saturated conditions. The water vapour activity, aw,
can be defined as the ratio of vapour partial pressures to the saturated partial
pressures for both cathode and anode interfaces.

Pw
aw  (5.33)
Psat

In order to have a fully hydrated membrane, both the fuel and the air are
humidified before being fed to the fuel cell. Water flows in both directions within
the membrane. When the Water crosses from the anode towards the cathode, it is
referred to as electro–osmotic drag. The electro–osmotic drag occurs due to the
dragging of water molecules with hydrogen ions. The number of dragged
molecules vary and are typically found to be in the range of 0.5– 5 water molecules
per hydrogen ion [66]. Whereas, when the water travels from the cathode towards

48
5 Modelling of PEM Fuel Cell Stack

the anode, it is called as back diffusion. This diffusion is a result of high

concentration gradients across the membrane, arising due to water production at
cathode side. Water flux due to electro–osmotic drag is given by:

i
J H 2O,drag  2ndrag (5.34)
nF

where:

mem
ndrag  ndrag
sat
(5.35)
22

The saturated electro–osmotic drag, ndrag

sat
, is assumed to be 2.5 for the current
study, as it has been experimentally evaluated between 2.3 to 2.7 for fully hydrated
membranes in equilibrium with liquid water [67].

Back diffusion, on the other hand is calculated by diffusion parameters:

 dry dmem
J H 2O ,diffusion  Dw (5.36)
Mm dz

Here, dry is the density of Nafion at dry conditions and Mm represents its
molecular weight. Dw denotes the water diffusivity and z is the axis along the
membrane thickness. Table 5.1 shows key figures for Nafion 117 used here:

Table 5.1: Key parameters for Nafion 117 membrane

Parameter Value
Membrane thickness, tm cm  0.0183

Density of dry membrane,  dry g cm3  3.28

Molecular weight of membrane, M m Kg mol  1.1

The net water transport through the membrane is the difference of electro–
osmotic drag and back diffusion:

 dry dmem i
J H 2O  Dw  2ndrag (5.37)
Mm dz nF

49
Modelling of PEM Fuel Cell Stack 5

The diffusion coefficient is a function of temperature and water content of the

membrane. The following correlation has been implemented here, which was
proposed by [73], [79], [80]:

  1 1 
Dw  D exp2416     (5.38)
  303 T 

To determine the value for Dλ, the following expression suggested by Springer et al.
[79], has been applied here. This equation is only valid for λ > 4, which is justified
for current simulations since the membrane is adequately hydrated and membrane
water content is always above 4.


D  10 6 2.563  0.33mem  0.02642mem  0.0006713mem  (5.39)

Some authors have considered correlations which are related to the electro–
osmotic drag e.g. Murahashi et al. proposed the following:

D  5.51  10 7  ndrag (5.40)

Others have applied the following equations based on experimental results and
consider membrane water content only while withdrawing temperature impact on
the water diffusion:

106 , mem  2
106 1  2mem  2, 2  mem  3
D   6 (5.41)
10 3  16.67mem  3, 3  mem  4.5
1.25 10 , mem  4.5

5.3.2 Nitrogen crossover

For PEMFC membranes, high proton conductivity and low ionic resistance has
been achieved by directing significant efforts towards minimizing membrane
thickness which subsequently promotes water crossovers across it to humidify the
anode side of the fuel cell. Moreover, high power density is also attributed to the
membrane thickness in addition to GDL and stack assembly. Limitations for
physical stability of the cell and gas crossovers have to be balanced by selection of a
membrane with reasonable thickness. It is however reported by Baik and Kim [47]
and Cheng et al. [81] that membrane thickness and equivalent weight does not
considerably effect the gas permeability coefficients, but influence of water content
in the membrane is significant enough. Mittlesteadt and Umbrell [82] suggested
that in fact N2 crossover is the sum of two parallel processes; gas diffusion through

50
5 Modelling of PEM Fuel Cell Stack

polymer and water phase of the ionomer respectively. Their simple model was
correlated to the functional form by Ahluwalia and Wang [45]. Here, a similar
permeation model of [59] with a scale factor of 8 and equations with the influence
of stack temperature and membrane water uptake is used.

 EN  1 1 
K N 2   N 2 (0.0295  1.21 f v  1.93 f v )  10 11  exp  2
2
   (5.42)
T 
 R  ref T  

The activation energy for nitrogen, E N is assumed to be 24 kJ/mol,  N is scale

2 2

factor, R is the universal gas constant, Tref is 303K and fv is the volumetric ratio of
water in the membrane and is given by;

mem  Vw
fv  (5.43)
Vmem  mem  Vw

where mem is the membrane water content, Vmem and Vw are molar volumes of dry
membrane and liquid water.

In a PEMFC, concentration gradient across the membrane is the driving force for
N2 diffusion from cathode to anode. Since concentration of a certain species could
be related to its partial pressure in a volume, the nitrogen flux is calculated from
the partial pressures of N2 in cathode and anode of the fuel cell.

PN 2 ,ca  PN 2 ,an
J N2  K N2 (5.44)
t mem

where tmem is the thickness of the membrane and is assumed to be constant in the
current simulations.

5.4 Flow Modelling

For flow model of anode and cathode, the model uses mass balance to calculate
the inlet and outlet properties of streams. The partial pressures of different species
are determined from the properties of the gas streams entering and leaving the
system, gas and water crossovers, products formed and depletion of reactants
during chemical reactions within the fuel cell stack model. Faraday’s law is used to
predict the molar flow rate required off the reactants for a specific current:

51
Modelling of PEM Fuel Cell Stack 5

I
n H 2  SH (5.45)
nF 2

where, SH2 is the hydrogen stoichiometric ratio for that particular current.

Molar balance for the anode outlet is given by:

 i  iloss 
n H 2 ,out  n H 2 ,in    Acell  N cells (5.46)
 2F 


n Ha 2O ,out  n Ha 2O ,in  J H 2O  Acell  N cells  (5.47)


n Na 2 ,out  n Na 2 ,in  J N 2  Acell  N cells  (5.48)

Molar balance for the cathode outlet is summed up in the equations given below:

1  i  iloss 
nO2 ,out  nO2 ,in    Acell  N cells (5.49)
2  2F 

 i  iloss 
n Hc 2O ,out  n Hc 2O ,in   
  J H 2O  Acell  N cells  (5.50)
 2F 


n Nc 2 ,out  n Nc 2 ,in  J N 2  Acell  N cells  (5.51)

nCO2 ,out  nCO2 ,in (5.52)

n Ar ,out  n Ar ,in (5.53)

5.5 Heat Transfer

The total energy into the fuel cell is consumed by the electrical power output, heat
removed by the coolant, heat loss at the stack surface and energy stored within the
stack itself. In the current model, a lumped thermal model proposed by [27] is
considered. The stack is regarded as a single thermal mass with a heat capacity.
With the assumption of stack temperature being equal to the coolant temperature
at the outlet, heat exchanged with the coolant and hence stack operating
temperature could be determined. The energy balance between the above–
mentioned modes can be given by:

52
5 Modelling of PEM Fuel Cell Stack

dT 
Ct   Ein  E out  E el  E loss (5.54)
dt

where, Ct is the thermal capacitance of the stack, E in is the total power delivered
by the fuel to the stack (kW), E el is the power consumed by the electrical load
(kW), E out is the heat transferred to the cooling water circulating in the stack (kW),
and E loss is the heat dissipated to the ambient (kW).

The subordinate components in the BoP, i.e. anode recirculation and water
pumps, air blower, mixers and heat exchangers are modelled using the default
mathematical models provided in Aspen Dynamics.

5.6 System Efficiency

The net power produced by the fuel cell is the product of current drawn and the
voltage generated. Since, average cell voltages are calibrated to define the IV curve
here, the net power produced by the stack is attained by multiplying the cell power
to the number of cells in the stack:

Pstack  Vcell  I   N cells (5.55)

For specific power output simulations at steady–state, the current (I) is assumed to
be constant, which calculates the flow rates of reactants based on the
stoichiometry. The voltage is determined by the electrochemical equations
summarized above. However, in case of transient analysis, the current and voltage
models are independent and are determined by the system controls.

Usually stack efficiency is calculated by the following equation:

Pstack
 stack  (5.56)
n fuel  LHV

However, since the fuel cell operates at higher stoichiometry, it is futile to calculate
the stack efficiency based on inlet and outlet conditions. Efficiency calculated in
this method will be lower than the actual stack performance. On the other hand,
cell efficiency will be a more suitable parameter to gauge the fuel cell performance.
In the present model, average cell voltage is determined, which is one of the
performance parameters, therefore stack efficiency is assessed on cell efficiency.
Another approach has been used here, in which the efficiency is determined by the

53
Modelling of PEM Fuel Cell Stack 5

difference of amount of fuel entering and leaving the system, as the fuel is
recirculated back into the anode feed:

Pstack
 stack  (5.57)
n fuel ,in  n fuel ,out   LHV

Overall system efficiency is defined as the ratio of net electrical power and net
calorific value of the hydrogen supplied to the system.

Pnet
 sys  inv (5.58)
n fuel  LHV

Here, inv is the power conditioning efficiency of the inverters and is assumed to be
96%. Net electrical power is the difference of power produced by the stack and
power consumed by all the auxiliary components:

Pnet  Pstack  Pblower  Ppumps  Pradiator  (5.59)

5.7 Model Validation

Simulations for the prescribed system were carried out and reliability of the
suggested model is verified and corroborated against design validation data
provided by fuel cell stack manufacturer, Ballard Power Systems [62]. Model
characteristics of the same system at various operating temperatures and power
loads have been studied in Hosseinzadeh and Rokni [65]. Figure 5.3a represents
adaptation of the devised model which corresponds to the polarization curve
obtained from operational data associated with the given PEMFC stack.
Experimental Calculated
Relative Error System efficiency
0,65
950 4
3 0,60
900
850 2 0,55
Error %

1
Efficiency
Vcell (mV)

800 0,50
0
750 ‐1 0,45
700 ‐2 0,40
650 ‐3
600 ‐4 0,35
550 ‐5 0,30
0 0,2 0,4 0,6 0,8 1 1,2 0 50 100 150 200 250 300
i (A/cm2) I (A)

Figure 5.3: (a) Comparison of stack operational data and calculated polarization curves for
PEM fuel cell with relative error, (b) System efficiency profile at operating current range.

54
 5 Modelling of PEM Fuel Cell Stack

As a convention here, the plots are demonstrated with both current and current
densities, which correspond to the same values on the x–axis. The calculated
relative error shows a good agreement between the model and data provided. As
suggested by the manufacturer, stack temperature range of 60–70°C has been used
in simulations of the current system. Selection of other parameters and operating
conditions is based upon table 3.1.

Figure 5.3b exhibits profile of overall system efficiency at corresponding currents.

A general trend of decreasing efficiencies with increasing loads can be observed,
which is characterised by an increase in ohmic overpotentials in the stack and high
power consumption of BoP for cooling of the system. However an exception to the
contrary can be noticed at very low currents, where high parasitic losses in BoP
render the system efficiency to be lower as well. Nevertheless, higher stack
efficiencies at low loads assist the system efficiency to be not as low when
compared to that at higher currents. The system operates at maximum efficiency at
25% of the nominal stack power, i.e. at a power load of 5.2 kW and current
withdrawal of approximately 60 A.

Data Calculated Error FC Power FC Heat

Stack Efficiency
1,4 5 25 0,7
1,2
3 20
Power & Heat (kW)
H2 flow (kg/hr)

1
0,6
Error (%)

Efficiency
0,8 15
0,6 ‐1 10
0,4 0,5
‐3 5
0,2
0 ‐5 0 0,4
50 100 150 200 250 300 0 0,2 0,4 0,6 0,8 1
Current (A) Current density (Acm–2)

Figure 5.4: (a) Calculated H2 consumption versus Ballard data with relative error, (b) Heat
and power generation with stack efficiency at corresponding current densities.

Hydrogen consumption for different operating loads is compared to the provided

experimental data as well, see Fig. 5.4a. Fuel utilization is found to be an excellent
match with the present model. In Fig. 5.4b, profiles for heat and power generated
by the stack are presented. At higher current densities, stack efficiency tends to
decrease, which is due to higher ohmic and concentration resistances. Power
production is subsequently declined with higher heat generation.

55
Modelling of PEM Fuel Cell Stack 5

56
Chapter 6

6 Exergy analysis of Fuel Cell Stack

6.1 Overview

As described in previous chapters, PEM fuel cells are electrochemical devices

which generate electricity at relatively higher power density and conversion
efficiency. However, like any other device, PEMFC are also subject to
thermodynamic irreversibilities that occur during the cell operations. Therefore,
these irreversibilities need to be assessed, analysed and consequently reduced to
ensure efficient operation of PEM fuel cells. An elementary approach to cater these
issues is the energy analysis of the system or component. However in an energy
analysis, which is based on the first law of thermodynamics, the loss in quality of
energy is ignored and all forms of energy are considered to be equivalent. On the
other hand, exergy analysis, based on the first and second law of thermodynamics,
shows the thermodynamic imperfections of a process, which comprises of all
energy losses and irreversibilities. One of the implications of 2nd law of
thermodynamics is that there is always some exergy destroyed when a process
involves a temperature change, which can be attributed to overall entropy increase
of the system plus its surroundings.

In order to evaluate measures to reduce exergy losses, one needs to have knowledge
of the processes that cause these losses. In the fuel cell stack, five processes can be
distinguished in which the exergy losses occur: heat generation and transfer
between the stack and its surrounding environment, mixing of species in inlet and
outlet streams, chemical reactions occurring within the fuel cells, and polarization
losses [83]. In a fuel cell system, other irreversibilities in the BoP components are
also present; such as blowers, humidifier, heat exchangers, etc. Since, fuel cells are
electrochemical devices and contain no mechanical parts, friction and losses
associated with it are a non–issue. The chemical energy of the fuel is directly
converted to electrical power output and heat is produced as a by–product of this

57
Exergy analysis of Fuel Cell Stack 6

chemical reaction. Moreover, polarization losses due to ohmic resistances in the

cells and electronic circuit are converted to heat as well.

In this study, an exergy analysis of a PEM fuel cell stack is performed to investigate
the effects of thermodynamic irreversibilities on the performance of PEMFC. One
of the objectives is to investigate the effects of various parameters such as operating
temperatures, current densities and reactant relative humidity on stack start–up,
which is a vital issue in automotive applications itself. Here, an exergetic analysis
on the fuel cell stack is conducted in time. The scope is not to investigate
irreversibility in the surrounding components.

6.2 Laws of thermodynamics and PEM fuel cells

A fuel cell generates electrical power ( W ) through chemical reactions between the
reactants and a certain amount of heat is released ( Q ) at a constant operating
temperature (Tf). The maximum amount of work that an internally reversible fuel
cell delivers can be written as:

W max  W int rev  TFC  T0   n f  S reaction (6.1)

Here, Wint rev is the work done by the fuel cell without presence of internal losses,
S reaction is the entropy change due to the reaction, and n f is the number of moles
of fuel. The electric power output and efficiency of internally reversible fuel cell
can be derived by the energy and entropy balance over the control volume and are
given by:

W int rev   n f  H reaction  n f  TFC  S reaction   n f  G reaction (6.2)

Greaction
 int rev  (6.3)
H reaction
0

where, H reaction and Greaction are enthalpy and Gibbs free energy of the reactions.
Hence, the maximum power that can be generated by any reversible fuel cell is
given by:

W max   n f  Greaction  TFC  T0   S reaction  (6.4)

It is observed in the equation above that external irreversibilities associated with

the reactants to reach the fuel cell operating temperature from the ambient and

58
6 Exergy analysis of Fuel Cell Stack

those linked to cooling the product from fuel cell temperature to the ambient are
expressed in the second part of right side of the equation. In an ideal case, where
it is assumed that the incoming reactants and released products are at their dead
state (environment conditions), the maximum work produced by the fuel cell is
demonstrated by [84]:


W 0,max  n f  Greaction
0
 (6.5)

In this case the maximum first law efficiency is given by Eq. 6.6, which is
independent of fuel cell temperature.

Greaction
0
 0,max  (6.6)
H reaction
0

In practice, the fuel cell like any other device is accomplished with internal
irreversibility. Efficiency of the fuel cell decreases with temperature, which is due
to external irreversibility. This drop cannot be overcome by elimination of internal
losses (activation, ohmic and concentration overpotentials). However, coupling a
fuel cell with external devices can extract more power from the system. Since,
PEMFC usually operate at low temperatures, heat generated within these modules
is of little or no use. On the other hand, SOFC operate in a range of 700–900°C
and thereby can be used as a topping cycle with a Stirling engine, gas or steam
turbines. For this particular reason, SOFC have higher exergetic efficiencies when
compared to PEMFC.

Table 6.1: Maximum achievable voltage and efficiency for H2–O2 fuel cells at different
temperatures
Temperature ΔGo Max EMF Efficiency
(°C) (kJ mol–1) (V) Limit
25 (liquid) –237.2 1.23 83%
80 (liquid) –228.2 1.18 80%
100 –225.2 1.17 79%
200 –220.4 1.14 77%
400 –210.3 1.09 74%
600 –199.6 1.04 70%
800 –188.6 0.98 66%
1000 –177.4 0.92 62%

59
Exergy analysis of Fuel Cell Stack 6

6.3 Exergy balance in PEM fuel cell

In context of the presented argument, a general exergy balance diagram of PEM
fuel cell can be plotted as shown in Figure 6.1. Here, E xmass , E xheat , E xwork and
E xst are exergy transfer by mass, heat, work and exergy stored in the fuel cell
respectively.

Figure 6.1: Schematic of exergy balance for a PEM fuel cell stack.

Thermodynamic concepts and their applicability for the current study are derived
from the literature [85]. General energy and exergy balance of a system are written
as follows:

E st  E in  E out  W FC  E loss  (6.7)

E x D  E x mass ,in  E x mass ,out  E x work  E x heat  E x st  (6.8)

where, E st and E x D are the stored energy respective exergy destruction rate, due to
irreversibilities in the fuel cell stack. It is evident from the above equation that
inlet exergy should be larger than outlet exergy and some part of it will be
destroyed within the stack, which is one of the implications of 2nd law of
thermodynamics.

While,

E xmass,in  i ni eit (6.9)

60
6 Exergy analysis of Fuel Cell Stack

Here eit is the total exergy of the flow stream (kJ/kmol), n is the molar flow rate
and i denotes the number of inlet or outlet streams.

E x work is equal to the electrical power produced ( W ); which incorporates all the
internal reversibilities i.e. polarization losses. In case of fuel cells, heat transfer with
the surroundings is only the heat loss to the ambient:

 T 
E xheat  1  0 Qloss (6.10)
 TFC 

The last term on right of Eq. (6.8), E xst is the exergy stored in the fuel cell stack
and is given by:

 dT  E st
E xst   FC  (6.11)
 dt  TFC

Exergetic efficiency of the stack is defined as the ratio of total electrical power
produced to the available exergy, which is this case is determined using:

 W 
 ex    x

 (6.12)
 E x mass ,in  E mass , out 

Generally, it is a point of interest to minimize entropy generation in a system and

thereby, reduce the destruction of exergy. In the exergy analysis of the proposed
system, entropy generation rate can be calculated by:

 E x 
S gen   D  (6.13)
 T0 

The total exergy of any flow stream is further broken down into physical and
chemical exergies. Kinetic and potential exergies are neglected here.

e t  e ph  e ch (6.14)

Physical exergy is associated with the temperature and pressure of the reactants and
the products in the fuel cell system. Simply, it is the amount of useful work that a
substance can deliver when brought reversibly from its state to the “restricted dead
state”. Physical exergies can be described by their enthalpy and entropy states as
below:

e ph  h  h0   T0 s  s0  (6.15)

61
Exergy analysis of Fuel Cell Stack 6

In the current study, restricted dead states are the same as surrounding
environment conditions i.e. 25°C and 1 bar of pressure.

The chemical exergy is associated with the departure of the chemical composition
of a system from that of the environment. The components of the media in the
system are first converted to reference compounds or products and eventually
diffuse to the environment which in this case is also the dead state. Since, inlet
and outlet flows are a mixture of gases, the exergy of the gaseous mixtures is
calculated by using:

eich   y i ei0 RT0   y i ln  y i  (6.16)

Here, ei0 is the standard molar chemical exergy for component species. The
following table 6.2, displays these exergy values, which are taken from literature
[86]. Enthalpies and entropies at specific conditions are calculated using Aspen
Dynamics’ built–in functions.

Table 6.2. Standard molar chemical exergy, of substances at 298.15 K and 1 bar.
Species Formula Exergy (kJ/kmol)
Hydrogen H2 236,100
Nitrogen N2 720
Oxygen O2 3970
Water (Vapour) H2O (g) 9500
Water (Liquid) H2O (l) 900

6.4 Simulation results

Equations characterising exergy of the PEMFC described above are implemented

in Aspen Custom ModelerTM in addition to the electrochemical, thermal, feed flow
and water transportation equations to constitute a dynamic fuel cell model. The
developed fuel cell component is then incorporated into Aspen Dynamics program
to evaluate intrinsic exergetic losses during transients and steady–state as well.
Thermodynamic and exergetic efficiencies along with net power of the system are
determined by current drawn and voltage produced by the stack. The total energy
into the fuel cell is consumed by the electrical power output, heat removed by the
coolant, heat loss at the stack surface and energy stored by the stack itself. In the
current model, a lumped thermal model proposed by [27] is considered.

62
6 Exergy analysis of Fuel Cell Stack

6.4.1 Exergy efficiency of PEM Fuel Cell

The exergy efficiency of the fuel cell is calculated from the ratio of power produced
to the available energy. In a system with fuel recirculation and higher fuel
stoichiometries, the stack efficiency has very low values compared to the system
efficiency, because a high percentage of fuel is wasted. Here, average cell efficiency
is considered to be the stack efficiency in order to compare it with exergetic
efficiency. Fig. 6.2a shows the energy and exergy efficiency profiles for average
operating current loads for the described system. A general decreasing trend of
both exergy and energy can be observed, though exergy efficiency of the stack is
higher than the energy efficiency.

Entropy Generation
Exergy Energy
0,75 Exergy destruction
35 15
0,70 30
12
25
0,65
Efficiency

S_gen (W/K)

Ex_des (kW)
20 9
0,60 15 6
0,55 10
3
5
0,50 0 0
0 30 60 90 120 150 180 210 0 0,15 0,3 0,45 0,6 0,75
Current (A) Current density (Acm–2)

Figure 6.2: (a) Energy and exergy efficiency of PEMFC at different current loads, (b) Exergy
destruction and entropy generation at increasing current densities.

Exergy destruction associated with the performance of fuel cell at increasing

current loads is represented in Fig. 6.2b. At higher currents, the stack efficiency
decreases; this raises the amount of entropy generation in the fuel cell.

63
Exerggy analysis of
o Fuel Cell Stack 6

Exergy
y In Exergyy Out FC P
Power Exergy Destroyeed
7
70 70
6
60 60
5
50 50
Exergy (kW)

4
40 40
3
30 30
2
20 20
1
10 10
0 0
30 60 80 100
1 150
0 200
Current (A)

FFigure 6.3: Total

T exergy entering andd leaving the fuel cell staack at differrent operatinng
currentss with the am
mount of fuuel cell powerr produced and
a exergy deestroyed.

Figurre 6.3 portrays the exxergy balan

nce for the fuel cell sttack, wheree the amouunt of
exerggy entering the stack is divided into electrric power generation,
g exergy losss and
exerggy leaving th
he system in
i the formm of outgoinng streams..

FC Power
F Delta_Ex
x % of Ex_des

30 40

25
Ex_Destruction %

35
20
Energy (kW)

15 30

10
25
5

0 20
20
0 50 80
0 110 14 40 170 20
00
Current (A)

Figure 6.4: Percentage of

o exergy desstruction at correspondin
ng FC powerr productionn.

Exerggy destroyed by PEM fuel cells ooperating att higher loaads is compparatively higher
h
to th
hat of lowerr load cond
ditions. Thhis can be noticed in Fig. 6.4, w where the exergy
e

64
6 Exergy analysis of Fuel Cell Stack

destruction ratio to the inlet power increases from 32% to 39% with increasing
currents from 30A to 200A.

6.4.2 Effect of stack current

The described system model is designed with a control oriented approach so that it
emulates a real time PEMFC operation. Input parameters, recommended by the
stack manufacturer such as operating temperatures, pressures and reactant
stoichiometries are embedded in the system controls and the system operates more
or less at the optimum. However, control strategies related to thermal and water
management play an important role in design of optimal stack behaviour in
transient conditions. Therefore, simulations are carried out to study the dynamic
response of the stack towards the influencing parameters.

Different start–up cases were investigated to analyse energetic and exergetic

efficiencies and factors affecting them. Currents of 40 and 100 A are applied for
two start–up scenarios. Initial stack temperature is that of surrounding conditions,
here 25°C is used. Figure 6.5 shows that at high current start–ups, both energy and
exergy efficiencies of the system decrease. At steady–state, exergetic efficiency of
the stack is found to be 70% and 66% when applied current is changed from 40 to
100 A.

(a) At current of 40A (b) At current of 100A

0.75

0.75
0.5 0.55 0.6 0.65 0.7

0.5 0.55 0.6 0.65 0.7

Energetic efficiency
Exergetic efficiency

Exergetic efficiency
Energetic effciency

0.0 100.0 200.0 300.0 400.0 500.0 0.0 50.0 100.0 150.0 200.0
Time (Seconds) Time (Seconds)

Figure 6.5: Energy and exergy efficiency of stack at system start–up: (a) at current of 40A,
(b) at current of 100A.

65
Exergy analysis of Fuel Cell Stack 6

It can be observed that besides initial fluctuations at the start–up, the exergetic
efficiency rises steadily until the stack reaches stable operating conditions. This is
due to the fact that voltage generated in the fuel cell rises steadily with the
operating temperature of the stack. This dependency is discussed later in the
following section. Another noticeable observation is the slope of elevation in these
efficiencies. The gap between energy and exergy efficiency increases with time as
stack temperature is raised from cold state. This is due to the direct proportionality
between physical exergies of the component species and the temperature of the
fuel cell.

6.4.3 Influence of operating temperature

The power generation capability of a PEMFC tends to be prominent with

increasing operating temperatures, as portrayed in Fig. 6.6. This is mainly due to
the rise in the cell voltage, which is a function of Gibbs free energy, and described
in Nernst equation [65]. Gibbs free energy is also a function of reactant partial
pressures and the operating temperature. However, in the present study, effects of
partial pressures are not discussed. It is evident from the Fig. 6.6a, that an increase
of temperature from 25°C to 65°C raises the voltage from 0.63 to 0.80 V. Low cell
voltages for the same amount of power would require high current densities;
leading to greater mass flow rates of the reactants. This will result in increased
difference between the total exergies of the reactants and the products, hence
lower the exergetic efficiency as demonstrated by [38].

(a) At current of 40A (b) At current of 100A

0.85

30.0 40.0 50.0 60.0 70.0 80.0

0.85

30.0 40.0 50.0 60.0 70.0 80.0

Operating temperature (C)

Operating temperature (C)

0.7 0.75 0.8

0.7 0.75 0.8

Cell Voltage (V)

Cell Voltage (V)

0.65

0.65
0.6

0.6

0.0 100.0 200.0 300.0 400.0 500.0 0.0 50.0 100.0 150.0 200.0 250.0
Time (Seconds) Time (Seconds)

Figure 6.6: Cell voltages and stack temperature at system start–up: (a) at current of 40A,
(b) at current of 100A.

66
6 Exergy analysis of Fuel Cell Stack

In contrast, higher operating temperatures produce more voltage potential across

the electrodes, consequently generating higher power output from the system. So
far, the influence of fuel cell operating temperature on energy and exergy efficiency
has been well established for start–up case, which is also dependant on other
varying operating factors. However, the magnitude of impact of any parameter on
the system is fully understood when it is subject to variation at steady–state
operations.

(a) Increase of Op. Temp (10 C) (a) Increase of Op. Temp (10 C)

10.0

90.0
1.0

Operating temperautre (C)

Heat produced (kW)
Exergy efficiency

0.75

80.0
7.5
Liquid fraction

60.0 70.0
5.0
0.25 0.5

2.5 0.0
0.0

250.0 300.0 350.0 400.0 450.0 250.0 300.0 350.0 400.0 450.0
Time (Seconds) Time (Seconds)

Figure 6.7: Effect of increase in stack operating temperature by 10°C (a) Exergetic efficiency
and liquid water fraction, (b) heat produced by the FC stack.

(b) Entropy Generation

100.0
20.0

Operating temperature (C)

Entropy generation (W/K)

90.0
10.0 15.0

70.0 80.0
5.0 0.0

300.0 400.0 500.0 600.0

Time (Seconds)

Figure 6.8: Entropy generation in the stack with change in stack operating temperature.

67
Exergy analysis of Fuel Cell Stack 6

Here, variation of operating temperature and its effect on other parameters is

investigated. Fig. 6.7a depicts the increase in exergetic efficiency and water vapour
with temperature increment from 67°C to 77°C at a current load of 100A. The
amount of liquid water at cathode is decreased to only 2% from the previous 10
%, thereby reducing the heat generation in the stack (Fig. 6.7b). The change in
entropy generation within the fuel cell with changing operating temperatures is
shown in Fig 6.8. Results show that the entropy production decreases with the
increase of operating temperature. In addition to the factors above, at higher
temperatures reaction kinetics is faster and irreversibilities associated to activation
polarization are considerably lowered, thereby reducing entropy generation in the
fuel cell.

6.4.4 Impact of ambient temperature

It is interesting to reason if increase in operating temperature is beneficial to

PEMFC, then how the system would react to variations in the surrounding
temperature. Fig. 6.9 portrays the effects of ambient temperature on exergetic
efficiency and entropy generation when a start–up current of 60A is drawn from
the system. It can be observed in Fig. 6.9a, that at 25°C ambient temperature, the
start–up exergetic efficiency is around 1% higher than to that of 15°C ambient
temperature in Fig. 6.9b.

Start-up at ambient temp of 25 C Start-up at ambient temp of 15 C

14.0

14.0
1.0

1.0
Cathode outlet vapour frac

Cathode outlet vapour frac

Entropy generation (W/K)

Entropy generation (W/K)

0.9

0.9
Exergetic efficiency

Exergetic efficiency
12.0

12.0
0.8

0.8
10.0

10.0
0.7

0.7
0.6

0.6
8.0

8.0
0.5

0.5
6.0

6.0

0.0 100.0 200.0 300.0 400.0 0.0 100.0 200.0 300.0 400.0
Time (Seconds) Time (Seconds)

Figure 6.9: Effect of ambient temperature on entropy generation and exergetic efficiency; (a)
ambient temperature of 25°C and (b) 15°C respectively.

68
6 Exergy analysis of Fuel Cell Stack

This is due to the amount of water condensation in the fuel cell cathode which has
been discussed previously. With a same current withdrawal of 60A, more water is
saturated when the surrounding temperature is lower. In the present case, 20% of
liquid water is present at cathode exhaust at 25°C, whereas 24% of water
saturation occurs at 15°C ambient temperature. Additionally, potential difference
generated across fuel cells is lower, hence a lower efficiency at start–up. In a similar
observation, it can be noted that entropy generated in system at 25°C is around
11.2 W/K.

On the other hand, entropy generation is determined to be 12.3 W/K at 15°C

start–up. Although there is a negative impact of lower ambient temperature on the
initial stage of fuel cell operations, steady–state operations show a positive effect.
Exergetic efficiency and destruction of exergy are found to be very close for both
cases, when the system operates at steady–state conditions. Increase in exergetic
efficiency for cold surroundings mounts to 1.5% to that of 25°C, while a 6%
reduction in entropy generation is observed when ambient is at 15°C.
Furthermore, its takes more time for the system to reach stable operations.

6.4.5 Effect of liquid water

Water management is a critical issue in PEM fuel cells and has significant impact
on overall performance of the stack. Removal of water from cathode is dependent
on stack temperature and pressure drop. Fuel cell water production increases with
current and is also dependent on the number of cells in the stack. Fig 6.10a shows
the start–up of the system at 40A. The amount of liquid water at cathode outlet is
around 17 % when stack operating temperature is close to the ambient and lowers
gradually as the stack reaches steady–state. On the other hand, when the stack is
operating at 100 A, the amount of liquid water percentage at cathode outlet is
23% approximately, which is higher than to when compared to 40A simulation.
This affects the exergetic efficiency of the fuel cell in a negative way. An abrupt
decrease at start–up can be seen in both figures below, although it is much
prominent at higher current densities, see Fig. 6.10b. This is the result of lower
stack temperatures at the starting sequence and higher current densities. The most
critical factor in lower exergetic efficiencies at start–up is the amount of heat
produced by condensation of water and chemical reactions occurring at the
cathode. As the system tends towards stability, increase in operating temperature
reduces liquid water at the outlet, thereby decreasing heat generation. Liquid
fractions at anode outlet are very low when compared to cathode outlet. They are
calculated to be in the range of 1.5 to 2.0 % for both 40A and 100A current
scenarios discussed above.

69
Exergy analysis of Fuel Cell Stack 6

(a) At current of 40A (b) At current of 100A

2.0 4.0 6.0 8.0 10.0

2.0 4.0 6.0 8.0 10.0

0.8

0.8
Heat produced (kW)

Heat produced (kW)

Exergy efficiency

Exergy efficiency
0.6

0.6
Liquid fraction

Liquid fraction
0.4

0.4
0.2

0.2
0.0

0.0
0.0 20.0 40.0 60.0 80.0 0.0 20.0 40.0 60.0 80.0
Time Seconds Time (Seconds)

Figure 6.10: Heat produced by liquid water and its effect on exergetic efficiency at system
start–up: (a) at current of 40A, (b) at current of 100A.

6.5 Summary

In this chapter, energetic and exergetic performances of PEMFC stack were

evaluated in time, considering the first and second law of thermodynamics and
associated irreversibilities were identified. Exergy analysis of PEMFC was
conducted for start–up case, in order to identify effect of underlying operating
parameters. Results show that the largest exergy destruction takes place when
system is started at high–currents and low ambient conditions. Also, cell voltages
increase with increasing stack temperature. Moreover, entropy generation is
reduced at high operating temperatures and with low presence of liquid water in
the cathode outlet. Furthermore, low ambient temperatures increase the exergetic
efficiency slightly when operating at steady state temperatures. Moreover, presence
of liquid water at cathode outlet is also one of the main contributors in high
entropy generation in the fuel cell stack. At high current start–ups, saturated water
accounts for almost quarter of the total cathode exhaust, generating additional
heat to be removed from the stack.

70
Chapter 7

7 System Analysis on Start–up

7.1 Premise

In order to make hydrogen powered fuel cell vehicles commercially viable, there
are many challenges which need to be overcome. The durability of PEMFCs is one
of these challenges that need to be catered to prolong the engine life to that of
currently used ICE vehicles. A common failure mode for automotive fuel cells is
the voltage degradation, which can be mainly attributed to voltage cycling during
transient operations, fuel starvation, and during system startup and shutdown [87],
[88]. Due to these start–up and shutdown cycles over a long period, the fuel cell
electrodes develop carbon corrosion which is one of the root causes for voltage
degradation [89].

Other issues point towards the practicality of fuel cell automobiles. The user will
expect to drive the vehicle almost immediately after turning it on, implying a short
warm–up time. However, environmental conditions dictate the heating times for
these fuel cells. Lower ambient temperatures impede the quick start–up by either
freezing of the residual water in the stack at sub–zero temperatures and secondly
the temperature–dependent performance of the fuel cell until it reaches higher
steady–state operating temperatures. Many key parameters that affect the start–up
operations of the fuel cell have been reported in the open literature; such as
ambient temperatures, water content in the membrane, operating loads and
corresponding current densities, reactant stoichiometry, and water saturation
within fuel cell. As already described in the literature review, a very large volume of
research has been conducted on cold starts with many propositions of
methodologies to prevent water freezing in the fuel cells; however, there is still a
need for models elaborating further details of the intricate processes and
mechanisms of local transport phenomena on the performance of the fuel cells
when starting at normal operating temperatures.

71
System Analysis on Start–up 7

Building on these needs, a start–up analysis is performed for the described

PEMFC system, starting at ambient temperature of 25°C and the focus has been
set to evaluate the effects of key parameters till the system reaches steady–state
operating temperatures. Thereby, a complete control–oriented system–level
dynamic model is proposed in this chapter which incorporates all necessary BoP
components for a PEMFC fuel cell stack by including electrochemical, thermal,
feed flows and water transportation models and a detailed control strategy. In
totality, here a sizable focus has been set for (i) comprehensive control strategy
regulating hydrogen and air feed flows, coolant inlet and fuel cell stack operating
temperatures, (ii) thermal model with liquid coolant circuit incorporated in it
which takes into account reactant and product phase changes, (iii) Thermal
management including effects of coolant controls and heat exchangers on fuel cell
stack performance, and (iv) Water management and effects of water crossover on
anode recirculation loop and fuel cell relative humidity.

7.2 System start–up

Fuel cell start–up sequence could be defined as the time required by a system to
reach stable conditions, which is achieved when the voltage and consequently
power reach a stationary value. Normally, the optimal operating temperature of a
PEM fuel cell is reported to be in the range of 60–80°C. Within this range,
chemical reactions occurring inside the fuel cell are relatively fast and facilitate
removal of water produced by these reactions. Whereas, at low temperatures, there
is a considerable rise in kinetic and ohmic losses in addition to reactant transport
losses caused by a high rate of water condensation [34]. Lower membrane
conductivity is also associated to these temperatures. Thus, it is necessary to elevate
the stack operating temperature as soon as possible during the start–up in order to
meet the external load requirements.

When the fuel cell is in standby stage i.e. when no current is drawn, auxiliary
components such as air blower, coolant pumps, radiator fan and hydrogen inlet
valve are closed. At system start–up, all of these components are switched on and
fuel valve is opened. Power from the stack is drawn only when the current load is
applied on the stack. In the current simulations, values of stream flows are
initialized to be non–zeros. In this way, a realistic start–up sequence could be
imitated and furthermore initial zero–condition in Aspen Dynamics could be
avoided. Initial temperature of the stack is assumed to be equal to ambient
temperature (25°C) and a start–up power load is introduced by ramping of current
at a rate of 20 A/sec.

72
7 System Analysis on Start–up

Operating temperature of the stack is maintained around 60–70°C with a pressure

range of 1.1–2.2 bar. Maximum power produced from the described stack is 21.2
kW corresponding to a current of 300A; however it is generally operated at lower
current ranges to attain higher efficiencies by reducing ohmic and concentration
overpotentials within the fuel cell stack. Other parameters and operating
conditions are selected from chapter 4.

7.2.1 Stack warm–up periods

Different start–up strategies are compared and the results are summarized below.
Currents at 40, 60, 80 and 100 amperes are drawn and heat–up times for these
four cases are presented in the Fig. 7.1. At start, the flow in cooling circuits is set
to the lowest since it is desired to raise the stack temperature to its optimal
operation. Coolant flow in the internal circuit is fixed to 290 kg/hr as
recommended by the manufacturer. Flow in the internal cooling circuit increases,
once the stack temperature difference increases by 15°C. Flow in external circuit is
regulated to maintain stack inlet temperature of internal cooling circuit. Air flow
in radiator is started to maintain temperature of external cooling loop to around
50°C. As can be seen in the figure below, when the stack is started at current of
40A at a ramp rate of 20 A/s, it takes approximately 425 seconds to reach stable
operating temperature.
900.0

900.0
80.0

80.0
External Coolant flow (kg/hr)
Internal coolant flow (kg/hr)

Coolant temperature [C]

Coolant temperature [C]

Air Radiator flow (kg/hr)

600.0

600.0
60.0

60.0
300.0

300.0
40.0

40.0
0.0

0.0

0.0 400.0 800.0 1200.0 0.0 400.0 800.0 1200.0

Time (Seconds) Time (Seconds)

Figure 7.1: Start–up at 40A; (a) Stack heat–up times and corresponding coolant flow rate,
(b) Flow rates in external cooling circuit and air radiator.

73
System Analysis on Start–up 7

The heat–up time for 60, 80 and 100 A is determined to be 300, 220 and 175
seconds respectively. The heat–up time becomes shorter as the applied current is
increased. High current densities at the cell surface result in a larger amount of
heat generation in a relatively short time, which rapidly raises the temperature of
the cells.
900.0

900.0
80.0

80.0
External coolant flow (kg/hr)
Internal coolant flow (kg/hr)

Coolant temperature (C)

Coolant temperature (C)

Air Radiator flow (kg/hr)
600.0

600.0
60.0

60.0
300.0

300.0
40.0

40.0
0.0

0.0

0.0 150.0 300.0 450.0 600.0 750.0 0.0 150.0 300.0 450.0 600.0 750.0
Time (Seconds) Time (Seconds)

Figure 7.2: Start–up at 60A; (a) heat–up time and coolant flow, (b) External cooler and
radiator flow rates.
900.0

900.0
80.0

80.0
External coolant flow (kg/hr)
Internal coolant flow (kg/hr)

Coolant temperature (C)

Coolant temperature (C)

Air Radiator flow (kg/hr)
600.0

600.0
60.0

60.0
300.0

300.0
40.0

40.0
0.0

0.0

0.0 200.0 400.0 600.0 0.0 200.0 400.0 600.0

Time (Seconds) Time (Seconds)

Figure 7.3: Start–up at 80A; (a) heat–up time and coolant flow, (b) External cooler and
radiator flow rates.

74
7 System Analysis on Start–up

Accordingly, the flow rates in the internal and external circuit are increased to
maintain the stack operating temperature. Once, the temperature of external
cooling circuit reaches the control value, air radiator is actuated to expel the heat
from the system. Figures 7.2, 7.3 and 7.4 further exemplify the heating times with
corresponding coolant flow rates at currents of 60A, 80A and 100A respectively.
900.0

900.0
80.0

80.0
External coolant flow (kg/hr)
Internal coolant flow (kg/hr)

Coolant temperature (C)

Coolant temperature (C)

Air Radiator flow (kg/hr)
600.0

600.0
60.0

60.0
300.0

300.0
40.0

40.0
0.0

0.0

0.0 150.0 300.0 450.0 0.0 150.0 300.0 450.0

Time (Seconds) Time (Seconds)

Figure 7.4: Start–up at 100A; (a) Stack heat–up times and corresponding coolant flow rate,
(b) Flow rates in external cooling circuit and air radiator.

7.2.2 Power generation and system efficiency

Generally, stack efficiency is higher to that of the complete system with BoP.
However, in a system with fuel recirculation, stack efficiency is attributed to fuel
stoichiometry, though there is no considerable effect on the system efficiency.
When the actual and stoichiometric mass flow of hydrogen differ least then less
fuel is wasted and stack efficiency is at its maximum. The same argument could be
generalized for system efficiency as well. This is because at higher air mass flows,
power consumption of auxiliary components increases.

Different start–up cases were investigated to analyse the system efficiency, net
power output and auxiliary power consumptions. Currents of 60 and 100 A are
applied for two start–up scenarios. Overall efficiency of the system is as much
affected by blower and radiator fan, as by the fuel cell stack itself. Fig. 7.5b shows
that at high current start–ups, power generated from the stack increases. However,
efficiency of the system decreases. As can be seen in the figure, system efficiency
decreases from 55% to 50% when applied current is changed from 60 to 100 A.
During start–up, power consumption of coolant pumps and air radiator is very low

75
System Analysis on Start–up 7

when compared to that of the air blower (Fig. 7.5c). Therefore, auxiliary power
consumption is dominated by the air blower. It is observed that power consumed
by the air blower increases during start–up and becomes constant after a few
seconds. Also, power produced by the stack increases at a sound pace and becomes
stable once steady state is reached. This elevation in the stack power is due to the
fact that the operating temperature of fuel cell increases till it reaches the desired
optimal temperature.

(a) System efficiency at 60A (b) System efficiency at 100A

9.0

0.8

9.0

0.8
Auxiliary Power (kW)

Auxiliary Power (kW)

Stack Power (kW)

Stack Power (kW)

0.5 0.6 0.7

0.5 0.6 0.7

System efficiency

System efficiency
6.0

6.0
3.0

3.0
0.4

0.4
0.0

0.3

0.0

0.3

0.0 150.0 300.0 450.0 600.0 750.0 0.0 150.0 300.0 450.0
Time (Seconds) Time (Seconds)

(c) Power consumption at 60A (d) Power consumption at 100A

1.0

1.0
0.2 0.4 0.6 0.8

0.2 0.4 0.6 0.8

Air Radiator (kW)

Air Radiator (kW)

Air Blower (kW)

Air Blower (kW)

Pumps (kW)

Pumps (kW)
0.0

0.0

0.0 150.0 300.0 450.0 600.0 750.0 0.0 150.0 300.0 450.0
Time (Seconds) Time (Seconds)

Figure 7.5: System efficiency and stack power; (a) at start–up current of 60A (b) at start–up
current of 100A, auxiliary power consumption at start–up of 60A (c) and 100A (d)
respectively.

76
7 System Analysis on Start–up

As seen in Fig. 7.5c and 7.5d, radiator fan is shut off at the start sequence, but is
activated once temperatures in the external cooling circuit are high enough to be
cooled down. Air flow in the radiator can be regarded as a linear function of its
electrical power and therefore once it is started by the controller, the system
efficiency decreases. Since, at high temperatures heat production in the stack is
quite high, the radiator fan consumes around 250 W for a current withdrawal of
100 A, whereas it only requires approximately 125 W when current is at 60 A.

7.2.3 Cell voltage at start–up sequence

Potential difference generated across the fuel cell is dependent on factors such as
current density, reactant partial pressures and temperature at which cell reactions
occur. Figure 8 elucidates some of these underlying factors which affect the voltage
and consequently power output of the stack. Here, during the PEMFC start
sequence, current is ramped to 80 A at a rate of 20 A/s. Since current density and
reactant pressures are regulated at very fast rate, they follow a constant profile
along the time axis.
80.0
0.85

7.5
0.1 0.15 0.2 0.25 0.3
Operating temperature (C)
Average Cell Voltage (V)

Current Density (A/cm2)

Stack Power (kW)

0.8

60.0
0.75

40.0
0.7 0.65

20.0

0.05

0.0 100.0 200.0 300.0 0.0 50.0 100.0 150.0 200.0 250.0
Time (Seconds) Time (Seconds)

Figure 7.6: (a) Cell voltage as a function of operating temperature with start–up current of
80A, (b) Stack power versus cell current density.

77
System Analysis on Start–up 7

70.0
0.82

7.0
Operating temperature (C)
0.72 0.74 0.76 0.78 0.8
Cell voltage (V)

6.8
Power (kW)
60.0

6.6
50.0

6.4
40.0
100.0 150.0 200.0 250.0
Time (Seconds)

Figure 7.7: Effect of sudden temperature drop on voltage and power.

However, the stack temperature, which is attributed to the coolant mass flows,
keeps on increasing until it reaches the desired operating temperature. As can be
seen in Fig. 7.6a, cell voltage increases with the operating temperature. This fact is
due to the reason being fast reaction kinetics at the electrodes of individual cell
sites. It can be observed that the voltage reduces abruptly when simulation time is
around 160 seconds. At this stage, the decrease in cell temperature and voltage is
caused by the sudden increase in coolant flow into the stack, which maintains the
stack operating temperature. In a similar manner, power produced by the stack
also increases and decreases with cell voltage, see Fig. 7.6b. Nevertheless, Fig. 7.7
shows a detailed view of the aforementioned instance of time. Evidently power
produced by the stack responds to a similar trend to that of voltage against
operating temperature of the stack.

In terms of voltage overpotentials, the voltage drop is characterized by activation,

ohmic and concentration losses. At start–up, activation polarization is the
dominant factor in voltage losses. Figure 7.8 displays the activation overpotential
with current ramp (20A/s) till it reaches 80A in 4 seconds. The peak of activation
loss can be observed at this instance of time. Since, exchange current density is
dependent on temperature and pressure of reactants; as temperature of the stack
rises, the overpotential tends to decline and becomes constant at steady–state
conditions. Ohmic overpotential, on the other hand increases with the current
drawn from the stack.

78
7 System Analysis on Start–up

0.006
0.5
0.8

0.1
Concentration Loss (V)
Activation Loss (V)
0.2 0.3 0.4

0.025 0.05 0.075

Cell Voltage (V)
0.75

Ohmic Loss (V)

0.004
0.7

0.002
0.65

0.1
0.0
0.6

0.0

0.0 50.0 100.0 150.0 200.0 250.0

Time Seconds

Figure 7.8: Contribution of different losses in cell voltage reduction.

With the increase in cell temperature, ionic resistance drops to very low values.
Concentration losses, which arise by the deficiency of reactants at reaction sites, do
not affect the voltage to a considerable extent. The current ramp up becomes
stable within a few seconds of start–up and does not fluctuate after that, therefore
the mass transfer overpotential does not contribute to the voltage drop effectively.

7.2.4 Fuel and oxidant flows

This section caters analysis of reactant flows into the fuel cell and investigates their
effect on cell performance as well as on adjacent streams. Cell performance is not
sensitive to fuel stoichiometry; whereas the cathode is more sensitive to air
stoichiometry. For this reason, simulations are carried out by employing operating
conditions and parameters mentioned previously (chapter 4). A start–up current of
60 A is used in this case. Figure 7.9 specifically portrays flow behaviour of fuel and
air into and out of anode and cathode channels respectively. It is observed that
fuel entering the stack contains water i.e. the fuel is humidified. Since, fuel from
anode exhaust is recirculated; it adds more unutilized hydrogen and water to the
inlet stream.

Composition of hydrogen in the entrance becomes stable after initial regulation.

However, the amount of water keeps on increasing due to the increase in stack’s
temperature and water cross–over from cathode. This occurrence shows similar
incremental trends as in case of stack temperature in Fig 7.1. Anode outlet flows
show similar trends, though the amount of hydrogen at the outlet is reduced as it
is consumed in the cell reactions.

79
System Analysis on Start–up 7

0.2 0.4 0.6 0.8 1.0 1.2 1.4 (a) (b)

0.2 0.4 0.6 0.8 1.0 1.2 1.4

Anode outlet flow (kg/hr)
Anode inlet flow (kg/hr)
Hydrogen flow (kg/hr)

Hydrogen flow (kg/hr)

Water flow (kg/hr)

Water flow (kg/hr)

0.0 100.0 200.0 300.0 400.0 500.0 0.0 100.0 200.0 300.0 400.0 500.0
Time (Seconds) Time (Seconds)

(c) (d)
20.0

20.0
Cathode outlet flow (kg/hr)
Cathode inlet flow (kg/hr)

Oxygen flow (kg/hr)

Oxygen flow (kg/hr)

Water flow (kg/hr)

Water flow (kg/hr)

15.0

15.0
10.0

10.0
5.0

5.0

0.0 100.0 200.0 300.0 400.0 500.0 0.0 100.0 200.0 300.0 400.0 500.0
Time (Seconds) Time (Seconds)

Figure 7.9: Fuel and air flows at inlet and outlet of the stack: (a) Fuel flow at anode inlet,
(b) Fuel flow at anode outlet, (c) Air flow at cathode inlet and (d) Air flow at cathode
outlet.

Air flow into the cell can also be seen to increase with time (Fig. 7.9c). As
temperature of the stack increases, more water is needed to maintain the humidity
at the desired level; here, a constant relative humidity of inlet air is assumed.
Oxygen and nitrogen remain more or less constant throughout the heat–up time.
Once, the stack reaches its steady–state, the air flow becomes constant. At cathode
outlet (Fig. 7.9d), the total flow increases correspondingly, however the oxygen
content is lower as it is consumed in the reaction as well, whereas water flow is
increased. The additional water in the outlet is the product of fuel cell reaction.
Higher stoichiometries for both fuel and oxidant are maintained to manage this
water production in the cathode and to dilute nitrogen crossover to the anode.

80
7 System Analysis on Start–up

7.2.5 Air humidification effects

In PEMFC systems, adequate hydration is necessary to facilitate ionic conduction

in the membrane and to avoid physical degradation over its extended usage. Cell
drying depends on a number of operating conditions including current density,
reactant flow rate, gas composition, relative humidity, inlet pressure, and cell
temperature. Dry operation will lead to stack performance degradation and
eventually, internal leaks. On the other hand, excess water accumulation in the
fuel cell might cause water flooding at the cathode region which is also not
desirable. Inlet humidities also contribute to liquid water saturation in the fuel
cell, which is reported in [90].

(a) (b)

0.5 0.6 0.7 0.8 0.9 1.0

1.0

Cathode outlet relative humidity

Cathode inlet relative humidity
Anode outlet relative humidity
Anode inlet relative humidity
Cathode outlet relative humidity
0.9
Cathode inlet relative humidity
Anode outlet relative humidity
Anode inlet relative humidity

0.5 0.6 0.7 0.8

0.0 50.0 100.0 150.0 200.0 0.0 50.0 100.0 150.0 200.0
Time (Seconds) Time (Seconds)

(c) (d)
1.0

0.5 0.6 0.7 0.8 0.9 1.0

Cathode outlet relative humidity
Cathode inlet relative humidity
Anode outlet relative humidity

Cathode outlet relative humidity

Anode inlet relative humidity

Cathode inlet relative humidity

Anode outlet relative humidity
Anode inlet relative humidity
0.9
0.8
0.7
0.6
0.5

0.0 50.0 100.0 150.0 200.0 0.0 50.0 100.0 150.0 200.0
Time (Seconds) Time (Seconds)

Figure 7.10: Effect of air inlet relative humidity on fuel inlet and outlet: (a) Air inlet
humidity of 95%, (b) 90%, (c) 85% and (d) 80% respectively.

81
System Analysis on Start–up 7

As mentioned above, the proposed system in this study utilizes an external

humidifier to moisten incoming air, whereas anode side is not humidified.
Recirculation of fuel serves the humidification purpose on the anode side of the
fuel cell. Here, the role of the inlet relative humidity at the cathode of fuel cell
electrodes and its interaction with adjacent flows is discussed and a comparison is
presented by a systematic analysis of these results. A constant relative humidity of
95%, 90%, 85% and 80% is selected at a start–up current of 120A for the present
simulations. Fig. 7.10 depicts relative humidity states at electrode inlet and outlets
respectively. The recirculation stream mixing with anode inlet stream contains
approximately 1% liquid water droplets.

It can be witnessed that anode inlet humidity increases from 90% to 94% when
cathode inlet air is humidified from 80% to 95%. Similarly, relative humidity at
anode exhaust is increased from 95% to 98%. It can be seen that time taken to
reach these humidity levels is more when air inlet humidity is kept lower. Since,
water is produced at the cathode; it is fully saturated at the outlet of the fuel cell.

(a) (b)
0.6 0.7 0.8 0.9 1.0
0.4 0.5 0.6 0.7 0.8 0.9 1.0

Anode outlet RH
Anode outlet RH

Anode inlet RH
Anode inlet RH

0.4 0.5

0.0 50.0 100.0 150.0 200.0 0.0 50.0 100.0 150.0 200.0
Time (Seconds) Time (Seconds)

Figure 7.11: Relative humidity of fuel at inlet and outlet: (a) with 1–2% liquid at anode
inlet (b) with no liquid at anode inlet.

While liquid droplets are not wanted in anode streams as they could cause water
flooding in the stack, they affect the humidity levels in the anode. Figure 7.11
elaborates on the time taken by the fuel cell to reach the maximum humidity levels
in the anode side. It can be noticed that the fuel cell reaches maximum humidity
levels within 100 s when the anode recirculation stream contains liquid water. On
the other hand, in the case where all the liquid water is assumed to be trapped by

82
7 System Analysis on Start–up

the knockout drum, the time required to reach highest humidity increases to
approximately 130 s.

7.2.6 Water saturation at cathode

Removal of water from the cathode is dependent on stack temperature and

pressure drop. Temperature is the more critical factor, since at high temperatures
water will be in the vapour state and easier to remove. Stack water production
increases with current and is also dependent on the number of cells. The exact
amount of liquid water product depends on the cathode outlet temperature.
Figure 7.12 provides information on the amount of liquid water at inlet and outlet
of the cathode channel. Fig. 7.12a shows the start–up of the system at 40A. As
seen, the amount of liquid water at cathode outlet is around 17 % when stack
operating temperature is close to the ambient and lowers to 11% when the stack
reaches steady–state at 65°C.

(a) (b)
80.0

80.0
0.05 0.1 0.15 0.2 0.25 0.3

0.05 0.1 0.15 0.2 0.25 0.3

Cathode outlet temperature [C]
Cathode outlet temperature [C]
60.0

60.0
Liquid fraction
Liquid fraction

40.0
20.0 40.0

20.0 0.0
0.0

0.0 150.0 300.0 450.0 0.0 75.0 150.0 225.0

Time (Seconds) Time (Seconds)

Figure 7.12: Percentage of liquid water at cathode outlet; (a) at start–up current of 40A
and (b) at start–up current of 100A.

On the other hand, when the stack is operating at 100 A (Fig. 7.12b), the amount
of liquid water percentage at cathode outlet is 23% approximately, which is much
higher than the case with 40A start–up. This is because of higher current densities
at start–up sequence when the stack temperatures are very low. When higher
currents are drawn, reactions within the fuel cell are accelerated which demand
more intake of fuel and oxidant. Consequently these reactions produce more
water when compared to production at low currents. Since the stack temperature

83
System Analysis on Start–up 7

is low at start–up, higher percentage of produced water is saturated at the fuel cell
cathode. At low currents, water removal is the dominant factor and reactant
stoichiometries are determined by the minimum flow rates required for water
removal which in the present case are more than adequate to provide the necessary
concentrations. At steady–state when stack operating temperature is around 67°C,
liquid water fraction in the outlet air is determined to be around 11%. Liquid
fractions at anode outlet are very low when compared to cathode outlet. They are
calculated to be in the range of 1.5 to 2.0 % for both 40A and 100A current
scenarios discussed above.

7.3 Summary

This chapter encompassed the dynamic analysis of PEMFC system during start–up
sequence. Main contributions of the suggested model are attributed to system
response methodology, which incorporates stack thermal behaviour in addition to
fuel cell electrochemistry, water crossover, mass and energy balances. It is observed
that system efficiency and voltage output are higher at low power start–ups but for
the fuel cell stack it takes longer time to reach stable operating conditions.
Furthermore, air radiator consumes more power at high currents once the system
is stable and due to this fact, system efficiency is reduced. In addition, it is shown
that cathode inlet water levels are adequate enough to humidify fuel stream, which
is recirculated into the anode. Finally, amount of liquid water in the cathode
outlet is considerably higher, at high current density start–ups, while water
removal from cathode exhaust requires additional attention at low stack operating
temperatures. While high start–up current densities cause issues at cathode,
humidity levels increase faster reducing the dry cell operating times.

84
Chapter 8

8 Transitory load changes –System

Analysis

8.1 Overview

As a power source for automotive applications, PEMFC systems are usually subject
to inflexible operating requirements when compared to stationary applications.
These systems have to operate at varying conditions related to temperatures,
pressures, power load and humidity. PEMFC dynamics are influenced by reactant
flows, heat management and water content in the streams as well as within the fuel
cell itself. All the auxiliary components, such as air and fuel supply system which
include compressors and control valves, and the thermal control system which
consists of heat exchangers, coolant pumps and air radiators need to be controlled
for optimum operation of fuel cell when the system experiences varying load
changes. Understanding the transient behaviour of a PEMFC therefore becomes
very beneficial in dynamic modelling of these power modules at a system–level.
Most of the studies available in literature focus on transient response of fuel cell
stack under different operating conditions; primarily on individual component
analysis. Therefore, a need for a control–oriented dynamic system model is
identified, which simulates a fuel cell stack under multiple varying operating
conditions and changing auxiliary components outputs. Dynamic characteristics of
PEMFC are also attributed to the heat management and water transportation,
which are critical issues and need to be optimized for efficient operations and
improve durability of the fuel cell stack. Investigations for effects of heat
exchangers on fuel cell stack performance and water crossover on anode
recirculation operations are therefore selected to be one of the primary objectives
here.

This chapter aims at analysis and investigation of a complete PEMFC system and
studies its transient response to varying load and operating conditions. Effects of

85
Transitory load changes –System Analysis 8

system controls and other components on the fuel cell performance are also
investigated. A thermal management strategy has been designed and its dynamic
impact on fuel cell stack has been conveyed here as well. Analysis of water
crossover in the fuel cell and its impact on anode recirculation operations has been
conducted and suitable findings are reported. Moreover, two–phase characteristics
of concerning material streams are determined which provide suitable insight to
saturated water issues in the fuel cell stack.

8.2 Results and discussion

In practice, the fuel cell is coupled with a battery and load management is
attributed to the operational strategy of the two combined. Therefore, external
load changes are not directed towards the fuel cell module straight away. The
battery acts as a buffer between the fuel cell and the external load, unless under
special circumstances when it reaches its lowest charge levels. This is when the fuel
cell becomes fully responsible for providing all the external power in addition to
charging the battery. Present simulations are carried out without the battery
component and load changes are assumed to be for the case when only the fuel
cell stack is affected. Results for transitory effects under variable load changes
(when a current corresponding to a specific power load is drawn from the stack
and varied at any occurrence of time) are discussed below:

8.2.1 Voltage loss during load change

An instance of load change, when current is ramped from 60 A to 100 A at a rate

of 20 amperes per second and vice versa, is presented in Fig. 8.1a and its effects on
cell voltage are examined. It can be observed that the cell voltage reduces abruptly
with current surge and vice versa. This decline in voltage potential at increase of
current density is attributed to the cell overpotentials. Fig. 8.1b elucidates the
losses that account for the voltage drop in individual cells. When the current is
ramped up from 60A to 100A, there is a certain lag between the consumption of
reactants and the supply. The time taken by the air blower and fuel valve to
provide additional air results in concentration overpotential, however the
magnitude of mass transfer loss is not considerably high to be the dominant factor.
This is due to the fact that current range of 100A equals current density of 0.35
Acm–2, which is well beyond the region governed by concentration losses.

86
8 Transitory load changes –System Analysis

20.0 40.0 60.0 80.0 100.0

0.84

0.005 0.00: 0.015 0.02

0.03
0.8

Concemtration Loss (V)

Activation Loss (V)
0.4
Cell Voltage (V)

Ohmic Loss (V)

current (A)

0.02
0.8

0.0

0.00:
0.76

-0.4 -0.8
0.72

0.0

0.0
750.0 800.0 850.0 900.0 950.0 750.0 800.0 850.0 900.0 950.0
Time (Seconds) Time (Seconds)

Figure 8.1: (a) Voltage change with stack current, (b) Variation in overpotentials at
transitory change of 60–100–60A.

It can be similarly observed that ohmic losses also increase with current density. At
the current ramp–up, the ohmic overpotential surges, however it can be noticed
that it decreases slightly and then gradually increase to become stable. Actually,
this is because of the temperature dependency of membrane ionic resistance.
Effect of operating temperature is further discussed in detail in the proceeding
section.

8.2.2 Thermal management of the system

Besides current density, cell voltage is also a function of operating temperature and
pressure. Since, pressures of the reactants are already regulated; here we will
discuss temperature dependency of developed potential differences in the cell.
Voltage increases with the elevation of operating temperature. This is due to fast
reaction kinetics at the electrodes of individual cell sites when operating at higher
temperatures. At the same instance of current ramp as more current is drawn;
more heat is produced by the reactions at cell sites, thus elevating the stack
temperature. Consequently, the voltage also increases after the initial dip and
follows the decreasing profile with the stack temperature as it is controlled towards
the desired operating temperature. The opposite could also be observed when the
current is reduced back to 60A.

87
Transitory load changes –System Analysis 8

1500.0
-80.0 -40.0 0.0 40.0 80.0 120.0
-80.0 -40.0 0.0 40.0 80.0 120.0
           

0.84

80.0
60.0 65.0 70.0 75.0 80.0 85.0

Internal coolant flow (kg/hr)

Operating temperature (C)

Stack temperature (C)

70.0
0.8
Cell voltage (V)

1000.0
Current (A)

Current (A)
0.76

60.0

500.0
0.72

50.0
0.68

40.0
750.0 875.0 1000.0 750.0 875.0 1000.0
Time (Seconds) Time (Seconds)
1500.0
-80.0 -40.0 0.0 40.0 80.0 120.0

     
External coolant flow (kg/hr)
Air radiator flow (kg/hr)
1000.0
Current (A)

500.0

750.0 875.0 1000.0

Time (Seconds)

Figure 8.2: Current ramp–up from 60–100A and vice versa: (a) Changes in cell voltages
and operating temperature, (b) Changes in coolant flow rate to maintain the fuel cell
operating temperature, (c) variation in external cooling flows and air radiator.

Since, temperature affects the voltage and overall system efficiency, it becomes
important to have a stable operating temperature and therefore an efficient
thermal management system. Figure 8.2b shows the variation in coolant flow rates
with temperatures. Similarly, it can also be seen in the Fig. 8.2c, that coolant flow
rates in both cooling circuits and an air radiator maintain the temperature of the
stack, though it is very slow compared to the reaction kinetics affecting the stack
voltage. Thermal control strategy and regulation of air radiator, internal and
external cooling circuits has already been detailed in chapter 4. Ramping of
current increases the operating temperature of stack thereby actuating controls to
contemplate this increase. Flow in the internal cooling circuit, which is coupled
with the fuel cell stack is increased to extract heat and maintain the desired

88
8 Transitory load changes –System Analysis

temperature difference between the stack inlet and outlet. Flow in the external
cooling circuit is pumped up to retain the inlet temperature to 60°C. Air flow in
radiator is also increased to maintain temperature of external cooling loop around
50°C by rejecting the heat to the surroundings. Rate of coolant flows in respective
streams is associated with the power consumed by coolant pumps and air radiator
which ultimately affects the overall system efficiency. It is also observed that air
radiator being a liquid–gas heat exchanger has the slowest reaction time out of the
three. Since temperature controls are slow compared to electrochemical reactions,
overall thermal control strategy has a fair impact on stack voltage which undergoes
frequent load changes.

8.2.3 System efficiency and power consumption

As can be seen in the Fig. 8.3a, when the stack current is altered from 60A to
100A, there is a reduction in system efficiency, mainly due to the associated
voltage drop. Overall efficiency of the system is as much affected by air blower and
radiator fan, as by the fuel cell stack itself. It can be observed that power produced
by the stack increases with current drawn. Also, power consumed by auxiliary
components increases, thereby reducing system efficiency from 55.5% to 50%.
Once decreased, there is an abrupt rise in efficiency due to increase in voltage and
it fluctuates at around 50% mark due to slow temperature controls and fluctuating
power consumption by air radiator.

Figure 8.3b displays the power generated by the fuel cell stack and consumption of
BoP components. Auxiliary power consumption is around 11–16% of the power
produced. Fig. 8.4a elaborates on power consumption in the air supply. Being a
major consumer in the system, power consumption of air blower rises with
increase in current and this incremental profile can be associated to the increased
mass flow of air required to maintain oxidant stoichiometry in the fuel cell. After
the initial rise, power consumption of air blower becomes constant; however the
consumption graph of air radiator takes more time to become stable (Fig. 8.4b).
This is due to the fact that flow in air radiator is manipulated to control the
temperature and is a slow process. Together, air blower and radiator consume 10%
to that of stack power, whereas a coolant and recirculation pumps account for 1.0–
1.3% when operating at 60A.

89
Transitory load changes –System Analysis 8

     
-120.0-80.0 -40.0 0.0 40.0 80.0 120.0
     

9.0
0.5 0.55 0.6 0.65 0.7 0.75

3.0 4.5 6.0 7.5

Auxiliary Power (kW)
System efficiency

Stack Power (kW)

PEMFC.Icell A

1.5
0.0
750.0 875.0 1000.0 750.0 875.0 1000.0
Time (Seconds) Time (Seconds)

Figure 8.3: (a) System efficiency variation with load change, (b) stack power output and
auxiliary power consumption.
100.0

           
0.2 0.4 0.6 0.8 1.0 1.2 1.4

0.4
Recirculation pump (kW)
50.0

Air Radiator (kW)

Air Blower (kW)
Current (A)

0.2
0.0 -50.0
-100.0

750.0 875.0 1000.0 750.0 875.0 1000.0

Time (Seconds) Time (Seconds)

Figure 8.4: Auxiliary power consumption versus current; (a) Air blower power, (b) Air
radiator and recirculation consumption.

Whereas, at higher currents of 100 A, power consumed by air blower and radiator
constitutes 15% of the total stack power. For a swift temperature control and a
thermally stable operation, air radiator requires more power or a larger heat
exchanger area which would further reduce efficiency during these load variations.
Power consumption of coolant pumps is not shown in the above figure because
increase in internal and external coolant flows affects the auxiliary power
consumption to a very small extent as compared to air compressor and radiator.
Anode recirculation pump also consumes very low power; however the peaks at
current surge are a noticeable detail, which is addressed later in the results.

90
8 Transitory load changes –System Analysis

8.2.4 Water management during load changes

In PEMFCs, water management is a fundamental issue since the performance of

fuel cell is strongly influenced by its internal water distribution. Figure 8.5a shows
net water diffusion in the cell when current is changed from 60A to 100A and
back. In the figure, positive values for water crossover designate transportation of
water from cathode to anode of the fuel cell and vice versa. With increasing
currents, more water is produced in the cathode which supports back–diffusion
towards the anode until the system reaches back to steady–state and there is almost
no net water crossover. Some negative peaks are also observed at the start of
current change, which are due to electro–osmotic drag. As more current is drawn,
there is a rapid increase in hydrogen ion flux towards the cathode, thereby
supporting water crossover through electro–osmotic drag. Back–diffusion rate is
increased with the production of water on the cathode side and subsequently water
flux is directed towards fuel cell anode. The reverse can said to be true when
current is ramped down from 100 A to 60A.

Effect of water distribution can also be observed in anode inlet and outlet relative
humidity. Whereas for the cathode, since it is assumed that air enters at a constant
relative humidity of 95%, the outlet humidity is always above 100%. Therefore,
results for only anode are discussed here.
150.0

-80.0 -40.0 0.0 40.0 80.0 120.0

           
0.92 0.96 1.0 1.04 1.08
-0.1 0.0 0.1 0.2 0.3 0.4
Net Water crossover (kg/hr)

Anode outlet RH (%/%)

Anode inlet RH (%/%)
50.0 100.0

Currentl (A)
Current (A)
0.0
-50.0

750.0 800.0 850.0 900.0 950.0 1000.0 750.0 800.0 850.0 900.0 950.0 1000.0
Time (Seconds) Time (Seconds)

Figure 8.5: Current ramp–up from 60–100A and vice versa: (a) Water crossover through
PEMFC membrane, (b) Effect of water transportation on anode RH.

91
Transitory load changes –System Analysis 8

From Fig. 8.5b, it could be further observed that relative humidity at anode outlet
decreases with a current surge, though an abrupt increase is detected at the start of
this change. Sudden ramping of the current consumes more hydrogen, leaving
higher molar fraction and partial pressure of water in the anode, thus the peak of
high relative humidity. With the rise in stack temperature, water activity on anode
side is reduced and water diffusion from cathode increases and consequently
stabilizes to almost zero net–water crossover with the control of operating
temperature. Factors contributing to the water crossover in the membrane are
elaborated in Fig. 8.6. The membrane water content, which is the measure of
amount of water per sulfonic group in the membrane, is a function of water
activity in the anode and cathode of the fuel cell. Since, the air entering the
cathode is humidified and contains additional water produced during cell
reactions; it is close to the maximum. Here, the average water activity at cathode is
not exactly 1, as the humidity levels at the inlet are around 95%. On the other
hand, water activity at the anode reduces with the increase in current. This raises
the gradient for back diffusion from the cathode towards the fuel cell anode. Fig.
8.6b depicts the average membrane content and back diffusion coefficient for
water transportation across the membrane. The diffusion coefficient is a function
of temperature as well as water content (see Eq. 5.39 and 5.40). As the membrane
water content is linearly stable, the diffusion profile inclines more towards the cell
temperature variations. The transportation phenomena in the fuel cell become
fairly complex with co–dependence of many parameters on temperature and
hydration of the membrane.
13.0
0.95 0.96 0.97 0.98 0.99 1.0

3.5
Diff_Coefficient cm2/s x 10
Cathode H2O acitivty
Anode H2O activity

PEM H2O content

3.25
12.0

3.0
11.0

2.75
10.0

750.0 800.0 850.0 900.0 950.0 750.0 800.0 850.0 900.0 950.0
Time (Seconds) Time (Seconds)

Figure 8.6: Current ramp–up from 60–100A and vice versa: (a) Water activity in fuel cell,
(b) Back diffusion coefficient for water in membrane.

92
8 Transitory load changes –System Analysis

Water removal from a fuel cell cathode is also dependent on stack temperature
and pressure drop. Temperature is the dominant factor of the two, and the
amount of saturated water depends on the cathode outlet temperature. Inlet
humidity also contributes to liquid water saturation in the fuel cell, as reported in
Wong et al. [90]. Figure 8.7 provides information on the amount of liquid water at
outlet of the cathode channel for two different current surge amplitudes.
50.0 100.0 150.0

     
-2.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0

0.35
0.2 0.25 0.3
Heat generated (kW)

Liquid fraction
Current (A)
0.0

0.15
-100.0 -50.0

0.1

750.0 875.0 1000.0

Time (Seconds)
50.0 100.0 150.0

     
80.0

0.35
Operating temperature (C)

0.2 0.25 0.3

75.0

Liquid fraction
Current (A)

65.0 70.0
0.0

0.15
-100.0 -50.0

60.0

0.1
55.0

550.0 660.0 770.0 880.0

Time (Seconds)

Figure 8.7: Liquid water at cathode outlet; (a) Current ramp–up from 60A to 100A and
vice versa, (b) Current ramp–up from 60A to 120A and back.

It can be seen in Fig. 8.7a, that there is an abrupt increase in amount of liquid
water at cathode when the current is ramped from 60A to 100A. When higher

93
Transitory load changes –System Analysis 8

currents are drawn, reactions within the fuel cell are accelerated which demand
more intake of fuel and oxidant. Consequently these reactions produce more
water when compared to production at low currents. Since the stack temperature
does not elevate till that instance, higher percentage of produced water is saturated
at fuel cell cathode. Heat produced by cell reactions then elevates the stack
temperature, thereby reducing saturated water at the outlet. Moreover,
condensation of water at cell sites produces additional heat which rapidly increases
the stack temperature. It can be further noticed in Fig. 8.7a that heat produced in
the stack lowers as water liquid fractions drop. On the other hand, when current is
reduced back to 60A, a similar but opposite profile is observed and the amount of
liquid water tends to increase with a sink in stack temperature. Therefore, at low
temperatures and currents, water removal is a dominant factor and stoichiometries
are determined by the minimum flow rates required for water removal which in
the present case are more than adequate to provide the necessary concentrations.
Figure 8.7b shows water saturation results when ramping of current is set from
60A to 120A instead. Although the data profile is analogous to that of Fig. 9a, it is
noted that amplitude of these peaks is higher when compared. Apparently, the
amount of liquid water at cathode exit is same for both cases when the fuel cell is
operating at steady state. Heat produced by condensation requires additional flow
of coolant to maintain stack operating temperature, yet it does not affect the
system efficiency to a greater extent as liquid pumps do not consume that much
power. As there is no external humidification apparatus for fuel and recirculation
of anode exhaust aides in humidifying the anode, water crossover has a significant
impact on anode operations which are discussed below.

8.2.5 Stream properties at fuel cell ports

Figure 8.8 depicts the variations in reactant inlet flows when the load on fuel cell
is varied. It is interesting to notice the fluctuations in the anode inlet and outlet.
When the current is ramped up from 60A to 100A, more hydrogen is consumed
increasing the utilization factor at that instance. This can be seen in Fig. 8.8b
where mass flow at anode exhaust suddenly decreases. Changes in temperature can
also be noted at these current surges as the air stream is cooled by external cooling
circuit. Control of fuel stoichiometry takes some time to readjust to the desired
level. Although, anode outlet flow is disturbed by fuel stoichiometry, it is also
affected by the water content in anode outlet (70% mass fraction), which depends
on the net water crossover within the cell. Initial peaks at both load changes are
also attributed to the electro–osmotic drag which is a function of current density.
Since anode inlet is supported by recirculation from anode exhaust, the changes
within the stack influence it considerably. In Fig. 8.8a, fuel flow at anode inlet

94
8 Transitory load changes –System Analysis

follows a similar trend to that of anode exhaust, as well as the inlet temperature.
Rapid reduction in temperature is due to lower volume of recirculation which is at
around 69°C whereas dry hydrogen from the tank is at 25°C. Fluctuations in
water content at anode inlet disturb fuel flow controls and relative humidity
constantly; they remain within acceptable ranges however. Figure 8.8c shows the
fuel stoichiometry changes and power consumed by recirculation pump, which are
affected by the depletion of hydrogen at anode cell sites and water diffusion during
the current surges.
-80.0 -40.0 0.0 40.0 80.0 120.0

40.0 50.0 60.0 70.0 80.0 90.0 100.0

(a)

0.0 40.0 80.0 120.0

40.0 50.0 60.0 70.0 80.0 90.0 100.0

(b)
2.5

2.5
Mass flow (kg/hr)

Mass flow (kg/hr)

Temperature (C)

Temperature (C)
2.0

2.0
Current (A)

Current (A)
1.5

1.5
1.0

1.0
-80.0 -40.0

750.0 875.0 1000.0 750.0 875.0 1000.0

Time (Seconds) Time (Seconds)
0.0 40.0 80.0 120.0

0.0 40.0 80.0 120.0

(c) (d)
70.0

35.0
4.0

0.2
Recirculation pump power (kW)
Hydrogen stoichiometry ()

Mass flow (kg/hr)

0.15

65.0

30.0
Temperature (C)
Current (A)

Current (A)

60.0

25.0
0.1
2.0

0.05

55.0

20.0
-80.0 -40.0

-80.0 -40.0

50.0
0.0

750.0 875.0 1000.0 750.0 875.0 1000.0

Time (Seconds) Time (Seconds)

Figure 8.8: Current ramp–up from 60–100A and vice versa:, (a) Stream properties at
anode inlet and (b) outlet respectively, (c) Effect on fuel stoichiometry and recirculation
pump, (d) Air flow into the cathode.

On cathode side (Fig. 8.8d), as expected, the air flow which is regulated by a
controller increases when the current is ramped up and steadies along with the
system. There is no specific temperature control of inlet air in this model, though
external coolant flows manipulate the air temperature which is further preheated

95
Transitory load changes –System Analysis 8

by the humidifier before entering the stack. Air temperature varies around 60°C
which is equal to the controlled temperature for coolant entering the stack;
however temperature difference at anode inlet is significantly higher during these
load variations. Such temperature gradients incur adverse effects as they form
thermal stresses in the stack with co–flow configuration and reduce its life cycle.

8.3 Summary

Main contributions of this chapter are ascribed to the dynamic system responses to
the transitory load changes, which is characterized mainly by heat management
and water transportation within the fuel cell. It is observed that thermal
management strategy greatly influences voltage output and system efficiency which
increase with stack operating temperature. Power consumed by liquid coolant
pumps is minimal and have no considerable effect on system efficiency, whereas
air radiator consumes most of the power in thermal management system. A 5%
increase in power consumption of air blower and radiator is observed during load
variations for the cases presented here. Furthermore, water crossover in the fuel
cell has shown a significant impact on PEMFC anode operations. Anode inlet
flows, humidity and recirculation pump are influenced by net water diffusion
during load changes. Temperature changes at anode inlet are considerably higher
during load variations and have a negative impact as they generate thermal stresses
in the stack and reduce its lifetime. Also, amount of saturated water at cathode is
dependent on operating temperature which apparently pertains to thermal
management strategy of the system.

96
Chapter 9

9 Nitrogen crossover & anode purging

methodologies

9.1 Premise

In a typical PEMFC, hydrogen is fed into the anode of the fuel cell stack and
oxygen/air into the cathode, while the water produced from cell reactions is
ejected out of the cathode outlet. Since the current generated in the fuel cell is
directly attributed to the rate of electrochemical reactions i.e. the number of
electrons dispatched per mole of hydrogen at anode and the number of moles of
oxygen for reduction at cathode, power produced by the stack is associated to the
rate at which the fuel and oxidant are supplied. Ideally, all of the hydrogen and
oxygen needed for any specific power of the fuel cell stack would be consumed and
only water and unused nitrogen of the air would leave the system. However, not all
hydrogen and oxygen delivered at the reaction stoichiometry reaches the cell sites
which can cause fuel starvation, leading to irreversible damage to the stack in
addition to a lower power output [65]. Therefore, higher stoichiometries of
reactants are fed into the system. In most systems, the unused hydrogen exceeding
the fuel stoichiometry is recirculated back into the feed stream, which serves in
humidification of the fuel and also increases the system efficiency.

As already mentioned, one of the benefits of PEMFCs is their high power density
which is essential for portable applications. Recent advancements in
manufacturing have made these stacks more compact by significantly reducing the
thickness of MEA, though it enhances gas crossover across the membrane resulting
in current losses within the fuel cell itself. One of these gases is nitrogen from the
air, which permeates from the cathode to anode and gets concentrated in the
anode due to recirculation. Similarly, water produced on the cathode side is also
transported towards the anode through the membrane. This build–up of nitrogen
and water reduces the concentration of hydrogen in anode channels and obstructs

97
Nitrogen crossover & anode purging methodologies 9

hydrogen molecules to reach cell reaction sites effectively, which adversely affects
the voltage developed in the cell [44], [48] . Also, H2 starved areas are developed
within the fuel cell which can also result in carbon corrosion [91], [92].
Furthermore, additional flow of these crossed over gases increases the power
consumption of recirculation pump, thereby adding to the efficiency loss due to
the voltage drop. In most systems with anode recirculation, N2 build–up is usually
avoided by frequently purging a certain percentage of the anode outlet gas before
mixing it with the inlet stream. An alternative approach is to continuously bleed a
small fraction of the recirculation stream. While both mentioned approaches
release some portions of nitrogen as well as water vapour from the recirculated
stream, it also purges unused hydrogen. In this context, the amount purge fraction
and time intervals of these purging techniques are of particular interest to the
authors. An optimum strategy would cease the build–up of N2 with a compromise
of minimum H2 wastage. Although, there are a few publications dealing with
nitrogen crossovers and associated purge processes, but to the best of authors’
knowledge, there is no available study on intricate details of nitrogen crossovers
and the effect of purge cycles on the fuel cell performance during transient
operations.

This chapter covers the dynamic characteristics of the PEMFC system operating
with purge cycles, and investigate its performance during transitory load changes.
N2 accumulation in fuel cell anode and its effect on system performance is
investigated, and in view of this buildup, different purging strategies are assessed.
At last, an alternate method for nitrogen detection and automatic purging is
proposed and compared with conventional methods during different stages of
external load changes. This model could be regarded as a basic foundation in
design and optimization of purge cycles for automotive systems running PEMFCs.

9.2 Simulation results and discussion

Using the PEMFC stack data and control parameters summarized in chapter 4,
simulations for the system are carried out using Aspen Dynamics program.
Reliability of the fuel cell model and dynamic behaviour of the system under
consideration have already been reported in authors’ previous studies [93], [94]
and [95]. These studies have reported transient response of different components
of the system as well as their mutual dependencies on control strategies, thermal
management and water transportation of the fuel cell stack under different
operating scenarios; such as system start–up and during transitory load changes.
Here, simulation results for the effect of nitrogen crossover and associated purging

98
9 Nitrogen crossover & anode purging methodologies

of anode recirculation are presented. Stack temperature range of 60–70°C has

been used in simulations of the current system.

9.2.1 Gas Buildup at anode

Based on the permeability of N2 and partial pressure gradient across the

membrane, simulations for the prescribed system were conducted and
corroborated against validation data by stack manufacturer [62]. Fig. 9.1 presents
the simulated net nitrogen crossover in the fuel cell stack against the provided
experimental data. The corresponding relative error advocates a reasonable
agreement between the simulations and stack trial data.

Experimental Simulations Error

0,009 10
8
N2 crossover (kg/hr)

6
0,008
4

Error (%)
2
0,007 0
-2
-4
0,006
-6
-8
0,005 -10
40 60 80 100 120 140 160 180
Current (A)

Figure 9.1: Calculated nitrogen crossover rates against provided experimental data.

Current drawn from the fuel cell is set to a constant 120 A with no purging of the
anode recirculation loop. Since pure H2 is used as the fuel, there is no presence of
nitrogen in anode inlet of the fuel cell at the start–up. Fig. 9.2a shows the N2
crossover from the cathode to anode side of the fuel cell during the start–up
operations of the stack. It is observed that as the temperature and water uptake of
the membrane increases, N2 permeance increases, thereby increasing nitrogen
crossover from the cathode to the anode side of the cell. An abrupt decline in
temperature is noticed at around 120 s, which is due to the control initializations
of coolant flows to maintain the stack operating temperature. Details of the
control setup and thermal management of the system are reported in [93]. N2
permeance decreases with the sudden reduction in temperature and subsequently
reducing the crossover at that instant of time. However, as the system stabilizes to
steady state, the permeance becomes constant but N2 crossover tends to decrease as

99
Nitrogen crossover & anode purging methodologies 9

can be seen in Fig. 9.2b. This is due to the fact that N2 accumulation in the anode
reduces the concentration gradient across the cells.

0.008

10.0

80.0

16.0
N2 perm (kmol/s.cm.bar) exp-14

Operating temperature (C)

Membrane water content

N2 crossover (kg/hr)
0.006

60.0

14.0
7.5
0.004

40.0

12.0
5.0
0.002

20.0

10.0
2.5 0.0
0.0

0.0

8.0

0.0 100.0 200.0 300.0

Time (Seconds)
0.0065

80.0
4.0 5.0 6.0 7.0
N2 perm (kmol/s.cm.bar) exp-14

Operating temperature (C)

N2 crossover (kg/hr)
0.006

60.0
0.0055

40.0
2.0 3.0
0.005

20.0
1.0

400.0 500.0 600.0 700.0 800.0

Time (Seconds)

Figure 9.2: N2 permeance and crossover rate at (a) fuel cell start–up and (b) Steady–state
operations.

Depletion of hydrogen in the anode due to nitrogen accumulation leads to a slow

decay of the cell voltage. Fig. 9.3a represents the voltage decay due to increase in
N2 fractions and corresponding decrease in H2 concentrations in the fuel inlet.
Although voltage decay shows a linear relation with nitrogen accumulation, its
impact over extended operations results in hydrogen starved regions in the cell
resulting in increased cathode overpotential and resistive losses [60]. Another issue
related to anode recirculation is the accumulation of water, which clogs flow fields
and results in lower concentrations of H2 as well. Fig. 9.3b points to the gradual
increase in liquid water at anode outlet.

100
9 Nitrogen crossover & anode purging methodologies

0.065
0.73

80.0
1.0
Anode inlet N2 (kmol/kmol)
Anode inlet H2 (kmol/kmol)

Stack temperature (C)

Liquid water (kg/hr)
Cell voltage (V)

0.75

60.0
0.728

0.06
0.5

40.0
0.25
0.726

0.055
0.0

400.0 500.0 600.0 700.0 800.0 400.0 500.0 600.0 700.0 800.0
Time (Seconds) Time (Seconds)

Figure 9.3: (a) Cell voltage decay due to N2 buildup and (b) Increment in liquid water at
anode exhaust.

9.2.2 Anode outlet bleed

As shown above, N2 accumulation could lead to dilution of hydrogen in the fuel.

Moreover, power consumption of recirculation pump increases due to additional
flow of nitrogen and water vapour. One way of preventing accumulation of these
gases is venting of anode exhaust often called as bleeding. Figure 9.4a presents
different percentages of anode bleed and nitrogen accumulation at anode inlet.

0.045

Bleed percentage 1% 2%
0.040 3% 4%

0.035
N2 molar fraction

0.030

0.025

0.020

0.015

0.010
300 350 400 450 500 550 600
Time (s)

101
Nitrogen crossover & anode purging methodologies 9

Bleed (3%)

0.025

0.1 0.08
0.005 0.00: 0.015 0.02
Purged H2O (kg/hr)
Purged N2 (kg/hr)
Purged H2 (kg/hr)

Purge fraction
0.04 0.06
0.02
0.0

0.0 200.0 400.0 600.0

Time (Seconds)

Figure 9.4: (a) N2 levels for different bleed fractions at anode inlet and (b) Composition of
bleed stream from system start–up.

It can be seen that N2 keeps on increasing when the vent out fraction is 1% and
remains constant around 3% of the anode exhaust. At this bleed rate, molar N2
remains close to 1.5% of the fuel entering the stack. Fig. 9.4b shows the
constituent flow rates of 3 % vented stream. A large amount of water is also vented
along with nitrogen. A constant loss of hydrogen at a rate of 0.011 kg/hr (2% of
the supplied fuel) would suffice to prevent gas accumulation in anode of the
PEMFC.

9.2.3 Anode outlet purge

A purge of the anode recirculation is another method to reduce nitrogen and

water levels in the fuel inlet. A purge is initiated by opening of a solenoid valve
placed at the anode outlet. This valve opens for a short duration (purge time),
removing a portion of gases in the stream. The time between these purges is
referred to as purge interval. Though any combination of these could be assumed,
here arbitrary values of 0.6s, 0.3s, 0.9s and 60s of purge valve opening time,
opening duration, valve closing time and purge interval are selected respectively.
Here, a steady state analysis is conducted by drawing a current of 120 A and the
valve opening is set to purge 20% of the exhaust stream. Modelling of the valve
and its pressure effects have been neglected here as it falls out of the scope of
study. However, [53] have studied the behaviour of pressure during purge process
and reported a uniform distribution of pressures for short purge times as used in
current simulations. In addition, it is reported that fluctuations in anode pressure
due to frequent purging, also supports liquid water removal from the channels.

102
9 Nitrogen crossover & anode purging methodologies

0.731

0.95

0.08

9.64

0.51
0.9
Anode inlet H2 (kmol/kmol)

Anode inlet N2 (kmol/kmol)

Auxiliary power (kW)

9.58 9.6 9.62
Stack Power (kW)

System efficiency
Cell voltage (V)

0.06

0.89

0.5
0.73

0.9

0.04

0.88

0.49
0.729

0.85

0.02

0.87

0.48
9.56
0.728

9.54

0.86
0.8

500.0 600.0 700.0 800.0 500.0 600.0 700.0 800.0

Time (Seconds) Time (Seconds)

Figure 9.5: Performance of Fuel cell for fixed purge interval of 60 s; (a) Voltage rejuvenation
with semi–stable profile and (b) system efficiency and power during purge sequences.

Fig. 9.5a shows a semi–stable cell voltage during these purge events. As soon as the
purge is initiated, it removes both liquid water and nitrogen gas and recovers the
voltage drop caused by reduced active cell sites. N2 and H2 fractions could also be
noticed to change with each purge. During the purge intervals, there is a steady
buildup of nitrogen which gradually decays the cell voltage which lowers the
efficiency. This can be seen in Fig. 9.5b where efficiency of the system increases
with each purge and slowly lowers before the next purge event. It is also noticed
that there is an abrupt decline in the system efficiency at purge instances. This is
due to the fact that unused hydrogen leaving the purge valve reduces the recyclable
amount, thus these drop lines. Change in power consumption of auxiliary
components is minimal as only low–power recirculation pump is affected by the
purge. Since the total purge time is only 1.8 s and purge interval is 60 s, it is
cumbersome to determine the average efficiency lost for long durations of fuel cell
operations. Nonetheless, amount of H2 vented to the atmosphere could be
determined to assess the economics of the operations. Fig. 9.6 explicates that
around 0.08 kg/hr of H2 is wasted per purge event with interval of 60 s, which
accounts for 1.2% of the total fuel supplied to the system.

103
Nitrogen crossover & anode purging methodologies 9

0.15 0.2 0.25

-0.06 -0.03 0.0 0.03 0.06 0.09 0.12

0.8
Purged H2 (kmol/kmol)

Purged N2 (kmol/kmol)

H2 ventout (kmol/hr)
0.75

0.05 0.1
0.7 0.65

0.0

450.0 500.0 550.0 600.0 650.0

Time (Seconds)

Figure 9.6: Molar fractions at purge routines and H2 vent out rate.

9.2.4 Automatic Anode purge

As mentioned earlier that in addition to reducing the fuel cell voltage, enrichment
of nitrogen could lead to corrosion of the electrodes. With the ageing of the cell
membrane it tends to get thinner and due to the degradation, micro cracks and
pinhole regions are generated which further increase N2 crossover. Another
approach simulated in the present investigation is the use of an automatic control
for optimization of purge intervals. For practical diagnosis of fuel cell systems,
nitrogen at anode inlet could be detected by arc emission spectroscopy methods.
Signal from these detectors is transmitted to the controller, where it adjusts the
parameters for effective purges. An optimum purge sequence would prevent N2
buildup at a minimum expense of released hydrogen. Optimization of the purge
routine could be carried out by manipulation of many variables such as valve
opening and closing times, purge time, purge interval and purge stream fractions.
Simulations for different combinations of these variables for different power loads
would require tremendous amount of time and effort. For the sake of comparison
purposes, only purge intervals are chosen to be the manipulated variable in the
present simulations.

According to [62], molar N2 fraction of 5% or less has no or minimal effect on

stack life. A purge initiation limit of 3% has been set i.e. the purge event will occur
only when N2 molar concentration of 3% or more is detected at anode inlet.
Simulations for two different currents are carried out here. Voltage recovery and
efficiency profiles for automatic purging are found to be similar to that of fixed
time interval as the purge process is same. However, N2 crossover in the cell
changes with current density as reported by [44]. Typically N2 crossover increases

104
9 Nitrogen crossover & anode purging methodologies

with current density, as more heat is produced to raise the temperatures and
increase nitrogen permeance. Yet in a system such as the one under investigation,
system controls maintain the temperature of the stack and prevent excessive gas
crossovers. Also, stoichiometric ratios of anode determine the flow rate of supplied
H2 and are inversely proportional to the N2 concentration in the anode exit stream
[47].

-0.06 -0.03 0.0 0.03 0.06 0.09 0.12

0.05 0.1 0.15 0.2 0.25
0.8
Purged H2 (kmol/kmol)

Purged N2 (kmol/kmol)

H2 ventout (kmol/hr)
0.7 0.75
0.65

0.0

450.0 600.0 750.0

Time (Seconds)
0.15 0.2 0.25

-0.06 -0.03 0.0 0.03 0.06 0.09 0.12

0.8
Purged H2 (kmol/kmol)

Purged N2 (kmol/kmol)

H2 ventout (kmol/hr)
0.75

0.05 0.1
0.7 0.65

0.0

450.0 600.0 750.0

Time (Seconds)

Figure 9.7: Automatic purge intervals and corresponding molar fractions of purges stream;
(a) at drawn current of 120 A, and (b) 60 A.

Figure 9.7a shows N2 and H2 levels in anode recirculation after purging and
corresponding flow rate of hydrogen purged to the atmosphere. When the current
drawn during steady state operation is 120 A, H2 is purged at a rate of 0.08 kg/hr,
similar to the rate at fixed interval purge in Fig. 9.6. However, automatic purging
interval is found to be around 85 s, saving 15 s per purge. In Fig. 9.7b, when the

105
Nitrogen crossover & anode purging methodologies 9

current is drawn at 60 A, H2 purge rate is reduced to 0.06 kg/hr and purge interval
is found to be 55 s. This is due to the fact that H2 flow rate at a low current density
is relatively smaller compared to that at a high current density, so is the anode exit
flow rate. Therefore, low currents will result in a high N2 levels in the anode
exhaust which will require frequent purging. Fuel losses associated with these
purges are reduced to 0.8% of the total fuel being fed to the system. Based on the
data from table 3.1, higher stoichiometric ratios are used at low currents.

9.2.5 Anode purge at load changes

PEMFC systems have to operate at varying load and operating conditions when
used in automotive applications. Nitrogen concentration would vary with changing
loads which requires an efficient purging strategy to cater these dynamics. For this
particular simulation, current load is changed from 120A to 60A and vice versa at
380 s and 410 s respectively. Results for different purging methods are presented
in Fig. 9.8 and Fig. 9.9.
150.0

0.005 0.006 0.007 0.008 0.009 0.01

N2 crossover (kg/hr)
100.0
Current (A)
50.0 0.0
-50.0

250.0 300.0 350.0 400.0 450.0 500.0

Time (Seconds)

Figure 9.8: N2 crossover through membrane for fixed purge interval routine during load
changes from 120A to 60A and vice versa.

106
9 Nitrogen crossover & anode purging methodologies

0.08 0.16 0.24 0.32 0.4 0.48 0.56

0.8

0.1
0.4 0.48 0.56 0.64 0.72
Purged H2 (kmol/kmol)

Purged N2 (kmol/kmol)
H2 ventout (kmol/hr)
-0.05 0.0 0.05
0.32

-0.1
250.0 300.0 350.0 400.0 450.0 500.0
Time (Seconds)
0.12
0.85

0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95

Anode inlet H2 (kmol/kmol)
Anode inlet N (kmol/kmol)
0.1
0.8
Cell voltage (V)

0.08
0.75

0.06
0.7

0.04 0.02
0.65

250.0 300.0 350.0 400.0 450.0 500.0

Time (Seconds)

Figure 9.9: Effect of load changes of 120–60–120 A on fixed purge interval; (a) H2 purge
rate (b) N2 concentration at anode inlet and cell voltage.

Fig. 9.8 portrays nitrogen crossover during the load changes for a fixed purge
interval of 60 s. As the current is decreased, nitrogen flux towards the anode is
reduced. N2 concentration at anode exit also decreases as shown in Fig. 9.9a.
However, N2 levels at anode inlet tend to increase (Fig. 9.9b). As explained above,
this is attributed to the low residual flows of H2 at low current densities. On the
other hand, when the current is elevated to 120A at 410 s, sudden shortage of H2
is observed until the time taken by flow valves to adjust. Reduced stoichiometry of
hydrogen results in N2 concentration spike as well. Voltage change as a function of
current density is evident in the figure, though voltage revival during purges is not
that pronounced.

107
Nitrogen crossover & anode purging methodologies 9

0.008
120.0

N2 crossover (kg/hr)
0.007
80.0
Current (A)

0.006
40.0

0.005
0.0
-40.0

250.0 300.0 350.0 400.0 450.0 500.0

Time (Seconds)

Figure 9.10: N2 crossover through membrane for automatic purge interval sequence during
load changes from 120 A to 60 A and vice versa.

Effects of load changes to the nitrogen crossover and system performance during
automatic purge are reported in Fig. 9.10. Nitrogen crossover shows similar trend
when compared to fixed time interval. When the current is lowered to 60 A, low
residual H2 increases nitrogen levels at anode exit. As the molar N2 concentration
reaches 3%, automatic purge control activates the purge process. Here again, only
purge intervals are varied for automatic purging.
0.0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
0.8

0.1
Purged H2 (kmol/kmol)

Purged N2 (kmol/kmol)

H2 ventout (kmol/hr)
0.75

0.05
0.7

0.0
0.65

-0.05
0.6

250.0 300.0 350.0 400.0 450.0 500.0

Time (Seconds)

108
9 Nitrogen crossover & anode purging methodologies

0.08

0.65 0.7 0.75 0.8 0.85 0.9 0.95

0.9

Anode inlet N2 (kmol/kmol)

Anode inlet H2 (kmol/kmol)

0.85
Cell voltage (V)

0.06
0.75 0.8

0.04
0.7

0.02
0.65

250.0 300.0 350.0 400.0 450.0 500.0

Time (Seconds)

Figure 9.11: Effect of load changes of 120–60–120 A on automatic purge interval; (a) H2
purge rate (b) N2 concentration at anode inlet and cell voltage.

It can be seen in Fig. 9.11a that several instances of purge occur between 380410
s. Purge rate of H2 varies with the unused fuel at the anode exit. Unlike fixed time
interval, N2 concentrations at anode inlet are retained to lower levels for automatic
purge controls (Fig. 9.11b).

Based on the results discussed above, the applicability of nitrogen detectors and
automatic purge shows promise as a diagnostic tool for prediction of gas crossovers
and act as a building block for devising purge strategies depending upon the load
and mode of application.

9.3 Summary

A study on dynamics of nitrogen crossovers in PEMFC system with recirculation

was conducted and its effects on purging strategy were discussed in chapter. The
results exhibit that with pure recirculation, voltage and system efficiency declines
due to nitrogen accumulation in fuel cell. Different purging methodologies were
simulated to address hydrogen dilution issue at reaction sites. Anode bleed out of
3% is found to be the limit for prevention of N2 buildup and retains the
concentration levels to less than 1%. Also, cell voltage degrades linearly with N2
buildup and rejuvenates at purge sequences. An alternate strategy for automatic
initiation of anode recirculation purge by employing nitrogen detectors was
simulated. Using the same purge time for various cases, it is shown that purge
interval is lowered for low currents mainly due to low H2 residual flows rates.

109
Nitrogen crossover & anode purging methodologies 9

Moreover, during transient load changes, automatic purge catered well to prevent
nitrogen levels from rising when compared to a fixed purge interval strategy.

110
Chapter 10

10 Concluding remarks

10.1 General summary

The fuel cell technology is one of the potential candidates for future
transportation systems. In order to do so, fuel cells have to be developed to the
extent that they are a viable alternative to current ICE vehicles in the near future.
In addition to be cost effective, FCVs have to ensure reliability and durability
when running under inflexible operating conditions. The need for transient
analysis of such systems becomes of prime importance in order to make fuel cells a
practical option for automobiles in the near future.

In the above context, the aim of this research was to investigate the dynamic
behaviour of a PEMFC system in order to improve system efficiency when it is
operating at different loads and operating conditions. A successful operation of
fuel cell would require proper handling of four main sub–systems; fuel and
oxidant supply, thermal and water management, and power conditioning modules.

These sub–systems need well–suited operating conditions which are controlled by

the auxiliary components and hence define the overall system performance and
effect the lifetime of the fuel cell stack. Current research was focused towards
addressing these issues, firstly by developing a PEMFC stack model and then
construction of the entire system with integrated controls to emulate a fuel cell
system operating under varying operating conditions and influence of external
parameters. System analyses were conducted for start–up sequence and transitory
load changes. Issues related to water and thermal management were investigated
with emphasis on species crossover mechanisms across the fuel cell. Finally, study
on different purging strategies was conducted in order to cater to water
management of the system.

111
Concluding remarks 10

10.2 Summary of analysis

10.2.1 Modelling of the PEM fuel cell system

The fuel cell model developed accommodates equations with electrochemical,

thermal, feed flow and species crossover models in addition to two–phase
calculations at fuel cell electrodes. Power output is calculated based on individual
voltage and current models with time varying external loads, reactant pressures and
stoichiometries, stack operating temperatures and reactant humidity levels. For
fuel cell voltage, the polarization curve was created on the basis of Nernst voltage
with activation, ohmic and concentration losses. Mass and energy conservation
equations with electrochemical relations were used to calculate flow rates,
concentrations and partial pressures of species in the reactant and product
streams. In order to shed light on the complex interplay of different transport
phenomena, detailed water and nitrogen crossover models were incorporated in
the fuel cell stack model. A simple humidifier model comprising of heat and mass
balance was built to achieve 95% humidity levels at cathode inlet of the fuel cell.
The fuel was humidified by recirculation of anode exhaust.

The prime function of the PEMFC stack is to provide stable and uninterrupted
power supply over a range of operating and environmental conditions. However,
fuel cell functionality is constrained within specific limitations pertaining to
operational integrity and stack lifetime. Optimum fuel cell performance depends
on a number of operating conditions including current density, reactant
stoichiometry, relative humidity, inlet pressures, and cell temperature. Each of
these conditions is at the behest of proper interaction between the controlling
auxiliary component and the fuel cell stack. Here, a control system was designed
and demonstrated with the ability to predict the dynamic requirements of PEM
fuel cell system during simultaneous changes of load and operating conditions.

Simulations for the prescribed system were carried out and reliability of the
suggested model was verified and corroborated against design validation data,
which were found to be in excellent agreement. Polarization curves, fuel
consumption and power output profiles were found to be in accordance with the
validation. Heat produced by the stack was also investigated by exergetic analysis of
the PEMFC stack and associated irreversibilities were identified. From the
preliminary results, a general conclusion could be drawn concerning the fuel cell
exergetic efficiency, which depends on reactant pressures, operating temperatures,
polarization overpotentials, current density and reactant stoichiometries. It was
deduced that stack start–up at high current densities reduces exergetic efficiency in

112
10 Concluding remarks

the initial instances, while the efficiency of the fuel cell can be increased through
increasing the fuel cell operating temperature and voltages. Moreover, entropy
generation is reduced at high operating temperatures and with low presence of
liquid water in the cathode outlet. Also, low ambient temperatures increase the
exergetic efficiency slightly when operating at steady state temperatures. However,
exergy destruction is higher at low start–up temperatures.

10.2.2 Start–up analysis of fuel cell system

In Proton Exchange Membrane Fuel Cells (PEMFC) systems, dynamic behaviour

of a start–up is of particular importance to ensure a short start–up time and an
efficient operation. A start–up analysis was performed for the described PEMFC
system, starting at ambient temperature of 25°C and the focus had been set to
evaluate the effects of key parameters till the system reaches steady–state operating
temperatures. Key contributions of the suggested model are attributed to system
response methodology, which incorporates stack thermal behaviour in addition to
fuel cell electrochemistry, water crossover, mass and energy balances. It was
observed that system efficiency and voltage output are higher at low power start–
ups but for the fuel cell stack it takes longer time to reach stable operating
conditions. In the context of fuel cell durability, quick start–ups to reach stable
operating temperatures would allow high performance of the fuel cell and limit the
operation time in the low performance states. Higher current density start–up
would reduce the warming up times for the stack, though the load management
during start–up would have to be properly managed. Another issue related to high
currents is the water saturation at cathode, which is considerably higher at high
current density start–ups. Removal of liquid water from cathode exhaust thus
requires additional attention at low stack operating temperatures. Furthermore, air
radiator consumes more power at high currents once the system is stable and due
to this fact, system efficiency is reduced. In the proposed system, only air is
humidified while the fuel is supposed to be adequately moistened by water
crossover through the cell and recycling of humidified exhaust. Simulations were
carried out to assess the extent of humidification in anode by assuming different
humidity levels of inlet air. It was determined that cathode inlet water levels are
adequate enough to humidify fuel stream. Though it is undesirable to have liquid
droplets in the recirculating stream, their presence tends to increase the relative
humidity at a faster pace. Small amounts of liquid water in anode would reduce
the dry operational conditions, but negatively affect the recirculation pumps.

113
Concluding remarks 10

10.2.3 System response during transitory load changes

As a power source for automotive applications, fuel cell systems usually operate at
continuously varying external requirements related to temperatures, pressures,
power load and humidity. PEMFC dynamics are influenced by reactant flows, heat
management and water content in the streams as well as within the fuel cell itself.
The power output from a fuel cell module is a combination of battery and the fuel
cell stack. Changes to the external load are handled by the battery with fuel cell
charging it until it reaches minimum set charge state from where the fuel cell is
responsible for providing power to the external load and for charging the battery as
well. Regardless of the size of the battery and load handling management of the
system, only load changes to the fuel cell stack were simulated here. Main
contributions are ascribed to the dynamic system responses to the transitory load
changes, which is characterized mainly by heat management and water
transportation within the fuel cell. It was observed that the thermal management
strategy greatly influences voltage output and system efficiency which increase with
stack operating temperature. Moreover, slow temperature controls affect the
stability of fuel cell operations. The changes in external load were few and far, only
to study the thermal controls. With the changes occurring more than a minute
apart, temperature changes stabilization was approximated to be within 30
seconds. High coolant mass flow rates were initiated in order to contemplate fast
electrochemical changes in the stack. With very high variations in coolant flow
rates, power consumed by liquid coolant pumps is minimal and had no
considerable effect on system efficiency. On the other hand, air radiator consumes
more during load changes. A 5% increase in power consumption of air blower and
radiator is observed during load variations for the cases presented here.
Furthermore, water crossover in the fuel cell has shown a significant impact on
PEMFC anode operations. Anode inlet flows, humidity and recirculation pump
are influenced by net water diffusion during load changes. Temperature changes at
anode inlet are considerably higher during load variations and have a negative
impact as they generate thermal stresses in the stack and reduce its lifetime. Also,
amount of saturated water at cathode is dependent on operating temperature
which apparently pertains to thermal management strategy of the system.

10.2.4 Effect of nitrogen crossover on purging strategy

Recent advancements in manufacturing of PEMFC stacks have made them more

compact by significantly reducing the thickness of MEA. On the contrary, thin
membranes enhances gas crossover across the cell resulting in current losses within
the fuel cell itself. Nitrogen and water produced on the cathode side are both

114
10 Concluding remarks

transported towards the anode through the membrane and reduce the
concentration of hydrogen in anode channels and obstructs hydrogen molecules to
reach cell reaction sites effectively, which adversely affects the voltage developed in
the cell. Here, model for nitrogen crossover and its build–up anode was developed
and its effects on fuel cell performance and anode recirculation were simulated,
and associated purging strategies were discussed. Nitrogen crossover model was
found to be in excellent agreement with the experimental data provided by the
stack manufacturer. The results exhibited that with pure recirculation, voltage and
system efficiency declined due to nitrogen accumulation in fuel cell. Purging of
anode recirculation is one of the approaches to relieve the stream off nitrogen and
unwanted water. However, unutilized fuel is also wasted with each purge. Different
purging methodologies were simulated to address hydrogen wastage during
purging and dilution in the fuel cell. For continuous bleeding of anode exhaust, a
bleed–out of 3% was found to be the limit for prevention of N2 buildup.
Concentration levels of nitrogen were also retained less than 1%. Apparently, cell
voltage degraded linearly with N2 buildup and rejuvenated at purge sequences.
These results are limited to partial fuel starvation only and do not take into
account the dilution of hydrogen in extreme conditions. An alternate strategy for
automatic initiation of anode recirculation purge by employing nitrogen detectors
was simulated. For practical diagnosis of fuel cell systems, nitrogen at anode inlet
could be detected and signal from these detectors is transmitted to the controller,
where it adjusts the parameters for effective purges. An optimum purge sequence
would prevent N2 buildup at a minimum expense of released hydrogen. Using the
same purge time for various cases, it is shown that purge interval is lowered for low
currents mainly due to low H2 residual flows rates. Moreover, during transient load
changes, automatic purge catered well to prevent nitrogen levels from rising when
compared to a fixed purge interval strategy. Build–up of nitrogen during fixed
purges is further increased if load changes are frequent. Whereas nitrogen
detection method proved to be feasible for predicting gas crossovers during purge
sequences at different loads and can be used as a base for optimizing and
development of anode purge strategies for PEMFC systems.

10.3 Improvements for future work

This work has treated some of the technical aspects of modelling a PEM fuel cell
system and controlling it to predict its behaviour with time. Even though this
model can be useful in optimizing and designing operational strategies for
different load cycles such as system start–up and purging states in PEMFC systems,
several parameters need to be validated with experimental data from an actual fuel

115
Concluding remarks 10

cell test bench. Most of the focus was inclined towards fuel cell stack in the current
research; however an extensive model validation from experimental setup would
certainly help in verification of individual auxiliary components and make the
model more insightful. The humidifier model used in this study is fairly simple
and uses a fixed input of relative humidity and calculates amount of water needed
for humidification by mass balance while the heat transfer is determined by the
energy balance. The dynamic effects of the actual humidifier are therefore needed
to simulate the actual humidity levels and temperature of the air entering the fuel
cell stack.

Within all the areas covered in this dissertation, more knowledge is required in
order to fully understand the inner workings of the fuel cell stack. Many challenges
still exist in understanding the precise mechanisms of transportation in the fuel
cell. Complex diffusion phenomena, such as water and nitrogen crossover
occurring within the stack require further deliberation. A multi–dimensional
model would be beneficial in determining these parameters in addition to several
spatially distributed variables. Pressure drops within the flow fields and
temperature distribution in the fuel cell is not uniform, which vary the current
density along the cell surface. Since, current density is a key parameter in defining
the performance of the fuel cell, information on its distribution would help in
identifying fuel starvation problems in more detail and predict transient behaviour
closer to the actual fuel cell.

Several controls were implemented in the system under investigation, which could
emulate the system response to a favourable extent. However, only PID controllers
with simple feedback loop were mostly incorporated with no tuning or
optimization conducted in this research. In conjunction with experimental
evaluation, advanced control strategies will further enhance the ability of this
system. Moreover, an extended exergy analysis of all the underlying components is
expected to help in identifying inefficient components during steady–state as well
as transient conditions, and contribute in a more robust control strategy of the
system.

Out of the several interesting plans, it is intended to extend the model with the
addition of battery and power conditioning component models. Currently, the
system efficiencies are calculated to be optimistically higher, because losses in the
external terminals and power conditioning components such as the inverter have
not been accounted for. Instead, a fixed value is assumed. A battery model in
connection with the fuel cell stack would be able to simulate load distribution on
these components and assist in studying the load management of the system.
Similarly, for detailed modelling of the purge process, there is need for a dynamic

116
10 Concluding remarks

model of purge valve. Finally, there is always a room for improvement in research
and the author considers it more than anyone as a motivation for further work.

117
Concluding remarks 10

118
 Bibliography

1 Renewable Energy Sources and Climate Change Mitigation. ; 2012.

2 Key World Energy Statistics. Paris: ; 2012.

3 Aspentech. Aspen plus dynamics.

2012.http://www.aspentech.com/products/aspen–dynamics.aspx

4 Hwang J–J, Kuo J–K, Wu W, Chang W–R, Lin C–H, Wang S–E. Lifecycle
performance assessment of fuel cell/battery electric vehicles. International
Journal of Hydrogen Energy 2013; 38:3433–3446.

5 Honda’s FCX fuel cell concept vehicle. AutoTechnology 2007; 7.

6 Friedlmeier G, Friedrich J, Panik F. Test Experiences with the

DaimlerChrysler Fuel Cell Electric Vehicle NECAR 4. Fuel Cells 2001;
1:92–96.

7 Thomas CE. Fuel cell and battery electric vehicles compared. International
Journal of Hydrogen Energy 2009; 34:6005–6020.

8 Houchins C, Kleen GJ, Peterson D, Garland NL, Ho DL, Marcinkoski J, et

al. U.S. doe progress towards developing low–cost, high performance,
durable polymer electrolyte membranes for fuel cell applications. Membranes
2012; 2:855–878.

9 Rodrigues A, Amphlett JC, Mann RF, Peppley BA, Roberge PR. Carbon
monoxide poisoning of proton–exchange membrane fuel cells. IECEC–97 –

119
Bibliography

proceedings of the thirty–second intersociety energy conversion engineering conference,

vols 1–4 1997; :768–773.

10 Baschuk, Li X. Carbon monoxide poisoning of proton exchange membrane

fuel cells. International Journal of Energy Research 2001; 25:695–713.

11 Ballard Power Systems. 2013.http://www.ballard.com/about–ballard/fuel–

cell–education–resources/how–a–fuel–cell–works.aspx

12 Gelfi, Stefanopoulou, Pukrushpan, Peng H. Dynamics of low–pressure and

high–pressure fuel cell air supply systems. 2003.

13 McCain BABA, Stefanopoulou AGAG, Kolmanovsky IVI V. On the

dynamics and control of through–plane water distributions in PEM fuel
cells. Chemical Engineering Science 2008; 63:4418–4432.

14 Ceraolo M, Miulli C, Pozio A. Modelling static and dynamic behaviour of

proton exchange membrane fuel cells on the basis of electro–chemical
description. Journal of Power Sources 2003; 113:131–144.

15 Amphlett JC, Mann RF, Peppley BA, Roberge PR, Rodrigues A. A model
predicting transient responses of proton exchange membrane fuel cells.
Journal of Power Sources 1996; 61:183–188.

16 Yerramalla S, Davari A, Feliachi A. Dynamic modeling and analysis of

polymer electrolyte fuel cell. 2002 IEEE Power Engineering Society Summer
Meeting, Vols 1–3, Conference Proceedings 2002; :82–86.

17 Pathapati PR, Xue X, Tang J. A new dynamic model for predicting transient
phenomena in a PEM fuel cell system. Renewable Energy 2005; 30:1–22.

18 Jia J, Wang Y, Li Q, Cham YT, Han M. Modeling and Dynamic

Characteristic Simulation of a Proton Exchange Membrane Fuel Cell. IEEE
Transactions on Energy Conversion 2009; 24:283–291.

19 Hu G, Fan J, Chen S, Liu Y, Cen K. Three–dimensional numerical analysis

of proton exchange membrane fuel cells (PEMFCs) with conventional and
interdigitated flow fields. Journal of Power Sources 2004; 136:1–9.

20 Kim K, Wang M, Von Spakovsky MR, Nelson DJ. Dynamic synthesis/design

and operation/control optimization under uncertainty of a PEMFC system.

120
 Bibliography

ASME International Mechanical Engineering Congress and Exposition, Proceedings

2009; 8:679–689.

21 Li Y, Xu S, Yang Z, Li Y. Experiment and simulation study on cold start of

automotive PEMFC. International Conference on Electric Information & Control
Engineering ICEICE, Wuhan 2011; :2166–2170.

22 Yan Q, Toghiani H, Lee Y–W, Liang K, Causey H. Effect of sub–freezing

temperatures on a PEM fuel cell performance, startup and fuel cell
components. Journal of Power Sources 2006; 160:1242–1250.

23 Mangold M, Piewek S, Klein O, Kienle A. A model for the freeze start

behavior of a pem fuel cell stack. Journal of Fuel Cell Science and Technology
2011; 8.

24 Meng H. A PEM fuel cell model for cold–start simulations. Journal of Power
Sources 2008; 178:141–150.

25 Sundaresan M, Moore RM. PEM Fuel Cell Stack Cold Start Thermal
Model. Fuel Cells 2005; 5:476–485.

26 Vasu G, Tangirala AK. Control–orientated thermal model for proton–

exchange membrane fuel cell systems. Journal of Power Sources 2008; 183:98–
108.

27 Khan MJ, Iqbal MT. Modelling and analysis of electrochemical, thermal,

and recetant flow dynamics for a PEM fuel cell system. Fuel Cells 2005;
5:463–475.

28 Shan Y, Choe S–Y. Modeling and simulation of a PEM fuel cell stack
considering temperature effects. Journal of Power Sources 2006; 158:274–286.

29 Del Real AJ, Arce A, Bordons C. Development and experimental validation

of a PEM fuel cell dynamic model. Journal of Power Sources 2007; 173:310–
324.

30 Ahn J–W, Choe S–Y. Coolant controls of a PEM fuel cell system. Journal of
Power Sources 2008; 179:252–264.

31 Jung J–H, Ahmed S. Dynamic model of PEM fuel cell using real–time
simulation techniques. Journal of Power Electronics 2010; 10:739–748.

121
Bibliography

32 Asghari S, Akhgar H, Imani BF. Design of thermal management subsystem

for a 5kW polymer electrolyte membrane fuel cell system. Journal of Power
Sources 2011; 196:3141–3148.

33 Beicha A. Modeling and simulation of proton exchange membrane fuel cell

systems. Journal of Power Sources 2012; 205:335–339.

34 Park S–K, Choe S–Y. Dynamic modeling and analysis of a 20–cell PEM fuel
cell stack considering temperature and two–phase effects. Journal of Power
Sources 2008; 179:660–672.

35 Blomen LJMJ. Review of an energy and exergy analysis of a fuel cell system.
Journal of Power Sources 1993; 41:239–252.

36 Saidi MH, Ehyaei MA, Abbasi A. Optimization of a combined heat and

power PEFC by exergy analysis. Journal of Power Sources 2005; 143:179–184.

37 Barbir F, Gómez T. Efficiency and economics of proton exchange

membrane (PEM) fuel cells. International Journal of Hydrogen Energy 1997;
22:1027–1037.

38 Kazim A. Exergy analysis of a PEM fuel cell at variable operating conditions.

Energy Conversion and Management 2004; 45:1949–1961.

39 Kazim. Exergoeconomic analysis of a PEM electrolyser at various operating

temperatures and pressures. International Journal of Energy Research 2005;
29:539–548.

40 Mert SO, Dincer I, Ozcelik Z. Exergoeconomic analysis of a vehicular PEM

fuel cell system. Journal of Power Sources 2007; 165:244–252.

41 Miansari M, Sedighi K, Amidpour M, Alizadeh E, Miansari M. Experimental

and thermodynamic approach on proton exchange membrane fuel cell
performance. Journal of Power Sources 2009; 190:356–361.

42 Leo TJ, Durango JA, Navarro E. Exergy analysis of PEM fuel cells for marine
applications. Energy 2010; 35:1164–1171.

43 Uyanga E, Sevjidsuren G, Bumaa B, Sangaa D, Altantsog P. Exergy analysis

of NEXATM PEM fuel cell module. 2011; 1:256–260.

122
 Bibliography

44 Kocha SS, Deliang Yang J, Yi JS. Characterization of gas crossover and its
implications in PEM fuel cells. AIChE Journal 2006; 52:1916–1925.

45 Ahluwalia RK, Wang X. Buildup of nitrogen in direct hydrogen polymer–

electrolyte fuel cell stacks. Journal of Power Sources 2007; 171:63–71.

46 Catalano J, Myezwa T, De Angelis MG, Baschetti MG, Sarti GC. The effect
of relative humidity on the gas permeability and swelling in PFSI
membranes. International Journal of Hydrogen Energy 2012; 37:6308–6316.

47 Baik KD, Kim MS. Characterization of nitrogen gas crossover through the
membrane in proton–exchange membrane fuel cells. International Journal of
Hydrogen Energy 2011; 36:732–739.

48 Weber. Gas–crossover and membrane–pinhole effects in polymer–

electrolyte fuel cells. Journal of the Electrochemical Society 2008; 155:B521–31.

49 Pukrushpan JT, Peng H, Stefanopoulou AG. Control–Oriented Modeling

and Analysis for Automotive Fuel Cel Systems. Journal of Dynamic Systems
Measurement & Control 2004; 126.

50 Zhu WH, Payne RU, Tatarchuk BJ. Critical flow rate of anode fuel exhaust
in a PEM fuel cell system. Journal of Power Sources 2006; 156:512–519.

51 Tang Y, Yuan W, Pan M, Li Z, Chen G, Li Y. Experimental investigation of

dynamic performance and transient responses of a kW–class PEM fuel cell
stack under various load changes. Applied Energy 2010; 87:1410–1417.

52 Zhai S, Zhou S, Sun P, Chen F, Niu J. Modeling study of anode water

flooding and gas purge for PEMFCs. Journal of Fuel Cell Science and Technology
2012; 9.

53 Gou J, Pei P, Wang Y. The dynamic behavior of pressure during purge

process in the anode of a PEM fuel cell. Journal of Power Sources 2007; 162.

54 Wang X, Tajiri K, Ahluwalia RK. Water transport during startup and

shutdown of polymer electrolyte fuel cell stacks. Journal of Power Sources
2010; 195:6680–6687.

55 Karimäki H, Pérez LC, Nikiforow K, Keränen TM, Viitakangas J, Ihonen J.

The use of on–line hydrogen sensor for studying inert gas effects and

123
Bibliography

nitrogen crossover in PEMFC system. International Journal of Hydrogen Energy

2011; 36:10179–10187.

56 Promislow K, St–Pierre J, Wetton B. A simple, analytic model of polymer

electrolyte membrane fuel cell anode recirculation at operating power
including nitrogen crossover. Journal of Power Sources 2011; 196:10050–
10056.

57 Hou Y, Shen C, Yang Z, He Y. A dynamic voltage model of a fuel cell stack

considering the effects of hydrogen purge operation. Renewable Energy 2012;
44:246–251.

58 Müller EA, Kolb F, Guzzella L, Stefanopoulou AG, McKay DA. Correlating

nitrogen accumulation with temporal fuel cell performance. Journal of Fuel
Cell Science and Technology 2010; 7:210131–2101311.

59 Siegel JB, Bohac S V, Stefanopoulou AG, Yesilyurt S. Nitrogen Front

Evolution in Purged Polymer Electrolyte Membrane Fuel Cell with Dead–
Ended Anode. Journal of the Electrochemical Society 2010; 157:B1081–93.

60 Yesilyurt S, Siegel JB, Stefanopoulou AG. Modeling and Experiments of

Voltage Transients of Polymer Electrolyte Membrane Fuel Cells With the
Dead–Ended Anode. Journal of Fuel Cell Science and Technology 2012; 9.

61 Choi JW, Hwang Y–S, Seo J–H, Lee DH, Cha SW, Kim MS. An
experimental study on the purge characteristics of the cathodic dead–end
mode PEMFC for the submarine or aerospace applications and performance
improvement with the pulsation effects. International Journal of Hydrogen
Energy 2010; 35:3698–3711.

62 Ballard Mark9 SSL. Product manual and integration guide. 2008.

63 Eikerling Research Group, SFU.

2013.http://www.sfu.ca/eikerlingresearch/overview/

64 Scientific Computing World. 2013.http://www.scientific–computing.com

/features/feature.php?feature_id=126

65 Hosseinzadeh E, Rokni M. Development and validation of a simple

analytical model of the Proton Exchange Membrane Fuel Cell (PEMFC) in a

124
 Bibliography

fork–lift truck power system. International journal of green energy 2013;

10:523–543.

66 Larminie J, Dicks A. Fuel cell systems explained. Chichester, West Sussex: J.

Wiley; 2003.

67 Spiegel C. Designing and building fuel cells. New York: McGraw–Hill; 2007.

68 Liu H, Laconti AB, Gasteiger HA, Zhang J. Factors impacting chemical

degradation of perfluorinated sulfonic acid ionomers. ECS Transactions
2005; 1:283–293.

69 Santarelli MG, Torchio MF, Cochis P. Parameters estimation of a PEM fuel

cell polarization curve and analysis of their behavior with temperature.
Journal of Power Sources 2006; 159:824–835.

70 Bockris JO. Handbook of fuel cell technology – Editor Carl Berger.

Prentice–Hall, 1968; $318.50. Electrochimica Acta 1969; 14:925–926.

71 Santarelli MG, Torchio MF. Experimental analysis of the effects of the

operating variables on the performance of a single PEMFC. Energy Conversion
and Management 2007; 48.

72 Mann RF, Amphlett JC, Hooper MAI, Jensen HM, Peppley BA, Roberge
PR. Development and application of a generalised steady–state
electrochemical model for a PEM fuel cell. Journal of Power Sources 2000;
86:173–180.

73 Zhang L, Pan M, Quan S. Model predictive control of water management in

PEMFC. Journal of Power Sources 2008; 180:322–329.

74 Spinelli P, Francia C, Ambrosio EP, Lucariello M. Semi–empirical

evaluation of PEMFC electro–catalytic activity. Journal of Power Sources 2008;
178:517–524.

75 Haji S. Analytical modeling of PEM fuel cell i–V curve. Renewable Energy
2011; 36:451–458.

76 Wang X, Nguyen T V, Hussey DS, Jacobson DL. Experimental study of

relative permeability of porous media used in PEM fuel cells. ECS
Transactions 2010; 33:1151–1162.

125
Bibliography

77 Martins LS, Ordonez JC, Vargas JVC. Experimental validation of a

simplified pemfc simulation model. ES2009: Proceedings Of The Asme 3rd
International Conference On Energy Sustainability, Vol 1 2009; :305–314.

78 Gurau V, Barbir F, Liu H. Analytical solution of a half–cell model for PEM

fuel cells. Journal of the Electrochemical Society 2000; 147:2468–2477.

79 Springer TE, Zawodzinski TA, Gottesfeld S. Polymer Electrolyte Fuel–Cell

model. Journal of the Electrochemical Society 1991; 138:2334–2342.

80 Nishiyama E, Murahashi T. Permeability of a PEMFC gas diffusion layer

based on morphology. ECS Transactions 2011; 41:2021–2031.

81 Cheng X, Zhang J, Tang Y, Song C, Shen J, Song D, et al. Hydrogen

crossover in high–temperature PEM fuel cells. Journal of Power Sources 2007;
167:25–31.

82 Mittelsteadt C, Umbrell M. Gas Permeability in Perfluorinated Sulfonic

Acid Polymer Electrolyte Membranes. Meeting abstracts– electrochemical society
– 2005.

83 Groot A De, Woudstra N. Exergy analysis of a fuel–cell system. Journal of the

Institute of Energy 1995; 68.

84 Wright SE. Comparison of the theoretical performance potential of fuel

cells and heat engines. Renewable Energy 2004; 29:179–195.

85 Bejan A. Advanced engineering thermodynamics. Hoboken,N.J.: Wiley; 2006.

86 Bejan A, Tsatsaronis G, Moran M. Thermal design and optimization. New

York: Wiley–Interscience; 1996.

87 Mathias, Makharia, Gasteiger, Conley, Fuller, Gittleman, et al. Two fuel cell
cars in every garage? Electrochemical Society Interface 2005; 14:24–35.

88 Yu PT, Kocha S, Paine L, Gu W, Wagner FT. The effects of air purge on the
degradation of PEM fuel cells during startup and shutdown procedures.
2004 AIChE Spring National Meeting, Conference Proceedings 2004; :521–527.

126
 Bibliography

89 Patterson TW, Darling RM. Damage to the cathode catalyst of a PEM fuel
cell caused by localized fuel starvation. Electrochemical and Solid–State Letters
2006; 9:A183–A185.

90 Wong KH, Loo KH, Lai YM, Tan S–C, Tse CK. A theoretical study of inlet
relative humidity control in PEM fuel cell. International Journal of Hydrogen
Energy 2011; 36:11871–11885.

91 Reiser CA, Bregoli L, Patterson TW, Yi JS, Yang JD, Perry ML, et al. A
reverse–current decay mechanism for fuel cells. Electrochemical and Solid–
State Letters 2005; 8:A273–A276.

92 Meyers JP, Darling RM. Model of carbon corrosion in PEM fuel cells. Journal
of the Electrochemical Society 2006; 153:A1432–A1442.

93 Rabbani A, Rokni M. Dynamic characteristics of an automotive fuel cell

system for transitory load changes. Sustainable Energy Technologies and
Assessments 2013; 1.http://dx.doi.org/10.1016/j.seta.2012.12.003

94 Rabbani A, Rokni M. Start–up Analysis of a PEM Fuel Cell System in

Vehicles. International Journal of Green Energy 2013;11:1–21.

95 Hosseinzadeh E, Rokni M, Rabbani A, Mortensen HH. Thermal and water

management of low temperature Proton Exchange Membrane Fuel Cell in
fork–lift truck power system. Applied Energy 2013; 104:434–444.

127
Bibliography

128
 Appendix A

Aspen Dynamics Modelling Language

A.1. PEM Fuel Cell component model

129
Appendix A

130




!
 




& '('
 ( ) * )
+ , '  -
. 
/
% '('
 ( 0 * )
1 ,  2345464478-
!# 0$454-
!! 
!
!
!&""'9)))::)
!+
!. ); ' ) 2 $4)):48-
!/ )< ' ) 2 $4))=48-
!% );< ' ) 2 $4))948-
!1 ) ' ) 2 $4))48-
# ); ' ) 2 $4))))48-
!  ' 0 2 $4 048-
 > ' 0 2 $4 >048-
 );, $ )?-
& )<, $ )?-
+ );<,$ )?-
. ), $ )?-
/ );,$ )?-
% ); $ 4;4-
1 )< $ 4<4-
# );<$ 4;<4-
! ) $ 44-
 );$ 4;4-

& )?' '   23);6 );<6 )78-
+ )?: '   23)<6 );<6 )78-
. )?@' '   2)?  )?'8-
/ )?@ '   2)?  )?:8-
% )?@ '   2)?  )?'  )?:8-
1
&#
&!""
& @' ' @

-

A.1
A.2
& @ ' @

-
&& <' ' <

-
&+ < ' <

-
&. @ ' @

-
&/ < ' <

-
&% >A ' < >A-
&1 ; ' < ;-
+#
+!"")
+
 ' () 2 $4
2")846 1.&%/8-
+ ( ' () 2 $4B2C")D846 %!&8-
+&
++""
=0
+.  ' ) 2 $4'))46
=8-
+/ D ' ) 2 $4'))46 1%!+8-
+%  '  2 $4')46
=8-
+1  '   2 $446
=6 #!8-
.#  '  2 $4)A46
=6 !!#8-
.! ' ' ' 2 $4'0:46
=6 ##%+%8-
. ' ' ' 2 $4 A46
=6 #+8-
. :0 ' : 2 $4:46
=6 ###+8-
.& ) ' ?: 2 $4
):A46
=6 ####!%8-
.+  '  2 $4'0)46
=6 &###8-
.. ': '  2 $4)46
=6 %#8-
./ , ' < , 2 $446 !!8-
.% E '  2 $4)046
=6 #8-
.1 (: ' ) 2 $4 46
=6 %#8-
/#
> '
9: 2 $4
)46
=6 !!##8-
/!  '  2 $4)46
=6 #8-
/ @ '  2 $4646
=6 ###8-
/ @ ' @6<  2 $446
=6 !8-
/& @? '  2 $4046 #+8-
/+ @)= '  2 $4
=))46
=6 !8-
/.  '  2 $446
=6 #8-
//  '  2 $446
=6 !#8-
/% F '  2 $4F46
=6 #+8-
/1   ' )  2 $4)46
=6 !#8-
%# ); ' :) 2 $4 ):)=;46
=6 #.!8-
%! )< ' :) 2 $4 ):)=<=46
=6 ###1/#8-
% );< ' :) 2 $4 ):)=>2846
=6 ###1+##8-
% );< ' :) 2 $4 ):)=>2846
=6 ####%%&8-
%& ) ' :) 2 $4 ):)=46
=6 ####/#8-
 %+ )< ' :) 2 $4 ):)=<46
=6 ##!1%/#8-
%. )'( ' :) 2 $4 ):)='46
=6 ##!!.&#8-
%/
%%"",9:
%1 2)?'8 '
9)9 2 $4
948-
1# 2)?'8 '
9)9 2 $4?G)948-
1! 2)?:8 '
9)9 2 $4
9:48-
1 2)?:8 '
9)9 2 $4?G)9:48-
1 2)?'8 '
9)9 2 $4
948-
1& 2)?'8 '
9)9 2 $4?G)948-
1+ 2)?:8 '
9)9 2 $4
9:48-
1. 2)?:8 '
9)9 2 $4?G)9:48-
1/ @2)?@8 '
9)9 2 $4')948-
1% @2)?@8 '
9)9 2 $4:)948-
11 2)?8 '
9) 2 $4
948-
!## 2)?8 '
9) 2 $4
948-
!#!
>2)?8 '
9: 2 $4
9:48-
!# B
2)?8 ' B)0 2 $4B)48-
!# B
2)?8 ' B)0 2 $4B)+48-
!#& E9 2)?8 '
 2 $4,H46
=6 #8-
!#+ 2)?'8 '  2 $446 #+8-
!#. 2)?:8 '  2 $4:46 #+8-
!#/ 2)?'8 '  2 $446 #+8-
!#% 2)?:8 '  2 $4:46 #+8-
!#1 02)?8 '  2 $4'046 #+8-
!!# 02)?8 '  2 $4'0:46 #+8-
!!!
<2)?8 '
9) 2 $4
948-
!!
<2)?8 '
9) 2 $4
9:48-
!!
<2)?8 '
9) 2 $4
948-
!!&
<2)?8 '
9) 2 $4
9:48-
!!+ 2)?8 ' ;  2 $4 46 #+8-
!!. 2)?8 ' ;  2 $4 :46 #+8-
!!/ 2)?8 ' ;  2 $4 46 #+8-
!!% 2)?8 ' ;  2 $4 46 #+8-
!!1 52)?'8 '
 2 $4')48-
!# 52)?:8 '
 2 $4:)48-
!! 52)?'8 '
 2 $4')48-
! 52)?:8 '
 2 $4:)48-
! 52)?8 '
 2 $4'0)48-
!& 52)?8 '
 2 $4:0)48-
!+ 52)?8 '
 2 $40)48-
!. E2)?8 '
 2 $4'G)48-

A.3
A.4
!/ E2)?8 '
 2 $4:G)48-
!% E2)?8 '
 2 $4G)48-
!1 52)?8 '
 2 $4'0)48-
!# 52)?8 '
 2 $4:0)48-
!! 52)?8 '
 2 $40)48-
! E2)?8 '
 2 $4'G)48-
! E2)?8 '
 2 $4:G)48-
!& E2)?8 '
 2 $4G)48-
!+
!.""<:0
!/ ; ' @6<  2 $4@:);46 !.8-
!% < ' @6<  2 $4@:)<=46 !%8-
!1 ( ' @6<  2 $4:)(46 #18-
!&#  ' ) 2 $4 A)46 (6 1%!+8-
!&!  ' @6< ) 2 $4')46 .!8-
!&  ' @6< ) 2 $4:)46 .!8-
!&  ' @6< ) 2 $49)46 .!8-
!&&  ' @6< ) 2 $49)46 .18-
!&+  ' @6< ) 2 $4'<)46 .18-
!&.  ' @6< ) 2 $4:<)46 .18-
!&/  ' @6< ) 2 $4<)46 #8-
!&%  ' @6< ) 2 $4<)46 (6 +8-
!&1 D ' ) 2 $4<)46 &!+8-
!+#  '  2 $4'046 !18-
!+!  '  2 $4'0:46 !.8-
!+  '   2 $446 #!+8-
!+  '   2 $4:46 #&8-
!+&  ' @6<  2 $446 #8-
!++  ' @6<  2 $4:46 !%8-
!+.  ' @6<  2 $446 !%+8-
!+/  ' @6<  2 $4:46 !&8-
!+%  ' @6<  2 $4946 !#8-
!+1  ' @6<  2 $4946 !#8-
!.# (;< ' )0 2 $4'048-
!.! (;< ' )0 2 $4'048-
!. (;< ' )0 2 $4:048-
!. (;< ' )0 2 $4:048-
!.& (;< ' )G 2 $4'G48-
!.+ (;< ' )G 2 $4'G48-
!.. (;< ' )G 2 $4:G48-
!./ (;< ' )G 2 $4:G48-
!.% (;< ' )G 2 $4G48-
 !.1 (;< ' )G 2 $4G48-
!/# B ' < 9 2 $4946 !.+8-
!/! ;B ' <  2 $4;46 !&#8-
!/  '
9) 2 $4
946 #%8-
!/  '
9) 2 $4
9:46 +#8-
!/&  '
9) 2 $4
946 !#8-
!/+  '
9) 2 $4
9:46 &#8-
!/.
 ' @6<
9) 2 $4)948-
!//
 ' @6<
9) 2 $4)9:48-
!/%
 ' @6<
9) 2 $4)948-
!/1
 ' @6<
9) 2 $4)9:48-
!%#
> '
9: 2 $4'0
9:46 +8-
!%!
> '
9: 2 $4'0
9:46 +#8-
!%
> '
9: 2 $4'0
9:46 /+8-
!%
> '
9: 2 $4'0
9:46 +#8-
!%&
> '
9: 2 $4'0
9:46/+8-
!%+
> '
9: 2 $4'0
9:46+#8-
!%.  '  2 $4>46 #8-
!%/  '  2 $4>:46 #8-
!%%  '  2 $4<9046 #8-
!%1  '  2 $4>046 #8-
!1# 9 '  2 $4'9046 #!8-
!1! 9 '  2 $4'9046 #/8-
!1 9 '  2 $4:9046 #8-
!1 9 '  2 $4:9046 #/8-
!1& 9 '  2 $4>046 #1&8-
!1+ 9 '  2 $4>0:46 #118-
!1. < '  2 $4'0<46 #8-
!1/ ; '  2 $4'0;46 !%8-
!1% ;< '  2 $4'0';<46 !%8-
!11 ;< '  2 $4'0:;<46 !%8-
## ?) '  2 $4>46 !#%8-
#! ?) '  2 $4>:46 !18-
# ?))) '  2 $4>))48-
#  ' 0 2 $4 948-
#& C ' =) 2 $4
=9)48-
#+ C ' =) 2 $4
=9):48-
#. C ' =) 2 $4
=9:48-
#/ C '
9)9 2 $4
99:48-
#% C) ' 9)9 2 $4
99:48-
#1 C '
9) 2 $4
99:A48-
!# C) '
9) 2 $4
99:A48-

A.5
A.6
!! C; ' =) 2 $4
=;'48-
! C< ' =) 2 $4
=<=:48-
! B '
 2 $4 B48-
!& B '
 2 $4(B48-
!+ B '
 2 $4B+48-
!. @ '  2 $4'=:48-
!/ @ '  2 $4:=:46 !1 /8-
!% @ '  2 $4'046 #%8-
!1 @) '  2 $4')48-
# @) '  2 $4:)48-
! @ ' @6<  2 $448-
 @ ' @6<  2 $448-
 , ' @6< , 2 $446 #.+8-
& , ' , 2 $446 !!/8-
+ , ' , 2 $4 +46 !!/8-
. , ' , 2 $4'0046 #&8-
/ , ' , 2 $4''0046 ##8-
% , ' , 2 $4:'0046 #&8-
1 ,:) ' , 2 $4<:)046 #!8-
# , ' , 2 $4'046 ###!8-
! , ' , 2 $4:046 ###!8-
 , ' , 2 $4046 ###!8-
 ,A ' , 2 $4 A048-
& ( ' ( 2 $4 :46 ##8-
+ ( ' ( 2 $4@:48-
. ( ' ( 2 $4@:62:))48-
/  '
 2 $4B048-
% ' '  2 $4'46 !#8-
1 ' '  2 $4'46 #+8-
&#  ' @6<  2 $4'046 #+8-
&! '
( ' @6<  2 $4'
(48-
& (; '  2 $4(0:)'@46 #1+8-
& (; '  2 $4(0:):@46 #18-
&& (; '  2 $4(0:)'<46 !#8-
&+ (; '  2 $4(0:):<46 !#8-
&. ,
 ' , 2 $4'046 !8-
&/ ,
 ' , 2 $4:046 !8-
&% ,
 ' , 2 $4046 !8-
&1 ?
 ' ?G 2 $4'G46 !8-
+# ?
 ' ?G 2 $4:G46 !8-
+! ?
 ' ?G 2 $4G46 !8-
+ ;0 ' :)0 2 $4'0):48-
 + ;0 ' :)0 2 $4:0):48-
+& ;0 ' :)0 2 $40):48-
++ ; ' :)G 2 $4'G):48-
+. ; ' :)G 2 $4:G):48-
+/ ; ' :)G 2 $4G):48-
+% ,
 ' , 2 $4'046 !8-
+1 ,
 ' , 2 $4:046 !8-
.# ,
 ' , 2 $4046 !8-
.! ?
 ' ?G 2 $4'G46 !8-
. ?
 ' ?G 2 $4:G46 !8-
. ?
 ' ?G 2 $4G46 !8-
.& ;0 ' :)0 2 $4'0):48-
.+ ;0 ' :)0 2 $4:0):48-
.. ;0 ' :)0 2 $40):48-
./ ; ' :)G 2 $4'G):48-
.% ; ' :)G 2 $4:G):48-
.1 ; ' :)G 2 $4G):48-
/#  ' ) 2 $4 :G948-
/!  ' ) 2 $4 :G948-
/ 0 ' ) 2 $4'0 :G948-
/  '  2 $4 48-
/&  '  2 $4 48-
/+  '  2 $4 48-
/.  '  2 $4 48-
//  '  2 $4 48-
/%  '  2 $4 948-
/1  ' ) 2 $4 '48-
%#  ' ) 2 $4 '48-
%!  ' ) 2 $4 :48-
%  ' ) 2 $4 :48-
%  ' ) 2 $4 48-
%&  ' ) 2 $4 48-
%+  ' ) 2 $4  '48-
%.  ' ) 2 $4  '48-
%/  ' ) 2 $4  :48-
%%  ' ) 2 $4  :48-
%1  ' ) 2 $4  48-
1#  ' ) 2 $4  48-
1!  ' ) 2 $4 '48-
1  ' ) 2 $4 '48-
1  ' ) 2 $4 :48-
1&  ' ) 2 $4 :48-

A.7
A.8
1+  ' ) 2 $4 48-
1.  ' ) 2 $4 48-
1/ ; ' :) 2 $4  :'48-
1% ; ' :) 2 $4  :'48-
11 ; ' :) 2 $4  ::48-
## ; ' :) 2 $4  ::48-
#! ; ' :) 2 $4  :48-
# ; ' :) 2 $4  :48-
# ;  ' :) 2 $4 :'48-
#& ;  ' :) 2 $4 :'48-
#+ ;  ' :) 2 $4 ::48-
#. ;  ' :) 2 $4 ::48-
#/ ;  ' :) 2 $4 :48-
#% ;  ' :) 2 $4 :48-
#1 @ '  2 $4 48-
!# < '  2 $4 48-
!!  '  2 $4 I48-
! B '  2 $4 48-
! =  ' 9 2 $4 =48-
!&
  ' ) 2 $4 '48-
!+ : ' :) 2 $4:=48-
!. : ' :) 2 $4:=48-
!/ : ' :) 2 $4:=:48-
!% : ' :) 2 $4:=:48-
!1 : ' :) 2 $4:=48-
# : ' :) 2 $4:=48-
! : ' :) 2 $4:)=48-
 : ' :) 2 $4:)=48-
 : ' :) 2 $4:)=:48-
& : ' :) 2 $4:)=:48-
+ : ' :) 2 $4:)=48-
. : ' :) 2 $4:)=48-
/ :0 ' :) 2 $4:)=048-
% :0 ' :) 2 $4:)=048-
1 :0 ' :) 2 $4:)=0:48-
# :0 ' :) 2 $4:)=0:48-
! :0 ' :) 2 $4:)=048-
 :0 ' :) 2 $4:)=048-
 : ' :) 2 $4:)=G48-
& : ' :) 2 $4:)=G48-
+ : ' :) 2 $4:)=G:48-
. : ' :) 2 $4:)=G:48-
 / : ' :) 2 $4:)=G48-
% : ' :) 2 $4:)=G48-
1 = ' :) 2 $4=48-
&# = ' :) 2 $4=:48-
&! = ' :) 2 $4=48-
& = ' :) 2 $4=:48-
& = ' :) 2 $4=48-
&& = ' :) 2 $4=48-
&+ =@ '  2 $4=:)48-
&. =< '  2 $4=:)48-
&/ =  '  2 $4 =:)48-
&% =H '  2 $4 =::)48-
&1 = '  2 $4 =:)48-
+# =  '  2 $4 =948-
+! E ' @6<  2 $4 =48-
+ ,)) ' ,) 2 $4
,)48-
+ ,9 ' ,) 2 $4
,)G948-
+& 0 '  2 $4
9))48-
++ A ' ) 2 $4
=)))46 =6 ! !&8-
+. A ' ) 2 $4)48-
+/ C '
9)9 2 $4
9:48-
+% C '
9)9 2 $4
9:A48-
+1 C) '
9)9 2 $4
9:48-
.# C) '
9)9 2 $4
9:A48-
.! ! ' @6<  2 $4,46 !#8 -
.  ' @6<  2 $4,46 !#8 -
.  ' @6<  2 $4,46 !#8 -
.&  ' <  2 $4:'):48-
.+
..
./"":
.%  J @ K )-
.1
/#""
9::)
/! 
>$  2
>8 J 
>:28-
/
,$ ,)) K (: J
>-
/
,?G;<$ ,9 K !### J !%#!+.-
/&
/+""')
/. D J  L /!+-
//
/%""<)

A.9
A.10
/1 )$  J <  @-
%# )$  J D-
%!
%"")
% )<D0$ D J  L /!+-
%&
%+""'
%. @'
:$  256 E6 ,
6 ;06 ;8 J : 2@'6 @'6 @'M8-
%/ @')$  J @'-
%% @'$  J @'-
%1 @';)9$ 24;48 J @'
K @'M24;48-
1# @'9)9$ 24;<48 J @'
K @'M24;<48-
1! @')9$ 2448 J @'
K @'M2448-
1 @')9$ @2)?@8 J @'
K @'M2)?@8-
1 @'?G$ ?
 J !  ,
-
1& @'G9$ 24;<48 J 24;<48 K ?
-
1+ @',?9$ !+-
1.
1/"":
1% @
:$  256 E6 ,
6 ;06 ;8 J 
: 2@6 @6 @M8-
11 @)$  J @-
&## @$  J @-
&#! @<)9$ 24<48 J @
K @M24<48-
&# @9)9$ 24;<48 J @
K @M24;<48-
&# @)9$ 2448 J @
K @M2448-
&#& @)9$ @2)?@8 J @
K @M2)?@8-
&#+ @?G$ ?
 J !  ,
-
&#. @G9$ 24;<48 J 24;<48 K ?
-
&#/ @,?9$ !+-
&#%
&#1""
&!# 
:$  256 E6 ,
6 ;06 ;8 J : 2@6 @6
@M8-
&!! )$  J @-
&! $  J @-
&! )9$ 2)?8 J @
K @M-
&!& $  28 J 
?G 2@6 @6 @M8-
&!+ ?G$ ?
 J !  ,
-
&!.
&!/""'
&!% <'
:$  256 E6 ,
6 ;06 ;8 J 
: 2<'6 <'6 <'M8-
&!1 <')9$ <'
J )28 L )2@8-
 &# <')$ <' J -
&! <'$ <' J -
& <';)9$ <'M24;48 K
=2!+6<'
8 J
=224;488-
& <'<)9$ <'M24<48 K
=2!+6<'
8 J
=2@24<488-
&& <'9)9$ <'M24;<48 K
=2!+6<'
8 J
=224;<488-
&+ <')9$ <'M2448 K
=2!+6<'
8 J
=224488-
&.
&/
   )?@ 
&% <')9$ <'M28 K
=2!+6<'
8 J
=2@288-
&1 

&# <' :$  2<':8 J  :) 2<'6 <'6 <'M8-
&! <', $  2(;<8 J  
, 2<'6 <'6 58-
& <'?G $  2(;<8 J  
G 2<'6 <'6 E8-
& <'?G$ ?
 J !  ,
-
&& <')$ <', K 2(;<K,
 L(;<K?
8 J !-
&+ ;9+$ 24;48 J 24;48  22@L@?8KK.8"2K
8-
&. <'G9$ 24;<48 J 24;<48 K ?
-
&/
&% @ 2JJ4548 :
&1 2448 J 2448 L C-
&&# 
&&! 2448 J 2448-
&& 
&&
&&& @ 2>JJ4548 :
&&+ 24;<48 J 24;<48 L C-
&&. 
&&/ 24;<48 J 24;<48-
&&% 
&&1
&+#"":
&+! <
:$  256 E6 ,
6 ;06 ;8 J 
: 2<6 <6 <M8-
&+ <)9$ <
J )28 L )2@8-
&+ <)$ < J -
&+& <$ < J -
&++ <;)9$ <M24;48 K
=2!+6<
8 J
=2@24;488-
&+. <<)9$ <M24<48 K
=2!+6<
8 J
=224<488-
&+/ <9)9$ <M24;<48 K
=2!+6<
8 J
=224;<488-
&+% <)9$ <M2448 K
=2!+6<
8 J
=224488-
&+1
&.#
   )?@ 
&.! <)9$ <M28 K
=2!+6 <
8 J
=2@288-

A.11
A.12
&. 

&. < :$  2<:8 J  :) 2<6 <6 <M8-
&.& <, $  2(;<8 J  
, 2<6 <6 58-
&.+ <?G $  2(;<8 J  
G 2<6 <6 E8-
&.. <?G$ ?
 J !  ,
-
&./ <)$ <, K 2(;<K,
 L(;<K?
8 J !-
&.% <=9.$ 24<48 J 24<48  2#+K2@L@8KK.8"2K
8-
&.1 <G9$ 24;<48 J 24;<48 K ?
-
&/#
&/! @ 2JJ4548 :
&/ 2448 J 2448  C-
&/ 
&/& 2448 J 2448-
&/+ 
&/.
&// @ 2>JJ4548 :
&/% 24;<48 J 24;<48  C L 22@L@?8KK.8"2K
8-
&/1 
&%# 24;<48 J 24;<48 L 22@L@?8KK.8"2K
8-
&%! 
&%
&%""
&%& 
:$  256 E6 ,
6 ;06 ;8 J : 2<6 <
6 <M8-
&%+ )$  J <-
&%. $ < J @  -
&%/ )$ <M J @M-
&%% :$  2<:8 J  :) 2<6 <6 <M8-
&%1 $  2(;<8 J  
?G 2<6 <6 <M8-
&1# )$ <, K (;< J !-
&1! ?G$ ?
 J !  ,
-
&1 $  28 J 
?G 2<6 <6 <M8-
&1 '0$ 0 K  J  L -
&1& 9$ <
J @
-
&1+
&1.""
9)
&1/
)@'$  2
>8 J 
>:2@'M8-
&1%
)@$  2
>8 J 
>:2@M8-
&11
)<'$  2
>8 J 
>:2<'M8-
+##
)<$  2
>8 J 
>:2<M8-
+#!
)@$  2
>8 J 
>:2@M8-
+#
)<$  2
>8 J 
>:2<M8-
 +#
+#& @'
9$
< J @'
K @'M K
>-
+#+ @')9$
 J )2
<8-
+#. @
9$
< J @
K @M K
>-
+#/ @)9$
 J )2
<8-
+#% <'
9$
< J <'
K <'M K
>-
+#1 <')9$
 J )2
<8-
+!# <
9$
< J <
K <M K
>-
+!! <)9$
 J )2
<8-
+!
+!""'
(
+!& '($ '
( K
 J
-
+!+
+!."" :)
+!/ ;:)$ ; J 24;48 " 224;4824;488-
+!% <=:)$ < J 24<48 " 224<4824<488-
+!1
+#""
+! '$ @ 2@ NJ !##8 :
+  J ##+-
+ 
+&  J ##+1&+#!#./.K E22!1/ !#8K2@O2&#!.888-
++ 
+.
+/ :$ @ 2@ NJ !##8 :
+%  J ##+-
+1 
+#  J ##11.+  .%% #&K@ L +%# #+K2@O8  !%.& #/K2@O8 L
1 !#K2@O&8-
+! 
+ @'$  J   -
+ @$  J   -
+& '0$  K  J 2 L 8-
++ '0:$  K  J 2 L 8-
+.
+/""
+% @'$  J @'M K
=2!+68-
+1 @$  J @M K
=2!+68-
+&# <'$  J <'M K
=2!+68-
+&! <$  J <M K
=2!+68-
+&
+&""'0

A.13
A.14
+&& '0';$  K ; J 24;48 L 24;48-
+&+ '0';<$  K ;< J 24;<48 L 24;<48-
+&. '0<$  K < J 24<48 L 24<48-
+&/ '0;<$  K ;< J 24;<48 L 24;<48-
+&%
+&1 '0'$  K 0 J  L -
++# '0$  K 0 J  L -
++!
++"">0
++ <$  2 8 J , 28-
++& @'$  2 8 J , 28-
+++ @$  2 8 J , 28-
++.  $  2 8 J , 28-
++/
++% @'$  J 24;<48-
++1 @$  J 24;<48-
+.# <$  J 24;<48-
+.!  $  J 24;<48-
+.
+.""(0;))
+.& @ 224;<48 P 8 :
+.+ (; J !#-
+.. 
+./ (:)@'$ (; K  J 24;<48-
+.% 
+.1
+/# @ 224;<48 P 8 :
+/! (; J !#-
+/ 
+/ (:)@$ (; K  J 24;<48-
+/& 
+/+
+/. @ 224;<48 P 8 :
+// (; J !#-
+/% 
+/1 (:)<'$ (; K  J 24;<48-
+%# 
+%!
+% @ 224;<48 P 8 :
+% (; J !#-
+%& 
+%+ (:)<$ (; K  J 24;<48-
 +%. 
+%/
+%%"">0
+%1 >0$ 9 J E K 2(;8 L 2!  E8 K 2(;8-
+1# >0$ 9 J E K 2(;8 L 2!  E8 K 2(;8-
+1!
+1""><0
+1 >$ ?) J ##& L !/!% K 9  1%+ K29O8 L . K29O8-
+1& >$ ?) J ##& L !/!% K 9  1%+ K29O8 L . K29O8-
+1+ >))$ ?))) K  J 2?) L ?)8-
+1. $  J !# .K2 E22&!.#K22!#"#8!#"D8888K2+.#K?)L##.&K2?)O
8####./!K2?)O88-
+1/ $ C K 2#K
8 J 2+K@K)8 K!#-
+1% :$ C K 2
> K )8 J (: K  K ! &K2))8-
+11 =$ C J C  C-
.## 9$ C J C K ' K .##-
.#! )9$ C) J C K
>24;<48K ' K .##- ""02)")O82A":8
.# 90$ C J C K -
.# 9)$ C) J C K
>24;<48-
.#&
.#+""<0
.#. >))$ 0 J 2?))) K ,98"2?)))K,9 L ,))8-
.#/ )$ A J ': K A K 2##1+ L 2!!K08  !1K20O88 K E22 "(8K22!"#82!"
D888-
.#% )9$ C K ) J A K .## K 202448  024488 K ' K !##-
.#1 )9$ C) J C K
>2448-
.!# )A$ C J C K -
.!! )A$ C) J C) K -
.!
.!""(
=
.!& ;
=$ C; K ' K .## J 2);8-
.!+ <
=$ C< K ' K .## J 2)<8-
.!.
.!/"":
.!% $ @ K2.##8 J  K
 K 224;4824;488K!###-
.!1 $ @ K 2'K!####8 J @-
.# $ @ K  J @-
.! ?$ @? J @ K ' K !####-
. ?)'$ @) J ; K @)=-
. ?)$ @) J < K @)=-
.&
.+""B

A.15
A.16
.. $  2B
8 J B
@ ?B' 26 @M8-
./ $ B J! . K 2B
24;<48  B
24;48  #+KB
24<488-
.% B0$  J 2( K 2D8 K ?<B2!"22;O2!88K2<O2#+88888-
.1 $ B J B L 2( K 2D8 K ?<B2!"22;O2!88K2<O2#+88888-
.#
.! $  2B
8 J B
@ ?B' 26 @M8-
. $ B J ! . K 2B
24;<48  B
24;48  #+KB
24<488-
.
.&"":
.+ $ , K 2K
8 J 2B8-
.. $ , K 2K
8 J 2B8-
./
.%""'0<0
.1 '$ ' J 2!F8K -
.&# '$ ' J 2!F8K -
.&! =: $ @ J #. K
 K E22F8KK
 K !!"2(KD88-
.& '$ , J ########!-
.& ':$ , K 2' K
8 J ( K 2D8 K ?<B22@L@8"2
=2!1 /6@888-
.&& '0$ , J , L ,-
.&+
.&.""<:)<0
.&/ $ ( J ########!-
.&% @$ ( J 2!%#K2!L##K2@L@8L##.K2D"#8OK2@L@8O+8K2)K!##8"22?)))
#.K2@L@88K E2!.&K22D#8"D8888-
.&1 @:)$ ( K 2'K!####8 J (-
.+# <:)0$ ,:) J 2@ L @ 8 K ' K!#### K 2( L (8-
.+!
.+""<0
.+ 0'$ , J 22(KD8"2K
88 K ?<B22!  2@"@)888-
.+& 0:$ , J 22(KD8"2K
88 K ?<B22!  2@"@)888-
.++ 0$ , J , L ,-
.+.
.+/""
 A9
.+% 0$ , J ,  ,  ,:)  ,-
.+1 A0$ ,A J , K -
..# 9 A$ B K !### J , K @ K -
..! $  K !+ J ,-
..
.."">A
..& >A9$ >A> J B-
..+ >A$ >A  J #-
...
 ../"";
..% ;B$ ;H J 2@':K@'
L @:K@
 <':K<'
 <:K<
8 L
-
..1 ;$ ;B J ;H-
./#
./!"" F
./ $  J 2@':K@'
L @:K@
L @: K@
8-
./ $  J 2<':K<'
L <:K<
L <: K <
8-
./& $   J   -
./+ $  K !# + J B K .##-
./. $  K !# + J 2:0K'K2D  D8K.##8-
.// $  K !# + J 2.###K##&K#/+KQ8-
./% F$  J  L  L  L -
./1
.%#"" B
.%! '@$  2 8 J  ) 2@'6 @'6 @'M8-
.% :@$  2 8 J  ) 2@6 @6 @M8-
.% '<$  2 8 J  ) 2<'6 <'6 <'M8-
.%& :<$  2 8 J  ) 2<6 <6 <M8-
.%+ @$  2 8 J  ) 2@6 @6 @M8-
.%. <$  2 8 J  ) 2<6 <6 <M8-
.%/
.%%  @$ @ K .## J K@'
L K@
L K@
-
.%1  <$ < K .## J K<'
L K<
L K<
-
.1# :$  J <  @-
.1! =$ =  J B K2D8-
.1 B$ B K D K .## J =  K!# %- ""2>"D8
.1
.1&  J   -
.1+  J   -
.1.  J   -
.1/  J   -
.1%  J   -
.11  J   -
/##
/#! ;  J @':  ;-
/# ;  J @:  ;-
/# ;  J <':  ;-
/#& ;  J <:  ;-
/#+ ;  J @:  ;-
/#. ;  J <:  ;-
/#/

A.17
A.18
/#%
/#1"" =F
/!#  '@$  2 8 J  ) 26 6 @'M8-
/!!  :@$  2 8 J  ) 26 6 @M8-
/!  '<$  2 8 J  ) 26 6 <'M8-
/!  :<$  2 8 J  ) 26 6 <M8-
/!&  @$  2 8 J  ) 26 6 @M8-
/!+  <$  2 8 J  ) 26 6 <M8-
/!.
/!/  :'@$  2;8 J  :) 26 6 @'M8-
/!%  ::@$  2;8 J  :) 26 6 @M8-
/!1  :'<$  2;8 J  :) 26 6 <'M8-
/#  ::<$  2;8 J  :) 26 6 <M8-
/!  :@$  2;8 J  :) 26 6 @M8-
/  :<$  2;8 J  :) 26 6 <M8-
/
/& : ='@$ : J 2@':  ;8  22D8K2   8"!# +8-
/+ : ='<$ : J 2<':  ;8  22D8K2   8"!# +8-
/. : =:@$ : J 2@:  ;8  22D8K2   8"!# +8-
// : =:<$ : J 2<:  ;8  22D8K2   8"!# +8-
/% : =@$ : J 2@:  ;8  22D8K2   8"!# +8-
/1 : =<$ : J 2<:  ;8  22D8K2   8"
!# +8-
/#
/! : =,'@$ :0 J 2524;48K ); L
=2!+6524;<488K );<8 L 2(K2D8K252
4;48K?<B2524;488 L
=2!+6524;<488K?<B2
=2!+6524;<48888"!# +8-
/ : =,'<$ :0 J 2524;48K ); L 524;<48K );<8 L 2(K2D8K2524;48K?<B2
524;488 L 524;<48K?<B2524;<4888"!# +8-
/ : =,:@$ :0 J 2524;<48K );<L52448K )L524<48K )<8 L 2(K2D8K25
24<48K?<B2524<488L52448K?<B2524488L524;<48K?<B2524;<4888"!# +8-
/& : =,:<$ :0 J 2524;<48K );<L52448K )L524<48K )<8 L 2(K2D8K2
524<48K?<B2524<488L52448K?<B2524488L524;<48K?<B2524;<4888"!# +8-
/+ : =,@$ :0 J 2524;<48K );<8 L 2(K2D8K2524;<48K?<B252
4;<4888"!# +8-
/. : =,<$ :0 J 2524;<48K );<8 L 2(K2D8K2524;<48K?<B252
4;<4888"!# +8-
//
/% : =?'@$ : J );<-
/1 : =?'<$ : J );<-
/&# : =?:@$ : J );<-
/&! : =?:<$ : J );<-
/& : =?@$ : J );<-
 /& : =?<$ : J );<-
/&&
/&+ : ='@$ : J :0K,
 L :K2!,
8-
/&. : ='<$ : J :0K,
 L :K2!,
8-
/&/ : =:@$ : J :0K,
 L :K2!,
8-
/&% : =:<$ : J :0K,
 L :K2!,
8-
/&1 : =@$ : J :0K,
 L :K2!,
8-
/+# : =<$ : J :0K,
 L :K2!,
8-
/+!
/+  ='@$ = J : L :-
/+  =:@$ = J : L :-
/+&  ='<$ = J : L :-
/++  =:<$ = J : L :-
/+.  =@$ = J : L :-
/+/  =<$ = J : L :-
/+%
/+1 =@ $ =@ J 2 =K@'
L =K@
L =K@
8-
/.# =< $ =< J 2 =K<'
L =K<
L =K<
8-
/.! =; $ =H J 2!  22D8"2D888K 2 8-
/. = $ = J 2Q"288K2 8-
/. = $ =  J =@  =<  2 8  =H  =-
/.& =$ =  J =@  =<-
/.+ =$ E J 2B K ###.8 " 2 =@  =<8-
/..
/./
/.% 

A.19
A.20
 Appendix A

A.2. Humidifier component model

151
Appendix A

152


 

 


& 
 '
 (  & 

) * +  ,
% -
.
/!!+0 

 
 11

2
# '
 +  '
 3 $4'

 1 45,
" '
6 +  '
 3 $4'

 745,
 '
6 +  '
 3 $4'

 0 45,
 '
 +  '
 3 $4'

  45,
 '
'6 +  '
 3 $4'

  745,
) '
+8 +  '
 3 $4'

  45,
% '
* $ '
9,
. '
6* $ '
9,
/ '
6*$ '
9,
2 '
* $ '
9,
# '
'6*$ '
9,
" '
+8* $ '
9,
  '
 $ 44,
 '
6 $ 464,
 '
6$ 464,
) '
 $ 44,
% '
'6$ 4'64,
. '
+8 $ 4+84,
/
2!!& 
#
" :-71 + : 
 & ,
  :+ + : 
 & ,
 6 -71 + 6  
 & ,
 6 + + 6  
 & ,
)
%!!& 
 
.
/ 8 + 8& 
 3 $48;1
  4< "#%5,
2
#!!: ; 
)"
) 
= +
01 3 $4+; 
  0171 4< %5,
)
=  +
01 3 $4+; 
  0171  4< %"5,

A.21
A.22
)
= +
01 3 $4+; 
  01 4< /%5,
))
=  +
01 3 $4+; 
  01  4< %"5,
)% & + &   3 $4=     4< "5,
). & + &   3 $4=    71 4< "5,
)/ &  + &   3 $4=      4< "5,
)2 &  + &   3 $4=    71  4< "5,
)# 8 +  3 $48;1
+ 4< "#%5,
%" 8 +  3 $48;1
-71 4< "#5,
%  8  +  3 $48;1
+  4< 7< "#%5,
% 8  +  3 $48;1
-71  4< "5,
% ' + '
 3 $41 45,
%) '  + '
 3 $41  45,
%% ' + '
 3 $4171 45,
%. '  + '
 3 $4171  45,
%/ 86  + 
; 3 $4+  ; 4<  5,
%2 86  + 
; 3 $4-71  ; 4<  5,
%#  + 0
 3 $4
 0   0 45,
."
 + 0
 3 $4
0   0 45,
.  
+  + 
   3 $46 
  + 4< 7< %"5,
. & + &   3 $4&   1
 4< 7< ""%5,
.
.)!!* 01 
.%
..
=3'
95 + 
01 3 $4
  0145,
./
63'
95 +  0
 3 $4
0  45,
.2
63'
95 +  0
 3 $4
0 71 45,
.#
6 3'
95 +  0
 3 $4
0  45,
/"
6 3'
95 +  0
 3 $4
0 71 45,
/ 
 + 0
 3 $4
0  45,
/
 + 0
 3 $4
0 71 45,
/
  + 0
 3 $4
0   45,
/)
  + 0
 3 $4
0 71  45,
/% 3'
95 +  0
 3 $4
 0  45,
/. 3'
95 +  0
 3 $4
 0 71 45,
//  3'
95 +  0
 3 $4
 0   45,
/2  3'
95 +  0
 3 $4
 0 71  45,
/# &3'
95 +  &   3 $4&    45,
2" &3'
95 +  &   3 $4&   71 45,
2  & 3'
95 +  &   3 $4&     45,
2 & 3'
95 +  &   3 $4&   71  45,
2 &3'
95 +  &   3 $4   
 45,
2) &3'
95 +  &   3 $4   
  45,
 2% & 3'
95 +  &   3 $4  
71 45,
2. & 3'
95 +  &   3 $4  
71 45,
2/  +  3 $4* 4< "5 ,
22
2#!!
  011

#"
#  '
=$ ' 3
=5 > 
=135,
#
#
 
:+ $ ' 3
=5 > 
=13:+ ?5,
#)
 
:-71$ ' 3
=5 > 
=13:-71?5,
#%
 
6 + $ ' 3
= 5 > 
=136 + ?5,
#.
 
6 -71$ ' 3
= 5 > 
=136 -71?5,
#/
#2!!+ 
##
"" :+ 
0$ 3'
95 > :+  @ :+ ?3'
95,
"  :+ *90 $  %,
"
"!!-71 
")
"% :-71
0$ 3'
95 > :-71 @ :-71?3'
95,
". :-71*90 $  %,
"/
"2!!+  
"#
" 6 + 
  $ 6 +  > 
+ ,
 6 + &  $ 6 + & > :+ &  &,
  3'
9  '
65 > 3'
9  '
65,
  3'
65 > 3'
65A  ,
) 6 +  > 
3 5,
% 6 + ? @ 6 +  >  ,
.
/ 6 + -1$ ' 36 + 15 > -1
 36 + < 6 + &< 6 + ?5,
2 6 + $ ' 386 5 > 
* 36 + < 6 + &< 6 + ?5,
# 6 + * @ 86  > ,
"
 !!-71  

 6 -71&  $ 6 -71& > :-71&  &,
)  3'
9  '
65 > 3'
9  '
65,
%  3'
65 > 3'
65  ,
. 6 -71 > 
3 5,

A.23
A.24
/ 6 -71? @ 6 -71 >  ,
2
# 6 -71-1$ ' 36 -7115 > -1
36 -71< 6 -71&< 6 -71?5,
" 6 -71$ ' 386 5 > 
* 36 -71< 6 -71&< 6 -71?5,
  6 -71* @ 86  > ,

!!=   1
)
%
8  $
 >  @
=3'
65,
.
/!!&   
2
# &34645 @ :+  > :+ & @ 34645,
)" &34645 @ :-71 > :-71& @ 34645,
)  & 34645 @ 6 +  > 6 + & @  34645,
) & 34645 @ 6 -71 > 6 -71& @  34645,
)
))!!=  ;   
)%
).   :+ $ ' 3&5 > *&   3:+ 5,
)/   :-71$ ' 3&5 > *&   3:-715,
)2   6 + $ ' 3& 5 > *&  36 + 5,
)#   6 -71$ ' 3& 5 > *&   36 -715,
%"
%  & > &3'
65,
% & > &3'
65,
% &  > & 3'
65,
%) &  > & 3'
65,
%%
%.!!8;

%/
%2 8 $ 8 @ & > &34645,
%# 8  $ 8  @ &  > & 34645,
." 871$ 8 @ & > &34645,
.  871 $ 8  @ &  > & 34645,
.
.
.)!!
0  

.%
.. :+ 
0$
6 > :+  @ :+ ? @
=,
./ :+ 
0$
 > 
3
65,
.2
 .# :-71
0$
6 > :-71 @ :-71? @
=,
/" :-71
0$
 > 
3
65,
/ 
/ 6 + 
0$
6  > 6 +  @ 6 + ? @
=,
/ 6 + 
0$
  > 
3
6 5,
/)
/% 6 -71
0$
6  > 6 -71 @ 6 -71? @
=,
/. 6 -71
0$
  > 
3
6 5,
//
/2
/#!!  B
2"
2   3'5 > '
* 3:+ < :+ &< :+ ?5,
2  3' 5 > '
* 36 + < 6 + &< 6 + ?5,
2  3'5 > '
* 3:-71< :-71&< :-71?5,
2)  3' 5 > '
* 36 -71< 6 -71&< 6 -71?5,
2% "2% @3
A
 5@3'A' 5@3:-716 -715>3
A
 5@3'A' 5@36 + :+ 5,
2.
2/-

A.25
A.26
 Appendix A

A.3. Control system codes

159
Appendix A

160

 

  
!!'  (  
 )  # *
+
,!!) !  -
% )    
 ./0  #1)12*
3 )     
 ./0  #1412*
$ )     
 ./0  #1512*
6 ) +    
 ./0  #17 12*
" ) ,    
 ./0  #1 ( 12*
 ) %    
 ./0  #1   8 12*
 ) 3    
 ./0  #1    8 12*
 ) $    
 ./0  #14  12*
+
, 5 
    
 . #15   12*
% 5  
    
 . #15  12*
3 5  
    
 . #15  12*
$ 5  +
    
 . #15  +12*
6 5  ,
    
 . #15  ,12*
" 5  %
    
 . #15  %12*
  5  3
    
 . #15  312*

!!9: 
+
,!!7 (8 
% 5 
; )   .6%3+,3$ ,+<9=&.6$ $6$<.) >. 3+2222*
3
$!! ( (8 
6 5  
; )   .,%$ $%,$6%%<9=&.%%+$,<.) >.+$+2222*
"
 !!7
 5  
; " "%  6 %3<.9=&.%,%$9+<.) > %,3222  ) +*

+!! (
, 5  +
; $,+  36+<.9=&. ,,9+<.) > +$$222  ) ,*
%
3!! ? 
$ 
  8
 . #15  8 12*
6
+"  .) @; "2 (
+  
 ; "*
+  .) A " 7 ) @ %"2 (

A.27
A.28
+ 
 ; %" !. B ,,3<9=&." + < ) 22*
++  .) A; %"2 (
+, 
 ; , B ."" %%%%%%3 < ) 2*
+% 
+3 
 ; "*
+$ 
+6
,"  .) % @; ,"2 (
,  5  ,
; "*
, 
, 5  ,
; 
  .) 3  ) %2*
,+ 
,,
,%
,3!! ?  8 
,$ 5  %
; ) %*
,6
%"!!89
%  5  3
; .",%3<) 2!.%$B) 2*
%
%9


 


 
 


 % 

# 
&
$
'!!
(

) 
(
  

(

* +  
,-./ .0.(.0./(.0.1.0..23&
4 +  # .(.&
5 &
"
    ,
#.
(6.3&

7


  

 
     

 ' ( ))*
+*,
-
$ .+ (
/0(*. *,1 "21 "" !3",4
% 5 """%4
3 .+ (
/0(*. *,1 "1 "" %!3",4
6 .   4
2 7 ( ))*8+*,
-
& .+ (
/0(*. *,1 "21 "" !3",4
" 7 ( ))*9*,
-
 .+ (
/0(*. *,1 "&31 ""!3",4
 7' ( ))*8*,
-
 7
$
%7

A.29
A.30
 Appendix A

A.4. Aspen Tasks, feed streams & efficiency

165
Appendix A

166



% 

&'%%(%)%&
*
+ ()    % ) ,  !-
' -./
0 1  2
 $/
" 1!  3/
# 1!  3/
4 1$!  3/
  15!  3/
  ,1(. 6 7( % ,18 1$8 19./
  ,() . 6 
 ) ,18 1$8 19./
 1) 6 () /
* 1):;!+/
+
07



 ' 


!(
 '')'*'(!
+
% )*   '
,  #-
. -/0
1 2  3
 &0
4 2#  50
$ 2#  50
6 2&#  50
" 27#  50
  ,2)/ 8 9)' ,2: 2&: 2;/0
  ,)* / 8 

,2: 2&: 2;/0
 2* 8 !)* 0
+ 2*<# %0
%
19

A.31
A.32

 


  

  &'  ( )) 

 *( +%
,,+-*.-/0 0 $ /1
# *( +%23.40  0 $$ /1
5
$ &    ( )) $1
6 *( +%23.40 $ 0 $ /1
!
7 &    ( )) $!1
  *( +%23.40  0 $ /1

2





 



 $
 %&

' () " *
+
, ( -

. /01 
 
 2
34 

"5/0156*
7
% 
 
 2
34 

"5
 %56*
8 #
 
 
 2
34 

"5#%56*
!
    
 2

"5 9

 56*
   % 2

"5 
:;56*
 #  %*
'
+ <

, # = #
*
.  > , = /01 >
%*
7  >  = 2#6*
8
!

A.33
A.34
 Appendix B

Paper I

171
Appendix B

Rabbani A, Rokni M, Hosseinzadeh E, Mortensen H. “Start–up Analysis of a PEM

Fuel Cell System in Vehicles”, International Journal of Green Energy, 2013; 11:
1–21.

172
 Paper I

Start–up Analysis of a PEM Fuel Cell System in Vehicles

Abid Rabbani1, Masoud Rokni1, Elham Hosseinzadeh1 and Henrik Hilleke

Mortensen2
1. Thermal Energy Section, Department of Mechanical Engineering, Technical
University of Denmark, 2800 Kgs. Lyngby, Denmark.
2. H2Logic A/S, 7400 Herning, Denmark.

ABSTRACT

Addressing large load fluctuation in automotive applications, dynamic analysis of a

Polymer Electrolyte Membrane Fuel Cell system is conducted here. Operations of a
comprehensive system–level control–oriented fuel cell model with all necessary auxiliary
components are demonstrated and simulation results for start–up scenario are
presented. It is shown that system stability is influenced by slow thermal management
controls. High loads at start–up affect voltage and system efficiency adversely. Cathode
inlet water levels are found to be adequate for humidification of recirculated fuel
stream. Liquid water at cathode outlet is considerably higher at high current density
start–ups, pertaining to water removal issues.

Keywords: Dynamic simulation, System modeling, fuel cell, PEMFC, water crossover.

1 Introduction
Hydrogen is expected to play a major role in future energy economy. Fuel cells have
shown a significant potential as an efficient solution for harnessing energy of hydrogen.
Therefore, fuel cells are recognized to be one of the future power supply systems. The
proton exchange membrane fuel cell (PEMFC) currently appears to be the preferred fuel
cell for a variety of mobile applications, mainly due to its relatively low operating
temperature, quick start–up, high power density and efficiency, system robustness and
low degradation due to corrosion.
Fuel cell operating requirements in vehicles are more inflexible than stationary
applications. These systems have to operate at varying conditions related to
temperatures, pressures, power load and humidity. All the auxiliary components
constitute the balance of plant (BoP). These auxiliary components, such as air and fuel
supply system which include air blower (compressor) and control valves, and the thermal
control system which comprises of heat exchangers, coolant pumps and air radiators are
essential for successful operation of the fuel cell system. Therefore, system level dynamic

B.1
Paper I

modelling becomes a useful tool in analysis of PEMFC systems. Such studies help in
improving system efficiency, design development and optimization of control strategies.
Very few researchers have reported on dynamic modelling of complete PEMFC systems
along with their BoP. Whereas, steady–state models of these systems are present in
abundance. A few others have studied these system components individually, where
most of the research focuses on the fuel cell or the stack itself.
A number of PEM fuel cell models are available in the literature. (Ceraolo, Miulli
and Pozio, 2003) developed a simplistic dynamic model based on cathode kinetics.
(Amphlett et al., 1996) extended their previous steady–state model and presented a
generalized transient model. Another bulk dynamic model catering to inverter load
effects is presented by (Yerramalla, Davari and Feliachi, 2002). (Pathapati, Xue and
Tang, 2005) developed a complete fuel cell model that included effects of double charge
layers at cell sites along with mass and energy balance equations for the gases and the
dynamics of flow and pressure in the channels. (Jia et al., 2009) also followed suit and
developed a dynamic model in MATLAB/Simulink to investigate fuel cell transient
electrical responses under various operating conditions. Though these models provide
valuable understanding of intricate fuel cell operations, they are mostly aimed at design
and optimization of individual cell components and operating parameters without going
into system level analysis.
Multi–dimensional models on the other hand, disseminate complex characteristics of
reactant flows and charge transportation within fuel cells. (Hu et al., 2004) represented a
three–dimensional computational PEM fuel cell model with conventional and inter–
digitized flow fields. Another three–dimensional mixed–domain PEM fuel cell model of
(Kim et al., 2009), which integrates intrinsic transport mechanisms, has been applied to
investigate effects of the fully coupled transport phenomena. These models might be
useful in predicting cell or stack performance, however due to high computational times
and limitations to integrate with BoP components, these models are not suitable for
system–level modelling.
Of all publications in open literature, relatively few studies account for control–
oriented PEMFC models. (Pukrushpan, Peng and Stefanopoulou, 2004) presented a
transient dynamic model, which includes in–blower flow and inertia dynamics,
manifold filling dynamics, reactant partial pressures and membrane humidity. A
control–oriented thermal dynamic model is also proposed by (del Real, Arce and
Bordons, 2007). (Ahn and Choe, 2008) investigated temperature effects on the overall
system performance and suggested strategies for coolant controls. Issues related to
temperature dynamics are also dealt and studied by (Vasu and Tangirala, 2008), which
could predict the effects of temperature and feed flows on system transient behaviour.
From a control perspective, it is important to develop a complete model that emulates
the real behaviour of PEMFC stack when subject to varying operating conditions.
Therefore, a need to develop a system–level model is identified that responds to
dynamic effects of all the necessary BoP components which is lacking in the
aforementioned studies.

B.2
 Paper I

While some researchers such as (Shan and Choe, 2006) and (Park and Choe, 2008)
presented a transient stack model and analysed the temperature distribution on fuel
cells, management of stack temperature by coolant flow controls becomes vital in
efficient operations and reliable performance of the fuel cell system. (Khan and Iqbal,
2005) proposed a transient model to predict efficiency in terms of voltage output and
heat losses by including heat transfer coefficients for the stack and an energy balance.
(Jung and Ahmed, 2010) developed a thermal stack model based on real–time simulator
in MATLAB/ Simulink environment. (Asghari, Akhgar and Imani, 2011) also designed
a thermal management system for a PEMFC which is oriented towards the flow fields
within the stack. Again these studies concentrate on the stack level without looking into
dynamic characteristics that are influenced by auxiliary components.
Start–up behaviour of PEMFC stacks at sub–zero temperatures has been studied and
reported by few researchers. (Youcai Li et al., 2011) conducted experiments and
validated their model based on the results. (Yan et al., 2006) investigated effects of sub–
freezing temperatures on fuel cell performance and start–up. A model for freeze start is
also developed by (Mangold et al., 2011) and compared to experimental data. (Meng,
2008) and (Sundaresan and Moore, 2005) both proposed thermal models for cold start
of PEMFC. However, start–up behaviour of a stack above the freezing temperatures
which accounts for multiple varying operating conditions and changing auxiliary
components outputs has not been reported to a greater extent.
Building on these modelling studies, a complete control–oriented system–level
dynamic model is proposed in this paper which incorporates all necessary BoP
components for a PEMFC fuel cell stack by including electrochemical, thermal, feed
flows and water transportation models and a detailed control strategy. In totality, this
work differs from previous studies such that here a sizable focus has been set for (i)
comprehensive control strategy regulating hydrogen and air feed flows, coolant inlet and
fuel cell stack operating temperatures, (ii) thermal model with liquid coolant circuit
incorporated in it which takes into account reactant and product phase changes, (iii)
Thermal management including effects of coolant controls and heat exchangers on fuel
cell stack performance, and (iv) Water management and effects of water crossover on
anode recirculation loop and fuel cell relative humidity.
Thus the objective of this study aims at design and modelling of a complete system
that incorporates all the essential auxiliary components with comprehensive control
strategies in order to emulate a real PEMFC system. Here, the focus is on the start–up
sequence of the proposed system.

2 System Configuration
Figure 1 shows the schematics of fuel cell system analysed in this study. It includes all
the components in the system, such as PEMFC stack, air blower and humidifier, pumps,
heat exchangers and radiator for the cooling circuit, flow valves and controllers.
Compressed air is cooled and humidified before entering the cathode side of the stack.
On the other side, pressurized hydrogen from storage tank is fed to the anode of fuel

B.3
Paper I

cell. Sin
nce a higheer fuel stoichiometry is maintain ned, anodee exhaust iss recirculatted back
to the ffeed stream
m via a recirrculation puump.
Heatt produced d in the staack is absoorbed by th
he coolant which circculates in a circuit
associatted with th
he stack and a heat exxchanger. An A externaal cooling lloop, conneected to
the aforrementioneed heat excchanger, in n turn coolss the water in the inteernal circuit.

FFigure 1. Scchematic of a complete PPEM fuel ceell system with auxiliary componentss.

This ciircuit also comprises of a heat exchangerr to precoo ol the blow wn air and d an air
radiatorr for heat rejection
r frrom the syystem. Floww of water is
i regulatedd by pump ps in the
respectiive circuits. In order to ensure a stable op peration of the system
m, a control system
is needded to regu ulate flows, temperattures and pressures
p in
i the feedd streams, cooling
circuits and plan nt componeents. Therrefore, in the t presentt study, PIID controlllers are
deployeed to emulaate the systtem in reall time and thereby an nalyse respoonse characcteristics
under vvarying opeerating con nditions. A Aspen Plus dynamics (Aspentechh 2012) co omprises
of builtt–in PID controllers
c and manippulators. Combinatio
C ons of thesse components are
used too devise co ontrols for entire systtem. Thesee controlleers collect data from various
compon nent inlet and outleets which are regarded as presssure, tempperature an nd flow
transmiitters, and manipulatte the correesponding componen nts to reachh the desireed state.
Due too this fact,, current system
s respponses and its behaaviour are attributed d to the
formulaated contrrol strateggy which is based on fuel cell stackk limitations and
recomm mendationss by the manufactuurer. Com mponent properties
p related too these
limitatiions are disscussed in the followiing subsecttions whereeas intricatte details of system
controlls are dealt in the relevvant sectioon later.

B.4
 Paper I

2.1 Fuel Cell stack

In this study, a Ballard fuel cell stack (Ballard Mark9 SSL, 2008) has been specifically
adapted. The model, which is based on equations adopted by (Hosseinzadeh and Rokni,
2012), contains some parameters which are attributed to the physical characteristics of
the system, as well as on operating conditions and membrane properties. Some of these
parameters which are obtained below in this section for the Ballard stack are shown in
Table 1. Current research aims to build up a system which meets the requirements of
actual stack running under recommended conditions.

Table 1. Parameter estimation for Ballard Fuel cell stack

Parameter Value
Number of electrons transferred per mole of
fuel, n e mole mol fuel  2
Number of electrons for the reaction rate, nel 1 for cathode and 4 for
mole mol fuel  anode
Internal current density, in A cm 2  0.002
Symmetry factor,  0.5
Membrane thickness, t m cm  0.0183
Density of the membrane–dry condition,  dry
3.28
g cm 
3

Molecular weight of membrane, M m Kg mol  1.1

During normal steady state operation, reactant pressure should be above coolant
pressure whereas for short transients and during start–up, coolant pressure may exceed
reactant pressure. It is recommended to operate the anode at a higher pressure than the
cathode. Nitrogen crosses over from the cathode to anode as cathode pressure increases
relative to the anode pressure. Ensuring that cathode pressure is lower than the anode
pressure will minimize nitrogen crossover and improve cell stability. In a system with
fuel recirculation, the consumption will be slightly above 1.0 stoichiometry as 1% to 2%
of the flow will be required for purging. In the present study, purge occurrences are
neglected. For a maximum service life and efficiency of the stack, Ballard recommends
operating conditions within which the stack should operate, (Ballard Mark9 SSL, 2008),
which are followed in this study. Figure 2 reports recommended reactant inlet pressures
and nominal pressure drops within the fuel cell stack.

B.5
 Paper I

Anode Pressure drop Cathode Pressure drop H2 Inlet pressure Air Inlet pressure

0,6 2,5

0,5 2,0
∆P (bar)

P (bar)
0,4 1,5
0,3
1,0
0,2
0,5
0,1
0 0,0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
(a) I (A) (b) I (A)

Figure 2. Stack data provided by Ballard (a) Nominal pressure drop within the fuel cell, (b) Inlet
pressures for reactants.

2.2 Humidifier
One of the key issues in PEMFC is the dehydration in the membrane. A fully
hydrated membrane supports the ionic crossover between the electrodes, as well as
extends its life. Therefore, reactants in the PEMFC need to be humidified before
entering the stack. In the present model, a humidifier utilizes the water produced by
chemical reaction inside the fuel cells to humidify the inlet air. A simple model based
on mass and energy balance is implemented here. Although some empirical models
could be considered, these vary over a wide range depending upon the types of
humidifiers used. Due to the lack of sufficient data, it is assumed that the inlet air is
optimally humidified after passing through the humidifier. The relative humidity of air
entering the cathode is set to 95% in the calculations; although other values can be
chosen. This assumption could be justified as it is very close to the real operational
conditions. On the anode side, there is no humidifier and the fuel is humidified by
means of water crossover and its recirculation. It is observed that water cross–over from
cathode to anode through the membrane is adequate enough to raise the relative
humidity in the anode exhaust to 100%. This recirculation, when mixed with inlet
stream delivers the desired humidity into the anode of the fuel cell.

2.3 Reactant feed systems

The fuel cell system is composed of blowers, pumps and valves that regulate flows of
material streams. A valve is placed between the hydrogen tank and inlet manifold of
anode which enables or disables the hydrogen supply. This regulatory valve adjusts high
hydrogen pressure of the tank to the desired operating pressure of the fuel cell. Since
the system does not operate on dead–end mode, the amount of hydrogen regulated by
this valve equals the stoichiometric hydrogen required of the fuel cell. Stoichiometric

B.6
 Paper I

ratio is defined as the amount of reactant supplied to the amount which is consumed in
the reaction.
Flow and pressure of oxidant into the cathode is regulated by the blower. The
amount of stoichiometric oxygen for the fuel cell reaction is manipulated by a controller
which regulates the electrical power of the blower, and hence compression and air flow.

2.4 Heat Exchangers and thermal management system

As can be seen in Fig.1 a network of heat exchangers constitute the thermal
management of the prescribed system. They extract the heat produced from the cell
stack and maintain the selected operating temperature which is essential for
performance and durability of the fuel cell. Although heat exchanger models used here
are predefined in Aspen Dynamics, some of the parameters have been assumed on the
basis of media entering the hot and cold sides of these heat exchangers. The heat
exchanger, which is connected to the internal cooling loop, has liquid water on both its
hot and cold side. Therefore, a UA value of 1.0 kW/K is assumed. Whereas, UA values
for air pre–cooler and radiator are approximated to be 0.05 kW/K and 0.3 kW/K
respectively. In performing the simulations, the pressure drop was assumed to be 0.05
on both sides of heat exchangers. The corrected LMTD is calculated in addition to the
corresponding inlet and outlet temperatures of hot and cold streams.

2.5 Pumps and Blowers

In the above proposed system, air blower, anode recirculation pump and water
pumps are one of the BoP components which are also regulated by the control system.
Aspen Dynamics TM contains models of these units in its library as well. Since the
nominal power of the PEMFC is only 21 kW, mass flow rates of fuel and air are very
low. Therefore, very low values of isentropic efficiencies are suggested in this paper. The
efficiency of a blower ranges from 15% to 40% in the calculations, depending on the air
mass flow. Calculated pump efficiencies are also very low for the cooling water circuits
and are determined to be around 70%.

3 Methodology
The characteristics of the PEMFC system described above are implemented in Aspen
Plus Dynamics TM (Aspentech 2012) which is a simulation tool for process modelling
and energy system analysis. It is a component based simulation tool with a built–in unit
operation modules library. The component library includes models of heat exchangers,
reactors, turbo machinery, decanters and separators, pressure relief valves, controllers
along with utility components and manipulators. These components are built upon by a
standard set of equations representing their physical properties, such as isentropic and
mechanical efficiencies in blowers and turbines. Also, user–based–custom–models can
be imported into the program.

B.7
Paper I

In Aspen plus Dynamics, the physical model of the complete system is devised by
connecting the relevant component models through work, heat or material streams and
by defining operating conditions for the system. The numerical solvers then convert
these inputs into set of dynamic mathematical equations to be solved simultaneously.
These equations include mass and energy conservation for all units and connecting
streams, as well as relations for thermodynamic properties of the fluids involved. In
addition, the program has the ability to run steady–state, initialization and dynamic
simulations for the same model by changing model or module specifications.
Occurrences of disturbances and operation constraints can be enforced by adding
control modules to the constructed system. An implicit integrator ensures stable
solution of the dynamic simulation and varies the integration step to ensure simulation
accuracy.
The PEMFC stack model presented in this study is based on a model developed by
(Hosseinzadeh and Rokni, 2012). They built a steady–state model for the same Ballard
fuel cell stack (Ballard Mark9 SSL, 2008) which is under investigation in the present
study. That study presented the effectiveness, validation and reliability of the model at
different operating conditions and conducted sensitivity analysis for effects of coolant
temperatures, humidity and reactant stoichiometry on system operations. However,
system behaviour at start–up and varying operating conditions was not studied, which
would have provided a detailed insight into transient performance of the system. In
contrast to their steady–state model, here a dynamic model of the prescribed system is
developed in Aspen Custom Modeller. This model constitutes equations for fuel cell
electro–chemistry, mass and energy balances, water crossover in membrane and fuel cell
thermodynamics. The model is implemented into ASPEN Plus Dynamics and system
controls are implemented in order to ensure stable operation of the plant during load
changes.
The thermodynamic efficiency and net power of the system are determined by the
current drawn and voltage produced by the stack. The average cell voltage of a fuel cell is
defined by an analytical expression:
Vcell  E  act  ohmic  conc (1)
where E is the theoretical voltage, act the activation overpotential, ohmic ohmic
overpotential and conc denotes concentration loss. Theoretical voltage is usually
expressed by Nernst equation (Spiegel, 2007):
 g 0f RT  aH 2O 
E  ln (2)
ne F ne F  aH 2 aO2 0.5 
where a is the activity of the species. Here the assumption of all reactants being ideal
follows that the activity of the gases is equal to their partial pressures and the activity of
liquid water is equal to 1. Then we have:
E
 g 0f
ne F

RT
ne F
 1
ln PH 2 PO2
0.5
 (3)

B.8
 Paper I

ne is the number of electrons transferred per mole of fuel which here is hydrogen in the
present case. Therefore, ne  2 is set; according the reactions taking place on the
cathode side.
Change in Gibbs free energy g 0f for the reaction below, is calculated at standard
pressure but is still a function of temperature.
g 0f  g 0f   H 2O
 
 g 0f H2

2
 
1 0
gf O2
(4)
The total activation losses in the cell are equal to sum of anode and cathode
contributions. Knowing this and assuming equal transfer coefficients in both electrodes,
the Butler–Volmer equation is simplified as:
RT  i  in  RT  i  in
 


 act   act ,c   act ,a  ln   F ln  i
(5)
 c F  i0 ,c  a  0 ,a


According to the base model, the value for internal current density, in is assumed to be
equal to 0.002 A / cm 2 . The equations below are valid for evaluating the transfer
coefficients on the anode and cathode side respectively:
 a    nel (6)
 c  1     nel (7)

The symmetry factor,   0.5 is chosen, nel is equal to 4 for anode and 1 for cathode,
see (Santarelli, Torchio and Cochis, 2006).
For exchange current density i0 , an analytical expression is chosen here, which
predicts the value of exchange current density at the anode and cathode separately.
 1     nel  F  E 
i0,a  nel  F  k a  exp  (8)
 RT 
    nel  F  E 
i0,c  nel  F  k c  exp  (9)
 RT 
The net water transport through the membrane is a combination of these two effects;
electro osmotic drag and back diffusion and is given as:
i  dry d
J H O  2ndrag  D (10)
2
2F Mm dz
Another parameter needed for calculation of net water flux is water diffusion which is a
function of membrane water content,  . Here, the expression suggested by (Springer,
Zawodzinski and Gottesfeld, 1991) is used:
 1 
 1
D 4  10 6 exp2416   
   2.563  0.33  0.02642  0.0006713  (11)
  303 T 
The total energy into the fuel cell is consumed by the electrical power output, heat
removed by the coolant, heat loss at the stack surface and energy stored within the stack
itself. In the current model, a lumped thermal model proposed by (Khan and Iqbal,
2005) is considered. The stack is regarded as a single thermal mass with a heat capacity.

B.9
Paper I

With the assumption of stack temperature being equal to the coolant temperature at the
outlet, heat exchanged with the coolant and hence stack operating temperature could be
determined. The energy balance between the above–mentioned modes can be given by:
C t  dT  Pin  Pout  Pel  Q loss (12)
dt
where, Ct is the thermal capacitance of the stack, Pin is the total power delivered by the
fuel to the stack (kW), Pel is the power consumed by the electrical load (kW), Pout is the
heat transferred to the cooling water circulating in the stack (kW), and Q loss is the heat
dissipated to the ambient (kW).
The subordinate components in the BoP, i.e. anode recirculation and water pumps,
air blower, mixers and heat exchangers are modelled using the default mathematical
models provided in Aspen Plus Dynamics.

4 Control System
This section presents the methodology of controlling system parameters and
operating conditions for the system to have a stable operation. Classic proportional–
integral (PI) controllers, which are widely used in industrial control systems, are
employed to regulate different components and flow streams. Principally, these
controllers calculate "error" value as the difference between a measured process variable
and a desired set–point, and attempt to minimize this error by adjusting the process
control inputs. Aspen Plus Dynamics has built–in PID controllers with options of
specifying process and output variable ranges, tuning and filtering of controls and
selection of ideal, series or parallel algorithms. Classical notation for output of the ideal
PID controller employed in this study is specified as:
 1 t d 
OP  K p  e(t )   e( )d  Td e(t )  (13)
 Ti 0 dt 
where, OP is the controller output, K p is the proportional gain which is set different
value for different components. Ti and Td are integral and derivative times respectively
and e is the calculated error between set point and processed variable at instantaneous
time t.  is the variable of integration taking values from 0 to present time. Control
strategies incorporating these PIDs and regulatory mechanisms for different operating
parameters are discussed accordingly.
Key parameters to be controlled in the proposed system are reactant inlet
stoichiometries, inlet pressures, coolant inlet and operating temperatures of the stack.
The control loops of hydrogen and air supply system should maintain the optimal
reactant ratio and prevent shortages that might occur during abrupt changes in the load
current. It is reported in (Hosseinzadeh and Rokni, 2012) that the aforementioned
PEMFC system is not sensitive to reduction in fuel flow during load fluctuations;
however if the hydrogen stoichiometry is below 1.0, then fuel starvation can cause
irreversible damage to the stack. On the other hand, oxygen starvation affects the fuel

B.10
 Paper I

cell to a greater degree and has a much larger effect on stack performance. Hence, an
adequate air supply is required.

H2 Stoichiometry Air Stoichiometry Coolant Delta T

7 12
6 10
Stoichiometric Ratio

5
8
4

∆T (◦C)
6
3
4
2
1 2

0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
(a) I (A) (b) I (A)

Figure 3. (a) Reactant stoichiometry used in the fuel cell control system, (b) Stack inlet and outlet
temperature difference maintained by coolant mass flow.

Figure 3(a) shows the fuel and air stoichiometries for the selected fuel cell stack which
are proposed by Ballard and in Fig. 3(b), recommended temperature difference between
stack inlet and outlet is displayed. It can be seen that at low current loads, high amounts
of excess reactant flows are desired. This is due to the fact that at low power
consumption and low pressures, water is formed by the reaction in the cathode side of
the cells and it needs to be ejected out of the stack, which is done by supplying high
amounts of air. While the amount of oxygen consumed depends on the stack current,
the amount of oxygen supplied to a fuel cell is directly related to the blower power.
Here, data from pressure and flow sensors is transmitted to the controller where an
algorithm based on Fig. 3(a) translates it to be the process variable and set point for the
PI controller, which regulates the blower power in order to maintain the desired oxygen
ratio. Similarly, an algorithm for controlling hydrogen flow is devised along with a PI
controller, which regulates the control valve opening for optimal fuel supply at the
desired pressures. Pressures in anode recirculation loop are governed by PI controllers
which collect data from anode upstream and manipulate recirculation pump power
accordingly. Figure 4 elucidates on processed and manipulated variable for different
control blocks implemented in the proposed system.

B.11
Paper I

Figure 4. Control blocks for the PEMFC system.

Thermal management in PEMFC systems is of vital importance, basically due to the

fact that heat produced in the selected fuel cell cannot be dissipated by convection and
radiation through the stack surface. A consistent and stable operation of around 70 °C
thus requires a liquid cooling system. Since the operating temperature of the fuel cell is
not very high, a low temperature difference with the ambient requires having a large
heat transfer surface. Therefore, an efficient thermal control system becomes of
substantial importance to ensure optimum system performance.
As shown in Fig. 4, the cooling system for the proposed fuel cell consists of internal
and external cooling circuits. As mentioned earlier, the coolant mass flow rate defines
the variance in the stack temperature or simply maintains the fuel cell operating
temperature. In this case, temperature in the stack can be controlled by the coolant flow
rate which acts as an input signal and is adjusted by the PI controller. Based on data
from Fig. 3(b), equations defining stack temperature as a set–point for controller are
developed. In addition, the controller simultaneously collects data from temperature
transmitter at the coolant outlet stream, which then changes the coolant flow
accordingly by sending output signals to the driving pump. Employment of such
algorithm in the system controller ensures a stable operation under normal steady–state
conditions, however for system start–up scenario, a different approach is required. This
is due to the fact that, at start–up sequence, temperature of the stack is equal to the
ambient and the above–mentioned control strategy would not allow the stack
temperature difference to increase more than 10 °C and hence increase the time to
reach steady–state operations, which is not desirable. In this regard, at start–up
sequence, the coolant flow is restricted to a minimum value until the desired stack

B.12
 Paper I

operating temperature is reached which is when the steady–state controls begin

functioning. Though, a lower coolant flow is needed to increase the stack heat–up time
at start, a very high temperature difference may cause damage to the fuel cells in the
stack. Henceforth, the maximum temperature difference is limited to 17 °C for all
scenarios here.
Temperature of the coolant entering the stack can similarly be controlled by the flow
of water in the external circuit. Control signal to the associated pump regulates the
electrical power of the pump and hence the coolant inlet temperature into the stack. In
a similar fashion, temperature of water in the external circuit is dependent on radiator
fan speed. PI controllers are used to regulate the fan speed as well.

5 Simulations
Reliability of the suggested model is verified by running simulations at different
power loads when the system reaches steady–state. Fig. 5 depicts adaptation of the
devised model which corresponds to the polarization curve obtained from the data
provided by the manufacturer (Ballard Mark9, SSL 2008) associated with the given
PEMFC stack. As suggested by the manufacturer, stack temperature range of 60–70°C
has been used in simulations of the current system. Model characteristics of the same
system at higher temperatures have been studied in (Hosseinzadeh and Rokni, 2012),
and show a good agreement with the stack data.

Experimental Calculated
1.000

900
Vcell (mV)

800

700

600

500
0,0 0,2 0,4 0,6 0,8 1,0 1,2
i (A/cm2)

Figure 5. Comparison of calculated polarization curve compared to the corresponding data from
the manufacturer (Ballard).

Start–up sequence
In case of PEMFC, dynamic behaviour of a start–up is of particular importance to
ensure a short start–up time and an efficient operation. Fuel cell start–up sequence
could be defined as the time required by a system to reach stable conditions, which is
achieved when the voltage and consequently power, reach a stationary value. Normally,

B.13
 Paper I

the optimal operating temperature of a PEM fuel cell is reported to be in the range of
60–80°C. Within this range, chemical reactions occurring inside the fuel cell are
relatively fast and facilitate removal of water produced by these reactions. Whereas, at
low temperatures, there is a considerable rise in kinetic and ohmic losses in addition to
reactant transport losses caused by a high rate of water condensation (Park and Choe,
2008). Thus, it is necessary to elevate the working temperature of a stack as quickly as
possible to meet the demands of the required load power.
When the fuel cell is in standby stage i.e. when no current is drawn, auxiliary
components such as air blower, coolant pumps, radiator fan and hydrogen inlet valve
are closed. At system start–up, all of these components are switched on and fuel valve is
opened. Power from the stack is only drawn when current load is applied on the stack.
In the current simulations, values of stream flows are initialized to be non–zeros. In this
way, a realistic start–up sequence could be imitated and furthermore initial zero–
condition in Aspen Dynamics could be avoided. Initial temperature of the stack is
assumed to be equal to ambient temperature (25°C) and a start–up power load is
introduced by ramping of current at a rate of 20 A/sec. Other parameters and operating
conditions are selected from section 3.

5.1 Thermal management of the system

Different start–up strategies are compared and there results are summarized below.
Currents at 40, 60, 80 and 100 amperes are drawn and heat–up times for these four
cases are presented in the Fig. 6 At start, the flow in cooling circuits is set to the lowest
since it is desired to raise the stack temperature to its optimal operation. Coolant flow
in the internal circuit is fixed to 290 kg/hr as recommended by the manufacturer. Flow
in the internal cooling circuit increases, once the stack temperature difference increases
by 15 °C. Flow in external circuit is regulated to maintain stack inlet temperature of
internal cooling circuit. Air flow in radiator is started to maintain temperature of
external cooling loop to around 50 °C.
1000.0

1000.0

(a) At Current 40A (b) At Current 60A

80.0

80.0
40.0 50.0 60.0 70.0

40.0 50.0 60.0 70.0

External Coolant flow (kg/hr)

External coolant flow (kg/hr)

Internal coolant flow (kg/hr)

Internal coolant flow (kg/hr)

Coolant temperature (C)

Coolant temperature [C]
750.0

750.0
Air Radiator flow (kg/hr)

Air Radiator flow (kg/hr)

500.0

500.0
250.0

250.0
30.0

30.0
0.0

0.0

0.0 200.0 400.0 600.0 800.0 1000.0 1200.0 0.0 150.0 300.0 450.0 600.0 750.0
Time (Seconds) Time (Seconds)

B.14
 Paper I

1000.0

1000.0
(c) At Current 80A (d) At Current 100A

80.0

80.0
40.0 50.0 60.0 70.0

40.0 50.0 60.0 70.0

External coolant flow (kg/hr)

External coolant flow (kg/hr)

Internal coolant flow (kg/hr)

Internal coolant flow (kg/hr)

Coolant temperature (C)

Coolant temperature (C)

750.0

750.0
Air Radiator flow (kg/hr)

Air Radiator flow (kg/hr)

500.0

500.0
250.0

250.0
30.0

30.0
0.0

0.0
0.0 200.0 400.0 600.0 0.0 150.0 300.0 450.0
Time (Seconds) Time (Seconds)

Figure 6. Stack heat–up times and corresponding coolant and radiator air mass flows: (a) at start–
up current of 40A (b) at start–up current of 60A (c) at start–up current of 80A (d) at start–up
current of 100A.
As can be seen in the figure below, when the stack is started at current of 40A at a ramp
rate of 20 A/s, it takes approximately 425 seconds to reach stable operating
temperature. The heat–up time for 60, 80 and 100 A is determined to be 300, 220 and
175 seconds respectively. The heat–up time becomes shorter as the applied current is
increased. Because of the high current density, a large amount of heat is produced in a
short time and the temperature rises rapidly through the cells.

5.2 Power generation and system efficiency

Generally, stack efficiency is higher to that of the complete system with BoP.
However, in a system with fuel recirculation, stack efficiency is attributed to fuel
stoichiometry, though there is no considerable effect on the system efficiency. When the
actual and stoichiometric mass flow of hydrogen differ least then less fuel is wasted and
stack efficiency is at its maximum. The same argument could be generalized for system
efficiency as well. This is because at higher air mass flows, power consumption of
auxiliary components increases.
Different start–up cases were investigated to analyse the system efficiency, net power
output and auxiliary power consumptions. Currents of 60 and 100 A are applied for
two start–up scenarios. Overall efficiency of the system is as much affected by blower
and radiator fan, as by the fuel cell stack itself. Fig. 7b, shows that at high current start–
ups, power generated from the stack increases. However, efficiency of the system
decreases. As can be seen in the figure, system efficiency decreases from 55% to 50%
when applied current is changed from 60 to 100 A. During start–up, power
consumption of coolant pumps and air radiator is very low when compared to that of
the air blower (Fig. 7c). Therefore, auxiliary power consumption is dominated by the air
blower. It is observed that power consumed by the air blower increases during start–up

B.15
Paper I

and becomes constant after a few seconds. Also, power produced by the stack increases
at a sound pace and becomes stable once steady state is reached. This eleviation in the
stack power is due to the fact that the operating temperature of fuel cell increases till it
reaches the desired optimal temperature.

(a) System efficiency at 60A (b) System efficiency at 100A

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0

0.8

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0

0.8
0.7

0.7
Auxiliary Power (kW)

Auxiliary Power (kW)

Stack Power (kW)

Stack Power (kW)

System efficiency

System efficiency
0.6

0.6
0.5

0.5
0.4

0.4
0.3

0.3
0.0 100.0 200.0 300.0 400.0 500.0 600.0 700.0 0.0 75.0 150.0 225.0 300.0 375.0 450.0
Time (Seconds) Time (Seconds)

(c) Power consumption at 60A (d) Power consumption at 100A

1.0

1.0
0.8

0.8
Air Radiator (kW)

Air Radiator (kW)

Air Blower (kW)

Air Blower (kW)

Pumps (kW)

Pumps (kW)
0.6

0.6
0.4

0.4
0.2

0.2
0.0

0.0

0.0 100.0 200.0 300.0 400.0 500.0 600.0 700.0 0.0 75.0 150.0 225.0 300.0 375.0 450.0
Time (Seconds) Time (Seconds)

Figure 7. System efficiency and stack power (a) at start–up current of 60A (b) at start–up current
of 100A, auxiliary power consumption at start–up of 60A (c) and 100A (d) respectively.

As seen in Fig. 7c and 7d, radiator fan is shut off at the start sequence, but is activated
once temperatures in the external cooling circuit are high enough to be cooled down.
Air flow in the radiator can be regarded as a linear function of its electrical power and
therefore once it is started by the controller, the system efficiency decreases. Since, at
high temperatures heat production in the stack is quite high, the radiator fan consumes
around 250 W for a current withdrawal of 100 A, whereas it only requires
approximately 125 W when current is at 60 A.

B.16
 Paper I

5.3 Cell voltage

Potential difference generated across the fuel cell is dependent on factors such as
current density, reactant partial pressures and temperature at which cell reactions occur.
Figure 8 elucidates some of these underlying factors which affect the voltage and
consequently power output of the stack. Here, during the PEMFC start sequence,
current is ramped to 80 A at a rate of 20 A/s. Since current density and reactant
pressures are regulated at very fast rate, they follow a constant profile along the time
axis. However, the stack temperature, which is attributed to the coolant mass flows,
keeps on increasing until it reaches the desired operating temperature. As can be seen in
Fig.8a, cell voltage increases with the operating temperature. This fact is due to the
reason being fast reaction kinetics at the electrodes of individual cell sites. It can be
observed that the voltage reduces abruptly when simulation time is around 160 seconds.
At this stage, the decrease in cell temperature and voltage is caused by the sudden
increase in coolant flow into the stack, which maintains the stack operating
temperature. In a similar manner, power produced by the stack also decreases with
voltage drop, although it is not that distinct in Fig. 8a. Nevertheless, Fig. 8b shows a
detailed view of the aforementioned instance of time. Evidently power produced by the
stack responds to a similar trend to that of voltage against operating temperature of the
stack.

(a) Start-up to steady-state (b) Close up: controller effect

0.85

80.0

10.0

70.0
0.3

7.0
0.8

6.9
40.0 50.0 60.0 70.0

0.79

50.0 55.0 60.0 65.0

0.25

Operating temperature (C)

Operating temperature (C)

Average Cell Voltage (V)

Current Density (A/cm2)

7.5
0.8

6.7 6.8
Stack Power (kW)

Cell voltage (V)

0.76 0.77 0.78

Power (kW)
0.2
0.75

5.0

6.6
0.15

6.5
0.7

2.5
0.1

30.0

0.75

45.0

6.4
0.65

0.05

20.0

0.74

40.0

0.0 50.0 100.0 150.0 200.0 250.0 300.0 125.0 150.0 175.0 200.0 225.0 250.0
Time (Seconds) Time (Seconds)

Figure 8. (a) Cell voltage as a function of stack operating temperature with start–up current of
80A, (b) Effect of sudden temperature drop on voltage and power.

5.4 Fuel and oxidant flows

This section caters analysis of reactant flows into the fuel cell and investigates their
effect on cell performance as well as on adjacent streams. Cell performance is not
sensitive to fuel stoichiometry; on the other hand, cathode is significantly more sensitive
to air stoichiometry. Therefore, simulations are carried out by employing operating
conditions and parameters mentioned previously (in section 3.1). A start–up current of

B.17
 Paper I

60 A is used in this case. Figure 9 specifically portrays flow behaviour of fuel and air into
and out of anode and cathode channels respectively.

(a) Fuel flow at anode inlet (b) Fuel flow at anode outlet
1.4

1.4
0.4 0.6 0.8 1.0 1.2

0.4 0.6 0.8 1.0 1.2

Anode outlet flow (kg/hr)
Anode inlet flow (kg/hr)
Hydrogen flow (kg/hr)

Hydrogen flow (kg/hr)

Water flow (kg/hr)

Water flow (kg/hr)

0.2

0.2
0.0 100.0 200.0 300.0 400.0 500.0 0.0 100.0 200.0 300.0 400.0 500.0
Time (Seconds) Time (Seconds)

(c) Air flow at cathode inlet (d) Air flow at cathode outlet
20.0

20.0
Cathode outlet flow (kg/hr)
Cathode inlet flow (kg/hr)

15.0

15.0
Oxygen flow (kg/hr)

Oxygen flow (kg/hr)

Water flow (kg/hr)

Water flow (kg/hr)

10.0

10.0
5.0

5.0

0.0 100.0 200.0 300.0 400.0 500.0 0.0 100.0 200.0 300.0 400.0 500.0
Time (Seconds) Time (Seconds)

Figure 9. Fuel and air flows at inlet and outlet of the stack: (a) Fuel flow at anode inlet, (b) Fuel
flow at anode outlet, (c) Air flow at cathode inlet and (d) Air flow at cathode outlet.
It is observed that fuel entering the stack contains water i.e. the fuel is humidified.
Since, fuel from anode exhaust is recirculated; it adds more unutilized hydrogen and
water to the inlet stream. Composition of hydrogen in the entrance becomes stable after
initial regulation. However, the amount of water keeps on increasing due to the increase
in stack’s temperature and water cross–over from cathode. This occurrence shows
similar incremental trends as in case of stack temperature in Fig 6. Anode outlet flows
show similar trends, though the amount of hydrogen at the outlet is reduced as it is
consumed in the cell reactions.
Air flow into the cell can also be seen to increase with time. As temperature of the
stack increases, more water is needed to maintain the humidity at the desired level; here,
a constant relative humidity of inlet air is assumed. Oxygen and nitrogen remain more
or less constant throughout the heat–up time. Once, the stack reaches its steady–state,

B.18
 Paper I

the air flow becomes constant. At cathode outlet, the total flow increases
correspondingly, however the oxygen content is lower as it is consumed in the reaction
as well, whereas water flow is increased. The additional water in the outlet is the product
of fuel cell reaction. Higher stoichiometries for both fuel and oxidant are maintained to
manage this water production in the cathode and to dilute nitrogen crossover to the
anode.

5.5 Air humidification effects

In PEMFC systems, sufficient hydration is essential to facilitate ionic conduction for
an efficient cell operation and to avoid physical degradation of the membrane. Cell
drying depends on a number of operating conditions including current density, reactant
flow rate, gas composition, relative humidity, inlet pressure, and cell temperature. Dry
operation will lead to stack performance degradation and eventually, internal leaks.
However, when excess water accumulates in the fuel cell, water flooding might be
resulted at the cathode region which is also not desirable. Inlet humidities also
contribute to liquid water saturation in the fuel cell, which is reported in (Wong et al.
2011). As mentioned above, the proposed system in this study utilizes an external
humidifier to moisten incoming air, whereas anode side is not humidified.
Recirculation of fuel serves the humidification purpose on the anode side of the fuel
cell.
Here, the role of the inlet relative humidity at the cathode of fuel cell electrodes and
its interaction with adjacent flows is discussed and a comparison is presented by a
systematic analysis of these results. A constant relative humidity of 95%, 90%, 85% and
80% is selected at a start–up current for the present simulations. Fig. 10 depicts relative
humidity states at electrode inlet and outlets respectively.
It can be witnessed that anode inlet humidity increases from 90% to 94% when
cathode inlet air is humidified from 80% to 95%. Similarly, relative humidity at anode
exhaust is increased from 95% to 98%. It can be seen that time taken to reach these
humidity levels is more when air inlet humidity is kept lower. Since, water is produced
at the cathode; it is fully saturated at the outlet of the fuel cell.

(a) (b)
0.5 0.6 0.7 0.8 0.9 1.0

0.5 0.6 0.7 0.8 0.9 1.0

Cathode outlet relative humidity

Cathode outlet relative humidity

Cathode inlet relative humidity

Cathode inlet relative humidity

Anode outlet relative humidity

Anode outlet relative humidity

Anode inlet relative humidity

Anode inlet relative humidity

0.0 50.0 100.0 150.0 200.0 0.0 50.0 100.0 150.0 200.0
Time (Seconds) Time (Seconds)

B.19
 Paper I

1.0 (c) (d)

0.5 0.6 0.7 0.8 0.9 1.0

Cathode outlet relative humidity
Cathode inlet relative humidity
Anode outlet relative humidity
Cathode outlet relative humidity

Anode inlet relative humidity

Cathode inlet relative humidity
Anode outlet relative humidity
Anode inlet relative humidity

0.8
0.6

0.0 50.0 100.0 150.0 200.0 0.0 50.0 100.0 150.0 200.0
Time (Seconds) Time (Seconds)

Figure 10. Effect of air inlet relative humidity on fuel inlet and outlet: (a) Air inlet humidity of
95%, (b) 90%, (c) 85% and (d) 80% respectively.

5.6 Water saturation at cathode

Water management is a critical issue since the performance of PEM fuel cell is
strongly influenced by its internal water distribution. Removal of water from the
cathode is dependent on stack temperature and pressure drop. Temperature is the more
critical factor, since at high temperatures water will be in the vapour state and easier to
remove. Stack water production increases with current and is also dependent on the
number of cells. The exact amount of liquid water product depends on the cathode
outlet temperature. Figure 11 provides information on the amount of liquid water at
inlet and outlet of the cathode channel. Fig. 11a shows the start–up of the system at
40A. As seen, the amount of liquid water at cathode outlet is around 17 % when stack
operating temperature is close to the ambient and lowers to 11% when the stack reaches
steady–state at 65 °C.
On the other hand, when the stack is operating at 100 A, the amount of liquid water
percentage at cathode outlet is 23% approximately, which is much higher than the case
with 40A start–up. This is because of higher current densities at start–up sequence when
the stack temperatures are very low. When higher currents are drawn, reactions within
the fuel cell are accelerated which demand more intake of fuel and oxidant.
Consequently these reactions produce more water when compared to production at low
currents. Since the stack temperature is low at start–up, higher percentage of produced
water is saturated at the fuel cell cathode. At low currents, water removal is the
dominant factor and reactant stoichiometries are determined by the minimum flow
rates required for water removal which in the present case are more than adequate to
provide the necessary concentrations. At steady–state when stack operating temperature
is around 67 °C, liquid water fraction in the outlet air is determined to be around 11%.

B.20
 Paper I

Liquid fractions at anode outlet are very low when compared to cathode outlet. They are
calculated to be in the range of 1.5 to 2.0 % for both 40A and 100A current scenarios
discussed above.

(a) At current 40A (b) At current 100A

80.0
80.0

0.3
0.3

0.25
Cathode outlet temperature [C]
0.25
Cathode outlet temperature [C]

60.0
60.0

0.2
Liquid fraction
0.2
Liquid fraction

40.0

0.15
40.0

0.15

0.1
0.1

20.0
20.0

0.05
0.05

0.0
0.0

0.0 150.0 300.0 450.0 0.0 75.0 150.0 225.0

Time (Seconds) Time (Seconds)

Figure 11. Percentage of liquid water at cathode outlet: (a) at start–up current of 60A and (b) at
start–up current of 100A.

6 Conclusions
In this work, a comprehensive dynamic model of a fuel cell system along with the all
necessary BoP components is presented. The objective of the proposed study is to design
and demonstrate a control system which has the ability to predict the dynamics of PEM
fuel cell system during simultaneous changes of operating conditions. Main
contributions of the suggested model are attributed to system response methodology,
which incorporates stack thermal behaviour in addition to fuel cell electrochemistry,
water crossover, mass and energy balances. Emulation of the presented system shows
that the results are in good agreement with the manufacturer’s data. It is observed that
system efficiency and voltage output are higher at low power start–ups but for the fuel
cell stack it takes longer time to reach stable operating conditions. Also, slow
temperature controls as opposed to fast electrochemical changes, affect fuel cell stability.
Furthermore, air radiator consumes more power at high currents once the system is
stable and due to this fact, system efficiency is reduced. In addition, it is shown that
cathode inlet water levels are adequate enough to humidify fuel stream, which is
recirculated into the anode. Finally, amount of liquid water in the cathode outlet is
considerably higher, at high current density start–ups, while water removal from cathode
exhaust requires additional attention at low stack operating temperatures. Thus, the
presented model can be used for optimizing and designing operational strategies for
PEMFC systems for automotive applications.

B.21
Paper I

Acknowledgement
The authors would like to thank the Danish Energy Agency for financial support and
our industrial partner, H2Logic, for their collaboration and technical support.

References
Ahn, Jong–Woo and Song–Yul Choe. 2008. Coolant controls of a PEM fuel cell system.
Journal of Power Sources 179, Nr. 1: 252–264.
Amphlett, J C, R F Mann, B A Peppley, P R Roberge and A Rodrigues. 1996. A model
predicting transient responses of proton exchange membrane fuel cells. Journal of
Power Sources 61, Nr. 1–2: 183–188.
Asghari, Saeed, Hooman Akhgar and Bagher Faghih Imani. 2011. Design of thermal
management subsystem for a 5kW polymer electrolyte membrane fuel cell system.
Journal of Power Sources 196, Nr. 6: 3141–3148.
Aspentech. 2012. Aspen plus dynamics. http://www.aspentech.com/products/aspen–
dynamics.aspx
Ballard Mark9 SSL. 2008. Product manual and integration guide.
Ceraolo, M, C Miulli and A Pozio. 2003. Modelling static and dynamic behaviour of
proton exchange membrane fuel cells on the basis of electro–chemical description.
Journal of Power Sources 113, Nr. 1: 131–144.
Hosseinzadeh, Elham and Masoud Rokni. 2012. Development and validation of a
simple analytical model of the Proton Exchange Membrane Fuel Cell (PEMFC) in a
fork–lift truck power system. International journal of green energy To be publ.
Hu, Guilin, Jianren Fan, Song Chen, Yongjiang Liu and Kefa Cen. 2004. Three–
dimensional numerical analysis of proton exchange membrane fuel cells (PEMFCs)
with conventional and interdigitated flow fields. Journal of Power Sources 136, Nr. 1:
1–9.
Jia, J, Y Wang, Q Li, Y T Cham and M Han. 2009. Modeling and Dynamic
Characteristic Simulation of a Proton Exchange Membrane Fuel Cell. IEEE
Transactions on Energy Conversion 24, Nr. 1: 283–291.
Jung, Jee–Hoon and Shehab Ahmed. 2010. Dynamic model of PEM fuel cell using real–
time simulation techniques. Journal of Power Electronics 10, Nr. 6: 739–748.
Khan, M J and M T Iqbal. 2005. Modelling and analysis of electrochemical, thermal,
and recetant flow dynamics for a PEM fuel cell system. Fuel Cells 5, Nr. 4: 463–475.
Kim, Kihyung, Meng Wang, Michael R Von Spakovsky and Douglas J Nelson. 2009.
Dynamic synthesis/design and operation/control optimization under uncertainty
of a PEMFC system. ASME International Mechanical Engineering Congress and
Exposition, Proceedings 8: 679–689.
Li, Youcai, Sichuan Xu, Zhigang Yang and Yongxiang Li. 2011. Experiment and
simulation study on cold start of automotive PEMFC: 2166–2170.

B.22
 Paper I

Mangold, Michael, Silvia Piewek, Olaf Klein and Achim Kienle. 2011. A model for the
freeze start behavior of a pem fuel cell stack. Journal of Fuel Cell Science and
Technology 8, Nr. 3.
Meng, Hua. 2008. A PEM fuel cell model for cold–start simulations. Journal of Power
Sources 178, Nr. 1: 141–150.
Park, Sang–Kyun and Song–Yul Choe. 2008. Dynamic modeling and analysis of a 20–
cell PEM fuel cell stack considering temperature and two–phase effects. Journal of
Power Sources 179, Nr. 2: 660–672.
Pathapati, P R, X Xue and J Tang. 2005. A new dynamic model for predicting transient
phenomena in a PEM fuel cell system. Renewable Energy 30, Nr. 1: 1–22.
Pukrushpan, Jay T, Huei Peng and Anna G Stefanopoulou. 2004. Control–Oriented
Modeling and Analysis for Automotive Fuel Cel Systems. Journal of Dynamic Systems
Measurement & Control 126, Nr. 1.
del Real, Alejandro J, Alicia Arce and Carlos Bordons. 2007. Development and
experimental validation of a PEM fuel cell dynamic model. Journal of Power Sources
173, Nr. 1: 310–324.
Santarelli, M G, M F Torchio and P Cochis. 2006. Parameters estimation of a PEM fuel
cell polarization curve and analysis of their behavior with temperature. Journal of
Power Sources 159, Nr. 2: 824–835.
Shan, Yuyao and Song–Yul Choe. 2006. Modeling and simulation of a PEM fuel cell
stack considering temperature effects. Journal of Power Sources 158, Nr. 1: 274–286.
Spiegel, Colleen. 2007. Designing and building fuel cells. New York: McGraw–Hill.
Springer, T E, T A Zawodzinski and S Gottesfeld. 1991. Polymer Electrolyte Fuel–Cell
model. Journal of the Electrochemical Society 138, Nr. 8: 2334–2342.
Sundaresan, M and R M Moore. 2005. PEM Fuel Cell Stack Cold Start Thermal Model.
Fuel Cells 5, Nr. 4: 476–485.
Vasu, G and A K Tangirala. 2008. Control–orientated thermal model for proton–
exchange membrane fuel cell systems. Journal of Power Sources 183, Nr. 1: 98–108.
Wong, K H, K H Loo, Y M Lai, Siew–Chong Tan and Chi K Tse. 2011. A theoretical
study of inlet relative humidity control in PEM fuel cell. International Journal of
Hydrogen Energy 36, Nr. 18: 11871–11885.
Yan, Qiangu, Hossein Toghiani, Young–Whan Lee, Kaiwen Liang and Heath Causey.
2006. Effect of sub–freezing temperatures on a PEM fuel cell performance, startup
and fuel cell components. Journal of Power Sources 160, Nr. 2: 1242–1250.
Yerramalla, S, A Davari and A Feliachi. 2002. Dynamic modeling and analysis of
polymer electrolyte fuel cell. 2002 IEEE Power Engineering Society Summer Meeting,
Vols 1–3, Conference Proceedings: 82–86.

B.23
Paper I

Nomenclature

E theoretical voltage V 
Vcell average cell voltage V 
Pel stack power kW 
Pin energy into the fuel cell kW 
Pout energy out of the fuel cell kW 
Q loss heat dissipated kW 
Ct stack thermal capacitance kW 
R universal gas constant J molK 
T temperature K 
F Faraday’s constant C molK
I current  A
Ncell number of cells  
g 0f change in Gibbs free energy J molK 
PH 2 hydrogen partial pressure  
PO2 oxygen partial pressure  
Mm mol. weight of membrane Kg mol 
J H 2O net water–diffusion flux mol scm2 
D water diffusion coefficient cm2 s 
aH 2 hydrogen activity  
a H 2O water activity  
a O2 oxygen activity  
i current density A cm2 
in internal current density A cm2 
i0 exchange current density A cm2 
i0,a anode exchange current density A cm2 
i0,c cathode exchange current density A cm2 
ka anode reaction rate mol scm2 
kc cathode reaction rate mol scm2 
ne electrons transferred mol e mol fuel 
nel number of electrons  
ndrag electro osmotic drag  
tm membrane thickness cm
Kp proportional gain

B.24
 Paper I

Ti integral time
Td derivative time
Greek symbols
 variable of integration
a anode transfer coefficient  
c cathode transfer coefficient  
 symmetry factor  
act activation overpotential V 
 act ,a anode activation overpotential V 
 act ,c cathode activation overpotential V 
conc concentration overpotential V 
ohmic ohmic overpotential V 
 membrane water content  
 dry membrane density g cm3 

B.25
Paper I

B.26
 Appendix C

Paper II

199
Appendix C

Rabbani, A & Rokni, M. “Dynamic characteristics of an automotive fuel cell

system for transitory load changes”, Journal of Sustainable Energy Technologies &
Assessments, 2013; 1: 34–43.

200
 Paper II

Dynamic characteristics of an automotive fuel cell system for

transitory load changes

Abid Rabbani and Masoud Rokni

Thermal Energy Section, Department of Mechanical Engineering, Technical University
of Denmark, 2800 Kgs. Lyngby, Denmark.

Abstract
A dynamic model of Polymer Electrolyte Membrane Fuel Cell (PEMFC) system is
developed to investigate the behavior and transient response of a fuel cell system for
automotive applications. Fuel cell dynamics are subjected to reactant flows, heat
management and water transportation inside the fuel cell. Therefore, a control–oriented
model has been devised in Aspen Plus Dynamics, which accommodates electrochemical,
thermal, feed flow and water crossover models in addition to two–phase calculations at
fuel cell electrodes. The model parameters have been adjusted specifically for a 21.2 kW
Ballard stack. Controls for temperatures, pressures, reactant stoichiometry and flows are
implemented to simulate the system behavior for different loads and operating
conditions. Simulation results for transitory load variations are discussed. Cell voltage
and system efficiency are influenced by current density and operating temperature as
well. Together, air blower and radiator consume 10% of the stack power at steady–state;
nevertheless their power consumption could reach 15% during load surges.
Furthermore, water crossover in the fuel cell has shown a significant impact on anode
inlet flows, humidity and recirculation pump during these load changes. Also, the
amount of water saturation at cathode is found to be abruptly fluctuating and its
removal from cathode is dependent on operating temperature and reactant
stoichiometry.

Key words: Dynamic simulation, system modeling, fuel cell, PEMFC, water crossover, system
control.

1. Introduction
Fuel cell systems have received substantial attention in recent years and research on
these systems has drastically increased mainly due to their inherent virtues of clean and
efficient mode of operation. Existing fuel cell systems are categorized based on the type
of electrolyte and preferred operating conditions. Among various types of fuel cells, the
Proton Exchange Membrane Fuel Cells (PEMFC) is currently the best choice for
portable power generation due to its relatively low operating temperature, quick start–
up, high power density and efficiency to name a few.

C.1
Paper II

As a power source for automotive applications, PEMFC systems are usually subject to
inflexible operating requirements when compared to stationary applications. These
systems have to operate at varying conditions related to temperatures, pressures, power
load and humidity. PEMFC dynamics are influenced by reactant flows, heat
management and water content in the streams as well as within the fuel cell itself. All
the auxiliary components, such as air and fuel supply system which include compressors
and control valves, and the thermal control system which consists of heat exchangers,
coolant pumps and air radiators need to be controlled for optimum operation of fuel
cell when the system experiences varying load changes. Understanding the transient
behaviour of a PEMFC therefore becomes very beneficial in dynamic modelling of these
power modules at a system–level.

Many PEM fuel cell models have been developed in recent years. However, very few of
these models are published on dynamic modelling of complete PEMFC systems along
with their BoP. Most of the available literature focuses on individual components of
these systems, mainly on the fuel cell stack. While, steady–state models of these systems
are present in abundance. A generalized dynamic model for fuel cell stack is reported by
Amphlett et al. [1]. Another bulk dynamic model used for developing a control system is
presented by Yerramalla et al. [2]. A simplistic dynamic model based on cathode kinetics
was developed by Ceraolo et al. [3]. Pukrushpan et al. [4] presented a transient dynamic
model and elucidated the dynamic characteristics of water transport in PEM fuel cells. A
complete PEMFC system model was developed by Pathapati et al. [5] which included the
dynamics of flow and pressure in the channels. Hu et al. [6] represented a three–
dimensional computational PEM fuel cell model with comparison of different flow
fields. In recent years, several improved models were published by Park and Choe [7]
and Jia et al. [8] to investigate fuel cell transient electrical responses under various
operating conditions.

Heat management in PEMFCs being a critical factor in its operations and performance
is accounted for in open literature as well. Issues related to temperature dynamics are
dealt and studied by Vasu and Tangirala [9], which could predict the effects of
temperature and feed flows on system transient behaviour. Khan and Iqbal [10]
proposed a transient model to predict efficiency in terms of voltage output, and a
thermal model including heat transfer coefficients and energy balance for the stack.
Shan and Choe [11] analysed the temperature distribution on fuel cells by developing a
two–dimensional model. Another control–oriented thermodynamic model is also
proposed by del Real et al. [12].Coolant control strategies were suggested by Ahn and
Choe [13] after investigation of temperature effects on the system. Jung and Ahmed [14]
developed a stack model based on real–time simulator in MATLAB/ Simulink
environment and validated it with experimental setup of Ballard Nexa fuel cell. A
thermal management system for a PEMFC was designed by Asghari et al. [15]. Influence
of temperature on fuel cell’s characteristics is also reported by Beicha [16].

C.2
 Paper II

The model presented in this study aims at analysis and investigation of a complete
PEMFC system and studies its transient response to varying load and operating
conditions. According to authors’ literature survey, no studies have been conducted on
system–level dynamic modelling of PEMFC system with all the necessary BoP
components. Previous studies focus on transient response of fuel cell stack under
different operating conditions; primarily on individual component analysis. Therefore, a
need for a control–oriented dynamic system model is identified, which simulates a fuel
cell stack under multiple varying operating conditions and changing auxiliary
components outputs. Dynamic characteristics of PEMFC are also attributed to the heat
management and water transportation that is scarcely reported in the open literature.
Investigations for effects of heat exchangers on fuel cell stack performance and water
crossover on anode recirculation operations are therefore selected to be one of the
primary objectives here.

Thereby in the entirety of this study, a sizeable focus has been set to devise a dynamic
model of the fuel cell stack, which accommodates the electrochemical, thermal, feed
flow and water transportation models. A complete system is constructed in Aspen Plus
Dynamics by incorporating all the essential auxiliary components and implementing
control strategies in order to emulate a real PEMFC system. Effects of these controls and
other components are also investigated in this work. A thermal management strategy has
been designed and its dynamic impact on fuel cell stack has been reported for the first
time. Analysis of water crossover in the fuel cell and its impact on anode recirculation
operations has been conducted and suitable findings are reported here. Moreover, two–
phase characteristics of concerning material streams are determined which provide
suitable insight to saturated water issues in the fuel cell stack. This study also takes into
consideration the BoP, such as air blower, valves, coolant pumps and air radiator;
making it a thorough tool for predicting PEMFC dynamics and to provide important
information for the design of control strategies.

In the current study, the focus is on complete system with all necessary auxiliary
components and their effect on system performance rather than effect of individual
component on the system. Thus, it differs substantially from previous studies in the
sense that not only dynamics of the fuel cell stack are included but responses of all other
auxiliary components are also incorporated by applying a detailed control strategy
design.

2. System overview
Layout of the proposed PEMFC system is shown in Fig. 1. The system comprises a
PEMFC stack, air compressor, humidifier, pumps, heat exchangers and radiator for the
cooling circuit, flow valves and controllers. Compressed air, which is fed into cathode of
the stack is cooled and humidified prior to its entrance. Pressurized hydrogen from
storage tank is regulated by a control valve into the fuel cell anode. Since the stack is not

C.3
Paper II

operateed at dead––end modee, a higher fuel stoichhiometry is maintaineed. Unutiliized fuel

from an node exhaust is recirrculated baack to the feed stream
m via a reecirculation
n pump,
thus alllowing the fuel to be humidified
h d.

Figuure 1. Schem
matic layout of a compleete PEM fueel cell system
m with auxiliiary components.

In ordeer to have a steady–staate operatioon, the fueel cell stack

k needs to bbe maintain ned at a
constannt operatinng temperatture. Thereefore, heatt rejected by b the stackk is absorb bed by a
liquid coolant which
w circu
ulates in a circuit associated
a with the stack and a heat
exchangger. An extternal coolling loop, cconnected to the aforrementioneed heat excchanger,
in turnn cools thee water in n the interrnal circuitt. This circuit also cconsists off a heat
exchangger to preccool air en ntering thee fuel cell and an airr radiator ffor heat reejection.
Flow of water is regulated by pumps in the resspective cirrcuits. In oorder to maintain m
nt stoichiom
reactan metry and d fuel celll operatingg temperatture at vaarying load ds, PID
controlllers are deployed to o regulate reactant anda coolannt flows. T This emulaates the
behavioour of fuel cell in reall time and hhelps in an
nalysis of syystem respoonse underr varying
operatinng conditiions. Aspen n Plus dynnamics con ntains buillt–in PID controllerss. These
controlllers collectt data from
m various coomponent inlet and outlets
o whiich are regaarded as
pressurre, temperrature and d flow traansmitters, and man nipulate thhe corresp ponding
compon nents to reach the desired state. Due to th his fact, currrent system
m responsess and its
behavioour are attrributed to the formuulated control strategyy which is based on fuel cell
stack llimitations and reccommendattions by the manu ufacturer. Detailed control
mechan nisms are illustrated in subsecction 2.6. Figure 1 only repreesents conttrollable
connecctions initiaating from the controoller. In caalculation as a well as rreality, the control
unit coollects inforrmation from variouus temperatture, pressu ure and floow transmitters in

C.4
 Paper II

addition to current and volltage data ffrom the fuel

f cell. Here
H a simpplified view
w of the
system is presenteed.

1.1 Fuel Cell staack

A Ballaard fuel celll stack [17]] has been specificallyy adapted in
n the curreent study. The
T fuel
cell stacck containss 110 cells with
w a cell area equall to 285 cm m2. Operatiing temperrature of
the staack is maiintained arround 60––70°C with h a pressuure range of 1.1– 2.2 2 bar.
Maximu um power produced from the described stack is 21.2 2 kW coorrespondiing to a
currentt of 300A; however
h it is generallyy operated at lower cu
urrent rangges to attain
n higher
efficien
ncies by red ducing ohmic and cconcentratiion overpo otentials wiithin the fuel
f cell
stack. TThis model, which iss based on n equationss adopted by b Hosseinnzadeh and d Rokni
[18], coontains som
me parametters influen ncing physiical charactteristics of the system, as well
as on ooperating co onditions and
a membbrane propeerties. Som me of these parameterrs which
are obttained for the Ballard d stack aree shown in n Table 1. Here, focuus has been n set to
build u up a systeem which meets thee requirem ments of actual
a stacck runningg under
recomm mended con nditions.

Table 1.. Parameter estimation for

f Ballard Fuel cell staack
Parammeter Value
Numbber of electtrons transfferred per m
mole of fueel, ne
mol m
mol fuel  2
e

he reaction rate, nel moole mol fuel 

Numbber of electtrons for th 1 for catthode and 4 for
anode
nal current density, in A cm 2 
Intern 0.002
Symmmetry factorr,  0.5
Membbrane thickkness, t m cm
m 0.0183
Densiity of the membrane–
m dry conditiion,  dry g cm 3  3.28
Moleccular weigh brane, M m K
ht of memb Kg mol  1.1

Duringg PEMFC steady–statee operation ns, coolant pressure shhould be loower than reactant
pressurres whereass during staart–up, cooolant presssure may exxceed reacttant pressuure. It is
recomm mended to operate thee anode at a higher prressure thaan the cathoode. Ensurring that
cathodee pressure is lower thhan the anoode pressurre will minimize nitroogen crosso over and
improvve cell stabbility. In a system w with fuel reecirculation
n, the connsumption will be
slightly above 1.0 stoichiommetry as gen nerally 1% to 2% of thet flow w will be requ
uired for
purgingg. In the prresent studdy, purge ooccurrencess are negleccted. For a maximum m service
life and
d efficiencyy of the staack, Ballardd suggests operating conditions
c s within whhich the
stack shhould operrate [17], which
w are uused in thee current investigatio
i on. Recomm mended
reactannt inlet preessures and nominall pressure drops witthin the ffuel cell sttack are
reporteed in Fig. 2..

C.5
 Paper II

Ano
ode Pressure drop Cath
hode Pressure d
drop H2 Inlet pressure Air Inlet p
pressure
0,6 2,5
5
0,5
2,0
0
0,4
∆P (bar)

P (bar)
1,5
5
0,3
1,0
0
0,2
0,1 0,5
5
0 0,0
0
0 50 10
00 150 200
2 250 300 0 50 100 15 0 200 250
2 300
(a) I (A) (b) A)
I (A
Figure 2. Design vaalidation daata from Balllard (a) Nominal pressu ure drop of rreactants within the
fuel cell, (b) Reaactant inlet ppressures corrresponding to
t current drrawn.

1.2 Humidifier
Dehydrration in fu uel cell memmbrane is aalso a criticcal issue in PEMFC syystems. In ordero to
reduce ionic resisstance in the t fuel ceell, a fully hydrated membranee is desired d which
enhancces ionic crrossover beetween the electrodess, as well as a extends its life. Thherefore,
the reaactants in PEMFC need n to bee humidifiied before entering the stack. In the
proposeed configu uration, a humidifier
h r is placed in conjun nction of ccathode in nlet and
exhaustt, where it utilizes thee water pro duced by chemical
c reaction insiide the fuell cells to
humidiify inlet airr. A simplisstic energy and mass balance
b mo odel is deveeloped to cater
c for
fuel celll cathode requiremen nts. Althouugh some empirical models m couuld be connsidered,
these vvary over a wide ran nge dependding upon n the typess of humiidifiers useed. It is
therefore assumed that inllet air is ooptimally humidified
h d after passsing throu ugh the
humidiifier. Relatiive humidity of air en ntering the cathode is arbitrarilyy set to 95% % in the
simulattions. This assumptio on could bbe justifiedd as it is veery close to real opeerational
conditions. On th he anode siide, fuel is humidifieed by mean ns of water crossover through
t
cell meembrane and a recircu
ulation of anode exxhaust into o the inleet stream, thereby
abandooning a need for separate hum midifier. Mo oreover, it can be obbserved thaat water
cross–oover from cathode
c to anode thrrough the membrane
m is adequatte enough to raise
the relaative humiidity in an node outlett to 100%. This exh haust, whenn recirculated and
mixed w with inlet stream maaintains thee desired humidity
h leevels at anoode inlet, which
w is
further discussed in results section.

1.3 Reactant feedd systems

Auxiliaary compon nents suchh as blowerrs, pumps and valvees regulate flows of material
m
streamss in PEMFC C systems. For hydroogen feed, a valve is placed
p betwween the hyydrogen
tank annd inlet maanifold of anode whiich enabless or disablees the hydrrogen supp ply. This
regulatoory valve adjusts
a high hydrogen n pressuree of the tank to the desired op perating
pressurre of the fu
uel cell. Siince the syystem doess not operaate on deaad–end mo ode, the
amountt of hydroggen regulated by this vvalve equalls the stoichiometric hhydrogen required
r

C.6
 Paper II

by the fuel cell. An air blo ower regullates the flow

fl and pressure
p off oxidant into
i the
cathodee. The amoount of stoichiometricc oxygen fo
or fuel cell reaction iss manipulated by a
controlller which regulates the
t electriccal power of the blow wer, therebby controlling the
compreession and air flow intto the systeem.

1.4 Heat Exchanngers and thermal manageement system

For heat manageement of the t prescriibed system m, a netwo ork of heaat exchanggers and
radiatorrs is deployyed. These heat exchaangers extrract heat prroduced byy the cell sttack and
maintaiin the seleected operaating tempperature which
w is esssential for performan nce and
durabillity of the fuel
f cell. Although
A heeat exchangger modelss used heree are predefined in
Aspen Dynamics,, some of the param meters have been assu umed on th the basis of media
enterinng the hot and
a cold siides of theese heat excchangers. The
T heat exxchanger, whichw is
conneccted to the internal co ooling loopp, has liquiid water onn both its hhot and coold side.
Therefoore, a UA value
v of 1.0 kW/K iss assumed. Whereas, UA values for air pree–cooler
and radiator are approxim mated to bbe 0.05 kW W/K and 0.3 kW/K K respectivvely. In
performming the sim mulations, the pressuure drop was assumed d to be 0.055 on both sides of
heat exxchangers. The correccted LMTD D is calcullated in ad
ddition to tthe corresp ponding
inlet an
nd outlet teemperaturees of hot an
nd cold streeams.

1.5 Pumps and Blowers

B
In the above prop posed systeem, air blowwer, anodee recirculattion pump and waterr pumps
are onee of the Bo oP compon nents which h are also regulated
r by
b the conttrol systemm. Aspen
Dynammics TM con ntains mod dels of thesse units in its libraryy as well. SSince the nominal
n
power oof the PEM MFC is onlly 21 kW, mass flow rates of fu uel and air are very loow. For
examplle, at an avverage load of 10 kW W, fuel and air flows are around 0.00014 kg/s k and
0.0088 kg/s resp pectively. Therefore,
T very low values off isentropiic efficienccies are
suggesteed in this paper. Th he efficienccy of a blo
ower ranges from 15 % to 48% % in the
calculattions, depeending on the
t air masss flow. Caalculated pu ump efficieencies are also
a very
low for the coolinng water circuits and aare determiined to be around
a 70% %.

1.6 Control Systeem

In ordeer to have a stable andd efficient ooperation, the system
m requires aan effective control
strategyy to regulatte system parameters
p and operaating condiitions. Typpical proportional–
integrall (PI) conntrollers, which
w are widely useed in industrial conntrol systems, are
employyed to regu ulate differrent compoonents and d flow streeams. Key parameterrs to be
controllled in thee proposed d system arre reactantt inlet stoiichiometriees, inlet prressures,
coolantt inlet and operating temperatur
t res of the sttack.

Figure 3a shows fuel

f and air stoichiom metries forr the selectted fuel ceell stack wh
hich are
proposeed by Ballaard and in Fig.
F 3b, reccommendeed temperatture differeence between stack
inlet an
nd outlet iss displayed
d. It can bee seen that at low currrent loadss, high amo ounts of
excess rreactant flo
ows are dessired. This is due to the
t fact thaat at low poower consu umption

C.7
Paper II

and low pressures, water is formed by reactions in cathode side of the cells and it needs
to be ejected out of the stack, which is done by supplying high amounts of reactants.
While the amount of oxygen consumed depends on the stack current, the amount of
oxygen supplied to a fuel cell is directly related to the blower power. Therefore, an
algorithm based on the above figure is developed to be the process variable for PI
controller, which regulates blower power to maintain the desired oxygen ratio. Similarly,
an algorithm for controlling hydrogen flow is devised along with a PI controller, which
regulates the control valve opening for optimal fuel supply.

H2 Stoichiometry Air Stoichiometry Coolant Delta T

7 12
6 10
Stoichiometric Ratio

5 8
4 ∆T (◦C) 6
3
4
2
2
1
0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
(a) I (A) (b) I (A)
Figure 3. Recommended operating conditions by Ballard, (a) Reactant stoichiometry in the fuel
cell, (b) stack inlet and outlet temperature difference maintained by coolant mass flow.

Thermal management in PEMFC systems is of vital importance, basically due to the fact
that heat produced in the selected fuel cell cannot be dissipated by convection and
radiation through the stack surface. A consistent and stable operation of around 70°C
thus requires a liquid cooling system. Since the operating temperature of the fuel cell is
not very high, a low temperature difference with the ambient requires having a large
heat transfer surface. Thus, an efficient thermal control system becomes of substantial
importance to ensure optimum system performance.

As shown in Fig.1, the cooling system for fuel cell consists of internal and external
cooling circuits. Also mentioned earlier, the coolant mass flow rate defines the variance
in the stack temperature or simply maintains the fuel cell operating temperature. In this
case, temperature in the stack can be controlled by coolant flow rate which acts as an
input signal and is adjusted by the PI controller. Based on data from Fig. 3b, equations
defining stack temperature as a set–point for controller are developed. In addition, the
controller simultaneously collects data from temperature transmitter at coolant outlet
stream, which then changes the coolant flow accordingly by sending output signals to
the driving pump. Employment of such algorithm in the system controller ensures a

C.8
 Paper II

stable operation under normal steady–state conditions, however for system start–up
scenario, a different approach is required.

Temperature of the coolant entering the stack can similarly be controlled by flow of
water in the external circuit. Control signal to the associated pump regulates electrical
power of the pump and hence the coolant inlet temperature into the stack. In a similar
fashion, temperature of water in the external circuit is dependent on radiator fan speed.
PI controllers are used to regulate the fan speed as well.

3. Fuel cell modelling

The characteristics of the PEMFC system described above are implemented in Aspen
Plus DynamicsTM which is a simulation tool for process modelling and energy system
analysis. The program contains a vast library of components and controls for standard
energy processes. The PEMFC stack model presented in this study is based on a model
developed by Hosseinzadeh and Rokni [18]. Concentration losses are neglected in the
present study, which is justified by the fact that the system does not run at such high
current densities where the concentration overpotentials becomes significant. Models
for fuel cell and humidifier are also implemented into the code and are based on
adopted mathematical models describing the voltages, current densities and their
dependence on operating pressures, temperatures and stoichiometric ratios of the
reactant gases. This model which incorporates governing equations for cell
electrochemical, polarization overpotentials, heat transfers and water diffusion across
the membrane is implemented into ASPEN Plus Dynamics and system controls are
implemented in order to ensure stable operation of the plant during load changes.
Thermodynamic efficiency and net power of the system are determined by the current
drawn and voltage produced by the stack. Total energy into the fuel cell is consumed by
electrical power output, heat removed by the coolant, heat loss at the stack surface and
energy stored by the stack itself. In the current model, a lumped thermal model
proposed by Khan and Iqbal [10] is considered. The stack is regarded as a single thermal
mass with a heat capacity. With the assumption of stack temperature being equal to
coolant temperature at the outlet, heat exchanged with the coolant and hence stack
operating temperature could be determined. Subordinate components in the BoP, i.e.
anode recirculation and water pumps, air compressor, mixers and heat exchangers are
simulated by using default mathematical models contained in Aspen Plus Dynamics
library. Table 2 shows a standard set of equations which constitute the model used in
the current study.

Table 2. Constitutive equations for PEMFC model.

Average cell voltage Vcell  E   act   ohmic   conc (1)

C.9
Paper II

 g 0f RT  a H 2O 
Nernst equation E  ln (2)
ne F ne F  a H 2 aO2 0.5 

Theoretical cell voltage E

 g 0f
ne F

RT
ne F
1
ln PH 2 PO2
 0 .5
 (3)

Change in Gibbs free

energy
 
g 0f  g 0f H 2O
 
 g 0f H2

2
 
1 0
gf O2
(4)

Activation losses in RT  i  in  RT  i  in 
 act   act ,c   act ,a  ln  ln (5)
PEMFC  c F  i0,c   a F  i0,a 

Ohmic overpotentials  ohmic  rel  rion   i (6)

Ionic resistance in the

rion 

C1  1  0.03i  0.062  T / 303  i 2.5
t
2
 (7)
cell ave  0.634  3i   expC2  T  303 / T  m
Membrane water
ave  0.043  17.18a w  39.85a w2  36a w3 (8)
content

Pw
Water vapour activity aw  (9)
Psat

Water profile inside cell c   a

mem  .z  a (10)
membrane tm

Transfer coefficient for  a    nel (11)

anode

Transfer coefficient for  c  1     nel (12)

cathode

Exchange current  1     nel  F  E 

i0,a  nel  F  k a  exp  (13)
density at anode  RT 

Exchange current     nel  F  E 

i0,c  nel  F  k c  exp  (14)
density at cathode  RT 
Net water  dry
i d
transportation J H 2O  2ndrag  D (15)
2F M m dz
through membrane
  1 1 
Water diffusion D  4  10  6 exp  2416     
  303 T   (16)
coefficient

2 .563  0 .33   0 .0264 2  0 .000671 3 

C.10
 Paper II

Fuel cell energy balance Ct  dT  Pin  Pout  Pel  Q loss (17)

dt

dN I
Faraday’s law  (18)
dt ne  F

4. Results and discussion

Simulations for the prescribed system were carried out and reliability of the suggested
model is verified and corroborated against design validation data by [17] at different
power loads. Model characteristics of the same system at various operating temperatures
and power loads have been studied in Hosseinzadeh and Rokni [18]. Figure 4a
represents adaptation of the devised model which corresponds to the polarization curve
obtained from operational data associated with the given PEMFC stack. The calculated
relative error shows a good agreement between the model and data provided. As
suggested by the manufacturer, stack temperature range of 60–70°C has been used in
simulations of the current system. Selection of other parameters and operating
conditions is based upon Fig. 2 and Fig. 3.

Experimental Calculated Relative Error System efficiency

0,65
950 4
3 0,60
900
2 0,55
850
Error %

1
Efficiency
Vcell (mV)

800 0,50
0
750 0,45
‐1
700 ‐2 0,40
650 ‐3
600 ‐4 0,35
550 ‐5 0,30
0 0,2 0,4 0,6 0,8 1 1,2 0 50 100 150 200 250 300
(a) i (A/cm2) (b) I (A)
Figure 4. (a) Comparison of Ballard operational data and calculated polarization curves for PEM
fuel cell with relative error, (b) Overall system efficiency profile at operating current range.

Fig. 4b exhibits profile of overall system efficiency at corresponding currents. A general

trend of decreasing efficiencies with increasing loads can be observed, which is
characterised by an increase in ohmic overpotentials in the stack and high power
consumption of BoP for cooling of the system. However an exception to the contrary
can be noticed at very low currents, where high parasitic losses in BoP render the system
efficiency to be lower as well. Nevertheless, higher stack efficiencies at low loads assist
the system efficiency to be not as low when compared to that at higher currents. Results
for transitory effects under variable load changes (when a current corresponding to a

C.11
Paper II

specificc power loaad is drawn

n from thee stack and
d varied at any occurrrence of tiime) are
discusseed below:

1.7 Cell voltage and

a temperatture variationns
An insttance of loaad change, when currrent is ram mped from 60 6 A to 1000 A at a raate of 20
amperees per secon nd and vicee versa, is ppresented in
i Fig. 5a and
a its effeects on celll voltage
are exaamined. It can be ob bserved thaat the cell voltage reduces abruuptly with current
surge aand vice versa. This decline
d in vvoltage pottential at increase of current deensity is
attributted to the cell
c overpotentials. H However, it is also notiiced that ceell voltage is also a
function of operaating temp perature annd pressuree. Since, prressures off the reactaants are
already regulated; here we will
w discusss temperatu ure depend dency of deeveloped potential
p
differen
nces in thee cell. Volttage increaases with thhe elevatio
on of operaating temp perature.
This is due to faast reaction kinetics at the eleectrodes of individu al cell sitees when
operatinng at higheer temperattures. At thhe same instance of current ram mp as more current
is draw
wn; more heat
h is prod
duced by th he reaction
ns at cell sites, thus eelevating thhe stack
temperature. Consequentlyy, the voltaage also inccreases afteer the initiaal dip and follows
the deccreasing proofile with the
t stack teemperaturee as it is coontrolled toowards the desired
operatinng temperaature. The opposite coould also beb observed d when the current is reduced
back too 60A.
1500.0
40.0 80.0 120.0

(b) Thhermal manage

ement
0.0 40.0 80.0 120.0

(aa) Cell Voltage at current ramp

p
80.0
65.0 70.0 75.0 80.0 85.0

0.84

Internal coolant flow (kg/hr)

g/ )
External coolant flow ((kg/hr)
( )
(C)

Air radiator flow (kg/hr)

Stack temperature (C)
70.0
g temperature

0.8

1000.0
Cell voltage (V)

Current (A)
Current (A)

0.76

60.0
Operating

0.0

500.0
0.72

50.0
-80.0 -40.0

-80.0 -40.0
p
60.0

0.68

40.0

750.0 800.0 850.0 900.0 950.0 1000.0 750.0 800.00 850.0 900.0 950.0 1000.0
Time (Secconds) Time (Seconds))

Figurre 5. Currennt ramp–up from 60–1000A and vicce versa: (a) Changes inn cell voltages and
opperating tem
mperature, (b)) Changes inn coolant floows to maintain the fuell cell operatiing
temperaturee.
Since, temperatu ure affects the voltaage and overall
o system efficie iency, it becomes
b
importaant to havve a stable operating temperatu ure and th herefore ann efficient thermal
manageement systeem. As can n be seen iin the Fig. 5b, coolan
nt flow ratees in both cooling
circuits and an airr radiator maintain
m thhe temperaature of thee stack, thoough it is very slow
comparred to thee reaction kinetics aaffecting th he stack vo oltage. Ram mping of current
increasees the operrating tempperature off stack thereby actuatting controols to conttemplate
this inccrease. Flow
w in the innternal coooling circu
uit, which is
i coupled with the fuel f cell
stack iss increasedd to extracct heat annd maintaiin the dessired tempperature difference

C.12
 Paper II

between the stack inlet and outlet. Flow in the external cooling circuit is pumped up to
retain the inlet temperature to 60°C. Air flow in radiator is also increased to maintain
temperature of external cooling loop around 50°C by rejecting the heat to the
surroundings. Rate of coolant flows in respective streams is associated with the power
consumed by coolant pumps and air radiator which ultimately affects the overall system
efficiency. It is also observed that air radiator being a liquid–gas heat exchanger has the
slowest reaction time out of the three. Since temperature controls are slow compared to
electrochemical reactions, overall thermal control strategy has a fair impact on stack
voltage which undergoes frequent load changes.

5.1 System efficiency and power consumption

As can be seen in the Fig. 6a, when the stack current is altered from 60A to 100A, there
is a reduction in system efficiency, mainly due to the associated voltage drop. Overall
efficiency of the system is as much affected by air blower and radiator fan, as by the fuel
cell stack itself. It can be observed that power produced by the stack increases with
current drawn. Also, power consumed by auxiliary components increases, thereby
reducing system efficiency from 55.5% to 50%. Once decreased, there is an abrupt rise
in efficiency due to increase in voltage and it fluctuates at around 50% mark due to slow
temperature controls and fluctuating power consumption by air radiator.
-120.0-80.0 -40.0 0.0 40.0 80.0 120.0

(a) System efficiency and power

100.0

(b) Auxiliary power consumption

4.0 6.0 8.0 10.0 12.0

0.55 0.6 0.65 0.7 0.75

1.2
Recirculation pump (kW)

0.2 0.4 0.6 0.8 1.0

Auxiliary Power (kW)

50.0
Stack Power (kW)

System efficiency

Air Radiator (kW)

Air Blower (kW)
PEMFC.Icell A

Current (A)
0.0 -50.0
2.0

0.5

-100.0
0.0

750.0 800.0 850.0 900.0 950.0 1000.0 750.0 800.0 850.0 900.0 950.0 1000.0
Time (Seconds) Time (Seconds)

Figure 6. (a) System efficiency, stack power output and power consumed, and (b) Auxiliary power
consumption.

Figure 6b elaborates on power consumption in the system. Being a major consumer in

the system, power consumption of air blower rises with increase in current and this
incremental profile can be associated to the increased mass flow of air required to
maintain oxidant stoichiometry in the fuel cell. After the initial rise, power
consumption of air blower becomes constant; however the consumption graph of air
radiator takes more time to become stable. This is due to the fact that flow in air
radiator is manipulated to control the temperature and is a slow process. Together, air

C.13
Paper II

blower and radiator consume 10% to that of stack power, whereas a coolant and
recirculation pumps account for 1.0–1.3% when operating at 60A.

On the other hand, at higher currents of 100 A, power consumed by air blower and
radiator constitutes 15% of the total stack power. For a swift temperature control and a
thermally stable operation, air radiator requires more power or a larger heat exchanger
area which would further reduce efficiency during these load variations. Power
consumption of coolant pumps is not shown in the above figure because increase in
internal and external coolant flows affects the auxiliary power consumption to a very
small extent as compared to air compressor and radiator. Anode recirculation pump also
consumes very low power; however the peaks at current surge are a noticeable detail,
which is addressed later in the results.

5.2 Effects of water transportation in fuel cell

In PEMFCs, water management is a critical issue since the performance of fuel cell is
strongly influenced by its internal water distribution. Figure 7a shows net water
diffusion in the cell when current is changed from 60A to 100A and back. In the figure,
positive values for water crossover designate transportation of water from cathode to
anode of the fuel cell and vice versa. With increasing currents, more water is produced
in the cathode which supports back–diffusion towards the anode until the system
reaches back to steady–state and there is almost no net water crossover. Some negative
peaks are also observed at the start of current change, which are due to electro–osmotic
drag.
150.0

(a) Water crossover in fuel cell

0.0 40.0 80.0 120.0

(b) Anode Humidity Levels

1.08
-0.1 0.0 0.1 0.2 0.3 0.4
Net Water crossover (kg/hr)

1.04
Anode outlet RH (%/%)
100.0

Anode inlet RH (%/%)

Current (A)

Currentl (A)

0.92 0.96 1.0

50.00.0

-80.0 -40.0
-50.0

750.0 800.0 850.0 900.0 950.0 1000.0 750.0 800.0 850.0 900.0 950.0 1000.0
Time (Seconds) Time (Seconds)

Figure 7. Current ramp–up from 60–100A and vice versa: (a) Water crossover through PEMFC
membrane, (b) Effect of water transportation on anode RH.

As more current is drawn, there is a rapid increase in hydrogen ion flux towards the
cathode, thereby supporting water crossover through electro–osmotic drag. Back–
diffusion rate is increased with the production of water on the cathode side and
subsequently water flux is directed towards fuel cell anode. The reverse can said to be

C.14
 Paper II

true when current is ramped down from 100 A to 60A. Effect of water distribution can
also be observed in anode inlet and outlet relative humidity. Whereas for the cathode,
since it is assumed that air enters at a constant relative humidity of 95%, the outlet
humidity is always above 100%. Therefore, results for only anode are discussed here.

From Fig. 7b, it could be further observed that relative humidity at anode outlet
decreases with a current surge, though an abrupt increase is detected at the start of this
change. Sudden ramping of the current consumes more hydrogen, leaving higher molar
fraction and partial pressure of water in the anode, thus the peak of high relative
humidity. With the rise in stack temperature, water activity on anode side is reduced
and water diffusion from cathode increases and consequently stabilizes to almost zero
net–water crossover with the control of operating temperature. As there is no external
humidification apparatus for fuel and recirculation of anode exhaust aides in
humidifying the anode, water crossover has a significant impact on anode operations
which are discussed below.

5.3 Fuel and oxidant flows

Figure 8 depicts the variations in reactant inlet flows when the load on fuel cell is varied.
It is interesting to notice the fluctuations in the anode inlet and outlet. When the
current is ramped up from 60A to 100A, more hydrogen is consumed increasing the
utilization factor at that instance. This can be seen in Fig. 8b where mass flow at anode
exhaust suddenly decreases. Changes in temperature can also be noted at these current
surges as the air stream is cooled by external cooling circuit.
120.0

40.0 50.0 60.0 70.0 80.0 90.0 100.0

(a) Stream properties at anode inlet

40.0 80.0 120.0

40.0 50.0 60.0 70.0 80.0 90.0 100.0

(b) Stream properties at anode outlet

2.5

2.5
40.0 80.0

Mass flow (kg/hr)

2.0

Mass flow (kg/hr)

Temperature (C)

2.0
Temperature (C)
Current (A)

Current (A)
1.5

1.5
0.0

0.0

1.0
1.0

-80.0 -40.0
-80.0 -40.0

750.0 800.0 850.0 900.0 950.0 1000.0 750.0 800.0 850.0 900.0 950.0 1000.0
Time (Seconds) Time (Seconds)

C.15
Paper II
120.0

Recirculation pump consumption (kW)

(c) Fuel stoichiometry

40.0 80.0 120.0

(d) Stream properties at cathode inlet

70.0

35.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

0.2
Hydrogen stoichiometry ()
40.0 80.0

0.15

65.0

30.0
Mass flow (kg/hr)
Temperature (C)
Current (A)

Current (A)

60.0

25.0
0.1
0.0

0.0

55.0

20.0
0.05
-80.0 -40.0

-80.0 -40.0

50.0
750.0 800.0 850.0 900.0 950.0 1000.0 750.0 800.0 850.0 900.0 950.0 1000.0
Time (Seconds) Time (Seconds)

Figure 8. Current ramp–up from 60–100A and vice versa:, (a) Stream properties at anode inlet
and (b) outlet respectively, (c) Effect on fuel stoichiometry and recirculation pump, (d) Air flow
into the cathode.

Control of fuel stoichiometry takes some time to readjust to the desired level. Although,
anode outlet flow is disturbed by fuel stoichiometry, it is also affected by the water
content in anode outlet (70% mass fraction), which depends on the net water crossover
within the cell. Initial peaks at both load changes are also attributed to the electro–
osmotic drag which is a function of current density. Since anode inlet is supported by
recirculation from anode exhaust, the changes within the stack influence it considerably.
In Fig. 8a, fuel flow at anode inlet follows a similar trend to that of anode exhaust, as
well as the inlet temperature. Rapid reduction in temperature is due to lower volume of
recirculation which is at around 69°C whereas dry hydrogen from the tank is at 25°C.
Fluctuations in water content at anode inlet disturb fuel flow controls and relative
humidity constantly; they remain within acceptable ranges however. Figure 8c shows the
fuel stoichiometry changes and power consumed by recirculation pump, which are
affected by the depletion of hydrogen at anode cell sites and water diffusion during the
current surges. On cathode side (Fig. 8d), as expected, the air flow which is regulated by
a controller increases when the current is ramped up and steadies along with the system.
There is no specific temperature control of inlet air in this model, though external
coolant flows manipulate the air temperature which is further preheated by the
humidifier before entering the stack. Air temperature varies around 60°C which is equal
to the controlled temperature for coolant entering the stack; however temperature
difference at anode inlet is significantly higher during these load variations. Such
temperature gradients incur adverse effects as they form thermal stresses in the stack
with co–flow configuration and reduce its life cycle.

5.4 Effects of water saturation at cathode

Water removal from a fuel cell cathode is dependent on stack temperature and pressure
drop. Temperature is more critical factor of the two, since at high temperature the water

C.16
 Paper II

will be in vapour state and easier to remove. Stack water production increases with
current and is also dependent on the number of cells. The exact amount of liquid water
product depends on cathode outlet temperature. Inlet humidity also contributes to
liquid water saturation in the fuel cell, as reported in Wong et al. [19]. Figure 9 provides
information on the amount of liquid water at outlet of the cathode channel for two
different current surge amplitudes.
50.0 100.0 150.0

50.0 100.0 150.0

(a) Ramping of current: 60-100-60A (b) Ramping of current: 60-120-60A
-2.0 0.0 2.0 4.0 6.0 8.0 10.012.014.0

0.15 0.2 0.25 0.3 0.35

60.0 65.0 70.0 75.0 80.0

0.15 0.2 0.25 0.3 0.35

Operating temperature (C)
Heat generated (kW)

Liquid fraction

Liquid fraction
Current (A)

Current (A)
0.0

0.0
-100.0 -50.0

-100.0 -50.0
0.1

0.1
55.0
750.0 800.0 850.0 900.0 950.0 1000.0 550.0 600.0 650.0 700.0 750.0 800.0 850.0
Time (Seconds) Time (Seconds)

Figure 9. Water crossover through PEMFC membrane; (a) Current ramp–up from 60A to 100A
and vice versa, (b) Current ramp–up from 60A to 120A and back.

It can be seen in Fig. 9a, that there is an abrupt increase in amount of liquid water at
cathode when the current is ramped from 60A to 100A. When higher currents are
drawn, reactions within the fuel cell are accelerated which demand more intake of fuel
and oxidant. Consequently these reactions produce more water when compared to
production at low currents. Since the stack temperature does not elevate till that
instance, higher percentage of produced water is saturated at fuel cell cathode. Heat
produced by cell reactions then elevates the stack temperature, thereby reducing
saturated water at the outlet. Moreover, condensation of water at cell sites produces
additional heat which rapidly increases the stack temperature. It can be further noticed
in Fig. 9a that heat produced in the stack lowers as water liquid fractions drop. On the
other hand, when current is reduced back to 60A, a similar but opposite profile is
observed and the amount of liquid water tends to increase with a sink in stack
temperature. Therefore, at low temperatures and currents, water removal is a dominant
factor and stoichiometries are determined by the minimum flow rates required for water
removal which in the present case are more than adequate to provide the necessary
concentrations. Figure 9b shows water saturation results when ramping of current is set
from 60A to 120A instead. Although the data profile is analogous to that of Fig. 9a, it is
noted that amplitude of these peaks is higher when compared. Apparently, the amount
of liquid water at cathode exit is same for both cases when the fuel cell is operating at
steady state. Heat produced by condensation requires additional flow of coolant to

C.17
Paper II

maintain stack operating temperature, yet it does not affect the system efficiency to a
greater extent as liquid pumps do not consume that much power.

5. Conclusions
In this study, a comprehensive dynamic model of a PEMFC system along with the
auxiliary components is presented and a substantial emphasis has been set to devise a
control–oriented dynamic model of the fuel cell stack, which accommodates the
electrochemical, thermal, feed flow and water transportation models. Main
contributions of the proposed model are ascribed to the dynamic system responses,
which is characterized mainly by heat management and water transportation within the
fuel cell.

It is observed that thermal management strategy greatly influences voltage output and
system efficiency which increase with stack operating temperature. Moreover, slow
temperature controls affect the stability of fuel cell operations. In order to contemplate
fast electrochemical changes in the stack, high coolant mass flow rates are applied.
Power consumed by liquid coolant pumps is minimal and have no considerable effect
on system efficiency, whereas air radiator consumes most of the power in thermal
management system. Further, work is required on air radiator to ensure thermal stability
of fuel cell operations and prolonged stack lifetime. A 5% increase in power
consumption of air blower and radiator is observed during load variations for the cases
presented here.

Furthermore, water crossover in the fuel cell has shown a significant impact on PEMFC
anode operations. Anode inlet flows, humidity and recirculation pump are influenced
by net water diffusion during load changes. Temperature changes at anode inlet are
considerably higher during load variations and have a negative impact as they generate
thermal stresses in the stack and reduce its lifetime. Also, amount of saturated water at
cathode is dependent on operating temperature which apparently pertains to thermal
management strategy of the system. At low currents and temperatures, reactant
stoichiometries are determined by the minimum flow rates required for water removal
from the stack.

Acknowledgement

The authors would like to thank the Danish Energy Agency for financial support and
our industrial partner, H2Logic, for their collaboration and technical support.

References

C.18
 Paper II

[1] J. C. Amphlett, R. F. Mann, B. A. Peppley, P. R. Roberge, A. Rodrigues, A model

predicting transient responses of proton exchange membrane fuel cells, J. Power
Sources 61(1996) 183–188.

[2] S. Yerramalla, A. Davari, A. Feliachi, Dynamic modeling and analysis of polymer

electrolyte fuel cell, IEEE Power Engineering Society Summer Meeting 1–3 (2002)
82–86.

[3] M. Ceraolo, C. Miulli, A. Pozio, Modelling static and dynamic behaviour of

proton exchange membrane fuel cells on the basis of electro–chemical
description, J. Power Sources 113 (2003) 131–144.

[4] J. T. Pukrushpan, H. Peng, A. G. Stefanopoulou, Control–Oriented Modeling

and Analysis for Automotive Fuel Cel Systems, J. Dynamic Systems Measurement
& Control 126 (2004) 14–25.

[5] P. R. Pathapati, X. Xue, J. Tang, A new dynamic model for predicting transient
phenomena in a PEM fuel cell system, Renewable Energy 30 (2005) 1–22.

[6] G. Hu, J. Fan, S. Chen, Y. Liu, K. Cen, Three–dimensional numerical analysis of

proton exchange membrane fuel cells (PEMFCs) with conventional and
interdigitated flow fields, J. Power Sources 136 (2004) 1–9.

[7] S.K. Park , S.Y. Choe, Dynamic modeling and analysis of a 20–cell PEM fuel cell
stack considering temperature and two–phase effects, J. Power Sources 179 (2008)
660–672.

[8] J. Jia, Y. Wang, Q. Li, Y. T. Cham, M. Han, Modeling and Dynamic

Characteristic Simulation of a Proton Exchange Membrane Fuel Cell, IEEE
Transactions on Energy Conversion 24 (2009) 283–291.

[9] G. Vasu, A. K. Tangirala, Control–orientated thermal model for proton–

exchange membrane fuel cell systems, J. Power Sources 183 (2008) 98–108.

[10] M. J. Khan, M. T. Iqbal, Modelling and analysis of electrochemical, thermal, and

reactant flow dynamics for a PEM fuel cell system, Fuel Cells 5 (2005) 463–475.

[11] Y. Shan, S.Y. Choe, Modeling and simulation of a PEM fuel cell stack considering
temperature effects, J. Power Sources 158 (2006) 274–286.

[12] A. J. del Real, A. Arce, C. Bordons, Development and experimental validation of

a PEM fuel cell dynamic model, J. Power Sources 173 (2007) 310–324.

C.19
Paper II

[13] J.W. Ahn, S.Y. Choe, Coolant controls of a PEM fuel cell system, J. Power
Sources 179 (2008) 252–264.

[14] J.H. Jung, S. Ahmed, Dynamic model of PEM fuel cell using real–time simulation
techniques, J. Power Electronics 10 (2010) 739–748.

[15] S. Asghari, H. Akhgar, B. F. Imani, Design of thermal management subsystem for

a 5kW polymer electrolyte membrane fuel cell system, J. Power Sources 196
(2011) 3141–3148.

[16] A. Beicha, Modeling and simulation of proton exchange membrane fuel cell
systems, J. Power Sources 205 (2012) 335–339.

[17] Ballard Mark9 SSL, Product manual and integration guide 2008.

[18] E. Hosseinzadeh, M. Rokni, Development and validation of a simple analytical

model of the Proton Exchange Membrane Fuel Cell (PEMFC) in a fork–lift truck
power system, Intl. J. Green Energy, In Press (2012) DOI:
10.1080/15435075.2012.678525.

[19] K. H. Wong, K. H. Loo, Y. M. Lai, S.–C. Tan, C. K. Tse, A theoretical study of

inlet relative humidity control in PEM fuel cell, Intl. J. Hydrogen Energy 36
(2011) 11871–11885.

Nomenclature
E theoretical voltage V 
Vcell average cell voltage V 
Pel stack power kW 
Pin energy into the fuel cell kW 
Pout energy out of the fuel cell kW 
Q loss heat dissipated kW 
Ct stack thermal capacitance kW 
R universal gas constant J molK 
T temperature K 
F Faraday’s constant C molK 
I current  A
N cell number of cells  
g 0f change in Gibbs free energy J molK 
PH 2 hydrogen partial pressure  

C.20
 Paper II

PO2 oxygen partial pressure  

Mm mol. weight of membrane Kg mol 
J H 2O net water–diffusion flux mol scm 2 
D water diffusion coefficient cm 2 s 
aH 2 hydrogen activity  
a H 2O water activity  
aO2 oxygen activity  
i current density A cm 2 
in internal current density A cm 2 
i0 exchange current density A cm 2 
i0 , a anode exchange current density A cm 2 
i0 , c cathode exchange current density A cm 2 
ka anode reaction rate mol scm 2 
kc cathode reaction rate mol scm 2 
ne electrons transferred mole mol fuel 
nel number of electrons  
ndrag electro osmotic drag  
tm membrane thickness cm

Greek symbols
a anode transfer coefficient  
c cathode transfer coefficient  
 symmetry factor  
 act activation overpotential V 
 act ,a anode activation overpotential V 
 act,c cathode activation overpotential V 
 conc concentration overpotential V 
 ohmic ohmic overpotential V 
 membrane water content  
 dry membrane density g cm3 

C.21
Paper II

C.22
 Appendix D

Paper III

223
Appendix D

Rabbani, A & Rokni, M. “Effect of nitrogen crossover on purging strategy in PEM

fuel cell systems”, accepted for Journal of Applied Energy, 2013.

224
 Paper III

Effect of nitrogen crossover on purging strategy in PEM fuel

cell systems

Abid Rabbani and Masoud Rokni

Thermal Energy Section, Department of Mechanical Engineering, Technical University
of Denmark, 2800 Kgs. Lyngby, Denmark.

Abstract
A comprehensive study on nitrogen crossover in polymer electrolyte membrane fuel cell
(PEMFC) system with anode recirculation is conducted and associated purging strategies
are discussed. Such systems when employed in automobiles are subjected to continuous
changes in load and external operating conditions, making it important to investigate
the dynamic performance of the system during transitory conditions. The model
developed here is able to predict nitrogen crossover in excellent agreement with the
design validation data of the stack. The results show that with pure recirculation, voltage
and system efficiency decline due to nitrogen accumulation in fuel cell. Different
purging techniques are simulated to address hydrogen dilution issue at reaction sites.
Anode bleed out of 3% is found to be limit for prevention of N2 buildup and retains the
concentration levels to less than 1%. An alternate strategy for automatic initiation of
anode recirculation purge was simulated by employing nitrogen detectors. It is observed
that purge interval is a direct function of current density and H2 residual flow rates.
Moreover, during transient load changes, automatic purge catered well to prevent
nitrogen levels from rising when compared to a fixed purge interval strategy. This model
can be used as a base for control and development of anode purge strategies for
automotive fuel cell systems.

Key words: Dynamic simulation, fuel cell, PEMFC, nitrogen crossover, anode purge.

Introduction
Hydrogen is being anticipated as the fuel of the future and fuel cells are gaining
significant importance in power generation technologies and in establishing the
hydrogen economy. Polymer Electrolyte Membrane Fuel Cells (PEMFCs) run on
hydrogen as a fuel, which only produces water when used with air as the oxidant. These

D.1
Paper III

types of fuel cells will prove to be vital in reducing our dependency on fossil fuels and
diminish harmful emissions into the atmosphere. Further advantages of PEMFCs
include low operating temperatures, high power density and swift start–ups making
them an ideal candidate for automotive applications. Conversely, these automobiles
operate in an environment subjected to continuous changes in load and external
operating conditions, which compromises durability and reliability of the fuel cells.
In a typical PEMFC, hydrogen is fed into the anode of the fuel cell stack and oxygen/air
into the cathode, while the water produced from cell reactions is ejected out of the
cathode outlet. Since the current generated in the fuel cell is a direct measure of the rate
of electrochemical reactions, power produced by the stack is associated with the rate at
which the reactants are consumed. Ideally, all of the hydrogen and oxygen needed for
any specific power of the fuel cell stack would be consumed and only water and unused
nitrogen of the air would leave the system. However, not all hydrogen and oxygen
delivered at the reaction stoichiometry reaches the cell sites. At a stoichiometry of 1.0,
the reactant partial pressure at the stack exit reaches very close to 0, which can cause fuel
starvation leading to irreversible damage to the stack in addition to a lower power
output [1]. Therefore, higher stoichiometries of reactants are fed into the system. In
most systems, the unused hydrogen is recirculated back into the feed stream, thereby
increasing the system efficiency.
As already mentioned, one of the benefits of PEMFCs is their high power density which
is essential for portable applications. Recent advancements in manufacturing have made
these stacks more compact by significantly reducing the thickness of Membrane
Electrode Assembly (MEA), though enhances gas crossover across the membrane
resulting in current losses within the fuel cell itself. One of these gases is nitrogen from
the air, which permeates from the cathode to anode and gets concentrated in the anode
due to recirculation. Similarly, water produced on the cathode side is also transported
towards the anode through the membrane. This build–up of nitrogen and water reduces
the concentration of hydrogen in anode channels and obstructs hydrogen molecules to
reach cell reaction sites effectively, which adversely affects the voltage developed in the
cell [2], [3] . Also, H2 starved areas are developed within the fuel cell which can also
result in carbon corrosion [4], [5]. Furthermore, additional flow of these crossed over
gases increases the power consumption of recirculation pump, thereby adding to the
efficiency loss due to the voltage drop [6]. In most systems with anode recirculation, N2
build–up is usually avoided by frequently purging a certain percentage of the anode
outlet gas before mixing it with the inlet stream. An alternative approach is to
continuously bleed a small fraction of the recirculation stream. While both mentioned
approaches release some portions of nitrogen as well as water vapor from the
recirculated stream, it also purges unused hydrogen. In this context, the amount purge
fraction and time intervals of these purging techniques are of particular interest to the
authors. An optimum strategy would cease the build–up of N2 with a compromise of
minimum H2 wastage. In this paper, we present a dynamic model of N2 build–up in

D.2
 Paper III

anode under recirculation and compare different purge strategies for optimal system
performance.
Literature available on water and gas crossover through PEMFC membranes is not
uncommon. Kocha et al. [2] characterized gas crossovers by applying an in–situ
electrochemical technique to determine hydrogen crossover rates and devised a model
to predict the extent of nitrogen accumulation along the anode flow fields. Ahluwalia
and Wang [7] modeled and analyzed the buildup of N2 in the recirculating anode gas
and the impact of this buildup on the performance of 90 kW PEMFC stack. Catalano et
al. [8] reported the effects of relative humidity on gas permeability and swelling in
Nafion membranes used in fuel cells. Baik and Kim [9] calculated nitrogen permeability
coefficients (NPC) by employing a mass spectrometer and analyzed N2 crossover under
open circuit voltage and power generation conditions. Ismail et al. [10] conducted a
parametric study to investigate the effect of transport properties of GDLs on fuel cell
performance. Weber [3] reported the effects of membrane pinholes and increase in gas–
permeation on PEMFC’s performance. It was also shown that water and thermal
management is also affected where there are large pinholes in the membrane.
There are a few published papers investigating the purge process modeling in general or
specifically pertaining to individual aspects of the process. Zhu et al. [11] experimentally
determined the critical flow rate in anode exhaust stream by manual purging of Nexa
Power module. Tang et al. [12] also investigated the transient response of the
aforementioned module and identified factors influencing its performance. Zhai et al.
[13] also presented their study on anode water flooding and simulated the gas purge
effect for medium current densities. Gou et al. [14] presented a one–dimensional
computational model to study the dynamic behavior of pressure in anode flow field
during the purge process. It was shown that different current densities had minimal
influence on the pressure drop and the pressure swing during this transient process. A
dynamic three–phase transport model was developed by Wang et al. [15], which lead in
identification of the optimum water uptake parameters for purge cycles during startup
from subfreezing temperatures and subsequent shutdowns. Karimaki et al. [16] setup a
test bench to study inert gas buildup effects on fuel cell stack and introduced the use of
online hydrogen sensor to measure gas crossover across the cell. Promislow et al. [17]
built a simple analytical model to describe the steady state profile of anode N2
concentration in PEMFCs with straight gas channels and identified an optimum bleed
rate for the specified stack.
Some others have published papers addressing the anodic dead–end mode of operations
in the fuel cell anode. Hou et al. [18] developed a dynamic voltage model for a ‘dead–
ended’ FC to simulate hydrogen purging in order to prevent water accumulation in the
anode. Muller et al. [19] also correlated N2 accumulation with a 20–cell temporal fuel
cell performance operating in a ‘dead–end’ mode by estimating permeability
characteristics of typical Nafion membranes. Siegel et al. [20] modeled a PEM fuel cell

D.3
Paper III

operating with a ‘dead–end’ mode and experimentally verified the evolution of liquid
water and nitrogen fronts along the length of anode channels. They further developed
the 1D model with periodically–purged anode channels and incorporate simple
resistance model for predicting nitrogen permeance and water transportation through
the membrane and presented their findings in Yesilyurt et al. [21]. Another
experimental study was conducted by Choi et al. [22] in which purge characteristics of a
cathodic dead–end mode PEMFC for a submarine were analyzed.
Although, there are a few publications dealing with nitrogen crossovers and associated
purge processes, but to the best of authors’ knowledge, there is no available study on
intricate details of nitrogen crossovers and the effect of purge cycles on the fuel cell
performance during transient operations. The present work attempts to simulate the
dynamic characteristics of the PEMFC system operating with purge cycles, and
investigate its performance during transitory load changes. For this purpose, a
comprehensive fuel cell model with all the necessary auxiliary components is built and
system simulations are carried out by employing a detailed control scheme in order to
emulate a practical operation environment. N2 accumulation in fuel cell anode and its
effect on system performance is investigated, and in view of this buildup, different
purging strategies are assessed. At last, an alternate method for nitrogen detection and
automatic purging is proposed and compared with conventional methods during
different stages of external load changes. This model could be regarded as a basic
foundation in design and optimization of purge cycles for automotive systems running
PEMFCs.

System configuration
Layout of the PEMFC system under investigation is shown in Fig. 1. The system
comprises a PEMFC stack, air compressor, humidifier, pumps, heat exchangers and
radiator for the cooling circuit, flow valves and controllers. Hydrogen from storage tank
is regulated by a control valve into the fuel cell anode. On the cathode side, compressed
air which is fed into stack is cooled and humidified prior to its entrance into fuel cell
flow fields. Since the stack is not operated at dead–end mode, a higher fuel
stoichiometry is maintained. Unused fuel from anode exhaust is recirculated back to the
anode inlet via a recirculation pump. Recycling of the exhaust stream eliminates the
need of fuel humidifier as it holds enough water to humidify the fuel. Liquid water
present in the anode exhaust is collected in a water trap, whereas the remaining water
present in the stream is purged with the other gases by a solenoid valve before mixing
with the inlet stream. In addition to electrical power, heat is produced by the stack. This
heat is absorbed by a liquid coolant which circulates in a circuit associated with the stack
and a heat exchanger. An external cooling loop, connected to the aforementioned heat
exchanger, in turn cools the water in the internal circuit. This circuit also consists of a

D.4
 Paper
P III

heat exxchanger to precool aiir entering the fuel ceell and an air
a radiator for heat reejection.
Flow off water is reegulated byy pumps in the respecctive circuitts.

Figure 1. Schematiic layout of the PEMFC

C system

PID coontrollers are

a deployed to regulaate reactan nt and coollant flows. These con ntrollers
collect data from various coomponent inlet and outlets
o whiich are reggarded as pressure,
p
temperature and flow transmitters, an nd manipu ulate the coorrespondinng compon nents to
reach thhe desired state. Duee to this faact, currentt system responses annd its behaavior are
attributted to the formulated
f d control strrategy whicch is based on fuel ceell stack lim
mitations
and reccommendattions by th he manufac turer. Detaailed contro ol mechaniisms are illuustrated
in subseequent subbsection.

Fuel Ce
Cell stack
A Ballaard fuel celll stack [23]] has been selected foor the curreent study. T
The fuel cell stack
containns 110 cellss with a cell area equual to 285 cm2. Maximum pow wer produceed from
the desscribed staack is 21.2 2 kW corr esponding to a currrent of 3000A; howevver it is
generallly operated d at lowerr current rranges to attain high her efficienncies by reducing
r
ohmic and concen ntration ovverpotentiaals within the
t fuel cell stack. Thhis model, which
w is
based oon equation ns adopted d by Hosseiinzadeh an nd Rokni [1 1], containss some parrameters
influen
ncing physiccal characteeristics of tthe system, as well as on operatiing conditions and
membrrane properrties. Here, focus hass been set to build up u a system m which meets
m the
requireements of actual stacck runningg under reecommended conditiions, see Table T 1.
Duringg PEMFC steady–statee operation ns, coolant pressure shhould be loower than reactant
pressurres whereas during start–up,
s ccoolant prressure maay exceed reactant pressure.
p

D.5
Paper III

Pressure in the anode is usually set higher than to that of the cathode, which will
minimize nitrogen crossover and ensure cell stability.

Table 1. Nominal operating conditions for PEMFC stack

Current (A)
Reactant Parameter
15 30 60 120 240 300
Fuel (Pure Hydrogen)
Inlet stoichiometry Min. 6.3 3.4 2.2 1.9 1.6 1.6
Nominal Inlet Pressure (kPa) 115 116 131 155 200 220
Pressure Drop (kPa) 9 13 13 14 16 18
Oxidant (Ambient Air)
Inlet stoichiometry Min. 5.1 2.4 1.8 1.8 1.8 1.8
Nominal Inlet Pressure (kPa) 108 110 117 138 180 200
Pressure Drop (kPa) 8 13 12 16 40 53
Relative Humidity (%) 95 95 95 95 95 95
Coolant
Inlet Temperature Max. (°C) 60 60 60 60 60 60
Outlet Temperature Max. (°C) 61 63 66 67 68 70
Minimum Coolant flow (lpm/cell) 0.05

Balance of Plant
A brief overview of balance of plant (BoP) components is presented in this section,
though a detailed description is reported in authors’ previous work [24]. In the
proposed configuration, a humidifier is placed in conjunction of cathode inlet and
exhaust, where it utilizes the water produced by chemical reaction inside the fuel cells to
humidify inlet air. It is assumed that inlet air is optimally humidified after passing
through the humidifier. Relative humidity of air entering the cathode is arbitrarily set to
95% in the simulations. This assumption could be justified as it is very close to real
operational conditions. Auxiliary components such as blowers, pumps and valves
regulate flows of material streams in PEMFC systems. For hydrogen feed, a valve is
placed between the hydrogen tank and inlet manifold of anode which enables or
disables the hydrogen supply. Since the system does not operate on dead–end mode, the
amount of hydrogen regulated by this valve equals the stoichiometric hydrogen required
by the fuel cell. An air blower regulates the flow and pressure of oxidant into the
cathode.
A network of heat exchangers and radiators is deployed for heat management of the
system. Although heat exchanger models used here are predefined in Aspen
DynamicsTM, some of the parameters have been assumed on the basis of media entering

D.6
 Paper
P III

the hott and cold sides

s of theese heat excchangers. The
T heat exxchanger, w which is connected
to the iinternal coooling loop,, has liquidd water on both its ho
ot and coldd side. Therrefore, a
UA vallue of 1.2 kW/K
k is asssumed. WWhereas, UA A values for air pre–ccooler and radiator
are appproximated d to be 0.1 kW/K an d 0.3 kW//K respectiively. Modeels for pum mps and
TM
T
blowerss are also contained
c in Aspen D Dynamics library. Since the nnominal power p of
the PEM MFC is only 21 kW, mass flow rates of fueel and air are a very low w. For exam mple, at
an averrage load of 10 kW, fuelf and airr flows are around 0.00014 kg//s and 0.00 088 kg/s
respectiively. Therrefore, veryy low valuees of isentropic efficiencies aree suggested d in this
paper. The efficiiency of a blower rranges from m 15% to o 48% in the calcu ulations,
dependding on thee air mass flow.f Calcuulated pummp efficienccies are alsso very low
w for the
coolingg water circuits and arre determin ned to be arround 70%%.

Controol System
A systeem requirees an effective contrrol strateggy to regullate system m parameters and
operatinng conditiions to ennsure a staable and effective
e opperation [225]. Proportional–
integrall (PI) con
ntrollers, which
w are widely useed in industrial conntrol systems, are
employyed to regu ulate differrent compoonents and d flow streeams. Key parameterrs to be
controllled in thee proposedd system arre reactantt inlet stoiichiometriees, inlet prressures,
coolantt inlet and operating temperatur
t res of the sttack.

Figure 2. Basic contrrol blocks for the PEMF

FC system

D.7
Paper III

While the amount of oxygen consumed depends on the stack current, the amount of
oxygen supplied to a fuel cell is directly related to the blower power. Therefore, an
algorithm based on the above figure is developed to be the process variable for PI
controller, which regulates blower power to maintain the desired oxygen ratio. Similarly,
an algorithm for controlling hydrogen flow is devised along with a PI controller, which
regulates the control valve opening for optimal fuel supply.
Fig. 2 describes the process and manipulated variables implemented in the control
scheme. Temperature in the stack can be controlled by coolant flow rate which acts as
an input signal and is adjusted by the PI controller. Based on data from table 1,
equations defining stack temperature as a set–point for controller are developed. In
addition, the controller simultaneously collects data from temperature transmitter at
coolant outlet stream, which then changes the coolant flow accordingly by sending
output signals to the driving pump. Employment of such algorithm in the system
controller ensures a stable operation under normal steady–state conditions, however for
system start–up scenario, a different approach is required. Temperature of the coolant
entering the stack can similarly be controlled by flow of water in the external circuit.
Control signal to the associated pump regulates electrical power of the pump and hence
the coolant inlet temperature into the stack. In a similar fashion, temperature of water
in the external circuit is dependent on radiator fan speed. PI controllers are used to
regulate the fan speed as well.

Model formulation
Nitrogen crossover
For PEMFC membranes, high proton conductivity and low ionic resistance has been
achieved by directing significant efforts towards minimizing membrane thickness which
subsequently promotes water crossovers across it to humidify the anode side of the fuel
cell [26]. Moreover, high power density is also attributed to the membrane thickness in
addition to GDL and stack assembly. Limitations for physical stability of the cell and gas
crossovers have to be balanced by selection of a membrane with reasonable thickness. It
is however reported by [9] and [27] that membrane thickness and equivalent weight does
not considerably effect the gas permeability coefficients, but influence of water content
in the membrane is significant enough. [28] suggested that in fact N2 crossover is the
sum of two parallel processes; gas diffusion through polymer and water phase of the
ionomer respectively. Their simple model was correlated to the functional form by [7].
Here, a similar permeation model of [20] with a scale factor of 8 and equations with the
influence of stack temperature and membrane water uptake is used.

D.8
 Paper III

 EN  1 1 
K N 2   N 2 (0.0295  1.21 f v  1.93 f v )  10 11  exp  2
2
   (1)
T 
 R  ref T 

The activation energy for nitrogen, E N is assumed to be 24 kJ/mol,  N is scale factor, R

2 2

is the universal gas constant, Tref is 303K and f v is the volumetric ratio of water in the
membrane and is given by;

 memVw
fv  (2)
Vmem   memVw

where mem is the membrane water content, Vmem and Vw are molar volumes of dry
membrane and liquid water.
In a PEMFC, concentration gradient across the membrane is the driving force for N2
diffusion from cathode to anode. Since concentration of a certain species could be
related to its partial pressure in a volume, the nitrogen flux is calculated from the partial
pressures of N2 in cathode and anode of the fuel cell.

PN 2 ,ca  PN 2 ,an
J N2  K N2 (3)
t mem

where t mem is the thickness of the membrane and is assumed to be constant in the
current simulations.
For flow model of anode and cathode, the model uses mass balance to calculate the inlet
and outlet properties of streams. The partial pressures of different species are
determined from the properties of the gas streams entering and leaving the system, gas
and water crossovers, products formed and depletion of reactants during chemical
reactions within the fuel cell stack model. Equations for water transportation across the
membrane and losses associated with hydrogen crossover are shown in the table 2.

Fuel cell stack model

Aspen DynamicsTM, which is a simulation tool for process modeling and energy system
analysis, is used to construct the system model of the configuration described above.
The program contains a vast library of components and controls for standard energy
processes, making it a suitable candidate for system level simulations. One such dynamic
model is presented by [29] . The PEMFC system studied here is based on previous
studies by [1], [24] and [30]. The model code is developed and implemented in Aspen
Custom ModelerTM. A simple model of humidifier is also executed into the code. This
model which incorporates governing equations for cell electrochemical, polarization
overpotentials, heat transfers and water diffusion across the membrane is employed into

D.9
 Paper III

Aspen Dynamics and system controls are implemented in order to ensure stable
operation of the plant during load changes. Thermodynamic efficiency and net power of
the system are determined by the current drawn and voltage produced by the stack.
Total energy into the fuel cell is consumed by electrical power output, heat removed by
the coolant, heat loss at the stack surface and energy stored by the stack itself. In the
current model, a lumped thermal model proposed by [31] is considered. The stack is
regarded as a single thermal mass with a heat capacity. With the assumption of stack
temperature being equal to coolant temperature at the outlet, heat exchanged with the
coolant and hence stack operating temperature could be determined. Subordinate
components in the BoP, i.e. anode recirculation and water pumps, air compressor,
mixers and heat exchangers are simulated by using default mathematical models
contained in Aspen Dynamics library. Table 2 shows a standard set of equations which
constitute the model used in the current study.

Table 2. Constitutive equations for PEMFC model.

Average cell voltage Vcell  E   act   ohmic   conc (4)

RT  a H 2O 
 g 0f
Nernst equation E ln (5)
ne F ne F  a H 2 aO2 0.5 

Theoretical cell voltage E

 g 0f
ne F

RT
ne F
1
ln PH 2 PO2 
 0. 5
 (6)

Change in Gibbs free

energy
 
g 0f  g 0f H 2O
 
 g 0f
1 0
2
gf O
H2

2
  (7)

RT  i  in  RT  i  in 
Activation losses in PEMFC  act   act ,c   act ,a  ln  ln (8)
 c F  i0,c   a F  i0,a 
Concentration RT  i  RT  i 
 conc   conc,c   conc,a  ln1   ln 1  (9)
overpotentials ne F  ilim,c  ne F  ilim,a 

Ohmic overpotentials  ohmic  rel  rion   i (10)

Ionic resistance in the cell rion 


C1  1  0.03i  0.062  T / 303  i 2.5
2
t
 (11)
ave  0.634  3i   expC 2  T  303 / T  m
Membrane water content  ave  0.043  17 .18 a w  39 .85 a w2  36 a w3 (12)

Pw
Water vapor activity aw  (13)
Psat
Water profile inside cell c   a
 mem  .z   a (14)
membrane tm

D.10
 Paper III

Transfer coefficient for  a    nel (15)

anode
Transfer coefficient for  c  1     nel (16)
cathode
Exchange current density at  1     nel  F  E 
i0,a  nel  F  k a  exp  (17)
anode  RT 
Exchange current density at     nel  F  E 
i0 ,c  nel  F  k c  exp  (18)
cathode  RT 
Net water transportation i  d
J H 2O  2ndrag  dry D (19)
through membrane 2F M m dz
  1 1 
D  4  10  6 exp  2416     
Water diffusion coefficient   303 T   (20)
2 .563  0 .33   0 .0264 2  0 .000671 3 
Fuel cell energy balance Ct  dT  Pin  Pout  Pel  Q loss (21)
dt
dN I
Faraday’s law  (22)
dt ne  F

Results and discussion

Using the PEMFC stack data and control parameters summarized in table 1, simulations
for the system are carried out using Aspen Plus Dynamics program. Reliability of the
fuel cell model and dynamic behavior of the system under consideration have already
been reported in authors’ previous studies [24], [32]. These studies have reported
transient response of different components of the system as well as their mutual
dependencies on control strategies, thermal management and water transportation of
the fuel cell stack under different operating scenarios; such as system start–up and
during transitory load changes. Here, simulation results for the effect of nitrogen
crossover and associated purging of anode recirculation are presented. Stack
temperature range of 60–70°C has been used in simulations of the current system.

Gas Buildup at anode

Based on the permeability of N2 and partial pressure gradient across the membrane,
simulations for the prescribed system were conducted and corroborated against
validation data by stack manufacturer [23]. Fig. 3 presents the simulated net nitrogen
crossover in the fuel cell stack against the experimental data. The corresponding relative
error advocates a reasonable agreement between the simulations and stack trial data.

D.11
 Paper III

Experimental Simulations Error

0,009 10
8

N2 crossover (kg/hr)
6
0,008
4

Error (%)
2
0,007 0
-2
-4
0,006
-6
-8
0,005 ‐10
40 60 80 100 120 140 160 180
Current (A)

Figure 3. Calculated nitrogen crossover rates against experimental data

Current drawn from the fuel cell is set to a constant 120 A with no purging of the
anode recirculation loop. Since pure H2 is used as the fuel, there is no presence of
nitrogen in anode inlet of the fuel cell at the start–up. Fig. 4a shows the N2 crossover
form cathode to anode side of the fuel cell during the start–up operations of the stack. It
is observed that as the temperature and water uptake of the membrane increases, N2
permeance increases, thereby increasing nitrogen crossover from the cathode to the
anode side of the cell. An abrupt decline in temperature is noticed at around 120 s,
which is due to the control initializations of coolant flows to maintain the stack
operating temperature. Details of the control setup and thermal management of the
system are reported in [24]. N2 permeance decreases with the sudden reduction in
temperature and subsequently reducing the crossover at that instant of time. However,
as the system stabilizes to steady state, the permeance becomes constant but N2 crossover
tends to decrease as can be seen in Fig. 4b. This is due to the fact that N2 accumulation
in the anode reduces the concentration gradient across the cells.
0.0065
0.008

10.0

80.0

16.0

80.0
N2 perm (kmol/s.cm.bar) exp-14

N2 perm (kmol/s.cm.bar) exp-14

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
Operating temperature (C)

Operating temperature (C)

Membrane water content
N2 crossover (kg/hr)

N2 crossover (kg/hr)
0.006

60.0

14.0
7.5

0.006

60.0
0.004

40.0

12.0
5.0

0.0055

40.0
0.002

20.0

10.0
2.5

0.005

20.0
0.0

0.0

0.0

8.0

0.0 100.0 200.0 300.0 400.0 500.0 600.0 700.0 800.0

Time (Seconds) Time (Seconds)

Figure 4. N2 permeance and crossover rate at (a) fuel cell start–up and (b) Steady–state operations

D.12
 Paper III

Depletion of hydrogen in the anode due to nitrogen accumulation leads to a slow decay
of the cell voltage. Fig. 5a represents the voltage decay due to increase in N2 fractions
and corresponding decrease in H2 concentrations in the fuel inlet. Although voltage
decay shows a linear relation with nitrogen accumulation, its impact over extended
operations results in hydrogen starved regions in the cell resulting in increased cathode
overpotential and resistive losses [21]. Another issue related to anode recirculation is the
accumulation of water, which clogs flow fields and results in lower concentrations of H2
as well. Fig. 5b points to the gradual increase in liquid water at anode outlet.

0.065

80.0
0.73

1.0
Anode inlet N2 (kmol/kmol)
Anode inlet H2 (kmol/kmol)

Stack temperature (C)

Liquid water (kg/hr)
0.729

0.75
Cell voltage (V)

60.0
0.728

0.06
0.5

40.0
0.727

0.25

0.055
0.726

0.0

400.0 500.0 600.0 700.0 800.0 400.0 500.0 600.0 700.0 800.0
Time (Seconds) Time (Seconds)

Figure 5. (a) Cell voltage decay due to N2 buildup and (b) Increment in liquid water at anode
exhaust

Anode outlet bleed

As shown above, N2 accumulation could lead to dilution of hydrogen in the fuel.
Moreover, power consumption of recirculation pump increases due to additional flow of
nitrogen and water vapor. One way of preventing accumulation of these gases is venting
of anode exhaust often called as bleeding. Although N2 and water are vented out of the
system, bleeding also wastes unused hydrogen resulting in lower system efficiency
realizing a need for an optimized bleed percentage. Here, different bleed fractions of the
anode recirculation stream are simulated and their impact on the system is analyzed. Fig.
6a displays the amount of N2 present in the anode inlet against different bleed fractions
at a drawn current of 120 A.

D.13
 Paper III

0.045
Bleed (3%)

0.025
2%

0.1
Bleed percentage 1%
0.040 3% 4%

Purged H2O (kg/hr)

0.005 0.00: 0.015 0.02

0.08
Purged N2 (kg/hr)
Purged H2 (kg/hr)
0.035

Purge fraction
N2 molar fraction

0.06
0.030

0.025

0.04
0.020

0.02
0.015

0.0
0.010
0.0 200.0 400.0 600.0
300 350 400 450 500 550 600
Time (Seconds)
Time (s)

Figure 6. N2 levels for different bleed fractions at anode inlet and (b) Composition of bleed stream
from system start–up

It can be seen that N2 keeps on increasing when the vent out fraction is 1% and remains
constant around 3% of the anode exhaust. At this bleed rate, molar N2 remains close to
1.5% of the fuel entering the stack. Fig. 6b shows the constituent flow rates of 3 %
vented stream. A large amount of water is also vented along with nitrogen. A constant
loss of hydrogen at a rate of 0.011 kg/hr (2% of the supplied fuel) would suffice to
prevent gas accumulation in anode of the PEMFC.

Anode outlet purge

A purge of the anode recirculation is another method to reduce nitrogen and water
levels in the fuel inlet. A purge is initiated by opening of a solenoid valve placed at the
anode outlet. This valve opens for a short duration (purge time), removing a portion of
gases in the stream. The time between these purges is referred to as purge interval.
Though any combination of these could be assumed, here arbitrary values of 0.6s, 0.3s,
0.9s and 60s of purge valve opening time, opening duration, valve closing time and
purge interval are selected respectively. Here, a steady state analysis is conducted by
drawing a current of 120 A and the valve opening is set to purge 20% of the exhaust
stream. Modeling of the valve and its pressure effects have been neglected here as it falls
out of the scope of study. However, [14] have studied the behavior of pressure during
purge process and reported a uniform distribution of pressures for short purge times as
used in current simulations. In addition, it is reported that fluctuations in anode
pressure due to frequent purging, also supports liquid water removal from the channels.
Fig. 7a shows a semi–stable cell voltage during these purge events. As soon as the purge
is initiated, it removes both liquid water and nitrogen gas and recovers the voltage drop
caused by reduced active cell sites.

D.14
 Paper III
0.731

0.925

0.08

9.64

0.51
0.9
Anode inlet H2 (kmol/kmol)

Anode inlet N2 (kmol/kmol)

Auxiliary power (kW)

9.62
0.9

Stack Power (kW)

System efficiency
0.06

0.89
Cell voltage (V)

0.5
0.73

0.825 0.85 0.875

9.6

0.88

0.49
0.04

9.58
0.729

0.02

0.87

0.48
9.56
0.728

9.54

0.86
0.8

500.0 600.0 700.0 800.0 500.0 600.0 700.0 800.0

Time (Seconds) Time (Seconds)

Figure 7. Performance of Fuel cell for fixed purge interval of 60 s, (a) Voltage rejuvenation with
semi–stable profile and (b) system efficiency and power during purge sequences

N2 and H2 fractions could also be noticed to change with each purge. During the purge
intervals, there is a steady buildup of nitrogen which gradually decays the cell voltage
which lowers the efficiency. This can be seen in Fig. 7b where efficiency of the system
increases with each purge and slowly lowers before the next purge event. It is also
noticed that there is an abrupt decline in the system efficiency at purge instances. This is
due to the fact that unused hydrogen leaving the purge valve reduces the recyclable
amount, thus these drop lines. Change in power consumption of auxiliary components
is minimal as only low–power recirculation pump is affected by the purge. Since the
total purge time is only 1.8 s and purge interval is 60 s, it is cumbersome to determine
the average efficiency lost for long durations of fuel cell operations. Nonetheless,
amount of H2 vented to the atmosphere could be determined to assess the economics of
the operations. Fig. 8 explicates that around 0.08 kg/hr (1.2% of supplied fuel) of H2 is
wasted per purge event with interval of 60 s.
0.12
0.25
0.8
Purged H2 (kmol/kmol)

Purged N2 (kmol/kmol)

0.0 0.03 0.06 0.09

H2 ventout (kmol/hr)
0.2
0.75

0.1 0.15
0.7

0.05

-0.06 -0.03
0.65

0.0

450.0 500.0 550.0 600.0 650.0

Time (Seconds)

Figure 8. Molar fractions at purge routines and H2 vent out rate

D.15
Paper III

Automatic Anode purge

As mentioned earlier that in addition to reducing the fuel cell voltage, enrichment of
nitrogen could lead to corrosion of the electrodes. With the ageing of the cell
membrane it tends to get thinner and due to the degradation, micro cracks and pinhole
regions are generated which further increase N2 crossover. Another approach simulated
in the present investigation is the use of an automatic control for optimization of purge
intervals. For practical diagnosis of fuel cell systems, nitrogen at anode inlet could be
detected by arc emission spectroscopy methods. Signal from these detectors is
transmitted to the controller, where it adjusts the parameters for effective purges. An
optimum purge sequence would prevent N2 buildup at a minimum expense of released
hydrogen. Optimization of the purge routine could be carried out by manipulation of
many variables such as valve opening and closing times, purge time, purge interval and
purge stream fractions. Simulations for different combinations of these variables for
different power loads would require tremendous amount of time and effort. For the
sake of comparison purposes, only purge intervals are chosen to be the manipulated
variable in the present simulations.
According to [23], molar N2 fraction of 5% or less has no or minimal effect on stack life.
A purge initiation limit of 3% has been set i.e. the purge event will occur only when N2
molar concentration of 3% or more is detected at anode inlet. Simulations for two
different currents are carried out here. Voltage recovery and efficiency profiles for
automatic purging are found to be similar to that of fixed time interval as the purge
process is same. However, N2 crossover in the cell changes with current density as
reported by [2]. Typically N2 crossover increases with current density, as more heat is
produced to raise the temperatures and increase nitrogen permeance. Yet in a system
such as the one under investigation, system controls maintain the temperature of the
stack and prevent excessive gas crossovers. Also, stoichiometric ratios of anode
determine the flow rate of supplied H2 and are inversely proportional to the N2
concentration in the anode exit stream [9].
Figure 9a shows N2 and H2 levels in anode recirculation after purging and
corresponding flow rate of hydrogen purged to the atmosphere. When the current
drawn during steady state operation is 120 A, H2 is purged at a rate of 0.08 kg/hr,
similar to the rate at fixed interval purge in Fig 8. However, automatic purging interval
is found to be around 85 s, saving 15 s per purge. In Fig. 9b, when the current is drawn
at 60 A, H2 purge rate is reduced to 0.06 kg/hr and purge interval is found to be 55 s.
This is due to the fact that H2 flow rate at a low current density is relatively smaller
compared to that at a high current density, so is the anode exit flow rate. Therefore, low
currents will result in a high N2 levels in the anode exhaust which will require frequent
purging. Fuel losses associated with current case are found to be 0.8% of the supplied
fuel. Based on the data from table 1, higher stoichiometric ratios are used at low
currents.

D.16
 Paper III

0.25

0.12

0.25

0.12
0.8

0.8
Purged H2 (kmol/kmol)

Purged N2 (kmol/kmol)

Purged H2 (kmol/kmol)

Purged N2 (kmol/kmol)
0.0 0.03 0.06 0.09

0.0 0.03 0.06 0.09

H2 ventout (kmol/hr)

H2 ventout (kmol/hr)
0.2

0.2
0.75

0.75
0.15

0.15
0.1

0.1
0.7

0.7
0.05

0.05
-0.06 -0.03

-0.06 -0.03
0.65

0.65
0.0

0.0
450.0 600.0 750.0 450.0 600.0 750.0
Time (Seconds) Time (Seconds)

Figure 9. Automatic purge intervals and corresponding molar fractions of purges stream (a) at
drawn current of 120 A, and (b) 60 A

Anode purge at load changes

PEMFC systems have to operate at varying load and operating conditions when used in
automotive applications. Nitrogen concentration would vary with changing loads which
requires an efficient purging strategy to cater these dynamics. For this particular
simulation, current load is changed from 120 A to 60 A and vice versa at 380 s and 410
s respectively. Results for different purging methods are presented in Fig. 10 and Fig. 11.
150.0

0.00:
N2 crossover (kg/hr)
100.0
Current (A)

0.008
50.0

0.006
0.0
-50.0

250.0 300.0 350.0 400.0 450.0 500.0

Time (Seconds)

Figure 10. N2 crossover through membrane for fixed purge interval routine during load changes
from 120 A to 60 A and vice versa

D.17
 Paper III

0.08 0.16 0.24 0.32 0.4 0.48 0.56

0.85

0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95

0.12
0.1
0.4 0.48 0.56 0.64 0.72 0.8

Anode inlet H2 (kmol/kmol)

Anode inlet N (kmol/kmol)
Purged H2 (kmol/kmol)

Purged N2 (kmol/kmol)
H2 ventout (kmol/hr)

0.1
0.05

Cell voltage (V)

0.8

0.08
0.75
0.0

0.04 0.06
-0.05

0.7
0.32

0.65

0.02
-0.1

250.0 300.0 350.0 400.0 450.0 500.0 250.0 300.0 350.0 400.0 450.0 500.0
Time (Seconds) Time (Seconds)

Figure 11. Effect of load changes of 120–60–120 A on fixed purge interval, (a) H2 purge rate (b)
N2 concentration at anode inlet and cell voltage

Fig. 10 portrays nitrogen crossover during the load changes for a fixed purge interval of
60 s. As the current is decreased, nitrogen flux towards the anode is reduced. N2
concentration at anode exit also decreases as shown in Fig. 11a. However, N2 levels at
anode inlet tend to increase (Fig. 11b). As explained above, this is attributed to the low
residual flows of H2 at low current densities. On the other hand, when the current is
elevated to 120A at 410 s, sudden shortage of H2 is observed until the time taken by
flow valves to adjust. Reduced stoichiometry of hydrogen results in N2 concentration
spike as well. Voltage change as a function of current density is evident in the figure,
though voltage revival during purges is not that pronounced.
120.0

0.008
N2 crossover (kg/hr)
0.007
80.0
Current (A)

0.006
40.0

0.005
0.0
-40.0

250.0 300.0 350.0 400.0 450.0 500.0

Time (Seconds)

Figure 12. N2 crossover through membrane for automatic purge interval sequence during load
changes from 120 A to 60 A and vice versa

D.18
 Paper III

0.09
0.0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

0.08

0.65 0.7 0.75 0.8 0.85 0.9 0.95

0.8

0.9

Anode inlet N2 (kmol/kmol)

Anode inlet H2 (kmol/kmol)

Purged H2 (kmol/kmol)

Purged N2 (kmol/kmol)

H2 ventout (kmol/hr)

0.85
0.75

Cell voltage (V)

0.04

0.06
0.8
0.7

0.75
-0.01

0.04
0.65

0.7
-0.06

0.65

0.02
0.6

250.0 300.0 350.0 400.0 450.0 500.0 250.0 300.0 350.0 400.0 450.0 500.0
Time (Seconds) Time (Seconds)

Figure 13. Effect of load changes of 120–60–120 A on automatic purge interval, (a) H2 purge
rate (b) N2 concentration at anode inlet and cell voltage

Effects of load changes to the nitrogen crossover and system performance during
automatic purge are reported in Fig. 12. Nitrogen crossover shows similar trend when
compared to fixed time interval. When the current is lowered to 60 A, low residual H2
increases nitrogen levels at anode exit. As the molar N2 concentration reaches 3%,
automatic purge control activates the purge process. Here again, only purge intervals are
varied for automatic purging. It can be seen in Fig. 13a that several instances of purge
occur between 380410 s. Purge rate of H2 varies with the unused fuel at the anode exit.
Unlike fixed time interval, N2 concentrations at anode inlet are retained to lower levels
for automatic purge controls (Fig. 13b). Due to these frequent purges however the
amount of wasted fuel will increase to that of fixed purges. Fuel wastage and therefore
efficiency of the system operating with automatic purges will depend on the transient
load changes.
For steady state case, the system with the anode outlet bleed has the highest amount of
fuel wastage at 2% of the supplied fuel. Whereas, the fixed purge and automatic purge
routines account for 1.2% and 0.8% of fuel loss respectively. However, during transient
load changes, the automatic technique loses more fuel due to frequent purging
compared to steady state operations. For the transient case discussed here, the fuel
wastage becomes equal for automatic and fixed purging. However, the nitrogen
concentration levels remain within the acceptable limits for the automatic purge only.
This helps in prevention of carbon corrosion by H2 starved areas formed due to N2
accumulation within the cells, which in turn results in prolonged stack lifetime. Based
on the results discussed above, the applicability of nitrogen detectors and automatic
purge shows promise as a diagnostic tool for prediction of gas crossovers and act as a
building block for devising purge strategies depending upon the load and mode of
application.

D.19
Paper III

Conclusions
A comprehensive study on dynamics of nitrogen crossovers in PEMFC system with
recirculation was conducted and its effects on purging strategy were discussed. The
model developed to predict nitrogen crossover was in good agreement with the design
validation data of the stack. The results exhibit that with pure recirculation, voltage and
system efficiency declines due to nitrogen accumulation in fuel cell. Different purging
methodologies were simulated to address hydrogen dilution issue at reaction sites.
Anode bleed out of 3% is found to be the limit for prevention of N2 buildup and retains
the concentration levels to less than 1%. Also, cell voltage degrades linearly with N2
buildup and rejuvenates at purge sequences. An alternate strategy for automatic
initiation of anode recirculation purge by employing nitrogen detectors was simulated.
Using the same purge time for various cases, it is shown that purge interval is lowered
for low currents mainly due to low H2 residual flows rates. Moreover, during transient
load changes, automatic purge catered well to prevent nitrogen levels from rising when
compared to a fixed purge interval strategy. This method has proven to be feasible for
predicting gas crossovers during purge sequences at different loads and can be used as a
base for optimizing and development of anode purge strategies for PEMFC systems.

Acknowledgements
The authors would like to thank the Danish Energy Agency for financial support and
our industrial partner, H2Logic, for their collaboration and technical support.

References

1. Hosseinzadeh E, Rokni M: Development and validation of a simple analytical model

of the Proton Exchange Membrane Fuel Cell (PEMFC) in a fork–lift truck power
system. International journal of green energy 2013; 10(5):523–43.

2. Kocha SS, Deliang Yang J, Yi JS: Characterization of gas crossover and its
implications in PEM fuel cells. AIChE Journal 2006, 52:1916–1925.

3. Weber: Gas–crossover and membrane–pinhole effects in polymer–electrolyte fuel

cells. Journal of the Electrochemical Society 2008, 155:B521–31.

4. Reiser CA, Bregoli L, Patterson TW, Yi JS, Yang JD, Perry ML, Jarvi TD: A reverse–
current decay mechanism for fuel cells. Electrochemical and Solid–State Letters 2005,
8:A273–A276.

5. Meyers JP, Darling RM: Model of carbon corrosion in PEM fuel cells. Journal of the
Electrochemical Society 2006, 153:A1432–A1442.

D.20
 Paper III

6. Bessarabov D: Gas Permeability of Proton–Exchange–Membranes. In PEM Fuel Cell

Diagnostic Tools. Edited by Haijiang Wang, Xian–Zi Yuan and HL. Boca Raton, Fla.:
Taylor & Francis; 2011:558.

7. Ahluwalia RK, Wang X: Buildup of nitrogen in direct hydrogen polymer–electrolyte

fuel cell stacks. Journal of Power Sources 2007, 171:63–71.

8. Catalano J, Myezwa T, De Angelis MG, Baschetti MG, Sarti GC: The effect of relative
humidity on the gas permeability and swelling in PFSI membranes. International
Journal of Hydrogen Energy 2012, 37:6308–6316.

9. Baik KD, Kim MS: Characterization of nitrogen gas crossover through the membrane
in proton–exchange membrane fuel cells. International Journal of Hydrogen Energy
2011, 36:732–739.

10. Ismail MS, Hughes KJ, Ingham DB, Ma L, Pourkashanian M: Effects of anisotropic
permeability and electrical conductivity of gas diffusion layers on the performance
of proton exchange membrane fuel cells. Applied Energy 2012, 95:50–63.

11. Zhu WH, Payne RU, Tatarchuk BJ: Critical flow rate of anode fuel exhaust in a
PEM fuel cell system. Journal of Power Sources 2006, 156:512–519.

12. Tang Y, Yuan W, Pan M, Li Z, Chen G, Li Y: Experimental investigation of dynamic

performance and transient responses of a kW–class PEM fuel cell stack under
various load changes. Applied Energy 2010, 87:1410–1417.

13. Zhai S, Zhou S, Sun P, Chen F, Niu J: Modeling study of anode water flooding and
gas purge for PEMFCs. Journal of Fuel Cell Science and Technology 2012, 9.

14. Gou J, Pei P, Wang Y: The dynamic behavior of pressure during purge process in
the anode of a PEM fuel cell. Journal of Power Sources 2007, 162.

15. Wang X, Tajiri K, Ahluwalia RK: Water transport during startup and shutdown of
polymer electrolyte fuel cell stacks. Journal of Power Sources 2010, 195:6680–6687.

16. Karimäki H, Pérez LC, Nikiforow K, Keränen TM, Viitakangas J, Ihonen J: The use
of on–line hydrogen sensor for studying inert gas effects and nitrogen crossover in
PEMFC system. International Journal of Hydrogen Energy 2011, 36:10179–10187.

17. Promislow K, St–Pierre J, Wetton B: A simple, analytic model of polymer electrolyte

membrane fuel cell anode recirculation at operating power including nitrogen
crossover. Journal of Power Sources 2011, 196:10050–10056.

D.21
Paper III

18. Hou Y, Shen C, Yang Z, He Y: A dynamic voltage model of a fuel cell stack
considering the effects of hydrogen purge operation. Renewable Energy 2012,
44:246–251.

19. Müller EA, Kolb F, Guzzella L, Stefanopoulou AG, McKay DA: Correlating
nitrogen accumulation with temporal fuel cell performance. Journal of Fuel Cell
Science and Technology 2010, 7:210131–2101311.

20. Siegel JB, Bohac S V, Stefanopoulou AG, Yesilyurt S: Nitrogen Front Evolution in
Purged Polymer Electrolyte Membrane Fuel Cell with Dead–Ended Anode. Journal
of the Electrochemical Society 2010, 157:B1081–93.

21. Yesilyurt S, Siegel JB, Stefanopoulou AG: Modeling and Experiments of Voltage
Transients of Polymer Electrolyte Membrane Fuel Cells With the Dead–Ended
Anode. Journal of Fuel Cell Science and Technology 2012, 9.

22. Choi JW, Hwang Y–S, Seo J–H, Lee DH, Cha SW, Kim MS: An experimental study
on the purge characteristics of the cathodic dead–end mode PEMFC for the
submarine or aerospace applications and performance improvement with the
pulsation effects. International Journal of Hydrogen Energy 2010, 35:3698–3711.

23. Ballard Mark9 SSL: Product manual and integration guide. 2008.

24. Rabbani A, Rokni M: Dynamic characteristics of an automotive fuel cell system for
transitory load changes. Sustainable Energy Technologies and Assessments 2013, 1.

25. Hwang J–J: Thermal control and performance assessment of a proton exchanger
membrane fuel cell generator. Applied Energy 2013, 108:184–193.

26. Wang Y, Chen KS, Mishler J, Cho SC, Adroher XC: A review of polymer electrolyte
membrane fuel cells: Technology, applications, and needs on fundamental research.
Applied Energy 2011, 88:981–1007.

27. Cheng X, Zhang J, Tang Y, Song C, Shen J, Song D, Zhang J: Hydrogen crossover in
high–temperature PEM fuel cells. Journal of Power Sources 2007, 167:25–31.

28. Mittelsteadt C, Umbrell M: Gas Permeability in Perfluorinated Sulfonic Acid

Polymer Electrolyte Membranes. MEETING ABSTRACTS– ELECTROCHEMICAL
SOCIETY –ALL DIVISIONS CD ROM EDITION– 2005.

29. Barelli L, Bidini G, Gallorini F, Ottaviano A: Dynamic analysis of PEMFC–based

CHP systems for domestic application. Applied Energy 2012, 91:13–28.

D.22
 Paper III

30. Hosseinzadeh E, Rokni M, Rabbani A, Mortensen HH: Thermal and water

management of low temperature Proton Exchange Membrane Fuel Cell in fork–lift
truck power system. Applied Energy 2013, 104:434–444.

31. Khan MJ, Iqbal MT: Modelling and analysis of electrochemical, thermal, and
recetant flow dynamics for a PEM fuel cell system. Fuel Cells 2005, 5:463–475.

32. Rabbani A, Rokni M: Start–up Analysis of a PEM Fuel Cell System in Vehicles.
International Journal of Green Energy 2013, 11:01–21.

Nomenclature

E theoretical voltage V 
Vcell average cell voltage V 
Pel stack power kW 
P in energy into the fuel cell kW 
Pout energy out of the fuel cell kW 
Q loss heat dissipated kW 
Ct stack thermal capacitance kW 
R universal gas constant J molK 
T temperature K 
F faraday’s constant C molK 
I current  A
N cell number of cells  
 g 0f change in Gibbs free energy J molK 
PH 2 hydrogen partial pressure  
PO2 oxygen partial pressure  
Mm mol. weight of membrane Kg mol 
J H 2O net water–diffusion flux mol scm 2 
D water diffusion coefficient cm 2 s 
E N2 activation energy kJ mol 
K N2 nitrogen permeance k mol s  cm  bar 
aH 2 hydrogen activity  
a H 2O water activity  
a O2 oxygen activity  
i current density A cm 2 
in internal current density A cm 2 

D.23
Paper III

i0 exchange current density A cm 2 

i0 ,a anode exchange current density A cm 2 
i0 ,c cathode exchange current density A cm 2 
ka anode reaction rate mol scm 2 
kc cathode reaction rate mol scm 2 
ne electrons transferred mol e mol fuel 
nel number of electrons  
n drag electro osmotic drag  
tm membrane thickness cm

Greek symbols
a anode transfer coefficient  
c cathode transfer coefficient  
N 2
perm scale factor  
 symmetry factor  
act activation overpotential V 
 act ,a anode activation overpotential V 
 act ,c cathode activation overpotential V 
 conc concentration overpotential V 
 ohmic ohmic overpotential V 
 membrane water content  
 dry membrane density g cm3 

D.24
 Appendix E

Paper IV

249
Appendix E

Rabbani, A & Rokni, M. “Dynamic Simulation of a Proton Exchange Membrane

Fuel Cell System for Automotive Applications”, Proceedings of Sustainable Energy
& Environmental Protection (SEEP) Conference, Dublin, 2012; 311–316.

250
Proceedings of SEEP2012, 05–08 June 2012, DCU, Dublin, Ireland Paper IV

Dynamic simulation of a proton exchange membrane fuel

cell system for automotive applications

R. A. Rabbani1 and M. Rokni2

1. Technical University of Denmark, Kgs. Lyngby, Denmark;
email: [email protected]
2. Technical University of Denmark, Kgs. Lyngby, Denmark;
email: [email protected]

ABSTRACT

A dynamic model of the PEMFC system is developed to investigate the behaviour and
transient response of the fuel cell system for automotive applications. The system
accounts for the fuel cell stack with coolant, humidifier, heat exchangers and pumps.
Governing equations for fuel cell and humidifier are implemented into the code and are
based on adopted mathematical models describing the voltages and current densities
and their dependence on operating pressures, temperatures and stoichiometric ratios of
the reactant gases. As a result, this model can predict both steady and transient states.
The model parameters have been adjusted specifically for a 21.2 kW Ballard stack [1].
This model also incorporates the effects of water cross–over in the fuel cell membrane.
Controls for temperatures, pressures, reactant stoichiometry and flows are implemented
to simulate the system behaviour for different loads and operating conditions.
Simulation results for system start–up and variable loads are discussed. Results for
system efficiency, auxiliary power consumption, feed flow effects and water crossover are
presented. Transitory effects of liquid water saturation at cathode are also determined.
This study can provide sufficient insight for further in–depth analysis of PEMFC and
prove to be a basis for efficient control and design methodologies.
Keywords: Dynamic simulation, PEM fuel cell, water crossover, Control system

1 INTRODUCTION

Fuel cells are recognized to be one of the future power supply systems. The proton
exchange membrane fuel cells (PEMFC) currently appear to be the preferred fuel cell for
a variety of mobile applications, mainly due to its relatively low operating temperature,

E.1
Paper IV Proceedings of SEEP2012, 05–08 June 2012, DCU, Dublin, Ireland

quick start–up, high power density and efficiency, system robustness and low
degradation due to corrosion.
Fuel cell operating requirements in vehicles are more inflexible than stationary
applications. These systems have to operate at varying conditions related to
temperatures, pressures, power load and humidity. All the auxiliary components
constitute the balance of plant (BoP). These auxiliary components, such as the air and
fuel supply system which include compressors and control valves, and the thermal
control system which comprise heat exchangers, coolant pumps and air radiators are
essential for the successful operation of the fuel cell system. Therefore, system level
dynamic modelling will be a useful tool in analysis of PEMFC systems.

There are quite a few PEM fuel cell models available in the literature. Fuel cell and stack
level transient modelling is performed in the dynamic model developed by [2]. A
simplistic dynamic model based on cathode kinetics was developed in [3]. A dynamic
model in MATLAB/Simulink was developed in [4] to investigate fuel cell transient
electrical responses under various operating conditions. Issues related to temperature
dynamics are dealt and studied by [5], which could predict the effects of temperature
and feed flows on system transient behaviour. [6] proposed a transient model to
predict efficiency in terms of voltage output, and a thermal model including heat
transfer coefficients and energy balance for the stack. A thermal management system
for a PEMFC was designed in [7] which was oriented towards the flow fields within the
stack. Start–up behaviour of PEMFC stacks at sub–zero temperatures have been studied
and reported by few researchers. [8] conducted experiments and validated their model
based on the results while [9] investigated effects of sub–freezing temperatures on fuel
cell performance and start–up.

Compared to all these studies, the model presented in this work aims at analysis and
investigation of a complete PEMFC system and study its transient response to operating
parameters such as temperature, humidity, pressure and reactant mass flows. One of the
objectives of this work is to devise a control–oriented dynamic model of the fuel cell
stack, accommodating the electrochemical, thermal and feed flow models. Therefore,
Aspen Plus DynamicsTM is used to develop a dynamic stack model with liquid coolant
circuit incorporated in it, which allows for a detailed analysis of the thermal interaction
with the surroundings. Moreover, the present model can provide suitable insight to
saturated water issues in the fuel cell stack. Two–phase characteristics of concerning
material streams are also determined.

2 SYSTEM CONFIGURATION

Figure 1 shows the schematics of fuel cell system analyzed in this study. It includes all
the components contained in the system, such as PEMFC stack, air compressor,
humidifier, pumps, heat exchangers and radiator for the cooling circuit, flow valves and

E.2
Proceedinngs of SEEP22012, 05–08
8 June 2012,, DCU, Dublin, Ireland Paper
P IV

controlllers. Comppressed air is cooled aand humid dified beforre enteringg to the catthode of
the stacck. On the other side, pressurizeed hydrogeen from storage tank iis fed to thhe anode
of fuell cell. Sinnce a high her fuel sstoichiomettry is maiintained, aanode exh haust is
recircullated back to the feed
d stream viaa a recirculaation pumpp.

Figu
ure 1. Schem
matic of a complete
c PPEM fuel ceell system with
w auxiliaary compon
nents

Heat p produced in n the stack k is absorbbed by thee coolant which

w circuulates in a circuit
associatted with thhe stack and a heat exxchanger. An A externaal cooling lloop, conneected to
the aforrementioned heat excchanger, in n turn coolls the waterr in the intternal circu
uit. This
circuit also comp prises of a heat exchaanger to prrecool air into
i the fuuel cell andd an air
radiatorr for heat rejection.
r Flow
F of watter is regulaated by pum
mps in the rrespective circuits.
c

In this study, a Baallard fuel cell

c stack [11] has been n specificallyy adapted. The model, which
is basedd on equattions adop pted by [100], howeverr, contains some paraameters wh hich are
attributted to the physical
p characteristiccs of the sysstem, as weell as on opperating con
nditions
and meembrane prroperties. SomeS of th
hese operatiing conditiions for thee Ballard stack
s are
shown in Table 1. Curren nt researchh aims to build up a system which meeets the
requireements of actual stacck runningg under recommendeed conditioons. The fuel f cell
2
stack coontains 1100 cells withh the cell aarea equal of 285 cm . Operatiing temperature of
the staack is maintained arounda 600–70°C. Maximum
M power prooduced fro om the
described stack is 21.2 kW correspondding to a current of 300A; 3 howwever it is generally
g
operateed at lower current raanges to redduce ohmicc and conccentration ooverpotenttials and
thus higgher efficieencies.

E.3
Paper IV Proceedings of SEEP2012, 05–08 June 2012, DCU, Dublin, Ireland

Table 1. Operating conditions for fuel cell stack recommended by manufacturer

(Ballard).
Stack Fuel inlet Air inlet Inlet Operating Fuel Air
power pressure pressure temperature temperature stoichiometry stoichiom
(kW) (bar) (bar) (°C) (°C) etry
1.4 1.15 1.08 60 61 6.3 5.1
2.7 1.16 1.10 60 63 3.4 2.4
5.1 1.31 1.17 60 66 2.2 1.8
9.7 1.55 1.38 60 67 1.7 1.8
18.0 2.00 1.80 60 68 1.6 1.8
21.2 2.20 2.00 60 70 1.6 1.8

Due to the lack of sufficient experimental data, it is assumed that the inlet air flow is
optimally humidified after passing through the humidifier. The relative humidity of the
air entering the cathode is set to 95% in the calculations; although other values can be
chosen. This assumption could be justified as it is very close to the real operational
conditions. On the anode side, there is no humidifier and the fuel can reach the desired
humidity by means of recirculation and water production at anode exit.

Figure 1 only represents controllable connections initiating from the controller. In

calculation as well as reality, the control unit collects information from various
temperature, pressure and flow transmitters in addition to current and voltage data from
the fuel cell. Here a simplified view of the system is presented.

3 METHODOLOGY

The characteristics of the PEMFC system described above are implemented in Aspen
Plus Dynamics TM which is a simulation tool for process modelling and energy system
analysis. The program contains a vast library of components and controls for standard
energy processes. The PEMFC stack model presented in this study is based on a model
developed by [10]. Concentration losses are neglected in the present study, which is
justified by the fact that the system does not run at such high current densities where
the concentration overpotentials becomes significant. Models for fuel cell and
humidifier are also implemented into the code and are based on adopted mathematical
models describing the voltages, current densities and their dependence on operating
pressures, temperatures and stoichiometric ratios of the reactant gases. This model
which incorporates governing equations for cell electrochemical, polarization
overpotentials, heat transfers and water diffusion across the membrane is implemented
into ASPEN Plus Dynamics and system controls are implemented in order to ensure
stable operation of the plant during load changes. The thermodynamic efficiency and

E.4
Proceedings of SEEP2012, 05–08 June 2012, DCU, Dublin, Ireland Paper IV

net power of the system are determined by current drawn and voltage produced by the
stack. The total energy into the fuel cell is consumed by the electrical power output, heat
removed by the coolant, heat loss at the stack surface and energy stored by the stack
itself. In the current model, a lumped thermal model proposed by [6] is considered. The
subordinate components in the BoP, i.e. anode recirculation and water pumps, air
compressor, mixers and heat exchangers are modelled using the default mathematical
models provided in Aspen Plus Dynamics.

4 CONTROL SYSTEM

This section presents methodology of controlling system parameters and operating

conditions for the system to have a stable operation. Classic proportional–integral (PI)
controllers, which are widely used in industrial control systems, are employed to regulate
different components and flow streams. Key parameters to be controlled in the
proposed system are reactant inlet stoichiometries, inlet pressures, coolant inlet and
operating temperatures of the stack.

As shown in Table 1, the recommended fuel and air stoichiometries for the selected fuel
cell stack and their inlet pressures, temperature difference between stack inlet and outlet
need to be controlled. It can be seen that at low current loads, high amounts of excess
reactant flows are desired. This is due to the fact that at low power consumption and
low pressures, water formed due to the reaction in the cathode side of the cells needs to
be ejected out of the stack, which is done by supplying high amounts of air. Therefore,
an algorithm based on the above figure is developed to be the process variable for the PI
controller, which regulates the compressor power in order to maintain the desired
oxygen ratio. Similarly, an algorithm for controlling hydrogen flow is devised along with
a PI controller, which regulates the control valve opening for optimal fuel supply.

Thermal management in PEMFC systems is of vital importance, basically due to the fact
that heat produced in the selected fuel cell cannot be dissipated by convection and
radiation through the stack surface. A consistent and stable operation of around 70°C
thus requires a cooling system, preferably with a liquid coolant. Since the operating
temperature of the fuel cell is not very high, a low temperature difference with the
ambient requires having a large heat transfer surface. Therefore, an efficient thermal
control system becomes of substantial importance to ensure optimum system
performance. In this case, the temperature in the stack can be controlled by the coolant
flow rate which acts as an input signal and is controlled by the PI controller. Based on
Ballard recommendations, an algorithm for start–up, shut–down and normal operations
is thus developed which sets the process variable for the controller. Coolant flow is
manipulated by sending output signals to the driving pump. Temperature of the coolant
entering the stack can similarly be controlled by the flow of water in the external circuit.
Control signal to the associated pump regulates the electrical power of the pump and

E.5
Paper IV Proceedings of SEEP2012, 05–08 June 2012, DCU, Dublin, Ireland

hence the coolant inlet temperature into the stack. In a similar fashion, the temperature
of water in the external circuit is dependent on radiator fan speed. PI controllers are
used to regulate the fan speed as well.

5 RESULTS AND DISCUSSION

Reliability of the suggested model has been verified and validated against experimental
data by [10], where the characteristics of the model were studied at various operating
temperatures and power loads. As suggested by the manufacturer, stack temperature
range of 60–70°C has been used in the simulations of the current system. In case of
PEMFC, dynamic behaviour of a start–up is of particular importance to ensure a short
start–up time and an efficient operation. In current simulations, the initial temperature
of the stack is assumed to be 25°C. Other parameters and operating conditions are
selected from section 2.

Here, results for a start–up case are summarised, when a current corresponding to a
specific power load is drawn from the stack. As can be seen in the Fig. 2a, when the
stack is started at current of 60 A, it takes approximately 300 seconds to reach stable
operating temperature. At start, the flow in cooling circuits is set to the lowest value
since it is desired to raise the stack temperature to its optimal operation. Coolant flow
in the internal circuit is fixed to 290 kg/hr as recommended by the manufacturer. Flow
in the internal cooling circuit increases, once the stack temperature difference increases
by 15°C. Flow in external circuit is regulated to maintain stack inlet temperature of
internal cooling circuit. Air flow in radiator starts to maintain temperature of external
cooling loop to around 50°C. It can be observed that the voltage reduces abruptly when
simulation time is around 155 seconds. At this stage, the decrease in cell temperature
and voltage is caused by the sudden increase in coolant flow into the stack, which aims
to maintain the stack operating temperature.

(a) Start-up thermal management

100.0
450.0 600.0
External coolant flow (kg/hr)
Internal coolant flow (kg/hr)

Air Radiator flow (kg/hr)

Stack Temperature (C)

75.0
300.0

50.0
150.0

25.0
0.0

0.0 150.0 300.0 450.0 600.0 750.0

Time Seconds

E.6
Proceedings of SEEP2012, 05–08 June 2012, DCU, Dublin, Ireland Paper IV

(b) Start-up efficiency and powers

0.51 0.54 0.57 0.6 0.63 0.66

6.0

0.6
Auxiliary Power kW
Stack Power (kW)

System Efficiency
0.5
4.0

0.4
2.0

0.3
0.0

0.2
0.0 150.0 300.0 450.0 600.0 750.0
Time Seconds

Figure 2. System start–up (a) Stack temperature and corresponding coolant flows at
current of 60A, (b) System efficiency, stack power output and power consumed by
auxiliary components

Overall efficiency of the system is as much affected by compressor and fan, as by the fuel
cell stack itself. Is can be observed from Fig. 2b, that power produced by the stack
increases with operating temperature untill it reaches steady–state conditions. Also,
power consumed by the air compressor becomes constant after a few seconds into the
start. Although, efficiency of the system follows a similar trend, it decreases once the
radiator fan is turned on at around 370 seconds. Together, air compressor and radiator
consume 10% to that of stack power, whereas a coolant and recirculation pumps
account for 1.0–1.3% of it. This is also noticed in figures above; increase in internal and
external coolant flows at time 200 and 370 seconds respectively, elevates the auxiliary
power consumption to a very small extent as compared to air compressor and radiator.
-80.0 -40.0 0.0 40.0 80.0 120.0

(b) Reactant flows at current ramp

0.0 20.0 40.0 60.0
Cathode inlet flow (kg/hr)

0.8 1.2 1.6 2.0 2.4 2.8

Anode inlet flow (kg/hr)
Current (A)

-20.0

750.0 800.0 850.0 900.0 950.0 1000.0

Time (Seconds)

E.7
Paper IV Proceedings of SEEP2012, 05–08 June 2012, DCU, Dublin, Ireland

In addition to the start sequence of the aforementioned model, transitory effects under
variable load are investigated as well. An instance of load change, when current is
ramped from 60 A to 100 A at a rate of 20 Amperes per second and vice versa, is
presented in Fig. 3a and its effect on cell voltage is examined.

(a) Cell Voltage at current ramp

65.0 70.0 75.0 80.0 85.0

0.84

-0.3-0.2-0.1 0.0 0.1 0.2 0.3 0.4

Operating Temperature (C)

Current Density (A/cm2)
Cell Voltage (V)
0.72 0.76 0.8
0.68

750.0 800.0 850.0 900.0 950.0 1000.0

Time (Seconds)

Figure 3. Current ramp–up from 60–100A and vice versa: (a) Changes in cell voltages
and operating temperature, (b) Variations in fuel and air inlet flows

Decline in voltage potential at increase of current density is a well–established fact.

However, it is also noticed that cell voltage is also a function of operating temperature
and pressure. Since, pressures of the reactants are already regulated to be at the
optimum; here we will discuss temperature dependency of developed potential
differences in the cell. Voltage decreases steadily with the decrease in operating
temperature. This fact is due to the reason being fast reaction kinetics at the electrodes
of individual cell sites. The opposite could also be observed when the current is reduced
back to 60A.

Fig. 3b depicts the variations in reactant inlet flows when the load on fuel cell is varied.
As expected, the air flow into the cathode, regulated by a controller, increases when the
current is ramped up and steadies along with the system. It is however interesting to
notice the fluctuations in the anode inlet. Since anode inlet is supported by
recirculation from anode exhaust, the changes within the stack influence it considerably.
Although, anode flows also tend to increase as the current is increased, they are affected
by the water content in the anode outlet (70% mass fraction), which depends on the net
water crossover within the cell. Initial peaks at both load changes are attributed to the
electro–osmotic drag which is a function of current density.

E.8
Proceedings of SEEP2012, 05–08 June 2012, DCU, Dublin, Ireland Paper IV

-80.0 -40.0 0.0 40.0 80.0 120.0

(b) Water saturation at cathode

0.1 0.15 0.2 0.25 0.3 0.35

-2.0 0.0 2.0 4.0 6.0 8.010.012.0
Heat generated (kW)

Liquid fraction
Current (A)

750.0 800.0 850.0 900.0 950.0 1000.0

Time (Seconds)
-50.0 0.0 50.0 100.0150.0200.0

(a) Water crossover in fuel cell

0.72 0.78 0.84 0.9 0.96 1.02

0.0 0.1 0.2 0.3 0.4

Net Water crossover (kg/hr)
Anode outlet RH (%/%)
Anode inlet RH (%/%)
Current (A)

750.0 800.0 850.0 900.0 950.0 1000.0

Time Seconds

Figure 4. Current ramp–up from 60–100A and vice versa: (a) Amount of liquid water at
cathode outlet, (b) Variation in net water crossover and anode RH.

Water management is a critical issue since the performance of PEM fuel cell is strongly
influenced by its internal water distribution. As discussed above, water crossover within
the cell influences anode inlet stream. Fig. 4a shows net water diffusion in the cell when
the current is changed from 60A to 100A and back. At high currents, more water is
produced in the cathode which supports back–diffusion towards the anode until the
system reaches back to steady–state and there is almost no net water crossover. The
peaks observed at the start of current change are due to electro–osmotic drag as
mentioned above. Effect of water distribution can also be observed in anode inlet and
outlet relative humidities. Whereas for the cathode, since it is assumed that air enters at
a constant relative humidity of 95%, the outlet humidity is always above 100%.
Therefore, results for only anode are discussed here.

Removal of water from the cathode is dependent on stack temperature and pressure
drop. Temperature is the more critical factor, since at high temperature the water will be
in the vapour state and easier to remove. Inlet humidities also contribute to liquid water
saturation in the fuel cell, which is reported in [11]. It can be seen in Fig. 4b, that the
amount of liquid water at cathode decreases at high currents. Heat produced by cell

E.9
Paper IV Proceedings of SEEP2012, 05–08 June 2012, DCU, Dublin, Ireland

reactions elevates the stack temperature, thereby reducing saturated water at the outlet.
On the other hand, reduction in stack current and temperature increases the liquid
water. Therefore, at low temperatures and currents, water removal is the dominant
factor and stoichiometries are determined by the minimum flow rates required for water
removal which in the present case are more than adequate to provide the necessary
concentrations. Moreover, condensation of water at cell sites produces additional heat
which requires higher coolant flows to maintain the stack operating temperature,
though it does not affect the system efficiency to a greater extent, since liquid pumps do
not consume that much power.

6 CONCLUSIONS

This work has presented a comprehensive dynamic model of a fuel cell system along
with the BoP. The model is oriented towards the control of associated operating
conditions for the fuel cell module. Main contributions of the proposed model are
attributed to the system response methodology, which incorporates stack thermal
behaviour in addition to fuel cell electrochemistry and flow mass and energy balance.
Emulation of the presented system shows that the results are in good agreement with
the manufacturer’s data. It is observed that voltage system efficiency increase with
operating temperature. Air radiator does not affect system start–up, though consumes
10% of stack power together with the air compressor. Furthermore, water crossover in
the stack affects the flows and humidity in anode recirculation. Also, removal of liquid
water from cathode is dependent on operating temperature. Further analysis is required
to study water diffusion in the cell, however based on these results; this model can be
used for optimizing and designing operational strategies for PEMFC systems for
automotive applications.

ACKNOWLEDGEMENT

The authors would like to thank the Danish National Advanced Technology
Foundation for financial support and our industrial partners, H2Logic, for their
collaboration and technical support.

REFERENCES

[1] Ballard Mark9 SSL, “Product manual and integration guide.” 2008.
[2] S.–K. Park and S.–Y. Choe, “Dynamic modeling and analysis of a 20–cell PEM
fuel cell stack considering temperature and two–phase effects,” Journal of Power
Sources, vol. 179, no. 2, pp. 660–672, 2008.
[3] M. Ceraolo, C. Miulli, and A. Pozio, “Modelling static and dynamic behaviour of
proton exchange membrane fuel cells on the basis of electro–chemical
description,” Journal of Power Sources, vol. 113, no. 1, pp. 131–144, 2003.

E.10
Proceedings of SEEP2012, 05–08 June 2012, DCU, Dublin, Ireland Paper IV

[4] J. Jia, Y. Wang, Q. Li, Y. T. Cham, and M. Han, “Modeling and Dynamic
Characteristic Simulation of a Proton Exchange Membrane Fuel Cell,” IEEE
Transactions on Energy Conversion, vol. 24, no. 1, pp. 283–291, 2009.
[5] G. Vasu and A. K. Tangirala, “Control–orientated thermal model for proton–
exchange membrane fuel cell systems,” Journal of Power Sources, vol. 183, no. 1, pp.
98–108, 2008.
[6] M. J. Khan and M. T. Iqbal, “Modelling and analysis of electrochemical, thermal,
and recetant flow dynamics for a PEM fuel cell system,” Fuel Cells, vol. 5, no. 4,
pp. 463–475, 2005.
[7] S. Asghari, H. Akhgar, and B. F. Imani, “Design of thermal management
subsystem for a 5kW polymer electrolyte membrane fuel cell system,” Journal of
Power Sources, vol. 196, no. 6, pp. 3141–3148, 2011.
[8] Y. Li, S. Xu, Z. Yang, and Y. Li, “Experiment and simulation study on cold start
of automotive PEMFC,” pp. 2166–2170, 2011.
[9] Q. Yan, H. Toghiani, Y.–W. Lee, K. Liang, and H. Causey, “Effect of sub–freezing
temperatures on a PEM fuel cell performance, startup and fuel cell components,”
Journal of Power Sources, vol. 160, no. 2, pp. 1242–1250, 2006.
[10] E. Hosseinzadeh and M. Rokni, “Development and validation of a simple
analytical model of the Proton Exchange Membrane Fuel Cell (PEMFC) in a
fork–lift truck power system,” International journal of green energy, vol. To be publ,
2012.
[11] K. H. Wong, K. H. Loo, Y. M. Lai, S.–C. Tan, and C. K. Tse, “A theoretical study
of inlet relative humidity control in PEM fuel cell,” International Journal of
Hydrogen Energy, vol. 36, no. 18, pp. 11871–11885, 2011.

E.11
262