Applied Thermal Engineering 112 (2017) 1382–1395

Contents lists available at ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

Modeling and analysis of single particle conversion of biomass in a

packed bed gasification system
Sandeep Kumar, S. Dasappa ⇑
Centre for Sustainable Technologies, Indian Institute of Science, Bangalore 560012, India

h i g h l i g h t s

Single particle biomass conversion analysis and modeling for packed bed gasification.

Pyrolysis and char reaction modeled.

Influence of thermo-physical properties analyzed.

Devolatilisation studied under inert and flaming conditions.

Modeling results validated with results from literature.

a r t i c l e i n f o a b s t r a c t

Article history: The present study focuses on developing a comprehensive one-dimensional spherically symmetrical
Received 28 August 2016 model for single particle biomass conversion - a precursor for a packed biomass gasification model.
Revised 20 October 2016 Pyrolysis, volatile combustion and char reactions with a range of reacting species and inert environment
Accepted 27 October 2016
are considered in the analysis. Flaming and inert pyrolysis are independently validated with experimen-
Available online 29 October 2016
tal results available in the literature. The influence of thermo-physical properties, like density, thermal
conductivity and particle size which affect the Biot number, heat flux, and surface to volume ratio of
Keywords:
the particle are studied. Superficial activation energy is proposed to normalize the conversion time with
Pyrolysis
Biomass gasification
data from different sets of experiments to address the effect of temperature by using a corrected time.
Single particle conversion Particle with Biot number close to unity suggests a departure from d2 law whereas larger size particle
Modeling shows close dependence on d2 law indicating a typical diffusion dominated process. The model is com-
Packed bed gasification pared with a broad range of experimental results available in the literature. The prediction of conversion
time, temperature profiles, weight loss rate and model results with normalization to physical parameters,
suggests that the predictions with such variability are good enough to be used as a robust model for the
packed bed gasification or combustion.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction understand the fundamental science behind pyrolysis process to

model any biomass combustion or gasification process.
Ever increasing energy demand and carbon emission have Many kinetic models have been reported in the literature which
become a primary concern, and meeting this escalating energy ranges from simplest models based on single step decomposition
need has resulted in renewed interest in green alternatives to fossil reaction to multi step and parallel reaction model [2–8]. Boateng
fuels [1]. Biomass is a natural substance which mainly contains cel- and Mtui [8] and Tamer et al. [9]. have modeled multi-step pyrol-
lulose, hemi-cellulose and lignin, with an average composition of ysis process in a fluidized bed gasifier based on Eulerian – Eulerian
C1H1.4O.6, with slight variations depending on the nature of the CFD model approach.
biomass. Combustion, pyrolysis and gasification are three thermo- The complex chemical mechanisms involved in pyrolysis are
chemical conversion routes for conversion of biomass to energy. not completely understood and the degradation pathway is a func-
Pyrolysis is the first stage in any thermal treatment of biomass, tion of heating rate, temperature, gaseous environment, pretreat-
be it combustion or gasification and it becomes important to ment, extent of inorganic impurities and catalysis. Based on the
heating rate pyrolysis is classified as slow and fast. As the name
⇑ Corresponding author. suggests, during slow pyrolysis, the biomass particles are subjected
E-mail address: [email protected] (S. Dasappa).

http://dx.doi.org/10.1016/j.applthermaleng.2016.10.170
1359-4311/Ó 2016 Elsevier Ltd. All rights reserved.
 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395 1383

Nomenclature

L characteristic dimension of particle, m Y i;s ith species concentration at gas film surrounded the
h heat transfer coefficient, W/(m2 K) particle surface
Bi Biot number KD mass transfer coefficient
A pre-exponential component of Arrhenius reaction rate e particle porosity
term CP specific heat, kJ/(kg K)
E activation energy, kJ/mol j thermal conductivity, W/(m K)
R universal gas constant, m3 atm K1 mol1 r particle radius, m
T temperature, K -_ 000i volumetric reaction rate of ith specie, kg/(m3 s)
t time, s q average particle density, kg/m3
m_ mass flow rate, kg/s H enthalpy of reaction, kJ/kg
YBio mass fraction of biomass in the particle during conver- qc density of wood char, kg/m3
sion process qb density of biomass, kg/m3
YChar mass fraction of char in the particle during conversion T1 ambient temperature, K
process TC core temperature, K
Yi mass fraction of ith species TS particle surface temperature, K
D diffusivity V volume of the biomass/char particle, m3
q density, kg/m3 Mg molecular mass of the mixture of gases, kg/kmol
m_ 00 mass flux, kg/(m2 s) Mi molecular mass of the ith specie, kg/kmol
n number of particles per unit volume rp pore radius of wood char, m
_p
m gasification rate of one particle, kg/s s tortuosity factor

to heating rates of the order about 10 K/s while in fast pyrolysis it close to the external temperature while a thermal gradient occurs
is in the range of 150–250 K/s [10]. inside the particle. A heat wave proceeds toward the center of the
Extensive studies on the slow pyrolysis of variety of wood sam- particle. Pyrolysis occurs quickly inside the thermal gradient. sreac-
ples have been conducted by Murty and co-workers [11–14] as a tion is the limiting conditions for small particles where the thermal
part of the fire research with focus on combustion mechanisms equilibrium between the particle and its surrounding is reached
and fire retardants. Murty has attempted to describe the physics much faster than pyrolysis completion. The devolatilisation rate
of pyrolysis and subsequently superimposed the chemical transfor- controls the whole pyrolysis. The devolatilisation in fixed bed gasi-
mations in the processes of heat and mass transfer [13]. The model fication process is often limited by internal and external heat
included with the solid particle subjected to an external heat flux transfer where the particles size is relatively larger. Physically
resulting in an inward conduction, outward diffusion of gaseous the particle has to be very small to increase the ratio of surface
products and the effects of local chemical enthalpy changes. The to volume and the external heat transfer while decreasing the
enhanced thermal energy leads to depolymerisation reaction internal thermal gradient - a case for carrying out kinetic studies.
towards production of smaller molecules. The thermally cracked The relative importance of the internal heat transfer to the external
molecules then diffuse both inwards ahead of the thermal wave heat transfer is defined by the Biot number (Bi), the ratio of their
and outwards through the hot char layer. Murty [13] reported with respective characteristic times:
the radiographs of the density of cellulose cylinders that the
decomposition wave passes radially through the cylinder. A ther- Bi ¼ ðsinternal =sexternal Þ ¼ hL=K
mal wave identified through the radial temperature measurements
suggests a chemical enthalpy change. It indicates that the thermal Biot numbers higher than 10 characterize a heat transfer lim-
conduction within the wood piece cannot be assumed as an inert ited by the internal conduction. When a solid biomass particle is
solid where thermal decomposition impacts the overall enthalpy heated by radiation and/or convection, a typical heat transfer pro-
change [13]. cess in a fixed bed gasifier, the heat is transferred from the surface
The internal heat transfer is defined by the energy conservation of the particle to the inner core through conduction [2]. Volatiles
inside the particle. The symmetry at the particle center is described and gaseous products participate in heat transfer and mass transfer
by the chemical decomposition and the external heat transfer at as well through the flow across the pores [3].
the surface, modeled by a global heat transfer coefficient. A charac- Sharma et al. [6] have developed a model of thermo-chemical
teristic time s associated with each process is related as, decomposition of biomass which includes the kinetics of both
the primary solid phase and the secondary gaseous phase (tar)
Internal heat transfer: sinternal = qcpL2/k reactions on a lumped basis. They compared the modeling results
External heat transfer: sexternal = qcpLh with the experimental results available in the literature for mass
loss of 25.4 mm diameter biomass particle with time at different
Chemical reaction: s ¼ Aeð RT Þ
E
reaction
temperatures. The model simulations were limited to the low tem-
perature pyrolysis zone within 415–834 °C range only. Sharma
If one of the characteristic times is much higher than the others,
et al. [6] compared the modeling results with the work of Park
the corresponding process is the limiting factor. When sexternal -
et al. [7] and found that negligible mass loss during the start of
 sinternal and sreaction, the external temperature is much higher
the process and predominant for the low temperature pyrolysis.
than the temperature inside the particle, which is uniform. When
The authors attribute the delay due to the heat conduction inside
pyrolysis temperature is reached, pyrolysis starts and goes to com-
the particle and state that for large size particles, conductive heat
pletion before thermal equilibrium occurs at the particle surface.
transfer determines the global devolatilisation rate.
On the other hand the limiting process is the internal heat transfer
Further, Sharma et al. [6] extended their modeling study to ana-
when sinternal  sexternal and sreaction, and the surface temperature is
lyze the effect of particle size on devolatilisation rate in an inert
1384 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395

and quiescent environment at constant temperature of 534 °C on experimental work on wood pyrolysis. Pyrolysis reactions, gas
cylindrical wood particles. They investigated for the devolatilisa- and tar evaluation were formulated in an Arrhenius type of expres-
tion rate and temperature rise of the core and surface of the cylin- sion, with the activation energy for the pyrolysis as 88.6 kJ/mol and
drical wood particles of size of 20 mm length and 1 mm, 2 mm and pre-exponent factor as 1.44  104. Pyrolysis reaction was derived
4 mm diameter respectively. They found out that 1 mm diameter with endothermicity of 0.42 MJ/kg. Pyrolysis rate, temperature
biomass particle behave like a thermally thin material and the core profile within the core and conversion rate were computed and
of the particle picks up the surface temperature quickly with neg- were in good agreement with the experimental results.
ligible time lag. Based on modeling results, Sharma et al. [6] found After completion of pyrolysis, char oxidation and reduction
out that the heat transfer resistance increases with the increase in reactions are of prime importance in gasification process towards
particle diameter. The temperature gradient was found to be very generation of combustible product gas. Dasappa [18] addressed
small for a 1 mm diameter particle which increases for 2 mm and the combustion of wood char particle in detail involving the C
4 mm diameter particle. They further analyzed that for biomass + O2, C + CO2 and C + H2O reactions, and extended the experimental
particle with size less than 1 mm, the heat transfer resistance work and the modeling to packed bed systems. He has addressed
can be ignored and a uniform temperature can be assumed the dependence upon the influence of particle diameter, tempera-
throughout the biomass particle during pyrolysis. ture and the reactive environment on the char conversion and also
Simmons and Ragland [15] have conducted combustion exper- established the kinetic and diffusion limits.
iments of millimeter sized wood cubes. Conditions similar to the In most of the practical thermo-chemical conversion devices, it
one in the furnace were simulated during the experiments includ- must be emphasized that the environment around the particles
ing flow variation over the particle with air as a reactant for the dynamically changes with time during the entire history of
flaming pyrolysis and char combustion. Parametric study with devolatilisation and char conversion process. Pyrolysis and char
respect to the variation in furnace temperature, flow velocity (Rey- gasification are the two major processes that finally leads to overall
nolds number) and oxygen concentration have been carried out. biomass conversion to fuel gas. Within the bulk volume of the
The study has shown that these parameters significantly influence packed bed, the various physical process that occur are heat trans-
the devolatilisation rate and char conversion rates. The total burn- fer between the fluid and the solid particles (convection) and
ing time of a 10 mm pine cubes was found to reduce from 81 to between the particles (conduction and radiation), fluid flow in-
68 s with a change in Reynolds number from 60 to 240. Using pine between the particles and homogenous gas phase reactions,
and oak wood chips as the lighter and denser wood respectively whereas the single particle conversion involves a more complex
and of similar size, they inferred that the burning rate per unit diffusion dominated species interaction and heterogeneous reac-
mass increases linearly with decrease in particle size and is higher tions within the particle surfaces apart from conductive heat trans-
for biomass with higher density (oak wood) compared to one with fer through the particle body. While the majority of the transports
lower density (pine wood). process that occurs between and within the particles are depen-
Literature shows substantial work on the inert pyrolysis and but dent on the properties of the fuel/biomass. It is important to under-
limited work has been done on the study of flaming pyrolysis; a stand the dependence of the thermo-physical properties on the
typical phenomenon experienced in a downdraft fixed bed gasifier pyrolysis process occurring within the packed bed.
where biomass undergoes pyrolysis in an oxidising environment. Literature reports extensive work on modeling of fluidised bed
One of the earlier works on the complete combustion of wood reactor [19–23] using both in house codes as well as established
flaming and glowing have been attempted by Mukunda et al. [5] software tools. Very limited work has been done on developing
through experiments and modeling studies. They have shown that mathematical model for fixed bed gasifiers. The fuel particles
classical d2 law prevails, and free convective effects need to be used in fluidised bed reactors lies in microns to millimeter size
addressed. This is similar to the arguments made by Murty and and fall under the category of thermally thin with low Biot num-
co-workers in their studies [11–14]. Parametric study on the pyrol- bers (less than 0.1), whereas the fixed bed reactors employ rela-
ysis process affecting pyrolysis, like the heat of decomposition and tively larger particles where the heat and species concentration is
thermal diffusivity has been studied. The literature reports general diffusion limited. The single particle biomass conversion analysis
behavior of the flaming time and glowing times but lacks on the work in literature is limited and lacks modeling of biomass con-
comparative study with experimental results with varying fuel version in varying ambient conditions and reacting environment.
properties, like density, thermal conductivity, etc. The packed bed models in the literature lack the independent
Models developed for pyrolysis focus on the set of complex validation and analysis of effects of thermo-physical properties
reactions leading to volatile characterization. It must be empha- of fuel at the particle level. Present work attempts to fill the
sized that the present work focus on development of mathematical gap while developing the mathematical model to address the
model of single particle biomass conversion suitable for packed simulation of single particle biomass conversion in an inert and
bed gasification process. Simple mass, species and energy balance reacting environment of a packed bed gasifier reactor. Present
is sufficient for present model as the transient products of pyroly- work adopts the char conversion model of Dasappa [18] incorpo-
sis is not of importance. Present work attempts to formulate the rating pyrolysis and char reactions to model biomass conversion
pyrolysis process through a global reaction. Prasad et al. [16] had using an in-house developed FORTRAN code. Present work ana-
performed kinetic investigation on pyrolysis and gasification char- lyzes and reports the effects of ambient temperature, density,
acteristics of Pongamia residue (de-oiled cakes) using thermo- thermal conductivity and particle diameter for pyrolysis process
gravimetry and downdraft gasifier. They identified the activation both in inert and reactive environment. The combination of
energies for pyrolysis in the range of 68.8–177.9 kJ/mol and pyrolysis (flaming) and char combustion (glowing) in a reactive
41.3–161.8 kJ/mol by differential method and FWO methods environment is further studied.
respectively, and the pre-exponential factor in the range of 105– The modeling results are compared and validated with the
1012 by differential method. The activation energy was found to experimental data taken from the literature towards evaluating
increase with conversion level and then reducing after 50% conver- the particle conversion time, weight loss rate at different input
sion level. The reduction in activation energy after 50% conversion data of temperature and particle diameter. Influence of the density
was attributed to slow thermal decomposition of lignin in de-oiled and thermal conductivity of the biomass on the overall pyrolysis
cake [16]. Di Blasi [17] has made careful observations during his process has also been addressed.
 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395 1385

Further, the complete combustion of particle of different diam- nomenon. Char oxidation and reduction is a complex phenomenon
eter in air is modeled towards predicting the conversion time, tem- governed by diffusion, adsorption and desorption of reactants and
perature profiles and weight loss rate which is discussed and products coupled with reaction. The porosity of the particle
compared with the available data from the literature. increases with volatile release dictates the further diffusion of
reactants. Porosity change with pyrolysis and with further char
reaction is also modeled.
2. Modeling of single particle analysis and significance of
The basic conservation equation governing the gasification pro-
pyrolysis model in present study
cess in the presence of gaseous species like O2, CO2, CO, H2O and N2
are set out. Pyrolysis, solid-gas reactions and gas-gas reactions are
Developing a model for a thermo-chemical conversion of bio-
incorporated in the model. Mathematical formulation employs fol-
mass in a packed bed downdraft gasification system involves both
lowing conservation equations to derive governing equation for
pyrolysis and char conversion process. Dasappa [18] had modeled
the model:
the process for char conversion used different reactants (CO2, H2O
and O2) with detailed reaction mechanism and extended to a

Mass conservation (gas and solid)
packed bed [24]. Pyrolysis is an important process during the gasi-

Species conservation equation (output gas species)
fication process where about 80% of the weight loss occurs and has

Energy conservation
a significant bearing on the overall packed bed gasification analy-
sis. While addressing the pyrolysis process which is further reacted
Fundamental governing equations are modeled in the present
with other species, the enthalpy of pyrolysis, rate of mass loss and
analysis by using unsteady spherically symmetric one dimensional
volatile release are of greater importance than the chemical com-
conservation equations. The assumptions made in this model are:
position of volatile yield. In order to simplify the modeling process
volatiles are considered as single specie which undergoes stoichio-
(i) Conversion process is in one dimensional
metric combustion in the bed. Rate of reaction of combustion of
(ii) Pressure gradient within the particle is neglected since the
volatile species is order of magnitude higher than the pyrolysis,
porosity of the particle is high
which makes the pyrolysis as a rate limiter validating the assump-
(iii) Uniformity of temperature between gas and solid (char)
tion of modeling the volatile as a single species.
Fig. 1 depicts the schematic diagram of the flow of process
When the solid biomass particle undergoes pyrolysis, it releases
occurring during a typical biomass particle undergoing thermo-
gaseous species (volatiles) and creates a porous char structure
chemical conversion process. With heat penetrating through radi-
behind. The porosity of biomass increases with the progress of
ation or convection to the surface initiates pyrolysis. The pyrolysis
pyrolysis. Further, the volatile undergoes combustion and release
process releases volatiles that either undergo combustion or issue
H2O and CO2, which in turn reacts with solid carbon (char) further
out as gas depending upon the reactive environment surrounding
increasing the porosity of the char. Hence, the average density (q)
the particle, leaving behind the porous char which undergoes the
of the porous char particle is related with the following relation.
reaction with the gas species diffusing inside the particle. The tem-
perature gradient, species concentration gradient and porosity q ¼ ðqC  Yc þ qb  Yb Þð1  Þ þ q ð1Þ
across the particle depth are of importance for particles of high Biot
Now at  ¼ 0 (no porosity) and Yc = 0 (only biomass), the rela-
numbers. Packed bed gasification employs biomass particle of
tion changes to q  ¼ qb ; means the average density of the particle
comparatively larger diameter (few millimeter to a few centimeter
is equal to the nonporous char density, as there are no gases. When
size). The Biot number varies between 0.4 and 4 for a particle
pyrolysis is complete, the porous structure is created with char as
diameter of 2–20 mm respectively. Therefore, the pyrolysis of lar-
the solid material behind, the density relation becomes
ger biomass particle in packed bed system falls in thermally thick
 ¼ qc ð1  Þ þ q. Again, at  ¼ 1 (after complete char conver-
q
regime. The heat transfer from the surface to core or the conduc-
sion), the relation changes to q  ¼ q; means the average density
tion of heat plays an important role in overall rate of pyrolysis
of porous char particle is equal to the gas density, i.e. the total solid
and particle conversion time. Heat transfer within the body is very
has converted to gases.
well modeled in the present study as the temperature gradients
Model formulation is adopted from Dasappa [18] and modified
become of utmost importance. In the gasification process, apart
to accommodate pyrolysis and thermal conductivity being formu-
from pyrolysis, char-gas reaction is equally important phe-
lated as a function of density (discussed in detail in Section 3.2.1).
Relevant aspects of the model are presented here.
For one-dimensional spherically symmetric systems, the gov-
erning equations are deduced as following:
Mass conservation equation (gas):
From the equation of continuity, conservation of mass associ-
ated with a fixed point in a flow can be expressed as;
@q
¼ r  ðqVÞ ð2Þ
@t
The control volume is a reacting system and as the porosity is
also changing with the time (due to the pyrolysis and char reac-
tion), the reaction term is added and gas density q replaced with
q. The above equation changes to;
@ðqÞ
¼ r  ðqVÞ þ -
_ 000 ð3Þ
@t b

where, - _ 000 3
b (kg/m s) is the volumetric reaction rate of the biomass
particle. For simplification, one dimensional spherically symmetric
Fig. 1. Schematic diagram of the processes occurring at the particle level. system has been considered in the present model. Hence, the
1386 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395

Eq. (3) translates to unsteady spherically symmetric one dimen- 2.1. Initial, interface and boundary conditions
sional mass conservation equation for porous char particle as
following: The set of governing equations (numbers) prove insufficient to
arrive at the solution as the number of unknown parameters is
@ðqÞ 1 @
¼ 2 ðqv r 2 Þ þ -
_ 000 ð4Þ more than the number of equations in the single particle analysis.
@t r @r b
Interface condition under valid assumptions provides extra set of
where, v is the superficial velocity of gas flow. equation facilitating the solution scheme formulation. A condition
 
Species conservation equation of quasi-steady state @t@ ðf Þ ¼ 0 is assumed in the gas phase spe-
cies and energy conservation equation which is justified for thin
The general vector form for mass conservation of ith species is interface. Reaction rates (of all heterogeneous and homogeneous
expressed as; reactions) are assumed zero (x _ 000
i ¼ 0) at interface implying only
@ðqY i Þ 0 transport of species. The energy and species equation for quasi-
_ 00i ¼ m
þrm _ 00i ð5Þ state gas phase translates respectively to:
@t


The term on RHS represents the net rate of mass production of _ P @T
mC 1 @ 2 @T
¼ r j ð12Þ
specie i per unit volume (i.e. volumetric reaction rate (-_ 000
i ) of ith 4pr 2 @r r2 @r @r
specie). As the porosity of particle is changing with the time, the


gas density term q is replace by q. Hence, the Eq. (5) can be writ- m_ @Y i 1 @ @Y i
¼ r 2
Dq ð13Þ
ten as follows: 4pr 2 @r r2 @r @r
@ðqYi Þ
_ 00i ¼ x
þ r:m _ 000 ð6Þ Integration constant is evaluated by assuming boundary condi-
@t i
tions at large diameter of the particle (r ! 1Þ, temperature is set
In the case of mass diffusion (no thermal diffusion or pressure to be T ! T1 and at the particle surface (r ¼ rS Þ, temperature is
diffusion) in a mixture, the general form of Fick’s law of diffusion considered to be (T ¼ TS Þ. Similarly for the species equation, the
can be used to evaluate the specie mass flux m _ 00i . For one- boundary conditions at r ! 1; Yi ! Yi1 and r ¼ rS ; Yi ¼ YiS (ith
dimensional spherically symmetric systems; the equation can be species at the surface of the particle). Assuming the Lewis number
deduced as following: to be unity, the final solution is derived as follows:


@ðqYi Þ 1 @ @Yi ðT  T 1 Þ ð1  gÞ
qv r2 Yi þ qDr2
0
¼ 2 þx
_ 00i ð7Þ ¼ ð14Þ
@t r @r @r ðT S  T 1 Þ ð1  gs Þ

Energy conservation equation

ðY i  Y i1 Þ ð1  gÞ
¼ ð15Þ
The heat flows in/out from the control volume through conduc- ðY is  Y i1 Þ ð1  gs Þ
tion and convection process. The energy conservation equation can    
_ P _ P
be described as; where, g ¼ exp  4mC
pjr and gs ¼ exp  4mC
pjrs . The subscript ‘s’

Heat Accumulation ¼ Heat production þ Heat Input refers to the value at the surface of the sphere. The solution of
energy and species-energy conservation equation can also be writ-
 Heat output ð8Þ ten in the following form:
Evaluating the individual terms in Eq. (8) and simplifying it to ðT  T 1 Þ ðY i  Y i1 Þ ð1  gÞ
the unsteady one-dimensional spherically symmetric systems, ¼ ¼ ð16Þ
ðT S  T 1 Þ ðY is  Y i1 Þ ð1  gS Þ
the final form of energy conservation equation becomes:

 Differentiating Eq. (16), at the surface (r ¼ rS ) will provide the
@ 1 @   2 @T
ðq
 CP TÞ ¼ q v r 2
CP T þ r j x
_ 000
b Hb ð9Þ interface conditions. At first the energy equation is considered
@t r2 @r @r
for obtaining the interface conditions; at the surface of the sphere:
The LHS term is unsteady term; the first and second terms on
ðT  T 1 Þ ð1  gÞ
RHS are convective and conductive heat transfer terms, and third ¼ ð17Þ
ðT S  T 1 Þ ð1  gS Þ
term on RHS is the reaction term. Hb is the sum total of enthalpy
of all the reactions taking place inside the biomass particle includ- Now, differentiating on both sides by at r = rs, the final expres-
ing pyrolysis and char reactions. sion is derived in differential form as:
Mass conservation equation (solid) @T
j ¼ C P Q ðT 1  T S Þ ð18Þ
The mass of the porous biomass particle is related with the
@r
h i
porosity of the particle, as described in Eq. (1). Differentiating Eq. where, Q ¼ m_ expðB0 Þ
and B0 ¼ 4mC
_
prs .
P
4pr2s f1expðB0 Þg
(1) and rearranging the terms leads to the following relation:
Similarly, the species equation has considered obtaining the
@ mS  @Yb @ interface condition,
¼ ðqb  qC Þð1  Þ  ðqb Yb þ qC Yc Þ ð10Þ
@t V @t @t @Y i
Dq ¼ Q ðY i1  Y iS Þ ð19Þ
where, mS is the mass of the porous char and V is the volume of @r
particle.
The term on LHS represents the rate of mass change per unit
3. Methods of solution
volume with time, i.e., volumetric reaction rate of biomass particle
(-
_ 000 3
b ) (kg/m s). Eq. (10) can be rearranged in the form;
The solution calls for integration of the parabolic system of par-
tial differential equations with the initial and boundary conditions
@Yb @ _
ðqb  qC Þð1  Þ  ðqb Yb þ qC Yc Þ ¼ - 000 ð11Þ reported in the previous section. The independent variable r is
@t @t b  
transformed into V ¼ 43 pr3 making the systems of equations fully
 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395 1387

conservative and removing the singularity at r = 0. The conserva- carbon (char) evaluated in our own experiments. The value of qc is
tion equations translate to: of the non-porous carbon density as suggested in that is chosen

 from Dasappa [18]. Values of rp, the pore radius, and s, the tortuos-
@ @ _ P T þ ð4pÞ2=3 ð3VÞ4=3 @T  Hb -
ðq
 CP TÞ ¼ mC _ 000 ð20Þ ity factor, are chosen from Groeneveld [26]. kc and kg are non-
@t @V @V b
porous char and gas thermal conductivity respectively chosen from

 Dasappa [18] which leads to porous char conductivity of 0.4–0.5 W/
@ @ _ i þ De qð4pÞ2=3 ð3VÞ4=3 @Yi þ -
ðqYi TÞ ¼ mY _ 000 ð21Þ (m K) and accounts for conduction and radiation inside the char.
@t @V @V i
Various other researchers [4,27–29] also used the similar values
For obtaining the mass loss rate, the equation of state is used to for thermal conductivity of char due to little variation in its intrinsic
get the relation. The Eq. (1) and equation of state (22) are used to nature, structure and composition (basically solid carbon).The ther-
calculate the mass flow rate. mal conductivity of gas phase, kg, is calculated taking into account
the presence of H2. Murty [13] observed the conductivity of mix-
PV ¼ nRT ð22Þ
tures increase by 20–50% with 10% mole fraction of H2 in the mix-
The governing equations and equation of state is used to derive ture. Unreacted char porosity is measured and is found similar to
a relation between different governing equations to converge at the one used by Dasappa [18].
solution.

 X 1 @ 3.2.1. Thermal conductivity of biomass
q ›ðq‰Þ q  ›e
þ 1 C  Mg ðqeY i Þ Various researchers [4,25,27,29] used the same value of thermal
q ›t q ›t M i @t conductivity of char, irrespective of the biomass species, due to its


q @ intrinsic nature, structure and composition (mainly carbon). On the
 ðq
 C P TÞ
q TC P @t other hand, thermal conductivity of biomass varies across different
¼0 ð23Þ species. Fig. 2 presents the thermal conductivity and diffusivity of
variety of biomass which shows the correlation between thermal
It can be clearly seen from the Eq. (23) that the 1st and 2nd term conductivity and density. The thermal conductivity, specific heat
contains (in bold) the LHS of mass conservation equations in gas and density data has been taken referred from Wood Handbook
and solid phase respectively. The values are derived from respec- published by U.S. Dept. of Agriculture [30]. It is evident from the
tive conservation equation and used to determine the mass flow data presented in Fig. 2 that the thermal conductivity increases lin-
_ The integration of the above equations is carried out by
rate (m). early with density and thermal diffusivity decreases with the
using the implicit Crank-Nicholson scheme. The present computa- increase in density. Thermal conductivity is highly dependent on
tions used 20 grids. It is found that the results are grid independent density apart from the presence of inorganic compound and orien-
by obtaining the results for 20/40 grid points. tation of grain to the heat flux.
Ragland et al. [31], in their extensive study on thermo-physical
3.1. Kinetics of pyrolysis and char reactions properties of variety of biomass, suggested a correlation for ther-
mal conductivity of wood as a function of density
The pyrolysis rate equation in Arrhenius form has been used
from the Di Blasi’s [25] work on pyrolysis of wood particles. The K wood ¼ 0:1941qw þ 0:0186 W m1 K1 ð25Þ
given rate equation rate signifies the conversion of biomass to where, Kwood is thermal conductivity of wood, and qw is respective
volatiles (gas) and left over carbon (char). The pyrolysis reaction wood density in g/m3. The correlation suggested by Ragland et al.
is assumed to be slightly endothermic with the heat of reaction [31] matches closely with the correlation derived from Wood Hand-
being 0.42 MJ/kg. book [30], as show in Fig. 2.


Epyr Enrico and Baldi [29] also used the formula suggested by Rag-
-000pyr ¼ Apyr X bio exp  ð24Þ land et al. [31] in their study on wood pyrolysis and modeling. Cor-
RT
relation of thermal conductivity and density, as suggested by
where, Apyr = 1.44  104 s1, Xbio = Biomass fraction, Ragland et al. [31], is used in the present study as well for evaluat-
Epyr = 88.6 kJ/mol. ing the thermal conductivity of wood.
During the devolatilisation process, biomass will convert to
char (Xchar) and gas. Xbio signifies the amount of biomass fraction 3.3. Modeling for flaming pyrolysis
available in the control volume at a given time. At any time,
(Xchar + Xbio) will be unity in the given control volume. The flaming pyrolysis conditions are simulated in the model
The scheme of char gas reaction (carbon with CO2, H2O and O2) and gas phase reactions are included in the model. A lump model
is adapted from Dasappa [18] and implemented. is adopted for evolution of volatile specie and its cracking to CO,
H2 and CH4. Assuming the 80% volatile content in the wood, it is
3.2. Choice of parameters cracked down to CH4, H2 and CO maintaining the elemental bal-
ance of C, H and O. Further, CO, CH4 and H2 undergo stoichiometric
The choice of physical, thermodynamic and transport properties combustion releasing heat of reaction. The calorific value of vola-
is based on the mean properties presented as follows: tile specie (C0.616H1.4O0.6) is evaluated as the difference of the
calorific value of biomass and char yield. The mild exothermicity
qc = 1900 kg/m3, qb = 650 kg/m3, rp (t = 0) = 50 lm, is the balance enthalpy of difference in calorific value of volatile
cp = 1.25 kJ/kg, Hc = 32.6 MJ/kg, kc = 1.85 W/m K, specie and that of CO, CH4 and H2.
kg = 0.071 W/m K s = 1.5 e = 0.88
C1 H1:4 O0:6 ðwoodÞ ! C0:616 H1:4 O0:6 ðvolatileÞ þ 0:384CðCharÞ
(for unreacted char)
 0:42 MJ=kg ð26Þ

C0:616 H1:4 O0:6 þ 0:008 O2 ! 0:016 CH4 þ 0:6 CO þ 0:668 H2

qb is the mean density for the Casuarina Equisetifolia woody biomass
þ 59:8 kJ=kg ð27Þ
measured and used for the present work. Hc is the calorific value of
1388 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395

Fig. 2. Thermal conductivity of various biomass species in comparison with their density [30].

CO þ 1=2 O2 ! CO2 þ 10:08 MJ=kg ð28Þ 4.2. Effect of temperature on conversion rate

Fig. 3 presents the experimental results of Park et al. [7] on the

H2 þ 1=2 O2 ! H2 O þ 141:65 MJ=kg ð29Þ effect of temperature on mass loss rate using wood spheres of
25.4 mm diameter, and compares with the present modeling sim-
ulation results. Weight loss from the particle was monitored with
CH4 þ 2 O2 ! CO2 þ 2H2 O þ 55:52 MJ=kg ð30Þ
time till the completion of pyrolysis. Simulations were carried
The reaction (26) occurs inside the particle and reaction (27) out using the wood properties (qb = 450 kg/m3, kw = 0.17 W/m K)
takes place outside the particle. Reaction (27) is followed for vola- and different ambient temperature set (688 K, 783 K and 879 K)
tile combustion, including volatile cracking and then combustion identified with the reference [7]. Quiescent conditions is consid-
of CO, CH4 and H2 (reactions 28–30). Availability of adequate ered (Re ? 0) with N2 in the ambient.
amount of O2 is assumed during flaming pyrolysis which is a valid The prediction from the model also show a similar weight loss
assumption for flaming pyrolysis of biomass burning in air as the profile and is found to be reasonably matching with the experi-
fresh flow of air continuously replaces the product of volatile and mental observations on the weight loss except in the case of
char combustion. In the absence of air (inert pyrolysis), the reac-
tion doesn’t proceed and conditions for inert pyrolysis are
achieved. Depending on the rate of volatile release from the parti-
cle (on mass basis), the total heat of reaction is calculated. The
enthalpy of the reactions is of prime importance in the present
case, as the heat released (or the physical flame) subsequently sup-
plies necessary energy to drive the pyrolysis and endothermic char
reduction reactions. The heat of volatile combustion along with the
interface conditions further dictates the heat transfer to the
particle.

4. Results and analysis

4.1. Pyrolysis in an inert environment

It is evident from the pyrolysis rate Eq. (24) that the devolatil-
isation or mass loss rate varies exponentially with the local tem-
perature. Analyzing the energy conservation Eq. (9) closely, the
temperature distribution within the particle is dependent on the
heat flux, thermal conductivity, particle size and density. In this
section, the effect of temperature, particle diameter, thermal con-
ductivity and density of the fuel on the pyrolysis rates are analyzed
and discussed. The results are compared and validated with the Fig. 3. Weight loss with time - comparison of modeling results with experimental
data available in the literature. results of Park et al. [7] on 25.4 mm wood sphere at different temperature.
 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395 1389

688 K. The weight loss is computed based on the loss of volatiles

from the sphere with time. An increase in the ambient temperature
reduces the time taken for devolatilisation.
A close observation of the weight loss profile indicates that the
conduction time delay during the initial heating period plays an
important role resulting in the difference in the devolatilisation
rate. With the increase in temperature the incipient time for
devolatilisation reduces. The time taken for the devolatilisation
(80% weight loss) decreases from 700 s to about 200 s with an
increase in furnace temperature from 688 K to 879 K.
This aspect is further analyzed by using the core and surface
temperature profile presented in Fig. 4. Fig. 4 presents the results
from the model, where the core and surface temperature of the
particle are plotted against time during the pyrolysis process at dif-
ferent ambient temperature. There is a visible lag between the
temperature rise at the surface and the core for all the cases. Con-
duction being the predominant heat transfer mechanism within
the particle, the core temperature lags behind the surface temper-
ature. Further, with increase in temperature, the net effect of
increased heat flux is clearly evident from Fig. 4; where the core
temperature increases at faster rate. The core temperature rises Fig. 5. Temperature rise with time at the surface and the core for a 20 mm diameter
to 350 K in 30 s with the ambient temperature set at 879 K com- poplar wood at 1600 K – comparison of present modeling result with Di Blasi [25]
pared to 110 s at 688 K. From the analysis, the ability of the model modeling result.

to predict the weight loss of a particle undergoing inert pyrolysis at

different temperature is established.
4.2.1. Superficial activation energy for pyrolysis
Fig. 5 compares the present modeling simulation results with
Towards addressing the approximation of the activation energy
the simulation results of Di Blasi’s [25] model where the kinetics
used in the present study, the data from the numerical experi-
of primary and secondary reaction kinetics for the pyrolysis has
ments conducted are used. An attempt has been made to collapse
been considered. The fuel properties for the simulation were cho-
the conversion with time data from different sets of experiments to
sen from Di Blasi [25] (20 mm poplar wood qb = 504 kg/m3,
address the effect of temperature by using a corrected time defined
k = 0.14 W/m K) and the ambient temperature was set at 1600 K.
as:
The prediction of surface temperature is in close agreement with
0 1
Di Blasi’s [25] results but some deviation is observed in the core
Ea
temperature. Di Blasi [25] used primary and secondary reaction T ¼ Texp@ 
0
A
mechanism for the pyrolysis where there action between the prod- R 1T  T1
ref
ucts of pyrolysis has influence on the temperature locally within
the particle, whereas the present model adopts the single step where, Tref is the reference temperature (950 K), and Ea is the super-
pyrolysis model and attributes the deviation in the temperature ficial activation energy for the pyrolysis process. Numerical experi-
profile to the enthalpy difference due to the different reaction ments towards estimating the activation energy for this case yields
model. the Ea as 20 kJ/mol. The difference between the true activation
energy used in the model 88.6 kJ/mol and the superficial activation
energy is argued to be due to the diffusion effects.
t 0c
tc ¼
T 01:8
Fig. 6 presents the normalized conversion time ðt c Þ of model
results plotted in Fig. 3 with temperature and compare it with
the conversion time at 950 K. It is evident that after normalizing
the conversion time with temperature, the conversion profiles
have collapsed to a single profile, close to the profile of 950 K. This
result validates the effect of temperature on the conversion time
simulated by the present model.

4.3. Effect of density and thermal conductivity

As discussed earlier, the thermal conductivity of biomass is a

function of density, and both density and thermal conductivity
influences the conversion rate of the biomass particle. It is clearly
evident from energy conservation Eq. (9) that the density (q  ) and
thermal conductivity plays important role in heat diffusion and
temperature profile of the biomass particle. One can infer from
the unsteady term on the RHS of energy conservation Eq. (9) that
the energy gain/loss on LHS will lead to higher temperature rise/-
fall for particles with low density whereas lower temperature rise/-
Fig. 4. Modeling result of temperature profile at the core and surface of the particle fall for denser particles. The temperature in turn significantly
of 25.4 mm diameter. affects the pyrolysis and char reaction rate, and hence overall con-
1390 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395

rates, for a range of values (0.1–0.2 W/(m K)) reported in the liter-
ature [30]. The simulations were carried out in inert pyrolysis con-
dition with theN2 flow at 1273 K, and wood with a density of
650 kg/m3. Fig. 8 presents the results which clearly shows that
conversion time decreases with the increase in thermal conductiv-
ity, and at kw = 0.2, the time for conversion varies with diameter
as  d1.2 1.7
o , while at kw = 0.1 the time for conversion varies as do .
Though the numerical simulation result analysis in this section
doesn’t predict any real scenario (as density and thermal conduc-
tivity are not independent), but these differences clearly suggest
that the internal thermal resistance have a significant contribution
towards the overall pyrolysis rates.
Varunkumar et al. [32] reported the increase in bed tempera-
ture using high-density pellet compared to the low-density wood.
They attributed only the density as the factor in the behavior with-
out any reference to the effect of thermal conductivity. From the
data, the high-density pellet also has higher ash content, implying
increased conductivity resulting in reduction of conductive resis-
tance. With increased pyrolysis rates due to faster penetration of
heat and with the assumption that oxidizer is available, in the case
of combustion, it is prudent to assume increased temperature.
Fig. 6. Weight loss with normalized conversion time of the modeling results on Various results cited above suggests that the choice of thermo-
25.4 mm wood sphere with reference temperature of 950 K. physical parameters, kinetic parameters are justified, with the pre-
dictions of results are in close agreement with the experimental
results.
version rates. Similarly, the thermal conductivity in the diffusion
term on LHS of energy conservation Eq. (9) directly affects the heat 4.4. Analysis of the density effect and Biot number on conversion rate
diffusion along the radial direction and the thermal gradient within
the particle. The present model is simulated for variation of density Density effect on the devolatilisation rate of biomass particle is
and thermal conductivity independently to analyze the influence evident from the results, as discussed above. An attempt has been
of each property of biomass on temperature profile and devolatil- made to normalize the devolatilisation time with density from dif-
isation rates. ferent sets of data available in the literature. Fig. 9 compare the
Numerical experiments were carried out by varying density of results for different furnace conditions where devolatilisation time
the fuel from 300 kg/m3 to 1200 kg/m3 with constant thermal con- is normalized with density for inert pyrolysis, based on the data
ductivity of 0.15 W/(m K), for 10, 15 and 20 mm diameter particle available in the literature. Table 1 provides the details of the data
in an inert environment of N2at 1273 K. It can be observed from used from the literature. The temperature at which the experi-
Fig. 7 that the conversion time is increasing with the density. The ments are conducted with the cited reference is identified against
conversion time with density profiles is found to be diverging with the each data point. The modeling results are presented for differ-
the increase in diameter implying increased density dependence of ent temperatures namely, 688 K, 783 K, 879 K and 1100 K which
larger particles. covers a range of experimental conditions. The general behavior
Similar numerical experiments were carried out to evaluate the of reduction in the normalized devolatilisation time of the particle
influence of thermal conductivity of biomass on overall conversion

Fig. 7. Variation in conversion time of biomass with varying density with constant Fig. 8. Variation in conversion time of biomass with varying thermal conductivity
thermal conductivity of 0.15 W/(m K) at different particle size at 1273 K. at different particle size.
 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395 1391

Fig. 9. Normalized devolatilisation time with particle size of different biomass species from the literature and validation of present modeling result for inert pyrolysis at
different temperatures (qc = 500 kg/m3 for modeling results, Index number for the data points - refer to Table 1).

Table 1
Index number and data used for Figs. 9, 10 and 13.

Data point Biomass Density (kg/m3) Particle size (mm) Temperature (K) Thermal conductivity (W/m K) Devolatilisation time (s) Reference
1 Beech wood 730 11.5 950 0.16 85 [29]
2 1050 70
3 1150 55
4 Pine wood 510 13.5 1250 0.17 70 [33]
5 Wood 500 25.4 688 0.116 650 [7]
6 783 350
7 879 225
8 Maple wood 630 25.4 638 0.14 800 [7]
9 688 725
10 736 450
11 783 375
12 831 350
13 879 250
14 Pine wood 370 13.8 1100 0.09 38 [15]
15 900 47

with an increase in the temperature is observed both in the exper- Nu ¼ 2 þ 0:589 Ra1=4 =Pr ð31Þ
iments and model. It can be observed from the data in Table 1 and
Fig. 9, that the data points from different literature with different where, Rayleigh number (Ra) depends on the flow field parameters
density but with small or no difference in temperature and particle and the temperature of the ambient and particle surface, Pr is the
diameter does fall to the similar location in the plot, validating the Prandtl number and kg is the thermal conductivity of ambient gas
normalization of devolatilisation time with density. It can be around the particle.
observed that the data point 5 and 9, 3 and 15, 7 and 13, 6 and For all the cases presented in Fig.10, i.e., varying diameter cou-
11 have very narrow difference in normalized devolatilisation pled with different ambient temperature, the Bi number is less
time. The modeling results simulated at density of 500 kg/m3 also than 10 suggesting that the heat transfer limited by the internal
closely matches with the data points from literature. conduction [27]. It is also evident that at high temperature and
Further, with the reduction in the diameter of the particle, the smaller particle diameter, Biot number is close to 1, indicating that
time for conversion approaches an asymptotic level. This is argued both internal and external resistance to heat transfer are compet-
to be based on the competition between the internal and external ing; suggesting departure from the d2 law for a typical diffusion
heat transfer process, a representative of Biot number. Biot number dominated process.
for the samples considered is analyzed and presented in Fig.10.
Biot number is defined as Bi = hL/k; where h is defined as the 4.5. Flaming pyrolysis
heat transfer coefficient at the surface, L is defined as the charac-
teristic length, the ratio of volume to surface for the sphere, and Simmons and Ragland [15], Mukunda et al. [5] and Gnanendra
k is the thermal conductivity of the biomass. Towards estimating et al. [34] have performed experiments on combustion of biomass
the heat transfer coefficient the experimental conditions data, Nus- sphere of different particle diameter. Mukunda et al. [5] used low
selt number correlation is used as indicated in the Eq. (31). Heat ash casuarina wood (density of 620 kg/m3) as a fuel and Gnanendra
transfer coefficient, h, is estimated using h = Nukg/L, where Nusselt et al. [34] used dense (density of 910 kg/m3) biogas sludge bri-
number, Nu is evaluated for flow over a sphere as, quette with 10% ash content. The model is simulated for 5–
1392 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395

Fig. 10. Normalized devolatilisation time and Biot number vs particle size of various biomass species from literature and present modeling result for inert pyrolysis at
different temperature (index number for the data points - refer to Table 1).

30 mm diameter particle under oxidising and quiescent conditions, and higher ash content in the biogas sludge. A deviation from the
similar to the flaming pyrolysis. In Fig. 11, the model results for d2 law is attributed to the presence of free convection [5,12].
flaming pyrolysis, where the pyrolysis gases react with the air to
produce a diffusion flame is compared with the experimental 4.6. Simulation of complete conversion of single particle under
results from the literature. The experimental results of flaming isothermal condition in air – influence of surface area to volume ratio
time of wood and biogas sludge briquettes have been taken from and heat flux on conversion rate
Mukunda et al. [5] and Gnanendra et al. [34] respectively. The
other properties of casuarina wood in the model were used as Commercial fixed bed biomass gasifier seldom uses spherical
per the experimental work of Mukunda et al. [5] i.e. particles, rather cylindrical or cubical pieces in general. It is imper-
kw = 0.168 W/(m K), cp = 1.4 kJ/(kg K) and the conductivity of bri- ative to analyze the effectiveness of the model, formulated for
quettes is estimated using correlation [31]. spherical geometry, on the other geometric shape particles. Sim-
Similar to the observation cited in the literature, the depen- mons and Ragland [15] studied pyrolysis and combustion of
dence on the diameter on flaming time is 2  103d1.7q for wood 10 mm pine wood cubes in air at a controlled temperature
while it is 1  103d1.6q for briquettes [34]. Gnanendra et al. [34] between 900 and 1200 K. While the excellent experimental data
argues the difference in the flaming time between these two spe- were available the authors had used cube for their studies. The
cies (wood and briquettes) to be due to the lower volatile content results from Simmons and Ragland [15] were chosen to simulate
the combustion of a particle in a furnace subject to different ambi-
ent temperature addressing both flaming as well as char combus-
tion. The code presently written has spherical co-ordinates and in
order to accommodate the results for the cube, the cube has been
converted to sphere of a certain diameter based on two criterion–
surface are equivalent and volume equivalent. Simmons and Rag-
land [15] conducted experiments with 10 mm cube whose surface
area and volume are 600 mm2 and 1000 mm3 respectively. A
sphere of 12.4 mm diameter would provide an equivalent volume
(i.e. equal mass) of a 10 mm cube while the surface area of 483
mm2 that is lower by about 15%, which implies lower heat flux
on the particle. Choosing 13.8 mm diameter sphere based on
equivalent surface area (allowing equal heat flux) of a 10 mm cube,
the volume estimates are of 1376 mm3 (i.e. higher mass to convert)
38% higher compared to the cube.
The model is simulated for the 12.4 mm and 13.8 mm wood
sphere with the properties of pine wood (density at 370 kg/m3
and thermal conductivity of 0.12 W/(m K)).The complete conver-
sion comprising of pyrolysis/flaming and char conversion is used
to simulate the combustion of the wood particle in the furnace
set at a different temperature in air. Fig. 12 illustrates the conver-
Fig. 11. Normalized devolatilisation time vs. particle size of various biomass
sion profile of the particle at different temperatures with respect to
species from the literature and present modeling result for flaming pyrolysis at time. The mass loss rate curve shows two distinct slopes; the initial
different temperature. phase of high mass loss rate is due to devolatilisation and flaming
 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395 1393

Fig. 12. Comparison of experimental weight loss profile of 10 mm pine cube [15] with modeling result of 12.4 mm and 13.8 mm pine wood sphere.

pyrolysis. The second phase of slow mass loss rate is identified as conversion process. Further, the time for conversion also reduces
char conversion phase that involves slow heterogeneous reactions with temperature. Comparing this results with Dasappa et al.
inside the porous char body. [35] on char conversion for a diameter of 13 mm diameter at
1273 K tc/q is 1.0 s m3/kg, and in the present case it is tc/q for the
4.7. Discussion on the flaming period 1200 K temperature is 0.95 s m3/kg, the difference within 5%.
The predominance of heat flux and surface area over volume
Fig. 12(a)–(d) compares weight loss with time for the modeling has been highlighted. It is important to note that the average bed
results and the experimental data from Simmons and Ragland [15]. temperature during packed bed gasification in downdraft configu-
The conversion profile of the 10 mm pine cube lies in between the ration is found to be in the range of 1000–1400 K, and the above
conversion profile of two different diameter spheres (12.4 mm and analysis provides the confidence to model spherical particles for
13.8 mm). With the increase in temperature, the weight loss pro- the wood cubes used in actual practice, with little margin of error.
file of both the spheres approaches with the one under considera- The above finding suggests to define the particle diameter in the
tion - a 10 mm cube. It is evident from Fig. 12(d) that at 1200 K, the model with similar surface area of the wood pieces used in exper-
small difference almost diminishes that exists at a lower tempera- iments. Apart from devolatilisation, as discussed in previous sec-
ture with respect to the weight loss profile of both the cube and tions, it was also important to validate the char conversion (after
spherical particle. pyrolysis) to establish the model as a robust solution towards fur-
Such behavior is justified based on the internal resistance to be ther packed bed modeling. From the analysis in the present section
nearly same irrespective of the external flux. The flaming time is and simulation of complete conversion of wood particle, both the
nearly same for both the diameters, suggesting the process is lim- sub-processes, pyrolysis and char conversion, has been established.
ited by conduction. Extending a similar analysis for comparison with the experi-
Beyond the flaming pyrolysis, the ambient air reacts with the ments carried by Mukunda et al. [5] and Gnanendra et al.[34],
char, and the remaining 20% weight loss occurs during the carbon the modeling simulations were performed on combustion of bio-
1394 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395

mass sphere for 5–30 mm diameter particle under flaming and qui-
escent conditions. The flow of air was maintained at Re ? 0, and
complete combustion was allowed to take place. The other proper-
ties of casuarina wood in the model were used as per the experi-
mental work of Mukunda et al. [5], i.e. kw = 0.168 W/(m K),
cp = 1.4 kJ/(kg K) and the conductivity of briquettes is estimated
using Eq. (25). The methodology for volatile combustion for simu-
lating the flaming pyrolysis is adopted.
Table 2 presents and compares the model results of flaming and
glowing time with respective experimental results. Ash has a sig-
nificant effect on the glowing time. The 10% ash was accounted
for in the model for the biogas sludge briquette, and the results
were reasonably accurate. The model predicts the glowing and
flaming time for woods as well as high ash, high density briquette
reasonably well under combustion.

4.8. Comparison with the experimental data accounting for various

biomass properties

As discussed above, the normalization of the devolatilisation

Fig. 13a. Devolatilisation time of inert pyrolysis for different biomass species listed
time with respect to temperature and density provides useful in Table 1.
information on understanding and validating the thermo-
physical properties effect on conversion of biomass.
Further, an attempt to normalize the conversion time with
respect to the particle size, temperature and properties of the fuel
– thermal conductivity and density is presented. Fig. 13(a) presents
the consolidated data of devolatilisation time from various exper-
iments from the literature and the model. The x-axis denotes the
sample numbers (details in Table 1), and the y-axis in Fig. 13(a)
presents the devolatilisation time. There is widespread in the
raw experimental data points having a varying particle size,
thermo-physical properties and experimental conditions. Fig. 13b
presents the consolidated data of normalized devolatilisation time
of the results presented in Fig. 13(b). There have been variations in
the temperature in various experiments that have been conducted.
The effect of temperature using a corrected time is considered for
the normalization as discussed above:
0 1
Ea
T ¼ Texp@ 
0
A ð32Þ
R 1T  T1
ref

where, Tref is the reference temperature (950 K), and Ea is the super-
ficial activation energy for the pyrolysis process. For this case, the Ea
is 20 kJ/mol. In the present modeling study, the dependence on par-
Fig. 13b. Normalized devolatilisation time of inert pyrolysis for different biomass
ticle diameter has been observed varying as tc  do1.5 and depen- species listed in Table 1.
dence on thermal conductivity as tc  ko1.45.
Normalizing with respect to the temperature, thermal conduc-
tivity, density and particle diameter is presented in Eq. (28), where
tc is the actual devolatilisation time and t0c is the normalized The single particle model results with correction factors, sug-
devolatilisation time. gests that the predictions with such variability are good enough
tc to be used as a robust model for the packed bed gasification and
t 0c ¼ ð33Þ combustion process.
T 0 qk
1:8 1:45 1:5
d0

Table 2
Comparison of flaming and glowing time of particle combustion.

Particle diameter (mm) Flaming time, tc Flaming time, tc Glowing time, tc Glowing time, tc
Experiment Model Experiment Model
Wood Briquette Wooda Briquetteb Wood Briquette Wooda Briquetteb
10 60 73 69 55 220 450 240 477
15 120 131 125 134 500 757 523 723
20 200 203 199 160 750 970 781 890
25 270 274 276 265 950 2154 978 2018
30 361 370 2710 2525
a
Mukunda et al. [5].
b
Gnanendra et al. [34].
 S. Kumar, S. Dasappa / Applied Thermal Engineering 112 (2017) 1382–1395 1395

5. Conclusion [8] A. Boateng, P.L. Mtui, CFD modeling of space-time evolution of fast pyrolysis
products in a bench-scale fluidized-bed reactor, Appl. Therm. Eng. 33–34
(2012) 190–198.
Present work addresses modeling of thermochemical conver- [9] M.I. Tamer, M. Abd ElSalam, M. Eliseu, R. Abel, Eulerian – Eulerian CFD model
sion of biomass particle in both the inert and reactive environment, on fluidized bed gasifier using coffee husks as fuel, Appl. Therm. Eng. 106
(2016) 1391–1402.
forming an integral part of the overall gasification system. One
[10] Meng Ni, Dennis Y.C. Leung, Michael K.H. Leung, K. Sumathy, An overview of
dimensional spherically symmetric model has been formulated hydrogen production from biomass, Fuel Process. Technol. 87 (5) (2006) 461–
using the fundamental governing equations of mass, energy and 472.
[11] K.A. Murty, P.L. Blackshear, Pyrolysis effects in the transfer of heat and mass in
species conservation, coupled with detailed pyrolysis, volatile
thermally decomposing organic solids, Symp. (Int.) Combust. 11 (1967) 517–
combustion, char gas heterogeneous reactions and water gas shift 523.
reaction. The numerical simulation results are compared and vali- [12] K.A. Murty, P.L. Blackshear, On the combustion of wood II: the influence of
dated with the experimental and modeling results available in the internal convection on the transient pyrolysis of cellulose, Combust. Sci.
Technol. 2 (1970) 5–9.
literature. [13] K.A. Murty, Thermal decomposition kinetics of wood pyrolysis, Combust.
Thin film approximation is invoked in the present study and Flame 18 (1972) 75–83.
enthalpy of reaction for the volatile combustion is incorporated [14] P.L. Blackshear, K.A. Murty, On the combustion of wood I: a scale effect in the
pyrolysis of solids, Combust. Sci. Technol. 2 (1970) 1–4.
in the energy equation for the simulation of flaming pyrolysis. [15] W. Simmons, K. Ragland, Burning rate of millimeter sized wood particles in a
The model is simulated for the flaming and inert pyrolysis, and furnace, Combut. Sci. Technol. 46 (1986) 1–15.
pyrolysis and char conversion process are independently validated [16] L. Prasad, P.M.V. Subbarao, J.P. Subrahmanyam, Pyrolysis and gasification
characteristics of Pongamia residue (de-oiled cake) using thermogravimetry
with the experimental results available in the literature. and downdraft gasifier, Appl. Therm. Eng. 63 (2014) 379–386.
Influence of the thermo-physical properties, namely thermal [17] C. Di Blasi, Comparison of semi-global mechanisms for primary pyrolysis of
conductivity, density and particle size of the fuel on the overall lignocellulosic fuels, J. Anal. Appl. Pyrol. 47 (1998) 43–64.
[18] S. Dasappa, Experimental and modeling studies on the gasification of wood-
conversion rate and temperature profile inside the particle has
char Ph.D. thesis, Indian Institute of Science, India, 1999.
been studied in-depth. The normalization of the conversion time [19] Qingang Xiong, Song-Charng Kong, A. Passalacqua, Development of a
with data from different sets of experiments proposing superficial generalized numerical frame work for simulating biomass fast pyrolysis in
fluidized-bed reactors, Chem. Eng. Sci. 99 (2013) 305–313.
activation energy addresses the effect of temperature by using a
[20] Qingang Xiong, S. Aramideh, Song-Charng Kong, Modeling effects of operating
corrected time. The close dependence of conversion time and tem- conditions on biomass fast pyrolysis in bubbling fluidized bed reactors, Energy
perature profile inside the particle on the Biot number, heat flux Fuels 27 (2013) 5948–5956.
and surface to volume ratio has been established. Particle with Biot [21] Qingang Xiong, Song-Charng Kong, Modeling effects of interphase transport
coefficients on biomass pyrolysis in fluidized beds, Powder Technol. 262
number close to unity, with both internal and external heat trans- (2014) 96–105.
fer competing with each other, suggests departure from the d2 law [22] Qingang Xiong, Fei Xu, E. Ramirez, S. Pannala, C. Stuart Daw, Modeling the
whereas larger size particle shows close dependence on d2 law sug- impact of bubbling bed hydrodynamics on tar yield and its fluctuations during
biomass fast pyrolysis, Fuel 164 (2016) 11–17.
gesting a typical diffusion dominated process. Normalizing [23] Qingang Xiong, J. Zhang, Fei Xu, G. Wiggins, C. Stuart Daw, Coupling DAEM and
devolatilisation time with respect to the temperature, thermal con- CFD for simulating biomass fast pyrolysis in fluidized beds, J. Anal. Appl. Pyrol.
ductivity, density and particle diameter using data from the litera- 117 (2016) 176–181.
[24] S. Dasappa, P.J. Paul, Gasification of char particles in packed beds: analysis and
ture and present numerical simulation is formulated and results, Int. J. Energy Res. 12 (25) (2001) 1053–1072.
presented. The capability of the model to predict complete burning [25] C. Di Blasi, Analysis of convection and secondary reaction effects within porous
of wood (pyrolysis and char reaction) has been established and solid fuels undergoing pyrolysis, Combust. Sci. Technol. 90 (1993) 315–340.
[26] M.J. Groeneveld, The co-current moving bed gasifier, Technical Report, Twente
compared with the literature reported values for various ambient
University of Technology, Netherlands, 1980.
conditions. [27] K.M. Bryden, K.W. Ragland, C.J. Rutland, Modeling thermally thick pyrolysis of
wood, Biomass Bioenergy 22 (2002) 41–53.
[28] R. Graham, M.A. Bergougnou, Fast pyrolysis of biomass, J. Anal. Appl. Pyrol. 6
References
(1984) 95–135.
[29] G. Enrico, G. Baldi, Analysis and modelling of wood pyrolysis, Chem. Eng. Sci.
[1] M. Francis, The greenhouse effect: a new source of energy, Appl. Therm. Eng. 66 (2011) 650–660.
27 (2007) 658–664. [30] W. Simpson, A. TenWolde, Chapter 3 - Physical Properties and Moisture
[2] B.V. Babu, A.S. Chaurasia, Pyrolysis of biomass: improved models for Relations of Wood, U.S. Department of Agriculture, Wood Hand book- Wood as
simultaneous kinetics and transport of heat, mass and momentum, Energy an Engineering Material, 1999, pp. 3.1–3.24.
Convers. Manage. 45 (2004) 1297–1327. [31] K.W. Ragland, D.J. Aerts, A.J. Baker, Properties of wood for combustion analysis,
[3] L.J. Curtis, D.J. Miller, Transport model with radiative heat transfer for rapid Bioresour. Technol. 37 (1991) 161–168.
cellulose pyrolysis, Ind. Eng. Chem. Res 27 (1988) 1775–1783. [32] S. Varunkumar, N.K.S. Rajan, H.S. Mukunda, Single particle and packed bed
[4] K.M. Bryden, M.J. Hagge, Modeling the combined impact of moisture and char combustion in modern gasifier stoves -density effects, Combust. Sci. Technol.
shrinkage on the pyrolysis of a biomass particle, Fuel 82 (2003) 1633–1644. 183 (11) (2011) 1147–1163.
[5] H.S. Mukunda, P.J. Paul, U. Shrinivasa, N.K.S. Rajan, Combustion of wooden [33] V.B. Silva, A. Rouboa, Using a two-stage equilibrium model to simulate oxygen
spheres – experiments and model analysis, Proc. Combust. Symp. (1984) air enriched gasification of pine biomass residues, Fuel Process. Technol. 109
1619–1628. (2013) 111–117.
[6] A. Sharma, V. Pareek, S. Wang, Z. Zhang, H. Yang, D. Zhang, A phenomenological [34] P. Gnanendra, D. Ramesha, S. Dasappa, Preliminary investigation on the use of
model of the mechanisms of lignocellulosic biomass pyrolysis process, biogas sludge, Int. J. Sustain. Energy 4 (31) (2012) 251–267.
Comput. Chem. Eng. 60 (2014) 231–241. [35] S. Dasappa, P.J. Paul, H.S. Mukunda, U. Shrinivasa, The gasification of wood-
[7] W.C. Park, A. Atreya, H.R. Baum, Experimental and theoretical investigation of char spheres in CO2-N2 mixtures: analysis and experiments, Chem. Eng. Sci. 2
heat and mass transfer processes during wood pyrolysis, Combust. Flame 157 (49) (1994) 223–232.
(2010) 481–494.