Pyrolysis of Biomass - Improved Models For Simultaneous

Energy Conversion and Management 45 (2004) 1297–1327
www.elsevier.com/locate/enconman
Pyrolysis of biomass: improved models for simultaneous

kinetics and transport of heat, mass
and momentum
B.V. Babu *, A.S. Chaurasia
Engineering Technology Group, Department of Chemical Engineering, Engineering Services Division,
Birla Institute of Technology and Science, Pilani, Rajasthan 333031, India
Received 12 June 2003; accepted 12 September 2003
Abstract
Understanding the physical phenomena of pyrolysis and representing them with an appropriate math-
ematical model is essential in the design of pyrolysis reactors and biomass gasifiers. Description of the
chemical processes of pyrolysis is coupled to an unsteady state, one dimensional, variable property model of
transport phenomena, including heat convection, conduction and radiation, volatiles and gas transport by
diffusion and convection and momentum transfer. In this study, a generalized reference model (Model I)
incorporating all the above effects is proposed. This is further improved by proposing two simplified models
(Models II and III) incorporating additional assumptions. A finite difference pure implicit scheme utilizing
a Tri-Diagonal Matrix Algorithm (TDMA) is employed for solving the heat transfer and mass transfer
model equations. A Runge–Kutta fourth order method is used for the chemical kinetics model equations.
Simulations are performed considering different geometries of equivalent radius ranging from 0.0001 to
0.017 m and temperatures ranging from 303 to 2800 K. The results obtained using these improved models
are in excellent agreement with the experimental data, much better than the agreement with earlier models
reported in the literature. The improved validated model, which is best suited for wide ranges of operating
conditions, is utilized to investigate the influence of particle size, particle shape, product distribution,
conversion time and heat of reaction.
2003 Elsevier Ltd. All rights reserved.
Keywords: Pyrolysis; Biomass; Kinetics; Heat transfer; Mass transfer; Momentum transfer; Simulation; Heat of
reaction; Modeling
*
Corresponding author. Tel.: +91-1596-245073x205/224; fax: +91-1596-244183.
E-mail address: bvbabu@bits-pilani.ac.in (B.V. Babu).
URL: http://bvbabu.50megs.com.
0196-8904/$ - see front matter 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.enconman.2003.09.013
1298 B.V. Babu, A.S. Chaurasia / Energy Conversion and Management 45 (2004) 1297–1327
Nomenclature
A1 ; A2 ; A3 frequency factor, 1/s

b geometry factor (slab ¼ 1, cylinder ¼ 2, sphere ¼ 3)
B virgin biomass
G1 (gases and volatiles)1
C1 (char)1
G2 (gases and volatiles)2
C2 (char)2
CB concentration of B, CB0 at initial condition, kg/m3
CG1 concentration of G1 , CG1 0 at initial condition, kg/m3
CC1 concentration of C1 , CC1 0 at initial condition, kg/m3
CG2 concentration of G2 , CG2 0 at initial condition, kg/m3
CC2 concentration of C2 , CC2 0 at initial condition, kg/m3
Cp specific heat capacity, J/kg K
CpG1 heat capacity of (gases and volatiles)1 , CpG1 0 at initial condition, J/mol K
D1 ; D2 constants defined by expressions of k1 and k2 , respectively, K
DeG1 effective diffusivity of (gases and volatiles)1 , DeG1 0 for initial effective diffusivity, m2 /s
E3 activation energy defined by expression of k3 , J/mol
h convective heat transfer coefficient, W/m2 K
k thermal conductivity, W/m K
kmG1 mass transfer coefficient of (gases and volatiles)1 across film, m/s
k1 ; k2 ; k3 rate constants, 1/s
l axial length of cylinder, m
L1 ; L2 constants defined by expressions of k1 and k2 , respectively, K2
M total number of equations used in simulation of model
n1 ; n2 ; n3 orders of reactions
p gas pressure, N/m2
P1 ; P2 variation constants of parameters defined in Table 3
Q defined by Eq. (35), m3 /kg
r radial distance, m
R radius for cylinder and sphere; half thickness for slab, m
Rc universal gas constant, J/mol
t time, s
T temperature, K
u gas velocity, m/s
Wi molecular weight of species i, kg/mol
x dimensionless radial distance
X conversion of biomass
Greek letters
DH heat of reaction, J/kg
Ds axial grid length
B.V. Babu, A.S. Chaurasia / Energy Conversion and Management 45 (2004) 1297–1327 1299
Dx radial grid distance

q density, q0 at initial condition, kg/m3
a thermal diffusivity, m2 /s
s dimensionless time
c constant defined by Eq. (26)
h normalized temperature
e emissivity coefficient
e00 void fraction of particle as defined by Eq. (26), e000 at initial condition
r Stefan Boltzmann constant, W/m2 K4
/ permeability, m2
g reaction progress variable
l viscosity, kg/m s
Dimensionless numbers
BiM Modified Biot number
Le Lewis number
Pr Prandtl number
Q00 Heat of reaction number
Re Reynolds number
Sh Sherwood number
Subscripts
g gas
0 initial
f final
V water vapor
L light hydrocarbons
B wood
m mean
1. Introduction
Biomass, being a readily available renewable energy source that reduces sulfur dioxide and
carbon dioxide emissions, is an attractive option as a fuel for power generation. The need for
biomass fuels is driven by the need to reduce carbon dioxide emissions and the need to eliminate
what has previously been considered a waste product in agricultural applications. An under-
standing of the chemical processes and transport mechanisms of biomass pyrolysis is important
for many applications, including optimization of boilers and large scale furnaces, determining
forest fire behavior and predicting the resistance of buildings to fire. The pyrolysis of biomass is
considered as one of the viable means for overcoming the so-called ‘‘energy crisis’’. It is a
renewable source of energy and has many advantages from an ecological point of view.
The pyrolysis process consists of the thermal degradation of biomass feedstock in the absence
of oxygen/air, leading to the formation of solid (charcoal), liquid (tar and other organics) and
gaseous products. Pyrolysis can be used as an independent process for the production of useful
energy (fuels) and/or chemicals. It also occurs as the first step in a gasification or combustion
process. Knowledge of the kinetics of pyrolysis of a solid particle is very significant in designing a
fluid–solid gasifier. In spite of these many advantages, little work has been reported in the liter-
ature on the mechanistic modeling of such kinetics. This may be attributed to the extreme
complexity of the pyrolysis process due to the presence of homogeneous and heterogeneous
reactions occurring consecutively along with heat, mass and momentum transfer operations
occurring simultaneously during the process.
2. Background
A general mathematical model of the pyrolysis process is not only difficult to formulate but also
tedious to solve. Fan et al. [1] developed a model for the pyrolysis process, which includes heat and
mass transfer effects in the particle. The reaction is considered to be first order with respect to the
initial particle concentration. Also, the product concentrations cannot be analyzed from the above
model, as the secondary reactions were not considered. Miyanami et al. [2] incorporated the effect
of the heat of reaction in the above model. The various modeling studies reported in the literature
till now present different versions or enrichments of the model suggested originally by Bamford
et al. [3]. According to this model, the equation for heat conduction in a pyrolyzing solid is
combined with those for heat generation assuming first order kinetics. However, the heat transfer
equation does not consider the effect of the change in density as a function of time. This model was
used by several researchers [4–7] and was modified by Kung [8] in order to incorporate the effects of
internal convection and variable transport properties. However, no specific kinetic mechanism was
suggested to predict the concentration of the various components produced during the pyrolysis.
Kansa et al. [9] included the momentum equation for the motion of the pyrolysis gases within the
solid, but a suitable kinetic mechanism has not been utilized by them, and the solution to the heat
and momentum balance equations is based on arbitrary boundary conditions.
Jalan and Srivastava [10] have solved the heat transfer equation by neglecting the effect of the
specific heat capacity and thermal conductivity of the char, which are functions of temperature as
reported by Koufopanos et al. [11]. The convective heat transfer coefficient is a function of
Reynolds number and Prandtl number as reported by Koufopanos et al. [11], but Jalan and
Srivastava [10] have considered a constant value of h ¼ 0:322 W/m2 K, neglecting the effect of
other parameters. Also, mass transfer effects are not incorporated in the model developed by Jalan
and Srivastava [10]. Keeping the above drawbacks of the existing models in view, Babu and
Chaurasia [12] developed a model for pyrolysis of a single biomass particle with the kinetic scheme
used by them in their earlier study [13–16]. Studies have been conducted on pyrolysis of biomass
and other substances by several researchers [17–26], but many of them neglected convective and
diffusive transport of the volatiles and gases.
Di Blasi [27–29] pointed out that a detailed transport model incorporating kinetics and heat and
mass transfer effects is necessary to predict the effects of the widely variable physical properties
(density, thermal conductivity, specific heat capacity) in the pyrolysis of biomass. However, she
used a different kinetic scheme wherein active cellulose is considered to be formed as an inter-
mediate product. However, as pointed out by Koufopanos et al. [11], it is very difficult to define
physically the components or the composition of the intermediates, and consequently, it is not
possible to measure its concentration experimentally in order to test the model rigorously. So, in
the present study, the model developed by Babu and Chaurasia [12], using a practically significant
kinetic scheme consisting of physically measurable parameters, is modified by incorporating the
convective and diffusion effects. The properties affecting the pyrolysis, such as porosity and per-
meability, are also incorporated in the model. In addition, DarcyÕs law and the ideal gas law are
also considered. In order to understand the role played by the assumptions usually made in the
mathematical description of large sample pyrolysis, models of different complexity are compared.
Pyle and Zaror [30] utilized a kinetic model assuming first order kinetics, based on the density
of the initial biomass. The effect of the change in density as a function of time of the biomass
particle is not considered by Koufopanos et al. [11] and Pyle and Zaror [30] while solving the heat
transfer model equation.
In this paper, a generalized and detailed mathematical model for pyrolysis of biomass has been
developed. Subsequently, two simplified models are obtained from the model by making related
simplifying assumptions, applicable for practical situations. These models are validated with
experimental data reported in literature. The model being developed considered the effect of
specific heat capacity and thermal conductivity of char as a function of temperature. A mean
value of the convective heat transfer coefficient, taken from the data reported in the literature
[28,30], is used throughout for the low and high values of final temperature, respectively, in which
the functional dependency on Reynolds number and Prandtl number is incorporated. The model
would enable the prediction of product yields of both char and volatiles and gases and can be used
for different particle geometries (slab, cylinder and sphere).
3. Description of mathematical model
The overall process of pyrolysis can be classified into primary and secondary stages. The
phenomena occurring in pyrolysis when a solid particle of biomass is heated in an inert atmo-
sphere is explained by the following mechanisms. Heat is first transferred to the particle surface by
radiation and/or convection and then to the inside of the particle. The temperature inside the
particle increases, causing (1) removal of moisture that is present in the biomass particle and (2)
the occurrence of pre-pyrolysis and main pyrolysis reactions, as discussed by Babu and Chaurasia
[13]. The heat changes due to the chemical reactions and phase changes contribute to a temper-
ature gradient as a function of time, which is nonlinear. Volatiles and gaseous products flow
through the pores of the particle and participate in the heat transfer process. The pyrolysis
reactions proceed with a rate depending upon the local temperature.
The pyrolysis reactions can be described by means of the scheme proposed by Koufopanos
et al. [11] and subsequently used by Babu and Chaurasia [13], as shown in Table 1. This scheme
indicates that the biomass decomposes to volatiles, gases and char. The volatiles and gases may
further react with char to produce different types of volatiles, gases and char where the compo-
sitions are different. Therefore, the primary pyrolysis products participate in secondary interac-
tions (Reaction 3), resulting in a modified final product distribution. Using this scheme, three
models are proposed in the present study under two categories namely, (1) generalized reference
model (Model I), and (2) simplified models (Models II and III).
Table 1
Mathematical model
Koufopanos et al. mechanism [11]
Particle model
Mass conservation for biomass, (gases and volatiles)1 (char)2 , (gases and volatiles)2 and (char)2 :
oCB
¼ k1 CBn1 k2 CBn1 ð1Þ
ot

oðCG1 e00 Þ oðCG1 uÞ b 1 oCG1 o2 CG1
þ ¼ DeG1 þ þ k1 CBn1 e00 k3 CGn21 CCn31 ð2Þ
ot or r or or2
oCC1
¼ k2 CBn1 k3 CGn21 CCn31 ð3Þ
ot
oCG2
¼ k3 CGn21 CCn31 ð4Þ
ot
oCC2
¼ k3 CGn21 CCn31 ð5Þ
ot
Enthalpy:

o b 1 oT o2 T oCG1 oT oq
ðCp qT Þ ¼ k þ 2 DeG1 CpG1 þ ðDH Þ ð6Þ
ot r or or or or ot
Initial conditions:
t ¼ 0; CB ¼ CB0 ; CG1 ¼ CC1 ¼ CG2 ¼ CC2 ¼ 0; T ðr; 0Þ ¼ T0 ð7Þ
Particle boundary conditions:

oCG1 oT
t > 0; r ¼ 0; ¼ 0; ¼0 ð8Þ–ð9Þ
or or r¼0

oCG1
t > 0; r ¼ R; DeG1 ¼ kmG1 ðCG1 0 CG1 Þ ð10Þ
or

oT
t > 0; r ¼ R; k ¼ hðTf T Þ þ reðTf4 T 4 Þ ð11Þ
or r¼R
Table 1 (continued)
Dimensionless forms of Eqs. (1)–(11):
oC B n n
¼ k1 C B1 k2 C B1 ð12Þ
ot
n n2 n3
oC G1 uR oC G1 DG1 b 1 oC G1 o2 C G1 R2 k1 C B1 e00 R2 k3 C G1 C C1
e00 þ ¼ þ þ ð13Þ
os a ox Le x ox ox2 a a
oC C1 n1 n2 n3
¼ k 2 C B k 3 C G1 C C 1 ð14Þ
ot
oC G2 n2 n3
¼ k3 C G1 C C1 ð15Þ
ot
oC C2 n2 n3
¼ k3 C G1 C C1 ð16Þ
ot

oh b 1 oh o2 h Q00 R2 k1 1 oC G1 oh
¼ þ 2þ þ DG1 C pG1 C B0 ð17Þ
os x ox ox a Le ox ox
s ¼ 0; C B ¼ 1; C G1 ¼ C C1 ¼ C G2 ¼ C C2 ¼ 0; hðx; 0Þ ¼ 1 ð18Þ
oC G1 oh
s > 0; x ¼ 0; ¼ 0; ¼0 ð19Þ–ð20Þ
ox ox

oC G1
s > 0; x ¼ 1; D G1 ¼ ShðC G1 0 C G1 Þ ð21Þ
ox
oh
s > 0; x ¼ 1; ¼ hBiM ð22Þ
ox
Koufopanos et al. correlation [11]:
h ¼ 0:322ðk=lÞPr1=3 Re0:5 ð23Þ
Darcy law and state equation:
/ op
u¼ ð24Þ
l ox
p ¼ CG1 Rc T =Wm ð25Þ

(continued on next page)
Table 1 (continued)
Other relations:
e00 ¼ e000 þ cð1 C B Þ; / ¼ g/B þ ð1 gÞ/C1 ; g ¼ CB =CB0 ð26Þ–ð28Þ
Conversion of biomass:
PM
C B0 i¼1 CB ðM þ 1Þ
X ¼ ð29Þ
C B0
3.1. Generalized reference model (Model I)
The generalized model incorporates all the possible effects of kinetics, heat transfer, mass
transfer and momentum transfer. The assumptions made in developing this model are as follows:
(1) The thermal and transport properties (porosity, thermal conductivity, specific heat, mass dif-
fusivity) vary with the conversion level.
(2) Heat transfer occurs by all three modes (i.e. conduction, convection and radiation).
(3) Gas phase processes occur under unsteady state conditions.
(4) Transport of mass takes place by convection and diffusion of volatile species.
(5) Pressure and velocity vary along the porous sample.
(6) Local thermal equilibrium exists between the solid matrix and the flowing gases.
(7) The system is one dimensional.
(8) There is no moisture content and no particle shrinkage.
Utilizing the kinetic scheme as described by Koufopanos et al. [11] and with the assumptions
stated above, the generalized model (Model I) is obtained and reported in Table 1. The equations
shown in Table 1 are written in dimensionless form with the help of the dimensionless groups
given in Table 2.
This generalized model, consisting of Eqs. (12)–(28), which is the most comprehensive one, is
named Model I (generalized reference model). However, under specific conditions, some of the
assumptions made in this generalized model are to be relaxed. Taking practical situations in the
pyrolysis of biomass into account, two simplified models are proposed, introducing specific
assumptions for each of them. The simulations obtained through this generalized model are used
as reference and compared with the simulations obtained using the simplified models. Simplifi-
cations are introduced only in the description of the physical processes.
3.2. Simplified models
These simplified models include additional assumptions other than those in the generalized
reference model (Model I). Many times, in practical situations, generalized models may not give
good predictions. In such cases, there is a need to relax some of the assumptions made in the
generalized models. Starting from the generalized model, the following two simplified models are
proposed in the present study for specific cases.
Table 2
Dimensionless groups
a ¼ k=qCp ð30Þ
x ¼ r=R ð31Þ
s ¼ at=R2 ð32Þ
h ¼ ðT Tf Þ=ðT0 Tf Þ ð33Þ
BiM ¼ ðR=kÞbh þ erðT 3 þ T 2 Tf þ Tf2 T þ Tf3 Þc ð34Þ
Q ¼ ðDH þ Cp T Þ=ðqCp ðT0 Tf ÞÞ ð35Þ
Q00 ¼ QCBn1 ð36Þ
C B ¼ CB =CB0 ð37Þ
C B0 ¼ CB0 =q0 ð38Þ
C G1 ¼ CG1 =CB0 ð39Þ
C G1 0 ¼ CG1 0 =CB0 ð40Þ
C C1 ¼ CC1 =CB0 ð41Þ
C G2 ¼ CG2 =CB0 ð42Þ
C C2 ¼ CC2 =CB0 ð43Þ
C pG1 ¼ CpG1 =CpG1 0 ð44Þ
DeG1 ¼ DeG1 =DeG1 0 ð45Þ
Le ¼ ðkÞ=ðq0 CpG1 0 DeG1 0 Þ ð46Þ
Sh ¼ ðkmG1 RÞ=DeG1 0 ð47Þ
3.2.1. First simplified model (Model II)

In most practical situations of industrial pyrolysis reactions, the contributions of the bulk
motion of gases inside the pores of the particle are insignificant. Because the resistance offered by
the pores in the solid particle is so high, the transport of these gases would take place essentially
by a diffusion mechanism but not by bulk motion (i.e. convection). Taking this situation into
consideration, the first simplified model (Model II) is proposed by making the additional
assumption that there is no bulk motion contribution (i.e. convective transport is neglected) for
the temperature profile and product yield predictions. In this treatment, the conservation equation
for the mass concentration of (gases and volatiles)1 (Eq. (13)) is modified by neglecting the second
term on the left hand side. In effect, the first simplified model (Model II) consists of Eqs. (12)–(23)
and (26).
3.2.2. Second simplified model (Model III)

Analogous to the above momentum transport, the corresponding convective part in the heat
transport of the gases within the particle also does not contribute significantly for the reasons
explained above. As the temperatures are also not too high, so as to have radiation effects, even the
radiative heat transfer of gases within the particles could also be neglected. In addition, as dense
particles and various kinds of hard wood are generally used in the pyrolysis of biomass gasification,
the effect of the porosity of the solid pellet is insignificant. Hence, the second simplified model
(Model III) is proposed based on the following two assumptions concerning practical applications:
(1) the basic mode of transfer inside the solid particle in the process of pyrolysis takes place by
conduction heat transfer only and (2) the effect of the porosity of the solid particle is negligible.
Based on the above assumptions, the conservation equation for the mass concentration of (gases
and volatiles)1 (Eq. (13)) and heat transfer model (Eq. (17)), respectively, become
oC G1 n n n
¼ k1 C B1 k3 C G21 C C31 ð48Þ
ot
oh b 1 oh o2 h Q00 R2 k1
¼ þ þ ð49Þ
os x ox ox2 a
So the second simplified model (Model III) consists of Eqs. (12), (14)–(16), (18), (20), (22), (23),
(48) and (49). Interestingly, this is similar to the model proposed by the authors in their earlier
study [12], which means that the generalized reference model reduced to the model proposed by
Babu and Chaurasia [12] under specific conditions.
4. Numerical solution and simulation
Eqs. (13) and (17) along with the initial and boundary conditions given by Eqs. (18)–(22), are
solved numerically by a finite difference method using a pure implicit scheme. The pure implicit
scheme is an unconditionally stable scheme, i.e. there is no restriction on time step in sharp
contrast with the Euler and Crank-Nicholson method discussed by Ghoshdastidar [31].
Eqs. (12)–(17) are solved simultaneously. Eqs. (12) and (14)–(16) are solved by a Runge–Kutta
fourth order method with both fixed step size and variable step size. It is found that the Runge–
Kutta fourth order variable step size method (RKVS) is faster than the Runge–Kutta fourth order
method with fixed step size (RKFS), as discussed by Babu et al. [32]. However, the RKVS method
does not give the solution for a particular and fixed interval of time. Hence, to obtain the solution
at a particular interval of time, which is required to compare the results of the present study with
those reported in the literature, the RKFS method is used in the present study. The discretized
form of Eqs. (13) and (17) are solved by the Tri-Diagonal Matrix Algorithm (TDMA), also
known as the Thomas Algorithm [33]. Simulations are performed considering different particle
geometries (slab, cylinder and sphere) of equivalent radius ranging from 0.0001 to 0.017 m and
temperatures ranging from 303 to 2800 K.
The values of the various parameters employed in the present study are listed in Tables 3 and 4.
It may be noted that the thermal diffusivity is taken to be constant and based on the initial
temperature of the wood, which is a representative value in pyrolysis. When separate values for
wood at different temperatures were tried in the simulation, it was found that for the same iter-
ation, the pyrolysis time was different for different radial positions. The pyrolysis time cannot be
different in a given iteration, and hence, a representative constant value corresponding to the
initial condition of the biomass is used (taken from Jalan and Srivastava [10]) and shown in
Table 3.
Table 3
Values of parameters used in the numerical solution of the model
Property Value Source
2
Convective heat transfer coefficient h ¼ 8:4 W/m K for T 6 673 K [30]
h ¼ 20:0 W/m2 K for T > 673 K [28]
Wood specific heat capacity Cp ¼ 112:0 þ 4:85ðT 273Þ (J/kg K) [11]
Char specific heat capacity Cp ¼ 1003:2 þ 2:09ðT 273Þ (J/kg K) [11]
Wood thermal conductivity k ¼ 0:13 þ 0:0003ðT 273Þ (W/m K) [11]
Char thermal conductivity k ¼ 0:08 0:0001ðT 273Þ (W/m K) [11]
Rate constant of reaction 1 k1 ¼ A1 exp½ðD1 =T Þ þ ðL1 =T 2 Þ (s1 ); where [11]
A1 ¼ 9:973 105 s1 ; D1 ¼ 17,254.4 K;
L1 ¼ )9,061,227 K2
Rate constant of reaction 2 k2 ¼ A2 exp½ðD2 =T Þ þ ðL2 =T 2 Þ (s1 ); where [11]
A2 ¼ 1:068 103 s1 ; D2 ¼ 10,224.4 K;
L2 ¼ )6,123,081 K2
Rate constant of reaction 3 k3 ¼ A3 exp½ðE3 =Rc T Þ (s1 ); where A3 ¼ 5:7 105 [11]
s1 ; E3 ¼ 81,000 J/mol
Heat capacity of (gases and volatiles)1 C pG1 ¼ 1 þ P1 ðh 1Þ; P1 ¼ 0:001 [1]
Effective diffusivity of DG1 ¼ ðhÞP1 expfP2 ð1 C B Þg; P1 ¼ 1:5; P2 ¼ 1:0 [1]
(gases and volatiles)1
Sherwood number Sh ¼ P1 ðhÞP2 ; P1 ¼ 50,000, P2 ¼ 0:75 [1]
Heat of reaction DH ¼ )255,000 J/kg [11]
Initial density of wood q0 ¼ 650 kg/m3 [11]
Initial thermal diffusivity of wood a ¼ 1:79 107 m2 /s [10]
Lewis number Le ¼ 2:0 [1]
Initial void fraction of particle e000 ¼ 0:5 [1]
Constant defined by equation of e00 c ¼ 0:3 [1]
Molecular weight of water vapor WV ¼ 18 103 kg/mol [34]
Molecular weight of light hydrocarbons WL ¼ 28 103 kg/mol [34]
Molecular weight of wood WB ¼ 110 103 kg/mol [34]
Viscosity l ¼ 3 105 kg/m s [35]
Permeability of wood /B ¼ 1 1014 m2 [35]
Permeability of char /C1 ¼ 5 1011 m2 [35]
Table 4
Nominal values of parameters employed in the present study
Parameters Values
Geometry factor b¼2
Initial concentration of B C B0 ¼ 1:0
Initial concentration of G1 C G1 0 ¼ 0
Initial concentration of C1 C C1 0 ¼ 0
Initial concentration of G2 C G2 0 ¼ 0
Initial concentration of C2 C C2 0 ¼ 0
Order of reaction 1 n1 ¼ 0:0–1.0
Order of reaction 2 n2 ¼ 1:5
Order of reaction 3 n3 ¼ 1:5
Particle radius range R ¼ (0.0001–0.017) m
Initial temperature T0 ¼ 303 K
Final temperature range Tf ¼ (303–2800) K
Emissivity coefficient e ¼ 0:95
Stefan Boltzmann constant r ¼ 5:67 108 W/m2 K4
5. Results and discussion
5.1. Models validation and comparison
The simultaneous kinetic, mass, heat and momentum transfer model developed in the present
study is compared with the experimental data reported by Pyle and Zaror [30], the Bamford,
Crank and Malan model [3] used by Pyle and Zaror [30] and the model developed by Jalan and
Srivastava [10]. Pyle and Zaror [30] conducted experimental studies on the course of pyrolysis of
cylindrical samples of wood. They performed the experiments in an inert constant temperature
environment and measured the instantaneous sample weight and radial temperature profile.
The major constituents of wood are cellulose, hemi-cellulose and lignin. There is some variation
in the relative abundance of these constituents in different species of wood, but as an approximate
guideline, cellulose is taken to be 50% and the other two constituents (hemi-cellulose and lignin) to
be 25% each by dry weight. Shafizadeh [23] studied the pyrolysis of cellulose at various temper-
atures. At temperatures less than or equal to 673 K, the dominant process is the reduction in the
degree of polymerization, and diffusion and convection effects are negligible. In the second step, at
temperatures above 673 K, there is formation of char, tar and gaseous products, and diffusion and
convection effects are predominant. So, a mean value of 8.4 W/m2 K [30] (in which the dependence
on Reynolds number and Prandtl number is incorporated and estimated employing Eq. (23)) was
used throughout for the convective heat transfer coefficient at the final temperature less than or
equal to 673 K. Similarly, a mean value of 20.0 W/m2 K [28] (again employing Eq. (23)) was used
throughout for the convective heat transfer coefficient at the final temperature greater than 673 K.
It may be noted that all the parameters used to test the present models are either both inde-
pendently and directly estimated or are taken from the literature.
Fig. 1 shows the temperature profile as a function of time at the centre (i.e. x ¼ 0) of the
cylindrical pellet of radius 0.003 m. This is compared with the profiles obtained by Jalan and
Fig. 1. Temperature profile as a function of time at the centre of the cylindrical pellet (R ¼ 0:003 m, T0 ¼ 303 K,
Tf ¼ 643 K, x ¼ 0).
Srivastava [10] and the experimental data obtained by Pyle and Zaror [30] at the centre of the
cylindrical pellet. It is found that all the models developed in the present study (Models I, II and
III) are in excellent agreement with the experimental data, better than the agreement with the
Jalan and Srivastava model [10].
The temperature profile as a function of time for the cylindrical pellet of radius 0.0075 m is
shown in Fig. 2. Here also, it is found that the models developed in the present study are in good
agreement with the experimental data, better than the agreement with the Jalan and Srivastava
model [10]. It may be noted that there is a deviation from the experimental data and Model I
predictions at higher values of the pyrolysis time. In fact, Model I underpredicts the experi-
mental data. This is due to the fact that as the pyrolysis proceeds, the cellulose polymers begin to
decompose rapidly, suggesting that secondary pyrolysis reactions are important [10]. As the
pyrolysis progresses, the velocity of the gases that are formed will increase due to the consid-
eration of both convection and diffusion effects (Model I). It escapes the cellulose matrix faster,
and hence, the residence time is reduced. So, secondary reactions of the volatiles produced by
the primary reactions cannot take place and, hence, the underprediction of particle temperature
at higher values of pyrolysis time. However, in Model II, it is assumed that there is no bulk
motion contribution, and in Model III, it is assumed that heat transfer takes place by conduc-
tion only, both of which lead to the prediction of an increase in temperature of the particle
in the simulated results at higher values of pyrolysis time. In fact, the volatiles and gases that are
formed by the primary reaction get sufficient time to undergo the secondary pyrolysis reaction,
resulting in the formation of volatiles and gases of different composition, and this leads to an
increase in temperature at the higher values of pyrolysis time. This is the reason why Models II
and III predict the results well even at high pyrolysis times, matching with the experimental
data.
Fig. 2. Temperature profile as a function of time at the centre of the cylindrical pellet (R ¼ 0:0075 m, T0 ¼ 303 K,
Tf ¼ 773 K, x ¼ 0).
The simulated results used in Figs. 1 and 2 are given in Tables 5 and 6, respectively, for
quantitative comparison of the present models with the Jalan and Srivastava model [10] and
experimental data of Pyle and Zaror [25]. The average percentage error and standard deviation
are also calculated. If Yexp is the experimental value of a quantity and Ycal is the value obtained
from the modeling equation, then the average percentage error Epa is defined as follows:
PN
½ðYexp Ycal Þ=Yexp 100
Epa ¼ i¼1 ð50Þ
N
The standard deviation (SD) is calculated by the following formula,
"P #
N 2 1=2
i¼1 ½ðYexp Ycal Þ=Yexp
SD ¼ ð51Þ
N 1
In all the cases, it is found that the average percentage error and standard deviation from
experimental data are significantly less in the prediction of the present models as compared to the
predictions of the Jalan and Srivastava model [10]. It may be noted from Table 6 that though the
SD for Model I ( ¼ 0.0392) is slightly higher than that for Models II and III ( ¼ 0.0186 and 0.0194,
respectively), it is still much less when compared to SD ( ¼ 0.0511) with the Jalan and Srivastava
model [10].
Fig. 3 shows the temperature profile as a function of dimensionless radial distance for the final
temperature of 643 K at the pyrolysis time of 4 min for the particle radius of 0.011 m. The
temperature profiles obtained in the present study are in much better agreement with the
experimental data [30] when compared to the other models [3,10]. The temperature profile at
t ¼ 11 min for the same particle radius of 0.011 m is shown in Fig. 4. At this high value of
pyrolysis time, it is found that the first simplified model (Model II) predictions are in excellent
agreement with the experimental data [30] as compared to the predictions with other models for
Table 5
Comparison of present models results with those of earlier models for various stages of pyrolysis (time) at the centre of
cylindrical pellet (R ¼ 0:003 m, T0 ¼ 303 K, Tf ¼ 643 K)
Time (s) Temperature (K)
P Model I Model II Model III J
0 303 303 303 303 303
20 397 401 401 400 387
40 493 495 495 493 478
60 541 553 554 552 533
80 581 589 590 588 574
100 609 609 612 610 602
150 641 633 636 634 630
200 648 640 642 640 639
Epa 0.93 0.93 0.75 1.56
SD 0.0125 0.0125 0.0106 0.0189
P ¼ Pyle and Zaror [30] (experimental); J ¼ Jalan and Srivastava model [10].
Table 6
Comparison of present models results with those of earlier models for various stages of pyrolysis (time) at the centre of
cylindrical pellet (R ¼ 0:0075 m, T0 ¼ 303 K, Tf ¼ 773 K)
Time (s) Temperature (K)
P Model I Model II Model III J
0 303 303 303 303 303
15 312 304 304 305 304
30 330 325 325 327 323
45 377 368 368 369 359
70 454 447 449 448 430
100 545 521 534 531 503
120 572 556 580 575 553
140 629 579 619 612 584
Epa 2.89 1.56 1.66 4.06
SD 0.0392 0.0186 0.0194 0.0511
P ¼ Pyle and Zaror [30] (experimental); J ¼ Jalan and Srivastava model [10].
the reasons mentioned above. The simulated results at the different radial positions for Figs. 3 and
4 are shown in Tables 7 and 8, respectively, for quantitative comparison of the present model with
the earlier models and the literature data, which substantiates the better predictions using the
present models. However, it may be noted both from Fig. 4 and Table 8 that Model I under-
predicts the experimental data for the reasons explained earlier, as this corresponds to a condition
of higher value of pyrolysis time. Fig. 5 shows the temperature profiles when the temperature is
increased from 643 K to 753 K as a function of radial distance for a time period of 2 min. In this
case also, the temperature profiles obtained in the present models are in better agreement with the
experimental data [30], which is also substantiated by the Epa and SD values in Table 9. Fig. 6
shows the conversion profile as a function of time with a cylindrical pellet of radius 0.011 m and
Fig. 3. Temperature profile as a function of radial distance (R ¼ 0:011 m, T0 ¼ 303 K, Tf ¼ 643 K, t ¼ 4 min).
final temperature of 753 K. As can be seen from the figure, the models developed herein are in
better agreement with the experimental data for conversion profile also. The average percentage
error and standard deviation from experimental data are less in the simplified models (Models II
and III), as seen from Tables 5–9 also. Thus, the experimental results of Pyle and Zaror [30] are in
accordance with the predictions of the theory based on the data presented here. Though all three
models proposed in the present study performed well under varied conditions, it is observed that
the second simplified model (Model III) is in excellent agreement with the experimental data as
compared to other models in all the situations.
Table 7
Comparison of present models results with those of earlier models for pyrolysis time ¼ 4 min (R ¼ 0:011 m, T0 ¼ 303 K,
Tf ¼ 643 K)
x Temperature (K)
P Model I Model II Model III J B
0 483 482 482 482 469 473
0.3 493 492 493 493 480 485
0.5 510 510 512 512 499 495
0.7 525 535 538 537 524 520
0.9 580 564 569 568 556 550
Epa 1.01 0.94 0.94 2.40 2.55
SD 0.0168 0.0156 0.0156 0.0305 0.0329
P ¼ Pyle and Zaror [30] (experimental); J ¼ Jalan and Srivastava model [10]; B ¼ Bamford, Crank and Malan model [3].
Table 8
Comparison of present models results with those of earlier models for pyrolysis time ¼ 11 min (R ¼ 0:011 m,
T0 ¼ 303 K, Tf ¼ 643 K)
x Temperature (K)
0 643 591 640 619 612 605
0.3 643 590 641 620 614 605
0.5 643 589 641 623 617 605
0.7 643 587 642 627 622 600
0.9 643 584 643 632 620 595
Epa 8.52 0.247 2.92 4.04 6.38
SD 0.0953 0.0033 0.0337 0.0457 0.0716
5.2. Simulation results
The improved validated model, which is best suited for a wide range of operating conditions, is
utilized to investigate the influences of particle size and particle shape on the time to reach 100 %
conversion and the product distribution. The effect of temperature on particle size and conversion
time is studied by considering wide ranges of particle radius and temperatures. As the pyrolysis
reactions are exothermic and/or endothermic in the different operating ranges of temperatures, the
understanding of the effect of heat of reaction in the reactor is equally important. So, extensive
simulations are also performed to examine the effect of heat of reaction number on the biomass
conversion, concentration and temperature profiles considering both exothermic and endothermic
reactions.
5.2.1. Effect of particle size and shape on conversion time and product yields
Fig. 7 shows the time to reach 100% conversion for the three different particle shapes when a
surface temperature of 643 K has been imposed. The spherical particle has the shortest conversion
time as can be expected based on its higher surface/volume ratio. Geometrically, the sphere has
Table 9
Comparison of present models results with those of earlier models for pyrolysis time ¼ 2 min (R ¼ 0:011 m, T0 ¼ 303 K,
Tf ¼ 753 K)
x Temperature (K)
0 420 414 417 408 404 400
0.3 440 437 442 432 426 420
0.5 490 476 486 474 464 450
0.7 530 530 549 533 522 510
0.9 620 591 623 607 594 580
Epa 1.93 1.59 2.12 3.60 5.54
SD 0.0285 0.0190 0.0259 0.0429 0.0644
more heat absorbing capacity as compared to the cylinder and slab, and it is the least for the slab.
Mathematically, this is reflected in the value of parameter b (b ¼ 1, 2 and 3 for slab, cylinder and
sphere, respectively) in the model Eqs. (13) and (17) and, hence, the observed trends. A significant
increase in conversion time is observed for the slab as compared to the cylinder and sphere.
Figs. 8–10 are concerned with the product distribution belonging to the final temperature of 643
K. They show the decrease in the yield of (volatiles and gases)1 with an increase in the yield of
(char)1 and (volatiles and gases)2 or (char)2 as the particle radius is increased. The yield of
(volatiles and gases)1 is maximum for the sphere and is least for the slab while the yield of (char)1
and (volatiles and gases)2 or (char)2 is more for the slab and is least for the sphere. This can be
explained by the decrease in average particle temperature at various radial positions as the particle
radius is increased. This decrease in temperature will be different for the various shapes as a result
of different heat penetrations and will be more for the slab and least for the sphere due to the
Fig. 6. Conversion profile as a function of time with cylindrical pellet (R ¼ 0:011 m, T0 ¼ 303 K, Tf ¼ 753 K).
Fig. 7. Time to reach 100% conversion as a function of particle radius for different geometries (T0 ¼ 303 K,
Tf ¼ 643 K).
higher surface/volume ratio as mentioned above. The decrease in temperature favors the char
yield as mentioned by Shafizadeh [23]. As the decrease in temperature is more for the slab, the
yield of (char)1 and (volatiles and gases)2 or (char)2 (Figs. 9 and 10) is more for the slab as
compared to the other geometries.
Fig. 8. Yield of (gases and volatiles)1 as a function of particle radius for different geometries (T0 ¼ 303 K, Tf ¼ 643 K).
Fig. 9. Yield of (char)1 as a function of particle radius for different geometries (T0 ¼ 303 K, Tf ¼ 643 K).
5.2.2. Effect of temperature on particle size and conversion time of pyrolysis

Fig. 11 shows the temperature profile as a function of radial distance for cylindrical pellets of
various radii at 900 K at the time of completion of pyrolysis for orders of reactions of n1 ¼ 1,
n2 ¼ n3 ¼ 1:5. It is found that as the particle radius increases, the time for completion of pyrolysis
ðtÞ also increases. The value of t for various particle sizes is shown above/below the temperature
profiles in Figs. 11, 12 and 16. The time required for completion of pyrolysis of the particle of
Fig. 10. Yield of (gases and volatiles)2 or (char)2 as a function of particle radius for different geometries (T0 ¼ 303 K,
Tf ¼ 643 K).
radius 0.00025 m is least (53 s), while it is maximum for the particle radius of 0.017 m (1128 s). For
smaller particle radii 0.00025 to 0.001 m, there is no change in temperature along the radial
position, while for the other particle radii, it is minimum at the centre and maximum at the wall.
Fig. 12 shows the temperature profile as a function of radial distance when the final temperature is
increased from 900 K to 1200 K. It is seen that as the temperature increases, the pyrolysis
completes faster for all the particle radii. The trends obtained are the same as those of Fig. 11 and
can be explained similarly.
Fig. 11. Temperature profile as a function of radial distance for various particle radii at 900 K at the time of completion
of pyrolysis (n1 ¼ 1, n2 ¼ n3 ¼ 1:5, T0 ¼ 303 K).
Fig. 12. Temperature profile as a function of radial distance for various particle radii at 1200 K at the time of com-
pletion of pyrolysis (n1 ¼ 1, n2 ¼ n3 ¼ 1:5, T0 ¼ 303 K).
Based on the observation made in Figs. 11 and 12 above that as the temperature increases, the
pyrolysis completes faster, the model is simulated for a wide range of temperatures ranging from
303 to 2800 K and particle radii ranging from 0.00025 to 0.017 m for the different geometries.
Some typical results are presented in Figs. 13–15 for particle radii of 0.005 m and 0.00025 m for a
cylinder and for a particle radius of 0.005 m for a sphere. Interestingly, it is found that as the
temperature increases, the time of completion of pyrolysis decreases up to a certain temperature,
giving the optimum pyrolysis time, and then increases with further increase in temperature. For
the particle radius of 0.005 m, the optimum time of completion of pyrolysis is found to be 88 s for
the final furnace temperature of 1645 K as shown in Fig. 13. This can be well explained by the
pyrolysis kinetics with the help of data presented in Tables 10 and 11 for optimum final furnace
temperatures of 1645 K and 2100 K, respectively. As the final furnace temperature is increased
from 1645 K to 2100 K, the decrease in concentration of biomass is more for 2100 K up to a
certain time (74 s). The maximum concentration of (char)1 is obtained in 38 s for the final furnace
temperature of 2100 K as compared to 46 s for 1645 K. So, after 38 s, the concentration of (char)1
decreases, and hence, the consumption of biomass also decreases for the final furnace temperature
of 2100 K. It is also observed that after 74 s, the decrease in concentration of biomass is less for
2100 K as compared to 1645 K because the (char)1 remaining is less for 2100 K as compared to
1645 K. So, the time of completion of pyrolysis to reach the desired final concentration of biomass
(i.e. zero) increases though the final furnace temperature is higher. A similar explanation could be
given for the results shown in Figs. 14 and 15 also. The optimum values of the parameters are
given in Table 12. From Table 12, it is observed that the final pyrolysis time and final pyrolysis
temperature are less for the sphere as compared to the cylinder for the same radius of 0.005 m
because as already observed above, it has a higher surface/volume ratio.
Fig. 16 shows the temperature profiles for the different geometries (slab, cylinder and sphere) as
functions of radial distance at the time of completion of pyrolysis. The radii of the cylinder and
sphere are taken as half the thickness of the slab. It is found that pyrolysis is faster in the case of
the sphere (pyrolysis time ¼ 95 s) and slowest for the slab (pyrolysis time ¼ 185 s), while the trends
Fig. 13. Time of completion of pyrolysis as a function of temperature for cylindrical particle of radius 0.005 m (n1 ¼ 1,
n2 ¼ n3 ¼ 1:5, T0 ¼ 303 K).
Fig. 14. Time of completion of pyrolysis as a function of temperature for cylindrical particle of radius 0.00025 m
(n1 ¼ 1, n2 ¼ n3 ¼ 1:5, T0 ¼ 303 K).
in temperature profiles obtained for all three geometries are same. The spherical particle has the
shortest conversion time as can be expected based on the higher surface/volume ratio as men-
tioned above.
5.2.3. Effect of heat of reaction number on biomass conversion

The effect of Q00 on the progress of pyrolysis of biomass is shown in Fig. 17 where the biomass
conversion is plotted against the pyrolysis time for different values of Q00 . A negative value stands
Fig. 15. Time of completion of pyrolysis as a function of temperature for spherical particle of radius 0.005 m (n1 ¼ 1,
n2 ¼ n3 ¼ 1:5, T0 ¼ 303 K).
Table 10
Simulated results with the optimum temperature (1645 K) for pyrolysis of cylindrical pellet of radius 0.005 m (n1 ¼ 1;
n2 ¼ n3 ¼ 1:5)
t T CB CG1 CC1 C G2 CC2
1.0 303.041416 650.000000 0.000000 0.000000 0.000000 0.000000
15.0 421.921449 649.965228 0.000003 0.034769 0.000000 0.000000
30.0 620.145975 535.048255 14.281696 99.831749 0.419150 0.419150
38.0 657.662178 385.407963 44.324157 179.732538 20.267671 20.267671
39.0 662.221267 366.467356 46.767841 184.470246 26.147278 26.147278
46.0 703.340007 230.944809 57.246921 184.793708 88.507281 88.507281
47.0 711.305786 210.836541 58.023858 180.265779 100.436911 100.436911
48.0 720.086499 190.575671 58.567103 174.558063 113.149581 113.149581
73.0 1220.20633 0.038576 2.747890 21.643747 312.784893 312.784893
74.0 1234.605217 0.026305 2.364912 21.253188 313.177797 313.177797
75.0 1248.412790 0.018079 2.043245 20.926480 313.506098 313.506098
76.0 1261.670650 0.012521 1.771999 20.651856 313.781812 313.781812
88.0 1387.710904 0.000253 0.414538 19.287125 315.149042 315.149042
for an exothermic reaction, while a positive value represents an endothermic reaction. It is ob-
served from the figure that the biomass conversion is more sensitive to the variation of Q00 when it
is negative than when it is positive. The biomass conversion increases with pyrolysis time for a
given value of Q00 .
The biomass conversion at the early stages (up to about 40 s) of the pyrolysis reaction does not
appear to be profoundly affected by different values of Q00 . However, the effect of Q00 on the
biomass conversion becomes significant at a later stage. For Q00 ¼ 0:5, the slope of the curve is
very steep, representing a very abrupt and fast increase in conversion with pyrolysis time. This is
Table 11
Simulated results with temperature ¼ 2100 K for pyrolysis of cylindrical pellet of radius 0.005 m (n1 ¼ 1; n2 ¼ n3 ¼ 1:5)
t T CB CG1 CC1 CG2 CC2
1.0 303.073899 650.000000 0.000000 0.000000 0.000000 0.000000
15.0 462.544327 649.857429 0.000469 0.142102 0.000000 0.000000
30.0 650.049746 504.244613 45.436203 96.004217 2.157483 2.157483
38.0 695.304910 36.019862 104.855115 131.66940 38.727808 38.727808
39.0 703.247479 311.758513 111.097880 129.60329 48.770158 48.770158
46.0 800.150481 120.676965 141.162004 65.012529 161.574251 161.574251
47.0 825.229412 93.974519 142.913731 50.361562 181.375094 181.375094
48.0 855.242754 69.844726 144.101093 36.229038 199.912571 199.912571
73.0 1636.091153 0.030624 153.513662 0.003811 248.225952 248.225952
74.0 1652.552961 0.025688 153.515172 0.003240 248.227950 248.227950
75.0 1668.394619 0.021661 153.516367 0.002772 248.229600 248.229600
76.0 1683.645462 0.018355 153.517318 0.002386 248.230971 248.230971
88.0 1828.787031 0.003332 153.521042 0.000549 248.237539 248.237539
89.0 1838.269943 0.002941 153.521113 0.000496 248.237725 248.237725
152.0 2072.129027 0.000007 153.521442 0.000009 248.239271 248.239271
Table 12
Optimum parametric values for different particle radii and geometries at the centre (n1 ¼ 1; n2 ¼ n3 ¼ 1:5)
Optimum parameters Cylinder Sphere
Particle radii (m) 0.00025 0.005 0.005
Final pyrolysis time (s) 35.0 88.0 78
Final temperature of furnace (K) 1100 1645 1455
Final pyrolysis temperature (K) 1099.99 1387.751 1350.09
Final concentration of initial biomass (kg/m3 ) 0 0 0
Final concentration of (char)1 (kg/m3 ) 0.00715 19.3174 64.3383
Final concentration of (volatile)1 (kg/m3 ) 366.985 0.41275 0.0299
Final concentration of (char)2 (kg/m3 ) 141.504 315.135 292.816
Final concentration of (volatile)2 (kg/m3 ) 141.504 315.135 292.816
due to the fact that the progress of the pyrolysis reaction is accelerated not only by heat con-
duction, convection and radiation from the surrounding environment but also by the heat gene-
rated inside the particle from the exothermic reaction. For the endothermic reaction (Q00 ¼ 10),
the sensible heat content of the particle is consumed as the heat of reaction. Therefore, heating of
the particle from the surrounding environment would not be as effective as in the case of the
exothermic reaction in raising the temperature of the particle. Thus, the rate of increase in the
conversion is relatively low at any given reaction time. The conversion of biomass is almost
complete at t ¼ 97:63 s for Q00 ¼ 0:5, and after this time, the conversion reaches its asymptotic
value and becomes constant from then on, but for Q00 ¼ 10, the conversion is only 48 % even at
t ¼ 97:63 s, complete (100%) at 213.65 s and constant from then onwards. The value of Q00 ¼ 0
(i.e. zero heat of reaction initially) does not imply an isothermal process, since heat transfer from
the surrounding environment would still significantly affect the rate of the pyrolysis reaction and,
hence, the steep increase in conversion with pyrolysis time.
Fig. 16. Temperature profile as a function of radial distance for different geometries at the time of completion of
pyrolysis (n1 ¼ 1, n2 ¼ n3 ¼ 1:5, R ¼ 0:005 m, T0 ¼ 303 K, Tf ¼ 900 K).
Fig. 17. Effect of Q00 on conversion of biomass (R ¼ 0:003 m, T0 ¼ 303 K, Tf ¼ 643 K).
5.2.4. Effect of heat of reaction number on temperature profile

The temperature profiles in the solid particle of biomass for the cases of Q00 ¼ 10 and )0.5 at
various stages of pyrolysis are shown in Figs. 18 and 19, respectively. It is seen that for the case of
Q00 ¼ 10, the temperature increases with radial distance. The temperature is a minimum at the
centre and a maximum at the wall of the particle. Note that the temperature gradient, dh=dx, is
consistently positive or equal to zero. In other words, the particle temperature never exceeds the
final temperature of 643 K. This situation is termed a Ôheat transfer controlled reactionÕ [2]. For
Q00 ¼ 0:5, however, the situation becomes different in that the temperature in the particle ap-
pears to increase beyond the surface temperature of 643 K when the conversion of biomass is near
completion (refer to temperature profile for 97.63 s in Fig. 19). In the case when the particle
temperature exceeds the surface temperature, the temperature gradient, dh=dx, at the surface
becomes negative, and thus, the heat conduction is towards the surroundings. Obviously, the heat
source from the surrounding environment, which is utilized to initiate the pyrolysis reaction, is no
longer required. The heat source could be viewed simply as a trigger for the early stages of the
reaction. This situation is termed the Ôself sustaining reactionÕ [2].
From both figures (Figs. 18 and 19), it is evident that initially, at t ¼ 0, the temperature profile
is flat as no pyrolysis reaction has begun. As time progresses (at t ¼ 20 s), there exists the tem-
perature gradient across the radial distance of the particle. As time progresses further (at t ¼ 40 s),
the temperature gradient decreases, and finally, at t ¼ 97:63 s for Q00 ¼ 0:5 and t ¼ 213:65 s for
Q00 ¼ 10, the temperature profile becomes almost flat, indicating that the pyrolysis is near com-
pletion. This effect can be visualized from Fig. 17 also for Q00 ¼ 0:5 and 10 for which at t ¼ 97:63
and 213.65 s, respectively, the conversion is almost complete.
5.2.5. Effect of heat of reaction number on biomass concentration profile

The concentration profiles of biomass are shown in Figs. 20 and 21 for the cases of Q00 ¼ 10 and
)0.5, respectively. It is seen that the concentration is more at the centre than at the surface as the
pyrolysis progresses. This may be attributed to the combined effects of the rate of heat transfer as
well as the rate of chemical reaction during the process. In the early stages of the reaction (up to
about 40 s), the decrease in concentration of biomass is nearly the same for both cases. At a given
radial position, after the pyrolysis time of 40 s, the decrease in concentration of biomass for the
case of Q00 ¼ 0:5 is much larger than that for the case of Q00 ¼ 10. The pyrolysis is nearly
complete at t ¼ 97:63 s for Q00 ¼ 0:5, but it is only 48% for Q00 ¼ 10 at t ¼ 97:63 s, as can be
visualized from Fig. 17 also, and the time of completion of pyrolysis is higher for Q00 ¼ 10 (213.65 s).
Fig. 18. Temperature profile as a function of radial distance for different values of t (Q00 ¼ 10, R ¼ 0:003 m, T0 ¼ 303 K,
Tf ¼ 643 K).
Fig. 19. Temperature profile as a function of radial distance for different values of t (Q00 ¼ 0:5, R ¼ 0:003 m,
T0 ¼ 303 K, Tf ¼ 643 K).
This is because the temperature in the particle for Q00 ¼ 0:5 is higher than that for Q00 ¼ 10 at
any given time. Consequently, the effective mass diffusivity of the volatiles and gases through the
particle is greater for the former case than for the latter case. It may also be noted that starting
from a concentration profile at t ¼ 0, the concentration gradient increases as time progresses, and
ultimately, the concentration profiles become flat again when reaching the nearly complete con-
version stage. These results are in conjunction with the results shown in Figs. 18 and 19 of the
temperature profiles.
Fig. 20. Concentration profile as a function of radial distance for different values of t (Q00 ¼ 10, R ¼ 0:003 m,
T0 ¼ 303 K, Tf ¼ 643 K).
Fig. 21. Concentration profile as a function of radial distance for different values of t (Q00 ¼ 0:5, R ¼ 0:003 m,
T0 ¼ 303 K, Tf ¼ 643 K).
6. Conclusions
• The simulated results obtained from all three models (Models I, II and III) developed in the
present study are in excellent agreement with the experimental data of Pyle and Zaror [30],
when compared to the mathematical model of Jalan and Srivastava [10] and the Bamford,
Crank and Malan model [3] used by Pyle and Zaror [30].
• Model I is a generalized reference model, which gives good prediction of the temperature pro-
files for lower temperatures and lower times of pyrolysis. However, the two improved simplified
models (Models II and III) predict well in the entire range of temperatures and pyrolysis times.
• Model III is in excellent agreement with the experimental data [30] as compared to the other
developed models (Models I and II) and the models reported in literature [3,10] for a wide
range of operating conditions.
• Simulations are performed using the models developed in the present study for different geo-
metries (slab, cylinder and sphere) for temperatures ranging from 303 to 2800 K and equivalent
particle radii ranging from 0.0001 to 0.017 m.
• The spherical particle has the shortest conversion time as compared to the slab and cylinder.
• The yield of (volatiles and gases)1 is a maximum for the sphere and is least for the slab, while
the yield of (char)1 and (volatiles and gases)2 or (char)2 is more for the slab and is least for the
sphere.
• As the temperature increases, the pyrolysis completes faster.
• Pyrolysis is faster in the case of the sphere (pyrolysis time ¼ 95 s) and slowest for the slab (pyro-
lysis time ¼ 185 s), while the trend in temperature profiles obtained for all three geometries are
the same.
• Some interesting trends have been obtained, especially with respect to the effect of final furnace
temperature on the completion of pyrolysis time. The results obtained using a wide range of
operating conditions in the present study show that the final pyrolysis time initially decreases
and then increases as the final temperature is increased, giving an optimum final pyrolysis time
corresponding to the final temperature.
• The above interesting trends are observed for the different geometries. These interesting trends
that were not reported earlier could be well explained from the kinetics by systematically gen-
erating huge amounts of data upon simulating the model equations.
• Two typical pyrolysis reactions, namely, heat transfer controlled reaction and self sustaining
reaction, are considered. The former is characterized by a positive value of the heat of reaction
number Q00 , e.g. 10, and the latter is characterized by a negative value, e.g. )0.5.
• For the value of Q00 ¼ 10 (endothermic reaction), the temperature profile increases monotoni-
cally to that of the steady state, and the rate of decrease in concentration of the biomass is very
much less.
• When Q00 is )0.5 (exothermic reaction), the temperature in the particle rapidly overshoots the
surface temperature after a certain concentration of biomass, and the rate of decrease in con-
centration of the biomass is very fast.
• In the early stages (up to about 40 s) of the reaction, the biomass conversion does not appear to
be significantly affected by different values of Q00 . Also, the decrease in concentration of the bio-
mass is nearly the same for heat of reaction numbers of 10 and )0.5 in the early stages of the
pyrolysis.
• The completion time for pyrolysis is less ( ¼ 97.63 s) for an exothermic reaction (Q00 ¼ 0:5) as
against a higher value ( ¼ 213.65 s) for an endothermic reaction (Q00 ¼ 10).
• The variation in temperature profile and concentration profile for the exothermic and endother-
mic pyrolysis reactions could be well explained by using the heat of reaction number.
• The results discussed above have a lot of practical importance and physical significance in
industrial pyrolysis applications. The results are also important and useful for the design of bio-
mass gasifiers, reactors etc.
References
[1] Fan LT, Fan LS, Miyanami K, Chen TY, Walawender WP. Can J Chem Eng 1977;55:47.
[2] Miyanami K, Fan LS, Fan LT, Walawender WP. Can J Chem Eng 1977;55:317.
[3] Bamford CH, Crank J, Malan DH. Proc Comb Phil Soc 1946;42:166.
[4] Tinney ER. In: 10th Int Symp on Combustion, Pittsburgh, PA, 1965. p. 925.
[5] Roberts AF, Clough G. In: 9th Int Symp on Combustion, Pittsburgh, PA, 1963. p. 158.
[6] Srivastava VK, Jalan RK. Energy Convers Manage 1994;35:1031.
[7] Tang WK, Neil WK. J Polym Sci 1964;6:65.
[8] Kung HC. Combust Flame 1972;18:185.
[9] Kansa EJ, Perlee HE, Chaiken RF. Combust Flame 1977;29:311.
[10] Jalan RK, Srivastava VK. Energy Convers Manage 1999;40:467.
[11] Koufopanos CA, Maschio G, Lucchesi A. Can J Chem Eng 1989;67:75.
[12] Babu BV, Chaurasia AS. Energy Convers Manage 2003;44:2251. Available from: http://bvbabu.50megs.com/
custom.html/#50.
custom.html/#47.
custom.html/#62.
[15] Babu BV, Chaurasia AS. Chem Eng Sci, submitted for publication.
[16] Babu BV, Chaurasia AS. Chem Eng Sci, submitted for publication.
[17] Babu BV, Chaurasia AS. In: Proc Int Conf on Multimedia and Design, Mumbai, India, 2002. p. 103. Available
from: http://bvbabu.50megs.com/custom.html/#51.
[18] Babu BV, Chaurasia AS. In: Proc Int Symp & 55th Annual Session of IIChE (CHEMCON-2002), OU, Hyderabad,
India, 2002. p. 105. Available from: http://bvbabu.50megs.com/custom.html/#52.
[19] Babu BV, Chaurasia AS. In: Proc Int Symp on Process Systems Engineering and Control (ISPSEC-2003)––for
Productivity Enhancement Through Design and Optimization, IIT-Bombay, Mumbai, India, 2003. p. 181.
Available from: http://bvbabu.50megs.com/custom.html/#55.
[20] Chaurasia AS, Babu BV. In: Int Conf on Energy and Environmental Technologies for Sustainable Development
(ICEET-2003), Jaipur, India, October 8–10, 2003. Available from: http://bvbabu.50megs.com/custom.html/#65.
[21] Chaurasia AS, Babu BV. In: Int Conf on Desert Technology-7 (DT-7), Jodhpur, India, November 9–14, 2003.
[22] Liliedahl T, Sj€
ostr€
om K. Biomass Bioenergy 1998;15:503.
[23] Shafizadeh F. J Anal Appl Pyrolysis 1982;3:283.
[24] Anthony DB, Howard JB. AIChE J 1976;22:625.
[25] Boutin O, Ferrer M, Lede J. Chem Eng Sci 2002;57:15.
[26] Beis SH, Onay O, € KocßKar OM.
€ Renewable Energy 2002;26:21.
[27] Di Blasi C. Fuel 1996;75:58.
[28] Di Blasi C. Fuel 1997;76:957.
[29] Di Blasi C. Prog Energy Combust Sci 1993;19:71.
[30] Pyle DL, Zaror CA. Chem Eng Sci 1984;39:147.
[31] Ghoshdastidar PS. Computer simulation of flow and heat transfer. New Delhi: Tata McGraw-Hill Publishing
Company Limited; 1998. p. 32–118.
[32] Babu BV, Angira R, Nilekar A. Comput Chem Eng, submitted for publication.
[33] Carnahan B, Luther HA, James OW. Applied numerical methods. New York: John Wiley & Sons; 1969.
[34] Hagge MJ, Bryden KM. Chem Eng Sci 2002;57:2811.
[35] Di Blasi C. Int J Heat Mass Transfer 1998;41:4139.

Pyrolysis of Biomass - Improved Models For Simultaneous

Caricato da

Informazioni sul documento

Descrizione originale:

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Pyrolysis of Biomass - Improved Models For Simultaneous

Caricato da

Copyright:

Formati disponibili

Energy Conversion and Management 45 (2004) 1297–1327

Pyrolysis of biomass: improved models for simultaneous

A1 ; A2 ; A3 frequency factor, 1/s

Dx radial grid distance

3. Description of mathematical model

t ¼ 0; CB ¼ CB0 ; CG1 ¼ CC1 ¼ CG2 ¼ CC2 ¼ 0; T ðr; 0Þ ¼ T0 ð7Þ

Particle boundary conditions:

Koufopanos et al. correlation [11]:

h ¼ 0:322ðk=lÞPr1=3 Re0:5 ð23Þ

Darcy law and state equation:

p ¼ CG1 Rc T =Wm ð25Þ

e00 ¼ e000 þ cð1 C B Þ; / ¼ g/B þ ð1 gÞ/C1 ; g ¼ CB =CB0 ð26Þ–ð28Þ

3.1. Generalized reference model (Model I)

3.2. Simpliﬁed models

BiM ¼ ðR=kÞbh þ erðT 3 þ T 2 Tf þ Tf2 T þ Tf3 Þc ð34Þ

Q ¼ ðDH þ Cp T Þ=ðqCp ðT0 Tf ÞÞ ð35Þ

Q00 ¼ QCBn1 ð36Þ

C B0 ¼ CB0 =q0 ð38Þ

C G1 ¼ CG1 =CB0 ð39Þ

C G1 0 ¼ CG1 0 =CB0 ð40Þ

C C1 ¼ CC1 =CB0 ð41Þ

C G2 ¼ CG2 =CB0 ð42Þ

C C2 ¼ CC2 =CB0 ð43Þ

C pG1 ¼ CpG1 =CpG1 0 ð44Þ

DeG1 ¼ DeG1 =DeG1 0 ð45Þ

Le ¼ ðkÞ=ðq0 CpG1 0 DeG1 0 Þ ð46Þ

Sh ¼ ðkmG1 RÞ=DeG1 0 ð47Þ

3.2.1. First simpliﬁed model (Model II)

3.2.2. Second simpliﬁed model (Model III)

4. Numerical solution and simulation

5. Results and discussion

5.1. Models validation and comparison

5.2. Simulation results

5.2.2. Eﬀect of temperature on particle size and conversion time of pyrolysis

5.2.3. Eﬀect of heat of reaction number on biomass conversion

5.2.4. Eﬀect of heat of reaction number on temperature proﬁle

5.2.5. Eﬀect of heat of reaction number on biomass concentration proﬁle

Potrebbero piacerti anche