A Distributed Activation Energy Model For Cellulose Pyrolysis in A FBR
A Distributed Activation Energy Model For Cellulose Pyrolysis in A FBR
A Distributed Activation Energy Model For Cellulose Pyrolysis in A FBR
⁎ ⁎⁎
Samreen Hameed a,c, Abhishek Sharma b,d, , Vishnu Pareek c,
a
Chemical, Polymer & Composite Materials Engineering Department, University of Engineering & Technology, New
Campus, Lahore, Pakistan
b
Department of Chemical Engineering, Manipal University Jaipur, Rajasthan, India
c
Western Australian School of Mines: Minerals, Energy and Chemical Engineering, Curtin University, WA,
Australia
d
Chemical & Environmental Engineering, School of Engineering, RMIT University, Melbourne, Victoria, Australia
a r t i c l e i n f o a b s t r a c t
Article history: Cellulose pyrolysis is used to give solid char, condensable vapors and non-condensable
Received 8 September 2022 gases. This is a complex process and to model this using CFD simulations, Euler-Euler
Received in revised form 10 January approach with multi-fluid is applied. The objective of the present study is to develop a CFD
2023 model to combine the reaction kinetics with the hydrodynamics to study the cellulose
Accepted 26 January 2023 pyrolysis in the fluidized bed reactor. The distributed activation energy model to compare
Available online 29 January 2023 the product yield with and without distribution of activation energy has been in
corporated. Simulations of cellulose pyrolysis in a fluidized bed reactor using two different
Keywords: kinetic schemes have been conducted. Model has been validated using global kinetic
Modeling scheme and then optimized kinetic parameters are introduced with a distribution of ac
Kinetic tivation energy. The predicted values of activation energies, frequency factor and stan
Fluidized bed reactor dard deviation impact the overall pyrolysis product yield. The yield of tar and char
Pyrolysis increases, however the fraction of gases decreases significantly because of higher acti
Cellulose vation energy. Tar yield is 82.1 wt% in the presence of DAEM, while the gas yield is 8 wt%
DAEM and char yield comes out to be 9.9 wt%.
© 2023 Published by Elsevier Ltd on behalf of Institution of Chemical Engineers.
Abbreviations: BFB, bubbling fluidized bed; CFD, computational fluid dynamics; DAEM, distributed activation energy model; EE,
Eulerian-Eulerian; FBR, fluidized bed reactor; TGA, thermo-gravimetric analysis
⁎
Corresponding author at: Department of Chemical Engineering, Manipal University Jaipur, Rajasthan, India.
⁎⁎
Corresponding author.
E-mail addresses: abhishek.sharma@jaipur.manipal.edu (A. Sharma), v.pareek@curtin.edu.au (V. Pareek).
https://doi.org/10.1016/j.cherd.2023.01.048
0263-8762/© 2023 Published by Elsevier Ltd on behalf of Institution of Chemical Engineers.
Chemical Engineering Research and Design 191 (2023) 414–425 415
1 (E E0 )2
Tar Vapors, Rg,T = k2 s1 s1 Xs1,A.Cell k4 g g Xg (3) f(E) = exp 2
2 2 (10)
Gas, Rg,G Where E0 is the mean activation energy and σ is the standard
deviation. Kinetic parameters, frequency factor, A and acti
= k3 s1 s1 Xs1,A.Cell (1 Y) + k 4 g g
vation energy, E were predicted using model fitting approach
g g
Xg + [ s1 s1 Xs1,A.Cell (k2 + k3)] +[ s1 s1 Xs1,A.Cell k3Y] in which the objective function was minimized by using
b c
fmincon in MATLAB as discussed in detail in the work of
(4) Hameed et al. (Hameed et al., 2022). These optimized kinetic
Char, R s1,C = parameters are given below in Table 3:
s1 s1 Xs1,A.Cell k3Y (5)
“Y” represents the ratio of char formed in the third reac 2.3. Multi-fluid model
tion, which is 0.35 (Xue et al., 2012).
All the reactions in this scheme are considered first-order The multiphase model used in this study describes the gas
and the reaction kinetics are determined according to and solid phases as inter-penetrating continua in Eulerian-
Arrhenius law, as: Eulerian framework. During the pyrolysis in a fluidized bed
k = Ae E/RT (6) reactor there exist three different phases: gas phase as pri
mary phase and the two solid phases as the secondary
The data of kinetic parameters and the heat of reaction is
phases. Conservation of mass, momentum, energy and the
presented in Table 1:
other basic laws is specified using the models described
Thermo-physical properties of species involved in the
below (Fluent, 2009). In addition to basic conservation laws,
pyrolysis of cellulose are presented in Table 2.
some other correlations necessary for the implementation of
multi-fluid model are also discussed in the following sec
2.2. Distributed activation energy model
tions:
k1 2.8 × 1019 2.42 × 108 (Di Blasi, 1994) 0 (Koufopanos et al., 1991)
k2 3.28 × 1014 1.965 × 108 (Di Blasi, 1994) 255 (Koufopanos et al., 1991)
k3 1.3 × 1010 1.505 × 108 (Di Blasi, 1994) -20 (Koufopanos et al., 1991)
k4 4.25 × 106 1.080 × 108 (Di Blasi, 1994) -42 (Curtis and Miller, 1988)
Chemical Engineering Research and Design 191 (2023) 414–425 417
Table 2 – Thermo-physical properties of species (Jalalifar et al., 2017; Lathouwers and Bellan, 2001).
Species Density, ρ Particle Molecular Weight, Heat Capacity, Dynamic Thermal Conductivity, k
(kg/m3) Diameter, ds (m) M.W (kg/kmol) Cp (J/kg.K) Viscosity, μ (J/kg.K)
(kg/ms)
Where, Hsn and qsn are the specific heat and conductive heat
flux of the nth solid phase. Hrsn is the source term which
incorporates the enthalpy change due to chemical reactions
in the solid phase. However, the inter-phase heat transfer
between the gas and solid phases is given by Hgs. Since there
is no radiative heat transfer in the system hence the terms
including radiative heat transfer have been neglected. These
terms can be defined as:
3 ( sn sn sn )
= (psn I + sn): vsn + sn + ln
2 t (25)
Table 5 – Pyrolysis product yield (wt%) comparison between experiment and simulations for pure cellulose.
Data Bio-oil (wt%) Char (wt%) Gases (wt%)
value of co-efficient of restitution was taken as 0.97 and the than the total volume as stated by Xue et al., (2012). The si
angle of internal friction was 55. For boundary conditions of mulated results showed an increase in amount of gases
walls, no slip was specified for the gas phase while for the that is the actual amount of gases leaving the fluidized bed
solid phase partial slip condition was used. reactor.
Energy conservation and species transport models were
used to account for the pyrolysis reactions. The reaction
terms have been integrated using a stiff ODE (Ordinary
Differential Equations) solver. For time discretization, second 4.2. Implementation of DAEM to CFD
order implicit method was used with the time step of 10−4
sec. QUICK algorithm has been used to calculate the volume Kinetic parameters used in the current model were predicted
fraction of each phase. Conservation equations for mo using thermogravimetric analysis and DAEM (Hameed, 2019).
mentum, energy and the species transport equations have The predicted kinetic parameters with the mean activation
been solved using second order (upwind) method. All simu energy and the standard deviation for pure cellulose are
lations were run for the simulation time of 150 s when summarized previously in Table 3. Kinetic parameters for
pseudo-steady state was achieved in the system. reaction 1 and 4 were specified with standard deviations of
activation energy; however, for reactions 2 and 3 only single
values of activation energy were used. These kinetic para
4. Results and discussion
meters have been used to understand the cellulose pyrolysis
behavior following the same reaction scheme as described in
4.1. Model validation
Figure1, in a fluidized bed reactor using CFD simulations.
The analytical solution of the DAEM is difficult therefore
In this study, the CFD model was validated using the ex
researchers have focused on the numerical solution of this
perimental data of Xue et al. (2012), and the same model has
model. Braun and Burnham used E0 3 andE0 + 3 as the
been used to calculate the pyrolysis product yield with dis
integral lower and upper limits respectively and solved the
tributed activation energy model which will be discussed in
model using 96 intervals for activation energy (Braun,
the next section in detail. In this part of the research, all
Burnham, 1987). Xiong et al., have previously carried out the
operating parameters and material properties have been
integration of DAEM with CFD, using different intervals such
kept same as of experimental data. The product yield (η) was
as: (E0 3 , E0 + 3 ), (E0 4 , E0 + 4 ), (E0 5 , E0 + 5 ) and
calculated using the equation:
(E0 6 , E0 + 6 ). They concluded that if the integration in
t tervals and their spaces are chosen optimally, then this
0 outlet
( X)dAdt + reactor
( X)dV
(t) = coupling does not increase the computational power prohi
Mcell_feed (26)
bitively (Xiong et al., n.d.). Keeping in view the outcome of
The comparison of the product yield obtained from si previous reported literature, the integration space chosen for
mulations using the literature kinetic data with the experi this case was (E0 3 , E0 + 3 ) with the 60 equal intervals.
mental data is presented in Table 5. Mean activation energies have been calculated and pre-ex
The comparison of the present simulation results with the ponential factor has also been modified using the Gaussian
experimental data shows that the predicted bio-oil yield was distribution function to incorporate the probability function.
comparable to the experimental data reported in literature. This has led to 124 chemical reaction equations. Reaction
The char and non-condensable gases yield, however, was terms along with the transport equations have been solved in
slightly over-predicted by the simulation when compared ANSYS FLUENT to predict the pyrolysis product yield. The
with the experimental results. Nevertheless, it is reasonable yield of bio-oil, char and gases is presented in Table 6.
to conclude that the product yields predicted by the model A comparison of product yield from experimental data
were in close agreement with the experimental data, thus and simulation results using Broido-Shafizadeh scheme,
validating the CFD model. The mass closure from the ex with and without DAEM is shown in Table 6. It is clear that
perimental data is around 96.7% which is expected due to the with DAEM the predicted bio-oil yield increased significantly.
error in calculating the mass percentage of gases. The cal This observation is in agreement with those reported in the
culated volume of non-condensable gases came out to be less literature (Xiong et al., 2016b) However, the yield of other
Table 6 – Pyrolysis product yield for pure cellulose with and without using DAEM.
Data Bio-oil (wt%) Char (wt%) Gases (wt%)
Fig. 4 – Time-averaged volume fraction profiles of three phases in the fluidized bed reactor.
starting to increase near the feed inlet where cellulose en bed region. As the tar particles moved along the bed height, it
tered in the reactor and started to pyrolyze. The rate of tar started to decompose to form gases, and its concentration
formation was higher than the gas formation reaction and decreased progressively. Since both tar and gases are elu
hence the concentration of tar increased more rapidly in the triated out of the reactor along with the gas phase, their
Fig. 5 – Instantaneous temperature profiles for gas, sand and cellulose phase in the fluidized bed reactor.
422 Chemical Engineering Research and Design 191 (2023) 414–425
Fig. 6 – Time-averaged temperature profiles of three phases Fig. 8 – Molar concentration of tar and gases.
in fluidized bed reactor.
Fig. 7 – Time-averaged velocity vectors of the three phases (a) gas (b) cellulose (c) sand.
Chemical Engineering Research and Design 191 (2023) 414–425 423
Fig. 9 – Mass fraction of pyrolysis product species (a) tar, (b) gas and (c) char.
Program of Higher Education Commission of Pakistan for the In the particles collision term there are two important
support to her PhD studies. parameters: ess is known as restitution coefficient and g0,ss
is the radial distribution function.
Appendix A For multiphase systems having N number of phases, the
same equation has been modified as:
Granular temperature model has been specified as phase
N dls3
property, neglecting the convective and diffusive terms, pl = l l l + l=1
2 (1 + els ) g0, ls l s l l
dl3 (38)
hence energy dissipation due to collisions and energy ex
change will be given as: The radial distribution is calculated according to Syamlal-
2 O′brien (Syamlal et al., 1993):
12(1 ess ) g0, ss 2 3/2
s = s s s N
ds (27) 1 3( k = 1 k / dk )
g0, kl = + 2
ds dl
(1 s) (1 s ) (ds + dl ) (39)
ls = 3Kls s (28)
Jalalifar, S., Abbassi, R., Garaniya, V., Hawboldt, K., 2018. Soria-Verdugo, A., Goos, E., García-Hernando, N., 2015. Effect of
Parametric analysis of pyrolysis process on the product yields the number of TGA curves employed on the biomass pyrolysis
in a bubbling fl uidized bed reactor. Fuel 234, 616–625. kinetics results obtained using the distributed activation en
Kirtania, K., Bhattacharya, S., 2015. Coupling of a distributed ac ergy model. Fuel Process. Technol. 134, 360–371. https://doi.
tivation energy model with particle simulation for entrained org/10.1016/j.fuproc.2015.02.018
flow pyrolysis of biomass. Fuel Process. Technol. 137, 131–138. Syamlal, M., 1987. The particle-particle drag term in a multi
https://doi.org/10.1016/j.fuproc.2015.04.014 particle model of fluidization. https://doi.org/MC/21353–2373,
Koufopanos, C.A., Papayannakos, N., Maschio, G., Lucchesi, A., NTIS/DE87006500.
1991. Modelling of the pyrolysis of biomass particles. Studies Syamlal, M., Rogers, W., O`Brien, T.J., 1993. MFIX documentation
on kinetics, thermal and heat transfer effects. Can. J. Chem. theory guide. Other Inf. PBD Dec 1993 1004, Medium: ED; Size:
Eng. 69, 907–915. 49 p. https://doi.org/METC-9411004, NTIS/DE9400087.
Lathouwers, D., Bellan, J., 2001. Yield optimization and scaling of White, J.E., Catallo, W.J., Legendre, B.L., 2011. Biomass pyrolysis
fluidized beds for tar production from biomass. Energy Fuels kinetics: a comparative critical review with relevant agri
15, 1247–1262. https://doi.org/10.1021/ef010053h cultural residue case studies. J. Anal. Appl. Pyrolysis 91, 1–33.
Lun, C.K.K., Savage, S.B., Jeffrey, D.J., Chepurniy, N., 1984. Kinetic https://doi.org/10.1016/j.jaap.2011.01.004
theories for granular flow: inelastic particles in Couette flow and Xiong, Q., Xu, F., Ramirez, E., Pannala, S., Daw, C.S., 2016a.
slightly inelastic particles in a general flowfield. J. Fluid Mech. Modeling the impact of bubbling bed hydrodynamics on tar
140, 223–256. https://doi.org/10.1017/S0022112084000586 yield and its fluctuations during biomass fast pyrolysis. Fuel
Schaeffer, D.G., 1987. Instability in the evolution-equations de 164, 11–17. https://doi.org/10.1016/j.fuel.2015.09.074
scribing incompressible granular flow. J. Differ. Equ. 66, 19–50. Xiong, Q., Zhang, J., Xu, F., Wiggins, G., Stuart Daw, C., 2016b.
Shafizadeh, F., 1982. Introduction to pyrolysis of biomass. J. Anal. Coupling DAEM and CFD for simulating biomass fast pyrolysis
Appl. Pyrolysis 3, 283–305. https://doi.org/10.1016/0165- in fluidized beds. J. Anal. Appl. Pyrolysis 117, 176–181. https://
2370(82)80017-X doi.org/10.1016/j.jaap.2015.11.015
Shafizadeh, F., Bradbury, A.G.W., 1979. Thermal degradation of Xue, Q., Heindel, T.J., Fox, R.O., 2011. A CFD model for biomass
cellulose in air and nitrogen at low temperatures. J. Appl. fast pyrolysis in fluidized-bed reactors. Chem. Eng. Sci. 66,
Polym. Sci. 23, 1431–1442. https://doi.org/10.1002/app.1979. 2440–2452. https://doi.org/10.1016/j.ces.2011.03.010
070230513 Xue, Q., Dalluge, D., Heindel, T.J., Fox, R.O., Brown, R.C., 2012.
Sharma, A., Wang, S., Pareek, V., Yang, H., Zhang, D., 2015. Multi- Experimental validation and CFD modeling study of biomass
fluid reactive modeling of fluidized bed pyrolysis process. fast pyrolysis in fluidized-bed reactors. Fuel 97, 757–769.
Chem. Eng. Sci. 123, 311–321. https://doi.org/10.1016/j.ces. https://doi.org/10.1016/j.fuel.2012.02.065
2014.11.019 Zhu, H., Dong, Z., Yu, X., Cunningham, G., Umashanker, J., Zhang,
Sonobe, T., Worasuwannarak, N., 2008. Kinetic analyses of X., Bridgwater, A.V., Cai, J., 2021. A predictive PBM-DEAM
biomass pyrolysis using the distributed activation energy model for lignocellulosic biomass pyrolysis. J. Anal. Appl.
model. Fuel 87, 414–421. https://doi.org/10.1016/j.fuel.2007. Pyrolysis 157. https://doi.org/10.1016/j.jaap.2021.105231
05.004