- Email: [email protected]
Predicting molecular composition of primary product derived from fast pyrolysis of lignin with semi-detailed kinetic model
Fuel 212 (2018) 515–522
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Full Length Article
Predicting molecular composition of primary product derived from fast pyrolysis of lignin with semi-detailed kinetic model Yuki Furutania, Shinji Kudob, Jun-ichiro Hayashib,c, Koyo Norinagad,
MARK
⁎
a
Interdisciplinary Graduate School of Engineering Sciences, Kyushu University, 6-1, Kasuga-koen, Kasuga, Fukuoka 816-8580, Japan Institute for Materials Chemistry and Engineering, Kyushu University, 6-1, Kasuga-koen, Kasuga, Fukuoka 816-8580, Japan c Research and Education Center of Carbon Resources, Kyushu University, 6-1, Kasuga-koen, Kasuga, Fukuoka 816-8580, Japan d Graduate School of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan b
G RA P H I C A L AB S T R A C T
A R T I C L E I N F O
A B S T R A C T
Keywords: Fast pyrolysis Lignin Product yield Semi-detailed kinetic model
A numerical approach is presented for predicting the yields of char and volatile components obtained from fast pyrolysis of three types of lignin (enzymatic hydrolysis lignin, EHL; organic extracted lignin, OEL; and Klason lignin, KL) in a two-stage tubular reactor (TS-TR) at 773–1223 K. The heating rate of lignin particle in the TS-TR was estimated at 102–104 K/s by solving the heat transfer equation. The pyrolytic behavior of lignin and the formation of products in the temperature rising process were predicted using a semi-detailed kinetic model consisting of 93 species and 406 reactions, and the predicted yields of 8 primary products (i.e., char, tar, CO, CO2, H2O, CH3OH, C2H6, and C3H6) were compared with experimental data for the critical evaluation. For EHL, the predicted yields of char and H2O were in good agreement with the experimental results at all temperatures. However, the numerical simulation overestimated tar yield and underestimated CO yield at high temperature probably due to a lack of the kinetic model of the tar cracking reaction. The predicted yields of CH3OH, C2H6, and C3H6 were close to the experimental values at high temperature by adding the detailed chemical kinetic model of the secondary vapor-phase reaction. Moreover, the model reproduced the experimental observation that among the three types of lignin the char yield increased in the order of EHL < OEL < KL, whereas the tar yield decreased.
⁎
Corresponding author. E-mail address: [email protected] (K. Norinaga).
http://dx.doi.org/10.1016/j.fuel.2017.10.079 Received 13 July 2017; Received in revised form 20 September 2017; Accepted 16 October 2017 0016-2361/ © 2017 Elsevier Ltd. All rights reserved.
Fuel 212 (2018) 515–522
Y. Furutani et al.
Fig. 1. Virtual compounds used to characterize initial lignin structures [47].
Table 1 Elemental compositions and equivalent compositions of EHL, OEL, and KL.
EHL OEL KL
Elemental composition (wt%)
Equivalent composition (wt%)
C
H
O
PLIG-C
PLIG-O
PLIG-H
63.3 62.0 62.9
5.9 5.7 5.3
28.9 31.6 31.4
44.1 36.8 55.3
0.0 23.3 41.3
55.9 39.9 3.4
time of 0.1 s for vapor-phase reactions [11–13]. In order to enhance the versatility of the DCKM, it is necessary to expand the DCKM to include the primary pyrolysis stage. For primary pyrolysis of lignin, lumped kinetic models of the decomposition reaction have been extensively reported in the literature [31–35]. However, these models did not allow prediction of the molecular composition of gases and tar components. Recently, Xiong et al. [36] performed computational fluid dynamics (CFD) simulations coupled with distributed activation energy mode (DAEM) reaction kinetics for lab-scale bubbling bed biomass pyrolysis reactor, and revealed that the coupled CFD–DAEM system does not significantly increase computational overhead. A reliable kinetic model, which predict the yields and the molecular composition of gas and tar as accurately as possible, has been required with the development of highly efficient CFD methods [36–45] and the increase of computational resource. Faravelli et al. [46] explored a semi-detailed kinetic model, which characterizes lignin structures using three virtual compounds and involves approximately 100 species and 400 reactions, to predict the molecular compositions of products derived from primary pyrolysis. Recently, Hough et al. [47] added eight reactions into the kinetic models established by Faravelli et al. [46], and compared the integral yields of char, tar, and gases predicted by the model and observed for slow pyrolysis experiments by thermogravimetric analysis (TGA) [48]. However, the model predictions have not yet been compared with the molecular compositions of volatiles derived from fast pyrolysis experiments, such as TS-TR experiments. This comparison is also an important step to optimize fast pyrolysis process, which generates much amount of volatile products from biomass and produces “bio-fuel” [5]. The purpose of this study is to examine whether the semi-detailed kinetic model established by Hough et al. [47] could reproduce the molecular composition of the primary products generated from fast pyrolysis with TS-TR experiments. First, characterization of lignin structures was described using virtual compounds. Second, the heating rate for lignin samples in the primary pyrolysis zone of the TS-TR was estimated. Finally, based on the estimated heating rate, the yields of char and volatiles were predicted using the semi-detailed kinetic model and compared with TS-TR experimental results [13] for the critical evaluation. This estimation would help to integrate the DCKM of both primary pyrolysis and secondary vapor-phase reactions.
Fig. 2. Elemental compositions of the six lignins.
1. Introduction After cellulose, lignin is the second most abundant component in biomass [1,2]. As a residue of the pulp and paper industry, huge amounts of lignin are available at low cost [2]. Although currently most lignin is burned to produce heat, thermal conversion processes, including pyrolysis, gasification and liquefaction, can be used to convert lignin or biomass into useful products, such as gas, char, liquid fuels and chemicals as well as heat [3–6]. Among these thermal conversion processes, pyrolysis is known as a common step to cause fragmentation of the lignin or biomass structure [7]. Pyrolysis is divided into two stages: (a) primary pyrolysis, where volatiles escape from biomass particles; and (b) secondary vapor-phase reactions, where the produced volatiles undergo further cracking, combine, or condense in the vapor phase. A lot of studies have been carried out to identify the pyrolysis products generated from secondary vapor-phase reactions and to establish lumped kinetic models [7–10]. Caballero et al. [10] established a lumped kinetic model for the global secondary reaction of Kraft lignin by assuming a first-order reaction. However, lumped kinetic models established based on global product categories, such as char, tar, and gases, cannot describe the formation mechanisms for specific products at the molecular level. A detailed chemical kinetic model (DCKM) of vapor-phase reactions based on elementary reactions [11–30] has been developed to overcome the limitations of the lumping approach and provide information on the pyrolysis behaviors of individual components. Our group revealed that the DCKM was able to reproduce not only the yields of major products but also those of minor products such as aromatic hydrocarbons, which were obtained with a two-stage tubular reactor (TSTR) connected to a gas chromatograph (GC) [11–13]. However, the DCKM has been limited to secondary vapor-phase reactions. Thus, the molecular composition of the volatiles derived from fast pyrolysis has to be obtained as a boundary condition with the TS-TR setting a residence 516
Fuel 212 (2018) 515–522
Y. Furutani et al.
outside the bounds of the triangle, the mass fraction of PLIG-O had a negative value. In this case, the mass fraction of PLIG-O was set to zero and the values of the other two components (i.e. PLIG-C and PLIG-H) were normalized to express the equivalent composition of EHL. Table 1 lists the elemental compositions and the equivalent compositions of EHL, OEL, and KL. 2.2. Estimation of heating rate No direct measurement of the sample heating rate has been made. Here, a numerical model for briefly estimating the heating rate and the results is presented. By assuming that the sample has a spherical shape, the temperature history (T) along the radial direction (r) in the lignin sample was estimated based on the following heat transfer equation:
∂T α ∂ ∂T = 2 ⎛r 2 ⎞ ∂t r ∂r ⎝ ∂r ⎠
Fig. 3. Average temperature profiles (Tav [K]) as a function of time (t [s]) at Ts = 773–1223 K.
(1)
The average temperature (Tav) in particle was expressed as:
2. Methods and modeling
Tav = 2.1. Virtual compounds for approximating lignin structures
1 R
∫0
R
Tdr
(2)
The following assumptions were made for this calculation. A detailed primary pyrolysis analysis of three types of lignin (enzymatic hydrolysis lignin, EHL; organic extracted lignin, OEL; and Klason lignin, KL) with TS-TR was reported by Yang et al. [13,49], and thus these lignins were employed in this study. We followed the approach of Hough et al. [47] by choosing virtual compounds to approximate various possible lignin structures. Three reference components based on a β-O-4 skeleton [50] (PLIG-C, PLIG-H, and PLIG-O, following the naming conventions of Hough et al. [47]) as shown in Fig. 1 were selected to characterize the structures of EHL, OEL, and KL. Fig. 2 gives a synoptic view of the elemental compositions of the six lignins (PLIG-C, PLIG-H, PLIG-O, EHL, OEL, and KL). The elemental compositions of OEL and KL was inside the bounds of a triangle whose vertices were the three reference compounds. Thus, the equivalent compositions of OEL and KL can be described in terms of the three reference compounds using a linear relationship based on mass conservation. On the other hand, as the elemental composition of EHL was
1. The thermal diffusivity (α) and particle size (R) are 1.83 × 10−7 m2/s and 0.56 mm, which are calculated by solving the following equation:
α=
k ρCp
4 m = ρ πR3 3
(3)
(4)
where the thermal conductivity (k), heat capacity (Cp), density (ρ), and mass (m) are 0.39 W/m/K [51], 1.60 J/g/K [52], 1.33 g/cm3 [53] and 1.00 mg [13], respectively. 2. The heat of reaction is negligibly small. 3. The initial temperature in the particle is 300 K. Fig. 4. Yields of (a) reactants, (b) main intermediates, main products of (c) char and (d) volatile obtained by heating at 103 K/s–1223 K. The chemical structures of intermediates and products are shown in Table 2.
517
Fuel 212 (2018) 515–522
Y. Furutani et al.
Table 2 Main components of (a) intermediates and (b) products (following the naming conventions of Faravelli et al. [46]).
(a)
(b)
518
Fuel 212 (2018) 515–522
Y. Furutani et al.
Fig. 5. Comparison of the predicted yields of (a) char, (b) tar, (c) CO, (d) CO2 and (e) H2O with experimental data [13].
Yield (wt%)
10 8
time, as derived from Eq. (2). According to the temperature curves in Fig. 3, the range of heating rate was estimated as 102–104 K/s.
102 [K/s] 103 [K/s] 104 [K/s] Experiment
2.3. Detailed chemical kinetic model for lignin pyrolysis
6
The semi-detailed kinetic model proposed by Hough et al. [47] was used to predict the molecular composition of the primary products generated from fast pyrolysis with TS-TR experiments [13]. Details of the reactions and sources of kinetic parameters are available in Hough et al.’s article [47]. DETCHEMBATCH code [11,12] was used to run the proposed kinetic models. Numerical simulations were performed under various heating rates (102, 103, and 104 K/s). After heating up to the experimentally-designated temperature (773–1223 K), the temperature was held isothermally until the char yield was constant. Fig. 4a–d shows the predicted thermal degradation behavior of EHL and product formation at 103 K/s heated up to 1223 K. Table 2 lists the chemical structures of the intermediates and products. Fig. 4a and b revealed that above 700 K the intermediate components of LIGH and LIGC were formed immediately upon degradation of PLIG-C and PLIGH. As shown in Fig. 4c and d, above 1000 K aromatic carbons (C10H2), VKETDM2, and H2O were generated as the main char, tar and gas products, respectively. Here, volatile species that were heavier than benzene were defined as tar, otherwise done as gas. The formation of aromatic carbons in char well reproduced the product characteristics revealed by temperature-dependent nuclear magnetic (NMR) spectra [54,55]. Similarly, Figs. S3–S9 in Supplementary material shows the predicted thermal degradation behavior of three lignins (EHL, OEL and
4 2 0
0 wt% 10 wt% 20 wt% 30 wt%
Fig. 6. The predicted change of the CO2 yield at 1223 K along with a fixed ratio of PLIG-C and PLIG-H and the mass fraction of PLIG-O increasing from 0.0 to 30.0 wt%. The experimental data [13] was also shown to compare with the numerical predictions.
4. When the particle is dropped into the bottom of the first isothermal zone (the primary pyrolysis zone) [13,49], the surface of the particle is exposed to the experimentally-defined pyrolysis temperature. Thus, the temperature at the wall (Ts) is fixed at 773–1223 K. Using these four assumptions, Eqs. (1) and (2) were solved by the finite difference method, as described in the Supplementary material. Fig. 3 shows the average temperature profile (Tav) as a function of 519
Fuel 212 (2018) 515–522
Y. Furutani et al.
2.5 2.0 1.5 1.0
Yield (wt%)
2.0
(b)
Fig. 7. Comparison of the predicted yields of (a) CH3OH, (b) C2H6, and (c) C3H6 with experimental data [13].
0.8 0.6 0.4 0.2
0.5 0.0
1.0
(a)
Yield (wt%)
Yield (wt%)
3.0
773K 923K 1023K 1123K 1223K
0.0
773K 923K 1023K 1123K 1223K
(c)
1.6 1.2 0.8 0.4 0.0
773K 923K 1023K 1123K 1223K
Fig. 8. Yields of (a) CH3OH, (b) C2H6, and (c) C3H6 predicted with and without DCKM of secondary vapor-phase reaction [13]. The experimental data [13] was also shown to compare with the numerical predictions.
3. Results and discussion
KL) and product formation under various temperature conditions. Experimental data obtained with the TS-TR connected to GC [12,13] was used to compare with numerical predictions. The TS-TR consists of two zones: (a) first zone for the primary pyrolysis, and (b) secondary zone for the vapor-phase reactions of volatiles. The sample was dropped into the bottom of the first zone, and then the generated volatiles were flown into the secondary zone with helium carrier gas. After passing through the secondary zone, the volatiles were directed into GC column and further identified by GC detector.
To critically evaluate the semi-detailed kinetic model for fast pyrolysis of lignin samples, the predicted yields were compared with experimental data. Fig. 5 shows a comparison between the product yields (char, tar, CO, CO2 and H2O) at the terminal points in Figs. S3–S7 and the experimental data [13] obtained using a TS-TR with a residence time of 0.1 s for vapor-phase reactions. The predicted chemical compositions of char and tar are described in Figs. S10–S19 in the 520
Fuel 212 (2018) 515–522
Y. Furutani et al.
Fig. 9. Comparison of the predicted yields of (a) char, (b) tar, and (c) gas for EHL, OEL, and KL at a temperature of 923 K with experimental data [49].
yields at 103 K/s in Fig. 7 were used as the initial species boundary. Fig. 8a–c shows a comparison of the predicted yields of CH3OH, C2H6, and C3H6 between with and without the DCKM of the secondary vaporphase reaction [13]. It is clear from Fig. 8a–c that the numerical predictions were improved especially at high temperature by considering the secondary vapor-phase reactions. Fig. 9a–c shows a comparison of the predicted char, tar, and gas yields for the three lignin samples (EHL, OEL, and KL) with experimental data at 923 K. The experimental yields included in Fig. 9 were obtained from TS-TR experiments with a residence time of 0.6 s for vapor-phase reactions [49]. Fig. 9a and b clearly indicate that the model predictions reproduce the experimentally observed trend for char and tar yields, with the char yield increasing in order of EHL < OEL < KL, whereas the tar yields decrease. However, a quantitative comparison could not be made because the effect of secondary vaporphase reactions could not be neglected at a residence time of 0.6 s.
Supplementary material. According to the data shown in Fig. 5, the heating rate has a small effect on the yields. At all temperatures our numerical results for char and H2O yields agreed well with the experimental measurements (Fig. 5a and e, respectively). However, the numerical simulation overestimated the tar yield and underestimated the CO yield at high temperature (Fig. 5b and c). These differences probably result from a lack of the kinetic model of the secondary vaporphase reaction, especially tar cracking reaction, which might not be negligible at high temperature. Thus, it is necessary to integrate the DCKM for the primary and secondary pyrolysis and to repeat the predictions for the yields of tar and CO. However, the kinetic parameters on the decomposition of monolignols being a major component of tar (Figs. S10–S14 in the Supplementary material) have not yet been investigated. At all temperatures the simulation underestimated CO2 yield (Fig. 5d). There are a number of possible reasons for this deviation. For instance, CO2 could be formed by the water-gas shift reaction (CO + H2O ⇌ CO2 + H2) in the vapor phase which are not participate in the current model. Further, hemicellulose with xylan units that can act as CO2 sources [56] might exist as an impurity in the sample. Other possible reason is that the equivalent composition of EHL did not include a PLIG-O component, which include a carboxyl group leading to CO2 formation, as shown in Table 1. Fig. 6 shows the change of the CO2 yield at 1223 K during the course of increasing the mass fraction of PLIG-O from 0.0 to 30.0 wt%. Because the CO2 yield increases with the increase of the mass fraction of PLIG-O, the addition of virtual compounds with the carboxyl group is expected to improve the prediction of the CO2 yield. Fig. 7a–c depicts the yields of methanol (CH3OH), ethane (C2H6), and propene (C3H6), respectively, which are minor products generated during EHL pyrolysis. As shown in Fig. 7, the semi-detailed kinetic model proposed by Hough et al. [47] does not exactly reproduce the yields of these minor products. The kinetic database for the thermal decomposition processes of CH3OH, C2H6, and C3H6 has been already included in the DCKM of secondary vapor phase reactions established by our group [13], although which does not include that of monolignols. By using this DCKM, numerical simulations of secondary vapor phase reactions of CH3OH, C2H6, and C3H6 were conducted at temperature of 773–1223 K and residence time of 0.1 s. The predicted
4. Conclusions The semi-detailed kinetic model of lignin primary pyrolysis proposed by Hough et al. [47] was used to predict the yield of char and volatile components (i.e., tar, CO, CO2, H2O, CH3OH, C2H6, and C3H6) obtained from EHL pyrolysis experiments in a TS-TR [13,49]. By solving the heat transfer equation, the heating rate of lignin particle in a TS-TR was estimated as 102–104 K/s. Numerical simulations were performed under various heating rates (102, 103, and 104 K/s) for temperatures of 773–1223 K. The numerical results for char and H2O yields were found to be in good agreement with the experimental results at all temperatures. However, the numerical simulation overestimated the tar yield and underestimated the CO yield at high temperatures, probably owing to the absence of secondary vapor-phase cracking reactions in the model. The predicted CO2 yield was improved by adding PLIG-O with the carboxyl group into the reference compounds. The predicted yields of minor products (CH3OH, C2H6, and C3H6) showed good correspondence with the experimental values at high temperature by adding the DCKM of secondary vapor-phase reactions [13] into the semi-detailed kinetic model of lignin primary pyrolysis [47]. The char and tar yields derived from EHL, OEL, and KL at 923 K agreed qualitatively, but not quantitatively, with the experimental data. In the 521
Fuel 212 (2018) 515–522
Y. Furutani et al.
[21] Li W, Law ME, Westmoreland PR, Kasper T, Hansen N, Kohse-Höinghaus K. Multiple benzene-formation paths in a fuel-rich cyclohexane flame. Combust Flame 2011;158:2077–89. [22] Kousoku A, Norinaga K, Miura K. Extended detailed chemical kinetic model for benzene pyrolysis with new reaction pathways including oligomer formation. Ind Eng Chem Res 2014;53:7956–64. [23] Norinaga K, Shoji T, Kudo S, Hayashi J-I. Detailed chemical kinetic modelling of vapour-phase cracking of multi-component molecular mixtures derived from the fast pyrolysis of cellulose. Fuel 2013;103:141–50. [24] Norinaga K, Sakurai Y, Sato R, Hayashi J-I. Numerical simulation of thermal conversion of aromatic hydrocarbons in the presence of hydrogen and steam using a detailed chemical kinetic model. Chem Eng J 2011;178:282–90. [25] Norinaga K, Hayashi J-I. Numerical simulation of the partial oxidation of hot coke oven gas with a detailed chemical kinetic model. Energy Fuels 2010;24:165–72. [26] Debiagi PEA, Gentile G, Pelucchi M, Frassoldati A, Cuoci A, Faravelli T, et al. Detailed kinetic mechanism of gas-phase reactions of volatiles released from biomass pyrolysis. Biomass Bioenergy 2016;93:60–71. [27] Furutani Y, Kudo S, Hayashi J-I, Norinaga K. Theoretical study on reaction pathways leading to CO and CO2 in the pyrolysis of resorcinol. J Phys Chem A 2017;121:631–7. [28] Altarawneh M, Dlugogorski BZ, Kennedy EM, Mackie JC. Theoretical study of unimolecular decomposition of catechol. J Phys Chem A 2010;114:1060–7. [29] Liu H, Chen J, Wang F, Wang Z, Wang L. Theoretical study on the thermodynamic properties and stability of polybrominated diphenyl sulfide catena. Acta Chim Sin 2010;68:540–50. [30] Altarawneh M, Dlugogorski BZ, Kennedy EM, Mackie JC. Thermochemical properties and decomposition pathways of three isomeric semiquinone radicals. J Phys Chem A 2010;114:1098–108. [31] Klein MT, Virk PS. Modeling of lignin thermolysis. Energy Fuels 2008;22:2175–82. [32] Nunn TR, Howard JB, Longwell JP, Peters WA. Product compositions and kinetics in the rapid pyrolysis of sweet gum hardwood. Ind Eng Chem Process Des Dev 1985;24:836–44. [33] Petrocelli FP, Klein MT. Model reaction pathways in Kraft lignin pyrolysis. Macromolecules 1984;17:161–9. [34] Caballero JA, Font R, Marcilla A. Study of the primary pyrolysis of Kraft lignin at high heating rates: yields and kinetics. J Anal Appl Pyrol 1996;36:159–78. [35] Pasquali CEL, Herrera H. Pyrolysis of lignin and IR analysis of residues. Thermochim Acta 1997;293:39–46. [36] Xiong Q, Zhang J, Xu F, Wiggins G, Daw CS. Coupling DAEM and CFD for simulating biomass fast pyrolysis in fluidized beds. J Anal Appl Pyrol 2016;117:176–81. [37] Xiong Q, Yang Y, Xu F, Pan Y, Zhang J, Hong K, et al. Overview of computational fluid dynamics simulation of reactor-scale biomass pyrolysis. ACS Sustain Chem Eng 2017;5:2783–98. [38] Aramideh S, Xiong Q, Kong S-C, Brown RC. Numerical simulation of biomass fast pyrolysis in an auger reactor. Fuel 2015;156:234–42. [39] Mellin P, Kantarelis E, Yang W. Computational fluid dynamics modeling of biomass fast pyrolysis in a fluidized bed reactor, using a comprehensive chemistry scheme. Fuel 2014;117:704–15. [40] Trendewicz A, Braun R, Dutta A, Ziegler J. One dimensional steady-state circulating fluidized-bed reactor model for biomass fast pyrolysis. Fuel 2014;133:253–62. [41] Xiong Q, Aramideh S, Kong S-C. Modeling effects of operating conditions on biomass fast pyrolysis in bubbling fluidized bed reactors. Energy Fuels 2013;27:5948–56. [42] Xiong Q, Aramideh S, Passalacqua A, Kong S-C. BIOTC: an open-source CFD code for simulating biomass fast pyrolysis. Comput Phys Commun 2014;185:1739–46. [43] Xiong Q, Kong S-C. High-resolution particle-scale simulation of biomass pyrolysis. ACS Sustain Chem Eng 2016;4:5456–61. [44] Xiong Q, Kong S-C. Modeling effects of interphase transport coefficients on biomass pyrolysis in fluidized beds. Powder Technol 2014;262:96–105. [45] Xiong Q, Kong S-C, Passalacqua A. Development of a generalized numerical framework for simulating biomass fast pyrolysis in fluidized-bed reactors. Chem Eng Sci 2013;99:305–13. [46] Faravelli T, Frassoldati A, Migliavacca G, Ranzi E. Detailed kinetic modeling of the thermal degradation of lignins. Biomass Bioenergy 2010;34:290–301. [47] Hough BR, Schwartz DT, Pfaendtner J. Detailed kinetic modeling of lignin pyrolysis for process optimization. Ind Eng Chem Res 2016;55:9147–53. [48] Jakab E, Faix O, Till F. Thermal decomposition of milled wood lignins studied by thermogravimetry/mass spectrometry. J Anal Appl Pyrol 1997;40–41:171–86. [49] Yang H, Appari S, Kudo S, Hayashi J-I, Kumagai S, Norinaga K. Chemical structures and primary pyrolysis characteristics of lignins obtained from different preparation methods. Nihon Enerugi Gakkaishi/J Jpn Inst Energy 2014;93:986–94. [50] Hage RE, Brosse N, Chrusciel L, Sanchez C, Sannigrahi P, Ragauskas A. Characterization of milled wood lignin and ethanol organosolv lignin from miscanthus. Polym Degrad Stab 2009;94:1632–8. [51] Eitelberger J, Hofstetter K. Prediction of transport properties of wood below the fiber saturation point – a multiscale homogenization approach and its experimental validation. Part I: Thermal conductivity. Compos Sci Technol 2011;71:134–44. [52] Hatakeyama T, Nakamura K, Hatakeyama H. Studies on heat capacity of cellulose and lignin by differential scanning calorimetry. Polymer 1982;23:1801–4. [53] Youssefian S, Rahbar N. Molecular origin of strength and stiffness in bamboo fibrils. Sci Rep 2015;5:11116. [54] Lou R, Wu SB. Products properties from fast pyrolysis of enzymatic/mild acidolysis lignin. Appl Energy 2011;88:316–22. [55] Sharma RK, Wooten JB, Baliga VL, Lin X, Chan WG, Hajaligol MR. Characterization of chars from pyrolysis of lignin. Fuel 2004;83:1469–82. [56] Shen DK, Gu S, Bridgwater AV. Study on the pyrolytic behaviour of xylan-based hemicellulose using TG-FTIR and Py-GC-FTIR. J Anal Appl Pyrol 2010;87:199–206.
future, the characterization of lignin structure based on IR or NMR spectra as well as elemental composition should be examined for adding the virtual compounds with the carboxyl group. Additionally, further work to obtain the kinetic parameters of monolignol decompositions based on ab initio calculations and transition state theory should be implemented for establishing the DCKM for predicting lignin pyrolysis behavior with both primary pyrolysis and secondary vapor-phase reactions. Acknowledgements This research was in part financially supported by KAKENHI (Grantin-Aid for Scientific Research (B): 17H03454). The authors are also grateful to the support by the Cooperative Research Program of “Network Joint Research Center for Materials and Devices”. All the computations in this study were performed on the PC cluster systems in our group and the high-performance computing system at the Research Institute for Information Technology, Kyushu University. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.fuel.2017.10.079. References [1] Rohella RS, Sahoo N, Paul SC, Choudhury S, Chakravortty V. Thermal studies on isolated and purified lignin. Thermochim Acta 1996;287:131–8. [2] Montané D, Torné-Fernández V, Fierro V. Activated carbons from lignin: kinetic modeling of the pyrolysis of Kraft lignin activated with phosphoric acid. Chem Eng J 2005;106:1–12. [3] Azadi P, Inderwildi OR, Farnood R, King DA. Liquid fuels, hydrogen and chemicals from lignin: a critical review. Renew Sustain Energy Rev 2013;21:506–23. [4] Davis KM, Rover M, Brown RC, Bai X, Wen Z, Jarboe LR. Recovery and utilization of lignin monomers as part of the biorefinery approach. Energies 2016;9:1–28. [5] Mu W, Ben H, Ragauskas A, Deng Y. Lignin pyrolysis components and upgradingtechnology review. Bioenergy Res 2013;6:1183–204. [6] Pandey MP, Kim CS. Lignin depolymerization and conversion: a review of thermochemical methods. Chem Eng Technol 2011;34:29–41. [7] Jegers HE, Klein MT. Primary and secondary lignin pyrolysis reaction pathways. Ind Eng Chem Process Des Dev 1985;24:173–83. [8] Hosoya T, Kawamoto H, Saka S. Secondary reactions of lignin-derived primary tar components. J Anal Appl Pyrol 2008;83:78–87. [9] Zhou S, Garcia-Perez M, Pecha B, McDonald AG, Kersten SRA, Westerhof RJM. Secondary vapor phase reactions of lignin-derived oligomers obtained by fast pyrolysis of pine wood. Energy Fuels 2013;27:1428–38. [10] Caballero JA, Font R, Marcilla A. Kinetic study of the secondary thermal decomposition of Kraft lignin. J Anal Appl Pyrol 1996;38:131–52. [11] Thimthong N, Appari S, Tanaka R, Iwanaga K, Kudo S, Hayashi J-I, et al. Kinetic modeling of non-catalytic partial oxidation of nascent volatiles derived from fast pyrolysis of woody biomass with detailed chemistry. Fuel Process Technol 2015;134:159–67. [12] Norinaga K, Yang H, Tanaka R, Appari S, Iwanaga K, Takashima Y, et al. A mechanistic study on the reaction pathways leading to benzene and naphthalene in cellulose vapor phase cracking. Biomass Bioenergy 2014;69:144–54. [13] Yang H, Appari S, Kudo S, Hayashi J-I, Norinaga K. Detailed chemical kinetic modeling of vapor-phase reactions of volatiles derived from fast pyrolysis of lignin. Ind Eng Chem Res 2015;54:6855–64. [14] Shoji T, Norinaga K, Mašek O, Hayashi J-I. Numerical simulation of secondary gas phase reactions of coffee grounds with a detailed chemical kinetic model. Nihon Enerugi Gakkaishi/J Jpn Inst Energy 2010;89:955–61. [15] Kohse-Höinghaus K, Oßwald P, Cool TA, Kasper T, Hansen N, Qi F, et al. Biofuel combustion chemistry: from ethanol to biodiesel. Angew Chem Int Ed 2010;49:3572–97. [16] Grana R, Frassoldati A, Faravelli T, Niemann U, Ranzi E, Seiser R, et al. An experimental and kinetic modeling study of combustion of isomers of butanol. Combust Flame 2010;157:2137–54. [17] Harper MR, Van Geem KM, Pyl SP, Marin GB, Green WH. Comprehensive reaction mechanism for n-butanol pyrolysis and combustion. Combust Flame 2011;158:16–41. [18] Black G, Curran HJ, Pichon S, Simmie JM, Zhukov V. Bio-butanol: combustion properties and detailed chemical kinetic model. Combust Flame 2010;157:363–73. [19] Labbe NJ, Seshadri V, Kasper T, Hansen N, Oßwald P, Westmoreland PR. Flame chemistry of tetrahydropyran as a model heteroatomic biofuel. Proc Combust Inst 2013;34:259–67. [20] Lucassen A, Labbe N, Westmoreland PR, Kohse-Höinghaus K. Combustion chemistry and fuel-nitrogen conversion in a laminar premixed flame of morpholine as a model biofuel. Combust Flame 2011;158:1647–66.
522
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Full Length Article
Predicting molecular composition of primary product derived from fast pyrolysis of lignin with semi-detailed kinetic model Yuki Furutania, Shinji Kudob, Jun-ichiro Hayashib,c, Koyo Norinagad,
MARK
⁎
a
Interdisciplinary Graduate School of Engineering Sciences, Kyushu University, 6-1, Kasuga-koen, Kasuga, Fukuoka 816-8580, Japan Institute for Materials Chemistry and Engineering, Kyushu University, 6-1, Kasuga-koen, Kasuga, Fukuoka 816-8580, Japan c Research and Education Center of Carbon Resources, Kyushu University, 6-1, Kasuga-koen, Kasuga, Fukuoka 816-8580, Japan d Graduate School of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan b
G RA P H I C A L AB S T R A C T
A R T I C L E I N F O
A B S T R A C T
Keywords: Fast pyrolysis Lignin Product yield Semi-detailed kinetic model
A numerical approach is presented for predicting the yields of char and volatile components obtained from fast pyrolysis of three types of lignin (enzymatic hydrolysis lignin, EHL; organic extracted lignin, OEL; and Klason lignin, KL) in a two-stage tubular reactor (TS-TR) at 773–1223 K. The heating rate of lignin particle in the TS-TR was estimated at 102–104 K/s by solving the heat transfer equation. The pyrolytic behavior of lignin and the formation of products in the temperature rising process were predicted using a semi-detailed kinetic model consisting of 93 species and 406 reactions, and the predicted yields of 8 primary products (i.e., char, tar, CO, CO2, H2O, CH3OH, C2H6, and C3H6) were compared with experimental data for the critical evaluation. For EHL, the predicted yields of char and H2O were in good agreement with the experimental results at all temperatures. However, the numerical simulation overestimated tar yield and underestimated CO yield at high temperature probably due to a lack of the kinetic model of the tar cracking reaction. The predicted yields of CH3OH, C2H6, and C3H6 were close to the experimental values at high temperature by adding the detailed chemical kinetic model of the secondary vapor-phase reaction. Moreover, the model reproduced the experimental observation that among the three types of lignin the char yield increased in the order of EHL < OEL < KL, whereas the tar yield decreased.
⁎
Corresponding author. E-mail address: [email protected] (K. Norinaga).
http://dx.doi.org/10.1016/j.fuel.2017.10.079 Received 13 July 2017; Received in revised form 20 September 2017; Accepted 16 October 2017 0016-2361/ © 2017 Elsevier Ltd. All rights reserved.
Fuel 212 (2018) 515–522
Y. Furutani et al.
Fig. 1. Virtual compounds used to characterize initial lignin structures [47].
Table 1 Elemental compositions and equivalent compositions of EHL, OEL, and KL.
EHL OEL KL
Elemental composition (wt%)
Equivalent composition (wt%)
C
H
O
PLIG-C
PLIG-O
PLIG-H
63.3 62.0 62.9
5.9 5.7 5.3
28.9 31.6 31.4
44.1 36.8 55.3
0.0 23.3 41.3
55.9 39.9 3.4
time of 0.1 s for vapor-phase reactions [11–13]. In order to enhance the versatility of the DCKM, it is necessary to expand the DCKM to include the primary pyrolysis stage. For primary pyrolysis of lignin, lumped kinetic models of the decomposition reaction have been extensively reported in the literature [31–35]. However, these models did not allow prediction of the molecular composition of gases and tar components. Recently, Xiong et al. [36] performed computational fluid dynamics (CFD) simulations coupled with distributed activation energy mode (DAEM) reaction kinetics for lab-scale bubbling bed biomass pyrolysis reactor, and revealed that the coupled CFD–DAEM system does not significantly increase computational overhead. A reliable kinetic model, which predict the yields and the molecular composition of gas and tar as accurately as possible, has been required with the development of highly efficient CFD methods [36–45] and the increase of computational resource. Faravelli et al. [46] explored a semi-detailed kinetic model, which characterizes lignin structures using three virtual compounds and involves approximately 100 species and 400 reactions, to predict the molecular compositions of products derived from primary pyrolysis. Recently, Hough et al. [47] added eight reactions into the kinetic models established by Faravelli et al. [46], and compared the integral yields of char, tar, and gases predicted by the model and observed for slow pyrolysis experiments by thermogravimetric analysis (TGA) [48]. However, the model predictions have not yet been compared with the molecular compositions of volatiles derived from fast pyrolysis experiments, such as TS-TR experiments. This comparison is also an important step to optimize fast pyrolysis process, which generates much amount of volatile products from biomass and produces “bio-fuel” [5]. The purpose of this study is to examine whether the semi-detailed kinetic model established by Hough et al. [47] could reproduce the molecular composition of the primary products generated from fast pyrolysis with TS-TR experiments. First, characterization of lignin structures was described using virtual compounds. Second, the heating rate for lignin samples in the primary pyrolysis zone of the TS-TR was estimated. Finally, based on the estimated heating rate, the yields of char and volatiles were predicted using the semi-detailed kinetic model and compared with TS-TR experimental results [13] for the critical evaluation. This estimation would help to integrate the DCKM of both primary pyrolysis and secondary vapor-phase reactions.
Fig. 2. Elemental compositions of the six lignins.
1. Introduction After cellulose, lignin is the second most abundant component in biomass [1,2]. As a residue of the pulp and paper industry, huge amounts of lignin are available at low cost [2]. Although currently most lignin is burned to produce heat, thermal conversion processes, including pyrolysis, gasification and liquefaction, can be used to convert lignin or biomass into useful products, such as gas, char, liquid fuels and chemicals as well as heat [3–6]. Among these thermal conversion processes, pyrolysis is known as a common step to cause fragmentation of the lignin or biomass structure [7]. Pyrolysis is divided into two stages: (a) primary pyrolysis, where volatiles escape from biomass particles; and (b) secondary vapor-phase reactions, where the produced volatiles undergo further cracking, combine, or condense in the vapor phase. A lot of studies have been carried out to identify the pyrolysis products generated from secondary vapor-phase reactions and to establish lumped kinetic models [7–10]. Caballero et al. [10] established a lumped kinetic model for the global secondary reaction of Kraft lignin by assuming a first-order reaction. However, lumped kinetic models established based on global product categories, such as char, tar, and gases, cannot describe the formation mechanisms for specific products at the molecular level. A detailed chemical kinetic model (DCKM) of vapor-phase reactions based on elementary reactions [11–30] has been developed to overcome the limitations of the lumping approach and provide information on the pyrolysis behaviors of individual components. Our group revealed that the DCKM was able to reproduce not only the yields of major products but also those of minor products such as aromatic hydrocarbons, which were obtained with a two-stage tubular reactor (TSTR) connected to a gas chromatograph (GC) [11–13]. However, the DCKM has been limited to secondary vapor-phase reactions. Thus, the molecular composition of the volatiles derived from fast pyrolysis has to be obtained as a boundary condition with the TS-TR setting a residence 516
Fuel 212 (2018) 515–522
Y. Furutani et al.
outside the bounds of the triangle, the mass fraction of PLIG-O had a negative value. In this case, the mass fraction of PLIG-O was set to zero and the values of the other two components (i.e. PLIG-C and PLIG-H) were normalized to express the equivalent composition of EHL. Table 1 lists the elemental compositions and the equivalent compositions of EHL, OEL, and KL. 2.2. Estimation of heating rate No direct measurement of the sample heating rate has been made. Here, a numerical model for briefly estimating the heating rate and the results is presented. By assuming that the sample has a spherical shape, the temperature history (T) along the radial direction (r) in the lignin sample was estimated based on the following heat transfer equation:
∂T α ∂ ∂T = 2 ⎛r 2 ⎞ ∂t r ∂r ⎝ ∂r ⎠
Fig. 3. Average temperature profiles (Tav [K]) as a function of time (t [s]) at Ts = 773–1223 K.
(1)
The average temperature (Tav) in particle was expressed as:
2. Methods and modeling
Tav = 2.1. Virtual compounds for approximating lignin structures
1 R
∫0
R
Tdr
(2)
The following assumptions were made for this calculation. A detailed primary pyrolysis analysis of three types of lignin (enzymatic hydrolysis lignin, EHL; organic extracted lignin, OEL; and Klason lignin, KL) with TS-TR was reported by Yang et al. [13,49], and thus these lignins were employed in this study. We followed the approach of Hough et al. [47] by choosing virtual compounds to approximate various possible lignin structures. Three reference components based on a β-O-4 skeleton [50] (PLIG-C, PLIG-H, and PLIG-O, following the naming conventions of Hough et al. [47]) as shown in Fig. 1 were selected to characterize the structures of EHL, OEL, and KL. Fig. 2 gives a synoptic view of the elemental compositions of the six lignins (PLIG-C, PLIG-H, PLIG-O, EHL, OEL, and KL). The elemental compositions of OEL and KL was inside the bounds of a triangle whose vertices were the three reference compounds. Thus, the equivalent compositions of OEL and KL can be described in terms of the three reference compounds using a linear relationship based on mass conservation. On the other hand, as the elemental composition of EHL was
1. The thermal diffusivity (α) and particle size (R) are 1.83 × 10−7 m2/s and 0.56 mm, which are calculated by solving the following equation:
α=
k ρCp
4 m = ρ πR3 3
(3)
(4)
where the thermal conductivity (k), heat capacity (Cp), density (ρ), and mass (m) are 0.39 W/m/K [51], 1.60 J/g/K [52], 1.33 g/cm3 [53] and 1.00 mg [13], respectively. 2. The heat of reaction is negligibly small. 3. The initial temperature in the particle is 300 K. Fig. 4. Yields of (a) reactants, (b) main intermediates, main products of (c) char and (d) volatile obtained by heating at 103 K/s–1223 K. The chemical structures of intermediates and products are shown in Table 2.
517
Fuel 212 (2018) 515–522
Y. Furutani et al.
Table 2 Main components of (a) intermediates and (b) products (following the naming conventions of Faravelli et al. [46]).
(a)
(b)
518
Fuel 212 (2018) 515–522
Y. Furutani et al.
Fig. 5. Comparison of the predicted yields of (a) char, (b) tar, (c) CO, (d) CO2 and (e) H2O with experimental data [13].
Yield (wt%)
10 8
time, as derived from Eq. (2). According to the temperature curves in Fig. 3, the range of heating rate was estimated as 102–104 K/s.
102 [K/s] 103 [K/s] 104 [K/s] Experiment
2.3. Detailed chemical kinetic model for lignin pyrolysis
6
The semi-detailed kinetic model proposed by Hough et al. [47] was used to predict the molecular composition of the primary products generated from fast pyrolysis with TS-TR experiments [13]. Details of the reactions and sources of kinetic parameters are available in Hough et al.’s article [47]. DETCHEMBATCH code [11,12] was used to run the proposed kinetic models. Numerical simulations were performed under various heating rates (102, 103, and 104 K/s). After heating up to the experimentally-designated temperature (773–1223 K), the temperature was held isothermally until the char yield was constant. Fig. 4a–d shows the predicted thermal degradation behavior of EHL and product formation at 103 K/s heated up to 1223 K. Table 2 lists the chemical structures of the intermediates and products. Fig. 4a and b revealed that above 700 K the intermediate components of LIGH and LIGC were formed immediately upon degradation of PLIG-C and PLIGH. As shown in Fig. 4c and d, above 1000 K aromatic carbons (C10H2), VKETDM2, and H2O were generated as the main char, tar and gas products, respectively. Here, volatile species that were heavier than benzene were defined as tar, otherwise done as gas. The formation of aromatic carbons in char well reproduced the product characteristics revealed by temperature-dependent nuclear magnetic (NMR) spectra [54,55]. Similarly, Figs. S3–S9 in Supplementary material shows the predicted thermal degradation behavior of three lignins (EHL, OEL and
4 2 0
0 wt% 10 wt% 20 wt% 30 wt%
Fig. 6. The predicted change of the CO2 yield at 1223 K along with a fixed ratio of PLIG-C and PLIG-H and the mass fraction of PLIG-O increasing from 0.0 to 30.0 wt%. The experimental data [13] was also shown to compare with the numerical predictions.
4. When the particle is dropped into the bottom of the first isothermal zone (the primary pyrolysis zone) [13,49], the surface of the particle is exposed to the experimentally-defined pyrolysis temperature. Thus, the temperature at the wall (Ts) is fixed at 773–1223 K. Using these four assumptions, Eqs. (1) and (2) were solved by the finite difference method, as described in the Supplementary material. Fig. 3 shows the average temperature profile (Tav) as a function of 519
Fuel 212 (2018) 515–522
Y. Furutani et al.
2.5 2.0 1.5 1.0
Yield (wt%)
2.0
(b)
Fig. 7. Comparison of the predicted yields of (a) CH3OH, (b) C2H6, and (c) C3H6 with experimental data [13].
0.8 0.6 0.4 0.2
0.5 0.0
1.0
(a)
Yield (wt%)
Yield (wt%)
3.0
773K 923K 1023K 1123K 1223K
0.0
773K 923K 1023K 1123K 1223K
(c)
1.6 1.2 0.8 0.4 0.0
773K 923K 1023K 1123K 1223K
Fig. 8. Yields of (a) CH3OH, (b) C2H6, and (c) C3H6 predicted with and without DCKM of secondary vapor-phase reaction [13]. The experimental data [13] was also shown to compare with the numerical predictions.
3. Results and discussion
KL) and product formation under various temperature conditions. Experimental data obtained with the TS-TR connected to GC [12,13] was used to compare with numerical predictions. The TS-TR consists of two zones: (a) first zone for the primary pyrolysis, and (b) secondary zone for the vapor-phase reactions of volatiles. The sample was dropped into the bottom of the first zone, and then the generated volatiles were flown into the secondary zone with helium carrier gas. After passing through the secondary zone, the volatiles were directed into GC column and further identified by GC detector.
To critically evaluate the semi-detailed kinetic model for fast pyrolysis of lignin samples, the predicted yields were compared with experimental data. Fig. 5 shows a comparison between the product yields (char, tar, CO, CO2 and H2O) at the terminal points in Figs. S3–S7 and the experimental data [13] obtained using a TS-TR with a residence time of 0.1 s for vapor-phase reactions. The predicted chemical compositions of char and tar are described in Figs. S10–S19 in the 520
Fuel 212 (2018) 515–522
Y. Furutani et al.
Fig. 9. Comparison of the predicted yields of (a) char, (b) tar, and (c) gas for EHL, OEL, and KL at a temperature of 923 K with experimental data [49].
yields at 103 K/s in Fig. 7 were used as the initial species boundary. Fig. 8a–c shows a comparison of the predicted yields of CH3OH, C2H6, and C3H6 between with and without the DCKM of the secondary vaporphase reaction [13]. It is clear from Fig. 8a–c that the numerical predictions were improved especially at high temperature by considering the secondary vapor-phase reactions. Fig. 9a–c shows a comparison of the predicted char, tar, and gas yields for the three lignin samples (EHL, OEL, and KL) with experimental data at 923 K. The experimental yields included in Fig. 9 were obtained from TS-TR experiments with a residence time of 0.6 s for vapor-phase reactions [49]. Fig. 9a and b clearly indicate that the model predictions reproduce the experimentally observed trend for char and tar yields, with the char yield increasing in order of EHL < OEL < KL, whereas the tar yields decrease. However, a quantitative comparison could not be made because the effect of secondary vaporphase reactions could not be neglected at a residence time of 0.6 s.
Supplementary material. According to the data shown in Fig. 5, the heating rate has a small effect on the yields. At all temperatures our numerical results for char and H2O yields agreed well with the experimental measurements (Fig. 5a and e, respectively). However, the numerical simulation overestimated the tar yield and underestimated the CO yield at high temperature (Fig. 5b and c). These differences probably result from a lack of the kinetic model of the secondary vaporphase reaction, especially tar cracking reaction, which might not be negligible at high temperature. Thus, it is necessary to integrate the DCKM for the primary and secondary pyrolysis and to repeat the predictions for the yields of tar and CO. However, the kinetic parameters on the decomposition of monolignols being a major component of tar (Figs. S10–S14 in the Supplementary material) have not yet been investigated. At all temperatures the simulation underestimated CO2 yield (Fig. 5d). There are a number of possible reasons for this deviation. For instance, CO2 could be formed by the water-gas shift reaction (CO + H2O ⇌ CO2 + H2) in the vapor phase which are not participate in the current model. Further, hemicellulose with xylan units that can act as CO2 sources [56] might exist as an impurity in the sample. Other possible reason is that the equivalent composition of EHL did not include a PLIG-O component, which include a carboxyl group leading to CO2 formation, as shown in Table 1. Fig. 6 shows the change of the CO2 yield at 1223 K during the course of increasing the mass fraction of PLIG-O from 0.0 to 30.0 wt%. Because the CO2 yield increases with the increase of the mass fraction of PLIG-O, the addition of virtual compounds with the carboxyl group is expected to improve the prediction of the CO2 yield. Fig. 7a–c depicts the yields of methanol (CH3OH), ethane (C2H6), and propene (C3H6), respectively, which are minor products generated during EHL pyrolysis. As shown in Fig. 7, the semi-detailed kinetic model proposed by Hough et al. [47] does not exactly reproduce the yields of these minor products. The kinetic database for the thermal decomposition processes of CH3OH, C2H6, and C3H6 has been already included in the DCKM of secondary vapor phase reactions established by our group [13], although which does not include that of monolignols. By using this DCKM, numerical simulations of secondary vapor phase reactions of CH3OH, C2H6, and C3H6 were conducted at temperature of 773–1223 K and residence time of 0.1 s. The predicted
4. Conclusions The semi-detailed kinetic model of lignin primary pyrolysis proposed by Hough et al. [47] was used to predict the yield of char and volatile components (i.e., tar, CO, CO2, H2O, CH3OH, C2H6, and C3H6) obtained from EHL pyrolysis experiments in a TS-TR [13,49]. By solving the heat transfer equation, the heating rate of lignin particle in a TS-TR was estimated as 102–104 K/s. Numerical simulations were performed under various heating rates (102, 103, and 104 K/s) for temperatures of 773–1223 K. The numerical results for char and H2O yields were found to be in good agreement with the experimental results at all temperatures. However, the numerical simulation overestimated the tar yield and underestimated the CO yield at high temperatures, probably owing to the absence of secondary vapor-phase cracking reactions in the model. The predicted CO2 yield was improved by adding PLIG-O with the carboxyl group into the reference compounds. The predicted yields of minor products (CH3OH, C2H6, and C3H6) showed good correspondence with the experimental values at high temperature by adding the DCKM of secondary vapor-phase reactions [13] into the semi-detailed kinetic model of lignin primary pyrolysis [47]. The char and tar yields derived from EHL, OEL, and KL at 923 K agreed qualitatively, but not quantitatively, with the experimental data. In the 521
Fuel 212 (2018) 515–522
Y. Furutani et al.
[21] Li W, Law ME, Westmoreland PR, Kasper T, Hansen N, Kohse-Höinghaus K. Multiple benzene-formation paths in a fuel-rich cyclohexane flame. Combust Flame 2011;158:2077–89. [22] Kousoku A, Norinaga K, Miura K. Extended detailed chemical kinetic model for benzene pyrolysis with new reaction pathways including oligomer formation. Ind Eng Chem Res 2014;53:7956–64. [23] Norinaga K, Shoji T, Kudo S, Hayashi J-I. Detailed chemical kinetic modelling of vapour-phase cracking of multi-component molecular mixtures derived from the fast pyrolysis of cellulose. Fuel 2013;103:141–50. [24] Norinaga K, Sakurai Y, Sato R, Hayashi J-I. Numerical simulation of thermal conversion of aromatic hydrocarbons in the presence of hydrogen and steam using a detailed chemical kinetic model. Chem Eng J 2011;178:282–90. [25] Norinaga K, Hayashi J-I. Numerical simulation of the partial oxidation of hot coke oven gas with a detailed chemical kinetic model. Energy Fuels 2010;24:165–72. [26] Debiagi PEA, Gentile G, Pelucchi M, Frassoldati A, Cuoci A, Faravelli T, et al. Detailed kinetic mechanism of gas-phase reactions of volatiles released from biomass pyrolysis. Biomass Bioenergy 2016;93:60–71. [27] Furutani Y, Kudo S, Hayashi J-I, Norinaga K. Theoretical study on reaction pathways leading to CO and CO2 in the pyrolysis of resorcinol. J Phys Chem A 2017;121:631–7. [28] Altarawneh M, Dlugogorski BZ, Kennedy EM, Mackie JC. Theoretical study of unimolecular decomposition of catechol. J Phys Chem A 2010;114:1060–7. [29] Liu H, Chen J, Wang F, Wang Z, Wang L. Theoretical study on the thermodynamic properties and stability of polybrominated diphenyl sulfide catena. Acta Chim Sin 2010;68:540–50. [30] Altarawneh M, Dlugogorski BZ, Kennedy EM, Mackie JC. Thermochemical properties and decomposition pathways of three isomeric semiquinone radicals. J Phys Chem A 2010;114:1098–108. [31] Klein MT, Virk PS. Modeling of lignin thermolysis. Energy Fuels 2008;22:2175–82. [32] Nunn TR, Howard JB, Longwell JP, Peters WA. Product compositions and kinetics in the rapid pyrolysis of sweet gum hardwood. Ind Eng Chem Process Des Dev 1985;24:836–44. [33] Petrocelli FP, Klein MT. Model reaction pathways in Kraft lignin pyrolysis. Macromolecules 1984;17:161–9. [34] Caballero JA, Font R, Marcilla A. Study of the primary pyrolysis of Kraft lignin at high heating rates: yields and kinetics. J Anal Appl Pyrol 1996;36:159–78. [35] Pasquali CEL, Herrera H. Pyrolysis of lignin and IR analysis of residues. Thermochim Acta 1997;293:39–46. [36] Xiong Q, Zhang J, Xu F, Wiggins G, Daw CS. Coupling DAEM and CFD for simulating biomass fast pyrolysis in fluidized beds. J Anal Appl Pyrol 2016;117:176–81. [37] Xiong Q, Yang Y, Xu F, Pan Y, Zhang J, Hong K, et al. Overview of computational fluid dynamics simulation of reactor-scale biomass pyrolysis. ACS Sustain Chem Eng 2017;5:2783–98. [38] Aramideh S, Xiong Q, Kong S-C, Brown RC. Numerical simulation of biomass fast pyrolysis in an auger reactor. Fuel 2015;156:234–42. [39] Mellin P, Kantarelis E, Yang W. Computational fluid dynamics modeling of biomass fast pyrolysis in a fluidized bed reactor, using a comprehensive chemistry scheme. Fuel 2014;117:704–15. [40] Trendewicz A, Braun R, Dutta A, Ziegler J. One dimensional steady-state circulating fluidized-bed reactor model for biomass fast pyrolysis. Fuel 2014;133:253–62. [41] Xiong Q, Aramideh S, Kong S-C. Modeling effects of operating conditions on biomass fast pyrolysis in bubbling fluidized bed reactors. Energy Fuels 2013;27:5948–56. [42] Xiong Q, Aramideh S, Passalacqua A, Kong S-C. BIOTC: an open-source CFD code for simulating biomass fast pyrolysis. Comput Phys Commun 2014;185:1739–46. [43] Xiong Q, Kong S-C. High-resolution particle-scale simulation of biomass pyrolysis. ACS Sustain Chem Eng 2016;4:5456–61. [44] Xiong Q, Kong S-C. Modeling effects of interphase transport coefficients on biomass pyrolysis in fluidized beds. Powder Technol 2014;262:96–105. [45] Xiong Q, Kong S-C, Passalacqua A. Development of a generalized numerical framework for simulating biomass fast pyrolysis in fluidized-bed reactors. Chem Eng Sci 2013;99:305–13. [46] Faravelli T, Frassoldati A, Migliavacca G, Ranzi E. Detailed kinetic modeling of the thermal degradation of lignins. Biomass Bioenergy 2010;34:290–301. [47] Hough BR, Schwartz DT, Pfaendtner J. Detailed kinetic modeling of lignin pyrolysis for process optimization. Ind Eng Chem Res 2016;55:9147–53. [48] Jakab E, Faix O, Till F. Thermal decomposition of milled wood lignins studied by thermogravimetry/mass spectrometry. J Anal Appl Pyrol 1997;40–41:171–86. [49] Yang H, Appari S, Kudo S, Hayashi J-I, Kumagai S, Norinaga K. Chemical structures and primary pyrolysis characteristics of lignins obtained from different preparation methods. Nihon Enerugi Gakkaishi/J Jpn Inst Energy 2014;93:986–94. [50] Hage RE, Brosse N, Chrusciel L, Sanchez C, Sannigrahi P, Ragauskas A. Characterization of milled wood lignin and ethanol organosolv lignin from miscanthus. Polym Degrad Stab 2009;94:1632–8. [51] Eitelberger J, Hofstetter K. Prediction of transport properties of wood below the fiber saturation point – a multiscale homogenization approach and its experimental validation. Part I: Thermal conductivity. Compos Sci Technol 2011;71:134–44. [52] Hatakeyama T, Nakamura K, Hatakeyama H. Studies on heat capacity of cellulose and lignin by differential scanning calorimetry. Polymer 1982;23:1801–4. [53] Youssefian S, Rahbar N. Molecular origin of strength and stiffness in bamboo fibrils. Sci Rep 2015;5:11116. [54] Lou R, Wu SB. Products properties from fast pyrolysis of enzymatic/mild acidolysis lignin. Appl Energy 2011;88:316–22. [55] Sharma RK, Wooten JB, Baliga VL, Lin X, Chan WG, Hajaligol MR. Characterization of chars from pyrolysis of lignin. Fuel 2004;83:1469–82. [56] Shen DK, Gu S, Bridgwater AV. Study on the pyrolytic behaviour of xylan-based hemicellulose using TG-FTIR and Py-GC-FTIR. J Anal Appl Pyrol 2010;87:199–206.
future, the characterization of lignin structure based on IR or NMR spectra as well as elemental composition should be examined for adding the virtual compounds with the carboxyl group. Additionally, further work to obtain the kinetic parameters of monolignol decompositions based on ab initio calculations and transition state theory should be implemented for establishing the DCKM for predicting lignin pyrolysis behavior with both primary pyrolysis and secondary vapor-phase reactions. Acknowledgements This research was in part financially supported by KAKENHI (Grantin-Aid for Scientific Research (B): 17H03454). The authors are also grateful to the support by the Cooperative Research Program of “Network Joint Research Center for Materials and Devices”. All the computations in this study were performed on the PC cluster systems in our group and the high-performance computing system at the Research Institute for Information Technology, Kyushu University. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.fuel.2017.10.079. References [1] Rohella RS, Sahoo N, Paul SC, Choudhury S, Chakravortty V. Thermal studies on isolated and purified lignin. Thermochim Acta 1996;287:131–8. [2] Montané D, Torné-Fernández V, Fierro V. Activated carbons from lignin: kinetic modeling of the pyrolysis of Kraft lignin activated with phosphoric acid. Chem Eng J 2005;106:1–12. [3] Azadi P, Inderwildi OR, Farnood R, King DA. Liquid fuels, hydrogen and chemicals from lignin: a critical review. Renew Sustain Energy Rev 2013;21:506–23. [4] Davis KM, Rover M, Brown RC, Bai X, Wen Z, Jarboe LR. Recovery and utilization of lignin monomers as part of the biorefinery approach. Energies 2016;9:1–28. [5] Mu W, Ben H, Ragauskas A, Deng Y. Lignin pyrolysis components and upgradingtechnology review. Bioenergy Res 2013;6:1183–204. [6] Pandey MP, Kim CS. Lignin depolymerization and conversion: a review of thermochemical methods. Chem Eng Technol 2011;34:29–41. [7] Jegers HE, Klein MT. Primary and secondary lignin pyrolysis reaction pathways. Ind Eng Chem Process Des Dev 1985;24:173–83. [8] Hosoya T, Kawamoto H, Saka S. Secondary reactions of lignin-derived primary tar components. J Anal Appl Pyrol 2008;83:78–87. [9] Zhou S, Garcia-Perez M, Pecha B, McDonald AG, Kersten SRA, Westerhof RJM. Secondary vapor phase reactions of lignin-derived oligomers obtained by fast pyrolysis of pine wood. Energy Fuels 2013;27:1428–38. [10] Caballero JA, Font R, Marcilla A. Kinetic study of the secondary thermal decomposition of Kraft lignin. J Anal Appl Pyrol 1996;38:131–52. [11] Thimthong N, Appari S, Tanaka R, Iwanaga K, Kudo S, Hayashi J-I, et al. Kinetic modeling of non-catalytic partial oxidation of nascent volatiles derived from fast pyrolysis of woody biomass with detailed chemistry. Fuel Process Technol 2015;134:159–67. [12] Norinaga K, Yang H, Tanaka R, Appari S, Iwanaga K, Takashima Y, et al. A mechanistic study on the reaction pathways leading to benzene and naphthalene in cellulose vapor phase cracking. Biomass Bioenergy 2014;69:144–54. [13] Yang H, Appari S, Kudo S, Hayashi J-I, Norinaga K. Detailed chemical kinetic modeling of vapor-phase reactions of volatiles derived from fast pyrolysis of lignin. Ind Eng Chem Res 2015;54:6855–64. [14] Shoji T, Norinaga K, Mašek O, Hayashi J-I. Numerical simulation of secondary gas phase reactions of coffee grounds with a detailed chemical kinetic model. Nihon Enerugi Gakkaishi/J Jpn Inst Energy 2010;89:955–61. [15] Kohse-Höinghaus K, Oßwald P, Cool TA, Kasper T, Hansen N, Qi F, et al. Biofuel combustion chemistry: from ethanol to biodiesel. Angew Chem Int Ed 2010;49:3572–97. [16] Grana R, Frassoldati A, Faravelli T, Niemann U, Ranzi E, Seiser R, et al. An experimental and kinetic modeling study of combustion of isomers of butanol. Combust Flame 2010;157:2137–54. [17] Harper MR, Van Geem KM, Pyl SP, Marin GB, Green WH. Comprehensive reaction mechanism for n-butanol pyrolysis and combustion. Combust Flame 2011;158:16–41. [18] Black G, Curran HJ, Pichon S, Simmie JM, Zhukov V. Bio-butanol: combustion properties and detailed chemical kinetic model. Combust Flame 2010;157:363–73. [19] Labbe NJ, Seshadri V, Kasper T, Hansen N, Oßwald P, Westmoreland PR. Flame chemistry of tetrahydropyran as a model heteroatomic biofuel. Proc Combust Inst 2013;34:259–67. [20] Lucassen A, Labbe N, Westmoreland PR, Kohse-Höinghaus K. Combustion chemistry and fuel-nitrogen conversion in a laminar premixed flame of morpholine as a model biofuel. Combust Flame 2011;158:1647–66.
522