292
Energy & Fuels 2005, 19, 292-297
Biomass Combustion Modeling in Fluidized Beds Panagiotis Grammelis* and Emmanuel Kakaras Laboratory of Steam Boilers and Thermal Plants, Mechanical Engineering Department, National Technical University of Athens, Athens, Greece Received July 8, 2004
A numerical code to calculate the burn-out time of a single biomass particle in a fluidized bed reactor was developed. Combustion modeling was performed by numerical calculation of energy and mass balances on the particle. Kinetic data for pyrolysis and char combustion stages, as determined by thermogravimetric analysis, were used. The results showed that the time needed for the completion of each partial combustion stage is less for biomass than for brown coal. The same remarks hold for all biomass species studied, whereas the predictions of the simulation model were in agreement with the results of the efforts of other researchers. The effect of uncertain parameters on model predictions was investigated, and it was proved that the particle diameter, moisture content, oxygen concentration, and especially the gas temperature have the greater influence on the calculated results. The combined use of the experimental data and the theoretical calculations provide a complete view of biomass/waste combustion behavior, which is absolutely necessary for the design and operation of large-scale units.
Introduction
Model Description
Computational fluid dynamics (CFD) codes are applied to simulate the combustion conditions inside the furnace and predict any potential implications during boiler operation. The information coming from the combustion submodels that are incorporated in CFD codes is important, because any inaccuracies will be carried over in subsequent calculations and results. Furthermore, there is a great variety in biomass/waste composition, which is a significant parameter when determining the design feedstock of the plant, demanding in-depth knowledge. Therefore, the development of a simple-to-use methodology that provides detailed, inadvance insight of biomass/waste combustion behavior, in great accuracy, is absolutely necessary to facilitate potential applications of these secondary fuels. The overall aim of this paper was to investigate whether a mathematical model using kinetic parameters from thermogravimetric analysis (TGA) can sufficiently predict the main characteristics of single biomass particle combustion in fluidized beds, and especially the burn-out time. Pyrolysis and char combustion tests in TGA provide useful information for the thermochemical conversion of biomass, in an easy and cost-effective way. At the same time, reaction kinetics for each of the two subprocesses can be determined using simple reaction models. The kinetic parameters may be adapted in a comprehensive combustion model that is used to calculate total burn-out times and other values that characterize biomass particle conversion. In this paper, the predictions of the mathematical model are described, whereas the results of the experiments that involve thermogravimetric studies and kinetic evaluation have been presented elsewhere.1-3
Combustion modeling of a biomass particle is based on the numerical solution of heat and mass conservation equations for the successive combustion stages, i.e., heating at low temperatures, drying, devolatilization, and volatiles combustion, followed by the burning of the residual char. A numerical code was developed to solve the differential equations that have been established to predict the burn-out time of a single secondary fuel (biomass) or coal particle, as well as the variation of particle characteristics, i.e., temperature, diameter, and density, in relation to time. During the model development, it was assumed that the particles are spherical and isothermal, the temperature of surrounding gases is constant, and the pyrolysis and combustion phenomena occur subsequently. The selection of a reaction scheme with proper rate constants that satisfactorily predicts the ultimate product yields distribution was the most difficult task in modeling the thermal degradation process of a biomass/ waste particle. The reaction schemes used both for the devolatilization and char combustion stages are described below. Devolatilization. The most applied pyrolysis model is the one-step global model in which the dry solid (S) decomposes into volatiles (V) and char (C):4-7
* Author to whom correspondence should be addressed. Phone: +30210-772.28.65. Fax: +30-210-772.36.63. E-mail:
[email protected].
k1
S 98 V + C
(1)
(1) Vamvuka, D.; Kakaras, E.; Kastanaki, Ε.; Grammelis, P. Fuel 2003, 82, 1949-1960. (2) Vamvuka, D.; Pasadakis, N.; Kastanaki, E.; Grammelis, P.; Kakaras, E. Energy Fuels 2003, 17 (3), 549-558. (3) Kastanaki, E.; Grammelis, P.; Vamvuka, D.; Kakaras, E., A Comparison of the Combustion Reactivity of Biomass, Hard Coal and Lignite Chars, Presented at the 1st European Combustion Meeting, Orleans, France, October 25-28, 2003. (4) Gronli, M. A Theoretical and Experimental Study of the Thermal Degradation of Biomass, Ph.D. Thesis, NTNU, Norway, 1996. (5) Gavalas, G. R. Coal Pyrolysis; Coal Science and Technology, Vol. 4: Elsevier: New York, 1982.
10.1021/ef049838o CCC: $30.25 © 2005 American Chemical Society Published on Web 12/03/2004
Biomass Combustion Modeling in Fluidized Beds
Energy & Fuels, Vol. 19, No. 1, 2005 293
Table 1. Proximate and Ultimate Analyses, and High Heating Value (as Received) of Biomass/Waste Species and Megalopolis Brown Coal Proximate Analysis (wt %)a
Ultimate Analysis (wt %, dry)
sample
moisture
volatiles
fixed carbon
ash
C
H
N
O
S
high heating value, HHV (MJ/kg)a
olive kernel cotton residues forest residues waste wood wood pellets Megalopolis brown coal
7.53 9.99 7.32 28.28 6.98 60.8
67.17 65.53 73.96 52.90 71.00 19.29
22.91 18.53 18.54 16.46 20.01 6.41
2.39 5.95 0.19 2.36 2.01 13.5
51.19 47.03 53.16 45.68 49.60 28.52
6.06 5.96 6.25 6.07 6.58 3.07
0.76 1.79 0.30 4.51 4.19 1.19
39.32 38.42 40.0 40.32 37.35 29.26
0.09 0.19 0.09 0.13 0.12 3.47
18.9 16.9 19.5 12.6 17.8 6.1
a
As received.
As an improvement to the one-step global model with a single rate constant, many researchers have used more-complex reaction schemes with parallel, consecutive, and/or competitive reactions. Several multistep pyrolysis models have been developed that take into account different groups of thermal decomposition reactions. The last and most-complicated approach, the distributed activation energy model, is applied for coal particles and assumes that devolatilization occurs through several first-order reactions that occur simultaneously. This model is characterized by the same preexponential factor and an activation energy following a distribution function f(E).4,5,8,9 With respect to biomass compounds, several decomposition zones can be distinguished in the volatiles release stage:4 Zone I (373-523 K). In this zone, extractives start to decompose. Zone II (523-623 K). The predominant reaction that occurs in this zone is hemicellulose decomposition. Zone III (623-773 K). In this zone, mainly cellulose and lignin decomposition occurs. Zone IV (>773 K). Mainly, lignin decomposition occurs in this zone. Assuming that each main constituent of biomass decomposes independently from each other, the consumption of the solid material then is calculated as the sum of the individual consumption of component i, according to
∂ ∂t
〈Fp〉 )
∑ ∂t 〈F ∂
p,i〉
∑ k (〈F
)-
i
p,i〉
- 〈FC∞,i〉)
(2)
where the reaction rate constants are calculated by the Arrhenius equation. More-complicated reaction schemes of biomass pyrolysis, which include the secondary reactions of tar to provide gases, have been proposed by other researchers.9-11 The approach of the separate decomposition of biomass main components was used in this mathematical model, because devolatilization is considered substage of the combustion process. In the numerical solution for the energy conservation equation, the heat storage and the cooling effect of the vapor and volatiles efflux are taken into account. Volatiles combustion is assumed to occur in the gas phase, in a thin boundary layer around the particle. Energy feedback during volatiles (6) Jamaluddin, A. S.; Wall, T. F.; Truelove, J. S. Modeling of Devolatilization and Combustion of Pulverized Coal under Rapid Heating Conditions. In Coal Science and Chemistry; Volborth, A., Ed.; Coal Science and Technology, Vol. 10; Elsevier: New York, 1987; pp 61-109. (7) Papageorgiou, N.; Kakaras, E. On the Mechanisms of Pyrolysis and Combustion of Brown Coal Particles in Fluidized Beds. Presented at the 2nd International Conference on Combustion Technologies for a Clean Environment, Portugal, July 19-22, 1993. (8) Agarwal, P. K.; Genetti, W. E.; Lee, Y. Y. Model for Devolatilization of Coal Particles in Fluidized Beds. Fuel 1984, 63, 11571165. (9) Samolada, M. C.; Vasalos, I. A. Fuel 1991, 70, 883-889. (10) Chan, W.-C. R.; Kelbon, M.; Krieger, B. B. Fuel 1985, 64, 15051513. (11) Di Blasi, C. Chem. Eng. Sci. 1996, 51 (7), 1121-1132.
Table 2. Kinetic Parameters and Stoichiometric Coefficients Applied for Modeling Olive Kernel Pyrolysis reaction
component
Ai (min-1)
Ei (kJ/mol)
ai
bi
1 2 3 4
hemicellulose cellulose lignin extractives
2.33 × 108 9.93 × 1017 2.63 × 101 2.03 × 1015
92.1 207.9 29.9 145.8
0.151 0.246 0.342 0.261
0.2 0.2 0.5 0.1
Table 3. Kinetic Parameters (Ai, Ei) and Component Contribution in Volatiles Release (γi) during Pyrolysis of Megalopolis Brown Coal reaction
Ai (min-1)
Ei (kJ/mol)
γi
1 2 3 4
1.07 × 7.18 × 105 2.37 × 103 3.54 × 102
34.1 89.7 65.4 75.1
0.327 0.146 0.144 0.383
102
combustion was taken into account through the effective flame temperature:12,13
TEFF ) TG +
kFfGHVY∞ cp,G
(3)
The main characteristics of biomass species studied in this work and Megalopolis brown coal are presented in Table 1. The kinetic parameters, which reflect the decomposition of each main constituent of biomass during devolatilization, were derived from TGA1-3 (see Tables 2 and 3). Char Combustion. The heterogeneous reaction of char oxidation is the last step of single-particle combustion. It essentially determines the duration of the entire process, because it lasts more than all other preceding steps. Many research efforts have been performed in modeling char combustion, taking into account both transport phenomena around and within the particle and intrinsic kinetics.12-23 Fundamentally, two types of char combustion models have been devel(12) Saastamoinen, J.; Richard, J. Drying, Pyrolysis and Combustion of Biomass Particles. Res. Thermochem. Biomass Convers. [Ed. Rev. Pap. Int. Conf.] 1988, 221-235. (13) Saastamoinen, J. Fundamentals of Biomass Drying, Pyrolysis and Combustion. Presented at the IEA Biomass Combustion Conference, Cambridge, U.K., 1994. (14) Williams, A.; Backreedy, R.; Habib, R.; Jones, J. M.; Pourkashanian, M. Fuel 2002, 81, 605-618. (15) Field, M. A.; Gill, D. W.; Morgan, B. B.; Hawksley, P. G. W. Combustion of Pulverised Coal; The British Coal Utilization Research Association: Leatherhead, Surrey, U.K., 1967. (16) Rajan Renga, R.; Wen, C. Y. AICHE J. 1980, 26 (4), 642-654. (17) La Nauze, R. D. Chem. Eng. Res. Des. 1985, 63, 3-33. (18) Vockrodt, S. 3-Dimensionale Simulation der Kohleverbrennung in Zirkulierenden Atmospha¨rischen Wirbelschichtfeuerungen, Ph.D. Thesis, VDI, Braunsweig, Germany, October 1995. (19) Vamvuka, D.; Woodburn, E. T. Int. J. Energy Res. 1998, 22, 657-670. (20) Agarwal, P. K.; Mitchell, W. J. Mass Transfer Processes Around Burning Char Particles in Fluidized Beds. In 22nd International Symposium on Combustion/The Combustion Institute; 1988; pp 279286. (21) Timothy, L. D.; Sarofim, A. F.; Beer, J. M. Characteristics of Single Particle Coal Combustion. In 18th International Symposium on Combustion; The Combustion Institute, Pittsburgh, PA, 1982; p 1123.
294
Energy & Fuels, Vol. 19, No. 1, 2005
Grammelis and Kakaras
Figure 1. Transient particle temperature and diameter, mass flow rate, and ratio of transient mass/initial mass in relation to time. (Reference conditions: Tg ) 1123 K, YO2 ) 10 vol %, Wg ) 2 m/s.) oped, i.e., global reaction models and intrinsic reactivity models.14 Despite their simplicity, the global models are widely applied in engineering calculations and they use apparent kinetic constants that are not necessarily based on real physical phenomena. The reaction zone is distinguished in three different regimes, depending on whether the char combustion process is chemical- or diffusion-controlled. The reaction rate constant for the char combustion (K) then is equal to
1 (1/KD) + (1/Kch)
Table 4. Kinetic Parameters of Brown Coal and Olive Kernel Char Combustion Ao (K s)-1
Ea (kJ/mol) Brown Coal
0.020 0.007
71.7 61.7 Olive Kernel
0.222 0.042
113.25 75.6
(4)
The activation energies and pre-exponential factors of both biomass and brown coal chars are given in Table 4.
where KD is the reaction rate of the oxygen diffusion at the particle surface and Kch is the reaction rate of the chemical reactions. Determination of kinetic constants when modeling char combustion is a more-complicated task, and the values that are calculated in the low-temperature TGA should be converted to reflect the high temperatures of the fluidized bed. It is well-known that the apparent activation energy under fluidized-bed conditions is equal to half of its value, when kinetics is considered the determinant factor.24 In this case, the optimum value of the pre-exponential factor must be evaluated in a way that describes the combustion process representatively, depending on the char sample. The apparent activation energy was estimated from the kinetic analysis of char combustion tests in TGA.3 The pre-exponential factor was calculated through nonlinear regression analysis, with reference to the burn-out time of brown coal char, resulting from the kinetic parameters proposed by Go¨rner.18 Based on the thermogravimetric studies,3 char combustion of Greek brown coal and olive kernel is modeled using two reactions.
Results and Discussion Comparison of Calculated Results for Biomass and Brown Coal Particle Burnout. The main results of the mathematical model, in relation to time, are illustrated in Figure 1. Comparing the combustion of biomass and brown coal, it is concluded that the temperature of the biomass particle increases faster, compared to brown coal, under the same heating conditions. This is due to the lower volumetric thermal capacity of the brown coal ((cpBrownCoalFBrownCoal)/(cpOl‚ KernelFOl‚Kernel) ≈ 1.8). The volatiles release rate is much higher in the case of olive kernel, and the process finishes earlier at a relatively low particle temperature. Taking into account that the great mass flow rate during biomass devolatilization prevents oxygen from reaching the particle surface, char combustion can be considered as a subsequent process. In contrast, the lower volatiles release rate of a brown coal particle, and its higher temperature after the volatiles release completion, indicate that the two subprocesses partially overlap. Large amounts of the volatiles escape from the particle boundary layer of a biomass particle without combustion, resulting in lower heat quantities available
K)
(22) Ilic, M.; Oka, S.; Grubor, B. D. Thermophys. Aeromech. 1995, 2 (4). (23) Hayhurst, A. N. Combust. Flame 1991, 85, 155-168. (24) Essenhigh, R. H.; Fortisch, D.; Klimesh, H. E. Energy Fuels 1999, 13 (5), 955-960.
Biomass Combustion Modeling in Fluidized Beds
Energy & Fuels, Vol. 19, No. 1, 2005 295
Table 5. Comparison of Present Model Results with Interational Literature reference
sample
methodology and conditions of tests/modeling
burnout time
comments
Turnbull et al.27 coal chars
combustion tests in fluidized bed and theoretical calculations: Tg ) 1173 K, dp ) 0.2-0.5 mm
tburnout ) 23 s (Phurnacite Breeze char), 32 s (Illinois char)
results derived from the combination of batch-type combustion tests and theoretical calculations
La Nauze17
petroleum coke, coal chars
theoretical calculations: Tg ) 1173 K, Wg ) 0.53 m/s, dp,char. ) 1.0 mm
petroleum coke: tburnout, experimental ) 91.3 s and tburnout, calculated ) 58-65 s
coal char combustion time of 150-240 s for Wg ) 0.214-0.383 m/s
Ntouros7
Megalopolis brown coal theoretical calculations: Tg ) 1123 K, YO2 ) 14 vol %, Wg ) 0.5 m/s, dp ) 1.0 mm
tburnout ) 45 s for Xc ) 100%; tpyrolysis) 4.35 s (Tg ) 1123 K), 4.8 s (Tg ) 1023 K), 6.35 s (Tg ) 923 K)
evaluation of kinetic parameters in pyrolysis using the exponential function
Vockrodt18 Vockrodt18 Vockrodt18
high-volatile coal numerical code: Tg ) 1123 K, tburnout ) 690 s; tpyrolysis ) 75 s Rhenish brown coal tburnout ) 107 s; tpyrolysis ) 75 s YO2 ) 3 vol %, Wg ) 8 m/s, Megalopolis brown coal dp ) 1.0 mm tburnout ) 45 s; tpyrolysis ) 2.8 s
Ross et al.25
six coals
pyrolysis tests in a fluidized bed and exponential function
tpyrolysis ) 1.65 s (dp ) 1 mm), 25.1 s (dp ) 6 mm)
Adanez et al.26
biomass chars (pine, beech, olive kernel)
tests in TGA and a fixed bed reactor; Tg ) 1073 K, Wg ) 0.3 m/s, dp ) 0.6-2.0 mm
tburnout ) 7.5 s (dp ) 0.6-0.8 mm), burnout times of infinitely 14.5 s (dp ) 1.0-1.25 mm), small mass based on 27.5 s (dp ) 1.6-2.0 mm) batch-type tests (5-40 mg)
this research
Megalopolis brown coal numerical code; Tg ) 1123 K, YO2 ) 10 vol %, Wg ) 2 m/s, dp ) 1.0 mm
tburnout ) 32.76 s for Xc ) 100%, tpyrolysis ) 1.78 s
this research
biomass (olive kernel)
tburnout ) 15.7 s for Xc ) 100%, tpyrolysis ) 0.77 s
for the particle thermal degradation during this phase. The particle temperature is significantly increased during char combustion. The opposite occurs when a brown coal particle is burned and its temperature greatly increases during volatiles combustion. However, for both fuels, the heat produced by combustion does not significantly increase the particle temperature above the gas temperature, because of the heat exchange between the particle and the gas. Although no differences are observed in the final particle temperature, the burn-out time of a biomass particle is significantly shorter than that of the brown coal particle. This is mainly due to the faster and moreprolific volatiles release of the biomass material, leading to a particle of lower temperature and much smaller density. Consequently, larger heat quantities due to convection and radiation are transmitted onto the particle, whereas the char reaction rate, which is inversely proportional to the particle density, becomes higher during this stage. In addition, taking into consideration that the different yields of volatiles alter the pore structureswhich has not been examined in this modelsand, thus, the reactivity of the remaining char, the biomass material burn-out time is expected to be even shorter.16 Validation of Mathematical Model Results. The calculated results of the developed model were compared with other published data, as shown in Table 5. The comparison was based on experimental or calculated results for the time needed to complete pyrolysis,25 char combustion,17,26,27 and total burnout7,18 of a solid fuel (25) Ross, D. P.; Heidenreich, C. A.; Zhang, D. K. Fuel 2000, 79, 873-883. (26) Adanez, J.; De Diego, L. F.; Garcia-Labiano, F.; Abad, A.; Abanades, J. C. Ind. Eng. Chem. Res. 2001, 40, 4317-4323. (27) Turnbull, E.; Kossakowski, E. R.; Davidson, J. F.; Hopes, R. B.; Blackshaw, H. W.; Goodyer, P. T. Y. Chem. Eng. Res. Des. 1984, 62 (July), 1217.
kinetic parameters of char combustion (from Go¨rner18) volatiles release times of 1.35-1.84 s for dp ) 1 mm found in the literature
kinetic parameters of devolatilization and char combustion calculated using TGA
particle in a fluidized bed installation. It is common knowledge that there are significant variations in experimental results, because they are dependent directly on the experimental methodology and the operating conditions.17 The calculated results of the present mathematical model are in agreement with those presented by other researchers (see Table 5). More specifically, Turnbull et al.27 reported burn-out times of 23-32 s for different coal chars. The respective time for Megalopolis brown coal char, as calculated with this mathematical model, is ∼31 s, which is within the specific range. The pyrolysis time of a single Megalopolis brown coal particle is 1.8 s, which ranges between 1.35 and 1.84 s, as proposed by Ross et al.25 for the same particle size (dp ) 1 mm). Vockrodt18 and Ntouros7 calculated increased times of volatiles release for Megalopolis brown coal, which correspond to 2.8 s and 4.35-6.35 s, respectively, depending on the surrounding gas temperature. Vockrodt18 applied the Suuberg model for lignite pyrolysis, which includes a complex reactions scheme to simulate the formation of different compounds. Ntouros7 used an exponential function to calculate the change in particle diameter, relative to time. Also, the heat transmitted onto the particle surface due to the simultaneous volatiles combustion in the surrounding area was not considered in any of these models. The heat released during the volatiles combustion accelerates the devolatilization process, and this was incorporated in the present model, resulting in shorter pyrolysis time periods. Moreover, Ross et al.25 proved that fuel devolatilization with an oxidant medium (air) instead of an inert gas (nitrogen) may cause a reduction in pyrolysis time, up to 40%, due to volatiles combustion. Shorter volatiles release time and increased particle temperature at the end of the pyrolysis process seriously influence the char conversion time. This is also
296
Energy & Fuels, Vol. 19, No. 1, 2005
Grammelis and Kakaras
Figure 2. Burn-out time of a single biomass/waste and brown coal particle, in relation to particle diameter. (Reference conditions: Tg ) 1123 K, YO2 ) 10 vol %, Wg ) 2 m/s.)
evident from the calculated results described in other research works.7,18 The particle temperature of the present model exceeds the surrounding gas temperature during the char combustion stage (also see Figure 1), as a result of energy conservation, which agrees well with previously published data. Fewer studies have been published that are related to the burn-out time of a single biomass particle. However, the calculated results for biomass combustion are not far from those anticipated due to the differentiation in fuel composition. The increased reactivity and much-higher volatiles quantity released during biomass pyrolysis result in shorter burn-out times. This was also confirmed during the devolatilization tests in the TGA.1,2 The char burn-out time calculated with the present model is ∼14.5 s, which is the value specified by Adanez et al.26 when burning biomass chars with a particle size fraction of 1-2.5 mm (see Table 5). The researchers investigated the effect of inert material mass in the fluidized bed and gas velocity during the combustion of beech char. The conversion of experimental results to infinitely small sample mass results in the same time values with the present model. Shorter burn-out times of biomass chars were recorded, in comparison to the coal samples during thermogravimetric studies.3 The differences were such as to account for a 30% reduction in biomass char burn-out time, compared to brown coal. Di Blasi et al.28 observed an increased reactivity of biomass char and mentioned a significant char burnout time decrease when the heating rate was increased from 20 K/min up to 80 K/min. Calculated Results for Different Biomass/Waste SpeciessSize and Temperature Effects. The mathematical model formulated to simulate the combustion of a single biomass/waste particle requires many input parameters and properties, the values of which are highly uncertain. To study the effect of uncertain model parameters on the model predictions, a sensitivity analysis was conducted. The numerical code was applied for different biomass waste species, with different properties (see Table 1). The kinetic parameters for both the pyrolysis and char combustion were determined through thermogravimetric studies, and their values are presented in Tables 6 and 7, respectively. (28) Di Blasi, C.; Buonanno, F.; Branca, C. Carbon 1999, 37, 12271238.
Table 6. Kinetic Parameters for Pyrolysis of Biomass Species Studied Hemicellulose sample cotton residues E (kJ/mol) A (min-1) γ (%) forest residues E (kJ/mol) A (min-1) γ (%) waste wood E (kJ/mol) A (min-1) γ (%) wood pellets E (kJ/mol) A (min-1) γ (%)
1st curve
2nd curve
95.1 4.19 × 108 17.6
cellulose
lignin
145.1 30.8 1.66 × 1012 2.47 × 101 48.3 34.1
108.2 124.6 233.8 34.5 1.99 × 1010 3.68 × 1010 7.23 × 1018 4.72 × 101 5.2 28.8 42.2 23.8 97.1 1.83 × 108 29.9
210.1 31.7 1.87 × 1017 2.59 × 101 43.0 27.2
93.7 9.48 × 107 29.5
199.7 28.3 2.04 × 1016 1.27 × 101 42.4 28.1
Table 7. Kinetic Parameters for Char Combustion of Biomass Species under Investigation sample
Ao (K s)-1
Ea (kJ/mol)
cotton residues forest residues waste wood wood pellets
0.1898 0.0145 0.0416 0.0298
91.05 65.0 74.85 72.65
The burn-out time of the different biomass/waste fuels and Megalopolis brown coal is illustrated as a function of the particle diameter in Figure 2. The burn-out time of a brown coal or biomass particle seems to be dependent on particle diameter linearly (see Figure 2). Differences in the burn-out times of brown coal and biomass particles are more apparent for large diameters. However, the burn-out times between the various biomass species are similar and become insignificant as the particle diameter decreases. The highest combustion time was obtained for cotton residues, whereas the lowest was observed for olive kernel. Burn-out times of the wood species ranged between the values calculated for cotton residues and olive kernel. The effect of bulk gas temperature was higher on the burn-out time calculations (Figure 3). When exceeding 1123 K, the burn-out times of all biomass species almost coincide. Some differences can be observed at lower temperatures, especially for cotton residues and olive kernel. Apart from the fuel properties, the sensitivity of the combustion model to particle characteristics (diameter,
Biomass Combustion Modeling in Fluidized Beds
Energy & Fuels, Vol. 19, No. 1, 2005 297
Figure 3. Burn-out time of a single biomass/waste and brown coal particle in relation to surrounding gas temperature. (Reference conditions: Tg ) 1123 K, YO2 ) 10 vol %, Wg ) 2 m/s.)
moisture), fluidized bed characteristics (gas temperature, gas velocity, volumetric oxygen concentration), and heat- and mass-transfer coefficients was examined. The biomass particle burn-out time was observed to be very sensitive to all parameters examined, except for gas velocity. Namely, small particles are heated much faster, mainly due to the higher convective heat-transfer coefficient, which is inversely proportional to the particle size. Moreover, the reactivity of the particle (in units of g cm-3 s-1) increases as the particle diameter decreases. Therefore, the time for the volatiles release and the burn-out time are significantly lower for smaller particles (from 62% up to 78%). In addition, the particle temperature after the devolatilization stage is higher for smaller particles but becomes slightly lower at the end of char combustion, resulting from the moreefficient heat exchange between the gas and the particle, when the particle size is small. The time for heating, drying, and volatiles release is reduced, when the fluidized bed temperature is increased, because of the higher heat flux transmitted onto the particle. In the same way, higher heating rates occur during char combustion and significantly lower burn-out times are obtained, as the char reactivity, which is directly dependent on the particle temperature, is increased. No temperature differences were observed at the end of the volatiles release, because the convective heat-transfer coefficient was gradually decreased from the cooling volatiles flow. Conclusions The purpose of this study is to develop a computational code for predicting the burn-out time of a single biomass particle. The research was based on previous experience for coal particles, taking into account the different biomass composition and characteristics. Kinetic parameters for devolatilization and char combustion used in the model were determined via thermogravimetric analysis (TGA). The results have shown that the time required for the heating, drying, evolution, and volatiles combustion and char combustion stages of particle conversion was significantly lower in the case of biomass species. The small volume heat capacity, the lower moisture content, and the higher volatile content of the biomass materials, compared to brown coal, accounted for this effect. The faster and more-prolific volatiles release, as well as the enhanced reactivity of biomass char, accelerates the fuel conversion even more,
resulting in remarkable differences of burn-out time calculations, in comparison to Greek brown coal. These findings are in agreement with those published by other researchers. The sensitivity of the present model to particle properties and mass- and heat-transfer coefficients, as well as operating conditions of fluidized beds, was tested. The results were greatly influenced by the temperature of the surrounding gas and, to a lesser extent, by the particle diameter, moisture content of the fuel, and bulk concentration of oxygen. Based on the aforementioned discussion, it can be concluded that the mathematical model developed in this work, which incorporates kinetic parameters as determined experimentally by TGA, successfully describes the behavior of different biomass particles during combustion. The methodology used by the present study could be applied in detailed computational codes for the effective design and operation of large-scale fluid bed units. Nomenclature Symbol A ai bi
HV KCh Kd Kf T t w YO2 F
Meaning frequency factor mass fraction of biomass main components contribution to char formation of biomass main components contribution to volatiles release of biomass main components specific heat capacity particle diameter activation energy stoichiometric mass ratio of pyrolysis products and oxygen in the flame heating value of volatiles chemical reaction rate constant diffusion rate constant empirical coefficient for the flame heat transfer temperature time velocity volumetric oxygen concentration density
Subscripts c eff g p ∞
char effective gas particle environment
ci CP d E fg
EF049838O