Global Kinetics of Wood Char Devolatilization and ... - ACS Publications

Oct 22, 2003 - Daniel C. Donato , John L. Campbell , Joseph B. Fontaine , Beverly E. Law ... Amir Tadros , Rodney Boyd , Peter Benyon , Charles Grima...
0 downloads 0 Views 145KB Size
Energy & Fuels 2003, 17, 1609-1615

1609

Global Kinetics of Wood Char Devolatilization and Combustion Carmen Branca and Colomba Di Blasi* Dipartimento di Ingegneria Chimica, Universita´ degli Studi di Napoli “Federico II”, P.le V. Tecchio, 80125 Napoli, Italy Received January 29, 2003

Thermogravimetric curves in air of wood chars obtained from different species and conventional or fast pyrolysis show a low-temperature shoulder (devolatilization) followed by a high-temperature peak (combustion). A n-order global reaction provides a very poor description of the differential curves and, in agreement with previous literature, requires relatively low activation energies (114.5 and 167 kJ/mol for conventional and fast pyrolysis chars, respectively). The combination with an additional first-order reaction for the devolatilization stage produces accurate predictions of both integral and differential curves. It also gives rise to much higher activation energies of the combustion reaction (183 or 229 kJ/mol, depending on the severity of the pyrolysis conditions). The activation energy of the devolatilization reaction also increases with the severity of the pyrolysis conditions (from 114.5 to 218.5 kJ/mol), whereas the differences in the char reactivity deriving from the wood species can be taken into account by preexponential factors and order of the combustion reaction. The use of two first-order reactions for the devolatilization stage does not improve significantly the accuracy of the predictions.

Introduction Combustion of wood takes place according to two stages:1 pyrolysis associated with the homogeneous combustion of volatile products, and heterogeneous combustion/gasification of char. Char reactivity is significantly affected by the nature of the solid fuel and the pyrolysis conditions which are responsible for pore structure, elemental composition (including ashes catalytically active), and aromaticity degree.2-3 Char conversion is slower by factors of 10-100 compared with the devolatilization stage of the virgin fuel,3-4 so that it largely affects the design and optimization of practical combustion systems. In particular, kinetic mechanisms of char combustion are needed to be coupled with the sub-models of other relevant phenomena, such as fluid flow, heat transfer, and gas-phase combustion, to construct advanced computational tools. The literature on the reactivities of coal chars and their kinetic modeling is huge. It has been extensively discussed in several reviews2-3 and is the subject of recent studies (see, for instance, refs 5 and 6).5-6 On the contrary, the kinetics and the characteristics of biomass/wood char combustion have been only partially addressed. Some studies have been carried out to inves* Corresponding author. Tel: 39-081-7682232. Fax: 39-081-2391800. E-mail: [email protected]. (1) Kanury, A. M. Combustion characteristics of biomass fuels. Combust. Sci. Technol. 1994, 97, 469-491. (2) Laurendeau, N. M. Heterogeneous Kinetics of coal char gasification and combustion. Prog. Energy Combust. Sci. 1978, 4, 221-270. (3) Smith, I. W. The combustion rates of coal chars: a review. In Nineteenth International Symposium on Combustion; The Combustion Institute: Pittsburgh, 1982; pp 1045-1065. (4) Ruth, L. A. Energy from municipal solid waste: a comparison with coal combustion technology. Prog. Energy Combust. Sci. 1998, 24, 545-564.

tigate the combustion rate under oxygen diffusion control,7-9 or to understand the details of the burning process for practical systems.10-14 In other cases,9,15-18 the analysis has been focused on the intrinsic kinetics of combustion, and so the mild thermal conditions of thermogravimetric systems have been selected. (5) Bews, I. M.; Hayhurst, A. N.; Richardson, S. M.; Taylor, S. G. The order, Arrhenius parameters, and mechanism of the reaction between gaseous oxygen and solid carbon. Combust. Flame 2001, 124, 231-245. (6) Hurt, R. H.; Calo, J. M. Semi-global intrinsic kinetics for char combustion modeling. Combust. Flame 2001, 125, 1138-1149. (7) Smith, F. G.; Pendarvis, R. W.; Rice, R. W. Combustion of cellulosic char under laminar flow conditions. Combust. Flame 1992, 88, 61-73. (8) Luo, M.; Stanmore, B. The combustion characteristics of char from pulverized bagasse. Fuel 1992, 71, 1074-1076. (9) Adanez, J.; de Diego, L. F.; Garcia-Labiano, F.; Abad, A.; Abanades, J. C. Determination of biomass char combustion reactivities for FBC applications by a combined method. Ind. Eng. Chem. Res. 2001, 40, 4317-4323. (10) Shafizadeh, F.; Sekiguchi, Y. Oxidation of chars during smoldering combustion of cellulosic materials. Combust. Flame 1984, 55, 171-179. (11) Mukunda, H. S.; Paul, P. J.; Rajan, N. K. Combustion of wooden spheres- experiments and model analysis. In Twentieth International Symposium on Combustion; The Combustion Institute: Pittsburgh, 1984; pp 1619-1628. (12) Dasappa, S.; Sridhar, H. V.; Paul, P. J.; Mukunda, H. S.; Shrinivasa, U. On the combustion of wood-char spheres in O2/N2 mixtures - experiments and analysis. In Twenty-fifth International Symposium on Combustion; The Combustion Institute: Pittsburgh, 1994; pp 569-576. (13) Wornat, M. J.; Hurt, R. H.; Yang, N. Y. C.; Headley, T. J. Structural and compositional transformations of biomass chars during combustion. Combust. Flame 1995, 100, 131-143. (14) Wornat, M. J.; Hurt, R. H.; Advise, K. A.; Yang, N. Y. C. Single particle combustion of two biomass chars. In Twenty-sixth International Symposium on Combustion; The Combustion Institute: Pittsburgh, 1996; pp 3075-3083. (15) Kashiwagi, T.; Nambu, H. Global kinetic constants for thermal oxidative degradation of a cellulosic sample. Combust. Flame 1992, 88, 345-368.

10.1021/ef030033a CCC: $25.00 © 2003 American Chemical Society Published on Web 10/22/2003

1610 Energy & Fuels, Vol. 17, No. 6, 2003

The majority of the kinetic models, using integral thermogravimetric (TG) data for parameter estimation, are based on a one-step global reaction with a powerlaw dependence on the oxygen partial pressure and the solid mass fraction. The differences in the nature of feedstocks and pyrolysis conditions make the comparison difficult. However, the reactivities are observed to increase as the heating rate is increased and/or the temperature is decreased during pyrolysis.17 Significant variation is also shown by kinetic parameters, with activation energies of 160 kJ/mol for cellulosic chars15 and 75-140 kJ/mol for other feedstocks.9,16-18 Despite the widely used one-step reaction, differential thermogravimetric (DTG) data10,19,20 show that combustion of lignocellulosic chars is a multistep process. Indeed, weight loss starts at relatively low temperature and occurs to a certain extent also in inert environment. This process is indicated as char devolatilization.20,21 More specifically, low-temperature cellulosic chars10 undergo devolatilization according to three sequential stages, corresponding to the pyrolysis of partially decomposed and intact glycosyl units and decomposition of paraffinic and carbonyl groups. Chars obtained at temperatures above 673 K show only the last stage. As expected, the amount of solid left at the conclusion of the devolatilization process (and burned in the presence of oxygen) becomes successively higher with the formation temperature. Only in one case20 has an attempt been made to formulate a kinetic mechanism which takes into account both devolatilization and combustion. Chars were obtained by means of high-pressure pyrolysis of macadamia nut shells and eucalyptus, and the estimated activation energies for the two sequential reactions are 105-110 and 146-148 kJ/mol, respectively. Successively,21 a multistep mechanism has been used to model in detail the devolatilization stage. In this work, thermogravimetry is employed for a systematic investigation of the combustion behavior of wood chars. The conditions of the experimental analysis have been determined so as to separate chemical kinetics effects from heat and mass transfer phenomena. The main objectives are (a) to formulate a global kinetic mechanism based on a few reactions and to estimate the associated kinetic constants, (b) to assess the role played by the number of the reaction steps on the activation energies, (c) to quantify the differences between char reactivities originating from wood species and pyrolysis conditions, and (d) to ascertain whether a single mechanism and one set of kinetic data can be used with sufficient accuracy in all cases. (16) Magnaterra, M.; Fusco, J. R.; Ochoa, J.; Cuckierman, R. Kinetic study of the reaction of different hardwood sawdust chars with oxygen, chemical and structural characterization of the samples. In Proceedings of the International Conference on Advances in Thermochemical Biomass Conversion; Bridgwater, A. V., Ed.; Blackie A. & P.: London, U.K., 1994; pp 116-130. (17) Janse, A. M. C.; de Jonge, H. G.; Prins, W.; van Swaaij, W. P. M. The combustion kinetics of char obtained by flash pyrolysis of pine wood. Ind. Eng. Chem. Res. 1998, 37, 3909-3918. (18) Di Blasi, C.; Buonanno, F.; Branca, C. Reactivities of some biomass chars in air. Carbon 1999, 37, 1227-1238. (19) Marcilla, A.; Conesa, J. A.; Asensia, M.; Garcia-Garcia, S. M. Thermal treatment and foaming of chars obtained from almond shells: kinetic study. Fuel 2000, 79, 829-836. (20) Varhegyi, G.; Szabo, P.; Antal, M. J.; Dai, X. Kinetic modeling of the gasification of biomass charcoals. In Proceedings of the 1st World Conference on Biomass for Energy and Industry; Kyritsis, S., Beenackers, A. A. C. M., Helm, P., Grassi, A., Chiaramonti, D., Eds.; James & James (Science Publisher) Ltd.: London, U.K., 2001; pp 1783-1785.

Branca and Di Blasi Table 1. Char Yield from the Pyrolysis of Several Woods (as percent of the initial dry mass) and Elemental Composition (as percent on a dry basis) of the Char Samples char yields % O (by [wt %] % C % H % N % ash difference)

char beecha (Fagus sylvatica) beechb Fagus sylvatica) Douglas fir (Pseudotsuga menziesii) pine (Pinus pinea) redwood (Sequoia sempervirens) chestnut (Castanea sativa) a

23.7

76.4

3.4

0.16

1.7

18.4

15.0

78.0

3.2

0.25

2.8

15.9

28.9

76.8

3.5

0.10

1.0

18.6

26.2

75.3

3.8

0.10

1.2

19.7

30.3

78.3

3.4

0.09

0.66

17.5

35.7

76.8

3.1

0.11

3.4

16.7

Conventional pyrolysis. b Fast pyrolysis.

Experimental Section Material. The tests consider chars obtained from two hardwoods (beech and chestnut) and three softwoods (Douglas fir, redwood, and pine). Pyrolysis was carried out for thick (40 mm diameter) wood cylinders radiatively heated along the lateral surface. Results concerning temperature dynamics, conversion times, and product yields were already presented elsewhere.22-23 Comparison between the different chars is made here for an external heat flux of 49 kW/m2, corresponding to a steady pyrolysis temperature of 800 K, with yields of char comprised between 24% (beech) and 35% (chestnut) of the initial dry mass of wood (Table 1). The elemental analysis (Table 1) shows that the C content is roughly the same in all cases (75-76%), except for redwood (78%), possibly as a consequence of the higher fixed carbon content of the virgin wood.24 These samples are indicated, in the following, as conventional pyrolysis chars. Char samples obtained from the pyrolysis of 2 mm thick particles in a fluidized sand-bed at a temperature of 800 K are also examined. The experimental system is the same as reported in ref 25 and consists of a steel reactor externally heated by a furnace. The system was batch operated with about 35-40 g of wood particles instantaneously fed, once the desired thermal conditions of the isothermal sand bed were achieved. The carbon content (Table 1) is slightly higher for the thinner particles (about 78% versus 76%), ensuing from the more severe thermal conditions established in this case (higher degree of devolatilization). These samples are indicated, in the following, as fast pyrolysis chars. Prior to thermogravimetric tests, char samples have been milled to powder (particle sizes below 80µm) and predried for 10h at 373K. Method. The thermogravimetric system has already been presented elsewhere,26,27 and only the main characteristics are summarized here. It consists of a furnace, a quartz reactor, a PID controller, a gas feeding system, an acquisition data set, (21) Varhegyi, G.; Szabo, P.; Antal, M. J. Kinetics of charcoal devolatilization. Energy Fuels 2002, 16, 724-731. (22) Di Blasi, C.; Gonzalez Hernandez, E.; Santoro, A. Radiative pyrolysis of single moist wood particles. Ind. Eng. Chem. Res. 2000, 39, 873-882. (23) Di Blasi, C.; Branca, C.; Santoro, A.; Gonzalez Hermandez, E. Pyrolytic behaviour and products of some wood varieties. Combust. Flame 2001, 124, 165-177. (24) Gronli, M. G.; Varhegyi, G.; Di Blasi, C. Thermogravimetric analysis and devolatilization kinetics of wood. Ind. Eng. Chem. Res. 2002, 41, 4201-4208. (25) Di Blasi, C.; Branca, C. Temperatures of wood particles in a hot-sand bed fluidized by nitrogen. Energy Fuels 2003, 17, 247-254. (26) Di Blasi, C.; Branca, C. Kinetics of primary product formation from wood pyrolysis. Ind. Eng. Chem. Res. 2001, 40, 5547-5556. (27) Branca, C.; Di Blasi, C.; Horacek, H. Analysis of the combustion kinetics and thermal behavior of an intumescent system. Ind. Eng. Chem. Res. 2001, 41, 2107-2114.

Wood Char Devolatilization and Combustion

Energy & Fuels, Vol. 17, No. 6, 2003 1611

and a precision balance. The furnace is a radiant chamber, which creates a uniformly heated zone, where a quartz reactor is located. The sample is exposed to thermal radiation by means of a stainless steel mesh screen, whose sides are wrapped on two stainless steel rods connected to a precision (0.1 mg) balance, which allows the weight of the sample to be continuously recorded. An air flow (nominal velocity of 0.5 × 10-2 m/s for the tests discussed in this study) establishes the proper reaction environment. Solid conversion is made to occur under known thermal conditions by means of feedback control of the sample temperature (measured by a close-coupled thin thermocouple), using the intensity of the applied radiative heat flux as the adjustable variable. It has been observed that char layer thicknesses up to 120 µm allow a good temperature control to be achieved, given maximum heating rates of 15 K/min and a final temperature of 873 K. Also, the weight loss curve is the same as the char layer thickness is decreased below 120 µm, indicating that spatial temperature gradients are negligible and oxygen diffusion is not the limiting process. Hence, the tests have been made for sample layers about 110 µm thick (3.5 mg distributed over a surface 25 × 5 mm2) with heating rates of 5, 10, and 15 K/min for a final temperature of 873 K for chars obtained from the conventional pyrolysis of beech wood. The influences of the wood species and pyrolysis conditions have been examined for a heating rate of 5 K/min and again a final temperature of 873 K. Each thermogravimetric test has been made in triplicate, showing good repeatability.

Results The thermogravimetric characteristics of the oxidation curves of wood chars are discussed first. Then kinetic mechanisms are proposed based on one, two, and three parallel reactions, the related kinetic constants are estimated, and the results are compared with previous literature. Thermogravimetric Analysis. Figure 1A reports the solid mass fractions (always defined on ash free basis) and the rates of volatile release as functions of temperature for conventional (beech and Douglas fir) and fast (beech) pyrolysis chars, obtained for a heating rate equal to 5 K/min. In all cases, the rate curves show a low-temperature shoulder followed by a peak. In accordance with previous investigations,10,19,20 the first and second zone can be associated with devolatilization and combustion of char, respectively. The characteristic temperatures for the two reaction zones are dependent on the wood species and the pyrolysis conditions, as indicated in Table 2 (definitions as in ref 24 and summarized in Figure 1B). In particular, Tshoulder values vary between 623 K (beech) and 664 K (reedwood) with corresponding mass fractions of 0.89 and 0.82. Variations for the peak rate characteristics are also relatively small (Tpeak values vary between 700 K (beech) and 738 K (Douglas fir)). Fast pyrolysis causes a displacement of the shoulder at slightly higher temperatures (641 K versus 623 K for beech wood), so that devolatilization and combustion present a wider overlap. The peak rate is higher, though the position is roughly the same. Finally, as shown in Table 3 (beech wood char), the characteristic temperatures (and the corresponding rates) increase with the heating rate, but the two reaction stages are still clearly visible. Thermogravimetric curves have also been measured of chars obtained from the conventional pyrolysis (40 mm cylinders) for different radiative heat fluxes (33-80 kW/m2 corresponding to steady pyrolysis tem-

Figure 1. (A) Solid mass fraction, on ash free basis, and rate of volatile release for conventional (beech and Douglas Fir) and fast (beech) pyrolysis chars, as measured in air for a heating rate of 5 K/min. (B) Definitions of characteristic temperatures, mass fractions, and devolatilization/combustion rates of thermogravimetric measurements. Table 2. Characteristic Parameters (Tshoulder, Yshoulder, Tpeak, Ypeak, -(dY/dt)peak) of the Thermogravimetric Curves of Conventional and Fast Pyrolysis Chars for a Heating Rate of 5 K/min char

Tshoulder [K]

Yshoulder

Tpeak [K]

Ypeak

-(dY/dt)peak × 103 [s-1]

beecha beechb pine chestnut Douglas fir redwood

623 641 655 653 656 664

0.89 0.92 0.87 0.90 0.88 0.82

699 701 718 735 738 724

0.34 0.35 0.31 0.39 0.37 0.34

1.25 1.36 1.18 0.91 0.89 1.02

Table 3. Characteristic Parameters (Tshoulder, Yshoulder, Tpeak, Ypeak, -(dY/dt)peak) of the Thermogravimetric Curves for Conventional Pyrolysis (beech) Chars for Heating Rates of 5-15 K/min h [K/min]

Tshoulder [K]

Yshoulder

Tpeak [K]

Ypeak

-(dY/dt)peak × 103 [s-1]

5 10 15

642 659 670

0.89 0.83 0.82

699 717 726

0.34 0.30 0.29

1.25 2.49 3.57

peratures of 700-950 K) and for fluid-bed pyrolysis (2 mm particles) at several temperatures of the sand bed (700-950 K). However, differences in the char reactivity are negligible between samples obtained with the same experimental system. It appears that the external temperature (always higher than the characteristic values of wood pyrolysis) is less important than the wood particle size. This is a major factor in relation to

1612 Energy & Fuels, Vol. 17, No. 6, 2003

Branca and Di Blasi

the actual heating rate established during the pyrolysis process. In other words, for both the experimental systems, the actual pyrolysis conditions are affected more by internal than external heat transfer conditions. These results confirm the analysis of ref 17, who point out that the heating rate during pyrolysis is decisive for the resulting char reactivity, whereas the final pyrolysis temperature is of minor importance, at least for temperatures in the range 773-873 K. Kinetic Mechanism. The most detailed mechanism proposed here consists of three parallel independent reactions (three-step model): +O2

+O2

+O2

A 98 V1, B 98 V2, C 98 V3

(a1-a3)

where A, B, and C are three char fractions, which produce the lumped volatile products, V1, V2, and V3, respectively. From the chemical point of view, reactions a1 and a2 are associated with char devolatilization (shoulder in the rate curve), and reaction a3 is associated with char combustion (peak in the rate curve). A parallel reaction (instead of a series reaction) mechanism has been chosen because it appears to be more flexible to take into account the overlap between the reaction zones. The rate of reactions a1 and a2 is assumed to present the usual Arrhenius dependence (A1, A2 preexponential factors and E1, E2 activation energies) on temperature and to be proportional to the mass fraction of components A and B:

R1 ) A1 exp(-E1/RT)YA R2 ) A2 exp(-E2/RT)YB

(1a) (1b)

For the char combustion rate, the rate of solid disappearance is generally related to the partial pressure of oxygen through an empirical exponent and the pore surface area available through the reaction volume. Given the relatively high air flow rate employed in the tests, it can assumed that the oxygen mass fraction remains constant during the reaction process. Consequently, its contribution is incorporated in the preexponential factor. Also, a simple power law (n) expression of the solid mass fraction18 is applied to describe the evolution of the pore surface area during the process:

R3 ) A3 exp(-E3/RT)YnC

(2)

As the sample temperature, T, is a known function of time, t:

T ) T0 + ht

(3)

where T0 is the initial temperature and h is the heating rate, the mathematical model can be expressed as

dYA ) -R1, YA(0) ) R dt dYB ) -R2, YB(0) ) β dt dYC ) -R3, YC(0) ) 1 - R - β dt

(4a-b) (5a-b) (6a-b)

where R and β are the volatile fractions released during devolatilization (components A and B). Simplifications in the mechanism of reactions a1-a3 can be introduced by assuming R ) 0, that is, devolatilization is described by only one reaction (two-step model) or R ) β ) 0, when the mechanism reduces to the widely used one-step global reaction of char combustion (one-step model). The simplifications are motivated by the relatively small volatile fraction released during the devolatilization stage (Yshoulder comprised between 0.82 and 0.92, see Tables 2-3) and by the difficulty in its detection when using only integral data for kinetic analysis. Moreover, given the extensive use of a single global reaction in previous studies, this approach is also considered for comparison purposes. The kinetic parameters are estimated through the numerical solution (implicit Euler method) of the mass conservation equations and the application of a direct method for the minimization of the objective functions, which consider both integral (TG) and differential (DTG) data. The details of the method have already been described elsewhere.27 The parameters to be estimated are the activation energies, the preexponential factors, the stoichiometric coefficients ,and the order of the combustion reaction. The fit between measured and calculated curves is defined in accordance with previous analyses28,29 as

xS/N × 100 (φi)exp,peak

(7)

∑ ((φi)exp - (φi)sim)2

(8)

% dev ) S)

i)1,N

where i represents the experimental (exp) or the simulated (sim) variable (φ is the solid mass fraction, Y, or the devolatilization rate, -dY/dt) at the time t (N is the number of experimental points, and the subscript “peak” indicates the maximum value). Kinetic Constants. The kinetic constants have been estimated for the conventional pyrolysis chars (beech wood) by means of thermogravimetric curves obtained for three heating rates (5, 10 and 15 K/min), to avoid compensation effects.30 The optimization procedure has been executed by requiring the same values of the activation energies, preexponential factors, and exponent, n, for all the curves. The stoichiometric coefficients have been allowed to vary with the heating rate. The initial guess of the kinetic parameters is derived from refs 20 and 21. The results of the kinetic analysis are summarized in Table 4 for the three models, whereas Figures 2 and 3 present a comparison between predicted and measured curves. As expected, the one-step model shows the worst performance, especially for the rate curves (deviations up to 18%), supporting the need to use simultaneously (28) Varhegyi, G.; Antal, M. J.; Szekely, T.; Szabo, P. Kinetics of the thermal decomposition of cellulose, hemicellulose, and sugar cane bagasse. Energy Fuels 1989, 3, 329-335. (29) Varhegyi, G.; Antal, M. J.; Jakab, E.; Szabo, P. Kinetic modelling of biomass pyrolysis. J. Anal. Appl. Pyrolysis 1997, 42, 73-87. (30) Conesa, J. A.; Marcilla, A.; Caballero, J. A.; Font, R. Comments on the validity and utility of the different methods for kientic analysis of thermogravimetric data. J. Anal. Appl. Pyrolysis 2001, 58-59, 617-633.

Wood Char Devolatilization and Combustion

Energy & Fuels, Vol. 17, No. 6, 2003 1613

Table 4. Kinetic Constants (A, E), Stoichiometric Coefficients (r, β), Reaction Order (n), and Deviations for Integral (TG) and Differential (DTG) Curves of Conventional Pyrolysis (beech) Chars, As Estimated by the One-, Two-, and Three-Step Model three-step model devolatilization

combustion

h [K/min] R β devDTG [%] devTG [%]

two-step model

E1)121.0 [kJ/mol] A1 ) 7.40 × 107 [s-1] E2 ) 140.0 [kJ/mol] A2 ) 7.90 × 108 [s-1] E3 ) 182.6 [kJ/mol] A3 ) 1.40 × 1011 [s-1] n ) 0.90 5 0.069 0.096 1.5 0.6

10 0.067 0.095 5.4 0.4

15 0.065 0.094 1.6 0.3

one-step model

E2 ) 114.5 [kJ/mol] A2 ) 1.22 × 107 [s-1] E3 ) 182.6 [kJ/mol] A3 ) 1.40 × 1011 [s-1] n ) 0.90 5 0 0.165 2.1 0.7

Figure 2. Mass fractions, on ash free basis, of conventional pyrolysis (beech) chars versus time for heating rates of 5-15 K/min as measured (symbols) and predicted (lines).

Figure 3. Rates of volatile release (chars from the conventional pyrolysis of beech) versus temperature for heating rates of 5-15 K/min as measured (symbols) and predicted (lines).

integral and differential data for the correct formulation of reaction mechanisms and the effective evaluation of kinetic constants. The activation energy is rather low (114.5 kJ/mol) and comprised in the range of values reported in the previous literature based on the same approach.16-18 Deviations for the other two models are small and not significantly affected by the use of one or two reactions for the description of the devolatilization stage. The widely different devolatilization and combustion rates

10 0 0.157 5.4 0.4

15 0 0.155 5.3 0.4

E3 ) 114.5 [kJ/mol] A3 ) 1.10 × 106 [s-1] n ) 0.86 5 0 0 18.0 4.6

10 0 0 16.6 2.5

15 0 0 13.2 4.90

Figure 4. Rate of volatile release (char from the conventional pyrolysis of beech) and components A and B versus time for a heating rate of 5 K/min as measured (symbols) and predicted (lines) by the two-step model.

are modeled by low and high activation energies. More precisely, for the first zone they are comparable with those found for the global one-step model (121-140 and 114.5 kJ/mol for the three- and two-step models, respectively). The amount of volatiles released is a weak function of the heating rate, and average values can be safely used in the engineering practice. The kinetic parameters for the combustion stage are independent of the detail degree in the description of the devolatilization stage. It is worth noting that the activation energy (182.6 kJ/mol) is significantly higher than that obtained by means of one-step global models and roughly the same as reported for the low-temperature zone of coal char or graphite combustion.5,6 Finally, the rate of the combustion reaction is nearly linear. An example of the time history of the component rates is shown in Figure 4 for the two-step model and a heating rate of 5 K/min (the results obtained for higher heating rates are qualitatively similar). The chief information that can be gained from this figure is that, despite the use of parallel reactions, the overlap between the evolution times of the volatile fractions is limited. This finding indicates that a series mechanism could also be proposed for the interpretation of thermogravimetric tests. On the other hand, simulations carried out with a two-step series mechanism do not introduce any significant variation on the estimated parameters. This finding is also in agreement with a recent thermogravimetric analysis on the kinetics of wood combustion,31

1614 Energy & Fuels, Vol. 17, No. 6, 2003

Branca and Di Blasi

Table 5. Kinetic Constants (A, E), Stoichiometric Coefficient (r, β), Reaction Order (n), and Deviations for Integral (TG) and Differential (DTG) Curves of the Conventional Pyrolysis Chars Obtained from Different Wood Species, As Estimated by the Two-Step Model and a Heating Rate of 5 K/mina char

A1 [s-1]

A2 [s-1]

n

R

β

beech pine redwood chestnut Douglas fir

1.22 × 107 4.68 × 106 2.91 × 106 3.50 × 106 2.72 × 106

3.00 × 1012 6.95 × 1010 5.02 × 1010 3.05 × 1010 2.82 × 1010

0.90 0.90 1.10 1.34 1.33

0 0 0 0 0

0.165 0.186 0.190 0.167 0.185

a

devDTG devTG [%] [%] 2.07 2.03 2.11 1.87 1.61

0.65 1.08 0.81 0.42 0.70

E1 ) 114.5 kJ/mol; E2 ) 182.6 kJ/mol.

which is based on a four-step series mechanism. The two steps employed for the combustion of the conventional pyrolysis char are well described by the same activation energies found here. From the physical point of view, a parallel mechanism could allow devolatilization and combustion to occur at comparable rates, a condition which might occur depending on char properties. The comparison between predictions and measurements shows that the two-reaction model provides a good description of the main characteristics of both the integral and differential curves. Moreover, given the same kinetics for the combustion reaction and the comparable devolatilization rates, no significant differences are expected in the predictions of practical combustion systems, obtained from the coupling of the twoor the three-step mechanism with transport equations. Hence, given its simplicity, the two-step reaction mechanism is recommended. It is also applied in the following to quantify the effects of wood species and pyrolysis conditions. Given the relatively small differences between chars derived from the conventional pyrolysis of several wood species, the evaluation of the kinetic parameters for the two-step model has been made through the simultaneous use of all the measurements (beech, chestnut, pine, Douglas fir, redwood) at 5 K/min. The optimization procedure has been executed by requiring the same energies for all the curves, whereas the other parameters have been allowed to vary. The kinetic parameters, listed in Table 5, report variations on the preexponential factors of about one and two orders of magnitude for the first and the second step, respectively. The fraction R varies between 0.165 and 0.185, and the order of the reaction varies between 0.9 and 1.34. From these parameters, it appears that the wood species is more important for the combustion than for the devolatilization stage. The agreement between predictions and measurements is good for both integral (Figure 5) and differential (Figure 6) data. Deviations are high (up to 5-18%) when the one-step mechanism is applied to conventional pyrolysis chars derived from different wood species, and the differences are taken into account by preexponential factors (2-8 × 105 - 1.1 × 109 s-1) and reaction order (0.86-1) (activation energy constant and equal to 114.5kJ/mol). On the other hand, when the constraint on the activation energy is removed, the range of estimated values is very narrow (115 - 119kJ/mol) and the deviations are barely affected. (31) Branca, C.; Di Blasi, C. Global intrinsic kinetics of wood oxidation. Fuel 2004, 83, 81-87.

Figure 5. Mass fractions, on ash free basis, of conventional pyrolysis chars, obtained from several woods, versus time for a heating rate of 5 K/min as measured (symbols) and predicted (lines) by the two-step model.

Figure 6. Rate of volatile release (chars from the conventional pyrolysis of several woods) versus time for a heating rate of 5 K/min as measured (symbols) and predicted (lines) by the twostep model.

The thermogravimetric curves of beech char, obtained from conventional and fast pyrolysis, cannot be evaluated simultaneously for any of the reaction mechanisms examined here. Indeed, in the latter case, a good fit between predictions and measurements is obtained only when all the parameters are allowed to vary, indicating that the pyrolysis conditions exert a strong effect on the reactivity of the resulting char. As shown in Table 6, deviations are small but the activation energies of the two reactions are significantly higher than those obtained for conventional pyrolysis chars. Moreover, the two values are nearly the same, thus suggesting that a one-step global model could perform better for the fast pyrolysis char (compared with conventional pyrolysis samples). Indeed, estimates lead to maximum deviations of 7% (against 18% of the other case) with an activation energy of 166 kJ/mol, a preexponential factor 8.06 × 105 s-1 and a reaction order of 0.96. Conclusions Thermogravimetric curves of wood char obtained in air at different heating rates for a final temperature of 873 K show two different reaction zones which can be

Wood Char Devolatilization and Combustion

Energy & Fuels, Vol. 17, No. 6, 2003 1615

Table 6. Kinetic Constants (A, E), Stoichiometric Coefficients (r,β), Reaction Order (n) and Deviations for Integral (TG) and Differential (DTG) Curves of the Conventional and Fast Pyrolysis (beech) Chars, As Estimated by the Two-Step Model and a Heating Rate of 5 K/min beech wood char

E2 [kJ/mol]

A2 [s-1]

E3 [kJ/mol]

A3 [s-1]

n

R

β

devDTG [%]

devTG [%]

conventional pyrolysis fast pyrolysis

114.5 218.5

1.22 × 107 1.33 × 1015

182.6 228.6

1.40 × 1011 4.85 × 1014

0.90 1.160

0 0

0.165 0.126

2.1 2.5

0.7 1.3

associated with devolatilization and combustion, respectively. The former process is quantitatively less important and has been neglected by the majority of previous kinetic analyses based on integral data only. It becomes successively less evident as the pyrolysis conditions are made more severe, but is not significantly affected by the wood species (at least for conventional pyrolysis chars). Three models of different complexity have been applied for the description of the thermogravimetric curves. The simplest approach, consisting of a global n-order reaction, provides a very poor description of differential data especially for the conventional pyrolysis chars. The corresponding activation energy increases with the severity of the pyrolysis conditions, but its value remains relatively low (114.5 and 165.6 kJ/mol for conventional and fast pyrolysis, respectively). The results obtained in this case are in complete agreement with previous literature. The introduction of a first-order rate reaction for the devolatilization process (activation energies of 114.5 or

218.5 kJ/mol, depending on conventional or fast pyrolysis) is sufficient to obtain accurate predictions of the differential curves. No significant improvement is provided by the use of two reactions for this stage. When char devolatilization is taken into account, the subsequent combustion zone is well described by activation energies much higher (183 kJ/mol for conventional pyrolysis chars) than those estimated for the global reaction. Again, the severity of the pyrolysis conditions is quantitatively important and causes a further augmentation in this parameter (229 kJ/mol). These values are comprised in the range of values reported for the low-temperature combustion of coal chars and graphite. The influences of the wood species on the char reactivity appear to be quantitatively less important than those of the thermal conditions established during pyrolysis. Indeed, for both the one- and the two-step mechanism, they can be taken into account by preexponential factors and order of the combustion reaction. EF030033A