A Two-phase One-dimensional Biomass Gasification Kinetics Model

Biomass and Bioenergy 21 (2001) 121–132

A two-phase one-dimensional biomass gasication kinetics model Daniele Fiaschi ∗ , Marco Michelini Dipartimento di Energetica “Sergio Stecco”, University of Florence, Via Santa Marta, 3, 50139 Firenze, Italy Received 29 May 2000; received in revised form 4 January 2001; accepted 6 April 2001

Abstract A mathematical model of biomass gasication kinetics in bubbling ,uidized beds has been developed. It is one-dimensional, as it is capable of predicting temperature and concentration gradients along the reactor axis, and considers two phases, a bubble and a dense phase. In addition to the reaction kinetics in the dense phase, mass transfer between the two phases and a quantitative estimation of local bubble and particle properties are included in the model. A theoretical optimization with respect to ER, pressure, bed height and gas velocity has also been performed. Finally, a comparison with experimental data from the literature was done, which showed a largely satisfactory agreement, though further validation is still required. c 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Biomass; Gasication; Kinetics; Fluidized bed

1. Introduction A correct design of the reactor, accounting also for the reaction kinetics, is fundamental to achieve an optimal conversion of the chemical energy present into the biomass feedstock. Several mathematical models have been proposed for biomass gasication kinetics [1– 6], with di9erent degrees of compliance, accuracy and adaptability to the di9erent reactor types. Generally, almost the whole of computational models consider the pyrolysis and sub-stoichiometric combustion

∗ Corresponding author. Tel.: +39-055-4796-436; fax: +39-0554796-342. E-mail address: danif@as.de.uni.it (D. Fiaschi).

as instantaneous because they are much faster than the gasication process [1,2]. Then, the related starting products are set on an experimental basis. The major problem that arises is to correctly tune the kinetic parameters on experimental–theoretical basis, which are able to depict a wide range of operating cases. Unfortunately, very poor detailed experimental data are available on this argument, and then it is diCcult to verify the general validity of the proposed mathematical models. The primary goal of this study has been to improve previous biomass gasication kinetics models [1,7], introducing the one-dimensionality and other di9usion phenomena, which are typical of bubbling ,uidized beds. The present kinetic model considers a onedimensional (axial) cylindrical ,uidized bed, divided

c 2001 Elsevier Science Ltd. All rights reserved. 0961-9534/01/$ - see front matter  PII: S 0 9 6 1 - 9 5 3 4 ( 0 1 ) 0 0 0 1 8 - 6

122

D. Fiaschi, M. Michelini / Biomass and Bioenergy 21 (2001) 121–132

Nomenclature A Ab Cb Cd D Db Dg dp dp0 dp0
h H hf j kbd kc ∗ Ki∞ Mb n1 n2 ns P ql r t

reactor cross sectional area (m2 ) bubble surface area (m2 ) reactant species molar concentration into the bubble phase reactant species molar concentration into the dense phase di9usion coeCcient bubbles diameter (m) gasier diameter (m) char particle diameter (mm) initial char particle diameter (mm) average particle size after the primary fragmentation (mm) specic enthalpy of the chemical species (kJ=kg) overall reactor height (not including freeboard) enthalpy of devaluation of the chemical species (kJ=kg) particle shrinking rate exponent bubble—dense phase mass transfer coeCcient mass transfer coeCcient elutriation rate (kg=m2 s) mass of solid particles in the ,uidized bed (kg) coeCcient accounting the primary fragmentation at the reactor inlet number of char particles coming through the burn-o9 number of species involved in reactions gasier pressure (bars) local external heat transfer (kJ=kg) dense phase overall reaction velocity (mol=s) generic instant of the gasication process (s)

into vertical elemental cylinders (with dense and bubble phases) up to the freeboard zone: the reactions occurring in this volume have been neglected (the fraction of char particles where the surface reactions take place and the residence time of these particles in this reactor’s zone are very low). The main steps of the calculation procedures are summarized as follows:

T tdis vb Vb vikin vit vmf v0 vg vt w xdi xib xid y ztot

local gasier temperature (K) time bound between dominating mass transfer and kinetics e9ects (s) bubble phase velocity (m=s) bubble volume (m3 ) reaction kinetics speed (mol=s) mass transfer speed (mol=s) minimum ,uidization velocity (m=s) inlet oxidizer velocity (m=s) ,uidizing gas velocity (m=s) solid particles entrainment velocity (m=s) total moles of H2 O added (biomass wetness + eventual steam=water additions) fraction of solid particles of di size at the generic time t ith species molar fraction in the bubble phase ith species molar fraction in the dense phase moles of oxidizing O2 per mole of biomass total moles of N2 added with the oxidizer (air + other)

Greek letters  moles of H2 per mole of C in the biomass formula  moles of O2 per mole of C in the biomass formula  coeCcient accounting the in,uence of Dg on vb bubble phase fraction b  initial molar ratio H2 Ovapour =CO2 produced in the pyrolysis=combustion process  local char particle density (kg=m3 ) Subscript tot total amount from bubbles and dense phases

(1) input of the biomass composition and the oxidizer type, ER, moisture; (2) input of the general reactor design parameters: length, diameter, thickness and wall materials; (3) calculation of the bottom temperature; (4) iterative process for each cylindrical element into which the ,uidized bed has been divided, involving bed hydrodynamics, heat and mass

D. Fiaschi, M. Michelini / Biomass and Bioenergy 21 (2001) 121–132

transfer and gasication kinetics, up to the bed top; (5) comparison of the results with experimental data.

2. Mathematical one-dimensional model for the uidized bed reactor The mathematical models for ,uidized bed reactors can be divided into three main groups, characterized by the number of phases accounted in the reactor: single-, double- and three-phase model [8]. The double-phase model has been the basis for the present study, with a dense phase (gas plus solid particles) and a bubble phase (mainly gaseous with much lower solid matter) [8]. The two-phase reactor (Fig. 1) is modeled as the sum of several elemental reactors of d z thickness. The related di9erential equations are solved versus the temperature and the syngas composition, along the gasier axis, for both dense and bubble phases. The model is one-dimensional, and then it is particularly suitable for bubble and circulating beds, where a high mixing degree is present. The reactor is supposed to be downdraft, both for the oxidant and biomass feeding. At a model level, the freeboard area has been considered to be chemically inert.

123

The gas ,ow entering the reactor at v0 speed (Fig. 1), is split into two phases: the dense phase, (characterized by the minimum ,uidization velocity vmf ), and the bubble phase (the velocity is v0 − vmf ). The solid particles were supposed to be present only in the dense phase; both phases follow a piston motion (compact ,ow without any mass transfer along the z direction). No heat exchange occurs along the axial direction. The mass balance in the starting bubble phase has the following expression: (v0 − vmf )

dCb = kbd b (Cb − Cd ): dz

Similarly, the overall mass balance for bubble and dense phases is (v0 − vmf )

dCd dCb + vmf + r(1 − b ) = 0: dz dz

The molar concentration of the ith component in the two phases has been expressed in terms of molar fraction: Cid =

xid nb ; (1 − b )Vtot

Cib =

xib nb ; b Vtot

then, the mass balance in the bubble phase referred to one mole of biomass is as follows:
 vb1 d xib xid xib : = k − bd (1 − b ) b b2 d z The dense phase mass balance, referred to one mole of biomass, results from the algebraic sum of the chemical kinetic rate and the bubble-dense phase transfer rate: vmf

d xid bxib = − vb1 − ri : dz dz

The boundary conditions at z = 0 assume the same concentration of each i species for both phases, then, adding this statement to the mass balance condition: xitot = xib + xid

Fig. 1. Two-phase model reactor.

xib xid = ; (1 − b ) b

the only solid species C (carbon) has been set to zero in the bubble phase (xCtot = xCd ). The bubble-dense phase mass transfer coeCcient kbd has been estimated by the two-phase theory [8]

124

D. Fiaschi, M. Michelini / Biomass and Bioenergy 21 (2001) 121–132

as the sum of two terms (convective and di9usive) through the bubble surface Ab : kbd =

0:5 0:25

(N + Ab kc ) vmf D g = 4:5 + 5:85 : Vb Db Db1:25

The bubble volume fraction b is expressed as mf b (h) = v0v−v [5], where vb (h) is the local rising b (h) particle speed, depending both on bubble and √ bed diameters. In this study, the expression vb = $ gDb has been adopted [5], where  is a coeCcient related to the gasier diameter Dg . The bed sand and the ,uidization mean in,uence these parameters (i.e. density and viscosity of the produced syngas). The evaluation of bubbles diameter is one of the topics for the proposed kinetic model. Several theoretical–experimental investigations [9] lead to many kinds of equations: however, they do not completely agree with each other, due to the in,uence of the specic operating conditions on the several kinetic parameters. However, some general behaviors can be considered to relate the bubbles size, the reactor pressure and the gas velocity: • at a constant pressure, increasing the gas speed leads to an increase in the bubbles size for boiling bed regime and a decrease at turbulent regime; • at xed v0 (or v0 − vmf ), the bubble size decreases with increasing pressure for both turbulent and boiling regimes (except for extremely low speed values). Being the main goal of this work, the evaluation of syngas composition versus residence time (i.e. along gasier height), an expression of the bubble size as a function of the above parameters and the bed level is required. Starting from the time averaged bubble diameter at a xed bed height, it is often convenient to evaluate the equivalent diameter De (mean bubble diameter along the whole ,uidized bed, H De = H1f 0 f Dbh dH ). The Rowe equation [10] and some coeCcients tuned on experimental data, have been adopted to have an expression Dbh used in the model [11]: Dbh = 0:38 × H 0:8 P 0:06 (vo − vmf )0:42 exp[ − 0:00014P 2 − 0:25(vo − vmf )2 − 0:1P(vo − vmf )]

while the equivalent diameter De = 0:2Hf0:8 P 0:06 (v0 − vmf )0:42 exp[ − 0:00014P 2 − 0:25(v0 − vmf )2 − 0:1P(v0 − vmf )]: 3. Abrasion and elutriation phenomena The abrasion has been modeled considering both primary and secondary fragmentations [12]: a multiplying constant n accounts for the particle diameter reduction in the bed. The bed is supposed to be perfectly mixed with uniform particle size, which reduction is attributed to the following contributions (fragmentation and spread of solids outside the reactor have been neglected): 1. chemical reaction kinetics, accounted by the j factor (j is the exponent of the particle shrinking rate expression −d(dp )=dt = kdjp ) is related to the limiting reactions mechanisms (for example, j = 0 if the reactions are kinetically limited); 2. abrasion, accounted by two multiplying constants n1 (primary fragmentation e9ect at the reactor inlet) and n2 (number of char fragments generated by the burn-o9 of one char particle). The real number of char fragments is assumed to be the mean between n2 and 1). Then, the following expression of mean Sauter particle diameter comes out: −1=3  n1 (n2 + 1) 3−j dp = dp0 : 4−j 2 For circulating beds, the particles rising from the bottom to the top of the reactor vary their size, with respect to the initial value; the average particle size after the primary fragmentation is related to its initial value by the following expression: (dp0
)3 = (dp0 )3 =n1 : The elutriation has been modeled following two different approaches [12,13]: • the rst one has been introduced into the existing model by calculating the instantaneous diameter and

D. Fiaschi, M. Michelini / Biomass and Bioenergy 21 (2001) 121–132

porosity of the particles and then the corresponding entrainment velocity: if this is lower than the operating reactor speed, the corresponding elutriated mass at a xed height is subtracted from the total mass of C. The elutriation rate is evaluated as in [13], for reactor levels higher than the Transport Disengaging Height (TDH): ∗ Ri = d=dt(xbi Mb ) = Ki∞ Axdi ; ∗ = 33(1 − vt =vg )2 Ki∞

evaluated with the Colakyan correlation [13]. The main inaccuracy of this procedure is due to the supposed particle’s uniform diameter: this implies that, at a certain height, the elutriation phenomena starts suddenly, which is quite diCcult to observe in real cases. • The second one considers abrasion and burn as independent phenomena, then the elutriation is mainly due to the rst one and the elutriated ,ow rate depends on the instantaneous particle diameter (v0 − vmf ), by the amount of C present in the bed and, nally, by the friction coeCcient [12]. Following the elemental reactor approach, the mass of char at a given height is obtained by subtracting from the initial mass of C the amount consummated by the chemical reactions and the elutriation. This second calculation scheme leads to a more gradual depiction of the elutriation phenomena. 4. Evaluation of the overall reaction kinetics The general biomass gasication reaction can be written as follows, starting from one mole of a generic biomass CH O : CH O + yO2 + ztot N2 + wH2 O = x1tot C + x2tot H2 + x3tot CO + x4tot H2 O + x5tot CO2 + x6tot CH4 + ztot N2 = (x1d )C + (x2d + x2b ) H2 + (x3d + x3b )CO + (x4d + x4b ) H2 O + (x5d + x5b ) CO2 + (x6d + x6b ) CH4 + (zd + zb )N2 . The chemical gasication reactions considered in the present mathematical model are [1,7]: C + CO2 = 2CO; C + 2H2 = CH4 ;

C + H2 O = H2 + CO; H2 O + CH4 = CO + 3H2 :

The pyrolysis and partial combustion are supposed to be much faster than the gasication, then the re-

125

lated solid (carbon) and volatile products (CO2 , CH4 , H2 O) are xed as the initial gasication species. Their amount depends on the composition and type of biomass considered. The only hydrocarbon accounted in the syngas has been CH4 , while the other minor species Cn Hm have been neglected and eventually included in the CH4 . With respect to the merely chemical models, the mass transfer contribution to the overall reaction speed has been added: vit = kc

6 xid : dp

Two serial e9ects then limit the gasication speed: chemical kinetics and mass transfer. Then, the rate of the ith chemical species follows (electrical model analogy): vit vikin : vi = vit + vikin The char particle diameter at the ith time, is given by the sum of the chemical reaction and abrasion e9ects dp (ti ) =

dp (ti−1 )(xi (ti )=xi (ti−1 ))1=3 : 1 + (nf − 1)p(1 − xi (ti )=xi (ti−1 ))

If the chemical reaction kinetic is the slowest process, the char particle reduces its density by increasing its porosity and leaving its external diameter dp almost unchanged. If at tdis time the chemical kinetic becomes the limiting e9ect, the expression of the density (accounting for the char consumption) when t ¿ tdis is   xi (tdis ) − xi (t) : (t) = (tdis ) 1 − xi (tdis ) The mass transfer e9ects are predominant in the lowest part of the reactor, where high temperature levels are present: then, tdis ¿ 0; obviously, tdis is a function of the temperature along the bed height. The temperature has been evaluated by a thermal balance along each of the elemental reactors into which the ,uidized bed has been divided (time interval tj+1 − tj ): ns 

xitot (tj )hitot +

itot =1

=

ns 

xitot (tj )hfitot

itot =1 ns 

xitot (tj+1 )hitot +

itot =1

ns 

xitot (tj+1 )hfitot + ql (tj ):

itot =1

126

D. Fiaschi, M. Michelini / Biomass and Bioenergy 21 (2001) 121–132

The heat exchange between the gas and the inert lling material (sand) has been neglected, considering the elemental virtual reactor as a homogeneous ,uid with uniform temperature [14]. However, the model also has the possibility of working at a xed value of temperature, constant along the whole bed height, like the previous basic models [1,7]. 5. Basic model results The composition of the product gas versus the bed height in both dense and bubble phases is shown in Figs. 2(a) and (b); the related basic data are reported in Table 1. Fig. 3 shows the resulting overall gas composition: the equilibrium composition is practically reached after half of the reactor height. It is interesting to remark the correlation between the bubble phase composition and the void fraction

Fig. 2. (a) Gas composition versus bed height in dense and bubble phases (dense phase); (b) Gas composition versus bed height in dense and bubble phases (bubble phase).

b (Fig. 4): at high values, (b ¿ 0:3) the inter-phase mass transfer is high, since a great amount of mass is present in the bubble phase. Successively, the gasication products such as H2 and CO are rapidly transferred from the dense phase (where they originate) to the bubble phase. Moreover, the mass transfer coeCcient is high enough to allow a fast stabilization of the concentration gradients and the related mass transfer between the two phases. When both the bubble phase fraction and the chemical reactions speed are consistently lowered (approach to the equilibrium), a partial bubble-dense phase transfer can be observed for the gasication products. Also, the pyrolysis products, such as CH4 , CO2 and H2 O, decay in the bubble phase; the dense phase, after the starting, increased due to the increase in its volume, moves toward the stabilization while approaching the equilibrium. Finally, Fig. 4 shows the non-linear trend of the minimum ,uidization velocity versus the reactor height: this is due to vmf , which depends on the di9erence between the lling sand and gas densities as well as on the inverse of gas viscosity, the latter decreasing with lowering temperature. The temperature behavior along the reactor axis is shown in Fig. 5: a rapid decay can be observed just after the Flame Pyrolysis (FP) zone, where the endothermic gasication reactions start and a large and rapid consumption of the combustion heat takes place. Along the remaining 85% of the reactor height, the temperature gradient is much lower, with values rang◦ ◦ ing between 900 C and 750 C, which are on line with those from commercial ,uidized bed gasiers. The above thermal balance can be regarded also as a rst principle verication of the model along the bed height. The temperature at the transition time tdis has been ◦ evaluated to be about 1610 C. Two di9erent char consumption mechanisms can be drawn at model level: char particle diameter reduction at the starting process, due to the surface reaction and the abrasion mechanisms; subsequently, after tdis , the reaction kinetic becomes dominant and the reacting char particles increase their porosity, with low reduction of the external diameter. It implies high values of dp at the reactor exit, but the related particles density is strongly reduced, so they are characterized by a great amount of voids, accounting for the e9ective char consumption (see Figs. 4 and 5). This mechanism is rather di9erent

D. Fiaschi, M. Michelini / Biomass and Bioenergy 21 (2001) 121–132

127

Table 1 Basic data for the gasication model run Biomass composition (sawdust)

CH1:41 O0:59



3

Humidity (dry basis) ER Oxidizer Bed height Bed diameter, dg Initial char particles diameter Char density Inert particles diameter Inert particles density

10% 0.33 Air 2:1 m 1:30 m 5 mm 1500 kg=m3 800
m 2600 kg=m3

Primary fragmentation factor n1 Secondary fragmentation factor n2 Oxidizer inlet speed vo Reactor wall thickness External convection Wall conductivity External temperature kat (elutriation coeCcient) Gasier pressure

1.1 1.8 1 m=s 0:25 m 10 W=m2 K 0:1 W=m K ◦ 30 C 5 × 10−6 1:013 bar

Fig. 3. Overall gas composition versus bed height.

Fig. 4. Bubble phase fraction, particle diameter, bubbles diameter and minimum ,uidization velocity versus bed height.

from those reported in [1,7], where the char density is considered to be constant and the char consumption is accounted by the reduction of the particle diameter dp . Obviously, particles with such high voids fraction tend to desegregate, following the “percolation frag-

Fig. 5. Gasication temperature, char particle density and bubbledense mass transfer coeCcient versus bed height.

mentation” mechanism; however, this aspect has not been accounted in the present model. Finally, Figs. 4 and 5 show also the behavior of the bubbles diameter Db and the inter-phase mass transfer coeCcient kbd along the reactor axis: the initial low values of Db promote the mass transfer between the two phases, due to the high bubble external specic surface, then the high kbd values can be observed. This parameter, however, remains high enough to allow a fast inter-phase mass transfer along the whole reactor, regardless of its initial rapid decay. 6. Sensitivity analysis: e'ects of the main operating variables The in,uence of the main parameters (Equivalence Ratio ER, pressure and surface gas velocity) on the gasication process has been investigated.

128

D. Fiaschi, M. Michelini / Biomass and Bioenergy 21 (2001) 121–132

Fig. 6. Molar fractions of the di9erent species versus ER (bed height = 0:25 m).

Fig. 7. A 70% char conversion ratio z(tC0:3 ) and syngas caloric value at t = tC0:3 versus ER.

The in,uence of ER is due to two important factors: 1. bottom reactor temperature increases with ER; 2. at xed values of temperature, the gasication reactions are predominant with respect to the combustion reactions, up to ER = 0:4. In this way, the gasication yield initially increases with ER, and then it starts decreasing when the combustion reactions become predominant. If the bed height is limited (and the residence time is therefore low, Fig. 6), relatively high ER values (¿0:4) are required to have a good syngas yield. For ER ¡ 0:2, the initial temperature is not suCcient to reach the 90% conversion of the char (C0:1 ). Accounting only the 70% degree of the char conversion (C0:3 ), the corresponding bed height decreases with increasing ER, but the gasication yield shows an opposite behavior (see Fig. 7). Then, the ER values giving a suCcient initial temperature to make the gasication process eCcient for limited bed heights (approximately 1:5 m) are between 0.3 and 0.4. This result is well evidenced by a two-phase model, because it emphasizes the importance of the mass-transfer e9ects for limited reactor heights, where the temperature is relatively high. The surface gas velocity does not have a direct in,uence on the chemical reaction rates, but only on the residence time, on the bubble phase reactor volume (b ) and on the inter-phase mass transfer coeCcient kbd (by b and Db ). For surface velocity close to the

Fig. 8. The tC0:1 bed height versus surface gas velocity with local and mean overall Db evaluation.

minimum bubbling value (about 0:5 m=s in this case) the bubble phase fraction is minimum and only the dense phase can be accounted for in the reactor. Increasing the surface gas velocity b increases: at v0 values close to 1:1 m=s, almost 100% of the lower reactor part is on a bubble phase, that undergoes a rapid decay along the bed height. However, it remains higher than in the cases with lower v0 . In this way, the reactor volume available for the gasication reactions decreases in the rst part of the gasier and the related gasication yield is lowered. Fig. 8 shows this e9ect in terms of char conversion level: the increase in the bubble phase fraction with v0 leads to an increase in the reactants molar fractions for this phase, so they cannot directly take part in the gasication until the dense phase transfer has been completed.

D. Fiaschi, M. Michelini / Biomass and Bioenergy 21 (2001) 121–132

Fig. 9. Inter-phase mass transfer coeCcient versus v0 at di9erent bed heights.

The inter-phase transfer coeCcient kbd decreases with increasing v0 , leading to a slower transfer of the reactants to the dense phase and, nally, of the products to the bubble phase (Fig. 9). Adopting the previously reported integral expression of bubble diameter De [11], an optimizing value of v0 in terms of char conversion z(C0:1 ) has been identied close to 1:1 m=s due to the optimal inter-phase mass transfer conditions (part of the reactants and products are in both phases). In this way, the reactants of the dense phase allow a high degree of solid–gas reactivity and, simultaneously, a consistent part of the products is transferred to the bubble phase, allowing the gasication reactions to be shifted toward the syngas products (Fig. 8). The gasication pressure in,uences the reaction kinetics and equilibrium, the bubbles diameter and the bed voids fraction b . Generally, an increase in pressure leads to a reduction of the bubble diameters and to an initial reduction of residence time: this, however, starts increasing at relatively high pressures (over 20 bar for the inquired biomass composition). The bed temperature increases with pressure: this is substantially due to the Buduard’s reaction (C + CO2 = 2CO), that is shifted to the left-hand side and leads to an overall reduction of the gasication process. Also the limiting mass transfer=chemical kinetics temperature increases with the gasication pressure (Fig. 10), showing that a more restricted part of the reactor is kinetically mass-transfer limited. Since the maximum temper-

129

Fig. 10. Char conversion ratio and temperature discriminating the dominant mechanism in the gasication process versus gasication pressure (bed height = 2 m).

ature at the beginning of the gasication process (with reference to the selected biomass, see Table 1) ◦ is 1648 C, for pressure values over 20 bar, the reactions are completely limited by the chemical-kinetics. The char conversion ratio versus the gasier pressure shows (for a 2 m height bed) an optimizing value of 5 bar (Fig. 10): this is still mainly due to the optimization of the Buduard’s reaction, together with the consistent inversion of the methanation reaction. At this pressure level, the maximum CO and H2 and the minimum CO2 and CH4 molar fractions are found (this result substantially agrees with those reported in [1]). 7. Comparison between single phase and double phase models The two phase and the single-phase (b = 0) approaches have been tested and compared. The same basic data have been assumed for both cases; the surface gas velocity v0 = vmin has been set for the single-phase runs. The presence of two phases generally lowers the overall reaction kinetics, especially during the rst few steps (Fig. 11), while the compositions tend to overlap moving toward the equilibrium conditions at the bed top. This behavior is due to the presence of all the reactants only in the dense phase, for the single-phase model. For the two-phase model, when the bubble fraction is suCciently reduced, the kinetic becomes similar to that of single-phase cases. The methanation reaction is mostly in,uenced by the presence of

130

D. Fiaschi, M. Michelini / Biomass and Bioenergy 21 (2001) 121–132

Fig. 11. Comparison between the e9ects of Chemical Kinetics + Mass Transfer (CK + MT) and Chemical Kinetics only (CK).

the bubble phase: in the reference single-phase models, hydrogen is not initially present, which leads to a fast inversion of this reaction with an immediate production of H2 and the subsequent rapid methane consumption. Given the relatively high starting temperature, the mass transfer e9ects have been found to be important up to tdis ; after this time, the results of the two approaches tend to resemble each other, since the chemical kinetic becomes the slowest process. On the whole, it can be concluded that only the initial gasication reactions steps are substantially in,uenced by the mass transfer e9ects, while at the top of the reactor only the chemical kinetic plays a dominant role on the overall reaction mechanisms (Fig. 11). 8. Model veri-cation A preliminary test validity of the developed kinetic model has been carried out using some experimental data from the literature. These data often missed some of the needed topic parameters for the simulation: in these cases, they have been reasonably assumed. The comparison of the numerical results with the experimental data, also allowed an improvement in the settings of some reactions’ chemical kinetic parameters, such as the activation energy and the frequency factor. Three of these test runs have been chosen to be reported in the present work. Keeping the same data entry of [5] (Table 2), the rst test was performed with two runs of wood chips

gasication (RUN 1, Table 2, Figs. 12 and 13). The temperature value at the bed top is always underestimated: this is due to the overestimation of the axial energy transfer between the elements into which the ,uidized bed has been divided at a model level. As a consequence of these low nal temperatures, the gasication process stops earlier, leading to an overall underestimation of the outlet char conversion ratio. Accuracy within 12% for the caloric value and the volumetric fractions of the di9erent gas species were registered; H2 is highly overestimated. The second test run was performed using the data collection from an agricultural wastes gasication process [15] (RUN 2, Table 2, Fig. 14). The parameter accounting the volatile matter at the gasication process start up has been reset superimposing the CH4 relative errors to zero. The only experimental available data were the caloric value and H2 ; CO and CH4 + Cn Hm volumetric fractions: the agreement of the calculations with the available experimental data is quite satisfactory, showing an overestimation of H2 , while the other volumetric fractions have been well predicted. Finally, a test run was performed with RDF (RUN 3, Table 2): the basic available data are shown in Table 2 [16]. Fig. 15 summarizes the related results, for both air and oxygen as oxidizers. In the rst case, the agreement between theoretical and experimental data is quite satisfactory, while it is poor in the second case. It is mainly due to the initial higher temperature, which leads to an over estimation of the Buduard reaction speed (and the related CO production) for oxygen gasication. A test with constant xed bed temperature has been performed for the oxygen gasication, which brought a consistent improvement to the predicted data. 9. Conclusions The present gasication kinetics model has been developed starting from previous zero-dimensional models. It is one-dimensional and biphasic, which allows the mass transfer e9ects to be accounted on the reaction kinetics mechanisms. The gasier freeboard is considered to be inert. The introduction of the two phases allowed to point out that the mass transfer activity is an increasing

D. Fiaschi, M. Michelini / Biomass and Bioenergy 21 (2001) 121–132

131

Table 2 Basic experimental test data conditions RUN 1 Wood chips 1.1 ER Moisture (%) Biomass composition Bed height (m) Bed diameter (m) Wall thickness (m) ◦ Wall conductivity (W=m C) Ext. ,uid convection coe9. (W=m2◦ C) ◦ Ext. Temp. ( C) Sand diameter (mm) Sand density (kg=m3 ) Char initial diameter (mm) Char initial apparent density (kg=m3 )

0.298 16.2

Wood chips 3.1

0.273 17.4 CH1:41 O0:59 1 0.3 0.1 0.03 6 25 0.5 2670 12 1500

RUN 2

RUN 3

Almond shell

Grass straw

Corn stalk

Pelletized RDF

0.132 33.0 CH1:59 O0:70

0.157 42.5 CH1:38 O0:65 0.75 0.66 0.25 0.03 6

0.162 43.5 CH1:53 O0:74

0.27– 0.41 5.8 CH1:6 O0:56 0.5 0.15 — 0.03 6

7

25 0.58 2670 6 1500

6

Fig. 12. Model verication for RUN 1.1.

Fig. 14. Model verication for RUN 2.

Fig. 13. Model verication for RUN 3.1.

Fig. 15. Model verication for RUN 3.

25 7 2600 9 1500

132

D. Fiaschi, M. Michelini / Biomass and Bioenergy 21 (2001) 121–132

function of void fraction b and it is lowered approaching the equilibrium. One of the main model’s features is the possibility to evaluate the temperature behavior along the bed height, that is also a good design tool when the parameters are well set with detailed experimental data. The surface gas velocity generally leads to an increase in the bubble diameter and to a decrease in the inter-phase mass transfer. For small height beds, increasing ER brings an increase in the gasication products due to the higher temperature and the subsequent reactions speed. Increasing the reactor pressure (up to a certain level depending on the biomass composition) implies a reduction of the bubble diameter, as well as the residence time of the reactants in the gasier. The temperature initially increases with pressure; then, a stabilization occurs, which brings the reactions to be only kinetically limited. The comparison between mass transfer and surface reactions kinetics e9ects on the whole gasication mechanisms has shown that the rst prevails at the starting process, due to the high temperature level; successively, when the temperature is stabilized, the latter one plays a dominant role. The model has shown a substantial agreement with other available gasication kinetic models. A preliminary validation with some experimental data from the literature has been carried out, showing, on the whole, quite encouraging results. A substantial improvement of the model e9ectiveness could be reached if more detailed (ad hoc) experimental data were available to allow a better setup of the main parameters. In this way, the whole model could be regarded as a start to be supported and improved by specially dedicated test campaigns. References [1] Wang Y, Kinoshita CM. Kinetic model of biomass gasication. Solar Energy 1993;51(1):19–25.

[2] Reed TB. Biomass gasication—principles and technology. Park Ridge, New Jersey, USA: Noyes Data Corporation, 1981. [3] Gaur S, Rao TR, Reed TB, Grover PD. Kinetics of corn cob char gasication in carbon dioxide. Fuel Science and Technology International 1992;10(9):1461–99. [4] Shi Chen J, Gunkel WW. Modeling and simulation of co-current moving bed gasication reactors—Part II. A detailed gasier model. Biomass 1987;14:75–98. [5] Bilodeau JF, Therien N, Proulx P, Czernik S, Chornet E. A mathematical model of ,uidized bed biomass gasication. The Canadian Journal of Chemical Engineering 1993;71(4):549– 57. [6] Ergudenler A, Ghaly AE, Hamdullahpur F, Al Taweel AM. Mathematical modeling of a ,uidized bed straw gasier. Part III, model verication. Energy Sources 1997;19(10):1099– 121. [7] Fiaschi D, Bettagli N, Desideri U. A Biomass Combustion— Gasication model: validation and sensitivity analysis. Journal of Energy Resources Technology, A.S.M.E. Transaction 1995;117:329. [8] Deemter V. Mixing. In: Davidson JF, Clift R, Harrison D, editors. Fluidization. 2nd ed. London: Academic Press, 1985, p. 331–55. [9] Darton RC. The physical behaviour of three-phase ,uidized beds. In: Davidson JF, Clift R, Harrison D, editors. Fluidization. 2nd ed. London: Academic Press, 1985, p. 495 –528. [10] Clift R, Grace JR. Continuous bubbling and slugging. In: Davidson JF, Clift R, Harrison D, editors. Fluidization. 2nd ed. London: Academic Press, 1985, p. 73–132. [11] Cai P, Schiavetti M, De Michele G, Grazzini GC, Miccio M. Quantitative estimation of bubble size in PFBC. Powder Technology 1994;80:99–109. [12] Arena U, Chirone R, D’Amore M, Miccio M, Salatino P. Some issues in modelling bubbling and circulating ,uidized bed in coal combustor. Powder Technology 1995;82:301–16. [13] Geldart D. Elutriation. In: Davidson JF, Clift R, Harrison D, editors. Fluidization. 2nd ed. London: Academic Press, 1985. [14] Xavier AM, Davidson JF. Heat transfer in ,uidized beds: convective heat transfer in ,uidized beds. In: Davidson JF, Clift R, Harrison D, editors. Fluidization. 2nd ed. London: Academic Press, 1985. [15] Shoeters J, Maniatis K, Buekens A. The ,uidized bed gasication of biomass: experimental studies on a bench scale reactor. Biomass 1989;19:129–43. [16] Lamorey GW, Jenkins BM, Gross JR. LP engine and ,uidized bed gas producer performance—ASAE. Paper No. 86-3072, 1986.