Solar Energy 74 (2003) 8591

A steady state model of gas-char reactions in a downdraft biomass gasier

D.L. Giltrap a , R. McKibbin b , G.R.G. Barnes a , *
b

Institute of Fundamental Sciences, Massey University, Palmerston North, New Zealand Institute of Information and Mathematical Sciences, Massey University, Auckland, New Zealand Accepted 10 February 2003

Abstract A phenomenological model of downdraft gasication under steady state operation is developed based on previously published values for the reaction kinetics in the reduction zone. The model predicts a product gas with a composition similar to that found experimentally, although the model over-predicts the methane concentration. The accuracy of the model is limited by the availability of data on the initial conditions at the top of the reduction zone. 2003 Elsevier Science Ltd. All rights reserved. Keywords: Downdraft gasication model

1. Introduction Thermochemical gasication is a process for converting solid fuels into gaseous form. The chemical energy of the solid fuel is converted into both the thermal and chemical energy of the gas. The chemical energy contained within the gas is a function of its chemical composition. Thus the chemical composition of the product gas determines its quality as a fuel. High concentrations of combustible gases such as H 2 , CO and CH 4 increase the combustion energy of the product gas. Downdraft gasication is a comparatively cheap method of gasication that can produce a product gas with very low tar content. Fig. 1 shows the major features of a downdraft gasier. In the region nearest the air inlet, aming pyrolysis processes occur. Highly exothermic combustion reactions provide the energy to pyrolyse (devolatilize) the solid fuel, and these two processes can occur nearly simultaneously. The temperature in the aming pyrolysis region is sufciently high to thermally crack the pyrolysis products into components of low molecular weight. In practice, factors such as the ow properties of the solid biomass and
* Corresponding author. Tel.: 1 64-6-356-9099; fax: 1 64-6350-5682. E-mail address: G.R.Barnes@massey.ac.nz (G.R.G. Barnes).

the requirement that input air be distributed evenly throughout the entire cross-section of the bed limit the maximum diameter of downdraft gasiers. Beneath the aming pyrolysis zone is the reduction zone where CO 2 and H 2 O are reduced by the following endothermic reactions on the char surface: Reaction 1: C 1 CO 2 2CO Reaction 2: C 1 H 2 OCO 1 H 2

Fig. 1. Schematic diagram of a downdraft gasier.

0038-092X / 03 / $ see front matter 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016 / S0038-092X(03)00091-4

86

D.L. Giltrap et al. / Solar Energy 74 (2003) 85 91

Nomenclature A Ai CRF cx Dp Ei fP Ki Nx n nx P Px R Rx ri T v m r Cross-sectional area of the gasier bed in m 2 Frequency factor for reaction i Char Reactivity Factor Molar heat capacity of the species x in J ? mol 2 1 K 2 1 Average particle diameter in the bed Activation energy for reaction i in J ? mol 2 1 Pyrolysis fraction Equilibrium constant for reaction i Number of moles of species x Molar concentration of all gaseous species in mol ? m 2 3 Molar concentration of species x in mol ? m 2 3 Total pressure in Pa Partial pressure of species x in Pa Gas constant in J ? mol 2 1 ? K 2 1 Net rate of creation of species x by chemical reactions in mol ? m 2 3 ? s 2 1 Rate of reaction i in mol ? m 2 3 ? s 2 1 Temperature in K Supercial gas velocity in m.s 2 1 Void fraction of the bed Fluid viscosity in kg ? m 2 1 ? s 2 1 Mass density of the uid in kg ? m 2 3

The chemical composition of a gasier product gas is often not equal to the equilibrium composition. Gumz (1950) used experimental data to calculate the degree of approach to equilibrium of the two reduction reactions above and the methane formation reaction: Reaction 3: C 1 2H 2 CH 4 The degree of approach to equilibrium of Reaction 3 was difcult to determine due to the presence of methane from pyrolysis of the fuel in experiments. Wang and Kinoshita (1993) modelled the kinetics of the reactions occurring in the reduction zone assuming a given gas residence time and temperature. This model is physically similar to a well-mixed uidised bed; however in a downdraft gasier the gas travels down the bed as it reacts, so there is one-dimensional variation in the gas temperature and composition. We used the reaction kinetics parameters found by Wang and Kinoshita to build a model of the gas composition and temperature within a downdraft gasier operating under steady-state conditions.

region where we assumed that all the oxygen from the air inlet had been combusted to CO 2 and that the pyrolysis products had been completely cracked. Solid carbon in the form of char was assumed to be present throughout the reduction region. The following reactions were considered: Reaction 1: C 1 CO 2 2CO Reaction 2: C 1 H 2 OCO 1 H 2 Reaction 3: C 1 2H 2 CH 4 Reaction 4: CH 4 1 H 2 OCO 1 3H 2 This is the same reaction scheme used in Wang and Kinoshita (1993), where it was found that including the watergas shift reaction (CO 1 H 2 OCO 2 1 H 2 ) had little effect on the nal result. We assumed that there was always a plentiful supply of carbon from the fuel, so the chemical species that needed to be tracked were CO 2 , CO, CH 4 , H 2 O and H 2 . N 2 from the inlet air was also tracked. Although N 2 does not participate in any of the reactions considered, it does dilute the nal product gas. The reaction rates were all considered to have an Arrhenius-type temperature dependence and to be proportional to the difference between the actual reactant / product ratio and the corresponding equilibrium ratio. Thus we obtained:

2. Chemical reactions The pyrolysis and cracking reactions were not considered in this model as the number of possible pyrolysis products along with all the possible reactions and intermediate products would make the model very complex. Instead we started the model at the top of the reduction

D.L. Giltrap et al. / Solar Energy 74 (2003) 85 91

87

2 E1 P2 CO r 1 5 A 1 exp ]] ? PCO 2 2 ] RT K2

S DS S DS S DS S DS

D D

(1)

Table 1 Net rate of production of the different gaseous species by chemical reactions in terms of the rates of reactions 1 to 4 Species R x (mol ? m 2 3 ? s 2 1 ) 0 2 r1 2r 1 1 r 2 1 r 4 r3 2 r4 2 r2 2 r4 r 2 2 2r 3 1 3r 4 r 1 1 r 2 2 r 3 1 2r 4

PCO .PH 2 2 E2 r 2 5 A 2 exp ]] ? PH 2 O 2 ]] RT K3 PCH 4 2 E3 ]] r 3 5 A 3 exp ]] ? P 2 H2 2 RT K4

(2)

(3)

3 PCO P H 2 E4 2 r 4 5 A 4 exp ]] ? PCH 4 ? PH 2 O 2 ]] RT K5

N2 CO 2 CO CH 4 H2O H2 Total number of gas molecules

(4)

where A i is the frequency factor for reaction i, Ei is the 21 activation energy of reaction i in J ? mol , R is the gas 21 21 constant in J ? mol ? K , T is the temperature in K, Px is the partial pressure of gaseous species x, and Ki is the equilibrium constant of reaction i. The values for the activation energies for these reactions were taken from Wang and Kinoshita (1993). However, we incorporated several factors into our constant pre-multipliers, such as char particle size and number of active carbon sites, that were considered separately by Wang and Kinoshita. So to calculate our frequency factors we multiplied the values found by Wang and Kinoshita by a Char Reactivity Factor (CRF ), that represents the relative reactivity of different char types. That is, we used A i 5 9 is the value found by Wang and A9 i ? C RF where A i Kinoshita.

d(n x v) ]] 5 Rx dz or dnx 1 dv ] 5 ] Rx 2 nx ] dz v dz

(6)

(7)

3. Mass and energy balances We assumed a cylindrical gasier bed of uniform crosssectional area A with negligible radial variation in the properties of both the bed and the gas. The axial distance is z. The gases were assumed to behave ideally. We then considered thin cross-sections of thickness Dz. For each chemical species in the gas, the change in the molar ow rate across Dz must equal the net rate of creation of species by chemical reactions within the volume. Therefore for small Dz we have: nx (z) Av(z) 5 n x (z 1 Dz) Av(z 1 Dz) 1 Rx A Dz (5)

Similarly, for steady-state operation, any change in the energy ow rate over distance Dz must be equal to the rate of energy released by reactions within the volume minus the rate of work done by the gas. This neglects changes in the kinetic and gravitational potential energy of the gas which are relatively insignicant. The thermal energy of N moles of gas species x is Nc x T where c x is the molar heat capacity of the gas in J ? K 2 1 ? mol 2 1 . For a chemical reaction the change in enthalpy DH in J / mol represents the change in the internal chemical energy of the system due to the reaction. Thus the rate of energy released due to chemical reactions is 2 o i r i ? DHi where i ranges over all the reactions being considered. So from the energy balance we obtained the following relation: vz 1D z A

SOn c TD
x x x

z 1D z

2 vz A

SOn c TD 5 2Or
x x x z i

? DHi A ? Dz 2 DsPAvd

(8)

where n x is the molar density of species x in mol ? m 2 3 , v is the supercial gas velocity in m ? s 2 1 , and R x is the net rate of formation of species x by chemical reactions in mol ? m 23 ? s 21. Eq. (5) can be used to generate expressions for the molar concentrations of each of the six gaseous species considered (CO 2 , CO, H 2 O, H 2 , CH 4 , N 2 ). Similarly the total molar density of all gases, n, can be expressed as the sum of the molar densities of each of the six component gases. Table 1 gives the net rates of formation of each gas species in terms of the reaction rates of Reactions 14. Rearranging Eq. (5) and taking the limit as Dz 0 gives:

where P is the total pressure (in Pa), the subscript x ranges over all chemical species and the subscript i ranges over all chemical reactions under consideration. Dividing by A Dz and taking the limit as Dz 0 gives: d v nx cx T x ]]]] 52 dz

SO

d(Pv) Or DH 2 ]] dz
i i i

(9)

Rearranging we get: dT 1 ] 5 ]]] 2 dz v ? n c

x x x

O S O
i

dP dv r i DHi 2 v ] 2 P ] 2 dz dz

O R c TD
x x x

(10)

88

D.L. Giltrap et al. / Solar Energy 74 (2003) 85 91

Eqs. (7) and (10) give a total of seven differential equations when all six gaseous species are considered. However, the state of the gasier at any given point is described by the concentration of each gas species, the temperature, the supercial gas velocity and the total pressure. So to complete our system of differential equations we still need expressions for dP / dz and dv / dz. The pressure gradient, dP / dz, is known to depend upon the supercial gas velocity. The following empirical formula relates the pressure gradient of a uid owing through a bed of solid particles to the physical properties of the bed and the uid (Ergun, 1952): dP 150m (1 2 )2 1.75r (1 2 ) 2 2 ] 5 ]]]] v 1 ]]]] v 3 dz D2 Dp 3 p (11)

computer software. We used the ODE45 function in MATLAB.

4. Initial values The model can predict the state of the gasier anywhere within the reduction region given a complete set of measurements of the nine system variables at a single point. In practice, measurements are usually made at the exit point of the gasier. However, as a predictive tool the model could be used to predict the gasier output for a given initial state at the top of the gasier.

4.1. Initial gas concentrations

where m is the uid viscosity, is the void fraction of the bed, Dp is the average particle diameter in the bed and r is the mass density of the uid. 2 We found coefcients for v and v empirically by measuring the pressure drop across a bed of pine chips of xed length at room temperature over a range of different air speeds. Of course, the bed of an operating downdraft gasier will vary in composition from fresh fuel at the top to smaller char particles just above the grate. To get an expression for dv / dz we took the ideal gas law: P 5 nRT (12) The initial position in this model is the top of the reduction region. That is the point at which all the oxygen from the air inlet has been consumed in combustion reactions. It is also assumed that all the pyrolysis and cracking reactions have been completed by this point. The gas at the top of the gasier comes from two different sources: from the inlet gas (usually air, but pure oxygen, steam or CO 2 are sometimes used) and from pyrolysis of the fuel. The exact proportion of each depends upon the rate of air owing into the gasier relative to the rates of the combustion, pyrolysis and cracking reactions. Rather than attempt to calculate this directly we introduced a pyrolysis fraction variable ( fP ), which can vary from 0 (no pyrolysis products) to 1 (pyrolysis products only). The air owing into the gasier is a mixture of N 2 and O 2 . We assumed that all the O 2 from the air inlet had been consumed by combustion reactions with the char (C 1 O 2 CO 2 ), while the N 2 remains inert. The pyrolysis products were assumed to have been cracked into an equivalent amount of CO, CH 4 and H 2 O. We took the pyrolysis products of Douglas r bark, with the empirical formula CH 3.03 O 1.17 (Graboski and Bain, 1981), as being typical of biomass. By an atomic balance we found that each mole of pyrolysis gas therefore contains 0.341 mol CO, 0.277 mol CH 4 and 0.382 mol H 2 O. This assumes a dry feedstock. Additional moisture content in the feedstock can be accounted for by increasing the proportion of H 2 O in the pyrolysis gas. Wet fuel will also lower the temperature within the gasier as additional energy is required to vaporise the water.

where n is the molar concentration of all gaseous species in mol ? m 2 3 , P is the total pressure in Pa, R is the gas 21 constant in J ? mol , and T is the temperature in K. Differentiating both sides with respect to z gives: dP dn dT ] 5 ]RT 1 nR ] dz dz dz (13)

An expression for dn / dz can be found by summing the expressions for dn x / dz from Eq. (7) for each gaseous species x. This gives expressions for dn / dz and dT / dz that depend upon dv / dz. Substituting Eqs. (7) and (10) into (13) and rearranging gives: r i DHi nx cx Rx dv 1 x x i ] 5 ]]]] ]]] 2 ]]] dz n T n c 1 nR

O
x

SO
O

O O
x

x x

v nx cx dP v x 2 ] ] 1 ]] 2 dz T P

DO D
Rx cx

(14)

4.2. Initial gas temperature

In the combustion region energy is required to dry the incoming fuel, drive the pyrolysis / cracking processes and to raise the temperature of the incoming gases. This energy is supplied by the exothermic combustion reactions. Thus the steady-state temperature will be the temperature at

This gives us a set of nine coupled rst-order ordinary differential equations in the system variables n x (where x ranges over the six different gas species considered), P, v and T, that can be solved using commercially-available

D.L. Giltrap et al. / Solar Energy 74 (2003) 85 91

89

which the rate of energy production matches the rate of energy consumption, and we expect lower temperatures for fuels with high moisture contents or a large fraction of volatiles. Senelwa (1997) found the average temperature just beneath the air inlets when gasifying an oven-dried biomass feedstock to be 955 8C (1228 K). We therefore used an initial temperature of 1200 K to test our model.

4.4. Pyrolysis fraction

The pyrolysis fraction fP is the effective fraction of the initial gas that comes from pyrolysis and cracking reactions. The value of fP will depend upon the proportion of volatile matter in the fuel, the rate at which air enters the system and the relative kinetics of the pyrolysis, cracking and combustion reactions. Rather than attempt to calculate this we examined the effect of different fP values on the model output. The air entering the gasier is 79% N 2 by volume. No N 2 is produced by the reactions considered and it does not participate (signicantly) in any reactions. So the N 2 concentration can be used to infer fp . Fig. 3 shows the N 2 concentration within the gasier for different values of fp . Air-blown downdraft gasiers typically produce gas containing | 50% N 2 (as a percentage of the dry gas volume). Therefore Fig. 3 shows that fP values between 0.4 and 0.5 produce results typical of downdraft gasication of biomass.

4.3. Char reactivity factor ( CRF )

The CRF for different chars were not known. However, the number of reactions occurring within a bed of xed length L is related to L, the CRF , the supercial gas velocity v, the reaction rate r i and the cross-sectional area of the bed A. Fig. 2 shows the temperature proles as the CRF value is varied using an fP of 0.5, v of 1 m ? s 2 1 , A of 3.14 3 10 2 2 m 2 and a starting temperature of 1200 K. As the CRF value increases the temperature within the gasier drops more rapidly as the gas undergoes more endothermic reactions in the time it takes to travel the length of the gasier. The outlet temperature found in Senelwa (1997) for a bed length of 0.275 m was 1034 K. Comparing this with the results in Fig. 2, a CRF of the order of 1000 produces a similar result. We have assumed that the CRF remains constant throughout the gasier bed. Experimental results suggest that the reactivity of char depends upon the thermal history and the degree of burnoff experienced by the char (Graboski, 1981). This would result in differences in the CRF of the char along the length of the bed. So the CRF used in this analysis should be considered an effective CRF .

5. Comparison with experimental results We tested our model against data collected from two different downdraft gasiers (Chee, 1987; Senelwa, 1997). Table 2 shows the parameters used in our model compared with those from the experiments. Fig. 4 shows the composition of the dry product gas predicted by the model along with those found experimentally. The model produced reasonable agreement with the experimental results for all components with the exception of CH 4 . This over-prediction of CH 4 is most likely the

Fig. 2. Temperature prole along gasier bed as the CRF value varied.

90

D.L. Giltrap et al. / Solar Energy 74 (2003) 85 91

Fig. 3. N 2 concentration along the gasier as a percentage of the dry gas volume for different initial pyrolysis fractions.

Table 2 Parameters used in experimental downdraft gasication and in the model simulation Parameter Bed length v T initial Moisture content CRF fP Chee (1987) Data not stated not stated not stated 5 wt % Senelwa (1997) Data 0.275 m 0.699 m / s 1228 K oven dried Model 0.275 m 1 m/s 1200 K 0 1000 0.5

result of the assumption that the O 2 in the air reacts only with the char. The pyrolysis products are cracked in a region of high temperature and in the presence of O 2 , so it is probable that some of the CH 4 produced will undergo combustion with O 2 (CH 4 1 2O 2 CO 2 1 2H 2 O). Fig. 5 shows the results when the initial gas concentration was altered under the assumption that all the O 2 in the air reacted with the CH 4 from the cracking of pyrolysis gas. The CH 4 concentration has dropped under this new assumption. Similarly the CO 2 level has also dropped as CH 4 combustion produces less CO 2 per mole of O 2 consumed.

Fig. 4. Composition of the dry product gas predicted by model compared with experimental results.

D.L. Giltrap et al. / Solar Energy 74 (2003) 85 91

91

Fig. 5. Composition of the dry product gas predicted by model compared with experimental results. Initial conditions of model assumed that the CH 4 produced by cracking of pyrolysis products reacted with O 2 entering the gasier from the air inlets.

6. Discussion The model prediction of the gas composition was in reasonable agreement with the experimental results with the exception that the model signicantly over-predicted the CH 4 concentration. The assumption that the CH 4 produced by pyrolysis is rapidly combusted with the oxygen at the air inlets reduced the amount of CH 4 predicted, but the prediction was still higher than the experimental concentrations found. The model output was quite sensitive to the initial temperature and gas composition assumed at the top of the gasier. We also assumed a constant CRF throughout the gasier bed. If the assumed CRF actually decreased along the bed length the effect would be approximately equivalent to reducing the bed length as the major effect is to decrease the number of reactions that occur during the gas residence time. However, if the actual CRF had a higher value over a small region at the top of the of the reduction region followed by a longer region of reduced reactivity, then the gas composition produced could be quite different due to more reactions occurring at higher temperature.

of approximations made. However, the model did signicantly over-predict the methane concentration in the product gas. The model could be improved with more data on the initial gas concentrations at the top of the reduction region, the relationship between the amount of pyrolysis products produced and temperature, and the variation of the char reactivity along the length of the gasier bed.

References
Chee, C.S., 1987. The air gasication of wood chips in a downdraft gasier. MSc Thesis, Kansas State University. Ergun, S., 1952. Fluid ow through packed columns. Chem. Eng. Prog. 48, 8994. Graboski, M., Bain, R., 1981. Properties of biomass relevant to gasication. In: Reed, T.B. (Ed.), Biomass Gasication Principles and Technology. Noyes Data Corporation, New Jersey, pp. 4171. Graboski, M., 1981. Kinetics of char gasication reactions. In: Reed, T.B. (Ed.), Biomass Gasication Principles and Technology. Noyes Data Corporation, New Jersey, pp. 154182. Gumz, W., 1950. In: Gas Producers and Blast Furnaces. John Wiley and Sons, pp. 7695. Senelwa, K., 1997. The air gasication of woody biomass from short rotation forests. Ph.D. Thesis, Massey University, New Zealand. Wang, Y., Kinoshita, C.M., 1993. Kinetic model of biomass gasication. Solar Energy 51, 1925.

7. Conclusion The steady state model produced a reasonable prediction of the output gas composition of a downdraft gasier operating under steady-state conditions given the number