Biomass gasification in a circulating fluidized bed

Abstract

This paper presents the results from biomass gasification tests in a pilot-scale (6.5-m tall × 0.1-m diameter) air-blown circulating fluidized bed gasifier, and compares them with model predictions. The operating temperature was maintained in the range 700–850°C, while the sawdust feed rate varied from 16 to 45kg/h. Temperature, air ratio, suspension density, fly ash re-injection and steam injection were found to influence the composition and heating value of the product gas. Tar yield from the biomass gasification decreased exponentially with increasing operating temperature for the range studied. A non-stoichiometric equilibrium model based on direct minimization of Gibbs free energy was developed to predict the performance of the gasifier. Experimental evidence indicated that the pilot gasifier deviated from chemical equilibrium due to kinetic limitations. A phenomenological model adapted from the pure equilibrium model, incorporating experimental results regarding unconverted carbon and methane to account for non-equilibrium factors, predicts product gas compositions, heating value and cold gas efficiency in good agreement with the experimental data.

Full text

Available online at www.sciencedirect.com
Biomass and Bioenergy 26 (2004) 171 – 193
Biomass gasication in a circulating uidized bed
X.T. Li
a
, J.R. Grace
a;∗
, C.J. Lim
a
, A.P. Watkinson
a
, H.P. Chen
b
, J.R. Kim
c
a
Department of Chemical and Biological Engineering, The University of British Columbia, 2216 Main Mall, Vancouver,
Canada V6T 1Z4
b
National Laboratory of Coal Combustion, Huazhong University of Science and Technology, Wuhan 430074, China
c
Hyundai Industrial Research Institute, Hyundai Heavy Ind. Co., Ltd., 1, Cheonha-Dong, Dong-Ku, Ulsan, South Korea 682-792
Received 10 October 2002; received in revised form 17 April 2003; accepted 28 April 2003
Abstract
This paper presents the results from biomass gasication tests in a pilot-scale (6.5-m tall × 0.1-m diameter) air-blown
circulating uidized bed gasier, and compares them with model predictions. The operating temperature was maintained in
the range 700–850
◦
C, while the sawdust feed rate varied from 16 to 45 kg=h. Temperature, air ratio, suspension density,
y ash re-injection and steam injection were found to inuence the composition and heating value of the product gas. Tar
yield from the biomass gasication decreased exponentially with increasing operating temperature for the range studied.
A non-stoichiometric equilibrium model based on direct minimization of Gibbs free energy was developed to predict the
performance of the gasier. Experimental evidence indicated that the pilot gasier deviated from chemical equilibrium due to
kinetic limitations. A phenomenological model adapted from the pure equilibrium model, incorporating experimental results
regarding unconverted carbon and methane to account for non-equilibrium factors, predicts product gas compositions, heating
value and cold gas eciency in good agreement with the experimental data.
? 2003 Elsevier Ltd. All rights reserved.
Keywords: Biomass gasication; Modelling; Elemental availability; Circulating uidized bed; Tar
1. Introduction
Biomass gasication produces fuel gas or synthe-
sis gas through the chemical conversion of biomass,
usually involving partial oxidation of the feedstock
in a reducing atmosphere in the presence of air, oxy-
gen and/or steam. Air-blown processes produce low
caloric value gases with a typical higher heating
value (HHV) of 4–7 MJ=Nm
3
, while oxygen- and
steam-blown processes result in gases with a HHV
∗
Corresponding author. Tel.: +1-604-822-3121; fax: +1-604-
822-6003.
E-mail address: jgrace@chml.ubc.ca (J.R. Grace).
of 10–18 MJ=Nm
3
[1]. According to a recent survey
[2], there are nearly 100 biomass gasication and/or
pyrolysis installations in Europe and North America.
Various types of gasiers have been explored for
biomass. Updraft moving-bed gasiers suer from
high tar yields in the product gas [3]. The inability to
maintain uniform radial temperature proles and to
avoid local slagging problems makes the moving bed
unsuitable for large installations [4]. Fluidized beds
now nd wide application in biomass gasication [5–
8]. However, due to their high degree of solids mixing
as well as particle entrainment, a single uidized bed
cannot achieve high solids conversion. The circulating
uidized bed (CFB) is a natural extension of the bub-
bling bed concept, with cyclones or other separators
0961-9534/$ - see front matter ? 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0961-9534(03)00084-9

172 X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193
Nomenclature
a air ratio, dened as ratio of actual air
supply to stoichiometric air requirement,
dimensionless
b element abundance vector, dimension-
less
b
0
initial element abundance vector, dimen-
sionless
b
∗
element abundance vector modied with
kinetic carbon conversion, dimension-
less
C carbon conversion, %
C
fa
carbon content in y ash, % of mass
C
f
carbon content in fuel, dry basis, %
E
1
cold-gas eciency excluding tars, %
E
2
cold-gas eciency including tars, %
F ratio of carbon in re-injected y ash to
carbon introduced with fuel, dimension-
less
g acceleration of gravity, 9:81 m=s
2
GCV gross caloric value of fuel, MJ/kg
H
0
298
heat of reaction at thermodynamic stan-
dard state, kJ/mol
HHV higher heating value of product gas,
MJ=Nm
3
, or tars, MJ/kg
h height of a section of the riser, m
K total number of components
L total number of feed streams
M moisture content of feed, kg/kg
(dry basis)
(m) number of iterations, dimensionless
˙m feed rate, kg/h
n moles of a given element or species, mol
N total number of species involved in the
system
P pressure dierence, kPa
T thermodynamic temperature, K
T
3
characteristic operating temperature
measured 3:95 m above primary air
inlet,
◦
C
T
0
eq
equilibrium temperature from best
t to experimental data, K
v
g
gas yield based on unit mass of feed,
Nm
3
=kg
y number of moles, mol
y
t
tar yield, kg/kg-fuel
Greek letters
 elemental availability, or fractional
achievement of equilibrium, dimen-
sionless

C
availability of carbon, dimension-
less

C;1
fraction of carbon converted into
gaseous species, dimensionless

C;2
fraction of carbon converted into
methane, dimensionless

H
availability of hydrogen, dimen-
sionless
 suspension voidage
 total number of phases in system
 density, kg=m
3
Subscripts
0 initial
C, H, N, O, S carbon, hydrogen, nitrogen, oxygen,
sulphur
eq equilibrium model prediction
f biomass feed
fa y ash
g product gas
meas measurement data
p particles
susp suspension
t tars
Superscripts
∗ modied value
o thermodynamic standard state
employed to capture and recycle solids in order to ex-
tend the solids residence time. The riser of a CFB gasi-
er operates in either the turbulent or fast uidization
ow regime. CFB gasication is now undergoing rapid
commercialization for biomass. Fundamental and
pilot studies are, nevertheless, required for scale-up,
as well as to ll gaps in understanding the underlying
principles.

X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193 173
Several dierent types of models [9–13] have been
developed for gasication systems—kinetic, equilib-
rium, and other. Unlike kinetic models that predict
the progress and product composition at dierent
positions along a reactor, an equilibrium model pre-
dicts the maximum achievable yield of a desired
product from a reacting system. It also provides a
useful design aid in evaluating the limiting possible
behaviour of a complex reacting system which is
dicult or unsafe to reproduce experimentally or in
commercial operation.
The objective of this study is to provide practical
data and a theoretical perspective for scale-up from
both pilot-scale experimental study and mathematical
modelling. An experimental study was completed on a
circulating uidized bed gasier to examine the eects
of operating parameters on the gas composition, gasi-
cation eciency and tar yield. The modelling work
started with a non-stoichiometric equilibrium model
based on free energy minimization to predict gasier
performance under ideal equilibrium conditions. The
present study not only gives experimental evidence
that real gasiers deviate from chemical equilibrium
in a number of ways, but also provides a phenomeno-
logical approach to correct the model by introducing
an elemental availability function which corrects for
non-equilibrium of certain components.
2. Experimental study
2.1. Experimental set-up and method
A schematic diagram of the pilot CFB gasier used
in the experiments appears in Fig. 1. The gasier
employs a riser 6:5 m high and 0:10 m in diameter,
a high-temperature cyclone for solids recycle and
ceramic bre lter unit for gas cleaning. Air was
supplied as the oxidant and uidizing agent after
passing through a start-up burner near the bottom of
the riser. Hot gas leaving the burner and pre-heated
air were mixed to preheat the bed and, if needed,
to maintain the suspension temperature at the de-
sired level. The temperatures of both the primary
and secondary air could be varied by adjusting the
total air supply and the fraction of each stream. The
start-up burner preheated the gasier to 400 –550
◦
C
before coal or biomass fuel could be fed to the riser
to further raise the temperature to the desired level.
The system was then switched to the gasication
mode.
Feed particles underwent moisture evaporation, py-
rolysis and char gasication primarily in the riser. The
fast uidization ow regime was maintained at the op-
erating temperature, with a typical supercial velocity
between 4 and 10 m=s, corresponding to an air ow
of 40 –65 Nm
3
=h, and a solids feedrate of 16–45 kg=h
for typical sawdust. The solids throughput was esti-
mated to be 0.7–2:0kg=m
2
s. Coarser particles in the
gas were captured by a high-temperature cyclone im-
mediately downstream of the riser. The solids captured
in the cyclone were recycled to the bottom of the riser
through an air-driven loop seal. Hot gas leaving the
cyclone at a temperature of 600–800
◦
C was cooled
by a two-stage water-jacketed heat exchanger and a
single-stage air preheater before entering the lter unit.
The gasier employs two independent feed systems,
one for the main fuel (biomass) and the other for aux-
iliary fuel (coal), used during start-up. The biomass
feed system consisted of two sealed hoppers, each of
volume 0:3m
3
, a rotary valve with a variable-speed
DC motor and controller, and a screw with a tapered
pitch and sleeve diameter to allow 10% compaction
of the sawdust volume to facilitate feeding. The upper
hopper could be relled while the lower hopper is in
service. A 0:2 m diameter pinch valve isolated the up-
per hopper from the working pressure in the riser, thus
allowing safe relling without interrupting the oper-
ation of the CFB reactor. The coal feed system (not
shown in Fig. 1) employed a hopper, a rotary valve
and a 10 mm dia. pneumatic conveying line connected
with a water-cooled injector 950 mm above the pri-
mary air inlet.
The steam injection system supplied 5 bar saturated
steam. The steam ow rate was measured with an
in-line annubar steam meter, calibrated by weighing
condensate water over a known time interval.
Process data, such as local temperatures and pres-
sures, were logged into a computer. The thermocou-
ple readings in the mid-section of the riser (T
3
) are
taken as the characteristic operating temperatures
unless specied otherwise. Gas sampling ports were
located near the inlet of the heat exchanger. Gas
samples were taken on average every 20 min and
analysed for H
2
, CO, CO
2
,CH
4
,N
2
and O
2
using a
Shimadzu gas chromatograph with a thermal con-
ductivity detector and a Supelco Carboxen-1000

174 X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193
Fig. 1. Schematic diagram of the CFB gasier.
Fig. 2. Tar sampling train: The rst empty bottle acts as a condenser. The three lled ones are tar impingers, with acetone as solvent.
Temperature varies from −3
◦
C to about 35
◦
C in the three impingers.

X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193 175
Table 1
Ultimate analysis of test fuels
a
Fuel type Cypress Hemlock SPF
b
Cedar PS
c
Mixed Average
Carbon wt% 51.6 51.8 50.4 52.3 49.1 48.9 50.9
Hydrogen wt% 6.20 6.20 6.25 6.11 7.26 7.86 6.60
Oxygen wt% 40.4 40.6 41.6 39.9 39.5 40.3 40.5
Nitrogen wt% 0.65 0.60 0.62 0.52 0.25 0.21 0.51
Sulphur wt% 0.46 0.38 0.34 0.39 0.50 0.07 0.34
Ash wt% 0.70 0.40 0.70 0.79 3.34 2.69 1.14
Moisture content wt% 9.7–22.0 8.8–15.0 10.0 10.6 10.1 4.2–6.7 15.0
Higher heating value MJ/kg 20.3 20.3 19.8 20.4 21.1 21.7 20.6
Higher heating value kcal/kg 4840 4850 4720 4880 5030 5170 4920
Stoichiometric air Nm
3
=kg 5.36 5.36 5.20 5.40 5.46 5.56 5.39
Dry bulk density kg=m
3
136 128 119 151 347 465 220
Mean particle diameter mm 1.49 0.92 0.82 0.67 0.38 0.43 0.79
a
All ultimate analyses, heating values and stoichiometric air volumes are on a dry basis.
b
SPF = spruce, pine and r mixed sawdust.
c
PS = 50 wt% pine bark 50 wt% spruce whitewood mix.
column. The gas and tar sampling device is shown
in Fig. 2. Tar yield was measured by in-line tar sam-
pling using a sampling train simplied from the Tar
Protocol [14], together with post-test direct tar collec-
tion from the stack. The tar sampling train employed
four 250 ml impingers, with acetone as the solvent,
working alternately at −3
◦
C and room temperature
(about 35
◦
C in the sampling area) in order to reduce
tar fog by forming larger droplets at room temperature
that are easier to capture in the next impinger.
Six sawdust species were tested; their ultimate anal-
yses and other relevant properties appear in Table 1.
Each sawdust was dried before being charged to the
hoppers. Bed ash collected from a previous run was
used as the starting bed material for each new run,
with silica sand making up for loss of solids. In some
runs, y ash collected from the outlet product stream
was pneumatically re-injected into the bottom of the
riser. The air used for re-injecting y ash was included
when calculating the air ratio. The carbon content of
the bed materials and re-injected y ash was accounted
for in the overall mass and energy balance. In the
last two runs, a nickel-based catalyst (C11-9 LDP,
S

ud-Chemie) was used for tar removal and methane
reforming. Its particle density is 2820 kg=m
3
. In each
of these two runs, a batch of about 11–14 kg of cat-
alyst, crushed and screened to 0.25 –1:7 mm in diam-
eter, was added to the riser by pneumatic conveying
immediately prior to switching the system to the gasi-
cation mode.
Fifteen test runs were conducted on the CFB gasi-
er, each with particular objectives, contributing to a
detailed parametric study of the eects of operating
temperature, air ratio, suspension density, steam injec-
tion, y ash re-injection, secondary air rate and cata-
lyst addition. The operating pressure in the system was
maintained at ∼ 1:05 bar, slightly higher than atmo-
spheric, except for the rst run, in which the pressure
was 1:65 bar. The inuences of the sawdust species
and moisture content were examined by comparing
results from dierent fuels.
A number of operating parameters can be used to
characterize gasication processes. The air ratio, a,
dened as the ratio of the actual air supply to the
stoichiometric air required for complete combustion,
is one such measure. We also use the O/C molar ratio
where steam or ash re-injection is involved. Since
sawdust has a high oxygen content, the minimum O/C
ratio is about 0.6, corresponding to air-free pyrolysis
conditions. The gas heating value is usually given as
the HHV of the dry product gas in MJ=Nm
3
. The tar
yield is expressed as the mass of tar per unit volume
of raw gas, in g=Nm
3
. An alternative denition [15]is
the mass of tar produced per unit mass of dry biomass
feed.
2.2. Temperature and gas composition proles
Fig. 3 shows radial and axial temperature distribu-
tions in the riser. The radial temperature prole for

176 X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193
400
500
600
700
800
900
1000
01234567
Height above primary air inlet, (m)
Temperature (
o
C)
time 1
sawdust feed port level
time 2
400
500
600
700
800
900
1000
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Radial position, r/R,(-)
Temperature, (
o
C)
Run 11
Run 12
(a)
(b)
Fig. 3. Measured temperature proles in the CFB gasier. (a)
Radial proles: —air ratio a=0:325, T
3
=789
◦
C;
•
—a=0:23,
T
3
= 701
◦
C; (b) axial proles, measured from Run 11 at r=R =1,
a =0:325, T
3
= 789
◦
C. For experimental operating conditions, see
Table 2.
Run 11 shown in Fig. 3(a) indicates that there could
be as much as a 45
◦
C dierence between the core
and wall region of the riser. Later measurements from
Run 12 from the opposite side, with the thermocouple
tip withdrawn 2 mm from the wall, showed improved
symmetry and less than a 15
◦
C centre-to-wall tem-
perature dierence. This temperature uniformity indi-
cates extensive radial mixing and radial heat transfer
in the riser, facilitating both homogeneous and hetero-
geneous reactions.
The axial prole of suspension temperature at con-
stant r=R = 1 is shown in Fig. 3(b). The tempera-
ture dierence across most of the riser height was less
than 100
◦
C, consistent with normal CFB reactors. The
measured temperature at the bottom of the riser was
600–700
◦
C for all test runs. The coarser particles set-
tled at the bottom and cooled there. However, intense
solids recycle minimized the temperature gradient.
In a circulating uidized bed operating in the fast
uidization ow regime, particles tend to migrate out-
wards toward the wall, driven by uid–particle inter-
actions and boundary eects, and descend along the
wall, while dilute upow is maintained in the inner
core [16]. As a result of the higher concentration of
particles in the wall region, there is a reducing region
there, with augmented CH
4
,H
2
and CO concentra-
tions as shown in Fig. 4(a).
The axial gas composition prole is plotted in
Fig. 4(b). The lower part of the riser mainly provided
pyrolysis of returning particles and evaporation of
moisture from fresh particles. For a =0:38, a major
rise in CO
2
content was observed over the 0.9 –2:0m
height interval where the partial oxidation of pyroly-
sis products resulted in a simultaneous decrease in the
concentrations of CO and other combustible species.
Gasication of char continued along the remainder
of the riser, raising the CO and H
2
contents again.
Although a cross-over of CO and CO
2
contents oc-
curred for a=0:38, this crossover was not repeated for
higher air ratio (a =0:46). The concentration of CH
4
never approached its equilibrium level due to the lim-
ited gas residence time in the riser. The measured N
2
content (not shown) decreased monotonically along
the riser height, indicating increasing conversion of
carbonaceous species.
2.3. Eects of air ratio and O/C molar ratio
Fig. 5 portrays the changes in the concentrations of
dierent species vs. air ratio, with the short straight
lines suggesting general trends rather than true lin-
ear relationships. The concentrations of CO
2
and H
2
O
would be expected to increase with increasing air ra-
tio, while the concentrations of reducing species, such
as CO, H
2
and CH
4
, decrease. Notwithstanding the
improved carbon conversion at higher air ratios, the
total fraction of combustible species decreases with
increasing air ratio, since the increase in the inert ni-
trogen far exceeds the gains in wood-borne species.

X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193 177
Fig. 4. Radial and axial gas composition proles: (a) radial proles.
Run 7, T
3
= 815
◦
C, a =0:45. Gas samples taken 5:09 m above
the primary air inlet; (b) axial proles. Solid lines and closed
points: Run 3, a =0:38, T
3
= 750
◦
C, Moisture content of sawdust
M =10:5%; dashed lines and open points: Run 15, a =0:46,
T
3
= 805
◦
C, M =4:2%. Gas samples taken at r=R = 1. See Table 2
for operating conditions.
The dry gas higher heating value at the standard
state of 101:3 kPa and 273 K can be estimated from
the gas composition by
HHV = (12:75[H
2
]+12:63[CO] + 39:82[CH
4
]
+63:43[C
2
H
4
]+···)=100; (1)
where the species contents are given in mol%, and
their heats of combustion, in MJ=Nm
3
. This equation
is based on heat of combustion data [17], assuming
0
5
10
15
20
25
0.1 0.2 0.3 0.4 0.5 0.6 0.7
Air ratio, (-)
Species molar contents, (%)
CO
CO
2
H
2
CH
4
CO
H
2
CO
2
CH
4
(a)
40
50
60
70
80
90
0.1 0.2 0.3 0.4 0.5 0.6 0.7
Air ratio, (-)
Species molar contents, (%)
N
2
N
2
(b)
Fig. 5. Eect of air ratio on instantaneous values of gas compo-
sition: fuel moisture M =6:6–22.0%. Solid lines and symbols for
riser temperatures T
3
= 700 ± 10
◦
C, dashed lines and open sym-
bols for T
3
= 820 ± 10
◦
C. Symbols: += × =CH
4
, =N =H
2
,
=
•
—CO, =—CO
2
, ♦=—N
2
.
ideal-gas behaviour for the gaseous species. Although
ethylene is listed in Eq. (1), the concentrations of ethy-
lene and higher hydrocarbons are often too low to be
detected. Since gas heating value can be signicantly
altered by the presence of even small percentages of
these hydrocarbons, accurate measurement of them is
crucial when the gasier operates at temperatures be-
low 700
◦
C or at elevated pressure.
Fig. 6 shows how the time-mean dry gas heat-
ing value varies with air ratio over the entire range
tested. It is seen that the time-mean gas heating value

178 X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193
HHV = 9.78exp(-2.86
a
)
R
2
= 0.91
0
1
2
3
4
5
6
7
0.1 0.2 0.3 0.4 0.5 0.6
Mean air ratio, (-)
Mean dry gas HHV, (MJ/Nm
3
)
Fig. 6. Eect of air ratio and feed rate on mean dry gas heating
value: T = 700–850
◦
C, M =6:6–15.0%. Data from test runs using
six sawdust species; feed rates: —16–27 kg=h; N—31–35 kg=h;
—40–49 kg=h.
can be approximated as an exponential function of the
air ratio
HHV = 9:78 exp(−2:86a)(0:22 6 a 6 0:54): (2)
The correlation coecient for this relationship is R
2
=
0:91. The standard error (SE) of the gas heating value
is shown by error bars, suggesting 7–8% of the mean
value. More scatter was observed in the instantaneous
gas heating value data. For a sample set of instanta-
neous gas HHV data representing a typical air ratio of
0.33, the standard deviation range corresponding to a
95% condence interval (CI) is 3:87 ± 0:59 MJ=Nm
3
.
For comparison, dierent ranges of feedrate are de-
noted by dierent symbols in Fig. 6. It appears that
feed rate has no signicant inuence on the trend for
the feedrate range tested. However, the methane con-
tent in the product gas was high (¿ 4%) when the
feedrate exceeded 40 kg=h, suggesting that the system
was reaching its throughput limit as the gas residence
time became inadequate for cracking of hydrocarbons.
The use of the O/C molar ratio is clearly preferred
to the air ratio for steam- or CO
2
-blown processes. In
view of this, an alternative correlation for the mea-
sured dry gas heating value was obtained:
HHV = 34:38 exp(−1:37[O=C])
(1:1 6 O=C 6 2:1): (3)
The correlation coecient of this equation is R
2
=
0:86. The gas heating value could equally be correlated
versus the O=[C+H]orO=[C+H=4] molar ratio,
when the molar abundance of hydrogen in the system
is comparable to that of carbon.
Three molar ratios are commonly used to char-
acterize the gas composition: CO=CO
2
,H
2
=CO and
CH
4
=H
2
. The O/C molar ratio varied between 1.1 and
2.1 for most cases tested. As more oxygen is supplied,
more CO
2
is formed, causing the CO=CO
2
ratio to de-
crease. Since the H
2
concentration in the raw gas is
mainly determined by the water–gas shift reaction, it
is less sensitive to the O/C ratio than is the CH
4
con-
centration. The H
2
=CO molar ratio increases slightly
with increasing O/C, while the CH
4
/H
2
molar ratio
decreases more sharply. It was found that air-blown
gasication of biomass usually resulted in a H
2
=CO
molar ratio less than 1, as in a previous study [18]ina
bubbling uidized bed gasier. Injection of steam as
gasifying agent increases the H
2
/CO molar ratio be-
cause moisture promotes both steam gasication and
the CO-shift reaction.
Modelling work [19] indicates that the equi-
librium methane concentration in the fuel gas is
less than 0.1% by volume for the temperature and
pressure ranges tested. This suggests that the high
CH
4
/H
2
ratio of the product gas (0.6–0.8) from
the pilot plant tests was not due to methanation.
Instead, it resulted from incomplete thermal crack-
ing of pyrolysis products and incomplete reforming
reactions.
2.4. Eect of operating temperature
Operating temperature plays an important role in
biomass gasication. Fig. 7 shows that the gas heating
value increases slightly with increasing temperature
for constant values of the air ratio. This is because of
improved carbon conversion at higher temperatures.
As shown in Fig. 7, the increase in gas HHV is approx-
imately 10% for an increase in operating temperature
from 700
◦
Cto800
◦
C. This increase of gas heating
value with increasing temperature indicates that the
gasier can benet from better thermal insulation and
from air preheating. System pressure showed little ef-
fect on gas composition for the limited range tested
(1.0–1:65 bar).

X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193 179
0
1
2
3
4
5
6
7
650 700 750 800 850
Tem
p
erature,
(
o
C
)
Dry gas HHV, (MJ/Nm
3
)
a = 0.22
a = 0.33
a = 0.37
a = 0.47
Fig. 7. Eect of operating temperature on dry gas heating value.
M =6:6–15.0%. Air ratios for each group of data points are given.
2.5. Eect of secondary air
Secondary air was found to have only a small ef-
fect on the gas composition. A previous study [20]
showed that secondary air contributed to tar removal,
but at the expense of lowering the gas heating value,
with a 14% decrease in the gas heating value as the
secondary air fraction increased from zero to 20%.
The mechanism proposed for tar removal due to ad-
dition of secondary air [20] is the formation of local
high-temperature zones where thermal cracking of tar
is promoted.
Experimental results obtained in this work indicate
that the gas heating value dropped from an average of
4:20 MJ=Nm
3
for no secondary air to 4:02 MJ=Nm
3
for 14.3% secondary air at constant total air ow,
less than a 5% decrease. This decrease was less pro-
nounced than reported previously [20]. However, in
the earlier work, there appeared to be an increase in
total air supply as the secondary air level increased.
Secondary air caused only a slight change in gas com-
position for the range of conditions investigated in
this work.
2.6. Eect of suspension density
In most CFB systems the solids gravity term is an
order of magnitude greater than terms due to wall fric-
tion, acceleration and gas gravity. The pressure drop
0
1
2
3
4
5
6
30 40 50 60 70 80 90 100
Overall suspensiondensity, (kg/m
3
)
Dry gas HHV, (MJ/Nm
3
)
Fig. 8. Eect of suspension density on gas heating value:
•
—Hemlock sawdust, a =0:337;T= 715–780
◦
C, M =14:7%;
—Pine and spruce mix, a =0:218, T = 676–735
◦
C, M =10:1%;
N—Mixed sawdust, a =0:258, T = 710–770
◦
C, M =6:6%.
can then be approximated closely by
P = 
p
g(1 − )h: (4)
Hence, the suspension density can be estimated as

susp
= 
p
(1 − )=P=gh: (5)
In the pilot tests, the suspension density was adjusted
by draining solids from the system with the air ratio
maintained constant. Suspension densities were mea-
sured below and above the secondary air injection
level. The overall suspension density in the riser was
taken as the weighted average of the two on a height
basis.
Fig. 8 shows the eect of the overall suspension
density on the gas heating value. The suspension
density at the bottom of riser was between 100 and
140 kg=m
3
. The gas heating value increased from
3.5 to 4:7MJ=Nm
3
as the overall suspension den-
sity increased from 42 to 93 kg=m
3
. The solid line
shows the best linear t for all data points, with a
correlation coecient R
2
=0:72. The positive in-
uence of suspension density on gas quality is be-
lieved to be due to an increase in the concentration
of solid reactants, together with enhanced solids
mixing.

180 X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193
0
0.5
1
1.5
2
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
F, (kg/kg)
HHV
far
/HHV
0
HHV
tar
/HHV
0
=1+0.6[1-exp(-F/0.7)]
R
2
= 0.74
Fig. 9. Eect of y ash re-injection on normalized gas heating
value: —SPF/cypress mix, a=0:35, T =700–740
◦
C, M =11:3%;
N—SPF/cypress sawdust, a =0:41, T = 720–760
◦
C, M =15:0%;
•
—Cedar/hemlock mix, a =0:40, T = 800–830
◦
C, M =12:6%.
2.7. Eect of y ash re-injection
Since y ash re-injection can increase suspension
density as well as the carbon concentration in the riser,
it should have a similar eect to raising the suspen-
sion density on gas quality and carbon conversion. To
facilitate discussion, we rst dene
F =
˙m
fa
C
fa
˙m
f
C
f
(6)
as the ratio of the carbon in re-injected y ash to the
carbon introduced with the fuel, where ˙m denotes dry
feed rate and C is the fractional carbon content. Sub-
scripts fa and f refer to y ash and fuel, respectively.
The carbon content in the ash retained in the riser was
less than 2% for all runs, while it varied from 12% to
63% in the y ash. This makes y ash a major source
of carbon loss if it is not recycled.
The gas heating value is plotted as a function of
the carbon recirculation ratio, F, in Fig. 9. A simple,
empirical correlation for the species and parameter
range tested is
HHV
far
=HHV
0
=1+0:6[1 − exp(−F=0:7)]: (7)
Here the subscripts far and 0 stand for cases with and
without y ash re-injection, respectively. The correla-
tion coecient for this relationship is R
2
=0:74. The
eect of re-injection diminishes as F approaches zero.
0
1
2
3
4
5
6
7
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6
O/C molar ratio,
(
-
)
Dry gasHHV,(MJ/Nm
3
)
HHV = 34.38exp(-1.37[O/C])
Fig. 10. Eect of steam injection rate on instantaneous dry gas
heating values for hemlock sawdust. T = 750–800
◦
C, a =0:38–
0.43, M =8:8–9.2%. Dashed line: best t for no steam injection;
solid line and data points: with steam injection.
The benet of carbon re-injection reaches a limit due
to kinetic limitations at a given temperature and solids
residence time. It is expected that y ash re-injection
would be more benecial at higher temperatures. Fly
ash re-injection had little eect on the product H
2
=CO
and CH
4
=H
2
ratios. However, measured gas composi-
tions indicate that for a given total oxygen/total car-
bon ratio, the CO=CO
2
ratio increased with increasing
re-injection.
2.8. Eect of steam injection and fuel-bound
moisture
Fig. 10 shows that steam injection can signicantly
improve gas quality at a given O/C molar ratio.
The steam injection rates tested were in the range
0–10:5kg=h, resulting in an increase of 0– 0.67 in the
O/C molar ratio from the baseline cases with no steam
injection. When steam is introduced, CO and H
2
are
formed as products of the endothermic steam-char re-
action. As a result, steam injection makes the dry gas
heating value higher than for purely air-blown pro-
cesses at the same O/C ratio. This eect can also be
seen from the product molar ratios, reecting progress
of the methane steam reforming and CO-shift re-
actions. However, in a gasication system without
an external heat source, steam injection lowers the

X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193 181
operating temperature, and this could lead to a higher
tar yield.
Steam, although injected at a higher level, caused
greater changes in the product ratios, and showed
much better reactivity, than an equal feed rate of fuel
moisture. To produce hydrogen-rich gas from small
units, it is desirable, where feasible, to employ steam
injection. Alternatively, feeding sawdust in the recy-
cle leg is likely to improve the ability of the mois-
ture in the feed material to participate in the chemical
reactions.
2.9. Eect of sawdust species and particle size
Notwithstanding dierences in wood type, dier-
ent sawdust species show greater uniformity in chem-
ical composition (Table 1) than coal and other solid
fossil fuels. For the six biomass species tested in the
present study, species eects on gas heating value
and carbon conversion were relatively minor from a
chemical point of view. Nevertheless, the various saw-
dust species behaved dierently during the gasica-
tion tests due to dierences in physical properties, e.g.
bre length, moisture content, shape and particle size,
caused by dierent methods of processing. For exam-
ple, cedar hog fuel, because of its long bre length,
tended to cause bridging in the feed hoppers. Blending
the ground cedar hog fuel with a more granular saw-
dust helped alleviate this tendency. The hemlock and
cypress sawdusts proved to be most suitable for feed-
ing because of their more favourable shape and low
bridging tendency, even at relatively elevated moisture
contents. Over the limited range tested in this work,
particle size eects on gas heating value and carbon
conversion were negligible.
2.10. Tar yield
The experimental data indicate that the tar concen-
tration primarily depends on the operating tempera-
ture. Ultimate analysis of tar samples shows that typ-
ical tars contain about 78% carbon, 6% hydrogen,
0.7% nitrogen, 12% oxygen, less than 0.5% sulphur,
and the rest being solids. The measured tar yield,
shown in Fig. 11, decreased sharply from 15:2g=Nm
3
at 700
◦
Cto0:4g=Nm
3
at 815
◦
C. This arises because
of increased tar cracking rates at higher temperatures.
The results are approximately linear on semi-log pa-
0.1
1
10
100
650 700 750 800 85
0
Operating temperature, (
o
C)
Tar yield (g/Nm
3
)
With catalyst
Fig. 11. Temperature dependence of tar yield and eect
of nickel-based catalyst: a =0:21–0.46, T = 700–815
◦
C,
M =4:18–14.7%.
•
—no catalyst; —with catalyst.
per suggesting an exponential decay function. Despite
measures to reduce sampling losses, the uncertainty in
tar measurement could be as large as 15% in practice.
Fortunately, the experimental trends were not aected
by such uncertainties. The results from this work are
in qualitative agreement with those of previous stud-
ies [21,22]. Another set of data [18] although tting
our trend line well at higher temperatures, shows con-
siderable scatter.
In addition to raising the operating temperature, it
has been reported [22,23] that further tar reduction
can be achieved by using commercial or mineral cat-
alysts. In Run 14, 62% tar removal was achieved at
a reactor temperature of 735
◦
C by adding 11–14 kg
nickel-on-alumina commercial steam-reforming cata-
lyst. The tar yield was reduced to 0:15 mg=Nm
3
when
the system operated at 800
◦
C with catalyst addition.
The catalyst activity increased as the operating tem-
perature rose, with the result that the H
2
content in-
creased with increasing temperature.
2.11. Mass and energy balances and gasication
eciency
Post-test mass and energy balances were performed
to determine the carbon conversion and thermal e-
ciencies. A nitrogen balance was chosen as the pri-
mary basis for the mass balance because nitrogen is

182 X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193
the most abundant element in an air-blown gasica-
tion process, while it is relatively easy to determine
accurately by experimental means, and it is largely in-
dependent of other elements. A secondary basis for
the mass balance calculations is the oxygen balance,
which helps diminish errors when the moisture content
in the product gas is unknown or cannot be measured
accurately. Alternatively, a hydrogen balance can be
used as a secondary basis [18].
The carbon conversion is dened as the fraction
of carbon in the feed converted to gaseous products.
A modied carbon conversion can be dened which
also accounts for the contribution of tars. The results
of mass balances and gasication eciency calcula-
tions appear in Table 2. For all fteen runs in the pi-
lot study, the overall mass balance gives 95 –101%
closure, with the corresponding ash balance giving
63–110% closure. Since the carbon balance is based
on other elemental balances, it is dicult to achieve
perfect closure.
The carbon conversion is determined from the prod-
uct gas composition and gas yield. The air (or O/C)
ratio was the primary factor inuencing carbon con-
version. A simple correlation for the experimental
carbon conversion vs. air ratio is
C =0:25+0:75[1 − exp(−a=0:23)]
(0:22 6 a 6 0:54): (8)
The correlation coecient (R
2
) of Eq. (8) is 0.86.
Temperature and residence time are not included in
this equation because of their relatively weak inu-
ences and the limited number of data points. Com-
parison of our test results with those of a previous
study [16] shows substantial agreement in the trend,
despite a dierence of ∼ 5% in the absolute carbon
conversion. The latter dierence likely arises from
dierences in reactor conguration, cyclone e-
ciency, fuel moisture content in the fuel and operating
temperature.
A cold-gas eciency is used to evaluate the gasi-
cation performance. The cold-gas eciency, E
1
, ex-
cluding the heating value of the condensables (tars),
is dened as the percentage of the fuel heating value
converted into the heating value of the product gas, i.e.
E
1
=
[HHV]
g
× v
g
[GCV]
f
× 100%; (9)
where [HHV]
g
(in MJ=Nm
3
) is the higher heating
value of the product gas, while [GCV]
f
(in MJ/kg)
denotes the gross caloric values of the fuel; v
g
is the
specic dry gas volume, in Nm
3
=kg fuel.
A modied cold-gas eciency, now including the
heating value of any tars, is
E
2
=
[HHV]
g
× v
g
+ [HHV]
t
× y
t
[GCV]
f
× 100%: (10)
Here y
t
is the specic tar yield in kg/kg-fuel,
and [HHV]
t
is the heating value of tar, taken as
30:1MJ=kg-tar, estimated from the tar analysis data.
E
1
and E
2
have both been used extensively in the
gasication literature [e.g. [24,25]].
Calculated gasication eciencies are given in the
last two rows of Table 2. In the present study, since
the enthalpy of the product gas and hot water pro-
duced by the heat exchangers is not utilized in any
downstream equipment, nor is the reactor externally
heated, the cold-gas eciency is adequate to assess the
performance of the process. However, external heat
may be required to maintain the operating tempera-
ture. The cold gas eciency, which does not account
for the external heat supply, then becomes insucient
to measure gasier performance.
The gasication eciencies E
1
and E
2
calculated
from the mass and energy balance are plotted in Fig. 12
0
20
40
60
80
100
1.2 1.4 1.6 1.8 2 2.
2
O/C molar ratio, (-)
E
1
and E
2
, (%)
Fig. 12. Variation of gasication eciency with O/C ratio. a=0:21–
0.54, T = 700–815
◦
C, M =4:2–22.0%. —E
1
, only product gas
considered;
•
—E
2
, tar also taken into account.

X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193 183
Table 2
Summary of mass balance and gasication eciency calculations
Run number 123456789101112131415
Sawdust species cyp. cyp. SPF hem. hem. hem. hem. SPF/C hem. c/h hem. PS mixed mixed mixed
Total gasication run time h 3.80 4.53 3.00 4.23 4.10 4.27 2.62 3.57 3.62 3.58 3.80 2.42 3.07 3.98 3.55
Sawdust consumption kg 91.0 103.8 80.7 110.6 118.6 108.2 88.1 126.3 112.3 95.4 120.8 117.3 140.0 164.8 55.2
Moisture content in
sawdust % 22.0 9.7 10.5 10.0 8.8 9.2 11.7 11.3 15.0 12.6 14.7 10.1 6.6 6.7 4.2
Throughput, dry basis kg=sm
2
0.87 0.97 1.13 1.10 1.23 1.08 1.39 1.47 1.23 1.09 1.27 2.04 1.99 1.81 0.70
Total air supplied Nm
3
310 255 177 281 226 225 142 207 207 179 186 125 188 251 135
Fly ash re-injection kg 000000016.8 20.0 5.2 00000
Total steam injection kg 00002.86 24.17 000000000
Mean suspension temp.
(T
3
) C 740 718 766 815 772 787 718 730 752 815 789 701 728 739 805
Primary air pressure bar 1.65 1.19 1.19 1.19 1.19 1.19 1.19 1.19 1.19 1.19 1.19 1.19 1.19 1.19 1.19
Time-mean air ratio – 0.536 0.445 0.402 0.522 0.376 0.427 0.340 0.350 0.411 0.399 0.337 0.218 0.258 0.294 0.460
Mean gas composition (dry basis)
H
2
% 5.6 3.1 3.2 3.0 4.0 3.8 5.5 3.9 3.5 4.1 4.2 5.4 5.1 7.3 5.9
N
2
% 68.0 68.1 67.1 68.4 61.8 65.2 59.5 62.5 64.8 64.6 62.6 53.9 56.3 55.4 64.6
CO % 6.9 11.0 10.7 9.6 14.7 12.6 16.6 15.1 13.4 12.3 14.6 21.4 19.9 17.9 10.0
CH
4
% 1.4 1.9 1.9 1.9 2.9 2.7 3.4 2.8 2.8 2.5 3.0 4.6 4.1 3.2 1.2
CO
2
% 18.1 15.9 17.1 17.1 16.5 15.7 15.0 15.6 15.6 16.5 15.7 14.7 14.5 16.3 18.3
Mean dry gas heating
value MJ=Nm
3
2.43 2.96 2.92 2.75 4.14 3.73 4.82 4.13 3.85 3.59 4.17 6.13 5.62 4.60 2.54
Gas yield Nm
3
=kg 3.30 2.92 2.48 3.10 2.59 2.75 2.34 2.27 2.46 2.51 2.13 1.72 2.06 2.35 3.24
Tar yield g=Nm
3
n. a. n. a. n. a. n. a. n. a. n. a. n. a. n. a. n. a. 0.41 1.39 15.13 10.26 2.35 0.04
Overall mass balance
closure % 98.3 99.6 97.7 98.4 97.8 98.7 98.0 96.9 98.7 97.5 98.7 100.5 100.7 100.1 95.4
Carbon conversion, gas % 97.7 102.0 92.9 99.2 98.6 93.0 91.9 85.8 91.1 87.6 81.9 81.6 89.8 94.9 98.3
Carbon conversion,
gas + tar % n. a. n. a. n. a. n. a. n. a. n. a. n. a. n. a. n. a. 87.7 82.4 85.7 93.1 95.8 98.3
Cold-gas eciency, E
1
% 53.2 51.8 44.8 52.2 64.3 60.3 71.4 51.0 50.8 53.6 58.7 63.3 64.7 60.5 44.2
Overall thermal eciency,
E
2
% n. a. n. a. n. a. n. a. n. a. n. a. n. a. n. a. n. a. 53.8 59.3 67.9 68.2 61.5 44.2
Notes: cyp: = cypress; SPF = spruce-pine-r mixture; hem: = hemlock; SPF=C = SPF=cypress mixture; c=h = mixed cedar-hemlock; PS = mixed pine bark-spruce.

184 X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193
against the O/C molar ratio. Results suggest that O/C
should be in the range 1.3–1.6 to maximize gasica-
tion eciency for the species and parameters tested.
Compared to carbon conversion, the gasication e-
ciencies give a better measure of the performance of
the gasier.
3. Equilibrium modelling
At chemical equilibrium a reacting system is at
its most stable composition, a condition achieved
when the entropy of the system is maximized, while
its Gibbs free energy is minimized. Two approaches
have been developed for equilibrium modelling: stoi-
chiometric and non-stoichiometric [26]. The stoichio-
metric approach requires a clearly dened reaction
mechanism incorporating all chemical reactions and
species involved. In a non-stoichiometric formulation,
on the other hand, no particular reaction mechanism
or species are involved in the numerical solution. The
only input needed to specify the feed is its elemen-
tal composition, which can be readily obtained from
ultimate analysis data. This method is particularly
suitable for problems with unclear reaction mecha-
nisms and feed streams like biomass whose precise
chemical compositions are unknown.
The pilot-scale gasier described above operated
under near-steady-state conditions and was subject to
axial and radial gradients of temperature and compo-
sition. In the equilibrium model, however, the reac-
tor is implicitly considered to be zero-dimensional,
i.e. no spatial distribution of parameters is considered,
nor are there changes with time because all forward
and reverse reactions have reached chemical equilib-
rium. The molar inow for any individual element
involved in the chemical reactions can then be writ-
ten as the sum of moles of that element in the vari-
ous feed streams. Tars are not included in the product
stream because of their low yield for the conditions
considered.
To simplify the problem, only 42 gaseous and 2
solid species involving C, H, O, N and S are con-
sidered in the present work. Other elements (e.g. Si
entering as SiO
2
, and mineral matter in the biomass)
are considered to be inert or independent of the re-
action system. Although carbon, oxygen and sulphur
may be present in mineral matter (e.g. as carbonates
and sulphates), and may be converted during gasica-
tion, inorganic C, O and S are minor contributors to
the elemental abundance of the system because of the
low ash content of the biomass concerned. Little error
therefore results when they are ignored in equilibrium
modelling.
Although our major concern in the present work
is the phenomenological model, which is adapted
from the pure equilibrium model, both models share
the same mathematical basis. The pure equilibrium
model is rst introduced. Followed by the justication
and methodology of incorporating non-equilibrium
factors.
3.1. Formulation of pure equilibrium model
The RAND algorithm which has been well-
documented in previous literature [26–28] is used
in the present work. Validation of the method was
described in our previous work [29], in which it was
used successfully to predict the performance of a coal
gasier. In the RAND algorithm, the change in moles
of a species in the mth iteration can be expressed
explicitly as a function of its current chemical poten-
tial, the phase distribution of the species at a given
system temperature and pressure, and the Lagrange
multiplier:
n
(m)
i
= n
(m)
i

K

k=1
a
ik
k
+ u
#
−
$
(m)
i
RT

for multi-species phases;
= u
#
n
(m)
i
for single-species phases
(i =1; 2;:::;N; k =1; 2;:::;K;
# =1; 2;:::;): (11)
These n
(m)
i
constitute a vector n
(m)
, which is the
change of number of moles for all species upon the
current iteration. N designates the total number of
species. $
(m)
i
and n
(m)
i
denote the chemical potential
and moles of species i in the mth iteration, respec-
tively. a
ik
is the coecient in the species-element ma-
trix;
k
is a function related to the Lagrange multiplier,
%
k
, i.e.
k
=
%
k
RT
(12)

X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193 185
u
#
is the phase split of n
i
, dened as
u
#
=
N

i=1
n
(m)
i#
=n
(m)
t
= n
(m)
t#
=n
(m)
t
; (13)
where subscript t means total and # refers to the phase
to which a species belongs.
The set of (K +) simultaneous algebraic equations
that are to be solved iteratively by the RAND algo-
rithm includes K linear equations regarding element
abundance:
K

k=1
N

i=1
a
ik
a
ij
n
(m)
i
k
+


#=1
b
(m)
j#
u
#
=
N

i=1
a
ij
n
(m)
k
$
(m)
i
RT
+ b
k
− b
(m)
k
(j =1; 2;:::;K): (14)
The  supplementary equations for dierent phases
are:
K

k=1
b
(m)
k#
k
− n
z#
u
#
=
N

i=1
n
(m)
i#
$
(m)
i#
RT
(# =1; 2;:::;): (15)
The initial element abundance vector b is calculated
from the feedstock. The kth element of the b-vector
at the mth iteration is
b
(m)
k
=
N

i=1
a
ik
n
(m)
i
: (16)
Mass balance constraints are imposed at every iter-
ation during solution of Eqs. (11)–(16), while the al-
gorithm iteratively minimizes the Gibbs free energy.
The dierence between the initial elemental abun-
dance vector and its current iteration value (b
k
−b
(m)
k
),
is added to the right-hand side of Eq. (14) to eliminate
error accumulation during the iteration process [26].
Finally, the new numbers of moles vector, n
(m+1)
,
is determined by
n
(m+1)
= n
(m)
+ !
(m)
n
(m)
; (17)
where !
(m)
is the step size parameter, 0 ¡!
(m)
¡ 1,
chosen such that all new moles generated from the
current iteration remain positive. The new molar frac-
tions of all species are then determined by
x
i
= n
i
=n
t
: (18)
Other quantities, such as elemental distributions, car-
bon conversion and water conversion, are all derived
from the variables in Eq. (18). The convergence crite-
rion is that the maximum absolute value of the changes
in the molar fractions for all species is less than a
pre-set upper limit, typically 1 × 10
−6
molar, to en-
sure good accuracy as well as a true global minimum
of the Gibbs free energy.
The energy balance of the process can be written
L

l=1
˙m
l
H
0
f ; 298; feed
+
L

l=1
˙m
l
H
feed
(T
feed;l
)
=
N

i=1
n
i
H
0
f ; 298; prod
+
N

i=1
n
i
H
prod
(T )
+H (T )(l =1; 2;:::;L) (19)
for any temperature T . The two terms on the left-hand
side are the total heat of formation and the total en-
thalpy of all feed streams, respectively. The rst two
terms on the right-hand side represent the total heat
of formation and total enthalpy of all product species,
respectively. The nal term denotes the system net en-
ergy excess production as a function of temperature.
This term can also be extended to account for any
heat removed from the reactor or provided by an ex-
ternal heat source. If H (T ) is positive, the system
adjusts its equilibrium temperature to a higher level.
The equilibrium temperature of the system is the tem-
perature at which H(T ) = 0. Eq. (19) is based on
1 kg of fuel (dry basis). Note that although m
l
and
n
i
have dierent units, all the terms on both sides of
the equation are converted into MJ by multiplying by
enthalpies expressed in MJ=Nm
3
, MJ/kg or kJ/mol as
needed.
The ultimate analysis of a typical sawdust used in
the model calculations, listed in the last column of
Table 1, is obtained by averaging the analyses of six
sawdust species. The heats of formation of the oxidant
and steam can be determined from standard thermo-
dynamic data [30]. The heat of formation of fuel, in
kJ/kg fuel as received, is calculated from the equation
[29]
H
0
f ; 298; fuel
= HHV − (327:63C
ar
+ 1417:94H
ar
+92:57S
ar
+ 158:67M
ar
); (20)

186 X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193
0
10
20
30
40
50
60
70
80
00.20.40.60.81
Air ratio, (-)
Species contents, (%)
N
2
H
2
CO
CO
2
H
2
O
C(s)
CH
4
Fig. 13. Variation of equilibrium gas composition with air ratio
for representative sawdust composition at a pressure of 1:013 bar.
Solid lines—1000 K, dashed lines—1100 K. No steam added.
where HHV denotes the higher heating value of the
fuel (kJ/kg, as-received basis). C
ar
, H
ar
, S
ar
and M
ar
represent the carbon, hydrogen, sulphur and moisture
contents of the fuel (wt%, as-received basis), respec-
tively. This equation assumes that CO
2
,H
2
O and SO
2
are the only combustion products.
3.2. Predicted species concentrations
Fig. 13 shows predictions of the dependence of the
molar contents of 7 major species on air ratio when the
representative sawdust is gasied at 1:013 bar. Five
species (H
2
,N
2
, CO, CO
2
and CH
4
) are given by
their molar fractions in the dry product gas. H
2
O(g)
is shown by its concentration in the wet gas, while
unconverted carbon, C(s), is represented by its molar
fraction in the overall equilibrium system. The numer-
ical predictions in the present work covered the tem-
perature range 600 –1600 K. However, to minimize
clutter, predictions are presented here only for 1000
and 1100 K.
As expected, major oxidizing species (CO
2
,H
2
O)
increase with increasing air ratio, while reducing
species (H
2
,CH
4
, CO) decrease. The model pre-
dicts that the equilibrium content of methane should
not exceed 3% at temperatures above 1000 K, even
with no air supplied. For typical gasication condi-
tions (a =0:3), the equilibrium CH
4
concentration
is only 0.02% for a temperature of 1100 K. Mea-
sured methane contents signicantly higher than this
level arise from incomplete cracking of the pyrolysis
products. The CO
2
and H
2
O concentrations decrease
slightly with increasing air ratio below 1000 K for
a¡0:2. This is due to the rapid disappearance of un-
converted carbon C(s) with increasing air ratio, which
causes rapid growth in the total moles in the gas phase.
At 1100 K, the initial fraction of unconverted carbon
is smaller than at 1000 K, and the decreasing trend
for CO and H
2
O vs. air ratio at low air ratios nearly
disappears.
3.3. Elemental distribution in products
The fate of elements in the feedstock can be
interpreted in terms of the distribution of fractions
converted to dierent species in the spectrum of nal
products, as shown in Fig. 14. These fractions must
add up to unity. Fig. 14(a) plots the predicted carbon
distribution versus operating temperature for a=0:3at
dierent pressures. The height under the lowest curve
for each pressure represents the molar fraction of
unconverted solid carbon, C(s). The narrow band be-
tween this curve and the next higher one for the same
pressure gives the mole fraction of hydrocarbons,
mostly CH
4
. The interval to the next curve for the
same pressure signies CO, while the region above the
highest corresponding curve is occupied by CO
2
. The
equilibrium model indicates that pressure should only
aect the carbon distribution for intermediate temper-
atures from ∼ 700 to 1250 K. Beyond this range, the
system pressure again becomes a secondary factor.
At 1000 and 1100 K, atmospheric pressure, and air
ratios below 0.2, a considerable portion of the carbon
may remain as solid carbon. For 0:2 ¡a¡0:3, there
is little change in the CO content due to the gradual
disappearance of solid carbon, suggesting that this is
a desired range of operation for producing CO-rich
gas.
The hydrogen distribution is shown in Fig. 14(b).
CH
4
occupies a signicant part of the equilibrium
product spectrum at low air ratios and temperatures
less than about 1000 K. The predictions show that an
equilibrium-controlled atmospheric biomass gasica-
tion process intended to produce hydrogen-rich gas
should operate in the temperature range from 1100
to 1300 K and at an air ratio from 0.15 to 0.25 to

X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193 187
0
0.2
0.4
0.6
0.8
1
600 800 1000 1200 1400 1600
Temperature, (K)
Molar fractions,(-)
C(s)
CH
4
CO
CO
2
1 5 20 bar
1 5 20 bar
1
5
20 bar
(a)
0
0.2
0.4
0.6
0.8
1
600 800 1000 1200 1400 1600
Temperature, (K)
Molar fractions, (-)
CH
4
H
2
H
2
O
1 5 20 bar
1 5 20 bar
(
b
)
Fig. 14. Predicted eects of temperature and pressure on elemental
distributions: (a) carbon distribution for a =0:3; (b) hydrogen
distribution for a =0:3.
maximize hydrogen production. For larger air ratios,
H
2
O starts to dominate the hydrogen distribution as a
net product of hydrogen oxidation.
The oxygen distribution in the system indicates that
more than half of the oxygen supply is used to pro-
duce CO
2
and water for an air ratio larger than 0.33.
Operation below this air ratio is therefore preferable.
In addition to fuel drying, good insulation, air preheat-
ing and natural gas augmentation may help maintain
an elevated system temperature.
The limited pressure inuence on the product dis-
tributions of the three most abundant elements (C,
H, and O) implies that high-temperature gasication
CH
4
CO
CO
2
H
O
C
H
2
O
K1100
1000
H
2
/CO molar ratio, (-)
Fig. 15. Carbon formation tendency in sawdust gasication: data
from CFB pilot tests (Runs 1–15) gasifying six sawdust species.
(T¿1200 K) does not require elevated pressures
since increasing pressure only increases the energy
consumption with little gain in equilibrium product
quality. The same holds for very low-temperature
processes (T¡700 K) such as those using super-
critical water as oxidizing agent, or anaerobic pro-
cesses. However, high pressure does concentrate the
gas phase and accelerate reactions while reducing
reactor volume and the time required to achieve
equilibrium.
3.4. Carbon formation tendency in biomass
gasication
Previous work [29] by the authors discussed the
nature of carbon formation in coal gasication sys-
tems and predicted carbon formation boundaries for
dierent operating temperatures and pressures. As in
that work, a biomass gasier can be represented by a
C–H–O ternary system since nitrogen is almost in-
ert, and the molar abundance of sulphur is two or
three orders of magnitude smaller than those of the
other four major elements. In Fig. 15, the elemental
abundance combinations from our experimental study
are represented by open points. Only three points
fall in the boundary zone, with none residing in the
carbon-forming regime. Therefore, the presence of

188 X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193
unconverted solid carbon must be attributed solely
to kinetic, mass transfer and gas–solid separation
limitations.
For a given elemental abundance combination,
when more air is supplied, the system moves along
a straight line toward the pure oxygen (O) vertex.
When the moisture content of the fuel increases or
when steam is injected, the system migrates towards
the H
2
O point. When y ash is re-injected, the abun-
dance of carbon increases without causing much
change in the abundance of H and O since y ash
consists mostly of carbon. Hence, the system travels
toward the C vertex, leading to an increased tendency
to form carbon.
3.5. Deviations from chemical equilibrium
Previous work [31] has reported deviations in gas
composition from equilibrium predictions for several
types of gasiers. The best-t equilibrium tempera-
ture, T
0
eq
based on an equilibrium model, tends to de-
viate from the reported operating temperature. The
best-t temperature is dened here as the temperature
which minimizes the sum of squares of the deviations
for ve principal gas species (H
2
, CO, CO
2
,CH
4
,N
2
),
i.e.
T
0
eq
= T at min




(y
eq
− y
meas
)
2



: (21)
The larger the temperature deviation, the farther the
system is from chemical equilibrium. At full chemical
equilibrium, the deviation between the actual temper-
ature and T
0
eq
should disappear. Therefore, the dier-
ence between the best-t T
0
eq
value and the representa-
tive actual temperature (e.g. cyclone exit temperature)
can be used as a measure of the approach to equilib-
rium.
Comparing the carbon conversion data from our pi-
lot tests with equilibrium predictions, one can nd a
similar shift towards the low-temperature side. The
deviation from equilibrium conditions indicates that
a higher operating temperature is required to reach
a given solid conversion than is thermodynamically
necessary.
Previous studies [32,33] have shown that high mea-
sured concentrations of methane from coal gasication
result from incomplete conversion of pyrolysis prod-
ucts; equilibrium molar concentrations of methane in
the o-gas are less than 0.1% for the entire parameter
range tested, whereas actual methane concentrations
are of the order of a few per cent. The high measured
methane concentration in the product gas cannot be
explained on an equilibrium or thermodynamic ba-
sis. Once again, the deviation must result from one or
more non-equilibrium factor, e.g. incomplete cracking
of pyrolysis products.
4. Modied model and comparison with
experimental results
4.1. Elemental availability and modied model
The element abundance vector of the feed can be
written
b
0
=(n
C
;n
H
;n
O
;n
N
;n
S
): (22)
It is often taken for granted that the amount of each
element participating in the chemical equilibrium is
exactly the same as in the feed. This is true when the
reaction kinetics and mass transfer processes do not
impede the achievement of equilibrium. However, this
assumption is not valid for real processes in which
reactions (mostly heterogeneous) are inuenced by
kinetics and/or mass transfer so that some elements
never achieve equilibrium.
The previous section showed deviations of our real
gasier from chemical equilibrium. This suggests that
an equilibrium-based model that does not consider
such deviations is liable to substantial error in predict-
ing gas composition and overall eciency. However,
equilibrium models are useful as design tools, able to
explore process behaviour under thermodynamic con-
trol. In order to correct for the deviations in real
gasication systems, a phenomenological model was
developed in this work by modifying an equilibrium-
based framework to account for key non-equilibrium
factors.
If the experimental carbon conversion and
methane yield are available, one can correct for
non-equilibrium eects by withdrawing the corre-
sponding carbon and hydrogen from the equilibrium
system. This method was applied successfully to coal
gasication [29], and it was also successful for steam
methane reforming [34] where hydrogen was prefer-
entially removed through perm-selective membranes.
A phenomenological model is then established which

X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193 189
β
H,2
= 23.5(1-
a
)
β
C,2
= 11.0(1-
a
)
0
5
10
15
20
25
0.2 0.3 0.4 0.5 0.6
Air ratio,
a
,
(
-
)
β
C,2
and
β
H,2
, (%)
Fig. 16. Eect of air ratio on the percentages of carbon and
hydrogen that remain as methane in the product gas. Data from
Runs 1–15. a =0:21–0.54, T = 700–815
◦
C, M =4:2–22.0%.
employs empirical data from the pilot-plant study to
account for factors which prevent equilibrium from
being achieved. In the present study, this approach
is extended to account for non-equilibrium eects of
pyrolysis products such as methane and carbon, and
the modied equilibrium model is then applied to
biomass gasication.
An availability function, , is imposed on each ele-
ment, leading to a modied element abundance vector
aecting the gas, i.e.:
b
∗
=(
C
n
C
;
H
n
H
;
O
n
O
;
N
n
N
;
S
n
S
): (23)
Based on experimental results from our pilot plant
tests (Table 2), the fraction of carbon converted into
gaseous species is

C;1
=0:25+0:75 exp(−a=0:23): (24)
However, a fraction of the carbon entering the gas
phase exists as methane, produced during the pyrol-
ysis stage and leaving the system without achieving
equilibrium. Mass balance calculations for the pilot
tests in this work (Fig. 16) suggest that this fraction
can be approximated by

C;2
=0:11(1 − a): (25)
Both 
C;1
and 
C;2
are based on the molar abundance
of carbon in the feed. The portion of carbon consumed
to produce methane must be deducted from the overall
Initial elemental abundance vector
Equilibrium
mainstream
C(s), CH
4
Mixed product stream
Feed
Products
Gasifier
Fig. 17. Schematic of kinetic modication of equilibrium model.
fraction of carbon in the gas phase. The availability
of carbon, i.e. the overall fraction of carbon entering
chemical equilibrium, is therefore

C
= 
C;1
− 
C;2
: (26)
Since one mole of methane contains four moles of
hydrogen atoms, the corresponding availability of hy-
drogen at equilibrium is

H
=1−
4
C;2
n
C
n
H
: (27)
The remaining error in the predictions can be at-
tributed to failure to achieve complete conversion for
other elements, as well as measurement errors. Since
there is no systematic method for handling incomplete
conversion of elements other than carbon, as a rst
approximation, we assume complete conversion for
all elements other than carbon and hydrogen. Thus,
Eq. (23) is reduced to
b
∗
=(
C
n
C
;
H
n
H
;n
O
;n
N
;n
S
): (28)
The modied equilibrium model is illustrated
schematically in Fig. 17. The reaction system is
assumed to be comprised of a mainstream in chem-
ical equilibrium and a bypass zone, controlled by
non-equilibrium factors. The eective abundances of
carbon and hydrogen in the equilibrium main stream
are clearly less than those computed from the feed.
Consequently, the eective air ratio exceeds that

190 X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193
0
10
20
30
40
50
60
70
0 0.2 0.4 0.6 0.8 1
Air ratio,
(
-
)
Species contents, (mol %)
N
2
H
2
O
CO
2
CH
4
CO
H
2
Fig. 18. Comparison between species molar contents predicted by
the modied equilibrium model and measured data for a tempera-
ture of 1100 K. Legends: —H
2
,
•
—CH
4
, —CO, —CO
2
,
+—H
2
O, ♦—N
2
. For experimental details see Table 2.
based on the overall stoichiometry. The modied b
∗
gives a much-improved approximation of the actual
element abundance entering equilibrium, leading to
substantially better predictions.
4.2. Comparison with experimental results
Fig. 18 shows the variation of gaseous species con-
tents with air ratio predicted by the modied model,
compared with the measured data. While N
2
is pre-
dicted to be similar to that from the pure equilibrium
model, signicant changes are found in H
2
,CH
4
, CO,
CO
2
and H
2
O. As in Fig. 13, all species are dry-gas
molar contents except for H
2
O. The modied model
predicts much higher CH
4
and H
2
O contents than the
pure equilibrium model. In addition, the predicted con-
centrations of H
2
and CO now go through maxima,
not seen in the pure equilibrium predictions.
Very little dierence is found between predictions
for 1000 and 1100 K, so only the latter are plotted.
Predictions for the CH
4
content agree very well with
experimental data. The predicted H
2
molar content re-
mains higher than the experimental data except for a
few cases. Similar deviations for H
2
were observed in
previous work [13]. The main reason for this deviation
is likely to be the fractional availability of water to the
water–gas shift reaction. This reaction is moderately
exothermic (H
o
298
= −41:1kJ=mol CO), so that its
equilibrium constant decreases with increasing tem-
perature. Therefore, the H
2
=CO molar ratio may fall
below 1 at high temperatures. The hydrogen produced
by the shift reaction is over-predicted if all the water in
the system is assumed to reach chemical equilibrium.
CO contents are under-predicted by the modied
model with a relative dierence of 20–25%, while pre-
dicted CO
2
contents are in good agreement with ex-
perimental data. Since H
2
contents are over-predicted,
while CO contents are under-predicted, the resulting
H
2
/CO molar ratios are signicantly higher than the
measured data, except for one test run with a particu-
larly high moisture content in the fuel (22.0%). Good
agreement is found between measured and predicted
N
2
and H
2
O contents in the product gas.
Part of the deviation also comes from the as-
sumption that all the unconverted carbon stays as
C(s) or pure graphitic carbon. However, in real pro-
cesses, the unconverted carbon usually occurs as coke
(CH
x
; 0 ¡x¡1), which decreases the availability
of both carbon and hydrogen. Further improvements
in this respect should be sought in future work. How-
ever, it is encouraging that the modied equilibrium
model gives much better predictions than the pure
equilibrium model.
Fig. 19 compares the predicted gas heating value
with the measured data. The dashed lines are cal-
culated from the pure equilibrium model, while the
solid line shows predictions of the modied model.
An exponential correlation of gas heating value with
air ratio (as shown in Fig. 6) seems quite reason-
able for tting experimental data over the range
0:2 6 a 6 0:5. There is good agreement between the
measured data and those predicted by the modied
equilibrium model. The error in the predicted H
2
=CO
molar ratio in the product gas has little inuence on
the resulting gas heating value, owing to the fortu-
itous similarity of the heats of combustion of H
2
and
CO (286 and 283 kJ=mol, respectively).
The measured and predicted cold gas eciencies
are compared in Fig. 20. Again, the dashed lines are
calculated from the pure equilibrium model, while
the solid line shows the prediction of the modied
model. A striking feature is the existence of a max-
imum cold gas eciency at a non-zero air ratio for
both versions of the model. The operating tempera-
tures in the experimental study varied from ∼ 1000 to

X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193 191
0
2
4
6
8
10
12
0 0.2 0.4 0.6 0.8 1
Air ratio,
(
-
)
Dry gas heating value, (MJ/Nm
3
)
Pure equilibrium model
1100 K
1000 K
900 K
800 K
Modified model
1100 K
Fig. 19. Eect of air ratio on predicted and experimental dry
gas heating value from sawdust gasication at 1:013 bar. Dashed
lines: pure equilibrium model predictions; solid line: modied
model prediction for 1100 K. Experimental data: a =0:22– 0.54,
M =4:2–22.0%, T =970–1090 K. —best cases;
•
—time-mean
values.
0
20
40
60
80
100
120
0 0.2 0.4 0.6 0.8 1
Air ratio,
(
-
)
Cold gas efficiency,E
1
, (%)
Pure equilibrium model
1100 K
900 K
1000 K
Modified model
1100 K
Fig. 20. Eect of air ratio on predicted and experimental cold-gas
eciencies from sawdust gasication at 1:013 bar. Dashed lines:
pure equilibrium model predictions for 1100 K. a =0:22– 0.54,
M =4:2–22.0%, T = 970–1090 K. —time-mean values.
1100 K. Therefore, maximum gasication eciency
could only be expected for a¡0:3, though the pure
equilibrium model predicts lower optimum air ratios
for dierent operating temperatures. The modied
model is much more successful. Despite scatter, the
experimental data are in substantial agreement with
the modied model predictions.
5. Conclusions
(1) Pilot plant tests of biomass gasication indicate
that the product gas composition and heating
value depend heavily on the air or O/C ratio and
suspension temperature. Gas heating value can
be increased by increasing the suspension den-
sity. Ash re-injection improved carbon conver-
sion, while steam injection improved the quality
(i.e. heating value) of the product gas.
(2) While carbon conversion increased with increas-
ing O/C molar ratio, the cold-gas gasication ef-
ciency decreased. Gasication eciency was
maximized within an optimum range of air ratio
(O=C=1:3–1.6, or a=0:25–0.33), while keeping
the tar yield relatively low. Tar yield decreased
exponentially with increasing operating temper-
ature. Addition of a reforming catalyst signi-
cantly reduced tar yield, while secondary air had
only a very limited eect on tar removal for a
constant total air ratio.
(3) The non-stoichiometric equilibrium model de-
veloped in this study predicts that the product gas
composition depends primarily on the air ratio.
Pressure only inuences the results signicantly
over a limited temperature range (750 –1200 K).
An air ratio of 0.2– 0.3 is most favourable for pro-
ducing CO-rich gas. To produce hydrogen-rich
gas at atmospheric pressure, the system should
operate in the temperature range from 1100 to
1300 K and at an air ratio of 0.15 –0.25.
(4) Experimental evidence is provided demonstrat-
ing that real gasication processes deviate from
chemical equilibrium. Any equilibrium-based
model that does not account for such deviations
is subject to error in predicting gas composi-
tion and overall eciency. The pure equilib-
rium model is therefore modied to account for
non-equilibrium factors. The concept of elemen-
tal availability is proposed, incorporating exper-
imental results regarding unconverted carbon
and methane.

192 X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193
(5) The modied model predicts product gas com-
positions, product heating value, gas yield and
cold gas eciency in good agreement with ex-
perimental data. Further work is recommended
to examine the role of tar, coke (containing hy-
drogen as well as carbon) and moisture in deter-
mining the nal gas composition.
Acknowledgements
The authors are grateful to Milaim Dervishaj and
Dr. Yonghua Li for their assistance in the experimental
study. Financial assistance from the Natural Sciences
and Engineering Research Council of Canada is grate-
fully acknowledged. Use of the facilities of the Pulp
and Paper Centre at UBC is also greatly appreciated.
References
[1] Schuster G, Loer G, Weigl K, Hofnauer H. Biomass
steam gasication—an extensive parametric modeling study.
Bioresource Technology 2001;77:71–9.
[2] Knoef HAM. Gaser inventory, http://www.btgworld.com/gi/,
2002.
[3] Beenackers AACM. Biomass gasication in moving beds,
a review of European technologies. Renewable Energy
1999;16:1180–6.
[4] Babu SP. Thermal gasication of biomass technology
development: end of task report for 1992 to 1994. Biomass
and Bioenergy 1995;9:5–15.
[5] Ergudenler A, Ghaly AE. Quality of gas produced from wheat
straw in a dual-distributor type uidized bed gasier. Biomass
and Bioenergy 1992;3:419–30.
[6] Caballero MA, Aznar MP, Gil J, Martin JA, Frances E,
Corella J. Commercial steam reforming catalysts to improve
biomass gasication with steam-oxygen mixtures. 1. Hot gas
upgrading by the catalytic reactor. Industrial and Engineering
Chemistry Research 1997;36:5227–39.
[7] Corella J, Orio A, Aznar P. Biomass gasication with
air in uidized bed: reforming of the gas composition
with commercial steam reforming catalysts. Industrial and
Engineering Chemistry Research 1998;37:4617–24.
[8] Liu P, Guo XF, Wu CZ, Chen Y, Arai N. Gasication
characteristics of biomass wastes in uidized bed gasier.
Journal of Propulsion and Power 2000;16:606–8.
[9] Vamvuka D, Woodburn ET, Senior PR. Modeling of an
entrained ow coal gasier. 1. Development of the model
and general predictions. Fuel 1995;74:1452–60.
[10] Chen CX, Horio M, Kojima T. Numerical simulation
of entrained ow coal gasiers. Part I: modeling of
coal gasication in an entrained ow gasier. Chemical
Engineering Science 2000;55:3861–74.
[11] Fiaschi D, Michelini M. A two-phase one-dimensional
biomass gasication kinetics model. Biomass and Bioenergy
2001;21:121–32.
[12] Chern SM, Walawender WP, Fan LT. Equilibrium modeling
of a downdraft gasier. 1. Overall gasier. Chemical
Engineering Communications 1991;108:243–65.
[13] Ruggiero M, Manfrida G. An equilibrium model for biomass
gasication processes. Renewable Energy 1999;16:1106–9.
[14] Knoef HAM. The UNDP/World Bank monitoring program
on small scale biomass gasiers (BTG’s experience on tar
measurement). Biomass and Bioenergy 2000;18:39–54.
[15] Gil J, Corella J, Aznar MP, Caballero MA. Biomass
gasication in atmospheric and bubbling uidized bed: eect
of the type of gasifying agent on the product distribution.
Biomass and Bioenergy 1999;17:389–403.
[16] Brereton CMH, Grace JR, Yu J. Axial gas mixing in a
circulating uidized bed. In: Basu P, Large JF, editors.
Circulating uidized bed technology II. New York: Pergamon
Press, 1988. p. 307–14.
[17] Lide DR. CRC Handbook of chemistry and physics, 74th ed.
Boca Raton, FL, USA: CRC Press, 1994.
[18] van der Drift A, van Doorn J, Vermeulen JW. Ten residual
biomass fuels for circulating uidized-bed gasication.
Biomass and Bioenergy 2001;20:45–56.
[19] Li XT. Biomass gasication in a circulating uidized
bed. PhD Thesis, Department of Chemical and Biological
Engineering, University of British Columbia, 2002.
[20] Pan YG, Roca X, Velo E, Puigjaner L. Removal of tar by
secondary air in uidized bed gasication of residual biomass
and coal. Fuel 1999;78:1703–9.
[21] Moersch O, Splietho H, Hein KRG. Tar quantication with
a new online analyzing method. Biomass and Bioenergy
2000;18:79–86.
[22] Rapagna S, Jand N, Kiennemann A, Foscolo PU.
Steam-gasication of biomass in a uidized bed of olivine
particles. Biomass and Bioenergy 2000;19:187–97.
[23] Sutton D, Kelleher B, Ross JRH. Review of literature
on catalysts for biomass gasication. Fuel Processing
Technology 2001;73:155–73.
[24] Hebden D, Stroud HJF. Coal gasication processes. In: Elliott
MA, editors. Chemistry of coal utilization 2nd suppl. vol.
New York: Wiley, 1981, p. 1599–752.
[25] Probstein RF, Hicks RW. Synthetic fuels. New York:
McGraw Hill, 1982.
[26] Smith WR, Missen RW. Chemical reaction equilibrium
analysis: theory and algorithms. New York: Wiley, 1982.
[27] White WB, Johnson SM, Dantzig GB. Chemical equilibrium
in complex mixtures. Journal of Chemical Physics
1958;28:751–5.
[28] Zeleznik FJ. Calculation of complex chemical equilibria.
Industrial and Engineering Chemistry 1968;60:27–57.
[29] Li XT, Grace JR, Watkinson AP, Lim CJ, Ergudenler
AE. Equilibrium modeling of gasication: a free energy
minimization approach and its application to a circulating
uidized bed coal gasier. Fuel 2001;80:195–207.
[30] Chase MW Jr, Davies CA, Downey DJ Jr, Frurip DJ,
McDonald RA, Syverud AD, JANAF thermodynamic tables,

X.T. Li et al. / Biomass and Bioenergy 26 (2004) 171 – 193 193
3rd ed. Journal of Physical and Chemistry References Data
1985;14:1–1856 (Suppl 1.).
[31] Watkinson AP, Lucas JP, Lim CJ. A prediction of
performance of commercial coal gasiers. Fuel 1991;70:
519–27.
[32] von Fredersdor CG, Elliott MA. Coal gasication. In: Lowry
HH, editor. Chemistry of coal utilization. New York: Wiley,
1963.
[33] Coates RL, Chen CL, Pope BJ. Coal devolatilization in a low
pressure, low residence time entrained ow reactor, in Coal
gasication. In: Massey LG, editor. Advances in chemistry
series 131. Washington DC: American Chemical Society,
1974, p. 92–107.
[34] Grace JR, Li XT, Lim CJ. Equilibrium modelling of catalytic
steam reforming of methane in membrane reactors with
oxygen addition. Catalysts Today 2001;64:141–9.