Documentos de Académico
Documentos de Profesional
Documentos de Cultura
ct
#
c;
Q
cz
"!R
N
, (1)
* moisture
c
+
ct
#
c;
Q
+
cz
"!m
+
, (2)
* gas-phase species
c
c
G
ct
#
c(
G
;
E
)
cz
"
c
cz
cD
G
E
c>
G
cz
#M
G
H
v
GH
R
H
#c
NG
,
i"O
`
, H
`
, CO, CO
`
, CH
"
, j"c1}c4, g1}g3, wg, (3}7)
* steam
c
c
'
`
'
ct
#
c(
'
`
'
;
E
)
cz
"M
'
`
'
H
v
'
`
'H
R
H
#
c
cz
cD
T
E
c>
'
`
'
cz
#m
+
#c
N'
`
'
, j"c1}c4, g1}g3, wg, (8)
* vapor-phase tar
c
c
2
ct
#
c(
2
;
E
)
cz
"
c
cz
cD
2
E
c>
2
cz
#v
2
R
N
!R
N`
, (9)
* nitrogen
`
`
"
E
!
G$`
`
G
, (10)
* total gas continuity
c
c
E
ct
#
c(
E
;
E
)
cz
"
G
H
v
GH
M
G
R
H
#m
+
#(1!v
!
)R
N
, (11)
i"N
`
, O
`
, H
`
, CO, CO
`
, CH
"
, H
`
O, j"c1}c4, g1}g3, wg,
* solid-phase energy
c(
G
G
H
G
)
ct
"
c
cz
zH
Q
c
Q
cz
#
c(;
Q
G
H
G
)
cz
!
H
R
H
AH
H
!Q
QE
#Q
QU
!m
+
, (12)
H
G
"c
QG
(
Q
!
"
), i"B, C, M, j"c5, g1}g3, p1,
* gas-phase energy
c
c(
G
G
H
G
)
ct
"
c
cz
zH
E
c
E
cz
!
c(;
E
G
H
G
)
cz
#Q
QE
#Q
EU
!
H
R
H
AH
H
, (13)
h
G
"c
EG
(
E
!
"
), i"N
`
, O
`
, H
`
, CO, CO
`
, CH
"
, H
`
O, T,
j"c1}c4, wg, p2,
Q
QE
"h
QE
v
N
(
Q
!
E
), Q
QU
"
4h
QU
D
(
U
!
Q
),
Q
EU
"
4h
EU
D
(
U
!
E
),
* ideal gas law, modi"ed Darcy law
P"
E
R
E
M
E
, (14)
B
E
j
cP
cz
";
Q
!;
E
. (15)
equilibrium temperature, but they cannot be used for
reactor design. Only in a very few cases chemical reaction
kinetics and transport phenomena have been properly
coupled to model conventional (Groeneveld & van
Swaaij, 1980) and strati"ed (Manurung & Beenackers,
1994) downdraft gasi"ers. However, the description of
the `#aming pyrolysisa (Reed & Markson, 1985), that is
biomass pyrolysis and combustion of volatile pyrolysis
products, is still based on equilibrium models or on
highly simpli"ed treatments, which assume the existence
of a stable combustion zone with in"nitely oxygen con-
version rate near the air inlet. Furthermore, these models
describe steady-state conditions. Thus, they do not allow
the prediction of the dynamic behaviour of strati"ed
gasi"ers and of the di!erent modes of stabilization of the
reaction front and thus the size of the reduction zone.
This study proposes a more advanced, dynamic model,
which includes "nite rate kinetics for biomass pyrolysis
and combustion of char, gaseous species and tars. The
model was used to simulate the e!ects of changes in
transport coe$cients, chemical kinetics and operating
conditions on the performances of the gasi"cation pro-
cess, in view of reactor design and optimization, espe-
cially in relation to the dynamic behaviour of downdraft
reactors, i.e. to top- or grate-stabilized operation.
2. Mathematical model
The strati"ed gasi"er model is based on mass and
energy balances for the solid phase and mass and energy
balances for the gas phase, written for a one-dimensional,
unsteady state system. Species considered are: oxygen,
nitrogen, hydrogen, steam, carbon dioxide, carbon mon-
oxide, methane and hydrocarbons (which also include
tars). The pressure drop in the reactor is modelled using
the generalized Darcy law but, given the large bed per-
meabilities, simulations have been carried out with the
assumption of constant pressure (the gas velocity is deter-
mined from the continuity equation and the density of
the mixture from the ideal gas law). Model equations are
listed in Table 1. As the concurrent #ows of solid and gas
descend across the reactor, several processes take place,
namely, moisture evaporation, biomass pyrolysis, char
combustion and gasi"cation, combustion of the gases
and thermal cracking of the tars, in accordance with the
schematic representation reported in Fig. 1. The main
features of submodels applied for these processes can be
derived from Table 2 and the relevant approximations
made are listed below.
Recently, moisture evaporation has been described in
some detail for a single biomass particle (Di Blasi, 1998).
For conditions similar to those of "xed-bed gasi"cation,
it has been shown that both gas-phase and liquid-phase
transport phenomena can play a controlling role. How-
ever, the single-particle e!ects are usually neglected in the
C. Di Blasi / Chemical Engineering Science 55 (2000) 2931}2944 2933
Fig. 1. Schematic of the strati"ed concurrent (downdraft) gasi"er.
Table 2
Moisture evaporation, chemical reactions and transport coe$cients
Evaporation/condensation
m
+
"v
N
k
K
(
T
!
'
`
'
), (16)
p
T
/p
TQ
"exp[(17.884!0.1423
Q
#23.63;10``
Q
)1.0327!67.41
;10`
Q
)"`3]. (17)
Pyrolysis reactions
B
I
N
Pv
A
CHAR#v
%
GAS
#v
2
AR, (p1)
AR
I
N`
PGAS
`
(p2)
R
N
"A
N
exp
!
E
N
R
Q
, (18)
R
N`
"cA
N`
exp
!
E
N`
R
E
2
. (19)
Gas-phase combustion
(tar) CH
```
O
""```
#0.867O
`
I
A
PCO#0.761H
`
O, (c1)
CH
"
#1.5O
`
I
A`
PCO#2H
`
O (c2)
2CO#O
`
I
A`
P2CO
`
, (c3)
H
`
#O
`
I
A"
P2H
`
O, (c4)
R
H
"A
H
exp
!
E
H
R
E
E
C"`
G
C
'
`
, j"c1, c2, i", CH
"
, (20)
R
A`
"A
A`
exp
!
E
A`
R
E
C
''
C"``
'
`
C
'
`
'
, (21)
R
A"
"A
A"
exp
!
E
A"
R
E
C
'
`
C
'
`
. (22)
Gas-phase water gas shift
CO#H
`
O
I
UE
& CO
`
#H
`
, (wg)
R
UE
"ck
UE
C
''
C
'
`
'
!
C
''
`
C
'
`
K
#
, (23)
k
UE
"A
UE
exp
!
E
UE
R
E
, K
#
"A
#
exp
E
#
R
E
. (24)
K
#
"A
#
exp
E
#
R
E
. (25)
Heterogeneous reactions of char
CH
?
O
@
#O
`
I
A`
P
2!2![#
:
2
CO#
2#[!
:
2
!1
CO
`
#:2H
`
O, (c5)
CH
?
O
@
#CO
`
I
E
P2CO#[H
`
O#
:
2
![
H
`
, (g1)
CH
?
O
@
#
2!
:
2
#[
H
`
I
E`
PCH
"
#[H
`
O, (g2)
CH
?
O
@
#(1![)H
`
O
I
E`
PCO#
1![#
:
2
H
`
, (g3)
formulation of reactor models, because the characteristic
times of moisture evaporation are orders of magnitude
shorter than those of char gasi"cation. Therefore, in most
cases the process is assumed to take place instanta-
neously (for instance Yoon, Wei & Denn, 1978) or to be
a di!usion limited process (Hobbs et al., 1993). The latter
treatment is retained in this study with the use of an
empirical expression for the vapour pressure, to take into
account the di!erences in evaporation rates for capillary
and bound water (Di Blasi, 1998) (p
TQ
(
Q
) is the
Clausius}Clapeyron expression).
It is well known that pyrolysis of large biomass par-
ticles is a process controlled by heat and mass transfer
(for instance, Di Blasi, 1996a). More speci"cally, the
conversion times are largely determined by the rate of
inward heat conduction, whereas product distribution is
highly dependent on intra-particle residence times of
volatiles and thus on the extent of secondary cracking.
Given the very steep temperature pro"les along the gasi-
"er axis in the concurrent con"guration, it is often (Bliek,
1984) assumed that intra-particle heat transfer resist-
ances are much lower than extra-particle axial resist-
ances. Thus, the particle is assumed to be thermally thin.
In this study, this treatment is still retained but apparent
2934 C. Di Blasi / Chemical Engineering Science 55 (2000) 2931}2944
Table 2 (Continued)
R
H
"v
N
C
G
(1/k
K
)#(1/k
H
)
, k
H
"A
H
exp
!
E
H
R
Q
, j"c5, g1!g3,
i"O
`
, CO
`
, H
`
, H
`
O (26}27)
r
A
R
`
"(1!,)X#,, (28)
X"
;
Q
;
Q"
, (29)
v
N
(r
A
)"
3(1!c)
r
A
, (30)
A"
c;
Q
cz
"!M
!
(R
A`
#R
E
#R
E`
#R
E`
). (31)
Solid/gas heat transfer coe$cient, mass transfer coe$cient
h
QE
"
2.06c
NE
E
;
E
c
Re"```Pr``, (32)
k
K
"
2.06c
NE
;
E
c
Re"```Sc``. (33)
Properties
zH
E
"cz
E
, zH
Q
"cz
PE
#c
z
Q
(z
Q
/(dz
PQ
)#1.43(1!1.2c))
, (34)
z
PE
"4o0.05`
E
, (35)
z
PQ
"4o0.85`
Q
, (36)
z
Q
"0.0013#0.05(
Q
/1000)#0.63(
Q
/1000)`
J
ms K
(37)
z
E
"4.8;10"'``
E
J
ms K
(38)
j"1.98;10`(
E
/300)``
kg
sm
(39)
Table 3
Devolatilization data: gas composition is expressed as percent of the
initial dry biomass
I II III
v
'
0.33 0.41 0.33
v
'
0.48 0.4 0.5
v
'
0.19 0.19 0.2
Gas 1 CO 7.5% 5.5% 11%
CO
`
13% 10.5% 11%
H
`
1% 0.02% 1%
CH
"
1.5% 1% 2%
H
`
O 25% 23% 25%
Gas 2 CO 9.5% 9.5% 13%
CO
`
5.7% 5.7% 3%
CH
"
3.8% 3.8% 4%
kinetics are used for the pyrolysis process. A one-step
global reaction is considered, where the fractions of
gases, tars and chars (Di Blasi, 1993) produced and the
gas composition should be speci"ed. Tars undergo sec-
ondary cracking in the void spaces of the bed (one-step
global reaction), to produce secondary gases, whose com-
position should again be speci"ed. The size (volume) and
the solid velocity of the spherical particles remain con-
stant as the bed density decreases during moisture evap-
oration and biomass devolatilization. Consequently, the
porosity of this portion of the bed varies. However, this
e!ect is not taken into account as single-particle simula-
tions (Di Blasi, 1997) show that porosity variation exerts
a negligible in#uence on the devolatilization character-
istics.
For primary wood pyrolysis, apparent activation ener-
gies are reported to be in the range 63}125 kJ/mol de-
pending on the wood species, sample size and heating
conditions (Roberts, 1970). Reference values of the ac-
tivation energy and pre-exponential factor are chosen as:
A
N
"1.516;10` s, E
N
"105 kJ/mol, determined
for beech wood cylinders 1}2 cm thick. Kinetic constants
for tar cracking from Liden, Berruti and Scott (1988) are
used. While the composition of the secondary gas has
been estimated on the basis of literature data obtained
for wood (Boroson, Howard, Longwell & Peters, 1989),
an experimental study (Di Blasi, Signorelli, Di Russo
& Rea, 1999) has been carried out to determine
primary pyrolysis product yields and gas composition for
conditions corresponding to downdraft gasi"cation.
Packed-beds (diameter 4 cm) of biomass particles have
been exposed to di!erent radiation intensities, which
correspond to maximum (surface) temperatures in the
bed in the range 650}1000 K. A continuous nitrogen #ow
across the bed reduces the residence times of volatile
products and makes the activity of extra-particle second-
ary reactions negligible. As expected, devolatilization
characteristics, which are representative of intra-particle
primary and secondary reactions, have been found to
depend both on the heating conditions and the biomass
type. Assuming a surface temperature of 850 K, three sets
of devolatilization data ha ve been considered for the
simulation of downdraft gasi"cation (Table 3). The "rst
data set (I) refers to 0.5}1 cm thick wood chips, the
second (II) to rice husks and the third (III) has been
considered for a parametric investigation of the e!ects of
variations in pyrolysis products on the gasi"cation char-
acteristics.
Combustion of volatile products is an important pro-
cess in downdraft gasi"cation. Here, the treatment pro-
posed by Bryden and Ragland (1996) for the "xed-bed
combustion of biomass is adopted. Hydrocarbons, which
include tars and methane, react with oxygen to form
water vapour and carbon monoxide, according to
a "nite-rate, global reaction. The formation of the reac-
tion intermediate, carbon monoxide, is important for the
correct prediction of the ignition delay time. Global,
C. Di Blasi / Chemical Engineering Science 55 (2000) 2931}2944 2935
"nite rate reactions are also considered for the combus-
tion of carbon monoxide and hydrogen. However, for
computational simplicity, instead of an in"nitely fast rate
of hydrogen consumption, a high pre-exponential factor
and a low activation energy have been introduced for
a second-order reaction rate (E"83 kJ/mol, A"
10` m`/s/mol). Tars are modelled as hydrocarbons
CH
```
O
""```
(Bryden & Ragland, 1996), with molecu-
lar weight equal to 95 (Corella, Aznar, Delgado & Aldea,
1991).
As for homogeneous reactions, the water gas shift
reaction is also considered and is described by "nite rate
kinetics, with the equilibrium constant reported by Yoon
et al. (1978) and the kinetic constants derived by Biba,
Macak, Klose and Malecha (1978).
Combustion and gasi"cation reactions of char are het-
erogeneous and are described by the unreacted core,
shrinking particle model, where a steep reaction zone
propagates through the isothermal char. Two mecha-
nisms responsible for the global reaction rate are con-
sidered: di!usion through the gas "lm, surrounding the
particle, and intrinsic chemical kinetics. To account for
the simultaneous e!ects of the two resistances, an e!ec-
tive volumetric reaction rate is introduced, based on the
assumption of a linear dependence of the reaction rate on
the oxidizing/gasifying species concentration. As a conse-
quence of the heterogeneous reactions, the particle dia-
meter shrinks and the density of the bed (and porosity)
remains constant, causing a gradual decrease in the solid
velocity. The minimum size of the particle and conse-
quently the maximum particle density number (v
N
(Eq.
(23))) depend on the ash content of the solid (Hobbs et al.,
1993).
For simplicity, chars are assumed to consist of pure
carbon (:"["0) and the products of heterogeneous
combustion to be carbon dioxide only ("1), though
this assumption is not valid in general (Hobbs et al.,
1993). The intrinsic kinetics of wood char combustion are
described according to the data determined for cellulosic
fuels (Kashiwagi & Nambu, 1992). Kinetics of char gasi"-
cation by carbon dioxide and steam are those reported
by Groeneveld and van Swaaij (1980). Hydrogasi"cation
is usually negligible under atmospheric conditions and its
rate is assumed to be slower by three orders of magnitude
compared with carbon dioxide and steam gasi"cation
(Hobbs et al., 1993).
Literature correlations are used for the e!ective ther-
mal conductivity of the bed (Goldman, Xieu, Oko, Milne
& Essenhigh, 1984), the e!ective bed-to-wall heat transfer
coe$cients (Hobbs et al., 1993) (not reported in Table 2),
the solid/gas heat transfer and the mass transfer coe$-
cients (Gupta & Thodos, 1963). However, the solid/gas
heat transfer coe$cient estimated from non-reacting
system data can exceed the experimental values in
reacting gasi"ers (Hobbs et al., 1993), as a consequence
of unsteady heat transfer. Therefore, the experimental
correlation is multiplied by empirical factors () with
values in the range 0.02}1 (Cho & Joseph, 1981;
Radulovic, Ghani & Smoot, 1995). In this study the
reference value for is taken equal to 1. Some di$culties
also exist in the application of the Gupta and Todos
correlation for the mass transfer coe$cient, mainly be-
cause of the di!erent scales of the conversion units (in
particular, there is large uncertainty for low Reynolds
numbers) and again the changes introduced by chemical
reactions. Apart from the correlation suggested by Bhat-
tacharya, Salam, Dudukovic and Joseph (1986), in all
cases the mass transfer coe$cient decreases as the par-
ticle size is increased, but di!erences become very high as
the particle diameter reduces to very low values. This is
critical, because in the unreacted-core, shrinking particle
model, the maximum temperature predicted in the oxida-
tion/reduction zone becomes highly dependent on the
correlation used. On the other hand, at high temper-
atures, "lm transfer becomes the controlling resistance.
To avoid unrealistic temperature values, corrective fac-
tors, which limit the maximum value of k
K
(Bhattacharya
et al., 1986) or reduce its value for all conditions (for
instance, 0.5 in Goldmann et al., 1984) have been sugges-
ted. Both procedures have been applied to the Gupta and
Todos correlation, used in this model. Given the proper
values of the correction factor or its maximum allowed in
the course of the reaction, the two procedures do not
result in any change from the qualitative point of view in
the predictions of process characteristics. Therefore, only
the e!ects associated with the changes in the maximum
value of k
K
(indicated as kH
K
) on the prediction of the
gasi"cation process will be discussed. The reference value
chosen for kH
K
is 0.045 m/s.
The variation of the gas thermal conductivity and
viscosity with temperature is described as in Purnomo,
Aerts and Ragland (1990). A constant value is assumed
for di!usivities (0.2;10" m`/s) and speci"c heats of
gaseous species (evaluated for a temperature of 1000 K)
and speci"c heats of biomass and char (1.34 kJ/kg) (Di
Blasi, 1996b). All the thermochemical data referred to or
listed in this section, with the inclusion of the devolatiliz-
ation data set I, will be indicated in the following as
reference data (a summary of kinetic constants is re-
ported in Table 4).
The numerical solution of the model equations is
based on operator splitting procedures and "nite-di!er-
ences approximations. The reactor is divided into a set of
elementary cylindrical cells, whose cross sections co-
incide with the reactor cross section, while the height can
be variable. The solution procedure is divided into three
stages, corresponding to chemical reaction processes,
heat exchange (between phases and with the reactor wall)
and transport phenomena. For each time step, in the "rst
two stages, the solution is calculated, for each control
volume, of linearized ordinary di!erential equations, by
means of a "rst-order implicit Euler method. In the third
2936 C. Di Blasi / Chemical Engineering Science 55 (2000) 2931}2944
Table 4
Reference value for kinetic constants
Pyrolysis
A
NJ
(s) 1.516;10` Roberts (1970)
E
NJ
(kJ/mol) 105 Roberts (1970)
A
N
`
(s) 4.28;10' Liden et al. (1988)
E
N
`
(kJ/mol) 107 Liden et al. (1988)
Ah
N