Sunteți pe pagina 1din 13

Biomass and Bioenergy 22 (2002) 41–53

Modeling thermally thick pyrolysis of wood


Kenneth M. Brydena; ∗ , Kenneth W. Raglandb , Christopher J. Rutlandb
a Department of Mechanical Engineering, Iowa State University, Ames, IA 50011-2161, USA
b Department of Mechanical Engineering, University of Wisconsin-Madison, Madison, WI 53706, USA

Received 21 September 2001; accepted 21 September 2001

Abstract
A general model of the pyrolysis of a wood slab is presented and validated with a set of heat release data. The model is
applied to particle half-thicknesses from 5 m to 5 cm, temperatures from 800 to 2000 K, and moisture contents from 0%
to 30%. Internal temperatures, pyrolysis rates and yields of tar, hydrocarbons and char are presented. Four pyrolysis regimes
are identi2ed, depending on external temperature and particle size: thermally thin—kinetically limited, thermally thin—heat
transfer limited, thermally thick, and thermal wave regimes.  c 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Modeling; Wood; Pyrolysis; Fire

1. Introduction contents with or without external combustion. In this


paper, this model has been utilized to examine the
Understanding the mechanisms and rates of wood eAect of wood size, moisture content, and temperature
combustion is important for modeling building 2res on the rate of pyrolysis, and formation of tar, volatiles,
and for designing wood stoves, furnaces, and boilers. and char.
Fires in buildings involve large wood slabs. Stoves
typically use stick wood and small logs. Furnaces
and boilers are designed for sawdust, woodchips,
chunkwood, or logs. Modeling combustion of these 2. Background
wood fuels involves applying chemical kinetics, heat
transfer, and mass transfer to drying, pyrolysis, char, Several researchers have developed detailed numer-
and external gas phase processes. A detailed computa- ical models of wood pyrolysis [2–11]. In general, these
tional model of wood combustion has been developed models include gas Bow within the particle in con-
and validated by Bryden [1]. This wood combustion junction with gas phase continuity, solid phase con-
model handles the full range of sizes and moisture tinuity, and energy conservation within the particle.
Multiple, competing two-step reactions are used to
model pyrolysis. The primary assumptions of these
models are one-dimensional heat and mass trans-
∗ Corresponding author. Tel.: +1-515-294-3891; fax: +1-515- port and local thermal equilibrium between the solid
294-3261. and the gaseous species. In general, several areas of
E-mail address: kmbryden@iastate.edu (K.M. Bryden). concern in wood pyrolysis are not modeled. These

0961-9534/02/$ - see front matter  c 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 9 6 1 - 9 5 3 4 ( 0 1 ) 0 0 0 6 0 - 5
42 K.M. Bryden et al. / Biomass and Bioenergy 22 (2002) 41–53

Nomenclature

A pre-exponential frequency factor  dynamic viscosity


dp eAective pore diameter  density
ė rate of chemical energy release ˜ uncorrected density (see, Ref. [1])
E activation energy  Stefan–Boltzman constant
f shrinkage factor !˙ i mass rate of production of species i per
G mass Bux unit volume
h heat transfer coeHcient
Ihk heat of reaction for reaction k Dimensionless numbers
Ih0i heat of formation of species i Bi Biot number
i internal energy Py second (external) pyrolysis number
k rate constant
Lc characteristic length Subscripts
p pressure C char
Q external radiative heat Bux cond conduction
R universal gas constant dry dry
t time eA eAective
T temperature g gas
Wi molecular weight of species i L light hydrocarbons
X speci2c gravity M moisture
Yi mass fraction of species i pyr pyrolysis
r recondensation
Greek letters rad radiation
 void fraction T tar
 reaction progress variable V water vapor
 permeability W wood
’ emissivity ∞ background
 thermal conductivity

include char shrinkage, diAusive transport of volatile particle. In her review Di Blasi [13] recommends the
species within the solid matrix, recondensation of use of two secondary reactions in which tar cracking to
volatile species, and char combustion and gasi2cation. lighter gases and tar repolymerization to char is mod-
A one-dimensional model for dry wood with char eled. This pyrolysis scheme is utilized in this paper.
shrinkage [4] and several one-dimensional models of In general, the primary pyrolysis reaction rates are
wet wood pyrolysis [7–11] have been presented. represented with an Arrhenius-type temperature de-
The kinetics of wood pyrolysis have been modeled pendence and 2rst-order dependence with respect to
by a number of researchers, and several reviews of this the mass of the unreacted wood. Chan, Kelbon, and
area have been completed [12,13]. The most complete Krieger [11] present kinetic rate parameters for ther-
wood pyrolysis models currently available use mul- mally thick combustion in which the pressure wave
tiple, competing two-step reactions. The most com- within the particle is not modeled and the resistance
mon primary reaction scheme is wood pyrolyzing to to Bow within the particle was assumed to be neg-
tar, gas, and char [11,14 –18]. By using the secondary ligible. Nunn et al. [16] present experimental-based
reactions, product yield becomes a function of tem- kinetic rate parameters which are for the overall de-
perature, heating rate, and residence time within the composition of gas. Similarly, Koufopanos et al. [15]
K.M. Bryden et al. / Biomass and Bioenergy 22 (2002) 41–53 43

have presented kinetics parameters for gas and tar as Each of the drying models has its advantages and
a group. Thurner and Mann [18] present parameters disadvantages, and none has a clear advantage over the
for decomposition to tar, gas, and char based on par- others. The energy balance model is simple and easy
ticles small enough that mass transfer and secondary to implement numerically. Its primary disadvantage is
reactions are negligible. In this study, the parameters that the drying zone is reduced to an in2nitely thin
of Thurner and Mann [18] are used for the tar, gas, moving surface and may not model drying and pyrol-
and char reactions. Secondary vapor phase decompo- ysis of small particles where the thickness of the dry-
sition of tar is generally modeled in the literature as ing zone is not negligible compared to the thickness
a homogeneous process proportional to the tar den- of the material. The kinetic scheme has two primary
sity with an Arrhenius-type temperature dependence advantages. The numerical results match the experi-
[13,19 –23]. Lede et al. [24] note the wide acceptance mental results in that temperature rises continuously
of the tar-to-gas kinetic rate parameters of Liden et al. into and out of a drying plateau, and the scheme can
[19], which have been used by Di Blasi [13] to model be implemented as one more Arrhenius-type relation-
pyrolysis under a wide range of conditions. Limited ship in a numerical code that handles many of these.
information is available on the tar-to-char reaction. The primary disadvantage of using an Arrhenius re-
Di Blasi [13] has estimated this reaction to proceed at lationship for drying is that many diAering physical
1
40 th the rate of the tar-to-gas reaction. phenomena are lumped into a single expression. Addi-
Moisture has a signi2cant impact on the overall tionally, only one-way coupling is maintained between
biomass combustion process, including changing the the drying rate and pressure evolution. In the algebraic
pyrolysis products and increasing the overall combus- model [2] no drying of the wood occurs at tempera-

tion time. Experimental studies (e.g. [26 –28]) provide tures ¡ 100 C. The algebraic expression is only appli-
temperature as a function of time within the wood un- cable to relatively dry materials with moisture contents
der thermally thick combustion conditions. In these ¡ 14%. Moisture content above this limit is handled
experiments, the wood at a particular location rises in the same manner as the thermal model of drying.
from ambient temperature to a plateau within a few Additionally, it is unlikely that equilibrium moisture

degrees of 100 C and then continues to rise. In dry conditions exist within wet wood exposed to combus-
wood there is little or no drying plateau, and as mois- tion level Buxes. The diAusion model accounts for both
ture content increases, the span of this plateau in- drying and recondensation within the wood, however,
creases. Currently, there are four basic ways to model the parameters have been experimentally veri2ed only
the drying of wood under combustion heat Buxes. The for low temperature wood drying and not for combus-
simplest of these is an energy balance in which the tion heat Buxes. Recondensation is accounted for by
drying front is assumed to be in2nitely thin and an assuming that the partial pressure of the water vapor

energy sink at 100 C accounts for the heat of vapor- is equal to its equilibrium pressure. Additionally, the
ization [8–10]. A second method [11] models drying pressure wave is tightly coupled to the drying rate.
of wood as an additional chemical reaction using an Because of this, areas that are not addressed in current
Arrhenius expression. A third method [2] uses an al- models such as cracking and the multi-dimensional
gebraic expression for temperature as a function of aspects of water movement that impact the pressure
moisture content to describe equilibrium moisture con- wave can have signi2cant impact on the drying rate,
tent in wood in an atmosphere of superheated steam. recondensation, and temperature pro2le.
A fourth method, the diAusion model, is based on re-
search and numerical modeling of low temperature

(¡ 200 C) drying of wood [25]. The movement of 3. Description of the mathematical model
unbound water and gas within the wood is modeled
using a modi2ed form of Darcy’s law, and bound wa- A wet wood slab of thickness L, subjected on
ter movement is modeled using a diAusion expression. one surface to a radiant heat Bux is modeled. The
Evaporation is driven both by the saturation pressure one-dimensional wood slab is a porous solid, and the
of the liquid water and the partial pressure of the water void volume is initially 2lled with humid air. Exter-
vapor [7]. nal heat transfer by radiation is assumed to be much
44 K.M. Bryden et al. / Biomass and Bioenergy 22 (2002) 41–53

Table 1
Parameters used for drying, recondensation, and pyrolysis reactions

Reaction A E Ref. Heat of reaction Ref.


(s−1 ) (kJ mol−1 ) (MJ kg−1 )

Pyrolysis reactions
1 k
Wood →gas 1:44 × 104 88.6 18 −0:42 [3]
k2
Wood →tar 4:13 × 106 112.7 18 −0:42 [3]
3 k
Wood →char 7:38 × 105 106.5 18 −0:42 [3]
k4
Tar →gas 4:28 × 106 107.5 19 0.04 [19]
k5
Tar →char 1 × 105 107.5 13 0.04 [19]

Process A E Ref. Heat of reaction Ref.


(s−1 ) (kJ mol−1 ) (MJ kg−1 )
Drying
6 k
Moisture→water vapour 5:13 × 1010 88 1,11 −2:44 [32]

Process k7 Ref. Heat of reaction Ref.


(cm−1 ) (MJ kg−1 )
Recondensation
7 k
Water vapour →moisture 125 1 2.44 [32]
dGV ◦
!˙ r = − = k7 G V ; T ¡ 95 C
where dx ◦
!˙ r = 0; T ¿ 95 C

greater than that by convection. The radiant Bux is convected inward, while the greater portion of py-
incident on the slab from a 2xed temperature external rolysis gases and water vapor pass through the char,
source. As the slab is heated, the energy required providing blowing at the surface. The pyrolysis gases
to dry the wet wood and the diAering eAective con- may react with oxygen within the boundary layer
ductivities between the unreacted, wet wood and the to form a Bame zone that feeds back energy to the
char result in a thermal wave with steep temperature surface. In this paper, radiant energy from external
and species gradients. Material properties (e.g., void surfaces dominates, and an inert external gas is as-
fraction, permeability, thermal conductivity, speci2c sumed. The solid phase equation set used to describe
heat), volume occupied by the solid, pressure, and drying and wood pyrolysis within the solid matrix
temperature all continuously change with time. Py- is based on the diAerential conservation equations of
rolysis of wood to gas, tar, and char; and secondary mass, species, and energy (Table 2).
pyrolysis reactions decomposing tar to gas and char This equation set is derived using the following
form a two-step, 2ve-reaction model for pyrolysis. assumptions: (1) wood is a one-dimensional porous
The complete set of reaction parameters for drying, reacting media with a well-de2ned surface free of
pyrolysis, and recondensation is given in Table 1. cracking; (2) gas and solid are in local thermal equilib-
Wood drying, pyrolysis, and combustion consist of rium; (3) diAusive transport of species within the pores
two separate domains, a porous solid and an external is negligible; (4) the inertial terms within the mo-
gas phase with the surface of the char providing the mentum balance equation are negligible, and the
interface between these two domains. As the surface Bow can be described using Darcy Bow; and (5) the
is heated, the wood is 2rst dried and then pyrolyzed shrinkage occurs only in the radial direction. The 2rst
by radiant and convective heat transfer. A small assumption must be applied with care. Although this
portion of the pyrolysis gases and water vapor are is a common assumption, there are few real cases
K.M. Bryden et al. / Biomass and Bioenergy 22 (2002) 41–53 45

Table 2
Summary of the equations used to model solid phase pyrolysis of wood

Conservation of wood @˜W =@t = !˙ W

Conservation of char @˜C =@t = !˙ C

Conservation of moisture @˜M =@t = !˙ M


     
@ fp @ p  @p !˙ V !˙ L !˙ T
Pressure evolution +f =R + +
@t T @y T  @y WV WL WT

Conservation of tar @(fYT g )=@t + f@(YT Gg )=@y = !˙ T

Conservation of water vapor @(fYV g )=@t + f@(YV Gg )=@ = !˙ V

Conservation of pyrolysis gases @(fg )=@t + f@Gg =@y = !˙ g


 
@(˜W iW + ˜C iC + ˜M iM ) @((YL hL + YT hT + YV hV )g u) @ @T 
Conservation of energy +f =f eA + Ih0i !˙ i
@t @y @y @y

Rate of production of wood !˙ W = − (k1 + k2 + k3 )˜W

Rate of production of char !˙ C = k3 ˜W + fk5 T

Rate of production of moisture !˙ M = − k6 ˜M + k7 GV

Rate of production of tar !˙ T = k1 ˜W − f(k4 + k5 )YT g

Rate of production of vapor !˙ V = k6 ˜M − k7 GV

Rate of production of gas !˙ g = (k1 + k2 )˜W − fk5 T + k6 M − k7 GV

Heat release rate ė = k1 ˜W Ih1 + k2 ˜W Ih2 + k3 ˜W Ih3 + fk4 YT g Ih4
+fk5 YT g Ih5 + (k6 ˜M − k7 GV )Ih6

where this can be fully justi2ed. Thermal and trans- where


port properties are strongly dependent on orientation. ◦
!˙ dry = 0; T ¡ 95 C;
Even so, signi2cant insight can be gained by consid- (1)

ering a one-dimensional model. !˙ dry = k6 ˜M ; T ¿ 95 C:
Assumption two is reasonable based on the Peclet
number [11,29]. By similar discussion it can be shown Chan et al. [11] suggest rate parameters for dry-
that diAusive transport of species within the wood ing of A6 = 5:13 × 106 s−1 and E6 = 88 kJ mol−1 .
pores is negligible [11]. Assumption four is reason- Using these parameters results in a drying plateau at a
able based on the small Reynolds number (¡ 0:01) higher temperature than expected for thermally thick
resulting from the small diameter of the wood pores wood combustion. In this work, the pre-exponential
[29]. Assumption 2ve is based on the assumption of a frequency was adjusted by multiplying A6 above by

one-dimensional media. Derivation of the solid phase 104 to provide a drying plateau between 100 C and

equations with shrinkage is presented in Ref. [1]. 120 C.


In this work drying is modeled using an Arrhenius- Recondensation of the water vapor within the un-
type expression. reacted wood is modeled by assuming that within the
cooler, unreacted regions of the wood, the water va-
k
Moisture →6 water vapor; por exists in a supersaturated state and is condensed
46 K.M. Bryden et al. / Biomass and Bioenergy 22 (2002) 41–53

at a rate proportional to the mass Bux of water vapor. values and correlations used for the physical param-
k
eters are presented in Table 3. The number of grid
Water vapor →7 moisture; points is chosen to provide good spatial resolution of
the steep temperature and species gradients encoun-
where tered. To achieve this, the number of grid points is
!˙ r = − dGV =d x = k7 GV ; T ¡ 95 C

varied depending on the thickness of the wood and

(2) the pyrolysis conditions. The number of grid points
!˙ r = 0; T ¿ 95 C: for the solid phase ranged from 21 to 1201. The time
steps for the solid phase are chosen to ensure stability
The proportionality constant is chosen by comparison
and accuracy of the solution. The time steps ranged
with the experimental results. A recondensation factor
from 1 × 10−4 to 1 s. Details of the numerical model
of 125 cm−1 is used in all cases [1].
are given in Ref. [1].

4. Numerical model
5. Validation of the model
The conservation equations for the solid domain are
solved using second-order accurate 2nite diAerencing. The model was validated using the experimental
A moving boundary incorporates both char shrink- work of Tran and White [27]. In these experiments
age and surface recession due to char combustion and wood blocks, built by gluing strips of wood together
gasi2cation. The diAerenced equations developed are to form 15 cm ×15 cm ×6:4 cm blocks, were exposed
implemented in a Fortran-based computer code. The on one of the 15 cm × 15 cm sides in air to a radiant

Table 3
Values and correlations of physical parameters used in the model

Property Correlation=value Ref.

Void fraction  = 1 − (W + C )=1500 − M =1000 [34]

Permeability  = W + (1 − )C where  = mW =mW; 0


W = 1 × 10−2 darcys [1]
C = 1 darcys [1]

Thermal conductivity eA = cond + rad


rad = =(1 − )’p dp 4T 3 [35]
’p = 0:9 Assumed
cond = M + W + (1 − )C
W = 0:200X + 0:0238 (W m−1 K −1 ) [33]
M = X (0:0040M ) (W m−1 K −1 ) [33]
C = 0:105 W m−1 K −1 [36]

Speci2c heat cW = 0:003867(T − 273:2) + 0:1031 (kJ kg−1 K −1 ) [37]


cC = 1:39 + 0:00036T (kJ kg−1 K −1 ) [38]
cpyr = 2:4 kJ kg−1 K −1 [1]
cM = 4:18 kJ kg−1 K −1 [32]

Molecular weight WV = 18 × 10−2 kg mol−1 [32]


WL = 26 × 10−2 kg mol−1 [1]
WV = 110 × 10−2 kg mol−1 [1]

Dynamic viscosity  = 4:847 × 10−7 T 0:64487 (kg m−1 s−1 ) [1,32]


K.M. Bryden et al. / Biomass and Bioenergy 22 (2002) 41–53 47

1.5
Southern Pine
1.0

0.5

0.0
20 30 40 50 60

Char Rate (mm min )


1.5

-1
Basswood
1.0

0.5

0.0
20 30 40 50 60
1.5
Red Oak
1.0
Fig. 1. Internal temperature history of 64 mm thick red oak exposed 0.5
on one surface to a 38 kW m−2 radiant heat Bux;—model results,
the shaded area is the range of experimental results from Ref. [27]. 0.0
20 30 40 50 60
-2
Heat Flux (kW m )

Fig. 2. Average charring rate as a function of the radiant heat Bux


heat Bux. The dominant direction of the heat Bux was for wood in air; —, model results; •, measurements of Ref. [27].
tangential to the annual rings. Temperatures were mea-
sured as a function of time at 6, 12, 18, 24, and 36 mm
from the surface. The char depth was assumed to cor- Fig. 2. The density, moisture content, and char shrink-

respond to a temperature of 300 C. When the tem- age coeHcients used in the model are those reported
perature at the thermocouple 36 mm from the surface by Tran and White [27] and are given in Table 4.

reached 300 C, the experiment was stopped and the The primary diAerences between these three woods
average charring rate determined. The radiant Bux re- are density and the char shrinkage. The less dense the
ported accounted for the increase in the temperature wood, the smaller the mass of water to vaporize and
of the heating environment and reradiation back to the the smaller the mass of wood to heat and pyrolyze,
sample due to the Bames. which results in faster charring rates. As the heat Bux
The measured and modeled internal temperatures rises, the char shrinkage decreases due to less time
as a function of time for test locations within a wood at temperature, resulting in increased char thicknesses
slab are shown in Fig. 1. The temperature at a location insulating the unreacted core from the Bux, raising
rises 2rst to the drying plateau and then rises smoothly the surface temperature, and increasing the convective
through the pyrolysis zone. When completely charred, heat loss. As a consequence, the rise in the charring
the temperature rise slows down and the temperature rate tends to Batten as the radiant heat Bux increases.
through the char layer tends toward a steady state tem- As shown, the match between the model and the exper-
perature pro2le between the hot surface and the cold imental data is good, even though generalized wood
unreacted core. As shown, the model matches well pyrolysis kinetics are used in the model.
with the drying plateau, the 2nal temperatures, and the
overall temperature pro2le. The temperature pro2le of
the model slightly leads the experimental data until 6. Modeling results
a depth of 36 mm where the model lags the experi-
mental data. It is suspected that this occurs because The numerical model was used to examine the
cracking of the solid matrix increases with time, thus impact of moisture content and particle size on the
increasing the conduction rate within the particle. This wood pyrolysis rate and regime. Generally, based on
eAect is not included in the model. transient conduction within the wood particle, wood
The average charring rate as a function of heat Bux pyrolysis has been divided into two separate regimes,
for red oak, basswood, and southern pine is shown in thermally thin and thermally thick. Within the
48 K.M. Bryden et al. / Biomass and Bioenergy 22 (2002) 41–53

Table 4
Wood properties used in the model

Species Density Ref. Shrinkage factor Ref. Shrinkage factor Ref.


(g cm−3 ) (drying) (pyrolysis)

Basswood 420 27 0.934 33 0:354 + 0:003Q [27]


Poplar 504 28 0.954 33 0.672 [28]
Red oak 660 27 0.956 33 0:446 + 0:005Q [27]
Southern pine 508 27 0.958 33 0:402 + 0:003Q [27]

thermally thin regime, pyrolysis is divided into kineti- impact the interpretation of the Biot number. In tran-
cally controlled pyrolysis and heat transfer controlled sient conduction the error associated with the lumped
pyrolysis based on the time scales of the thermally capacitance method (thermally thin assumption) is
thin combustion and pyrolysis kinetics [30]. Based on small if Bi ¡ 0:1 [31]. For the case of solid particle
the modeling results, it is suggested that in addition drying, Ragland and Borman [32] recommend that the
to thermally thin and thermally thick, a third regime particle be considered to be thermally thin if Bi ¡ 0:2.
be added—the thermal wave regime. Based on a review of the model output, the higher
The pyrolysis cases discussed here are for a 2xed limit of Bi ¡ 0:2 is appropriate. This results in maxi-
background temperature that provides a radiant Bux mum temperature diAerence across the particle during
that dominates the heat transfer to the particle. Oxygen pyrolysis of about 30 –35 K. In this case, the pyroly-
levels in the freestream are set to 0% with no gas phase sis rate obtained from using the average temperature
or external char reactions. The half-thicknesses exam- varies by ¡ 10% from the pyrolysis rate obtained con-
ined range from 50 m to 5 cm. Background tempera- sidering the temperature gradient. This is reasonable
tures range from 800 to 2000 K, and moisture contents based on the available kinetics and physical properties.
range from 0% to 30%. This covers a wide range of Since moisture increases the conductivity of wood, the
wood combustion interests, including 2res with rela- Biot number increases as the particle dries. Because
tively low temperatures; combustion of large particles, of this, the limiting Bi number for thermally thin py-
such as wood chunks or logs in a 2xed bed at high rolysis occurs after the wood is dried, and hence the
temperature; wood chip gasi2cation and combustion; wood particle size limit for thermally thin combustion
and suspension burning. is independent of the moisture content.
The Biot number, based on the conductivity of dry In thermally thin pyrolysis of a thin wood slab
wood and modi2ed to account for radiant heat transfer, exposed to relatively low temperature (e.g. 50 m
can be used to assess the relative rates of internal and half-thickness and 800 K), the resistance to heat trans-
external heat transfer. fer within the particle is small compared to the exter-
nal heat transfer. Drying and pyrolysis occur in series
Bi = hrad Lc =; (3) (Fig. 3). Moisture content and particle size do
not aAect the pyrolysis rate. At 2000 K, a 15 m
where
half-thickness (30 m thick) is the largest size for a
hrad = ’(Ts + T∞ )(Ts2 + T∞
2
): (4) wood slab that the thermally thin assumption is valid.
This size limit increases to 250 m half-thickness as
The surface temperature, Ts , is not known indepen- the temperature decreases to 800 K. For wood par-
dently from the model and varies during pyrolysis ticles that can be treated as cylinders the thermally
in most cases; therefore, it is assumed to be equal thin limit is a diameter of 60 m and 1 mm at 2000
to background temperature. This overestimates hrad ; and 800 K, respectively. For wood particles that can
to account for this, hrad is divided by a factor of 2. be treated as spheres the thermally thin limit is a
This approximation places hrad within about 25% of diameter of 90 m and 1:5 mm at 2000 and 800 K,
the actual value and therefore does not signi2cantly respectively.
K.M. Bryden et al. / Biomass and Bioenergy 22 (2002) 41–53 49

0.25 0.020
Total
Mass Flux (kg m s )

0.20
-2 -1

0.015

Mass Flux (kg m s )


-2 -1
Tar Total
0.15 Tar
0.010
Water Vapor
0.10
Light Hydrocarbons
+ Tar
0.005
0.05
Water

0.00 0.000
0.0 0.1 0.2 0.3 0.4 0 50 100 150
Time (s) Time (s)

Fig. 3. Mass Bux vs. time for poplar with a 50 m half-thickness Fig. 5. Mass Bux vs. time for poplar with a 0:5 cm half-thickness
and 30% moisture exposed to an 800 K radiant temperature Bux and 20% moisture exposed to a 1000 K radiant temperature.
in nitrogen.

0.07

Mass Flux (kg m s )


700 2 -1 0.06
Original distance from the centerline
600 0.05

500 0.5 cm 0.04


Temperature ( C)

Water vapor
o

400 0.4 cm 0.03


0.3 cm Light Hydrocarbons
Total
300 0.2 cm 0.02

200 0.1 cm 0.01


Tar
100 0.00
0 100 200 300 400
0 Time (s)
0 50 100 150 200
Time (s) Fig. 6. Mass Bux vs. time for poplar with a 2 cm half-thickness
and 30% moisture exposed to a 1500 K radiant temperature in
Fig. 4. Internal temperatures vs. time for poplar with a 0:5 cm nitrogen.
half-thickness and 20% moisture exposed to a 1000 K radiant
temperature.
With a relatively large particle and high background
temperatures (e.g. 2 cm half-thickness and 1500 K) a
In thermally thick pyrolysis (e.g. 0:5 cm half- thermal pyrolysis wave results (Fig. 6). The internal
thickness and 1000 K) the relative rates of internal rate of heat transfer is slow relative to the external
and external heat transfer are similar. Throughout the rate of heat transfer, and there is a large temperature
drying and pyrolysis process there is a temperature gradient across the particle. At a given location there
gradient across the particle and at a given location is a transient temperature plateau due to drying. How-
there is a transient temperature plateau due to drying ever, in contrast to the thermally thick mode, pyroly-
(Fig. 4). The drying and pyrolysis zones in the par- sis and drying occur simultaneously throughout most
ticle overlap only slightly (Fig. 5). As a result, the of the pyrolysis. A thermal wave, which dries and py-
moisture content has limited impact on the product rolizes the particle as it moves through the particle,
yield. In contrast, particle size does aAect the product is established and moves to the center of the particle.
yield by increasing the holdup time in the particle and The temperature within the pyrolized char region rises
the temperature of the tars and pyrolysis gases within quickly compared to the non-reacted moist wood core
the particle. because of the smaller heat capacity of the char. This
50 K.M. Bryden et al. / Biomass and Bioenergy 22 (2002) 41–53

diAerence in heat capacity, combined with the energy 1.0


required to dry the wood, thins the drying and pyroly-

Product Yield (mass fraction)


sis regions and brings them relatively close together. 0.8
1500 K 2000 K
No current guidelines are available for the divi-

Tars
0.6
sion between the thermally thick and thermal wave
800 K 1000 K
regimes. For the purposes of this study, the pyrolysis
0.4
wave regime was determined to be when at least 20% 2000 K
of the solid core is still undried and at least 20% of 0.2 1000 K
1500 K

the wood has been pyrolyzed. This de2nition ensures

Light HC
that drying and pyrolysis occur simultaneously for a 0.0 800 K
signi2cant portion of the combustion process. As the
0.03 0.1 1 10 30
char layer thickens, the heat transfer rate into the un-
Half-Thickness (mm)
reacted core slows down, and the drying and pyrolysis
rates gradually slow down (Fig. 6). After the particle Fig. 7. Pyrolysis product yields vs. thickness for poplar parti-
is dried, the temperature of the core rises quickly and cles exposed to various radiant temperatures and moisture levels
the small portion of the remaining wood is rapidly py- of—0% and, - - - 30%.
rolized. Product yield is a function of both moisture
content and particle thickness. Higher moisture con-
tent lowers the temperature of the pyrolysis region and carbons are inversely coupled, while the char yield re-
the temperature within the char, thus reducing the rate mains relatively constant. For small particles tar yield
of secondary reactions. This increases the tar yield and increases as a function of temperature from about 60%
decreases the light hydrocarbon yield. As particle size at 800 K to about 74% at 2000 K. Conversely, light
increases, the char layer thickens, providing more time hydrocarbon yield decreases as a function of tempera-
for secondary reactions and decreasing the tar yield. ture from about 10% at 800 K to about 5% at 2000 K.
This division between the thermally thick and the As particle size increases, the secondary reactions
pyrolysis wave regime occurs at Bi ≈ 10 based on reduce the tar yield and increase the light hydrocarbon
the conductivity of the dry wood. This limit is a weak yield. The impact of the secondary reactions increases
function of particle size and moisture content increas- with temperature. At 800 K the tar yield is reduced by
ing from Bi = 8 for 1 mm particles with 20% moisture about one-third from approximately 60% for 50 m
to Bi = 15 for 4 cm particles with 0.1% moisture. To particles to 40% for 3 cm particles, while the light
understand why this limit is a function of the Bi num- hydrocarbon yield is increased by about three times
ber, remember that the transient non-dimensionalized from approximately 10% for 50 m particles to 30%
temperature pro2le in a homogeneous, non-reacting for 3 cm particles. At 2000 K the tar yield is reduced
plane wall is a function of the Bi number [31]. As the by about 90% from about 74% for 50 m particles to
Biot number increases the temperature drop across the about 6% at 3 cm, and the light hydrocarbon yield is
particle increases. In moist wood the aAect is accen- increased by about 12 times from about 5% for 50 m
tuated by the energy required to dry the particle and particles to about 60% at 3 cm. At high temperatures,
the change in material properties as the wood is py- product yield is a weak function of moisture content.
rolyzed to char. The moisture slows the inward move- At higher moisture contents the temperature in the
ment of the thermal wave while the lower density of char layer is slightly reduced, reducing the rate of the
the char allows the temperature pro2les within the char secondary reactions.
to respond more quickly than the temperature pro2les Pyrolysis time as a function of particle size, tem-
within the unreacted wood. As a consequence, in large perature, and moisture content is summarized in
wood particles the pyrolysis and drying zones are ad- Fig. 8. As shown, the impact of moisture content on
jacent to each other and driven by a thermal wave. the total pyrolysis time is small compared to increases
The impact of particle size, temperature, and mois- in pyrolysis times due to changes in background
ture content on pyrolysis product yield is summarized temperature and particle thickness. As the parti-
in Fig. 7. As shown, the yields of tar and light hydro- cle becomes thinner at 800 K, the overall pyrolysis
K.M. Bryden et al. / Biomass and Bioenergy 22 (2002) 41–53 51

10
4
thermally thin, kinetically controlled combustion can
800 K
1000 K
be assumed within the region Bi ¡ 0:2 and Py ¿ 5.
Additionally, within a furnace or 2re environment the
3
10
1500 K
2
Py' = 5 2000 K assumption of Py ¿ 5 may be able to be relaxed at
10 higher temperatures. However, particle sizes smaller
Region 1
1
800 K
than 1 m are required to reach this regime at temper-
Time (s)

10
atures of 1500 K or higher. As shown, most pyrolysis
Region 4
10
0
1000 K within high temperature environments occurs in the
Region 2 thermally thick and pyrolysis wave regimes.
Fig. 8 provides insight to a number of practical
-1
10 Bi = 10
Region 3
-2
systems. For example, consider co-2ring very 2nely
Bi = 0.2 1500 K
10 chipped wood with coal. Assume Bat particles 0:2×2×
-3 10 mm in size. The half-thickness of the characteristic
10
0.01 0.1 1 10 100 length is 0:1 mm and typical background temperatures
Half-Thickness (mm) are between 1500 and 2000 K placing the pyrolysis
in the thermally thick regime. Moisture content of the
Fig. 8. Pyrolysis time vs. half-thickness for poplar particles exposed feed stock will not aAect the pyrolysis gas yield and
to various radiant temperatures and moisture levels of—0% and,
will have a limited impact on the pyrolysis time. Con-
- - - 30%. Region 1 is the kinetically limited, thermally thin
pyrolysis regime, Region 2 the heat transfer limited, thermally thin sider wood gasi2cation of poplar chips. Typical wood
pyrolysis regime, Region 3 the thermally thick pyrolysis regime, chip sizes are in the range of 0:3 × 1:5 × 3 cm, the
Region 4 the thermal wave pyrolysis regime. half-thickness of the characteristic length is 1:5 mm.
Using a background temperature of 1000 K, the py-
rolysis occurs in the thermally thick regime. Changes
time tends towards a constant time independent of in moisture content will impact pyrolysis time but not
moisture content or thickness. This is expected; in the product yield. Both stick wood pyrolysis and 2res
this region the particle is thermally thin and kineti- involve very thick particles (half-thickness ¿ 2 cm)
cally controlled. As temperature increases, this region and are in the thermal wave regime. In the case of 2res,
is reached at smaller and smaller particle sizes be- the thermal wave is so evident that pyrolysis rates are
cause the rate of pyrolysis increases exponentially as given in cm s−1 [33].
the temperature rises, requiring higher external heat
transfer rates to reach the kinetically limited region.
The relative rates of external heating and pyrolysis 7. Conclusions
can be assessed by considering the external pyrolysis
number, Py . The second or external pyrolysis number Results from a detailed computational model show
is that the pyrolysis rate of wood depends primarily
on radiant surface temperature and particle size, and
Py = hrad =(kpyr W cW d): (6)
secondarily on particle moisture. Predicted pyrol-
The thin dotted lines represent the pyrolysis time ysis times ranged from 10 ms for small particles
for an in2nitely thin, kinetically controlled particle. at 2000 K to 80 min for large particles at 800 K.
The diAerence between the thin dotted line and the Yields of tar and light hydrocarbons depend on par-
solid line at the same temperature can then be seen ticle size and temperature and are inversely coupled,
as the error associated with the in2nitely thin, ki- while the char yield remains relatively constant. Four
netically controlled assumption at various particle pyrolysis regimes are denoted: (1) thermally thin—
half-thicknesses. Pyle and Zaror [30] recommend a kinetically limited, Bi ¡ 0:2 and second Py ¿ 5;
limit of Py ¿ 10 for the thermally thin, kinetically (2) thermally thin—heat transfer limited, Bi ¡ 0:2
controlled regime. As shown, the error in pyrolysis and Py ¡ 5; (3) thermally thick, 0:2 ¡ Bi ¡ 10; and
times at Py = 5 is ¡ 25% at 800 K. At higher temper- (4) the thermal wave regime, Bi ¿ 10. Most practical
atures this error is signi2cantly smaller. Based on this, applications are in regimes 3 or 4.
52 K.M. Bryden et al. / Biomass and Bioenergy 22 (2002) 41–53

Acknowledgements kinetics, thermal and heat transfer eAects. The Canadian


Journal of Chemical Engineering 1991;69:907–15.
The authors wish to thank Robert White of the US [16] Nunn TR, Howard MR, Longwell JB, Peters WA. Product
compositions and kinetics in the rapid pyrolysis of sweet
Forest Service Forest Products Laboratory in Madison, gum hardwood. Industrial and Engineering Chemistry Process
WI, for sharing his detailed experimental results. Design and Development 1985;24:836–44.
[17] Sha2zadeh F, Chin P. Thermal deterioration of wood. In:
Goldstein IS, editor. Wood technology. Washington: ACS
References Press, 1977. p. 57– 61.
[18] Thurner F, Mann U. Kinetic investigation of wood pyrolysis.
[1] Bryden KM. Computational modeling of wood combustion. Industrial and Engineering Chemistry Process Design and
Ph.D. thesis, University of Wisconsin-Madison, 1998. Development 1981;20:482–8.
[2] Alves SS, Figueiredo JL. A model for pyrolysis of wet wood. [19] Liden AG, Berruti F, Scott DS. A kinetic model for the
Chemical Engineering Science 1989;44:2861–9. production of liquids from the Bash pyrolysis of biomass.
[3] Di Blasi C. Analysis of convection and secondary reaction Chemical Engineering Communications 1988;65:207–21.
eAects within porous solid fuels undergoing pyrolysis. [20] Boroson ML, Howard JB, Longwell JP, Peters AW. Product
Combustion Science and Technology 1993;90:315–40. yields and kinetics from the vapor phase cracking of wood
[4] Di Blasi C. Processes of Bames spreading over the surface pyrolysis tars. AIChE Journal 1989;35:120–8.
of charring fuels: eAects of the solid thickness. Combustion [21] Kosstrin HM. Direct formation of pyrolysis oil from
and Flame 1994;97:225–39. biomass. In: Proceedings of the Specialists Workshop on
[5] Di Blasi C. Predictions of wind-opposed Bame spread Fast Pyrolysis of Biomass. CO: Copper Mountain, 1980.
rates and energy feedback analysis for charring solids p. 105 –21.
in a microgravity environment. Combustion and Flame [22] Diebold JP. The cracking kinetics of depolymerized biomass
1995;100:332–40. vapors in a continuous, tubular reactor. M.S. thesis, Colorado
[6] Di Blasi C. Heat, momentum, and mass transport through Schools of Mines, 1985.
a shrinking biomass particle exposed to thermal radiation. [23] Antal MJ. Vapor phase pyrolysis of biomass derived
Chemical Engineering Science 1996;51:1121–32. volatile matter. In: Overend RP, Milne TA, Mudge LK,
[7] Melaaen MC, GrHnli MG. Modeling and simulation of moist editors. Fundamentals of thermochemical biomass conversion.
wood drying and pyrolysis. In: Bridgwater AV, Boocock London: Elsevier, 1982. p. 511–37.
DBG, editors. Developments in thermochemical biomass [24] Lede J, Diebold JP, Peacocke GVC, Piskorz J. The
conversion. London: Blackie, 1997. p. 132–46. nature and properties of intermediate and unvaporized
[8] Ragland KW, Boerger JC, Baker AJ. A model of chunkwood biomass pyrolysis materials. In: Bridgwater AV, Boocock
combustion. Forest Products Journal 1988;38(2):27–32. DBG, editors. Developments in thermochemical biomass
[9] Saastamoinen JJ. Model for drying and pyrolysis in an conversion. London: Blackie, 1997. p. 27–42.
updraft gasi2er. In: Bridgwater AV, editor. Advances in
[25] Ouelhazi N, Arnaud G, Fohr JP. A two-dimensional study
thermochemical biomass conversion. London: Blackie, 1993.
of wood plank drying. The eAect of gaseous pressure below
p. 186–200.
the boiling point. Transport in Porous Media 1992;7:39–61.
[10] Simmons WW. Analysis of single particle wood
[26] Bamford CH, Crank J, Malan DH. The combustion of wood.
combustion in convective Bow. Ph.D. thesis, University of
Part I. Proceedings of the Cambridge Philosophical Society
Wisconsin-Madison, 1983.
1946;42:166–82.
[11] Chan WR, Kelbon M, Krieger BB. Modeling and
experimental veri2cation of physical and chemical processes [27] Tran HC, White RH. Burning rate of solid wood measured in a
during pyrolysis of a large biomass particle. Fuel heat release rate calorimeter. Fire and Materials 1992;16:197–
1985;64:1505–13. 206.
[12] Antal MJ. Cellulose pyrolysis kinetics: the current state of [28] White RH. Charring rates of diAerent wood species. Ph.D.
knowledge. Industrial and Engineering Chemistry Research thesis, University of Wisconsin-Madison, 1988.
1995;34:703–17. [29] Kansa EJ, Perlee HE, Chaiken RF. Mathematical model
[13] Di Blasi C. Modeling and simulation of combustion processes of wood pyrolysis including internal forced convection.
of charring and non-charring fuels. Progress in Energy and Combustion and Flame 1997;29:311–24.
Combustion Science 1993;19:71–104. [30] Pyle DL, Zaror CA. Models for the low temperature pyrolysis
[14] Font T, Marcilla A, VerdVu E, Devesa J. Kinetics of wood particles. In: Bridgwater AV, editor. Thermochemical
of the pyrolysis of almond shells and almond shells processing of biomass. London: Butterworths, 1984.
impregnated with CoCl2 in a Buidized bed reactor and p. 201–16.
in a pyroprobe 100. Industrial and Engineering Chemistry [31] Incropera FP, Dewitt DP. Fundamentals of heat and mass
Research 1990;29:1846–55. transfer, 4th ed. New York: Wiley, 1996.
[15] Koufopanos CA, Papayannakos N, Mashio G, Lucchesi A. [32] Borman GL, Ragland KW. Combustion engineering.
Modeling the pyrolysis of biomass particles: studies on New York: McGraw-Hill, 1998.
K.M. Bryden et al. / Biomass and Bioenergy 22 (2002) 41–53 53

[33] Forest Products Laboratory, Wood handbook: wood as an (International) on Combustion. Pittsburgh: The Combustion
engineering material, agricultural handbook 72. Washington: Institute, 1976. p. 1459 –70.
US Department of Agriculture, 1987. [37] TenWolde A, McNatt JD, Krahn L. Thermal properties
[34] Siau JF. Transport processes in wood. New York: Springer, of wood and wood panel products for use in buildings.
1984. DOE=USDA-21697=1, Oak Ridge National Laboratory, Oak
[35] Panton RL, Rittman JG. Pyrolysis of a slab of Ridge, 1988.
porous material. In: 13th Symposium (International) on [38] Stull DR. JANAF thermochemical tables. NSRDS-NBS 37.
Combustion. Pittsburgh: The Combustion Institute, 1971. US Government Printing OHce, Washington, 1971.
p. 881–91.
[36] Lee CK, Chaiken RF, Singer JM. Charring pyrolysis of
wood in 2res by laser simulation. In: 16th Symposium

S-ar putea să vă placă și