This article was downloaded by: [University of Illinois at Urbana-Champaign]

On: 29 October 2013, At: 15:26

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Chemical Engineering Communications

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/gcec20

A KINETIC MODEL FOR THE PRODUCTION OF LIQUIDS

FROM THE FLASH PYROLYSIS OF BIOMASS
a

A.G. LIDEN , F. BERRUTI & D.S. SCOTT

Department of Chemical Engineering , University of Minnesota# , 19, 151 mundson Hall 421
Washington Ave, S.E. Minneapolis, Minn, 55455, U.S.A.
b

Department of Chemical and Petroleum Engineering , The University of Calgary , Calgary,

Alberta, T2N 1N4, Canada.
c

Department of Chemical Engineering , University of Waterloo , Waterloo, Ontario, N2L 3G1,

Canada.
Published online: 27 Aug 2010.

To cite this article: A.G. LIDEN , F. BERRUTI & D.S. SCOTT (1988) A KINETIC MODEL FOR THE PRODUCTION OF LIQUIDS FROM
THE FLASH PYROLYSIS OF BIOMASS, Chemical Engineering Communications, 65:1, 207-221, DOI: 10.1080/00986448808940254
To link to this article: http://dx.doi.org/10.1080/00986448808940254

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions

Chem. Eng. Comm. 1988, Vol. 65, pp. 207-221

Reprints available directly from the publisher.
Photocopying permitted by license only.
1988 Gordon and Breach Science Publishers S.A.
Printed in the United States of America

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

A KINETIC MODEL FOR THE PRODUCTION OF

LIQUIDS FROM THE FLASH PYROLYSIS OF
BIOMASS
A.G. LIDEN,! F. BERRUTI,zt and D.S. SCOTT]
1 Department of Chemical Engineering, University of Minnesota, #19, 151
Amundson Hall, 421 Washington Ave. S.E. Minneapolis, Minn., 55455 U.S.A.
(612) 625-9013. 2 Department of Chemical and Petroleum Engineering, The
University of Calgary, Calgary, Alberta, T2N 1N4, Canada (403) 220-6529.
3 Department of Chemical Engineering, University of Waterloo, Waterloo,
Ontario, N2L 3G1, Canada (519) 885-1211.

(Received Seplember 10, t987; in final form October 19, 1987)

A kinetically based prediction model for the production of organic liquids from the flash pyrolysis of
biomass is proposed. Wood or other biomass is assumed to be decomposed according to two parallel
reactions yielding liquid tar and (gas + char). The tar is then assumed to further react by secondary
homogeneous reactions to form mainly gas as a product.

The model provides a very good agreement with the experimental results obtained using a pilot
plant fluidized bed pyrolysis reactor.
The proposed model is shown to be able to predict the organic liquid yield as a function of the
operating parameters of the process, within the optimal conditions for maximizing the tar yields, and
the reaction rate constants compare reasonably well with those reported in the literature.
KEYWORDS
Flash biomass pyrolysis
Tar
Reaction model,
Fluidized bed reactor

INTRODUCfION
The thermal conversion of biomass to organic liquid products can be accomplished using several techniques. One possible method is biomass flash pyrolysis.
Biomass flash pyrolysis is a rapid heating-rate process occurring at moderate
temperatures (typically from 400C to 600C) in which a high yield of liquid
tar/oil product is obtained (Scott and Piskorz, 1982, 1984). The process requires
vapor residence times less than 1 second and solids residence times slightly higher
but of the same order of magnitude (Scott and Piskorz, 1982).
A fluidized bed reactor has been used successfully in the flash pyrolysis process
to produce a liquid oil from a wide variety of biomass feedstocks, with organic
liquid yields as high as 60-70% by weight (on a moisture free basis) of the
material fed, using hardwoods such as aspen-poplar and maple (Scott and
Piskorz, 1982, 1984, 1985).

t To whom correspondence should be addressed.

207

208

A.G. LIDEN, F. BERRUTI AND D.S. SCOTT

GAS

BIOMASS

TAR (DILl

<

GAS
CHAR

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

CHAR
prtmory
reactLons

FIGURE I

secondary

reoctl.ons

Overall pyrolysis reaction mechanism proposed by Shafizadeh (1975).

The rapid thermal decomposition of biomass by flash pyrolysis occurs by a

sequence of very complex reactions, so that the development of a complete
reaction model can become a difficult task. Consequently, a number of
simplifications are needed if such a reaction pattern has to be modelled, even
approximately.
A simple general approach to a kinetic model, proposed by many workers, was
to divide the products of biomass pyrolysis into three major groups: gas, tar (or
oil) and char. For example, Shafizadeh (1975) proposed an overall reaction
mechanism of this sort, as schematically represented in Figure 1.
The kinetic investigations of wood pyrolysis reported in the literature can be
divided into three main groups:
1. Studies to determine the overall decomposition rate of the biomass, as
measured by rates of weight loss.
2. Studies to determine the rate constants for the formation of different pyrolysis
products such as tar, char and gas, without considering any secondary reaction
which may occur.
3. Investigations to determine rate constants for assumed primary pyrolysis
reactions as well as secondary reactions of the volatile products.

A number of investigations of biomass pyrolysis reactions have been described

in the literature, (summarized by Antal (1983) in a very extensive review), which
used thermo gravimetric analysis (TGA) to determine the overall biomass
decomposition rate (Broido (1976), Fairbridge et ai. (1978), Kala et ai. (1979),
Tran and Charanjit (1979), Shafizadeh and Bradbury (1979), Jain et ai. (1980),
Antal et ai. (1980. Other investigations employed techniques other than TGA,
such as flash radiant heating (Shivadev and Emmons (1974, a fluidized bed
reactor (Barooah and Long (1976, a tube furnace (Min (1977), and electrical
heating of filter paper strips (Lewellen et ai. (1977.
Interest in specific types of products of pyrolysis has led to proposals for more
complex models. Bradbury et ai. (1979) proposed a reaction model taking into
account the formation of two groups of products: volatiles and (char + gas).
Thurner and Mann (1981) studied the pyrolysis of oak sawdust in a tube furnace

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

A KINEMATIC MODEL FOR FLASH PYROLYSIS

209

and divided the products into three groups: gas, tar (oil) and char. They
calculated rate constants and concluded that the reaction was chemically
controlled in their method for biomass particles smaller than 2 mm and
temperatures between 325 and 385C.
A recent and extensive study was conducted by Hajaligol et al. (1982). Kinetic
parameter values were reported from their investigation of cellulose pyrolysis for
a large number of low molecular weight compounds which were assumed to be
ultimate decomposition products.
Kosstrin (1980) reported kinetic values for both primary and secondary
pyrolysis reactions for a number of different biomass materials. The secondary
reactions of the tar were assumed to be first order surface reactions.
In general, from a survey of the published works, it can be seen that the
reported kinetic parameter values show a considerable scatter, possibly because a
wide variety of different techniques and operating conditions have been used by
the various researchers.
Important factors influencing the nature of the kinetics of the pyrolysis
reactions include the residence time of the volatiles in the solid biomass matrix,
the residence time of the volatiles in the reactor, heating rate of the biomass and
the temperature range of the operation. Interpretation of kinetic parameters for
different pyrolysis conditions is not completely satisfactory as yet and requires a
better knowledge of the actual reaction mechanisms. Examples of kinetic studies
reported in the literature are summarized in Table I.

TABLE I
Some kinetic studies reported in the literature

Author
Biomass

Method
Temperature
Heating rate
Kinetic eqn.

Parameters

Lewellen et at.

Thurner and Mann

( 1977)

(1981)

Cellulose

Oak sawdust

Electric heater
250-1000"C
400-1O,000"C/S
-dW/dt=kW
k = k"exp[-E,,/RT]

Tube furnace
325-385"C

k" = 6.79 109 (S-I)

Eo = 139.8 kJ/mole

-dW/dt=k,W

Kosstrin (1980)
Sawdust, paper, rice straw,

pine bark
Fluidized bed reactor
430-960"C
"Fast"
primary step: dW /dt = k(W - W)
secondary step: first order

surface reaction
= 1 = char
Sawdust: W = 0.3
k Ol = 7.377 lOs (S-I) primary: k o = 1013 (S-I)
E OI = 106.5 kJ/mole
E" = 183.3 kJ/mole
t=2=tar
secondary: k o = 3.8 m/s
6
k 02=4.125 10 (S-I)
Eo = 72.80 kJ/mole'
E02 = 112.7 kJ/mole
Paper: W = 0.604
t = 3 = gas
primary: k" = 8. 9 . 1012 (S-I)
E" = 183.3 kJ/mole
k OJ = 1.435 . 10' (S-I)
E 0 3 = 88.6 kJ/mole
secondary: k o = 5.0 m/s
E" = 65.7 kJ/mole
t

210

A.G. LIDEN, F. BERRUTI AND D.S. SCOTT

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

KINETIC MODELLING
The pilot plant unit used for obtaining kinetic data is schematically shown in
Figure 2. A fluidized bed reactor was employed for carrying out the pyrolysis
reaction. The biomass was fed into the fluid bed containing sand with a twin
screw feeder. The fluidizing gas consisted of recycled gaseous products after they
passed through a preheating chamber to reach a temperature of about 650C. The
operating conditions for the reactor were set so that gases, vapors and char
products left the reactor from the top, while the sand was retained into the bed.
The char product was separated in a cylclone and the condensable liquids in two
condensers in series. A portion of the product gases was then recycled and used
as fluidizing gas and the remaining part collected for analyses and mass balances.
A detailed description of the equipment used has been reported elsewhere (Scott
and Piskorz (1984.
Typical experimental results in terms of product yields versus temperature
which were obtained from the fluidized bed pilot plant pyrolysis reactor (Scott
and Piskorz (1982, 1984, 1985)) are shown in Figure 3. It is evident from the
general trend of the experimental results, particularly for the yields of liquids,
that a kinetic model able to predict such product yields must consider both
primary and secondary reaction steps if the observed maximum is to be described
by the model.
As discussed earlier, Kosstrin (1980) appears to be the only investigator who
reported in the literature kinetic parameters for both primary and secondary
pyrolysis reaction steps.
Very recently Diebold (1985) presented kinetic parameters for the secondary
vapor cracking reaction using a model slightly more complex than the one used by
Kosstrin. In Diebold's model, the primary vapor formed could react either to

FLUID 8EO

REACTOR
FIRST
CONDENSER

SECOND
CONDENSER

FillER

FURNACE
TAR PRODUCTS

LIGHT TAR

ANl\lYSIS

FLOW CONTROLLER

PRESSURE REGUL'TOR

COMPRESSOR

FIGURE 2 Schematic representation of the Waterloo Flash Pyrolysis Pilot Plant.

A KINEMATIC MODEL FOR FLASH PYROLYSIS

211

0
0

"'o

TAR loLLl
CHAR

.J

",0

.
.nlil

..:

GAS

Ec.

-ci

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

co

'"
(flo

-'~

w
>-<

'-'0

:::oi<l
0
0

""
CL

0
0

iOO

i50

550

500

600

650

T ( Q Cl
FIGURE 3 Experimental results in terms of product yields versus temperature obtained from the
Waterloo fluidized bed pilot plant using hardwood as biomass feed. A gas residence time of 0.5
seconds has been used for all runs (Scott and Piskorz , 1982. 1984, 1985).

yield gas or to form "secondary tar", described as an aromatic tar containing

residual oxygen.
The model presented in this paper is built on the same approach as Kosstrin's
model. However, the secondary reactions are not assumed in this work to be
surface reactions, but rather homogeneous gas reactions. The model, which is
based on the Waterloo flash pyrolysis process carried out in the fluidized bed
operating in the bubbling mode, includes a number of assumptions.
1. The biomass (wood) is assumed to be decomposed according to two parallel
reactions, one yielding tar (oil), and the other (char + gas). The tar is then
assumed to undergo further secondary reactions giving mainly gas as a product,
an assumption supported by the experimental results shown in Figure 3. This
simplified reaction mechanism which is assumed as the basis of the present model
is described by the scheme shown in Figure 4. The reaction kinetics represented
by the kinetic model illustrated in Figure 4 can be expressed according to the
following equations, assuming first order kinetics for all processes,
rw =kC w

(1)

rl=k1CW

(2)

rz = kZC T

(3)

and
where k = k,
the reactor.

+ k 3 , as illustrated in Figure 4 and

C w is the wood concentration in

212

A.G. LIDEN, F. BERRUTI AND D.S. SCOTT

PROPOSED REACTION MECHANISM

TAR

GAS

(T)

"

(G)

\1000

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

( \I )

(GAS

CHAR)

(G)

FIGURE 4

(C)

Proposed reaction mechanism. basis of the present model.

2. Modelling of a fluidized bed reactor operation is not a straightforward task

and several approaches can be taken. In the present work, the gas and particle
flows are treated separately. For the gas flow, the following assumptions are
made;
-perfect plug flow of the gas phase,
-isothermal conditions,
-steady-state operation (i.e. no gas accumulation in the bed),
-negligible increase in gas flow due to products generation.
Using the above assumptions, a differential material balance for the tar can be
written as

o ac;

- - - d = -k,Cw

+ k 2C r

(4)

Being Q not a function of the axial coordinate z, the differential time (dt) can be
defined as
dt =A dz l Q

(5)

which, substituted into Eq. (4), yields

sc;

---;jt = k , Cw

- k 2Cr

(6)

Equation (6) can be integrated for the time interval t = 0 to t = G, with the
boundary condition

Cr(O) = 0

(7)

to give the tar concentration at the reactor outlet

- k C [1- exp( -k 2 T a)]
Cr ( Ta ) - I '"
k

(8)

To find the wood concentration in the bed (C w ) , the following assumptions are

A KINEMATIC MODEL FOR FLASH PYROLYSIS

213

made:

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

-steady-state operation (experimental results confirmed no wood accumulation

in the reactor),
-isothermal conditions (the heat up time has been calculated and shown to be
negligible when compared to the reaction time),
-perfect mixing within the reactor, but all particles having approximately the
same residence time Ts (PFR output).
The last assumption is based upon experimental results, showing very narrow exit
age distribution curves E(t), reported elsewhere by Berruti et at. (1986, 1987).
The wood material balance can be written as:
INPUT-OUTPUT-REACTION = 0

(9)

If a time interval ilt is considered, the wood input is defined as Sm. The reaction

term is:
REACTION = kCwV dt

(10)

and a general expression for the output is:

OUTPUT =

Sm exp( -kt)E(t) dt

(11)

For a PFR output, the exit age distribution curve E(t) is:

= c5(t -

E(t)

(12)

Ts )

and therefore Eq. (11) becomes:

OUTPUT = Sm exp( -kTs)

(13)

By substitution of the input, output and reaction terms into Eq. (9), the following
expression for the wood concentration is obtained:
C _ m[1 - exp( -kTs )]
wVk

(14)

where m = Sm] t!.t = wood feed rate. Equation (14) can then be substituted into
Eq. (8) to give the tar concentration at the bed outlet:
- k}m[1 - exp( -kTs)] [1- exp( -kzTc)]
CT ( Tc ) .
kV
kz

(15)

The maximum theoretical tar yield, or ultimate tar yield, can be defined as

<p* = k-f k:

(16)

which substituted into Eq. (15) yields

CT(Tc)

= <p* m[l- exp( -kTs)] [1- exp( -kzTc)]

V

kz

(17)

214

A.G. LIDEN, F. BERRUTI AND D.S. SCOTT

During the time T:G' rilT:G weight units of biomass have been fed to the bed. The
tar yield can then be expressed by division of Eq. (17) with rilT:GI V. Thus, the tar
yield, denoted as cp, can be expressed by the following equation:

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

cp = CP*[1 -

exp( -kr:s )] [1- exp( -kzT:G)]

kzT:G

(18)

If the value of kr:s is larger than 7, the second factor on the right hand side of Eq.
(18) will be larger than 0.999. Further, if kr:s is at least 10 times larger than kzT:G'
Eq. (18) can be approximated by

cp =

[1- exp( -kzT:G)]

kzT:G

(19)

The residence time of the biomass particles in the fluidized bed reactor is of the
order of a few seconds (Berruti et ai. (1986)) and the overall decomposition
constant (k) has been shown to have at 500C a value of about 4 S-1 (Kosstrin
(1980)).
The gas residence time used in flash pyrolysis is of the order of 0.5 seconds, and
k z has been reported by Diebold (1985) with values of the order of 0.2 S-I. Thus,
the use of the simple form of Eq. (19) and 500C appears to be justifiable. The
basic assumption behind Eq. (19) is that primary tar-forming reactions occur
much faster than secondary reactions and thus the tar yield is largely determined
at short reaction times by the extent of the decomposition step.
If the kinetic coefficients k and k z are assumed to follow an Arrhenius law, then
the reaction model described by Eq. (19) consists of three parameters: a
preexponential factor, an activation energy and a theoretical maximum tar yield,
as described by the ratio CP* of the primary tar-forming decomposition step to the
char-forming step.
Experimental data, shown in Table II, from the flash pyrolysis pilot plant using
Brockville whole-tree poplar wood, and clean lEA aspen-poplar wood, have
been employed for the purpose of finding values of the three model kinetic
parameters characteristic of the pyrolysis behaviour of both these very similar
poplar woods.
The reactor was assumed to contain a homogeneous mixture of sand and wood
throughout the entire volume. This assumption is based on the very well known
good mixing feature of the fluidized bed reactor. A nonlinear iterative program in
SAS (Statistical Analysis System) was used for the parameter fitting. The results
obtained by application of Eq. (19) using all the available data in the temperature
range 500C to 650C were the following:

CP* = 0.703
k zo = 4.28 . 106 S-I
z = 107.5 kl/mole

(20)

It should be pointed out that the experimental information was not sufficient to
confirm a unique validity for the model, nor to give highly accurate values of the
model parameters. The obtained parameters should then be seen as preliminary

(um)

recovery

Overall

Char
Gas

95.2

55.82
5.30
25.46
8.57

Tar (oil)

Water

1.68

0.69

6.3

(kg/hr)

Feed rate

Gas Res.
Time (s)

content %

Moisture

595

450

Temp. C

Part. size

33

Run no.

95.1

65.78
4.30
13.21
11.85

1.34

0.60

6.3

-595

500

31

96.5

61.96
5.30
10.64
18.56

1.46

0.56

6.3

-595

550

32

99.\

61.97
6.61
18.29
9.25

2.14

0.75

6.3

-595

490

36

98.4

61.7
10.4
13.04
13.18

1.22

0.67

4.5

-1100

509

37

94.2

55.97
10.5
J1.43
16.34

2.22

0.53

4.5

-1100

550

38

3.61

0.55

5.2

-1100

466

48

1.71

0.58

5.2

-1100

450

49

99.4

50.43
11.8
26.96
10.20
99.8

58.15
9.75
22.47
9.41

100.4

56.44
10.6
23.56
9.82

Yields (m.L basis) (%)

3.93

0.68

5.2

-1100

460

47

97.2

58.99
10.2
10.63
17.39

4.12

0.57

5.2

-1000

555

50

99.7

62.94
10.3
14.41
12.03

2.24

0.64

5.2

-1000

497

51

99.1

60.17
8.78
20.01
10.10

1.97

0.51

5.2

-1000

455

52

Some experimental results (pilot plant) using Brockville poplar (whole tree)

TABLE"

100.1

9.61
29.34

11.2

49.93

2.02

0.48

5.2

-1000

588

53

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

98.4

58.45
12.8
15.33
11.75

1.47

0.49

5.2

-1000

497

54

100.5

62.89
9.66
16.46
11.53

2.10

0.47

5.2

-1000

504

58

100.3

57.6
13.2
10.6
18.8

0.85

0.69

5.2

-1000

551

60

A.G. LIDEN, F. BERRUTI AND D.S. SCOTT

216
0

g
TAR YIELD DATA PDINTs (Scott ond

I/)
.j

PLskorz, 1982, 1981, 1985)

1/)0

o .
.ali!

---- MDDEL PREDICTION

.:

- iii
Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

Eo

'"
U10

--'~

,..

..... 0

'-'

5~
a

"" d

Q..

100

150

500

550

600

650

T( DC)

FIGURE 5 Model prediction (solid line) compared to tar yield experimental results, for a gas
residence time of 0.5 seconds.

approximations in a very promising model. However, for prediction purposes

over the range of the data available, the model represented by Eq. (19) appears
to give a good fit. Furthermore, yields calculated by extrapolation of the model
indicate a very reasonable trend, as shown in Figure 5.
At temperatures below the optimum temperature of 500C, the tar yield will be
determined by the primary reaction step, that is the formation of the tar fraction
from the wood being pyrolyzed. In these conditions, it can be assumed that the
secondary reactions will not play any important role. As the residence time of the
wood particles is not precisely known for the fluidized bed pyrolysis reactor used,
it is not possible to accurately estimate the primary reaction parameters from the
available experimental data. However, Kosstrin (1980) reported primary decomposition parameters which appear to be appropriate also for the operating
conditions of the flash pyrolysis process used in this work (Table I).
Using the kinetic parameter values of Kosstrin (1980), an estimate of the
residence time of the biomass particles in the fluidized bed reactor can be made.
From Eq. (18), an expression for the solids residence time can be explicitly
obtained by simple algebraic rearrangement:

(21)
From the available data for runs at temperatures below the optimal pyrolysis
temperature for high liquid yields of about 500C (Table II), estimates of ts can

A KINEMATIC MODEL FOR FLASH PYROLYSIS

217

TABLE III

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

Estimates of solids residence times from Eq. (21)

Run no.

T(C)

k (5-')

k 2 (S-I)

Tar yicld

Solids res. time (s)

30
33
49
52

425
450
450
455

0.194
0.578
0.578
0.660

0.0387
0.0735
0.0735
0.0831

0.559
0.558
0.564
0.6017

8.4
2.9
3.0
3.2

be calculated, as reported in Table III. The results reported in Table III show
that the residence time for the biomass particles of Run 30 is higher than the
residence time for the other runs. A more detailed study of the residence time
distribution of low density particles in a fluidized bed reactor which has been
reported elsewhere (Berruti et al. (1986 shows that the wood particles are not
entrained from the bed until their densities are below a certain limiting value.
Thus, at lower temperatures, a longer residence time would be expected because
of the lower decomposition rate of the wood. However, it is reasonable to assume
a residence time of about 3 seconds for temperatures around 450C. For lower
temperatures, the residence time will be longer and the model represented by Eq.
(18) will thus underestimate the wood conversion. At temperatures below 450C,
char formation becomes a more important route for wood pyrolysis, and the
kinetic model assumed may lose some of its validity. Thus, until further data is
available, only limited extrapolation of the model represented by Eq. (18) is
recommended outside the temperature region of 450C to 600C and a range of
gas residence times of 0.4 to 0.7 seconds.
o

r-----------......---..,....---..---.....,

=g~

~'3 ~ ~
We

,...

-------------------L--

-~

a:
'"
oL-_ _---'
..... 0

0.2

-'-

o.s

GAS RESIDENCE TIME

0.6

_ _--l

0.8

(s)

FIGURE 6 Influence of the gas residence time on the tar yicld at 500 and 6OQC, according to the
model prediction.

218

A.G. LIDEN, F. BERRUTI AND D.S. SCOTI

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

Model predictions for some different operating conditions are shown in Figures
5 to 6. The graphs are slightly extended outside the recommended applicable
region to show that the trend is quite reasonable. Experimental results obtained
from the pilot plant pyrolysis unit are also shown in Figure 5, where the tar yield
is plotted versus temperature at a gas residence time of 0.5 seconds. The tar yield
dependence on gas residence time is shown in Figure 6 at 500 and 600C. At
500C, the tar yield is fairly independent of gas residence time, whereas at 600C
the tar yield clearly decreases with increasing gas residence time.

DISCUSSION
The model presented in this work appears to be a useful kinetically based
prediction model for the tar yield from fluidized bed flash pyrolysis of wood
particles.
It is of interest to compare the kinetic parameters determined in the present
work with other reported parameters for secondary cracking reactions. Kosstrin
(1980) expressed the secondary kinetic parameters on a surface basis. In order to
compare his results to those found in this work, the rate constants obtained by
Kosstrin must be converted to a volume basis. Assuming that the particle size
used by Kosstrin was about 700 jlm, the secondary' reaction parameters reported
for sawdust become:
k zo = 3.26.104 S-I

(22)

= 72.80 kJ/mole

Diebold (1985) reported results of secondary cracking of wood pyrolysis

volatiles for experiments carried out at temperatures above 600C. In order to
obtain a comparison of kinetic parameters an extrapolation of these results is
required. A comparison of rate constants for the secondary tar decomposition
reactions obtained in this work to those estimated from the results of Kosstrin
(1980) and Diebold (1985) is shown in Table IV. If the differences in feedstock
and experimental procedure are taken into account, the agreement is surprisingly
good, thus emphasizing that the derived kinetic model can be expected to yield
very reasonable results.
TABLE IV
Comparison of secondary rate constants

Kosstrin (1980)

500
600
700
(') Extrapolations.

0.40
\.44

4.03

k 2 (S-I)
Diebold (1985)

This work

0.19(')
0.89
3.07

0.23
1.58
7.23 (')

A KINEMATIC MODEL FOR FLASH PYROLYSIS

219

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

CONCLUSIONS AND RECOMMENDATIONS

A kinetically based prediction model has been presented which is shown to be
able to predict the tar yield from the fluidized bed flash pyrolysis of hardwood,
within the temperature range of 450C to 600C, and for gas residence times of
0.4 to 0.7 s.
Wood or other biomass is assumed to undergo decomposition according to two
parallel reactions yielding tar (oil) and (char + gas). The tar is then assumed to
further react by secondary homogeneous reactions to form mainly gas as a
product.
The tar yield can be expressed by the general relationship:

s)] ext

ljJ = ljJ*[I- exp( -kr

[1 -

-k 2 't'c) ]

(18)

20

where

ljJ*
k

= 0.703

= 10'3 exp(-183.3

103/(RT (S-I)

k z = 4.28 106 exp( -107.5 103/(RT(s-l)

Care should be taken in using the various parameters given for modelling
pyrolysis reactions occurring at conditions different than those used for the
derivation of the model itself since model parameters cannot be correlated as yet
to the physical or chemical properties of the biomass used. However, as shown by
Liden (1985), the model should be valid for most biomass particles under 2 mm in
size, when pyrolyzed in a fluidized bed reactor at temperatures of 450C or
higher.
The model provides very reasonable predictions within the range of operating
conditions studied, and the secondary reaction rate constant (k z) compares very
well with other secondary reaction constants reported in the literature.

ACKNOWLEDGEMENTS
The research reported was supported through a Natural Sciences and Engineering
Council of Canada grant. The authors are grateful to Mr. Jan Piskorz for having
provided the experimental flash pyrolysis data and to Mr. Peter Majerski for his
technical help.

NOMENCLATURE

A
CT
Cw

reactor cross sectional area (rrr')

tar concentration (kg/rn")
wood concentration (kg/nr')

A.G. LIDEN, F. BERRUTI AND D.S. SCOTT

220
Eo

E2
E(I) .

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

kl

k2
k 20
k3

m
Q
r

T
V

activation energy for primary decomposition step (kJ/mole)

activation energy for secondary reaction step (kJ /mole)
exit age distribution curve
overall wood decomposition rate constant (S-I)
tar formation rate constant (S-I)
tar decomposition rate constant (S-I)
preexponential factor for the decomposition step (s-')
(gas + char) primary formation rate constant (S-I)
feed rate (kg/s)
volumetric gas flow rate (m 3/s)
rate of reaction (kg/m? . s)
time (s)
temperature
total reactor volume (rrr')
axial reactor coordinate (m)

eC)

Greek Leiters
<5

rp*
rp
1:0
1:s

dirac function
ultimate tar yield (maximum theoretical tar yield)
observed tar yield
gas residence time (s)
solids residence time (s)

Subscripts

C
G
S
T
W
1
2

char
gas
solid
tar
wood
primary reaction step
secondary reaction step

REFERENCES
Antal, M.J. Jr., Friedman, H.L., and Rogers, F.E., Comb. Svi. and Tech., 21, 141-152 (1980).
Antal, M.J. Jr., Adv. in Solar Energy, K.W. Boer and J.A. Duffie Eds., American Solar Energy
Society, N.Y., 61-111, (1981).
Barooah, J.N., and Long, V.D., Fuel, 55, 116-120, (1976).

Downloaded by [University of Illinois at Urbana-Champaign] at 15:26 29 October 2013

A KINEMATIC MODEL FOR FLASH PYROLYSIS

221

Berruti, F., Liden, A.G., and Scott, D.S., Proc. 36th Can. Chern. Eng. ConL, Sarnia, Ont., (1986).
Berruti, F., Liden, A.G., and Scott, D.S., Chem. Eng. Sci. (Accepted for publication, July 13, 1987),
(1987).
Bradbury, A.G.W., Sakai, Y., and Shafizadeh, F., J. Appl. Polymer Sci., 23,3271-3280, (1979).
Broido, A., in "Thermal Uses and Properties of Carbohydrates and Lignins", Shafizadeh, F.,
Sarkanen, K.V., and Tillman, D.A. Eds., Academic Press, N.Y., 19-35, (1976).
Diebold, J.P., Master Thesis, Colorado School of Mines, Golden, Co., U.S.A., (1985).
Fairbridge, c., Ross, R.A., and Sood, S.P., J. Appl. Polymer Sci., 22,497-510, (1978).
Hajaligol, M.R., Howard, J.B., Longwell. J.P., and Peters, W.A., Ind. Eng. Chem. Proc. Des. Dev.,
2l, 457-465, (1982).
Jain, R.K., Bhatnagar, S., Krishnan, L., and Bhatnagar, H.L., Indian Iourn. Chem., 19A,
1171-1174, (1980).
Kala, c., Gut, I.S., and Bhatnagar, H.L., J. Indian Chem. Soc., 56,890-894, (1979).
Kosstrin, H., Proc. Spec. Workshop on Fast Pyrolysis of Biomass, Copper Mountain, Colorado,
U.S.A., 105-121, (1980).
Lewellen, P.C., Peters, W.A., and Howard, J.B., Proc. Int. Symp. Combust., 16, 1471-1480, (1977).
Liden, A.G., M.A.Sc. Thesis, University of Waterloo, Waterloo, Ont., Canada, (1985).
Min, K., Comb. and Flame, 30,285-294, (1977).
Scott, D.S., and Piskorz , J., The Can. Jour. Chem. Eng., 60,666-674, (1982).
Scott, D.S., and Piskorz, J., The Can. Jour. Chem. Eng., 62,404-412, (1984).
Scott, D.S., Piskorz, J., and Radlein, D., Ind. Eng. Chem. Proc. Des. Dev., 24(3), 581-5888, (1985).
Shafizadeh, F., Appl. Polym. Symp., 28, 153-174, (1975).
Shafizadeh, F., and Bradbury, A.G.W., J. Appl. Polym. Sci., 23, 1431-1442, (1979).
Shivadev, U.K., and Emmons, H.W., Comb. and Flame, 22, 223-236, (1974).
Thurner, F., and Mann, U., Ind. Eng. Chem. Proc. Des. Dev., 20(3), 482-488, (1981).
Tran. D.O., and Charanjit, R., AlChE Symp. Ser., 75(184), 41-49, (1979).