Fuel Vol. 75 No. 8, pp.

987-998, 1996
Copyright © 1996ElsevierScienceLtd
Printed in Great Britain. All rights reserved
ELSEVIER PH: S0016-2361(96)00030-0 0016-2361/96 $I 5.00+ 0.00

Pyrolysis characteristics of biomass and

biomass components

K. Raveendran, Anuradda Ganesh and Kartic C. Khilar*

Energy Systems Engineering, Department of Mechanica/ Engineering,
*Department of Chemica/ Engineering, Indian Institute of Technology,
Bombay--400 076, India
e-mai/." aganesh@me.iitb.emeL/n
(Received 22 May 1995; revised 2 January 1996)

Biomass pyrolysis studies were conducted using both a thermogravimetric analyser and a packed-bed
pyrolyser. Each kind of biomass has a characteristic pyrolysis behaviour which is explained based on its
individual component characteristics. Studies on isolated biomass components as well as synthetic biomass
show that the interactions among the components are not of as much significance as the composition of the
biomass. Direct summative correlations based on biomass component pyrolysis adequately explain both the
pyrolysis characteristics and product distribution of biomass. It is inferred that there is no detectable
interaction among the components during pyrolysis in either the thermogravimetric analyser or the packed-
bed pyrolyser. However, ash present in biomass seems to have a strong influence on both the pyrolysis
characteristics and the product distribution. Copyright © 1996 Elsevier Science Ltd.
(Keywords: biomass; biomass components; pyrolysis)

For an agriculture-based economy like that of India, the
prospect of being able to convert widely available
biomass materials into various forms of fuel is most
attractive. In every thermochemical conversion route,
pyrolysis plays a vital role. Hence there is renewed
interest in understanding the complex pyrolysis process.
In the past few decades much work has been done to
study the influence of the operating parameters, such as
temperature, heating rate, pressure and residence time,
on pyrolysis 1. However, there still remains a need to
study the effect of the feedstock properties on the
process. Different kinds of biomass, although consisting
of the same major constituents, have different com-
positions. The influence of biomass composition and ash
content and composition on pyrolysis characteristics is
the object of the present study. This paper discusses the
pyrolysis characteristics and product distribution of
biomass and their correlation with those of the
individual components.
Most of the pyrolysis characteristics reported in
literature are for woody materials. A few attempts have
been made at correlating the pyrolysis characteristics of
biomass with those of its constituents, viz. cellulose,
hemicellulose, lignin and extractives 1. Thermal analysis
curves for wood often exhibit three peaks and have led
researchers to believe that the mechanism of wood
pyrolysis is a superposition of the mechanisms of its
components 2-7. Shafizadeh and McGinnis 8 argued that
qualitatively the thermal behaviour of wood reflected the
behaviour of its components.
Beall 9 showed that the same major components
isolated from wood by different methods behave
differently; and Roberts 1° showed that the structural
properties of the components influence the pyrolysis
characteristics, so he did not attempt to correlate the
pyrolysis behaviour of biomass with that of its com-
ponents. Other workers/1-13 attempted to correlate
biomass pyrolysis characteristics with those of its
components using simple overall pyrolysis kinetic
models. Though good agreement was seen at a particular
heating rate, specific for each biomass, the agreement
was not satisfactory at other heating rates. This
discrepancy was attributed to the influence of heating
rate on char formation / . Antal 1, in his exhaustive review
on pyrolysis of biomass, addressed the problem of
'whether the pyrolysis of lignocellulose can be
represented as a simple superposition of the behaviour'
of its components'. He proposed that a mathematical
superposition of the components' t.g.a, curves should
explain their interaction adequately.
It is appropriate to mention here some of the work
carried out on copyrolysis of various biomass and coal
mixtures which give insight into the interaction of
biomass during pyrolysis. Klose and Stuke 14 reported
no interaction between coal and biomass during
copyrolysis in t.g.a. However, Nikkhah et al. 15, in their
detailed copyrolysis studies of various biomass and coal
mixtures in a batch reactor, reported increased gas yields,
as well as increased heating value and hydrocarbon
content of the pyrolysis gases. McGee et al) 6 reported
copyrolysis studies of the mixtures of poly(vinyl
chloride) (PVC) and wood/straw, to simulate municipal
solid waste pyrolysis char. They found that the inter-
action between PVC and wood/straw increased the char

Fuel 1996 Volume 75 Number 8 987

Pyrolysis characteristics of biomass and biomass components: K. Raveendran et al.

yield but reduced the char reactivity. Copyrolysis studies was determined using standard geological rock analysis
conducted by Khan et al. ~7 on mixtures of coals and methods2°'2l; the results are presented in an earlier
heavy petroleum residues and by Saxby and Sato ~8 on paper 22.
Australian oil shale and lignite, all in a packed-bed All biomass samples taken for both t.g.a, and PBP
pyrolyser (PBP), revealed the prevalence of synergetic studies were ground <250 #m.
effects; they also showed that the initial composition of Isolated biomass components and synthetic biomass
the feedstock mixture had a direct bearing on the product also were studied. Holocellulose, cellulose, hemi-
distribution and properties. cellulose, lignin and extractives were isolated from
On critical analysis of these studies, the interesting fact wood by methods described in the literature ~9. Synthetic
emerges that most of the studies conducted in a t.g.a. biomass was prepared to simulate the results of original
show no interaction among biomass components or whole biomass (with and without ash).
between biomass and coal, whereas the studies con-
ducted in packed-bed batch reactors show interaction in
both cases. Synthetic biomass preparation. Synthetic biomass
Given the range of knowledge reported above, the samples were prepared by mixing each of the individual
following objectives were set for the present study: biomass constituents (cellulose, xylan, lignin, extractives
and ash) proportionately. The proportions of individual
(1) to investigate comprehensively the pyrolysis constituents were obtained from summative analysis.
characteristics and product distribution of various Standard cellulose, lignin and xylan samples were
kinds of biomass and their components in both t.g.a. obtained commercially for this purpose. Extractives
and PBP, and to compare the results to attempt to were isolated from each kind of biomass according to
reconcile the discrepant information and inferences TAPPI standard Tllm. The ash was obtained by
reported in the literature; combusting the corresponding biomass in a muffle
(2) to develop a mathematical model to predict the furnace. To simulate deashed biomass, ash was excluded
pyrolysis characteristics and product distribution of from the mixture of other constituents.
biomass from the characteristics of the individual
components; Pyrolysis
(3) to study the influence of the composition of the ash To fulfil the objectives of the study, pyrolysis
as a parameter within (1) and (2). experiments were conducted in both a t.g.a, and a PBP.
In the t.g.a, experiments a small sample (5-10mg) was
taken, thus reducing the effect of secondary reactions
EXPERIMENTAL and heat and mass transfer on the product yield, whereas
in PBP experiments a much larger sample was used
Feedstock (10-25 g), so that secondary reactions and other heat and
Biomass. Fourteen commonly available kinds of mass transfer effects played greater roles.
biomass in the region were selected for the present study.
These included a variety of potential biomass fuels T.g.a. studies. A DuPont 9900 thermal analyser was
available from energy crops (wood) as well as residues used for all the pyrolysis studies. Dynamic t.g.a, studies
of agricultural (husks, straw etc.) and food processing were carried out at linear heating rate of 50Kmin -1,
activities (bagasse, nut shells etc.). The chemical com- covering the temperature range from atmospheric to
position of these materials was determined by methods 1273K. All the studies were carried out in an inert
described in the literature ~9. The composition of ash atmosphere of flowing nitrogen (50 cm 3 min-1).

GAS .EATER7 [ REACt0R

/ ~ I F "H00K T THt=RMOCOUP!.E 8 CLEANING'PORT
I~.'.~ ...... "'" "J"" "~ I!'"'] [ l J-*,-i 1 PRESSURE REGULATOR 9 PRODUCT VAPOUR
'|;J. . . . . . . . ~ ' ~ T , RO*AME,,R ,0'EMPE"ATURE = * . , . 0 " E .
I'. Jel | e ~ w°",t*l')~l~ I; "b2~LI 3 PYROLYSISREACTOR It TEMPERATURE INDICATOR
I'~,~lff"l."i~:-.:~ -- I'::~'111"J~ ,ELECTR,CALFU..ACE' ,, CO.OE.SE" TURE
I:: ,il l l : 1 ; ;':;i v , CERA.,C woo, .0., .EDWATERTA.,
I::,:~:,,I~I.~F;I-';.::.I I~ ~1:1::'.~1 ..SU,AT.O. ,, WATER C,ReO,AT,O.P~MP
~ ~, ~. .~ t / , r s .eatlnSEUEMe"T
tl'~" ~" -,, -~ ~.
' v " ,~*
: ,,~'.: "- s CERAMIC TUBE
..b-. o - 16 GAS FLOW METER

~,-¢ .... •
. . . . •1' ~',' J

t ,--" ........... I // - II \\ ' ~""-~_~_~ ATMOSP.ER~,

® @

Figure 1 Schematic of the packed-bed pyrolyser test set-up

988 Fuel 1996 Volume 75 Number 8

 Pyrolysis characteristics of biomass and biomass components." K. Raveendran et al.

Table I Pyrolysischaracteristicsof biomass componentsin t.g.a.~

Yield (wt% daf) Max. Temp. Initial Temp. at

rate at max. decomp. involution
Volatiles Char (wt% K -l) rate (K) temp. (K) point (K)
Whatman cellulose 97.5 2.5 1.7 682 573 713
Wood cellulose 86.0 14.0 1.7 667 573 697
Alkali lignin 59.4 40.6 0.3 695 413 773
Acid lignin 52.9 47.1 0.4 693 473 700
Hemicellulose 68.0 32.0 0.5 573 448 550
Xylan 70.0 30.0 0.9 584 463 500
Extractives 73.0 26.9 0.4 668 393 575
a Heating rate 50 K min-1

Untreated biomass, isolated biomass components and 1,00 r r r

Acid l i g n i n
synthetic biomass were all subjected to these analyses. o + llgnin
AIk=tl
0 Extractivcs
If w h a t r n a n ce(lulos¢
• Wood c e l l u l o s e
PBP studies. Experiments were conducted in a 0,75 • Herni cellulose
• Xylon
packed-bed pyrolyser designed for the purpose, with g
provision for collecting the pyrolysis products. Figure 1 u
shows a diagram of the pyrolysis reactor set-up, consist- ~ 0,50
ing of an electrically heated stainless steel tubular
pyrolysis reactor and another electrically heated stainless
o,
steel tubular gas heater. Nitrogen from the cylinder was
0.25
first heated to the operating temperature in the gas heater
and entered the pyrolysis reactor tangentially at the top,
flushed the pyrolysis vapour and left through the bottom .==. I=, ~ t
of the reactor through a double-walled glass condenser 0.00

tube to a train of wash-bottles immersed in an ice bath. + Acid tignin

1.50
Cold water was circulated through the condenser. A DTG O Alkali lignin
o Extmctlves
programmable PID controller was used to control the A Whatmon cellutos~
1.25
temperatures of the reactor and heater. Thermocouples v Wood cellulose
• Hemi cellulose
were incorporated at three places in the reactor and one in 1.00 • Xylan
the sample boat to monitor the sample temperature. E
m
Five representative kinds of biomass were chosen for 0.75
investigations in this reactor. Isothermal experiments
773K in an inert atmosphere of flowing nitrogen >
O,SO
(500cm 3 min -1) were conducted for each of these five
Q
biomass samples, isolated biomass components and 0'25

mixtures of the components.

A known weight of ~ 1 0 - 2 5 g of powdered samples 0.00
150 250 350 450 550 650 750 850
was taken in the stainless steel mesh 'boat' and placed in Temperature ( C )
the furnace, which was already maintained at the
Figure 2 T.g.a.and d.t.g, curvesof biomass components
predetermined temperature of 773 K with the nitrogen
flow. The volatiles evolved were collected and quenched,
and the non-condensable gases were passed through a
presented in Table 1. Several observations can at once be
flow meter. The experiments were continued until the
highlighted:
evolution of gases ceased.
The furnace was switched off and the flow of nitrogen
was maintained until the sample attained room tempera- (1) cellulose decomposes within a narrow temperature
ture. The char remaining in the furnace was carefully range, 573-703 K;
removed and weighed. The yield of liquids was obtained (2) cellulose decomposition rate is the highest and char
by weighing the wash-bottles before and after the yield the lowest;
experiment, and the yield of gas was obtained by (3) Whatman cellulose has a higher rate as well as a
difference. Repeatability was ensured by repeating the narrower temperature range for decomposition than
experiments until the same product yield fractions were cellulose extracted from wood;
obtained with a precision of 4-2%. (4) lignin on the other hand decomposes over a wider
temperature range, 523-823 K;
(5) the char yield from lignin is the highest, in the region
RESULTS A N D DISCUSSION of 45-50 wt%;
(6) hemicellulose and xylan are thermally the most
Pyrolysis characteristics of biomass components in t.g.a. unstable and start decomposing at a much lower
Figure 2 presents the integral and differential thermo- temperature than every other component, with
grams of the basic components. The pyrolysis data are ~30 wt% char yield;

Fuel 1996 V o l u m e 75 Number 8 989

Pyrolysis characteristics of biomass and biomass components: K. Raveendran et al.

Table 2 Pyrolysis characteristics of whole biomass in t.g.a,a

Yield (wt% daf) Max. Temp, Initial Temp. at

rate at max. decomp, involution
Volatiles Char (wt% K-l) rate (K) temp. (K) point (K)

Bagasse 79.7 20.3 0.9 677 483 688

Cashewnut shell 81.1 18.9 0.6 638 473 673
Coconut coir 69.8 30.2 0.8 672 513 673
Coconut shell 70.7 29.3 0.8 615 518 678
Coir pith 56.8 43.2 0.6 622 483 663
Corn cob 73.5 26.5 1.1 603 533 653
Corn stalks 70.9 29.1 1.1 634 498 653
Cotton gin waste 80.6 19.4 1.3 679 523 688
Groundnut shell 68.7 31.3 0.7 662 493 683
Millet husk 70.1 29.9 0.9 653 523 663
Rice husk 70.0 30.0 0.8 666 518 663
Rice straw 74.7 25.3 1.0 651 518 673
Subabul wood 76.3 23.7 0.9 683 498 663
Wheat straw 72.8 27.2 0.9 604 493 663
Heating rate 50 K min-1

Table 3 Distribution of volatiles released during biomass pyrolysis in t.g.a,a

Distribution of volatiles (temp. ranges)

Moisture
Zone I Zone II Zone III Zone IV Zone V Total
<373K 373-523 K 523-623 K 623-773 K >773 K volatiles

Bagasse 5.6 3.2 27.2 41.8 4.5 79.7

Cashewnut shell 1.3 4.7 37.7 27.2 11.5 81.1
Coconut coir 8.3 2.2 25.0 35.1 7.6 69.9
Coconut shell 7.4 2.8 30.3 33.1 4.4 70.7
Coir pith 14.6 1.0 22.0 30.6 3.2 56.8
Corn cob 5.5 2.5 28.6 38.9 3.5 73.5
Corn stalks 5.4 2.3 27.2 30.4 11.0 70.9
Cotton gin waste 7.4 1.7 27.6 42.8 8.6 80.7
Groundnut shell 8.0 4.4 20.9 36.7 6.8 68.7
Millet straw 8.3 4.2 30.6 30.4 4.9 70.1
Rice husk 7.7 6.2 22.7 37.5 8.2 70.0
Rice straw 9.7 5.1 24.1 40.3 5.2 74.7
Subabul wood 6.2 3.6 24.2 44.5 4.0 76.3
Wheat straw 6.5 3.2 36.0 24.3 6.3 72.8
a Moisture, wt% ash-free biomass; volatiles, wt% daf biomass; heating rate 50 K min-l

(7) extractives decompose in a similar way to lignin b u t
at a slightly higher rate a n d at a slightly lower
temperature.
F r o m the above observations, a ' z o n a t i o n ' scheme can
be envisaged, relating to the d e c o m p o s i t i o n status of
each c o m p o n e n t . The 'zones' are postulated as:
zone I: <373 K m a i n l y m o i s t u r e evolution
zone II: 373-523 K extractives start d e c o m p o s i n g
zone III: 523-623 K predominantly hemicellulose decom-
position
zone IV: 6 2 3 - 7 7 3 K m a i n l y cellulose a n d lignin decom-
position
zone V: >773 K m a i n l y lignin d e c o m p o s i t i o n
Pyrolysis characteristics of biomass in t.g.a.
The fourteen potential b i o m a s s fuels selected for this
study were subjected to pyrolysis in the t.g.a. The
results are presented in Table 2. I n accordance with
the z o n a t i o n hypothesis, the volatiles released o n
pyrolysis were classified as shown in Table 3. T h a t
each k i n d of b i o m a s s has its o w n typical decom-
position characteristics is evident from these tables.
F o r example coir pith releases only 5 6 w t % as
volatiles, whereas bagasse a n d c o t t o n gin waste each
release ~ 8 0 w t % . The m a x i m u m rate o f decomposi-
tion of c o t t o n gin waste is m o r e t h a n twice that of
coir pith. Even t h o u g h it is evident from Table 3 that
m o s t of the volatiles are released in zone IV, the

990 Fuel 1996 Volume 75 Number 8

 Pyrolysis characteristics of biomass and biomass components. K. Raveendran et al.

t.O0 J ~'.'N"~. I , 100

J~"

Synthetic \ x . ~i Synthetic U
O
'~ 0.75 . \ll I . . . . Untreated .. \\,~ .... Untreated , II -Untreated
o
0.75
?, /~l I " \ i~
m

- ,,}'\I COR.CO . RICE HUSK I WOOD

0.50
,ti', o.0.50
.C
Oa

'~' 0.25
0.25 >
>

0.00 -- ~___ / I , .
i 1.00 o
150 350 550 750
• Synthetic A
o Temperature ( C )
A
- Untrzotzd - ~\/,l - - Untreated
8 o.75 --
035
u
COfR PITH RICE STRAW
O

u. 0,50
0.50

0.zs
"~, 0.25 .~
>o
0,00
___¢' ," ,

150 0.00
3 SO 550 750~150 350 SSO SO
Temperature (C) Temperature ( C )

Figure 3 T.g.a. and d.t.g, curves of synthetic and untreated biomass

numerical values range from 28wt% for coir pith to Table 4 Crystallinity index of cellulose
almost 81 wt% for corn stalk.
Cellulose Cr-I
The above observations bring about the specificity of
individual biomass types. This specificity is held to be Whatman 34.4
attributable to the differences in composition of the Chromatograph 91.5
biomass. To investigate whether the effects of the Wood 68,9
Coir 49,6
individual components of a biomass are simply additive, Corn cob 68.9
synthetic biomass prepared by mixing the relevant Rice husk 56.5
components was studied. Figure 3 presents the weight Rice straw 52.9
loss and derivative weight loss curves for synthetic G r o u n d n u t shell 54.7
biomass samples (the curve for fresh biomass is also
shown for comparison).
It is appropriate to mention here that the curves of the cellulose (defined in terms of Cr-I, which incidentally is
synthetic biomass matched well (with minor deviations of the same order of magnitude for all types of biomass),
explained later) the curves for fresh biomass only when has an effect on the pyrolysis characteristics of biomass.
wood cellulose was used for preparing the synthetic
samples. The deviation of the fresh biomass curve was Mathematical correlation for pyrolysis in t.g.a.
much greater for synthetic biomass containing Whatman An attempt was made to represent mathematically the
cellulose. To explain this, X-ray diffraction (XRD) correlation between the pyrolysis characteristics of
analyses were carried out to determine the crystallinity biomass and those of its components. This simple
of celluloses isolated from various kinds of biomass. The additive correlation assumes that the overall pyrolysis
crystallinity index (Cr-I) was calculated by the method behaviour ofa biomass is the weighted sum of the partial
suggested by Segal and co-workers 23'24. Table 4 gives the contributions of its components, the relative proportions
Cr-I values obtained for different types of cellulose. It in the total composition defining the respective weight
may be noted that crystallinity of cellulose extracted losses of the components:
from biomass lies between the Cr-I values of Whatman
cellulose and chromatograph cellulose. It is known 25-28 [AWbIT]C = AWc]Z ° Xc + AWhIT" Xh + AW]IT" Xl
that crystallinity influences the pyrolysis characteristics. + AWeIT'Xe (1)
Hence, to synthesize biomass it is necessary to use
cellulose having a crystallinity similar to that in the where [AWbtr]c is the correlated weight loss for any
natural biomass. Hence wood cellulose was used. given biomass and X¢, Xh, Xl, Xe are the initial fractions
The above results indicate the interesting fact that the of cellulose, hemicellulose, lignin and extractives,
components themselves play individually significant respectively, present in the corresponding biomass. The
roles in determining the pyrolysis characteristics of values of local weight loss at a temperature T for
biomass. Also, the basic structure or degree of poly- biomass and for all components, designated as AWblT,
merization of the biomass is less significant than its Awcl T, Awh[T, AWl[T and AwelT respectively, are
composition. In other words, the way in which com- obtained from individual t.g.a, curves.
ponents are bound (chemically) is not as important as the The correlated data for biomass are obtained through
actual amounts of individual components present in a the following steps based on Equation (1):
particular biomass. However, the chemical structure of (1) The t.g.a, weight loss curve for each component is
the individual components, for example the type of accurately read in a fixed temperature range to define

Fuel 1996 Volume 75 Number 8 991

Pyrolysis characteristics of biomass and biomass components: K. Raveendran et al.

1O 0 ~,.-.~ Im , ~ w ~ --" 1,00

"~N~. /t $1r1= 0.179 ~\ SE,--" 0.130 ~ ' I S[1 ': 0"11'!
075-. , ' [, SE2=0.0098 . \',~ $£2= 0"0097 $E2=0"0086 075 "~"

. . . . •

/., ,....) f~
000 _.. , ,o " = - J ,
I ~ $EI = 0.137 ~:4"~x. ~' SEI = 0 ,It49 ISO 350 550 750"
\,, s~2 = o.oo9,. ":~ !I st.; -- o.oo-9,.. ~ r.,',,p,~a,,,r,(C)
o \ ~, :l ' Rice straw 3.75 ~ _ Experiment=t
: ~ t] ; ---Correlatedl Eqns.(l) ~,. (2)
~0.S0- / ~ ] [ ~ .. i~!:! _ _ . . . . Correlated, Eqns.(3)~,. (4)
0.50 =~n $E1- Standard error, Eqn$. (1) ~, (2)
/il i : ~ / /\[1 "~ SE2-Stondarderror , Eq~ (3). (4)

~02s- //i~: ~ -:. fi ~ ~-. - "

ooo.--- , --.
150 350 550 750~150 350 SSO 750
Temperature ( C ) Temperature( C)
Figure 4 T.g.a.and d.t.g, curvesof untreatedbiomass,and correlateddata

the weight loss at that temperature; these weight
losses are used as weightings.
(2) The weighted sum of the weight loss of biomass over
the specified temperature range is computed by
multiplying the weight losses of the individual
components obtained in step (1) by the correspond-
ing initial fractions of the components present in the
biomass and summing.
(3) The resulting weighted sum is taken as a measure of
weight loss at the specified temperature for the total
biomass, which is summed over the temperature
range 423-1023 K to obtain the complete weight loss
curve.
Figure 4 compares the calculated weight loss with
typical t.g.a, curves for few kinds of biomass, as well as
the derivative curves for these cases. The equation for the
derivative of weight loss curve is as follows (as may be
deduced from Equation 1):
dw
q- (~)1T'XI-}- (~tt)e T"Xe
where [(dw/dt)b[7-]c is the correlated derivative weight
loss for any given kind of biomass and the values of
the derivative weight loss at temperature T for the
biomass and for all its components at a temperature
are designated a s (dw/dt)blT, (dw/dt)clT, (dw/dt)llT,
(dw/dt)hlr, and (dw/dt)elr, and are directly obtained
from the individual d.t.g, curves.
As can be observed, the correlated data are
comparable with the experimental data, with a standard
error of 0.13-0.18. The deviation in the correlated data is
mainly in the region where the devolatilization rate is
high and where a major fraction of the devolatilization
occurs. This deviation may be attributed to the influence
of ash elements, as shown below, besides some
unaccountable factors such as density and porosity.
To calculate the weight loss data using Equations (1)
and (2), wood cellulose t.g.a, data were used (Figure 2).
It is important to mention here that the cellulose chosen
for study should have similar crystallinity, as discussed
above. When Whatman cellulose data were used in
Equations (1) and (2) the calculated weight losses did not
match the experimental data. Figure 5 shows the
experimental and calculated (using Whatman cellulose)
t.g.a, data. As can be seen, the deviation obtained is quite
significant, which is mainly attributed to the difference in
the cellulose crystallinity.
To put the above inferences on a firmer basis,
correlation studies were also conducted with synthetic
biomass samples prepared with and without ash; the
results are shown in Figures 6 and 7. It is interesting and
encouraging to find that the correlated data match very
well with the calculated data in the case of synthetic
biomass without ash. The deviation seen in the case of
synthetic biomass with ash is obviously due to the
influence of ash.
dw
To account for the influence of ash, a multiplicative
correction factor was introduced in the above
correlations:
(2)
[eXWblr]E = [aWb[r]C
x A- ~ XdT: xexp
(3)
[(~-~)b TJE= [(-~)b TIC
X (A- [()(a~nlxnz]
Slj
(4)
where [AWbIT]E and [(dw/dt)blr]E are the experimental
weight loss data and derivative weight loss data for any
given biomass, Xa, Xsi and X1 are the initial silica-free

992 Fuel 1996 Volume 75 Number 8

 Pyrolysis characteristics of biomass and biomass components." K. Raveendran et al.

1.00 1.00
i ! $E -- 0 o2241 ~\ ^ 5E=0,2253 N SE = 0"2656
0,75 ,, - l - _ 0-75 u
f I coRN cob _\/ ~ .ICE Husk "~I~, wooo

i 0,50 - ~l - / \\' - , ~ _
0.S0 -~
~l I 1~, l, I ?,~1
~
0.25 - - O.2S .~
i
I '. ~ . . . _ . m _ = _ = ~ _ = ~__ , ,,
0,00 ~ i ~ I " ~ i ~ - ~ - - 0-00
- ~'X " "-~'x. /~ ,so ~so sso 7s
Temperature (C)
c 0,75 0,75 u
o
Experimental) Eqns. (1) ~. (2)
Correlated (Whatmo Cellulose)
"= o.s, 0.50 .~ SE-Standard error

.c II
T, 0.2 o.2s -~

0,00 0,00
150 350 SSO 750,150 350 SSO 750
Temperature (C) Temperature (C)

Figure 5 T.g.a. and d.t.g, curves of untreated biomass,and correlated data

1.0 ~ ~ ~ t,oo
5E = 0,0683 5E = 0.0732 5E=0.0142
g o.7s Synthetic corn cob Synthetic rice husk " Synthtti¢ wood " 0.75
wdhout ash
with out ash without ash I
u~
-- O.SO

0-25 .... ., I 0.25

1 !SE
°

0.00 o.oo

5E : 0"0 473 : 0.0732 ~so "-¢'~ 350Temperature(cSSO

) 7SO
"~ 0.7 Synthetic coir pith Synthetic rice straw" 0.?S ;"
o " ' I Experimental
u without ash I without ash .... Correlated, E q n s . ( t l & ( 2 )
o
== SE - Standard error
~ O.S, - 050

E .=
en 0.25 0.25

\ -~ _ /• r~
0.00F=- ~'0.00
iS0 350 550 "/5OptSO 350 SSO 750
Temperature(C) Temperature ( C )

Figure 6 T.g.a. and d.t.g, curves of syntheticbiomasswithout ash, and correlated data

ash, silica and lignin fractional contents in dry biomass action of certain ash elements in the presence of pyrolysis
on a weight basis, and A = 0.5, n 1 = 8.5 and n2 = 7.0. products such as water and carbon dioxide. The
Constants A, n 1 and n2 were obtained by non-linear inorganic ash elements are known to play a catalytic
optimization, using least-squares criteria for con- role in gasifying char. Silica in ash does not play any
vergence, using the Box complex algorithm 29. As can catalytic role; however, it alters the thermal properties
be seen from Figures 4 and 7, the weight loss data and the pore structure of char and hence its reactivity.
predicted using Equations (3) and (4) match the
experimental results well. Mathematical correlation for pyrolys& in PBP: prediction
Equations (3) and (4) show the combined effect of ash of product distribution
elements and biomass constituents, particularly the To investigate further, studies were carried out in the
lignin component. This may be explained in a similar packed-bed pyrolyser with five representative types of
manner to that in the earlier paper on the influence of biomass and also the standard biomass components such
ash 22, that lignin present in biomass forms more char as cellulose, xylan and lignin. This was intended to
during pyrolysis and this char is gasified by catalytic provide insight into the interaction, if any, among the

Fuel 1996 Volume 75 Number 8 993

Pyrolysis characteristics of biomass and biomass components: K. Raveendran et al.

1,00 -~ ~ n -- "~"~ I I ~ r i 1.00

~'~x , SEI =0.2Or. '"%, 5E! = 0.932 "~. SEt = 0,981
;o,,- ,,,=o.oo,, \., o oo,. o oo,, _o,,7
~tl.~, 1, Synthetic corn ¢ob ~ / ~ Synth¢tic ricehusk '~ ! Synth¢tic woo¢l --
k ~1 ~'l w,,h a.h j~ll w.h ash e,\.~L .,,h a.h
o.,o- t " ti\4 -oso
_ .'. i "...

o~s- ). i I "-~ ...... /..:,, - ~ /-i ~ ~ - ~ o

- J , "~"~------ooo ~
°°°L ,_- o. o.,. ,,o ,o

.9 / ~ ~ Synthetic coir pith ~ Synthcticrice straw 3.75 Expcrimentol

~" / A~'~ w,l, a.~ / ~ * " " °'" z ---¢orre,a,ed. E,.s..~= C2~
0150I- t ~ _ -- ~ I - 0.50 "~ ...... Correlettetl, Eqll,, (3)& (/~-)
SE1- StandarderrorpEqBs(1)&(?.)
~ 0.25[ //I ' " ' ' ~ //i: ~ : SE2-Standard error, EqBS (3'&'&)
; t ..#..-5' k " i .... 0.,, ._.
0.00~ I I~ I ~ 13 13n
150 350 550 750,150 350 550 750
Temperature(C) Temperature(C)

Figure 7 T.g.a. and d.t.g, curves of synthetic biomass with ash, and correlated data

Table 5 Pyrolysis product yields (wt% daf) in PBP experiments product distributions obtained from most of the biomass
types are intermediate between those obtained with
Volatiles Char Liquids Gas
cellulose and lignin. In other words, the product yields
Biomass obtained with cellulose and with lignin hold as the upper
Coir pith 70.5 29.5 29.5 41.0 and lower limits respectively for the product yields that
Corn cob 79.9 20.1 37.4 42.5 can be obtained with most types of biomass, for a given
Groundnut shell 72.9 27.1 40.5 32.5
Rice husk 82.7 17.3 41.2 41.5 operating condition. This is an important criterion to be
Rice straw 78.8 21.2 47.0 31.8 considered for any biomass pyrolysis process design.
Subabul wood 80.7 19.3 22.6 58.1
Components Additive correlation for product distributhgn. Similar to
Cellulose 88.9 11.1 46.8 43.0 the correlation discussed above for t.g.a, data, an additive
Lignin 58.3 41.7 26.8 30.5 model is suggested for predicting the product distribution
Xylan 79.3 20.7 40.5 38.8 obtained in the PBP. On similar lines as explained earlier,
this model also assumes that the additive overall product
distribution of biomass is the weighted sum of the partial
components during pyrolysis. The information on contributions of its components, with the relative propor-
product distribution would be useful in design and tions in the total composition defining the respective
development of any pyrolysis process. weights of the components:

Biomass and components pyrolysis product distribu-

(rb, i)c =rc, i.xc+ rx, i . ( x h + x o ) + rl,;.(xL) (5)
tion. Table 5 presents the product distribution obtained where Y is the product yield and subscripts b, c, x, h, and
for biomass on pyrolysis in the PBP. The char yields 1 denote the yields from biomass, cellulose, xylan,
from biomass vary from ~17wt% for rice husk to hemicellulose and lignin as obtained by additive analysis,
30wt% for coir pith. The liquid yields vary from on the dry and ash-free basis; subscript i denotes char,
,,~22wt% for wood to almost double that figure for liquid or gas accordingly.
groundnut shell and rice husk. The gas yields vary The extracts and hemicellulose are seen to behave
from ~30wt% for groundnut shell to ~58wt% for similarly, especially in terms of char yield (cf. Figure 2).
wood. As seen earlier in the t.g.a, studies, here also bio- Hence extractives can be combined with hemicellulose
mass showed specificity in terms of product distribution. (Xh + Xe) for calculating the product yield.
(Note: for the sixth set of data, for rice straw, see below.) Table 6 presents the theoretical product distribution
To gain more information to explain the specificity of calculated by Equation (5) and the deviation from the
biomass or the variation in product distribution, experimental values. As can be seen, the deviation is
experiments with biomass components were conducted greatest for wood and least for groundnut shell. To
with the individual isolated components. The product explain the deviation between the experimental and
distributions obtained for the biomass components calculated product yields, the possible influence of two
studied are also presented in Table 5. As can be seen, factors was considered: (1) interaction among the
wood cellulose gives the lowest char yield and the highest components; and (2) influence of ash elements. Each
liquid and gas yields, whereas lignin gives the highest factor was investigated separately.
char yield and the lowest liquid and gas yields. The
product distribution from xylan lies between that of (1) Interaction among components. Since cellulose and
cellulose and lignin. It is also interesting to note that the lignin together form ,,~70wt% of the biomass, these two

994 Fuel 1996 Volume 75 Number 8

 Pyrolysis characteristics of biomass and biomass components." K. Raveendran et al.

Table 6 Calculated product yieldsa

Char Liquid Gas

Calc. Diff. Dev. Calc. Diff. Dev. Calc. Diff. Dev.

Coir pith 24.4 -5.2 - 17.5 46.8 17.3 58.9 36.5 -4.5 - 10.9
Corn cob 21.1 1.0 5.2 42.4 4.9 13.1 40.7 - 1.8 -4.3
Groundnut shell 24.0 -3.1 - 11.4 38.8 - 1.7 -4.5 38.0 5.5 17.1
Rice husk 21.2 3.9 22.4 41.5 0.3 0.7 39.9 -1.7 -4.0
Rice straw 21.1 -0.1 -0.5 41.8 -5.2 -11.0 39.0 7.2 22.7
Wood 21.9 2.6 13.3 39.2 16.6 73.2 38.0 -20.0 -34.5

a Calc., calculated product yield (wt% daf) by Equation (5);

Diff., difference between experimental and calculated values (wt% daf);
Dev., relative deviation from experimental data (%)

Table 7 Pyrolysis product yields (wt%) of component mixtures ash and silica contents of the biomass, is in accord
Cellulose:
with previous studies22. The following relation describes
lignin ratio Volatiles Char Liquids Gas this function:
3:1 81.9 18.1 40.5 41.3 (Yb, i)E ----=(Yb,i)C x [A XI nl j[-~2 Xsn3] (6)
1: 1 70.2 29.8 33.1 37.0
1:3 63.9 36.1 31.5 32.8 where ( Yb,i)E is the product yield obtained experimentally
and (Yb,i)C is defined by Equation (5).
Figure 9 shows a plot of %,i vs. [A XInl Xan2 Xs~3],
60 I I I I I t where 'ffb,i (Yb, i)E/(Yb, i)C and the values of nl, n2, n3
=
(C-" L ) - C E L L U L O S E °LIGNIN RATIO
and A are presented in Table 8 along with standard error
..,/.
(0:1) (1:3) (1:1) (3;1) (1:0) and R 2 values. These values of nl, n2, n3 and A are
',.9
LIJ
difterent for the char, liquid and gas yields. The above
~= 4O correlation was developed using data obtained for the
five biomass materials used in the PBP experiments. To
Q
,..I validate the correlation, experiments were conducted for
_w
one more kind of biomass, namely rice straw; yields of
~ 20 char, liquids and gases matched well with those obtained
U
from the correlation.
o
O. t, - GAS Y I E L D
It can be inferred from the results of both t.g.a, and
~ ,,,! .... i .... i .... i .... I .... i ....
PBP experiments that the secondary reactions or any
- 20 0 20 40 60 80 100 120 other effect due to the increased sample mass mainly
CE L L U L O S E ( ' / , WEIGHT) influence the product distribution and that the inter-
I . l l , i . . . . I ) i , i I . . . . I , i i , I . . . . i . . . . i
action of the individual components remains unaffected.
120 100 80 60 40 20 0 -20 This is clear from the fact that the t.g.a, and PBP results
LIGNIN 1% WEIGHT) can be used to develop correlations to obtain pyrolysis
characteristics. The differences lie in the correlation
Figure 8 Product yields of cellulose-lignin mixtures vs. cellulose and
lignin contents coefficients and not in the observed trend.

Simplified overall correlation. To be able to predict the

major components were mixed in different proportions. biomass pyrolysis product yields using Equation (6), it is
The experimental product distributions obtained are necessary to know not only the chemical composition of
presented in Table 7. the biomass but also the pyrolysis product yields from
As seen for biomass product distribution above, here the individual components at a specified temperature.
also the product yields obtained lie between the upper To simplify this further, another correlation was
and lower limits obtained from cellulose and lignin. It attempted, keeping in view the influences of ash and of
can also be readily seen from Table 7 that as the cellulose cellulose and lignin, which contribute the major fraction
content of the mixture decreases, the char yield increases by weight. This is expressed as:
and the liquid and gas yields decrease. Figure 8 presents
the product distribution of mixtures as a function of their
cellulose or lignin content. It is clear that the product
yields are direct functions of the initial composition of where the constants A, nl, n2 and n 3 were obtained by
the mixture. Hence it may be said that there is no regression analysis and are presented in Table 8 along
detectable effect of interaction among the components with the standard error and R 2 values obtained. The
on pyrolysis product distribution. correlations were developed with the experimental data
obtained for five kinds of biomass and were verified with
(2) Influence of ash. To explain the influence of ash, a the data obtained for rice straw. Figure 10 shows the
similar approach to that used for the correlation in product yields as a function of composition, and as can
the t.g.a, study was used here. The influence of ash be seen, the calculated yields are in good agreement with
was found to be a function of the lignin, silica-free the experimental values.

Fuel 1996 Volume 75 Number 8 995

Pyrolysis characteristics of biomass and biomass components: K. Raveendran et al.

1'60
WD
1,4,0

~ 1.20
Gs cP
GS ~ A ~ C P
~,~ 1.00 . . . . .
.... /2~- - -6 / "CB

"80 I
l I ', I L I I I i I I I
,.2 -°''u o.s 0"8 1-o 1-2 1.4 0.4. 0'6 0'8 1"0 1"2 1,4- 1"6

1o,
...............
5S o R H . . . . . . . . . . .

.0"8 8
.g oc/cop e
~. 0,92

0'56 o CP aWD
I WDI I i ! If I I I I I
1'300 0'6 0'8 1'0 0"2 0'6 0'6 0'8 1'0 1"2 1"4.
CP- COIR PITH CP RH-RICE HUSK CP
RS-RICE STRAW o j
CB-CORN
-GS- COB
GROUNDNUT SHELL f o G S -wo -WOOD s
1.10
J¢
u . . . . . . . . . 3; f-oG8
WD / I -
0.90
R Ho,,,~/ Jl
o I
I
RH I
0"70 i ~ i I I I I I I I I I I II ] I I I
0,75 -85 -95 1"05 1.15 0"8 "9 1"0 1-1 1'2 1.5
n 1 n2 n3
X t .Xa "Xsi C -X• -Xsi .C

Figure 9 Ratio of product yields (experimental and correlated) vs, cellulose, lignin, ash and silica contents

Table 8 Regression coefficients for Equations (6) and (7)

Correlation coefficients
Standard
A n1 n2 n3 error R2

Additive correlation (Eqn 6)

Char 2.584 0.378 0.343 -0.048 0.13 0.79
Liquid 1.888 0.694 0.273 0.030 0.23 0.79
Gas 0.349 -0.562 -0.363 0.001 0.15 0.86

Overall correlation (Eqn 7)

Char 49.90 -0.0258 0.258 -0.044 0.03 0.99
Liquid 102.86 0.7028 0.335 0.033 0.16 0.90
Gas 13.93 -0.4168 -0.337 -0.002 0.15 0.87
Char (t.g.a.) 56.43 -0.2664 0.079 0.048 0.12 0.91

The above correlations give a good fit for char yields synthetic biomass studies. The influence of ash is
obtained in the t.g.a, studies also. The correlation reflected in the correlation developed.
coefficients obtained for t.g.a, data are also presented It is also shown that the individual components taken
in Table 8. As can be seen from Figure 10, the t.g.a, data for study should be similar to those present in biomass.
also follow a similar trend to that of the PBP data. In this respect the crystallinity index is found to be a
useful measure for comparing various celluloses.
The specificity of biomass is also seen in the product
CONCLUSIONS
distribution obtained in the packed-bed pyrolyser, which
Thermogravimetric studies show that each kind of is further explained by the product distribution from the
biomass has unique pyrolysis characteristics, by virtue individual components. In this case also, a simple
of the specific proportions of the components present additive correlation incorporating the influence of ash
in it. A simple additive correlation to define this has been developed which describes the product
phenomenon has been developed, which suggests that distribution well. These results are supported by studies
there is no detectable interaction among the components on mixtures of components. The results re-emphasize the
during pyrolysis. These results are supported by fact that the feedstock composition and properties play a

996 Fuel 1996 Volume 75 Number 8

 Pyrolysis characteristics of biomass and biomass components. • K. Raveendran et al.

60
WD

~" so

>.
40

30
S RH

20 i i i

SO
!o ,'o s'o ' ,o ! 30~ ' 4 ~0 : 510 i 60

GS RHH~"
o.. 4(]

i/I
' ~ 313
>-

2( 1 I
I ;tl5 I 3iS ! 415 20 30 40 5b
30
oGS

m 2S
13.
"E
~ 20
oRH'~ * oWO

15 I I I I I I I
I ,119 I I I 217 I
15 23 1S 19 23 27 30
45 RH-RICE HUSK
CP- COIR PITH CP
RS-RICE STRAW CP
,¢ 40 • CB-CORN COB o o
G5- GROUNDNUT SHELL WD-WOOD
p.
i-
o
~>- 30

I I t / I ! I
20 J 2,5 i ~ i i w 37.5
20 3 35 37.5 20 25 30 35

x l. x 2. x , in3
. c
rxq "1 x2.2 x, 3. c
L~?J "
Figure 10 Product yields as a function of cellulose, lignin, ash and silica contents

significant role in determining pyrolysis characteristics 2 Akita, K. Rep. Fire Res. lnst. Japan 1956, 9, 10
a n d p r o d u c t distribution, a n d these are generically specific 3 Stamm, A. J. Ind. Eng. Chem. 1956, 48, 413
4 Arseneau, D. F. Can J. Chem. 1961, 39, 1915
for each k i n d o f b i o m a s s , u n d e r given o p e r a t i n g 5 Radislav, D. and Frantisek, R. Holz Roh- Werkstoff 1962, 20,
conditions. 473
Results o f b o t h t.g.a, a n d P B P experiments show t h a t 6 Tang, W. K. Forest Service Research Paper FPL 71, US Dept.
s e c o n d a r y reactions o r a n y o t h e r effect due to an increase of Agriculture, 1967
7 Shafizadeh, F. and DeGroot, W. F. In 'Fuels and Energy from
in s a m p l e m a s s m a i n l y influence the p r o d u c t distribu-
Renewable Resources' (Eds F. Shafizadeh, K. Sarkanen and D.
tion; the i n t e r a c t i o n o f the i n d i v i d u a l c o m p o n e n t s Tillman), Academic Press, New York, 1977
r e m a i n s unaffected. 8 Shafizadeh, F. and McGinnis, G. D. Carbohyd. Res. 1971, 16,
273
9 Beall, F. C. Wood Sci. Technol. 1971, 5, 156
ACKNOWLEDGEMENTS 10 Roberts, A. F. In 'Thirteenth Symposium (International) on
T h e a u t h o r s wish to t h a n k P r o f e s s o r W o l f g a n g Klose, Combustion', The Combustion Institute, Pittsburgh, 1971,
p. 893
F G T h e r m o d y n a m i k , K a s s e l University, G e r m a n y , for 11 Mellotte, H. and Richard, J. R. In 'Energy from Biomass', EUR
useful discussions a n d v a l u a b l e suggestions. T h e y also 8245, Commission of the European Communities, 1983,
t h a n k D r K. I. G a s n a s e k a r a n a n d D r A. S u n d a r e s a n , pp. 523-529
R e s e a r c h Scholars in the C h e m i s t r y D e p a r t m e n t , I.I.T., 12 Vovelle, E. and Mellotte, H. In 'Energy from Biomass', EUR
8245, Commission of the European Communities, 1983, pp.
B o m b a y , for their help in o b t a i n i n g the X R D patterns. 925-930
13 Ward, S. M. and Braslaw, J. Combust. Flame 1985 21,261
14 Klose, W. and Stuke, V. Fuel Process. TechnoL 1993, 36, 283
REFERENCES 15 Nikkhah, K., Bakhshi, N. N. and MacDonald, D• G. In 'Energy
from Biomass and Wastes XVI' (Ed. D. L. Klass), Institute of
Antal, M. J., Jr. In 'Advances in Solar Energy', Vol. 2 (Eds Gas Technology, Chicago, 1993
K. W. Boer and J. A. Duffle), American Solar Energy Society, 16 McGee, B., Norton, F., Snape, C. E. and Hall, P. J. Fuel 1995,
New York, 1983, pp. 175-239 74, 28

Fuel 1996 Volume 75 Number 8 997

Pyrolysis characteristics of biomass and biomass components: K. Raveendran et al.

17 Khan, M. R., Heshieh, F. Y. and Heaclky, L. Am. Chem. Soc. 23 Segal, L., Creely, J. J., Martin, A. E., Jr and Conrad, C. M. Tex.
Div. Fuel Chem. Preprints 1989, 34, 1167 Res. J. 1959, 21, 786
18 Saxby, J. D. and Sato, S. Fuel 1990, 69, 1109 24 Loeb, L. and Segal, L. Tex. Res. J. 1955, 21, 516
19 Browning, B. L. 'Methods of Wood Chemistry', Vols I & II, 25 Chatterjee, P. K. and Conrad, C. M. Tex. Res. J. 1966, 36, 487
Wiley, New York, 1970 26 Kilzer, F. J. and Broido, A. Pyrodynamics 1965, 2, 151
20 Shapino, S. and Brannock, W. W. 'Rapid Analysis of Silicate, 27 Broido, A. and Weinstein, M. Combust. Sci. Technol. 1970, 1,
Carbonate and Phosphate Rocks', US Geological Survey 279
Bulletin 1144-A, Washington, 1962 28 Weinstein, M. and Broido, A. Combust. Sci. Technol. 1970, 1,
21 Boar, P. L. and Ingram, L. K. Analyst 1970, 95, 124 287
22 Raveendran, K., Ganesh, A., Khilar, K. C. Fuel 1995, 74, 1813 29 Box, M. J. Computer J. 1965, 8, 42

998 Fuel 1996 Volume 75 Number 8