You are on page 1of 6

Fuel Vol. 76, No. 13, pp.

1277-1282, 1997
© 1997 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
ELSEVIER Plh S0016-2361(97)00106-3 0016-2361/97 $17,00+0.00

Effects of particle size, heating rate


and pressure on measurement of
pyrolysis kinetics by
thermogravimetric analysis

Veronika Seebauer, Josef Petek and Gernot Staudinger


Institut for Verfahrenstechnik, Technical University Graz, Inffeldgasse 25, A-8010 Graz, Austria
(Received 22 October 1996; revised 24 April 1997)

Experiments using varied particle size, heating rate and pressure with a high-volatile bituminous coal were
undertaken to determine possible effects of these experimental parameters on coal pyrolysis experiments. The
results show that secondary reactions can influence the results to a large extent. For the particle size in particular it
is established that the optimum is not a bed of very fine particles, since detailed gas analysis by FT-i.r.
spectroscopy reveals a clear trend in the composition of the product gas which can be related to secondary tar
cracking in the bed. Furthermore, it is concluded that vapour pressure equations cannot be used directly to describe
the evolution of tar. © 1997 Elsevier Science Ltd.

(Keywords: coal pyrolysis kinetics; thermogravimetric analysis; secondary reactions)

Thermogravimetric analysis (t.g.a.) is widely used to hot product gases, its housing is flushed with a separate gas
determine the rate of decomposition reactions occurring in flow. Nitrogen is used for both the gas flow through the
solid fuels under the action of heat. Several publications 1-4 balance and the reactor sweep gas. These are both controlled
have reported mass loss data obtained at constant heating by mass flow controllers. The gas stream from the reactor
rate, often combined with detailed gas analysis of the including the product gas is mixed with the balance flush
volatiles. The results of such measurements often form the and flows through a high-pressure filter to precipitate the
basis for determining kinetic parameters which are used in condensable substances which are summarized by the term
more or less elaborate models for the pyrolysis of coal, 'tar'. Because of the filter temperature of 15°C, water is not
biomass or other solid fuels 5-8. condensed. The cylindrical sample basket, - 9 mm in
Experimental techniques used by other authors for the diameter and 30 mm long, is made of Incoloy 800 wire
investigation of pyrolysis are the wire-mesh reactor, various mesh. The temperature of the sample is measured by a
kinds of entrained-flow reactor such as the drop-tube thermocouple placed upstream, a few millimetres beneath
reactor, fluidized beds or fixed beds. Each method has the sample basket. The main characteristics of the t.g.a, are
certain advantages and limitations 9. summarized in Table I and a more detailed description of its
The advantages of a t.g.a, experiment are the accurate design and operation is given elsewhere 10.¿~.
measurement of the sample temperature and the possibility
of combination with an extensive gas analysis system. Gas analysis
However, due to the large sample mass needed to achieve
good quality in gas analysis, such experiments risk being The composition of the uncondensable pyrolysis products
influenced by the operating parameters of particle size, was analysed by Fourier transform infrared (FT-i.r.)
heating rate and pressure. spectrometry and thermal conductivity hydrogen analysers.
The present paper aims at elucidating the effects of these A spectrometer with a long-path minicell (path length
parameters on the pyrolysis of coal and compares the 1.8 m) was used. The F/'-i.r. range between 4000 and
findings with models of vapour-liquid equilibrium for tar. 400 cm -I was scanned with a resolution of 1 cm -I in the
single-scan mode. Focus was placed on the quantitative
analysis of the main components, namely carbon monoxide,
EXPERIMENTAL carbon dioxide, methane, ethane, ethene, propane and
propene.
The rmogravimetric analyser ( t.g.a. ) Since one experiment yields up to 350 spectra, a quick but
The t.g.a, works in the temperature range up to 1100°C accurate method to determine the correlation of the
and pressures up to 4 MPa. As shown in Figure 1, the measured spectra with the actual concentrations had to be
sample is suspended by a microbalance located above an developed. Specific problems arose from the fact that the
electrically heated reactor. To protect the balance from the gas concentrations were rapidly changing and that certain

Fuel 1997 Volume 76 Number 13 1277


Measurement of pyrolysis kinetics by thermogravimetric analysis: V. Seebauer et al.

balanceflush (N2)

samplelock 4======1

coolant expansionvalve

sample holder ~ I1,111,11 Iill i H t° ga~analysia


in the reaet°r i ~ H
coo,an

reactor flush (N2) I ~ high-pressure


filter

Figure 1 Schematic of the thermogravimetric analyser (t.g.a.)

Table 1 Characteristics of the thermogravimetric analyser was connected to the system to check the validity of the F r -
i.r. data.
Temperature -< 1 IO0°C
Pressure - 4 MPa Samples
Heating rate 2 - 6 0 K min -~
Sample mass ~1 g The fuel investigated was Westerholt (Germany) high-
Reactor sweepgas 2 . 4 L m i n -~ N2(s.t.p.) volatile bituminous coal, the proximate and ultimate
Balance flush 1.2 L min -r N2 (s.t.p.) analyses of which are shown in Table 3. The coal samples

Table 2 Gas species analysed by FT-i.r. spectroscopy


Wavenumber (cm -t) Range 1 (ppmv) Range 2 (ppmv) Comment
CO 2172 600 6000
CO2 617 -- 3200 at high concentrations
CO2 668 315 -- at low concentrations
CH4 3086 1030 5200 influenced by other hydrocarbons
C2H6 2895 500 5200 influenced by CH4, C3H8, C3H6
C 2H 4 949 140 1500 influenced by C3H6
C 3H 8 2967 380 1500 influenced by CH 4, C2H 6, C3H6
C3H6 912 140 1500 influenced by C 2H4
H zO 1844 -- --

species interact in the absorption spectra. The spectra were were crushed in a mortar, sieved into fractions between 100
evaluated by the application of calibration curves obtained and 2000/zm and dried for 2 h at 105°C in an oven.
with special gas mixtures and from dilution series of single
gases. The peak height at a wavenumber significant for the RESULTS AND DISCUSSION
gas species considered was filtered from the multifile
containing the complete record of the spectra over time. In Effect of particle size
some cases the peak height was influenced by the presence To investigate the influence of particle size the following
of one or more other gases, so that special algorithms had to parameters were kept constant: pressure 0.1 MPa; final
be used to deconvolute the spectra. The relation between temperature 1000°C; heating rate 5 K min -~ started at
absorption and concentration was not linear in all cases. 100°C. The particle size was varied in five fractions
Table 2 gives an overview of the gases analysed by FF-i.r. between 100 and 2000/~m. A cylindrical particle with a
spectroscopy. Since symmetric molecules consisting of two diameter of 9 mm and a length of 25 mm, referred to as the
atoms are not infrared-active, the gas stream also flowed 'single particle' was also pyrolysed for comparison.
through two thermal conductivity hydrogen analysers. Two The total volatile yield was calculated from the mass loss
devices were used so as to cover a wide range of in the t.g.a. 'Detected gases' are the sum of CO, CO2,
concentration with good resolution at low concentrations. methane, ethane, ethene, propane, propene and hydrogen.
The data obtained had to be corrected for sensitivity to water 'Tar + water' is the difference between the total volatiles
and hydrocarbons. In addition, a carbon monoxide analyser and detected gases.

1278 Fuel 1997 Volume 76 Number 13


Measurement of pyrolysis kinetics by thermogravimetric analysis: V. Seebauer et al.

Table 3 Proximate and ultimate analyses (wt% daf) of Wester- 0,12


holt coal ~ I00-125~m
Moisture" 2.4
Ash h 3.4 0,09 +400-500 pm r ~/7~\
Volatiles 40.4 E ~ 8oo_1ooo~Jm ]
Fixed C 59.6 O
o
C 85.0 "5 0,06
H 5.6
O 6.7
N 1.53
S 1.15 o 0,03

"As-received basis n,'


bDry basis
0,00
0 200 400 600
Temperature (°C)
800 1000

4%
[] 100-12s.m Figure 3 Effect of particle size on formation of CO
[I l i j200-2 0.m
/I I I 11 illl j 8OO-lOOOpm 40/0 I
t111 Ililll =16oo-2ooo,m,, I I r-~l~ r-12 K/rain
[] Single particle
o /I il HEll ~-~ m3 K/rain

/l|il IIlll 120 K/rain I


;~%11111 d~ Ii1111 , 4 o K/min : I~
0% , ~ : : ' : ' ', , ; /111 /II!11 1ffll60K/min Ili~

Figure 2 Effect of particle size on yields of gaseous components 1°/


0% lillllllll
o
. .
~
. .
~ ~
.E
rlillm]
~
, --~i --' [i~S
~
, i'ililM
~
Q.
,

No significant differences in total volatiles, detected


Figure 4 Effect of heating rate on yields of gaseous components
gases or tar + water were found between the different
milled fractions. All showed a similar value of - 3 2 wt% daf
total volatile yield. For the single particle, the total volatile
yield was much higher, namely 37 wt%; in this case the revealing a sequence of three significant peaks. The curve is
detected gases decreased and tar + water increased reduced in height with increasing particle size and
dramatically. Closer examination of the evolved gases decreasing void space. The single particle shows the
showed differences between the size fractions (Figure 2): lowest rate of CO formation. Thus the formation of CO
CO, CO; and H2 decreased with increasing particle size, can be directly related to the bed porosity, which seems
whereas methane, ethane and propane increased. to play an important role in the secondary reactions of the
The initial porosity of the single particle was - 4 vol.% tar.
and therefore the residence time of the product gas in the
pores was less than in the voidage of the bed, which was in Effect of heating rate
the range 48-63 vol.% (fraction 100-125 #m, 63 vol.%; To investigate the effect of heating rate, the 400-500 #m
200-250/~m, 57 vol.%; 400-500/~m, 53 vol.%; 800- size fraction was used at a pressure of 0.1 MPa with a
1000 #m, 49 vol.%; 1600-2000/~m, 48 vol.%). The differ- temperature range of 100-1000°C. A restriction in experi-
ences in bed porosity between the smaller fractions were ments at high heating rates is the minimum time interval of
much more pronounced than between the larger fractions. 32 s between two FT-i.r. measurements. At 60 K min -1 the
According to the theory of secondary reactions, the tar is whole evolution curve of a gas species consists of only 30
cracked during transport in the pores of the coal particles, data points, which causes inaccuracies in the evaluation, but
leading to the formation of CO, methane and soot. Therefore the weight signal is registered separately every 2 s by the
less tar and more CO and methane should be observed for t.g.a.
larger particles, with longer pores. However, the opposite The yield of volatiles increases with increasing heating
was observed. On the other hand, cracking of the tar can also rate, but because of the small range of heating rates applied
take place within the fixed bed, in the voids between the the differences in ultimate yield are low. According to
particles. The larger the particles, the higher is the apparent Gibbins-Matham and Kandiyoti 12 the increase in total
density and the lower the bed porosity. Hence the larger the volatile yield is mainly due to enhanced tar production.
particles, the shorter is the residence time of the tar in the The reason for this behaviour is the existence of two
bed. The single particle shows the highest ultimate yield competing reactions both consuming the 'metaplast' phase:
because the tar is swept away by the flushing gas tar evaporation and char formation. The higher the heating
immediately and there is no time for secondary reactions rate, the less time is available for tar cracking and
on the hot surface of the particle. consequently the more tar and the less methane are
Figure 3 shows the evolution of CO versus temperature, produced.

Fuel 1997 V o l u m e 76 N u m b e r 13 1279


Measurement of pyrolysis kinetics by thermogravimetric analysis: V. Seebauer et al.

0,20 0,20
-*- 2 K/min f-t ~0.1 MPa 1
3 K/min ~. ~. 0,16 ~ 0 . 4 MPa /
0,16 o~
J~
--~'1 MPa L
E~
E~ - - 20 K/min ~ 2 MPa I
'5 v 0,12 3 ~ o,12
._~~
) 0,0,
3 MPa I
._~ ~, ~ 4 MPa j
) §o,o6 Z:2::: ,
rr 0,04
(~
tic 0,04

0,00
0,00 r ----
0 200 400 600 800 1000
200 400 600 800 100 Temperature (°C)
Temperature (°C)
Figure 8 Effect of pressure on formation of methane
Figure 5 Effect of heating rate on formation of methane

3,0
+0.1 MPa 1
2,5
40%

30%
4-
~ 2,0 "-04Pa/ I
-,,-1 MPa
~ 2 MPa
~3MPa
h
/
/
=o'O
~ 1,5
4 MPa ]

3,5 1,0
~20%
0,5
,,,,% ~ ~.
10% 0,0
200 400 600 800 1000
Temperature (°C)
0% Figure 9 Effect of pressure on formation of tar + water
Total volatiles Detected gases Tar + water

Figure 6 Effect of pressure on total yields


the kinetic parameters and the available time. Since
pyrolysis experiments at higher heating rates have a shorter
reaction time (between 100 and 1000°C), the product
4%
evolution (which does not occur instantaneously) is shifted
El0.1 MPa
to higher temperatures.
/ " 0 ' 4 MPa
3% In those cases where the rates of formation of a gas
species did not differ even though different heating rates
l~mill~ I:m mill / • 2 MPa were used, heat transfer limitation in the experimental
2% system is the most probable reason.

t]m[Illl [ N4MPa
"o

Effect of pressure
1%
The effect of pressure is the most distinct: it had strong
influence on the yields of total volatiles, detected gases
0%
[j 11111,lU =milL,I| 1=111,E,Hilll, r~,.~, I!H 1=11,~, tl I=111 and tar -t- water (Figure 6). The total volatile yield
o ~ ~ ~ ~ ~ ~ ~: decreased by - 2 0 % , from 37 to 30 wt% daf. The tar +
0 ~ ~ 0a

n o_
water fraction showed an even more distinct dependence,
since the detected gases slightly increased with increasing
Figure 7 Effect of pressure on yields of gaseous components pressure.
Pressure exerts the greatest influence on those compounds
which according to Solomon et al. 13 are related to cross-
In this investigation the sum of gases formed remained linking reactions of the metaplast. Such reaction products
more or less constant, but a shift between the species could are CO2 and methane, which increase with pressure, as can
be observed (Figure 4): hydrocarbons and hydrogen be seen in Figure 7. In contrast, ethane, ethene, propene
decreased with increasing heating rate, whereas CO2 and H2 decreased, while CO remained more or less
increased. constant. Solomon et al. 13 define two sorts of cross-linking
As an example of the effect of heating rate on the reactions. Low-temperature cross-linking takes place
evolution of all pyrolysis product gases, the evolution of between 200 and 450°C and is related to the formation of
methane is shown in Figure 5, where a shift in the peak CO2 and water. Moderate-temperature cross-linking occurs
towards higher temperatures can be observed. This shift can between 450 and 600°C, associated with the formation of
be explained by the reaction kinetics: the amount of product methane.
formed depends on the concentration of the reacting agent, The formation of methane is shown in Figure 8 as a

1280 Fuel 1997 Volume 76 Number 13


Measurement of pyrolysis kinetics by thermogravimetric analysis: V. Seebauer et a l.

function of pressure. At different pressures the peaks start at Table 4 Constants in the vapour pressure equation (Equation (1))
approximately the same temperature of 350°C, but with given by different authors
increasing pressure the formation of methane takes a longer oe (atm) • 7
time and becomes more pronounced at higher temperatures.
Unger and Suuberg 15 5756 255 0.586
A possible explanation for this behaviour is that at higher Niksa 16 69.2 1.6 1
pressures the tar evaporation is suppressed and thus cross- Niksa and Kerstein 17 296077 200 0.6
linking reactions yielding methane gain importance. Fletcher et al. ~4 (FGP) 87 060 299 0.59
Propane showed the same trend as methane.
The other hydrocarbons investigated did not show a
relation to cross-linking reactions, because they appeared at
the same temperature, independent of pressure. Furthermore
4 MPa
the yield of e.g. ethane decreased in approximately the same
ratio as the tar + water yield was reduced (Figure 9), so the
0.1 MPa
formation of ethane was proportional to the tar production.
This was also true for ethene and propene, which showed
0.04 MPa
the same behaviour as ethane.
Comparing the formation of methane and tar + water, it is
evident that methane and also propane originate from 0.001 MPa
sources other than tar and the other hydrocarbons.

Comparison with models of vapour-liquid equilibrium of


tar 0 100 200 300 400 500 600 700
To describe the effects of heating rate and pressure on the Molar weight (g/mole)
ultimate tar yield and on the composition of the condensable Figure 1O Molecular size range of tar tractions reaching the
pyrolysis products, several models for coal pyrolysis indicated vapour pressure in the temperature range 300-600°C,
contain mechanisms which consist of the following steps: calculated from Equation (1)
• thermal degradation of the coal structure yielding macro-
molecules or clusters of different size ('bridge-breaking', stage of evolution towards higher temperatures. According
'metaplast formation'), to the FGP equation a molecule of size 500 g m o l - 1reaches
followed by the competing steps of a partial pressure of 0.01 atm at a bulk pressure of 1 atm at a
• evaporation of the resulting small and medium-sized temperature of 458°C, whereas the same partial pressure at
molecules determined by vapour-liquid equilibria 40 atm (i.e. a vapour pressure of 0.4 atm) is reached at
and 677°C. On the other hand, tar contains a large fraction of
• condensation of the fragments to form large highly aro- components with a molecular mass < 200 g mo1-1 which
matic structures ('cross-linking', 'char formation'). already develop a considerable vapour pressure at tempera-
Fletcher et al. ~4 give a summary of several models ~5-17, tures well below 300°C.
applying a vapour pressure equation of the form: In Figure 10 the size range of molecules is shown which
according to Equation (1) reach a certain vapour pressure--
PV = e~ e x p ( - ~8(MW)7-) (1) 4, 0.1, 0.04 and 0.00l M P a - - i n the temperature range 300-
600°C. As the composition of the boiling fraction changes
dramatically with pressure, the trend which was found in
where pv denotes the vapour pressure of tar molecule i Figure 9, i.e. that the tar evolution decreases with increasing
(atm), MW is the molecular mass (g mol -j) and T is the pressure but remains at the same temperature location
temperature (K). The constants or, ~ and 3' given by the (between 300 and 600°C at a heating rate of 5 K min-I),
authors concerned are listed in Table 4. Fletcher et al. cannot be explained by a simple evaporation mechanism.
showed that their model (FGP) performed best in compar- Therefore other mechanisms have to be included which
ison with measured values of vapour pressures of narrow- help to resolve these discrepancies. It is obvious that at first,
boiling fractions of coal liquids and with the boiling points tar precursors have to be formed before any tar can evolve.
of organic compounds at different pressures, while other Kinetic models that include the breaking of bridges in the
equations turned out to be fitted statistically to yield the molecular structure of the coal or simply the formation of
measured molecular weight distribution of the tar. Therefore 'metaplast' succeed in the description of the onset of tar
further comparison focuses on the FGP constants. evolution, since the kinetic parameters of these scission
Figure 9 shows the rate of formation of the tar + water mechanisms determine the evolution of substances of low
fraction (calculated from the volatiles evolution measured molecular mass.
by the weight loss and the amount of gases detected in the A more complex mechanism has to be invoked to explain
gas analysers) in the pressure range 0.1-4 MPa. With the fact that the peaks of the evolution of tar shift to lower
increasing pressure a clear decrease in the total yield of this temperature with increasing pressure and that tar evolution
fraction can be observed, but also a slight shift in the onset ends at -600°C, independent of the bulk pressure.
of evolution and in the peak position towards lower Reactions among the molecules in the mobile phase which
temperature. Yet the overall temperature range in which combine the aromatic structures of the tar precursors to form
the evolution of tar 4- water takes place remains char represent the most probable mechanism that gives rise
approximately the same, i.e. between 300 and 600°C. to the phenomena observed at different pressures and
If the mechanism for the evolution of tar were simply the heating rates. At higher pressures these so-called 'cross-
evaporation of molecules of different sizes from a liquid linking' reactions lead to a higher char yield since the tar
phase, an increase in pressure should result in a shift in the precursors are kept in the mobile phase for longer and at

Fuel 1997 V o l u m e 76 N u m b e r 13 1281


Measurement of pyrolysis kinetics by thermogravimetric analysis: V. Seebauer et al.

higher temperature, so that more char is formed. On the using such vapour pressure equations always link the tar
other hand the bridge-breaking mechanism takes place in a evolution to other reactions in the metaplast phase, such as
more interconnected structure, so it can only produce bridge-breaking or cross-linking, or to overall assumptions
smaller fractions which evaporate at lower temperature. concerning mass transport (entrainment of large molecules).
High heating rates cause the evaporation of larger fragments Since the parameters of these processes are not experimen-
before the cross-linking reactions become effective, and tally accessible and thus represent only a fit to the actual
thus the total volatile yield and in particular the tar yield are outcome of the experiment, the authors suggest deriving
very sensitive to the heating rate. directly a simple kinetic model for the evolution of tar and
However, a major problem remains with the mechanisms the formation of char with the parameters heating rate and
described above: neither the process of bridge-breaking nor pressure.
the cross-linking reactions can be directly quantified in
pyrolysis experiments, since only the overall process of the
gas and tar evolution can be recorded. Of course, ACKNOWLEDGEMENTS
experimental data on e.g. solvent swelling of the coal and This work was supported by the Austrian Science Founda-
char or the molecular weight distribution of the tar can give tion FWF (Project No. S06803) and the Austrian Federal
considerable information about the amount of cross-linking Ministry of Science and Transport.
or bridge-breaking, but in the end it is the total volatile yield
of the pyrolysis experiment that is decisive. Furthermore, it
must be considered that the more parameters a model
combines, on the one hand the easier it is to conform with REFERENCES
experimental results, but on the other hand the more difficult
1 Van Krevelen, D. W., Schors, A., Bos, H., Groenewege, M.
it is to verify the assumptions concerning a single step in the P. and Westrik, R., Fuel, 1956, 35, 230.
sequence of mechanisms. 2 Van Heek, K. H., VDI-Forschungshefte, 1982, 612.
3 Pottgiesser, C., Ph.D. thesis, RWTH Aachen, 1980.
4 Tromp, P. J. J., Ph.D. thesis, University of Amsterdam,
CONCLUSIONS
1987.
The main conclusion to be drawn from this study is that 5 Wutti, R., Petek, J. and Staudinger, G., Fuel, 1996, 75, 843.
t.g.a, experiments alone cannot be used to derive kinetic 6 Solomon, P. R., Hamblen, D. G., Carangelo, R. M., Serio,
parameters for pyrolytic reactions. Although such experi- M. A. and Deshpande, G. V., Energy and Fuels, 1988, 2,
ments offer the possibility of very precise analysis of the 405.
7 Grant, D. M., Pugmire, R. J., Fletcher, T. H. and Kerstein, A.
evolving non-condensable gases, it is evident that there are
R., Energy and Fuels, 1989, 3, 175.
large effects of the bed of particles. For example, the use of 8 Niksa, S., Combustion and Flame, 1995, 100, 384.
smaller particles led to a higher extent of secondary 9 Solomon, P. R., Serio, M. A. and Suuberg, E. M., Progress
reactions, i.e. lower tar yields and higher yields of the in Energy and Combustion Science, 1992, 18, 133.
gases produced by cracking reactions (e.g. carbon dioxide, 10 MiJhlen,H.-J. and Sulimma, A., Thermochimica Acta, 1986,
methane). Nevertheless, the information about the composi- 103, 163.
tion of the primary pyrolysis products (although it may be 11 Miihlen, H.-J. and Sulimma, A., Fuel Processing
altered to a certain extent by secondary reactions) still Technology, 1987, 15, 145.
constitutes a basic fact for the modelling of pyrolysis 12 Gibbins-Matham, J. and Kandiyoti, R., Energy and Fuels,
reactions. It is suggested that the use of a single particle in a 1988, 2, 505.
13 Solomon, P. R., Serio, M. A., Deshpande, G. V. and Kroo,
t.g.a, experiment at low heating rates leads to gas analysis
E., Energy and Fuels, 1990, 4, 42.
data which are less influenced by secondary reactions. 14 Fletcher, T. H., Kerstein, A. R., Pugmire, R. J., Solum, M. S.
Future work will be dedicated to the combination of t.g.a. and Grant, D. M., Energy and Fuels, 1992, 6, 414.
results with data for other reactor types. 15 Unger, P. E. and Suuberg, E. M., American Chemical
From the data on the effect of pressure, it must be Society Division of Fuel Chemistry Preprints, 1983, 28, 278.
concluded that the vapour pressure equations cited above 16 Niksa, S., AIChE Journal, 1988, 34, 790.
cannot be directly used to describe the evolution of tar as a 17 Niksa, S. and Kerstein, A. R., Energy and Fuels, 1991, 5,
function of pressure and heating rate. Actually the models 647.

1282 Fuel 1997 Volume 76 Number 13

You might also like