You are on page 1of 9

Fuel 82 (2003) 1139–1147

www.fuelfirst.com

Pyrolysis of Miscanthus Giganteus and wood pellets: TG-FTIR


analysis and reaction kineticsq
W. de Jonga,*, A. Pironea, M.A. Wójtowiczb
a
Department of Mechanical Engineering and Marine Technology Section Thermal Power Engineering, Delft University of Technology,
Mekelweg 2, NL-2628 CD, Delft, The Netherlands
b
Advanced Fuel Research, Inc., 87 Church Street, East Hartford, CT 06108-3728, USA
Received 15 January 2002; revised 17 October 2002; accepted 11 December 2002; available online 15 January 2003

Abstract
Characterisation of two biomass fuels (pelletised Miscanthus Giganteus and wood) was performed using thermogravimetric analysis with
measurement of products by means of Fourier transform infrared spectroscopy (TG-FTIR). Three heating rate profiles were applied (10, 30
and 100 8C/min), with a final temperature of 900 8C. HCN and HNCO were found to be the major N-products, while the NH3 fraction was
detected to a minor extent. Kinetic parameters were obtained from the TG-FTIR results using a model based on parallel first-order reactions
with a Gaussian distribution of activation energies. On the basis of the above kinetic analysis and product yields, input files for the functional
group – devolatilisation, vaporisation, cross-linking biomass-pyrolysis model were prepared. The fit of model parameters to TG-FTIR
product-evolution data was found to be generally good, but the model-predicted yields for some species did not fit experimental data at all
heating rates. Further improvements in the model are needed to resolve above difficulty.
q 2003 Elsevier Science Ltd. All rights reserved.
Keywords: Biomass pyrolysis; TG-FTIR characterisation; Kinetic analysis; Functional group–devolatilisation, vaporisation, cross-linking model

1. Introduction nitrogen oxides [11]. The reduction of such harmful


emissions significantly relies on the knowledge of early
Both the growing awareness of the decreasing avail- nitrogen evolution from biomass pyrolysis, which is found
ability of fossil fuels and the increasing pressure on the to be the primary stage for both gasification and combustion
environment from production and combustion of fossil processes involving coal or biomass-type fuels. However, to
fuels, have nowadays led to a deeper interest in sustainable date, few detailed kinetic data are available in the accessible
heat and power generation using biomass [1]. In particular, literature on the evolution of individual products during
biomass gasification can form a solid basis of new flash pyrolysis of biomass. Even in the case of low heating
promising energy conversion technologies such as fuel rate pyrolysis, such data are limited. The lack of data,
cells and integrated gasification combined cycle (IGCC) combined with the large variety and diversity of biomass
installations. However, the produced gas from biofuel-fed feedstocks, is one of the main sources of inaccuracies and
gasifiers often includes significant amounts of fixed nitrogen difficulties experienced when modelling emission behaviour
species, mainly ammonia (NH3) and hydrogen cyanide of biomass thermal conversion processes.
(HCN), which are released from biomass-bound nitrogen In the present research study pyrolysis experiments for
(fuel-N) during biomass devolatilisation, occurring at the two different types of biomass (pelletised Dutch Miscanthus
Giganteus and wood) were performed. Pelletised fuels were
initial pyrolysis stage [2 – 17]. These gas-phase species pass
chosen for this study as a result of their main advantage of
through end-use systems of the product gas, where they can
high energy density on a volumetric basis, which makes
poison catalysts, fuel cell internals or may undergo further
transport costs and investment costs for fuel storage and
oxidation and be emitted as pollutant species, mainly
process feeding smaller than in the case of non-pelletised
* Corresponding author. Tel.: þ 31-15-2789476; fax: þ 31-15-2782460. fuels [18].
E-mail address: w.dejong@wbmt.tudelft.nl (W. de Jong). The pyrolysis analysis was performed by means of a TG-
q
Published first on the web via Fuelfirst.com –http://www.fuelfirst.com FTIR instrument, which combines thermogravimetric
0016-2361/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0016-2361(02)00419-2
1140 W. de Jong et al. / Fuel 82 (2003) 1139–1147

analysis (TGA) with evolved product analysis by Fourier relatively fast carrier-gas flow rates so that the volatile
transform infrared (FTIR) spectroscopy. The use of low products of pyrolysis are swept into the gas cell immediately
heating rate experiments for the study of pyrolysis kinetics after they are formed. It is presumed, therefore, that the
in general, and TGA in particular, is well established with a gases quantified by the FTIR analyser are essentially
number of papers and books having been published on the products of primary pyrolysis. Specifically, for the analysis
subject (see for example Refs. [19 – 43] and references reported in this study, approximately 13 – 16 mg of the
quoted therein). The advantages and drawbacks of the biomass sample (particle size , 500 mm) was heated in
technique are discussed in an extensive review by Solomon helium at 10 8C/min, first to 80 8C to dry the sample for
et al. [39]. In general, successful extrapolations of kinetic 20 min, and then to 900 8C for pyrolysis. Upon reaching
expressions to high heating rates are expected. 900 8C and holding the temperature for 3 min, the sample
Employing the TG-FTIR system, kinetic rates for species was immediately cooled to 250 8C over a 20 min period.
evolution were obtained under low heating rate conditions. After cooling, a small flow of oxygen was added to the
In particular, kinetic rate parameters and pool sizes of helium sweep gas and the temperature was ramped to
species precursors were fitted into experimental data in 900 8C at 30 8C/min to combust the remaining char. This
order to convert the TG-FTIR results into input files to be profile was repeated at pyrolysis heating rates of 30 and
used within the functional group – devolatilisation, vapor- 100 8C/min for the samples, the elemental analysis of which
isation, cross-linking (FG –DVC) biomass pyrolysis model is reported in Table 1. In all experiments, FTIR absorbance
[44]. It is worth to emphasise that these rates, after being spectra were obtained every 30 s. Concentrations of all the
extrapolated, can be used to predict via the FG –DVC volatiles species, except for tar, showing absorption in the
model, the conversion behaviour of main compounds at infrared spectrum were obtained using quantification
higher heating rates and temperatures in practical conver- routines based on calibration runs performed with pure
sion processes. compounds. Tar evolution patterns and yields were
determined by difference using the sum of gases quantified
by FTIR and the balance curve obtained thermogravime-
2. Experimental section trically. The instrument, however, can quantify tars, but this
has so far only been worked out for coal samples. As the
2.1. TG-FTIR experiments infrared spectra of biomass tars differ significantly from coal
tars, a generic reference spectrum for biomass tars has not
been implemented yet.
The TG-FTIR system used in this study consists of a
sample suspended from a balance in a helium gas stream
within a furnace. As the sample is heated, the evolving 2.2. Experimental results and discussion
volatile products are carried out of the furnace directly into a
5.1 cm diameter gas cell (heated to 155 8C), where the A summary of the TG-FTIR experiments for wood
volatiles are analysed by FTIR spectroscopy. The system pellets (WP) and Miscanthus Giganteus (MG) is presented
continuously monitors: (1) the time dependent evolution of in Table 2. Observations resulting from the analysis of TG-
the gases; and (2) weight of the non-volatile material FTIR data for the biomass samples are discussed here.
(residue). The apparatus and its functional principles have The practically constant char yields at different heating
been described elsewhere [45,50]. A detailed description of rates could imply that cross-linking reactions are relatively
the TG-FTIR technique and some of the applications can be unimportant, at least under the conditions used in this study.
found in Refs. [45 – 49,51]. The instrument and the It could also mean that the major factors determining char
measurements were designed in such a way so that formation, i.e. bond breaking and cross-linking, are in
secondary reactions are minimised. This is accomplished balance with each other. Tar yields increase as the heating
through the use of small sample and particle sizes as well as rate increases. This happens at the expense of decreasing

Table 1
Elemental analysis for the two different biomass types applied in this study. Data for WP (‘A-quality’ purchased from Labee company, The Netherlands) were
obtained from experiments performed at IVD, University of Stuttgart (Germany). Data for pelletised Dutch MG are averaged values from analyses carried out
at IVD, at KTH University (Sweden) and at the Energy Research Centre of The Netherlands (ECN)

Sample Moisturea Asha Cb Hb Nb Sb Ob,c

Wood pellets 8.2 0.26 52.1 5.2 0.33 0.11 42.2


Miscanthus 8.7 ^ 1.47 2.5 ^ 0.44 49.4 ^ 1.07 5.5 ^ 0.37 0.64 ^ 0.14 0.16 ^ 0.06 44.0 ^ 0.77
a
As-received basis.
b
Dry and ash free basis.
c
Determined by difference.
W. de Jong et al. / Fuel 82 (2003) 1139–1147 1141

Table 2 Bagasse and straw, while a value of 21 wt% for Miscanthus


Summary of TG-FTIR experiments for WP and pelletised Dutch MG. All straw.1 However, precise data for MG have not been
yields are expressed in weight percent of the original biomass on a dry, ash-
determined, thus making it unsure whether or not the
free basis
differences in char yields can be related unequivocally to the
Sample WP WP WP MG MG MG diverse lignin content of the biomass types.
Comparable fractions of HCN, HNCO and NH3 are
Heating rate (8C/min) 10 30 100 10 30 100 observed, although the comparison of the data in Table 2
Volatile matter (from TGA) 86.2 86.2 86.2 80.2 79.4 81.9
does not reveal a consistent pattern. Studies accomplished
Char 13.8 13.8 13.8 19.8 20.6 18.1
Tar 37.2 38.8 48.4 18.1 21.9 28.3 for biomass conversion are presented in Refs. [11,13,54],
Methane (CH4) 1.27 1.30 1.15 1.18 1.14 1.16 showing that at low heating rates NH3 is generally the
Water (pyr.) (H2O) 10.6 16.0 13.6 17.7 18.5 20.6 dominant N-product. The pyrolysis experiments considered
Carbon monoxide (CO) 7.48 6.77 5.10 8.55 7.15 5.65 in this study were performed indeed at low heating rates,
Carbon dioxide (CO2) 5.90 5.08 4.88 10.4 9.43 9.11
obtaining results in disagreement with the mentioned
Ethylene (C2H4) 0.15 0.23 0.09 0.52 0.27 0.13
Hydrogen cyanide (HCN) 0.09 0.09 0.05 0.22 0.14 0.11 references as HCN yields were found to be mostly higher
Ammonia (NH3) 0.00 0.02 0.03 0.13 0.10 0.11 than those of NH3 (Table 2). According to Ref. [13], HCN
Isocyanic acid (HNCO) 0.11 0.02 0.05 0.20 0.11 0.07 formation may go to completion much more rapidly than
Carbonyl sulphide (COS) 0.18 0.11 0.00 0.31 0.16 0.11 that of NH3, which is found to be the main N-product when
Sulphur dioxide (SO2) 0.17 0.12 0.40 0.00 0.00 0.00
secondary reactions are considered in conversion processes
Formaldehyde (CH2O) 3.66 2.93 3.15 1.12 1.12 1.16
Acetaldehyde (CH3CHO) 9.28 8.11 3.97 10.2 9.02 7.70 characterised by longer residence times. This could partly
Methanol (CH3OH) 0.98 0.74 0.64 1.42 1.13 1.07 explain the high concentrations of HCN found in this study.
Formic acid (HCOOH) 2.68 1.73 0.82 2.12 1.66 0.37 In fact, TG-FTIR experiments were designed to minimise
Acetic acid (CH3COOH) 2.78 2.37 2.28 4.24 3.59 3.35 secondary reactions.
Phenol (C6H5OH) 1.46 0.65 0.73 2.31 1.83 1.99
Acetone (CH3OCH3) 2.05 2.15 1.63 1.80 1.86 2.10

3. Modelling section
yields of lighter gaseous species (mostly CO, CO2 and 3.1. Kinetic analysis
acetaldehyde). This can be explained by differences in
kinetic effects associated with the release of tar and light A reasonable starting assumption is that the reaction rates
species. Apparently, at higher heating rates, there is not to be determined follow first-order kinetics with distributed
enough time allowed for the evolution of light species and activation energy of width s (Gaussian distribution is
thus, tar fragments leaving the biomass carry with them assumed). In particular the rate constant ki for release of the
precursor material that could potentially result in light gas ith functional group (also called: pool) is given by an
compound formation. The heating rate does not have a Arrhenius expression of the form [55]
significant effect on the yields of the following species:
methane, formaldehyde and acetone. This behaviour ki ¼ Ai expðð2Ei ^ si Þ=RTÞ ð1Þ
suggests that the above species may have different where Ai is the pre-exponential factor, Ei is the activation
precursors than CO, CO2 and acetaldehyde. energy and si the width of distribution in activation
Furthermore, a potential relation between char formation energies.
and CH4, CH2O and CH3OCH3 release could be advanced The technique usually employed at AFR to process TG-
since their relatively constant yields when varying the FTIR data for determining kinetic parameters, is the non-
heating rates. However, these suppositions have to be isothermal method based on the temperature at which the
further investigated. The effect of the heating rate is more rate of volatile evolution is maximum (Tmax) [56,57].
complex (e.g. a maximum or a minimum is observed) in the Typically, Tmax is measured at several heating rates (in our
case of the following species: ethylene (WP), phenol and case at 10, 30 and 100 8C/min) and the activation energy E
water (WP). and the pre-exponential factor A can be extracted from the
Pyrolysis of Miscanthus produces higher yields of char following equation [56]
and lower yields of volatile matter as compared to wood. 2
This different behaviour may be explained in terms of lnðM=Tmax Þ ¼ lnðAR=EÞ 2 E=ðRTmax Þ ð2Þ
different biochemical composition between the two differ- where M represents the varying heating rate.
ent biomass types. Generally, in biomass, lignin shows a While the value of E is accurately determined even for
relatively large contribution to the production of fixed wide distributions, the value of A usually requires a slight
carbon than other main constituents such as cellulose or adjustment (typically within a factor of two) [56]. The width
hemi-cellulose. The lignin content in WP (pine wood) is of the distribution, s, can then be determined from the width
about 23 wt% [52]. In Ref. [53], values between 10 and
1
40 wt% are presented for various herbaceous species such as http://www.ecn.nl/phyllis
1142 W. de Jong et al. / Fuel 82 (2003) 1139–1147

Table 3 of the peak representing the rate of weight loss. In the Tmax
Kinetic parameters and precursor pool size (yields) for main and N-species method, some difficulties can be encountered when peaks
determined for WP and pelletised Dutch MG. Numbers before species
are not well resolved; in such cases, substantial shift in Tmax
names indicate precursor-material sources (pools). Some species evolve as
a single peak (one source), the evolution of other species is bimodal or can occur. Another limitation is associated with the
trimodal (two or three sources) presence of small, multiple maxima superimposed on a
broader peak, i.e. when the assumption of the first order
Sample Precursor pool A (s21) E/R (K) s/R Yield per
kinetics is not fully supported. In this case, the applicability
identifier per (K) precursor
evolved pool of the Tmax method is reduced. Once A, E and s values are
species (wt% daf) determined, the sizes of precursor pools (i.e. the concen-
trations of the precursor species) for individual species are
Wood pellets 1-CO2 5.20 £ 1011 18,300 900 0.95 determined by adjusting the simulated peak heights so that
2-CO2 5.20 £ 1011 20,500 500 3.10 they best fit the TG-FTIR data. If necessary, minor
3-CO2 2.80 £ 1012 24,830 900 0.90
adjustments to A, E and s are made to improve the fit to
1-CO 5.20 £ 1011 20,300 500 1.90
2-CO 6.50 £ 1012 26,300 3200 3.05 the data. In the present study, an attempt was made to
3-CO 2.80 £ 1012 35,930 3300 1.80 employ the Tmax method, but limited data resolution,
1-H2O 5.20 £ 1011 19,900 1200 11.94 especially at 100 8C/min, caused significant uncertainties
2-H2O 4.40 £ 1012 26,000 2500 3.00 in Tmax determination. For this reason, A, E and s values
1-CH4 6.50 £ 1011 19,200 800 0.05
were fitted into experimental data using a trial-and-error
2-CH4 6.50 £ 1011 22,500 800 0.60
3-CH4 3.10 £ 1012 27,700 2100 0.60 approach. Care was taken to ensure that pre-exponential
1-Tars 5.20 £ 1011 20,500 500 38.00 factors were generally consistent with the transition-state
1-HCN 2.80 £ 1012 21,230 1900 0.05 theory ðA < 1011 – 1016 s21 Þ: The results from the kinetic
2-HCN 2.80 £ 1012 27,000 2500 0.04 analysis for the biomass samples considered in this study are
1-HNCO 2.80 £ 1012 17,930 900 0.02
given in Table 3. In addition to the kinetic parameters
2-HNCO 2.80 £ 1012 28,930 3800 0.02
1-C2H4 2.80 £ 1012 25,930 800 0.20 explained in Eq. (1), Table 3 contains estimated sizes of
1-CH3OH 5.20 £ 1011 17,800 800 0.40 precursor pools (sources of the precursor material) for each
2-CH3OH 2.30 £ 1011 21,030 700 0.39 major peak and shoulder observed in TG-FTIR data. A more
1-CH2O 5.20 £ 1011 18,400 500 1.20 detailed discussion of that concept is given in Section 3.2. In
2-CH2O 5.20 £ 1011 20,500 200 2.10
the case of WP, no kinetic data for ammonia release were
1-CH3CHO 5.20 £ 1011 20,400 500 7.00
1-HCOOH 5.20 £ 1012 20,800 800 1.70 determined as the concentrations of this component were
1-CH3COOH 5.20 £ 1011 19,300 900 2.30 found to be rather low and noisy.
1-C6H5OH 7.40 £ 1018 32,500 2500 0.75
1-CH3OCH3 5.20 £ 1012 20,900 800 1.90 3.2. FG –DVC model results and discussion
Miscanthus 1-CO2 5.20 £ 1011 16,300 900 0.98
2-CO2 5.20 £ 1011 19,500 500 7.50 A general description of the FG – DVC model for coal-
3-CO2 2.80 £ 1012 24,830 900 0.93 pyrolysis can be found in Refs. [44,55,58]. The FG – DVC
1-CO 5.20 £ 1011 19,550 400 3.90
model was originally developed by AFR to describe coal
2-CO 6.50 £ 1012 26,300 3200 1.85
3-CO 2.80 £ 1012 35,930 3300 1.40 thermal decomposition, by predicting product distribution,
1-H2O 5.20 £ 1011 19,100 400 14.14 extract yields, cross-link density, molecular weight distri-
2-H2O 4.40 £ 1012 24,000 2500 4.50 bution and fluidity as a function of coal rank, heating rate
1-CH4 6.50 £ 1011 19,500 600 0.15 and pressure. The FG – DVC code combines a FG model for
2-CH4 6.50 £ 1011 22,300 500 0.23
gas evolution and a statistical depolymerisation, vaporisa-
3-CH4 3.10 £ 1012 26,900 1600 0.76
1-Tars 5.20 £ 1011 19,700 500 21.34 tion and cross-linking (DVC) model for tar and char
1-HCN 2.80 £ 1012 20,530 1900 0.07 formation. The FG subroutine is used to describe the gas
2-HCN 2.80 £ 1012 28,000 3100 0.07 evolution and the elemental and functional group compo-
1-HNCO 2.80 £ 1012 16,930 900 0.02 sitions. The DVC subroutine is employed to determine the
2-HNCO 2.80 £ 1012 25,930 2800 0.09
amount and molecular weight of macromolecular frag-
1-NH3 5.20 £ 1011 18,600 800 0.07
2-NH3 5.20 £ 1011 25,000 2500 0.03 ments. The lightest of these fragments evolve as tar.
1-C2H4 2.80 £ 1012 24,930 800 0.21 The model was later extended to biomass feedstock [44].
1-CH3OH 5.20 £ 1011 17,500 500 0.20 There are two major differences between the coal and
2-CH3OH 2.30 £ 1011 19,930 400 0.95 biomass version of the model. First, the DVC part of the
1-CH2O 5.20 £ 1011 17,100 500 0.33
model is de-emphasised in the biomass version in favour of
2-CH2O 5.20 £ 1011 19,500 200 0.82
1-CH3CHO 5.20 £ 1011 19,500 300 9.02 the FG description of gas and tar evolution. Secondly,
1-HCOOH 5.20 £ 1012 20,500 500 1.66 unlike in the coal version, where tar competes for the
1-CH3COOH 5.20 £ 1011 18,700 900 3.59 precursor material with gaseous species, tar is treated just
1-C6H5OH 7.40 £ 1018 29,100 2100 2.03 like any other volatile species in the biomass model. The
1-CH3OCH3 5.20 £ 1012 20,500 800 1.86
evolution of each species is associated with the peak found
W. de Jong et al. / Fuel 82 (2003) 1139–1147 1143

Fig. 1. Comparison between TG-FTIR measurements and FG–DVC modelling results for rates and yields obtained for the pyrolysis of WP. Heating rate was
30 8C/min. Lines without markers: FG–DVC model predictions for yields (secondary axis) and rates (primary axis). Lines with B markers: experimental TG-
FTIR measurements for yields (secondary axis) and rates (primary axis).
1144 W. de Jong et al. / Fuel 82 (2003) 1139–1147

Fig. 2. Comparison between TG-FTIR measurements and FG– DVC modelling results for rates and yields obtained for the pyrolysis of MG. Heating rate was
30 8C/min. Lines without markers: FG –DVC model predictions for yields (secondary axis) and rates (primary axis). Lines with B markers: experimental TG-
FTIR measurements for yields (secondary axis) and rates (primary axis).
W. de Jong et al. / Fuel 82 (2003) 1139–1147 1145

in the TG-FTIR measurement. For some compounds, fitting rate. Yields of char, CH4, CH2O and CH3OCH3 were
the wide pyrolysis peaks by the FG-submodel required more independent of the heating rates applied. N-species release
than one single source (or pool). For instance, for CO three generally showed a higher HCN yield compared to NH3.
precursor pools were used (Table 3). Also, significant amounts of HNCO were detected.
Each source evolves according to However, regarding nitrogen species, the difficulty of
dWi =dt ¼ ki Wi ðcharÞ ð3Þ measuring low concentrations should be underlined:
relatively high noise levels might affect the significance of
where Wi is the gas species, Wi(char) is the amount of the these values.
functional group source remaining in the char and ki is the Kinetic and composition (pool size) parameters were
rate constant given by Eq. (1). obtained for each volatile species and used to simulate TG-
Input files for the FG – DVC biomass model for the two FTIR results by means of the biomass version of the FG –
different biomass samples involved in this work were DVC model. N-species evolutions for Miscanthus were
developed by AFR according to the TG-FTIR experiments predicted in better agreement than for wood, due to the low
and kinetic analysis described above. A set of TG-FTIR data N-content in the wood and related detectability of the FTIR.
as well as FG – DVC model fits to these data using the input The locations of the peak for the pyrolysis rates for main
files mentioned are presented in Figs. 1 and 2. For all the species (CO, CO2, H2O, CH4) were correctly simulated by
experiments, the balance curve is plotted as a function of the model, although it was difficult to fit model-predicted
time, with a thermocouple measured temperature given on yields to experimental data at all heating rates. It is expected
the primary axis. Evolution and weight curves are presented that a new version of the FG – DVC biomass model, to be
for main and N-gas species evolving from the samples. The available in 2002, will include the mechanism for the
evolution curves are given in ‘weight percent per minute’ competition between gas and tar, improving the predictive
(primary axis), while product yield curves, which are the capabilities of the model. Furthermore, in order to validate
integral of the evolution curves and show the total amount the rate parameters and pool sizes presented in this study, it
of the species evolved, are presented in the units ‘weight is recommended to carry out a comparison between model
percent’ (secondary axis). predictions and experimental results under different pyrol-
In general, the model correctly simulates the locations of ysis conditions, especially at higher heating rates. For this
peak pyrolysis rates for individual species (CO, CO2, etc.). purpose, e.g. a drop tube furnace or a heated grid reactor can
For several species, however, it was difficult to fit model- be applied.
predicted yields to experimental data at all heating rates. In
these cases, a compromise had to be made by having the fits
underpredict experimental data at one heating rate and
overpredict at other. We believe that above difficulty was
caused by the inability of the biomass model to account for Acknowledgements
the competition between gas and tar for the available
precursor pools. It was noted that at high heating rates there The Dutch Agency for Energy and Environment,
was not enough time allowed, within specific temperature NOVEM, is acknowledged for funding the research in the
windows, for the evolution of light species. As a framework of biomass gasification modelling via the project
consequence, tar fragments leaving the biomass carried ‘Biomassa vergassing-CFB-STEG’ (contract EWAB
with them precursor material that could potentially result in 355298/2010). Mr Drenth of Agromiscanthus (The Nether-
light-gas formation. lands) is thanked for supply of the Miscanthus Giganteus
In particular, N-species evolutions for Miscanthus pellets. Finally, Dipl.-Ing. (FH) Bernd Janisch at I.V.D.
obtained from the FG –DVC model are in rather good (University of Stuttgart) is acknowledged for performing
agreement with TG-FTIR measurements. On the other hand, proximate and ultimate analyses of the fuels.
there are some discrepancies for the fixed nitrogen
compounds evolution for WP pyrolysis, probably due to
the TG-FTIR limits of detectability for such low References
concentrations.
[1] Johansson TB, Kelly H, Reddy AKN, Williams RH. Renewable
energy: sources for fuel and electricity. California: Island Press; 1993.
4. Conclusions ISBN 1-55963-138-4.
[2] Björkman E, Larsson C. The formation of nitrogen compounds
Biomass pyrolysis kinetics of wood and Dutch MG at gasification conditions, with and without internal catalysts.
TPS-report 96/48 Termiska Processer AB, Studsvik, Sweden; 1996.
pellets were studied using TG-FTIR analysis. Tar release
[3] de Jong W, Ünal Ö, Andries J, Hein KRG, Spliethoff
appeared to increase with increasing heating rate. On the H. Pressurised gasification experiments and modelling of biomass
other hand, light gaseous species (mostly CO, CO2 and and fossil fuels in fluidised bed gasifiers with hot gas ceramic
CH3CHO) showed an adverse dependence on the heating filters. Proceedings of the Sixth International Conference of
1146 W. de Jong et al. / Fuel 82 (2003) 1139–1147

Technologies and Combustion for a Clean Environment, vol. 3, [24] Ciuryla VT, Weimer RF, Bivans DA, Motika SA. Ambient-pressure
Porto, Portugal; 2001. p. 1327–37. thermogravimetric characterization of four different coals and their
[4] El Asri R, Komonov AA, De Ruyck J. Fluidized bed flash pyrolysis of chars. Fuel 1979;58:749– 54.
biomass. Proceedings of the Second Olle Lindström Symposium on [25] Gaur S, Reed TB. Thermal data for natural and synthetic fuels. New
Renewable Energy–Bioenergy, Stockholm, Sweden; 1999. p. 121– 7. York, USA: Marcel Dekker; 1998.
[5] Glarborg P, Alzueta MU, Dam-Johansen K. Kinetic modelling of [26] Coats AW, Redfern JP. Kinetic parameters from thermogravimetric
hydrocarbon/nitric oxide interactions in a flow reactor. Combust data. Nature 1964;201:68–9.
Flame 1998;115:1–27. [27] Gyulai G, Greenhow EJ. A new integral method for the kinetic
[6] Hämäläinen JP, Aho MJ, Tummavuori JL. Formation of nitrogen analysis of thermogravimetric data. J Therm Anal 1974;6:279– 91.
oxides from fuel-N through HCN and NH3: a model compound study. [28] Doyle J. Kinetic analysis of thermogravimetric data. J Appl Polym Sci
Fuel 1994;73:1894 –8. 1961;5(15):285– 92.
[7] Hämäläinen JP, Aho MJ. Conversion of fuel nitrogen through HCN [29] Zsako J. Kinetic analysis of thermogravimetric data. III. Improvement
and NH3 to nitrogen oxides at elevated pressure. Fuel 1996;75: of Doyle’s isothermal method. J Therm Anal 1970;2:145– 9.
1377–86. [30] Zsako J. Kinetic analysis of thermogravimetric data. VI. Some
[8] Kilpinen PT, Leppälahti JK, Coda Zabetta EG, Hupa M. Gas-phase problems of deriving kinetic parameters from TG curves. J Therm
conversion of NH3 to N2 in gasification. Part I: a kinetic modelling Anal 1973;5:239–51.
study on the potential of the method. IFRF Combust J 1999; Article [31] Satava V, Skvara F. Mechanism and kinetics of the decomposition of
Number 199901, ISSN 1562-479X. solids by a thermogravimetric method. J Am Ceram Soc 1969;52(11):
[9] Kilpinen PT, Coda Zabetta EG, Hupa M. A kinetic modelling study on 591 –5.
the potential staged combustion in gas turbines for the reduction of [32] Freeman ES, Carroll B. The application of thermoanalytical
nitrogen oxide emission from biomass IGCC plants. Energy Fuels techniques to reaction kinetics. The thermogravimetric evaluation
2000;14:751 –61. of the kinetics of the decomposition of calcium oxalate monohydrate.
[10] Kilpinen PT, Coda Zabetta EG. Gas-phase conversion of NH3 to N2 J Phys Chem 1958;62:394 –7.
in gasification. Part II: testing the kinetic model. IFRF Combust J [33] Varhegyi G. Integration of the rate constant and linearization of the
2001; Article Number 200104, ISSN 1562-479X. kinetic equations in non-isothermal reaction kinetics. Thermochim
[11] Leppälahti J, Koljonen T. Nitrogen evolution from coal, peat and Acta 1978;25:201–7.
wood during gasification: literature review. Espoo, Finland: VTT [34] Antal MJ, Friedman HL, Rogers FE. Kinetics of cellulose pyrolysis in
technical research centre of Finland; 1994. nitrogen and steam. Combust Sci Technol 1980;21:141–52.
[12] Leppälahti J. Behaviour of fuel-bound nitrogen in gasification and in [35] Arseneau DF. Competitive reactions in the thermal decomposition of
high temperature NH3 removal processes. Espoo, Finland: VTT cellulose. Can J Chem 1971;49:632 –8.
technical research centre of Finland; 1998. [36] Domburgs G, Rossinskaya G, Sergeeva V. Study of thermal stability
[13] Li C-Z, Tan LL. Formation of NOx and SOx precursors during the of b-ether bonds in lignin and its models. Proceedings of the Fourth
pyrolysis of coal and biomass. Part I. Effects of reactor configuration International Conference on Thermal Analysis, Budapest, Hungary,
on the determined yields of HCN and NH3 during pyrolysis. Fuel vol. 2; 1974. p. 211–19.
2000;79:1883 –9. [37] Klein MT, Virk PS. Modelling of lignin thermolysis. ACS Div Fuel
[14] Li C-Z, Tan LL. Formation of NOx and SOx precursors during the Chem Prepr 1981;26(3):77–88.
pyrolysis of coal and biomass. Part III. Further discussion on the [38] Friedman HL. Kinetics of thermal degradation of char-forming
formation of HCN and NH3 during pyrolysis. Fuel 2000;79: plastics from thermogravimetry. Application to a phenolic plastic.
1899–906. J Polym Sci, Part C 1964;6:183–95.
[15] Padban N. PFB air gasification of biomass, investigation of product [39] Solomon PR, Serio MA, Suuberg EM. Coal pyrolysis: experiments,
formation and problematic issues related to ammonia, tar and alkali. kinetic rates and mechanisms. Prog Energy Combust Sci 1992;18:
Doctoral dissertation. Department of Chemical Engineering II, Lund 133 –220.
University, Sweden; 2000. [40] Miura K. A new and simple method to estimate f ðEÞ and k0 ðEÞ in the
[16] Vriesman P, Heginuz E, Sjöström K. Biomass gasification in a distributed activation energy model from three sets of experimental
laboratory-scale AFBG; influence of the location of the feeding point data. Energy Fuels 1995;9:302–7.
on the fuel-N conversion. Fuel 2000;79:1371 –8. [41] Miura K, Maki T. A simple method for estimating f ðEÞ and k0 ðEÞ in
[17] Zhou J, Masutani SM, Ishimura DM, Turn SQ, Kinoshita CM. Release the distributed activation energy model. Energy Fuels 1998;12:
of fuel-bound nitrogen during biomass gasification. Ind Engng Chem 864 –9.
Res 2000;39:626–34. [42] Miura K, Maki T. Simplified method to estimate f ðEÞ in the
[18] Reed TB, Bryant B. Densified biomass a new form of solid fuel with a distributed activation energy model for analyzing coal pyrolysis
forward for the 21st century. Golden, Colorado, USA: The Biomass reaction. J Chem Engng Jpn 1998;31(2):228–35.
Energy Foundation (BEF); 1978. [43] Maki T, Miura K. Analysis of pyrolysis reactions of various coals
[19] van Krevelen DW, van Heerden C, Huntjens FJ. Physicochemical including argonne premium coals using a new distributed activation
aspects of the pyrolysis of coal and related organic compounds. Fuel energy model. Energy Fuels 1997;11:972–7.
1951;30:253 –9. [44] Chen Y, Charpenay S, Jensen A, Wójtowicz MA, Serio MA.
[20] Jüntgen H, van Heek KH. Progress made in the research of bituminous Modeling of biomass pyrolysis. 27th Symposium (International) on
coal. Paper presented at the annual meeting of the DGMK, Salzburg; Combustion, Pittsburgh, PA: The combustion Institute; 1998. p.
1968 (translated by Belov and Associates, Denver, CO, APTIC-TR- 1327–34.
0779 (1970)). [45] Carangelo RM, Solomon PR, Gerson DJ. Application of TG-FTIR to
[21] Burnham AK, Oh MS, Crawford RW, Samoun AM. Pyrolysis study hydrocarbon structure and kinetics. Fuel 1987;66:960.
of argonne premium coals: activation energy distributions and [46] Whelan JK, Solomon PR, Deshpande GV, Carangelo RM.
related chemistry. Energy Fuels 1989;3:42–55. Thermogravimetric Fourier transform infrared spectroscopy (TG-
[22] Jüntgen H. Review of the kinetics of pyrolysis and hydropyrolysis in FTIR) of petroleum source rock—initial results. Energy Fuels
relation to the chemical constitution of coal. Fuel 1984;63:731–7. 1988;2:65.
[23] Weimer RF, Ngan DY. Rates of light gas production by devolatilisa- [47] Bassilakis R, Zhao Y, Solomon PR, Serio MA. Sulphur and nitrogen
tion of coals and lignite. ACS Div Fuel Chem Prepr 1979;24(3): evolution in the argonne coals: experiment and modeling. Energy
129–40. Fuels 1993;7:710–20.
W. de Jong et al. / Fuel 82 (2003) 1139–1147 1147

[48] Carangelo RM, Serio MA, Solomon PR, Charpenay S, Yu ZZ, [54] Glarborg P, Jensen DA, Johnsson JE. Some aspects of fuel
Bassilakis R. Coal pyrolysis: measurements and modeling of product nitrogen conversion in solid fuel fired systems. Proceedings of the
evolution kinetics and char properties. ACS Div Fuel Chem Prepr Sixth International Conference of Technologies and Combustion
1991;36(2):796. for a Clean Environment, invited lecture num. 3, Porto, Portugal;
[49] Solomon PR, Hamblen DG, Serio MA, Yu ZZ, Charpenay S. A 2001.
characterisation method and model for predicting coal conversion [55] Solomon PR, Hamblen DG, Serio MA, Yu ZZ, Charpenay S.
behaviour. Fuel 1993;72(4):469–88. Can coal science be predictive. ACS Div Fuel Chem Prepr 1991;
[50] Bassilakis R, Carangelo RM, Wójtowicz MA. TG-FTIR analysis of 36(1):211– 67.
biomass pyrolysis. Fuel 2001;80:1765 –86. [56] Braun RL, Burnham AK. Analysis of chemical reaction kinetics using
[51] Solomon PR, Serio MA, Carangelo RM, Bassilakis R, Yu ZZ, a distribution of activation energies and simpler models. Energy Fuels
Charpenay S, Whelan J. Analysis of coal by thermogravimetry– 1987;1:153 –61.
Fourier transform infrared spectroscopy and pyrolysis modeling. [57] Teng H, Serio MA, Wójtowicz MA, Bassilakis R, Solomon PR.
J Anal Appl Pyrol 1991;19:1–14.
Reprocessing of used tires into activated carbon and other products.
[52] Miller RS, Bellan J. A generalised biomass pyrolysis model based on
Ind Engng Chem Res 1995;34:3102–11.
superimposed cellulose, hemi-cellulose and lignin kinetics. Combust
[58] Solomon PR, Hamblen DG, Carangelo RM, Serio MA, Deshpande
Sci Technol 1997;126:97–137.
GV. A general model of coal devolatilisation. ACS Div Fuel Chem
[53] Klass DL. Biomass for renewable energy, fuels, and chemicals.
Prepr 1987;32(3):83.
London, UK: Academic Press; 1998.

You might also like