See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/223729538

Thermogravimetric analysis and kinetics

modeling of isothermal carbonization of olive
wood in inert atmosphere

Article in Thermochimica Acta · November 2005

DOI: 10.1016/j.tca.2005.09.018

CITATIONS READS

35 127

4 authors, including:

Najla Grioui Kamel Halouani

University of Sfax Institut Préparatoire aux Etudes d'Ingénieurs…
16 PUBLICATIONS 93 CITATIONS 61 PUBLICATIONS 394 CITATIONS

SEE PROFILE SEE PROFILE

Foued Elhalouani
University of Sfax
97 PUBLICATIONS 401 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Mediterranean Agricultural Wastes: Environmentally Sustainable Resource for an Innovative

Renewable Energy Technology View project

Olive Oil Wastes to Electricity: Catalytic pyrolysis for waste valorization towards green diesel for
power generation View project

All content following this page was uploaded by Kamel Halouani on 07 February 2017.

The user has requested enhancement of the downloaded file.

 Thermochimica Acta 440 (2005) 23–30

Thermogravimetric analysis and kinetics modeling of isothermal

carbonization of olive wood in inert atmosphere
Najla Grioui a , Kamel Halouani a,∗ , André Zoulalian b , Foued Halouani c
a Micro-Electro-Thermal Systems – Industrial Energy Systems Group (METS-IESG), Institut Préparatoire aux Etudes d’Ingénieurs de Sfax (IPEIS),
B.P. 805, 3000 Sfax, Tunisie
b Laboratoire d’Etudes et de Recherches sur le Matériau Bois (LERMAB), Université Henri Point Carré Nancy 1 (UHP), B.P. 239,

54506 Vandoeuvre lès Nancy Cedex, France

c Ecole Nationale d’Ingénieurs de Sfax (ENIS), B.P. 3038, Sfax, Tunisie

Received 3 May 2005; received in revised form 9 September 2005; accepted 30 September 2005

Abstract
The kinetics of olive wood carbonization is investigated by means of isothermal thermogravimetric analysis method. Measurements were
carried out in a thermobalance for different fixed temperatures between 498 and 648 K. A two-stage semi-global kinetic model consisting of four
sequential steps was proposed to derive kinetic parameters. The olive wood is classified in three pseudo-components. For the first two, similar
thermal degradation mechanisms take place in a single reaction step. For the third, the thermal degradation takes place in two consecutive steps.
The isothermal conditions allow the kinetic constants (activation energy and pre-exponential factors) to be estimated by means of the analytical
solution of the mass conservation equations. An overall good agreement was obtained with activation energy values available in the literature.
© 2005 Published by Elsevier B.V.

Keywords: Carbonization; Olive wood; Kinetic parameters; TGA; Modeling

1. Introduction place at temperatures lower than 473 K. For temperature range of

473–553 K, hemicelluloses are converted essentially into gases
Biomass, especially wood, has traditionally been an impor- and acetic acid. This step corresponds to the wood roasting.
tant source of energy particularly attractive nowadays because of Above 553 K, the lignin and cellulose decompose to give three
its inherent nature of being environmentally friendly and renew- products: gas, tar and char. Therefore, the knowledge of the
able. Wood has also been considered as a potential feedstock for kinetic schemes of the wood carbonization is essential. Partic-
gasification to produce a mixture of H2 and CO (syngas). In this ularly, kinetic analyses of wood carbonization, also called low
process, the wood is converted into char as an intermediate prod- temperature pyrolysis, under isothermal conditions have been
uct which is subsequently or simultaneously gasified. The wood carried out by several authors to represent the kinetic schemes
carbonization in which the high yield charcoal is the principal of this process [1–8]. The simplest treatment describes the pro-
product is effectively an initial stage in any gasification pro- cess by means of a one-step global reaction for degradation
cess. It is characterized by a slow heating rate, a relatively low in which the activation energies vary toughly between 60 and
temperature (600–700 K) and a long residence time. Indeed, the 170 kJ mol−1 [1–4]. These models are not applicable for simu-
mechanism of wood carbonization shows the presence of several lating wood carbonization because they assume a constant ratio
decomposition phases when the temperature increases. A wood of the charcoal to volatiles yield [9]. A one-step multi-reaction
drying phase with elimination of some volatile compounds takes kinetic mechanism of beech wood was described by [5] as a two-
stage temperature process in the range of 523–673 K. Two sets
of kinetic parameters (activation energies of 17–115 kJ mol−1 )
∗
were needed depending on a limit temperature of 603 K. This
Corresponding author. Tel.: +216 98 954 415; fax: +216 74 246 347.
E-mail addresses: Najla.Grioui@fss.rnu.tn (N. Grioui),
kinetic scheme cannot be extended to systems different from
Kamel.Halouani@ipeis.rnu.tn (K. Halouani), Andre.Zoulalian@lermab. the one on which it is based [9]. Multistage, semi-global models
uhp-nancy.fr (A. Zoulalian), Foued.Halouani@enis.rnu.tn (F. Halouani). are developed to describe the kinetic of isothermal wood degra-

0040-6031/$ – see front matter © 2005 Published by Elsevier B.V.

doi:10.1016/j.tca.2005.09.018

TCA-74101; No. of Pages 8

24 N. Grioui et al. / Thermochimica Acta 440 (2005) 23–30

Table 1 - a micro-balance, based on a compensation permitting to mea-

Elementary composition of Tunisian olive wood sure the mass of the sample placed in a basket of 5 cm3
Elementary composition Mass fraction (%) volume, continuously balanced and thus maintained at a fixed
Carbon 45.08
position during the experiment. This allows to avoid even-
Hydrogen 6.21 tual intra-particle temperature gradients in isothermal con-
Nitrogen 0.4 ditions in the oven. The maximal mass variation is 200 mg
Oxygen 45.39 and the mass measurement precision is 0.4 ␮g. The oven
Sulphur Inferior to 0.3 temperature is adjustable with a precision of 0.1 K and can
reach a maximum of 1473 K. In order to establish an inert
atmosphere during all experiments, a controlled argon flow
dation [6–8]. For carbonization techniques where the product
fixed to 7.2 l h−1 at 273 K and 1 atm, sweeps the measure-
gases energy is recycled in the process [10], models of this type
ment cell that is purged for 20 min before starting the heating
can be successfully applied to simulate thermal conversion since
program.
they include the description of the primary degradation of the
virgin wood and the secondary degradation of the primary pyrol-
2.3. Experimental methodology
ysis products. The large differences between reaction schemes
and kinetic constant values proposed by previous authors show
The initial mass of olive wood powder in the basket is about
that further analysis is needed to study the multi-stage thermal
170 mg. The experiments start with a drying phase during which
degradation kinetics of wood under isothermal conditions.
the temperature is raised from room temperature to 423 K at a
The main objective of this study is to propose a two-stage,
heating rate of 20 K min−1 with a holding time of 20 min. Then,
semi-global kinetic model of olive wood carbonization. Isother-
the same heating rate is applied to the oven until reaching the
mal weight loss curves of olive wood, measured in an inert
fixed studied temperature (pre-heated phase). This methodol-
atmosphere for a temperature range of 498–623 K, are used to
ogy is followed to reach the isothermal stage in the pre-heated
elaborate a reaction mechanisms scheme. A comparison is also
wooden powder without any intra-particle gradients and in order
made between the estimated activation energy values and those
to shorten the first thermal dynamic stage [6]. The subsequent
available in the literature corresponding to isothermal carboniza-
thermal decomposition is carried out in isothermal conditions
tion conditions.
during 5 h. The same experiment is repeated for seven fixed
temperatures (T = 498, 523, 548, 573, 598, 623 and 648 K). Mea-
2. Experimental study surements of the mass of the wooden powder are taken in a
fixed position during 5 h at each temperature level. In these con-
2.1. Material specification ditions, the diffusion time is always inferior than the reaction
time. Indeed, for a wood particle of e = 0.5 mm thickness and a
A hard olive wood, from Sfax in Tunisia, is used as raw diffusion coefficient D between 10−7 and 10−8 m2 s−1 , the dif-
material in all experiments. It was crushed by a robot and the fusion time τ = e2 /D is respectively equal to 2.5 and 25 s. This
obtained powder was sieved through two sieves. The particle time is always very small compared to the reaction time which
size of the sieved powder was between 0.5 and 1 mm. Table 1 exceeds 500 s in all the studied cases. The temperature and the
shows the obtained results of the olive wood elementary analysis concentration gradients are consequently negligible in the bas-
realized in the Analysis Central Service of CNRS in France. ket which can be assumed to have a uniform temperature and
By neglecting the sulphur and nitrogen fractions in the above concentration.
analysis data, the empirical formula of the used wood can be
represented by: CH1.67 O0.75 . 2.4. Measurements

2.2. Apparatus description The residual weight curves of the olive wood, for the seven
studied temperatures T = 498, 523, 548, 573, 598, 623 and 648 K,
A Setaram thermobalance apparatus is used for the thermo- are plotted in Fig. 1. In all the experiments, the values of the
gravimetric analysis. This apparatus records the different data total mass used are those which correspond to the fixed tem-
concerning the evolution of the temperature and the mass loss. peratures. In fact, to separate the mass loss during the dynamic
The thermobalance consists of three main parts: stage (non-isothermal) of decomposition from that of the isother-
mal stage, particularly for higher temperatures (598, 623 and
- a controller which allows the transfer of the experimental data 648 K), the relative values of total mass before reaching the
(mass, temperature) to a computer which records and treats fixed temperature are not taken into account in the estimation
them by Setsys software (e.g. DTG), of the kinetic parameters. Only the start time at the beginning
- an oven composed of an electrical resistance in graphite and of the transformation is obtained by extrapolation of the mass
a double envelope cooled by a circulation of water in order to values corresponding to the fixed temperature. This time of ori-
control the heat flow and then the temperature. A PDI regulator gin is estimated with a rather good accuracy for the majority of
allows to vary the rate of temperature change between 0.2 and the studied temperatures by an exponential extrapolation at the
50 K min−1 , beginning of the experimental curves.
 N. Grioui et al. / Thermochimica Acta 440 (2005) 23–30 25

Fig. 1. Experimental mass loss curves of olive wood for the different studied temperatures.

The residual mass curves presented in Fig. 1 show that the tulated to reflect the sum of the thermal response of its three main
rate of the kinetics degradation process increases with the tem- components; inorganic components catalyze the conversion pro-
perature. At high temperatures (598, 623 and 648 K), a very cess in an unpredictable variety of ways [9,16].
fast mass loss stage is observed after which the mass loss In Fig. 2 we have examined experimentally the evolution
m(t)/mA0 becomes very low. This rate becomes almost equal to of the mass loss rate of the olive wood during carbonization
zero for temperatures above 648 K. The residual mass obtained between 298 and 1173 K, with a slow heating rate (1.5 K min−1 )
at 648 K represents 32% of the mass of the initial dried sample. to avoid spatial gradients of temperature. This figure describes
the thermal degradation behavior of the olive wood with increas-
3. Kinetic modeling ing temperature and confirms the presence of three peaks in
wood decomposition: between 353 and 423 K, the loss of mass
Several kinetic models of wood pyrolysis [7,11–15] sug- is due to the drying phase. From 453 K, the wood decomposition
gest to divide the wood into two or three parts. The models starts by a first peak at 490 K. Then, a second peak appears at
where these three parts correspond to the three main components 543 K and finally, a third important peak appears at the tem-
of wood (hemicelluloses, cellulose, lignin) are not satisfactory perature of about 600 K. This decomposition continues with
because the thermal degradation behavior of wood has been pos- a low rate for the higher temperatures. At 723 K, the rate of

Fig. 2. Evolution of the mass loss rate of the olive wood for a temperature range of 293–1173 K.
26 N. Grioui et al. / Thermochimica Acta 440 (2005) 23–30

decomposition is very low and becomes negligible at 923 K. dmA2

= −k2 mA2 (6)
The temperature values where the three peaks are observed in dt
this experiment do not correspond necessarily to the decomposi- dmA3
tion temperature of the three principal components of the wood: = −k3 mA3 (7)
hemicelluloses, cellulose and lignin, separately. dt
Based on experimental results and the wood carboniza- dmC2
= γ2 k2 mA2 (8)
tion mechanisms mentioned above, we propose to develop a dt
two-stage, semi-global multi-reaction kinetic model of olive dmB
wood carbonization. In our model, the wood is subdivided = k3 βmA3 − k4 mB (9)
dt
into three pseudo-components A1 , A2 , A3 ; each of them cor-
responds to a specific kinetic law and a mass fraction α1 , α2 and dmC3
= γ 3 k4 m B (10)
α3 = 1 − α1 − α2 , respectively. At low temperature (T ≤ 523 K), dt
only the degradation of A1 prevails and the product is gaseous. where k1 , k2 , k3 and k4 are the rate constants of reactions (1),
In fact, this level of temperature (490 K) corresponds to the elim- (2), (3) and (4), respectively, and mA1 , mA2 , mA3 , mC2 , mB and
ination of some volatile wood compounds during carbonization. mC3 represent the mass of the constituents A1 , A2 , A3 , C2 , B
At high temperature, the degradation of pseudo-components A1 and C3 , respectively. Here, m(t) represents the total mass of the
and A2 is complete and for A3 , the transformation includes two sample at the time t, which is
reaction steps in which the product of the first step (B) is con-
verted into charcoal (C3 ) and gases (G4 ) in the second step. m(t) = mA1 (t) + mA2 (t) + mA3 (t) + mC2 (t) + mC3 (t) + mB (t)
Indeed, for high temperatures, it can be seen from Fig. 1 that (11)
after a very fast mass loss stage, there is a very weak variation
of the mass as function of the time. This abrupt change in the The analytical resolution of the balance equations leads to the
evolution of the mass loss is necessarily related to the presence following equation:


of two consecutive or parallel stages of wood decomposition. m(t)
The proposed kinetic scheme is as follows: = α2 γ2 + α3 γ3 β + α1 × exp(−k1 t) + α2 (1 − γ2 )
m A0




A1 (1st pseudo component of olive wood) → G1 (gas) (1) βk3 βγ3 k4
× exp(−k2 t) + α3 1− +
k3 −k4 k3 −k4
A2 (2nd pseudo component of olive wood)

α3 βk3
→ γ 2 C2 (charcoal) + G2 (gas) × exp(−k3 t) + (1 − γ3 ) × exp(−k4 t)
(2) k3 − k 4
(12)
A3 (3rd pseudo component of olive wood)
where mA0 is the mass of the dried wooden powder at 423 K.
→ βB (intermediate solid) + G3 (gas) (3)
3.1. Determination of the model parameters
B (intermediate solid) → γ 3 C3 (charcoal) + G4 (gas) (4)
The two first pseudo-components A1 and A2 degrade in a single Eq. (12) shows that the mass of the sample during the pyrol-
stage reaction by giving gas products (G1 ) and (G2 ), respec- ysis at fixed temperature in an inert atmosphere depends on 10
tively, and a non-degradable solid residue “charcoal” (C2 ). For parameters (α1 , α2 , α3 , γ 2, γ 3 , β, k1 , k2 , k3 and k4 ). From the
A3 the thermal degradation takes place in two consecutive reac- experimental curves, it is not possible to determine all these
tion stages. The first stage leads to an intermediate product (B) parameters without carrying out a sequential analysis. Based on
comparable to a solid (large hydrocarbon molecule) and gases the experimental mass loss rate (Fig. 2) beyond 550 K, we can
(G3 ). In the second stage the intermediate product (B) is trans- observe that the third peak of the mass degradation rate is more
formed into a non-degradable solid “charcoal” (C3 ) and gas intense than the first two. This indicates that the mass fraction α3
products (G4 ). The constituent B is not really a volatile interme- of A3 is higher than (α1 + α2 ) of the pseudo-components A1 and
diate product which can be degraded but its degradation rate is A2 . The same experimental results show that the fraction α2 of
relatively low compared to the ones of the pseudo-components A2 is higher than α1 of A1 . After several test simulations of the
A1 , A2 and A3 . The mass fractions (β) of the intermediate prod- mass loss evolution, we have fixed the sum of the mass fractions
uct (B) and those of the non-degradable solid “charcoal” (γ 2 and of the pseudo-components A1 and A2 to 0.4 (α1 + α2 = 0.4) and
γ 3 ) depend on temperature. the mass fraction of A3 to 0.6 (α3 = 0.6). These values of α1 + α2
By assuming that the kinetic is described by first order laws and α3 conform to the best description of the experimental evo-
for the four reactions, the mass balance of the solid and slightly lution of the mass loss.
volatile components A1 , A2 , A3 , C2 , B and C3 , can be respec- The degradation of the three pseudo-components A1 , A2 and
tively written as A3 of the olive wood occurs in different temperature ranges. At
low temperature (498 K ≤ T ≤ 523 K), the number of unknown
dmA1 parameters is limited to two (k1 and k2 ) and to four parameters
= −k1 mA1 (5)
dt (k3 , k4 , β and γ 3 ) at high temperatures (598 K ≤ T ≤ 648 K). For
 N. Grioui et al. / Thermochimica Acta 440 (2005) 23–30 27




the intermediate temperatures (523 K ≤ T ≤ 598 K), we deter- βk3 βγ3 k4
mine the values of the parameters k2 , k3 , β and γ 3 , and keep + α3 1 − +
k3 − k 4 k3 − k 4
the same value of γ 2 and extrapolate the expressions of k1 and

k4 . This semi empirical procedure is certainly not the best, but α3 βk3
× exp(−k3 t)+ (1 − γ3 ) × exp(−k4 t)
its merit is that it does not require a simultaneous determina- k3 −k4
tion of the all parameters. Then, we consider the three following (15)
domains:
At low temperatures, Eq. (13) can be written for t → ∞
- at low temperatures (498 K ≤ T ≤ 523 K), only reactions (1) as m(t)/mA0 = α2 γ2 + α3 + α2 (1 − γ2 ) × exp(−k2 t), because
and (2) take place. In this case, m(t) can be written as reaction (1) is achieved. At 523 K, the measured residual mass

 defined by m∞ /mA0 = α2 γ2 + α3 allows the calculation of the
m(t) term ␣2 × ␥2 in Eq. (13) equal to about 0.1 at 523 K. By fitting the
= α2 γ2 + α3 + α1 × exp(−k1 t)
m A0 slope of the experimental curve Ln((m(t)/mA0 ) − α3 − α2 γ2 ) =
−k2 t + Ln(α2 (1 − γ2 ))for t → ∞ we determine the value of
+ α2 (1 − γ2 ) × exp(−k2 t) (13)
k2 . The values of k1 and α1 are obtained by representing at
- for high temperatures (598 K ≤ T ≤ 648 K), the decomposition t → 0 the experimental curve Ln((m(t)/mA0 ) − α3 − α2 γ2 ) =
of the olive wood for the two first pseudo-components A1 −k1 t + Lnα1 because reaction (2) is not yet released. Then
and A2 are total and only the transformation of A3 remains the values of α1 , α2 , k1 , k2 and γ 2 are obtained. The val-
(reactions (3) and (4)). The variation of the total mass is given ues of α1 = 0.16 and α2 = 0.24 are used for all temperature
by ranges of carbonization. These values are in agreement with

 the amplitudes of the corresponding first and second peaks
m(t) in Fig. 2. The value of charcoal fraction obtained from A2
= α 2 γ2 + α 3 γ3 β
m A0 of the wood is equal to γ 2 = 0.42 assumed constant for T ≥



 523 K.
βk3 βγ3 k4
+ α3 1 − + For high temperatures, Fig. 1 shows that the final asymptotic
k3 − k 4 k3 − k 4 value of the residual mass is about 0.3, taking into account the


α3 βk3 mass fraction α2 × γ 2 = 0.1 of the non-degradable product (C2 )
× exp(−k3 t)+ (1 − γ3 ) × exp(−k4 t) “charcoal” produced from A2 , the value of the mass fraction of
k3 − k 4
(14) charcoal (C3 ) produced from A3 (α3 × β × γ 3 ) is equal to 0.2.
It can noted here that the fraction of non-degradable product
- between 523 and 598 K, the four reaction stages take place “charcoal” from α3 was the double of that resulting of α2 which
simultaneously and following the assumption proposed, the corresponds appreciably to the identical proportions taking into
variation of the total mass is given by account the value of α2 = 0.24. The same method used above for

 low temperature with Eq. (13) is used with Eq. (14) to determine
m(t)
= α2 γ2 + α3 γ3 β + α1 × exp(−k1 t) k3 , k4 , γ 3 and β. As shown in Fig. 3, the parameters γ 3 and
m A0 β are functions of temperature. In fact, the mass fraction γ 3
+ α2 (1 − γ2 ) × exp(−k2 t) of the non-degradable solid (C3 ) increases to detriment of β

Fig. 3. Variation of the parameters β and γ 3 as function of the temperature.

28 N. Grioui et al. / Thermochimica Acta 440 (2005) 23–30

Fig. 4. Arrhenius plot: k1 , k2 , k3 and k4 as function of 1/T.

which represents the mass fraction of the intermediate solid (B) all optimized parameters in the model, the comparison between
transformed into non-degradable solid “charcoal” (C3 ). experimental and calculated curves giving the residual mass
By assuming that k1 , k2 , k3 and k4 depend on the temperature according to the time for all the studied temperatures is rep-
according to the Arrhenius law, the values of the constants rate k1 resented in Fig. 5. By considering the variability of wood, we
and k4 are extrapolated in the intermediate range of temperature can deduce that the kinetic model represents accurately all the
(Fig. 4). experimental curves.
The experimental curves (Fig. 1) in this range of temper- All of the optimized values of the model parameters are given
ature are used to determine the optimized parameters k2 , k3 , in Table 2 in which e is the average relative deviation between
γ 3 and β which depend on temperature. To determine the the obtained modeling values and the experimental data:
value of k2 , Eq. (15) can be written for t → 0 as m(t)/mA0 =  N  
1 − [α1 k1 + α2 (1 − γ2 )k2 ]tbecause reactions (3) and (4) have 1  mcal − mexp 
e= 2 × 100 (16)
not yet occurred. The values of γ 3 and β are extrapolated in N mcal + mexp
i
the intermediate range of temperature (Fig. 3). The optimized
values of γ 3 , β and k3 can be determined by fitting the experi- The slopes and the origin ordinates of the straight lines of
mental and theoretical curves of m(t)/mA0 . Indeed, by using the the Arrhenius plot ki (i = 1, 2, 3, 4) reported in Fig. 4 allow the

Fig. 5. Verification of the accuracy of model parameters.

 N. Grioui et al. / Thermochimica Acta 440 (2005) 23–30 29

Table 2
Kinetic parameters of olive wood carbonization for a temperature range of 498–648 K
T (K) k1 (s−1 ) k2 (s−1 ) k3 (s−1 ) k4 (s−1 ) γ2 β γ3 e (%)

498 2.8 × 10−4 2.4 × 10−5 1.22 × 10−6 1.5 × 10−6 0.95 – – 1.00
523 9.5 × 10−4 8.65 × 10−5 8.63 × 10−6 2.71 × 10−6 0.42 – – 0.12
548 2.9 × 10−3 2.25 × 10−4 4.5 × 10−5 4.64 × 10−6 0.42 0.66 0.5 1.00
573 7.94 × 10−2 7.4 × 10−4 2.6 × 10−4 7.56 × 10−6 0.42 0.54 0.62 1.00
598 2 × 10−2 1.8 × 10−3 1.4 × 10−3 9 × 10−6 0.42 0.51 0.66 0.60
623 4.72. × 10−2 4.21 × 10−3 5.2 × 10−3 3.16 × 10−5 0.42 0.391 0.85 0.90
648 1.04 × 10−1 9.34 × 10−3 1.27 × 10−2 1.95 × 10−5 0.42 0.39 0.85 1.00

Table 3 m(t)
Kinetic constants of the four reactions = α2 γ2 + α3 γ3 β + α1 × exp(−k1 t) + α2 (1 − γ2 )
m A0
Reactions Ea (kJ mol−1 ) k0 (s−1 )



βk3 βγ3 k4
(1) 105.89 3.5 × 107 × exp(−k2 t) + α3 1 − +
k3 − k 4 k3 − k 4
(2) 106.78 3.72 × 106


(3) 169.56 7.23 × 1011 α3 βk3
(4) 51.04 3.4 × 10−1 × exp(−k3 t) + (1 − γ3 ) × exp(−k4 t)
k3 − k 4
(17)

where the values of α1 = 0.16, α2 = 0.24 and α3 = 0.6 are indepen-

activation energies Eai and the pre-exponential factors k0i of the dent of the temperature and γ 2 = 0.42 at T ≥ 523 K, ki (i = 1–4)
global reactions to be respectively estimated (Table 3). represents respectively the reaction rate of the four reactions cal-
culated as function of the temperature from the Arrhenius law
3.2. Model validation ki = k0i exp(−Eai /RT ), where the values of k0i and Eai are given
in Table 3. The parameters β and γ 3 are given by the following
Here we propose a comparison of the obtained values of correlations:
the activation energies Eai with those available in the literature.
Overall the comparisons with literature results can be considered β = −2.4 × 10−3 T + 1.9252 (18)
good, taking into account the differences in the experimen-
γ3 = 3.8 × 10−3 T − 1.5807 (19)
tal technique (isothermal versus dynamic thermogravimetry),
the feedstock properties/characteristics, the reaction tempera-
tures, the mathematical treatment of the data and the kinetic 4. Conclusion
mechanisms [8]. Indeed, for the first reaction (or the first pseudo-
component A1 ), typical activation energies are of the order of Global degradation kinetics of olive wood carbonization has
84–140 kJ mol−1 [1,17–20], against a value of 106 kJ mol−1 been investigated by means of isothermal thermogravimetry.
estimated here. Activation energies for the second reaction are Weight loss has been measured and used to elaborate a two-
reported to vary between 73 and 121 kJ mol−1 [1,17–20] against stage, semi-global multi reaction kinetic model of olive wood
a value of 107 kJ mol−1 estimated here. For the third reaction, carbonization in the temperature range from 473 to 673 K. The
the activation energy is of the order of 170 kJ mol−1 which isothermal conditions have allowed the kinetic constants (acti-
is relatively high compared to those available in the literature vation energies and pre-exponential factors) to be estimated by
112 ≤ Ea ≤ 133 kJ mol−1 [1,18–20]. The higher value estimated means of the analytical solution of the mass conservation equa-
here can be due to the difficulty, for isothermal conditions, to sep- tions. Overall the comparisons of the values of the activation
arate the decomposition of the pseudo component A3 from that energies with literature results can be considered good. Finally,
of other components (A2 , particularly) [8]. Finally, for the fourth the proposed model is formulated in a general equation given at
reaction, corresponding to the decomposition of the intermedi- any time and temperature of the production rate of charcoal (C2
ate solid (B), an activation energy of 51 kJ mol−1 is obtained. and C3 ), tar (B) and gaseous (G1 –G4 ) during wood carboniza-
The lower value estimated here compared to those in the lit- tion. The proposed model will be applied in the next stage to
erature (80–108 kJ mol−1 ) [20,21] can be due to the slow rate different species of wood sawdust and will be integrated in a
of the relative transformation in the studied temperature range complete modeling of carbonization of a thick wood particle.
(498–643 K) which corresponds to a diffusional transport of
a strongly absorbed molecule or an internal rearrangement of Acknowledgement
chemical structure rich in carbon.
Finally, to facilitate the utilization of our model by other The authors wish to thank the Analysis Central Service of
researchers, we propose to represent all our results by the fol- CNRS (Vernaison, France) for realization of the elementary
lowing general analytical equation: analysis of our Tunisian olive wood.
 30 N. Grioui et al. / Thermochimica Acta 440 (2005) 23–30

References [11] A.F. Roberts, Combust. Flame 14 (1970) 261.

[12] M.G. Grönli, Ph.D. Thesis, Norwegian University of Science and Tech-
[1] F. Thurner, U. Mann, Ind. Eng. Chem. Process. Des. Dev. 20 (1981) nology, Norway, 1996.
482. [13] G. Varhegby, M.J. Antal, E. Jakab, P. Szabo, J. Anal. Appl. Pyrol. 42
[2] S.W. Ward, J. Braslaw, Combust. Flame 61 (1985) 261. (1997) 73.
[3] B.M. Wagenaar, W. Prins, W.P.M. van Swaaij, Fuel Process. Technol. [14] J.J.M. Orfao, F.J.A. Antunes, J.L. Figueiredo, Fuel 78 (1999) 349.
36 (1994) 291. [15] L. Helsen, E. Van den Bulck, J. Anal. Appl. Pyrol. 53 (2000) 51.
[4] J. Reina, E. Velo, L. Puigjaner, Ind. Eng. Chem. Res. 37 (1998) 4290. [16] M.J. Antal, G. Varhegby, Ind. Eng. Chem. Res. 34 (1995) 703.
[5] J.N. Barooahm, V.D. Longm, Fuel 55 (1976) 116. [17] R. Bibao, J.F. Mastral, M.E. Aldea, J. Ceamos, J. Anal. Appl. Pyrol. 42
[6] C.A. Koufopanos, G. Maschio, A. Lucchesi, Can. J. Chem. Eng. 67 (1997) 189.
(1989) 75. [18] W.-C.R. Chan, M. Kelbon, B. Krieger, Fuel 64 (1985) 1505.
[7] C. Di Blasi, M. Lanzetta, J. Anal. Appl. Pyrol. 40–41 (1997) 287. [19] R. Front, A. Marcilla, E. Verdu, J. Devessa, Ind. Eng. Chem. Res. 29
[8] C. Branca, C. Di Blasi, J. Anal. Appl. Pyrol. 67 (2003) 207. (1990) 1846.
[9] C. Di Blasi, Chem. Eng. Sci. 51 (7) (1996) 1121. [20] M.G. Grönli, M.C. Melaaen, Energy Fuels 14 (2000) 791.
[10] K. Halouani et, H. Farhat, Renew. Energy 28 (2003) 129. [21] A.G. Liden, F. Berruti, D.S. Sott, Chem. Eng. Commun. 65 (1988) 207.

View publication stats