Вы находитесь на странице: 1из 11

Energy 93 (2015) 1731e1741

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Torrefaction of wood and bark from Eucalyptus globulus and Eucalyptus


nitens: Focus on volatile evolution vs feasible temperatures
rez a, *, Cristina Segura a, Vero
Luis E. Arteaga-Pe  nica Bustamante-Garca b,
Oscar Go mez Capiro c, Romel Jime nez c, **
a
Technological Development Unit, University of Concepcio n, Coronel, Chile
b rez del Estado de Durango, Durango, Mexico
Forest Sciences Faculty, University of Jua
c
Chemical Engineering Department, University of Concepcio n, Concepcion, Chile

a r t i c l e i n f o a b s t r a c t

Article history: Torrefaction is a thermal pretreatment leading to the improvement of most of the fuel properties of
Received 20 July 2015 biomass, namely energy density, HHV (higher heating value), grindability and hydrophobicity. The aim of
Received in revised form this study is to identify the most feasible temperature to carry out torrefaction of Eucalyptus globulus and
13 September 2015
nitens, based on chemical evidences associated to the release of volatiles during thermal treatment of
Accepted 4 October 2015
Available online 19 November 2015
biomass. With that end: (i) Devolatilization kinetics, (ii) Effects of temperature and residence time and
(iii) volatiles composition during torrefaction of both wood and bark were analyzed. In all cases DTG
(derivative thermogravimetric curves) exhibited the typical shape of lignocellulosic materials, with three
Keywords:
Torrefaction
decomposition phases and two reaction zones. Values of activation energies for hemicellulose decom-
Eucalyptus position, were in agreement with those reported in the literature (121e170 kJ/mol). Carboxylic acids,
Volatiles evolution water and phenolic compounds showed two peaks, which were associated to torrefaction (below 310  C)
Feasible temperature and pyrolysis (310e410  C) respectively. The most feasible temperatures for torrefaction were estimated
as a function of these peaks, and it ranged between 295  C and 310  C for all samples. Main volatile
species at the torrefaction peaks were distributed as Water > Acetic Acid > CO2 > Others, while Levo-
glucosan formation was marginal, due to the catalytic effect of inorganics.
2015 Elsevier Ltd. All rights reserved.

1. Introduction potential with more than 15 million of hectares of native and wood
plantations yielding around 35 m3/ha/y. Main wood species being
The migration from actual fossil-based economies to more exploited by the Chilean forest industry are: Pinus radiata (61%),
sustainable bio-based ones have been identied as the most Eucalyptus globulus (29%) and Eucalyptus nitens (10%) [3]. Euca-
conscious way to reach sustainable development standards. Energy lyptus is used nationally and exported overseas as feedstock for
security concerns, excessive fossil fuel consumption, increasing cellulose plants, and in less extent, for energy production. For the
pollutant emissions and incipient climate change are the main production of cellulose, eucalyptus chips should be bark-free, hence
drivers for such a change. Chile, has experienced a fast economic bark is removed in eld or in pre-processing stations. This sub-
growth in the last decades, with a proportional effect on the energy product (11e12 vol.% of processed wood) it is being wasted as
demand (1.42 EJ in 2013) and pollutants emissions [1,2]. Further- there are no feasible alternatives for its exploitation.
more, the consumption of primary energy in Chile heavily relies on These facts have created a positive scenario for introducing
imported fossil fuels, mainly oil (30%), coal (22%) and natural gas renewable energy sources into the Chilean electric market [4],
(13%) [1]. Paradoxically, the country accounts for a huge bioenergy especially for biomass. Main barriers for such a change are inherent
to the biomass itself, because some of its properties are inconve-
nient for energy applications, namely its hydrophobic character,
high moisture content (30e50% as received), and its heterogeneity.
* Corresponding author. Tel.: 56 41 266 1855; fax: 56 41 275 1233. Other disadvantages of biomass are its tenacious and brous
** Corresponding author. Tel.: 56 41 266 4762.
structure that makes difcult to design and control the feeding
rez), romeljimenez@udec.cl
E-mail addresses: l.arteaga@udt.cl (L.E. Arteaga-Pe
nez).
(R. Jime systems.

http://dx.doi.org/10.1016/j.energy.2015.10.007
0360-5442/ 2015 Elsevier Ltd. All rights reserved.
1732 L.E. Arteaga-Perez et al. / Energy 93 (2015) 1731e1741

Nomenclature ni reaction order for devolatilization of


pseudocomponent i (i 1 (hemicellulose), 2
HHV higher heating value (MJ/kg) (cellulose) and 3 (lignin))
Ea_i activation energy for devolatilization of Std. Dv. standard deviation (%)
pseudocomponent i (i 1 (hemicellulose), 2 T temperature ( C)
(cellulose) and 3 (lignin)) (kJ/kmol) TGA thermogravimetric analysis
CAW carboxylic acids peaks for wood E. nitens and t time (min)
E. globulus (a.u) WPB water peaks for Bark of E. nitens and E. globulus (A)
CAB carboxylic acids peaks for bark from E. nitens and WPW water peaks for wood of E. nitens and E. globulus (A)
E. globulus (a.u) wi weight fraction (i hemicellulose, cellulose and
X600 char yield at 600  C (%) lignin)
EDF energy densication factor
EY energy yield (%) Greek letters
M mass (g) calc calculated
dX/dt mass derivative (%/min) daf dry ash free
MS mass spectrometry db dry basis
MY mass yield (%) exp experimental
OF objective function f nal
Ko_i pre-exponential factor for devolatilization of o initial
pseudocomponent i (i 1 (hemicellulose), 2 torr torreed
(cellulose) and 3 (lignin)) (min1)

Torrefaction is a technology aiming to reduce and/or eliminate Laminaria japonica and they concluded that same energy densi-
most of above-mentioned difculties. It is a thermochemical pre- cation as for terrestrial biomasses, can be obtained for algae at
treatment akin to mild pyrolysis, as it occurs at relatively low considerably lower temperatures (94% energy yield at 150  C) and
temperatures, 200e350  C, in a non- or low- oxidizing atmosphere residence times. Similarly, Poudel et al. [10] reported on the inter-
[5e7]. During such treatment, the tenacious brous of the original changeability of time and temperature during torrefaction, to
hydrophilic material is largely destroyed, through the breakdown obtain the same values of energy yield for several combinations of
of hemicellulose and to a lesser degree of cellulose and lignin. The these operation variables. From their study, a feasibility tempera-
remaining solid (biochar) becomes brittle and easier to grind. With ture range between 290  C and 330  C, was identied, supported on
the removal of light volatile fractions that contain most of the ox- the energy yield (>80%) and on the liberation of CO, CO2 and CxHy.
ygen, the material becomes more hydrophobic and its heating Poudel's [10] study did not included the effect of other volatiles
value increases by 5e25% at the expense of a mass loss of ~30% such as carboxylic acids and water, on the properties of the torre-
[8e10]. The main virtues of torrefaction are its high energy ef- ed material. More recently, Strandberg et al. [23] studied the effect
ciency, economic feasibility to replace wood pellets and its subse- of temperature and residence time on the torrefaction of spruce
quent exibility [11,12]. This process has gained interest in the last 5 wood, concluding that temperature and time are not totally inter-
years, which have been evidenced by a stepwise increment in the changeable, as their effect depends on the nature of the biomass.
number of peer-reviewed papers, summarized in several review They found a region to interchange these parameters, but they did
reports [6,11e13]. not make references to the relation between this feasible area and
When biomass undergoes torrefaction, temperature and resi- the chemical transformations occurring in the feedstock during
dence time are the most important parameters, as they dene the torrefaction.
severity of the treatment [14e18]. The higher the torrefaction time In spite of the vast literature reporting on the effect of main
and temperature, the higher the degree of biomass decomposition. operation parameters affecting the mass yield, products composi-
Temperature, in the extreme case, could cause the complete con- tion and its fuel properties during biomass torrefaction; until now
version of cellulose into more simple molecules, leading to a solid there is not a consensus in how to determine the most feasible
product like lignite [19]. Moreover, by increasing either tempera- temperatures to carry out the treatment. As prior discussed, this
ture or residence time, the heating value of the produced solid temperatures depends on the feedstock, as different biomass
increases in reference to raw material (due to the reduction of the sources will decompose at different extent through different
atomic H/C and O/C ratios), while mass yield is continuously mechanisms, under the same conditions [5]. These facts suggest
reduced. As was recently reported by Sermyagina et al. [20], this that the problem should be addressed by studying the thermal
behavior establishes a trade-off between both product heating decomposition process using as tracers, the volatile composition,
value and process mass yield. The solution to such a problem, can mass yield and the energy densication factor. The prior statement
be reached by using a more global criteria like the energy yield, inspired our study, aimed to establish the most feasible tempera-
dened as the product of EDF (energy densication factor) and the tures to carry out the torrefaction of E. globulus and E. nitens of
mass yield. Most common approaches, propose establishing the wood and bark.
proper combination of residence time and temperature, by xing
the desired energy yield. In a recent study, Chen et al. [21] exam- 2. Experimental
ined this relation in the torrefaction of microalgae residues (Chla-
mydomonas sp). They concluded that both parameters produced a 2.1. Samples collection and preparation
linear effect on energy and mass yields and suggested that oper-
ating at higher temperatures and short residence times is always Biomass samples (E. globulus and E. nitens) were collected from a
more feasible. Uemura et al. [22], studied the torrefaction of 15-year-old plantation, property of Forestal Rio Calle-Calle
L.E. Arteaga-Perez et al. / Energy 93 (2015) 1731e1741 1733

(Chile). Experiments were carried out for these species due to their 2.3.2. Devolatilization kinetics
economic importance in the Chilean forest industry (at about 40% Branca and Di Blasi [31] studied the kinetics of biomass com-
of the market). All samples were taken at the same time to avoid bustion processes through two series- and parallel-reaction
the effect of aging in the eld. Furthermore, wood and bark comes models. Each of those models consisted in four reactions (3 for
from trees with similar morphological characteristics (age, height devolatilization of the main components of lignocellulosic biomass,
and diameter) and they are representative of eld and processing and 1 reaction for char burn-off) all giving similar results for the
stations. estimated kinetic parameters. In this study, we have neglected the
burn-off reaction due to the TGA analysis is carried out under inert
2.2. Biomass characterization atmosphere (N2). Therefore, a three parallel reaction model (Eqs.
(1)e(3)), corresponding to the devolatilization of hemicellulose,
The samples were characterized before and after torrefaction, cellulose and lignin [32] was assumed:
using standard practice for proximate analysis (D3172-89) [24]
k1
and elemental analysis was performed in a Leco True Spec Hemicellulose!V1 (1)
analyzer, according to the standard (ABNT-NBR-8112) [25].
Moreover, caloric values were determined in a Parr 6400 auto- k2
Cellulose!V2 (2)
mated calorimeter (ASTM D5865-13) [26]. Characterization of raw
samples is presented in Table 1. The reported values are mean of
three replicates, for which, the standard deviation (s) is also Lignin!V3
k3
(3)
provided.
Biomass composition in terms of cellulose, hemicellulose and The rate of these three reactions can be represented by a general
lignin was deduced by atomic balance according to the procedure power law expression of nth order:
reported by Burhenne et al. [14]. The chemical formula of cellulose  
dai
(C6H10O5) and hemicellulose (C5H10O5) are accepted to be the same Ki exp Eai=RT 1  ai ni i 1; ; 3 (4)
for all biomasses (Table 2). On the contrary, there is no a general dt
formula for lignin as it is product from polymerization of three
where R is the universal gas constant (kJ kmol1 K1) and T is the
monolignols (p-coumaryl, conyferil and synapyl). Accordingly,
absolute temperature. The conversion degree (a) is dened as the
lignin is classied as H (p-hydroxiphenil) type, G (guaiacyl) type
mass fraction of decomposed solid or released volatiles:
and S (syringil) type. For Lignin we have assumed the elemental
formulas reported by Miranda et al. [27]; who suggested two rep- mo  mi
resentations for wood and bark respectively, based on their a (5)
mo  mf
compositional differences (see Table 2).
As the components fractions calculated by atomic balance are Where mo and mf are the initial and nal masses of solid respec-
sensitive to the elemental composition, the data is checked during tively and mi is the mass of solid at any time.
kinetic analysis by curve tting to a three-component devolatili- According to the adopted model, the overall conversion rate
zation model [29,30]. results from the contribution of the three pseudo-components:

2.3. Experimental setups and procedures da X 3


da
wi i (6)
dt i1
dt
2.3.1. TGA-MS (thermogravimetric analysis coupled to mass
spectrometry)
Where wi is the mass fraction of each pseudo-component.
Thermal decomposition and volatiles evolution of both wood
and bark, were analyzed using TGA-MS. Weight loss as function of
2.3.2.1. Numerical method. Differential thermogravimetric data
temperature was recorded in a thermobalance (Netzsch, model
acquired from TGA experiments was represented as a conversion
STA 409 PC) from 20  C to 600  C at 10  C/min under nitrogen
rate as a function of temperature. This function was used to adjust
atmosphere (70 mL/min). Volatiles composition at the outlet of
the kinetic parameters (Ai, Eai and ni) by simulating the individual
the thermobalance was recorded by a mass spectrometer (QMS
olos, Netzsch) by following the evolution of principal contribution of each pseudo-component to the total conversion
403C Ae
rate. This was carried out by a nonlinear optimization procedure,
products [2 (H2), 18 (H2O), 28 (CO), 32 (CH3OH), 44 (CO2), 46
which minimizes the following objective function:
(CH2O2), 60 (C2H4O2), 96 (C5H4O2)]. Two runs of this analyses
were performed in order to verify the reproducibility of the "    #2
assays. XN
dai dai
OF  (7)
J1
dt exp dt calc

Table 1
Ultimate analysis and heating value of E. globulus and E. nitens (wood and bark). Where subscripts exp and calc refers to experimental and calculated
values and N is the number of experimental points.
Biomass Ultimatea Chemical formula
b
C H N O Ash HHV
2.3.3. Torrefaction. Lab-scale facility
(%) (%) (%) (%) (%) (MJ/kg)
After studying the fundamentals of thermoconversion of
E. globulus 43.5 6.2 0.02 50.28 1.4 18.17 C3.63H6.22N0.01O3.14
E. globulus and E. nitens based on results from TGA-MS analyses,
Bark. E. globulus 43.6 8.14 0.43 47.83 9.0 13.34 C3.63H8.1N0.03O2.99
E. nitens 48.5 5.3 1.3 44.9 0.28 18.36 C4H5.3N0.1O2.8 torrefaction is studied in a Lab-scale facility. Lab-scale tests were
Bark. E. nitens 41.4 5 1.6 51.9 6.7 13.24 C3.45H5N0.11O3.25 performed within the torrefaction temperatures region, dened on
a
Standard Deviations (s) %C 0.25, %H 0.208, %N 0.06, %Ash 0.08,
the basis of the decomposition phases identied during the TGA-
HHV(MJ/kg) 0.14. MS study. The laboratory setup (See Fig. 1) consists of a custom-
b
Oxygen is calculated by difference from C, H, N. built reactor placed within three heating zone electric oven
1734 L.E. Arteaga-Perez et al. / Energy 93 (2015) 1731e1741

Table 2
Chemical composition on dry basis of E. globulus and E. nitens (wood and bark).

Biomass Atomic balance Literature Ref.

Cellulose (%) Hemicellulose (%) Lignin (%) Cellulose (%) Hemicellulose (%) Lignin (%)

Wood 51.1 26.7 22.2 51 26 23 [27,28]


Bark 48 26 26 48 26.1 25.9 [27,28]
Formula (in wood) C6H10O5 C5H10O5 C9H8.5O2.5(OHph)0.53(OCH3)1.74 S/G Lignin: 80/20
Formula (in Bark) C6H10O5 C5H10O5 C9H9.1O2.3(OHph)0.31(OCH3)1.53 S/G Lignin: 73/23

(3.2 kW). The setup is also equipped with a gas preheating system Liquids collected during torrefaction, were stored at 3e5  C until
(rst zone), a high temperature lter (0.174 m in length and 0.026 m they were analyzed. The water content of the aqueous fraction was
in diameter), a water-cooled condenser running at 4  C, and an determined by titration according to Karl-Fisher methodology in a
electrostatic precipitator (operating voltage: 15 kV). The reactor is a moisture titrator (Mettler Toledo C30) [33]. Moreover, quantitative
0.54 m long stainless steel (316S) with an inner diameter of organic composition was measured in a (High Precision Liquid
0.0508 m. It is operated vertically and is equipped with an internal Chromatograph) HPLC. This HPLC is equipped with an auto-
support (porous frit with perforations of 0.001 m) where biomass sampler (SIL-20A), a pump (LC-20AT), an oven (CTO-20AC) and
samples are placed. The temperature prole inside the reactor is two detectors, a UV (SPD-20AV) and a refractive index detector
monitored at three points, using K-type thermocouples: at the (RID-10) respectively. Two series-connected columns (Rezex ROA-
bottom (T1), in the biomass bed (T2) and at the volatiles exit (T3). Organic Acid H) and a guard column (SUPELGUARD C6H10) were
Heat transfer is favored by a constant ow of pre-heated nitrogen used. The mobile phase was H2SO4 (0.0075 M). Approximately 0.5 g
(1e2 L/min). All the lines and equipment within the experimental of a sample was dissolved in 5 ml of the mobile phase and then it
setup are well insulated. was well mixed with a vortex mixer for 20e30 min. The resulting
Two sets of experimental conditions were studied: residence solution, was then ltered through a (polyethersulfone) PES
times of 15 and 30 min, and temperatures of 250 and 280  C for four- Membrane 0.45 mm lter; 10 mL were injected into the HPLC. The
category variables (E. nitens and E. globulus wood and bark) corre- oven temperature in all the analyses was 75  C.
sponding to a 22 factorial design each. Temperature was increased Permanent gases (namely CO, CO2 and CH4) were analyzed by an
at a rate of 10  C/min from room temperature to the torrefaction AEROGRAPH GC (model A-90-P), equipped with a packed column
temperature and residence time was computed once the reaction carboxen 1000 of 4.6 m Length.
temperature was reached. After each run, the nitrogen ow was All the experiences were fully randomized to avoid the effect of
interrupted and the reactor was rapidly cooled by an external ow lurking variables and two replicates of each analysis were carried
of air. Experiments were carried out using 20 g of biomass (previ- out.
ously sieved to 2e4 mm size); particle sizes were selected to reduce
the transport resistances, as reported in Bates et al. [16]. 2.5. Quantication parameters

The effect of torrefaction process is evaluated through mass (Eq.


2.4. Methods for solid, liquid and gases analyses (8)) yield, energy density factor (Eq. (9)) and energy yield (Eq. (10))
[15,17,18].
After each run, solids were analyzed according to the procedures
above described for proximate analysis and higher heating value. Mass Yield : MY Mt =Mo daf (8)

Fig. 1. Experimental setup. (1) Nitrogen preheater, (2) reactor, (3) high temperature lter, (4) condenser coupled to a cooling bath, (5) electrostatic precipitator, (6) gas sampling
point.
L.E. Arteaga-Perez et al. / Energy 93 (2015) 1731e1741 1735

hemicellulose, cellulose and lignin [14,34,35], hence their estab-


Energy density factor : EDF HHVt =HHVo daf (9)
lishment is crucial for following the kinetics of the process. Values
of standard deviation between the run and it replicates are
Energy Yield : EY MY*EDF (10) embedded in the same gure.
First phase, below 200  C, is common for all tested samples and
it corresponds to the removal of both physical and chemical
3. Results and discussion bounded moisture [5]. The second, termed as active pyrolysis
zone, starts roughly at 200  C and last until Toffset (~400  C). Spe-
3.1. TGA-MS cically, in this second region, a difference between both TGA and
DTG curves for wood and bark is evident. The lower DTG peaks for
Thermogravimetric experiments are analyzed through the bark, could be explained by its higher content of ash (84e94%
characteristic points dened by Gronli et al. [32] as: higher than wood) and also to the compositional differences among
Tonset (hc) is the extrapolated onset temperature calculated from both lignins (S/G: 80/20 for wood and S/G: 73/27 for bark). S-type
the partial peak that results from the decomposition of the hemi- lignin is more reactive than H-type and G-type; therefore it is rst
cellulose component. consumed during pyrolysis. Furthermore, Fig. 2 shows a slight shift
(dX/dt)sh and Tsh are the overall maximum of the hemicellu- to the left (7.5e12  C) in Tonset for ash, which generally is associated
lose decomposition rate and the corresponding temperature, with the catalytic effect of some inorganics [36] or with the higher
respectively. presence of tanins in the bark [14]. Above 400  C, the rate of weight
(dX/dt)peak and Tpeak are the overall maximum of the mass loss loss is slower, indicating the establishment of a third zone,
rate and the corresponding temperature, respectively. commonly known as passive pyrolysis. Main characteristics
Toffset is the extrapolated offset temperature of the (dX/dt) points for the thermal decomposition of E. globulus and E. nitens are
curves. This value describes the end of the cellulose decomposition summarized in Table 3 and used as reference for studying the
and of active pyrolysis zone. devolatilization kinetics. Values in Table 3 are the mean and stan-
X600 is the percentage of char yield at 600  C. dard deviation of those deduced from two replicates of the TGA
study.
3.1.1. TGA characterization curves The solid product remaining after TGA tests (X600) is 8.7e8.9%
Mass loss and derivative mass loss curves (Fig. 2) exhibit the higher for bark. This difference relies on the ash content of the
typical behavior of lignocellulosic materials, with three decompo- samples and it is relevant for torrefaction and further energy con-
sition phases. These phases are generally related to the content of version processes, as was previously discussed by Almeida et al.
[15].

3.1.2. Volatile evolution


Volatile species from torrefaction are of immense importance
for process synthesis and for energy management within torre-
faction plants. Most common practices propose using these com-
pounds in post-combustion units to fulll energy requirements for
drying and/or torrefaction [6,12]. Therefore, their yield and
composition may inuence on the plant layout and also on the
energy integration strategy to reach the so-called auto-sustainable
energy point. Volatiles comprises a condensable phase and per-
manent gases, their proportion and composition depends mainly
on feedstock nature (biomass source) and reaction conditions
(temperature and residence time) [8,37,38]. Main organics detected
by Mass Spect. during thermal decomposition of Eucalyptus wood
and bark are represented in Fig. 3a and b. All signals are normalized
by the initial weight of biomass, no corrections for heating rate
were needed because all tests were carried out at 10  C/min.
Maximum peaks for carboxylic acids and phenolic compounds
are registered at very similar temperatures among the studied
species and their respective samples of wood and bark. Most of the
compounds are detected in the active pyrolysis zone (200e400  C)
and they exhibit a double peak, the rst corresponding to devola-
tilization stages at lower temperatures (<310  C) and the second to
the pyrolysis (300e400  C) of cellulose. These peaks are essential to
estimate the most feasible operation temperatures for torrefaction,
as they can relate the chemical effect of the process severity with
the solid product energy content. Acetic acid (C2H4O2), formic acid
(CH2O2), phenol (C6H6O) and furfural (C5H4O2) are the main
products from Eucalyptus torrefaction. Their formation during the
treatment of other biomass sources have been also reported (wil-
low and beech [39], Nigerian woods [40], bamboo [41], wood beech
[42]). Water and acetic acid are formed at low temperatures; the
former by dehydration and liberation of side hydroxyl groups from
cellulose and the later by hydrolysis of acetyl groups (COCH3) from
Fig. 2. TGA and DTG curves for wood and bark from E. nitens (a) and E. globulus (b). hemicellulose. Hemicellulose in eucalyptus (hardwood) is basically
1736 L.E. Arteaga-Perez et al. / Energy 93 (2015) 1731e1741

Table 3
TGA characteristic points and char yield at 600  C for E. globulus and E. nitens.

Parameters E. globulus Bark. globulus E. nitens Bark. nitens

Tonset ( C) 217.5 1.1 210.0 1.1 222.2 1.3 210.2 1.3


Tsh ( C) 293.5 1.1 301.5 1.1 298.2 1.4 293.2 1.3
Tpeak ( C) 358.0 1.2 359.5 1.3 363.4 2.1 362.3 2.1
Toff ( C) 400.1 2.1 397.8 1.8 396.6 1.7 400.0 1.8
(dX/dt)sh (%/min) 4.02 0.03 2.6 0.02 2.9 0.01 4.2 0.02
(dX/dt)peak (%/min) 12.9 0.08 8.6 0.2 10.6 0.2 13.8 0.09
X600 (%) 21.9 0.2 30.8 0.1 21.2 0.1 29.9 0.2

tests was marginal (See Table 4). This could be explained by the
inuence of some inorganics, particularly potassium (K) [44]. It is
well known that potassium in biomass is mainly found dissolved in
the organic phase (e.g. as carbonate, chloride or hydroxyde) or as a
cation attached to the carboxylic or other functional groups. Po-
tassium is released during thermal decomposition of biomass, in
two stages, the rst taking place between 250 and 500  C, for po-
tassium associated to the organic phase and above 500  C for
inorganic potassium (KCl) [45]. Then the potassium species
dramatically promotes the degradation of Levoglucosan at low
temperatures (<350  C) [46]. This fact was recently corroborated by
Hwang et al. [47] who studied the effect of potassium content on
the formation of Levoglucosan during the pyrolysis of Yellow
Poplar. They reported that Levoglucosan concentration decreased
by 50% by adding only 0.5 mg/g of potassium to a demineralized
sample. According to Miranda et al. [27] and Lo pez-Gonz alez et al.
[48], Ca and K are the major inorganics in bark from E. globulus and
nitens, which supports the previous hypothesis. Table 4 shows the
maximum peaks composition of the condensable fraction of vola-
tiles, at a typical torrefaction temperature (280  C).
Formation of CO, CO2 and CH2O2 below 300  C is associated with
the decomposition of xylans [40]; hence, their presence in the
volatiles produced from biomass torrefaction is a function of the
chemical composition of feedstock.

3.2. Devolatilization kinetics

Fig. 4a and b, summarize the adjustment of the experimental


DTG curve for E. globulus and E. nitens, to a rst order multicom-
ponent devolatilization model. Two set of experimental data used
to deduce the kinetic parameters were compared statistically by an
F-test considering the hypothesis that there are no statistical dif-
ferences among their standard deviations. Since the condence
intervals (in all cases) for the ratio of the variances contained the
Fig. 3. Weight loss and volatile evolution during Eucalyptus thermal decomposition.
value 1 the hypothesis was not rejected, hence there are not sig-
(a) E. nitens and (b) E. globulus. Continuous and dashed lines corresponds to wood and nicant statistical differences between the standard deviation of
bark respectively. both samples at a 95% condence level. Values of standard devia-
tion are embedded within Fig. 4a and b, and the deduced kinetic
parameters (pre-exponential factor and activation energies) are
formed by glucuronoxylans (11e22 wt%) [43], which are the most presented in Table 5.
reactive components within the torrefaction temperature range, Mass fractions of pseudo-components determined by atomic
and hence they degrade faster than any other solid wood compo- balance (Table 2) were used as starting search point for nonlinear
nent [5,9]. Acetoxy- and methoxy-groups attached to the poly- regression between experimental and modeled data. Calculated
sugars (in particular to xylose units), trends to form acetic acid and curves were then optimized by minimizing the objective function
methanol when the wood is heated to temperatures above 200  C, represented by Eq. (7). The estimated hemicellulose and cellulose
this explain the double-peak in the acetic acid prole. Between fractions were around 25e27.9% and 47.3e49.9% respectively,
250  C and 398  C, more complex structures (from cellulose) start which is in good agreement with literature reports (s < 2%) [48].
decomposing and polymerizing through cross-linking reactions For Lignin a good adjustment was also reached, but slightly higher
and scission of molecular bonds. In particular, the cleavage of the difference (1e5%) in reference to the values obtained by Miranda
1,4 glucosidic linkage in the cellulose structure initiates the for- et al. [27] was found. The last could be explained by the approxi-
mation of Levoglucosan (C6H10O5) which concentration is expected mations of the model, which neglects the effect of extractives (they
to increase in certain range of temperatures to further polymerize are considered within lignin). In spite of these differences, the
to other complex organics such as hydroxyacetaldehyde and acetal. optimized correlation curves superpose the experimental DTG with
Nevertheless, curiously, Levoglucosan formation during TGA-MS less than 1% deviation.
L.E. Arteaga-Perez et al. / Energy 93 (2015) 1731e1741 1737

Table 4
Maximum peaks of volatiles from torrefaction of E. globulus and E. nitens at 280  C.

Compound Formula M Ion signal (1010 a.u)

E. globulus Bark. globulus Bark. nitens E. nitens

Hydrogen H2 2 0.9 0.01 0.8 0.03 0.7 0.03 1.3 0.01


Methyl radical CH3- 15 0.8 0.02 0.9 0.02 0.7 0.02 1.2 0.01
Water H2O 18 3.2 0.01 4.0 0.03 2.7 0.03 6.2 0.01
Carbon monoxide CO 27 5.7 0.04 5.5 0.02 5.0 0.02 8.7 0.02
Methanol CH3OH 32 0.9 0.03 1.0 0.03 1.2 0.03 1.9 0.02
Carbon dioxide CO2 44 1.6 0.02 1.5 0.03 0.9 0.02 2.7 0.02
Formic acid CH2O2 46 (1.3 0.01)E-2 (1.2 0.03)E-2 (7.5 0.01)E-3 (2.7 0.02)E-2
Acetic acid C2H4O2 60 (1.6 0.02)E-2 (1.5 0.03)E-2 (4.9 0.04)E-3 (3.9 0.05)E-2
Hydroxyacetone C3H6O2 74 (2.5 0.01)E-3 (2.5 0.02)E-3 (2.3 0.03)E-3 (2.4 0.04)E-3
Lactic acid C3H6O3 90 (2.2 0.01)E-3 (2.2 0.03)E-3 (2.1 0.03)E-3 (2.1 0.03)E-3
Phenol C6H6O 94 (6.1 0.01)E-3 (8.4 0.02)E-1 (4.4 0.02)E-3 (9.5 0.02)E-3
Furfural C5H4O2 96 (4.2 0.01)E-3 (8.8 0.02)E-1 (2.6 0.02)E-3 (8.9 0.03)E-3
Di-anhydro-hexosan C6H8O4 144 (2.2 0.03)E-3 (4.0 0.02)E-3 (2.0 0.02)E-3 (2.1 0.06)E-3
Levoglucosan C6H10O5 162 (2.2 0.02)E-4 (5.5 0.02)E-4 (2.0 0.02)E-4 (2.1 0.01)E-4

Furthermore, the lower activation energies for cellulose and


hemicellulose decomposition in bark samples, is a response to their
content of ashes. As earlier discussed, some inorganics present in
the ashes (K, Mg and Ca) inuences on the decomposition mech-
anism of biomass, mainly between 250 and 500  C, where they are
released (e.g. potassium). This catalytic effect, results in lower
activation energies and higher carbon yields (X600) at the end of the
tests (See Fig. 2 and Table 5).
Comparing kinetic parameters between studies is a complicated
task, since they can vary widely due to the reaction conditions,
sample nature and even with the calculation method [30]. The re-
sults obtained in this study are similar to that of Bach et al. [49]
(Spruce and Birch), Jeguirim et al. [29] (Arundo donax) and Orfao
et al. [50] (Pine and Eucalyptus) who reported on devolatilization
kinetics of woody biomass, using similar procedures. Among the
several papers reporting on the kinetics of Eucalyptus wood py-
rolysis [50e52], this is the rst study reporting on the devolatili-
zation kinetics coupled with Mass Spect. for both wood and bark.

3.3. Torrefaction assays at lab-scale

Temperature and time intervals were selected to represent both


medium and severe torrefaction ranges. Nevertheless, selection
of specic temperature depends on several factors (scale, speci-
cation and end use of products, etc ). Here, we report on mass and
energy yields, to further relate them and volatile composition with
the feasibility temperature region.

3.3.1. Effect of operation variables. Temperature and residence time


Temperature is the main operation parameter in torrefaction
processes, it dene the thermal degradation of the feedstock and
inuences on mechanism and kinetics of reactions. Chen & Kuo
[53], explained the torrefaction of lignocellulosic materials through
Fig. 4. Modeling of DTG experimental vs pseudocomponents predicted proles. (a) the individual decomposition of its constituents and suggested that
E. nitens, (b) E. globulus.
synergic effect between hemicellulose, cellulose, and lignin can be
neglected. Results of main measures of merit for torrefaction of
wood and bark from E. globulus and E. nitens are presented in Fig. 5.
Estimated activation energies for hemicellulose decomposition, As reaction temperature rises, both mass (90e65%) and energy
were similar for all the studied materials, ranging between 121 and (95e75%) yields are reduced for both wood and bark, due to the
171 kJ/mol. These values respond to their similar contents of dehydration and devolatilization of light organics. Hemicellulose is
hemicellulose as well as to the composition of this polymer. Dif- a lineal polymer which rarely degrade below 220  C, therefore at
ferences between activation energies for bark and wood could be this temperature the heating value and other fuel properties of
also explained by the higher content of extractives in the bark. biomass are affected slightly. Above 250  C, carboxylic acids,
Extractives are mixtures of a variety of chemicals (terpenes, tan- methanol, furfurals and permanent gases (CO and CO2) start pro-
nins, fatty acids, resins, etc.), hence they decompose thermally in a ducing through depolymerization of hemicellulose (near 90% to
wide range of temperatures, such as happens with lignin. 280  C) and partially by cellulose and lignin decomposition (See
1738 L.E. Arteaga-Perez et al. / Energy 93 (2015) 1731e1741

Table 5
Kinetic parameters of devolatilization model for E. globulus and E. nitens (wood and bark).

Parameters E. globulus Bark. globulus E. nitens Bark. nitens

Ea1 (kJ/mol) 144.9 1.9 130.8 2.0 171.2 2.0 121.9 0.6
Ko1 (min1) (1.9E03 1.7) (6.0E02 2.3) (8.2E03 3.4) (1.4E04 1.8)
Ea2 (kJ/mol) 228.2 1.3 226.5 0.7 248.9 0.4 217.9 2.0
Ko2 (min1) (5.20E08 1.0) (4.8E08 6.0) (6.2E08 13) (6.4E07 8.0)
Ea3 (kJ/mol) 91.3 4.1 83.5 2.7 101.6 4.7 90.9 1.6
Ko3 (min1) 5.3 0.3 17.3 0.01 492.0 2.41 0.3
Hemicellulose (%wt.) 27.9 0.07 25.6 0.1 25.3 0.2 26.8 0.1
Cellulose (%wt.) 49.9 0.02 49.9 0.3 47.4 0.1 47.3 0.5
Lignin (%wt.) 22.1 0.05 24.4 0.4 26.7 1.0 25.8 0.4

Fig. 6). From 250  C, carbonyl and carboxyl groups in cellulose


(44 wt.%) are transformed into volatile fractions which can also
react in homogeneous phase to more complex compounds. On the
other hand, lignin, is a three-dimensional phenolic polymer and it
decomposes in a wide range of temperatures (160  Ce900  C),
hence some of phenolic compounds formed at T < 280  C could be
produced from aromatic rings of lignin. Differences of mass and
energy yields between bark samples have the same order of
magnitude than for wood, which relates with their composition
similarities. Lower values of energy yields found for bark can be
associated to its composition, mainly to the ash content as was
previously discussed. Residence time also inuences the mass and
energy yields producing and inverse proportional effect, at a lower
extent than temperature [6].

3.3.2. Volatiles composition


Composition and HHV (higher heating value) of volatiles
released during torrefaction of E. globulus and E. nitens (wood and
bark) were determined at 280  C and 30 min residence time. These
conditions were selected, based on its major effect on the energy
densication factor (1.17e1.27 at 280  C and 1.07e1.15 at 250  C). Fig. 6. Volatile composition and HHV at severe torrefaction conditions. T 280  C and
t 30 min.
Main volatile species (Fig. 6) collected after each treatment were
distributed as Water > Acetic Acid > CO2 > Others (methanol, car- K  
X OH
boxylic acids, etc.), which is similar to that reported by Nocquet AI yk 
et al. [38] and Bates et al. [8]. C db (11)
k1
HHV values of volatiles were higher for wood than for bark, where k kith compound in the volatiles kswater
which is a direct indicator of the energy densication degree and of
biomass degradation during torrefaction. As biomass decomposes, AI (atomic indicator) accounts for the H, O and C atoms being
volatiles are continuously enriched with organics from devolatili- released during torrefaction. By analyzing the denition of AI, it
zation/depolymerization stages while the H/C:O/C ratios (van could be inferred that is desirable to perform torrefaction in a
Krevelen criteria) in the remaining torreed product move towards temperature range at which the released species have higher
the coal region. This behavior can be followed by analyzing the (H O)/C atomic ratios. Nevertheless, AI is a complex function of
atomic distribution in the volatiles through the following indicator: the volatiles and it can vary not only as a consequence of biomass

Fig. 5. Torrefaction Mass and energy Yields. (a) E. nitens, (b) E. globulus.
L.E. Arteaga-Perez et al. / Energy 93 (2015) 1731e1741 1739

a statistical approach to identify a two dimensional region (T vs t)


at which energy yield does not change. Nevertheless, they sug-
gested that this region is not commonly encountered for all
biomass due to their compositional differences, hence they pro-
pose addressing the problem from the phenomenological point of
view.
Here, a simplied procedure based on TGA coupled to Mass
Spect., is used to identify most feasible temperatures to carry out
torrefaction. The identication of such temperature relies on the
denition of torrefaction and pyrolysis peaks [53], which were
identied on the basis of chemical evidences (evolution of volatiles)
and considering the prior denition of the atomic indicator (Eq.
(11)).
Most important products during low-temperature (<350  C)
devolatilization of E. globulus and E. nitens (wood and bark) were
water and carboxylic acids (Fig. 7a and b). The sum of carboxylic
acid signals (CA-signals) is proportional to the AI (ranging from 3 to
4 in the case of formic and acetic acid) while water (W-signal)
peaks above 220  C represents the rst stages of torrefaction
(elimination of chemically bounded moisture [5]), hence they are
used here to identify the feasible temperatures.
As shown, both CA-signals and W-signals for E. nitens and
E. globulus (wood and bark) exhibit two maximum at certain values
of temperatures, rst corresponding to torrefaction (between 290
and 310  C) and the second to low-temperature pyrolysis (between
350 and 373  C). Mass and Energy yields and energy densication
factor for each sample at the so-called peaks are presented in
Table 6. The values encountered here for torrefaction peaks are in
the same order of magnitude of those reported by Poudel et al. [10]
and Strandberg et al. [23].
Furthermore, main measures of merit (MY, EY and EDF) at the
Fig. 7. Identication of torrefaction and pyrolysis peaks, (a) E. nitens, (b) E. globulus. identied feasible torrefaction temperature are encouraging,
especially the energy yield, which ranged between 88 and 94% in all
cases, that is above the theoretical value (90%) indicated in most of
decomposition; as homogeneous phase reactions, metallic species the literature. These values are references because they were ob-
and other parameters can also inuence on it. Therefore, in this tained during a TGA study, for more conscious analysis, differences
study this parameter is used as a qualitative index to follow between TGA-based temperatures and values deduced at higher
degradation and not as a quantitative measure of merit, to compare scales should be checked based on transport limitations [54]. The
treatments. interchangeability of reaction time and temperatures was not
Values of AI in Fig. 6, suggest a higher degradation for bark than addressed as it clearly can be deduced by integrating the rate
for wood, which should be related to its higher content of ash. expression (Eq. (4)), kinetic parameters (Table 5) at the torrefaction
Furthermore, from the process engineering view point, as volatiles temperatures previously reported.
increases their HHV, more heat can be recovered by post-
combustion to fulll energy requirements in both torrefaction
4. Conclusions
and drying processes (~2.9 kW/kg of released moisture) [5,11]. That
is why, feasible operation temperature should be selected by
The objective of this study was to develop a comprehensive
considering (i) feedstock nature, (ii) volatiles composition and (iii)
evaluation of torrefaction of wood and bark from E. nitens and
system mass and energy balances.
E. globulus. These species were selected for their economic
importance in the Chilean forest market (~40%). Results from TGA-
3.3.3. Feasible operation temperatures MS evidenced the importance of carboxylic acids on process per-
As earlier discussed, there is a marked interest in nding the formance, basically due to their potential effect on both: product
feasible temperatures that guarantees a certain energy yield and quality and global energy balance. Two peaks of carboxylic acids
product quality after torrefaction. Strandberg et al. [23], proposed were found in the active pyrolysis zone (250e400  C). These peaks

Table 6
Feasible temperatures for torrefaction of E. globulus and E. nitens (wood and bark).

Torrefaction peak Pyrolysis peak

T HHV MY EY EDF T HHV MY EY EDF


( C) (MJ/kg) (%) (%) e ( C) (MJ/kg) (%) (%) e

E. globulus 305 0.6 19.8 81.5 88.9 1.09 359 0.8 25.1 51.5 71 1.38
Bark globulus 308 0.8 14.4 83.1 89.8 1.08 370 0.9 19.1 48.3 69 1.43
E. nitens 298 0.2 19.3 89.0 93.5 1.05 362.3 1.1 26.4 46.8 67.5 1.44
Bark nitens 309 0.3 14.2 85.0 91.3 1.07 372 1.3 19.2 47.0 68.2 1.45
1740 L.E. Arteaga-Perez et al. / Energy 93 (2015) 1731e1741

were associated to torrefaction and pyrolysis respectively and [21] Chen W-H, Huang M-Y, Chang J-S, Chen C-Y, Lee W-J. An energy analysis of
torrefaction for upgrading microalga residue as a solid fuel. Bioresour
were used to dene the most feasible torrefaction temperature.
Technol 2015;185:285e93. http://dx.doi.org/10.1016/j.biortech.2015.02.-
Energy yield at such a temperature, ranges between 93.5% and 095.
88.9%, and it was reached in a relatively narrow interval [22] Uemura Y, Matsumoto R, Saadon S, Matsumura Y. A study on torrefaction of
(298e310  C). Further research is required for studying the effect Laminaria japonica. Fuel Process Technol 2015;138:133e8. http://dx.doi.org/
10.1016/j.fuproc.2015.05.016.
of transport limitations on applying these criteria for process [23] Strandberg M, Olofsson I, Pommer L, Wiklund-Lindstro m S, berg K, Nordin A.
design and scaling-up. Effects of temperature and residence time on continuous torrefaction of
spruce wood. Fuel Process Technol 2015;134:387e98. http://dx.doi.org/
10.1016/j.fuproc.2015.02.021.
Acknowledgments [24] D3172-89 A. Standard practice for proximate analysis of coal and coke. 2013.
[25] ABNT 8112 Ae AB de NTe N. Carva ~o vegetal e analise imediata 1968.
[26] 58655-13 A. Standard test method for gross caloric value of coal and coke.
This work has been nancially supported by the project FONDEF 2013.
B09I1015 and Basal CONICYT PFB-27 of the Technological Devel- [27] Miranda I, Gominho J, Mirra I, Pereira H. Fractioning and chemical charac-
terization of barks of Betula pendula and Eucalyptus globulus. Ind Crops Prod
n, Chile. Authors would like
opment Unit of University of Concepcio 2013;41:299e305. http://dx.doi.org/10.1016/j.indcrop.2012.04.024.
to acknowledge to Francisco Niefercol, bachelor student from [28] Esteves Costa CA, Rodrigues Pinto PC, Rodrigues AE. Evaluation of chemical
Environmental Chemistry Department of University Federico Santa processing impact on E. globulus wood lignin and comparison with bark
lignin. Ind Crops Prod 2014;61:479e91.
Mara for giving us support during experimental analysis. [29] Jeguirim M, Trouve  G. Pyrolysis characteristics and kinetics of Arundo donax
using thermogravimetric analysis. Bioresour Technol 2009;100:4026e31.
http://dx.doi.org/10.1016/j.biortech.2009.03.033.
References [30] White JE, Catallo WJ, Legendre BL. Biomass pyrolysis kinetics: a comparative
critical review with relevant agricultural residue case studies. J Anal Appl
[1] Energ M, Ministerio de Energa. Balanc Energe tico Nac. 2014. http://www. Pyrol 2011;91:1e33. http://dx.doi.org/10.1016/j.jaap.2011.01.004.
minenergia.cl/documentos/balance-energetico.html. [31] Branca C, Blasi Di. Parallel and series-reaction mechanisms of wood and char
[2] Daz-Robles L, Fu JS, Vergara-Ferna ndez A, Etcharren P, Schiappacasse LN, combustion. Therm Sci 2004;8:51e64.
Reed GD, et al. Health risks caused by short term exposure to ultrane par- [32] Gronli MG, Varhegyi G, Di Blasi C. Thermogravimetric analysis and devolati-
ticles generated by residential wood combustion: a case study of Temuco, lization kinetics of wood. Ind Eng Chem Res 2002;41:4201e8.
Chile. Environ Int 2014;66:174e81. http://dx.doi.org/10.1016/ [33] Van de Voort FR, Sedman J, Cocciardi R, Juneau S. An automated FTIR method
j.envint.2014.01.017. for the routine quantitative determination of moisture in lubricants: an
[3] INFOR. Forestry Institute. Forest statistics. 2014. http://wef.infor.cl/. alternative to Karl Fischer titration. Talanta 2007;72:289e95. http://
[4] Pacheco M. Agenda de energa. Un desafo pas, Progereso para todos. Min- dx.doi.org/10.1016/j.talanta.2006.10.042.
isterio de Energa de la Repblica de Chile; 2013. http://www.minenergia.cl/ [34] Anca-Couce A, Berger A, Zobel N. How to determine consistent biomass py-
documentos/otros-documentos/agenda-de-energia-un-desao-pais.html. rolysis kinetics in a parallel reaction scheme. Fuel 2014;123:230e40. http://
[5] Basu P. Biomass gasication, pyrolysis and torrefaction. Practical design and dx.doi.org/10.1016/j.fuel.2014.01.014.
theory. 2nd ed. New York: Elsevier Ltd; 2013. [35] Li Z, Zhao W, Meng B, Liu C, Zhu Q, Zhao G. Kinetic study of corn straw py-
[6] Chen W-H, Peng J, Bi XT. A state-of-the-art review of biomass torrefaction, rolysis: comparison of two different three-pseudocomponent models. Bio-
densication and applications. Renew Sustain Energy Rev 2015;44:847e66. resour Technol 2008;99:7616e22. http://dx.doi.org/10.1016/
http://dx.doi.org/10.1016/j.rser.2014.12.039. j.biortech.2008.02.003.
[7] Pighinelli ALMT, Boateng A, Mullen C, Elkasabi Y. Evaluation of Brazilian [36] Vamvuka D, Karouki E, Sfakiotakis S. Gasication of waste biomass chars by
biomasses as feedstocks for fuel production via fast pyrolysis. Energy Sustain carbon dioxide via thermogravimetry. Part I: effect of mineral matter. Fuel
Dev 2014;21:42e50. http://dx.doi.org/10.1016/j.esd.2014.05.002. 2011;90:1120e7. http://dx.doi.org/10.1016/j.fuel.2010.12.001.
[8] Bates RB, Ghoniem AF. Biomass torrefaction: modeling of volatile and solid [37] Prins MJ, Ptasinski KJ, Janssen FJJG. Torrefaction of wood. J Anal Appl Pyrol
product evolution kinetics. Bioresour Technol 2012;124:460e9. http:// 2006;77:35e40. http://dx.doi.org/10.1016/j.jaap.2006.01.001.
dx.doi.org/10.1016/j.biortech.2012.07.018. [38] Nocquet T, Dupont C, Commandre J-M, Grateau M, Thiery S, Salvador S. Vol-
[9] Prins MJ, Ptasinski KJ, Janssen FJJG. Torrefaction of wood. J Anal Appl Pyrol atile species release during torrefaction of wood and its macromolecular
2006;77:28e34. http://dx.doi.org/10.1016/j.jaap.2006.01.002. constituents: part 1 e experimental study. Energy 2014;72:180e7. http://
[10] Poudel J, Ohm T-I, Oh SC. A study on torrefaction of food waste. Fuel dx.doi.org/10.1016/j.energy.2014.02.061.
2015;140:275e81. http://dx.doi.org/10.1016/j.fuel.2014.09.120. [39] Prins MJ, Ptasinski KJ, Janssen FJJG. More efcient biomass gasication
[11] Tumuluru J, Sokhansanj S, Wright CT, Boardman RD, Hess JR. Review on via torrefaction. Energy 2006;31:3458e70. http://dx.doi.org/10.1016/
biomass torrefaction process and product properties and design of moving j.energy.2006.03.008.
bed torrefaction system model development. 2011 Louisville, Kentucky, [40] Lasode O, Balogun AO, McDonald AG. Torrefaction of some Nigerian ligno-
August 7eAugust 10, 2011. 2011. http://dx.doi.org/10.13031/2013.37192. cellulosic resources and decomposition kinetics. J Anal Appl Pyrol 2014;109:
[12] Nhuchhen D, Basu P, Acharya B. A comprehensive review on biomass torre- 47e55. http://dx.doi.org/10.1016/j.jaap.2014.07.014.
faction. Int J Renew Energy Biofuels 2014;2014:1e56. http://dx.doi.org/ [41] Chen W-H, Liu S-H, Juang T-T, Tsai C-M, Zhuang Y-Q. Characterization of solid
10.5171/2014.506376. and liquid products from bamboo torrefaction. Appl Energy 2015. http://
[13] Van der Stelt MJC, Gerhauser H, Kiel JH, Ptasinski KJ. Biomass upgrading by dx.doi.org/10.1016/j.apenergy.2015.03.022.
torrefaction for the production of biofuels: a review. Biomass Bioenergy [42] Nocquet T, Dupont C, Commandre J-M, Grateau M, Thiery S, Salvador S. Vol-
2011;35:3748e62. http://dx.doi.org/10.1016/j.biombioe.2011.06.023. atile species release during torrefaction of biomass and its macromolecular
[14] Burhenne L, Messmer J, Aicher T, Laborie M-P. The effect of the biomass constituents: part 2 e modeling study. Energy 2014;72:188e94. http://
components lignin, cellulose and hemicellulose on TGA and xed bed pyrol- dx.doi.org/10.1016/j.energy.2014.05.023.
ysis. J Anal Appl Pyrol 2013;101:177e84. http://dx.doi.org/10.1016/ [43] Chen Z, Hu TQ, Jang HF, Grant E. Modication of xylan in alkaline treated
j.jaap.2013.01.012. bleached hardwood kraft pulps as classied by attenuated total-internal-
[15] Almeida G, Brito JO, Perre  P. Alterations in energy properties of eucalyptus reection (ATR) FTIR spectroscopy. Carbohydr Polym 2015;127:418e26.
wood and bark subjected to torrefaction: the potential of mass loss as a http://dx.doi.org/10.1016/j.carbpol.2015.03.084.
synthetic indicator. Bioresour Technol 2010;101:9778e84. http://dx.doi.org/ [44] Richards GN, Zheng G. Inuence of metal ions and of salts on products from
10.1016/j.biortech.2010.07.026. pyrolysis of wood: applications to thermochemical processing of newsprint
[16] Bates RB, Ghoniem AF. Biomass torrefaction: modeling of reaction thermo- and biomass. J Anal Appl Pyrol 1991;21:133e46.
chemistry. Bioresour Technol 2013;134:331e40. http://dx.doi.org/10.1016/ [45] Yu C, Zhang W. Progress in thermochemical biomass conversion. 1st ed. Ox-
j.biortech.2013.01.158. ford: Blackwell Science Ltd.; 2001.
[17] Medic D, Darr M, Shah A, Potter B, Zimmerman J. Effects of torrefaction pro- [46] Eom IY, Kim JY, Kim TS, Lee SM, Choi D, Choi IG, et al. Effect of essential
cess parameters on biomass feedstock upgrading. Fuel 2012;91:147e54. inorganic metals on primary thermal degradation of lignocellulosic biomass.
http://dx.doi.org/10.1016/j.fuel.2011.07.019. Bioresour Technol 2012;104:687e94. http://dx.doi.org/10.1016/
[18] Benavente V, Fullana A. Torrefaction of olive mill waste. Biomass Bioenergy j.biortech.2011.10.035.
2015;73:186e94. http://dx.doi.org/10.1016/j.biombioe.2014.12.020. [47] Hwang H, Oh S, Cho TS, Choi IG, Choi JW. Fast pyrolysis of potassium
[19] Li M-F, Li X, Bian J, Xu J-K, Yang S, Sun R-C. Inuence of temperature on impregnated poplar wood and characterization of its inuence on the for-
bamboo torrefaction under carbon dioxide atmosphere. Ind Crops Prod mation as well as properties of pyrolytic products. Bioresour Technol
2015;76:149e57. http://dx.doi.org/10.1016/j.indcrop.2015.04.060. 2013;150:359e66. http://dx.doi.org/10.1016/j.biortech.2013.09.132.
[20] Sermyagina E, Saari J, Zakeri B, Kaikko J, Vakkilainen E. Effect of heat inte- [48] Lo pez-Gonzalez D, Fernandez-Lopez M, Valverde JL, Sanchez-Silva L. Gasi-
gration method and torrefaction temperature on the performance of an in- cation of lignocellulosic biomass char obtained from pyrolysis: kinetic and
tegrated CHP-torrefaction plant. Appl Energy 2015;149:24e34. http:// evolved gas analyses. Energy 2014;71:456e67. http://dx.doi.org/10.1016/
dx.doi.org/10.1016/j.apenergy.2015.03.102. j.energy.2014.04.105.
L.E. Arteaga-Perez et al. / Energy 93 (2015) 1731e1741 1741

[49] Bach Q-V, Tran K-Q, Skreiberg , Khalil RA, Phan AN. Effects of wet torre- termination des parame
[52] Kifani-Sahban F, Belkbira L, Zoulalian A. De tres cine
-
faction on reactivity and kinetics of wood under air combustion conditions. tiques de la pyrolyse lente de l'eucalyptus Marocain. Thermochim Acta
Fuel 2014;137:375e83. http://dx.doi.org/10.1016/j.fuel.2014.08.011. 1996;289:33e40.
~o JJM, Antunes FJ, Figueiredo JL. Pyrolysis kinetics of lignocellulosic
[50] Orfa [53] Chen W-H, Kuo P-C. Torrefaction and co-torrefaction characterization of
materialsdthree independent reactions model. Fuel 1999;78:349e58. http:// hemicellulose, cellulose and lignin as well as torrefaction of some basic con-
dx.doi.org/10.1016/S0016-2361(98)00156-2. stituents in biomass. Energy 2011;36:803e11. http://dx.doi.org/10.1016/
 P. Inverse analysis of
[51] Cavagnol S, Sanz E, Nastoll W, Roesler JF, Zymla V, Perre j.energy.2010.12.036.
wood pyrolysis with long residence times in the temperature range 210 to [54] Bates RB, Ghoniem AF. Modeling kinetics-transport interactions during
290  C: selection of multi-step kinetic models based on mass loss residues. biomass torrefaction: the effects of temperature, particle size, and moisture
Thermochim Acta 2013:574. http://dx.doi.org/10.1016/j.tca.2013.09.009. content. Fuel 2014;137:216e29. http://dx.doi.org/10.1016/j.fuel.2014.07.047.

Вам также может понравиться