- Email: [email protected]
A comprehensive study on pyrolysis kinetics of microalgal biomass
Energy Conversion and Management 131 (2017) 109–116
Contents lists available at ScienceDirect
Energy Conversion and Management journal homepage: www.elsevier.com/locate/enconman
A comprehensive study on pyrolysis kinetics of microalgal biomass Quang-Vu Bach, Wei-Hsin Chen ⇑ Department of Aeronautics and Astronautics, National Cheng Kung University, Tainan 701, Taiwan
a r t i c l e
i n f o
Article history: Received 3 August 2016 Received in revised form 14 October 2016 Accepted 31 October 2016 Available online 10 November 2016 Keywords: Microalgal biomass Pyrolysis Kinetic modeling Activation energy Thermogravimetric analysis
a b s t r a c t Pyrolysis of microalgal biomass for biofuels production has attracted much attention. However, detailed degradation mechanism and kinetics of the process have not been fully explored yet. In this study, a nonisothermal pyrolysis of microalga Chlorella vulgaris ESP-31 is thermogravimetrically investigated. Several kinetic models, from a single reaction to seven parallel reactions, are tested to fit the experimental pyrolysis data for finding out the optimal pyrolysis model. The results show that the pyrolysis behavior of the microalga is somewhat different from that of lignocellulosic biomass, stemming from the inherent difference in their compositions. Overall, the kinetic modeling processes show that increasing the number of reactions improves the model fit quality. Curve fitting results indicate that the models consisting of three and less than three reactions are not suitable for microalga pyrolysis. The four-reaction model, via considering the pyrolysis of carbohydrate, protein, lipid and others, can be employed for modeling the thermal degradation; however, it cannot precisely predict the thermal degradation of the shoulder and the small peak. The conducted seven-reaction model further partitions the decomposition processes of carbohydrate and protein into two stages, and explains the thermal degradation well. The model indicates that the devolatilization peak is attributed to the combined degradation of Protein I and Carbohydrate II. The seven-reaction model offers the highest fit quality and is thus recommended for predicting the microalga pyrolysis processes. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Biomass is a potential source of renewable energy, which is receiving a great deal of attention due to its advantages over other alternative energy sources [1]. Biomass derived fuels can be categorized into first, second, and third generation biofuels. First generation biofuels are produced from food crops such as sugarcane, corn, potato, wheat, and sugar beet. This directly affects human food supplies and biodiversity [2]. Second generation biofuels are produced from non-food or lignocellulosic biomass and biomass residues such as switchgrass, grass, jatropha, miscanthus, husk, wood chips, leaves and stump. Non-food biomass does not threaten food supplies; however, its planation will compete with arable lands for food crops and influence nutrient cycles and soil conservation [3]. The disadvantages of first and second generation biofuels can be overcome by exploiting algal biomass, which is considered as the potential feedstock for third generation biofuels. Unlike land-based lignocellulosic biomass, algal biomass can be
⇑ Corresponding author. E-mail addresses: (W.-H. Chen).
[email protected],
http://dx.doi.org/10.1016/j.enconman.2016.10.077 0196-8904/Ó 2016 Elsevier Ltd. All rights reserved.
[email protected]
cultivated in fresh, saline, or waste water [4], and thus requires no land allocation. Moreover, compared to terrestrial biomass, algal biomass, including microalgae and macroalgae, has higher growth rate and photosynthetic efficiency, and hence can absorb more CO2 during its growth [5,6]. This enables more efficient reduction in the greenhouse gas emissions [7], and even achieves CO2 utilization [8]. Many thermochemical conversion processes can be employed to convert microalgal biomass into energy-dense biofuels such as bio-oils and biochars [9–11]. Pyrolysis is a thermal conversion route in which microalga is heated at elevated temperatures (400–800 °C) and in an oxygen-free atmosphere [12]. The pyrolysis products include bio-oils and biochars as well as a small fraction of permanent gases, and their distribution usually depends on the operating parameters such as temperature (or heating rate) and residence time [13]. Pyrolysis at high temperatures with short residence times yields more bio-oil. On the other hand, biochar production is favored for pyrolysis at low temperatures with long residence times. Both bio-oils and biochars can be used directly for combustion to produce heat and power. Bio-oils can be further upgraded to produce liquid transport fuels and bio-chemicals, while biochars can be used as activated carbon, soil enhancer, fertilizer, etc.
110
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116
Nomenclature Abbreviations daf dry and ash free basis db dry basis DTG differential thermogravimetric TG thermogravimetric TGA thermogravimetric analysis wt weight LTSC/HTSC low/high thermal stable components Comp component FWO Flynn-Wall-Ozawa model KAS Kissinger-Akahira-Sunose model DAEM distributed activation energy model SSGM single-step global model Cal. calculated data Exp. experimental data
As an important conversion of microalgal biomass into biofuels, pyrolysis has received great attention recently. The number of works on microalgae pyrolysis has been increasing in order to assess the chemical, physical and fuel properties of microalgae bio-oils and to compare with lignocellulosic biomass bio-oils. It has been reported that bio-oils produced from microalgal biomass are more stable than those from lignocellulosic biomass [10]. Moreover, microalgae bio-oils contain less oxygen and have greater heating value than lignocellulosic biomass bio-oils [14]. In addition to researches on the properties of microalgae bio-oils, kinetic studies have been also conducted to investigate the pyrolysis behavior of microalgae. In practice, pyrolysis kinetics is highly related to chemical reaction control and reactor design. For this purpose, thermogravimetric technique has become a proven method and been widely used by many researchers [15–21]. It has been pointed out that microalgae pyrolysis undergoes three stages. In the first stage, water is removed from microalgae. The mass loss in this stage is associated with the moisture content of the feedstock. Main microalgal components, including carbohydrate, protein and lipid, are thermally degraded in the second stage, which accounts for most of the mass loss during pyrolysis. In the last stage, only slight mass loss is observed which is attributed to the decomposition of carbonaceous matters in the solid residues. In some studies [22], the thermal decomposition of lipid is partitioned from that of carbohydrate and protein, whereby four-stage thermal degradation of microalgae is defined. Although pyrolysis kinetic researches of microalgal biomass are active, most of the available studies employed simplified kinetics models. In other words, deep knowledge about the reaction mechanism from the obtained kinetic data is insufficient. A number of recent studies regarding pyrolysis kinetic of various microalgae are summarized in Table 1 in which a variety of methods were conducted. As can be seen from the table, the available studies employed several approximation and transformation of the Arrhenius expression to estimate the values of activation energy and pre-exponential factor at different conversion rates. As a result, the values of the two kinetic parameters are normally in a range, from which mean values can be calculated and they represent the whole microalgal biomass, even though the microalgae contain several components whose reactivity and kinetic parameters are different. In addition, these models cannot reproduce the predicted thermogravimetric curves or provide any information about the fit quality between the modeled and experimental data. Consequently, the evaluation of these models is difficult. More complex kinetic models, e.g., multiple reaction models, are available and
Symbols A c kðTÞ or k m m0 mf R T t
a
pre-exponential factor (s1) contribution factor reaction rate constant (s1) sample mass at any time (g) initial sample mass (g, at 105 °C) final residual mass (g, at 700 °C) universal gas constant (=8.314 Jmol1K1) absolute temperature (K) conversion time (s) conversion degree
Subscript i ith component
Table 1 Summary of recent pyrolysis kinetic studies on various microalgae.
a
Feedstock
Methoda
Main results
Refs.
Dunaliella tertiolecta
FWO, KAS
Mean activation energy: 145.7 kJ/mol in KAS method and 146.4 kJ/mol in FWO method
[15]
Chlorella spp.
FreemanCarroll
Activation energy: 71.3– 79.2 kJ/mol Pre-exponential factor: 1.47–1.62 106 min1
[16]
Chlorella vulgaris
FWO, KAS, regression
Mean activation energy: 66.7 kJ/mol in KAS method and 61.7 kJ/mol in FWO method
[17]
Chlorella pyrenoidosa (CP) and Spirulina platensis (SP)
Vyazovkin
Activation energy: 8.85– 114.5 kJ/mol for CP and 74.35–140.1 kJ/mol for SP
[18]
Scenedesmus almeriensis (SC), Nannochloropsis Gaditana (NG) and Chlorella vulgaris (CV)
Órfão
Activation energy: 63.5 kJ/mol for CV, 128.1 kJ/mol for NG, 79.6 kJ/mol for SC
[19]
Nannochloropsis oculata (NO) and Tetraselmis sp. (TS)
Simplified DAEM
Highest activation energy: 152 kJ/mol for NO and 334 kJ/mol for TS
[20]
Chlorella pyrenoidosa
SSGM, simplified DAEM
Mean activation energy: 143.7 kJ/mol in SSGM and 100.6 kJ/mol in simplified DAEM
[21]
Abbreviations are explained in Nomenclature.
have been successfully applied to analyze the pyrolysis behaviors of several lignocellulosic biomass materials [23–26]. However, applicability of these models for microalgae pyrolysis is still unclarified because the main components of microalgae (i.e., carbohydrates, proteins and lipids) differ from those of lignocellulosic biomass (i.e., hemicellulose, cellulose and lignin). In order to address this issue, a variety of microalga pyrolysis models are tested in this study, from which more comprehensive information and better understanding of microalga pyrolysis mechanism can be obtained. In this study, various kinetic models from a single to multiple parallel reactions are tested in order to figure out the optimal pyrolysis model which is applicable for microalgal biomass pyrolysis. For this purpose, microalga Chlorella vulgaris ESP-31 was pyrolyzed in a nitrogen atmosphere along with a non-isothermal
111
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116
mode by means of a thermogravimetric analyzer. Different kinetic models, presented in next section, are applied to fit to the experimental data. Extracted kinetic data including activation energy and pre-exponential factor are presented, and the fit quality of each model is evaluated. 2. Kinetic modeling
[9]. The LTSC consists of carbohydrate and protein due to their low decomposition temperatures, while the HTSC belongs to lipid as a consequence of its higher decomposition temperature [22]. Accordingly, their thermal decompositions are expressed as:
8 k1 < LTSC ! Char 1 þ Volatile 1 Microalga k2 : HTSC ! Char 2 þ Volatile 2
ð4Þ
This part presents the kinetic models employed for microalga pyrolysis. In fact, global kinetic models, assuming single and multiple parallel reactions, have been successful applied for the pyrolysis of lignocellulosic biomass [23–26]. However, some modifications are needed when they are employed for microalgal biomass due to the inherent difference in the compositions of the two materials. Reaction mechanisms as well as explanations and assumptions of different models are provided below, while a summary of components in the models is presented in Table 2.
Considering the thermal stability of microalgal components [9], it is assumed that the first group consists of carbohydrate and protein, while lipid belong to the second one. Each reaction also obeys the Arrhenius equation, namely, Eq. (2). In addition, in this proposed model or subsequent multiple parallel reaction models, it is assumed that individual reactions are independent of each other, implying that there is no interaction or synergistic effect between reactive components [28]. The total conversion of the microalga is thus the sum of two component conversions multiplying with corresponding contribution factors (c).
2.1. Single reaction model
da da da þ c2 ¼ c1 dt dt LTSC dt HTSC
This is one of the simplest kinetic models for biomass pyrolysis [27], which can be applicable for microalgae. It assumes that all components in a microalga have the same reactivity, and the thermal degradation of the microalga produces char and volatile: k
Microalga ! Char þ Volatile
ð1Þ
The reaction rate constant (k) is described by the Arrhenius expression:
da Ea Ea f ðaÞ ¼ A exp ð1 aÞ ¼ kðTÞ f ðaÞ ¼ A exp dt RT RT
ð2Þ
where a is the conversion degree, t is the conversion time, A is the pre-exponential factor, Ea is the activation energy of the reaction, R is the universal gas constant, and T is the absolute temperature. The conversion degree (a) is defined as the mass fraction of decomposed solid:
a¼
m0 m m0 mf
ð3Þ
where m0 and mf are the initial and final masses of solid, and m is the mass of solid at any time. 2.2. Multiple parallel reaction models 2.2.1. Two reactions In this two-reaction model, microalgal components are classified into two groups based on their reactivity: low thermal stable components (LTSC) and high thermal stable components (HTSC)
ð5Þ
As the name implies, c is a factor accounting for the contribution of a specific component to the total mass loss during pyrolysis. 2.2.2. Three or more reactions
8 k1 > Comp 1 ! Char 1 þ Volatile 1 > > > > > k2 > > Comp 2 ! Char 2 þ Volatile 2 > > < Microalga : > > >: > > > > : > > > : kN Comp N ! Char N þ Volatile N
ð6Þ
This is upgraded versions of the foregoing model. In order to have better fit quality between the modeled and experimental data, the number of components is increased to three or more in the corresponding models, in which each reaction represents a component. This arises from the fact that a microalga consists of various components with different reactivity. Generally, a microalga comprises three main components (i.e., carbohydrate, protein, and lipid) and other minor components. Consequently, a threereaction model is employed to elucidate the thermal degradation of the three main components only. Furthermore, the minor components (known as ‘‘others”) are included in a four-reaction model. In a multiple reaction model, up to seven reactions are employed, considering that the main components may consist of several
Table 2 Summary of models selection. Comp. 1 Single reaction
d
Two reactions Carbohydrate & Protein Lipid
d
Three reactions Carbohydrate Protein Lipid Four reactions Carbohydrate Protein Lipid Others
Comp. 2
Comp. 3
Comp. 4
d d d d d d d d
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116
constituents with different reactivity. The total conversion of the microalga in these models is the sum of all partial reactions, and written as: M da X dai ci ¼ dt dt i¼1
ð7Þ
where ci denotes the contribution factor of each component and M is the number of the components. Each component has three independent parameters (i.e., Ea, A and c), and thus there are 9, 12 and 21 parameters used in the model fitting for three, four, and seven reactions.
3.3. Numerical method Collected experimental data from the thermogravimetric analysis (TGA) were transformed to differential thermogravimetric (DTG) data, which presented the conversion rate ddta versus temperature T. Drying period was excluded from the modeling because it is not relevant to the decomposition of microalgal components. Different models were then applied to fit the experimental DTG data via a curve fitting process, which is based on non-linear least squares method. Curve fitting procedure and kinetic parameters extraction can be found in other studies [26,31] where Excel was employed as the main software. The fit quality was determined as below:
0
3. Experimental 3.1. Material
1
qffiffiffiffi
B Fit ð%Þ ¼ @1 h da
OF N
C A 100%
i
ð8Þ
dt exp max
The feedstock in this study is microalga Chlorella vulgaris ESP31, which was collected from a fish pond in southern Taiwan and cultivated at the National Cheng Kung University. Cultivation, collection and storage procedure of the microalga can be found elsewhere [29,30]. The chemical composition and important fuel characteristics, including proximate and ultimate analyses as well as higher heating value, of the microalga are listed in Table 3. 3.2. Thermogravimetric experiment
OF ¼ min
" N X daj j¼1
where
daj dt
exp
#2 daj dt exp dt cal
and
daj dt
cal
ð9Þ
represent the experimental and modeled
conversion rates, respectively, and N is the number of experimental points. 4. Results and discussion
An SDT Q600 thermogravimetric analyzer (TA Instruments) was employed for the microalga pyrolysis in this study. In each experiment, about 5 mg of the microalga was loaded into a 150 ll alumina crucible, which was transferred to the chamber of the analyzer thereafter. The tested sample was first heated from room temperature to 105 °C at a heating rate of 20 °C/min, and then held at 105 °C for 30 min for drying. Afterwards, a constant heating rate of 10 °C/min was applied to heat the sample to 700 °C. This slow heating rate has been chosen to ensure that the experiments were in kinetically controlled regime. In addition, a number of past studies [26,27,31–33] have shown that this heating profile was good for kinetic study. Besides, the size of the microalga is in the order of 10 lm so that diffusion effects in the course of thermal degradation are minimized. Nitrogen at a flow rate of 100 ml/min was supplied to ensure an inert atmosphere for the experiment. Three repetitions were run to ascertain the experimental quality, and good data reproducibility was achieved.
4.1. Microalga pyrolysis behavior Fig. 1 presents the pyrolytic TGA and DTG curves of the microalga. On account of excluding the drying period from the data set, the figure shows two reactive stages for the microalga instead of three stages as normally reported in the literature. Almost no mass loss is exhibited at temperatures less than 150 °C. Thereafter, a significant mass loss (up to 70 wt%) is observed in the first stage at temperatures within 150–450 °C, mainly stemming from the thermal degradation of carbohydrate, protein, and lipid in the microalga [9]. When the temperature is higher than 450 °C in the second stage, the mass of the microalga decays slightly, indicating the decomposition of carbonaceous matters in the solid residues. The DTG curve shows a sharp devolatilization peak (with a height of 2.10 103 s1) at about 290 °C, accompanied by a small peak
100
-1
Conversion rate (dα/dt x 10 , s )
2
80
3
Table 3 Characterization of microalga Chlorella vulgaris ESP-31. Characteristics
Value
Proximate analysis (wt%, dry-basis) Fixed carbon Volatile matter Ash
16.39 74.59 9.02
Ultimate analysis (wt%, dry-ash-free) C H N O
53.01 8.67 3.26 35.05
Higher heating value (MJ/kg, dry basis)
22.02
Chemical composition (wt%, dry basis) Carbohydrate Lipid Protein Others
56.92 14.83 22.50 5.75
1.5 60 1 40 0.5
20
0 100
200
300
400
500
600
0 700
o
Temperature ( C) Fig. 1. Thermal degradation of microalga during pyrolysis.
Mass fraction (%)
112
113
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116
This section presents kinetic modeling results for different models described in Section 2. Curve fitting results are presented in Figs. 2–6, while extracted kinetic data and fit quality values are tabulated in Tables 4 and 5. Pros and cons of each model are discussed in detail. 4.2.1. Single reaction model Curve fitting for the single reaction model is presented in Fig. 2, which shows a poor fit between the experimental and modeled data. It can be observed that the peak of the modeled curve is far from that of the experimental curve, and no shoulder occurs on the modeled curve. Consequently, the model offers very low fit quality (=87.51%). However, the activation energy (=38.84 kJ/mol) and pre-exponential factor (=6.94 100 s1) are within the ranges reported in the literature (Table 1). Overall, it can be seen that the single reaction model is not suitable for modeling the microalga pyrolysis kinetics. 4.2.2. Two-reaction model Fig. 3 demonstrates the curve fitting of the two-reaction model. Compared with the single reaction one, this model yields a much better fit quality (=96.46%). Overall, the modeled curve is composed of the reactions of the LTSC at temperature range of 160– 325 °C and those of the HTSC at 160–510 °C. The estimated values of the activation energy of the LTSC and HTSC are respectively 135.83 and 53.39 kJ/mol, while their pre-exponential factors are 3.97 1010 and 5.09 101 s1, respectively. These kinetic parameters lie in the reported ranges. However, some drawbacks of the model can be pointed out from the figure. The two peaks (i.e., the devolatilization peak and the accompanied small peak) are not well modeled because of the underestimations. In particular,
Conversion rate (dα/dt x 103, s-1)
4.2. Kinetic modeling
2
Exp. Cal. LTSC HTSC
1.5
1
0.5
0 100
200
300
400
500
600
Temperature (oC) Fig. 3. Curve fitting for two-reaction model.
2
Conversion rate (dα/dt x 103, s-1)
(at around 385 °C) on the right of the main peak and a shoulder (at around 220 °C) on the left. Compared with lignocellulosic biomass materials, the pyrolysis of the microalga is somewhat different: a clear shoulder appears on the left of the peak for hardwoods, but almost no shoulder can be observed for softwoods [32,33]. For the microalga, a small peak locates on the right of the devolatilization peak, as a consequence of the difference in the compositions of the two types of biomass.
Exp. Cal. Carbohydrate Protein Lipid
1.5
1
0.5
0 100
200
300
400
500
600
Temperature (oC) Fig. 4. Curve fitting for three-reaction model.
Exp. Cal.
no curve covers the shoulder. The results indicate that the tworeaction model is still not good enough for the microalga pyrolysis.
3
-1
Conversion rate (dα/dt x 10 , s )
2
1.5
1
0.5
0 100
200
300
400
500
Temperature (oC) Fig. 2. Curve fitting for single reaction step model.
600
4.2.3. Three-reaction model The assumptions in the previous models are not suitable for a complex material like microalga; therefore, the three-reaction model separates the decomposition of three main microalgal components (i.e. carbohydrate, protein and lipid) into three parallel reactions. Results from this model are presented in Fig. 4, showing a reasonable fit quality (=98.44%). Currently, no study includes the thermal decomposition of all main microalga components (e.g. carbohydrate, protein and lipid); however, other works on similar compounds from different materials can be beneficial to help understanding the thermal reactivity of these components separately. Pavlath et al. [34] studied the pyrolysis of carbohydrate and showed that they started degrading from about 200 °C and lasted up to 350 °C. The thermal degradation suggested two reaction peaks, consisting of a minor one at 241–261 °C (caused by the degradation of glucose, maltose and cellobiose) and a major
114
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116
Conversion rate (dα/dt x 103, s-1)
2
one at 316–353 °C (caused by the degradation of previous components and amylose, cellulose). In other studies, Samouillan et al. [35] and Dandurand et al. [36] reported that the decomposition of some proteins spread at temperatures between 220 and 380 °C, and their degradation peaks were in the range of 285– 301 °C, which was virtually lower than that of carbohydrate. Moreover, Pena Muniz et al. [37] pointed out that lipids mostly degraded in the temperature range of 270–580 °C with a peak at around 400 °C. Applying the survey above for Fig. 4, it is easy to assign the modeled components to the actual microalgal components. That is, the three peaks in the temperature ranges of 110– 420 °C, 210–310 °C and 150–515 °C can be allocated to the thermal degradation of carbohydrate, protein and lipid in the microalga, respectively. Accordingly, the kinetic parameters of these components are listed in Table 4, showing that the activation energy of protein is the highest (=208.80 kJ/mol) and that of carbohydrate is the lowest (=40.36 kJ/mol), while the value of lipid is 48.46 kJ/mol. In comparison with the single and two-reaction models, the three-reaction model is able to accurately predict the devolatilization (highest) peak, thereby significantly improving the model fit quality. However, the predictions of the small peak and the shoulder from this model are still unsatisfactory, resulting from the appearances of small gaps between the modeled and experimental curves.
Exp. Cal. Carbohydrate Protein Lipid Others
1.5
1
0.5
0 100
200
300
400
500
600
o
Temperature ( C) Fig. 5. Curve fitting for four-reaction model.
Conversion rate (dα/dt x 103, s-1)
2
Exp. Cal Carbohydrate I Carbohydrate II Protein I Protein II Lipid Others Intermediates
1.5
1
4.2.4. Four-reaction model In order to further improve the fit quality of the three-reaction model, a fourth component is introduced in the four-reaction model. The curve fitting result of this model is plotted in Fig. 5. In examining the position of the extra component, it can be attributed to the so-called ‘‘others” in the microalga. However, the reactivity of this minor component is unclear and not welldocumented in the literature yet. The extracted kinetic parameters of carbohydrate, protein and lipid for this model in Table 4 are close to the values in the three-reaction model. In addition, the activation energy and pre-exponential factor of the ‘‘others” component are respectively 197.73 kJ/mol and 4.00 1013 s1. However, the modeled curve in Fig. 5 still shows some variations at the shoulder and the small peak, compared with the experimental curve. This indicates that the thermal degradation of the microalga in these areas may include the decomposition of many components. It should be mentioned that, even though the model offers a good fit (=98.44%), the contribution factors of protein (=0.34) and lipid (=0.38) appear to be overestimated because their percentages in the microalga are respectively 22.5 and 14.8 wt%
0.5
0 100
200
300
400
500
600
Temperature (oC) Fig. 6. Curve fitting for seven-reaction model.
Table 4 Estimated kinetic data from different models (numbers in brackets show standard deviations). Kinetic parameters
Fit (%) 1
A (kJ/mol)
E (s
38.84 (0.18)
6.94 100 (3.90 101)
–
Two reactions Carbohydrate & protein Lipid
135.83 (0.85) 53.39 (0.20)
3.97 1010 (3.91 108) 5.09 101 (2.17 102)
0.47 (0.01) 0.49 (0.01)
Three reactions Carbohydrate Protein Lipid
40.36 (0.63) 208.80 (1.17) 48.46 (0.20)
1.06 101 (3.05 102) 3.55 1017 (1.05 1015) 1.73 101 (2.25 102)
0.21 (0.01) 0.34 (0.01) 0.42 (0.02)
Four reactions Carbohydrate Protein Lipid Others
39.82 (0.38) 208.86 (1.10) 48.61 (0.20) 197.73 (1.30)
1.06 101 (4.52 103) 3.54 1017 (9.32 1014) 1.72 101 (6.75 102) 4.00 1013 (1.97 109)
0.21 0.34 0.38 0.03
)
c
Single reaction
87.51 (0.10) 96.46 (0.42)
98.44 (0.09)
98.68 (0.13) (0.01) (0.01) (0.02) (0.01)
115
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116 Table 5 Estimated kinetic data from seven-reaction model (numbers in brackets show standard deviations). Kinetic parameters
Seven reactions 1. Low thermal resistant carbohydrate (Carbohydrate I) 2. High thermal resistant carbohydrate (Carbohydrate II) 3. Low thermal resistant protein (Protein I) 4. High thermal resistant protein (Protein II) 5. Lipid 6. Others 7. Intermediate products (Intermediates)
Fit (%) 1
A (kJ/mol)
E (s
74.00 (0.34) 221.73 (1.79) 207.04 (1.11) 220.94 (1.49) 112.91 (1.10) 150.56 (1.44) 47.40 (0.23)
4.00 105 (1.58 103) 3.68 1018 (1.56 1015) 4.00 1017 (8.68 1014) 4.50 1017 (3.07 1015) 5.95 106 (2.53 104) 5.75 1010 (1.92 108) 1.04 101 (7.06 101)
)
c 99.37 (0.33)
(Table 3). In order to solve these issues, more complex model is conducted in the next section.
4.2.5. Seven-reaction model Considering that the degradation peaks of carbohydrate and protein are in a wide range of temperature [34–36], they may contain several constituents with different reactivity. To address this, each component is further partitioned into two substances: low and high thermal resistant substances (group I and II, respectively). Thus, the first four components are introduced and listed in Table 5 (i.e., Carbohydrate I, Carbohydrate II, Protein I and Protein II). Lipid has long aliphatic chains so that its structure and reactivity are much simpler than those of carbohydrate and protein [37]. Therefore, only one reaction is needed to describe the thermal degradation of lipid in the microalga. Another two components in this model are the so-called ‘‘others” and intermediate products. The latter is formed during the degradation of the microalga at low temperatures. The curve fitting result of the seven-reaction model is presented in Fig. 6. The model can precisely predict the shoulder and the small peak, showing a very good fit quality (=99.37%). Moreover, the contribution factors of the modeled components in Table 5 are reasonable, compared with the percentages of the components in Table 3. The total contributions of proteins and lipid are respectively 0.25 and 0.15, which are very close to their percentages in the microalga (=22.5 and 14.8 wt%, respectively). In addition, the contributions of carbohydrate and intermediate products are 0.26 and 0.28, respectively. These values might indicate that the intermediate products are mainly formed by the decomposition of carbohydrate, considering that the percentage of carbohydrate in the microalga is 56.9 wt%. More importantly, the thermal degradation mechanism of the microalga can be recognized from the distributions of the component curves shown in Fig. 6. The figure depicts that the shoulder exhibited is mainly due to the decomposition of Carbohydrate I, while the small peak is mainly attributed to the decomposition of intermediates and others. The devolatilization peak is resulted from the combined degradation of Protein I and Carbohydrate II, not by a single one as assumed by previous models. For microalga Chlorella vulgaris in this study, these two peaks are overlapped so that only one peak can be observed during the pyrolysis. This finding can help explaining the pyrolysis of other microalgal species such as Chlorella pyrenoidosa, showing double peaks with almost the same intensity and very close distance [18,21]. In the case of microalga Chlorella pyrenoidosa, Protein I and Carbohydrate II may be separated and create the double peaks. In view of the high fit quality and clear physical meaning of the involved components, this seven-reaction model is recommended for modeling microalga pyrolysis. However, a disadvantage of this model compared with the others is its complexity, which normally requires higher computational resource. It is known that the main components in microalgal biomass include carbohydrates, proteins, and lipids. Though only one
0.06 0.20 0.21 0.04 0.15 0.03 0.28
(0.01) (0.02) (0.02) (0.01) (0.01) (0.01) (0.02)
microalga species (Chlorella vulgaris ESP-31) was tested in this study, the models have been developed based on the three basic components in microalgae. Therefore, the models are also applicable to other microalgae. On account of differences in the compositions of different microalgae, the obtained kinetics data, including the activation energy and pre-exponential factor, may be different. But the feasibility of the developed kinetic models remains the same.
5. Conclusion The pyrolysis characteristics of microalga Chlorella vulgaris ESP31 has been studied and modeled from thermogravimetric analysis, where two degradation stages (excluding drying) has been observed. The DTG curve suggests that a devolatilization peak accompanied by a shoulder and a small peak is exhibited. Several kinetic models, from a single reaction to seven parallel reactions, have been conducted to figure out the pyrolysis mechanism. Overall, the extracted kinetic parameters are in good agreement with available data in the literature. The results show that the single, two-, and three-reaction models cannot fit the peaks and shoulder well so that they are not suitable for microalga pyrolysis. Increasing the number of reactions improves the fitness between the experimental data and the model. It was found that the fourreaction models can be satisfactory; however, they cannot precisely predict the thermal degradation around the shoulder and the small peak. To further improve the fit quality of the curve fitting, a hypothesis of individually dividing the thermal degradation of carbohydrate and protein into two stages was proposed. As a result, the seven-reaction model is recommended due to its highly precise curve fitting, i.e. DTG curve is consistent with the model and reactions. The contribution factors of the modeled components are reasonable when compared with the percentages of the components in the microalga. More importantly, the thermal degradation mechanism of the microalga can be recognized from the distributions of the component curves. The curve fitting reveals that the shoulder is due to the decomposition of Carbohydrate I, while the small peak is mainly contributed by the simultaneous decomposition of intermediates and others. In addition, the devolatilization peak is attributed to the combined degradation of Protein I and Carbohydrate II. This developed seven-reaction model can be employed for microalga pyrolysis prediction and reactor design.
Acknowledgments The authors also acknowledge the financial support from the Ministry of Science and Technology, Taiwan, R.O.C., under the contracts MOST 102-2221-E-006-288-MY3 and MOST 105-2811-E006-003 for this research.
116
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116
References [1] Saidur R, Abdelaziz EA, Demirbas A, Hossain MS, Mekhilef S. A review on biomass as a fuel for boilers. Renew Sustain Energy Rev 2011;15:2262–89. [2] Mohr A, Raman S. Lessons from first generation biofuels and implications for the sustainability appraisal of second generation biofuels. Energy Policy 2013;63:114–22. [3] Antizar-Ladislao B, Turrion-Gomez JL. Second-generation biofuels and local bioenergy systems. Biofuels Bioprod Biorefin 2008;2:455–69. [4] Demirbas MF. Biofuels from algae for sustainable development. Appl Energy 2011;88:3473–80. [5] Aresta M, Dibenedetto A, Barberio G. Utilization of macro-algae for enhanced CO2 fixation and biofuels production: development of a computing software for an LCA study. Fuel Process Technol 2005;86:1679–93. [6] Bach Q-V, Sillero MV, Tran K-Q, Skjermo J. Fast hydrothermal liquefaction of a Norwegian macro-alga: screening tests. Algal Res. 2014;6(Part B):271–6. [7] Gutiérrez-Arriaga CG, Serna-González M, Ponce-Ortega JM, El-Halwagi MM. Sustainable integration of algal biodiesel production with steam electric power plants for greenhouse gas mitigation. ACS Sustainable Chem. Eng. 2014;2:1388–403. [8] Cuellar-Bermudez SP, Garcia-Perez JS, Rittmann BE, Parra-Saldivar R. Photosynthetic bioenergy utilizing CO2: an approach on flue gases utilization for third generation biofuels. J. Clean. Prod. 2015;98:53–65. [9] Chen W-H, Lin B-J, Huang M-Y, Chang J-S. Thermochemical conversion of microalgal biomass into biofuels: a review. Bioresour Technol 2015;184:314–27. [10] Suali E, Sarbatly R. Conversion of microalgae to biofuel. Renew Sustain Energy Rev 2012;16:4316–42. [11] Chen W-H, Peng J, Bi XT. A state-of-the-art review of biomass torrefaction, densification and applications. Renew Sustain Energy Rev 2015;44: 847–66. [12] Chen W-H, Lin B-J. Characteristics of products from the pyrolysis of oil palm fiber and its pellets in nitrogen and carbon dioxide atmospheres. Energy 2016;94:569–78. [13] Bridgwater AV. Review of fast pyrolysis of biomass and product upgrading. Biomass Bioenergy 2012;38:68–94. [14] Kim SW, Koo BS, Lee DH. A comparative study of bio-oils from pyrolysis of microalgae and oil seed waste in a fluidized bed. Bioresour Technol 2014;162:96–102. [15] Shuping Z, Yulong W, Mingde Y, Chun L, Junmao T. Pyrolysis characteristics and kinetics of the marine microalgae Dunaliella tertiolecta using thermogravimetric analyzer. Bioresour Technol 2010;101:359–65. [16] Rizzo AM, Prussi M, Bettucci L, Libelli IM, Chiaramonti D. Characterization of microalga Chlorella as a fuel and its thermogravimetric behavior. Appl Energy 2013;102:24–31. [17] Agrawal A, Chakraborty S. A kinetic study of pyrolysis and combustion of microalgae Chlorella vulgaris using thermo-gravimetric analysis. Bioresour Technol 2013;128:72–80. [18] Gai C, Zhang Y, Chen W-T, Zhang P, Dong Y. Thermogravimetric and kinetic analysis of thermal decomposition characteristics of low-lipid microalgae. Bioresour Technol 2013;150:139–48.
[19] López-González D, Fernandez-Lopez M, Valverde JL, Sanchez-Silva L. Pyrolysis of three different types of microalgae: kinetic and evolved gas analysis. Energy 2014;73:33–43. [20] Ceylan S, Kazan D. Pyrolysis kinetics and thermal characteristics of microalgae Nannochloropsis oculata and Tetraselmis sp.. Bioresour Technol 2015;187:1–5. [21] Hu M, Chen Z, Guo D, Liu C, Xiao B, Hu Z, et al. Thermogravimetric study on pyrolysis kinetics of Chlorella pyrenoidosa and bloom-forming cyanobacteria. Bioresour Technol 2015;177:41–50. [22] Chen W-H, Wu Z-Y, Chang J-S. Isothermal and non-isothermal torrefaction characteristics and kinetics of microalga Scenedesmus obliquus CNW-N. Bioresour Technol 2014;155:245–51. [23] Lanzetta M, Di Blasi C. Pyrolysis kinetics of wheat and corn straw. J Anal Appl Pyrol 1998;44:181–92. [24] Branca C, Di Blasi C. Global kinetics of wood char devolatilization and combustion. Energy Fuels 2003;17:1609–15. [25] Branca C, Albano A, Di Blasi C. Critical evaluation of global mechanisms of wood devolatilization. Thermochim Acta 2005;429:133–41. [26] Bach Q-V, Tran K-Q, Skreiberg Ø, Trinh TT. Effects of wet torrefaction on pyrolysis of woody biomass fuels. Energy 2015;88:443–56. [27] Tran K-Q, Bach Q-V, Trinh TT, Seisenbaeva G. Non-isothermal pyrolysis of torrefied stump – a comparative kinetic evaluation. Appl Energy 2014;136:759–66. [28] Lu K-M, Lee W-J, Chen W-H, Lin T-C. Thermogravimetric analysis and kinetics of co-pyrolysis of raw/torrefied wood and coal blends. Appl Energy 2013;105:57–65. [29] Chen W-H, Huang M-Y, Chang J-S, Chen C-Y. Torrefaction operation and optimization of microalga residue for energy densification and utilization. Appl Energy 2015;154:622–30. [30] Tran D-T, Yeh K-L, Chen C-L, Chang J-S. Enzymatic transesterification of microalgal oil from Chlorella vulgaris ESP-31 for biodiesel synthesis using immobilized Burkholderia lipase. Bioresour Technol 2012;108:119–27. [31] Bach Q-V, Tran K-Q, Skreiberg Ø, Khalil RA, Phan AN. Effects of wet torrefaction on reactivity and kinetics of wood under air combustion conditions. Fuel 2014;137:375–83. [32] Bach Q-V, Tran K-Q, Skreiberg Ø. Combustion kinetics of wet-torrefied forest residues using the distributed activation energy model (DAEM). Appl Energy. [33] Bach Q-V, Tran K-Q, Skreiberg Ø. Comparative study on the thermal degradation of dry- and wet-torrefied woods. Appl Energy. [34] Pavlath AE, Gregorski KS. Atmospheric pyrolysis of carbohydrates with thermogravimetric and mass spectrometric analyses. J Anal Appl Pyrol 1985;8:41–8. [35] Samouillan V, Dandurand-Lods J, Lamure A, Maurel E, Lacabanne C, Gerosa G, et al. Thermal analysis characterization of aortic tissues for cardiac valve bioprostheses. J Biomed Mater Res 1999;46:531–8. [36] Dandurand J, Samouillan V, Lacoste-Ferre MH, Lacabanne C, Bochicchio B, Pepe A. Conformational and thermal characterization of a synthetic peptidic fragment inspired from human tropoelastin: signature of the amyloid fibers. Pathol Biol (Paris) 2014;62:100–7. [37] Pena Muniz MA, Ferreira dos Santos MN, da Costa CEF, Morais L, Lamarão MLN, Ribeiro-Costa RM, et al. Physicochemical characterization, fatty acid composition, and thermal analysis of Bertholletia excelsa HBK oil. Pharmacognosy Mag. 2015;11:147–51.
Contents lists available at ScienceDirect
Energy Conversion and Management journal homepage: www.elsevier.com/locate/enconman
A comprehensive study on pyrolysis kinetics of microalgal biomass Quang-Vu Bach, Wei-Hsin Chen ⇑ Department of Aeronautics and Astronautics, National Cheng Kung University, Tainan 701, Taiwan
a r t i c l e
i n f o
Article history: Received 3 August 2016 Received in revised form 14 October 2016 Accepted 31 October 2016 Available online 10 November 2016 Keywords: Microalgal biomass Pyrolysis Kinetic modeling Activation energy Thermogravimetric analysis
a b s t r a c t Pyrolysis of microalgal biomass for biofuels production has attracted much attention. However, detailed degradation mechanism and kinetics of the process have not been fully explored yet. In this study, a nonisothermal pyrolysis of microalga Chlorella vulgaris ESP-31 is thermogravimetrically investigated. Several kinetic models, from a single reaction to seven parallel reactions, are tested to fit the experimental pyrolysis data for finding out the optimal pyrolysis model. The results show that the pyrolysis behavior of the microalga is somewhat different from that of lignocellulosic biomass, stemming from the inherent difference in their compositions. Overall, the kinetic modeling processes show that increasing the number of reactions improves the model fit quality. Curve fitting results indicate that the models consisting of three and less than three reactions are not suitable for microalga pyrolysis. The four-reaction model, via considering the pyrolysis of carbohydrate, protein, lipid and others, can be employed for modeling the thermal degradation; however, it cannot precisely predict the thermal degradation of the shoulder and the small peak. The conducted seven-reaction model further partitions the decomposition processes of carbohydrate and protein into two stages, and explains the thermal degradation well. The model indicates that the devolatilization peak is attributed to the combined degradation of Protein I and Carbohydrate II. The seven-reaction model offers the highest fit quality and is thus recommended for predicting the microalga pyrolysis processes. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Biomass is a potential source of renewable energy, which is receiving a great deal of attention due to its advantages over other alternative energy sources [1]. Biomass derived fuels can be categorized into first, second, and third generation biofuels. First generation biofuels are produced from food crops such as sugarcane, corn, potato, wheat, and sugar beet. This directly affects human food supplies and biodiversity [2]. Second generation biofuels are produced from non-food or lignocellulosic biomass and biomass residues such as switchgrass, grass, jatropha, miscanthus, husk, wood chips, leaves and stump. Non-food biomass does not threaten food supplies; however, its planation will compete with arable lands for food crops and influence nutrient cycles and soil conservation [3]. The disadvantages of first and second generation biofuels can be overcome by exploiting algal biomass, which is considered as the potential feedstock for third generation biofuels. Unlike land-based lignocellulosic biomass, algal biomass can be
⇑ Corresponding author. E-mail addresses: (W.-H. Chen).
[email protected],
http://dx.doi.org/10.1016/j.enconman.2016.10.077 0196-8904/Ó 2016 Elsevier Ltd. All rights reserved.
[email protected]
cultivated in fresh, saline, or waste water [4], and thus requires no land allocation. Moreover, compared to terrestrial biomass, algal biomass, including microalgae and macroalgae, has higher growth rate and photosynthetic efficiency, and hence can absorb more CO2 during its growth [5,6]. This enables more efficient reduction in the greenhouse gas emissions [7], and even achieves CO2 utilization [8]. Many thermochemical conversion processes can be employed to convert microalgal biomass into energy-dense biofuels such as bio-oils and biochars [9–11]. Pyrolysis is a thermal conversion route in which microalga is heated at elevated temperatures (400–800 °C) and in an oxygen-free atmosphere [12]. The pyrolysis products include bio-oils and biochars as well as a small fraction of permanent gases, and their distribution usually depends on the operating parameters such as temperature (or heating rate) and residence time [13]. Pyrolysis at high temperatures with short residence times yields more bio-oil. On the other hand, biochar production is favored for pyrolysis at low temperatures with long residence times. Both bio-oils and biochars can be used directly for combustion to produce heat and power. Bio-oils can be further upgraded to produce liquid transport fuels and bio-chemicals, while biochars can be used as activated carbon, soil enhancer, fertilizer, etc.
110
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116
Nomenclature Abbreviations daf dry and ash free basis db dry basis DTG differential thermogravimetric TG thermogravimetric TGA thermogravimetric analysis wt weight LTSC/HTSC low/high thermal stable components Comp component FWO Flynn-Wall-Ozawa model KAS Kissinger-Akahira-Sunose model DAEM distributed activation energy model SSGM single-step global model Cal. calculated data Exp. experimental data
As an important conversion of microalgal biomass into biofuels, pyrolysis has received great attention recently. The number of works on microalgae pyrolysis has been increasing in order to assess the chemical, physical and fuel properties of microalgae bio-oils and to compare with lignocellulosic biomass bio-oils. It has been reported that bio-oils produced from microalgal biomass are more stable than those from lignocellulosic biomass [10]. Moreover, microalgae bio-oils contain less oxygen and have greater heating value than lignocellulosic biomass bio-oils [14]. In addition to researches on the properties of microalgae bio-oils, kinetic studies have been also conducted to investigate the pyrolysis behavior of microalgae. In practice, pyrolysis kinetics is highly related to chemical reaction control and reactor design. For this purpose, thermogravimetric technique has become a proven method and been widely used by many researchers [15–21]. It has been pointed out that microalgae pyrolysis undergoes three stages. In the first stage, water is removed from microalgae. The mass loss in this stage is associated with the moisture content of the feedstock. Main microalgal components, including carbohydrate, protein and lipid, are thermally degraded in the second stage, which accounts for most of the mass loss during pyrolysis. In the last stage, only slight mass loss is observed which is attributed to the decomposition of carbonaceous matters in the solid residues. In some studies [22], the thermal decomposition of lipid is partitioned from that of carbohydrate and protein, whereby four-stage thermal degradation of microalgae is defined. Although pyrolysis kinetic researches of microalgal biomass are active, most of the available studies employed simplified kinetics models. In other words, deep knowledge about the reaction mechanism from the obtained kinetic data is insufficient. A number of recent studies regarding pyrolysis kinetic of various microalgae are summarized in Table 1 in which a variety of methods were conducted. As can be seen from the table, the available studies employed several approximation and transformation of the Arrhenius expression to estimate the values of activation energy and pre-exponential factor at different conversion rates. As a result, the values of the two kinetic parameters are normally in a range, from which mean values can be calculated and they represent the whole microalgal biomass, even though the microalgae contain several components whose reactivity and kinetic parameters are different. In addition, these models cannot reproduce the predicted thermogravimetric curves or provide any information about the fit quality between the modeled and experimental data. Consequently, the evaluation of these models is difficult. More complex kinetic models, e.g., multiple reaction models, are available and
Symbols A c kðTÞ or k m m0 mf R T t
a
pre-exponential factor (s1) contribution factor reaction rate constant (s1) sample mass at any time (g) initial sample mass (g, at 105 °C) final residual mass (g, at 700 °C) universal gas constant (=8.314 Jmol1K1) absolute temperature (K) conversion time (s) conversion degree
Subscript i ith component
Table 1 Summary of recent pyrolysis kinetic studies on various microalgae.
a
Feedstock
Methoda
Main results
Refs.
Dunaliella tertiolecta
FWO, KAS
Mean activation energy: 145.7 kJ/mol in KAS method and 146.4 kJ/mol in FWO method
[15]
Chlorella spp.
FreemanCarroll
Activation energy: 71.3– 79.2 kJ/mol Pre-exponential factor: 1.47–1.62 106 min1
[16]
Chlorella vulgaris
FWO, KAS, regression
Mean activation energy: 66.7 kJ/mol in KAS method and 61.7 kJ/mol in FWO method
[17]
Chlorella pyrenoidosa (CP) and Spirulina platensis (SP)
Vyazovkin
Activation energy: 8.85– 114.5 kJ/mol for CP and 74.35–140.1 kJ/mol for SP
[18]
Scenedesmus almeriensis (SC), Nannochloropsis Gaditana (NG) and Chlorella vulgaris (CV)
Órfão
Activation energy: 63.5 kJ/mol for CV, 128.1 kJ/mol for NG, 79.6 kJ/mol for SC
[19]
Nannochloropsis oculata (NO) and Tetraselmis sp. (TS)
Simplified DAEM
Highest activation energy: 152 kJ/mol for NO and 334 kJ/mol for TS
[20]
Chlorella pyrenoidosa
SSGM, simplified DAEM
Mean activation energy: 143.7 kJ/mol in SSGM and 100.6 kJ/mol in simplified DAEM
[21]
Abbreviations are explained in Nomenclature.
have been successfully applied to analyze the pyrolysis behaviors of several lignocellulosic biomass materials [23–26]. However, applicability of these models for microalgae pyrolysis is still unclarified because the main components of microalgae (i.e., carbohydrates, proteins and lipids) differ from those of lignocellulosic biomass (i.e., hemicellulose, cellulose and lignin). In order to address this issue, a variety of microalga pyrolysis models are tested in this study, from which more comprehensive information and better understanding of microalga pyrolysis mechanism can be obtained. In this study, various kinetic models from a single to multiple parallel reactions are tested in order to figure out the optimal pyrolysis model which is applicable for microalgal biomass pyrolysis. For this purpose, microalga Chlorella vulgaris ESP-31 was pyrolyzed in a nitrogen atmosphere along with a non-isothermal
111
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116
mode by means of a thermogravimetric analyzer. Different kinetic models, presented in next section, are applied to fit to the experimental data. Extracted kinetic data including activation energy and pre-exponential factor are presented, and the fit quality of each model is evaluated. 2. Kinetic modeling
[9]. The LTSC consists of carbohydrate and protein due to their low decomposition temperatures, while the HTSC belongs to lipid as a consequence of its higher decomposition temperature [22]. Accordingly, their thermal decompositions are expressed as:
8 k1 < LTSC ! Char 1 þ Volatile 1 Microalga k2 : HTSC ! Char 2 þ Volatile 2
ð4Þ
This part presents the kinetic models employed for microalga pyrolysis. In fact, global kinetic models, assuming single and multiple parallel reactions, have been successful applied for the pyrolysis of lignocellulosic biomass [23–26]. However, some modifications are needed when they are employed for microalgal biomass due to the inherent difference in the compositions of the two materials. Reaction mechanisms as well as explanations and assumptions of different models are provided below, while a summary of components in the models is presented in Table 2.
Considering the thermal stability of microalgal components [9], it is assumed that the first group consists of carbohydrate and protein, while lipid belong to the second one. Each reaction also obeys the Arrhenius equation, namely, Eq. (2). In addition, in this proposed model or subsequent multiple parallel reaction models, it is assumed that individual reactions are independent of each other, implying that there is no interaction or synergistic effect between reactive components [28]. The total conversion of the microalga is thus the sum of two component conversions multiplying with corresponding contribution factors (c).
2.1. Single reaction model
da da da þ c2 ¼ c1 dt dt LTSC dt HTSC
This is one of the simplest kinetic models for biomass pyrolysis [27], which can be applicable for microalgae. It assumes that all components in a microalga have the same reactivity, and the thermal degradation of the microalga produces char and volatile: k
Microalga ! Char þ Volatile
ð1Þ
The reaction rate constant (k) is described by the Arrhenius expression:
da Ea Ea f ðaÞ ¼ A exp ð1 aÞ ¼ kðTÞ f ðaÞ ¼ A exp dt RT RT
ð2Þ
where a is the conversion degree, t is the conversion time, A is the pre-exponential factor, Ea is the activation energy of the reaction, R is the universal gas constant, and T is the absolute temperature. The conversion degree (a) is defined as the mass fraction of decomposed solid:
a¼
m0 m m0 mf
ð3Þ
where m0 and mf are the initial and final masses of solid, and m is the mass of solid at any time. 2.2. Multiple parallel reaction models 2.2.1. Two reactions In this two-reaction model, microalgal components are classified into two groups based on their reactivity: low thermal stable components (LTSC) and high thermal stable components (HTSC)
ð5Þ
As the name implies, c is a factor accounting for the contribution of a specific component to the total mass loss during pyrolysis. 2.2.2. Three or more reactions
8 k1 > Comp 1 ! Char 1 þ Volatile 1 > > > > > k2 > > Comp 2 ! Char 2 þ Volatile 2 > > < Microalga : > > >: > > > > : > > > : kN Comp N ! Char N þ Volatile N
ð6Þ
This is upgraded versions of the foregoing model. In order to have better fit quality between the modeled and experimental data, the number of components is increased to three or more in the corresponding models, in which each reaction represents a component. This arises from the fact that a microalga consists of various components with different reactivity. Generally, a microalga comprises three main components (i.e., carbohydrate, protein, and lipid) and other minor components. Consequently, a threereaction model is employed to elucidate the thermal degradation of the three main components only. Furthermore, the minor components (known as ‘‘others”) are included in a four-reaction model. In a multiple reaction model, up to seven reactions are employed, considering that the main components may consist of several
Table 2 Summary of models selection. Comp. 1 Single reaction
d
Two reactions Carbohydrate & Protein Lipid
d
Three reactions Carbohydrate Protein Lipid Four reactions Carbohydrate Protein Lipid Others
Comp. 2
Comp. 3
Comp. 4
d d d d d d d d
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116
constituents with different reactivity. The total conversion of the microalga in these models is the sum of all partial reactions, and written as: M da X dai ci ¼ dt dt i¼1
ð7Þ
where ci denotes the contribution factor of each component and M is the number of the components. Each component has three independent parameters (i.e., Ea, A and c), and thus there are 9, 12 and 21 parameters used in the model fitting for three, four, and seven reactions.
3.3. Numerical method Collected experimental data from the thermogravimetric analysis (TGA) were transformed to differential thermogravimetric (DTG) data, which presented the conversion rate ddta versus temperature T. Drying period was excluded from the modeling because it is not relevant to the decomposition of microalgal components. Different models were then applied to fit the experimental DTG data via a curve fitting process, which is based on non-linear least squares method. Curve fitting procedure and kinetic parameters extraction can be found in other studies [26,31] where Excel was employed as the main software. The fit quality was determined as below:
0
3. Experimental 3.1. Material
1
qffiffiffiffi
B Fit ð%Þ ¼ @1 h da
OF N
C A 100%
i
ð8Þ
dt exp max
The feedstock in this study is microalga Chlorella vulgaris ESP31, which was collected from a fish pond in southern Taiwan and cultivated at the National Cheng Kung University. Cultivation, collection and storage procedure of the microalga can be found elsewhere [29,30]. The chemical composition and important fuel characteristics, including proximate and ultimate analyses as well as higher heating value, of the microalga are listed in Table 3. 3.2. Thermogravimetric experiment
OF ¼ min
" N X daj j¼1
where
daj dt
exp
#2 daj dt exp dt cal
and
daj dt
cal
ð9Þ
represent the experimental and modeled
conversion rates, respectively, and N is the number of experimental points. 4. Results and discussion
An SDT Q600 thermogravimetric analyzer (TA Instruments) was employed for the microalga pyrolysis in this study. In each experiment, about 5 mg of the microalga was loaded into a 150 ll alumina crucible, which was transferred to the chamber of the analyzer thereafter. The tested sample was first heated from room temperature to 105 °C at a heating rate of 20 °C/min, and then held at 105 °C for 30 min for drying. Afterwards, a constant heating rate of 10 °C/min was applied to heat the sample to 700 °C. This slow heating rate has been chosen to ensure that the experiments were in kinetically controlled regime. In addition, a number of past studies [26,27,31–33] have shown that this heating profile was good for kinetic study. Besides, the size of the microalga is in the order of 10 lm so that diffusion effects in the course of thermal degradation are minimized. Nitrogen at a flow rate of 100 ml/min was supplied to ensure an inert atmosphere for the experiment. Three repetitions were run to ascertain the experimental quality, and good data reproducibility was achieved.
4.1. Microalga pyrolysis behavior Fig. 1 presents the pyrolytic TGA and DTG curves of the microalga. On account of excluding the drying period from the data set, the figure shows two reactive stages for the microalga instead of three stages as normally reported in the literature. Almost no mass loss is exhibited at temperatures less than 150 °C. Thereafter, a significant mass loss (up to 70 wt%) is observed in the first stage at temperatures within 150–450 °C, mainly stemming from the thermal degradation of carbohydrate, protein, and lipid in the microalga [9]. When the temperature is higher than 450 °C in the second stage, the mass of the microalga decays slightly, indicating the decomposition of carbonaceous matters in the solid residues. The DTG curve shows a sharp devolatilization peak (with a height of 2.10 103 s1) at about 290 °C, accompanied by a small peak
100
-1
Conversion rate (dα/dt x 10 , s )
2
80
3
Table 3 Characterization of microalga Chlorella vulgaris ESP-31. Characteristics
Value
Proximate analysis (wt%, dry-basis) Fixed carbon Volatile matter Ash
16.39 74.59 9.02
Ultimate analysis (wt%, dry-ash-free) C H N O
53.01 8.67 3.26 35.05
Higher heating value (MJ/kg, dry basis)
22.02
Chemical composition (wt%, dry basis) Carbohydrate Lipid Protein Others
56.92 14.83 22.50 5.75
1.5 60 1 40 0.5
20
0 100
200
300
400
500
600
0 700
o
Temperature ( C) Fig. 1. Thermal degradation of microalga during pyrolysis.
Mass fraction (%)
112
113
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116
This section presents kinetic modeling results for different models described in Section 2. Curve fitting results are presented in Figs. 2–6, while extracted kinetic data and fit quality values are tabulated in Tables 4 and 5. Pros and cons of each model are discussed in detail. 4.2.1. Single reaction model Curve fitting for the single reaction model is presented in Fig. 2, which shows a poor fit between the experimental and modeled data. It can be observed that the peak of the modeled curve is far from that of the experimental curve, and no shoulder occurs on the modeled curve. Consequently, the model offers very low fit quality (=87.51%). However, the activation energy (=38.84 kJ/mol) and pre-exponential factor (=6.94 100 s1) are within the ranges reported in the literature (Table 1). Overall, it can be seen that the single reaction model is not suitable for modeling the microalga pyrolysis kinetics. 4.2.2. Two-reaction model Fig. 3 demonstrates the curve fitting of the two-reaction model. Compared with the single reaction one, this model yields a much better fit quality (=96.46%). Overall, the modeled curve is composed of the reactions of the LTSC at temperature range of 160– 325 °C and those of the HTSC at 160–510 °C. The estimated values of the activation energy of the LTSC and HTSC are respectively 135.83 and 53.39 kJ/mol, while their pre-exponential factors are 3.97 1010 and 5.09 101 s1, respectively. These kinetic parameters lie in the reported ranges. However, some drawbacks of the model can be pointed out from the figure. The two peaks (i.e., the devolatilization peak and the accompanied small peak) are not well modeled because of the underestimations. In particular,
Conversion rate (dα/dt x 103, s-1)
4.2. Kinetic modeling
2
Exp. Cal. LTSC HTSC
1.5
1
0.5
0 100
200
300
400
500
600
Temperature (oC) Fig. 3. Curve fitting for two-reaction model.
2
Conversion rate (dα/dt x 103, s-1)
(at around 385 °C) on the right of the main peak and a shoulder (at around 220 °C) on the left. Compared with lignocellulosic biomass materials, the pyrolysis of the microalga is somewhat different: a clear shoulder appears on the left of the peak for hardwoods, but almost no shoulder can be observed for softwoods [32,33]. For the microalga, a small peak locates on the right of the devolatilization peak, as a consequence of the difference in the compositions of the two types of biomass.
Exp. Cal. Carbohydrate Protein Lipid
1.5
1
0.5
0 100
200
300
400
500
600
Temperature (oC) Fig. 4. Curve fitting for three-reaction model.
Exp. Cal.
no curve covers the shoulder. The results indicate that the tworeaction model is still not good enough for the microalga pyrolysis.
3
-1
Conversion rate (dα/dt x 10 , s )
2
1.5
1
0.5
0 100
200
300
400
500
Temperature (oC) Fig. 2. Curve fitting for single reaction step model.
600
4.2.3. Three-reaction model The assumptions in the previous models are not suitable for a complex material like microalga; therefore, the three-reaction model separates the decomposition of three main microalgal components (i.e. carbohydrate, protein and lipid) into three parallel reactions. Results from this model are presented in Fig. 4, showing a reasonable fit quality (=98.44%). Currently, no study includes the thermal decomposition of all main microalga components (e.g. carbohydrate, protein and lipid); however, other works on similar compounds from different materials can be beneficial to help understanding the thermal reactivity of these components separately. Pavlath et al. [34] studied the pyrolysis of carbohydrate and showed that they started degrading from about 200 °C and lasted up to 350 °C. The thermal degradation suggested two reaction peaks, consisting of a minor one at 241–261 °C (caused by the degradation of glucose, maltose and cellobiose) and a major
114
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116
Conversion rate (dα/dt x 103, s-1)
2
one at 316–353 °C (caused by the degradation of previous components and amylose, cellulose). In other studies, Samouillan et al. [35] and Dandurand et al. [36] reported that the decomposition of some proteins spread at temperatures between 220 and 380 °C, and their degradation peaks were in the range of 285– 301 °C, which was virtually lower than that of carbohydrate. Moreover, Pena Muniz et al. [37] pointed out that lipids mostly degraded in the temperature range of 270–580 °C with a peak at around 400 °C. Applying the survey above for Fig. 4, it is easy to assign the modeled components to the actual microalgal components. That is, the three peaks in the temperature ranges of 110– 420 °C, 210–310 °C and 150–515 °C can be allocated to the thermal degradation of carbohydrate, protein and lipid in the microalga, respectively. Accordingly, the kinetic parameters of these components are listed in Table 4, showing that the activation energy of protein is the highest (=208.80 kJ/mol) and that of carbohydrate is the lowest (=40.36 kJ/mol), while the value of lipid is 48.46 kJ/mol. In comparison with the single and two-reaction models, the three-reaction model is able to accurately predict the devolatilization (highest) peak, thereby significantly improving the model fit quality. However, the predictions of the small peak and the shoulder from this model are still unsatisfactory, resulting from the appearances of small gaps between the modeled and experimental curves.
Exp. Cal. Carbohydrate Protein Lipid Others
1.5
1
0.5
0 100
200
300
400
500
600
o
Temperature ( C) Fig. 5. Curve fitting for four-reaction model.
Conversion rate (dα/dt x 103, s-1)
2
Exp. Cal Carbohydrate I Carbohydrate II Protein I Protein II Lipid Others Intermediates
1.5
1
4.2.4. Four-reaction model In order to further improve the fit quality of the three-reaction model, a fourth component is introduced in the four-reaction model. The curve fitting result of this model is plotted in Fig. 5. In examining the position of the extra component, it can be attributed to the so-called ‘‘others” in the microalga. However, the reactivity of this minor component is unclear and not welldocumented in the literature yet. The extracted kinetic parameters of carbohydrate, protein and lipid for this model in Table 4 are close to the values in the three-reaction model. In addition, the activation energy and pre-exponential factor of the ‘‘others” component are respectively 197.73 kJ/mol and 4.00 1013 s1. However, the modeled curve in Fig. 5 still shows some variations at the shoulder and the small peak, compared with the experimental curve. This indicates that the thermal degradation of the microalga in these areas may include the decomposition of many components. It should be mentioned that, even though the model offers a good fit (=98.44%), the contribution factors of protein (=0.34) and lipid (=0.38) appear to be overestimated because their percentages in the microalga are respectively 22.5 and 14.8 wt%
0.5
0 100
200
300
400
500
600
Temperature (oC) Fig. 6. Curve fitting for seven-reaction model.
Table 4 Estimated kinetic data from different models (numbers in brackets show standard deviations). Kinetic parameters
Fit (%) 1
A (kJ/mol)
E (s
38.84 (0.18)
6.94 100 (3.90 101)
–
Two reactions Carbohydrate & protein Lipid
135.83 (0.85) 53.39 (0.20)
3.97 1010 (3.91 108) 5.09 101 (2.17 102)
0.47 (0.01) 0.49 (0.01)
Three reactions Carbohydrate Protein Lipid
40.36 (0.63) 208.80 (1.17) 48.46 (0.20)
1.06 101 (3.05 102) 3.55 1017 (1.05 1015) 1.73 101 (2.25 102)
0.21 (0.01) 0.34 (0.01) 0.42 (0.02)
Four reactions Carbohydrate Protein Lipid Others
39.82 (0.38) 208.86 (1.10) 48.61 (0.20) 197.73 (1.30)
1.06 101 (4.52 103) 3.54 1017 (9.32 1014) 1.72 101 (6.75 102) 4.00 1013 (1.97 109)
0.21 0.34 0.38 0.03
)
c
Single reaction
87.51 (0.10) 96.46 (0.42)
98.44 (0.09)
98.68 (0.13) (0.01) (0.01) (0.02) (0.01)
115
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116 Table 5 Estimated kinetic data from seven-reaction model (numbers in brackets show standard deviations). Kinetic parameters
Seven reactions 1. Low thermal resistant carbohydrate (Carbohydrate I) 2. High thermal resistant carbohydrate (Carbohydrate II) 3. Low thermal resistant protein (Protein I) 4. High thermal resistant protein (Protein II) 5. Lipid 6. Others 7. Intermediate products (Intermediates)
Fit (%) 1
A (kJ/mol)
E (s
74.00 (0.34) 221.73 (1.79) 207.04 (1.11) 220.94 (1.49) 112.91 (1.10) 150.56 (1.44) 47.40 (0.23)
4.00 105 (1.58 103) 3.68 1018 (1.56 1015) 4.00 1017 (8.68 1014) 4.50 1017 (3.07 1015) 5.95 106 (2.53 104) 5.75 1010 (1.92 108) 1.04 101 (7.06 101)
)
c 99.37 (0.33)
(Table 3). In order to solve these issues, more complex model is conducted in the next section.
4.2.5. Seven-reaction model Considering that the degradation peaks of carbohydrate and protein are in a wide range of temperature [34–36], they may contain several constituents with different reactivity. To address this, each component is further partitioned into two substances: low and high thermal resistant substances (group I and II, respectively). Thus, the first four components are introduced and listed in Table 5 (i.e., Carbohydrate I, Carbohydrate II, Protein I and Protein II). Lipid has long aliphatic chains so that its structure and reactivity are much simpler than those of carbohydrate and protein [37]. Therefore, only one reaction is needed to describe the thermal degradation of lipid in the microalga. Another two components in this model are the so-called ‘‘others” and intermediate products. The latter is formed during the degradation of the microalga at low temperatures. The curve fitting result of the seven-reaction model is presented in Fig. 6. The model can precisely predict the shoulder and the small peak, showing a very good fit quality (=99.37%). Moreover, the contribution factors of the modeled components in Table 5 are reasonable, compared with the percentages of the components in Table 3. The total contributions of proteins and lipid are respectively 0.25 and 0.15, which are very close to their percentages in the microalga (=22.5 and 14.8 wt%, respectively). In addition, the contributions of carbohydrate and intermediate products are 0.26 and 0.28, respectively. These values might indicate that the intermediate products are mainly formed by the decomposition of carbohydrate, considering that the percentage of carbohydrate in the microalga is 56.9 wt%. More importantly, the thermal degradation mechanism of the microalga can be recognized from the distributions of the component curves shown in Fig. 6. The figure depicts that the shoulder exhibited is mainly due to the decomposition of Carbohydrate I, while the small peak is mainly attributed to the decomposition of intermediates and others. The devolatilization peak is resulted from the combined degradation of Protein I and Carbohydrate II, not by a single one as assumed by previous models. For microalga Chlorella vulgaris in this study, these two peaks are overlapped so that only one peak can be observed during the pyrolysis. This finding can help explaining the pyrolysis of other microalgal species such as Chlorella pyrenoidosa, showing double peaks with almost the same intensity and very close distance [18,21]. In the case of microalga Chlorella pyrenoidosa, Protein I and Carbohydrate II may be separated and create the double peaks. In view of the high fit quality and clear physical meaning of the involved components, this seven-reaction model is recommended for modeling microalga pyrolysis. However, a disadvantage of this model compared with the others is its complexity, which normally requires higher computational resource. It is known that the main components in microalgal biomass include carbohydrates, proteins, and lipids. Though only one
0.06 0.20 0.21 0.04 0.15 0.03 0.28
(0.01) (0.02) (0.02) (0.01) (0.01) (0.01) (0.02)
microalga species (Chlorella vulgaris ESP-31) was tested in this study, the models have been developed based on the three basic components in microalgae. Therefore, the models are also applicable to other microalgae. On account of differences in the compositions of different microalgae, the obtained kinetics data, including the activation energy and pre-exponential factor, may be different. But the feasibility of the developed kinetic models remains the same.
5. Conclusion The pyrolysis characteristics of microalga Chlorella vulgaris ESP31 has been studied and modeled from thermogravimetric analysis, where two degradation stages (excluding drying) has been observed. The DTG curve suggests that a devolatilization peak accompanied by a shoulder and a small peak is exhibited. Several kinetic models, from a single reaction to seven parallel reactions, have been conducted to figure out the pyrolysis mechanism. Overall, the extracted kinetic parameters are in good agreement with available data in the literature. The results show that the single, two-, and three-reaction models cannot fit the peaks and shoulder well so that they are not suitable for microalga pyrolysis. Increasing the number of reactions improves the fitness between the experimental data and the model. It was found that the fourreaction models can be satisfactory; however, they cannot precisely predict the thermal degradation around the shoulder and the small peak. To further improve the fit quality of the curve fitting, a hypothesis of individually dividing the thermal degradation of carbohydrate and protein into two stages was proposed. As a result, the seven-reaction model is recommended due to its highly precise curve fitting, i.e. DTG curve is consistent with the model and reactions. The contribution factors of the modeled components are reasonable when compared with the percentages of the components in the microalga. More importantly, the thermal degradation mechanism of the microalga can be recognized from the distributions of the component curves. The curve fitting reveals that the shoulder is due to the decomposition of Carbohydrate I, while the small peak is mainly contributed by the simultaneous decomposition of intermediates and others. In addition, the devolatilization peak is attributed to the combined degradation of Protein I and Carbohydrate II. This developed seven-reaction model can be employed for microalga pyrolysis prediction and reactor design.
Acknowledgments The authors also acknowledge the financial support from the Ministry of Science and Technology, Taiwan, R.O.C., under the contracts MOST 102-2221-E-006-288-MY3 and MOST 105-2811-E006-003 for this research.
116
Q.-V. Bach, W.-H. Chen / Energy Conversion and Management 131 (2017) 109–116
References [1] Saidur R, Abdelaziz EA, Demirbas A, Hossain MS, Mekhilef S. A review on biomass as a fuel for boilers. Renew Sustain Energy Rev 2011;15:2262–89. [2] Mohr A, Raman S. Lessons from first generation biofuels and implications for the sustainability appraisal of second generation biofuels. Energy Policy 2013;63:114–22. [3] Antizar-Ladislao B, Turrion-Gomez JL. Second-generation biofuels and local bioenergy systems. Biofuels Bioprod Biorefin 2008;2:455–69. [4] Demirbas MF. Biofuels from algae for sustainable development. Appl Energy 2011;88:3473–80. [5] Aresta M, Dibenedetto A, Barberio G. Utilization of macro-algae for enhanced CO2 fixation and biofuels production: development of a computing software for an LCA study. Fuel Process Technol 2005;86:1679–93. [6] Bach Q-V, Sillero MV, Tran K-Q, Skjermo J. Fast hydrothermal liquefaction of a Norwegian macro-alga: screening tests. Algal Res. 2014;6(Part B):271–6. [7] Gutiérrez-Arriaga CG, Serna-González M, Ponce-Ortega JM, El-Halwagi MM. Sustainable integration of algal biodiesel production with steam electric power plants for greenhouse gas mitigation. ACS Sustainable Chem. Eng. 2014;2:1388–403. [8] Cuellar-Bermudez SP, Garcia-Perez JS, Rittmann BE, Parra-Saldivar R. Photosynthetic bioenergy utilizing CO2: an approach on flue gases utilization for third generation biofuels. J. Clean. Prod. 2015;98:53–65. [9] Chen W-H, Lin B-J, Huang M-Y, Chang J-S. Thermochemical conversion of microalgal biomass into biofuels: a review. Bioresour Technol 2015;184:314–27. [10] Suali E, Sarbatly R. Conversion of microalgae to biofuel. Renew Sustain Energy Rev 2012;16:4316–42. [11] Chen W-H, Peng J, Bi XT. A state-of-the-art review of biomass torrefaction, densification and applications. Renew Sustain Energy Rev 2015;44: 847–66. [12] Chen W-H, Lin B-J. Characteristics of products from the pyrolysis of oil palm fiber and its pellets in nitrogen and carbon dioxide atmospheres. Energy 2016;94:569–78. [13] Bridgwater AV. Review of fast pyrolysis of biomass and product upgrading. Biomass Bioenergy 2012;38:68–94. [14] Kim SW, Koo BS, Lee DH. A comparative study of bio-oils from pyrolysis of microalgae and oil seed waste in a fluidized bed. Bioresour Technol 2014;162:96–102. [15] Shuping Z, Yulong W, Mingde Y, Chun L, Junmao T. Pyrolysis characteristics and kinetics of the marine microalgae Dunaliella tertiolecta using thermogravimetric analyzer. Bioresour Technol 2010;101:359–65. [16] Rizzo AM, Prussi M, Bettucci L, Libelli IM, Chiaramonti D. Characterization of microalga Chlorella as a fuel and its thermogravimetric behavior. Appl Energy 2013;102:24–31. [17] Agrawal A, Chakraborty S. A kinetic study of pyrolysis and combustion of microalgae Chlorella vulgaris using thermo-gravimetric analysis. Bioresour Technol 2013;128:72–80. [18] Gai C, Zhang Y, Chen W-T, Zhang P, Dong Y. Thermogravimetric and kinetic analysis of thermal decomposition characteristics of low-lipid microalgae. Bioresour Technol 2013;150:139–48.
[19] López-González D, Fernandez-Lopez M, Valverde JL, Sanchez-Silva L. Pyrolysis of three different types of microalgae: kinetic and evolved gas analysis. Energy 2014;73:33–43. [20] Ceylan S, Kazan D. Pyrolysis kinetics and thermal characteristics of microalgae Nannochloropsis oculata and Tetraselmis sp.. Bioresour Technol 2015;187:1–5. [21] Hu M, Chen Z, Guo D, Liu C, Xiao B, Hu Z, et al. Thermogravimetric study on pyrolysis kinetics of Chlorella pyrenoidosa and bloom-forming cyanobacteria. Bioresour Technol 2015;177:41–50. [22] Chen W-H, Wu Z-Y, Chang J-S. Isothermal and non-isothermal torrefaction characteristics and kinetics of microalga Scenedesmus obliquus CNW-N. Bioresour Technol 2014;155:245–51. [23] Lanzetta M, Di Blasi C. Pyrolysis kinetics of wheat and corn straw. J Anal Appl Pyrol 1998;44:181–92. [24] Branca C, Di Blasi C. Global kinetics of wood char devolatilization and combustion. Energy Fuels 2003;17:1609–15. [25] Branca C, Albano A, Di Blasi C. Critical evaluation of global mechanisms of wood devolatilization. Thermochim Acta 2005;429:133–41. [26] Bach Q-V, Tran K-Q, Skreiberg Ø, Trinh TT. Effects of wet torrefaction on pyrolysis of woody biomass fuels. Energy 2015;88:443–56. [27] Tran K-Q, Bach Q-V, Trinh TT, Seisenbaeva G. Non-isothermal pyrolysis of torrefied stump – a comparative kinetic evaluation. Appl Energy 2014;136:759–66. [28] Lu K-M, Lee W-J, Chen W-H, Lin T-C. Thermogravimetric analysis and kinetics of co-pyrolysis of raw/torrefied wood and coal blends. Appl Energy 2013;105:57–65. [29] Chen W-H, Huang M-Y, Chang J-S, Chen C-Y. Torrefaction operation and optimization of microalga residue for energy densification and utilization. Appl Energy 2015;154:622–30. [30] Tran D-T, Yeh K-L, Chen C-L, Chang J-S. Enzymatic transesterification of microalgal oil from Chlorella vulgaris ESP-31 for biodiesel synthesis using immobilized Burkholderia lipase. Bioresour Technol 2012;108:119–27. [31] Bach Q-V, Tran K-Q, Skreiberg Ø, Khalil RA, Phan AN. Effects of wet torrefaction on reactivity and kinetics of wood under air combustion conditions. Fuel 2014;137:375–83. [32] Bach Q-V, Tran K-Q, Skreiberg Ø. Combustion kinetics of wet-torrefied forest residues using the distributed activation energy model (DAEM). Appl Energy. [33] Bach Q-V, Tran K-Q, Skreiberg Ø. Comparative study on the thermal degradation of dry- and wet-torrefied woods. Appl Energy. [34] Pavlath AE, Gregorski KS. Atmospheric pyrolysis of carbohydrates with thermogravimetric and mass spectrometric analyses. J Anal Appl Pyrol 1985;8:41–8. [35] Samouillan V, Dandurand-Lods J, Lamure A, Maurel E, Lacabanne C, Gerosa G, et al. Thermal analysis characterization of aortic tissues for cardiac valve bioprostheses. J Biomed Mater Res 1999;46:531–8. [36] Dandurand J, Samouillan V, Lacoste-Ferre MH, Lacabanne C, Bochicchio B, Pepe A. Conformational and thermal characterization of a synthetic peptidic fragment inspired from human tropoelastin: signature of the amyloid fibers. Pathol Biol (Paris) 2014;62:100–7. [37] Pena Muniz MA, Ferreira dos Santos MN, da Costa CEF, Morais L, Lamarão MLN, Ribeiro-Costa RM, et al. Physicochemical characterization, fatty acid composition, and thermal analysis of Bertholletia excelsa HBK oil. Pharmacognosy Mag. 2015;11:147–51.