Contents lists available at ScienceDirect

Bioresource Technology

journal homepage: www.elsevier.com/locate/biortech

Thermal oxidative degradation kinetics of agricultural residues usingdistributed activation energy model and global kinetic model

Xiu'e Rena,b, Jianbiao Chena,⁎, Gang Lia, Yanhong Wanga, Xuemei Langa, Shuanshi Fana

a Key Laboratory of Enhanced Heat Transfer and Energy Conservation, Ministry Education, School of Chemistry and Chemical Engineering, South China University ofTechnology, Guangzhou 510640, Guangdong, ChinabDepartment of Chemistry and Pharmacy, Zhuhai College of Jilin University, Zhuhai 519041, Guangdong, China

A R T I C L E I N F O

Keywords:Agricultural residuesThermal oxidative degradationKinetic analysisDistributed activation energy model (DAEM)Global kinetic model (GKM)

A B S T R A C T

The study concerned the thermal oxidative degradation kinetics of agricultural residues, peanut shell (PS) andsunflower shell (SS). The thermal behaviors were evaluated via thermogravimetric analysis and the kineticparameters were determined by using distributed activation energy model (DAEM) and global kinetic model(GKM). Results showed that thermal oxidative decomposition of two samples processed in three zones; theignition, burnout, and comprehensive combustibility between two agricultural residues were of great difference;and the combustion performance could be improved by boosting heating rate. The activation energy rangescalculated by the DAEM for the thermal oxidative degradation of PS and SS were 88.94–145.30 kJmol−1 and94.86–169.18 kJmol−1, respectively. The activation energy obtained by the GKM for the oxidative decom-position of hemicellulose and cellulose was obviously lower than that for the lignin oxidation at identical heatingrate. To some degree, the determined kinetic parameters could acceptably simulate experimental data.

1. Introduction

Biomass is an organic compound generated by photosynthesis,which is the world’s fourth largest primary energy source behind coal,petroleum, and natural gas, accounting for 10–15% of the total energyconsumption (Saidur et al., 2011; Bach and Chen, 2017). Unlike otherrenewable sources such as solar, marine, wind, and geothermal en-ergies, biomass is independent of climate and locality, and can be ob-tained easily due to its wide distribution (Chen et al., 2012b; Ellabbanet al., 2014). Furthermore, biomass is the only carbon–based alternativeenergy, which can not only be burned to generate heat and produceelectricity, but can also be converted into some high-value chemicalsfor industrial production (Chen et al., 2012a; Yahya et al., 2015).However, not all biomass resources are applicable to the energy utili-zation. Along with the problem of population increasingly growing,governments around the world have established strict policies thatbiomass energy should be harvested from non–edible plants and pro-duced on non–arable lands (Gai et al., 2015; Bach and Chen, 2017).Alternatively, agricultural residues are abundantly available, which canbe used as the feedstock for the next generation of biofuels and che-micals (White et al., 2011). Actually, a majority of agricultural residueswere not properly used, if the situation is improved, it will bring a veryimportant society and economy benefit.

Recently, with the growth of population and the improvement ofliving standards, there is a rapid increase in the demand for ediblevegetable oils. In China, peanut and sunflower are the two most popularoil crops. The dominant peanut producing regions include three pro-vinces, namely Shandong, Henan, and Hebei. The sunflower are pop-ular in the regions of Inner Mongolia, Jilin, Xinjiang, Gansu, and Hebei.According to the China Statistical Year book reported by the NationalBureau of Statistics of China (NBSC), the planting area and productionof peanut in 2016 were 4727 million mu and 1729 million tons, re-spectively (National Bureau of Statics of China, 2017). With regard tosunflower, the planting area and production in 2016 reached 1490million mu and 251 million tons, respectively. Consequently, as theprocessing residues of the two crops, peanut shell (PS) and sunflowershell (SS) are abundantly available every year (Chen et al., 2009; Wanget al., 2012). In the past, a considerable part of two residues were ca-sually abandoned or burned, which seriously damaged the ecologicalenvironment and wasted the biomass resources. In order to address theproblems brought by the unreasonable disposal of agricultural residues,researchers have focused on transforming them into heat, solid, liquidor gas fuels, and electricity via various thermochemical conversion(TCC) methods (Ma et al., 2012; Papari and Hawboldt, 2015; Wanget al., 2017). As a simple, efficient, and cheap TCC method, combustionis responsible for the vast majority of energy production in the world,

https://doi.org/10.1016/j.biortech.2018.04.047Received 28 March 2018; Received in revised form 9 April 2018; Accepted 11 April 2018

⁎ Corresponding author.E-mail address: jianbiaochen@scut.edu.cn (J. Chen).

Bioresource Technology 261 (2018) 403–411

Available online 14 April 20180960-8524/ © 2018 Elsevier Ltd. All rights reserved.

T

which can realize agricultural residues resourceful utilization, improvethe eco-environment, and enhance economy benefit (Jiang et al., 2017).

To date, the combustion (or named as thermal oxidative degrada-tion) behaviors and kinetics with regard to various agricultural residueshave been widely analyzed and reported, which could provide certaintechnical guidance for the design, operation, and scaling of the biomasscombustor. Shen et al. (2011) have studied the thermal degradationmechanisms of agricultural residues from two wood species under inertand oxidative conditions. Compared to that under inert condition, theweight loss of two residues under oxidative condition had a significantincrease, indicating that the thermal reactivity of two residues wasgreatly enhanced due to the occurrence of oxidation reactions. As re-ported by López et al. (2014), the combustion kinetics of corn, sun-flower, rape, microalgae and their blends were compared. The resultsshowed that the activation energy from corn combustion was lowerthan that from the blends (corn-sunflower, and corn-microalgae),whereas higher than that from the corn-rape blend. With respect to theagricultural residues from olive trees, the activation energies calculatedby the model-free methods were confirmed by the model-fitting methodwhen a first order reaction was considered (Garcia-Maraver et al.,2015). According to Álvarez et al. (2016), the database of performanceparameters and kinetic data for the combustion of most common agri-cultural residues were provided. The results indicated that the com-bustion process of agricultural residues could be divided into two zones:thermal oxidative degradation of cellulose and hemicellulose, and thecombustion of lignin fraction. In the previous work, the thermal oxi-dative degradation behaviors and kinetics of four typical oil–plant re-sidues have also been reported. It was found that the combustioncharacteristic parameters were quite different from each other, and thekinetic data were also in a great difference (Chen et al., 2017b). Theresults of above literature have fully suggested that agricultural re-sidues as an interesting feedstock have a high resource potential.However, a systematic study on the thermal oxidative degradationbehaviors and kinetics of PS and SS has not yet been thoroughly per-formed.

In this paper, the thermal oxidative degradation behaviors withrespect to the characteristic parameters of PS and SS were evaluatedusing thermogravimetric analysis in air atmosphere at heating rates of5, 10, 20, and 40 Kmin−1. Additionally, the kinetic parameters for the

thermal oxidative degradation process of PS and SS were calculated bythe distributed activation energy model (DAEM) and global kineticmodel (GKM).

2. Materials and methods

2.1. Raw materials

As two common agricultural residues, peanut shell (PS) and sun-flower shell (SS) were picked as the feedstock to perform thermal oxi-dative degradation experiments. On basis of the Laboratory AnalyticalProcedures, the basic properties with regard to proximate analysis andultimate analysis of the two samples were characterized. As shown inTable 1, PS and SS both have a high content of volatile matter and acomparatively low contents of fixed carbon and ash. As seen from ul-timate analysis, the carbon and hydrogen contents in SS were higherthan those in PS, but the contents of oxygen, nitrogen, and sulfur in PSwere relatively higher.

2.2. Thermal oxidative degradation experiments

Thermal oxidative degradation tests of PS and SS were carried out ina NETZSCH STA 449C simultaneous thermogravimetric analyzer (TGA).Prior to the experiments, the PS and SS samples were ground andsieved. To remove the heat and mass transfer limitations, the PS and SS

Nomenclature

Abbreviations

PS Peanut shellSS Sunflower shellTGA Thermogravimetric analysis/analyzerTCC Thermochemical conversionDTG Differential thermogravimetryDTA Differential thermal analysisGKM Global kinetic modelDAEM Distributed activation energy model

Symbols

−Rv Average weight loss rate, % min−1

−Rp Maximum weight loss rate, % min−1

Ci Ignition index, % min−3

Cb Burnout index, % min−4

h(O2) Oxygen partial pressure functionΔt1/2 Time interval at half value of −Rp, sV Weight loss percentage at time t, %V∞ Total weight loss percentage, %ΔT1/2 Temperature interval at half value of −Rp, K

CCI Comprehensive combustibility index, %2 min−2 K−3

R Universal gas constant, J mol−1 K−1

ti Ignition time, stp Peak time, stb Burnout time, sw0 Initial weight, mgw Weight at time t, mgw∞ Final weight, mgTi Ignition temperature, KTp Peak temperature, KTb Burnout temperature, Kdx/dt Conversion rate, % min−1

x Conversion degree, %f(x) Differential kinetic modelg(x) Integral kinetic modelT Absolute temperature, Kk(T) Reaction rate constantE Activation energy, kJ mol−1

A Pre-exponential factor, s−1

β Heating rate, Kmin−1

R2 Correlation coefficientWL Weight loss percentage, %

Table 1Proximate analysis and ultimate analysis of PS and SS.

Proximate analysis (wt%, ara) Ultimate analysis (wt%, dafb)

PS SS PS SS

Fixed carbon 18.92 20.17 Carbon (C) 44.42 49.46Ash 2.87 2.23 Hydrogen (H) 4.87 6.18Volatile matter 70.79 69.75 Oxygen (O)c 48.92 42.87Moisture 7.42 7.85 Nitrogen (N) 1.45 1.22

Sulfur (S) 0.34 0.27

a Dry basis.b Dry ash free basis.c By difference.

X. Ren et al. Bioresource Technology 261 (2018) 403–411

404

samples with the particle size smaller than 0.30mm and the initialweight less than 5.0 mg were heated up from 300.0 to 1000.0 K at theheating rates of 5, 10, 20, and 40 Kmin−1. During the experiments, thefurnace of TGA is flushed with 50.0 mLmin−1 dry air to keep an oxi-dative atmosphere. For eliminating the effects of air buoyancy, the TGbaselines were obtained under identical conditions without PS or SS inthe crucible of TGA. Furthermore, all the thermal oxidative degradationtests were repeated three times to guarantee the reproducibility of theexperimental data.

2.3. Thermal oxidative degradation characteristic parameters

To make the study meaningful and applicable in practice that em-ploys agricultural residues as fuels, the thermal oxidative degradationcharacteristic parameters calculated from weight remain and weightloss rate profiles were provided, including (1) Ti, the ignition tem-perature; (2) Tp, the peak temperature; (3) Tb, the burnout temperature;(4) −Rp, the maximum weight loss rate; (5) −Rv, the average weightloss rate; (6) ΔT, the temperature interval at half value of −Rp. Thedefinitions of these parameters were provided in Supplementary data(Chen et al., 2015; Meng et al., 2013).

In addition, the ignition index Ci (Li et al., 2011), burnout index Cb

(Chen et al., 2017b), and comprehensive combustibility index CCI (Xieet al., 2018) were also recommended to analyze the thermal oxidativedegradation performance of PS and SS at different heating rates. Theseindices can be calculated as functions of the thermal oxidative de-composition characteristic temperature, time, and reaction rate.

=−×

CR

t tip

i p (1)

=−× ×

CR

t t tΔbp

1/2 p b (2)

=− × −

×CCI

R RT T

( ) ( )p v

i2

b (3)

where ti, tp, tb, and Δt1/2 are the the ignition time, peak time, burnouttime, and time interval at half value of −Rp, respectively. Ti, Tb, −Rp,and −Rv have been clarified in the preceding part.

2.4. Thermal oxidative degradation kinetic theory

The thermal oxidative degradation kinetics analysis of agriculturalresidues is conducive to acquiring more information from the thermo-gravimetric tests (Zhu et al., 2015). The determined kinetic parameterscan be used to predict the thermal oxidative degradation process andoptimize the thermal conversion behaviors.

2.4.1. The distributed activation energy modelThe distributed activation energy model (DAEM) has been suc-

cessfully used for describing the thermal degradation of various solidmaterials (Cai et al., 2014; Hu et al., 2016). Whether in the inert oroxidative atmosphere, the whole thermal degradation process consistedof a series of irreversible first-order reactions occurring successively.The corresponding kinetic formula is

∫− =∞

∞VV

E T f E dE1 Φ( , ) ( )0 (4)

where V=w0–w, V∞=w0–w∞; =E TΦ( , ) exp ∫− −( )( )dTexpAβ T

T ERT0

***Here, w0, w, and w∞ are the initial weight, weight at time t, and final

weight, respectively. V is the weight loss percentage at time t; V∞ is thetotal weight loss percentage; E, A, β, T, and R represent the activationenergy, pre–exponential factor, heating rate, absolute temperature, anduniversal gas constant, respectively; f(E) is the normalized distributionfunction of the activation energy.

As pointed out by Miura (1995), Φ (E, T) is a step function, which

can be approximated as:

⎜ ⎟≅ ⎛⎝

− ⎛⎝

− ⎞⎠

⎞⎠

E T ARTβE

ERT

Φ( , ) exp exp2

(5)

Through a series of simplification after substituting Eq. (5) into Eq.(4), the DAEM can be written as follows:

⎛⎝

⎞⎠

= ⎛⎝

⎞⎠

+ −β

TARE

ERT

ln ln 0.60752 (6)

In Eq. (6), the plot between ln[β/T2] and 1/T is a straight line forthe same V/V∞ at different heating rates corresponding to the occur-rence of the first-order reaction. The activation energy E and pre-ex-ponential factor A could be determined by the slope and intercept of thelinear-fitting plot, respectively.

2.4.2. The global kinetic modelThe global kinetic model (GKM) is another effective method to de-

termine the kinetic triplets, which is based on the pre–assumption of acertain reaction model. As given in Supplementary data, the most fre-quently used reaction models are chemical order reaction, nucleation,diffusional and contracting geometry, autocatalytic, et al. Utilizing thethermal data from the notable weight loss zone(s), the thermal oxida-tive degradation process of PS and SS could be simulated by the con-version rate equation (Papari and Hawboldt, 2015; Chen et al., 2017a).

= ⇒ = ⎛⎝

− ⎞⎠

dxdt

k T f x h dxdT

Aβ

ERT

f x( )· ( )· (O ) ·exp · ( )2(7)

where dx/dt is the conversion rate, a function of the reaction model f(x), rate constant k(T), and O2 partial pressure function h(O2); x is theconversion degree for each thermal oxidative degradation zone, wherex=(w0–w)/(w0–w∞); For non-isothermal tests performed at heatingrate β= dT/dt, the conversion rate equation can be rearranged as theright of Eq. (7).

The integral form of Eq. (7) is (Vlaev et al., 2008; Du et al., 2014):

∫ ∫= = ⎛⎝

− ⎞⎠

=g x dxf x

Aβ

ERT

dT Aβ

P u( )( )

exp ( )x

T

T

0 0 (8)

where g(x) is the integral form of f(x); u= E/RT and P(u) is the tem-perature integral.

As one of the most common global kinetic method, Coats–Redfernmethod can be employed to calculate the kinetic data of the thermaloxidative degradation process of PS and SS.

⎡⎣

⎤⎦

= ⎡⎣⎢

⎛⎝

− ⎞⎠

⎤⎦⎥

−g xT

ARβE

RTE

ERT

ln( )

ln 1 22 (9)

The relevant temperature integral is:

∫= ⎛⎝

− ⎞⎠

≅ ⎛⎝

− ⎞⎠

⎛⎝

− ⎞⎠

P u ERT

dT RTE

RTE

ERT

( ) exp 2 1 2 expT

T

0 (10)

Here, 2RT/E was very close to zero, namely 1–2RT/E≈ 1, Eq. (9)could be simplified as:

⎡⎣

⎤⎦

= ⎡⎣⎢

⎤⎦⎥

−g xT

ARβE

ERT

ln( )

ln2 (11)

In Eq. (11), when the best kinetic model was selected, the plot be-tween ln[g(x)/T2] and 1/T would provide the highest correlationcoefficient R2. E and A can be determined from the slope and interceptof the regression line, respectively.

3. Results and discussion

3.1. Thermal oxidative degradation process

In the current section, the thermal oxidative degradation processesof PS and SS in air at 5, 10, 20, and 40 Kmin−1 were described with the

X. Ren et al. Bioresource Technology 261 (2018) 403–411

405

weight remaining (Thermogravimetric, TG), weight loss rate(Differential thermogravimetric, DTG), and heat flow (differentialthermal analysis, DTA) profiles. As shown in Figs. 1 and 2, the thermaloxidative degradation process of PS and SS could be roughly dividedinto three zones. The first zone was related to the moisture release, i.e.Biomass (solid, wet)→ Biomass (solid, dry)+H2O (gas); the secondzone was related to the release and combustion of hemicellulose andcellulose, i.e. Biomass (solid, dry)+O2 (gas)→A1 (solid)+B1 (gas);and the third zone corresponded to the combustion of lignin fraction,i.e. A1 (solid)+O2 (gas)→ C (ash)+ B2 (gas). It has been reportedthat the thermal decomposition of hemicellulose and cellulose hap-pened at 493.0–588.0 K and 588.0–673.0 K, respectively (Yang et al.,2007). Moreover, the lignin fraction was more difficult to decompose in433.0–1173.0 K, and the temperature at the maximum weight loss ratewas about 800.0 K. As reported by Álvarez et al. (2016), the thermaloxidative degradation behaviors of 28 common biomass samples werealmost identical. During the combustion process, regardless of moistureevaporation, there were two steps respectively related to the combus-tion of cellulose and lignin. In the previous study, the thermal oxidativedegradation process of four oil-plant residues were also composed ofthree zones (Chen et al., 2017b). Therefore, the thermal reactionschemes proposed in this article were further verified by the previousliterature with regard to various biomass materials. The similar thermaloxidative decomposition schemes for various biomass samples could beascribed to their similar chemical components, including such asmoisture, hemicellulose, cellulose, and lignin.

Furthermore, the heat events during the thermal oxidative decom-position of PS and SS were also analyzed by DTA profiles. As presentedin Fig. 3, there were two notable exothermic peaks in DTA profilesduring the thermal oxidative degradation of PS and SS. Logically, theendothermic peak would accordingly appear in the first zone, whereasit was not reflected from DTA profiles. This was because the moisturecontent in samples was so small that the evaporation did not need muchexternal heat. As for the second and third combustion zone of PS andSS, the locations of exothermic peaks in DTA profiles corresponded wellto those of weight loss rate peaks in DTG profiles. It was observed thatthe heights of exothermic peaks for the second combustion zone of PSwere higher than those for the third zone, while the heights for thesecond combustion zone of SS were lower than those for the third zone.The reason for this difference might be that PS had a higher content ofvolatiles and lower content of fixed carbon than those in SS, which werehighly related to the second and third combustion zone, respectively.

It could be seen in Fig. 1 that an increase in heating rate merelyshifted weight remaining profiles to higher temperatures, while notchanging the pattern of thermal oxidative degradation of PS and SS.With respect to the DTG and DTA, as seen from Figs. 2 and 3, theprofiles also extended to higher temperatures. The temperature shiftswere called thermal hysteresis. It was due to that the higher heatingrate led to a lower heat transfer efficiency during the heating process.The similar findings were also reported in the thermal oxidative de-gradation of lignocellulosic biomass (eucalyptus wood, fir wood, andpine bark) (López-González et al., 2013), rice straw (Xie and Ma, 2013),bio-ferment residue (Du et al., 2013), Tetraselmis suecica (Tahmasebiet al., 2013), olive oil production chain residues (Buratti et al., 2016),karanja (Pongamia pinnata) fruit hulls char (Islam et al., 2016), andbamboo (Liang et al., 2017) at different heating rates.

3.2. Thermal oxidative degradation characteristics

For quantitatively analyzing the effects of heating rates and biomassspecies on the thermal conversion behaviors, the thermal oxidativedegradation characteristic parameters of PS and SS in air atmosphere atheating rates of 5, 10, 20, and 40 Kmin−1 were determined, and theresults were presented in Table 2. As expected, with β increasing from 5to 40 Kmin−1, the thermal oxidative degradation characteristic tem-peratures of PS and SS (Ti, Tp, and Tb) increased. It was found that the

thermal degradation temperatures of PS in the second zone at diverse βswere lower than those of SS, while the temperatures of PS in the thirdzone were higher. The phenomenon might be due to the differences ofvolatile matter and fixed carbon contents in the samples. As for−Rp, Ci,and Cb, with β increasing, the values increased clearly, indicating that abetter ignition and burnout performance could be obtained at higher β.Compared with PS, the values of −Rp, Ci, and Cb for SS were almosthigher at the same β, especially at high β.

The comprehensive combustibility indices CCIs of PS and SS at 5,10, 20, and 40 Kmin−1 were illustrated in Table 3. It could be observedthat, whether PS or SS, the values of CCI grew significantly when theheating rate increased from 5 to 40 Kmin−1. The values of CCI2 fromthe thermal oxidative decomposition of PS and SS were larger one orderthan those of CCI3 at the identical β. Consequently, the overall com-bustion performance of samples was highly depended on the thermaloxidative degradation of volatiles. With β going from 5 to 40 Kmin−1,the value of CCI for PS and SS rose from 3.72× 10−8 to 2.06× 10−6%K−3 min−2, and 3.78×10−8 to 2.52×10−6% K−3 min−2, respec-tively. With the value of β increasing, the combustion performances ofPS and SS were improved 55.38 and 66.67 times, respectively. Com-pared with PS, the CCIs from the thermal oxidative decomposition of SSwere more sensitive to the increase in β. The influences of heating ratesand biomass species on the thermal conversion behaviors of various

(a)

(b)

Fig. 1. Weight remaining profiles of the thermal oxidative degradation of (a) PSand (b) SS at heating rates of 5, 10, 20, and 40 Kmin−1.

X. Ren et al. Bioresource Technology 261 (2018) 403–411

406

biomass materials in the TGA were very common, but data for thermaloxidative degradation of PS and SS have not been reported yet.

3.3. Thermal oxidative degradation kinetics analysis

3.3.1. Kinetic parameters from DAEMIn this paper, the thermal oxidative degradation kinetics of PS and

SS were analyzed by the DAEM, where the reactions for the overalloxidation process were supposed to be single kinetic model corre-sponding to the consecutive conversions. As suggested by the KineticsCommittee of the International Confederation for Thermal Analysis andCalorimetry, the conversion degree V/V∞ or x used in the determina-tion of kinetic parameters by the DAEM were in a broad range of 5–95%with an increment of 5% (Vyazovkin et al., 2011). According to Eq. (6),the experimental data and fitting lines obtained by the DAEM from thethermal oxidative degradation of PS and SS for given values of V/V∞

from 0.05 to 0.95 at heating rates of 5, 10, 20, and 40 Kmin−1 wereshown in Fig. 4. Further, the values of slope, activation energy E, andcorrelation coefficient R2 could be calculated, and the results weregiven in Table 4. As presented in Table 4, the R2 values were in therange of 0.9794–0.9999, which indicated that the values of E calculatedby the DAEM were credible.

As seen from Table 4, the trend of E versus x from the thermal

oxidative degradation of PS was similar to that of SS except V/V∞ lessthan 0.15, which were also available in Supplementary data. It wasbecause that the initial stage of thermal degradation reactions of solidsamples were unstable, the values of E were lack in regularity. Thedifferences in the trend of E versus x between PS and SS at V/V∞ lessthan 0.15 became understandable. With regard to the middle range ofV/V∞, the E values for the thermal oxidative degradation of PS and SSwere relatively invariable. At the end of the thermal oxidative processof PS and SS, the higher values of E could be ascribed to the less vo-latiles and more stable chemical structure of lignin. It was found thatthe values of E for the thermal oxidative decomposition of PS and SSwere 88.94–145.30 kJmol−1 and 94.86–169.18 kJmol−1 in the con-version range of 5–95%, respectively. In the previous study (Shen et al.2011), the E range for the combustion of wood was about175–235 kJmol−1 at the conversion of 10–65% and 300–770 kJmol−1

at the conversion of 70–95%. Xie and Ma (2013) found that the E rangefor the combustion of rice straw was from 95 to 186 kJmol−1 byFriedman and from 97 to 204 kJmol−1 by OFW at the conversion of20–80%. Taking the differences in the biomass species, experimentalconditions, and kinetic calculation methods into consideration, the Eranges for PS and SS in this study showed good agreement with theconclusion published in the literature.

Fig. 2. Weight loss rate profiles of the thermal oxidative degradation of (a) PSand (b) SS at heating rates of 5, 10, 20, and 40 Kmin−1.

(a)

(b)

Fig. 3. Differential thermal analysis profiles of the thermal oxidative degrada-tion of (a) PS and (b) SS at heating rates of 5, 10, 20, and 40 Kmin−1.

X. Ren et al. Bioresource Technology 261 (2018) 403–411

407

3.3.2. Kinetic parameters from GKMTo obtain kinetic parameters from GKM, the conversion degrees x of

the individual zone of PS and SS should be recalculated first, whichwere the function of the temperature. The values of x used for kineticsanalysis were in the range of 5–95%. The kinetic parameters for theglobal reaction of the oxidative decomposition of PS and SS at distinctheating rates were illustrated in Table 5. It could be observed that theR2 data for various oxidative decomposition zones were in the range of0.9742–0.9941, indicating that the mechanism models selected for ki-netics analysis were reliable. As listed in Table 5, whether PS or SS, theE value for the second zone was clearly lower than that for the thirdzone at identical heating rate. Compared with lignin, the structures forhemicellulose and cellulose were more amorphous and random, thustheir oxidative degradations happened at lower temperatures, and theminimum energies required to break the chemical bonds betweenatoms were lower. The E values for the second zone of PS at differentheating rates were slightly smaller than those of SS, whereas the valuesfor the third zone of PS at distinct heating rates were higher than thoseof SS. It was because the contents of volatiles and fixed carbon in PS

sample were respectively higher and lower than those in SS sample. Asreported by Álvarez et al. (2016), the kinetics analysis on the thermaloxidative degradation of 28 biomass samples have also been performedby Coats–Redfern method. It was found that the E values for the com-bustion of volatiles and lignin in the range 17.5–155.0 kJ mol−1 and16.2–237.0 kJmol−1, respectively.

3.3.3. Comparison of two kinetic methodsThe thermal oxidative decomposition of agricultural residues is a

complicated process due to involving a set of chemical reactions andphysical processes. Therefore, the single kinetic method cannot accu-rately determine the kinetic parameters of biomass oxidative degrada-tion. It is more reliable to evaluate the thermal data using multiplekinetic methods together with differences comparing (Vyazovkin et al.,2011; Gai et al., 2013; Chen et al., 2017a).

The DAEM and GKM are common model–free method and mod-el–fitting method, respectively. Because of different assumptions adopted,the results obtained from above two approaches presented a certain dis-crepancy as shown in Fig. 5. In the DAEM, the value of E obtained is a

Table 2Thermal oxidative degradation characteristics of PS and SS at 5, 10, 20, and 40 Kmin−1.

Samples βa Zone 2 Zone 3

Ti2b −Rp2c Tp2d Tb2e Ci2

f Cb2g Ti3b −Rp3

c Tp3d Tb3e Ci3f Cb3

g

PS5 431.6 3.80 578.2 656.0 2.98× 10−3 6.41× 10−5 656.2 1.07 724.8 760.0 1.92× 10−4 2.30× 10−5

10 436.2 8.02 599.8 677.2 2.29× 10−2 8.39× 10−4 677.4 2.83 746.6 775.0 1.87× 10−3 7.21× 10−4

20 446.8 16.49 612.8 710.4 1.67× 10−1 1.42× 10−2 710.6 7.34 761.8 800.0 1.81× 10−2 1.29× 10−2

40 461.8 31.69 621.6 720.2 1.06 2.44× 10−1 720.4 12.42 768.0 845.2 1.29× 10−1 6.14× 10−2

SS5 409.6 3.70 566.2 655.2 3.22× 10−3 6.10× 10−5 655.4 1.34 726.8 845.2 2.36× 10−4 1.28× 10−5

10 445.2 7.75 577.2 661.2 2.03× 10−2 1.12× 10−3 661.4 3.59 734.4 785. 2.39× 10−3 5.62× 10−4

20 462.6 16.64 601.6 686.0 1.44× 10−1 1.53× 10−2 686.2 12.52 750.0 810.2 3.13× 10−2 2.60× 10−2

40 474.8 37.85 617.6 696.0 1.12 3.42× 10−1 696.2 17.01 760.4 850.8 1.77× 10−1 1.13× 10−1

a β, heating rate, K min−1.b Ti, the ignition temperature, K; Ti2 and Ti3 are the ignition temperature of the second and third zone, respectively.c Tp, the peak temperature of each zone, K; Tp2 and Tp3 are the peak temperature of the second and third zone, respectively.d −Rp, the maximum weight loss rate of each zone, wt% min−1; −Rp2 and −Rp3, are the maximum weight loss rate of the second and third zone, respectively.e Tb, the burnout temperature, K; Tb2 and Tb3 are the burnout temperature of the second and third zone, respectively.f Ci, the ignition index, wt% min−3; Ci2 and Ci3 are the ignition index of the second and third zone, respectively.g Cb, the burnout index, wt% min−4; Cb2 and Cb3 are the ignition index of the second and third zone, respectively.

Table 3Comprehensive combustibility index CCI of PS and SS at 5, 10, 20, and 40 Kmin−1.

Samples β Zone 2 Zone 3 CCId

−Rv2a ΔT2b WL2c CCI2d −Rv3

a ΔT3b WL3c CCI3d

Kmin−1 % min−1 K % % K−3 min−2 % min−1 K % % K−3 min−2 % K−3 min−2

PS5 1.50 87.4 65.62 4.66× 10−8 0.87 32.4 18.00 2.86×10−9 3.72× 10−8

10 2.99 104.3 69.36 1.86× 10−7 1.98 19.3 19.40 1.58×10−8 1.49× 10−7

20 6.01 101.0 72.30 6.99× 10−7 3.90 20.8 17.66 7.08×10−8 5.75× 10−7

40 13.01 104.4 67.03 2.69× 10−6 7.16 63.8 22.44 2.03×10−7 2.06× 10−6

SS5 1.42 83.3 70.79 4.76× 10−8 0.55 52.6 19.48 2.02×10−9 3.78× 10−8

10 3.07 78.8 64.01 1.81× 10−7 1.73 30.0 21.22 1.80×10−8 1.41× 10−7

20 6.38 96.7 65.18 7.23× 10−7 3.58 15.1 22.04 1.18×10−7 5.70× 10−7

40 14.13 90.8 62.66 3.41× 10−6 6.60 45.2 24.85 2.72×10−7 2.52× 10−6

a −Rv, the average weight loss rate, wt% min−1; −Rv2 and −Rv3 are the average weight loss rate of the second and third zone, respectively.b ΔT, the temperature interval at half value of −Rp of each zone; ΔT2 and ΔT3 are the temperature interval at half value of −Rp of the second and third zone,

respectively.c WL, the weight loss of each zone; WL2 and WL3 are the weight loss of the second and third zone, respectively.d CCI, the comprehensive combustibility index, wt%2 min−2 K−3; CCI2 and CCI3 are the comprehensive combustibility index of the second and third zone,

respectively.

X. Ren et al. Bioresource Technology 261 (2018) 403–411

408

function of the conversion degree x. Multistep reaction mechanism of thecombustion process could be identified from the knowledge of the re-lationship between E and x. While in the GKM, such as Coats-Redfernmethod, the E value was treated as a constant in the whole process. Hence,the errors of the kinetic parameters from model-fitting approaches may berelatively larger than those from model-free approaches. As seen fromFig. 5, no matter PS or SS, both DAEM and GKM could give acceptablecorrelations to the experimental data, especially at the second thermaloxidative degradation zone. However, the errors between theoretical cal-culation and experimental results were relative large at the third zone.Whether PS or SS, with the heating rate increasing, the discrepancies at thethird zone decreased. The similar results have also been reported by Chenet al. (2017a) in the cattle manure degradation. The errors between ex-perimental results and simulated curves by the DAEM and GKM were didexist. It might be due to that the DAEM method requires the values ofactivation energy to be distributed continually, and the whole thermaldecomposition process should obey a set of first–order reactions occurringsuccessively. As for GKM method, it was based on the preliminary as-sumption of reaction model, and then calculating the activation energyand pre–exponential factor.

(a)

(b)

Fig. 4. Experimental (scatter) and fitting (line) values obtained by the DAEMfrom the thermal oxidative degradation of (a) PS and (b) SS at different heatingrates.

Table 4Slope, activation energy E, and correlation coefficient R2 obtained by the DAEM from the thermal oxidative degradation of PS and SS.

Samples V/V∞/% Slope E/kJmol−1 R2 Sample V/V∞/% Slope E/kJmol−1 R2

PS 0.05 −17.476 145.30 0.9875 SS 0.05 −11.409 94.86 0.98800.10 −16.021 133.21 0.9950 0.10 −12.934 107.54 0.99920.15 −14.281 118.74 0.9900 0.15 −13.140 109.25 0.99870.20 −14.202 118.08 0.9853 0.20 −13.064 108.62 0.99880.25 −13.490 112.16 0.9916 0.25 −12.851 106.85 0.99860.30 −12.923 107.44 0.9871 0.30 −12.848 106.82 0.99850.35 −12.713 105.70 0.9859 0.35 −12.687 105.49 0.99830.40 −12.645 105.14 0.9876 0.40 −13.263 110.28 0.99480.45 −12.732 105.86 0.9885 0.45 −13.537 112.55 0.99620.50 −12.619 104.92 0.9906 0.50 −13.895 115.53 0.99320.55 −11.697 97.26 0.9845 0.55 −13.819 114.90 0.99290.60 −10.697 88.94 0.9837 0.60 −13.757 114.38 0.99290.65 −11.003 91.48 0.9905 0.65 −12.254 101.89 0.99880.70 −11.486 95.50 0.9940 0.70 −11.750 97.70 0.99400.75 −11.785 97.99 0.9957 0.75 −11.961 99.45 0.99510.80 −12.323 102.46 0.9994 0.80 −13.533 112.52 0.98940.85 −13.063 108.61 0.9999 0.85 −17.110 142.26 0.98020.90 −13.928 115.80 0.9999 0.90 −20.347 169.18 0.97940.95 −16.124 134.06 0.9994 0.95 −20.348 169.18 0.9805

Average 109.93 115.75

Table 5Kinetic parameters obtained by the GKM from the thermal oxidative degrada-tion of PS and SS at 5, 10, 20, and 40 Kmin−1.

Samples β Temperaturerange ΔT/K

Activationenergy E/kJmol−1

Pre-exponentialfactor A/s−1

Kineticmodel f(x)

CorrelationcoefficientR2

PS 5 431.6–656.0 72.18 4.80× 103 1−x 0.9835656.2–760.0 227.21 4.90× 1014 (1−x)2 0.9797

10 436.2–677.2 67.67 2.42× 103 1−x 0.9843677.4–775.0 252.31 2.11× 1016 (1−x)2 0.9768

20 446.8–710.4 62.68 9.85× 102 1−x 0.9802710.6–800.0 349.46 6.94× 1022 (1−x)2 0.9815

40 461.8–720.2 63.59 1.66× 103 1−x 0.9941720.4–845.2 238.62 6.35× 1014 (1−x)2 0.9815

SS 5 409.6–655.2 70.59 4.40× 103 1−x 0.9816655.4–845.2 138.96 6.18× 106 1−x 0.9854

10 445.2–661.2 79.46 4.43× 104 1−x 0.9828661.4–785.2 152.12 4.56× 108 1−x 0.9883

20 462.6–686.0 77.64 3.58× 104 1−x 0.9820686.2–810.2 189.82 2.35× 1011 1−x 0.9901

40 474.8–696.0 87.72 1.63× 105 1−x 0.9925696.2–850.8 164.77 3.47× 109 1−x 0.9926

X. Ren et al. Bioresource Technology 261 (2018) 403–411

409

(a) (b)

(c) (d)

(e) (f)

(g) (h)

Fig. 5. Experimental data and calculated values obtained by the DAEM and GKM from the thermal oxidative degradation of (a–d) PS at 5, 10, 20, and 40 Kmin−1,(e–g) SS at 5, 10, 20, and 40 Kmin−1.

X. Ren et al. Bioresource Technology 261 (2018) 403–411

410

4. Conclusions

The thermal oxidative degradation process of peanut shell andsunflower shell was composed of moisture release, hemicellulose andcellulose combustion, and lignin oxidation. As revealed by the com-bustion characteristic parameters, the thermal characteristics werefound to be influenced by biomass species and heating rates. Theaverage activation energies calculated by the DAEM for peanut shelland sunflower shell were 109.93 kJmol−1 and 115.75 kJmol−1, re-spectively. In the GKM, the activation energy for hemicellulose/cellu-lose combustion was obviously lower than that for lignin oxidation atidentical heating rate. To some degree, the calculated kinetic para-meters could acceptably simulate experimental data.

5. Notes

The authors declare no competing financial interest.

Acknowledgements

Financial support for this work that has been provided by theFundamental Research Funds for the Central Universities, UnitedKingdom (2017BQ062 and D2155190) is gratefully acknowledged.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in theonline version, at https://doi.org/10.1016/j.biortech.2018.04.047.

References

Álvarez, A., Pizarro, C., García, R., Bueno, J.L., Lavín, A.G., 2016. Determination of ki-netic parameters for biomass combustion. Bioresour. Technol. 216, 36–43.

Buratti, C., Mousavi, S., Barbanera, M., Lascaro, E., Cotana, F., Bufacchi, M., 2016.Thermal behaviour and kinetic study of the olive oil production chain residues andtheir mixtures during co-combustion. Bioresour. Technol. 214, 266–275.

Bach, Q.V., Chen, W.H., 2017. Pyrolysis characteristics and kinetics of microalgae viathermogravimetric analysis (TGA): a state-of-the-art review. Bioresour. Technol. 246,88–100.

Cai, J., Wu, W., Liu, R., 2014. An overview of distributed activation energy model and itsapplication in the pyrolysis of lignocellulosic biomass. Renew. Sustain. Energy Rev.36, 236–246.

Chen, C., Ma, X., He, Y., 2012a. Co-pyrolysis characteristics of microalgae Chlorella vul-garis and coal through TGA. Bioresour. Technol. 117, 264–273.

Chen, G., He, S., Cheng, Z., Guan, Y., Yan, B., Ma, W., Leung, D.Y., 2017a. Comparison ofkinetic analysis methods in thermal decomposition of cattle manure by themo-gravimetric analysis. Bioresour. Technol. 243, 69–77.

Chen, J., Mu, L., Cai, J., Yao, P., Song, X., Yin, H., Li, A., 2015. Pyrolysis and oxy-fuelcombustion characteristics and kinetics of petrochemical wastewater sludge usingthermogravimetric analysis. Bioresour. Technol. 198, 115–123.

Chen, J., Wang, Y., Lang, X., Ren, X.E., Fan, S., 2017b. Comparative evaluation of thermaloxidative decomposition for oil-plant residues via thermogravimetric analysis:thermal conversion characteristics, kinetics, and thermodynamics. Bioresour.Technol. 243, 37–46.

Chen, L., Xing, L., Han, L., 2009. Renewable energy from agro-residues in China: solidbiofuels and biomass briquetting technology. Renew. Sustain. Energy Rev. 13,2689–2695.

Chen, W.-H., Lu, K.-M., Tsai, C.-M., 2012b. An experimental analysis on property andstructure variations of agricultural wastes undergoing torrefaction. Appl. Energy 100,318–325.

Du, Y., Jiang, X., Lv, G., Ma, X., Jin, Y., Wang, F., Chi, Y., Yan, J., 2014. Thermal behaviorand kinetics of bio-ferment residue/coal blends during co-pyrolysis. Energy Convers.Manage. 88, 459–463.

Du, Y., Jiang, X., Ma, X., Liu, X., Lv, G., Jin, Y., Wang, F., Chi, Y., Yan, J., 2013. Evaluationof cofiring bioferment residue with coal at different proportions: combustion char-acteristics and kinetics. Energy Fuels 27, 6295–6303.

Ellabban, O., Abu-Rub, H., Blaabjerg, F., 2014. Renewable energy resources: currentstatus, future prospects and their enabling technology. Renew. Sustain. Energy Rev.39, 748–764.

Gai, C., Dong, Y., Zhang, T., 2013. The kinetic analysis of the pyrolysis of agriculturalresidue under non-isothermal conditions. Bioresour. Technol. 127, 298–305.

Gai, C., Liu, Z., Han, G., Peng, N., Fan, A., 2015. Combustion behavior and kinetics of low-lipid microalgae via thermogravimetric analysis. Bioresour. Technol. 181, 148–154.

Garcia-Maraver, A., Perez-Jimenez, J.A., Serrano-Bernardo, F., Zamorano, M., 2015.Determination and comparison of combustion kinetics parameters of agriculturalbiomass from olive trees. Renew. Energy 83, 897–904.

Hu, M., Chen, Z., Wang, S., Guo, D., Ma, C., Zhou, Y., Chen, J., Laghari, M., Fazal, S., Xiao,B., Zhang, B., Ma, S., 2016. Thermogravimetric kinetics of lignocellulosic biomassslow pyrolysis using distributed activation energy model, Fraser-Suzuki deconvolu-tion, and iso–conversional method. Energy Convers. Manage. 118, 1–11.

Islam, M.A., Auta, M., Kabir, G., Hameed, B.H., 2016. A thermogravimetric analysis of thecombustion kinetics of karanja (Pongamia pinnata) fruit hulls char. Bioresour.Technol. 200, 335–341.

Jiang, X., Chen, D., Ma, Z., Yan, J., 2017. Models for the combustion of single solid fuelparticles in fluidized beds: a review. Renew. Sustain. Energy Rev. 68, 410–431.

Li, X.G., Lv, Y., Ma, B.G., Jian, S.W., Tan, H.B., 2011. Thermogravimetric investigation onco-combustion characteristics of tobacco residue and high-ash anthracite coal.Bioresour. Technol. 102, 9783–9787.

Liang, F., Wang, R., Jiang, C., Yang, X., Zhang, T., Hu, W., Mi, B., Liu, Z., 2017.Investigating co-combustion characteristics of bamboo and wood. Bioresour.Technol. 243, 556–565.

López, R., Fernández, C., Fierro, J., Cara, J., Martínez, O., Sánchez, M.E., 2014. Oxy-combustion of corn, sunflower, rape and microalgae bioresidues and their blendsfrom the perspective of thermogravimetric analysis. Energy 74, 845–854.

López-González, D., Fernandez-Lopez, M., Valverde, J.L., Sanchez-Silva, L., 2013.Thermogravimetric-mass spectrometric analysis on combustion of lignocellulosicbiomass. Bioresour. Technol. 143, 562–574.

Ma, L., Wang, T., Liu, Q., Zhang, X., Ma, W., Zhang, Q., 2012. A review of thermal–-chemical conversion of lignocellulosic biomass in China. Biotechnol. Adv. 30,859–873.

Meng, F., Yu, J., Tahmasebi, A., Han, Y., 2013. Pyrolysis and combustion behavior of coalgangue in O2/CO2 and O2/N2 mixtures using thermogravimetric analysis and a droptube furnace. Energy Fuels 27, 2923–2932.

Miura, K., 1995. A new and simple method to estimate f(E) and k0(E) in the distributedactivation energy model from three sets of experimental data. Energy Fuels 9,302–307.

National Bureau of Statics of China, 2017. China Statistical Yearbook. http://www.stats.gov.cn/tjsj/ndsj/2017/indexch.htm.

Papari, S., Hawboldt, K., 2015. A review on the pyrolysis of woody biomass to bio-oil:focus on kinetic models. Renew. Sustain. Energy Rev. 52, 1580–1595.

Saidur, R., Abdelaziz, E.A., Demirbas, A., Hossain, M.S., Mekhilef, S., 2011. A review onbiomass as a fuel for boilers. Renew. Sustain. Energy Rev. 15, 2262–2289.

Shen, D.K., Gu, S., Jin, B., Fang, M.X., 2011. Thermal degradation mechanisms of woodunder inert and oxidative environments using DAEM methods. Bioresour. Technol.102, 2047–2052.

Tahmasebi, A., Kassim, M.A., Yu, J., Bhattacharya, S., 2013. Thermogravimetric study ofthe combustion of Tetraselmis suecica microalgae and its blend with a Victorian browncoal in O2/N2 and O2/CO2 atmospheres. Bioresour. Technol. 150, 15–27.

Vlaev, L., Nedelchev, N., Gyurova, K., Zagorcheva, M., 2008. A comparative study of non-isothermal kinetics of decomposition of calcium oxalate monohydrate. J. Anal. Appl.Pyrol. 81, 253–262.

Vyazovkin, S., Burnham, A.K., Criado, J.M., Pérez-Maqueda, L.A., Popescu, C.,Sbirrazzuoli, N., 2011. ICTAC Kinetics Committee recommendations for performingkinetic computations on thermal analysis data. Thermochim. Acta 520, 1–19.

Wang, S., Dai, G., Yang, H., Luo, Z., 2017. Lignocellulosic biomass pyrolysis mechanism: astate-of-the-art review. Prog. Energy Combust. Sci. 62, 33–86.

Wang, X., Chen, H., Yang, H., Wang, J., Xin, F., 2012. The influence of alkali and alkalineearth metal compounds on pyrolysis of peanut shell. Asia-Pac. J. Chem. Eng. 7,463–468.

White, J.E., Catallo, W.J., Legendre, B.L., 2011. Biomass pyrolysis kinetics: a comparativecritical review with relevant agricultural residue case studies. J. Anal. Appl. Pyrolysis91, 1–33.

Xie, C., Liu, J., Zhang, X., Xie, W., Sun, J., Chang, K., Kuo, J., Xie, W., Liu, C., Sun, S.,Buyukada, M., Evrendilek, F., 2018. Co-combustion thermal conversion character-istics of textile dyeing sludge and pomelo peel using TGA and artificial neural net-works. Appl. Energy 212, 786–795.

Xie, Z., Ma, X., 2013. The thermal behaviour of the co-combustion between paper sludgeand rice straw. Bioresour. Technol. 146, 611–618.

Yahya, M.A., Al-Qodah, Z., Ngah, C.Z., 2015. Agricultural bio-waste materials as potentialsustainable precursors used for activated carbon production: a review. Renew.Sustain. Energy Rev. 46, 218–235.

Yang, H., Yan, R., Chen, H., Lee, D.H., Zheng, C., 2007. Characteristics of hemicellulose,cellulose and lignin pyrolysis. Fuel 86, 1781–1788.

Zhu, X., Chen, Z., Xiao, B., Hu, Z., Hu, M., Liu, C., Zhang, Q., 2015. Co-pyrolysis behaviorsand kinetics of sewage sludge and pine sawdust blends under non-isothermal con-ditions. J. Therm. Anal. Calorim. 119, 2269–2279.

X. Ren et al. Bioresource Technology 261 (2018) 403–411

411