Thermal behavior research for co-combustion of furfural residue and oil shale semi-coke

Applied Thermal Engineering 120 (2017) 19–25

Contents lists available at ScienceDirect

Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

Thermal behavior research for co-combustion of furfural residue and oil shale semi-coke Hong Qin, Wanli Wang, Hongpeng Liu, Lidong Zhang, Qing Wang ⇑, Chuangye Shi, Kewei Yao Engineering Research Center of Oil Shale Comprehensive Utilization Ministry of Education, Northeast Electric Power University, Changchun Road 169, Jilin 132012, China

h i g h l i g h t s
Combustion behavior of oil shale semi-coke and furfural residue was studied.
The effect of heating rate and mixing ratio were researched.
The mutual effect of blends occurred in the process of co-combustion.
The kinetics parameters were calculated by Coats-Redfern method.

a r t i c l e

i n f o

Article history: Received 18 December 2016 Revised 22 March 2017 Accepted 25 March 2017 Available online 27 March 2017 Keywords: Furfural residue Semi-coke Co-combustion Synergy Kinetics

a b s t r a c t The thermal behavior of Longkou oil shale semi-coke, furfural residue and their blends was researched in this paper. The experiment was carried out using TG-FTIR to research the combustion mechanism, under different heating rate (10, 20, 40 and 80 °C/min) and simulation of air condition (80% nitrogen: 20% oxygen). The oil shale semi-coke was blended with furfural residue in proportions from 20% to 100%. Two following factors were studied: heating rate and mixing ratio. The combustion behavior of semi-coke can be promoted furfural residue. And the synergistic effect was calculated by means of interaction coefficient f and the Relative Error of Mean Square root. The combustion of two samples can be divided into two stages, and three stages of co-combustion. Synergy occurred in every stage. At the same time, the cocombustion kinetics parameters were calculated by the Coats-Redfern method. The activation energy of the mixtures first decreased and then increased with the degree of reaction. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction In China, oil shale is still regarded as an essential supplement for limited petroleum resources [1]. Semi-coke generated from the process of oil shale retorting, which still contains a certain calorific value, is always abandoned. Generally, every ton of shale oil production produces about 10–30 tons of semi-coke [2]. So much semi-coke discharge not merely occupies abundant land, but does harm to environment [3], from which the leachate contains sulfide, hydrocarbons, poly-cyclic aromatic hydro-carbon [4], and trace amounts of other toxic elements. Until now no large-scale utilization of these solid wastes has been found in China. Some researchers suggested burning petroclastic shale, retorting and excess gas to generate electricity [5,6]. Opik et al. [7] proposed that mixing some oil shale into semicoke was necessary to supplement insufficient heat value from

⇑ Corresponding author. E-mail address: [email protected] (Q. Wang). http://dx.doi.org/10.1016/j.applthermaleng.2017.03.111 1359-4311/Ó 2017 Elsevier Ltd. All rights reserved.

semi-coke for stable combustion. Arro et al. [8] compared different combustion modes and proposed to burn the mixture of semi-coke and oil shale by means of fluidized bed combustion. Kaijuvee et al. [9] studied the adsorption property of sulfur of oil shale, semi-coke and their mixture when burning in the fluidized bed, in which he concluded the semi-coke can absolutely adsorb the SO2 generated. Qin et al. [10] studied the co-combustion kinetics of shale oil sludge and semi-coke, and found the oil shale sludge improved the combustion of semi-coke. Furfural residue, generated after extracting furfural from corncob, is a kind of biomass waste. There generates about 12–15 tons of residue per ton of furfural production. Furfural residue is rich in cellulose, hemicellulose and lignin, which own great recycling value [11]. However, its high salt content may pollute the atmosphere, soil and river if abandoned directly. Therefore, effective measures should be taken to eliminate these detrimental influence in the furfural industry. It is noteworthy that the two solid wastes have a certain calorific value, which are available for energy utilization by putting them

20

H. Qin et al. / Applied Thermal Engineering 120 (2017) 19–25

into boiler. Semi-coke of oil shale is hard to ignite solely because of a lower calorific value contained. While it can be solved by mixing the two materials because furfural residue contains much calorific value close to bituminous coal, which supplements insufficient heat of semi-coke. In addition, co-combustion of furfural residue and oil shale semi-coke may hopefully become components into non-hazardous. Thus semi-coke of oil shale and furfural residue being made energy use may be an ideal solution. Till now few reports have been found for co-combustion research of semi-coke and furfural residue. While a number of studies have been conducted for co-combustion of other fuels. Liu et al. [12] studied thermal behavior and the synergy of oil shale semi-coke with torrefied cornstalk, and concluded that cocombustion increased with greater degree of heat-treated cornstalk, and interpretation of the MR curve indicated that the influence of the torrefied cornstalks on the combustion process is observed mainly during the second phase [13]. Gayan et al. [14] found that blending biomass improved combustion efficiency when burning coal in a circulating fluidized bed. Sahu et al. [15] selected different biomass carbo-coal mixing with semi-coke to research combustion kinetics through DSC-TGA experiment. The results showed the burnout rate increased with burning time for combustion of coal mixed with less than 50% biomass. Varol et al. [16] found the ignition temperature of biomass and low grade lignite was close to pure biomass when burning in a thermogravimetric analyzer. Wang et al. [17] found the temperatures of ignition and burnout reduced with the increasing rice straw proportion. In actual engineering project, combustion of biomass is a low cost low risk of renewable energy utilization way. It’s significant for complementing the shortage of fossil fuels by the burning of these two abandoned residues, and reducing the emissions of traditional pollutant (SO2, NOx) as well as greenhouse gas (CO2, CH4). The achievement of this study may provide reference for renewable energy application. Also it is meaningful for conventional fossil fuel consumption in energy production in China. 2. Experimental section The experiments were performed on a thermogravimetric Analysis TGA/DSC1 (Mettler-Toledo, Switzerland) and Fourier Transform Infrared Spectrometer (Nicolet IS10, Thermo Scientific, United State). 10–20 mg of each sample is prepared in the tests. Four heating rates were arranged, i.e. 10, 20, 40, 80 °C/min, and the reaction temperature at 50–950 °C. The gas atmosphere was synthetic air with 50 mL/min (O2:N2 = 1:4). The spectral detection ranged between 4000 and 400 cm1, with 4 cm1 of resolution, and 20 kHz of 200 cycles of scanning rate. The semi-coke was made by retorting Longkou oil shale. Furfural residue was selected from a furfural factory located in Shuangyang District, Changchun city. The samples were ground into fine particle (<0.2 mm). Owing to high moisture contained in the furfural residue, a drying process was needed. The furfural residue was blended with semi-coke in certain proportions. The data in detail are listed in Table 1 where FR and SC are respectively the abbreviations of furfural residue and semi-coke. Table 2 shows the fuel property of two fuels, containing proximate, ultimate analyses and heat value. There are some differences between the two materials. The volatile contained in semi-coke is

less while abundant in furfural residue. On the contrary, the ash content of semi-coke is much higher than that of furfural residue. The ash and volatile content of the two fuels forms complementary trend, so it is possible to promote properties for each other during their combustion.

3. Results and discussion 3.1. Combustion of pure samples The curves of combustion for furfural residue and oil shale semi-coke under a heating rate of 20 °C/min are shown in Fig. 1, where Fig. 1a and b are respectively TG-DTG curves of semi-coke and furfural residue. Fig. 1a shows the combustion of semi-coke can be divided into two stages, thermal decomposition of volatile (360–655 °C) and fixed carbon burnout (655–810 °C). From the figure one can see the sample weightlessness is small because of low content of surplus volatile and fixed carbon for semi-coke. The combustion mainly concentrates on release and thermal decomposition of volatile at 360–655 °C in the first stage, with only 16% of weightlessness. While the second stage mainly shows smaller weight loss than the first with only 8% of weightlessness at 655–810 °C, in which a peak occurs at 760 °C. Fig. 1b shows two stages are also contained in the whole combustion process of furfural residue, namely burning of volatile and fixed burning out of carbon. The first stage occurs at 260– 370 °C, where the weight loss reaches 49%. And 43% of weightlessness takes place at the second stage (370–680 °C). Unlike semicoke, furfural residue contains abundant components materials volatile and fixed carbon, which lead to so much range variation in mass of sample. It is also deduced that furfural residue behaves superior to semi-coke. From the DTG curve of furfural residue, it is shown two peaks distributed in the process. It’s noteworthy that the amplitude of first one is larger than the latter, while the first peak width smaller than the latter. This phenomenon just confirmed the judgment of two stages distribution. Generally, fluidized bed combustion not only need sufficient burning, but a substantial non-combustible solid materials save considerable heat to maintain bed temperature. The ash of semicoke may possess thermal storage capacity. Because of the disparate behaviors between the two samples during combustion, there might be possible to realize favoring combustion characteristics to each other when mixes them together.

3.2. Effect of mixing ratio Fig. 2 shows TG-DTG curves of four kinds of blended samples under the heating rate of 20 °C/min. From the figure b, one can see that there are three peaks during the mixture co-combustion process, therefore the co-combustion can be divided into three distinct phases: The combustion stage of volatile in the furfural residue, releasing and thermal decomposition of volatile of semi-coke and burning of fixed carbon of furfural residue, and, burning of fixed carbon in the semi-coke. With the increase of furfural residue proportion, the weightlessness rate of samples significantly increased in the TG curve; and the DTG curve shows the weightlessness rate increase in every stage, especially in the first two

Table 1 Proportion of sample mixtures. Serial number

S1

S2

S3

S4

S5

S6

FR:SC

0:10

2:8

4:6

6:4

8:2

10:0

21

H. Qin et al. / Applied Thermal Engineering 120 (2017) 19–25 Table 2 Proximate, ultimate analyses and calorific value of samples. Sample

Proximate analysis

Semi-coke Furfural slag

Qnet,ar/(J/g)

Mad

Vad

Aad

FCad

1.23 2.90

12.75 69.44

76.46 4.31

9.56 23.35

4622 19,763

0.0000 100

-0.0002

-0.0008

TG DTG

80

-0.0010

TG/%

TG/%

-0.0006

70

0

200

400

600

800

Had

Oad

Nad

Sad

15.79 24.28

0.84 2.32

4.9 18.56

0.61 0.26

0.89 0.83

100

0.000

80

-0.002 -0.004

60

-0.006

40

-0.008

TG DTG

20

-0.0012

Cad

-0.010

0

1000

0

200

400

600

T/°C

T/°C

(a) S1

(b) S6

800

1000

DTG(%/min)

90

DTG(%/min)

-0.0004

Ultimate analysis

-0.012

Fig. 1. The TG-DTG curves of S1 and S6.

80

-0.002

DTG/(%/min)

0.000

TG/%

100

S2 S3 S4 S5

60 40 20

-0.004

S2 S3 S4 S5

-0.006 -0.008 -0.010

200

0

400

600

800

1000

0

200

400

600

T/°C

T/°C

(a) TG

(b) DTG

800

1000

Fig. 2. The TG-DTG curves of S2 to S5 under different mixing ratios.

TG and DTG curves were obtained. However, the evolution trend of these combustion curves are similar. With the increase of heating rate, these curves move towards higher temperature, and the initial and terminal temperature of each phase become higher. The reason of this phenomenon may be explained as follows: (1) The high heating rate shortens the reaction time, which cause deficiency of oxygen, so the reaction is not sufficient. (2) High

stages, which indicates that furfural residue can improve the combustion behavior of semi-coke. 3.3. Effect of heating rate Fig. 3 gives the TG-DTG curves of sample S3 under different heating rates (10, 20, 40, 80 °C). As is shown in the fig. 3, different

0.000

100

10°C/min 20°C/min 40°C/min 80°C/min

TG/%

80

-0.001

DTG/(%/min)

90

70 60 50 40

-0.002

10°C/min 20°C/min 40°C/min 80°C/min

-0.003 -0.004 -0.005

0

200

400

600

800

1000

-0.006

0

200

400

600

T/°C

T/°C

(a) TG

(b) DTG

Fig. 3. The TG and DTG curves of S3 under different heating rates.

800

1000

22

H. Qin et al. / Applied Thermal Engineering 120 (2017) 19–25

temperature cause high ignition and burnout temperature of samples. (3) The temperature gradient between exterior and interior of the sample particles become larger under high heating rate, which slows down the precipitation of volatile.

In order to analyze the experimental and theoretical curves of TG and DTG based on the same temperature, theoretical calculation formulas are as follows:




 dM dM dM ¼ xsc þ xfr dt cal dt sc dt fr

Rexp DT 1=2;exp =T p;exp Rcal DT 1=2;cal =T p;cal

RMS ¼

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u
 uPn xiexp xi 2 cal u t i¼1 xical n

Under the condition of linear heating rate, b = dT/dt, the equation can be written as:

ð6Þ

ð1Þ

For combustion process at low heating rate, the reaction rate can be regarded as being controlled by the chemical power factor, the relationship between reaction rate and temperature follows the Arrhenius law, so the equation can be written as:

ð2Þ


 da A E f ðaÞ ¼ exp  RT dT b

ð7Þ

a ¼ ðm0  mT Þ=ðm0  mf Þ

ð8Þ

where M is the rest quality percentage of sample at t time, %; xsc and xfr are the quality percentage of two fuels in the mixture, %; (dM/dt)sc and (dM/dt)fr are the weight loss ratio, %/min. In addition, in order to analyze the synergy in the combustion process, interaction coefficient f [18] and relative root mean square error RMS [19] were used to research the mutual effect in the combustion process. Among them, f can be used to identify whether mixture is beneficial to combustion. When f > 1, the two promote each other; conversely, f < 1, the mutual inhibition occurs. And the RMS can reflect the degree of mutual effect. The formulas of two indicators are listed as follows:

f ¼

ð5Þ

da A ¼ kðTÞf ðaÞ dT b

4. The mutual effect of mixture co-combustion

M cal ¼ xsc M sc þ xfr M fr

da A ¼ kðTÞf ðaÞ dt b

ð3Þ

where a is the conversion degree, %; A is the frequency Arrhenius factor, min1; E is the activation energy, kJ/mol; R is the universal gas constant; f(a) is a function depending on the reaction mechanism; m0, mT, mf are respectively the initial mass, the mass under T and remainder mass of sample at the end of reaction. By using Coats-Redfern integral method, the mechanism formula is:

f ðaÞ ¼ ð1  aÞn

In this case, determination the form of f(a) is converted to obtaining the reaction index n,

 ln

ð4Þ

where R is the value of weightlessness peak, %; DT1/2 is the half peak width, °C; Tp is the peak temperature, °C; Xi is the weight loss rate value of the i point, %/min; n is the number of experiment points we have chosen. Theoretical and experimental TG-DTG curves of four blends are shown in Fig. 4 and synergy evaluation indicators are then obtained in Table 3. From the results it may be found that mixture combustion is not a simple superposition of each single fuel burning, but a process with mutual influence. Table 3 shows the value of f varies with different mixing ratio of blend. There is mutual effect during the whole process, while the RMS of furfural slag in the volatile decomposition stage is lower than the last two stages. In the first two stages, the variation trends of these samples are almost equal. The RMS increases with enhancing furfural slag ratio. The value of f indicates that the two fuels promoted each other only for sample S4 in the process of combustion, whereas a contrary trend was observed for other samples. In the co-firing stage, the two components of S3 and S4 promoted each other, while mutual inhibition takes place only for S5 of fixed carbon combustion phase, the rest promoted each other. As a whole, appropriate increase of furfural slag in the mixture is advantageous to the combustion. As for the blends of 80% furfural residue, mutual inhibition occurs between the two components. The reason may be overmuch oxygen consumption needed by abundant organics content in furfural slag results in competition for oxidization of the two fuels. 5. The kinetic analysis of co-combustion 5.1. Coats-Redfern method analysis Kinetics parameters were calculated based on the basic equations [20,21] as follows:

ð9Þ

GðaÞ



T2

¼ ln


 AR E  bE RT

ð10Þ

where the equation of G(a) is:

"

ln

2

T ð1  nÞ 

ln

1  ð1  aÞ1n

 lnð1  aÞ T

2

#

¼ ln

 ¼ ln

   AR 2RT E 1  bE E 2:3RT

   AR 2RT E 1  bE E 2:3RT

ðn–1Þ

ðn ¼ 1Þ

ð11Þ

ð12Þ

Select n of different value individually as reaction series for each stage in the combustion. Set 1/T as X, make linear regression between the left of (11) or (12) and 1/T. Choose the reaction series when the fitting coefficient is the maximum for each sample, then calculate activation energy (E) and frequency Arrhenius factor (A) according to the slope and intercept. There are all two peaks appear in the DTG curves of furfural residue and semi-coke, so it is essential to calculate their kinetic parameters stage by stage. Table 4 shows the kinetics parameters of samples in different stages. According to a former kinetic research of biomass and coal [22], reaction series n is selected for different samples and stages to calculate the kinetic parameters in the combustion. As shown in Table 4, the activation energy E and frequency Arrhenius factor A are different in every stage of combustion. There are two reaction stages for whatever pure semi-coke or furfural residue in the process, in which two reaction energies and frequency factors were calculated. While different temperature regions are demonstrated between the two samples. Coincidentally, the temperature region takes place at first stage of semi-coke is nearly same with that at the second one of furfural residue. In the first reaction stage, the activation energy of furfural residue is larger than that of semi-coke, while the semi-coke’s is larger than furfural residue in the same temperature range. Compared to the semi-coke, the temperature range of furfural residue is lower, while the activation energy and activated molecule are larger than semi-coke. According to the proximate analyses, furfural residue

23

H. Qin et al. / Applied Thermal Engineering 120 (2017) 19–25

100 0.000

S2,cal S2 S2,cal S2

200

0

-0.002 400

600

800

600

-0.002

80

DTG

S4,cal S4

-0.006

S4,cal S4

400

600

800

-0.002 TG

40

DTG

-0.004

S5,cal S5 S5,cal S5

20

1000

1000

0.000

60

0

800

0

-0.006 -0.008

200

400

600

T/°C

T/°C

(c) S4

(d) S5

800

DTG/(%/min)

TG/%

100

DTG/(%/min)

0.000

-0.004

200

400

(b) S3

TG

0

200

0

(a) S2

60

20

-0.004

T/°C

80

40

40

-0.003

T/°C

100

-0.002

S3,cal S3 S3,cal S3

50

1000

DTG

TG

70 60

TG/%

60

TG/%

-0.001

DTG

TG

-0.001

80

DTG/(%/min)

80

0.000

90

DTG/(%/min)

TG/%

100

1000

Fig. 4. The theoretical and experimental TG-DTG curves of samples.

Table 3 Synergy evaluation parameters of different blending samples. Sample

First stage

S2 S3 S4 S5

Second stage

Third stage

f

RMS

f

RMS

f

RMS

0.904 0.974 1.031 0.979

0.385 0.242 0.075 0.048

0.951 1.147 1.005 0.987

0.381 0.308 0.230 0.293

1.126 1.045 1.048 0.992

0.490 0.676 0.873 1.368

Table 4 The kinetic parameters of the samples. Sample

b °C/min

Temperature °C

n

R

E kJ/mol

A min1

S1

40 40

376–655 655–810

1 1

0.995 0.993

72.522 27.278

2719.485 32.750

S2

40 40 40

270–388 388–645 645–794

0.5 3 4

0.991 0.999 0.993

84.118 23.040 25.541

1.590  106 12.436 2.661

S3

40 40 40

280–378 378–648 648–766

3 3 5

0.988 0.997 0.989

107.518 20.124 26.210

5.685  108 3.122 15.285

S4

40 40 40

282–390 390–670 670–757

3 3 5

0.994 0.996 0.992

125.319 22.357 23.628

4.386  1010 9.219 14.285

S5

40 40 40

255–374 374–676 676–732

1 2 5

0.992 0.991 0.995

105.625 21.813 27.459

7.955  108 10.694 40.531

S6

40 40

265–385 385–675

2/3 0.5

0.995 0.998

111.423 4.800

3.329  109 0.089

owns much more volatile than semi-coke, which mainly burns in the first stage. So this just explain the largest activation energy consisting in the first stage. As for blends, E becomes larger then smaller with the increase of semi-coke.

In the second stage of pure samples, the semi-coke’s activation energy is larger. Because of high ash contained, semi-coke is hard to be ignited although low volatile is left, so it is reasonable to illustrate the activation energy is larger than furfural residue.

24

H. Qin et al. / Applied Thermal Engineering 120 (2017) 19–25

peak. The releasing time of CO2 is obviously advanced after blending with furfural residue, in addition, the releasing amount becomes larger. With the increase of furfural residue, the peak of CO2 gas gradually become the largest when the furfural residue amount reaches 80%. At the same time, the releasing time gradually advanced with the increase of furfural residue, but the releasing amount is not easy to compare from the figure. In order to better analyze the influence of different mixing ratio on the releasing amount of CO and CO2, set the integral area as the releasing amount of CO and CO2 by integrating the curves in Fig. 6, ACO and ACO2 respectively account for the integral area of CO and CO2 curves, and tsCO2 illustrates the time when the CO2 gas began to release, teCO2 is the end of releasing time. The results are shown in Table 5. As is shown in the Table 5, the numerical value of ACO/ACO2 is small when the semi-coke burns alone, but when furfural slag is added in, the value becomes larger. And with the increase of the furfural slag ratio, the value of ACO/ACO2 decreases. At the same time, the releasing time shortens overall. When the ratio reaches 60%, the value is lower than the semi-coke.

While in the latter stages of blended samples (S2 to S5), activation energy becomes much smaller than the first stage. Because the heat from burning of volatile in furfural residue can help other combustible components burning in the later stages, in which the combustible substance are almost burned out, so it may be the reason for the activation energy in the two stages being much smaller than the first one. 6. FTIR analysis of flue gas Fig. 5 shows the combustion three-dimensional spectra of S4, the B, C, D, E curves are respectively corresponding to the absorption curves of 5, 10, 15, 20 min in co-combustion process. As is shown in Fig. 5, there are several components in flue gas during combustion, such as H2O, CH4, CO2, CO, SO2, NO2. Fig. 6 shows the releasing curves of CO2 and CO of the six samples. As shown in the figure, the releasing curve of S1 is a narrow

7. Conclusions

0.12 0.08 0.04 0.00

SO2 NO

H2O

2

-0.04

CO

This paper investigated co-combustion behavior of furfural slag and Longkou oil shale semi-coke. Some representative conclusions are summarized from this paper:

absorbance

CO2

(1) The combustion process of two fuels can be divided into two stages: the volatile stage and the fixed carbon stage. Furfural slag burns more sufficient than semi-coke. As for co-combustion process, three stages are contained: combustion of volatile in furfural slag, decomposition stage for volatile of semi-coke and combustion of fixed carbon of furfural slag, and burning out stage for the fixed carbon of semi-coke. (2) Too low or high heating rate shortens the reaction time, which is not beneficial to combustion. The combustion characteristic can be promoted by increasing mixing ratio of furfural residue.

B C D E

500 1000 1500 2000 2500 3000 3500

wavenumber/cm-1 Fig. 5. Infrared absorption three dimensional spectra in process of combustion of sample S4.

0.12 S1 S2 S3 S4 S5 S6

0.08 0.06 0.04

0.016

S1 S2 S3 S4 S5 S6

0.012

absorbance

absorbance

0.10

0.008 0.004

0.02

0.000

0.00 0

5

10

15

20

25

-0.004

30

0

5

10

15

20

25

t/min

t/min

(a) releasing curves of CO2

(b) releasing curves of CO

30

Fig. 6. Releasing curves of CO2 and CO of samples.

Table 5 Data comparisons of TG-FTIR at different proportions. Parameters

S1

S2

S3

S4

S5

S6

tsCO2/min teCO2/min teCO2  tsCO2 ACO2 ACO ACO/ACO2

11.0 26.2 15.2 0.3013 0.0384 0.1274

10.0 25.0 15.0 0.4056 0.0813 0.2004

10.0 24.8 14.8 0.5574 0.0965 0.1731

10.4 24.9 14.5 0.8559 0.1008 0.1178

8.0 22.2 14.2 0.8827 0.1006 0.1140

6.5 20.2 13.7 0.7008 0.0713 0.1017

H. Qin et al. / Applied Thermal Engineering 120 (2017) 19–25

(3) The co-combustion of furfural slag and semi-coke is not simple superposition but a complicated process, there exist synergies between the two samples in every combustion process. Synergy behaves difference with variation of the proportion of samples. (4) The kinetic study for the co-combustion process showed that blending with furfural residue can promote the combustion of semi-coke, the co-combustion is a complicated process. The activation energy is large in the first stage of all samples but much smaller in the latter stages because combustible substance are almost burned out after the first stage.

Acknowledgements Authors are grateful for the support from Jilin Provincial Department of Education Projects (2015, No. 251). This research was also supported by Jilin Science and Technology Development Plan Project (20150204012SF), and Changjiang Scholars Program (IRT13052). References [1] Zhongliang Yao, Xiaoqian Ma, Zihuan Wang, et al., Characteristics of co-combustion and kinetic study on hydrochar with oil shale: a thermogravimetric analysis, Appl. Therm. Eng. 110 (2017) 1420–1427. [2] Indrek Külaots, Jillian L. Goldfarb, Eric M. Suuberg, Characterization of Chinese, American and Estonian oil shale semicokes and their sorptive potential, Fuel 89 (11) (2010) 3300–3306. [3] M.F. Yardim, E. Ekinci, V. Minkova, et al., Formation of porous structure of semicokes from pyrolysis of Turkish coals in different atmospheres, Fuel 82 (4) (2003) 459–464. [4] Andres Trikkel, Rein Kuusik, Ants Martins, et al., Utilization of Estonian oil shale semicoke, Fuel Process. Technol. 89 (8) (2008) 756–763. [5] Wang Qing, Bai Jingru, Sun Baizhong, et al., Strategy of Huadian oil shale comprehensive utilization, Oil Shale 22 (3) (2005) 305–315.

25

[6] X.M. Jiang, X.X. Han, Z.G. Cui, New technology for the comprehensive utilization of Chinese oil shale resources, Energy 32 (5) (2007) 772–777. [7] I. Opik, N. Golubev, A. Kaidalov, et al., Current status of oil shale processing in solid heat carrier UTT (Galoter) retorts in Estonia, Oil Shale 18 (2) (2001) 99– 108. [8] H. Arro, A. Prikk, T. Pihu, et al., Utilization of semi-coke of Estonian shale oil industry, Oil Shale 19 (2) (2002) 117–125. [9] T. Kaljuvee, R. Kuusik, A. Trikkel, et al., Behavior of sulphur compounds at combustion of oil shale semicoke, Oil Shale 20 (2) (2003) 113–125. [10] Hong Qin, Yaokui Yue, Lidong Zhang, et al., Study on co-combustion kinetics of oil shale sludge and semicoke, Energy Fuels 30 (3) (2016) 2373–2384. [11] Ying Wang, Zhenyu Xu, Xu Song, et al., The preparation of low-cost adsorbent for heavy metal based on furfural residue, Adv. Manuf. Process. 32 (1) (2017) 87–92. [12] Hong-peng Liu, Wen Xue Liang, Qin Hong, et al., Thermal behavior of cocombustion of oil shale semi-coke with torrefied cornstalk, Appl. Therm. Eng. 109 (Part A) (2016) 413–422. [13] Hong-peng Liu, Wenxue Liang, Hong Qin, et al., Synergy in co-combustion of oil shale semi-coke with torrefied cornstalk, Appl. Therm. Eng. 109 (Part A) (2016) 653–662. [14] Pilar Gayan, Juan Adanez, F. Luis, de Diego, et al., Circulating fluidised bed cocombustion of coal and biomass, Fuel 83 (3) (2004) 277–286. [15] S.G. Sahua, P. Sarkara, N. Chakrabortyb, et al., Thermogravimetric assessment of combustion characteristics of blends of a coal with different biomass chars, Fuel Process. Technol. 91 (3) (2010) 369–378. [16] M. Varol, A.T. Atimtay, B. Bay, et al., Investigation of co-combustion characteristics of low quality lignite coals and biomass with thermogravimetric analysis, Thermochim. Acta 510 (1–2) (2010) 195–201. [17] Qing Wang, Xudong Wang, Hongpeng Liu, et al., Study of the combustion mechanism of oil shale semi-coke with rice straw based on Gaussian Multipeak fitting and peak-to-peak methods, Oil Shale 30 (2) (2013) 157–172. [18] Baizhong Sun, Theory and Experimental Investigation on Combustion Characteristics of Oil Shale and Semi-Coke Blends, North China Electric Power University, 2009. [19] Hongpeng Liu, Combustion Characteristics of Oil Shale Semi-Coke and Cornstalk Blends, Tianjin University, 2012. [20] U. Ozveren, Z.S. Ozdogan, Investigation of the slow pyrolysis kinetics of olive oil pomace using thermo-gravimetric analysis coupled with mass spectrometry, Biomass Bioenergy. 58 (2013) 168–179. [21] R. Hiltena, J.P. Vandenbrinkb, A.H. Patersonc, et al., Linking isoconversional pyrolysis kinetics to compositional characteristics for multiple Sorghum bicolor genotypes, Thermochim. Acta 577 (2014) 46–52. [22] Hongbo Lu, Huiyu Dai, Combustion Characteristics and Kinetic Analysis on Biomass and Coal Blends, Northeast Electric Power University, 2013.