Pyrolysis kinetics of biomass (herb residue) under isothermal condition in a micro fluidized bed

Energy Conversion and Management 93 (2015) 367–376

Contents lists available at ScienceDirect

Energy Conversion and Management journal homepage: www.elsevier.com/locate/enconman

Pyrolysis kinetics of biomass (herb residue) under isothermal condition in a micro fluidized bed Feiqiang Guo a, Yuping Dong b,c,⇑, Zhaochuan Lv b, Pengfei Fan b, Shuai Yang b, Lei Dong c a

School of Electric Power Engineering, China University of Mining and Technology, 221116 Xuzhou, PR China School of Mechanical Engineering, Shandong University, 250061 Jinan, PR China c Shandong Baichuan Tongchuang Energy Company Ltd., 250101 Jinan, PR China b

a r t i c l e

i n f o

Article history: Received 14 September 2014 Accepted 16 January 2015 Available online 3 February 2015 Keywords: Reaction kinetics Herb residue Pyrolysis Micro fluidized bed Isothermal

a b s t r a c t Herb residue is one of the most important industrial biomass in China in terms of availability and potential for use as a bioenergy resource. The kinetics of the thermal decomposition of this fuel in an inert atmosphere was evaluated using a micro fluidized bed. The isothermal differential analysis was applied for determination of kinetic parameters for the major gas components formation including reaction order, activation energy and pre-exponential factor. The temperature inside the micro fluidized bed was steady and the pyrolysis reaction of herb residue finished in around 10 s at 600–800 °C. The reaction time for complete releasing of individual gas components was shorter at higher temperature. Experimental results showed that under the conditions studied, the values of activation energy for generating H2, CO, CO2 and CH4 were 18.90, 12.05, 10.48 and 11.31 kJ/mol respectively, corresponding to the values of pre-exponential factor in the range of 0.88–1.38 s1. The results indicated that H2 was the most difficult to form due to the highest activation energy, while generating CO was the easiest corresponding to the lowest activation energy. Compared with TGA and other analysis approaches, the kinetic parameters obtain by the micro fluidized bed were significantly lower benefiting from its quick reaction features. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction An increasing demand for energy with the fast economic growth has revived interest in the development of renewable energy sources. Alternative fuels derived from biomass are regarded as a promising energy resource which can be used as substitutes for petroleum or natural gas [1,2]. Moreover, biomass resources, such as wood, agricultural products, and industrial residue, are CO2-neutral energy resources with low content in nitrogen and sulfur, and the corresponding fuels are therefore seen as clean energy compared with fossil fuels [3]. Herb residue studied in this text represents a kind of concentrated industrial biomass resource in China, and it was recently tested by the authors to produce fuel gas using a fluidized bed gasifier [4,5]. Pyrolysis which is seen as an important step in biomass gasification process is a fundamental thermochemical conversion process that can be used to convert biomass directly into fuels. Thus, pyrolysis has proved itself to be an important technique in biomass utilization and has attracted significant attention [6,7]. ⇑ Corresponding author at: School of Mechanical Engineering, Shandong University, 250061 Jinan, PR China. Tel.: +86 516 83592000. E-mail address: [email protected] (Y. Dong). http://dx.doi.org/10.1016/j.enconman.2015.01.042 0196-8904/Ó 2015 Elsevier Ltd. All rights reserved.

A thorough understanding of pyrolysis kinetics of biomass is a key component in the efficient design of biomass conversion processes. Thermogravimetric analysis (TGA) has been generally used to analyze biomass pyrolysis kinetics based on monitoring the mass variation of a spot sample under a specified heating program. The kinetics of thermal decomposition of a variety of biomass samples, such as wood [8], agricultural wastes [9,10], process residues [11–13] and organic wastes [14] were studied by non-isothermal thermogravimetry method under different atmosphere. Most of the values of activation energy ranged from 100 to 300 kJ/mol in different conversion range. Based on the thermogravimetric analysis method, many mathematical approaches have been developed to deduce the kinetic parameters [15], while these approaches are usually on the basis of preliminary assumption of a certain reaction order and reaction model [16,17]. In addition, TGA can be carried out in isothermal conditions as well, but there always a small mass loss before the pyrolysis reaching the given temperature, resulting in a certain error for the measuring method [18,19]. According to Yu et al. [20], there are still many drawbacks when testing biomass pyrolysis by TGA, such as inevitable gas diffusion and temperature deviation caused by the highly endothermic or exothermic reactions. Thus, the overall kinetics deduced

368

F. Guo et al. / Energy Conversion and Management 93 (2015) 367–376

from thermogravimetric analysis method can hardly reflect the process intrinsic characteristics of biomass pyrolysis. Fluidized bed reactor has been verified having advantages in fast on-line feed, minimized diffusion inhibition, quick heating for isothermal conditions and testing at arbitrary temperatures and in various gaseous atmospheres [20,21]. Kinetic parameters can be achieved using the fluidized bed reactor as well based on measuring the releasing of major gas components (CO, H2, CO2 and CH4) in the thermal degradation of the tested sample. Also, it can be pointed out that the values of reaction order, activation energy and pre-exponential factor for generating the gas components should be close to intrinsic characteristics of biomass pyrolysis as the interfacial diffusion limitations for biomass particles reactions is minimized. The aim of the present work was to investigate the characteristics and kinetics under isothermal condition in a micro fluidized bed reactor to describe the thermal decomposition process of herb residue. This study undertook analyses of product gas generating characteristics by a mass spectrometer, and the direct Arrhenius plot method was employed to derive kinetic parameters for forming four major gas components (H2, CO, CO2 and CH4), including the reaction order, activation energy and pre-exponential factor. Parallel tests using the same samples in TGA were conducted and the kinetic parameters were compared to further estimate performance of the micro fluidized bed. It is hoped that the obtained pyrolysis kinetics can give a better understanding of the biomass pyrolysis process for generating gas components. 2. Experimental 2.1. Raw material properties The biomass material used in this study was herb residue originated from Henan Wanxi Pharmaceutical Co., Ltd, located in the southwest of Henan province of China. The herb residue samples were air-dried, crushed and then sieved before the pyrolysis tests. The chemical analysis of the fuel simples is summarized in Table 1, showing that the volatile and carbon contents are around 74% and 15% respectively. The proximate analysis and ultimate analysis were carried out through the methods described in our previous report [22]. As suggested by Yu et al. [23], the influence of the intra-particle diffusion would be negligible for the fuels with sizes below 120 lm. Therefore, the fuel particles in 80–120 lm was used in this article by pulverizing and sieving. 2.2. Apparatus and procedure The pyrolysis experiments in this article were performed using a micro fluidized bed reactor system to obtain the pyrolysis gas data as a function of time under isothermal conditions. The schematic Table 1 Chemical composition of herb residue. Fuel Proximate analysis (wt.%, db) Volatile Fixed carbon Ash Ultimate analysis (wt.%, daf) Carbon Hydrogen Oxygen Nitrogen Sulfur Others ar.: as received basis; db: dry basis; daf: dry ash free basis.

Herb residue 74.30 14.95 10.76 42.40 6.20 47.39 1.86 0.15 2.00

diagram of the experimental apparatus is presented in Fig. 1 and its main components are: a fluidized bed reactor of 20 mm in diameter, an on-line pulse sample feeding system, a temperature and pressure sensor and a mass spectrometer (AMETEK, American) for on-line gas analyzing. The reactor designed in this article is 150 mm in height and consists of two porous plates to separate it into three zones. The lower zone is designed to realize uniform distribution of the fluidized gas, like the wind chamber of the fluidized bed. The middle zone of 40 mm in height is the zone between the two porous plates where pyrolysis reactions occur. The upper zone of 60 mm in height is the zone above the upper porous plate where fine particles escaping from the lower stage are caught to realize complete reaction. Two branches were designed at the middle of the reactor, one for fuel particles injection by compressed gas, the other one for temperature and pressure sensors installing. Quartz sand with mean diameter of 0.25 mm was used as the fluidization medium in the reactor and argon gas was used as carrier gas during the experiments. The temperature of the furnace, carrier gas flow rate and actions of pulse sample injection are all controlled by a computer. Meantime, the temperatures inside the reactor, pressures at the reactor inlet and outlet and the date of the product gas from the mass spectrometer are logged into the computer. Before each test, three grams of quartz sand was put into the lower layer and one gram was put into the upper layer. The minimum fluidization velocity (Umf) under ambient condition was first detected by measuring the pressures at the gas inlet and bottom of the upper porous plate. The value of Umf detected was 0.019 m/s, corresponding to the gas flow rate of 355 mL/min. In order to ensure the good fluidization of sand, the flow rate of the fluidizing gas was 500 mL/min, and thus the superficial velocity of the fluidizing gas was around 0.027 m/s under ambient condition. The reactor was heated by the furnace in fluidization state to the desired temperature (600–850 °C) for the herb residue pyrolysis. After that, about 10 mg fuel sample was injected into the reactor and the pyrolysis gas (CO, H2, CO2 and CH4) was measured by the mass spectrometer continuously. To assure the reliability of the test results, each test is repeated for three times. 2.3. Kinetic methods 2.3.1. Isothermal kinetic method A diverse set of possible mechanisms of biomass decomposition have been elucidated in previous literature reviews. In this work, the kinetic study was to describe the gas components generation under isothermal condition, and the mechanisms should belong to the multi-component mechanism which was considered originally proposed in [24,25]. This mechanism was available for predicting the formation rates and the yields of reaction products or solid- and gas-phase intermediates. The mechanism was verified in [26,27] to study the biomass pyrolysis kinetics based on series reactions taking into account the presence of several zones in the isothermal weight loss curves. Based on this, integral data concerning generation of the product gas under isothermal condition was measured by a mass spectrometer in this study. The corresponding final kinetic parameter values are estimated so as to get the best fit with the combustible gas yield by herb residue pyrolysis. In this study, quick heating for isothermal reaction conditions can be realized by the micro fluidized bed reactor. Referring to Yu et al. [20] and Manyà et al. [28], the kinetic parameters of biomass pyrolysis in isothermal process were calculated generally using the shrinking core model. During the experiments, the variation of the concentrations of the product gases with time under different reaction temperatures can be measured by the mass spectrometer, and then the activation energy can be gained using the model expressed as a function of the gases accumulated production.

369

F. Guo et al. / Energy Conversion and Management 93 (2015) 367–376

Exhaust

Mass spectrometer

Flow sensor

Needle vavle

Gas filter

Pressure sensor Electric furnace Temperature and pressure sensor

Electromagnetic valve

Display interface Herb residue sample

Mass flowmeter

Quartz tubular reactor

Gas valve

Mass flowmeter

Compressed gas

Fig. 1. Schematic diagram of the micro fluidized bed reactor.

mt ¼

12 

mt0 ¼

Rt

ui  qv dt 22:4

0

12 

ð1Þ

R t0

ui  qv dt 22:4

0

ð2Þ

Rt

x¼

ui  qv dt mt t ¼ R t0e  100% mt0 ui  qv dt

ð3Þ

t0

where mt and mt0 are the conversions of i (H2, CO, CO2 and CH4) at time t and the reaction end (t0); ui represents the concentration of gas species i, and qv denotes the flow rate of the product gas from the reactor.

dx ¼ kðTÞf ðxÞ dt

ð4Þ

where k(T) is the reaction rate constant defined by the Arrhenius equation and it is a constant in isothermal process, which means it can be separated from f(x). f(x) is the model function equation in the differential format and is expressed as follows:

f ðxÞ ¼ ð1  xÞn

ð5Þ

dx ¼ kðTÞð1  xÞn dt

ð6Þ

where n is the reaction order. Taking natural logarithms of the each   sides from Eq. (6), a linear equation is achieved correlating ln dx dt and ln(1  x), as shown in Eq. (7). The values of ln (k(T)) and n which correspond to the intercept and slope of the linear equation respectively can be calculated at different temperatures.

    dx ¼ ln kðTÞð1  xÞn ¼ ln ðkðTÞÞ þ n lnð1  xÞ ln dt

logarithms of the both sides of Eq. (8). Since both the terms of E/R and ln A are constants, a straight line should be able to be obtained by the linear regression of ln (k(T)) versus 1/T. The activation energy (E) and the pre-exponential factor (A) can be obtained from the slope and the intercept of the straight line, respectively.

  E kðTÞ ¼ A exp  RT

ð8Þ

   E E ¼ lnðkðTÞÞ ¼ ln A exp  þ ln A RT RT

ð9Þ

where E is the activation energy, kJ/mol; A denotes to the pre-exponential factor, 1/s; T refers to the temperature, K; R represents the gas constant, 8.314 J/(mol K). 2.3.2. Non-isothermal kinetic method The Flynn–Wall–Ozawa (FWO) [30] method and Coats–Redfern method [31] are the two most common model-free methods of determination of kinetic parameters under non-isothermal conditions by thermogravimetric analysis. In these methods, the fraction of conversion a was defined as:

a¼

w0  wt w0  w1

where w0, wt and w1 are sample masses at the beginning, at time t and at the end of reaction, respectively. In the pyrolysis process of biomass, the global kinetics reaction can be described as:

da ¼ kðTÞf ðaÞ dt GðaÞ ¼

Z

kðTÞdt ¼

ð7Þ

According to El-Sayed and Mostafa [29], direct Arrhenius plot method was used in this article to calculate activation energy and pre-exponential factor. Eq. (9) was obtained by taking natural

ð10Þ

ð11Þ Z

  Z a E da dt ¼ A exp  RT 0 f ðaÞ

ð12Þ

where G(a) is the model function equations in the integral formats. In programmed heating method, the temperature T is described as:

T ¼ T 0 þ bt

ð13Þ

370

F. Guo et al. / Energy Conversion and Management 93 (2015) 367–376

where b is the linear heating rate and it is a constant. The approximation used by the FWO method is expressed in Eq. (14). Substituting of Eq. (14) into integral Eq. (12), and then taking the logarithm of both sides leads to the expression for the FWO method of Eq. (15):

Z

T

T0

   E E dT ¼ 0:0048e1:0516 exp  RT R

ln b ¼ ln



ð14Þ

 AE E  2:315  0:4567 RGðaÞ RT

ð15Þ

  GðaÞ T2

¼ ln

  AR  E=ðRTÞ bE

ð16Þ

The mechanism function G(a) implicates the reaction mechanism, and twenty empirical mechanism functions were set up and used in this method [9,20]. Eq. (16) implies that the value of  ln GðT 2aÞ is subject to a linear correlation with 1/T, and the slop  . Thus, the activation and intercept correspond to E/R and ln AR bE energy and pre-exponential factor can be deduced from the slope

(a)

o

Temperature ( C)

804

3.1.1. Temperature and pressure drop profiles The micro fluidized bed reactor used in the study realizes quick heating for isothermal conditions and the injected fuel particles can pyrolyze continuously at a stable temperature. The temperature in the main reaction zone (the zone between two porous plates) of the reactor was measured every 1 s to monitor the temperature variation during each test. The temperature profile for a typical experiment at T = 800 °C are presented in Fig. 2(a). It is apparent that the temperature profile maintains almost steady state in the 10 min with the variation of measuring data within 1 °C, which implies that the temperature inside the micro fluidized bed was hardly influenced by pyrolysis reactions. The pressure drop mainly comes from the resistance of the two porous plates and the fluidizing sand and fuel particles. Fig. 2(b) shows the changing trend of pressure drop obtained in the same ten minutes at T = 800 °C and it can be seen that its value remained fairly stable with variation within 350 Pa.

802 800 798 796 15700

(b) Pressure drop (Pa)

3. Results and discussion 3.1. Isothermal experiments in the micro fluidized bed

For a given conversion, a, the points of ln(b) versus 1/T can be fitted to a straight line, allowing the activation energy to be determined from the slope of the correlation line. In terms of the Coats–Redfern method, the employed empirical approximation is the following expression:

ln

and intercept of the straight line. The activation energy values obtained by Coats–Redfern method can be compared with that by FWO method to judge the accuracy of the adopted mechanism function G(a). The corresponding G(a) by which the activation energy obtained was the closest to the result given by the FWO equation can be selected to estimate the pre-exponential factor.

15600

15500

15400 0

2

4

6

8

10

Time/(min) Fig. 2. Temperature (a) and pressure drop (b) profiles inside the reactor at T = 800 °C.

3.1.2. Gas components releasing characteristics In this study, pyrolysis analysis of fuel particles is conducted at six temperatures of 600 °C, 650 °C, 700 °C, 750 °C, 800 °C and 850 °C. The releasing characteristics of the pyrolysis gas components measured by the mass spectrometer of two typical experiments are plotted in Fig. 3, showing that the releasing pattern of biomass differs a little at different temperature. The releasing order of gas species is CO2, CH4, CO and H2 during herb residue pyrolysis at 600 °C. The intensity of H2 is not obvious at this temperature, while a noticeable change of the releasing order of H2 was observed when the temperature was raised to 850 °C. The difference among the releasing of the gas species was not significant, which indicated that the difficulty of the gas species releasing differs little in herb residue pyrolysis at temperatures from 600 °C to 800 °C. In addition, it can be seen from Fig. 3 that the pyrolysis reaction in the fluidized bed finished in around 10 s, and this completion time was much shorter than the experimental data from TGA and other methods [32,8], indicating that higher reaction rate 3.00E-007

o

1.40E-007

T=600 C

CH4 1.20E-007

CH4

2.50E-007

H2

H2 CO

CO

2.00E-007

Intensity (%)

1.00E-007

Intensity (%)

CO2

o

T=850 C

CO2

8.00E-008 6.00E-008

1.50E-007 1.00E-007

4.00E-008

5.00E-008

2.00E-008 0.00E+000 128

0.00E+000 130

132

134

Time (s)

136

138

140

128

130

132

134

Time (s)

Fig. 3. The releasing characteristics of the major gas species during pyrolysis.

136

138

140

371

F. Guo et al. / Energy Conversion and Management 93 (2015) 367–376

100

100

H2

CO 80

80

60 o

600 C o 650 C o 700 C o 750 C o 800 C o 850 C

40

20

Conversion x (%)

Conversion x (%)

o

600 C o 650 C o 700 C o 750 C o 800 C o 850 C

60

40

20

0

0 0

2

4

6

8

10

12

0

2

4

8

10

12

100

100

CH4

CO2 80

60

o

600 C o 650 C o 700 C o 750 C o 800 C o 850 C

40

20

0

COnversion x (%)

80

Conversion x (%)

6

Time (s)

Time (s)

o

600 C o 650 C o 700 C o 750 C o 800 C o 850 C

60

40

20

0 0

2

4

6

8

10

12

Time (s)

0

2

4

6

8

10

12

Time (s)

Fig. 4. Relative conversion of pyrolysis gas components varying with time at different temperatures.

was achieved using the micro fluidized bed reactor. Compared with fixed bed reactor, the stable temperature of the medium inside the reactor realizes better heating transfer characteristics which can lower the temperature gradients that may exist in the particles [33]. Thus, a more equal and stable decomposition process or releasing rate occurs at the given temperatures in the micro fluidized bed reactor. 3.1.3. Pyrolysis gas components conversion versus reaction time Setting the time when fuel particles were injected as the beginning of pyrolysis reaction, the relative conversion x of the four gas components (H2, CO, CO2 and CH4) was calculated according to Eqs. (1)–(3), and the largest gas yield at the end of the reaction corresponds to the conversion of 100%. The calculated results of conversion x of the four gas components versus time at different reaction temperatures are shown in Fig. 4. It can be seen that the time for complete reaction(x = 100%) of the gas components decreased with elevating temperature, indicating that higher temperature promotes the pyrolysis reaction. The degree of increment in the reaction rate of H2 which can be seen from the slope of the curves varies obviously with an increase in temperature, particularly in the range from 600 °C to 700 °C, which indicates that the temperature significantly affected the formation of H2, while the influence of temperature on the other three gas components was not that obvious as H2. The variation law of the reaction rate is similar with Yu et al. [23], and the values of activation energy are the fundamental reason influencing the releasing characteristics of these four major gas components. 3.1.4. Variation of reaction order of gas components Since four data sets were achieved for conversion x of the  major gas components versus reaction time in Fig. 4, ln dx versus dt ln(1  x) at five reaction temperatures for each individual gas component was calculated according to Eqs. (4)–(7) and plotted in

Fig. 5. It is obvious that the entire curve can be divided into three parts at a given temperature and this can be explained by three reaction stages during the pyrolysis reaction after the sample particles were injected into the micro fluidized bed. The first stage is defined as the fast heating stage of the pyrolysis process which is shown as the zone before the colored zone in Fig. 5. The injected sample particles mixed with the high temperature fluidized medium (quartz sand) and reached the desired temperature by heat transfer instantly. The fuel particles began to pyrolysis during this fast heating stage and the conversion of gas components was unsteady,   which in turn resulted in the irregularity of correlation of ln dx and ln(1  x). It was generally believed that the reactions dt occurred mainly on the surface of fuel particles [34,35], and the changing trend was almost the same at different temperature. The relative range of this stage is different for different gas component, denoting that the releasing of these gas components was different at the same temperature. The steady temperature of the fuel particles was achieved after the fast heating stage, and then the pyrolysis reaction reached the   major decomposition stage. In this stage, the curves of ln dx verdt sus ln(1  x) of each gas component showed good linear relationship, indicating that the pyrolysis reaction was mainly controlled by chemical kinetics. This stage reflects the reaction process of sample particles at the given temperature and therefore can represent the intrinsic reaction of the pyrolysis process. At the end of stage two, curves in Fig. 5 appeared inflection and tended to be vertical which represented that the pyrolysis reaction of sample particles was nearly completed, and it was seen as the ending stage of the pyrolysis process. Based on above analysis, the major decomposition stage is the main phase of biomass pyrolysis and the kinetic parameters are calculated corresponding to this stage. The regression lines of this stage are calculated according to Eq. (7) for the individual gas component at a given temperature and plotted in Fig. 6. The intercept

372

F. Guo et al. / Energy Conversion and Management 93 (2015) 367–376 o

600 C o 650 C o 700 C o 750 C o 800 C o 850 C

-1 -2 -3

-3 -4

-5 -6 -7

CO

-5 -6 -7

-8

-8

-9 -10

o

600 C o 650 C o 700 C o 750 C o 800 C o 850 C

-2

ln (dx/dt)

ln (dx/dt)

-4

-1

H2

-9

-8

-7

-6

-5

-4

-3

-2

-1

0

-10

1

-8

-6

ln (1-x)

-1

o

ln (dx/dt)

-4

CO2

600 C o 650 C o 700 C o 750 C o 800 C o 850 C

-2

-4

ln (dx/dt)

-3

-2

0

-4

-2

0

o

600 C o 650 C o 700 C o 750 C o 800 C o 850 C

-2

-4

ln (1-x)

-5 -6

CH4

-6

-8

-7 -8

-10

-9 -10

-8

-6

-4

-2

-10

0

-8

-6

ln (1-x)

ln (1-x) Fig. 5. Curves of ln

dx dt

versus ln(1  x) for individual components at different temperatures.

-1.0

-1

CO

-1.5

H2

-2

-2.0 ln (dx/dt)

o

600 C o 650 C o 700 C o 750 C o 800 C o 850 C

-3.0 -3.5 -4.0

o

600 C o 650 C o 700 C o 750 C o 800 C o 850 C

-4

-5

-6

-4.5 -7

-2.5

-2.0

-1.5

-1.0

-0.5

-5

-4

-3

-2

-1

ln (1-x)

ln (1-x) -1

-1

CO2 -2

-2 -3

o

ln (dx/dt)

ln (dx/dt)

ln (dx/dt)

-3

-2.5

o

600 C o 650 C o 700 C o 750 C o 800 C o 850 C

-4

-5

-3

-2

-1

ln (1-x) Fig. 6. Regression lines of ln

-3

-4

-6 -4

600 C o 650 C o 700 C o 750 C o 800 C o 850 C

0

-5 -3.5

-3.0

-2.5

-2.0

-1.5

ln (1-x) dx dt

versus ln(1  x) for individual components at different temperatures.

-1.0

-0.5

373

F. Guo et al. / Energy Conversion and Management 93 (2015) 367–376

and slope of each line correspond to rate constant ln(k(T)) and reaction order n under different reaction temperatures. The calculated ln(k(T)) and n for the four gas components at different temperature are listed in Table 2, showing that the R-squared was above 0.96 for all the curves. According to Eq. (5), the reaction order (n) gives different formulations of the term f(x), suggesting that the reaction mechanism varies with different reaction order. It can be seen that the reaction Table 2 ln(k(T)) and n of the four gas components at different temperatures. 600 °C

650 °C

700 °C

750 °C

800 °C

850 °C

H2 ln(k(T)) n R2 SD

1.16 1.10 0.96 0.22

0.99 1.26 0.96 0.18

0.84 1.23 0.99 0.09

0.80 1.28 0.99 0.10

0.66 1.36 0.99 0.09

0.56 1.38 0.98 0.12

CO ln(k(T)) n R2 SD

0.57 0.97 1.00 0.08

0.48 1.10 1.00 0.06

0.42 1.15 1.00 0.08

0.31 1.08 1.00 0.10

0.25 1.27 0.99 0.20

0.21 1.31 1.00 0.06

CO2 ln(k(T)) n R2 SD

0.83 0.97 0.99 0.14

0.62 1.13 0.99 0.10

0.61 1.16 0.99 0.14

0.60 1.18 0.99 0.13

0.56 1.23 0.97 0.25

0.47 1.27 0.98 0.18

CH4 ln(k(T)) n R2 SD

0.58 0.88 0.99 0.09

0.51 0.92 0.97 0.16

0.39 0.98 0.96 0.22

0.38 1.06 0.97 0.21

0.34 1.17 0.98 0.13

0.20 1.25 0.96 0.18

order values of H2, CO, CO2 and CH4 were between 1.1–1.38, 0.97–1.31, 0.97–1.27 and 0.88–1.25 respectively at the given temperatures. The formation of gas components was determined by different reactions, leading to the difference in the formation mechanisms of these four gas components during biomass pyrolysis. Furthermore, pyrolysis process of biomass consists of a very complex set of reactions involving the formation of radicals, determining the generating characteristics of gas components. The pyrolysis mechanisms of biomass pyrolysis were therefore subjected to different pyrolysis conditions which depended on the temperature in this study. As a result, the reaction orders of each individual gas component varied with the variation of temperature as well, as shown in Table 2, implying that the generating mechanisms of the gas components changed at different reaction temperature. The reaction order gained under isothermal conditions by the micro fluidized bed reactor was higher compared with non-isothermal differential thermal gravimetric analysis [36,29], which meant that kinetic parameters may vary obviously by different analytical method due to the complicated physical and chemical process of biomass pyrolysis. 3.1.5. The results of the kinetic parameters After obtaining the values of ln(k(T)) for each gas component at different temperatures in Table 2, the linear regression method was applied to correlate ln(k(T)) and 1000/T according to Eq. (9), and the results were shown in Fig. 7. ln A and E/R can be achieved as the intercept and slope of the regression lines, and then the activation energy (E) and pre-exponential factor (A) together with R-squared were obtained for each individual gas component and listed in Table 3. It can be seen that the values of R-squared are higher than 0.91 for all the gas components.

CO

-0.2

H2

-0.6

-0.3

ln k (T)

lnk (T)

-0.8

ln k(T)=-2.27*(1000/T)+1.46 2 R =0.98

-0.4

lnk(T)=-1.45*(1000/T)+1.09 2 R =0.99

-1.0

-0.5

-1.2 0.85

0.90

0.95

1.00

1.05

1.10

-0.6 0.85

1.15

0.90

0.95

1000/T

1.00

1.05

1.10

1.15

1000/T

-0.2 -0.5

CO2

CH4 -0.3

-0.7

ln k (T)

ln k (T)

-0.6

ln k(T)=-1.26*(1000/T)-0.64 2 R =0.91

-0.4

-0.5

ln k(T)=-1.36*(1000/T)+0.97 2 R =0.92

-0.8

0.85

0.90

0.95

1.00

1000/T

1.05

1.10

1.15

-0.6 0.85

0.90

0.95

1.00

1.05

1000/T

Fig. 7. Linear fitting of ln(k(T)) and 1000/T for each gas component at different temperatures.

1.10

1.15

374

F. Guo et al. / Energy Conversion and Management 93 (2015) 367–376

As is well known, activation energy is the obstacle that must be overcome before a chemical reaction is generated and it represents the difficulty for forming the gas components during biomass pyrolysis. For the four major gas components, H2 has the largest activation energy with a value of 18.90 kJ/mol, indicating that releasing H2 was more difficult in pyrolysis reaction. The releasing characteristics of these gas components were also reflected in Fig. 3, which showing that the intensity of H2 measured was not obvious at lower temperature (600 °C). In comparison, it is much easier to release CO2 owing to the lowest activation energy of 10.48 kJ/mol. 3.2. TGA experiments In order to compare the parameters obtained in the micro fluidized bed, parallel tests in TGA (TGA–SDTA 851e) were also conducted using the same herb residue under non-isothermal conditions. Experiments were carried out at heating rates of 5, 10, 20 and 30 °C/min and 10 mg the same herb residue sample was performed each time. Pure N2 was used as the carrier gas at a steady flow rate of 150 mL/min during the experiments. The TG and DTG curves of herb residue at different heating rates from 5 to 30 K/min are shown in Fig. 8. The entire herb residue pyrolysis process can be divided into three stages, occurring moisture evaporation, main devolatilization and continuous slight devolatilization respectively. It was apparent that heating rate is one of the most important parameters influencing the pyrolysis characteristics, and different trends in the rates of weight losses took place when heating rate changed between 5 and 30 K/min due to the heterogeneous structure of biomass [30]. In order to determine the parameters of the herb residue pyrolysis, the FWO method (Eq. (15)) was first used to estimate the activation energy on the conversion fraction. Conversion fractions varying from 20% to 80% were employed at different heating rates, and the regression lines are illustrated in Fig. 9. The values of Table 3 Kinetic parameters obtained for generating the four major gas components. Formed gas

E (kJ/mol)

A (1/s)

R2

H2 CO CO2 CH4

18.90 12.05 10.48 11.31

4.30 2.96 1.90 2.63

0.98 0.99 0.91 0.92

0.0

dm/dt (%/min)

100

90

70

1.6

-0.2

1.4 -0.3

-0.4

1.2

60

30

600

800

1000

Temperature (K)

o

40

After the kinetic parameters are obtained, kinetic parameters obtained in this work, by both the micro fluidized bed and TGA, were compared with other results reported in literatures using different raw materials and under different operating conditions, as shown in Table 4. Kinetic parameters studies under isothermal condition were tried by different authors using fluidized bed, tubular reactor and fixed bed as reactor. The results reported in this study can be compared with results of kinetic analysis conducted by Yu et al. [23], using the similar fluidized bed for the pyrolysis of beer lees. A tubular reactor was designed and set up by Lv et al. [37] for the fast pyrolysis of biomass to simulate the fast heating rate in the fluidized bed. The heating rate of the samples exceeded 1000 °C/min and the resulting E and A thus differed a little from this study, with E varying in 15–25 kJ/mol and A in 0– 5 s1. The activation energy values obtained by Encinar et al. [38] for generating H2, CH4 and CO using a fixed bed reactor was obviously higher due to more inhibitions from the gas diffusion in the fixed bed compared with the fluidized bed reactors. Many pyrolysis tests have been performed by TGA under nonthermal condition as a result of its heating via programmed heating method, and the heating rate of the samples is usually between 5 and 30 °C/min which is much lower than the rates realized in the isothermal measurements. The values of activation energy and pre-exponential factor were thus different significantly compared

-0.1

400

50

3.3. Kinetic parameters comparison

5 C/min o 10 C/min o 20 C/min o 30 C/min

lg ( ß)

Mass (%)

80

activation energy at different conversion was obtained from slop of the regression lines, and the resulting data were tabulated in Fig. 9 via digital values. It can be suggested the pyrolysis of herb residue proceeds with varied reaction mechanisms as a result of the different activation energy values from 106.91 to 127.22 kJ/mol. Based on the activation energy values obtained by FWO approach, Coats–Redfern method was then used to estimated different pairs of the Arrhenius parameters, A and E, with the twenty different models G(a) which are assumed to describe the proposed reaction mechanism and can be seen in the literatures[9,20]. Substitution of different models G(a) into equation Eq. (16) and the obtained activation energy values were compared with that obtained by FWO approach to examine the accuracy of the various function models. The activation energies determined by the mechanism function models G(a) = a2 and G(a) = (1  a)ln(1  a) + a are acceptably close to the values given by the FWO equation, with values from 112 to 130 kJ/mol at different heating rate. The preexponential factor was in turn determined to be about 106 1/s based on these two function models according to the Coats– Redfern equation.

E/(kJ/mol) 106.91

30%

110.52

40%

113.86

50%

116.5

60%

120.16

70%

123.69

80%

R > 0.99 20% 30% 40% 50% 60% 70% 80%

1.0

0.8

600

800

1000

Temperature (K) Fig. 8. TG and DTG curves of herb residue for heating rates in 5–30 °C/min.

127.22 2

0.6 400

20%

1.3

1.4

1.5

1.6

1.7

1.8

1.9

2.0

2.1

1000/T Fig. 9. Regression lines of ln b and 1/T for different conversions.

375

F. Guo et al. / Energy Conversion and Management 93 (2015) 367–376 Table 4 Comparison of the measured kinetic parameters with literature report by different authors. Condition

Reactor

Fuel

Heating rate

Pyrolysis product

Temperature (°C)

E (kJ/ mol)

A (1/s)

Source

Isothermal

Fluidized bed

Herb residue

Instantaneous

600–850

Beer lees

Instantaneous

Tubular reactor Fixed bed

Pine sawdust Cellulose Cynara cardunculus L.

Instantaneous Instantaneous

H2 CH4 CO

300–800

12.05 18.90 10.48 11.31 12.36 28.25 10.91 12.49 24.45 15.12 92.00 48.00 16.00

2.96 4.30 1.90 2.63 1.60 7.92 1.32 1.47 4.81 0.71 – – –

This study

Fluidized bed

CO H2 CO2 CH4 CO H2 CO2 CH4 Gas mixture

TGA TA–MS

Herb residue Coffee hulls

5–30 °C/min 5 °C/min

25–900 25–900

TGA

10 °C/min

200–1400

105–130 80.44 97.52 105.36 137.52 293–461

TGA

Five agricultural wastes Pine

Weight loss CO H2 CO2 CH4 Weight loss

10–20 °C/min

Weight loss

100–600

160–270

106 4.64 5.96 7.34 10.70 8.2  105– 1.4  106 1011–1016

TGAS

Agricultural residues

2–10 °C/min

Weight loss

200–600

120–250

1011–1018

Nonisothermal

with the values obtained by the micro fluidized bed in this study. The resulting E achieved by Huang et al. [39] was in the range of 80–140 kJ/mol for generating gases through differential route by using the thermal analysis–mass spectrometry (TA–MS), and the results were similar to that values obtained in this study by TGA method, whereas the pre-exponential factors were in the same order with those from the fluidized bed. The other reported E and A for the TGA tests obtained by detecting the weight loss of the biomass samples in Table 4 were even higher, with E in 160– 500 kJ/mol and A in orders of 105–1018 s1. According to the comparison in Table 4, it can be suggested that the differences that appear in results of kinetic studies is subject to the reactor and heating rate. Heating of samples can be considered instantaneous under isothermal approach, leading to lower activation energy E and the frequency factor than programmed heating method. The fluidized bed reactor having the advantage in minimized diffusion inhibition results in the lower E and A compared with fixed bed under similar isothermal condition. Thus, the measured kinetic parameters of biomass pyrolysis in the micro fluidized bed should be closer to the intrinsic chemical kinetics. Furthermore, the biomass fuel undergoes quick heating after it was fed into the reactor when a fluidized bed was used for biomass conversion during the real applications. The obtained pyrolysis kinetics using the micro fluidized under isothermal condition can give a better description of the biomass pyrolysis process for generating gas components.

4. Conclusion In this study, the detailed kinetic analysis of the isothermal pyrolysis of herb residue was performed on the basis of the pyrolysis gas releasing characteristics for individual gas components using a micro fluidized bed. The following conclusions were made. A steady temperature was achieved using the micro fluidized bed with the variation of measuring date within 1 °C at a given temperature of 800 °C. The pyrolysis reaction of herb residue in the micro fluidized bed finished in around 10 s at reaction temperatures from 600 to 800 °C, and the complete reaction time was obviously shorter than that in the literature reports, indicating that the pyrolysis reactions in the micro fluidized bed were obviously

500–900

700

Yu [23]

Lv [37] Encinar [38]

This study Huang [39]

Wilson [40] Soria-Verdugo [41] Sonobe [42]

quicker than the other methods. Higher temperature promotes the pyrolysis reactions, resulting in a shorter time for the complete reaction of the gas components. The values of activation energy for generating H2, CO, CO2 and CH4 are 18.90 kJ/mol, 12.05 kJ/mol, 10.48 kJ/mol and 11.31 kJ/mol respectively, corresponding to the values of pre-exponential factor in the range of 0.88–1.38 s1. The values of activation energy represents the difficulty for the four major gas components formation, and the results showing that H2 was the most difficult and CO2 was the easiest to generate in biomass pyrolysis. The results were much lower compared with the values obtained by TGA and other methods, and it can be guessed that the kinetic parameters achieved in this articles was closer to the intrinsic reaction process of biomass pyrolysis as a result of the better heat and mass transfer characteristics of the micro fluidized bed. Acknowledgements This work was financially supported by the National Natural Science Foundation of China (51406226), National Key Foundation for Exploring Scientific Instrument (2011YQ12003906) and China Postdoctoral Science Foundation (2014M551693). We gratefully acknowledge the valuable cooperation with Prof. Xu Guangwen and the members of his laboratory. References [1] White JE, Catallo WJ, Legendre BL. Biomass pyrolysis kinetics: a comparative critical review with relevant agricultural residue case studies. J Anal Appl Pyrol 2011;91:1–33. [2] Bildirici ME. Economic growth and biomass energy. Biomass Bioenergy 2013;50:19–24. [3] Masnadi MS, Habibi R, Kopyscinski J, Hill JM, Bi X, Lim CJ. Characterization and co-pyrolysis kinetics of biomass and fossil fuels. Fuel 2014;117:1204–14. [4] Guo FQ, Dong YP, Dong L, Jing YZ. An innovative example of herb residues recycling by gasification in a fluidized bed. Waste Manage 2013;33:825–32. [5] Guo FQ, Dong YP, Zhang TH, Dong L, Guo CW, Rao ZH. Experimental study on herb residue gasification in an air-blown circulating fluidized bed gasifier. Ind Eng Chem Res 2014;53:13264–73. [6] Tchapda AH, Pisupati SV. A review of thermal co-conversion of coal and biomass/waste. Energies 2014;7:1098–148. [7] Mettler MS, Vlachos DG, Dauenhauer PJ. Top ten fundamental challenges of biomass pyrolysis for biofuels. J Cover: Energy Environ Sci 2012;5:7797–809. [8] Slopiecka K, Bartocci P, Fantozzi F. Thermogravimetric analysis and kinetic study of poplar wood pyrolysis. Appl Energy 2012;97:491–7.

376

F. Guo et al. / Energy Conversion and Management 93 (2015) 367–376

[9] Mishra G, Bhaskar T. Non isothermal model free kinetics for pyrolysis of rice straw. Bioresour Technol 2014;169:614–21. [10] Aboyade AO, Hugo TJ, Carrier M, Meyer EL, Stahl R, Knoetze JH, et al. Nonisothermal kinetic analysis of the devolatilization of corn cobs and sugar cane bagasse in an inert atmosphere. Thermochim Acta 2011;517:81–9. [11] Jankovic´ B. The pyrolysis of coffee paper cup waste samples using nonisothermal thermo-analytical techniques. The use of combined kinetic and statistical analysis in the interpretation of mechanistic features of the process. Energy Convers Manage 2014;85:33–49. [12] Guerrero MRB, da Silva Paula MM, Zaragoza MM, Gutiérrez JS, Velderrain VG, Ortiza AL, et al. Thermogravimetric study on the pyrolysis kinetics of apple pomace as waste biomass. Int J Hydrogen Energy 2014;39:16619–27. [13] Du YY, Jiang XG, Lv GJ, Ma XJ, Jin YQ, Wang F, et al. Thermal behavior and kinetics of bio-ferment residue/coal blends during co-pyrolysis. Energy Convers Manage 2014;88:459–63. [14] Jayaraman K, Gökalp I. Pyrolysis, combustion and gasification characteristics of miscanthus and sewage sludge. Energy Convers Manage 2015;89:83–91. [15] Conesa JA, Domene A. Biomasses pyrolysis and combustion kinetics through nth order parallel reactions. Thermochim Acta 2011;523:176–81. [16] Soria-Verdugo A, Garcia-Gutierrez LM, Blanco-Cano L, Garcia-Hernando N, Ruiz-Rivas U. Evaluating the accuracy of the distributed activation energy model for biomass devolatilization curves obtained at high heating rates. Energy Convers Manage 2014;86:1045–9. [17] Ranzi E, Cuoci A, Faravelli T, Frassoldati A, Frassoldati A, Migliavacca G, et al. Chemical kinetics of biomass pyrolysis. Energy Fuels 2008;22:4292–300. [18] Senneca O. Kinetics of pyrolysis, combustion and gasification of three biomass fuels. Fuel Process Technol 2007;88:87–97. [19] Guo J, Lua AC. Kinetic study on pyrolysis of extracted oil palm fiber. Isothermal and non-isothermal conditions. J Therm Anal Calorim 2000;59:763–74. [20] Yu J, Zeng X, Zhang JW, Zhong M, Zhang GY, Wang Y, et al. Isothermal differential characteristics of gas–solid reaction in micro-fluidized bed reactor. Fuel 2013;103:29–36. [21] Yu J, Liu Z, Zhu JH, Dong L, Xu GW. Kinetics and mechanism of solid reactions in a micro fluidized bed reactor. AIChE J 2010;56:2905–12. [22] Guo FQ, Dong YP, Guo CL. Effect of design and operating parameters on the gasification process of biomass in a downdraft fixed bed: an experimental study. Int J Hydrogen Energy 2014;39:5625–33. [23] Yu J, Yao CB, Zeng X, Geng S, Li D, Wang Y, et al. Biomass pyrolysis in a microfluidized bed reactor: characterization and kinetics. Chem Eng J 2011;168:839–47. [24] Ward SM, Braslaw J. Experimental weight loss kinetics of wood pyrolysis under vacuum. Combust Flame 1985;61:261–9.

[25] Di Blasi C. Modeling chemical and physical processes of wood and biomass pyrolysis. Prog Energy Combust Sci 2008;34:47–90. [26] Di Blasi C, Branca C. The effects of water leaching on the isothermal degradation kinetics of straw. Ind Eng Chem Res 2000;39:2169–74. [27] Branca C, Di Blasi C. Kinetics of the isothermal degradation of wood in the temperature range 528–708 K. J Anal Appl Pyrol 2003;67:207–19. [28] Manyà JJ, Arauzo J. An alternative kinetic approach to describe the isothermal pyrolysis of micro-particles of sugar cane bagasse. Chem Eng J 2008;139:549–61. [29] El-Sayed SA, Mostafa ME. Pyrolysis characteristics and kinetic parameters determination of biomass fuel powders by differential thermal gravimetric analysis (TGA/DTG). Energy Convers Manage 2014;85:165–72. [30] Ozawa T. A new method of analyzing thermogravimetric data. Bull Chem Soc Jpn 1965;38:1881–6. [31] Coats AW, Redfern JP. Kinetic parameters from thermogravimetric data. Nature 1964;201:68–9. [32] Ounas A, Aboulkas A, El harfi K, Bacaoui A, Yaacoubi A. Pyrolysis of olive residue and sugar cane bagasse: non-isothermal thermogravimetric kinetic analysis. Bioresour Technol 2011;102:11234–8. [33] Milosavljevic I, Suuberg EM. Cellulose thermal decomposition kinetics: global mass loss kinetics. Ind Eng Chem Res 1995;34:1081–91. [34] Yu J, Yue JR, Liu ZE, Li D, Xu GW, Zhu JH, et al. Kinetics and mechanism of solid reactions in a micro fluidized bed reactor. AIChE J 2010;56:2905–12. [35] Zeng X, Wang F, Wang YG, Li A, Yu J, Xu GW. Characterization of char gasification in a micro fluidized bed reaction analyzer. Energy Fuels 2014;28:1838–45. [36] Gai C, Dong YP, Zhang TH. The kinetic analysis of the pyrolysis of agricultural residue under non-isothermal conditions. Bioresour Technol 2013;127:298–305. [37] Lv PM, Chang J, Wang TJ, Wu CZ. A kinetic study on biomass fast catalytic pyrolysis. Energy Fuels 2004;18:1865–9. [38] Encinar JM, González JF, González J. Fixed-bed pyrolysis of Cynara cardunculus L. Product yields and compositions. Fuel Process Technol 2000;68:209–22. [39] Huang YF, Chiueh PT, Kuan WH, Lo SL. Pyrolysis kinetics of biomass from product information. Appl Energy 2013;110:1–8. [40] Wilson L, Yang WH, Blasiak W, John GR, Mhilu CF. Thermal characterization of tropical biomass feedstocks. Energy Convers Manage 2011;52:191–8. [41] Soria-Verdugo A, Garcia-Hernando N, Garcia-Gutierrez LM, Ruiz-Rivas U. Analysis of biomass and sewage sludge devolatilization using the distributed activation energy model. Energy Convers Manage 2013;65:239–44. [42] Sonobe T, Worasuwannarak N. Kinetic analyses of biomass pyrolysis using the distributed activation energy model. Fuel 2008;87:414–21.