Dependency of the combustion behavior of energy grass and three other types of biomass upon lignocellulosic composition

The combustion characteristics of four kinds of biomass fuels (energy grass, sawdust, corn cob, and walnut shell) are investigated in this article. All the samples are heated from room temperature to 800°C at multiple heating rates of 10, 20, and 30°C/min. The effect of hemicellulose, cellulose, and lignin components on the pyrolysis and combustion processes of energy grass is explored by comparison to those of the other three types of biomass. The hemicellulose and cellulose content of samples could improve the devolatilization performance during biomass combustion. Furthermore, the comprehensive combustion index suggested herein indicates that the combustion performance of energy grass or walnut shell is limited by their high ash content or their low ratio of cellulose to lignin. Kinetic parameters are obtained by combining the isoconversional method (OFW and KAS models) and the method of master‐plots. The apparent activation energy of the devolatilization stage is higher than that of the char oxidization stage, which is mainly influenced by the lignocellulosic composition. © 2017 American Institute of Chemical Engineers Environ Prog, 37: 815–823, 2018


INTRODUCTION
Biomass is an only renewable carbon source, gaining particular attention in energy generation for its neutral CO 2 conversion, low NO X and SO 2 emissions, and high content of volatiles [1]. Generally, biomass can be converted into heat or liquid fuel by three thermal conversion processes namely pyrolysis, gasification, and combustion [2]. Direct combustion is one of the most extensively employed technologies for commercial or industrial utilization of biomass, and is responsible for about 95-97% of the world's bioenergy produced [3][4][5]. Thermal analysis is regarded as a useful and reliable tool to determine the thermal properties and kinetics of biomass during combustion [6,7]. The kinetic parameters are essential for the design, control, and optimization of industrial equipment [8]. Iso-conversional methods, such as Ozawa-Flynn-Wall (OFW) or Kissinger-Akahira-Sunose (KAS), are commonly used in numerous kinetic studies on biomass combustion without prior knowledge of reaction mechanisms [9][10][11].
Biomass is generally considered as an organic fuel derived from plants, including wood, agricultural wastes, herbaceous crops, and short-rotation energy crops [3,12]. Up to now, most studies have focused on the combustion of agricultural or woody biomass using thermogravimetric analysis (TGA), such as wood [13,14], pine sawdust [15], capsicum stalks [16], straw [17], sunflower [18], corn cob and stover [19], grape marc (i.e., skin, seed, and stalk) [20]. In comparison to other traditional biomass fuels, energy crops are a promising alternative that are cost-effective, and do not generally require particularly fertile soil good soil or high levels of fertilizer and pesticide application [21][22][23][24]. Energy grass (A. donax) is a kind of energy crop artificially cultivated and has following advantages: high calorific value (22.76 MJ/kg), high production (over 15 years after planting as a perennial plant), and lower land requirement (growing rapidly in sandy, saline-alkali, or industrial waste land) [25]. Because of the economical and environmental advantages of the combustion of energy grass, it is pertinent to investigate its thermal properties. However, previous studies on energy grass have been limited and have tended to focus on the combustion characteristics.
The chemical composition plays crucial role in the thermal conversion of biomass. Cellulose, a polysaccharide with the generic formula C 6 H 10 O 5 , is the major component of biomass cell walls; hemicellulose is another cell wall component represented by the generic formula C 5 H 8 O 4 ; and lignin is the aromatic compound built from three highly cross-linked units [26][27][28][29]. So far, few investigations focused on the combustion characteristics related to the biomass components. Gani and Naurse [30] found that the cellulose and lignin content were important to evaluate the pyrolysis characteristics. Cheng et al. [31] and Kai et al. [32] used artificial biomass components (microcrystalline cellulose, xylan, and lignin) to explore the contribution of lignocellulosic components to the thermal process. They found that hemicellulose combusted easily and the weight loss of lignin covered a broad temperature range.
The aim of this work is to investigate the combustion characteristics of energy grass in comparison to those of sawdust, corn cob, and walnut shell. The combustion characteristics are explored as a function of the hemicellulose, cellulose, and lignin contents. The kinetic parameters of the four types of biomass are calculated by methods of isoconversion master-plots.

Sample Preparation
The samples used in this study are energy grass (EG), sawdust (SD), corn cob (CC), and walnut shell (WS). EG, that is, A. donax, is selected from Changping district in Beijing, and SD, CC, and WS are from Heilongjiang Province in China. The four kinds of biomass are ground and sieved to pass an aperture of 200 lm to reduce the resistance to mass and heat transfer. All the samples are dried in an oven set at 1058C for 2.0 h. Their proximate and ultimate analyses are conducted based on the ASTM methods. Furthermore, the contents of hemicellulose, cellulose, and lignin are determined by employing the method stipulated by the National Renewable Energy Laboratory (NREL) [33]. The lignocellulosic compositions are measured three times and the average values are obtained in this study.

Experiments
The characteristics of pyrolysis and combustion are evaluated using thermogravimetric analysis. Each sample of mass 6 6 0.2 mg is weighed and placed into an alumina crucible. The pyrolysis process is conducted under nitrogen. To remove the air and ensure an inert environment, the samples are flushed by flowing nitrogen, and then heated from room temperature to 8008C at a heating rate of 208C/min with a flow rate of 80 mL/min. The combustion process is instead carried out in oxidative atmosphere (20% oxygen and 80% nitrogen) at multiple heating rates of 10, 20, and 308C/min. All the experiments are conducted three times to assure the repeatability.

Definition of Characteristic Parameters
To clearly describe the pyrolysis and combustion processes, the following parameters are defined by thermogravimetic (TG) curves [21,34]: 1. DTG max : the maximum rate of weight loss during thermal decomposition, indicating combustibility of the sample. DTG max1 and DTG max2 are introduced if there are two peaks in TG profile. 2. T P : the temperature which corresponds to the maximum degradation rate, which is an indicator of reactivity.
A lower T P indicates better ignition performance during the combustion process. 3. T i : the initial temperature of the thermal conversion process (e.g., the ignition temperature for combustion). As defined by Biagini et al. [34], the initial temperature is determined by the tangential method and is derived from the earliest maximum degradation rate (DTG max1 ). As shown in Fig. 1, a vertical line is drawn through point A (DTG max1 ) intersecting with TG curve at point B. Subsequently an intersection between a tangent through point B and an extended TG initial horizontal line is made at point C, whose corresponding temperature is considered to be the beginning of thermal conversion process. 4. T e : the end temperature of thermal degradation (i.e., the burnout temperature for combustion), which is also determined through determination of the tangent, but derived from the latest degradation rate (DTG max2 ). T e represents the completion of thermal degradation. Therefore, the vertical line is drawn through point D in Fig.1, and the tangent line through point E is intersected with the extended TG final horizontal line. 5. D i : the ignition index [18], which is a measurement of ignition performance of combustion and described as Eq. (1): 6. S n : the comprehensive combustion index [16], which is used to evaluate general performance of combustion and can be defined as Eq. (2):

Kinetic Methods
As mentioned above, the OFW and KAS models are employed to determine activation energy of the combustion of biomass, and the master-plot method is used to determine the reaction mechanism herein. The degree of conversion of biomass can be written as Eq. (3): where m 0 and m f represent the initial and final masses of the sample, respectively, while m t is the mass at any time.
The fundamental rate equation is generally expressed as Eq. (4): where t is the time, T is the temperature, and f(a) is the reaction function. k(T) denotes the temperature-dependent rate constant, which is defined in terms of the Arrhenius equation: where A is the pre-exponential factor, E is the activation energy, and R is the universal gas constant. At a constant heating rate b 5 dT/dt, Eqs. (4) and (5) can be transformed and combined to give the following expression: Upon integration of Eq. (6), the following is obtained: The OFW model is described by Eq. (8): The KAS model is expressed as Eq. (9): Based on Eqs. (8) and (9), the apparent activation energy (E) at a given conversion rate can be obtained from linear correlations of ln(b) and ln(b/T 2 ) versus1/T [9]. The heating rates of 10, 20, and 308C/min are selected to calculate the value E herein. The method of master-plots is regarded as an effective way to determine the reaction mechanism and reaction order n [35,36]. The integrated function G(a) can be approximated to Eq. (10) as T 0 being zero has little impact on the righthand side of Eq. (7).
where P(u) (u 5 E/RT) is the temperature integral. As P(u) does not have an analytical solution, the approximate value can be obtained from Doyle [37]: The generalized master plots method is suitable for different heating schedules [38]. The kinetic triplets (i.e., kinetic models), A and E are constant for a single-step process [39].
Substituting the value a 5 0.5 into Eq. (10) yields The following equation is easily derived from Eqs. (10) and (12): where G(a) is constant for a given kinetic model, G(a)/ G(0.5) denotes the theoretical value, while P(u)/P(u 0.5 ) is inferred from experimental data. Therefore, the appropriate mechanism function is obtained from Eq. (13). Different expressions of common reaction mechanisms are listed in Table 1 [35].

Biomass Characterization
The results of proximate, ultimate, and compositional analyses of biomass samples are shown in Table 2. It is essential to give a summary of chemical analysis due to the correlation with thermal performance of biomass [40]. It is obvious that energy grass (EG) has large amount of ash, whilst sawdust (SD), corncob (CC), and walnut (WS) have small content of ash. Lower ash content may be more beneficial to combustion process. Besides, the sequence of volatile content is CC > SD > WS > EG. Table 2 also shows C, H, O, N, and S content of biomass, herein the O content is calculated by difference. The content of N and S of EG is a little higher than those of the other three biomasses. In addition, some differences on lignocellulosic components among the biomass samples SD, CC, and WS are demonstrated with high content of cellulose, hemicellulose, and lignin, respectively.

Pyrolysis Characteristics
The pyrolysis process is the initial step of biomass combustion. As shown in Fig. 2a, EG (energy grass) undergoes the smallest weight loss, which is in agreement with the results of proximate analysis (Table 2). There are two peaks in the first stage of pyrolysis except for SD (sawdust) in Fig.  2b. One local maxima peak is observed at about 3008C and the main peak at around 3508C, which represents the thermal decomposition of hemicellulose and cellulose, respectively. This is consistent with the fact that hemicellulose decompose over the temperature range of 150-3508C, while cellulose  [41]. It is worth noting that the first peak of CC (corn cob) is larger than that of WS (walnut shell) or EG mainly due to the higher content of hemicellulose. There is no shoulder peak for SD, indicating that the decomposition of hemicellulose is overlapped by that of cellulose and lignin. The pyrolysis characteristic parameters of the four types of biomass are summarized in Table 3. It is noted that the pyrolysis of EG starts and ends at 278.98C and 406.58C, respectively, which are lower than those of the three other types of biomass. The maximum weight loss rate of EG or WS is lower while that of SD or CC is higher, indicating the poorer reactivity of EG or WS. This observation is related to their volatiles content. Figure 3 shows the combustion profiles of the four types of biomass as a function of temperature at heating rates of 10, 20, and 308C/min. It is evident that the combustion process can be divided into three stages responding to their DTG profiles. The first stage is dehydration from room Table 2. Proximate, ultimate, and compositional analyses of biomass samples.

Mechanisms Symbol f(a) G(a)
Order of reaction First-order F 1 1 2 a 2ln(1 2 a) Second-order F 2 (1 2 a) 2 (1 2 a) 21    , which moves to a slightly lower temperature versus pyrolysis in an inert atmosphere. The maximum devolatilization rate of combustion is a little higher than that of pyrolysis. This is because mild heterogeneous oxidation promotes the pyrolytic abstraction of volatile matter [39]. In addition, the main peak appears around 3008C and the shoulder peak of EG is not distinct during the combustion process, which is probably because the hemicellulose peak has merged with the cellulose peak. This phenomenon can be explained by the alkali ions causing a reduction in the decomposition temperature of cellulose [42]. The last stage is attributed to the oxidation of char in the temperature range of approximately 380$6008C. A series of combustion parameters of the four types of biomass, including the ignition index and comprehensive combustion index, are listed in Table 4. It is noted that the ignition temperature of EG is lower than that of WS or SD. This might be attributed to the fact that the higher lignin content in woody plants delayed the ignition [16]. Moreover, in comparison to EG, CC ignites earlier and exhibits the highest reaction rate in the devolatilization stage, as the ignition performance is improved by higher cellulose and  hemicellulose content [43]. At a given heating rate, the reactivity (maximum weight loss rate, DTG max ) sequence of devolatilization stage is CC > SD > WS > EG, which is consistent with the hemicellulose and cellulose content in fuels. The ignition index (D i ) of EG shows the worst ignition performance. Furthermore, the comprehensive combustion index (S n ) is introduced to evaluate the combustion performance. As shown in Fig. 4, S n increases with elevated heating rate for each biomass. The higher S n of CC and SD indicate better combustion performance, while S n of WS and EG is lower.
Comparing the difference of chemical composition among the four biomasses, the lowest S n of EG may be attributed to the largest ash content. Both WS and CC exhibits a lower ash content than does EG. However, WS has a poorer combustion performance, which probably results from lower ratio of cellulose to lignin content (0.39) than that of CC (2.1). Therefore, to evaluate the combustion performance of biomass, both ash composition and the ratio of cellulose to lignin content should be taken into consideration. Figure 3 presents the effect of heating rate on combustion performance. It can be seen that both the temperature ranges of the devolatilization and char oxidation stages are becoming wider with an increase in heating rate. As the heating rate increases from 108C/min to 308C/min, the maximum weight loss rates (DTG max1 ) of EG, SD, CC, and WS increase from 6.1%/min, 13%/min, 11.5%/min and 9.5%/min to 18.6%/min, 34%/min, 41.6%/min, and 23.2%/min, respectively. Combining with Table 4, the ignition temperature (T i ) and end temperature (T e ) also move to higher values as the heating rate increases for all the samples, which results from a particle gradient temperature due to limited thermal conductivity. Similar results were also obtained in previous studies [34,44].

Isoconversional Method
OFW and KAS models are used to analyze the kinetics due to their ability to give a relatively accurate value of activation energy that is independent of the reaction mechanism [9]. Two stages of thermal degradation are investigated assuming single-step reactions for the solid-state process. Taking energy grass (EG) as an example, Figure 5 shows that the plot of ln b and ln (b/T 2 ) versus1/T with respect of conversion rate both give an approximately linear relationship.
The activation energy of EG can be obtained by determining the slope of the fitted lines. The other three types of biomass (SD, CC, and WS) are also divided into two stages, and activation energies are acquired using the above method. Almost all the samples present an excellent linear dependency such that correlation coefficients are >0.99. In this study, the average values of E are calculated by OFW and KAS models as they are appropriate to assess the kinetic parameters of thermal process. As listed in Table 5, the average E value of devolatilization stage is higher than that of char oxidation stage. With respect to the lignocellulosic composition of biomass, this might arise from the fact that lignin, whose decomposition rate is lower than cellulose and hemicellulose components, is condensed to char [45]. There is little difference among activation energies (E) of the four types of biomass in devolatilization stage, but for the char oxidation stage, samples of EG and CC show higher E than do SD and WS. This might be attributed to the fact that cellulose with the highest E has an obvious effect on global kinetics and the E value of hemicellulose was higher than that of lignin [46]. SD and WS contain large amounts of lignin, which explains their lower E in the char oxidation stage.

The Method of Master-Plots
The average activation energy (E) is calculated by the isoconversional methods, and then P(u) is obtained using Eq. (11). Figure 6a shows the P(u)/P(u 0.5 ) versus a of the devolatilization and char oxidation stages (EG) at different heating rates. It is revealed that the master-plots of P(u)/P(u 0.5 ) against a are in very close agreement for the different heating rates. Similar profiles are obtained for the other three samples, indicating that the kinetics of biomass thermal degradation can be approximated as a single-step reaction model. In addition, the theoretical master plots of G(a)/ G(0.5) versus a are compared with the experimental curves P(u)/P(u 0.5 ) in Fig. 6b. It is found that the EG-oxidation stage matches the theoretical master plot of the first-order model F 1 , while the experimental master plot of EGdevolatilization stage lies between F 2 and F 3 models. Furthermore, the F 2.2 model is the most appropriate to describe the devolatilization stage of energy grass by plotting more F n models, as shown in Figure 6c. Similarly, the kinetic models of the other types of biomass (SD, CC, and WS) are determined by comparing the experimental and theoretical materplots (Fig. 6d). The pre-exponential factor (A) is estimated by the intercept of the fitted straight lines (Eqs. (8) and (9))  based upon the determined E and G(a). The average A value and corresponding kinetic models are summarized in Table  6. It is evident that A in the devolatilization stage is much higher than that of char oxidation stage for all samples in this study.

CONCLUSIONS
In this work, the combustion characteristics of energy grass are investigated in comparison with sawdust, corn cob, and walnut shell using TGA. The kinetic parameters of combustion are obtained by combining the iso-conversional and mater-plots methods. The main results can be summarized as follows: Compared with the pyrolysis under nitrogen, the devolatilization stage of combustion moves to a slightly lower temperature, and the maximum devolatilization rate is higher under air. Lignocellulosic composition plays an important role in the ignition performance. Ignition index increases with increasing hemicellulose and cellulose content of biomass.
The comprehensive combustion index can be used to describe the combustion performance of biomass. Poor combustion behavior of energy grass is largely due to high ash content, but for walnut shell, this can be attributed to the low ratio of cellulose to lignin content. The apparent activation energy of the devolatilization stage is higher than that of the char oxidization stage for the four types of biomass studied. Compared with sawdust and walnut shell, energy grass and corn cob show higher activation energy. The reaction mechanism of devolatilization and char oxidation stages can be expressed by the master-plot method for the four types of biomass.