The energy consumption and pellets’ characteristics in the co-pelletization of oil cake and sawdust

Abstract

Improvement of the co-pelletization of biomass (cedarwood and camphorwood) and oil cake was carried out in the present work. The characteristics of the raw materials were determined by chemical and physical analysis. The co-pelletization process was studied by quantifying the energy consumption of compaction and extrusion processes. The physical properties of the pellets, including moisture adsorption, pellet energy density and hardness, were also determined. The blending of biomass with oil cake would increase the density of the pellets, and thus reduce their costs during production, transport and storage. The energy consumption of compaction decreased upon increasing the castor bean cake content, while the energy consumption of extrusion and the pellets’ energy density showed the opposite tendency. Moreover, the maximum compression force and hardness increased with the increase in the blend ratio of CAS (castor bean cake). This thermal analysis suggests that adding CAS can enhance char combustion and can lead to greater reactivity during char combustion. 10–20% of castor bean cake by mass would be a suitable proportion in co-pelletization with biomass, in terms of energy consumption and the pellet properties.

---

1. Introduction

Biomass pellets have been considered as a high-quality fuel suitable for many industrial and residential applications.^1,2 The densities of pellets can be 4–10 times higher than those of sawdust or straw, which is helpful in reducing the costs of transport and storage.^1,3 However, the increasing price of raw materials, the limited adaptability of the pellet unit and the unsatisfactory quality of pellets hinder the further development of the pelletization industry.⁴ Those challenges are aggravated in areas accompanied with other biomass industries, such as biomass-power generation, paper making, and artificial boards etc. In East Asia, due to the competition for feedstock among several biomass industries, the price of raw materials has climbed to 64–108 USD/t, which is higher by far compared to that in Europe and North America. Major industrial enterprises in East Asia have to use various kinds of biomass of different shapes, including hard wood, soft wood and straw, so the dies and rollers in the pellet unit have to be fed with different kinds/shapes of biomass, consequently shortening their operating life.⁵ Therefore, much research has concentrated on the co-pelletization of mixtures of materials.^6–9

Blends of raw materials could contribute to the production of higher-quality pellets for combustion.⁶ The biomass additives used in blending are usually bentonite, lignosulfonate, modified cellulose binders, proteins, etc.^10,11 The additives form a bridge, film or matrix to improve the pellet quality or minimize the variation in pellet quality.^10–12 However, the additives are more expensive than conventional biomass, which may affect the economic feasibility of energy from biomass.^11,13 Therefore, suitable additives should be chosen from solid wastes or other byproducts.

Much research on pelletizing additives has focused on the identification of combustion behaviour and mechanical durability.^14,15 However, few works have been conducted to find a type of practical additive, considering the optimization of energy consumption, pellet density, pellet hardness and moisture uptake.

In this study, castor bean cake, a kind of residue in the oil expression process, was utilized as an additive in the process of co-pelletization. On account of the diversity of biomass in terms of both physical and chemical properties, experiments were carried out to evaluate the effects of biomass type (softwood and hardwood) and the blending proportions of oil cake on the energy consumption of pelletization and the properties of the pellets.

2. Materials and methods

2.1 Materials

Woody sawdust of cedarwood (CED) and camphorwood (CAM) used in this work were selected as raw materials, representing softwood and hardwood respectively. The selected oil cake was castor bean cake (CAS), which was obtained from the castor bean expression process in a pilot study of Hunan Academy of Forestry (Changsha, China). Proximate analysis was carried out according to the Standard Practice for the Proximate Analysis of Coal and Coke (GB/T212-2001) using a Vario EL III elemental analyzer (Germany). Three replicates were measured for each raw material. The characteristics of those raw materials are presented in Table 1.

Table 1 Chemical analysis of raw materials (wt% dry basis)

Analysis		Cedarwood	Camphorwood	Castor bean cake
a Elemental analysis (dried and ash-free base).
Elemental analysis^a (%)	C	48.95	48.18	44.21
	H	5.91	6.09	5.93
	O	44.49	45.03	41.38
	N	0.65	0.70	7.92
	S	—	—	0.56
Fiber analysis (%)	Hemicelluloses	12.05	20.82	28.98
	Cellulose	36.22	38.87	20.73
	Lignin	27.61	24.40	23.75
Proximate analysis (%)	Moisture	7.63	6.67	7.42
	Volatile matter	74.49	79.02	64.62
	Fixed carbon	16.68	12.53	19.82
	Ash	1.20	1.78	8.14
Protein analysis (%)		—	—	6.24

---

The raw materials were dried by air (40 °C, 48 h), ground by a pulverizer and then screened into fractions of particle size below 0.45 mm. The average sawdust moisture content was 5% after drying. The moisture content of the feedstock was adjusted to 10% by adding deionized water, and then the feedstock was maintained in plastic bottles at 4 °C for 48 h to mix them uniformly.

2.2 Pelletizing process and characteristics analysis

As shown in Fig. 1, ADWD-10, a piston-cylinder unit installed with a punch and die set, was used for pelletization. A die set of 7 mm inside diameter and 120 mm length with a piston of 6.8 mm in diameter and 80 mm in length was installed to make a single pellet. The die was encircled with a heating tap connected to a temperature controller to preheat the cylinder, simulating the heat generated in industrial pelletization.¹⁶ A removable plug was used to plug the end of the die. The transducer was applied to measure the compression force and displacement during pelletization and the data was recorded by a computer.

Fig. 1 The pictures of the pellet die setup: the ADWD-10 universal press, the schematic diagram of the apparatus and the pellets.

CAS was added to CED and CAM with proportions of 0%, 5%, 10%, 15%, 20%, and 25%, respectively. Prior to pelletization, the die set was heated up to 110 °C. The prepared material was loaded in the ADWD-10 at the same dosage, with the top hole of the die set filled with approximately 0.8 g of the prepared sample. The loaded sample was compressed at a rate of 2 mm min⁻¹ until the desired pressure was achieved. A maximum force of 4000 N was applied and held for 30 s. After that, the pellet was extruded from the die set and cooled down immediately. Seven pellets were made for each sample, these were measured in length and weight and kept in plastic bottles for two weeks.

2.3 Moisture absorption

The pellets for each blended sample were dried at 105 °C for 48 h before the moisture absorption test. These tests were carried out in a humidity chamber (30 °C and 90% relative humidity). The weight of the pellets was determined with intervals of 20 min in the first five hours. After then, the final moisture absorption was measured after 2 days. The kinetics of moisture absorption is presented using the ASABE formulation as follows:^14,17

M − M_e/M_i − M_e = e^−kt, (1)

where M is the instantaneous moisture content (decimal, dry basis), M_e is the equilibrium moisture content (decimal, dry basis), and M_i is the initial moisture content (decimal, dry basis). The coefficient k is an absorption constant (min⁻¹), and t is the exposure time (min).

2.4 Mechanical hardness

Analysis of the pellet mechanical hardness was carried out by the ADWD-10 pressure unit. The stored pellets were horizontally placed on a flat surface under the compression bar. The pressure force was loaded at the centre of the pellet. The compression bar was run at a certain rate of 2 mm min⁻¹ until the pellet was broken, in which the collapse force reached the maximum pressure. The collapse force and corresponding distances were recorded by a computer. Three replicates were conducted for each sample. The Meyer hardness was calculated using the following equation:¹⁸

MH = F/[π(Dh − h²)] (2)

where MH is the Meyer hardness (N mm⁻²); F is the maximum pressure (N); D is the pellet diameter (mm) and h is the corresponding indentation depth of maximum pressure (mm).

2.5 Thermogravimetric analysis (TG)

Thermogravimetric analysis was performed using a TG-60 analyzer (JAP) for evaluation of the combustion characteristics of materials. In this study, the temperature program consisted of two steps: the first, from room temperature to 105 °C at a 10 °C min⁻¹ heating rate, and then rising up to 800 °C at a 15 °C min⁻¹ heating rate. Air was used as the purging gas at a flow rate of 100 mL min⁻¹. Masses between 3 and 5 mg for each sample were used to evaluate the combustion behaviour, in order to reduce the interference of mass and heat on the test.

3. Results and discussion

3.1 Energy consumption

It was reported by previous research that the processing of hardwood sawdust consumed more energy than softwood sawdust.¹⁹ However, as shown in Fig. 2a and b, the opposite results were observed in the present work. Error bars in the figures represent the standard deviations of seven replicates, which indicate the heterogeneous character of wood powder in terms of mechanical properties. The energy consumptions were 19.6 kJ kg⁻¹ and 15.6 kJ kg⁻¹ for CED and CAM, respectively. There are two explanations for these opposite results, in terms of micro and macro viewpoints. On the micro level, the hemicellulose content in CAM is higher than that in CED (Table 1). The hemicellulose can act as a viscoelastic interface in the lignocellulose structure due to the larger quantity of hydrogen boding in hemicelluloses.¹⁷ CAM, with a higher hemicellulose content consequently has stronger tenacity, with a higher modulus of elasticity in the elastic region.¹⁷ The higher tenacity of sawdust contributed to lower energy consumption during compaction.¹ As a result, a lower amount of energy is thus required during the compaction process for CAM than CED. Meanwhile, on the macro level, a higher energy consumption means greater consumption due to friction during pelletization. In Fig. 3a1–3, more intertwined fibers were identified on the surface of CED compared with the flat surface of CAM. The friction between the CED particles was hence increased due to the large contact area in the intertwined fibers, resulting in a higher compaction energy consumption for the CED runs.

Fig. 2 The energy consumption of compression (a) and extrusion (b) as a function of content of CAS.

Fig. 3 SEM micrographs of raw materials and pellets. (a1) Raw CED; (a2) raw CAM; (a3) raw CAS; (b1) CED; (b2) 90% CED + 10% CAS; (b3) 75% CED + 25% CAS; (c1) CAM; (c2) 90% CAM + 10% CAS; (c3) 75% CAM + 25% CAS.

In Fig. 2a, lower compaction energy consumptions were observed as the CAS content increased for CED, while no significant change was observed for CAM. This can be due to the different fiber structures between CED and CAM.²⁰ The lubricating function of CAS was utilized in the intertwined fiber structure of CED, while there was no clear evidence for this in the flat structure of CAM. The CAS was squeezed into the gaps and voids between the biomass particles, and consequently coated the intertwined fiber surface of CED with residual oil, proteins and starch which can act as lubricants, resulting in the decline of friction.⁸ Meanwhile, lower extrusion energy consumptions were obtained in all CED–CAS runs, while the extrusion energy consumption increased with the increase in CAS content in the CAM–CAS runs (shown in Fig. 2b). The denatured proteins and gelatinized starch in CAS immobilized in the fiber matrix of CED contributed to the enhancement of hydrogen bonding and “solid bridges” among particles, leading to the improvement in the pellets’ plasticity, reducing the extrusion energy consumption.¹ Moreover, higher extrusion energy consumptions were obtained with the increase in CAS content in the CAM–CAS runs (Fig. 2b). This may be due to the viscidity of proteins and starch in CAS during pelletization.¹⁷ In the compaction process, several denatured proteins and gelatinized starch in CAS were softened, and subsequently were squeezed into the space between the pellet and the inner wall of die, as the flat surfaces of CAM particles lack a fiber matrix. The protein and starch were plasticized and adhered to the surface of the inner wall of the die, resulting in higher friction during extrusion. Furthermore, the higher ash content in CAS was another reason for the higher friction (Table 1). Therefore, the extrusion energy consumption in the CAM–CAS run increased with the increment of the CAS content.

Furthermore, as shown in Fig. 2, the variations in compaction and extrusion energy consumption between CED–CAS and CAM–CAS were “narrowed” with the CAS content between 10–20%. In industrial utilization, the limited feedstock species were more suitable for specific pelletization equipment due to the variations of physical and chemical characteristics of biomass. The properties of mixing materials can be made uniform by the addition of CAS, resulting in homogeneous energy consumption and an increase in equipment service life.²¹

3.2 Pellet density

In Table 2, the diameter and length resiliency for all pellets are approximately equal, indicating that the pure and blended pellets had good dimensional stability during storage. However, these results are distinguished with previous research.²² In previous work, the resiliency mainly occurred in the length direction with a decrease from 25% to 10% occurring with the increase in content of waste wrapping paper during storage.²² The interlocking bonds in pellets were strengthened by the additives, contributing to the increasing resistance to the disruptive force and the reduction of length resiliency.^9,22 In the present study, the additive of CAS can also strengthen the interlocking bonds, similar to the function of waste wrapping paper. Those differences may be due to the higher compression force used in this study, leading to stronger interlocking bonds in the more stable matrix of CAS and biomass resisting the disruptive force.^6,9,16

Table 2 Mass and high heating values of raw materials

Samples	Size^a (mm)		Resiliency^b (mm)		Mass density^a (kg m^−3)		Mass density^b (kg m^−3)		HHV (MJ kg^−1)	Energy density (GJ m^−3)
	Diameter	Length	Diameter	Length	Average	SD	Average	SD
a Measured after removal.b Measured after 20 days.
CED
100% CED	7.45	15.48	0.10	0.72	1086.24	3.86	1095.89	4.81	19.57	21.45
95% CED + 5% CAS	7.44	15.44	0.18	0.97	1088.04	7.74	1088.97	16.21	19.58	21.32
90% CED + 10% CAS	7.43	15.25	0.10	0.89	1106.39	7.11	1112.63	5.87	19.48	21.67
85% CED + 15% CAS	7.44	15.58	0.01	1.01	1095.91	14.38	1101.79	14.68	19.37	21.34
80% CED + 20% CAS	7.44	15.22	0.18	1.09	1113.23	10.85	1124.60	13.97	19.35	21.76
75% CED + 25% CAS	7.45	15.14	0.14	0.89	1122.31	10.78	1132.37	14.41	19.40	21.97
CAM
100% CAM	7.46	21.04	0.02	0.34	1012.43	32.22	1025.15	30.67	18.86	19.33
95% CAM + 5% CAS	7.46	20.99	0.04	0.95	1019.09	21.67	1026.58	6.16	18.89	19.39
90% CAM + 10% CAS	7.43	20.03	0.04	0.62	1057.17	19.86	1075.01	22.32	19.05	20.48
85% CAM + 15% CAS	7.44	20.65	0.01	0.42	1045.79	11.96	1061.04	8.87	18.87	20.02
80% CAM + 20% CAS	7.44	19.97	0.01	0.42	1072.29	40.48	1085.91	35.50	19.08	20.72
75% CAM + 25% CAS	7.44	19.87	0	0.80	1096.12	8.10	1098.66	8.93	18.96	20.83

---

No significant changes in the mass density were observed for the pellets after 20 days’ storage, implying that the length resiliency can significantly affect the mass density of pellets during storage. However, the mass densities of CED–CAS and CAM–CAS increased with increasing CAS content (Table 2). Such trends in the pellet density were mainly due to the softened protein and starch flowing into the voids and gaps between biomass particles and forming tighter structures as shown in Fig. 3b1–3 and c1–3. The energy density of CED–CAS and CAM–CAS increased from 21.45 GJ m⁻³ to 21.97 GJ m⁻³ and 19.33 GJ m⁻³ to 20.83 GJ m⁻³, respectively. These results indicate that the CAS blended with biomass can improve the energy-quality of pellets in industrial applications.

3.3 Moisture adsorption

The adsorption of moisture can “loosen” the hydrogen bonding and solid bridges within pellets.^16,23 Dusts and fines were thus produced from the moisture uptake, causing ignition and explosion during handling, storage and transportation.²⁴ In Fig. 4 and Table 3, the estimates for the adsorption rates (in the first five hours) are 0.00442 min⁻¹ and 0.00492 min⁻¹ for CED and CAM, respectively. The higher hemicellulose content in CAM may cause the higher adsorption rate. In addition, the closer structure of CED, with compressed fibers, resisted the moisture uptake compared with CAM. However, the equilibrium moisture adsorption (measured in the 48^th hour) was 13.10% for CED and 11.63% for CAM. These results indicate that the intertwined fiber structure of CED can provide more sites to absorb moisture compared to that of CAM. Moreover, there was no clear variations for CED–CAS and CAM–CAS with increasing of CAS content in terms of moisture adsorption.

Fig. 4 The instantaneous moisture content of pellets densified from blends of biomass and CAS. (a) CED; (b) CAM.

Table 3 Data on moisture adsorption

Samples	CED			CAM
	Equilibrium moisture adsorption	Absorption rate k (min^−1)	R-Square	Equilibrium moisture adsorption	Absorption rate k (min^−1)	R-Square
0% CAS	13.10	0.00442	0.98	11.63	0.00492	0.99
5% CAS	13.55	0.00601	0.96	11.88	0.00529	0.98
10% CAS	13.61	0.00511	0.98	11.32	0.00469	0.99
15% CAS	14.20	0.00526	0.95	11.27	0.00468	0.99
20% CAS	14.14	0.00422	0.98	10.90	0.00405	0.98
25% CAS	14.12	0.00436	0.98	10.75	0.00432	0.98

---

3.4 Mechanical hardness

Fig. 5 shows that the maximum collapsing force and Meyer hardness of CED were significantly higher than those for CAM. This may be attributed to the diversity in the particle structures mentioned in Section 3.1. As shown in Fig. 3a1–3, more intertwined fibers were identified on the surface of CED compared with the flat surface of CAM. The fiber matrix and solid bridge between CED particles were hence strengthened due to the large contact area in the intertwined fibers, resulting in a higher maximum collapsing force and Meyer hardness for CED runs. Meanwhile, in Fig. 5, the average maximum collapsing forces of CED–CAS and CAM–CAS increased from 63 N to 100 N and from 23 N to 53 N with increasingly higher CAS content, respectively. It can be implied that the binders between particles were enhanced by CAS due to the adhesive function of the proteins and starch in CAS.^9,22 This is consistent with previous research.²⁵ It was reported that the collapsing forces of alfalfa pellets which contained 17.6% and 22% protein were 425 N and 507 N, respectively.²⁵ However, there were different trends in the Meyer hardness between CED–CAS and CAM–CAS with increasing CAS content. In Fig. 5a, the Meyer hardness of CAM–CAS increased from 3.93 N mm⁻² to 6.52 N mm⁻². This may be due to the enhancement of hydrogen bonding and solid bridges with the addition of CAS.^3,23 However, a slight decline was detected for the Meyer hardness of CED–CAS in the range of 5.09–5.67 N mm⁻² upon the CAS content increasing from 5% to 15%.

Fig. 5 Meyer hardness and the collapsing force of pellets as a function of content of CAS. (a) Meyer strength; (b) collapsing force.

The Meyer hardness is determined by the maximum collapsing force versus distance during compression, reflecting the resistance to deformation. As shown in Fig. 6a, the distance of the indentation depth increased at the same compression force with the increasing CAS content for CED, while the opposite trend was observed for CAM. Therefore, a lower Meyer hardness was obtained at CAS contents of 10% and 15% in the CED–CAS runs. On the micro level, the protein and starch contained in CAS, which was softened and acted as an interlocking binder between biomass particles, can increase the elasticity of the pellet matrix, resulting in a lower resistance to deformation.^9,23 In addition, the Meyer hardness in the CED–CAS runs was enhanced slightly and leveled off around 5.2 N mm⁻² with the increase in CAS content from 15% to 25%. The further increase in the maximum collapsing force contributed to the slight increase in Meyer hardness. However, a further increase in Meyer hardness could not be achieved due to the resistance of the pellet to deformation, which can be confirmed by the images in Fig. 6a.

Fig. 6 Collapsing force versus distance graph of pellets as a function of content of CAS. (a) CED; (b) CAM.

3.5 Combustion characteristic

It is evident from the TG and DTG curves shown in Fig. 7 that the two types of biomass (CED and CAM) depicted three step weight losses: moisture evaporation, oxidative degradation and char combustion. The combustion parameters were characterized by ignition temperature, maximum DTG and temperature at maximum DTG; these are summarized in Table 4. The higher maximum DTG in the DTG curves (Fig. 7 and Table 4) represents the strongly oxidative pyrolysis region and the temperature at the maximum DTG is considered as a standard for the reactivity of materials.²⁶ The initial weight loss is attributed to the dehydration in the samples at temperatures of 25–100 °C. In the region of oxidative degradation, the ignition temperatures were 223–214 °C and 222–211 °C for the CED and CAM sequences, respectively. The lower ignition temperature is due to the residual oil in CAS, which can act as the comburent during combustion. It was also found that the maximum DTG value decreased from 1.25 to 0.97 s⁻¹ and from 1.12 to 0.47 s⁻¹ for CED and CAM with increasing contents of CAS, respectively. These results indicated that the high fixed carbon content (19.82%) and the ash content (8.14%) in CAS can slow down the releasing and burning of combustibles in biomass by covering the volatile matter. Furthermore, the close-knit structure formed by CAS may reduce the level of release of volatiles.⁹ As shown in Fig. 7, there is a increase in the maximum DTG in the char combustion region with the increment of CAS content for both biomass materials. It was reported that char combustion could be improved by adding coal, which can lead to higher HHV and stable combustion.²⁷ This result suggested that adding CAS can enhance char combustion and lead to greater reactivity during char combustion. This can be explained by two reasons, as follows: firstly, the higher fixed carbon content in CAS can improve the char combustion compared with that in biomass. Besides, the char combustion rate can be enhanced by mutual interaction caused by adding CAS. As a result, the addition of CAS has a similar function to coal in terms of combustion.

Fig. 7 TG and DTG curves of raw and torrefied pellets in a nitrogen flow rate of 50 mL min⁻¹ and a air flow rate of 100 mL min⁻¹ at a heating rate of 15 °C min⁻¹. (a) TG; (b) DTG.

Table 4 Combustion characteristics of raw biomass, torrefied biomass and their blends with CAS

Samples	Oxidative degradation			Char combustion
	Temperature range (°C)	Tpeak (°C)	Maximum DTG (% s^−1)	Temperature range (°C)	Tpeak (°C)		Maximum DTG (% s^−1)
100% CED	221–375	330	1.25	375–493		465	0.03
95% CED + 5% CAS	223–367	338	1.14	367–495		467	0.04
90% CED + 10% CAS	220–361	335	1.19	361–499		462	0.06
85% CED + 15% CAS	215–359	339	1.26	359–497		460	0.05
80% CED + 20% CAS	222–360	336	1.19	360–507		459	0.25
75% CED + 25% CAS	214–360	320	0.97	360–500		436	0.34
100% CAM	213–363	319	1.12	363–479		446	0.03
95% CAM + 5% CAS	218–354	337	1.12	354–477		447	0.03
90% CAM + 10% CAS	213–352	323	1.06	352–492		451	0.05
85% CAM + 15% CAS	219–357	322	0.93	357–507		446	0.06
80% CAM + 20% CAS	222–348	330	0.71	348–496		439	0.09
75% CAM + 25% CAS	211–365	316	0.47	365–532		433	0.17

---

4. Conclusions

There is a significant decrease in the compaction energy consumption for biomass with increasing contents of CAS. Furthermore, the denaturation of protein in CAS during compression can improve the formation of solid bridges and hydrogen bonding among particles, resulting in an increasingly higher maximum compression force or Meyer hardness of the pellets. Moreover, the addition of CAS can enhance char combustion, which led to greater reactivity during char combustion. However, due to the low output and the higher ash content in CAS, this additive should be added in a relatively low proportion of 10–20% by mass, which can be suitable for pellet production. The oil cake, a kind of residue obtained from the castor bean expression process, meets requirements of economic feasibility and commercial production.

Acknowledgements

The authors gratefully acknowledge the financial support provided by the National Natural Science Foundation of China (No. 21407046, No. 31470594), and the Key Laboratory of Renewable Energy Chinese Academy of Sciences (No. y407k91001).

Notes and references

1. H. Li, X. H. Liu, R. Legros, X. T. Bi, C. J. Lim and S. Sokhansanj, Appl. Energy, 2012, 93, 680–685.
2. Y. Z. Lu, C. W. Du, C. B. Yu and J. Zhoua, Anal. Methods, 2014, 25, 86–91.
3. Z. Liu, A. Quek and A. R. Balasubramanian, Appl. Energy, 2014, 113, 1315–1322.
4. A. Pirraglia, R. Gonzalez, D. Saloni and J. Denig, Energy Convers. Manage., 2013, 66, 153–164.
5. J. Berghel, S. Frodeson, K. Granström, R. Renström, M. Ståhl, D. Nordgren and P. Tomani, Fuel Process. Technol., 2013, 112, 64–69.
6. M. V. Gil, P. Oulego, M. D. Casal, C. Pevida, J. J. Pis and F. Rubiera, Bioresour. Technol., 2010, 101, 8859–8867.
7. H. Li, L. B. Jiang, C. Z. Li, J. Liang, X. Z. Yuan, Z. H. Xiao, Z. H. Xiao and H. Wang, Fuel Process. Technol., 2015, 132, 55–61.
8. M. Ståhl and J. Berghel, Biomass Bioenergy, 2011, 35, 4849–4854.
9. L. Jiang, J. Liang, X. Z. Yuan, H. Li, C. Z. Li and Z. H. Xiao, Bioresour. Technol., 2014, 166, 435–443.
10. N. Kaliyan and R. Morey, Biomass Bioenergy, 2009, 33, 337–359.
11. H. J. Ahn, H. S. Chang, S. M. Lee, D. H. Choi, S. T. Cho, G. S. Han and I. Yang, Renewable Energy, 2014, 62, 18–23.
12. W. H. Chen, W. Y. Cheng, K. M. Lu and Y. P. Huang, Appl. Energy, 2011, 88, 3636–3644.
13. J. Chau, T. Sowlati, S. Sokhansanj, F. Preto, S. Melin and X. T. Bi, Appl. Energy, 2009, 86, 616–621.
14. P. S. Lam, S. Sokhansanj, X. T. Bi, C. J. Lim and S. H. Larsson, Bioresour. Technol., 2012, 116, 396–402.
15. D. Bergström, S. Israelsson, M. Öhman, S. A. Dahlqvist, R. Gref and C. Boman, Fuel Process. Technol., 2008, 89, 1324–1329.
16. P. Y. Lam, P. S. Lam, S. Sokhansanj, X. T. Bi, C. J. Lim and S. Melin, Fuel, 2014, 117, 1085–1092.
17. P. S. Lam, S. Sokhansanj, X. T. Bi, C. J. Lim and S. Melin, Energy Fuels, 2011, 25, 1521–1528.
18. L. Tabil, S. Sokhansanj, W. Crerar, R. Patil, M. Khoshtaghaza and A. Opoku, Can. Agric. Eng., 2002, 44, 55–63.
19. N. P. K. Nielsen, J. K. Holm and C. Felby, Energy Fuels, 2009, 23, 3211–3216.
20. D. R. Morris, F. R. Steward and C. A. Gilmore, Energy Convers. Manage., 2000, 41, 1557–1568.
21. S. Patel, A. M. Kaushal and A. K. Bansal, Pharm.Res., 2007, 24, 111–124.
22. L. J. Kong, S. H. Tian, C. He, C. M. Du, Y. T. Tu and Y. Xiong, Appl. Energy, 2012, 98, 33–39.
23. N. Kaliyan and R. V. Morey, Bioresour. Technol., 2010, 101, 1082–1090.
24. C. Serrano, E. Monedero, M. Lapuerta and H. Portero, Fuel Process. Technol., 2011, 92, 699–706.
25. L. Tabil and S. Sokhansanj, Can. Agric. Eng., 1997, 39, 17–23.
26. Q. Yi, F. Qi, G. Cheng, Y. Zhang, B. Xiao and Z. Hu, J. Therm. Anal. Calorim., 2012, 112, 1475–1479.
27. C. Zhou, G. Liu, S. Cheng, T. Fang and P. K. Lam, Bioresour. Technol., 2014, 166, 243–251.

---