Biofuels derived from cellulosic biomass offer one of the best near- to midterm alternatives to petroleum-based liquid transportation fuels. Biofuel conversion is mainly done through a biochemical pathway in which size reduction, pelleting, pretreatment, enzymatic hydrolysis, and fermentation are main processes. Many studies reveal that biomass particle size dictates the energy consumption in the size reduction. Biomass particle size also influences sugar yield in enzymatic hydrolysis, and biofuel yield in fermentation is approximately proportional to the former enzymatic hydrolysis sugar yield. Most reported studies focus on the effects of biomass particle size on a specific process; as a result, in the current literature, there is no commonly accepted guidance to select the overall optimum particle size in order to minimize the energy consumption and maximize sugar yield. This study presents a comprehensive experimental investigation converting three types of biomass (big bluestem, wheat straw, and corn stover) into fermentable sugars and studies the effects of biomass particle size throughout the multistep bioconversion. Three particle sizes (4 mm, 2 mm, and 1 mm) were produced by knife milling and were pelletized with an ultrasonic pelleting system. Dilute acid method was applied to pretreat biomass before enzymatic hydrolysis. Results suggested 2 mm is the optimum particle size to minimize energy consumption in size reduction and pelleting and to maximize sugar yield among the three particle sizes for big bluestem and wheat straw biomass. Nevertheless, there is no significant difference in sugar yield for corn stover for the three particle sizes.
Today's economy and society are overwhelmingly reliant upon liquid transportation fuels. This reality will not change dramatically in the near future. However, over 90% of the liquid transportation fuel used in the U.S. is petroleum based (gasoline and diesel) and more than half of the petroleum is imported [1–4]. Meanwhile, consuming petroleum-based transportation fuels contributes to the accumulation of greenhouse gases (CO2, SO2, and NOx) in the atmosphere [5,6]. Therefore, it is imperative to develop alternative liquid transportation fuels that are domestically produced and environmentally friendly.
The second-generation biofuels produced from cellulosic biomass (forest and agricultural residues and dedicated energy crops) are alternatives to conventional transportation fuels. Compared to the first-generation biofuels (fuels produced from feedstocks such as corns and sugarcanes, which can potentially be made into food/feed), producing biofuels from the inedible cellulosic biomass creates less competition with food/feed production for the limited agricultural farmland and other resources . Producing and using cellulosic biofuels can turn residues and wastes into energy, enhance the nation's energy security, create new employment opportunities, and improve rural economies . Cellulosic biofuel can be used on its own as a sustainable liquid transportation fuel or blended with conventional transportation fuel to meet the government's mandate of 16 billion gallons of cellulosic biofuel produced annually by 2022 in the Renewable Fuel Standard as part of the Energy Independence and Security Act [7,8]. There are over 1 billion dry tons of cellulosic biomass that can be sustainably harvested in the U.S. every year. This amount is sufficient to produce 90 billion gallons of biofuel that can displace about 30% of the nation's current liquid fuel consumption annually . Furthermore, cellulosic biofuel has the potential to reduce greenhouse gas emissions. Among all the other renewable energy sources, biomass is the only sustainable carbon carrier. Biomass grows by absorbing carbon dioxide, water, and nutrients and converts them into hydrocarbons through photosynthesis. When biomass is consumed as a fuel, carbon is cycled in the atmosphere .
The composition of cellulosic biomass varies by species. The three main components are cellulose (40–50%), hemicellulose (20–30%), and lignin (15–20%) [11,12]. Surrounding cellulose fibrils (an agglomeration of glucan chains with a large number of glucose units) are hemicellulose and lignin that interweave to form the networking structure as shown in Fig. 1. Cellulose and hemicellulose chains can be broken into sugars, which can then be fermented into biofuels by enzymes. Lignin is a large group of aromatic polymers and contains no sugar components. Lignin can be used to make value-added chemicals and biobased products or burned to produce bioenergy [11–13].
Figure 2 illustrates the major steps of converting cellulosic biomass feedstocks into biofuels. These steps are divided into the feedstock preprocessing stage conducted in a field or at a depot and the bioconversion stage performed at a biorefinery. Size reduction is a necessary process with the current conversion technologies, as raw cellulosic biomass feedstocks (such as the stems of herbaceous biomass or chunks of woody biomass) cannot be directly converted into biofuels efficiently [14,15]. This process decreases cellulosic biomass particle size by mechanical methods such as milling, cutting, and chipping. The large physical size and the strong structure of cellulosic biomass make size reduction a highly energy-intensive process . Pelleting applies mechanical forces to compress biomass particles produced by size reduction into uniformly sized pellets or briquettes. Pelleting can increase the volumetric density of biomass from 40–200 kg/m3 to 600–1400 kg/m3, which will significantly lower the cost and improve the feedstock flowability in biomass logistics [17,18]. After being transported into a biorefinery, cellulosic biomass is pretreated first. The purpose of pretreatment is to break down the matrix formed by hemicellulose and lignin to make the feedstock more accessible to enzymatic hydrolysis; thus, pretreatment can accelerate the bioconversion and increase the yield of fermentable sugars. Finally, sugars are fermented into biofuels, such as bioethanol [19,20].
Biomass particle size (after size reduction) is a curial input process parameter with impacts on both the feedstock preprocessing and bioconversion stages. For example, biomass particle size dictates the energy consumption in the size reduction process . In order to produce biomass with a smaller particle size, more energy is usually consumed [22–26]. Biomass particle size also influences the yield of fermentable sugar in hydrolysis [27–30], and biofuel yield after fermentation is almost proportional to the former hydrolysis sugar yield .
Numerous studies have been conducted to investigate the effects of biomass particle size in producing cellulosic biofuel. However, as summarized in Table 1, the reported relationships are inconsistent. More importantly, it has been found that most reported studies focus on the effects of biomass particle size on a specific process or a single output parameter; references on the effects of biomass particle size throughout both the feedstock preprocessing and bioconversion stages are quite limited. As a result, in the current literature, there are no commonly accepted guidelines on how to select biomass particle size in order to conserve energy in the feedstock preprocessing stage while still achieving good biofuel yield in the bioconversion stage. This study presents a comprehensive experimental investigation on the conversion of three biomass materials into fermentable sugars and studies the effects of biomass particle size on various evaluation parameters throughout the multistage biofuel manufacturing process.
The cellulosic biomass materials used in this study were wheat straw, corn stover, and big bluestem. The moisture content of the materials was conditioned at approximately 7%. All materials (whole stems) were processed through a knife mill (SM 2000, Retsch, Inc., Haan, Germany) powered by a 1.5 kW three-phase electric motor. The milling chamber of the knife mill is shown in Fig. 3. Three sieves with sizes of 4 mm, 2 mm, and 1 mm were used to control the particle size received. In each size reduction experiment, 100 g of biomass was gradually fed into the milling chamber. The knife mill was stopped after 10 s when all the biomass was loaded into the milling chamber. The weight of the particles collected was measured, and the size reduction time was recorded.
As illustrated in Fig. 4, cellulosic biomass particles were compressed into pellets using an ultrasonic machine. This process combines an ultrasonic generation system and a pneumatic loading system. The ultrasonic generation system (Stationary USM, Sonic-Mill, Albuquerque, NM) converts the 60 Hz of power line frequency to 20 kHz of electrical power. Afterward, the 20 kHz electrical power is converted into mechanical vibration by a piezoelectric converter. The high-frequency mechanical vibration is then amplified and transmitted to the end surface of the pelleting tool. Biomass particles were compressed into a pellet inside a mold with a cylindrical cavity of 20 mm diameter. The pelleting action is driven by a pneumatic cylinder. In this study, the compressed air pressure for pelleting was set at 345 kPa (50 psi). The vibration amplitude of the pelleting tool is controlled by the power supply unit. In this study, the ultrasonic power was set at 50% (of the maximum ultrasonic power the power supply unit can generate), and the pelleting duration was kept at 90 s. Each pellet was made of 2 g of biomass. The moisture content of the biomass particles for pelleting was approximately 7% and pelleting experiments were conducted in ambient temperature (20–22 °C).
Dilute sulfuric acid pretreatment was employed in this study. Before each pretreatment test, 5 g (dry weight) of pellets and 150 mL of 2% (w/v) sulfuric acid were loaded into a 600 mL glass liner of a pressure reactor (4760A, Parr Instrument Co., Moline, IL). Samples were pretreated at 120 °C for 30 min. After pretreatment, biomass particles were collected and washed with warm distilled water on a suction filtration system to remove the residual acid. The moisture content of the collected biomass was measured to determine the dry weight of biomass for the subsequent hydrolysis.
Two independent hydrolysis flasks were prepared for each experimental condition. The volume of the hydrolysis slurry was 40 mL, which consisted of 2 g dry biomass (5% w/v), cellulase enzyme complex (Accellerase 1500, DuPont, Wilmington, DE), and sodium azide (0.02% w/v) in a buffer solution (pH = 4.8). The enzyme-to-biomass ratio was 1 mL for 1 g of dry biomass. Supernatant liquid from each flask was collected after 24 and 48 h of hydrolysis for sugar content measurement. The hydrolysis temperature was kept at 50 °C, and the agitation speed of the water bath shaker was 110 rpm.
Evaluation Parameters and Measurements
Energy consumption in this investigation was the electrical energy consumed in size reduction and pelleting. A Fluke 189 multimeter and a Fluke 200 AC current clamp (Fluke Corp., Everett, WA) were used to conduct the measurement. Current data were collected by the fluke software. The sampling rate was two readings per second. The software recorded the average current (IAVG) in each test. The line-to-neutral voltage (V) did not fluctuate much during tests, so it was regarded as a constant of 120 V. The electrical energy consumed during each test (Et) was calculated using the following equations: size reduction (three-phase power) Et = √3IAVG·V·t (J) and pelleting (single-phase power) Et = IAVG·V·t (J). Dividing Et by the weight (w) of the biomass materials received after the tests gives energy consumption (E) per unit weight as E = Et (kJ)/w (g). For each experimental condition in size reduction and pelleting, eight independent energy consumption measurements were performed.
Pelleting Density and Durability.
During the handling and transportation of biomass pellets from feedstock preprocessing facilities to biorefineries, the pellets will experience many different impacts, which can decrease pellet quality. Pellet density and durability were employed to evaluate the pellet quality in this study. Density of a pellet was calculated by the ratio of its weight to its volume. Eight pellets were randomly chosen for density measurements. Three density values (initial density, out-of-mold density, and 24-h density) were measured for each pellet. The initial density was the pellet density at the end of each test while it was still in the mold. The inner diameter of the mold cavity was taken as the pellet diameter, and the pellet height was obtained from a digital readout attached on the pelleting tool. Out-of-mold density was the pellet density right after being unloaded from the mold, and 24-h density was the pellet density measured 24 h after being unloaded. The latter two densities were measured to reflect the amount of spring-back (volume expansion) and the stability of a pellet during storage; these pellet dimensions were measured by a digital caliper. Durability tests were performed on an ASABE standard pellet durability tester (Seedburo Equipment Co., Des Plaines, IL) (Fig. 5). First, test pellets (stored in ambient conditions for 24 h) were loaded into the tester chamber. Then, the chamber was rotated at 50 rpm for 10 min. After each test, pellets were transferred and shaken through a U.S. No. 6 standard sieve to separate whole pellets from crumbs and loose particles. Durability is defined as the ability of densified pellets to remain intact when handled. The durability index was calculated as percentage between the mass pellets after and the mass of pellets before durability testing.
Biomass-based sugar yield was employed in this study. It is calculated as the amount of glucose yield (g) produced by 1 g of dry biomass in hydrolysis. Sugar yield measurement was performed on a high pressure liquid chromatography system (Agilent Technologies, Inc., Santa Clara, CA) with a Rezex™ RCM-Monosaccharide column (Phenomenex, Inc., Torrance, CA) and a Refractive Index Detector RID-G1362A (Agilent Technologies, Inc., Santa Clara, CA) at 40 °C. Water was used as the mobile phase at a flow rate of 0.6 mL/min. Chromatograph temperature was maintained at 80 °C. Samples were filtered through 0.2 μm hydrophilic polytetrafluoroethylene syringe filters before analysis. Two independent samples were measured under each experimental condition.
For all the measurements (except pellet durability index), independent experiments were carried out for at least three replicates, and data were averaged and presented with experimental standard errors. Analysis of variance using the least significant difference test was performed with sas software (SAS Institute, Inc., Cary, NC) at a significance level of 0.05 to compare the means of data collected under each experimental condition.
Results and Discussion
Size Reduction Energy Consumption.
As shown in Fig. 6, for all the biomass materials used in this study, size reduction energy consumption increased as biomass particle size decreased. This observed trend has been consistently reported in many studies with both the same size reduction equipment [21,32–35] and other machines [22–26,36–40]. In addition, to produce a specific amount of biomass particles of the same size, a size reduction machine with a higher power rating usually consumes less energy due to its higher material handling efficiency [32–36,53]. It was also noticed that when producing smaller biomass particles, the average current only increased slightly compared to that when producing larger biomass particles. The significant increase in energy consumption was mainly caused by the elongated time required to produce the same weight of biomass particles. In this study, for wheat straw and corn stover, the energy consumed to produce 1 mm biomass particles doubled the energy consumption to produce 2 mm particles. This significant increase in energy consumption when biomass particle size decreased from 2 mm to 1 mm was also reported when conducting size reduction of other biomass materials such as switchgrass [32,33], Miscanthus [32,33], and barley straw . Comparatively, for every particle size produced, big bluestem consumed the least amount of energy; producing 1 mm big bluestem only consumed 30% more energy than producing 2 mm big bluestem. It was observed that big bluestem particles had the best flowability among the three biomass materials. Particles smaller than the openings on the sieves could be discharged efficiently out of the milling chamber, which shortened the time required to produce the same amount of particles as required by the other two types. Contrarily, there is one report that found the opposite trend that producing larger particles consumed more energy . This observation was mainly caused by the biomass flowability issue of larger particles (45 mm and 85 mm) in that they did not move through the grinder (HG-200, Vermeer Corp., Pella, IA) as easily as the smaller particles (20 mm and 30 mm) .
Pelleting Energy Consumption.
The effects of biomass particle size on pelleting energy consumption are shown in Fig. 7. It was noticed that pelleting wheat straw particles consumed more energy than the other two materials. Pelleting 4 mm and 2 mm particles consumed more energy than pelleting 1 mm particles under the same conditions for wheat straw. This observation agreed with the predicted trend of an empirical model developed for this process . Since larger biomass particles initially yield a lower bulk density of the particles in the mold and require more energy to be condensed into a pellet than do smaller particles . For corn stover and big bluestem, biomass particle size did not significantly affect the pelleting energy consumption (P > 0.05). Song et al. also found that the energy consumption did not change significantly when pelleting 0.45 mm, 1 mm, 2 mm, and 4 mm particles; however, pelleting 8 mm particles consumed a larger amount of energy . Svihus et al.  studied the effects of biomass particle size on pelleting energy consumption with a high capacity pelleting machine (RPM350, Münch-Edelstahl GmbH, Hilden, Germany). The results indicated that there were no significant differences in energy consumption when pelleting 3 mm and 6.1 mm particles. Zhang et al.  also reported that the energy consumption were approximately the same when pelleting 3.2 mm and 9.5 mm corn stover and sorghum stalk particles on a ring-die pelleting machine (CPM 2000, California Pellet Mill Co., Crawfordsville, IN).
The pellet density data are shown in Fig. 8. Columns in the diagram show the out-of-mold density values. The upper bounds of the error bars are the initial density values and the lower bounds are the 24-h density values. It can be seen that pellets experienced a 15–25% spring-back in their volumes when taken out of the mold. They became more stable over the next 24 h with less than 5% spring-back occurring. Only for wheat straw pellets, the amount of spring-back increased as the particle size became larger. Wheat straw pellets had a much higher density than the other two materials, yet the density of corn stover pellets was the lowest. In general, pellets made from 2 mm particles had a slightly higher average density than that of pellets made from 1 mm particles, while pellets made from 4 mm particles had the lowest density among the three particle sizes. Mani et al.  examined the effects of compressive force, particle size, and moisture content on the density of pellets made from corn stover, wheat straw, barley straw, and switchgrass. It was found that all these variables significantly affected the pellet density, the only exception being that no significant effect was observed regarding particle size on wheat straw pellet density. Three particle sizes, 0.8, 1.6, and 3.2, were produced on a hammer mill, and smaller particles (except wheat straw) produced denser pellets. Mani et al. explained that different viscous and elastic characteristics of biomass materials contributed to the different densification behaviors since cellulosic biomass is often regarded as a viscoelastic material [63–65].
Pellet durability results are shown in Fig. 9. Pellets made from 1 mm and 2 mm biomass particles had similar durability, which was higher than that of pellets made from 4 mm biomass particles. Wheat straw pellets had the highest durability among the three biomass materials. Pellets made from 1 mm and 2 mm wheat straw particles had a nearly identical durability of 94%. The durability of pellets made from 4 mm wheat straw slightly decreased to 93%. Pellets made from 1 mm and 2 mm corn stover particles had a durability of 90%, and the durability decreased to 87% with 4 mm corn stover particles. Big bluestem pellets were less durable compared to the other two materials. Their durability results dropped from 86% to 80% as particle size increased from 1 mm to 4 mm. It has been broadly reported that finer particles generally correspond with greater durability as larger particles may serve as fissure points to initiate cracking or splitting in pellets . With the same pelleting mechanism as used in this study, Zhang et al. by doing a full factorial design of experiments found that 2 mm wheat straw particles produced more durable pellets than 1 mm particles . With conventional pelleting methods (without ultrasonic vibration), Singh and Kashyap reported that decreasing rice husk particle size from 5 mm to 4 mm increased pellet durability from 84% to 95% . Franke and Rey found that the smallest particles in their study, ranging from 0.5 mm to 0.7 mm, were the most feasible in generating high quality pellets . Hoover et al. reported that 4 mm corn stover particles produced pellets with a higher durability than the 6 mm particles . In a study of alfalfa pellet durability, Hill and Pulkinen  noted that a decrease in particle size from 6.4 mm to 2.8 mm increased the durability by more than 15%. Several other researchers observed that a mixture of particle sizes is more beneficial producing durable pellets. When being condensed, the mixture can increase the interparticle bonding and remove interparticle spaces more efficiently than uniformed particles .
Figure 10 shows the sugar yield results measured after 24-h (a) and 48-h (b) hydrolysis. For the same biomass material, 48-h sugar yield was approximately 10–15% higher than 24-h sugar yield, aside from 1 mm and 2 mm wheat straw, whose 24- and 48-h sugar yields were almost the same. Results in this study reveal that 2 mm particles had the highest sugar yield for all three biomass materials. Big bluestem had the highest sugar yield at all three particle size levels, with the 2 mm particles yielding about 20% more sugar after 48-h hydrolysis than the other two sizes. For wheat straw, the sugar yield differences between the 2 mm particles and the other two sizes were about 10%. The sugar yield differences among the corn stover particles of three size levels were not significant (P > 0.05).
Many studies have been conducted to investigate the effects of biomass particle size on hydrolysis sugar yield; however, results from these studies are inconsistent. As summarized in Table 1, the three possible relationships have all been reported. Two literature reviews present comprehensive summaries of reported studies regarding the effects of biomass particle size on sugar yield [22,66]. Observations from the literature reveal that the inconsistent relationships between biomass particle size and sugar yield could be a result of different (1) sugar yield definitions (sugar content versus conversion efficiency), (2) particle size ranges (micro versus millimeter), (3) types of biomass (pure cellulose versus lignocellulosic biomass), and (4) different pretreatment methods (hydrothermal pretreatment versus chemical pretreatment). It is also possible that in some cases, particle size is a weak predictor of hydrolysis sugar yield and has little effects on a substrate that has already been susceptible to hydrolysis .
This study presented a comprehensive experimental investigation on the conversion of corn stover, wheat straw, and big bluestem cellulosic biomass into fermentable sugars. The three biomass materials were processed through the feedstock preprocessing stage (size reduction and pelleting) and bioconversion stage (pretreatment and hydrolysis). Table 2 summarizes the effects of biomass particle size on various evaluation parameters throughout the multistage biofuel manufacturing process. The following conclusions were drawn: (1) For big bluestem, particle size did not significantly affect pellet density and the energy consumption of pelleting. Generally, smaller particle tended to have better durability. For sugar yield, 2 mm big bluestem was significantly higher than the other two particle sizes. (2) For corn stover, particle size negatively affected the energy consumption of size reduction significantly. Nevertheless, particle size did not have a significant impact on pellet density or sugar yield. (3) For wheat straw, smaller particle size significantly increased the energy consumption in size reduction and pelleting. Two millimeter had significantly higher sugar yield. However, particle size had no significant effects on pellet density. The effects of particle size on bioethanol production critically depended on the biomass type. Two millimeter particle size was recommended for big bluestem and wheat straw under the conditions employed in this study, whereas 4 mm corn stover produced similar fermentable sugar while consuming less energy than the other two particle sizes.
This study serves as an example to show that in order to realize cost-effective biofuel manufacturing, research in the feedstock preprocessing stage and the bioconversion stage should not be isolated from each other. More quantitative process optimization is needed from a system engineering perspective to provide practical guidance on selecting the optimal particle size. Observed trends have to be validated on a production scale platform. On the other hand, how mechanical impacts during the feedstock preprocessing stage alter the compositional and structural features of biomass that affect the biomass susceptibility to enzymes has to be further investigated. From these two perspectives, the fundamental connection linking the two stages in biofuel manufacturing can be established.
The authors would like to acknowledge undergraduate research assistants Nick Eisenbarth, Lane Sorell, and Catharine Lei, for their help in conducting experiments and measurements for this work.
The U.S. National Science Foundation (NSF).
Division of Civil, Mechanical and Manufacturing Innovation (1562671).