See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/271421386

Farm-Scale Biomass Pelletizer Performance for Switchgrass Pellet Production

Conference Paper in Applied Engineering in Agriculture · January 2013

DOI: 10.13031/aim.20131598515

CITATIONS READS
0 577

6 authors, including:

Hojae Yi Gregory Roth

Pennsylvania State University Pennsylvania State University
84 PUBLICATIONS 744 CITATIONS 95 PUBLICATIONS 2,924 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Hojae Yi on 22 September 2017.

The user has requested enhancement of the downloaded file.

 FARM-SCALE BIOMASS PELLETIZER PERFORMANCE
FOR SWITCHGRASS PELLET PRODUCTION
D. Ciolkosz, R. Hilton, C. Swackhamer, H. Yi, V. M. Puri, D. Swomley, G. Roth

ABSTRACT. The impact of feedstock characteristics (moisture content, additive content) and operational parameters (die
temperature, pelletizer speed) on pellet quality, plugging tendency, and pelletizer energy use was investigated for a small-
scale pelletizer, suitable for on-farm use (~70 kg h-1 rated output). Ground switchgrass (Panicum virgatum) was prepared
in small (<1 kg) batches and run through a flat plate, rotating die pelletizer with 6 mm die holes and a die thickness of
25 mm. The resulting pellet quality and die flow (plugging tendency) were assessed using a subjective pellet quality scale.
Results indicate that successful pelleting conditions were most consistently achieved by using a "premix" consisting of
ground switchgrass and Distillers Dried Grains in a 3:7 ratio (mass basis) to condition the die, followed by the actual
feedstock mixture. The highest quality pellets were obtained from switchgrass with a moisture content ranging from 12%
to 18% (wet basis). Adding between 1% and 4% vegetable oil improved pellet appearance, while adding starch (1% To
5%) to the switchgrass feedstock did not yield quality improvements. Reducing the operating speed of the pelletizer
resulted in improved quality of the pellets. Feedstock moisture content was positively correlated to pellet production rate
and negatively correlated with pelletizer energy use.
Keywords. Biomass densification pellets.

P
elletizing is a process of creating a densified quality, often measured as the diametral compressive
cylindrical compact by extruding granular material strength of the pellet and its resistance to crumbling when
through a rigid die. Biomass pelletizing is placed in a rotating test container (ASABE Standards,
typically achieved using rollers to press ground 2012). Variables of investigation have included:
material through a die that is shaped either as a ring or a • moisture content,
flat plate, with holes to allow for passage and densification • steam addition,
of the ground biomass (fig. 1). While definitions vary, for • particle size distribution,
the purposes of this article, the term “pellets” will refer to • feedstock age,
densified cylindrical biomass compacts 5 to 10 mm in • process temperature,
diameter. Densified compacts greater than 25 mm in • process pressure,
diameter will be referred to as “briquettes” (PFI, 2008;
• feedstock additives,
Ciolkosz, 2009).
• die Length-to-Diameter (L/D) ratio.
Studies have investigated the pelleting of various types
of biomass, including wood, compost, grasses, straw, crop
residue, and torrefied material (Bergman and Kiel, 2005;
Mani et al., 2006; Finney et al., 2009; Tumuluru et al.,
2010, Stelte et al., 2011b). Typically, studies have focused
on the impact of feedstock characteristics on fuel pellet

Submitted for review in June 2014 as manuscript number MS 10803;

approved for publication by the Machinery Systems Community of
ASABE in April 2015. Presented at the 2013 ASABE Annual
International Meeting as Paper No. 1598515.
The authors are Daniel Ciolkosz, ASABE Member, Research
Associate, Ryan Hilton, Research Assistant, Clay Swackhamer,
Research Assistant, Ho-Jae Yi, ASABE Member, Research Associate,
Virendra M. Puri, ASABE Member, Professor, Drew Swomley,
Research Assistant, The Pennsylvania State University, Department of
Agricultural and Biological Engineering, University Park, Pennsylvania;
and Gregory Roth, Professor, The Pennsylvania State University,
Department of Crop Science, University Park, Pennsylvania.
Corresponding author: Daniel Ciolkosz, The Pennsylvania State Figure 1. Schematic of key components and processes of a biomass
University, Department of Agricultural and Biological Engineering, 249 pelletizer. A) axial compression of loose granular material, B) axial
Ag Engineering Building, University Park, PA 16802; phone: 814-863- and transverse compression and movement, C) Axial compression and
3484; email: [email protected]. movement, and D) ejection and relaxation of pellets. (Not to scale)

Applied Engineering in Agriculture

Vol. 31(4): 559-567 © 2015 American Society of Agricultural and Biological Engineers ISSN 0883-8542 DOI 10.13031/aea.31.10803 559
 The variability of results from various studies suggests measured a specific energy consumption rate of 300 MJ t-1
that the pelletizing process is highly sensitive to variations using the same equipment (Kaliyan et al., 2009). Specific
in procedure, and that the type of pelletizing equipment, as energy consumption of the process is relevant as it gives a
well as the manner in which it is used, may play an more precise indication of the energy used to densify the
important role in the performance of the pelletizer. biomass (as opposed to total energy use of the device), and
Experimental manufacture of pellets is often carried out can in turn be used to determine the ratio of energy used in
either using a slow, controlled densification process, or by the process vs. the energy content of the feedstock.
using industrial scale pelleting equipment (i.e., Kaliyan and Several mathematical/numerical models have been
Morey, 2006; Arshadi et al., 2008). While the slower developed that attempt to characterize the biomass
process is often more controlled and easier to characterize, pelletizing process. Nielsen (2009) proposes a three-step
the much faster process used in commercial pelleting model for biomass pelletization consisting of three distinct
equipment is likely to have different performance steps: compression of the granular material above the die,
characteristics due to the different rates of compaction and flow through the narrowing “inlet throat” of the die, and
different geometries of the dies. friction resistance as the material passes through the
Key mechanisms that are identified as being relevant to straight section of the die. Several studies have employed
the biomass pelleting process include: finite element models to simulate the process (Hu et al.,
1. Moisture, which is shown to have a significant 2010; Lu et al., 2012; Ye et al., 2013), yielding insight for
impact, by reducing the glass transition temperature die design. Axial pressure within the die is modeled to
of the lignin (Nielsen et al., 2009), and affecting the decrease exponentially as the biomass travels through the
frictional characteristics of the material within the die (Krizan et al., 2009) dependent on feedstock
die, ultimately impacting pellet durability and density characteristics. On the feedstock side, statistical models
(Larsson et al., 2008). Steam is reported to be an have been developed that characterize the density of the
especially effective means for achieving improved ground biomass as a function of pressure (Tumuluru et al.,
performance for biomass (Leaver, 1988), 2010), although their application to the pelletizing process
2. Heating of lignin above its glass transition tempera- has not been widely reported.
ture, allowing for the (mobilized) lignin to form While the impact of variables on pellet quality is of
covalent bonds between adjacent particles as well as significant importance, the impact on equipment
to experience plastic deformation forming mechani- performance has been less thoroughly studied. Problems
cal “bridging” structures within the pellet (Kaliyan such as die clogging, failure to form pellets (“wash
and Morey, 2009; Stelte et al., 2011a), through”), and excessive energy use can make the pelleting
3. Longitudinal compression of the ground biomass process quite challenging. For example, Holm et al. (2006)
within the die, resulting in increased density of the report that feedstock type, additive composition, and
material. Resistive pressure is theorized to be a func- method of mixing can all impact the performance of the
tion of the pressure-friction characteristic of the pelletizer. They also noted the beneficial impact of Brewers
compressed material within the die (Krizan et al., Spent Grains on the pelletizing process for wood pellet
2009), and production, although the mechanism for this improvement
4. The Poisson ratio of the granular feedstock is was not identified.
suggested to be a key mechanism by which resisting The apparent mechanism of pellet formation and
pressure is built up in the die - impacting pellet quali- movement through the die involves resisting force
ty, energy requirements, and tendency of the die to developing within the feedstock matrix as pressure is
clog (Holm et al., 2006). applied from above. The maximum resisting force of a
The pelleting of switchgrass using ring die pelletizer differential length of matrix is likely a function of the
equipment has received some attention in the literature. applied force from the roller, the properties of the feedstock
Kaliyan et al. (2009) tested the impact of die aspect ratio matrix, and the “back pressure” applied from below. If the
when using a ring die pelletizer with 9.5 mm diameter die conditions experienced by the matrix are not sufficient to
openings, finding that a greater Length to Diameter (L/D) cause plastic deformation and agglomeration of the
ratio resulted in higher bulk density and durability of particles, “wash through” of feedstock occurs and pellets
switchgrass pellets. They also found that preheating the are not formed. If the maximum available downward force
feedstock was not necessary, concluding that friction from the roller is greater than the resisting force of the
between the particles and between the particles and the die matrix, the pellet material moves through the die. If not,
was sufficient to heat the pellets to the point where natural clogging of the die occurs. However, higher quality pellets
binding agents were activated (i.e., the glass transition tend to be associated with higher compressive forces
temperature of the binder, ~75°C). They measured a (Kaliyan and Morey, 2009).
specific energy consumption rate (excluding the power One of the research opportunities in this area is to study
used to run the machine when empty) of 403 to 414 MJ t-1. and better understand the relationship between the applied
Jannasch et al. (2004) pelletized switchgrass in a compressive force, properties of the feedstock, design of
commercial-scale ring-die pelletizer. They found that and manner in which the equipment is used, and the
reduced feedstock particle size results in increased pellet resistive force provided by the feedstock. This will allow
hardness, and reported a specific energy consumption rate for better control of equipment design and operation so that
of 268 MJ t-1. In an earlier study, Samson et al. (2000) the relatively small operational window of “good pellets”

560 APPLIED ENGINEERING IN AGRICULTURE

can be achieved more reliably and less desirable outcomes completely uniform. The sample was not subjected to a
can be avoided. waiting/hold period at this point, but was immediately used
Small-scale pelletizing equipment is, in some respects, for pelleting. DDG, starch, and oil were likewise added and
well suited for farm-scale production of biomass pellets. Its mixed manually (if called for). Cornstarch, when used, was
small size, portability, and affordability make it an always mixed before the addition of water, as this
attractive entry level device for small-scale pellet facilitated even mixing. Cornstarch and canola oil were
producers. They tend to be different from industrial scale never used in the same testing mix; they were always used
devices in several respects. First, the infeed mechanism separately.
tends to be manual rather than metered. Second, the rollers As a follow-up experiment, six additional tests were run
tend to be of smaller diameter, which can impact the at six moisture contents ranging from 15% to 29%, during
magnitude and direction of resulting compressive forces. which pellets were collected on a 15 s time interval and
Third, while industry recommendations for herbaceous evaluated separately for each time period. This allowed for
biomass often specify a die hole length-to-diameter ratio of evaluation of the variation in pellet quality and energy use
8 to 12 (Leaver, 1988), the dies on small pelletizers can over the course of the pelleting operation. Power use by the
often be thinner (i.e., less deep), which can have critical pelletizer was measured using a three-phase power
impacts on the formation and properties of the pellets. transducer (Ohio Semitronics Model W-061C, Hilliard,
Anecdotal discussions with operators of small pelletizers Ohio), and voltage and current were also measured (Veris
suggest that these devices are particularly prone to Hawkeye Model 922, Tualatin Ore., and CR Magnetics
problems with producing consistent high quality pellets, Model CR4550-250, St Louis, Mo.). Readings were taken
and that the mechanisms controlling their performance are on a 1 s interval and stored in a datalogger (Campbell
only vaguely understood (i.e. Shang et al., 2014). Scientific Model CR1000, Logan, Utah) for later analysis.
Because of these unique differences and the lack of a Correlation analysis was used to identify relationships
more rigorous understanding of the pelletizing process, between energy use, pellet throughput (mass produced
there is a need for continued work for assessing the during the 15 s period), and pellet quality.
performance of small-scale pelletizer devices under The pelletizer used for this study is a Pellet Pros (Model
different feedstock and operating conditions. This study PP220, Dubuque, Iowa) consisting of a 75 mm diameter
seeks to help address this issue by examining the impact of roller and a 150 mm diameter flat plate die, 25 mm thick
moisture content, use of additives, and equipment speed on with 6 mm diameter die holes. The inlet taper is 5 mm deep
equipment performance and pellet quality when making with an angle of 22°, and the exit taper is 1 mm deep.
switchgrass pellets. Pelletizer speed was controlled using a variable frequency
drive (Automation Direct Model GS2-25PO, Cumming
Ga.).
METHODOLOGY Prior to each experimental run, the die and housing were
Table 1 lists the selected experimental treatments for the brushed clean and the die holes were unplugged using a
study to investigate the impact of five operational variables hammer and punch. It should be noted that the innermost
on pelletizer performance. ring of holes on the die were left plugged because the roller
The rationale for selecting these variables and corre- does not fully pass over them during operation. The hopper
sponding ranges is based on the operational capabilities of was then fixed to the feed mouth to finish prepping the
the pelleting equipment and the common operating machine for the pelletizing of switchgrass mixtures.
conditions reported in other biomass densification studies. An earlier study indicated that it is very difficult for the
Switchgrass (Panicum virgatum cv. “Cave in Rock”) was switchgrass to pelletize properly in the machine, although
obtained from a local farmer in baled form, and was then once the machine started to make pellets, it tended to
ground in a Munson Knife Mill (Model SCC-10-S, Utica, operate more effectively (Rooney et al., 2012). One
N.Y.) with a 6.3 mm size screen. Moisture content of the important operational detail that was discovered was that
ground material was measured based on mass loss in a slow addition of the feedstock to the device was not
sample when placed in a drying oven at 105°C for 24 h. effective - probably due to heating and drying of the
Approximately 500 g of ground switchgrass were then feedstock material prior to its being pressed into the die.
prepared for testing. Moisture content was adjusted by Instead, adding the material all at once yielded better
manually adding liquid water to the sample using a spray results. More importantly, however, it was found that a
bottle and mixing by hand until the material appeared mixture of switchgrass and DDG, when added to the

Table 1. Experimental variables for study of forming pellets using switchgrass.

Pelletizer
Test Moisture Content DDG Content Corn Starch Content Oil Content Die Temp Speed
Variable (wet basis) (% by mass) (% by mass) (% by mass) at Start (Hz)
Moisture Content 5% - 30% 0% 0% 0% 20°C 5.7
[a]
Added DDG 10%, 20% 0, 10, 20, 30, 40% 0% 0% 20°C 5.7
Added corn starch 14%-17% 0% 0, 1, 2, 4, 6, 10% 0% 20°C 5.7
Added oil 14% 0% 0% 0, 1, 2, 3, 4, 5, 6% 20°C 5.7
Pelletizer speed 14% 0% 0, 1, 2, 4, 6% 0, 1, 2, 3, 4, 5, 6% 20°C 11.3 or 5.7
[a]
DDG=Distillers Dried Grains.

31(4): 559-567 561

pelletizer at the start of the experimental run, caused the Statistical analysis of the impact of moisture content,
subsequent test sample to pelletize much more reliably. As starch, and oil on pellet quality and die flow included
a result, this approach (use of a “premix” followed by the correlation and regression analysis. Analysis of the impact
sample treatment) was used for all experimental treatments of pelletizer speed was carried out using a paired t-test
reported here. Preliminary tests (data not shown, see Hilton between the two speeds tested. Analysis of energy use and
and Swomley, 2012) indicated that the pre-mix was most pellet quality utilized correlation analysis-calculating
effective when a high percentage of DDG was used, and a correlation between energy use, production rate, and pellet
mixture of 70% DDG and 30% switchgrass (mass basis) quality and by regressing production rate and energy use
was selected for use. versus moisture content.
Evaluation of the resulting pellets was carried out using
two subjective pellet evaluation scales: one for pellet
quality and one for the “die flow” of the process (table 2). RESULTS
A score ranging from 0 to 10 was given to the results of IMPACT OF MOISTURE CONTENT
each experimental treatment, providing an evaluation of the The correlation coefficient between moisture content
quality of the final product and the likelihood of the die to and pellet quality is only 0.119-suggesting a very weak
clog with material and cease functioning. Descriptions of relationship. However, examining the data (fig. 2) shows
ratings of 1, 5, and 10 were developed, and operators were that varying the moisture content of the feedstock actually
asked to use their judgment to assign a whole number score has a noticeable impact on the pelletizing process. Die
based on those descriptions. Typically, 200 to 300 g of temperature was measured during these tests using an infra-
pellets were produced by a single treatment for subsequent red sensor, with die temperatures in the 80°C to 90°C
evaluation. Comparison of subjective ratings of pellet range, and no apparent correlation to feedstock moisture
quality to durability measured by the standard “tumbling content.
box” test (ASABE Standards, 2012) has indicated a positive Pellet quality had higher ratings when moisture content
correlation between the two measures, with greater of the test mixture ranged between approximately 12% and
sensitivity from the subjective ratings (i.e. a 60% variation 25%. Outside this range, pellet quality was lower. Tests
in subjective rating corresponded to a 40% variation in below 12% moisture generally failed to form pellets,
durability rating – Vendetti and Crawford, 2013). Also, resulting in “washthrough” of the die. Tests at high
multiple individuals using the scale gave evaluations in moisture contents either failed to form pellets or else
good agreement with one another, further indicating the clogged the die. Die flow ratings tended to decrease as
robustness of the measure. moisture content increased, according to the following
regression equation:

Table 2. Subjective rating scale descriptions.

FD = -0.3043*M + 11.647 (1)
Score: 0 5 10 where
Pellet No pellets made Small pellet formed; Large, strong, tightly
quality brittle and not tightly packed pellet equal in
FD = die flow rating,
rating packed quality to commercial M = moisture content (%, wet basis),
grade wood pellets R2 = 0.5829.
Die flow Die clogged Die clogged within Virtually no clogging; Based on pellet quality and die flow, the optimum
rating immediately 1 min of starting to all material ran through
(< 20 pellets run (<50 pellets die
moisture content appears to be between 12% and 18% for
formed) formed) small-scale pelletizers.

10 10
9 9
8 8
Pellet Quality Rating (0-10)

7 7
Die Flow Rating (0-10)

6 6
5 5
4 4
3 3
2 2
1 1
0 0
0 5 10 15 20 25 30 35
Moisture Content (%)

Figure 2. Impact of feedstock moisture (percent, w.b.) on pellet quality (diamonds) and die flow (squares).

562 APPLIED ENGINEERING IN AGRICULTURE

IMPACT OF ADDING DISTILLERS DRIED GRAINS (DDG) IMPACT OF PELLETIZER SPEED
Adding DDG to switchgrass at 20% moisture content Several experimental runs were conducted with the die
resulted in creation of pellets that were not very strong, rotating at full speed (11.3 Hz), then repeated but with the
whereas mixtures with DDG and 10% moisture were more die at half speed (5.7 Hz), in order to investigate the impact
successful. Increasing the fraction of DDG in the mix is of equipment speed on pelletizer performance (fig. 4). Our
associated with a slight reduction in pellet quality, but this results showed that none of the full speed tests produced
trend is not statistically significant (p=0.12). pellets with a greater quality than tests run at half speed.
The mean pellet quality for tests run at half speed is 7.36,
IMPACT OF ADDING STARCH and is 5.64 for the tests run at full speed, while the p-value
Adding corn starch to the test mixture did not have a for the paired t-test is <0.001. However, the die flow rating
noticeable impact on either pellet quality or die flow. The tended to be higher for tests run at full speed, with a mean
correlation between starch content and pellet quality, and die flow of 7.21 at half speed and 8.21 at full speed (again,
between starch content and die flow, are 0.154 and 0.183, p<0.001).
respectively. Small granules of hardened starch were It was found that, when tests were run at half speed the
observed in the pellets, which may indicate that the starch material fed into the hopper had to be manually agitated to
was not uniformly distributed in the test material. The achieve good performance. Otherwise, the pelletizer
pellets that formed tended to exhibit many minor transverse stopped producing pellets and the testing mixture would
cracks, suggestive of longitudinal expansion of the pellet “wash through” the die without forming pellets. This may
after it left the die. The mean pellet quality and die flow be due to formation of pockets of steam at the surface of
ratings for starch addition were 6.75 and 7.75, respectively. the die that disrupted the flow of feedstock into the die.
When tests were run at full speed, vibration of the
IMPACT OF ADDING CANOLA OIL machine caused the feedstock in the hopper to be more
Results of the testing of canola oil’s impact on pelletizer vigorously agitated by the motion of the pelletizer,
performance are summarized below. presumably preventing steam pockets from developing.
The use of canola oil as an additive did not noticeably
improve pellet quality or die flow-in fact, a slight TIME INTERVAL ANALYSIS
downward trend may be noticed in figure 3 (p value for Pellet quality, mass of pellets produced and energy use
linear regression = 0.087). Unlike pellets with added starch, varied a great deal over the course of each experimental
pellets using oil in the feedstock did not exhibit run. Figure 5 shows an example of this variation, in this
longitudinal cracks but instead had a shiny appearance. The instance for the 20% moisture treatment.
pellets received mean ratings of 6.4 and 7.4 for pellet Summarized results from the time interval runs are
quality and die flow, respectively. The best results were shown in table 4. The “Weighted Pellet Quality” data are
observed when a test mixture was used with a moisture weighted according to the mass of pellets produced during
content of 14% mixed with 1% canola oil. Pellets that were each time interval, as follows:
produced using this mixture attained a higher quality than
tests run using no additives.
9
8
7
Pellet Quality Rating

Table 3. Impact of DDG on pelletizer performance.

6
10% Moisture 20% Moisture
% DDG Pellet Quality Die Flow Pellet Quality Die Flow 5
0% 7 8 5 7 4
10% 6 7 4.5 7 3
20% 6.5 7 5 7 2
30% 6.5 8 5 7 1
40% 6.5 8 4.5 7
0
0 2 4 6 8 10 12
Die Roller Frequency (HZ)
p
10
9
8
7
Die Flow Rating

6
5
4
3
2
1
0
0 2 4 6 8 10 12
Die Roller Frequency (HZ)

Figure 3. Impact of canola oil on pelletizer performance. Figure 4. Impact of die speed on pelletizer performance.

31(4): 559-567 563

 9 60

Pellet Quality (1-10, squares), Energy

8

Mass of Pellets (g, triangles)

50
7

Use (kw, diamonds)

6 40
5
30
4
3 20
2
10
1
0 0

0-15

120-135
150-165
180-195
210-225
240-255
270-285
300-315
330-345
360-375
390-405
420-435
450-465
90-105
30-45
60-75
Time Series (seconds)

Figure 5. Time series data for 20% moisture feedstock.

Table 4. Performance of time interval runs. 9.00 0.900

Weighted Mass of Mass of
Moisture Pellet Elapsed Feedstock Pellets Energy 8.00 0.800
Content Quality Time Processed Produced Used
(%) (0-10) (s) (g) (g) (kwh) 7.00 0.700
14 5.5 315 1088.2 369.9 0.132

kwh per kg of Pellets Produced

17 4.7 450 1156.5 314.1 0.172
kg per hr of Pellets Produced
6.00 0.600
20 4.5 480 1123.8 701.0 0.230
23 4.0 465 1289.6 1024.7 0.214
26 5.6 390 1148.2 875.9 0.255 5.00 0.500
29 3.3 555 1098.2 984.0 0.197
4.00 0.400

QW = (Σ Mi*Qi) / ΣMi (2) 3.00 0.300

where 2.00 0.200

QW = weighted pellet quality (1-10),
Mi = mass of pellets produced during “ith” time interval 1.00 0.100
(g),
Qi = pellet quality of “ith” time interval (1-10). 0.00 0.000
10 15 20 25 30
Moisture Content (%)
Analysis of average energy and throughput for the six
time interval runs indicates that as moisture content of the Figure 6. Pellet production rate (triangles) and energy use (squares)
feedstock increases, the rate of pellet production increases, vs. moisture content, with best fit line for pellet production rate (solid)
and energy use (dashed).
and the specific energy use (kwh per kg) decreases (fig. 6).
Best fit equations for these trends are as follows:
When assessing 15 s interval data from these runs, The
RP = -0.4447 + 0.2873 * M (3) strongest correlation is between energy use and pellet
quality, especially at the lower moisture contents tested
where (<20% m.c.), where the correlation coefficient is 0.898.
RP = pelletizer production rate (kg h-1,) The same trend is evident for correlation between pellet
M = moisture content (%), quality and mass of pellets produced, with a strong
R2 = 0.5496, correlation (0.863) when the moisture content is below
E =0.6632 - 0.0159 * M (4) 20%, but lower for wetter samples. This suggests that
energy use for high moisture feedstocks may be dictated by
where factors other than resistance to flow within the die.
E = pelletizer energy use (kwh kg-1),
M = moisture content (%),
R2 = 0.4961.
This suggests that moisture may be acting as a lubricant, DISCUSSION
reducing friction either between particles or between the These results suggest several interesting things about the
matrix and the die wall. In addition, moisture is known to pelleting process. Using a “premix” of switchgrass and
reduce the glass transition temperature of the natural DDG followed by the test mixture, was successful at
occurring binders and plasticize compounds that are acting achieving repeatable performance from the device. This
as lubricants (Kaliyan and Morey, 2009). result bears resemblance to the findings of Holm et al.
(2006), who noted improved pellet durability when brewers

564 APPLIED ENGINEERING IN AGRICULTURE

spent grains were added to the feeedstock. However, DDG binder in pellets (Thomas et al., 1998). Perhaps the added
did not appear to improve pellet quality in this study. The starch was not appropriately mixed or else not in high
working hypothesis regarding DDG and ground enough quantity to impact pellet quality. Effectiveness of
switchgrass that developed over the course of these starch as a binder depends on uniform coverage of the
experiments was that the DDG/Switchgrass mix is more biomass particles with a film of starch, as well as
readily able to form a “dynamic plug” -a matrix of material “gelatinization” or rupturing of the starch granule, which
within the die that resists flow enough to allow back may not have occurred during the tests.
pressure to be developed, but does not resist flow so much The subjective pellet quality and die flow ratings
that the device clogs. The mechanism by which this occurs developed in this study proved themselves to be rapid and
is not clear, but does not appear to be linked to starch or oil practical methods to assess the results of the experiments.
content in the DDG. Instead, this phenomenon may involve While subjective assessments are not always considered
the shape of the DDG causing the switchgrass fibers to be ideal for scientific work, in this case they provide a
oriented more randomly, thus increasing their tendency to necessary and useful method to obtain information not
lock together and form bridging structures within the die. otherwise available. Methods have been established for
The photos in figure 7 show un-densified chopped measuring some physical properties of biomass pellets, but
switchgrass, DDG, and a mixture of the two materials. these methods are either slow and costly (i.e. compressive
The finding that pellet quality improved when the strength testing) or are not successful for weak pellets (i.e.
pelletizer was run at half speed was interesting. The likely tumbling box durability tester). Furthermore, the “quality”
mechanism for this effect is that the pellets spent a longer of a pellet is a subjective term in and of itself, and
residence time in the die, perhaps allowing the particles to experienced personal evaluation is probably the best single
relax stress in their densified position and reduce the amount approach for obtaining a full assessment of the pel-
of springback that occurred when the pellets left the die. let-especially for small-scale operations or production for
Alternately, additional time in the die could have allowed the personal use, where a quick, inexpensive assessment is
lignin in the biomass to more fully heat and better adhere to needed. “Die Flow” is another characteristic that can be
adjacent particles. This explanation is consistent with the evaluated subjectively, but does not subject itself readily to
findings of Kaliyan et al. (2009), in which a thicker die (i.e. a physical-reductionist approach. Development of
longer residence time) was found to increase pellet bulk improved testing methods that evaluate pellets more
density and durability. Some models of the pelleting process completely and in a cost-effective manner would be useful,
(Tumurulu et al., 2010) do not include a time component, but but in the meantime, subjective analysis appears to be an
it appears that such an inclusion may be in order. This appropriate and valuable method for quick and easy
finding raises the interesting possibility of using a thinner die assessment.
at a slower throughput rate, which may have system or Power consumption was recorded for a limited portion
energetic benefits. of the tests carried out in this study, so this study does not
Moisture content has often been reported as a key provide a complete picture of power consumption
parameter impacting switchgrass pellet quality (Tumuluru characteristics for the different mixtures. However, higher
et al., 2010), and results from this study suggest the same. moisture content feedstocks tended to result in lower
Pellet quality and die flow were best with moisture content energy use and higher rates of production, suggesting in
in the range of 12% to 18%. The die tended to clog at very turn that moisture may have a lubricant effect on the pellets
high moisture content, and pellet quality at extremes of and/or by lowering the glass transition temperature of
moisture content was also very poor. natural binders to enable stronger inter-particle binding as
Oil and starch were both investigated as possible they pass through the die. Energy use and pellet quality are
additives-oil did improve switchgrass pellet appear- positively correlated, which suggests, reasonably, that
ance-reducing cracking of the pellet, but did not impact greater energy use is translating into more effective
overall quality appreciably. Starch addition did not have a densification. However, the correlation between pellet
discernable positive impact on pellet quality or die flow. quality and mass of pellets produced is a somewhat
This is surprising, since gelatinized starch is a well-known surprising, but welcome result, indicating that high quality

Figure 7. Close-up photos of (L to R) chopped switchgrass, distillers dried grains (DDG), and a mix of switchgrass and DDG. A mm scale is
shown in the leftmost photo.

31(4): 559-567 565

and high rates of production may not be mutually Finney, K., Sharifi, V., & Swithenbank, J. (2009). Fuel pelletization
exclusive. with a binder: Part II—The impacts of binders on the
The results and experience obtained in this study suggest combustion of spent mushroom compost-coal tailing pellets.
that the pelleting process is highly dependent on a feedback Energy & Fuels, 23(6), 3203-3210.
http://dx.doi.org/10.1021/ef900021t.
mechanism, in which the “back pressure” resisting forward Grinstead, C. M., & Snell, J. L. (1997). Chapter 11: Markov chains.
movement of the pellet in the die is dependent on the In Introduction to Probability (2nd ed.). Providence, R.I.:
quality of the pellet that is being formed, and the quality of American Mathematical Society.
the pellet that is being formed is dependent in part on the Hilton, R., & Swomley, D. (2012). Small scale biomass pelletizer
back pressure that is resisting forward movement of the performance. Project Report. University Park, Pa.: Penn State
pellet in the die. Achieving a stable operating point for Department of Agricultural and Biological Engineering.
small-scale pelletizers gives every indication of being a Holm, J., Henriksen, U., Hustad, J., & Sorensen, L. (2006). Toward
challenging task. Maintaining that stable operating point an understanding of controlling parameters in softwood and
via long duration runs may be challenging as well. The hardwood pellets production. Energy & Fuels, 20(6), 2686-
2694. http://dx.doi.org/10.1021/ef0503360.
existence of a feedback mechanism in the system suggests Hu, J., Xu, G., Liu, J., Lei, T., & Shen, S. (2010). Finite element
the possibility of developing control methods that can more modeling simulation in the straw pellet cold compressing
effectively stabilize and even optimize the operation of molding process. Bioresources, 5(4), 2447-2456.
small-scale pelletizers, but this will require greater insight Jannasch, T., Quan, Y., & Samson, R. (2004). A process and energy
into the controlling mechanisms within the process. A analysis of pelletizing switchgrass. Ste. Anne de Bellevue,
Markov-chain model (Grinstead and Snell, 1997) may lend Quebec, Canada: Resource Efficeint Agricultural Production
itself to characterizing this process, due to the cyclical (REAP-Canada). Retrieved from www.reap-canada.com.
nature of the applied force from the roller to the granular Kaliyan, N., & Morey, R. (2006). Densification characteristics of
feedstock. corn stover and switchgrass. ASABE Paper No. 066714. St.
Joseph, Mich.: ASABE.
Kaliyan, N., & Morey, R. (2009). Factors affecting strength and
durability of densified biomass products. Biomass and
CONCLUSIONS Bioenergy, 33(3), 337–359.
Switchgrass pellet quality and pelletizer performance http://dx.doi.org/10.1016/j.biombioe.2008.08.005.
were found to be linked to feedstock and operational Kaliyan, N., Morey, R. V., White, M. D., & Doering, A. (2009).
Roll press briquetting and pelleting of corn stover and
characteristics of small pelletizers. Conclusions from this
switchgrass. Trans. ASABE, 52(2), 543-555.
study include: Krizan, P., Soos, K., & Vukelic, L. (2009). Counter pressure
• A two-step process was successful at creating effecting on compacted briquette in pressing chamber. J. Prod.
switchgrass pellets, in which a premix of DDG and Eng., 12(1), 63-66.
switchgrass is followed by the test mixture. Larsson, S., Thyrel, M., Geladi, P., & Lestander, T. (2008). High
• Moisture content had a noticeable effect on pellet quality biofuel pellet production from pre-compacted low
quality and die flow, with best results occurring in density raw materials. Bioresource Tech., 99(15), 7176-7182.
the range between 12% and 18% (wet basis). http://dx.doi.org/10.1016/j.biortech.2007.12.065.
Leaver, R. (1988). The Pelleting Process (2nd ed.). Muncy, Pa.:
• Using cornstarch as an additive to our testing mixture Sprout-Bauer Inc.
did not improve the quality of switchgrass pellets. Liu, B., Liang, X., & Guo, H. (2012). Analysis of biomass briquette
• Using canola oil as an additive to our testing mixture machine ring mold length-diameter ratio of finite element based
did not improve the quality of switchgrass pellets. on ANSYS. Key Eng. Materials, 501, 463-466.
• Increased feedstock moisture content increased http://dx.doi.org/10.4028/www.scientific.net/KEM.501.463.
machine throughput and reduced energy use. Mani, S., Tabil, L., & Sokhansanj, S. (2006). Effects of compressive
• Switchgrass pellet quality is positively correlated force, particle size and moisture content on mechanical
properties of biomass pellets from grasses. Biomass and
with energy use as well as machine throughput. Bioenergy, 30(7), 648-654.
http://dx.doi.org/10.1016/j.biombioe.2005.01.004.
Nielsen, N. (2009). Effect of fiber orientation on compression and
REFERENCES frictional properties of sawdust particles in fuel pellet
Arshadi, M., Gref, R., Geladi, P., Dahlqvist, S., & Lestander, T. production. Energy & Fuels, 23, 3211-3216.
(2008). The influence of raw material characteristics on the PFI. (2008). PFI standard specification for residential/commercial
industrial pelletizing process and pellet quality. Fuel Proc. densified fuel. Arlington, Va.: Pellet Fuel Institute.
Tech., 89, 1442-1447. Rooney, D., Ciolkosz, D., Yi, H., & Puri, V. (2012). Performance
http://dx.doi.org/10.1016/j.fuproc.2008.07.001. characteristics of small scale biomass pelletizing equipment.
ASABE Standards. (2012). S269.5: Densified products for bulk 2012 Meeting of the Northeast Agricultural and Biological
handling—Definitions and method. St. Joseph, Mich.: ASABE. Engineering Convention (NABEC). St. Joseph, Mich.: ASABE.
Bergman, P., & J. Kiel. (2005). Torrefaction for biomass upgrading. Samson, R., Duxbury, P., Drisdelle, M., & Lapointe, C. (2000).
14th European Biomass Conf Exhibition. Paris, France. 17-21 Assessment of pelletized biofuels. Ste. Anne de ellevue, Quebec,
October, 2005. ECN Report #ECN-RX-05-180. Canada: Resource Efficiency Agricultural Production (REAP-
Ciolkosz, D. (2009). Manufacturing fuel pellets from biomass. Penn Canada). Retrieved from www.reap-canada.com.
State Renewable and Alternative Energy Fact Sheet Series
#UC203. University Park, Pa.: The Pennsylvania State
University.

566 APPLIED ENGINEERING IN AGRICULTURE

 Shang, L., Nielsen, N., Stelte, W., Dahl, J., Ahrenfeldt, J., Holm, J., Thomas, M., van Vliet, T., & van der Poel, A. F. B. (1998).
Arnavat, M., Bach, L., Henriksen, U. (2014). Lab and bench Physical quality of pelleted animal feed 3. Contribution of
scale pelletization of torrefied wood chips—Process feedstuff components. Animal Feed Sci. Tech., 70(1-2), 59-78.
optimization and pellet quality. Bioenergy Research. (7) 87-94. http://dx.doi.org/10.1016/S0377-8401(97)00072-2.
Stelte, W., Clemons, C., Holm, J., Sanadi, A., Ahrenfeldt, J., Shang, Tumurulu, J. S., Wright, C., Kenney, K. L., & Hess, J. R. (2010). A
L., & Henriksen, U. (2011b). Pelletizing properties of torrefied technical review on biomass processing: densification,
spruce. Biomass and Bioenergy, 35(11), 4690-4698. preprocessing, modeling, and optimization. ASABE Paper No.
http://dx.doi.org/10.1016/j.biombioe.2011.09.025. 1009401. St. Joseph, Mich.: ASABE.
Stelte, W., Holm, J., Sanadi, A., Barsberg, S., Ahrenfeldt, J., & Vendetti, V., & Crawford, J. C. (2013). Switchgrass pellet
Henriksen, U. (2011a). A Study of bonding and failure durability standardization. 2013 NEWBio Bioenergy Consortium
mechanisms in fuel pellets from different biomass resources. Ann. Symp.State College, Pa.
Biomass and Bioenergy, 35(2), 910-918. Ye, T., Li, H., Wang, L., & Fan, X. (2013). Finite element analysis
http://dx.doi.org/10.1016/j.biombioe.2010.11.003. on biomass pelletizing process. Proc. ICCEAE 2012, AISC 181,
(pp. 39-44). Springer-Verlag. Berlin Heidelberg.

31(4): 559-567 567

View publication stats