Food Bioprocess Technol

DOI 10.1007/s11947-016-1807-x

ORIGINAL PAPER

Thermo-Physical Properties of Distillers’ Spent Grain Pellets

at Different Moisture Contents and Condensed Distillers’ Soluble
Concentrations
Rani Puthukulangara Ramachandran 1 & Jitendra Paliwal 1 & Stefan Cenkowski 1

Received: 6 June 2016 / Accepted: 20 September 2016

# Springer Science+Business Media New York 2016

Abstract The thermo-physical properties of distillers’ spent Introduction

grain (DSG) pellets are the key input parameters for the heat
and mass transfer modeling of the drying process and for the As per the annual report of Husky Energy, western Canada’s
design of the suitable drying and storage systems. The main largest producer of ethanol, approximately 260,000 tons of
thermo-physical properties like particle density, thermal con- dried distillers’ grain with solubles (DDGS) is produced each
ductivity, and specific heat capacity of DSG pellets were de- year, which is about 35–40 % of the grain input per annum
termined using standard laboratory methods. The effects of (Husky Energy 2016). Distillers’ spent grain (DSG) is the
moisture content, percentage of condensed distillers’ solubles major by-product from ethanol plants, corresponding to about
(also called solubles), and temperature on these selected prop- 85 % of the total by-products generated (Xiros et al. 2008). It
erties were determined. The average particle density of the accounts for 10–20 % of the total income for ethanol indus-
DSG pellets with 0, 10, 30, and 50 % solubles was found to tries when used as animal feed and is also a good source of
be in the range of 898.8–1136.7 kg/m3. It was observed that fiber and protein when utilized in baked human foods
the particle density of DSG pellets increased with an increase (Johnson et al. 2011). Transportation and long-term storage
in condensed distillers’ soluble concentration and decreased of this wet slurry of DSG, obtained after wet milling method
with an increase in moisture content. Thermal conductivity of ethanol production, is highly challenging because of its
(0.17–0.42 W/(mK)) and specific heat (1.76–3.47 kJ/(kgK)) high initial moisture content (approximately 80 % wet basis).
of the DSG pellets increase linearly with an increase in mois- The storage and handling of wet DSG could be improved by
ture content, soluble concentration of the sample, and temper- drying it to safe storage moisture content of 10–12 % (Johnson
ature of the drying medium. Three multiple linear regression et al. 2014; Bourassa et al. 2015). Hence, wet DSG is centri-
equations were developed for predicting these properties as a fuged to separate the coarse grain fraction and the semi-solid
function of moisture content, soluble concentration, and tem- fraction called condensed distillers’ solubles (CDSs) or simply
perature with R2 value ≥0.86. solubles (Stroem et al. 2009; Bourassa et al. 2015). The high
amount of cellulose, hemicelluloses, lignin, protein, and lipids
(Sousa et al. 2007) makes this by-product attractive for added
Keywords Distillers’ spent grains . Condensed distillers’ value by drying.
solubles . Particle density . Thermal conductivity . Specific Superheated steam drying has been proven to be suitable
heat for drying wet DSG due to its superior heat transfer properties
when compared to air and elimination of fire hazard during the
drying process (Zielinska et al. 2009; Zielinska and
Cenkowski 2012; Johnson et al. 2013). Generally, the wet
* Stefan Cenkowski
DSG is poured over dry DSG granules and dried using rotary
[email protected] drum dryers in order to provide greater surface area for heat
and mass transfer. In this study, DSG at 25–35 %wb. moisture
1
Biosystems Engineering, University of Manitoba, content is compacted to form cylindrical pellets. These pellets
Winnipeg, Manitoba, Canada are then dried using superheated steam with wet DSG poured
 Food Bioprocess Technol

over them. The compaction process eliminates mixing with condensation is very important in the heat and mass transfer
already over-dried DSG granules while developing a substan- mechanism of superheated steam (SS) drying (Hamawand
tial contact area with the drying medium and thereby improv- 2011; Johnson et al. 2013; Hamawand et al. 2014). The con-
ing the drying process. Various elements of the drying process densation stops when the surface temperature of the sample
like constant rate drying, falling rate drying, moisture diffu- exceeds the saturation steam temperature. The temperature
sion, surface water evaporation, etc. can be studied using and the heat capacity of the sample influence this local con-
mathematical modeling of the heat and mass transfer phenom- densation (Iyota et al. 2007). The sensible heat gain occurring
ena that occurred during drying (Hamawand 2013). Detailed during condensation is supplied by the latent heat of conden-
information on the transport phenomena (heat and mass trans- sation of steam (Pakowski and Adamski 2011). The duration
fer) of drying process can be obtained with the help of ad- of condensation is governed by the temperature rise in the
vanced computer simulation techniques. Thermo-physical sample, which in turn is influenced by the heat capacity of
properties like thermal conductivity, specific heat, and particle the sample. Also, some amounts of condensed water might be
density of samples to be dried are the most important input absorbed by the sample causing an increase in the moisture
variables required for validating the mathematical model content of the sample and changing its thermo-physical prop-
(Yang et al. 2002; Suvarnakuta et al. 2007) and can be used erties (Emami et al. 2007).
in computational modeling of superheated steam drying Besides the use in heat and mass transfer modeling, these
(Erriguible et al. 2006). Also, these properties are essential thermo-physical properties are also being utilized for various
for optimization of various processing techniques like drying, processing operations like stabilizing, pelletizing, freezing,
mixing, storage, and handling (Mosqueda et al. 2014). These thawing, cooking, pasteurization, and sterilization and storage
thermal properties vary with different factors such as moisture (Nesvadba 1982; Sreenarayanan and Chattopadhyay 1986;
content, particle density, and composition of the material Drouzas et al. 1991). Hence, a complete understanding of
(Waszkielis et al. 2013, 2014). the major thermo-physical properties of the DSG sample like
The high heterogeneity of the compacted DSG in terms of particle density, thermal conductivity, and specific heat is im-
particle size and high porosity creates many challenges in the portant in the heat and mass transfer modeling of the SS dry-
measurement of its thermo-physical properties. Johnson et al. ing of DSG pellets. The precision of computational modeling
(2015) reported that an increase in the percentage of CDS in a of heat and mass transfer mechanism of superheated steam
mixture of coarse grain fraction of wet DSG and CDS in- drying is highly influenced by the accuracy of input parame-
creases the overall strength of a DSG pellet. They reported ters of the sample as well as the heat transfer medium and the
that the particle size of the CDS being about 56 % smaller boundary conditions for energy and mass balance equations
than that of the coarse grain fraction of wet DSG improved the (Defraeye 2014). Hence, the main objective of this study is to
packing of particles during compaction. They also reported determine the effect of moisture content and soluble concen-
that the CDS contains 48.8 % of crude protein and 5.76 % tration on the important thermo-physical properties of DSG
of starch, whereas the wet DSG contains only 29.4 % crude pellets such as thermal conductivity, specific heat, and particle
protein and 3.77 % starch. The higher percentage of starch density as affected by different temperatures of heat transfer
(about 52 % more) and protein (65 % more) in the CDS, when medium.
compared to wet DSG, justifies the lower porosity and higher
particle density (Tumuluru et al. 2010) and strength of the
DSG pellets with solubles (Johnson et al. 2014). Thomas Materials and Methods
et al. (1998) reported that the starch and protein can act as a
binding agent in the compaction process. The higher percent- Sample Preparation for Compacting
age of protein and starch present in the CDS helps to improve
the strength and durability of the pellet. The change in com- The DSG samples used in this study were a mixture of 90 %
position and density of the pellet due to the addition of differ- mashed corn and 10 % mashed wheat obtained after
ent levels of solubles also affects the overall thermal conduc- bioethanol production. The whole stillage was collected from
tivity of the pellet. In addition, the increase in the percentage a local distillery, Mohawk Canada Limited (Minnedosa, MB).
of condensed solubles causes an increase in the moisture dif- The whole stillage contained coarse grain fraction, condensed
fusivity during drying (Zielinska and Cenkowski 2012). solubles, and water. The wet stillage was kept in a freezer
Unlike hot air drying, superheated steam drying has a distinct (Eaton Viking, T. Eaton Company Limited, Toronto,
phenomenon called initial condensation that occurs in the Canada) at −15 °C in a sealed plastic pail (polypropylene pails
warming up period of drying. Due to the lower initial temper- from Pioneer Plastics) to prevent spoilage from further fer-
ature of the sample to be dried, the superheated steam coming mentation. Before sample preparation, a part of the sample
in contact with the sample is subjected to temporary conden- was removed from the freezer and allowed to thaw at room
sation on the sample surface. This phenomenon of initial temperature. The sample was then mixed thoroughly to ensure
Food Bioprocess Technol

uniform thawing. Once thawed, the sample was subjected to temperature (Stroem et al. 2009). A separate set of samples
centrifugation using a Sorvall General Purpose RC-3 labora- was then prepared at 25, 35, and 45 % moisture content for
tory centrifuge (Thermo Fisher Scientific, Waltham, MA) at each of the composition mentioned above. The prepared sam-
2200 rpm for 10 min. Approximately 750 mL of the sample ples were stored in a freezer at −15 °C.
was placed in each of the four centrifuge bowls for separation. For making the compacted DSG pellets, approximately
The supernatant liquid (thin stillage), top layer thin grain frac- 4.25 ± 0.05 g of thawed raw material at a given moisture
tions (solubles), and bottom layer of the coarse grain fraction content and percent solubles were placed in a cylindrical steel
of DSG were then separated and weighed. Each of the frac- mold with a diameter of 12.2 ± 0.01 mm and height of
tions was stored separately in airtight containers (polypropyl- 80 ± 0.01 mm. The mold with the sample was then placed
ene 5, Pioneer Plastics) which were then stored at −15 °C until on the platform of a universal testing machine (Model 3366
testing. Universal Testing Systems, Instron Corp., Norwood, MA). A
cylindrical die with a diameter of 12.08 ± 0.01 mm and a
Composition and Initial Moisture Content length of 83.65 ± 0.01 mm were attached to a load cell of
10 kN on the upper jaw of the universal testing machine.
Ten replicates of the centrifugation trials were conducted to Each sample of DSG was formed with a compaction load of
determine the percentage composition of the raw sample in 6820 N, the equivalent of 60.3 ± 0.1 MPa, at a compression
terms of coarse grain fraction, solubles, and thin stillage. The rate of 50 mm/min (Mani et al. 2003; Johnson et al. 2014).
mass of the separated coarse grain fraction, solubles, and thin After applying the compressive load, each individual pellet
stillage was recorded for each trial. The percentage composi- was held at the same load for 300 s by retaining the die inside
tion of each of the fraction with respect to the whole mass was the mold. This was done to maintain a constant pressure on the
calculated. The moisture content of the coarse grain fraction formed pellet and to reduce the springback effect (Tumuluru
and the solubles were determined using the air-oven drying et al. 2010). After 300 s of holding time, the die was removed
method (AACC 2000). Three replicates of each sample (2 g from the mold and the pellet was pushed out of the mold using
each) were dried in a laboratory oven (Thermo Electron the die.
Corporation, Waltham, MA) at 135 °C for 2 h (Johnson
et al. 2013, 2014). The whole stillage used for the study Particle Density of Pellets
contained 49.5 ± 0.5, 9.5 ± 0.5, and 40.5 ± 0.5 % w/w of the
coarse grain fraction, solubles, and supernatant liquid, respec- Dimensions of each DSG pellet with different moisture con-
tively. All the moisture contents were determined in wet basis tents and soluble concentrations were measured using a digital
(%wb.). The initial moisture content of the whole stillage was Vernier caliper (Pro-Point), with 0.01 mm accuracy. The mass
89.2 ± 0.05 %. The initial moisture content of coarse grain of each pellet was also measured using an electronic weighing
fraction and solubles after centrifugal separation was balance with an accuracy of 0.001 g (PGW 453e, Adam
77.8 ± 0.08 and 82.9 ± 0.1 %, respectively. Equipment Inc., Danbury, CT). The calculated volume and
average mass of the pellets were used to determine the particle
Compaction of Distillers’ Spent Grain density of the individual pellets. The experiments were repeat-
ed with five times for each moisture content and soluble
Different amounts of solubles were added to a known mass of concentration.
coarse grain fraction of DSG to obtain various desirable com-
positions of DSG with solubles. Different sets of DSG sam- Determination of Specific Heat
ples were prepared with 0, 10, 30, and 50 % solubles. A
control sample was represented by DSG without any solubles. Specific heat of the compacted DSG was determined using a
The moisture content of each mixture was also measured modified method of mixtures (Stitt and Kennedy 1945;
using standard air-oven drying method. Each of these samples Mannheim et al. 1957; Mohsenin 1980; Muir and
with different soluble concentration and coarse grain fraction Viravanichai 1972). The apparatus used in this study consisted
of DSG was gently mixed using a spatula for about 2 min and of two units: a cylindrical canister made of pure copper and an
dried in a laboratory oven at a lower temperature (60 ± 5 °C). insulated thermos flask (calorimeter) attached to a mechanical
A small amount of mixed sample (about 20 g) capable of shaker as shown in Fig. 1. The pellet was placed in a copper
producing 4–5 pellets where only prepared at a time to prevent canister before mixing with the heat transfer fluid (distilled
excessive drying of the sample when stored for longer time. water) within the 18/8 stainless steel thermos flask (Stanley,
Gentle mixing with the spatula would prevent the heating of Pacific Market International, USA) to avoid moisture absorp-
the sample. Lower temperature was used for drying the sam- tion and pellet deformation. The inner dimensions of the cy-
ple to ensure that the properties of the samples are not altered lindrical canister were made in such a way that it could ac-
as the stickiness of the distillers’ spent grains increases with com modate two D SG pell et s w ith a diameter of
 Food Bioprocess Technol

volume of the canister. The temperature of the copper canister

DAQ and the pellet was then checked once more before placing
Thermo-flask
them into the heat transfer fluid (distilled water) to ensure that
both were at the same temperature. Similarly, the polystyrene
Polystyrene Insulation base and the lid were dismantled from the mechanical shaker
and placed in an oven at 50 °C for about 5 min. This was done
to minimize the heat loss during the experiment. Hot water at
Thermocouple wire one of the three set temperatures was previously prepared and
placed inside the calorimeter resting inside the polystyrene
Copper canister with sample insulation of the mechanical shaker as shown in Fig. 1. The
water temperature was recorded 3 to 5 min prior to each ex-
Mechanical shaker
perimental trial to ensure that it remained constant at the de-
Fig. 1 Experimental setup for measuring the specific heat of DSG pellets
sired value. Prior to the start of each trial, the initial tempera-
(modified method of mixtures) ture of the canister and pellets (equal in temperature) and the
hot water inside the calorimeter were recorded. The canister
was then dropped into the calorimeter and the insulation lid
12.4 ± 0.05 mm and a height of 28.9 ± 0.05 mm. The copper was closed immediately. A mechanical shaker was employed
canister consisted of a 1.4 ± 0.02 mm thick wall and a lid with to shake the assembly at a constant speed. The shaker assem-
a slightly larger diameter than the canister. A soft silicon gas- bly was fabricated using a rotary shaker motor (Stern Pinball
ket was inserted between the lip of the canister and the top of Motors, USA) attached to a flat platform. The temperature of
the cap to ensure no penetration of fluid into the canister. the distilled water was recorded at every 1 s interval using a
The calorimeter was made of a double wall vacuum data acquisition unit. The temperature of the sample canister
thermo-flask with an insulated lid. A thin plastic tube with a unit and the water was allowed to reach equilibrium before
diameter of 0.8 ± 0.05 mm was placed through a hole drilled stopping the experiment. Separate sets of experiments were
into the lid of the thermos. A thin wire T-type thermocouple done in triplicates for each moisture content (25, 35, and
(Omega Engineering Inc., USA) with a diameter of 45 %), soluble concentration (0, 10, 30, and 50 % w/w), and
0.5 ± 0.03 mm was placed inside the plastic tube so that the water temperature (40, 60, and 80 °C). Experiments were done
tip of the thermocouple was approximately located in the cen- in triplicates for each factor (temperature, moisture content,
ter of the heat transfer fluid within the thermo-flask. The top of and soluble concentration) and for every level in each factor.
the plastic tube was sealed perfectly after placing the thermo- To calculate the calorimeter constant of the experimental
couple wire inside it ensuring that the temperature-sensing end setup (water equivalent of the calorimeter and its associated
of the thermocouple was only exposed to the heat transfer parts), separate trials were conducted with a known mass of
fluid and not the surrounding environment. Similarly, the open pure copper (3.5 mm diameters) and aluminum granules
end of the tube on the thermos flask lid was also covered with (1 mm diameter) with specific heat of 0.386 and 0.9 kJ/
thermal resistant tape in order to minimize the heat loss. The (kgK), respectively. Separate calibration tests were conducted
entire calorimeter, including the thermocouple, was then for each heat transfer fluid temperature, and corresponding
placed into a block of extruded polystyrene insulation and calorimetric constants were determined. To estimate the can-
covered with an insulated lid. A mechanical shaker at the base ister constant, a separate set of experiments was done with the
of the unit ensured that the fluid within the thermo-flask canister being empty for each temperature. Specific heat of
remained at a uniform temperature. The thermocouples were DSG pellets was calculated using Eq. 1, as described below,
then connected to a data acquisition system (DAQ). All data by considering the calorimetric constant at each heat transfer
obtained by the DAQ (Agilent—34970A, USA) was then fluid temperature and the corresponding canister constant.
transmitted through a USB interface to a computer. By doing
so, the temperature changes of the heat transfer fluid could be
recorded over time. ½ðM w −M th ÞC w ðT ew −T iw Þ−M can C can ðT ew −T ican Þ
Cp ¼ ð1Þ
After compaction, the pellet was carefully placed vertically M pellet ðT ew −T ican Þ
inside a copper canister. The initial temperature of the pellets
was checked using the thermocouple. The canister with the where Mth is the mass of water equivalent of the calorimeter
pellet was placed in an environmental chamber set at (calorimetric constant) in kg; Mpellet is the mass of DSG pellets
25 ± 1 °C. The copper canister and pellet reached equilibrium in kg; Mw is the mass of heat transfer fluid (water) in kg; Cw,
with the environmental chamber temperature after 3 to 5 min. Ccan, and Cp are the specific heat of water, copper canister, and
Also, the resting time inside the environmental chamber DSG pellets, respectively, in kJ/(kgK); and Tiw, Tew, and Tican
allowed the pellets to slightly expand and occupy the entire are the initial temperature of water, equilibrium temperature of
Food Bioprocess Technol

the water sample mixture, and initial temperature of copper the top cover and the copper rod in the lid stopper assem-
canister, respectively, in K. bly were aligned so that a small space of 7.2 ± 3 mm
The temperature of the copper canister and pellet was the between the plug and the rod was left for placing the test
same before the start of the experiment. The temperature his- sample (DSG pellet). Two thin T-type thermocouples were
tory of the water was recorded continuously at 1-s intervals installed about 2 mm below the surface of the rod and the
using data acquisition system. All recorded data was later used plug by securing them in a narrow hole drilled into the
for determination of specific heat of DSG pellets using Eq. 1. copper. In order to minimize the heat dissipation due to
The effect of thermal leakage and energy added by shaking the air gap around the sample, the lid of the thermo-flask
was also considered by using the graphical method as men- was covered with a styrofoam insulation cover.
tioned by Stitt and Kennedy (1945). This was accomplished The thermo-flask was filled with the heat transfer fluid
by calculating the temperature difference before and after (distilled water and crushed ice mixture/water with set tem-
mixing. A line with magnitude equal to this difference was perature). The temperature of the heat transfer fluid inside the
drawn approximately through the middle of the temperature thermo-flask was allowed to equilibrate to a constant value for
curve where a sudden deviation in temperature due to mixing about 10 min. The thermo-flask with the heat transfer fluid
was observed. The temperature corresponding to the end- was gently mixed with the help of a mechanical shaker.
points of this line was chosen as the initial and final tempera- C om pa ct ed c yl i nd r i ca l p el l et s o f ap pr ox i m a t e l y
tures of the mixing fluid. 12.5 ± 1 mm diameter and 7.2 ± 3 mm thickness were made
using the universal testing machine (Model 3366 Universal
Determination of Thermal Conductivity Testing Systems, Instron Corp., Norwood, MA). The pellet
was placed between the copper rod and the plug. The temper-
The thermal conductivity of cylindrical DSG pellets was ature history of both the top and the bottom of the pellet were
determined using the modified Fitch apparatus (Fitch recorded using two thermocouples. As per the quasi-steady
1935; Dutta et al. 1988; Zuritz et al. 1989; Rahman conduction heat transfer through the DSG pellet, the energy
1991), which was designed in the Department of balance in the copper plug and its analytical solution
Biosystems Engineering, University of Manitoba (Mohsenin 1980) are given by Eqs. 2 and 3:
(Fig. 2). It consists of a 540 mL the 18/8 stainless steel
Ak p ðT cr −T Þ dT
thermos flask (Stanley, Pacific Market International, ¼ mcp C p;c ð2Þ
USA) covered with a stopper made of extruded polysty- x dt
rene foam. A copper rod of 15 mm diameter and 115 mm with initial condition T = T0, when t = 0.
length was secured to a hole drilled in the center of the lid

so that one end of the copper rod is exactly in line with ðT cr −T 0 Þ Ak p t
ln ¼
 ð3Þ
the outer end of the polystyrene stopper. The lid stopper ðT cr −T Þ x mcp C p;c
assembly with the copper rod seals the thermo-flask. The
top cover for the Fitch apparatus was made with polysty- where A is the heat transfer area in m2; kp is the thermal
rene with a copper plug of the same diameter as that of conductivity of pellets in W/(mK); T, T0, and Tcr are the in-
the rod and 16 mm depth at its center. The copper plug in stantaneous temperature of the copper plug and sample, initial
temperature of copper plug, and temperature of heat source
(copper rod) in K; x is the thickness of the pellet in m; mcp is
Polystyrene cover
the mass of the copper plug in kg; and Cp,c is the specific heat
Copper plug of copper in J/kgK.
DSG pellet The thermal conductivity of the DSG pellets was calculated
Thermocouple wires from of the slope of the semi-log plot of temperature gradient
versus time and by equating the slope to the left-hand side of
DAQ ðAk p tÞ
the Eq. 3, i.e., x m C . The experiments were repeated five
ð cp p;c Þ
times for each moisture content (25, 35, 45 %), soluble con-
Copper rod
centration of DSG pellets (0, 10, 30, 50 % w/w), and heat
transfer fluid temperature (0, 40, 80 °C).
Insulated thermo-flask

Analysis of Data

Fig. 2 Experimental setup for measuring the thermal conductivity of The effect of moisture content and soluble concentration on
DSG pellets (modified Fitch apparatus) the thermo-physical properties of the DSG pellets at different
 Food Bioprocess Technol

temperatures was studied using Statistical Software (SAS 9.4). with different compressive pressure. They also reported
A method of least squares to fit a general linear model (GLM that the addition of 30 % solubles to the DSG sample
procedure) and a regression analysis were conducted on the increased the density of the pellet by about 10 % even
experimental data using SAS (version 9.4, SAS Institute Inc., at lower compressive pressure of 60.6 MPa. At the same
Cary, NC). A regression analysis with two independent vari- time, their results showed that the soluble concentration
ables (soluble concentration and moisture content) and three did not have an effect on pellet strength at higher com-
independent variables (soluble concentration, moisture con- pressive pressures (above 90 MPa). Statistical analysis of
tent, and temperature) was used for describing the effect of the experimental values using general linear model anal-
each factor on particle density, thermal conductivity, and spe- ysis shows that both the independent variables (soluble
cific heat of the DSG pellets, respectively. The predicted concentration and moisture content) have a significant ef-
values were compared with the experimental values by using fect (P < 0.001) on the particle density of the DSG pellets.
R2 value and mean relative percentage error (Jian et al. 2013) Since there was no significant interaction between the
determined using the following equation. independent variables, a simple linear regression analysis
was used to predict the particle density of the DSG pel-
1 X jPi −Ei j
n
lets. The prediction model for the particle density of DSG
MRP ¼  100 ð4Þ
n i¼1 E i pellets as a function of soluble concentration and moisture
content is shown in Eq. 5. The prediction agrees well with
where MRP is the mean relative percentage error; Pi and Ei are the experimental values with an R2 value of 0.86. The
the instantaneous predicted and experimental values of the mean relative percentage error for the predicted values
property tested; and n is the number of data points compared. was 3.4 %.

D ¼ 213:3 S−45:5 MC þ 1011:2 ð5Þ

Results and Discussion
where D is the particle density of the pellet in kg/m3; S is
the % solubles in a fraction; and MC is the moisture
Density and Expansion of DSG Pellets
content in %wb. As shown in the Fig. 3, the particle
density of the DSG pellets increased with an increase in
The particle density of DSG pellets with different soluble
soluble concentration supporting the conclusions of
concentrations (0, 10, 30, and 50 %) ranged between
Johnson et al. (2015). This increase in particle density
898.8–1073.4, 981.4–1057.7, 995.8–1114.92, and
with the increase in soluble concentration might be due
1039.7–1136.7 kg/m3, respectively, for the different mois-
to the superior binding properties of solubles, high
ture contents (Fig. 3). Increasing the percentage of solu-
amount of starch and protein, and its lower particle size
bles eliminated the micro-pores in the pellet. A similar
as reported by Mosqueda et al. (2013) and Johnson et al.
range of values (i.e., 940–1090 kg/m3) was also reported
(2014). The micro-porosity of the pellet due to the small
by Johnson et al. (2013) for the DSG pellets compacted
air pockets within the pellet will also affect the density of
1200 the pellet. The increase in soluble concentration in dried
1150 DSG granules leads to a decreased porosity (Mosqueda
et al. 2014) because of the fact that the bulk porosity of
1100
Density (kg/m3)

dried solubles were about 10 % lesser than that of the

1050
dried DSG powder (Zielinska 2016). A heterogeneous
1000 mixture of particles with different sizes has greater bond-
950 ing because of the higher number of inter-particle bonds
900 formed during compaction (Payne 1978). In contrast, the
850
moisture content has a negative effect on the particle den-
sity of the pellets (Serrano et al. 2011; Samuelsson et al.
800
0 10 20 30 40 50 60 2012) because higher moisture content samples tend to
% Solubles expand quickly than the lower moisture content samples
(Mohseni and Zaske 1976). A similar negative correlation
pred(25) pred(35) pred(45)
between the hardness and the moisture content of the
25% MC 35% MC 45% MC
DSG pellets was reported by Johnson et al. (2015). The
Fig. 3 Effect of moisture content and soluble concentration on the strength and durability of the densified products increased
particle density of the DSG pellets (where pred(25), pred(35), and
pred(45) show the predicted density of pellets with 25, 35, and 45 % with increasing moisture content until an optimum is
moisture content, respectively) reached (Kaliyan and Morey 2009). The DSG pellets at
Food Bioprocess Technol

higher moisture content (>25 %) have more moisture model for the prediction of specific heat capacity of DSG pellets
trapped in the inter-particle spaces, and these water mol- is shown in Eq. 6.
ecules prevent the complete compression and release of
C p ¼ 0:019 MC þ 0:011 S þ 0:023 T þ 0:2716 ð6Þ
natural binders from solid particles owing to the
incompressibility of water (Pickard et al. 1961).
where Cp is the specific heat of DSG pellets in kJ/(kgK); S
Specific Heat of DSG Pellets is the % solubles in a fraction; MC is the moisture content in
%wb.; and T is the temperature of the heat transfer fluid in °C.
The average specific heat of DSG pellets for temperatures of 40, The specific heat capacity of the DSG pellets increases
60, and 80 °C ranged between 1.76–2.62, 1.85–3.32, and 2.67– linearly with an increase in moisture content, temperature,
3.47 kJ/(kgK), respectively. The specific heat values of these and soluble concentration (Fig. 4a–d). A similar trend in the
high moisture DSG pellets (25–45 % moisture content) were specific heat with an increase in temperature and moisture
found to be insignificantly higher than the dried distiller’s gran- content was also reported by Jian et al. (2013) for canola
ules (1:1 ratio of wheat and corn) which were in the range of 1.7– seeds and by Ashtiani et al. (2014) for sesame seeds. The
2.6 kJ/(kgK) (Zielinska 2016). The higher values of specific heat mean relative percentage error and R2 value for the developed
were mainly attributed by the higher moisture content of DSG model were 5.5 and 0.86, respectively. A comparison of pre-
pellets since water has higher specific heat than dried grains. The dicted and experimental values of specific heat capacity of the
experimental values of specific heat capacities of the DSG pellets DSG pellets shows that the general linear model as described
with different percentage of solubles (0, 10, 30, 50 %) at different by Eq. 6 has better predictability of the dependent variable
moisture content (25, 35, 45 %) and at three different tempera- (specific heat).
tures of heat transfer fluid (40, 60, 80 °C) are shown in Fig. 4a–d.
As per the results of statistical analysis, all the three independent
variables (moisture content, soluble concentration, and tempera- Thermal Conductivity of DSG Pellets
ture of heat transfer fluid) have a significant effect (P < 0.001) on
the dependent variable (specific heat capacity). The interaction Thermal conductivity of a material depends on the structural
terms between the independent variables were neglected from the anatomy of the sample, as well as its composition. Hence, it is
model since they were insignificant at P > 0.05. The regression highly challenging to get an accurate measurement of thermal

Fig. 4 a–d Effect of moisture a 4

b 4
content (25, 35, and 45 %) and
temperature (40, 60, and 80 °C) 3.5 3.5
on the specific heat capacity of
DSG pellets with different 3 3
Cp (kJ/kgK)
Cp (kJ/kgK)

percentage of solubles (where a,

b, c, and d represent the graphs 2.5 2.5
for 0, 10, 30, and 50 % of
2 2
solubles, respectively and
pred(40), pred(60), and pred(80) 1.5 1.5
shows the predicted specific heat
of pellets at 40, 60, and 80 °C 1 1
temperatures, respectively) 20 30 40 50 20 30 40 50
% MC % MC

c 4
d 4

3.5 3.5

3 3
Cp (kJ/kgK)
Cp (kJ/kgK)

2.5 2.5

2 2

1.5 1.5

1 1
20 30 40 50 20 30 40 50
% MC % MC
 Food Bioprocess Technol

conductivity of a sample when compared to the other thermal equation of best fit was determined with the help of multiple
properties (Rahman 1991). The variation of thermal conduc- regression analysis avoiding the interaction terms since they
tivity of DSG pellets at different moisture contents (25, 35, were found to be insignificant. The prediction equation for
45 %) and temperature of heat transfer fluid (0, 40, 80 °C) for thermal conductivity as a function of temperature, moisture
each percentage composition of solubles (0, 10, 30, 50 %) is content, and soluble concentration is given by Eq. 7.
shown in Fig. 5a–d. The values of thermal conductivity of the
DSG pellets with different soluble concentration and moisture k p ¼ 0:0012 T þ 0:016 S þ 0:0064 MC−0:0022 ð7Þ
contents for the three different temperatures of heat transfer
fluid (0, 40, 80 °C) were 0.17–0.30, 0.20–0.35, and 0.22–
0.42 W/(mK), respectively. These thermal conductivity values where kp is the thermal conductivity of the DSG pellet in
were higher than the values reported for dried DSG powder W/(mK); S is the soluble concentration in fraction; MC is the
which was in the range of 0.04–0.05 W/(mK) (Mosqueda moisture content in %wb.; and T is the temperature of the heat
et al. 2014) because of the fact that the these DSG pellets were transfer fluid in °C. The thermal conductivity of the DSG
at significantly higher moisture content (i.e., 25–45 % mois- pellets increases with increase in moisture content, soluble
ture content). The presently reported values were found to be concentration, and temperature (Fig. 5a–d). Similar results
closer to the thermal conductivity values of corn kernels showing a positive effect of temperature (Jancazak 2013)
which were reported to be in the range of 0.15 to 0.35 W/ and moisture content on the thermal conductivity of material
(mK) for a moisture range of 10–35 % (Kustermann et al. were reported by Elansari and Hobani (2009) and Modi et al.
1981). At each moisture content and temperature, there was (2014). The increase in thermal conductivity values with an
a gradual increase in the value of thermal conductivity with an increase in moisture content is attributed to the higher conduc-
increase in soluble concentration up to 30 %, beyond which tive heat transfer of water molecules (Sreenarayanan and
the value seems to be approximately constant. All the three Chattopadhyay 1986). Emami et al. (2007) reported that the
independent variables (moisture content, soluble concentra- thermal conductivity is greatly influenced by the atomic activ-
tion, and temperature) have a significant effect on the depen- ity inside the material. Hence, at a higher temperature, there is
dent variable (thermal conductivity) with their P values a higher atomic activity which improves the ability of the
<0.001, <0.010, and <0.001, respectively. The mathematical material to transfer heat energy. A similar trend in thermal

Fig. 5 a–d Effect of moisture a 0.5

b
content (25, 35, and 45 %) and 0.5
temperature (0, 40, and 80 °C) on
the thermal conductivity of DSG 0.4 0.4
pellets with different soluble
kp (W/m K)

concentrations (where a, b, c, and

kp (W/m K)

d represents the graphs for 0, 10, 0.3 0.3

30, and 50 % of solubles,
respectively and pred(0),
0.2 0.2
pred(60), and pred(80) show the
predicted specific heat of pellets
at 40, 60, and 80 °C temperatures, 0.1 0.1
respectively) 20 30 40 50 20 30 40 50
% MC % MC
c 0.5
d 0.5

0.4 0.4
kp (W/m K)

kp (W/m K)

0.3 0.3

0.2 0.2

0.1 0.1
20 30 40 50 20 30 40 50
% MC % MC
Food Bioprocess Technol

conductivity with increase in temperature was reported by (40 °C). There was an increase in the specific heat value of
Mahapatra et al. (2011) and Jancazak (2013). The increase DSG pellets (9–20 %) with an increase in soluble concentra-
in thermal conductivity values of DSG pellets with the in- tion for different temperatures of heat transfer fluid.
crease in soluble concentration can be explained by the corre- The values of thermal conductivity also showed a positive
lation between soluble concentration and particle density as correlation with the moisture content, temperature, and solu-
mentioned in Eq. 7. The DSG pellets with a higher percentage ble concentration. The average value of thermal conductivity
of solubles have a higher density, which decreases the air of DSG pellets ranged between 0.17–0.42 W/(mK) for differ-
pockets inside the pellet. Since air is a poor conductor of heat ent moisture contents (25, 35, 45 %) and soluble concentration
than solids, air pockets inside the material (i.e., higher poros- (0,10, 30, 50 %). The moisture content of a pellet was pre-
ity) lead to decreased thermal conductivity and vice versa. dominantly affected by the thermal conductivity than other
Hence, at constant moisture content, higher density pellets independent variables. Multiple linear regression equations
will have higher thermal conductivity. Also, the high amount for all the three properties show a good agreement with the
of starch and protein in the solubles increases the thermal experimental data. Hence, these equations can be used for
conductivity as starch and protein have higher thermal con- predicting the values of these properties at any moisture con-
ductivities than cellulosic fiber (Choi and Okos 2004). The tent, soluble concentration, and temperature with appreciable
prediction equation for thermal conductivity was validated accuracy. These equations can be used as important tools to
with the experimental data by using R2 value and mean rela- predict thermo-physical properties of DSG pellets, as these
tive percentage error. The value of thermal conductivity for properties are essential for the numerical modeling of heat
different moisture content, temperature, and soluble concen- and mass transfer mechanisms. Also, these properties will
tration is well predicted by the multiple regression equation serve as essential data for optimization of different processing
(Eq. 7) with an R2 value of 0.89. The mean relative percentage and storage techniques and for the design of suitable drying
error calculated for the prediction equation was 6.7 %, which system for distillers’ spent grains.
indicates good agreement between the experimental and pre-
dicted values of thermal conductivities. Acknowledgments The authors acknowledge the Natural Sciences and
Engineering Research Council of Canada and University of Manitoba
Graduate Fellowship for their financial support. Thanks are also due to
NSERC-funded summer research assistants, Jennifer Pieniuta, Clifford
Conclusion Dueck, and Craig Heppner, for their technical assistance.

Thermo-physical properties are the key input parameters re-

quired for mathematical modeling of heat and mass transfer References
phenomena of the drying process. Therefore, thermal conduc-
tivity, specific heat, and particle density were determined as a AACC Method 44-15A (2000). Moisture-air-oven methods. St. Paul:
function of moisture content, soluble concentration, and tem- Approved Methods of the American Association of Cereal
Chemists.
perature using standard laboratory methods. The particle den-
Ashtiani, S., Emadi, B., Sanaeimoghadam, A., & Aghkhani, M. (2014).
sity of DSG pellets was in the range of 898.8–1136.7 kg/m3 Effect of moisture content and temperature on thermal behaviour of
for moisture content ranging from 25 to 45 % and soluble sesame seed. The Annals of the University Dunarea de Jos of Galati
concentration ranging from 0 to 50 %. The particle density Fascicle VI—Food Technology, 38(1), 87–103.
of DSG pellets increased by 8.3, 10.8, and 19 % with an Bourassa, J., Ramachandran, R. P., Paliwal, J., & Cenkowski, S. (2015).
Drying characteristics and moisture diffusivity of distillers’ spent grains
increase in soluble concentration (0–50 %) for 25, 35, and dried in superheated steam. Drying Technology, 33, 2012–2018.
45 % moisture content, respectively. But the average particle Choi, Y., & Okos, M. R. (2004). Thermal conductivity of foods. In D. R.
density of the DSG pellets was found to decrease (10–18 %) Heldman (Ed.), Encyclopedia of agricultural, food, and biological
with an increase in moisture content. The superior binding engineering (pp. 1004–1010). New York: Marcel Dekker Inc.
Defraeye, T. (2014). Advanced computational modelling for drying pro-
properties of protein and starch and finer particle size of sol- cesses—a review. Applied Energy, 131, 323–344.
ubles explain the increase in density with an increase in solu- Drouzas, A. E., Maroulis, Z. B., Karathanos, V. T., & Saravacos, G. D.
ble concentration. (1991). Direct and indirect determination of the effective thermal
Specific heat of the DSG pellets was positively correlated diffusivity of granular starch. Journal of Food Engineering., 13(2),
91–101.
with the moisture content, soluble concentration, and temper- Dutta, K. S., Nema, V. K., & Bhardwaj, R. K. (1988). Thermal properties of
ature. The average value of the specific heat of DSG pellets gram. Journal of Agricultural Engineering Research, 39(4), 269–275.
ranged between 1.76 and 3.47 kJ/(kgK) for different moisture Elansari, A. M., & Hobani, A. I. (2009). Effect of temperature and mois-
contents (25, 35, 45 %) and soluble concentration (0,10, 30, ture content on thermal conductivity of four types of meat.
International Journal of Food Properties, 12, 308–315.
50 %). However, at higher temperatures (80 °C), the effect of Emami, S., Tabil, L. G., & Tyler, R. T. (2007). Thermal properties of
moisture content on the specific heat capacity of the DSG chickpea flour, isolated chickpea starch, and isolated chickpea pro-
pellets was minimal when compared to at lower temperatures tein. Transactions of the ASABE, 50(2), 597–604.
 Food Bioprocess Technol

Erriguible, A., Bernada, P., Couture, F., & Roques, N. (2006). Simulation Mosqueda, M. R., Tabil, L. G., & Muthukumarappan, K. (2013). Effects
of superheated steam drying from coupling models. Drying of condensed distillers solubles and drying temperature on the
Technology, 24, 941–951. physico-chemical characteristics of laboratory-prepared wheat dis-
Fitch, A. L. (1935). A new thermal conductivity apparatus. American tillers grain with solubles. International Journal of Agricultural and
Journal of Physics, 3(3), 135–136. Biological Engineering, 6(2), 73–85.
Hamawand, I. (2011). Effect of colloidal particles associated with the Mosqueda, M. R., Tabil, L. G., & Muthukumarappan, K. (2014).
liquid bridge in sticking during drying in superheated steam. Physico-chemical characteristics of wheat distillers dried grain with
International Journal of Engineering, 24(2), 119–126. solubles sourced from a Saskatchewan ethanol plant. Canadian
Hamawand, I. (2013). Drying steps under superheated steam: a review Biosystems Engineering, 56, 3.1–3.15.
and modelling. Journal of Energy and Environment Research, 3(2), Muir, W. E., & Viravanichai, S. (1972). Specific heat of wheat. Journal of
107–125. Agricultural Engineering Research, 17(4), 338–342.
Hamawand, I., Yusaf, T., & Bennett, J. (2014). Study and modelling Nesvadba, P. (1982). Methods for the measurement of thermal conduc-
drying of banana slices under superheated steam. Asia-Pacific tivity and diffusivity of foodstuffs. Journal of Food Engineering,
Journal of Chemical Engineering, 9, 591–603. 1(2), 93–113.
H u s k y E n e r g y ( 2 0 1 6 ) . h t t p : / / w w w. h u s k y e n e r g y. Pakowski, Z., & Adamski, R. (2011). On prediction of the drying rate in
com/huskyproducts/canada.asp. Accessed on May and 1 Aug 2016. superheated steam drying process. Drying Technology, 29, 1492–1498.
Iyota, H., Nishimura, N., Yoshida, M., & Nomura (2007). Simulation of Payne, J. D. (1978). Improving quality of pellet feeds. Milling Feed and
superheated steam drying considering initial steam condensation. Fertilizer, 162(5), 34–41.
Drying Technology, 19(7), 1425–1440. Pickard, G. E., Roll, W. M., & Ramser, J. H. (1961). Fundamentals of hay
Jancazak, D. (2013). The effect of temperature, composition and phase of wafering. Transactions of ASABE, 4(1), 65–68.
the composting process on the thermal conductivity of the substrate. Rahman, S. (1991). Evaluation of precision of the modified Fitch method
Ecological Engineering Journal, 61, 354–357. for thermal conductivity measurement of foods. Journal of Food
Jian, F., Jayas, D. S., & White, N. D. G. (2013). Specific heat, thermal Engineering, 14, 71–82.
diffusivity, and bulk density of genetically modified canola with Samuelsson, R., Larsson, S., Thyrel, M., & Lestander, T. A. (2012).
high oil content at different moisture contents, temperatures, and Moisture content and storage time influence the binding mecha-
storage times. Transactions of the ASABE, 56(3), 1077–1083. nisms in biofuel wood pellets. Applied Energy, 99, 109–115.
Serrano, C., Monedero, E., Lapuerta, M., & Portero, H. (2011). Effect of
Johnson, P., Cenkowski, S., & Paliwal, J. (2011). Bulk density and angle of
moisture content, particle size and pine addition on quality parame-
repose of distiller ’ s spent grain under different drying methods and
ters of barley straw pellets. Fuel Processing Technology, 92, 699–
soluble concentrations. Winnipeg, Manitoba: CSBE/SCGAB 2011
706.
Annual Conference Inn at the Forks 10-13 July 2011. Paper No.
Sousa, S. C. L., Silva, J. P., Ramos, A. M. M., & Simões, R. M. S. (2007).
CSBE11-318.
Pulping and papermaking potential of brewery’s spent grain.
Johnson, P., Paliwal, J., & Cenkowski, S. (2013). Superheated steam
Cellulose Chemistry and Technology, 41(2), 183–191.
drying characteristics of single cylindrical compacts produced from
Sreenarayanan, V. V., & Chattopadhyay, P. K. (1986). Thermal conduc-
wet distiller’s spent grain. In Proceedings of CSBE/SCGAB 2013
tivity and diffusivity of rice bran. Journal of Agricultural
Annual Conference, Saskatchewan 7–10 July 2013. Paper No.
Engineering Research., 34, 115–121.
CSBE13–11.
Stitt, F., & Kennedy, E. K. (1945). Specific heats of dehydrated vegeta-
Johnson, P., Paliwal, J., & Cenkowski, S. (2014). Analysis of the disin-
bles and egg powder. Food Research International, 10(5), 426–436.
tegration of distiller’s spent grain compacts as affected by drying in
Stroem, L. K., Desai, D. K., & Hoadley, A. F. A. (2009). Superheated
superheated steam. Drying Technology, 32, 1060–1070.
steam drying of brewer’s spent grain in a rotary drum. Advanced
Johnson, P., Paliwal, J., & Cenkowski, S. (2015). Effect of solubles on Powder Technology, 20, 240–244.
disintegration of distiller’s spent grain compacts during superheated Suvarnakuta, P., Devahastin, S., & Mujumdar, A. S. (2007). A mathemat-
steam drying. Drying Technology, 33(6), 671–683. ical model for low pressure superheated steam drying of biomaterial.
Kaliyan, N., & Morey, R. V. (2009). Factors affecting strength and dura- Chemical Engineering and Processing, 46(7), 675–683.
bility of densified biomass products. Biomass and Bioenergy, 33(3), Thomas, M., Van-Vliet, T., & Van-der-Poel, A. F. B. (1998). Physical
337–359. quality of pelleted animal feed. 3. Contribution of feedstuff compo-
Kustermann, M., Scherer, R., & Kutzbach, H. D. (1981). Thermal con- nents. Animal Feed Science and Technology, 70(1–2), 59–78.
ductivity and diffusivity of shelled corn and grain. Journal of Food Tumuluru, J. S., Tabil, L., Opoku, A., Mosqueda, M. R., & Fadeyi, O.
Process Engineering, 4, 137–153. (2010). Effect of process variables on the quality characteristics of
Mahapatra, A. K., Lan, Y., & Harris, D. L. (2011). Influence of moisture pelleted wheat distiller’s dried grains with solubles. Biosystems
content and temperature on thermal conductivity and thermal diffusivity Engineering, 105(4), 466–475.
of rice flours. International Journal of Food Properties, 14, 675–683. Waszkielis, K. M., Wronowski, R., Chlebus, W., Bialobrzewski, I., Dach,
Mani, S., Tabil, L. G., & Sokhansanj, S. (2003). An overview of com- J., Pilarski, K., & Jancazak, D. (2013). The effect of temperature,
paction of biomass grinds. Powder Handling and Processing, 15(2), composition and phase of the composting process on the thermal
160–168. conductivity of the substrate. Ecological Engineering Journal, 61,
Mannheim, H. C., Steinberg, M. P., Nelson, A. I., & Kendall, T. W. (1957). 354–357.
The heat content of bread. Food Technology, 11(7), 394–388. Waszkielis, K. M., Białobrzewski, I., Nowak, K. W., Dzadz, & Dach, J.
Modi, S. K., Durga, P. B., & Basavaraj, M. (2014). Effect of moisture (2014). Determination of the thermal conductivity of composted
content and temperature on thermal conductivity of Psidium material. Measurement: Journal of the International Measurement
guajava L. by line heat source method (transient analysis). Confederation, 58, 441–447.
International Journal of Heat and Mass Transfer, 78, 354–359. Xiros, C., Topakas, E., Katapodis, P., & Christakopoulos, P. (2008).
Mohseni, N., & Zaske, J. (1976). Stress relaxation and energy require- Hydrolysis and fermentation of brewer’s spent grain by
ments in compaction of unconsolidated materials. Journal of Neurospora crassa. Bioresource Technology, 99(13), 5427–5435.
Agricultural Engineering Research, 21, 193–205. Yang, W., Sokhansanj, S., Tang, J., & Winter, P. (2002). Determination of
Mohsenin, N. N. (1980). Thermal properties of foods and agricultural thermal conductivity, specific heat and thermal diffusivity of borage
materials. New York: Gordon and Breach Science Publishers. seeds. Biosystems Engineering, 82(2), 169–176.
Food Bioprocess Technol

Zielinska, M. (2016). Thermophysical properties of laboratory-prepared Zielinska, M., Cenkowski, S., & Markowski, M. (2009). Superheated
corn/wheat dried distiller’s grains and dried distillers solubles steam drying of distillers’ spent grains on a single inert particle.
dehydrated with superheated steam and hot air. Drying Drying Technology, 27(12), 1279–1285.
Technology, 34(10), 1147–1161. Zuritz, C. A., Sastry, S. K., McCoy, S. C., Murakami, E. G., & Blaisdeh,
Zielinska, M., & Cenkowski, S. (2012). Superheated steam drying char- J. L. (1989). A modified Fitch device for measuring the thermal
acteristic and moisture diffusivity of distillers’ wet grains and con- conductivity of small food particles. Transactions of ASAE, 32(2),
densed distillers’ solubles. Journal of Food Engineering, 109(3), 711–718.
627–634.