Journal of Stored Products Research 59 (2014) 254e259

Contents lists available at ScienceDirect

Journal of Stored Products Research

journal homepage: www.elsevier.com/locate/jspr

Experimental study and discrete element method modeling of

temperature distributions in rapeseed stored in a model bin
R. Rusinek*, R. Koby1ka
Institute of Agrophysics, Polish Academy of Sciences, ul. Doswiadczalna 4, 20-290 Lublin, Poland

a r t i c l e i n f o a b s t r a c t

Article history: Rapeseed is one of the major sources of vegetable oil. Improving processing and storage conditions is
Accepted 22 March 2014 very important in order to get high quality product. Rapeseed stored in silos is at constant risk of
Available online 14 April 2014 deterioration of quality. The self-heating phenomenon is the main reason for deterioration during
storage. Therefore, the aim of the present work was to describe the temperature distribution during this
Keywords: phenomenon and determine whether the self-heating area in a silo could be predicted. The discrete
Temperature distribution
element method (DEM) was used to predict temperature distributions in rapeseed in a cylindrical storage
Self-heating
bin and to describe the self-heating process of rapeseed. Model validation was carried out by comparing
Granular assembly
Discrete element method
the results of the model with experimentally measured grain temperatures at different points in a model
Silo silo. The predicted and measured temperatures were found to be in good agreement. DEM may be useful
for predicting temperature distributions in a silo and can describe the self-heating phenomenon.
2014 Elsevier Ltd. All rights reserved.

1. Introduction temperature for a long time; on the other hand, it is a drawback

The self-heating phenomenon is a well-known reason for the
deterioration of the quality of freshly harvested seeds. Some factors
that inuence self-heating are high moisture content, a high tem-
perature of the harvested seeds and excessive metabolic activity of
insects living in the crop (Mills, 1989; Jia et al., 2001; Arbogast et al.,
2004: Ramirez et al., 2010). Self-heating is a destructive process
(Gawrysiak-Witulska et al., 2011) that is, unfortunately, rather
common in practice during the storage of grain and seeds. The
physical consequences of self-heating include the formation of an
area of increased temperature, caking of the seeds, and cessation of
the outow of the material from the silo.
One method to reduce the risk of deterioration of stored grain
is to monitor the temperature in the silo and to provide aeration
when necessary. The ability to detect this phenomenon depends on
the thermal diffusivity of the bulk material. For example, rapeseed,
according to the studies of Moysey et al. (1977), has a poor thermal
diffusivity, ranging from 8.7
108 to 10.14
108 m2/s (the ther-
mal diffusivities of some other bulk materials, according to
ASAE (2008), are as follows: corn, 10.22
108 m2/s; rice,
10.99
108 m2/s; and hard wheat, 11.5
108 m2/s). On the one
hand, this is an advantage because precooled seeds hold their low
* Corresponding author. Tel.: 48 81 744 50 61, fax: 48 81 744 50 67.
E-mail address: rrusinek@ipan.lublin.pl (R. Rusinek).
because of the low level of heat transmission. The poor migration of
heat accelerates the dynamics of the development of self-heating
and makes self-heating difcult to detect and locate (Ileleji et al.,
2006). Heat transmission between seeds may occur in three
ways: by seed-to-seed contact, by convection of the air in the
intergranular spaces, and by radiation between the surfaces of
particles or across voids (Vargas and McCarthy, 2001). Unless there
is forced air movement in a particular direction, the temperature
eld in the case of heat exchange by convection is governed by the
gravitational eld, heated air tending to move upwards. Two
important determinants of convectional heat exchange in seeds are
the bulk density of the medium and the moisture content. The
density is inuenced by the method by which the container was
lled (Horabik and Rusinek, 2001), the moisture content of the
seeds (Wia cek and Molenda, 2011), and the geometry and the
structure of the seeds (qukaszuk et al., 2009). Lower porosity causes
a higher resistance to airow, but it increases heat transfer (and
therefore the thermal conductivity) by contact. Some authors (e.g.,
Smith and Sokhansanj, 1990; Iguaz et al., 2004) claim that heat
transmission by seed-to-seed contact dominates in small-grain
cereals such as wheat, rice, and rapeseed, whereas the effect of
convection is more important for larger particles such as apples and
potatoes. According to Muramatsu et al. (2007), the thermal con-
ductivity of brown rice kernels increases linearly with an increase
in the moisture content; similar results were obtained for cumin
seed by Singh and Goswami (2000).

http://dx.doi.org/10.1016/j.jspr.2014.03.009
0022-474X/ 2014 Elsevier Ltd. All rights reserved.
 R. Rusinek, R. Kobyka / Journal of Stored Products Research 59 (2014) 254e259
In practice, the safe storage of seeds requires their ventilation
(Rusinek et al., 2012). Safety and measurement equipment is used
to control the conditions of storage, and the condition monitoring is
usually based upon measurements of the temperature in the
intergranular space (Jia et al., 2001). For the storage of dry matter,
ISO 4112:1990 (ISO, 1990) recommends that temperature sensors
should be arranged geometrically in a silo in the shape of a grid. The
present work is centered on the description of temperature in-
creases that indicate self-heating.
Testing destructive phenomena such as self-heating in labora-
tory experiments is demanding of time and resources. However, the
development of computer technology and numerical algorithms in
recent decades provides us with an opportunity to devise a more
convenient alternative. Numerical modeling of heat transfer inside
a mass of material is usually based on assumptions of continuum
behavior and homogeneity, using, for example, the nite element
method or computational uid dynamics (a basic description of
such models can be found, for example, in Wang and Sun, 2003).
Recently, there have been a number of numerical simulations of
heat transfer, pressure distributions, and airows in grain silos (Xu
and Burfoot, 1999; Jia et al., 2000a,b; Moran et al., 2006; Juan et al.,
2006; Chourasia and Goswami, 2007). However, although such
approaches may give accurate time-averaged results, they do not
take account of interparticle interactions in the granular bed, nor of
the behavior of distinct particles (Chaudhuri et al., 2006).
In recent years, one of the more popular approaches for dealing
with granular media has been the discrete (or distinct) element
method (DEM). This was originally proposed by Cundall and Strack
(1979), but has been extensively developed and used over the last
20 years. The DEM was originally developed to simulate the me-
chanical response of a granular bed. However, it can easily be
extended to study other physical properties such as heat transfer (Li
and Mason, 2000; Chaudhuri et al., 2006; Kwapinska et al., 2008;
Nguyen et al., 2009).
In the work reported here, the DEM software package LIGGGHTS,
developed by Kloss et al. (2012) was used. The aim was to evaluate
the potential application of DEM to predicting phenomena such as
self-heating. The objectives of this study were as follows.
 to determine the progress of the self-heating phenomenon with
time and to nd out whether, and if so how fast, a grid of
temperature sensors situated in the bed can detect an area of
self-heating;
 to adapt the DEM model to predict the temperature distribution
in a rapeseed storage bin and to describe the self-heating phe-
nomenon; and
 to validate the simulation model based on experimental data.
2. Methods
2.1. Measuring system
The system was composed of 20 sensors. Each sensor measured
the temperature and relative humidity (r.h.) of the air in the
intergranular space. The temperature and r.h. values were trans-
mitted over single wires to a control module. All data e sensor
numbers, temperature and r.h. values, measurement numbers, and
additional information e were archived in the memory of the
control module. After the test, control module was connected to the
PC to copy the data. Each of the humidity and temperature sensors
was built into a square plastic box with a size of 10
10
15 mm.
The casing had 1 mm diameter holes in it, sufcient to ensure a free
ow of air. This paper presents the results of the temperature
measurements only.
255
2.2. Model silo
A measurement station was developed for the study presented
here, composed of a model cylindrical silo with a capacity of 3.85 m3
(w2500 kg), which was lled with winter rapeseed of the variety
Suzy. A cylindrical container with a height H and diameter D of 1.7 m
(so that H/D 1) was constructed from steel sheeting. The surface of
the seedbed was covered with plywood. The container was lled
centrally, directly from a screw conveyor. The sensors were placed
every 0.15 m in one plane in both vertical and horizontal directions.
In the rst test (Fig. 1a), the sensor network was adjusted so as to
place the intersection of the horizontal and vertical axes at the
center of the model container, where the most rapid progression of
the heating phenomenon might be expected to occur. In this
experiment, the silo was lled with three layers of seeds, each 0.5 m
high: the rst layer, at the bottom, had a moisture content of 7%
(w.b.), the second layer had a moisture content of 12.7%, and the last
layer, at the top, had a moisture content of 7% again. The moisture
content of the seeds was determined by oven drying at 103  1  C
for 72 h (ASAE, 2006).
The second test (Fig. 1b) was performed with an internal heat
source. An electric heater (0.17 kW, 230 V, consisting of a heating
mat 0.8 m long and 0.5 m wide, which generated a maximum
temperature of 40  C) was installed at the bottom of the container.
The temperature of the heater was increased in accordance with
the typical progression of self-heating: for the rst six days, the
increase was about 1  C per day, and for subsequent days, 2  C per
day. A digital programmer and temperature sensors in the bulk
material were used to control the increase of the temperature in the
silo. Three replications of the experiment were performed. Tem-
perature in the laboratory was controlled at 21  C.
2.3. Numerical DEM simulation
The DEM algorithm is based on four main steps, namely:
 detection of particleeparticle and particleewall contacts;
 application of forces (contact, gravity, etc.) to each particle;
 constructing and solving a set of differential equations based on
Newtons second law of motion for every particle;
 updating the particle parameters (e.g., positions and velocities)
after a time step and repeating the whole cycle.
In the most basic form of the three-dimensional DEM, the sys-
tem of equations in the third step consists of three equations for the
spatial coordinates and three for the rotational coordinates. How-
ever, this system can easily be extended with additional equations,
describing other particle properties transferred during contact, for
example charge and temperature.
The mathematical model used in the present work for calcu-
lating the contact forces between colliding particles was based on
Hertzs theory, extended with a tangential force based on the work
of Mindlin and Deresiewicz (see, e.g., Di Renzo and Di Maio, 2004).
In the case of a static bed discussed here the normal component of
the contact force is calculated as follows:
4 * p* 3=2
N E R dij ; (1)
3
where the reduced parameters representing the Youngs modulus
E*, and particle radius R* are calculated as follows:
 
 
1 1  v2i 1  v2j
; (2)
E* Ei Ej
256 R. Rusinek, R. Kobyka / Journal of Stored Products Research 59 (2014) 254e259

Fig. 1. Model silo: (a) test 1, self-heating; (b) test 2, with internal source (0, 1, 2,.,19, temperature sensors; distance between sensors 15 cm).

A numerical experiment was performed in a cylindrical bin

1 1 1 (Fig. 2) with a diameter D and height H equal to 0.068 m (i.e., H/
; (3)
R* Ri Rj D 1), lled with 33,000 spherical particles having the material
parameters of rapeseed (Wia cek and Molenda, 2011; Timbers,
Also, n is Poissons ratio, d is the overlap of spheres in contact
1975): particle radius R, 0.001 m; density r, 1050 kg/m3; Youngs
(d Ri Rj  d > 0, where d is the distance between the centers of
modulus E, 410 MPa; Poissons ratio n, 0.17; coefcient of restitution
spheres i and j), The indices i and j are the numbers labeling the
e, 0.4; coefcient of friction m, 0.379; thermal capacity c, 99.185 J/
particles (the particle id) (i s j).
kg K; and thermal conductivity k, 1.2823 J/K s m. The at bottom of
The model used for heat transmission between two particles
the container was split into two equal parts (see Fig. 2); one part
was the same as the one used by Chaudhuri et al. (2006). An
was used as a heat source (the hatched part of the bottom), and the
evaluation of this model with all the assumptions behind it can be
other was left at the environmental temperature (21  C).
found, for example, in Vargas and McCarthy (2001) and the refer-
Boundary conditions were assumed at constant temperature to
ences cited therein. The rate of heat conduction Q in this model is
reect laboratory environment during the testing. Because of the
calculated from every single particleeparticle and particleewall
highly different scales between simulation and an experimental
contact (the heat transmission through pores, for example by
setup, particles in heated area very close to the cold wall gain more
convection and radiation, is assumed to be negligible, especially for
heat from other particles because of more contact points than they
storage without aeration e Smith and Sokhansanj, 1990) using the
are able to lose through single contact with silo wall. The constant
following scheme (Chaudhuri et al., 2006):
temperature of the walls was stabilization factor rather than an
element able to carry the heat.
new
Qij hc;ij DTij (4)
Having in mind an equation describing thermal energy transfer
Here, T is the temperature, and the heat transfer coefcient hc is inside the continuous material:
expressed as

4ki kj q
hc;ij Acontact;ij ; (5)
ki kj

where the coefcients k are the thermal conductivities of the par-

ticles (in units of J/K s m), and A is the contact area (in m2) between
two particles (assumed to be spheres), calculated on the basis of
Hertzs theory (Dintwa et al., 2008):
!2=3
3NR*
Acontact;ij p : (6)
4E*

The temperature change is calculated according to the following

formula:

dTi X
old
mi ci Qij Qsource;i (7)
dt contact;ij

where mi is the mass of the particle, ci is the specic thermal ca-

pacity (in J/kg K), and Qsource,i is the rate of heat generation due to
heat sources. Fig. 2. Numerical setup. The hatched area was used as a heater.
 R. Rusinek, R. Kobyka / Journal of Stored Products Research 59 (2014) 254e259
dQ
rcp vDT
dt
where, dQ/dt is heat ux, r and cp e bulk density and thermal ca-
pacity of the material, DT e temperature change and v e velocity.
Based on Buckingham P-Theorem, the terms
dQ
dt
P (9)
rcp vDT
should remain constant for all systems, and therefore in our case
dQ dQ
dte dts
re cpe ve DTe rs cps vs DTs
where indexes e and s denotes an experimental and simulation
coefcients, respectively. Assuming similar conditions in laboratory
experiment and numerical simulations namely dQ/dte dQ/dts,
re rs, cpe cps and DTe DTs and rewriting velocity as length/
time, one might get the relation between times in numerical versus
laboratory experiment:
te ls
ts
le
To reect the conditions of the laboratory experiment, the
temperature was raised every 345 s of simulation time, which,
based on formula (11), corresponds to 24 h of the real experiment
on the scale of the model silo.
The progress of the temperature increase was adjusted in
accordance with the self-heating phenomenon.
3. Results and discussion
Some measured values of the temperature in the self-heating
experiment are shown in Fig. 3, and the measured and simulated
temperatures of the rapeseed at different distances from the silo
bottom are shown in Figs. 4e6. The phenomenon of self-heating
was axisymmetric. Larsson et al. (2012) obtained similar results
for wood pellet stored in approximately 4500 tons capacity silo. The
most active area, where the self-heating evolved most rapidly, was
Fig. 3. Distribution of temperature along the horizontal axis in the self-heating
experiment: values for sensors 11e15 only.
257
(8)
(10)
Fig. 4. Experimental (mean and standard deviation) and DEM simulated values for
sensors close to the wall.
near the geometric center of the layer of seeds with a moisture
content of 12.7%. This area was monitored by sensors 11e20 along
the horizontal axis and 3e6 along the vertical axis. The proximity of
(11)
the container wall facilitated more effective heat emission from the
seed mass, which was recorded by sensors 11 and 20. Most prob-
ably, part of the heat was absorbed by the steel wall of the
container. Similar results were obtained for the measured tem-
perature close to the wall in the experiment with internal heat
generation and in the simulation experiment.
Results of research on self-heating show that agglomeration and
lumping of matter under real conditions starts at temperature
levels of 30e35  C. It may be concluded from this that preventive
measures make sense only in the early stages of the development of
the phenomenon. If none of the measurement probes in the silo is
in the immediate area of danger, the occurrence of the process may
not be detected. Similar results were obtained by Ileleji et al. (2006)
for corn stored in a 12.5 t capacity bin.
The temperature of the stored rapeseed followed the trend of
increasing with the internal heat generation; however, the changes
in the values were different between the silo center and near the
wall. The temperature at the center at a height of 0.05D increased to
higher values than those measured by the corresponding sensors
near the wall at the same height. Similar trends were observed in
the DEM simulated values (Figs 4 and 5). This trend was conrmed
in research and computer simulation by Jia et al. (2000a, b, 2001)
for wheat and rice stored in a steel bin.
Figure 6 depicts the heat distribution inside the stored grain bulk
calculated using DEM at the very end of the simulation experiment.
Similar values of heat distribution were obtained using DEM by
Chaudhuri et al., (2006) in work on heat transfer in a rotating vessel.
Figure 6 is composed of four diagrams, where each diagram presents
the heat distribution inside a slice (of thickness 1 mm). The values of
the temperatures of the particles with their centers inside the given
slice were averaged, and a contour plot (isolines) was drawn. The
background of each diagram represents a real view of the particles
inside the slice, marked with colors representing temperatures from
21  C (blue) to 27  C (red). The top left picture (Fig. 6a) shows an
axial cross-section parallel to the x-axis (0  y  0.001 m) of the silo.
(For interpretation of the references to color in this gure, the reader
is referred to the web version of this article.) One may notice the
strong asymmetry of the temperature distribution, caused by the
heat source being placed only in the left half of the bottom. The
inuence of the silo walls and of the cold part of the bottom is
258 R. Rusinek, R. Kobyka / Journal of Stored Products Research 59 (2014) 254e259

clearly visible as well. Because of this inuence, the area of increased

Fig. 5. Experimental (mean and standard deviation) and DEM simulated values for
sensors close to the vertical axis of symmetry.
temperature tends to move not only vertically with time but also
horizontally toward the center. As a result, the area of rapeseed with
a temperature below 21.5  C is almost symmetrical. Figure 6b pre-
sents an axial cross-section parallel to the y-axis (0  x  0.001 m) as
seen from the direction where the heat source is located. All parti-
cles in this diagram are from the region above the unheated part of
the bottom. As can be seen, the particles in this region barely
exceeded 27  C, although the heat source operated at a maximum
temperature of 37  C. Figure 6c depicts a cross-section in the same
direction, but this time shifted to the area of the heat source
(i.e., 0.007 m  x  0.008 m), which leads to an area with a highly
increased temperature. Considering the heating-element tempera-
ture of 37  C, one may notice that such a high temperature is not
distributed widely, and at a height corresponding to about two
particle diameters the temperature decreases to about 33  C.
Figure 6b and c shows a very good symmetry of the temperature
eld; however, slight uctuations can be observed in Fig. 6b, which
leads to the conclusion that the progress of heat transport to some

Fig. 6. DEM predicted temperature distribution: (a) axial cross-section parallel to x-axis; (b) axial cross-section parallel to y-axis; (c) axial cross-section parallel to y-axis, shifted
toward the heat-source area by approximately 0.11D; (d) horizontal cross-section at a height of about 0.25D.
 R. Rusinek, R. Kobyka / Journal of Stored Products Research 59 (2014) 254e259
extent depends on the internal structure of the grains (e.g., force
chains) as well. Figure 6d presents a horizontal cross-section at a
height of approximately 0.25D (0.0165 m  z  0.0175 m). An
asymmetry in the temperature eld due to the nonaxial placement
of the heat generator has occurred here. The temperature dissipa-
tion at this height is also very high, and only a small area in the
middle exceeds 26  C.
4. Conclusion
Results showed that the maximum distance between the heat
source and the most distant sensor that recorded changes in tem-
perature was in the range of 0.5e0.6 m. According to ISO 4112:1990,
which assumes that each sensor monitors an area of radius 1.5 m,
the sensitivity of the measurement system was insufcient. The
system of temperature probes in silos that is proposed by norm
when there is unfavorable spatial location of the self-heating
phenomenon, may not serve its purpose.
The predicted temperatures obtained using the DEM are in fairly
good agreement with the measured values. The model of heat
transmission between two particles that we used to describe heat
transfer during the storage of rapeseed was able to predict the
distribution of the temperature in the model silo. The model can be
used satisfactorily to predict the temperature increase due to self-
heating in a silo without aeration. Our work demonstrates the
potential of DEM modeling for predicting changes in temperature
in bulk materials.
The main advantage of evaluating heat transfer at the particle
interactions level over continuum approach is that in the latter
case, heat propagates in all directions with the same intensity,
while in the former case propagation of heat depends on the
strength of interaction forces. In granular materials in the same
small region there are particles which are carrying minimal pres-
sure, while their neighbors may carry much bigger pressures
forming so called force chains. Using the discrete particle approach
we are able to detect such uctuations of heat inside small regions
of particles.
DEM in its current form requires a lot of improvement before it
could be considered as a reliable tool for an engineering analysis.
The two main drawbacks of this method are the relatively small
number of elements it can efciently treat as compared to sizes of
bulks in real scale silos, and the varied shape of the particles. DEM
in its current form may only use sphere based particles (spheres or
clusters of spheres), which limits number of contact points and in
consequence might give less adequate results for properties carried
through contact (e.g. heat transfer, interaction forces etc.). Future
development of DEM must include solving such problems, by
nding out faster and more accurate numerical algorithms to be
able to reect all conditions encountered in the real technological
storage process.
References
Arbogast, R.T., Weaver, D.K., Kendra, P.E., Chini, S.R., 2004. Temperature variation in
stored maize and its effect on capture of beetles in grain probe traps. J. Stored
Prod. Res. 40, 135e150.
ASAE, 2006. Moisture measurement e unground grain and seeds. American Society
of Agricultural and Biological Engineers.
ASAE, 2008. Thermal properties of grain and grain products, D234.4 May 2003
(R2008). American Society of Agricultural and Biological Engineers.
Chaudhuri, B., Muzzio, F.J., Tomassone, M.S., 2006. Modeling of heat transfer in
granular ow in rotating vessels. Chem. Eng. Sci. 61, 6348e6360.
259
Chourasia, M.K., Goswami, T.K., 2007. CFD simulation of effects of operating pa-
rameters and product on heat transfer and moisture loss in the stack of bagged
potatoes. J. Food Eng. 80, 947e960.
Cundall, P.A., Strack, O.D., 1979. A discrete numerical model for granular assemblies.
Geotechnique 29, 47e65.
Di Renzo, A., Di Maio, F.P., 2004. Comparison of contact-force models for the
simulation of collisions in DEM-based granular ow codes. Chem. Eng. Sci. 59,
525e541.
Dintwa, E., Tijskens, E., Ramon, H., 2008. On the accuracy of the Hertz model to
describe the normal contact of soft elastic spheres. Granular Matter 10, 209e
221.
Gawrysiak-Witulska, M., Siger, A., Wawrzyniak, J., Nogala-Ka1ucka, M., 2011.
Changes in tocochromanol content in seeds of Brassica napus L. during adverse
conditions of storage. J. Am. Oil Chem. Soc. 88, 1379e1385.
Horabik, J., Rusinek, R., 2001. Pressure ratio of cereal grains determined in a uniaxial
compression test. Int. Agrophys. 16, 23e28.
Iguaz, A., Arroqui, C., Esnoz, A., Virsed, P., 2004. Modelling and validation of heat
transfer in stored rough rice without areation. Biosyst. Eng. 88, 429e439.
Ileleji, K.E., Maier, D.E., Bhat, C., Woloshuk, C.P., 2006. Detection of a developing hot
spot in stored corn with a CO2 sensor. Appl. Eng. Agric. 22, 275e289.
ISO, 1990. Cereal and pulses e guidance on measurement of the temperature of
grain stored in bulk. ISO. ISO 4112.
Jia, C.C., Sun, Da-Wen, Cao, C.W., 2000a. Mathematical simulation of temperature
elds in a stored grain bin due to internal heat generation. J. Food Eng. 43, 227e
233.
Jia, C.C., Sun, Da-Wen, Cao, C.W., 2000b. Finite element prediction of transient
temperature distribution in a grain storage bin. J. Agric. Eng. Res. 76, 323e330.
Jia, C.C., Sun, Da-Wen, Cao, C.W., 2001. Computer simulation of temperature
changes in a wheat storage bin. J. Stored Prod. Res. 37, 165e177.
Juan, A., Moran, J.M., Guerra, M.I., Couto, A., Ayuga, F., Augado, P.J., 2006. Estab-
lishing stress state of cylindrical metal silos using nite element method:
comparison with ENV 1993. Thin-Walled Struct. 44, 1192e1200.
Kloss, C., Goniva, C., Hager, A., Amberger, S., Pirker, S., 2012. Models, algorithms and
validation for opensource DEM and CFD-DEM. Prog. Comput. Fluid Dyn. 12,
140e152.
Kwapinska, M., Saage, G., Tsotsas, E., 2008. Continuous versus discrete modelling of
heat transfer to agitated beds. Powder Technol. 181, 331e342.
Li, J., Mason, D.J., 2000. A computational investigation of transient heat transfer in
pneumatic transport of granular particles. Powder Technol. 112, 273e282.
Larsson, S.H., Lestander, T.A., Crompton, D., Melin, S., Sokhansanj, S., 2012. Tem-
perature patterns in large scale wood pellet silo storage. Appl. Energy 92, 322e
327.
qukaszuk, J., Molenda, M., Horabik, J., Montross, M.D., 2009. Variability of pressure
drops in grain generated by kernel shape and bedding method. J. Stored Prod.
Res. 45, 112e118.
Mills, J.T., 1989. Spoilage and heating of storing agricultural products: prevention,
detection, and control. Canadian Government Publishing Centre, Ottawa,
Ontario, Canada.
Moran, J.M., Andres, J., Robles, R., Aguado, P.J., 2006. Effects of environmental
temperature changes on steel silos. Biosyst. Eng.g 94, 229e238.
Moysey, E.B., Shaw, J.T., Lampman, W.P., 1977. The effect of temperature and
moisture on the thermal properties of rapeseed. Trans. ASAE, 768e771.
Muramatsu, Y., Tagawa, A., Sakaguchi, E., Kasai, T., 2007. Prediction of thermal
conductivity of kernels and a packed bed of brown rice. J. Food Eng. 80, 241e
248.
Nguyen, V.D., Cogne, C., Guessasma, M., Bellenger, E., Fortin, J., 2009. Discrete
modeling of granular ow with thermal transfer: application to the discharge of
silos. Appl. Therm. Eng. 29, 1846e1853.
Ramirez, A., Garcia-Torrent, J., Tascon, A., 2010. Experimental determination of self-
heating and self-ignition risks associated with the dusts of agricultural mate-
rials commonly stored in silos. J. Hazard. Mater. 175, 920e927.
Rusinek, R., Tys, J., Horabik, J., 2012. A concept for a steel tank for the drying and
safe storage of rapeseed. Rev. Construct. 4, 89e91 (in Polish).
Singh, K.K., Goswami, T.K., 2000. Thermal properties of cumin seed. J. Food Eng. 45,
181e187.
Smith, E.A., Sokhansanj, S., 1990. Natural convection and temperature of stored
produce e a theoretical analysis. Can. Agric. Eng. 32, 91e97.
Timbers, G.E., 1975. Properties of rapeseed 1. Thermal conductivity and specic
heat. Can. Agric. Eng. 17, 81e84.
Vargas, W.I., McCarthy, J.J., 2001. Heat conduction in granular materials. AIChE J. 47,
1052e1059.
Wang, L., Sun, Da-Wen, 2003. Recent developments in numerical modelling of
heating and cooling processes in the food industry e a review. Trends Food Sci.
Technol. 14, 408e423.
Wia cek, J., Molenda, M., 2011. Moisture-dependent physical properties of rapeseed
e experimental and DEM modeling. Int. Agrophys. 25, 59e65.
Xu, Y., Burfoot, D., 1999. Simulating the bulk storage of foodstuffs. J. Food Eng. 39,
23e29.