G Model

PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS

Particuology xxx (2017) xxxxxx

Contents lists available at ScienceDirect

Particuology
journal homepage: www.elsevier.com/locate/partic

Development of a rarefaction wave at discharge initiation in a storage

siloDEM simulations
R. Kobyka , J. Horabik, M. Molenda
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: The generation of a rarefaction wave at the initiation of discharge from a storage silo is a phenomenon of
Received 17 June 2016 scientic and practical interest. The effect, sometimes termed the dynamic pressure switch, may create
Received in revised form 31 October 2016 dangerous pulsations of the storage structure. Owing to the nonlinearity, discontinuity, and heterogeneity
Accepted 17 March 2017
of granular systems, the mechanism of generation and propagation of stress waves is complex and not
Available online xxx
yet completely understood.
The present study conducted discrete element simulations to model the formation and propagation of
Keywords:
a rarefaction wave in a granular material contained in a silo. Modeling was performed for a at-bottom
Granular ow
Dynamic pressure switch
cylindrical container with diameter of 0.1 or 0.12 m and height of 0.5 m. The effects of the orice size
Discrete element method and the shape of the initial discharging impulse on the shape and extent of the rarefaction wave were
Silo discharge examined. Positions, velocities, and forces of particles were recorded every 105 s and used to infer the
Stress wave location of the front of the rarefaction wave and loads on construction members. Discharge through
Rarefaction wave the entire bottom of the bin generates a plane rarefaction wave that may be followed by a compaction
wave, depending on the discharge rate. Discharge through the orice generates a spherical rarefaction
wave that, after reection from the silo wall, travels up the silo as a sequence of rarefactioncompaction
cycles with constant wavelength equal to the silo diameter. During the travel of the wave along the bin
height, the wave amplitude increases with the distance traveled. Simulations conrmed earlier ndings
of laboratory and numerical (nite element method) experiments and a theoretical approach, estimating
the speed of the front of the rarefaction wave to range from 70 to 80 m/s and the speed of the tail to range
from 20 to 60 m/s.
2017 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of
Sciences. Published by Elsevier B.V. All rights reserved.

Introduction non-uniform distribution of pressure and dynamic overpressures.

The start of discharge of granular solids from a silo results in
a sharp stress redistribution, which can be observed as a ramp of
silo pressure. The opening of the discharge gate reduces the ver-
tical pressure and may be associated with an increase in lateral
pressure. Zhang, Britton, and Jaremek (1993) reported an increase
in lateral pressure exceeding 40% and peaking within 0.7 s of the
discharge time for wheat exiting a model silo having a smooth
and corrugated wall. Dynamic phenomena during gravity discharge
often appear as vibrations (called silo music) or shocks (called silo
quakes) (Tejchman & Gudehus, 1993; Wilde et al., 2008). These
self-excited effects may result in severe dynamic loads on con-
struction members, such as pressure pulsations and shocks and the
Corresponding author. Fax: +48 817445067.
E-mail address: rkobylka@ipan.lublin.pl (R. Kobyka).
Vibrations and shocks arise from changes in material properties
during the ow, such as changes in density, cohesion, degree of
mobilization of internal and wall friction (Molenda, Horabik, &
Ross, 1995; Muite, Quinn, Sundaresan, & Rao, 2004; Roberts &
Wensrich, 2002), and the slipstick behavior during shear ow
(Stasiak, Molenda, Horabik, Mueller, & Opalinski, 2014; Wensrich,
2002).
Under industrial conditions, an initial phase of discharge is fre-
quently accompanied by the propagation of pressure waves from
the orice up through the material (Roberts & Wensrich, 2002).
Pressure waves (i.e., compression and rarefaction waves) and asso-
ciated discontinuities in velocity elds in the granular material
during discharge are inherent elements of pulsations and shocks
in silo structures (Wensrich, 2002). The phenomena of granular
solids leading to the emergence and propagation of stress waves
have received growing interest owing to the potential to clarify the
state of the granular medium (Hostler & Brennen, 2005) and owing

http://dx.doi.org/10.1016/j.partic.2017.03.006
1674-2001/ 2017 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by Elsevier B.V. All rights reserved.

Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
G Model
PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS
2 R. Kobyka et al. / Particuology xxx (2017) xxxxxx
to a broad range of industrial applications regarding the dynamic
aspects of handling bulk materials (Wang, Wensrich, & Ooi, 2012).
In recent decades, the propagation of pressure waves in granular
solids has been extensively investigated experimentally (Ben-Dor,
Britan, Elperin, Igra, & Jiang, 1997; Brzsnyi & Kovcs, 2011;
Hostler & Brennen, 2005; Muite et al., 2004; Wensrich, 2002), and
theoretically taking an analytical approach (Boutreux, Raphael, &
De Gennes, 1997; Cowin & Comfort, 1982; Ocone & Astarita, 1995),
and numerical approaches (Herbold & Nesterenko, 2012) based on
the nite element method (FEM) (Wang et al., 2012; Wensrich,
2003), and discrete element method (DEM) (Lamei & Mirghasemi,
2011; Melin, 1994; Rong, Negi, & Jofriet, 1995; Wensrich, 2002;
Wensrich & Stratton, 2011). Such works have provided insight into
mechanisms leading to the emergence, formation, and propagation
of the pressure wave.
The mechanism of the creation and propagation of the stress
wave in granular solids is not yet completely understood as it
involves complicated issues of the nonlinearity and heterogeneity
of granular systems. The speed of the wave propagation increases
with the overall constraining pressure p. For higher pressures, if
the elastic interaction is the Hertz contact law, the speed increases
like p1/6 (Hostler & Brennen, 2005; Melin, 1994), while the depen-
dence is stronger for low pressures, like p1/4 , possibly owing to
the dominant role of conical asperities in the transition of forces
between particles (Goddard, 1990). The dependence of the wave
propagation on pressure originates from the barotropy of the mate-
rial (Wensrich & Stratton, 2011). If the material behind a wave is at a
higher pressure than that in front of the wave, the front of the wave
tend to break, similar to a breaking wave on a beach (Boltachev,
Kaygorodov, & Volkov, 2009; Ocone & Astarita, 1995; Wensrich &
Stratton, 2011).
By making constitutive assumptions on the density dependence
of the particulate-phase pressure, Ocone & Astarita (1995) showed
that rarefaction waves smooth out as they propagate while com-
pression waves reinforce each other to become shocks. Behind the
front of a rarefaction wave, material dilates such that the speed
of perturbation is lower and the wave prole becomes smoother.
Wensrich (2002) showed that a rarefaction wave that is broad
and round is immediately followed by a compression wave that
is sharper.
When particles start to discharge from a silo, a rarefaction
wave is generated by gravity-driven movements of particles los-
ing support from the bottom. The bottom layer of particles starts
to accelerate at the beginning of discharge. The contact forces
with upper layers weaken and are nally lost. The process moves
up the column as the next layers of particles lose their support
from particles below. A traveling disturbance of particle velocities,
contact forces and distances between particles arises. In terms of
continuum mechanics, the rarefaction wave can be treated as a dis-
turbance of the elds of stress, velocity, and density traversing the
bedding.
Wensrich (2002) followed the approach of Boutreux et al. (1997)
of using the dynamic version of Janssens equation and applied a
hypoplastic model of material to develop the analytical formula for
the amplitude of a rarefaction wave. He showed that the amplitude
of the rarefaction wave that originates at the discharge orice grows
exponentially as it travels up the silo:
 + (h, t) = 0+ eh/2H0 ,
where  + (h, t) is the stress disturbance in the axial direction at
distance h and time t, 0+ is the initial stress disturbance at the silo
discharge orice, h is the distance traveled by the wave, t is time,
H0 = D/4k is the Janssen characteristic depth of the column, D is
the diameter of the column,  is the wall friction coefcient, and k
is the lateral-to-vertical pressure ratio.
Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
Table 1
Material and container parameters.
Material:
Youngs modulus, E (MPa) 868
Poissons ratio,  0.18
Restitution coefcient, e 0.5
Particle-wall friction coefcient, pw 0.3
Particle-particle friction coefcient, pp 0.482
Particle solid density,  (kg/m3 ) 1290
Bulk density, b (kg/m3 ) 804.8 (793.8)
Void fraction 0.376 (0.385)
Mean particle size, d (mm) 3.79
Particle size range (mm) 3.743.84
Particle number 83,000 (118,000)
Time step, t (s) 107
Container:
Youngs modulus, E (MPa) 2 105
Poissons ratio, 0.3
Height, H (m) 0.5
Diameter, D (m) 0.1 (0.12)
Orice diameter, D0 (m) 0.036, 0.1 (0.12)
Orice speed, u0 (m/s) 0.005, 0.05,
The exponential growth of the amplitude of a rarefaction wave
has been conrmed experimentally (Wensrich, 2002) and numeri-
cally (Wang et al., 2012; Wensrich, 2002, 2003). As the rarefaction
wave dilates the material, there is a decrease in vertical and lateral
pressure and a decrease in density directly behind the wave front.
The exponential growth of the amplitude of the rarefaction wave
obtained from numerical simulations was found to agree well with
experimental data (Roberts, 2012; Roberts & Wensrich, 2002).
A wide range of measurements of the wave speed in gran-
ular beds has been reported in the literature; i.e., from 50 to
500 m/s for glass and sand (Ben-Dor et al., 1997; Brzsnyi &
Kovcs, 2011; Hostler & Brennen, 2005; Wensrich, 2002). A prob-
able reason for the broad range of wave speeds given by the
authors is a difference in the bulk properties of materials used in
experiments and modeling parameters; i.e., the initial bulk den-
sity, degree of nonlinearity of the forcedisplacement relationship
for particleparticle interaction, and the stressstrain relationship,
continuum mechanics equivalent. Additionally, different frictional,
viscous or plastic damping and different boundary conditions and
construction parameters of the bin affect the wave speed. The stress
history may be another factor in play.
The purpose of the present study was to examine the formation
and propagation of a rarefaction wave in a granular material ll-
ing a at-bottomed cylindrical model silo employing the DEM, and
to compare ndings with those of experiments and nite element
modeling. Some, not yet clearly explained, issues are addressed; i.e.,
the effects of the orice size and the shape of the initial discharge
impulse on the shape and extent of a rarefaction wave.
DEM modeling
A series of DEM simulations was performed with an assembly
of 83,000 (and 118,000) spherical particles with a random uniform
distribution of diameters of 3.79 0.05 mm. The HertzMindlin
theory of elastic frictional collisions between particles (and
between particles and a wall) was used for simulations following
the work of Mindlin and Deresiewicz (e.g., Di Renzo & Di Maio,
(1) 2004). A damping term in the contact model was adopted as pro-
posed by Tsuji, Tanaka, and Ishida (1992); i.e., it was assumed that
the coefcient of restitution has a constant value for given particle
properties. The material parameters of particles were taken as those
measured for wheat kernels by Wiacek, 2008; i.e., Youngs mod-
ulus E = 868 MPa, Poissons ratio  = 0.18, coefcient of restitution
e = 0.5, and coefcient of particlewall friction pw = 0.3 (Table 1).
Numerical experiments were carried out for two at-bottomed
G Model
PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS
R. Kobyka et al. / Particuology xxx (2017) xxxxxx 3

cylindrical containers with diameters D = 0.1 and (L) 0.12 m and

having height H = 0.5 m (i.e., H/D = 5 and 4.17). Two diameters of the
container orice were tested: (O) D0 = 0.036 m and the (C) entire bin
diameter (i.e., 0.1 or 0.12 m). DEM software (LIGGGHTS) developed
by Kloss, Goniva, Hager, Amberger, and Pirker (2012) was used for
numerical simulations.

Filling, rest and discharge

To form a granular bedding in a replicable manner, the prepara-

tory cylindrical container located immediately above the test
container was equipped with a conical hopper having a 13 incli-
nation to the horizontal plane and an orice 0.036 m in diameter.
Particles were generated in random locations in the preparatory
container. When the rst batch of 5000 particles had settled, the
orice was opened and the material owed down to ll the test
container. Throughout the discharge, subsequent particles were
generated to maintain a continuous stream for lling. After lling,
the system was allowed to equilibrate for 1.3 (or 1.7) s.
The particles were discharged through the entire cross-section
of the bin or through the orice with a diameter of 0.036 m.
Three cases of gravity-driven discharge were studied as earlier Fig 1. Characteristics of rarefaction (h1 , h2/2 , h2 , h3 ) and compaction (h3 , h4 , h5 )
adopted by Wensrich (2003) and Wang et al. (2012), but with an wave proles. (h1 : the rarefaction wave head (a < 105 m/s2 ); h2/2 : the position of
extended range of velocity. They were two cases of ow with a half the minimum particle acceleration amin /2; h2 : the position of minimum particle
controlled velocity of the discharge gate of (1) u0 = 0.005 m/s and acceleration amin ; h3 : switch from a rarefaction to compaction wave or the position of
the rarefaction wave tail; h3 : the position switching from a rarefaction to compaction
(2) u0 = 0.05 m/s, and (3) the immediate removal of the orice or
wave; h4 : the position of maximum particle acceleration amax ; h5 : the position of
the oor of the bin (u0 = ). The velocity of the outlet u(t) followed maximum particle velocity vmax ; and v/h: slope of the linear part of the velocity
the kinematic boundary condition applied by Wensrich (2003) and prole of the rarefaction wave).
Wang et al. (2012) in experiments and numerical simulations (Eq.
(2)):
h3 , the position switching from a rarefaction to compaction
 
u0 1 cos t for t < 0.005 s
wave;
u (t) = 2 0.005 . (2) h4 , the position of maximum particle acceleration amax ; and
h5 , the position of maximum particle velocity vmax .
u0 for t 0.005 s
Velocities of characteristic points of waves U1 ,. . ., U5 were cal-
The same initial bedding was used in all variants of discharge culated as traveled distances h1 ,. . ., h5 divided by the travel times.
simulations to avoid noise introduced by various initial packings. In the central part of the rarefaction wave, the velocity of particles
Effects of the silo diameter, outlet diameter, and discharge velocity increased linearly with decreasing distance to the axis. The slope
on the generation of the pressure switch were examined. of the velocity change v/h in this linear part of the wave prole
(see Fig. 1) and the wave amplitude were used as two characteristic
measures of the evolution of the wave.
Method of data analysis

To obtain the relationships of particle velocity v versus h/D (see Results

Fig. 1), average values were taken for particles having centers in
a region that was three mean particle diameters d deep and wide
and had a height of 1.5d. Such test regions were set axially in the
container at a height ranging from 0 to 0.5 m with steps of 1.5d. The
central difference formula was used to calculate acceleration:
v (t + t) v (t t)
a (t) = . (3)
2t
Travel of the rarefaction wave up the silo height was described
using four characteristic positions (Fig. 1):
h1 , the position of the wave head (i.e., the height at which par-
ticle acceleration a exceeded 105 m/s2 );
h2/2 , the position of half the minimum particle acceleration
amin /2;
h2 , the position of the minimum particle acceleration amin ; and
h3 , the position of the rarefaction tail (a > 10 m/s2 in the case
of immediate removal of the entire bottom of the bin or a switch
from a rarefaction to compaction wave). Parameter h3 was dou-
ble checked; h3 was the position of the minimum particle velocity
vmin while h3 was the position of the change in sign of particle
acceleration.
The compaction wave was characterized by three positions in
the same way as the rarefaction wave (Fig. 1):
Sensitivity of the wave appearance to simulation parameters
Preliminary tests were performed to examine the effect of vari-
ation of various parameters (i.e., the simulation time step t,
coefcient of restitution e, Youngs modulus E (Fig. 2), particle size
distribution and shape of the bottom (Fig. 3)) on temporary proles
of the particle velocity along the silo height and on the wave speed.
The time step applied in simulations was 107 s (i.e. 0.4% of the
Rayleigh time step2.5 105 s (Huang, Nydal, & Yao, 2014). A 10-
fold increase (106 s) or decrease (108 s) in the time step did not
affect the velocity prole along the width of the rarefaction wave
(Fig. 2(a)). An increase or decrease in the coefcient of restitution
e of 0.1 only slightly affected the shape of the velocity prole, and
only in the area of greatest velocity change; i.e., the area of gener-
ation of the compaction wave (Fig. 2(b)). The strongest effect was
that of a change in Youngs modulus (Fig. 2(c)). A fourfold increase
in Youngs modulus increased the velocity of the front of the wave
(U1 ) by a factor of 1.5, while a fourfold decrease reduced U1 by
approximately 26%. This trend is consistent with the experimental
ndings of Duffy and Mindlin (1957) (and later Hostler & Brennen
(2005)) that the wave speed in granular bed scales with the elas-
tic wave speed in the material of the particles ((E/)0.5 , where E is

Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
G Model
PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS
4 R. Kobyka et al. / Particuology xxx (2017) xxxxxx

Fig. 2. Temporary proles of particle velocity along the bin height after 9 ms (upper row) and 13.5 ms (bottom row) discharge from a bin 0.12 m in diameter (D0 = 0.12 m,
u0 = 0.05 m/s) for different simulation parameters of: (a) time step t, (b) coefcient of restitution e, and (c) Young modulus E.

Youngs modulus and  is the density of the particles). Velocity U3 of
the position switching from a rarefaction to compaction wave was
much less sensitive to a change in E than the speed U1 of the front
of the rarefaction wave. The amplitude of the wave decreased for
both a fourfold increase and fourfold decrease in Youngs modulus
(Fig. 2(c)).
An increase in the range of distribution of particle diameters
increased the amplitude of the rarefaction wave and attened the
velocity prole (Fig. 3(a)). An increase in the polydispersity of the
material thus increases the ability of the material to dilate. Neither
the velocity U1 of the front of the rarefaction wave nor the veloc-
ity U3 of the position switching from the rarefaction to compaction
wave was affected by the degree of polydispersity of the granular
material. A change in shape of the silo bottom from a at bottom
to a hopper bottom considerably increased particle velocity along
the wavefront and the shape of the wave prole in the bottom part
of the silo. The reection of the wavefront from inclined walls of
the hopper increased the wave amplitude (Fig. 3(b)). The velocity
of the wavefront in a hopper discharge only slightly exceeded that
for the at-bottomed setup (75.1 versus 71.7 m/s). The remaining
considered velocities were lower than their at-bottomed equiva-
lents.

Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
G Model
PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS
R. Kobyka et al. / Particuology xxx (2017) xxxxxx 5

Fig. 3. Comparison of temporary proles of particle velocity along the bin height after 1.5 and 3 ms (upper row), and 4.5 and 6 ms (lower row) discharge from the bin 0.12 m
in diameter (D0 = 0.036 m, u0 = , dp 0.05 mm) for (a) a silo with a at bottom with an increased polydispersity range (dp 0.1 mm), and (b) a hopper silo with an angle of
inclination to the bottom plane of 60 .

General description of the waves
Numerical simulations of the start of discharge suggested that
rarefaction and compaction waves were observed in all cases of the
experiment, while the compaction wave was absent in two cases:
the lowest discharge rate (0.005 m/s) through the orice and the
immediate opening of the entire bottom of the bin. The compaction
wave was associated with the slowing down of particles behind the
rarefaction wavefront as a result of the controlled discharge rate,
converging ow in the hopper above the orice and wall friction. In

Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
G Model
PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS
6 R. Kobyka et al. / Particuology xxx (2017) xxxxxx

Fig. 4. Frames from visualization of the propagation of the rarefaction wave in the silo 0.12 m in diameter initiated by the immediate opening of (a) the entire bottom and
(b) the orice 0.036 m in diameter.

the case of discharge through the orice 0.036 m in diameter, the
converging ow in the hopper retarded particles and a compaction
wave was clearly visible. In the case of the immediate removal of the
entire bottom of the bin, the pressure below the rarefaction wave
vanished and there was no support to slow the particles above.
Fig. 4 shows frames from the visualization of propagation of the
rarefaction wave in the silo of D = 0.12 m for the immediate open-
ing of (a) the entire bottom of the bin and (b) the orice 0.036 m
in diameter. The most important difference between the two dis-
charge modes is that, in the case of the discharge through the
entire silo cross-section, a single one-dimensional rarefaction wave
appears and may be followed by a compaction wave, while dis-
charge through the orice initiates a spherical rarefaction wave that
changes into a periodic sequence of rarefactioncompaction waves
as it travels up the bin. The radius of the spherical wave increases
till the wavefront meets the bin wall and, after reection from the
wall, the rarefaction impulse propagates up the bin as a dome with
a radius slightly larger than the bin radius. The sequences of frames
presented in Fig. 4(b) illustrate that the compaction wave following
the front of the rarefaction wave grows from the center of the spher-
ical wave of rarefaction. The rst rarefactioncompaction sequence
is followed by subsequent, much smaller rarefactioncompaction
sequences.
Illustrations of temporary proles of particle velocities along the
bin height for selected times elapsed from the start of discharge
(Fig. 5(a)(c)) show differences in the development of a rarefaction

Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
G Model
PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS
R. Kobyka et al. / Particuology xxx (2017) xxxxxx 7

Fig. 5. Temporary proles of particle velocity along the bin height at different discharging times from the bin 0.12 m in diameter for: (a) (C1L) D0 = 0.12 m, u0 = 0.005 m/s; (b)
(C2L) D0 = 0.12 m, u0 = 0.05 m/s; (c) (C3L) D0 = 0.12 m, u0 = ; (d) (O1L) D0 = 0.036 m, u0 = 0.005 m/s; (e) (O2L) D0 = 0.036 m, u0 = 0.05 m/s; (f) (O3L) D0 = 0.036 m, u0 = .

wave among three rates of discharge through the entire bottom
of the bin. In the case of the lowest discharge rate (0.005 m/s,
Fig. 5(a)), the wave is the broadest and most rounded although
the wave amplitude develops approximately twice as quickly as
that in the case of a discharge rate of 0.05 m/s (Fig. 5(b)). In the
case of immediate removal of the bottom, the particle velocity
increases smoothly and continuously (Fig. 5(c)) while the particle
acceleration changes quickly with a short peak much higher than
the acceleration of gravity owing to elastic compulsion between
particles (Fig. 6(c)). Converging ow in the vicinity of and above
the orice changes the development of the wave (Fig. 5(d)(f)) as
compared with one-dimensional ow through the cross-section of
the entire bottom of the bin (Fig. 5(a)(c)). The oscillations of the
particle velocity resulting from rarefactioncompaction sequences

Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
G Model
PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS
8 R. Kobyka et al. / Particuology xxx (2017) xxxxxx

Fig. 6. Temporary proles of particle acceleration along the bin height at different discharge times from the bin 0.12 m in diameter for: (a) (C1L) D0 = 0.12 m, u0 = 0.005 m/s;
(b) (C2L) D0 = 0.12 m, u0 = 0.05 m/s; (c) (C3L) D0 = 0.12 m, u0 = ; (d) (O1L) D0 = 0.036 m, u0 = 0.005 m/s; (e) (O2L) D0 = 0.036 m, u0 = 0.05 m/s; (f) (O3L) D0 = 0.036 m, u0 = .

become more visible as the discharge rate increases. The sequences
of rarefactioncompaction take the form of sinusoidal oscillations
with constant wavelength (Fig. 6(d)(f)). The amplitude of the rst
sequence is the highest and decreases quickly in each following
sequence. The amplitude of the rarefactioncompaction sequence
increases with the discharge rate. The wavelength is modied by
the discharge rate. The wavelength is about 0.6D for the discharge
rate of 0.005 m/s (Fig. 6(d)) and 1D for higher discharge rates
(Fig. 6(e) and (f)).
Travel of the wave
The speed of the rarefaction wave is proportional to a power
function of overall pressure p (Goddard, 1990; Hostler & Brennen,
2005; Melin, 1994). This means that the wave starts with the high-
est speed at the bin bottom and slows as it travels up the bin height
through a vertical column of particles. A detailed comparison of
the wave speeds was performed for the average value of the speed
determined from an analysis of linear regression (Fig. 7). The aver-
age speed of the rarefaction wave (characterized by velocities U1 ,
U2/2 , U2 , and U3 ) and that of the compaction wave (characterized by
U3 , U4 , and U5 ) show large differences among investigated cases.
The smallest difference was observed in the velocity of the wave
head U1 (Table 2); i.e., from 69.5 m/s in the case of the lowest dis-
charge rate of the orice to 82.4 m/s in the case of the immediate
removal of the entire bottom of the bin. Reported results (Hostler
& Brennen, 2005; Melin, 1994) suggest that the speed of the rar-
efaction wave is sensitive to material parameters only to a lesser

Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
G Model
PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS
R. Kobyka et al. / Particuology xxx (2017) xxxxxx 9

Fig. 7. Positions of characteristic points of rarefaction (h1 , h2 , h3 ) and compaction (h3 , h4 , h5 ) waves traveling up the silo in the case of immediate removal of the orice: (a)
(C3) D = D0 = 0.1 m, (b) (O3L) D = 0.12 m, D0 = 0.036 m.

Table 2
Velocities of rarefaction (U1 , U2 , U3 ) and compaction waves (U3 , U4 , U5 ) determined using characteristic measures h1 , h2 , h3 , h4 , and h5 (see Fig. 1).

Label Bin diameter (m) Disch. rate u0 (m/s) Rarefaction wave Compaction wave

Uhead Uamin /2 Uamin Uvmin Uswitch/tail Uamax Uvmax

(m/s) (m/s) (m/s) (m/s) (m/s) (m/s) (m/s)

U1 U2/2 U2 U3 U3 U4 U5

Orce diameter = bin diameter

C1 0.1 0.005 71.9 52.6 54.0 58.3 57.8
C1L 0.12 0.005 74.5 53.3 55.6 58.1 51.6
C2 0.1 0.05 75.0 52.6 19.5 20.6 21.9 21.0 22.2
C2L 0.12 0.05 77.4 53.8 19.5 20.3 22.3 22.2 16.4
C3 0.1 79.9 62.7 56.1 82.5
C3L 0.12 82.4 61.7 54.1 77.8
Orce diameter = 0.036 m
O1 0.1 0.005 69.5 49.6 47.3 55.5 51.5 45.1
O1L 0.12 0.005 71.5 50.1 45.7 50.1 57.7 43.9
O2 0.1 0.05 72.9 54.1 53.2 61.7 59.6 55.8 55.9
O2L 0.12 0.05 74.7 54.4 54.3 52.1 53.4 54.8 53.7
O3 0.1 78.1 57.2 53.2 49.6 49.4 49.7 67.7
O3L 0.12 80.1 57.7 53.8 52.7 53.7 55.1 57.6

extent. The highest value of velocity obtained for the case of the
plane wave in the larger bin suggests that factors restraining the
formation of the wave (i.e., the orice dimension, discharge rate,
and bin size) reduce also the speed of the wave. The highest differ-
ences in velocity (from 20 to 60 m/s) were found for U3 (of position
switching to a compaction or wave tail), while slightly lower were
those observed for U2 (of position of amin ). In the case of the low-
est switch velocity U3 (C2, C2L), the most intense development of
the rarefaction wave was observed with a continuous increase in
wavelength.
The dependence of temporary values of speed of the wave head
Uhead , determined from the linear regression line of 31 subsequent
points of the h1 (t) relation (Fig. 7), on the overall pressure p was
approximated by the power function (Fig. 8)
 
U1 /U10 = 1 A 1 p/p0 , (4)

where p is the mean pressure calculated from micro-variables fol-

lowing Chang and Liao (1990), p0 is the mean pressure at h = 0, U10
is the initial wave speed at h = 0, and and A are approximation
parameters. Because values of variables U1 and p were determined
for different locations along the height h, to coincide with the same
 extrapolated according
values of h, the p(h) relationship was  to the
exponential function p (h) = 1700 1 e9.31(0.5h) to t the same
heights h as the U1 (h) relation.
Fig. 8. Approximation of the wave speed U1 by a power function of conning pres-
sure p.
The exponent of the nonlinear regression line was found to
range from 0.221 for free discharge (C3, O3) to 0.276 for velocity-
controlled discharge and the corresponding coefcient A ranged

Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
G Model
PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS
10 R. Kobyka et al. / Particuology xxx (2017) xxxxxx
from 0.7 to 0.586 (Fig. 8). The average value of the exponent was
found to be close to the value of 0.25 reported by Goddard (1990)
for the case of low pressure. Similar uctuations of the temporary
speed of waves in each of the studied cases result from the use of
identical initial bedding. The courses of relative temporary speeds
of the wave head U1 /U10 versus relative pressure p1 /p0 follow the
same pattern, although they can be separated into two groups: (1) a
slower change in the case of the controlled velocity of discharge (C1,
C2, O1, and O2) and (2) a faster change in the case of the immediate
removal of the entire bottom of the bin or orice (C3 and O3).
Analysis of positions of characteristic points h1 ,. . ., h5 and the
evolution of proles of velocity (Fig. 5) and acceleration (Fig. 6)
as the wave traveled up the bin shed light on three effects. The
rst effect is a decrease in the velocity of propagation of the wave
tail as compared with the velocity of the wave head in the fully
developed rarefaction wave that increases continuously in width
(Fig. 5(b) versus Fig. 5(a)). The second is transition from rarefac-
tion to free discharge (the point at which the plots for h2 and h3
in Fig. 7(a) cross). The third is the difference under free-discharge
conditions between discharge through the orice (Fig. 6(f)) and that
through the entire bottom of the bin (Fig. 6(c)).
Evolution of amplitudes of rarefaction and compaction waves
The amplitude of particle velocity within the rarefaction wave
generally increased with the traveled distance (v(h3 ) in Fig. 9(a), (b),
(e), and (f)). However, in two cases, a decrease in the wave ampli-
tude with the traveled distance was observed (v(h3 ) in Fig. 9(c) for
the amplitude of acceleration and Fig. 9(d) for the amplitude of
velocity). The rarefaction wave is followed by a compaction wave
(v(h5 ) in Fig. 9(a), (b), (e), and (f)) or not (Fig. 9(c) and (d)).
Four cases of the evolution of amplitudes of rarefaction and
compaction waves were observed as follows.
Case one: a rarefaction wave followed by a compaction wave
(Fig. 9(a), (b), (e), and (f))
This situation occurs when the orice is moved down quickly
with acceleration similar to that of gravity (Fig. 9(b) and (e)) or
when free discharge through the orice is allowed and ow is con-
trolled by the orice aperture (Fig. 9(f)). In the studied case, a 9-fold
decrease in the cross-section of the discharge orice resulted in at
least a 60-fold decrease in the wave amplitude and the amplitude
was thus nearly constant as the wave traveled up the silo. When
the particle velocities, controlled by the velocity of the orice, were
additionally hampered by converging ow produced by the orice,
amplitudes of both waves (rarefaction and compaction) were small,
nearly the same, and almost constant (Fig. 9(e) compared with
Fig. 9(b)). For the lowest velocity of discharge through the entire
bottom of the bin (Fig. 9(a)), the compaction following the rarefac-
tion wave was found to be distributed almost uniformly between
the end of the rarefaction wave and the bin bottom (Fig. 6(a)). In
that case, the compaction wave was described only by the velocity
of the switch from rarefaction to compaction (Table 2).
Case two: a vanishing compaction wave (Fig. 9(d)) or no
compaction wave (Fig. 9(c))
In the case of the lowest discharge rate through the ori-
ce (0.005 m/s, D0 = 0.036), the particle velocity increases mainly
monotonically. The local minimum velocity was visible only for a
short traveled distance of the wave, although the minimum and
maximum of acceleration were clearly visible (Fig. 6(d)) and corre-
sponding velocities U3 and U4 were determined (Table 2). Particles
did not have enough free space to increase their velocity above the
velocity of the orice and only changes in the rate of increase in
particle velocity without compaction were therefore observed. The
switch velocity, U3 , and velocity U4 were determined for the com-
Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
paction wave (Table 2). The opposite situation of a limited reduction
in the discharge rate occurred in the case of discharge through the
entire bottom of the bin (Fig. 9(c)). Wall friction was a limited obsta-
cle for the discharge in this case. Here, an acceleration wave rather
than a velocity wave arose and was not followed by a compaction
wave.
Case three: a decrease in amplitude of the rarefaction wave
(Fig. 9(d))
In the case of the lowest rate of discharge through the orice
(0.005 m/s), the ow of particles is so slow that the amplitude of
the rarefaction wave (local minimum velocity) decreases with the
traveled distance. Initially, a low wave amplitude vanishes as the
wave travels as a result of adjustment to a low discharge rate.
Case four: free discharge in the vertical columnacceleration
wave (Fig. 9(c))
This is the case of the second-order wave of rarefaction; i.e., a
discontinuity of the acceleration wave appears. Across the wave
of acceleration, the particle velocity, bulk density and pressure
are continuous but their derivatives are not (Nunziato, 1978). For
an outlet diameter equal to the bin diameter, the particle veloc-
ity increases nearly continuously as a result of free gravitational
ow slightly hampered by wall friction and viscous damping. In
the initial part of discharge, an additional increase in particle veloc-
ity arises from elastic compulsion forces acting between particles.
Short-range elastic repulsion forces are a reason of the sudden
increase in particle acceleration much above the acceleration of
gravity for a short period of time and for a short distance trav-
eled by the wave. After a short peak, the acceleration decreases
rapidly to the acceleration of gravity. With an increase in the dis-
tance traveled by the wave, the elastic compulsion forces decay
and are balanced by wall friction and viscous damping. Acceler-
ation of particles decreases asymptotically to the acceleration of
gravity. Applying the same measures as in the case of the wave
of velocity (Fig. 1), the amplitude of the acceleration wave was
determined as the acceleration at h2 and the wave tail (h3 ) as the
acceleration increased above 10 m/s2 . The amplitude of the accel-
eration wave separates the region of loosening resulting from the
joint action of the elastic compulsion forces and gravity ow from
the region of loosening resulting from gravity ow alone. The loos-
ening of material continuously intensies from the wave head to
the bottom.
Along the rarefaction and compaction wave, there is a change
in volumetric strain v and a corresponding change in bulk density
of the material, which is related to the change in particle velocity
according to the rule of continuity of a material (Nunziato, 1978;
Wensrich, 2003):
[v] + U [v ] = 0, (5)
where [v] is the change in particle velocity along the wave, [v ] the
change in volumetric strain along the wave, and U the speed of the
wave.
In the case of a discharge rate of 0.05 m/s through the entire
bottom of the bin (Fig. 5(b), C2L), a highest amplitude of parti-
cle velocity vmin of 0.1 m/s and a speed of the rarefaction wave
of 77.4 m/s (Table 2) were found, while the maximum volumetric
strain v was of 0.0013. The compaction wave following the wave of
rarefaction reduced the strain to 0.00065, and the strain remained
nearly constant till the orice.
Wensrich (2002) used an analytical model and found that a
rarefaction disturbance grew exponentially (Eq. (1)), which was
later conrmed in FEM simulations by Wang et al. (2012). We thus
adopted a relationship between the amplitude of the disturbance
G Model
PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS
R. Kobyka et al. / Particuology xxx (2017) xxxxxx 11

Fig. 9. Evolution of rarefaction and compaction wave amplitudes in the bin 0.12 m in diameter for: (a) (C1L) D0 = 0.12 m, u0 = 0.005 m/s; (b) (C2L) D0 = 0.12 m, u0 = 0.05 m/s;
(c) (C3L) D0 = 0.12 m, u0 = ; (d) (O1L) D0 = 0.036 m, u0 = 0.005 m/s; (e) (O2L) D0 = 0.036 m, u0 = 0.05 m/s; (f) (O3L) D0 = 0.036 m, u0 = . Note that the wave is an acceleration
wave in case (c) and a velocity wave in all other cases.

of axial velocity vmin (h) obtained in DEM simulations and the mag-
nitude of the disturbance initiating a rarefaction wave u0 :
vmin (h)
= e2kh/D ,
u0
with the pressure ratio k and apparent coefcient of wall fric-
tion  calculated from stress components for the corresponding
location of the wave head.
Fig. 10 compares analytical calculations and DEM simulations
of the growth of the amplitude of velocity of the rarefaction wave
as the wave travels up the silo. Values of the velocity amplitude
obtained in DEM simulations agree with an analytical model only
in one case; i.e., for a discharge rate of 0.05 m/s through the entire
bottom of the bin. In the case of a bin 0.1 m in diameter, the
DEM simulation results coincide with analytical values in the much
lower range of traveled distance as compared with the case for the
bin 0.12 m in diameter. It is suggested that the number of par-
(6)
ticles in the smaller bin approaches the limit necessary for the
simulation of that process. The 10-fold reduction of the discharge
rate (to 0.005 m/s) resulted in an approximately 20-fold decrease
in the amplitude of particle velocity, and the normalized ampli-
tude vmin /u0 is therefore approximately half that of the analytical
prediction. Similarly, the reduction in the orice diameter from
 
0.12 to 0.036 m resulted in an approximately 60-fold reduction
in the velocity amplitude vmin D02 / u0 D2 . The amplitude growth,
determined from DEM simulation and analytical predictions for the

Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
G Model
PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS
12 R. Kobyka et al. / Particuology xxx (2017) xxxxxx
 of the
Fig. 10. Comparison of analytical results and DEM simulations  normalized
amplitude of the rarefaction wave of particle velocity vmin D02 / u0 D2
travels up the silo.
corresponding value of the averaged Janssen characteristic depth
(H0 in Eq. (1)) of about 0.16, for a discharge rate of 0.05 m/s through
the entire bottom of the bin corresponds well with experimen-
tal and modeled data of the wave amplitude growth collected and
reviewed by Wensrich (2003).
Discussion
The generation of a rarefaction wave at the initiation of dis-
charge from a storage silo is a phenomenon of great scientic
and practical interest. The effect, sometimes termed the dynamic
pressure switch, may create severe pulsations of bin structures, par-
ticularly owing to a resonance effect. The rise of arches of particles
followed by their collapse after the fall of particles below them was
proposed by Muite et al. (2004) as the leading mechanism of the
generation of rarefaction waves. In that way, large systems of par-
ticles rearrange the network of their contacts to sustain the change
in load (Roux, 2000).
The DEM has become a promising tool for the examination of
interactions in granular bedding, providing insight into mecha-
nisms of stress transmission that explain the effect of properties
of individual particles on the generation and propagation of stress
waves. The speed of the rarefaction wavefront obtained in our DEM
simulation was found to range from 70 to 80 m/s, which agrees well
with reported experimental values. The highest wave speed was
found in the case of the least-restrained ow; i.e., after immediate
removal of the entire bottom of the bin.
Wensrich and Stratton (2011) proposed a barotropic effect on
the shape of waves accompanying discharge commencement. In
materials with nonlinear forcedisplacement interactions between
particles, the compaction wave takes the form of a shock wave,
while the rarefaction wave smoothens and broadens during the
propagation. Numerical simulations conrmed the interpretation
of a barotropic effect on dynamic phenomena at the commence-
ment of silo discharge. The pronounced example observed in
simulations was the formation of the one-dimensional rarefaction
wave during discharge through the entire bottom of the bin. The
wavefront smoothened as it traveled up the bin height while the
wave amplitude increased. As a result, the speed of the wave tail
was found to be less than one-quarter the speed of the wavefront.
A difference was observed in the proles of the wavefront for
two examined modes of bin discharge. Discharge through the entire
bottom of the bin initiated propagation of a plane wave while dis-
charge through the orice initiated a spherical wave that changed
shape to part of an arch after being reected from the bin wall.
Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
The region near and above the orice, where the ow is the most
rapid (Hirshfeld & Rapaport, 2001), initiates the spherical wave of
rarefaction. The motion is dominated by what occurs near the cen-
ter of the container. Reection of the rarefaction wave from the bin
wall stops further growth of the radius of the wave. The compaction
wave following the wave of rarefaction emerges in the center of
the sphere of the initial rarefaction wave and both waves travel up
the bin height. In that way, regular rarefactioncompaction cycles
arise with nearly constant wavelength and frequency. The interac-
tion of rarefaction and compaction waves with a bin structure (i.e.,
reection from the wall or travel upward along the wall) creates
dangerous load ramps. This may result in severe pulsations of the
bin structure (Muite et al., 2004; Wensrich, 2002, 2003). Damping
properties of the granular material may reduce stress pulsations to
some extent, as reported by Tejchman and Klisinski (2001). These
authors performed nite element simulations for the rapid ow
of granular materials in a silo and demonstrated that the inclu-
sion of inertial forces and viscous damping decreased calculated
as the wave
displacements, velocities and stress amplitudes.
Conclusions
We modeled the formation and propagation of a rarefaction
wave in a granular material lling a at-bottomed cylindrical model
silo employing the DEM. We compared our ndings with those of
experiments and nite element modeling. The following conclu-
sions are drawn from our modeling.
The DEM as a method based on realistic forcedisplacement
interactions between particles is a useful tool for analysis of the
development of rarefaction and compaction waves.
Simulations conrmed earlier ndings of laboratory and numer-
ical experiments estimating the speed of the front of the
rarefaction wave ranging from 70 to 80 m/s and that of its tail
ranging from 20 to 60 m/s.
The shape of the front of the rarefaction wave is sensitive to the
dimension of the discharge orice.
As a rarefaction wave travels along the silo height, its amplitude
increases with distance traveled, except in the case of a very low
discharge rate. Only in the case of a discharge rate of 0.05 m/s
through the entire cross-section of the bin was the rate of ampli-
tude growth exactly equal, which was predicted analytically and
by FEM modeling.
The gradient of velocity along the rarefaction wave decreases
with the distance traveled by the wave owing to the barotrophy
of the material, except in the case of a very low discharge rate.
References
Ben-Dor, G., Britan, A., Elperin, T., Igra, O., & Jiang, J. P. (1997). Experimental inves-
tigation of the interaction between weak shock waves and granular layers.
Experiments in Fluids, 22(5), 432443.
Boltachev, G. S., Kaygorodov, A. S., & Volkov, N. B. (2009). Densication of the granular
medium by the low amplitude shock waves. Acta Mechanica, 207(3), 223234.
Brzsnyi, T., & Kovcs, Z. (2011). High-speed imaging of traveling waves in a gran-
ular material during silo discharge. Physical Review E, 83(3), 032301.
Boutreux, T., Raphael, E., & De Gennes, P. G. (1997). Propagation of a pressure step in a
granular material: The role of wall friction. Physical Review E, 55(5), 57595773.
Chang, C. S., & Liao, C. L. (1990). Constitutive relation for a particulate medium with
the effect of particle rotation. International Journal of Solids and Structures, 26(4),
437453.
Cowin, S. C., & Comfort, W. J. (1982). Gravity-induced density discontinuity waves
in sand columns. Journal of Applied Mechanics, 49, 497500.
Di Renzo, A., & Di Maio, F. P. (2004). Comparison of contact-force models for the
simulation of collisions in DEM-based granular ow codes. Chemical Engineering
Science, 59(3), 525541.
Duffy, J., & Mindlin, R. D. (1957). Stress-strain relations and vibrations of a granular
medium. Journal of Applied Mechanics, 24, 585593.
Goddard, J. D. (1990). Nonlinear elasticity and pressure-dependent wave speeds
in granular media. Proceedings of the Royal Society of London A: Mathematical
Physical and Engineering Sciences, 430(July (1878)), 105131.
G Model
PARTIC-1019; No. of Pages 13 ARTICLE IN PRESS
R. Kobyka et al. / Particuology xxx (2017) xxxxxx
Herbold, E. B., & Nesterenko, V. F. (2012). Propagation of rarefaction pulses in par-
ticulate materials with strain-softening behavior. AIP Conference Proceedings,
1426(March (1)), 14471450.
Hirshfeld, D., & Rapaport, D. C. (2001). Granular ow from a silo: Discrete-particle
simulations in three dimensions. The European Physical Journal E: Soft Matter and
Biological Physics, 4(2), 193199.
Hostler, S. R., & Brennen, C. E. (2005). Pressure wave propagation in a granular bed.
Physical Review E, 72(3), 031303.
Huang, Y. J., Nydal, O. J., & Yao, B. (2014). Time step criterions for nonlinear dense
packed granular materials in time-driven method simulations. Powder Technol-
ogy, 253, 8088.
Kloss, C., Goniva, C., Hager, A., Amberger, S., & Pirker, S. (2012). Models, algorithms
and validation for opensource DEM and CFDDEM. Progress in Computational
Fluid Dynamics, an International Journal, 12(23), 140152.
Lamei, M., & Mirghasemi, A. A. (2011). A discrete element model for simulating
saturated granular soil. Particuology, 9(6), 650658.
Melin, S. (1994). Wave propagation in granular assemblies. Physical Review E, 49(3),
23532361.
Molenda, M., Horabik, J., & Ross, I. J. (1995). Dynamic load response in a model bin
at the start of grain discharge. Transactions of the ASAE, 38(6), 18691873.
Muite, B. K., Quinn, S. F., Sundaresan, S., & Rao, K. K. (2004). Silo music and silo quake:
Granular ow-induced vibration. Powder Technology, 145(3), 190202.
Nunziato, J. W. (1978). Propagation of plane waves in granular media. In Proceedings
of the USJapan Seminar on ContinuumMechanical and Statistical Approaches in
the Mechanics of Granular Materials (pp. 291300).
Ocone, R., & Astarita, G. (1995). Compression and rarefaction waves in granular ow.
Powder Technology, 82(3), 231237.
Roberts, A. W. (2012). Review of silo loadings associated with the storage of bulk
granular materials. In Proceedings of the International Conference of Agricultural
Engineering-CIGR-AgEng 2012: Agriculture and Engineering for a Healthier Life (pp.
8-12). July.
Please cite this article in press as: Kobyka, R., et al. Development of a rarefaction wave at discharge initiation in a storage siloDEM
simulations. Particuology (2017), http://dx.doi.org/10.1016/j.partic.2017.03.006
13
Roberts, A. W., & Wensrich, C. M. (2002). Flow dynamics or quaking in gravity
discharge from silos. Chemical Engineering Science, 57(2), 295305.
Rong, G. H., Negi, S. C., & Jofriet, J. C. (1995). Simulation of ow behaviour of bulk
solids in bins. Part 2: Shear bands, ow corrective inserts and velocity proles.
Journal of Agricultural Engineering Research, 62(4), 257269.
Roux, J. N. (2000). Geometric origin of mechanical properties of granular materials.
Physical Review E, 61(6), 68026832.
Stasiak, M., Molenda, M., Horabik, J., Mueller, P., & Opalinski, I. (2014). Mechani-
cal properties of potato starch modied by moisture content and addition of
lubricant. International Agrophysics, 28(4), 501509.
Tejchman, J., & Gudehus, G. (1993). Silo-music and silo-quake experiments and a
numerical Cosserat approach. Powder Technology, 76(2), 201212.
Tejchman, J., & Klisinski, M. (2001). FE-studies on rapid ow of bulk solids in silos.
Granular Matter, 3(4), 215230.
Tsuji, Y., Tanaka, T., & Ishida, T. (1992). Lagrangian numerical simulation of plug ow
of cohesionless particles in a horizontal pipe. Powder Technology, 71(3), 239250.
Wang, Y., Wensrich, C. M., & Ooi, J. Y. (2012). Rarefaction wave propagation in tapered
granular columns. Chemical Engineering Science, 71, 3238.
Wensrich, C. (2002). Experimental behaviour of quaking in tall silos. Powder Tech-
nology, 127(1), 8794.
Wensrich, C. (2003). Numerical modelling of quaking in tall silos. International Jour-
nal of Mechanical Sciences, 45(3), 541551.
Wensrich, C. M., & Stratton, R. E. (2011). Shock waves in granular materials: Discrete
and continuum comparisons. Powder Technology, 210(3), 288292.
Wiacek, J. (2008). Modeling of quasi-static effects in grain bedding using discrete ele-
ment method. Doctoral dissertation. Lublin, Poland: Institute of Agrophysics PAS.
Wilde, K., Rucka, M., & Tejchman, J. (2008). Silo music-mechanism of dynamic ow
and structure interaction. Powder Technology, 186(2), 113129.
Zhang, Q., Britton, M. G., & Jaremek, R. (1993). Dynamic loads during discharge for
wheat, barley and canola in a smooth and a corrugated-walled model bin. Journal
of Agricultural Engineering Research, 56(2), 111119.