Mechanics of Materials 41 (2009) 707–714

Contents lists available at ScienceDirect

Mechanics of Materials
journal homepage: www.elsevier.com/locate/mechmat

Numerical investigation of progressive development of granular pile

with spherical and non-spherical particles
Chong Zhou, Jin Y. Ooi *
Institute for Infrastructure & Environment, School of Engineering & Electronics, University of Edinburgh, King’s Building, Edinburgh EH9 3JL, UK

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

Article history: The base pressure distribution underneath a granular pile is carefully investigated using
Received 15 December 2008 the discrete element method. Simulations were performed using both spherical and non-
spherical particles in several numerical models of granular piles. A pressure profile with
a significant pressure dip under the apex was predicted for all three models under certain
conditions. The spherical assembly appears to produce a more erratic pressure profile with
a less significant pressure dip than the non-spherical particles. The history of the pile prep-
aration is shown to be important whilst the base deflection is shown to be not a require-
ment for the dip to form. The results are in good agreement with some published
experimental observations. The issue of lengthscale in terms of the dimension of the charg-
ing source relative to the dimension of the pile is explored. The granular fabric in terms of
the magnitude of contact force, the contact orientation and the particle orientation within
a granular pile have also been described.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction Smid and Novosad, 1981), and good reviews of the prob-
lem can be found elsewhere e.g. (Atman et al., 2005; Cates
The handling and processing of granular solids are vital et al., 1998; Savage, 1998). There is little consensus on the
operations in a wide range of industries. It has been esti- fundamental physics and mechanics assumptions made
mated that roughly 50% of the products and at least 75% between the many mathematical models of this apparently
of the raw materials in industry are granular or particulate simple system, and quite contradictory results are some-
in form (Nedderman, 1992). Granular solids are commonly times claimed. Numerous factors have been suggested to
stored in open stockpiles, particularly in the mining indus- explain the pressure dip observed under the apex of a pile.
try. The design of gravity reclaim system for a stockpile re- These include the presence of a base deflection (Zhou et al.,
quires knowledge of the base pressure distribution 2003), pile construction history (Vanel et al., 1999), forma-
underneath the stockpile. One classic granular mechanics tion of granular skeleton (Cates et al., 1998), particle shape
problem relating to stockpiles is that a significant dip in (Zuriguel and Mullin, 2008), particle segregation (Head and
the vertical pressure on the base is observed underneath Rodgers, 1997) and ‘reduced local density due to increased
the apex of the pile, at the location where intuition would filling rate’ (Smid and Novosad, 1981). However the inter-
expect a maximum pressure. For this reason, the base pres- play between these factors is not at all clear and a compre-
sure distribution is sometimes described as an M-shaped hensive understanding of this phenomenon remains
distribution. This problem has been the subject of numer- elusive. This paper presents a numerical investigation of
ous analytical e.g. (Cates et al., 1998; Reydellet and Clé- granular piles comprising of spherical and non-spherical
ment, 2001), numerical e.g. (Luding, 1997; Zhou et al., particles.
2003) and experimental studies e.g. (Geng et al., 2001; It is noted that the great majority of granular pile exper-
iments in which a significant pressure dip was reported
came from using real solids that were non-spherical and/
* Corresponding author. Tel.: +44 (0) 131 6505725.
E-mail address: [email protected] (J.Y. Ooi). or rough frictional particles. These include sand and

0167-6636/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mechmat.2009.01.017
708 C. Zhou, J.Y. Ooi / Mechanics of Materials 41 (2009) 707–714

fertilizer granules (Smid and Novosad, 1971), elliptical par- sliding contact friction was considered for all cases. The
ticles (Zuriguel and Mullin, 2008) and iron ore pellets (Ooi commercial code EDEM was used to simulate spherical
et al., 2008). The literature also shows that the majority of particles because the rolling friction model of Zhou et al.
experiments using 3D spheres or 2D circular disks pro- (1999) has been fully implemented in EDEM and is not
duced a less significant pressure dip and often with consid- available in PFC3D.
erable pressure fluctuation (Brockbank et al., 1997; Geng To explore the particle shape effect, non-spherical par-
et al., 2001; Zuriguel and Mullin, 2008). These results have ticle in this paper was created by overlapping two uniform
led some to believe that the pressure dip beneath a granu- spheres (2 mm in radius) together to give an aspect ratio of
lar pile is not a robustly reproducible phenomenon and its 1.5. The particles were centrally charged onto a rigid hori-
formation can be sensitive to numerous factors. zontal frictional base to form piles of varying dimensions.
The discrete element method which models the granu- The three groups of simulation (single layer pile, wedge-
lar solids at particle scale level is used in this study to shaped pile and conical pile) were conducted and pre-
investigate the influence of several factors on the pressure sented in the next three sections. Table 1 lists the material
dip phenomenon. Spherical and non-spherical particles parameters used in this study.
were modelled to explore the influence of particle shape. The base pressures were averaged on a base segment of
The question of 2D and 3D representation of a granular pile length 10 times the mean particle diameter, with the seg-
was explored using three different numerical models, ment boundary moving by one particle diameter for each
namely single layer model, the planar wedge model and evaluation (moving average technique). The averaging pro-
the 3D conical model. The numerical results show that a cess on 10 diameters was chosen to give a balance between
pronounced pressure dip below the apex of the pile can the very high scatter if the segment is small, and the loss of
form in all three models, when the pile was created by detail if it is large. The change of only one diameter indi-
pouring particles from a concentrated jet onto a frictional cates what an imaginary pressure cell might observe if
rigid base. The influence of the pile construction history placed at each point. These evaluated base pressure distri-
in the form of the dimension of the pouring jet was also butions are presented and discussed below.
explored.
The question of lengthscale was explored by studying 3. Single layer pile model
the evolution of the base pressure as the pile grows larger.
Many of the published DEM simulations of granular piles In this DEM model, all particles were generated to fall
were relatively small in size and often produced less than randomly with particle centres aligned in a vertical plane.
convincing evidence on the pressure dip phenomenon, so The non-spherical paired-sphere particles were generated
an attempt is made here to evaluate the pressure dip in with random orientation of the particle longitudinal axis
terms of the particle size and pile size more explicitly. in the prescribed plane. The simulation thus simulates
The structural evolution of force chains within the pile is the pile formation from a single layer of particles and
also investigated to better understand the force bearing resembles a 2D situation. The pouring jet was given a
network supporting the weight of the assembly. width of 10R to model a concentrated pouring jet but large
enough to avoid mechanical arching. The jet was kept as
2. Implementation of DEM model close to the top of the growing pile as possible to reduce
the impact energy on the pile. The distance between the
The present work is based on the discrete element apex of the pile and the jet source was maintained at a
method originally proposed by Cundall and Strack (1979) height in between 8 and 15 particle diameters. The charg-
which has been covered extensively in the literature e.g. ing rate was set at 500 particles per second. When the tar-
(Herrmann and Luding, 1998; Yu, 2004) and will not be get number of particles has been generated, the assembly
elaborated here. The DEM calculations were performed was simulated for a further 3 s to reach its equilibrium
using the PFC3D code (Itasca Consulting Group Inc., state before data was analysed for the pile. The equilibrium
2003), apart from one simulation on spheres which was state was confirmed by examining both average unbal-
performed using the EDEM code (DEM Solutions Ltd., anced force and total kinetic energy in the whole system,
2007). Both commercial codes deploy an explicit numerical which were less than 1  106 N and 5  1011 J, respec-
scheme in which the movement and interaction between a tively. To explore the scaling effect, we continued to build
finite number of particles are modelled at individual parti-
cle level as proposed by Cundall and Strack (1979).
Particle contact is modelled using a spring-dashpot
Table 1
model with a tangential slider satisfying Columb friction. Values of the DEM parameters used.
The widely used non-linear visco-elastic Hertz-Mindlin
Name of the variable Symbol Value
no slip contact model (Tsuji et al., 1992) was adopted in
all simulations reported in this paper. For simulations Sphere radius [mm] R 2
Particle density [kg/m3] qp 3000
involving spheres, rolling friction was found to be neces-
Poisson’s ratio m 0.3
sary to produce a pile with a realistic angle of repose, so Young’s modulus [MPa] E 10
a simple rolling friction model first proposed by Zhou Damping coefficient D 0.8
et al. (1999) was used, in which a rolling friction torque Sliding friction between particle and particle lp,p 0.3
acts opposite to the relative rotation between particles Sliding friction between particle and base lp,w 0.2
Rolling friction for the spheres rp,p/w 0.3
and between particles and boundary. A classical Coulomb
 C. Zhou, J.Y. Ooi / Mechanics of Materials 41 (2009) 707–714 709

up the granular pile gradually by pouring more particles resistance and allowed a stable pile to be formed. The sim-
onto the pile and evaluate the pressure distributions at ulation without rolling friction was conducted (not shown
increasing pile sizes. here) and confirmed this occurrence. This also agrees with
Fig. 1 shows the predicted vertical pressure distribution previous studies of Zhou et al. (1999) and others in the use
acting on the horizontal base of single layer granular piles of rolling friction to form stable granular pile. A detailed
for single sphere and paired-sphere assemblies at 3000, discussion on the various rolling friction models in DEM
6000 and 9000 particles. The pressure is normalised with simulations and the influence on granular solid behaviour
the maximum computed value Pmax and the base coordi- can be found in Ai et al. (2007).
nate is normalised with the radius of the pile Rp. For the The granular piles were further explored by studying
non-spherical paired-sphere assembly, the M-shaped ver- the characteristics of the contact force network within
tical pressure distribution with a central pressure dip is the pile. For single layer model, all particle contacts oc-
clearly evident even for the smallest 3000 particle heap. cur within the prescribed plane, so the particle contact
The dip became more pronounced as the pile grew larger. orientation can be evaluated. For the non-spherical parti-
On the other hand for the spherical assembly, the pressure cle, contact orientation is defined as the contact angle
dip was not present at 3000 particles, could be discerned made by the constituent spheres of the two particles in
from the 6000 particle case and became pronounced only contact as shown in Fig. 2. Fig. 3 shows the distribution
for the 9000 particle case. There is also a significant loss of the contact orientation for all the particle–particle
of symmetry in the base pressure for the spheres due to contacts in the pile presented as radial plots. In these
significant local rupture events that perturbed the symme- plots, the frequency of occurrence for each contact angle
try (see for example the local plateau of pressure midway is normalised with the maximum value in the total con-
between the apex and the edge of pile). This is related to tact network and 0° represents two particles in contact
the crystalline structure that has formed in the monosized horizontally whilst 90° represents vertical contact
sphere assemblies. It should also be noted that further sim- orientation.
ulations using other combinations of input parameters Since each contact always appears in pair, there exists a
(not shown here) did not always lead to a clear M-shaped 180° rotational symmetry. For the monosized sphere, a
pressure distribution, which is being investigated further. crystalline packing structure has formed producing contact
The results in Fig. 1 suggest that non-spherical particles orientation predominantly in the 60° and 120° orienta-
with a greater degree of particle interlocking produced a tions. For the pile composed of paired-sphere particle,
more robust pressure dip. The base deflection as proposed the contact orientations show a periodic distribution
by some researchers is not a prerequisite for the formation which can be approximated by a second-order Fourier
of the pressure dip since these simulations have predicted expansion with a preferential vertical contact orientation.
convincing pressure dips with a rigid horizontal base. The This distribution compares well with the DEM calculations
comparison between spherical and non-spherical assembly of Nouguier-Lehon et al. (2005) where polydispersed
is also relevant since the great majority of the published polygonal particles were simulated, indicating that this
simulations on granular piles are based on 3D spheres or form of contact distribution may prevail in most naturally
2D circular disks. Sphere with its perfect particle symmetry deposited granular assemblies of non-spherical particles.
has the propensity for excessive particle rotation that can The significant difference in contact orientation between
obscure the phenomena pertaining to real granular solids spherical and non-spherical system shown here suggests
and would not have formed a stable pile on a horizontal caution in extending the results from studies based on
plane. As explained above, rolling friction was introduced 3D spheres and 2D disks to real granular solids that are
for the spherical assembly here which increased the rolling not perfect spheres.

1.1 1.1
Normalised Base Pressure: Pi / P max
Normalised Base Pressure: Pi / P max

1.0 1.0
3K 3K
0.9 6K 0.9 6K
0.8 9K 0.8 9K
0.7 0.7

0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3
0.2 0.2
0.1 0.1
0.0 0.0
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

Normalised Radius r/R p Normalised Radius r/R p

(a) single sphere (b) paired-sphere

Fig. 1. Base pressure distributions for 3000, 6000, 9000 single layer piles.
710 C. Zhou, J.Y. Ooi / Mechanics of Materials 41 (2009) 707–714

in this single layer model with 9000 particles. The orienta-

tion of the paired-sphere particle is defined as the angle
that its long axis makes with the normal to the base as
illustrated in Fig. 2. It can be observed in Fig. 4 that the par-
ticles are mainly oriented horizontally, being the optimally
most stable position. This also explains the dominant ver-
tical contact orientation noted above. Very similar particle
orientation distribution has been observed in photoelastic
granular pile experiments on elliptic cylinders as shown
in Fig. 4b (Zuriguel and Mullin, 2008).

4. Planar wedge model

Next, we extend the single layer model to wedge-

shaped granular pile, which has been reported in studies
of the pressure distribution beneath a sandpile (Vanel
Fig. 2. Contact orientation and particle orientation for non-spherical
particles in contact. et al., 1999; Zhou et al., 2003). The wedge pile was simu-
lated by confining with front and rear frictionless vertical
walls to form a wedge pile with a thickness of 16 times
It is also interesting to explore the orientations of the the particle radius. Simulations were carried out with
particles in the granular pile formation. Fig. 4 shows the paired-sphere particles being centrally charged over the
particle orientation distribution for the paired–sphere pile whole thickness onto a rigid horizontal base.

Fig. 3. Distribution of contact orientation in the single layer pile with 9000 particles.

Fig. 4. Comparison of particle orientation distribution in simulation and experiment.

 C. Zhou, J.Y. Ooi / Mechanics of Materials 41 (2009) 707–714 711

Several researchers have studied the influence of heap to the dimension of the pile is an important scaling issue
construction history on the formation of the pressure dip which is under further investigation.
e.g. (Vanel et al., 1999). Recent 2D photoelastic experi-
ments carried out by Geng et al. (2001) provides evidence 5. Conical pile model
of the effect of width and falling height of the pouring jet.
Here, planar wedge DEM models of granular piles were This section presents the full 3D conical pile simulations
simulated using two pouring jet dimensions to investi- using paired-sphere particles. The common proposition in
gate this effect. Fig. 5 compares the granular piles and explaining the pressure dip is some form of arching effect
the associated vertical base pressures underneath the arising from the formation process which results in a sig-
pile formed from two pouring jet dimensions: a wide ex- nificant part of the weight at the central zone being pro-
tended pouring jet (80R width) and a narrow concen- gressively carried by the outer radial zone. It is thus
trated jet (8R width). The results show that the angles interesting to compare the results from the single layer
of repose arising from the two pouring dimensions are model and the planar wedge model to this full 3D conical
the same at 32°, so the difference in the base pressure pile where the arching action is significantly different.
distribution cannot be attributed to the repose angle or Fig. 6a shows the vertical base pressure under conical
self weight of the material. It is evident that for this pair piles of 5000, 10,000, 15,000 and 20,000 particles. Axial
of simulations, the run with a concentrated jet displayed symmetry was assumed in the pressure evaluation with
a distinct dip whereas the one with the extended pour- the averaging segmental area being an annulus section,
ing jet did not. This supports the proposition that the so only half of the pressure profile is shown. Similar mov-
method of pile construction play a significant role in ing averaging was deployed so that every statistically pos-
the formation of the pressure dip. The results suggest sible pressure evaluation has been plotted. The averaging
that the concentrated jet charging leading to material areas started from a circle with a radius of 6R at the centre
avalanching down the conical slope during formation of the conical pile, and then incrementally shifted by 1R.
may be important in the pressure dip phenomenon. With The results showed the evolution of the pressure dip as
the extended pouring source, this process may have been the pile built up, which appears to have not been reported
somewhat disturbed to the point that the resulting pack- by any previous numerical studies. The base pressure ap-
ing structure no longer exhibits the M-shaped pressure pears to approach an asymptotic M-shaped pressure distri-
distribution. Further experimental results (Ooi et al., bution as the pile size increases. The results may be
2008) has indicated that even for an extended pouring compared with the physical stockpile experiments of Ooi
jet, the pressure dip will eventually take form as the pile et al. (2008) using mini iron ore pellets (Fig. 6b) where
grows larger. The dimension of the charging jet relative much similarity can be discerned. Also, Ooi et al.’s experi-

Width of pouring jet = 80R Width of pouring jet = 8R

1.1 1.1
1.0 1.0
Normalised Base Pressure: Pi / P max
Normalised Base Pressure: Pi / P max

0.9 0.9
0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0.0 0.0
-1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.00 -1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.00
Normalised Radius r/Rp Normalised Radius r/R p
(a) extended pouring jet (b) concentrated pouring jet

Fig. 5. Base pressure profiles for 3D wedge pile of 5000 paired-sphere particles using two different pouring jets.
712 C. Zhou, J.Y. Ooi / Mechanics of Materials 41 (2009) 707–714

r/dp
0 37 74 111 148 18
4.5

4 Test 1 (Pouring jet Rj=16mm)

3.5

Pressure (kPa)
3

2.5

1.5

0.5

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Normalised radius r/Rp

(a) conical pile simulations (b) iron ore pellets experiments (Ooi et al., 2008)

Fig. 6. Base pressure profiles for 3D conical stockpiles: (a) current DEM simulation; (b) experimental results.

ments showed a dip which approaches 50% of the maxi- To further probe the magnitude of the contact force
mum base pressure measured at a radial coordinate of within the pile, the probability density function for the
0.3Rp: this also compares reasonably well with the DEM contact force F (normalized with the mean contact force
prediction here. Comparing with the single layer and pla- for the whole population) is shown in a semi-logarithmic
nar wedge models, the conical pile appears to produce a plot in Fig. 9. The data shows that the strong contact forces
larger central dip which is probably due to the greater de- (defined as F > 1) are well described by an exponential
gree of arching in a conical pile compared with the planar function of the form P(F) = CenF indicating that the occur-
arching in the previous two models. rence of the strong forces decreases exponentially with the
It is reasonable to assume that the averaging length- magnitude of the force and there is a significant scatter at
scale of 6R chosen above can have a significant influence the tail end where the contact forces are large. For the
on the computed pressure distribution especially where weak contact forces (defined as F < 1), the distribution
the pressure is changing rapidly and the numerical pile is deviates away from the exponential. These weak contact
relatively small. The sensitivity to averaging lengthscale forces constitute the majority of the contacts in the system
is explored further using 4R, 5R, 6R and 7R averaging seg- (>70% of all contacts), which is in agreement with the
ments and the results are shown in Fig. 7. The scatter in experimental observations of Geng et al.’s (2001) and the
the computed pressure increases as the averaging length numerical results of Radjai et al. (1998). Similar contact
decreases. This is expected since at small lengthscale, the force distribution has been reported in other densely
discrete nature of the complex local contact force structure packed granular solid systems (Coppersmith, 1997; Liu
dominates, resulting in highly fluctuating averaged forces. et al., 1995).
To assist in characterising the pressure distribution and the
magnitude of the central pressure dip, we propose an 6. Conclusion
empirical equation of the form:
h pr i Discrete element simulations of granular piles using
p ¼ acos b  cer=d ð1Þ spherical and non-spherical particles have been presented.
2
For spherical assembly, a stable granular pile with a realis-
with four fitting coefficients a, b, c and d and performed
tic angle of repose is predicted only when rolling friction is
non-linear regression best fit to each of the four cases,
incorporated. The resulting base pressure profile appears
which are also plotted in Fig. 7. The best fit trends for the
more erratic with a less robust central pressure dip than
four averaging lengths turn out to be very similar: the cen-
the non-spherical particles. We suggest that sphere with
tral minimum pressure, the maximum base pressure and
its perfect symmetry has the propensity for excessive rota-
its radial location are not particularly sensitive to the
tion that can obscure the phenomena pertaining to real
choice of these four averaging lengths trialed, which lends
granular solids. The rest of the paper focused on non-
credence to base pressure evaluation here. Further infor-
spherical particle in the form of a paired-sphere shape with
mation on the number of contacts per averaging segment
an aspect ratio of 1.5. The results show that the paired-
is shown in Fig. 8 for 20 K particle granular pile with 6R
sphere assembly produced a robust M-shaped pressure
averaging length. The number of contacts per segment in-
profile with a significant dip under the apex when a suffi-
creases almost linearly as the averaging area increases,
cient pile size was achieved. Apart from the smallest 5000
with more than 100 contact forces per averaging length
particle conical pile simulation, all paired-sphere simula-
for radial coordinate greater than 20% of pile radius. The
tions reported in this paper produced a central pressure
average contacts per segment turn out to be relatively con-
dip.
stant at 0.043 contact per mm2.
 C. Zhou, J.Y. Ooi / Mechanics of Materials 41 (2009) 707–714 713

1.2 1.2

1.0 1.0
Base Pressure (KPa)

Base Pressure (KPa)

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Normalised Radius r/Rp Normalised Radius r/Rp

(a) averaging length = 4R (b) averaging length = 5R

1.2 1.2

1.0 1.0

Base Pressure (KPa)

0.8 0.8
Base Pressure (KPa)

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Normalised Radius r/Rp Normalised Radius r/Rp
(c) averaging length = 6R (d) averaging length = 7R

Fig. 7. Base pressure evaluated with different averaging segmental lengths.

500
No. of contacts per averaging segment

-1
400 10

300
-2
10
P(Fi)

200

-3
10
100

-4
0 10
0.0 0.2 0.4 0.6 0.8 1.0 0 1 2 3 4 5 6 7 8
Normalised radius r/Rp Fi / Fmean

Fig. 8. Number of contacts in averaging segment for a 3D conical pile Fig. 9. Normalized contact force distribution function P(Fi) for a 3D
with 20,000 particles. conical pile with 20,000 particles.

The results support the proposition that the M-shaped

pressure distribution is a robust and natural pressure distri- an asymptotic M-shaped pressure distribution as the pile
bution for a typical stockpile where the scale is considerably size increases, which is in good agreement with experimen-
larger than the mean particle size. In addition, the results tal observations. The dimension of the pouring jet relative to
indicate that base deflection is not prerequisite for the pres- the dimension of the pile is an important scaling issue which
sure dip whereas a relatively concentrated pouring jet with is under further investigation. The granular fabric in terms of
respect to the pile dimension appears to be essential. The magnitude of the contact force, contact orientation and par-
base pressure distribution under progressive development ticle orientation within a granular pile have also been
of the granular pile suggests the base pressure approaches described.
714 C. Zhou, J.Y. Ooi / Mechanics of Materials 41 (2009) 707–714

Typical industrial stockpiles can easily contain over Head, D.A., Rodgers, G.J., 1997. Slowly driven sandpile formation with
granular mixtures. Physical Review E 56, 1976.
1010 particles. Recent advances in parallel implementation
Herrmann, H.J., Luding, S., 1998. Modeling granular media on the
of DEM methodology coupled with increasingly cheap and computer. Continuum Mechanics and Thermodynamics 10, 189–231.
powerful CPUs mean that it is now quite possible to model Itasca Consulting Group Inc., 2003. PFC3D – Particle Flow Code in 3
large stockpiles of the order of 105–107 particles or more. Dimensions, Version 3.0, Minneapolis, USA.
Liu, C.H., Nagel, S.R., Schecter, D.A., Coppersmith, S.N., Majumdar, S.,
However the present study suggests that the pressure dip Narayan, O., Witten, T.A., 1995. Force fluctuations in bead packs.
phenomenon develops from relatively early stages of pile Science 269, 513–515.
formation. One of the largest scale experimental investiga- Luding, S., 1997. Stress distribution in static two-dimensional granular
model media in the absence of friction. Physical Review E 55, 4720.
tion of stockpile confirmed that this is indeed the case McBride, W., 2006. Base Pressure Measurements under a Scale Model
(McBride, 2006). The capability of DEM to predict the pres- Stockpile. Particulate Science and Technology 24, 59–70.
sure dip phenomenon with a promising agreement with Nedderman, R.M., 1992. Statics and Kinematics of Granular Materials.
Cambridge University Press, Cambridge, U.K.
the experiments shows the potential of using this tech- Nouguier-Lehon, C., Vincens, E., Cambou, B., 2005. Structural changes in
nique to explore the particle scale mechanics and the asso- granular materials: the case of irregular polygonal particles.
ciated granular fabric which produce bulk responses that International Journal of Solids and Structures 42, 6356–6375.
Ooi, J.Y., Ai, J., Zhong, Z., Chen, J.F., Rotter, J.M., 2008. Progressive pressure
are still poorly understood. measurement beneath a granular pile with and without base
deflection. In: Chen, J.F., Ooi, J.Y., Teng, J.G. (Eds.), Structures and
Acknowledgements Granular Solids – from Scientific Principles to Engineering
Application. CRC Press/Balkema, Edinburgh, UK, p. 87.
Radjai, F., Wolf, D.E., Jean, M., Moreau, J.-J., 1998. Bimodal Character of
The first author holds a Dorothy Hodgkin Postgraduate Stress Transmission in Granular Packings. Physical Review Letters 80,
Award sponsored by the UK Engineering and Physical 61.
Sciences Research Council (EPSRC) and Shell. The authors Reydellet, G., Clément, E., 2001. Green’s Function Probe of a Static
Granular Piling. Physical Review Letters 86, 3308.
are very grateful for the helpful discussions with Mr. Jun Savage, S.B., 1998. Modeling and granular material boundary value
Ai, Dr. Jianfei Chen and Prof. Michael Rotter. problems. In: Hermann, H.J., Hovi, J.P., Luding, S. (Eds.), Physics of
Dry Granular Media. Kluwer Academic, Boston, Mass, Dordrecht, pp.
25–98.
References Smid, J., Novosad, J., 1971. Pressure cell for measuring normal and shear
stresses. Powder Technology 4, 322–327.
Ai, J., Chen, J.F., Ooi, J.Y., Rotter, J.M., 2007. Assessment of rolling friction/ Smid, J., Novosad, J., 1981. Pressure distribution under heaped bulk solids,
resistance models in discrete element modeling. Internal Report, Proceedings 1981 Powtech Conference, Institution of Chemical
University of Edinburgh. Engineers Symposium Series No. 63, pp. 12.
Atman, A.P.F., Brunet, P., Geng, J., Reydellet, G., Claudin, P., Behringer, R.P., Tsuji, Y., Tanaka, T., Ishida, T., 1992. Langrangian numerical simulation of
Clément, E., 2005. From the stress response function (back) to the plug flow of cohesionless particles in horizontal pipe. Powder
sand pile ‘‘dip”. The European Physical Journal E – Soft Matter V17, Technology 71, 239–250.
93–100. Vanel, L., Howell, D., Clark, D., Behringer, R.P., Clément, E., 1999. Memories
Brockbank, R., Huntley, J.M., Ball, R.C., 1997. Contact force distribution in sand: experimental tests of construction history on stress
beneath a three-dimensional granular pile. Journal de Physique I 7, distributions under sandpiles. Physical Review E – Statistical
1521–1532. Physics, Plasmas, Fluids, and Related Interdisciplinary Topics, 60.
Cates, M.E., Wittmer, J.P., Bouchaud, J.P., Claudin, P., 1998. Development of Yu, A.B., 2004. Discrete element method: an effective way for particle
stresses in cohesionless poured sand. Philosophical Transactions of scale research of particulate matter Engineering Computations.
the Royal Society A: Mathematical, Physical and Engineering Sciences International Journal for Computer-Aided Engineering 21, 205–
356, 2535–2560. 214.
Coppersmith, S.N., 1997. Force fluctuations in granular media Physica D. Zhou, Y.C., Wright, B.D., Yang, R.Y., Xu, B.H., Yu, A.B., 1999. Rolling friction
Nonlinear Phenomena 107, 183–185. in the dynamic simulation of sandpile formation. Physica A:
Cundall, P.A., Strack, O.D.L., 1979. Discrete numerical model for granular Statistical Mechanics and its Applications 269, 536–553.
assemblies. Geotechnique 29, 47–65. Zhou, Y.C., Xu, B.H., Zou, R.P., Yu, A.B., Zulli, P., 2003. Stress distribution in
DEM Solutions Ltd., 2007. EDEM discrete element code, version 1.2, a sandpile formed on a deflected base. Advanced Powder Technology
Edinburgh, UK. 14, 401–410.
Geng, J., Longhi, E., Behringer, R.P., Howell, D.W., 2001. Memory in two- Zuriguel, I., Mullin, T., 2008. The role of particle shape on the stress
dimensional heap experiments. Physical Review E – Statistical, distribution in a sandpile. Proceedings of the Royal Society A:
Nonlinear, and Soft Matter Physics, 64. Mathematical, Physical and Engineering Sciences 464, 99–116.