PTEC-11980; No of Pages 10

Powder Technology xxx (2016) xxxxxx

Contents lists available at ScienceDirect

Powder Technology

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

A coupled FEM/DEM model for pipe conveyor systems: Analysis of the

contact forces on belt
Q.J. Zheng a,, M.H. Xu a, K.W. Chu a, R.H. Pan b, A.B. Yu a,c
a
Laboratory for Simulation and Modelling of Particulate Systems, Department of Chemical Engineering, Monash University, Clayton, Vic 3800, Australia
b
Longking Bulk Materials Science and Engineering Co., Xiamen 361000, China
c
Monash University-Southeast University Joint Research Institute, Suzhou, China

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

Article history: The use of Pipe Belt Conveyor (PBC) has gained more and more popularity in bulk solids handling. Compared to
Received 30 May 2016 the traditional trough conveyors, PBC can allow tighter curves and steeper gradients of conveyor routes and thus
Received in revised form 20 September 2016 better suits the applications in difcult terrains. It has many unique mechanical characteristics which are not well
Accepted 27 September 2016
understood yet. This paper proposes a coupled nite element model (FEM) and discrete element method (DEM)
Available online xxxx
model to investigate the mechanics of a PBC system, with particular reference to the distribution of contact force
Keywords:
in the pipe section. This FEM/DEM model considers comprehensively the formation of pipe from a at belt and
Bulk solids handling the microscopic structure of discrete particles under gravity, and thus can well describe the states of both convey-
Pipe belt conveyor or belt and bulk solids. The predictions of the contact forces are shown to agree well with the previous experi-
Finite element method mental data under different conditions. Using this approach, the distribution of contact forces under different
Discrete element method load conditions, i.e. zero gravity (ZG), empty pipe (VF 0%), volume ll of 40% (VF 40%) and 80% (VF 80%) are ob-
Contact force tained and their dependencies on the rotation angle of pipe are investigated.
Granular materials 2016 Elsevier B.V. All rights reserved.

1. Introduction
The Pipe Belt Conveyor (PBC), invented by Japan Pipe Conveyor
(JPC) in early 1980's, has become more and more popular in bulk solids
handling [13]. This type of conveyor uses a pipe-shaped belt to enclose
granular materials during conveyance, which can effectively prevent
dust emission or other environmental pollutions. It is particularly suit-
able for difcult terrains due to its high adaptability to tight curves
and steep gradients of conveying route. However, PBC differs from the
conventional trough conveyor in many mechanical aspects and requires
special considerations in design. For example, the choice of the trans-
verse stiffness of belt is more important in PBC than in trough
conveyorinsufcient stiffness may result in undesirable collapse and
misalignment of pipe while over large stiffness can signicantly in-
crease the energy requirement and capital investment [4]. More funda-
mental researches are needed in this area in order to achieve better
operational safety and economic prots.
In designing a PBC, several points are usually important: the coef-
cient of main friction between belt and rolling idlers, and how to ensure
the electric motor and belt run steadily without abrupt uctuations. The
main friction is mainly concerned in the power calculation of PBC, which
is in theory determined by the properties of belt and granular material
Corresponding author.
E-mail address: Qijun.Zheng@monash.edu (Q.J. Zheng).
[513]. Nuttall et al. [9] derived an analytical model of the rolling resis-
tance between roller and belt by taking into account the viscoelasticity
of the cover material. Wheeler et al. [14] studied another type of resis-
tance in PBC, i.e. exural resistance, resulting from the successive lift/
sag of granular material when passing through the idler panel. The sec-
ond important issue is mainly related to the dynamic characteristic of a
PBC. Such a belt may be deemed as a long elastic spring supplemented
with viscous dampers and distributed masses, including the masses of
belt and bulk material as well as the rotational inertia of idlers and ac-
cessories. Therefore, a large stress wave is often generated in the belt
when starting or braking the conveyor, which should be properly
smoothed by adjusting the loading curve [1518].
Investigating the contact force between belt and idler rolls is of fun-
damental signicance to understand the behaviours of conveyor, be-
cause these forces determine the value of rolling resistance [7] and
affect the overall energy consumption and dynamic behaviours of PBC.
It is also useful to understand the mechanism behind the abnormal ro-
tation of pipe during conveyance [19]. Htte et al. [20] measured the
quasi-static contact forces in a curved pipe and quantied their depen-
dencies on the radius of curve, the belt construction, and the belt ten-
sion. Michalik and Zajac [21] designed a computer integrated system
for the evaluation of the contact forces on idlers and the strains at differ-
ent locations of a belt. Using a similar test rig, Molnar et al. [22,23] sur-
veyed the idler forces in the transition region of conveyor and
formulated an empirical equation based on the measured data. Another

http://dx.doi.org/10.1016/j.powtec.2016.09.070
0032-5910/ 2016 Elsevier B.V. All rights reserved.

Please cite this article as: Q.J. Zheng, et al., A coupled FEM/DEM model for pipe conveyor systems: Analysis of the contact forces on belt, Powder
Technol. (2016), http://dx.doi.org/10.1016/j.powtec.2016.09.070
2 Q.J. Zheng et al. / Powder Technology xxx (2016) xxxxxx
work by Molnr et al. [24] highlighted the asymmetrical distribution of
contact force due to the asymmetry of the belt tension. Zamiralova and
Lodewijks [2527] reviewed the usable experimental devices for deter-
mining the contact force including the six-point bending test, the quasi-
static test equipped with different frames and the dynamic test. Using
the six-point bending test device, Zamiralova and Lodewijks [19] inves-
tigated the inuences of pipe diameter and overlap location on the spa-
tial distribution of contact force. The force behaviours of pipe were
shown to be rather complex and sensitive, although the effects of fric-
tion and longitudinal tension had been excluded in their measurements.
Recently, Liu et al. [28,29] also proposed a stress discontinuity approach
to predict the pressure prole on conveying belt and veried their ap-
proach against experimental measurement.
Many studies of PBC have been conducted by means of nite ele-
ment method (FEM) in the literature. del Coz Daz [30] investigated
the warping effect of belt in a curved pipe and its correlation to the
stress and strain distributions. Fedorko et al. [31] simulated the mould-
ing process of belt and estimated the contact forces acting on the idler
rolls. Pang and Lodewijks [32] utilised FEM to calculate the contact
forces of a short pipe and compared the numerical results with experi-
mental measurements. However, all these FEM studies accounted for
the structural components of conveyor only, i.e. the belt and idlers,
but not granular materials and their dynamics during conveying. Re-
cently, a coupled FEM and discrete element method (DEM) approach
has been used to predict the belt deection of conventional trough con-
veyor [33] and the bulk exural resistance [34]. However, to date no
such attempts have been made for pipe conveyors.
Fig. 1. (a) Industrial installation of pipe conveyor; (b) the loading point and (c) the general concept and structural arrangement [3].
Please cite this article as: Q.J. Zheng, et al., A coupled FEM/DEM model for pipe conveyor systems: Analysis of the contact forces on belt, Powder
Technol. (2016), http://dx.doi.org/10.1016/j.powtec.2016.09.070
This paper investigates the contact forces of pipe conveyor by use of
a coupled FEM/DEM approach, in which the conveying belt is modelled
by FEM and granular materials by DEM. In what follows, we will present
in sequence the theory and the implementation of FEM/DEM approach,
the model validation against previous experiments, the outcome of con-
tact forces under a variety of loads, and nally a summary of the major
ndings.
2. Model setup
2.1. FEM model of conveying belt
The belt manufactured for conveying purpose is usually a composite
of steel cord (or fabrics) and rubber covers, as shown in Fig. 1. The steel
cord serves mainly to carry the longitudinal tension, while the rubber
together with some lateral reinforcements sustains the transverse
bending of belt. The conveying belt has three major mechanical features
that need to be reected in FEM model. First, the belt is in essence
orthotropic, with disparate stiffness in the longitudinal and transverse
directions as mentioned above. Secondly, the belt has remarkable vis-
cosity due to the constituent of rubber, and can dissipate energy when
dynamically tensioned or bent. This viscosity comes into play mainly
in dynamic conditions, whilst is relatively less important in quasi-static
situations where the strain rate of material is slow. Thirdly, the bending
stiffness is non-uniform along the belt width, but usually reduced in the
edge area in order to ensure a good seal of pipe. However, there is by far
no standard with regard to the reduced value of stiffness and how wide
 Q.J. Zheng et al. / Powder Technology xxx (2016) xxxxxx 3

the edge area should be. In this work, it is approximately estimated by a The contact force between particles is described by a spring-dashpot
calibration against experimental results [19]. model in the normal direction, and a spring-dashpot-slider model along
Based on the above considerations, the belt is treated as an the tangential direction. The normal contact stiffness kn derived from
orthotropic elastic shell in FEM, described by the laminate constitutive the Hertzian theory [40] varies with the relative approach of particles,
model [35]: given by
8 9 2 38 9
>
>
>
11 >
>
>
1=E1 21 =E2 31 =E3 0 0 0 >
>
>
11 >
>
> 4  p
>
>
> > 6 12 =E1
22 >
> 1=E2 32 =E3 0 0 0 7 >
7>
> >
> kn E Re n 4
< = 66 < 22 > = 3
33 13 =E1 23 =E2 1=E3 0 0 0 7
6 7 33
1
> > 6
> 12 > 6 0 0 0 1=G12 0 0 7 7>
> 12 >
>
>
>
>
>
> 4
13 >
>
> 0 0 0 0 1=G13 0 5>
>
>
> >
>
> where E = 1/((1 2p1)/Ep1 + (1 2p2)/Ep2) is the effective contact
: ; : 13 > ;
23 0 0 0 0 0 1=G23 23 modulus and Re = Rp1Rp2/(Rp1 + Rp2) is the effective contact radius. Ep1,
Ep2, p1, p1 are the Young's moduli and Poisson's ratios of contacting
where ij, ij denote the logarithmic strain and Cauchy stress respective- spheres. n denotes the overlapping distance between two spheres.
ly. E1, E2, E3 are the elastic moduli along the length, width, and thickness In theory, the tangential contact between two elastic frictional
of belt. There is little information about the Poisson's ratios 12, 23, 13 spheres involves a complicated stick-slip mechanism. In DEM simula-
and shear moduli G12, G13, G23 of conveying belt in the literature [19] be- tion, however, some simplied models are also usable without much in-
cause it is impractical to perform a standard shear test on an inhomoge- uence on the accuracy of contact force [41,42]. The initial tangential
neous thin belt. For simplicity, the Poisson's ratios are taken to be the stiffness derived by Cattaneo [43] and Mindlin [44] under condition of
same as that of usual rubbers, and the shear moduli are evaluated in constant normal force is adopted here, given by
terms of the Poisson's ratio and the transverse modulus E2 using the
p
equation G12 = E2/2(1 + 12). The geometrical parameters and density kt 8G Re n 5
of the belt are listed in Table 1, which are the same as those used in
the experiment of Zamiralova and Lodewijks [19]. where G = 1/((2 p1)/Gp1 + (1 p2)/Gp2) is the effective shear
The idler rolls are treated as rigid shells in the FEM model since the modulus. Gp1 , Gp2 are the shear moduli of the two spheres in contact.
idlers are usually made of steel and far stiffer than the belt. According The damping coefcient used is a dimensionless fraction of the critical
to the Hertzian contact theory, the shape of idler may inuence the p
damping 4mkn , and has the same value in the normal and tangential
area and stiffness of the contact with belt [36]. However, since such con-
directions [35]. The rolling friction is not used in the DEM model. In-
tact details are not the focus of this study, the cylindrical idlers are sim-
stead, a mild non-sphericity of particle is introduced to provide resis-
plied to plates to reduce the computational cost.
tance to particle rotation and decrease the owability of granular
system, which is implemented by adhering two component spheres
2.2. DEM model of granular materials
with a rigid coupling constraint. For details, see Table 2 and Section 3.3.
It is noted that a granular material conveyed by PBC may also be
The DEM proposed by Cundall and Strack [37] in 1980's is a rst-
modelled as a continuum and described with elastoplastic theory [45
principle approach to simulate granular materials. It takes into account
47]. The continuum modelling is more efcient than DEM in terms of
the interactions between particles or between particles and surround-
computational cost although it may disregard some microscopic infor-
ing media and tracks the mobilization of each particle in light of the
mation of granular ow. Such a continuum approach is not used here
Newtonian law. No assumptions on the constitutive relationship of
but may be an attractive option for industrial applications.
granular materials are needed in this approach. A good review of its the-
oretical background and major applications has been presented by Zhu
2.3. Interaction between structural and discrete elements
et al. [38,39]. Briey, the governing equations of DEM are expressed as

dV i X In the FEM/DEM, as mentioned above, the structural components of

mi F ij mi g 2 PBC such as belt and idler rolls are simulated by FEM, while the granular
dt j materials inside the pipe are handled with DEM. The nite elements and
discrete particles interact mainly through the contacts at the interfaces/
di X boundaries. In the previous coupled DEM and FEM simulation [48,49],
Ii M ij 3
dt j there is usually a need to communicate information between DEM
and FEM solvers, i.e. passing the deformed mesh of FEM into DEM as
where mi, Ii are respectively the mass and rotational inertia of particle i.
Vi and i are the translational and rotational velocities. g is the gravity Table 2
acceleration. Fij and Mij denote respectively the force and torque on Parameters of the glued-sphere particles used in DEM.
particle i resulting from neighboring particle j. Baseline Ranges of
Parameters values variation

Radius of component sphere, Rp (mm) 8

Table 1 Aspect ratio of DEM particle, dmax/dmin 1 11.25a
Parameters of conveying belt. Density of component sphere, (kg/m3) 2500 20502500b
Particle elastic modulus, Ep (MPa) 10
Parameters Baseline values Ranges of variation
Poisson's ratio, p 0.3
Nominal diameter of pipe, D (mm) 500 450530 Interparticle sliding friction, p 0.3 0.10.7
Belt width, w (mm) 1794 Normal and tangential damping coefcients, cn 0.3
Belt thickness, h (mm) 22.2 and ct
Belt length, L (mm) 152 a
Note that the rolling friction is not available in the package of Abaqus. To overcome the
Belt density, (kg/m3) 1267
excessive rotation of spheres, a moderate non-sphericity of particles is introduced in the
Transverse elastic modulus, E2 (MPa) 23
simulation via a glued-sphere approach, i.e. rigidly coupling the translational and rotation-
Longitudinal elastic modulus, E1 (MPa) 400
al freedoms of two component spheres to form a non-spherical particle shape.
Poisson's ratio, 12 0.4 b
In the glued-sphere assembly, some volumetric overlap may exist between the com-
Shear modulus, G12, G13, G23 8.3
ponent spheres leading to an increase in the apparent density of DEM particle. The weight
Coefcient of friction at belt surface, 0.05 0.010.3
of bulk solids is important in this study; to keep it comparable for different particle shapes,
Decrease ratio of modulus at belt margin, 0.5 0.31
the density of sphere is varied in accordance with the aspect ratio.

Please cite this article as: Q.J. Zheng, et al., A coupled FEM/DEM model for pipe conveyor systems: Analysis of the contact forces on belt, Powder
Technol. (2016), http://dx.doi.org/10.1016/j.powtec.2016.09.070
4 Q.J. Zheng et al. / Powder Technology xxx (2016) xxxxxx

a boundary for particles, and exerting the DEM results of wall forces on
the structural elements in return. In doing so, correctly treating the
boundaries and wall forces is important to ensure the smooth transfer
of momentum and energy between particles and structures. In this sim-
ulation, since the DEM particles and FEM structures are handled with
the same software package (Abaqus 6.14), the coupling procedure can
be simpler. What a user needs to do is to dene particles as an ele-
ment-based surface and explicitly include this surface in the general
contact domain of the model (it is not included by default) [35]. This
is done in the input le prior to the simulation. Then the general contact
algorithm will monitor the displacements of the nodes (both DEM and
FEM) during the simulation looking for possible penetrations into sur-
faces. When penetrations are detected, the contact forces will be deter-
mined according to the dened contact stiffness/damping and exerted
onto relevant contacting bodies. One question here might be exactly
how the contact forces are applied to the FEM mesh, because normally
the contacting point of particles and structure is not right on the
node of FEM mesh, thus some re-distribution of the force is required.
There is a routine methodology in FEM to handle this issue, using the
so-called shape function to transform the concentrated force acting on
the elemental surface to equivalent forces on the elemental nodes. A de-
tailed discussion of this treatment can be found in [48,50]. Upon know-
ing the nodal forces, the nodal displacements of structural mesh at the
new time step can be calculated with FEM. In this way, a real-time
and two-way coupling of FEM and DEM can be realized.
It should be borne in mind when setting up the coupled model that
the denitions of contact in FEM and DEM are somewhat different. In
FEM, the contact surfaces of solids and structures are physically
impenetrable. The concept of contact stiffness is introduced mainly for
a numerical reason, i.e. to decide the tolerance of penetration of two sur-
faces in the numerical algorithm. Likewise, the contact damping is used
to ensure the stability of numerical solution only. By contrast, in DEM,
the contact stiffness and damping have explicit physical
meaningsthey stand for the mechanical response of a particle under
external loads. Therefore, dening the contact properties is no longer
a trivial numerical issue when DEM particles are included. They must
be determined based on the contact mechanics described in Eqs. (4)
and (5).
The time increments in FEM and DEM should be approximately
matched; otherwise, the coupling method will choose the smaller
time increment to ensure the numerical stability causing great waste
of computational resource. The elastic modulus of DEM particles and
the characteristic length of FEM elements can be adjusted to meet this
goal without much inuence on the reliability of simulation results.
2.4. Simulation conditions and procedures
Fig. 2 illustrates the procedures for simulating pipe conveyor with
coupled FEM/DEM approach. To faithfully represent the mechanical
state of the belt, it is important to note that the belt is stress-free
when it is at and involves stresses when it is bent to form a pipe.
Thus the modelling should start with a at belt; otherwise, if the belt
is initially a pipe, the stress resulting from belt bending should be spec-
ied simultaneously which is, however, unknown. There are several
ways to mould the at belt into a pipe in FEM, for example, by exerting
two torques at the belt edges [31]. The moulding process is here

Fig. 2. The coupled FEM/DEM simulation of a nite-length model PBC: (a) initial congurations of DEM particles (blue), belt (green) and an auxiliary cylinder (red) to aid the moulding of
belt (the idler rolls are placed beyond the viewpoint initially and will be moved inward later). Although not shown, two rigid planes exist in the front and rear of the PBC model to prevent
the longitudinal ow of particles, playing a similar role to the periodic boundary condition used in DEM; (b) bending the belt around the cylinder by imposing a pressure on the top surface;
(c) rotating the pipe to a certain angle; (d) shifting the idlers to enclose the pipe; (e) removing the red cylinder and applying gravity; (f) the nal conguration of PBC. (For interpretation of
the references to color in this gure legend, the reader is referred to the web version of this article.)

Please cite this article as: Q.J. Zheng, et al., A coupled FEM/DEM model for pipe conveyor systems: Analysis of the contact forces on belt, Powder
Technol. (2016), http://dx.doi.org/10.1016/j.powtec.2016.09.070
 Q.J. Zheng et al. / Powder Technology xxx (2016) xxxxxx
Table 3
Load cases.
Loads Particle number Effective gravity force
Zero gravity (ZG) Nil Nil
Volume ll 0% (VF 0%) Nil Belt mass 101.9 N
Volume ll 40% 2890 Belt mass 101.9 N +
(VF 40%) bulk solids 150.8 N
Volume ll 80% (VF 80%) 5920 Belt mass 101.9 N + bulk solids
310.4 N
completed with the help of an auxiliary cylinder and a pressure acting
on the top surface of the belt as shown in Fig. 2ae (the radius of the cyl-
inder is somewhat smaller than that of pipe to tolerate the thickness of
belt). The idlers are arranged far away from the belt at the beginning so
as not to impede its movement. They are shifted later on to enclose the
belt at the end of moulding step. Thereafter, the pressure on the belt top
and the auxiliary cylinder are dismissed as they are no longer necessary
for retaining the pipe shape. To account for the effect of overlap location
on contact force, the pipe may be rotated by a certain angle during this
process as shown in Fig. 2b and c.
The spherical particles are placed loosely inside the cylinder at the
initial stage as illustrated in Fig. 2a. Due to the absence of gravity and
the protection of the auxiliary cylinder, these particles keep stationery
in the course of belt moulding. Upon the removal of cylinder and the ap-
plication of gravity, the particles settle down quickly on the conveying
belt and form a relatively dense packing, as shown in Fig. 2ef. Different
lling ratios of the pipe have been considered in this work as shown in
Table 3.
The model of pipe conveyor considered here is basically quasi-static.
Some dynamic effects are neglected in this model, such as the periodic
sag/opening of the belt and active/passive state transition of granular
materials when passing the idler hexagon. The practical background of
this static set-up is the six-point bending test device used in industries
to assess the tness between the transverse stiffness of belt and pipe di-
ameter. Studying the belt-idler interaction in this static situation can
provide general guidelines for industrial design. Furthermore, we con-
sider that the dynamic effects mainly come from the sag of the belt. In
conventional trough conveyor, the belt sag is relatively large and the
vertical jumps and rearrangement of particles are signicant. In pipe
conveyor, however, the sag ratio is very small because of the circular
shape and the dynamic forces should be insignicant. More important-
ly, we select this set-up for the convenience of model validation. In a dy-
namic situation, many factors can come into effect, and consequently, it
Fig. 3. (a) The conveyor belt simulated by FEM (colors show the transverse stress 22 on the outer side of belt in units Pa); (b) the spherical particles modelled by DEM in case of volume ll
80% (colors shows the resultant normal contact forces of particle in units N).
Please cite this article as: Q.J. Zheng, et al., A coupled FEM/DEM model for pipe conveyor systems: Analysis of the contact forces on belt, Powder
Technol. (2016), http://dx.doi.org/10.1016/j.powtec.2016.09.070
5
is rather difcult to quantitatively validate the FEM/DEM approach.
Comparatively, the static set-up is well-controlled, allowing to explore
some important factors, e.g. the belt stiffness and friction, and gain
more insights into the pipe conveyor system.
3. Results and discussion
3.1. Stress and force information obtained by FEM/DEM
Fig. 3 shows the mechanical states of the belt and bulk solids simu-
lated by FEM/DEM under the load condition of VF 80%. The structural
components and DEM particles are shown separately in this gure be-
cause their mechanical states are represented by different qualities, i.e.
the bulk stress and inter-particle force respectively. Special attention
is paid to the transverse stress 22 of the belt because of its close rele-
vance to the contact force between belt and idlers. As seen, a remarkable
transverse stress is produced along the width of belt except for the over-
lapping area around Idler 1 where, as mentioned in Section 2.1, a small-
er stiffness is used. It is also found that the belt stress often undergoes
local decreases in the immediate vicinity of the contact points with
idlers, e.g. Idler 3, 4, 5 while reaching local maximums in the corners be-
tween idlers, much like the localised deformation of a thin structural
shell under indentation [51].
Fig. 4 plots the prole of transverse stress along the belt width. It is
seen that the stress equals zero at the free ends of belt near Idler 1,
then it increases rapidly from Idler 1 to Idler 2 (or 6), and varies period-
ically across the three bottom idlers (3, 4, 5). Of particular interest, the
stress prole seems not affected by the lling level of bulk solids the
stress curves of VF 0%, VF 40% and VF 80% are almost coincident, al-
though the difference in the gravity of bulk solids between the three
cases can be as large as 300 N. It is suggested that although the bulk
solids impose a distributed pressure on the carrying-side surface of
belt, the pressure is largely transmitted to the idlers through the contact
points without much inuence on the belt deformation and stress.
Therefore, the belt-idler contact force Fi can be regarded as a superposi-
tion of two independent termsone resulting from the elastic recovery
of belt, and the other from the gravity of bulk solids and belt, as shown
in subsequent sections.
3.2. Validation of contact force in case of VF 0%
This subsection validates FEM/DEM approach against the experi-
ment of Zamiralova and Lodewijks [19] under the condition of an
6 Q.J. Zheng et al. / Powder Technology xxx (2016) xxxxxx
Fig. 4. Distributions of transverse stress 22 along the belt width for different volume lls.
empty pipe, i.e. VF 0%. For comparison, the simulation parameters take
values from the experiment whenever possible. The transverse modulus
E2 of belt is approximated as
K parameters discussed above.
E2 12  3
6
h
where K=21.23 Nm is the uniform bending stiffness measured in ex-
periment [19]. h=22.2 mm is the thickness of belt.
The coefcient of friction between belt and idlers is not explicitly
given in the paper [19]. Several values are therefore tried in this work
with the results presented in Fig. 5a. It can be seen that the existence
of friction does have an inuence on the contact forces, and the experi-
mental data can be well reproduced with a small friction coefcient
such as = 0.01 or 0.05. This is reasonable because in the experiment
the idlers were wrapped with foil papers and thus the coefcient of fric-
tion ought to be small [19]. Unless otherwise stated, = 0.05 is used
throughout this work.
As mentioned, the belt often involves inhomogeneous transverse
stiffness E2 along the width to enable a good seal of pipe. Therefore, a re-
duced modulus E2 is used in our model for the overlapping area, which
Fig. 5. The contact force evaluated by FEM/DEM: (a) for different friction coefcients of belt surface; and (b) for different transverse moduli of the overlapping area (approximately 0.25 m
from the edge).
Please cite this article as: Q.J. Zheng, et al., A coupled FEM/DEM model for pipe conveyor systems: Analysis of the contact forces on belt, Powder
Technol. (2016), http://dx.doi.org/10.1016/j.powtec.2016.09.070
is approximately 0.25 m from the edge. Fig. 5b shows the results of con-
tact forces obtained with different reduced moduli. As seen, the reduced
modulus in overlap effects a decrease of contact force at Idler 2 and 6
and a sharp rise at Idler 1. Reducing the transverse modulus by half
can satisfactorily reproduce the experimental results of contact force,
which is therefore adopted in all subsequent simulations.
Fig. 6 shows the variation of belt-idler contact force with the change
in the diameter of the pipe. Normally, for a given belt, the smaller pipe
diameter causes the larger bending curvature of the belt and thus re-
quires greater contact force from the idler. But this trend does not
apply to all the cases and the change of overlap length can also exert
complicated inuences. For example, the force on Idler 2 in the pipe of
500 mm is lower than that of either bigger pipe (530 mm) or smaller
pipe (450 mm). But such complex behaviours in the empty pipe may
become secondary in loaded pipes as will be shown later.
Fig. 7 shows the variation of belt-idler contact forces with the rota-
tion angle of pipe . , in fact, measures the angular position of the
overlap relative to the vertical direction. When = 0, the overlap is lo-
cated on the top of the pipe, and if = 180, the overlap is rotated to the
bottom. The vast change of contact force with the rotation of pipe shows
the high sensitivity of this hyperstatic problem. Solving this problem
usually requires a predictive model that can take into account the belt
deformation. Using one set of parameters, the current FEM model quan-
titatively captures the complex variation of contact force shown in Figs.
5 and 6. The good agreements between simulation and experimental re-
sults justify the validation of the FEM model and the estimation of belt
3.3. The contact forces of loaded pipes
The contact forces under load conditions of ZG, VF 0%, VF 40% and
80% are presented in Fig. 8. As discussed in Section 3.1, the contact forces
mainly come from two mechanisms: the elastic deformation of belt and
the gravity of material and belt. In the ZG situation, the gravity is zero,
thus the contact forces result exclusively from the bending deformation
of belt. The largest contact force occurs on the top idler, where the belt
opens most widely and hence require the strongest constraining force.
In the VF 0% condition, the slag of belt under self-weight to a certain ex-
tent compensates the opening tendency and alleviates the forces on
upper idlers (1, 2, 6). On the other hand, the forces on the lower idlers
(3, 4, 5) increase in order to balance the additional belt weight. In
cases of VF 40% and VF 80%, the forces on lower Idler (3, 4, 5) are further
 Q.J. Zheng et al. / Powder Technology xxx (2016) xxxxxx 7

Fig. 6. The distributions of contact force varying with the diameter of pipe. The same belt is used for the three cases.

increased in proportion to the volume content of bulk solids as shown in
Fig. 8, fullling approximately the principle of superposition.
Fig. 9 shows the distribution ratio of contact forces in the pipe, i.e. F i
6
=1 F i , where Fi denotes the force on the i idler. We see that the distri-
bution ratio is very sensitive to the pipe rotation angle in cases of ZG and
VF 0%. In practices, other subtle factors including the length of overlap
and the hysteresis of belt could also come into effect, posing a signicant
difculty for precisely evaluating the contact forces. This however
seems not a serious problem in heavily loaded pipe. As shown in Fig.
9c and d, with the increase of solid ll, the relative ratio of force on
upper Idler (1, 2, 6) decreases and becomes less sensitive to the pipe

Fig. 7. The distributions of contact force varying with the pipe rotation angle.

Please cite this article as: Q.J. Zheng, et al., A coupled FEM/DEM model for pipe conveyor systems: Analysis of the contact forces on belt, Powder
Technol. (2016), http://dx.doi.org/10.1016/j.powtec.2016.09.070
8 Q.J. Zheng et al. / Powder Technology xxx (2016) xxxxxx
Fig. 8. The contact forces under different load conditions.
rotation. It is therefore easier to establish the model of contact force
when the gravity dominates. According to the current FEM/DEM results,
for a normal PBC with a ll rate of about 80%, the force ratios on Idler 1
6 are roughly [0.1, 0.0, 0.25, 0.4, 0.25, 0.0], or [0.0, 0.0, 0.3, 0.4, 0.3, 0.0]
depending on the stiffness of belt. The FEM/DEM model should be useful
in many practices where the mechanical behavior of belt and the parti-
cle ow are both important.
Fig. 9. The distribution ratios of contact force in pipe of different rotation angles under various loads: (a) ZG, (b) VF 0%, (c) VF 40%, (d) VF 80%. The ratio is given by F i =1 F i where Fi
denotes the force on the i idler roll.
Please cite this article as: Q.J. Zheng, et al., A coupled FEM/DEM model for pipe conveyor systems: Analysis of the contact forces on belt, Powder
Technol. (2016), http://dx.doi.org/10.1016/j.powtec.2016.09.070
Fig. 10 further shows the effects of particle properties such as sliding
friction and particle shape on the distribution of contact forces. The non-
sphere is created by adhering two component spheres through a multi-
ple point constraint (MPC) in Abaqus, which rigidly couples all the
translational and rotational freedoms of two component spheres. Its
shape is controlled by the aspect ratio, dmax/dmin where dmin and dmax
are respectively the longer and shorter side length of the particle. Aspect
ratio varies from 1.0 and 1.25 to produce a moderate non-sphericity of
particles in the system. It can be observed that the particle friction and
shape exert a mild inuence on the contact force. A notable difference
between the systems of high ( = 0.1 or 0.5) and low owability
( = 0.7 or non-sphere) occurs at the bottom idler 4. The former is
more like the hydrostatic situation so that the bottom idler is subjected
to the highest pressure. In the latter case, the material is more frictional
and can form some force arches to alleviate the pressure at the bottom.
But in general, the effect is not so strong in the scope of this study. It is
perhaps because the curved shape of belt (boundary condition for par-
ticles) is not favorable for the formation of force arch, and the particles
considered in this work are relatively coarse. More studies are needed
to explore the cases of ner particles in the future.
4. Conclusions
This work proposes a coupled FEM/DEM model for simulating pipe
conveyor systems and utilises this model to investigate the fundamental
contact force acting between conveying belt and structural components.
The obtained results are compared with the previous experimental
measurement in case of an empty pipe. Different load conditions are
6
 Q.J. Zheng et al. / Powder Technology xxx (2016) xxxxxx
Fig. 10. Effects of particle properties on the contact force distribution. The aspect ratio of the non-sphere is 1.25, as shown in the right inset gure. More details are described in text and
Table 2.
considered in the simulation and their inuences on the contact force
are discussed. The following conclusions can be drawn from this work:
1) The FEM/DEM model is an effective tool to analyse the mechanics of
pipe conveyor. It considers comprehensively the formation of pipe
from a at belt and the microscopic structure of discrete particles
under gravity, and thus can well describe the states of both conveyor
belt and conveyed bulk solids. As an application, it is employed in
this work to evaluate the contact forces between the belt and
supporting idlers. The obtained results agree qualitatively with the
previous experimental measurements in a variety of pipe diameters
and rotation angles.
2) The belt-idler contact force results from both the elastic recovery of
belt and the weight of granular materials. There is an only weak in-
terplay between the two factors since it is observed from Fig. 4 that
the belt stress does not change much with the volume lls of bulk
solids. Therefore, the principle of superposition applies approxi-
mately in this contact problem as demonstrated in Fig. 8.
3) The contact force is highly sensitive in cases of zero gravity and
empty pipe. Many subtle factors such as the friction of belt surface,
the softening of belt edge and the position of overlap can remarkably
affect its distribution. Such sensitivity, however, decreases with the
increasing content of bulk solids in the pipe. In heavily loaded pipe,
e.g. volume ll of 80%, the gravity of material becomes dominant
and the distribution ratios of contact force approach stable values.
The particle properties such as sliding friction and particle shape
may also inuence the force transmission within the particle bed
and change the contact force of bottom idler mildly.
The pipe considered in this work is shorter than the usual distance of
idler rings in order to compare the previous six-point bending experi-
ment. But since the solid mass and transverse stiffness of belt are both
proportional to pipe length, this short model may be roughly viewed
as a unit length of pipe conveyor and the obtained force ratio should
have some general applicability. For example, it can be used to estimate
the contact forces according to the true idler distance and densities of
material and belt.
Please cite this article as: Q.J. Zheng, et al., A coupled FEM/DEM model for pipe conveyor systems: Analysis of the contact forces on belt, Powder
Technol. (2016), http://dx.doi.org/10.1016/j.powtec.2016.09.070
9
Acknowledgement
The authors are grateful to the nancial supports of this work from
Australian Research Council and Longking Bulk Materials Science and
Engineering Co. (Project ID: LP130100365). This research was under-
taken with the assistance of resources from the National Computational
Infrastructure (NCI), which is supported by the Australian Government.
References
[1] A. Imai, Pipe conveyor development, benchmark and future trend, Bulk Solids
Handl. (2011) 19.
[2] G. Lodewijks, Research and development in closed belt conveyor systems, Bulk
Solids Handl. 20 (2000) 465470.
[3] S. Ramjee, P. Staples, Pipe conveyors for infrastructure projects: innovative solution
for conveyor systems, Bulk Solids Handl. 35 (2015) 2027.
[4] A.H. Ansari, Pipe conveyors advantages, Chem. Eng. World 40 (2005) 4749.
[5] F.S. Wang, Indentation rolling resistance of conveyor belts based on Maxwell model,
Adv. Mater. Res. 479 (2012) 15261529.
[6] T.J. Rudolphi, A.V. Reicks, The importance of non-linear strain consideration in belt
conveyor indentation loss, Bulk Solids Handl. 32 (2012) 5257.
[7] C. Wheeler, P.J. Munzenberger, Indentation Rolling Resistance Measurement,
Beltcon 16 Johannesburg, South Africa, 2008.
[8] T.J. Rudolphi, A.V. Reicks, Viscoelastic indentation and resistance to motion of con-
veyor belts using a generalized Maxwell model of the backing material, Rubber
Chem. Technol. 79 (2006) 307319.
[9] A.J.G. Nuttall, G. Lodewijks, A.J. Klein Breteler, Modelling rolling contact phenomena
in a pouch belt conveyor system, Wear 260 (2006) 10811089.
[10] W.D. Kim, W.S. Kim, C.S. Woo, H.J. Lee, Some considerations on mechanical testing
methods of rubbery materials using nonlinear nite element analysis, Polym. Int.
53 (2004) 850856.
[11] G. Lodewijks, Determination of rolling resistance of belt conveyors using rubber
data: fact or ction? Bulk Solids Handl. 23 (2003) 384391.
[12] D.L. Chertok, J.M. Golden, G.A.C. Graham, Hysteretic friction for the transient rolling
contact problem of linear viscoelasticity, J. Appl. Mech. 68 (2001) 589595.
[13] Q.J. Zheng, H.P. Zhu, A.B. Yu, Finite element analysis of the rolling friction of a viscous
particle on a rigid plane, Powder Technol. 207 (2011) 401406.
[14] C.A. Wheeler, A.W. Roberts, M.G. Jones, Calculating the exure resistance of bulk
solids transported on belt conveyors, Part. Part. Syst. Charact. 21 (2004) 340347.
[15] G. Lodewijks, Non-linear dynamics of belt conveyor systems, Bulk Solids Handl. 17
(1997) 5767.
[16] Y.F. Hou, Q.R. Meng, Dynamic characteristics of conveyor belts, J. China Univ. Min.
Technol. 18 (2008) 629633.
[17] A.J.G. Nuttall, G. Lodewijks, Dynamics of multiple drive belt conveyor systems, Part.
Part. Syst. Charact. 24 (2007) 365369.
10 Q.J. Zheng et al. / Powder Technology xxx (2016) xxxxxx
[18] G. Suweken, W.T. Van Horssen, On the transversal vibrations of a conveyor belt with
a low and time-varying velocity. Part I: the string-like case, J. Sound Vib. 264 (2003)
117133.
[19] M.E. Zamiralova, G. Lodewijks, Measurement of a pipe belt conveyor contact forces
and cross section deformation by means of the six-point pipe belt stiffness testing
device, Measurement 70 (2015) 232246.
[20] S. Hotte, L. Overmeyer, T. Wennekamp, Research on the form force behaviour of a
pipe conveyor in different curve radii, Bulk Solids Handl. (2011).
[21] P. Michalik, J. Zajac, Using of computer integrated system for static tests of pipe con-
veyer belts, 13th International Carpathian Control Conference (ICCC) 2012,
pp. 480485.
[22] V. Molnr, G. Fedorko, B. Stehlkov, P. Michalik, M. Weiszer, A regression model for
prediction of pipe conveyor belt contact forces on idler rolls, Measurement 46
(2013) 39103917.
[23] V. Molnr, G. Fedorko, B. Stehlkov, . Kudels, N. Huskov, Statistical approach for
evaluation of pipe conveyor's belt contact forces on guide idlers, Measurement 46
(2013) 31273135.
[24] V. Molnr, G. Fedorko, B. Stehlkov, M. Tomakov, Z. Hulnov, Analysis of asym-
metrical effect of tension forces in conveyor belt on the idler roll contact forces in
the idler housing, Measurement 52 (2014) 2232.
[25] M.E. Zamiralova, G. Lodewijks, Pipe conveyor test rigs: design, application and test
results part C, Bulk Solids Handl. 35 (2015) 4249.
[26] M.E. Zamiralova, G. Lodewijks, Pipe conveyor test rigs: design, application and test
results part A, Bulk Solids Handl. 34 (2014) 4045.
[27] M.E. Zamiralova, G. Lodewijks, Pipe conveyor test rigs: design, application and test
results part B, Bulk Solids Handl. 34 (2014) 3846.
[28] X. Liu, Y. Pang, G. Lodewijks, A stress discontinuity approach to model the stress pro-
le on a loaded conveyor belt, Powder Technol. 273 (2015) 102110.
[29] X. Liu, Y. Pang, G. Lodewijks, Theoretical and experimental determination of the
pressure distribution on a loaded conveyor belt, Measurement 77 (2016) 307316.
[30] J.J. del Coz Daz, P. Garcia Nieto, J. Vilan Vilan, A. Martn Rodrguez, J. Prado Tamargo,
A. Lozano Martinez-Luengas, Non-linear analysis and warping of tubular pipe con-
veyors by the nite element method, Math. Comput. Model. 46 (2007) 95108.
[31] G. Fedorko, V. Ivano, V. Molnr, N. Huskov, Simulation of interaction of a pipe
conveyor belt with moulding rolls, Procedia Eng. 48 (2012) 129134.
[32] Y. Pang, G. Lodewijks, Pipe belt conveyor staticscomparison of simulation results
and measurements, Bulk Solids Handl. 33 (2013) 5256.
[33] M. Dratt, A. Katterfeld, C.A. Wheeler, Prediction of conveyor belt deection by cou-
pling of FEM and DEM simulations, Bulk Solids Handl. 30 (2010) 380384.
Please cite this article as: Q.J. Zheng, et al., A coupled FEM/DEM model for pipe conveyor systems: Analysis of the contact forces on belt, Powder
Technol. (2016), http://dx.doi.org/10.1016/j.powtec.2016.09.070
[34] M. Dratt, A. Katterfeld, C. Wheeler, Determination of the bulk exure resistance via
coupled FEM-DEM simulation, 8th International Conference on Conveying and Han-
dling of Particulate Solids (CHOPS 2015, 2015.
[35] ABAQUS: User's Manual, Dassault Systmes Simulia Corp, Providence, RI, 2015.
[36] Q.J. Zheng, Z.Y. Zhou, A.B. Yu, Contact forces between viscoelastic ellipsoidal parti-
cles, Powder Technol. 248 (2013) 2533.
[37] P.A. Cundall, O.D.L. Strack, Discrete numerical-model for granular assemblies,
Geotechnique 29 (1979) 4765.
[38] H.P. Zhu, Z.Y. Zhou, R.Y. Yang, A.B. Yu, Discrete particle simulation of particulate sys-
tems: theoretical developments, Chem. Eng. Sci. 62 (2007) 33783396.
[39] H.P. Zhu, Z.Y. Zhou, R.Y. Yang, A.B. Yu, Discrete particle simulation of particulate sys-
tems: a review of major applications and ndings, Chem. Eng. Sci. 63 (2008)
57285770.
[40] H. Hertz, Ueber die Berhrung fester elastischer Krper, J. Reine Angew. Math. 92
(1882) 156171.
[41] C. Thornton, S.J. Cummins, P.W. Cleary, An investigation of the comparative behav-
iour of alternative contact force models during elastic collisions, Powder Technol.
210 (2011) 189197.
[42] Q.J. Zheng, H.P. Zhu, A.B. Yu, Finite element analysis of the contact forces between a
viscoelastic sphere and rigid plane, Powder Technol. 226 (2012) 130142.
[43] C. Cattaneo, Sul contatto di due corpi elastici: distribuzione locale degli sforzi, Rend.
Accad. Naz. Lincei 27 (1938) 342348.
[44] R.D. Mindlin, Compliance of elastic bodies in contact, J. Appl. Mech.-T ASME 16
(1949) 259268.
[45] Q.J. Zheng, A.B. Yu, Modelling the granular ow in a rotating drum by the Eulerian
nite element method, Powder Technol. 286 (2015) 361370.
[46] Q.J. Zheng, A.B. Yu, Finite element investigation of the ow and stress patterns in
conical hopper during discharge, Chem. Eng. Sci. 129 (2015) 4957.
[47] Q.J. Zheng, A.B. Yu, Why have continuum theories previously failed to describe sand-
pile formation? Phys. Rev. Lett. 113 (2014) 068001.
[48] M. Dratt, P. Schartner, A. Katterfeld, C. Wheeler, C. Wensrich, Coupled DEM and FEM
Simulations for the Analysis of Conveyor Belt Deection| NOVA, The University of
Newcastle's Digital Repository, 2012.
[49] K. Murugaratnam, S. Utili, N. Petrinic, A combined DEMFEM numerical method for
shot peening parameter optimisation, Adv. Eng. Softw. 79 (2015) 1326.
[50] Y.C. Chung, J.Y. Ooi, Linking of discrete element modelling with nite element anal-
ysis for analysing structures in contact with particulate solid, Powder Technol. 217
(2012) 107120.
[51] A. Nasto, A. Ajdari, A. Lazarus, A. Vaziri, P.M. Reis, Localization of deformation in thin
shells under indentation, Soft Matter 9 (2013) 67966803.