-1-

Published in Powder Handling & Processing, Vol. 13, No. 2, April/June 2001.
COMPUTER SIMULATION OF GRANULAR MATERIAL: VIBRATING
FEEDERS
Guillermo Rein Soto-Yarritu and ngel de Andrs Martnez
Instituto de Investigacin Tecnolgica, Universidad Pontificia Comillas
C/ Santa Cruz de Marcenado 26. 28015 Madrid. Spain
Phone: +34 91 542 2800, Fax.: +34 91 542 3176
email: [email protected]
Key Words: Granular Material, Discontinuous Deformation Analysis, Numerical Simulation,
Vibrating Feeder, Gravity Table Separator.
ABSTRACT
This paper presents a numerical model for simulation of granular material movement. The proposed computer model
is based on Discontinuous Deformation Analysis, a numerical method recently adapted for simulation of particles
motion. Gravity, impacts between particles, impacts with the boundaries, friction and cohesion are considered. The
motion of particles is induced by the vibration of the deck on which they rest and boundary conditions are set to
model a vibrating feeder for granular material transport. The model is compared to a simple analytical model used by
industries, showing that qualitative results agree. Numerical results were applied to optimize the design of a gravity
table separator machine for compost of which already a first prototype has been constructed.

1.- INTRODUCTION
Granular material can be defined as the material formed by multiple solid particles, whatever the size
of the particles. It represents the most common form of raw material in industry world-wide. In
weight, it is estimated that at least 75% of the raw material and 50% of the products in the chemical
industry (Nedderman 1992) are handled as granular material. Chemical, pharmaceutical, food,
mining, metallurgy, construction and many other industries use it continuously in its applications and
processes. Sugar, salt, flour, cement, sand and pills are worth examples of granular material.
Although its simple appearance, granular material presents a wide range of behaviors that are still far
from being full understood and explored. This form of material is considered by some experts as a
fourth state of matter; it can behave as a solid (it undergoes deformation under loads); as a liquid (it
flows and is poured); as a gas (it has compressibility and is formed by particles without bonds);or
more dramatically in its unique way. In spite of this fascinating properties, scientists and researchers
have paid little attention to it.
In opposition to fluid dynamics, granular dynamics was put partially aside until computer revolution
in the eighties allowed a more successful approach to the topic. Just one decade ago, industrial
structures related to granular mechanics as silos, stockpiles, feeders or conveyors were based on
simple and low performance experimental results with little theoretical basis. Nowadays, technology
and computer-based simulations permit a more efficient approach towards structure optimization
based on more precise and rigorous results.
Among machines for material handling (conveyors, tables, bins, elevators, etc), this paper focuses on
vibrating feeders. These transports move material inducing movement by a vibrating deck in contact
with the particles.
A gravity table is a machine for separation of granular material in dense and less dense (see Fig. 1)
by two physical actions: vibrating transport and fluidization. This paper is based on the research
conducted for optimization of a gravity table separator, focusing on the vibrating feeder which drives
the discharge of the dense phase.

-2-

Fig. 1. Schematic operating gravity table separator.

2.- DISCONTINUOUS DEFORMATION ANALYSIS
A new numerical method was presented in the mid eighties (Shi et al. 1984) called Discontinuous
Deformation Analysis (DDA). In some aspects, this method is similar to the finite element method
(FEM) and in other aspects to the discrete element method (DEM). It was first developed for
applications in geomechanics for the study of landslides and rocks motions (Fig. 2a and 2b) (Shi
1993). Today, it has been successfully applied to other engineering fields as for example, bulk
material discharge (Fig. 2c) (Fernndez 1997).

b)

a)

c)

Fig. 2. DDA examples: a) landslides; b) tunnel; c) bulk material discharge.

DDA is based on a system of discontinuous elements where displacements of each block are the
unknowns that are obtained solving the equilibrium equations. These equations are formulated
through minimizing the total potential energy of the system (as FEM). The formulation is
fundamentally discontinuous in the sense of the elements (as DEM). Advantages of this method are
(Thomas 1996): dynamic equilibrium is always reached; penetration of bodies and traction on
surfaces are not permitted at each time step; and solution stability is reached without introducing
artificial damping.
This method was further developed (Ke et al. 1995) to a formulation specially adapted to two
dimensional rigid disks and rigid moving boundaries. The model presented in this paper is based on
this last formulation of the DDA method.
Detailed formulation of the method can be found in (Ke et al. 1995).
In the reference coordinate system of a disk, u1 and u 2 are the displacements of point i, identified by
its coordinates ( x, y ) . Each disk has three degrees of freedom: two translations (U1 ,U 2 ) and one
rotation U 3 . Relation between point displacements and disk movement is given by the first order
approximation (1).
i

u1

u 2 ( x,y)

U1
1 0 ( y 0 y ) 1 0 ( y 0 y )
=
U 2 =
[Di ]
0 1 ( x x0 ) U 0 1 ( x x0 )
3

(1)

-3-

If there are n disks, a system of n equations is formed in terms of the

submatrices with the correspondent stiffness [K ij ] and forces [Fi ] submatrices:

[K11 ]

[K 21 ]
[K ]
31
M

[K n1 ]

[K12 ] [K13 ]
[K 22 ] [K 23 ]
[K 32 ] [K 33 ]
M

[K n2 ] [K n3 ]

K
K
K
O
K

[K1n ] [D1 ] [F1 ]

[K 2 n ] [D2 ] [F2 ]
[K 3n ] [D3 ] = [F3 ]
M M

[K nn ] [Dn ]

displacements

(2)

[Fn ]

[K ] = [K ]
ij

[Di ]

(3)

ij

l =1

[Fi ] = [Fi ]l ; with l enegy sources

(4)

l =1

Each row of equation (2) is obtained minimizing the potential energy E (5) produced by all the
forces acting (eq. (3) and (4)). Submatrices [Kii ] depends on the material properties of disk i and

[K ], where (i j ) , is defined by contact between disks i and j.

ij

E
= 0, j = 1,2,3 for disk i
U ij

(5)

Surface contact is modeled by means of two visco-elastic elements (Fig.3) which are added to the
global system matrix (2) in the time step when contact is detected (penalty method). This
methodology fulfills constraints of no tension between blocks and no penetration of one block into
another.

Fig 3. Contact elements: a) normal and b) tangential.

Contact and interaction between multiple bodies introduces high non linearity into the problem. To
control this non linearity, at each time step the algorithm checks contact states and constrains. In
case constrains are not completely fulfilled, calculation process is repeated changing the state of
contacts from open to close or vice versa and reducing the time increment until convergence is
reached.
Disadvantages of DDA method are time and memory requirements. Comparing extreme cases to
FEM, DDA is five times slower and memory needed is four times larger.
3.- APPLICATION TO GRANULAR MATERIAL TRANSPORT IN VIBRATING FEEDERS
The model was applied to the design of a gravity table separator using the simulation results. Some
machine parameters were externally imposed by mechanical design criteria and values were directly
introduced into the model. Material properties were determined by consulting available literature
and studies of the interesting mechanical design parameters were conducted. In order to better
improve the material model, particles have been arranged forming different clusters (Fig. 4a).
The model simulates a 2D rectangular compartment that contains a plane deck inclined an angle
with the horizontal direction (Fig. 4b). The deck is the feeder which is excited with a sinusoidal
vibration in the angle direction whereas a sufficient quantity of particles with similar properties
to the real material is placed on the deck. Deck starts to move and particles react. The objective is to

-4maximize the horizontal discharge velocity. Angle of the feeder must be small but not zero,
because the gravity separation phenomenon needs a slight inclination in order to be efficient and
move forward only the dense phase of the material.

a)

b)
Fig. 4. a); Particle clusters; b) Geometrical parameters of the model.

Initial conditions were set to zero velocity for both the disks and the boundaries.
In the simulations (Fig. 5), discharge velocity is computed deriving numerically the horizontal
position of the center of gravity respect to time. The graph of the velocity versus time (Fig. 6) shows
a short unsteady period followed by a longer quasi-steady period, consequently, simulations were
carried out until useful information about the steady regime could be drawn.

Fig. 5. Sequence of instances for =10, =22 and =6.18 (eq. (6)).

Fig. 6. Horizontal velocity of the center of gravity for =10, =22 and =6.18 (eq. (6)).

-5-

The model permits to analyze machine parameters and material properties effects through
parametrical and sensitivity studies. Here is a brief presentation of the different parameters that the
model requires and therefore they can be studied. The material used in the simulations represents the
dense phase of compost. Compost is the fermented organic fraction of solid urban waste.
See Table 1 for material properties. Material is formed of 175 disks forming clusters. The density
represents the mean density of glass, plastic and light metals. Values for friction and cohesion of
particles were found in literature. The damping of the system results in an overall restitution
coefficient near 0.8.
Table 1. Material properties
Shape and size

Diameter between 5 and 10 mm

Density

2500 kg/m3

Particle friction coefficient

0.84 ()

Particle cohesion

700 N/m2 ()

Restitution coefficient

near 0.8

See Table 2 for machine operating parameters. Angle for deck inclination respect horizontal,
vibration amplitude and vibration frequency were imposed by mechanical design criterion. The
resulting non-dimensional acceleration number (6) is the ratio of vibrating acceleration to gravity
acceleration g . Angle of vibration respect horizontal is a variable parameter during operation.
The optimum value for is discussed in the next section. The friction of the deck represents that of
a metal but increased due to the holes of the screen on the deck surface. Adhesion of the deck were
set to a low value.
Table 2. Machine Operating Parameters
Angle

Vibration amplitude A

6 mm

Vibration frequency f

16 Hz

6.18

Optimum angle

30

Deck friction coefficient

0.65

Deck adhesion

60 N/m2

4 A(f )2
=
g

(6)

4.- COMPUTER SIMULATION RESULTS

4.1 Validation
In order to validate the model and its results, a simple analytical model (Rachner 1964) has been
used which computes the trajectory of one frictionless particle excited by a vibrating deck under
gravity effect. This analytical model is conservative and non linear, although it only uses the
parameters amplitude, frequency and angle. Therefore it does not take into account critical
()

Nedderman 1992

-6-

granular phenomena like multiple particles or friction. Due to these characteristics, the analytical
model is not acceptable for quantitative validation. In spite of this, the analytical model is applied for
engineering applications by many industries, and thus it can be used for qualitative validation of the
proposed model.

Fig 7. Results comparison as a function of angle for =0, A =6 mm and f =16 Hz.
It has been compared the horizontal velocity for a similar configuration ( =0) for both models.
Results show (Fig. 7) that both models behave in a similar way, being the maximum discrepancy 38%
and the mean discrepancy 16%. The optimum angle for the proposed model is 23.75 and for the
analytical model is 24, with a slight difference of 1%. It is necessary to point out that each
engineering department modifies output velocities of the analytical model multiplying by near-one
coefficients related to their experience-based knowledge. These coefficients were not available for
this work.
4.2 Applied Results
Simulating the dense phase of compost and introducing the machine parameters imposed by
mechanical design criteria ( =8), the model has permitted to calculate the optimum angle of
vibration. The optimum angle, which gives the maximum output velocity, is located near 30 (Fig.
8). The increase in 8 of inclination by angle has decreased the discharge velocity in 33%
comparing to Fig. 7.

-7Fig 8. Discharge velocity results as a function of the angle of vibration for =8 and =6.18.
5.- CONCLUSIONS
DDA method has been successfully applied to granular material transport simulations. Results of the
model compare qualitatively well to a simple analytical model. Applied to a gravity table separator
for compost, the model has allowed to determine the optimum angle of vibration. Using the model,
mechanical specifications of the machine has been proved to be correct and forecasted discharge
velocities for dense phase are better than those of machines currently in use.
This research has permitted the construction of a first prototype (Fig. 9) that will be followed by a
commercial machine.

Fig. 9. Manufactured prototype for testing

ACKNOWLEDGEMENTS
This work was carried out in collaboration with the Spanish engineering firm Urbar Ingenieros S.A.
The authors thank the financial support of the Spanish Ministerio de Industria y Energa and
Comunidad de Madrid.
6.- REFERENCES
1. Fernndez, Rosa M. (1997). Simulacin analtica y numrica de la descarga de ridos.
Bachelors Thesis , ETSII Universidad Pontificia Comillas.
2. Gerstel, A. and Scheublin, J. R. (1994). Computer simulation program for vibrating equipment,
Bulk Solids Handling, Vol. 14, No. 3, pp. 573-577.
3. Ke, T. C. and Bray, J. D. (1995). Modeling of particulate media using discontinuous
deformation analysis, Journal of Engineering Mechanics, Vol. 121, No. 11, pp. 1234-1243.
4. Nedderman, R. M. (1992). Statics and Kinematics of Granular Materials, Cambridge,
Cambridge University Press.
5. Shi, G. H. and Goodman, R. E. (1984). Discontinuous Deformation Analysis, Proceedings of
25th U.S. Symposium on rock Mechanics, pp. 269-277.
6. Shi, G. H. (1993). Block System Modeling by Discontinuous Deformation Analysis,
Southampton, Topics in Engineering, Computational Mechanics Publications, Vol. 11.
7. Thomas, Patricia A., Bray, Jonathan D. and Te-Chih, Ke (1996). Discontinuous Deformation
Analysis for soil mechanics. Discontinuous Deformation Analysis (DDA) and simulations of
discontinuous media. M. Reza Salami & Don Banks. TSI Press. pp. 454-461.

-8-

AUTHORS SHORT CV

Guillermo Rein Soto-Yarritu

He was born in Madrid, Spain, in 1975. He received the degree of
Mechanical Engineer from Universidad Pontificia Comillas,
Madrid, in 1999. He studied his sixth and last academic year at the
Mechanical Department of University of Texas at Austin, as part of
the GE4 exchange program. From October 1999 up to now he is a
research assistant and a graduate student at the Instituto de
Investigacin Tecnolgica of Universidad Pontificia Comillas. He
is working in numerical methods applied to computer simulations
of granular material at the Industrial Systems research group. He is
interested in numerical methods, thermosciences and mechanical
design.

Prof. Dr. Angel de Andrs Martnez

He was born in Madrid, Spain, in 1964. He studied Engineering
in the Universidad Pontificia Comillas, Madrid, where he
received an honorific mention at the end of his studies. Received
his Ph. D. degree in the UPCO in April 1997. Presently he is
working as a Professor at the Mechanical Engineer Department of
the UPCO and partially as a researcher at Instituto de
Investigacin Tecnolgica in the Industrial Systems research
group, where he is studying numerical methods applied to
mechanical engineering problems in general and to fracture
mechanics in particular. He also worked for the private industry
during several years before joining the University as a full time
teacher and researcher.