EDEM Hysteretic Spring model

Contents
HYSTERETIC SPRING MODEL IMPLEMENTATION DETAILS ........................................................................ 2
EXAMPLE SIMULATION: COMPRESSION TESTS ......................................................................................... 4
REFERENCES ............................................................................................................................................. 6

Hysteretic Spring model implementation

details
The Hysteretic Spring contact model allows plastic deformation behaviors to be included
in the contact mechanics equations, resulting in particles behaving in an elastic manner
up to a predefined stress. Once this stress is exceeded, the particles behave as though
undergoing plastic deformation. The result is that large overlaps are achievable without
excessive forces acting upon them, thus representing a compressible material.
Hysteretic Spring normal force calculation is based on the Walton-Braun theory
described in the following references, (Walton and Braun 1986), (Walton and Braun
1986).

Figure 1 Force-displacement curve for the Hysteretic Spring model

Figure 1 shows a schematic of the Hysteretic Spring contact model force displacement
relationship. Note that the unloading force goes to zero before the displacement
recovers to the initial contacting point. The quantity δ0 represents the residual ‘overlap’
that remains because of the plastic deformation (i.e., flattening) in the contact region.
The exact location of each old contact spot is not ‘remembered,’ so that
particles become like new undeformed spheres once they separate. Subsequent
contacts will load along a new loading path with slope, K1. Any reloading before
separation follows the slope K2 until the original loading curve K1 is reached, whereupon
the lower slope is followed until unloading occurs.

In particular, the normal force is calculated as follows

K1 n for loading ( K1 n  K 2 ( n   0 ))


Fn  K 2 ( n   0 ) for unloading/ reloading ( n   0 )
0 ( n   0 )
 for unloading

where K1 and K2 are the loading and unloading stiffness respectively, δn, is normal
overlap and δ0 is the residual overlap. Loading stiffness, K1, is related to the yield
strength of each material participating in contact, Y1 and Y2, as follows (Walton O. ,
2006):

K1  5 R* min(Y1 , Y2 )

Here R* is the equivalent radius of the two particles participating in the contact.
The following expression for coefficient of restitution

K1
e
K2

allows for the determination of K2 from K1 (Walton O. , 2006). Residual overlap is

updated every timestep according to the following rule:

 K1
 n (1  K ) for loading ( K1 n  K 2 ( n   0 ))
 2
 0   0 for unloading/ reloading ( n   0 )
 for unloading ( n   0 )
 n


The damping of normal force is controlled by normal damping factor, bn, as follows:

4m* k
F  bn
n
d
vnrel
  
2

1  
 ln e 
Here k is either K1 or K2. The rest of the variables and factors in the above expression
are explained in the Linear Spring contact model section.

The tangential forces will be implemented as per the Linear Spring contact model but
with modified tangential stiffness. In particular, tangential force depends on the stiffness
factor, γt,, as follows:

Ft   min( t K1 t  Ft d , Fn )

Unlike for the normal component, tangential damping force is not scaled by damping
factor:

4m* t k
Ft d   vtrel
  
2

1  
 ln e 

A default value for the Yield Strength is estimated from Young’s modulus E and Radius
of the particle R according to the algorithm suggested in (Walton O. , 2006). First, the
possible default value is calculated from:

4 E 1
Y
15 R 70

If this value is less than 100Pa it is replaced with a value of 0.003 * E.

Example Simulation: Compression tests

To test the model the following scenario was simulated. Spherical particles with diameter
of 12 mm and density 1000 kg/m3 were settled in cylindrical container with base
diameter 150 mm and height 350 mm. The top lid is initially positioned just above the fill
level. Top lid initially moves down with velocity = 0.1 m/s for 0.4 s and then moves up
with the same velocity. Several compression test are simulated, first using Hertz-Mindlin
(no slip) contact model with particle Shear Modulus 1e6 Pa, then Hysteretic Spring
contact model with Yield Strength 1e5 Pa and 1e4 Pa. Coefficient of restitution is set to
0.1 in all cases.
Figure 2 Left: Initial state of granular material before compression. Middle: final state of
material after compression using Hertz-Mindlin (no slip) contact model. Right: final state of
material after compression using Hysteretic Spring contact model.

The results of the simulations illustrate the behavioral difference between these two
contact methods is the application of a loading and unloading axial force to a constrained
cylinder of particles. During unloading the elastic particles in simulation with Hertz-
Mindlin contact model will return to approximately the starting height prior to
compression. Using the hysteretic elastic-plastic model, the switch to plastic failure
means that a greater level of compression can be achieved between the particles. Upon
unloading, there is little if no readjustment of the particles as the compressive force is
removed. This means that the final fill level in the cylinder is significantly lower than both
that of the starting condition and that of the final condition in simulations with Hertz-
Mindlin contact model, as is shown in Figure 2. Figure 3 shows the results for the force
on the lid as a function of the lid position calculated in these compression test
simulations. The results show significant drop of force in the unload phase due to
plasticity for the simulation with Hysteretic Spring model.
Figure 3 Simulation results for the force on the lid as a function of lid position. Hysteretic
Spring model shows immediate drop of the force to zero at the beginning of
decompression due to plastic deformation.

References
Walton, O. (2006). (Linearized) Elastic-Plastic contact model. DEM Solutions.

Walton, O. R., & Braun, R. L. (1986). Stress calculations for Assemblies of Inelastic
Spheres in Uniform Shear. Acta Mechanica , 63, 73-86.

Walton, O. R., & Braun, R. L. (1986). Viscosity and Temperature Calculations for
Assemblies of Inelastic Frictional Disks. J. Rheology , 30 (5), 949-980.