5 Development of a Dynamic-Physical Process Model for Sieving
151
is the parameter determination for small and arbitrarily shaped non-spherical particles [59, 60]. Even if the values are measured with high accuracy, the results in
bulk experiments and simulations may differ, due to computational limitations with
which the particle size and shape is resolved in the simulations [61–63]. Another
possibility is to combine both methods by first determining the parameters with the
second method, followed by an parameter adaptation with small scale experiments
according to the first method as proposed by Elskamp et al. in [30].
2.5.1 Determination of DEM Parameters at the Single Particle Scale
Two of the most important parameters for DEM simulations are the particle size and
shape, which must be determined in advance to other parameters. While the particle
size usually can be measured easily, the representation of particle shape is more challenging although it can be relied on the approaches as outlined in Sect. 2.1. In the past
some researchers represented simple model shapes by manually assembling spheres
when relying on the multi-sphere method. Among others, Markauskas et al. [64]
used 3 to 50 spheres of different sizes to represent ellipsoid particles. Particles with
a slightly more complex shape were reported by Pasha et al. [65], which represented
maize grains with a clump of several spherical particles generated by an automated
optimization process based on 3D X-ray tomography data. In contrast, Williams et al.
[66] evaluated descriptions of the irregular particle shapes of iron ore using a digital image segmentation technique and generated corresponding non-spherical DEM
particles. Mollon and Zhao [67] presented a method for producing 3D non-spherical
particles based on three 2D contours of the cross sections of random realistic grains
of sand. A similar method was derived in [30] using an automated shape adaption
algorithm.
Other important parameters for DEM simulations, which can be determined by
direct measurement at the individual particle level, are the particle density, stiffness,
Young’s and shear modulus, sliding and rolling friction, damping and the coefficient
of restitution.
If, as here, a linear contact model is used, particle stiffness is specified based on an
appropriate calibration with various small scale experiments (e.g. shear tests) or by
sensitivity analyzes. In contrast, when using Hertz-Mindlin models it is required to
calculate contact stiffness based on Young’s modulus, shear modulus, and Poisson’s
ratio. The modulus of elasticity is obtained by uniaxial compression tests in which
individual particles are compressed [68–70].
Several researchers have reported the particle-wall sliding friction obtained with
different approaches. One method for determining slip friction between two particles
was described by Senetakis et al. [71], who designed a device to perform shear tests for
small displacements, loads and non-spherical particles. Barrios et al. [63] determined
the contact coefficient of sliding friction with a rotating pin-on-disk tribometer. In
addition, the coefficient of sliding friction can be obtained using a direct shear box
(e.g. Jenike shear cell) in which a wall is replaced by the desired wall material and
the particles are sheared over it (see e.g. [72–75]).
151
is the parameter determination for small and arbitrarily shaped non-spherical particles [59, 60]. Even if the values are measured with high accuracy, the results in
bulk experiments and simulations may differ, due to computational limitations with
which the particle size and shape is resolved in the simulations [61–63]. Another
possibility is to combine both methods by first determining the parameters with the
second method, followed by an parameter adaptation with small scale experiments
according to the first method as proposed by Elskamp et al. in [30].
2.5.1 Determination of DEM Parameters at the Single Particle Scale
Two of the most important parameters for DEM simulations are the particle size and
shape, which must be determined in advance to other parameters. While the particle
size usually can be measured easily, the representation of particle shape is more challenging although it can be relied on the approaches as outlined in Sect. 2.1. In the past
some researchers represented simple model shapes by manually assembling spheres
when relying on the multi-sphere method. Among others, Markauskas et al. [64]
used 3 to 50 spheres of different sizes to represent ellipsoid particles. Particles with
a slightly more complex shape were reported by Pasha et al. [65], which represented
maize grains with a clump of several spherical particles generated by an automated
optimization process based on 3D X-ray tomography data. In contrast, Williams et al.
[66] evaluated descriptions of the irregular particle shapes of iron ore using a digital image segmentation technique and generated corresponding non-spherical DEM
particles. Mollon and Zhao [67] presented a method for producing 3D non-spherical
particles based on three 2D contours of the cross sections of random realistic grains
of sand. A similar method was derived in [30] using an automated shape adaption
algorithm.
Other important parameters for DEM simulations, which can be determined by
direct measurement at the individual particle level, are the particle density, stiffness,
Young’s and shear modulus, sliding and rolling friction, damping and the coefficient
of restitution.
If, as here, a linear contact model is used, particle stiffness is specified based on an
appropriate calibration with various small scale experiments (e.g. shear tests) or by
sensitivity analyzes. In contrast, when using Hertz-Mindlin models it is required to
calculate contact stiffness based on Young’s modulus, shear modulus, and Poisson’s
ratio. The modulus of elasticity is obtained by uniaxial compression tests in which
individual particles are compressed [68–70].
Several researchers have reported the particle-wall sliding friction obtained with
different approaches. One method for determining slip friction between two particles
was described by Senetakis et al. [71], who designed a device to perform shear tests for
small displacements, loads and non-spherical particles. Barrios et al. [63] determined
the contact coefficient of sliding friction with a rotating pin-on-disk tribometer. In
addition, the coefficient of sliding friction can be obtained using a direct shear box
(e.g. Jenike shear cell) in which a wall is replaced by the desired wall material and
the particles are sheared over it (see e.g. [72–75]).