Penetration in coarse granular packing - Benchmark

Benchmark Lead(s)

Hao Shi @Hao Shi

Hrachya Kocharyan @Hrachya Kocharyan

Confirmed collaborator(s)

Please feel free to join and add your code name to the table below. This benchmark case is very basic so it can be easily performed in any code.

DEM Code name

Responsible person:

EDEM

@Hao Shi

Mercury

@Thomas Weinhart

YADE

@Danny van der Haven @Karol Brzeziński

BlazeDEM

@Nicolin Govender

Ansys RockyDEM & LIGGGHTS

@Manuel Moncada

Benchmark Type

Target Result

(e.g. How close to an analytical solution?)

Compare the penetration behaviour in a predefined coarse granular packing.

Expected outputs/quantities:

(a) Macroscopic scale (time evolution, data saved for every 0.02s):

• Total penetration resistance - depth,

• Tip resistance - depth

• Sleeve/shaft resistance - depth

(b) Microscopic scale (end of penetration, t=2s):

• Averaged coordination number with standard deviation (example: CN=5.678, STD=1.234)

• Contact force chain close to the bottom of the plate (visualisation image only, see examples below, normal force magnitude with colourbar, range 0.2-1N for no-rolling and 1-7N for TypeC rolling model, view from +X axis)

(c) Simulation hardware specs and runtime

• CPU/GPU specifications that were used to perform the case (example: Intel Xeon w5-3423, 2.26GHz, 12 cores; NVIDIA RTX A6000 48G)

• Total Runtime for the 2 [s] simulation (example: 10.2 hours)

Benchmark Description

Granular coarse materials such as gravel, grains and minerals are common in geotechnical, mining and bulk handling applications. A common method to investigate the penetration resistance in soils/sand-bed is the Cone Penetration Test (CPT). For coarse granular materials, plate penetration is more appropriate to represent the penetrations happening in the real bulk material handling processes, e.g., excavation or grabbing.

This benchmark case represents the rigid plate penetrating through granular media. The system contains two sets of geometries: plate and container, as shown in the figure below. The container has a mesh composed of 10 triangular elements, while the plate is separated into two parts: side walls (48 elements) and top-bottom walls (8 elements). In this way, the forces acting on different parts of the plate can be recorded easily. All the geometries are provided in three separate stereolithography (.stl) files

Open

At the start of the simulation, the container was filled with 137,425 particles of type M1 (detailed properties given at the bottom of the page) with a median diameter of 4 mm. The particles were randomly generated with a normal particle size distribution (PSD) using a mean value of 1 and a standard deviation of 0.05. The packing is then relaxed under gravity to reach the resulting fully settled state. Two benchmark runs are performed: the first one using solely the Hertz-Mindlin no-slip model and the second one using the Hertz-Mindlin no-slip model combined with Type C rolling friction model.
The plate penetrates vertically downwards in the negative z-direction with a constant velocity of 50 mm/s. The total penetration time is 2 seconds, resulting in a final penetration depth of 100mm, which is around half of the height of the packing. The timestep is fixed at 1 microsecond, which is about 10% of Rayleigh's timestep. The forces acting on the plate geometries during the whole penetration process are recorded with a sampling interval (saving count) every 0.02 seconds.

Benchmark Geometry and Boundary Conditions

A cubic box with open top is used to hold the granular packing, this open box is provided with stl file format.

A steel plate consists of top/bottom walls and side walls is provided in stl files.

The initial granular packing is provided in data format and is explained in more detail in the initial configuration section.

File name

Description

Location/link

domain_box.stl

Cubic domain box to hold all the particles.

plate_top_bottom_walls.stl

Top and bottom walls for the plate geometry

plate_side_walls.stl

Side walls for the plate geometry

ini_Packing_Penetration.data

Initial granular packing

Presentation (20260203 WG4 meeting at Kayseri):

https://doi.org/10.5281/zenodo.18609843

Additional notes:

Particle Type

Select all applicable

• Spheres [x]

• Sphere clumps

• Ellipsoidal

• Super-quadrics

• Level-set

• Potential surfaces

• Polygonal

• Polyhedra

• Sphero-polyhedra

• Other (specify):

Mixture Type

Select all applicable

• Monodisperse

• Bidisperse

• Polydisperse [x]

• Uniformly graded

• Other (specify):

Material properties - Artifical granular material M1 and steel

Provide particle density/densities, Young’s modulus and Poisson’s ratio (where appropriate), as well as properties that are required to define particle shape (i.e. radii, longest/shortest axis, sphere separation distance for clumps, etc.). Add/remove rows if required.

Material description

M-sphere

Steel

Median particle diameter d_50

4 [mm]

-

PSD (normal distribution) mean

1

-

PSD (normal distribution) STD

0.05

-

Density

2000 [kg/m3]

7500 [kg/m3]

Young’s Modulus

0.5 [GPa]

210 [GPa]

Poisson’s ratio

0.2

0.2

Coefficient of restitution

0.5 (M1 - M1)

0.4 (M1 - Steel)

Coefficient of sliding friction

0.3 (M1 - M1)

0.2 (M1 - Steel)

Coefficient of rolling friction

0.2 (M1 - M1)

0.2 (M1 - Steel)

Contact model(s) and properties

Normal direction: Hertzian model

Tangential direction: Mindlin model with/without Type C rolling friction (optional if your code doesn’t have Type C rolling model, which contains a mechanical spring torque and a viscous damping torque).

Open

Ref: http://dx.doi.org/10.1016/j.powtec.2010.09.030; http://dx.doi.org/10.1016/j.powtec.2011.10.057

ON-DEM Contact Model Database Reference:

Simulation setup

Parameters for simulation and analysis. Add rows if required (e.g. number of cycles, unbalanced force ratio, etc.)

Time-step

1e-6 [s]

Simulation time

2 [s]

Data output interval

0.02 [s]

Number of particles

137,425

Plate penetration velocity

0.05 [m/s]

Initial configuration

To minimise the error caused during the particle packing generation process, the initially settled particle packing is provided with a single data file: “ini\_Packing\_Penetration.data”:

Open

which includes full information regarding the positions and velocities of all particles at timestep zero. The data file format used in MercuryDPM is adopted here for the sake of convenience. The first line is structured as below:

Open

Open

Output data format

Two benchmark simulation runs need to be performed: one without the rolling friction model (set rolling friction to zero) and one with TypeC rolling friction model (set rolling friction to 0.5). For both simulations, the final snapshot (t=2s) containing particle locations and velocities is saved using the data format specified in the Initial Configuration section.

At each sampling interval (saving count) of 0.02 seconds, the total forces acting on the two plate geometries (side and bottom walls) are also recorded in a separate output data file in the following order:

Open

Note we use SI base units for all the quantities: [m], [s], [kg]

Output files (see details in the Target results section above):

• force_plate.txt [data for expected output (a)]

• extra_data.txt [coordination number as in output (b) and specs as in output (c)]

• force_chain_no_rolling.png [output (b), one image for each simulation at t=2s, max size 2 MB per image]

Please create a .zip archive of your output files and send it to: h.shi-1@utwente.nl & hkocharyan@aua.am

Other considerations and notes

We intentionally omit the dynamic quantities, such as granular temperature, because the process here is very slow/quasi-static. The main aim is to investigate whether a similar bulk material behaviour (especially resistance forces) can be reproduced similarly across different DEM codes.

Tentative schedule

Expression of interest by:

Feb 28, 2026

Start date (data received from):

Mar 1, 2026

Submission deadline for data:

May 31, 2026

Data analysis to be completed by:

Jul 31, 2026

Tentative publication date for draft submission

Aug 31, 2026