Every numerical method relies on the accurate choice of models, solver settings, and material parameters in order to be able to mimic real-world behavior. This also applies to Discrete Elements Method (DEM) simulations. You could use standard material properties, but adjusting those material interaction parameters using automated calibration methods is a key step for accurate simulations.
You could use standard material properties, but if you want to simulate reality, it is important to understand that the materials actually vary from site to site. Adjusting those material interaction parameters using automated calibration methods is a key step for accurate simulations. Even with basic materials, friction and restitution coefficients between particles and particles and boundaries have to be adjusted in order to accurately predict the bulk flow behavior. When extra forces come into play, such as adhesion forces, those additional parameters also need to be selected and properly specified.
In general, the calibration process consists of reproducing simple experimental tests and verifying the matching between numerical results and experimental data. Multiple simulations are run on the simple cases in order to find the set of parameters that best fit the experimental values. Then those same parameters can be used in the more complex simulation with confidence. With enough time and patience and an experienced engineer, this can be done with manual iterations. But, by far, the best way is with automated solutions that can select and vary the input parameters automatically based on sensitivity analysis and meta modeling techniques. These algorithms have built in intelligence to automatically reduce the number of cases to be run and they can be trusted to seek out those critical coefficients as efficiently as possible.
Snapshot of the performed angle of repose simulation
As you know, ANSYS wants your tools to work together to help you design the best products possible. We love it when our partners work together in the Workbench environment to bring you solutions that do this. For instance, the integration between ANSYS Optislang and Rocky DEM within ANSYS Workbench makes calibration of a DEM problem a straight forward and efficient process.
Metamodel of Optimal Prognosis (MOP) for the simulated angle of repose, which is an average over the pile and depends on the dynamic and the rolling friction.
Want to know more about Rocky DEM – ANSYS optiSLang?
Join the ANSYS in Action webinar on August 17th to learn how the automated calibration of a numerical simulation with experimental results can help to improve the credibility of the simulation results with all stakeholders. In this webinar, we will demonstrate the setup of an automated analysis for angle-of-repose and drawn-down-angle simulations performed using Rocky DEM. Come and see how easy it is to use sensitivity analysis and meta-modeling techniques to find the best fit of particle-specific data with the experimental data.
While we will demonstrate how to apply Rocky DEM inside ANSYS optiSLang, the presented workflow can be applied to any FEM, CFD or electromagnetic simulation model.