Calibration of DEM Parameters for Polypropylene Powder Using Schulze Ring Shear Tester: An Efficient Upscaled Computational Approach

Characterization is essential to understand complex powder behavior. Among different tools, the Schulze Ring Shear Tester (RST) measures powder bulk properties under quasi-static conditions. It consists of an annular trough hosting the powder sample and an annular lid covering it. The normal force applied by the lid compresses the powder, and shear stress is the particle system's response to the applied trough rotation [1]. Besides characterization, shear test data can be used for calibrating particle simulation methods to replicate and predict experiments for the same material.<br/>The Discrete Element Method (DEM) is used to simulate various granular materials, however, simulating the shear cell filled with fine polypropylene (PP) powder is computationally infeasible due to the large number of particles. Therefore, we employ two techniques to mitigate computational limitations in DEM: partial ring periodic boundary conditions and coarse-grained particles [2, 3].<br/>The RST is used to gather density and shear stress data of PP and is the focus of our numerical setup. We approach improved DEM, beginning with coarser particles in a small angular segment of the shear cell. Then, the resolution is progressively increased to model more realistic, finer particle sizes while also incorporating larger cell segments. Comprehensive data from both experiment and simulation are required to ensure accurate calibration. These include the yield locus, the time (strain) evolution of shear stress, and the bulk density for different confining stresses. The challenge is to find a valid set of DEM interaction parameters to match experimental data. To address this, we use Grain-Learning, an iterative Bayesian optimization framework, to calibrate the DEM parameters [4]. The calibrated coarse-grained DEM parameters can reproduce experimental data for this material (PP) under various consolidation normal stresses. Other materials will be considered in ongoing and future work.<br/><br/>REFERENCES<br/>[1] D. Barletta, M. Poletto, and A. C. Santomaso, “Chapter 4. Bulk Powder Flow Characterisation Techniques” in Powder Flow, A. Hassanpour, C. Hare, and M. Pasha, Eds., Cambridge: Royal Society of Chemistry, 2019, pp. 64–146. doi: 10.1039/9781788016100-00064.<br/>[2] T. Weinhart et al., “Fast, flexible particle simulations—an introduction to MercuryDPM,” Comput. Phys. Commun. 249, 107129, 2020.<br/>[3] M. Sakai and S. Koshizuka, “Large-scale discrete element modeling in pneumatic conveying,” Chem. Eng. Sci. 64(3), 533–539, 2009. doi: 10.1016/j.ces.2008.10.003.<br/>[4] H. Cheng, T. Shuku, K. Thoeni, P. Tempone, S. Luding, and V. Magnanimo, “An iterative Bayesian filtering framework for fast and automated calibration of DEM models”, Comput. Methods Appl. Mech. Eng. 350, 268–294, 2019. doi: 10.1016/j.cma.2019.01.027.<br/><br/>