2024 AIChE Annual Meeting
(342f) Decoupling Calibration and Coarse-Graining for DEM Simulation of Fine Particles
Authors
This methodology involves employing a surrogate system for the calibration of parameters. The surrogate system and the system of interest should have identical parameter sensitivity, which can be determined by running a set of quick simulations. The surrogate system should be such that the multitude of simulations needed for calibration may be conducted readily (e.g., using periodic BC).
Once the parameters for the resolved system (i.e. without coarse-graining) are identified using the surrogate system, the effect of coarse-graining on the parameters is studied using the same/another suitable surrogate system. Thus, the parameters corresponding to a coarse-grained system for a given coarse-grain ratio are obtained, and this final parameter is used for the system of interest to generate results that reproduce experimental results quantitatively. This methodology is shown in Fig. 1.
We demonstrate this method using Lactose Fast Flo 316 powder and a rotary tablet press feeder as the system of interest. A wedge-shaped hopper with a collection pan at the bottom was considered for the surrogate system to calibrate the properties of the resolved system, where the discharge rate and angle of repose were the observables. An Artificial Neural Network (ANN) based calibration methodology was used [1] to obtain the calibrated parameter space. The effect of coarse-graining on the parameters was studied using a simple shoe-die pair. Finally, the calibrated parameters corresponding to a given CG ratio are used for the DEM simulation of the tablet press feeder.
References:
[1] Forgber, T., Khinast, J. G., & Fink, E. (2022). A hybrid workflow for investigating wide DEM parameter spaces. Powder Technology, 404, 117440.