Framework for DEM Model Calibration and Validation

Övrigt konferensbidrag, 2015

The discrete element method (DEM) is used for modelling and simulating particle flow behaviour in research and industrial applications. If the results from such a simulation are to be trusted in a research project or be used as knowledge basis in engineering decision making, the modeller will need to apply some kind of calibration and validation approach. Even though most researchers and engineers apply some kind of method for calibration of the model and validation of the results, no general consensus or calibration methodology framework has been established that governs the quality or accuracy. DEM simulations are now used for modelling a vast number of machines and processes in minerals processing. It is hence of essence that the academia and industry continues the process of discussing a methodology for calibrating rock and ore materials. This paper aims at joining this discussion by presenting a general framework. The approach is based on the calibration framework proposed by Hofmann (2005) and adopting it to calibration of DEM models. The framework is based on calibration and validation on three different levels ranging from the basic single contact model parameter calibration; to a mid-level flow property test; to a full scale experiment. A calibration device has been developed in order to facilitate multiple flow regime experiments with directly observable particle flow paths. In order to efficiently perform both calibration experiments and simulations a design of experiments approach is applied. The actual calibration is conducted by minimizing the error between the experimental domain and the simulation domain by applying multi-objective optimization methods.