DEM Calibration: A Complex Optimization Problem

By using discrete element method we can explicitly simulate individual particles of a granular assembly. Although for adequate results - just like in case of other numerical simulation methods in engineering - calibration of the input parameters of the model has a huge importance and calibration of micromechanical parameters is exactly what makes the discrete element method to be difficult to use for industrial modeling purposes. In many cases the calibration itself consumes more time and resources, than modeling the mechanical process itself, usually it uses trial and error methods. Aim of this paper is the presentation of development of a highly autonomous, standard shear test based calibration algorithm. For modeling the calibration process in a rectangular shape shear box Yade DEM software was used. For automatization, the process must be rigorously controlled; the trial and error process must be superseded by sensitivity test based "gradient" methods. As a conclusion the validation of the developed automatic calibration method is also shown. For further researches it would be an interesting opportunity the implementation of artificial intelligence within this iteration based calibration algorithm, to speed up the process.