CALIBRATION OF PARAMETERS FOR PARTICLE SIMULATION OF BUILDING MATERIALS, USING STOCHASTIC OPTIMIZATION PROCEDURES

摘要

The Discrete-Element-Method (DEM) offers new capabilities for modelling and simulating granular and flowable building materials during the entire processing cycle. Particle simulation has already been used successfully to describe several processing steps of fresh concrete, e.g. modelling the flow behaviour of self compacting concrete, the mixing procedure of concrete within various mixer designs, the form filling process and the compaction of fresh concrete. A necessary and essential task within the modelling process is the description of the real material behaviour with adequate simulation parameters. Therefore simple experiments are processed, which are remodelled within the simulation. The simulation parameters are iteratively adjusted until there is an adequate match between real and simulated behaviour. This calibration procedure is time-consuming and difficult, so the goal is to find a new efficient method to determine adequate model parameters. Therefore the empirical parameter search was substituted by a systematic optimization strategy to find representative parameters. The usage of this strategy is shown by the slump flow test of flowable concrete. The sensitivity analysis determines the influence of each simulation parameter on the flow span and computes the coefficient of determination. The optimization uses the relevant parameters to find the best fit.