Analysis of the influence of a pseudo-random number generator type on the kinetics of the cellular automata recrystallization model

Abstract

Models based on discrete simulation methods, i.e., cellular automata (CA) or Monte Carlo (MC), often contain probabilistic elements in the code to capture the stochastic character of material behaviour. They can be related to various physical aspects reflected in the model, e.g., heterogeneous energy distribution or random crystallographic orientation assignment to subsequent grains. Therefore, the paper's main aim is to analyze the influence of such random elements in the CA recrystallization model used to predict the material microstructure morphology after thermo-mechanical treatment. The first part of the work presents the available tools for pseudo-random number generation and the main elements of the investigated CA recrystallization model. Different variants of input data preparation prior to the simulation, including deformation energy distribution, are shown. The role of the investigated pseudo-random number generators (rand, lcg, mt, ranlux) and their distribution (uniform, normal, lognormal) is highlighted in this part. Finally, the input data are used to perform simulations of recrystallization progress, and the effect of pseudo-random number generation parameters on transformation kinetics and growth of recrystallized grains is evaluated. With this work, we have demonstrated that the choice of the generator type, particularly the method of its distribution, influences the simulation results in the form of final material morphology.