Efficient generation of random fiber distribution by combining random sequential expansion and particle swarm optimization algorithms

Representative volume elements (RVEs) with random fiber distribution are widely used in micro-mechanics for determining the properties of unidirectional fiber-reinforced composites from their microstructure. In this paper, the random sequential expansion (RSE) and particle swarm optimization (PSO) algorithms are combined to develop an efficient methodology for generating such RVEs. This methodology can successfully eliminate the biased regions of dense fiber population and unreal matrix-rich corners that usually appear in created RVEs and resolve the complications in selection of the input parameters in the RSE algorithm. Statistical analysis of the generated microstructures and two- and three-dimensional finite element analyses are performed to validate the capability of the proposed algorithms.