Abstract
An effective computer generation method is presented in this paper to more perfectly and rapidly generate the random distribution domains with large numbers of grains (pores). At first, the geometries of heterogeneous grains and the stationary random distribution model with large numbers of grains are defined. Second, the effective computer generation method, including compactness algorithm and selection algorithm, is described in detail. Then the effectiveness of the generation method and the comparison with the take-and-place method are given, and some examples with different geometries of grains in 2- and 3-dimension cases are illustrated. The computer generation method in this paper has been applied to the computation of effective heat transfer behavior for the composites of the random distribution with large numbers of grains, and some numerical results are demonstrated. The generation method in this paper is able to make the generated samples hold better stochastic property, and it is also suitable to generating samples subjected to non-uniform probability model.