LAUSR

Abstract The elastic behavior of materials is dominated by the microstructure, which will evolve with the temperature in the fabricating process. Recently, we have proposed an efficient and simple algorithm for generating spatially-correlated random fields, reflecting more realistic microstructures. Using a mesoscale Finite Element Model, we have studied the elastic behavior of materials with the pores of various concentrations and spatial correlations. Considering the distributions of pore sizes, we are able to predict the modulus of elasticity with the relative deviation of 5%–10%. Based on structural recognition and Monte-Carlo tree search, we have determined the main characteristic of microstructure which is dominant to the elastic properties, where the relative deviation of modulus prediction can be further reduced to be around 3%–6%. Our finding will provide a better understanding and prediction of mechanical behavior based on the microstructures of materials.