Numerical Design of Random Micro-Heterogeneous Materials with Functionally-Graded Effective Thermal Conductivities Using Genetic Algorithms and the Fast Boundary Element Method.

M. DONDERO, A. CISILINO, J. P. TOMBA

CMES-COMPUTER MODELING IN ENGINEERING & SCIENCES

TECH SCIENCE PRESS

Lugar: Forsyth, GA, USA; Año: 2011 vol. 1974 p. 1 - 21

1526-1492

This paper introduces a numerical methodology for the design of randommicro-heterogeneous materials with functionally graded effective thermal conductivities(ETC). The optimization is carried out using representative volume elements(RVEs), a parallel Genetic Algorithm (GA) as optimization method, and aFast Multipole Boundary Element Method (FMBEM) for the evaluation of the costfunction. The methodology is applied for the design of foam-like microstructuresconsisting of random distributions of circular insulated holes. The temperature fieldalong a material sample is used as objective function, while the spatial distributionof the holes is the design variable. There are presented details of the FMBEM andthe GA implementations, their customizations and tune up, and the analysis for thesizing of the RVE. The effectiveness of the proposed methodology is demonstratedfor two examples. Computed results are experimentally validated using ad-hocdevised experiments.