Fast multipole BEM and Genetic Algorithms for the design of foams with functional-graded thermal conductivity

M. DONDERO; A.P. CISILINO; A. RODRIGUEZ CARRANZA; G. STAVROULAKIS

Recent Advances in BEM to honor Professor D.E. Beskos

Springer

Lugar: Berlin; Año: 2009; p. 75 - 91

A numerical tool for the design of foam-like microstructures with customized heat conduction properties is presented in this paper. The spatial variation of the local material properties is achieved via the optimization of the spatial distribution of the gas cells. A Genetic Algorithm (GA) is used for this purpose. Although very effective and versatile GAs are computationally expensive. This problem is tackled in two ways here: a Fast Multipole Boundary Element Method algorithm is used for the thermal analysis of the microstructures, and the GA is implemented in parallel using a cluster of PCs. Two examples illustrate the performance of the developped implementation.