A validated shape optimization approach for maximizing the convective heat transfer from generic heat sink geometries

Abstract

Improving the effectiveness of heat transfer devices is one of the most challenging and determinant processes for industrial situations involving energy transformation. Thermal simulations can enable engineers to try and assess different approaches to deal with a heat transfer situation, at a very early stage in the design process of a device. The enhancement of the heat transfer potential of a device, based on the optimization of its shape, has been reported by many authors. This thesis presents the development of a shape optimization procedure for maximizing the heat transfer from complex geometries by (i) establishing and validating a robust and low-cost computational fluid dynamics (CFD) method to predict the fluid flow and heat transfer of steady and unsteady convective heat transfer phenomena, and by (ii) incorporating the CFD model in an optimization loop which aims at searching for the optimal design that maximizes the heat transfer, using a genetic algorithm. In parallel, this study could enable the improvement of our understanding of coral growth as these living organisms rely on mass transfer processes to develop. In order to achieve the development of a shape optimization procedure that can be applied to complex geometries, and additionally, that can provide results comparable to situations specific to the living environment of corals, the investigation was done progressively. The idea is to study flow and heat transfer configurations around geometries of increasing complexity resulting in a shape optimization methodology applicable to a wide range of shape complexities. A genetic algorithm is used to search for optimal solutions over a high-dimension search-space that represents all the possible designs of a complex geometry. The study of how genetic algorithms operate in combination with CFD simulations is conducted in order to assess the shape optimization methodology.