Filtering in microchannel heat sink topology optimization

Abstract

In recent years, microchannel heat sinks have increasingly been designed using mathematical approaches such as topology optimization. This tactic can lead to novel heat sink lay-outs with enhanced performance over existing designs. Physics dictate a preference towards very short and small channels for the cooling fluid, benefiting the heat transfer performance. However, the feature size depends on the coarseness of the numerical grid used in the optimization, and very fine designs can simply not be manufactured using existing technologies. In structural optimization, mesh-independent filtering methods are used to ensure grid independence and manufacturability. In this research, design and sensitivity filtering are tested within liquid cooling microchannel heat sink topology optimization. The optimized designs are evaluated with a focus on the level of grid independence and manufacturability that can be achieved through incorporation of these methods. Additionally, a mutual comparison of both filter methods is performed. Results indicate that for unconstrained optimization problems and under certain conditions relating the filters, it can be proven that both methods will lead to identical results.