Population-based optimization for heat sink design in electronics cooling applications

Smooth and scale-roughened plate-fin heat sinks for electronic device cooling are considered. Developments in modeling momentum and heat transport in heterogeneous and hierarchical devices with full conjugate effects included provide the ability to rapidly obtain nonlocal descriptions of the flow and temperature fields in such devices. Such modeling, based on Volume Averaging Theory (VAT), directly incorporates the device morphology into the governing field equations, allowing geometric optimization to be based on theoretically correct governing equations that are quickly solved and rigorously derived from the fundamental Navier-Stokes and solid and fluid thermal energy equations. A number of optimization methods for heat sink designers who model heat sinks with VAT can be envisioned due to VAT´s singular ability to rapidly - compared to Direct Numerical Simulations (DNS) - provide detailed solutions. Design of Experiment (DOE) has been used in the past, and more recently Genetic Algorithms (GAs) and Particle Swarm Optimizers (PSOs) have appeared attractive for multi-parameter thermal-fluid device optimization. In this study, optimization employing a GA and a PSO on two types of heat sinks modeled with VAT is carried out and the capabilities of the two optimization methods are discussed. It is found that the GA and PSO methods are both effective in locating the heat sinks´ optimum configurations and that the computational time they take to do so is on the order of just several minutes when using a typical laptop computer. This study demonstrates the usefulness of population-based optimization methods in optimizing transport phenomena in heterogeneous and hierarchical heat transfer devices when VAT-based mo deling is emplo yed.