Genetic Algorithms Based Methodology for Optimizing Thermal Performances of Micro-Channel Cooling Systems on Chip

摘要

This work is focused on analyzing the thermal transfer in micro-channel heat exchangers integrated onto the backside of VLSI chips. The analysis reveals a multiple dependency of thermal performance on design parameters, thus making difficult the optimum design of micro-channel geometry for maximum thermal transfer. To optimize the micro-channels, a Genetic Algorithm based methodology is proposed. The method is based on the ability of GAs in finding the extremes of a function when these can not be deduced by analytical methods. Initially, the analytical modeling of the heat transfer is considered, and then the GA is applied to find the geometrical characteristics of micro-channels with the condition that the chip temperature rise resulting from the dissipated power, to be of minimum value. The results are in good agreement with those in numerical simulations, other analytical models, and experimental available data, providing us a promising method in the study of heat transfer enhancement in micro-channel heat exchangers.