Artificial Neural Network Trained, Genetic Algorithms Optimized Thermal Energy Storage Heatsinks for Electronics Cooling

A thermal response model for designing thermal energy storage heatsink utilized for electronics cooling is developed in this paper. In this study, thermal energy storage (TES) heatsink made out of aluminum with paraffin as the phase change material (PCM) is considered. By using numerical simulation, stabilization time and maximum operating temperature to transition temperature difference is obtained for varying fin thicknesses, fin height, number of fins and PCM volume. The numerical simulation results were then compared with existing experimental work. The numerical results matched the melting temperature variation obtained by the experimental work. The validated numerical results are then used to train the artificial neural networks (ANN) to predict stabilization time and maximum operating temperature to transition temperature difference for new fin thicknesses, fin height, number of fins and PCM volume. Finally the optimization of the fin thickness, fin height, number of fins and PCM volume of the thermal energy storage heatsink is obtained by embedding the trained ANN as a fitness function into genetic algorithms (GA). The objective of optimization is to maximize stabilization time and to minimize maximum operating temperature to transition temperature difference. Finally the optimized results for the TES heatsink is used to build a new computer model for numerical analysis. The final optimized model results and the validated preliminary model results are then compared. The final results will show a significant improvement from the validated model. Further the study will show that by combining ANN and GA, a superior tool for optimization is realized.