Investigating thermal performance enhancement in perforated pin fin arrays for cooling electronic systems through integrated CFD and deep learning analysis

Abstract

Pin fin heat sinks have garnered considerable attention within the realm of thermal management for high-heat-flux electronics systems. This study advances the understanding of perforated pin fin systems, offering novel insights into heat transfer enhancement. The research explores circular and square perforation geometries located at varying positions along the pin fin length, aiming to enhance heat dissipation and extend the lifespan of electronic components. A parametric analysis examines the impact of perforation quantity on thermal performance with a rigorous comparison against non-perforated pin fins. The investigation incorporates machine learning techniques, including artificial neural networks, to predict cooling system thermal efficiency. Using computational fluid dynamics simulations and Artificial Neural Network modeling, the obtained results indicate the optimal design mode is the use of 3 square holes on the pin fin, which leads to an increase in thermal efficiency by 16.63% compared to the case without pin fins. Additionally, the mean absolute error value calculated across all data extracted from the computational fluid dynamics simulation for predicting thermal efficiency was 2.25%. This low error value further demonstrates the accuracy of the simulation data and neural network model in predicting the heat transfer performance of the pin fin designs.

Keywords

• Deep Learning
• Thermal Performance
• Perforated Pin Fin
• Forced Convection
• Heat Dissipation