Investigating Thermal Performance of Substrate Board through Forced Convection and Machine Learning

Abstract

In this study, the thermal performance of substrate board exposed to forced convection with different heat source configurations was investigated. Seven asymmetric integrated circuit chips (heat sources) positioned at various points on the substrate board were cooled through steady-state experiments using laminar forced convection heat transfer mode. The objective was to determine the optimal layout of the seven integrated circuit chips on the board for lowering the maximum temperature. The optimal configuration was determined experimentally and was further validated by employing a machine-learning optimization strategy. Various correlations have been proposed to investigate the effect of the substrate board arrangement on the integrated circuit (IC) Chip temperature and heat transfer coefficient. These findings imply that the size and configuration of the substrate board, input heat flux, and placement of the IC chips have a significant impact on their temperature. Because the heat is discretely placed in this scenario, the temperature of the integrated circuit (IC) chips is the lowest for higher values of the non-dimensional parameter λ. This aids in efficiently reducing the temperature of chips through cooling. Another important factor in the cooling of IC chips is air velocity. The maximum temperature reduction is 14.02% at an air velocity of 3.5 m/s.