Assessing Thermohydraulic Performance in Novel Micro Pin-Fin Heat Sinks: A Synergistic Experimental, Agile Manufacturing, and Machine Learning Approach

Abstract

As advancements in technology and rapid product development redefine engineering paradigms, this study examines the influence of innovative and bio-inspired designs on heat transfer efficiency. The research evaluates the thermohydraulic performance of new biomorphic pin fins employing various strategic approaches and agile manufacturing techniques to optimise the design process. Experimental assessments were conducted on four hybrid pin fin configurations within Reynolds Numbers ranging from 101 to 507 and power outputs of 150W and 250W. The investigation focused on how different geometrical features impact critical performance metrics, including the Nusselt Number, thermal resistance, and pressure drop. Results indicate a significant enhancement in heat transfer performance, ranging from 25% to 45%, compared to traditional designs, even at lower Reynolds Numbers and energy consumption levels. Additionally, new empirical correlations were developed specifically for these hybrid designs. Machine learning models demonstrated high accuracy in predicting the Nusselt Number, using Reynolds and Prandtl Numbers as key variables, achieving a mean absolute percentage error (MAPE) of less than 3.5% and an R² value exceeding 0.95. Among the models evaluated, XGBoost, Random Forest, and Polynomial Regression exhibited superior performance with both real and synthetic data. This study underscores the potential of unconventional biomorphic geometries, highlighting the benefits of agile manufacturing and cutting-edge technologies in optimising resource use and improving predictive accuracy. The findings advocate for a reassessment of traditional heat sink designs and propose promising directions for future research in advanced sustainable thermal management.