Machine Learning and Numerical Analysis for Predicting and Optimizing Heat Sink Performance in Interrupted Minichannels

Abstract

In present research article, researcher investigates geometric modifications in minichannel heat sinks to maximize thermal efficiency while minimizing energy requirements. Main focuses of this study is on interrupted channel designs, where preliminary findings indicate substantial heat transfer improvements using liquid coolants under laminar flow conditions particularly with V-type geometries distributed across channel widths. During the research a comprehensive finite element analysis is conducted to evaluate the thermal performance of various interrupted mini-channel configurations, including rectangular and V-type geometries. Numerical simulations provide insights into key performance metrics, including Reynolds number (Re), Nusselt number (Nu), heat transfer coefficient (h), geometric shape, and flow conditions. After numerical analysis six machine learning models namely Gradient Boosting, Support Vector Regression (SVR), and Ridge Regression, alongside Random Forest, Linear Regression and Lasso Regression are developed to carry out thermal performance optimization. An objective function is formulated to maximize both Nu and h simultaneously. This approach identifies optimal heat sink geometries while revealing performance patterns across different Reynolds number ranges. The research findings provide design guidelines for next-generation minichannel heat sinks with enhanced thermal management capabilities at reduced pumping power requirements.