Comparative predictive analysis using ANN and RCA for experimental investigation on branched and conventional micro heat pipe

Abstract

The demand for compact, high-performance electronic devices rises, making thermal management in limited space crucial. This study presents a novel branched micro heat pipe (MHP) to solve this pressing problem. This novel method improves thermal performance for small-scale electronic cooling applications, surpassing conventional cooling methods. Innovative design has branched condenser of 12.7 mm evaporator, 24.6 mm adiabatic section and 12.7 mm condenser, each, compared to conventional MHP of identical size. Investigation encompasses a study of various parameters including working fluids, wettability characteristics, fill ratio, heat input, and cross-sectional geometries. The significant improvement in thermal performance by applying a hydrophobic surface treatment to the condenser section, increasing capillary force and condensate transport. Ammonia and de-ionized water excel thermal performance at lower and higher applied heat input (Qh), respectively. At higher Qh (>2 W), ammonia and toluene in the conventional MHP experience faster drying compared to the branched MHP, especially for lower fill ratio (0.6 FR). The branched MHP outperforms the conventional MHP, showing improved fluid circulation and performance, for fill ratios ranging from 0.65 to 0.4 and effectively transporting maximum heat flux in range of 10–14.4 W/cm2 with square mono-grooved MHP exhibits inferior performance compared to the trapezoidal and triangular mono-grooved. To provide a framework for future studies, we establish an artificial neural network (ANN) and statistical prediction model of thermal performance using the non-dimensional Kutateladze number (Ku), showcasing a robust correlation with experimental results and demonstrating only a minor deviation of ± 3 %.