Enhanced thermal performance analysis of flat miniature heat pipes using ANN and advanced wick-fluid configurations

Abstract

Heat pipes' high heat transfer capabilities with small temperature gradients make them key for efficient thermal management. The performance of flat miniature heat pipes (FMHPs) is examined in this study, with particular attention paid to the effects of operational parameters, working fluids, and wick structures. Thermal resistance, capillary limit, boiling limit, and temperature distribution under varied heat fluxes and ambient conditions were assessed using numerical simulations and artificial neural network (ANN) models. The findings show that the highest capillary limit (roughly 1 kW) and lowest thermal resistance (0.34–0.56 K/W) are obtained from sintered copper wicks that use water as the working fluid. With a slight decrease in capillary limit (roughly 10 %), the use of CuO nanofluids further reduces thermal resistance by up to 7 % at 10 vol%. Vapor velocities can reach 0.63 m/s in hotspot conditions, producing pressure gradients of roughly 18.63 kPa. The system's overall thermal uniformity is enhanced by multi-core heat loading. To predict maximum temperature, thermal resistance, and heat transfer with high accuracy (R2 > 0.99), an ANN model was created. When wick-working fluid combinations were compared using a normalized Figure of Merit (FOM), it was found that water with mesh screen generated the highest FOM (1.0), while water and sintered copper remained the best option for high-flux applications. This work offers a thorough framework for optimizing FMHPs through machine learning techniques, advanced modeling, and new performance metrics, promoting better thermal management in high-performance systems such as electronics cooling.