A comparative performance analysis of machine learning techniques for optimizing heat transfer between parallel disks

This research rigorously investigates the heat transfer dynamics between parallel disks through a combined experimental and machine learning methodology, focusing on heat flux, Reynolds number, and gap ratio. Conducting 100 experiments across varied Reynolds numbers, heat fluxes, and gap ratios, the study identifies optimal parameter values where the Nusselt number maximizes, attributing this to enhanced convective heat transfer. An artificial neural network (ANN) model, refined using Teaching-Learning-Based Optimization (TLBO) and JAYA algorithms, accurately predicts the Nusselt number, confirming experimental findings and providing a robust tool for optimizing heat transfer systems in applications like gas turbines and heat exchangers. The study underscores the critical importance of precise flow parameter control, offering significant advancements in the design and optimization of engineering systems involving parallel disks. Results demonstrate a highest average Nusselt number of 48.132 at a gap ratio of 20.38, heat flux of 444.30 W/m 2 , and Reynolds number of 99.99, validating the reliability of the proposed optimization models.