Preliminary framework for physics-informed AI application to heat transfer in minichannel heat exchangers

This paper outlines a preliminary concept for applying physics-informed machine learning to analyse and correlate flow boiling heat transfer in minichannel heat exchangers. The framework is conceptual and is intended to guide future integration of experimental data with data-driven modelling. The Random Forest method is proposed as a candidate algorithm due to its interpretability and ability to capture nonlinear interactions among key dimensionless parameters. The study presents the planned workflow, feature engineering principles, and validation strategy to ensure physical consistency. The expected result is a preliminary methodological framework linking experimental and theoretical perspectives on heat transfer. Future work will extend this concept through comprehensive data acquisition, benchmarking, and analytical correlation development.