Development of a validated data-driven fin geometric design framework for microchannel heat exchangers

Microchannel heat exchangers (MCHXs) are widely used in compact and high-efficiency cooling systems for electronics, automotive, heating, ventilation and air-conditioning applications. Accurately predicting air-side performance across diverse fin geometries remains challenging. Conventional empirical correlations for the Colburn j -factor and Fanning friction factors ( f -factor) offer computational efficiency but are restricted to narrow geometric ranges. Machine learning (ML) models trained on computational fluid dynamics (CFD) data provide greater flexibility but often lack direct experimental validation. Previous studies have also relied mainly on offline validation using generic fin baselines, limiting reliability for practical design. This study proposes a validated, data-driven framework that integrates Design of Experiments, laboratory testing, CFD simulations, regression-based correlations, and ML models aimed at high-fidelity MCHXs performance evaluation. Eleven fin configurations were experimentally tested under dry conditions, generating air-side performance data with low uncertainties. Validated CFD simulations reproduced experimental results within ±20 % for both metrics and were used to establish a comprehensive database. Regression-based correlations derived from this database achieved high accuracy (R² > 0.95) across multiple fin structures. Among the ML models for various combinations of fin types, artificial neural networks and Gaussian process regression delivered the best performance (R² ≈ 0.99 for both j and f ). An uncertainty propagation analysis combining experimental, CFD, and ML contributions yielded overall errors of 7.30 % for j and 4.14 % for f , closely matching the observed differences between experimental results and ML predictions. The proposed framework enables accurate air-side performance prediction and rapid geometry optimization of MCHXs, supporting the development of energy-efficient HVAC systems.