Preprint of "A reduced-order prediction of natural convection heat transfer coefficient including boundary layer phenomena for electric conductors in medium-voltage switchgears"

The proposes a simulation-driven correlation for the convective heat transfer coefficient that explicitly accounts for boundary layer separation through a separation ratio. A reduced-order model was developed using parametric, coupled electromagnetic–thermal simulations, providing detailed distributions of losses, temperature field, and convective heat transfer coefficient under various loads in natural convection conditions. The Nusselt number correlation was identified via data fitting regression based on a set of parametric mock-up simulations. The correlation was consequently implemented into full-scale multiphysics models for four switchgear panel variants in the current range from 1000 A to 4000 A. The reduced-order model was compared with full-scale simulations of several industrial switchgears, which were validated using a standardised temperature rise test. The dataset contains numerical data and graphical representation of data related to the study.