Computational fluid dynamics-derived correlations for power transformer thermal-hydraulic performance prediction

Abstract

The cooling performance of power transformer (PT) windings depends on both fluid properties and winding geometry. This study performed 108 steady-state CFD simulations on a core-type, oil-directed PT to assess different axial (6-8 mm) and radial (4-6 mm) channel dimensions with mineral oil and natural esters across multiple Reynolds numbers. Numerical predictions were validated against experimental data, confirming the accuracy of the simulations. Results showed that pressure drop trends were unaffected by heat transfer and correlated well with fluid properties under isothermal conditions. Transformer thermal performance was mainly influenced by Reynolds number and axial channel width; natural esters produced lower winding temperatures than mineral oil, but at a higher-pressure drop. Using a porous-bed flow analogy, new correlations were developed for pressure drop and winding-to-oil temperature gradients, achieving deviations below 0.7% and 1.8 degrees C, respectively. These correlations provide accurate and computationally efficient alternatives to CFD during design, supporting both new transformer development and retrofitting with alternative fluids.