Estimation of the windage loss and heat transfer characteristics inside the finite length of electrical machines’ airgap based on CFD and MLA

The performance of electric machines heavily depends on the airgap length, as it affects magnetic energy transfer. A larger airgap increases the magnetic circuit reluctance, reducing output power but making heat removal easier. A numerical approach estimates airgap heat transfer and windage loss, validated against analytical correlations based on Taylor-Couette flow, with the inner cylinder rotating and the outer stationary. Heat transfer and windage loss correlations are developed for various airgap ratios (G) and aspect ratios (AR). Skin friction coefficients for different airgap geometries are estimated to calculate windage loss for high Reynolds and Taylor numbers. The airgap ratio significantly impacts heat transfer, while the aspect ratio strongly affects windage loss. Machine Learning Algorithms (MLAs) are trained and tested on 1200 data points from high-fidelity Computational Fluid Dynamics (CFD) and Computational Heat Transfer (CHT). Comparisons of Artificial Neural Network (ANN), Random Forest (RF), and Support Vector Regressor (SVR) performances against CFD data show that ANN predicts skin friction coefficients best, while SVM excels in predicting windage loss and the Nusselt number.