Confidence Levels of Performance Map Based Drive Cycle Analysis for Induction Motors

Modern electric drives operate under highly transient and uneven torque–speed conditions defined by standardized drive cycles, where efficiency is typically estimated using steady-state efficiency maps whose predictive reliability under dynamic operation is not fully quantified. This thesis investigates the accuracy of steady-state efficiency map–based methods versus analytic time-stepping approaches for induction motor drive cycle analysis using a laboratory-scale machine and experimentally reproduced down-scaled drive cycles. Analytical, finite-element, and experimental modeling approaches are evaluated under identical drive cycles, including electro-thermal effects, and validated against measurements. The study systematically quantifies the trade-off between computational effort and prediction accuracy while analyzing the influence of grid resolution, grid placement, mesh density, and temperature. Results show that analytic time-stepping achieves the highest agreement with experiments, whereas steady-state efficiency maps offer a computationally efficient alternative but are sensitive to discretization and modeling choices. Finally, the work explores accelerated evaluation strategies using optimal sampling and data-driven surrogate models to enable efficient and confidence-based drive cycle performance prediction.