Modern product design demands reliability and cost-effectiveness. This dissertation introduces a novel simulation method to optimize eBike Drive Units, a relatively new product class whose diverse real-world loads and inherent uncertainties make robust design particularly challenging. By integrating field measurement load data with a data-driven approach based on FEM simulation data and machine learning, this work develops an efficient probabilistic fatigue calculation. This approach significantly reduces computational costs for reliability analysis, enabling robust design principles to be applied early in development. A case study validates the method's ability to assess uncertainty effects and support design decisions, offering vital insights for future eBike development.
Marco Steck (Wed,) studied this question.