Understanding End-of-Line NVH Scatter Bands in Electric Drives with High-Precision Simulations

An important characteristics of battery electric vehicles (BEVs) is their noise signature. Besides tire and wind noise, noise from auxiliaries as pumps, the electric drive unit (EDU) is one of the major contributors. The dynamic and acoustic behavior of EDUs can be significantly affected by production tolerances. The effects that lead to these scatter bands must be understood to be able to control them better and thus guarantee a consistently high quality of the products and a silent and pleasant drive. The paper discusses a simulation driven approach to investigate production tolerances and their effect on the NVH behavior of the EDU, using high precision transient multi-body dynamic analysis. This approach considers the main effects, influences, and the interaction from elastic structures of electric motor and transmission with accurate gear contact models in a fully coupled way. It servs as virtual end of line test, applicable in all steps of a new EDU development, by increasing front loading. Various parameters such as clearances, gear microgeometry, bearing deviation, misalignment, unbalance, electrical excitation, and control effects can be investigated for their sensitivity and impact on transfer and response. Such a model is used for dynamic analyses of specific use cases and operating conditions. The important part of this paper is the demonstration of the applicability of such a fully physical and complex approach for large-scale DoE to investigate the required tolerance space for the defined parameters and the parameter combinations without the need of model simplification or transfer to frequency domain and by making use of Cloud resources. The derived data is further on used to train a surrogate data model to cover the whole parameter space. The effect of changes on specific NVH KPI’s like mechanical orders, resulting in gear whine, and their separation from electrical orders, as well as the specific root cause of a detected phenomenon can be analyzed.