In comparison to internal combustion engines, which usually have low frequency, broadband excitations, in electric vehicles, tonal excitations from the electric drivetrain are noticeable and disturbing. As the acoustic and structural dynamic behavior, often referred to as noise, vibration, and harshness (NVH), strongly influences customers’ quality perceptions, optimizing it is a key challenge in development. This study investigates the influence of static rotor–stator eccentricity on the NVH behavior of an electric drivetrain using a transient elastic multibody simulation (eMBS) model incorporating non-linear gear meshing, bearing contact, and electromagnetic forces. The analysis identifies the 36th order excitation of the electric machine as the dominant source, leading to a maximum total acceleration level of 152 dB. Two specific excitation directions were found to reduce this amplitude most effectively. However, varying the amount of static eccentricity in these directions resulted in only minor vibration reductions (<1.5 dB). The findings indicate that the symmetric mode shapes of the cylindrical housing govern the response, indicating that addressing the excitability of housing modes by developing asymmetric housing designs could offer a more effective approach for NVH optimizations of electric drivetrains.

The virtual testing of a technical product in terms of its acoustic behavior enables the engineer to evaluate the design without the need for physical prototypes. To predict the acoustic behavior virtually, models should consider both the dynamic excitations of the system and the transfer behavior of sound from the excitation source to the user. The transfer behavior is influenced by the resonance frequencies and the damping of the structure, the latter of which mainly occurs in the joints between the components. Stiffness and damping values are required to design the transfer behavior of the joints. For this, load conditions must be determined. To model load and displacement states occurring in the joint in dynamic applications, a distributed approach can be used by modelling in the FEM. This makes it possible to consider the influence of the microgeometry in the form of the surface roughness on the transfer behavior of the joint. To use the continuously distributed load and displacement states in concentrated model elements or as design parameters in the form of linearized stiffness and damping values, the load conditions must be linearized in the considered operating point. This paper therefore presents a method for the model-based derivation of linearized stiffness and damping values, which assesses the influence of the design of the joint on stiffness and damping on the basis of experimentally determined surface roughness. For this, distributed models in the FE are parameterized based on a measured surface roughnesses. Based on these models, linearized stiffness and damping values are determined for different joint geometries. This makes it possible for the first time to derive model-based linearized stiffness and damping values for different joint geometries. The Brake-Reuss beam is used as an academic object of investigation. Transferring the method to industrial systems can optimize costs and resources in the product development process.

Virtual testing of a technical product with respect to its acoustic behavior enables engineers to evaluate the design without the need for physical prototypes. In order to virtually test the acoustic behavior, the effects of both, the dynamic excitations of the system and the transfer behavior of sound from the excitation source to the user must be considered in simulation models. The transfer path is primarily influenced by resonance frequencies and damping. In structures connected via bolted joints, damping within the joint is the dominant source of damping along the transfer path. Therefore, appropriate structural dynamic models of bolted joint stiffness and damping are necessary. These models must represent both the normal and tangential stiffness and damping in bolted joints. Currently, experimental investigations are still required for the parameterization of bolted joint models. As an example, for the normal behavior, optical 3D measurements of the joint surfaces are conducted to obtain their topology. The surface topology is then used to identify parameters of an exponential contact law through elasto-plastic material modeling. To eliminate the need for complex measurements during model parameterization, this article presents a method for predicting the normal contact stiffness in bolted joints based on numerically generated surfaces. For this purpose, an algorithm is implemented to generate surfaces with different roughness properties. Based on these surfaces, the exponential contact law through elasto-plastic material modeling is performed. This calculation is used to determine the contact stiffness at different pressures. A Design of Experiments is conducted in which the surface roughness is varied to calculate the normal contact stiffness. Therefore, the surface parameters S _A , S _sk and S _ku are varied. S _A describes the average roughness and thus the size of the real contact area, S _sk characterizes the asymmetry of the height distribution and provides information on load-bearing peaks or valleys, and S _ku quantifies the sharpness of the distribution and thus the proportion of peaks. These properties significantly determine the microscopic deformation at the contact and therefore directly affect the stiffness and damping values of bolted joints. The resulting normal contact stiffness values are then used to create a regression model. This makes it possible, for the first time, to derive normal contact stiffness values for different surface roughnesses without performing any physical measurements. The normal contact stiffness values calculated from the regression model are used to parameterize a structural dynamic finite element model of the Brake-Reuß-Beam bolted structure. The model predicts the eigenfrequencies with a deviation of about 1.9% compared to experimental results, confirming the accuracy of the proposed approach.