Structural Dynamic Modeling and Simulation of Acoustic Sound Emissions of Electric Traction Motors

The acoustic behavior of electric drive systems is one of the main comfort criteria of electromobility. Due to its high-pitched sound emissions, the electric motor plays an important role. The corresponding noise is predominantly determined by the vibrational behavior of the electric machine given by the structural transfer function. The early phase consideration of the vibrational behavior of electric machine structures becomes even more relevant if one takes into account the strong requirements towards lightweight design and spatial restrictions inside vehicle applications. One of the most important tools inside the early stage development is the structural dynamic simulation. In order to be able to sustainably predict the vibrational behavior of an electric machine, the corresponding simulation model needs to sufficiently represents all acoustically relevant structural effects and at the same time remain practical and numerically solvable in a reasonable amount of time. This conflict is dealt with in this dissertation. The acoustic behavior of electric machines is strongly coupled to the vibrational behavior of the electric machine stator. The microscopic representation of the strongly heterogeneous stator structure is elaborate and requires a large computational effort. Therefore, so-called homogenized substitutional materials are typically employed in structural dynamical simulations of electric motors. The homogenized materials intend to represent the effective stiffness and damping properties of the underlying heterogeneous structure by an anisotropic substitutional material. Typically, the corresponding effective stiffness and damping properties of the homogeneous material are reversely obtained from experimental investigations on the particular structure. However, this approach presumes the physical existence of prototypes that can be tested. In this thesis, different so-called homogenization techniques will be investigated that allow the identification of homogenized material properties based micromechanical models of the underlying heterogeneous structure. Therefore, various numerical and analytic approaches will be investigated. The resulting modeling approaches will be validated based on different experimental analyses on an exemplary stator structure and subsequently be employed in a comprehensive acoustic simulation of an entire electric drive train. However, the simulation and optimization of the mostly broadband acoustic behavior of electric motors remains time-consuming. In order to efficiently predict the acoustic behavior of electric machines the use of model order reduction methods can be advantageous. Model order reduction methods typically involve mathematical algorithms that yield the effective reduction of the model’s degrees-of-freedom. In this thesis, different model order reduction techniques will be applied and evaluated regarding their usability in the area of vibrational simulations of electric machines. A particularly efficient model order reduction could be achieved by using so-called Krylovsubspaces. By employing the Krylov-subspace method the solution time for particular operation points of the electric machine could be reduced to less than 10% of the original solution time. The integrated modeling procedure, presented in this thesis, yields the sustainable and efficient representation of the vibrational behavior of electric machines. It allows the early phase evaluation and optimization of the acoustic behavior of different electric machine designs. This thesis differs from similar research so far that a generic approach was used to make the representation of the global dynamic behavior of the electric machine possible. The process includes micromechanical models which add a unique robustness and sustainability to the approach