Efficient Prediction and Analysis of the Noise Radiated by an Electric Powertrain

Reducing the emitted noise from vehicles is a primary issue for automotive OEMs due to the constant evolution of the noise regulations. In the context of electric powertrains, virtual prototyping has proven to be a cost-efficient alternative to the build-test process, especially in early design stage and/or if optimization is targeted. Due to the multiphysics nature of the model, the full simulation chain involves multiple components, each having its own specific modelling attributes. The difficulty then resides in the parts assembly, solving issues like mesh-to-mesh projections, time to frequency-domain transformation, 2d-axisymmetric to 3d mapping, data formatting and management, unit and local coordinate systems… This paper presents an environment that allows for the prediction and analysis of the noise radiated by electric automotive powertrains. The stator-rotor electro-magnetic behavior is represented by time-dependent forces applied on stator teeth. Transfer functions from structural modes to acoustic pressure describe the vibro-acoustic behavior and ensure a fast synthesis of the radiated noise. It is demonstrated how and where harmonic and space order decompositions are introduced in the computational process to deliver efficient and powerful analysis means to drive design decisions. As such, the workflow operation does not require deep expertise neither in electro-magnetic nor vibro-acoustic simulation. The proposed workflow, implemented in the Actran acoustic simulation environment, is finally used to obtain and analyze results on a typical industrial electric powertrain model.