This paper describes an optimization methodology de dicated to the minimization of the noise and vibrations in e le ctric motors. It relies on a numerical workflow which is de scribed in this paper, coupling models belonging to the fields of e lectromagnetics, structural dynamics and acoustics. Afte rwards, a deterministic optimization method is described and applied to automotive traction motors. As the optimization of e le ctric motors aims at manufacturing motors which are silent whe n operated, an enhancement of the optimization to consider the effect of manufacturing tolerances, materials properties dispe rsion and control uncertainties on the vibratory or acoustic le ve ls to minimize is finally described and this robust optimization method is applie d on a practical case.

This study presents a sensitivity analysis methodology used for electric motor design. This innovative approach evaluates both global effects of parameter variations in their design range and of parameter deviations in their tolerance intervals on design objectives. For the purpose of robust optimisation, this method helps to select the most influent design parameters and uncertain parameters, which are not necessarily the same. Suitable for any design approach, this method is particularly useful in dealing with objectives defined by non-linear and non-regular functions, such as electric motor acoustic criteria. In this study, the method is applied to the sensitivity evaluation of electromagnetic tangential excitations responsible for acoustic emissions in an electric motor. The sensitivity of output mean torque is also investigated. The sensitivity analysis shows that acoustic criteria appear generally more sensitive to parameter deviations than mean torque. Parameter deviations can be even more influent on acoustic criteria than larger parameter variations in their design range. As can be expected from the sensitivity results, the study eventually shows that the acoustic optimisation of the electric motor faces robustness issues.