How to improve e-NVH simulation and tests correlation?


This article shows how to improve the comparison between MANATEE software outputs (simulated noise level radiated by electric motor) and e-NVH measurements.

Note that due to the multiphysic complexity of e-NVH phenomena, one may still have +/-3 dB accuracy from 500 Hz to 5000 Hz even with the most accurate modelling strategies and fitting work.

Improvement of test results

Before improving numerical simulation accuracy, it is important to compare exactly the same quantities. As an example, MANATEE acoustic output includes sound power level but sound pressure level is more easily measured; sound power level and sound pressure levels are indeed different, and sound pressure level measurements can include the following effects that must be carefully taken into account in simulations:

  • distance from acoustic point source: one should make the difference between center of the electrical machine and outer surface of the housing in test and MANATEE simulations
  • directivity: sound level measurements may vary a lot around the electrical machine at a fixed distance, MANATEE may give an average sound pressure level or a local sound pressure level depending on the acoustic model
  • presence of reverberation field: MANATEE sound pressure results can include reflection effects through room constant but it should be accurately measured
  • presence of background noise: before comparing simulation to tests, acoustic measurements should be carefully filtered to remove any parasitic noise source. Mechanical and aerodynamic noise may appear at same frequencies as electromagnetic noise is some particular cases.

For best fit between simulation and tests it is recommended to perform sound power level measurements using intensimetry and separate electromagnetic sources from non magnetic sources by specific tests.

Improvement of simulation results

The differences between simulation and tests can come at different levels of simulation :

  • calculation of electromagnetic forces: best electromagnetic model is based on MANATEE coupling with FEMM but real machine may include static and dynamic eccentricity, uneven airgap, uneven magnetization which significantly enrich Maxwell excitation spectrum and may cause differences between tests and simulations (new orders, new resonances). Such differences may also come from unbalanced currents, Rotor Slot Harmonic or unexpected current harmonics due to strong electromechanical coupling. Note that experimental phase currents, magnetization pattern and magnetic forces can be enforced in MANATEE.
  • calculation of vibration levels: best structural model is based on MANATEE coupling with FEA (e.g. Optistruct) using EVS method but by default MANATEE assumes a constant damping of 2% although damping depends on structural mode, frequency and temperature. One should know the operational damping of the electrical machine to be able to correctly compare test and simulation results. Note that experimental modal parameters can be enforced in MANATEE, as explained in the article on how to improve modal comparison between tests and MANATEE.
  • calculation of acoustic noise level: best acoustic model is based on MANATEE coupling with FEA (e.g. Actran) but by default MANATEE only calculates airborne noise due to outer structure vibration under magnetic forces. One may have to include the excitation of the rotor during the EVS process to include structure-borne noise and get simulation results closer from experimental ones. Note that modal radiation efficiency can be enforced in MANATEE.

See also

See MANATEE software industrial validation cases.

EOMYS engineering consulting team can also help you to obtain a good correlation between simulation and tests.

Read more