This article addresses various aspects and limitations associated with the Simcenter 3D method, Inverse Numerical Acoustics (INA) and also provides general guidelines for the users to ensure good results.
INA aspects and limitations
INA depends on many aspects such as number of acoustic nodes vs number of microphones, pressure field relevance etc. INA is an undetermined solution so there are no clear guidelines that ensure 100% accuracy. The ideal case would be to have equal number of microphones as the number of vibrating nodes and each microphone should ideally bring in new/unique information about the vibrations or relevant pressure field. However, it is not practical to have so many microphones or measurement channels in a test so what is suggested is to follow the general guidelines given below by making sure
to have a good number of microphones nearby radiating surface and that
they provide good coverage of the radiating surface.
At lower frequencies, having a higher number of microphones does not hurt. A possible drawback could be that too many microphones could bring in some redundant information into the Acoustic Transfer Matrix (ATV) matrix, which needs to be inverted. But for that reason, we apply the regularization and remove the smallest singular values in the Singular Value Decomposition (SVD) of the ATV matrix, before inverting.
General guidelines recommended for INA
Position of the microphones with respect to the radiating surface – nearby the radiating surface (better to be <𝜆/3 where 𝜆 is the acoustic wavelength corresponding to the maximum frequency) to capture the information of evanescent waves
Position of the microphones with respect to each other – spaced neither too far from each other (better to be <𝜆/2) to include the information with enough resolution on ATV nor too close to each other to have linear dependency of the ATV matrix rows
Distribution of the microphones – ensure good coverage of all radiating surfaces
Microphones should not be located on flat planes – better to follow the contour of the structure
Test data needed for INA should be collected in an anechoic chamber – if the test was done in a semi-anechoic chamber, the floor should be modeled with infinite plane in simulation for computing ATVs
If possible, it is better to measure the vibration at some points on the structure, to use as reference while checking INA results
Model should not have many sharp features like edges, corners – better to have unique element normals unlike those corresponding to the elements that form such corners and edges
Microphones should not be at the exact same location as the outer surface nodes of the acoustic mesh, to avoid merging of nodes in the INA simulation