Discussion Forum

Hi, I've been using the Pressure Acoustics interface to evaluate total power absorbed in a domain and have come across some unexpected behaviour. This has been driving me nuts for about two days, but I've finally figured out the problem is that the values of acoustic intensity returned at boundaries with PMLs do not match the behavior one would expect based on how other features function.

What I've been doing is integrating the boundary-normal component of intensity over an enclosing boundary to get power absorbed. Since air is modeled as lossless, this should give identical results for any choice of boundary (subject to discretisation error). The boundary integrals in question are similar to the exterior field integrals (they are also boundary integrals that involve p and the boundary-normal component of grad p). Since these often seem to be evaluated at the boundary of a PML in examples without issue, I assumed this would be the case for my integrals involving intensity too, but it isn't.

Please see the attached model for an example. The same boundary integral is evaluated over an inner boundary that's in contact with an obstacle, a middle boundary that's in free air, and an outer boundary that's in contact with a PML. All three of these should return the same result but they don't - the inner and middle ones match, but the outer one that's in contact with the PML returns different values. Two boundary conditions on the obstacle are shown: In one case the obstacle is rigid so no absorption should occur. In the other it has a positive real surface impedance so absorbs sound power. In the first case the outer boundary integral returns a negative value, indicating power generation, proving it is the incorrect one.

Are you able to comment on why this works correctly in the exterior field calculations but not in what I have done. Is there a fix?

Incidentally, if there's a better way than what I've implemented to compute the boundary-normal component of intensity then I'd be interested to know it.

Thanks.