Discussion Forum

Roland,

you can certainly add material in the optimization domain. Check the truss example in the application library to get an idea how it needs to be done.

Cheers Edgar

Material is added or removed, but only in the optimization domain.

Roland,

you can certainly add material in the optimization domain. Check the truss example in the application library to get an idea how it needs to be done.

Cheers Edgar

Hi Edgar,

Thanks for your response. At this point I don't have much time left to run the optimization again. However, maybe you could clarify if the method I used would be accurate:

What I've done is used topology optimization to remove air material from the inner region of the loudspeaker, which creates voids, using the edges of the enclosure as a void boundary. My understanding is that COMSOL would then treat these voids as sound hard boundaries. So, as long as the material of the enclosure is acoustically rigid enough, this should be essentially the same as adding extra enclosure material into the air domain.

I'm just curious to see how accurate the method I used would be. I went over the results with my professor who seemed to think it would be accurate enough for the scope of my project.

Thanks,

Roland Baxter

Hi Roland,

I have never done topology optimization this way, so I cannot tell if it works and gives accurate enough results. I am wondering which way the optimizer creates voids in the model and assigns a boundary condition. At the end of the day, give it a whirl and compare to a more conservative approach.

Cheers Edgar

is used topology optimization to remove air material from the inner region of the loudspeaker, which creates voids, using the edges of the enclosure as a void

It is not that COMSOL treats the voids (or rather its boundaries) as sound hard walls, it is that YOU will have to interpret them as being sound hard. In the topology optimization, you have the two extremes air and "dense and 'stiff' fluid". So a very binary design, if you can get there, will have a distribuition of those extremes. There is large impedance jump going from one to the other, and so you interpret that as putting a mechanical structure where the one extreme is, and the rest is air. You will have to test in the end how well that proxy worked, especially if you final result was not very binary.