Discussion Forum

Hi everybody,

I'm pretty new at this COMSOL thing, so I could use any advise you can give me on the following problem:

I'm trying to calculate the ratio of strain energy stored in a thin film on the surface of a disk to the total strain energy in the whole disk/film system. The film is made of tantala (Y = 140 GPa, Poisson's = 0.23, density = 2390 Kg/m^3), and the disk is made of fused silica.

I've built the disk, which can probably just be modeled as a cylinder, 3inches in diameter and 1/10th inch thick, but I've added some small bevels because that's what we're actually measuring. The bevels don't really affect the mode frequencies anyway.

I can calculate the eigenfrequencies pretty easily, and they match what I expect. I can get the total strain energy by either doing a global evaluation of the total strain energy, or doing a volume integral of the strain energy density.

I can calculate the surface strain energy by doing a surface integral over the surface where the coating would be, and this gives me the linear energy density (in J/m), which is okay, because I can just multiply it by the thickness of the coating to get a total energy for the film. Unfortunately, this seems to give me the energy assuming the coating layer is silica. I have tried to change the material of just this boundary to tantala, but it doesn't seem to change the answer. Am I doing something wrong here?

I have also tried to make the tantala layer as a shell attached to the surface. I've used pinning constraints on the surface and edges of the shell so that it conforms to the shape of the silica disk.

However, when I try to calculate the energy using global evaluations of the shell strain energy, it gives me nonsensical answers. For instance: There are basically two modes at each frequency, rotated by 45 degrees from each other (aside from the radially symmetric modes, where there's only one mode at that frequency) . This is expected, and it's exactly what we see in measurements. But while the twinned modes should give the same strain energy because they're identical in all but rotation, the solver gives energies that are different by 2-3 orders of magnitude, and also much smaller than expected.

I can't seem to get anything by integrating over the shell strain energy densities either. It generally gives me zeros.

Any ideas what's up? Have I not properly constrained my shell? Is there an easier way to do this? At this point, I'm about ready to just export the displacements and calculated the energies myself.