Learning Forum

[casey-harwood]

I'm trying to run a fairly simple simulation of a canonical bridge span with fixed-fixed end conditions. Rectangular flat plate for a deck, with two girders. I created an acoustic fluid domain with the bridge located at the surface of the fluid (mimicking a bridge that is about to be over-topped by a flood), so the underside of the bridge deck, half of the bridge deck sides, and the entirety of the girders are in contact with the fluid. See below:

I am just trying to get natural frequencies and, by comparison with the dry resonant frequencies, infer the modal added mass ratios.

I went through my usual steps: formed a single part with shared topology, turned down the growth rate in the acoustic domain to 1.1, and set a suitably small element size on the bridge structure. Mesh below:



The bridge structure is currently set to aluminum alloy (this is a research-scale model), and the acoustic domain is set to liquid water or air, so I can compare dry and wetted conditions. I am not considering wave effects at this time, so free surface boundaries and gravitational acceleration are left out. Faces of the bridge with +/- Z normals are constrained (fixed-fixed).

The simulation executes normally, but my results are totally unphysical.

I'm getting natural frequencies that are 5-10 times higher when the fluid domain is set to water than when it is set to air (or when it's suppressed entirely).

Thinking I must have flubbed something in the setup, I created a new project from scratch, regenerated my geometry, and went through the setup again, only to get exactly the same results. Varying the mesh density doesn't seem to have any effect.

In other words, I'm getting a negative added mass. Or, what I suspect is really happening is that there's an artificial stiffness being introduced. I don't know how that would be the case, however.

Any ideas what I may be running into?