[gdbb89]

Hi All,

I've created a 3D acoustics FEA program in MATLAB, and I believe I'm not applying free (i.e., open/no wall) boundary conditions correctly. Per all the literature I've read, an open cavity has its boundary conditions defined by setting the boundary's pressure p to p = 0. In an FEA formulation, this is done by removing the corresponding nodes' rows and columns from the system matrices. However, in my results, it appears that there's some amount of reflection occurring at the walls, as if some information is passing through the 1-element BC and reflecting information back.

Here is my attempt: https://imgur.com/a/kg8YZIK

I'm using 8-node hex elements in a structured grid. The shape functions are linear.

And a description of the simulation:

I've applied a 1 ms half-sine pulse of volume source rate excitation of amplitude 1 in^3/sec applied at the center of a 3D rectangular prism.

The excitation occurs at the geometric center of the box. The video shown is a y-plane cross-section at the middle of a box. The box has dimensions 50 in (L) x 20 in (W) x 30 in (H). In the top video, all six faces are open (p = 0) and therefore should not reflect acoustic pressure waves. In the second, all faces are rigid and therefore should reflect acoustic pressure waves.

Air is used with a mass density of 1.193E-7 slinch/in^3 and a celerity of 13,504 in/sec. Damping is modeled using Rayleigh Damping equivalent to [C] = 100*[M] + 1E-7*[K].

The solution is a direct solution; modal synthesis was not used. Numerical integration was carried out using the Newmark-beta Method.

Could someone please help me figure this out, or simply validate that this is how the physical phenomenon of what I'm simulating would actually evolve over time?

Thanks!