Compressible solver with dynamic mesh relaxation zone velocity
 

Hello,

We combined the compressibleInterDyMFoam solver from v2106 with the wave relaxation techniques from waves2Foam ala waveDyMFoam.

The solver works very well and as expected in terms of wave generation and propagation whenever the grid is static or the grid motions are relatively small. However, as the dynamic mesh motion increases in amplitude (for example moving up and down sinusoidal in direction of g, ie, heave in the naval architecture world), the relaxation zones tend to introduce velocity artifacts that show up as vortex-like structures that can eventually pollute the entire computational domain. Often the wave elevation is not affected but the velocity field becomes very distorted.

I am attaching a video comparing the static grid case vs the forced-heave case. The forced heave is achieved using the oscillatingLinearMotion function. You can see that downstream around x=80m (and eventually upstream), a vortex-like structure is generated and it sticks around for the duration of the simulation while also getting advected around.

video

I am also attaching a screenshot with the structures highlighted in green:
https://i.imgur.com/L4PX1n5.png

In these simulations, I am using a single relaxation zone (inlet) via the cylindrical shape, so the solved-for regions is just centered around the center of the domain. The same behavior occurs with an outlet zone set to potentialCurrent for example. I have also looked at various lengths of the relaxation zone, domain sizes, fvScheme and fvSolution settings, but nothing seems to help.

It seems like the relaxation step may need to consider additional terms in the compressible multiphase setup with dynamic mesh motion. I'm wondering if anyone has seen a similar behavior or can point to ways of resolving this problem?