Spurious Buoyancy Plume from Immersed Boundary with FFT Pressure Solver

[liuchihl]

This issue is related to another one I previously posted (#4561), but it represents a different problem. Specifically, this highlights how using an FFT-based pressure solver in combination with an immersed boundary can lead to issues.

The model simulates tidal flow interacting with topography. Below are two videos with the same configuration, except one with the CG pressure solver and the other with the FFT pressure solver. Note that the total buoyancy B = $\bar{B}$+b, where $\bar{B}$ is the background buoyancy field and b is the perturbation buoyancy.

(1) FFT-based pressure solver:
An artificially generated buoyant water is created from the immersed boundary, causing a plume-like feature.

internal_tide_theta.0.0036_Nx.500_Nz.250_t.460_slices_xz-u-v-w-B_B_zoomin_FFT.mp4

(2) CG pressure solver:
The CG solver solves the issue.

internal_tide_theta.0.0036_Nx.500_Nz.250_t.452_slices_xz_B_zoomin_CG.mp4

[glwagner]

Very cool example illustrating the need for the CG solver! @liuchihl, should we convert this to a discussion? (Unless there are code changes that are being suggested, which could result in closing this issue.) What do you think?

[glwagner]

Also tagging @xkykai

[liuchihl]

Sure, I agree, converting to a discussion makes sense to me.