grid-aligned nonlinear thin shell instability

[BenWibking]

Describe the bug
We seem to be very susceptible to producing thin shells that are aligned with the grid axes when the thin-shell radius is not well resolved ($\Delta x \gg 10^{-2}$ pc).

To Reproduce
Steps to reproduce the behavior:

1. Run the metal advection problem
2. Examine the density structure near the midplane
3. See thin shells in square, grid-like patterns

Additional context
This appears to be a known problem. See, e.g., section 3.6 "Grid-forced instabilities of underresolved dense shells": https://academic.oup.com/mnras/article/507/1/1492/6348113#290180922.

Figure 2 of above:

cc @markkrumholz @aditivijayan

[BenWibking]

@markkrumholz @aditivijayan It might be the case that nothing can be done about this (except to increase the resolution whenever feasible).

Have you noticed this in other simulations?

[markkrumholz]

@aditivijayan and I did see this, but it was a very short-lived feature that only appeared relatively early in the simulation before the outflow was well-established and when SNe were still going off largely in quite dense gas. Since we drop that portion of the simulation from the analysis in any event (since it's before the flow comes into statistical steady state), we didn't make any effort to fix it.

[BenWibking]

Ok, thanks. I'm not actually worried about this for the global properties of these simulations, I just wanted to keep track of this for the future.