MESAHub
Sound speed limiter on TDC?
@orlox suggested that we impose a (smooth) limit on the convection speed in TDC, and implemented this in the tdc_gamma_no_cs branch.
I'd just like to have a bit of discussion about the physics involved here before we merge that in. In particular, I'd like to hear from @rsmolec on whether or not there is actually physics limiting convection to the sound speed, and to how consistent any additional physics we might include to create such a limit is with the physics already in TDC.
@rsmolec will certainly know better than I, but isn't it usually the case that if convective velocities approach the sound speed, some assumptions in MLT are already invalid? That's why I've always had some respect for MLT++... it knew that things were probably wrong, so it just swapped those wrong assumptions for other wrong assumptions that were numerically more stable.
Was MLT++ about supersonic convection? I always thought it was about locally super-Eddington convection, but I could well be confused...
My immediate reaction to 'MLT + supersonic' is that I get worried about the various uses of Cp, because the pressure fluctuations become large, but I don't think that changes anything quantitatively.
I also get worried that maybe the turbulent dissipation is no longer ~vc^3 and instead something like vc^3 * (vc/cs)^n for some n. That wouldn't quite serve as a limiter, but would make it harder to get strongly supersonic flows...
But I'm definitely naive here...
I think you're right that the primary trigger for MLT++ was super-Eddington rather than supersonic convective velocities. My recollection is that super-Eddington convective luminosity would usually coincide with supersonic convective velocities near the surface. Deeper in the star, though, I think it can still be super-Eddington without being supersonic.
So, my comment may be a non-sequitur for the scenario at hand.
generalizing adam's question just a bit, is there anywhere where supersonic convection takes place within stars?
for one, mach numbers in the oxygen and silicon shells at the onset of core-collapse peak at M~ 0.3 depending on mass.
For RSP/nonlinear pulsation calculations, this was not an issue. Strong shocks do develop in some models (I don't have Mach numbers in mind, but I can check that for a few models) and for these I found that we cannot simply get rid of artificial viscosity in RSP. It has to stay there with large cutoff parameter. Otherwise there are no specific additions to TDC implemented in pulsation codes to deal with supersonic convection. There is a convective flux limiter to deal with violation of diffusion approximation (not implemented in RSP, see Section 2.4.2 of my PhD thesis, and a limiter/correction term for inefficient convection (in RSP implemented differently, through Drad term), but these are different. What is the motivation?
I think the motivation is that @orlox saw strongly supersonic convection in ccsn_IIp when run with TDC and was concerned about that, but he'll know more than I do...
Can provide a few examples of cases that produce supersonic convection with Cox next week. It's not just crazy exploding stuff, happens as well in much more run-of-the-mill stars.
ok. an hr diagram colored or regioned by the max convective speed might be interesting.