[FLASH-USERS] flash 4.6 CCSN run problem

[kiyarash Taghiniyarami]

Dear Nilotpal Mukherjee,

I have had the same problem exactly near the bounce(in 2D) and postbounce.
Pumping energy in the zones for my case runs, finally leading to the error
such as imaginary sound speed, zero energy and density of EOS.
I tried to use a more diffusive hydro scheme (as an example from PPM to
WENO or set the artificial viscosity to higher values) for a few thousand
time steps when these errors pop up and then set back hydro setting to the
default. But that is a really tedious way of running the simulation. So I
got rid of the errors but also got the entropy up to even 50 or 60 in some
zones which seem to be unphysical.

When I used the artificial viscosity of 1 or 2 from the start of the
postbounce and use just PPM scheme, I could manage to run without any
errors up to 500 ms (then I stopped it by myself) postbounce and got a more
meaningful physical result according to Evan and Sean paper in 2015 (Shock
radius for 27M star reached to 400km in 221 ms post bounce for heat factor
1.0 and effective angular resolution of 0.75 and did not get any strange
high entropy in the zones  as the earlier case ).
That is what I found by accident. The evolution of shock radius did not
change by changing the artificial viscosity but the entropy in the
post-shock region was different.

Hope the others share their ideas on that.



Best,
Khodayar





On Tue, Jul 5, 2022 at 8:33 PM Nilotpal Mukherjee <laptonil3 at gmail.com>
wrote:

> Dear Flash users,
>
> I have been trying to run the CCSN setup found in FLASH 4.6, but we
> encountered a few issues, similar to what others have reported. Here is a
> short summary of what we did so far, and where we are stuck.
>
>
> We ran 1D CCSN simulation found in flash till just before the
> core-bounce(~228 seconds). We then took the data from the last checkpoint
> file, and then used this as the starting data for a 2D run.
>
> We encountered the same issue with the EOS convergence. Our EOS was the
> LS220_240r_140t_50y_analmu_20120628_SVNr28, though we tried and encountered
> the same issue with some of the other EOS's mentioned in the 2013 paper. We
> also put in temperature floors, so when temperature fell below the
> threshold, energy would be artificially supplied to bring the temperature
> back up. This was suggested as a patch for the EOS convergence issue that
> kept coming up shortly after core-bounce.
>
> After putting the floor in, the simulation did run for a bit longer,
> before it encountered some other errors and aborted. We figured out that
> the "nuc_low_eos" was causing some sort of error, so we decided to not rely
> on that and use "nuc_eos_full" for all the calculations. We were not able
> to single out the problem with the nuc_low_eos.
>
> We saw Nans appear in the leakage scheme, rt_calc_Leak.F90 i.e the dye
> value was coming out to be a Nan. What was also surprising was that the
> Nans seemed to have some sort of angular dependence, i.e Nans would appear
> along certain arcs of calculation (the ye graph is attached ). We then just
> decided to set dye equal to 0 whenever it gave a Nan and carry on with the
> simulation , this caused the simulation to run the longest (about 130 ms)
> after bounce. We are currently trying to figure out why these Nans are
> appearing.
>
> I would really appreciate any help/advice on how to solve the issue.
>
> Sincerely,
>
> Nilotpal Mukherjee
>
> _______________________________________________
> flash-users mailing list
> flash-users at flash.rochester.edu
>
> For list info, including unsubscribe:
> https://flash.rochester.edu/mailman/listinfo/flash-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20220705/703426b8/attachment.htm>