[dkswnsgud98]

Hello,



I am running the simulation for the high speed flow over structure (above mach 10), and I am keep getting this error message: "Rank 2) WARNING: [3466] Temperature extrapolation error found at rank 2 at cell 2205, located at <0.479343, 4.96801, 0>, in region 0, whose level=8, ii=333, jj=508, kk=384"

Could you let me know how can I check that exact spot so that I can take a look and fix it?



I wonder if I can see that mesh or cell where the error occur.



Thank you in advance.

[MFGT]

You can create a vertex at that location to look at in Studio.

[ksrivast]

Hello Jay,

You may consider turning on the hidden parameter write_critical_posts_flag = 1. CONVERGE will write out post file when temperature extrapolation errors arise. The post files can be evaluated in Tecplot for CONVERGE at the printed location(s), in more detail.

Be wary of cases with too many recoveries/temperature extrapolation errors. CONVERGE may dump out a lot of post files which may exceed your storage limits.

Sincerely,

[grath]

Originally Posted by ksrivast

Hello Jay,

You may consider turning on the hidden parameter write_critical_posts_flag = 1. CONVERGE will write out post file when temperature extrapolation errors arise. The post files can be evaluated in Tecplot for CONVERGE at the printed location(s), in more detail.

Be wary of cases with too many recoveries/temperature extrapolation errors. CONVERGE may dump out a lot of post files which may exceed your storage limits.

Sincerely,

Hi Kislaya,

Sorry for my intervention, but I think I have this type of problem. I have a lot of memory issues with my case.
Sometimes, it crashes during post.h5, and sometimes due to "too many files opened...".
I also have some lines with "recoveries/temperature extrapolation errors" in my log files.

Could you explain a bit more about this subject, please?


Have a good day.
Sincerely.

[ksrivast]

Greetings,

We recommend you have a discussion with you IT team if you face issues with memory (unless you see a memory leak reported in your memory_usage.out file) and I/O file accessing.

Recoveries are ok to have during CONVERGE simulations. The solver cuts the time-steps dynamically if it faces difficulties converging to the specified tolerances with large time-steps. Frequent/Persistent/Too many recoveries are worth paying attention to, especially if they occur right before the case crashes. They might indicate issues in your solution. If log_level is set to 3 in inputs.in, the converge.log will report the XYZ location of such recoveries. The hidden flag mentioned above will allow you to write out a post file at the recovered time-step.

Hope this helps.

Sincerely,