Restart & mapping
 

Hi,

I am running a casa of motoring data using RANS simulation and I wanted to start from a certain point (restart.rst) in order to split this simulation in two different simulations: the first using LES, the other one holding input file of RANS in order to catch some differences!

Then I saved a restart file from the first RANS and then I have used this file as restart.rst file for both the aformentioned simulation, and since I want to create mapping I used mapflag, write_map.in, map.dat and so on...

However theproblem seems to be the restart.rst file itself. In fact simulation goes ahead until it reaches the point:

Reading restart map data from restart file restart.rst .........

JOB ABORT invoked by rank 0:
ERROR: Possibly Corrupt File, seeking data of size 51436220279 bytes, found restart file of size 51436220271 bytes

I'm quiet sure the restart.rst is entirely copied in the correct folder!

Hello Andrea,

What version of CONVERGE are you using? Are you using the same CONVERGE version for the case from which you got this restart.rst file and these new cases?

Sincerely,

Srivastava

Right!!!

It should be the problem!! Thanks a lot! I was totally focused on the inputs and I wouldn't even think to this!!!!

Thanks a lot!

Right!!!

It should be the problem!! Thanks a lot! I was totally focused on the inputs and I wouldn't even think to this!!!!

Thanks a lot!

Kislaya,

a question. What about running LES starting from a restart obtained with RANS. For instance let's suppose I have two engine cycles using RANS, then I stop it an I continue with LES for another cycle, or maybe to have a combustion process using LES with data initialized by RANS. What I should do?

I set restart_flag =0 and I used map_flag=2. The system gives me the same error: corrupted file. I suppose it is related to the switching between RANS and LES. Is there a specific way to bypass this issue?

Hello Andrea,

With the RANS model, have a map file (map*.out) generated. Map files can be generated at user specified instances and are always generated at the end of the simulation. By enabling Initial conditions & Events > Mapping variables > Cell mapping (map_flag=1 in inputs.in), you can use this map fie, renamed as map.dat, to initialize your LES run. This should help you bypass the issue.

Please do note the map.dat file will not contain any information on time. So remember to manually change to the start_time parameter in inputs.in to match the time the map file was generated for appropriate initialization.

Hope this helps.

Sincerely,

Srivastava

I'm also with this problem but in my case I'm not using different versions of Converge. Does anybody know how to solve it?


I'm simulating a simple duct with jump area, with velocity fluctuation in the inlet, NSCBC condition in inlet and outlet, without turbulence.



JOB ABORT invoked by rank 0:
ERROR: Possibly Corrupt File, seeking data of size 210 bytes, found restart file of size 200 bytes

Hello Bianca,

What version of CONVERGE are you using?

Have you changed any input files for this restart run compared to the original run?

Sincerely,

Hi Kislaya,


I'm using CONVERGE 2.4 and I didn't change nothing in my simulation. I just restart after run for the first time, with the same parameters. and I'm using write 3 restart files on the first simulation.

What version of v2.4 are you using? You can find this info at the header of your logfile or converge.start file generated during runtime.

It is 2.4.22.

Hello,

There's a possibility that your restart file is corrupted. Can you open a terminal from your working directory and execute command

tail -f <restartXXXX.rst>

The restart file is binary except for the first and the last line. The above command will print out the last line and if the restart file finished writing out correctly then you should see "ENDofRSTRT" being printed on your screen.

If this does not get printed, then the restart file is corrupted and you will have to use a different restart file.

If you see "ENDofRSTRT" then some input parameters have been changed before restarting. I suggest you diff the two sets of inputs to find out.

Thanks,

Hello,

I tested and showed "ENDofRSTRT".

The only thing that I changed in the input to restart the simulation was the total time. Is it should be a problem?
So do you suggest to create another folder to save the inputs for the restarted simulation? Or just rename the new inputs files?


Thanks

I believe you changed the end time of simulation? This shouldn't be done while restarting. Restart the simulation without any change and then while the simulation is running, you can change the end time as long as you have "reread_input" turned ON.

Alternatively you can do a "restart mapping" using the same restart file and that will let you change the end time while restarting. But note that this is a mapping operation and your grid is not preserved when mapping.

Thanks,

Ah ok, I will try this way.


But another question: so to restart the simulation I should change in the terminal the restart_flag to 1 or open converge studio, change in Simulation Parameters>Run Parameters>Run Mode>Restart and then export again the input files to the same folder of the previously simulation?

It does the same thing. You can simply open inputs.in and change the restart_flag=1 and set restart_number to 1 as well.

I tried both ways but still not working.
I also created a simple case to test the restart option and it gave the same error. Is it should be a error with the license?

Can you send your *.in and *.dat files to support@convergecfd.com using your official email? You do not have to send the restart files as the file size is too large to attach to an email. We will check why restarting or restart mapping is not working.

That restart number should be based on latest restart file or should be given by default 1? because out of a set of restart files ,there r numbers assigned to it like restart0029,restart0030, etc. so what should v do, if dat is the case?

Restart number is your own choice, doesnt depent on the number of the restart file. :)