[RonSwanson]

Hello all,

I have been trying to run SU2 and I believe I am unable to restart properly SU2 simulations. I will reach a point in the simulation where I have a low residuals, for instance prior to stopping I have a max residual of -7.5 and on restart I have a max residual of +2, with no parameters changed in .cfg other than switching restart from NO to YES. I am treating the software as I would use Fluent where I can cancel a simulation, change parameters and start where I left, perhaps this is incorrect.

I have now installed SU2 on three different computers (linux) and had no luck either restarting generally or when using the continuous_adjoint.py script. I am certain I am following the correct procedure but it is quite likely I have missed something.

Initially I was interested in using the restart function for convenience and transitioning from 1st to 2nd order, now however I want to test the adjoint and optimisation and clearly restarting is essential.

In order to rectify I tried a number of methods of stopping and then restarting:

-Terminating solution in terminal
-Defining a number of iterations in the .cfg
-Reaching convergence

I have also
-installed on a number of computers
-Tried versions 3.2.5 and 3.2.7
-Ran one of the continuous adjoint test cases


Has anybody had issues like this before? Thank you.

Regards,
Ross.

[fpalacios]

Hi Ross,
Sorry for the late reply, in order to properly restart SU2. You should typically rename the restart_flow.dat file to solution_flow.dat

In other words we use
RESTART_FLOW_FILENAME= restart_flow.dat

to write a restart file and

SOLUTION_FLOW_FILENAME= solution_flow.dat

to read a solution file. but you can use the same name for both. The same applies to the adjoint restart.

Thanks for using SU2,
Francisco Palacios
SU2 lead developer

[RonSwanson]

Thank you Francisco,

I have since moved from using the developers version to the 3.2.8 stable release and the problem has been resolved. I presume it was to do with my installation methods.

Many thanks,
Ross

[varghese9]

does the same apply for typical wing section module .as stated in jj alonso paper the wing is first kept in pitching for 3 cyles then the free response is measured for 2d flutter solution removing the holding pin .can i use the pitching case first as direct analysi for time equal t= 0.33 sec and later setup the 2d config file for flutter using the class solver typical wing section and use the last 2 output from pitching case to start the flutter response by switching to restart option with changing the file name of pitching case to solution _flow .dat from restart_flow .dat and run the response if not what should be the correct procedure.let me know if possible ----