[blank665]

Hey guys,

I'm trying to simulate a problem in a server cluster. I start the solver via script over the .def file. The problem is, even when I set a maximum iteration number of lets say 100, the simulation goes on and on, until the server cluster shuts down my job due to walltime limit. Because of that no .res file is created and the simlation was in vain.

This is the script im using (provided by my institute):

https://i.imgur.com/VEn3QqR.png

Is there a way or a batch command to put into the script, so that the result file is created lets say after every 10 iterations ? Or a simpler way would be a command for stopping my simulation correctly after 500 iterations.

Kind Regards

[Gert-Jan]

Don't add links in your posts. No one knows what's exactly behind.
Post additional info on the forum itself. Please go to "Go advanced" to see how. See below.....

[ghorrocks]

There is a FAQ on images: https://www.cfd-online.com/Wiki/Ansy...n_the_forum.3F

To answer your question:
If you set up the simulation to have a maximum of 100 iterations in CFX-Pre then it will stop at 100 and produce a res file. If your simulation is going beyond 100 iterations then you have not set this.

You have options to save res files every 10 iterations, at any iteration number or many other options. It is all in the Convergence parameters in CFX-Pre.

A final comment:
You appear to be running in PBS, but you just call the cfx5solve command which probably will not run inside PBS. Are you sure this correct?

[Opaque]

When you are running on a cluster, and have limited time available for a given simulation, you must set the maximum "wall clock time" to a smaller value than the allocated time so the simulation can finish smoothly.

The alternative is to use backup files and let it fail and recover from the last backup file. Though it works, you will be wasting the effort from the last backup file to the point of failure.

The cluster queue is doing what is intended to do, to finish the job immediately once it runs out of time.