Discussion Groups

Simulation CFD

Reply
Active Contributor
mikesnidersnider
Posts: 30
Registered: ‎09-03-2013

Crash Recovery

181 Views, 7 Replies
01-27-2014 12:45 PM

I have had several Design Studies crash.  A search of the folders and files created by these crashed studies found meshes created prior to the crash and data.  Please show me where I may find a procedure or tell me how to reload these good folders/files into the crashed or a new study.  I have tried coping the "design study" folder to a new study without success.  I have tried coping *.res.XXX  where XXX is the cycle count without success.  I have tried to start CFD from zero and iterate zero times and the mesher starts then there are no results.  I get "exception" message when the scenario file is copied from one study to another.   Perhaps I do not understand the file structure or Simulation CFD terminology or the way the code loads files and the difference between a CFDST file and a CFZ.  Clearly I need training in these areas.

 

I thank you in advance for any assistance with this matter

 

Mike

Please use plain text.
*Expert Elite*
OmkarJ
Posts: 406
Registered: ‎10-02-2012

Re: Crash Recovery

01-28-2014 01:24 AM in reply to: mikesnidersnider

I have found that the reason for the corrupt simulation is the scenario that you were working last time it was working. Just deleting the folder of that scenario can help you open the simulation again. But if this doesn't work, following may work:

 

1) Open cfz file, this will create all your scenarios/design studies

2) Open the scenario and run it for 0 iterations, mesh will be formed

3) Copy res.sxxx file in the crashed folder to this new folder. 

4) When you try to run the simulation, it will give you the option to run from sXXX iterations. Run here for 0 iterations

5) All your results will mapped to the new mesh. 

 

That said, you can't see your convergence plots because the st file in the new scenario contains the information of only 0 iteration (and not all XXX iterations). You can access the st file from crashed folder to see the results, in excel etc, if you wish. Copying this file to new scenario doesn't help in displaying the convergence plots.

Please use plain text.
Active Contributor
mikesnidersnider
Posts: 30
Registered: ‎09-03-2013

Re: Crash Recovery

01-28-2014 04:28 PM in reply to: OmkarJ

Hi OmkarJ,

 

Thank you for your reply.  I just crashed again.  I was looking at results real time and trying to figure out how to move the triade or XYZ arrows off of a results plane and a crash occurred.  My network consists of a computer “PC” which is a licensing server and a second computer “Thor” (Thor has two Xeons E-5s and 64 gigs of ram) which is my work station and solver computer.  I use Thor to run Inventor, Fusion and Simulation CFD.  Would there be fewer crashes if I use PC to monitor real time calculations  of SimCFD instead of Thor and still have Thor solve the problem?  I cannot resist evaluating interim results as they are be calculated.  (Kid in the candy store….)  Would this make my network and the software more sailor/kid proof?  Also a scenario may have five or more iterations of mesh adaptations in it when the crash occurs.  Would I have to run your procedure five times?  Obviously the last iteration is most valued but I am **** and want to have access to all of them.  Finally, I press save (the floppy disk icon) after I submit the scenario for solution and I am not sure if any hard drive activity occurs.  Perhaps this could be a future development to improve sailor/kid recovery. 

 

Thanks again and my you be blessed.

 

Mike

Please use plain text.
*Expert Elite*
OmkarJ
Posts: 406
Registered: ‎10-02-2012

Re: Crash Recovery

01-29-2014 01:23 AM in reply to: mikesnidersnider

From what I understand, Thor (fancy name btw!) is your workstation which is also a solver computer, thus simulation is solved locally. This in principle should be more robust. If in case, you use a remote solver in network then that would be a bit less reliable, as has been my experience. In your case, PC seems to be only the licencing server. As long as the network is consistent and licence is accessible, you should have no problems. That said, the natuer of error on the crash may also point to corrective actions in your practices.

 

For adaptive meshing, of course you would want only the last scenario to be recovered, as the whole point of mesh adaptation cycles is to achieve adequate mesh distribution, which is already saved in the cfz file for last scenario.

 

As for having the access to all other iterations, I have not yet found a way to make SimCFD recognize the st file of old scenario that contains the information about all iterations. If you just copy, it doesn't detect. If you copy and run for 0 iterations, it deletes the file and replaces with 0 iterations. If you are hell bent on seeing the iterations, say for observign the convergence etc, you can just open the st file with notepad, paste in excel and plot the iterations.

 

Please use plain text.
Active Contributor
mikesnidersnider
Posts: 30
Registered: ‎09-03-2013

Re: Crash Recovery

01-29-2014 10:36 AM in reply to: OmkarJ

Hi OmkarJ,

 

Thanks for you time.  Good Job!

 

Mike

Please use plain text.
Active Contributor
mikesnidersnider
Posts: 30
Registered: ‎09-03-2013

Re: Crash Recovery

05-16-2014 06:50 AM in reply to: mikesnidersnider

Hi Everyone,

 

I have crashed again with a scenario running.  How does one recover when an analysis is submitted and running as a service in the background?  I have tried running the CFZ file to create a "support" scenario and copying all the files in the crashed file to the support scenario with no results.  The running job continues writing files to the crashed folder.  I would like to monitor the analysis as it proceeds.  Your assistance would be most appreciated.

 

Captain Crash aka Mike Snider

Please use plain text.
Employee
srhusain
Posts: 57
Registered: ‎04-02-2014

Re: Crash Recovery

05-16-2014 08:26 AM in reply to: mikesnidersnider

I don't know if this will work for you but assuming that you did not modify anything in the original design study folder (where the .cfdst file resides) and sub solders, you can try the following steps with the UI closed:

  1. In the scenario folder(s)- these are the subfolders under the design folder(s), find and remove the files if existing: "prep_auto.xvs", "run_auto.xvs", and "post_auto.xvs".
  2. Open the design study in the UI

Hope this helps.

Please use plain text.
Active Contributor
mikesnidersnider
Posts: 30
Registered: ‎09-03-2013

Re: Crash Recovery

05-16-2014 09:06 AM in reply to: srhusain

Hi srhusain,

 

Thanks for your reply.  I found and deleted the .xvs file per your recomendation and Sim. CFD 2014 would not respond.  Windows reported the following:

 

Description:

  A problem caused this program to stop interacting with Windows.

 

Problem signature:

  Problem Event Name:                        AppHangB1

  Application Name:                             SimCFD.exe

  Application Version:                           14.0.2013.301

  Application Timestamp:                     513044b5

  Hang Signature:                                  3903

  Hang Type:                                          0

  OS Version:                                          6.1.7601.2.1.0.256.48

  Locale ID:                                             1033

  Additional Hang Signature 1:           39034447abef8a39d26d1851d0bbd0b9

  Additional Hang Signature 2:           3eca

  Additional Hang Signature 3:           3eca0a81865ff0b0d18315ffe5713c78

  Additional Hang Signature 4:           3903

  Additional Hang Signature 5:           39034447abef8a39d26d1851d0bbd0b9

  Additional Hang Signature 6:           3eca

  Additional Hang Signature 7:           3eca0a81865ff0b0d18315ffe5713c78

 

Read our privacy statement online:

 

 

If the online privacy statement is not available, please read our privacy statement offline:

 

 

 Interestingly a similar report was generated prior to my removing the .xvs file.

 

Thanks for any help you can provide.

 

Mike

 

Please use plain text.