MES Simulation Locks-up Before Starting Analysis

[Alan_H]

I have a simulation (MES with Nonlinear Material Models) that I'm attempting to run but freezes before starting the actual analysis.  Here's a quick summary of what I'm seeing from a thread standpoing:

1)  AlgFrame.exe running (click perform analysis)

2)  AnalysisServer.exe thread starts and subsequent starts several other child threads in sequence, the last of which is apak4.exe.  (all threads appear to exit normally)

3)  apak4.exe spawns an instance of arun4.exe.

the problem appears that the AlgFrame.exe thread starts grabbing processor time again at this point (25% or 1 core) and grindes forever with nothing happening.  arun4.exe never starts to run on any core (no cpu time).  "ds.alg before kill.txt" contains log up to this point.

I've let my system sit overnight and still nothing.  Came in the next moring and AlgFrame.exe was still grinding away at 25% cpu utilization.  Here's the REALLY weird part.  I can kill the AlgFrame.exe thread and arun4.exe starts processing up until the first convergence at which point it errors out.  "ds.alg after conv.txt" contains the log at this point

Any ideas?

[zhuangs · Accepted solution]

 If running a model from interface, AlgFrame will always use CPU. And your model may be too large for your computer.  First, you can check whether the "Number of threads/cores" in "Paramters" panel is set to "All", and then here you can increase the "Percent memory allocation" to 80% or so.

If this can still not work, you might need to run from command line:

First, "Check model" from the interface.  And the "Check model" work is done, close FEA Editor.

Second, go to the design scenerio folder and open a command window. and then input

              "C:\Program Files\Autodesk\Algor Simulation 2012"\arun4 ds -run

After the simulation, you can open the model and see the results.

Hope this can help.

-Shoubing

[Alan_H]

Zhuangs,

Confirmed that I was running with all processors and bumped my memory up to 80% as suggested.  Same problem.  Did the second suggestion (solve from command prompt) and everything appears to run fine.  Once the system converges the first time tho I again get the recursive I/O operation error again.  I've attached a grab from the command prompt. 

If the model is to large shouldn't the system indicate something to that affect?

Alan

[Alan_H]

Shoubling,

Souring about getting your name wrong on that pevious post.  I tried something else that worked.  I created a new design scenario and deleted the original scenario (this way there wouldn't be any results in the DS).  I then did a "Check Model" from within the interface and subsequently closed it and ran the analysis at the command prompt.  It's just made it past the 2nd convergence so I think that fixed the problem.

Must have been something wrong with the results that had been created in the previous attempts that was causing the system from writing the results out (thus the I/O errors)

Thanks for the help.

Alan

[bjorn_fallqvist]

I've seen this also every now and then. Just close analysis_server.exe from the task manager. This helps me at least. A new analysis window opens, and the analysis starts.

[zhuangs]

I am wondering whether this is because "Perform analysis" is double-clicked.  I encountered this problem a few times for some large models when I wrongly double-clicked  "Perform analysis". But if I close one from the task manager, it will be fine.

-Shoubing

---

@bjorn_fallqvist wrote:

I've seen this also every now and then. Just close analysis_server.exe from the task manager. This helps me at least. A new analysis window opens, and the analysis starts.

---