Program exception error during analysis

[Alan_H]

Ran into a new error today while running a non-linear simulation.  The specific error was:

forrtl: severe (157): Program Exception - access violation

It's not causing Simulation as a whole to crash, simply the ansysis.   I was solving a 2D MES simulation (with both bonded and frictional surface contacts) when the error occured. I'm currently running Simulation Mechanical 2012 on an XP (32 bit, SP3 installed) system w/ 4 Gig of RAM.  I've also included a copy of the ds.alg log file in case it would be helpfull since it contains the full error message and subsequent data.

Any ideas or workarounds would be greatly appreciated. 

[S.LI]

Hi,

Does this crash happen randomly or consistent at the same place?

Is it possible for you to share the model, so that we can investigate it deeply?

Thanks a lot and sorry for the inconvenience.

[Alan_H]

It happened consistantly with a single model (Again, never seen anything like this one).  I recreated everthing from scratch and I was able to get a solution, only not with my original model (and yes, I saved it expecting that question).

Alan

[zhuangs]

Hi Alan,

You are using "Frictional contact" in your model.  Please set up a user-defined contact distance with a larger value (than the automatic one (compare the value of contact distance - C_dist in ds.al file) for the contact pairs with frictional contact.  This might help

-Shoubing

---

@Alan_H wrote:

Ran into a new error today while running a non-linear simulation.  The specific error was:

forrtl: severe (157): Program Exception - access violation

It's not causing Simulation as a whole to crash, simply the ansysis.   I was solving a 2D MES simulation (with both bonded and frictional surface contacts) when the error occured. I'm currently running Simulation Mechanical 2012 on an XP (32 bit, SP3 installed) system w/ 4 Gig of RAM.  I've also included a copy of the ds.alg log file in case it would be helpfull since it contains the full error message and subsequent data.

Any ideas or workarounds would be greatly appreciated. 

---

[Alan_H]

Shoubing,

This did work for the simulation I was running at the time; however, I seem to be getting this error a lot with 2012 (don't remember ever seeing it in 2011).  It just seems weird that a contact tolerance issue in an anysis is causing access violations.

Part of the problem is there is no indication which surfaces with frictional contact are causing the problem.  Makes it rather difficult to debug.

Alan

[zhuangs · Accepted solution]

Hi Alan,

The access violations do not result from a contact tolerance issue.  However a bad contact tolerance might result in slow convergence with tiny steps, which results in more access to those files, where the access violation might happen.

I also encountered the access violations related to ds.t8, ds.al, etc, especially on VISTA or WIN7.

You are right.  The log file has no indication which surfaces with frictional contact are causing the problem.  And we have already noticed this problem and are planning to add some handling and make the product more debuggable to the user soon.

Shoubing

---

Alan_H wrote:

Shoubing,

This did work for the simulation I was running at the time; however, I seem to be getting this error a lot with 2012 (don't remember ever seeing it in 2011).  It just seems weird that a contact tolerance issue in an anysis is causing access violations.

Part of the problem is there is no indication which surfaces with frictional contact are causing the problem.  Makes it rather difficult to debug.

Alan

---

[Alan_H]

Shoubing,

Thanks for the feedback.  The tactic I've been using to work around this is to go with all frictionless contacts and  convert a single contact at a time to frictional contact between iterations.  Takes a while but at least this way I can typically figure out which set of surfaces is causing the problem.

Alan

[zhuangs]

Hi Allen,

The tactic you are using sounds great.  Sometimes I am using a similar method to investigate a model.

In the future the Autodesk Simulation product will help handle the "WARNING: NEUTRAL POINT IS USED BEFORE SET" warning better and report a diagnostic method with contact pair ID to the user.

Shoubing

---

Alan_H wrote:

Shoubing,

Thanks for the feedback.  The tactic I've been using to work around this is to go with all frictionless contacts and  convert a single contact at a time to frictional contact between iterations.  Takes a while but at least this way I can typically figure out which set of surfaces is causing the problem.

Alan

---