Stress Analysis freezes!

[SteveJoynson9171]

I am having problems with my stress analysis. With some large components, it cannot complete a stress analysis.  I have increased the minimum and average element size but still nothing.  It will freeze halfway through (see pic attached) alothough it still says its ongoing but with no change for hours.  Is this something wrong my my computers processing capacity? as it only occurs with large components, although it has worked when more complex components.

Please Help!

[VinodBalasubramanian]

Hi,

Are you using a 32 bit machine or a 64 bit. If you are using 32 bit, you can step up your RAM usage by enabling 3GB Switch as per the document below,

http://usa.autodesk.com/adsk/servlet/ps/dl/item?siteID=123112&id=9583842&linkID=9240697

You could also try using the lower graphics settings by going to Application Options -> Hardware -> Compatibility View. This might take a lesser memory resource.

Was your stress analysis working fine earlier or you are having issues with large components from the beginning. 

Keep me posted with your answers. I will get back to you soon.

______________________________________________________________

If my post answers your question, please click the "Accept as Solution" button. This helps everyone find answers more quickly!

[SteveJoynson9171]

Tried your initial recommendation with no luck.

I have previously analysed assemblies containing more parts etc without problem, im not sure if it is the was I constructed the assembly? it only contains 3 different parts?!

I have analysed things since without problem also, this problem i would say has occured 3-4 times out of 30-40 assemblies I have tried

[VinodBalasubramanian]

Can you try the following addin utility and reset your stress analysis module to factory default and test whether you could able to run it successfully with your assembly.

http://usa.autodesk.com/adsk/servlet/ps/dl/item?siteID=123112&id=13404176&linkID=9242018

Hope it helps.

If not do let us know, if you can provide a sample dataset for us to investigate further.

[SteveJoynson9171 · Accepted solution]

tried that to no avail

regards

[henderh]

Hi Steve,

  If you could post the parts and assembly we can take a closer look.  If you don't want to post the the dataset publicly, it can be emailed it to me directly.

Thanks, -Hugh

hughDOThendersonATautodeskDOTcom

[davef]

Hello.  I know this is an older thread.  If there's a better place to post my concern - please let me know.

I am having the exact same experience with Inventor stalling roughly half way through large models.  Smaller models pose no problems.  I'm using Win 7 64, Inventor 2014, 32 GB ram.

Is there a solution or recommendation somebody can offer?  Thanks.

[GSE_Dan_A]

Dave,

There are several Features that can cause an Analysis to become sluggish.  If ANY of your parts have either enbossed text or actual threads (coil and revolutiuons on a bolt), it will cause the FEA to bog down.  I would look to exlude these features from your part(s) if they are present. 

[davef]

Hi Dan,

Thanks for the quick response. No embossed text in the model, but I do have threads.  The threads are not actual threads (modeled coil/revolutions on a bolt or in a hole), the threads in the assembly (male and female) were created with Inventor's thread feauture.  You know, the one that puts a pretty picture of a thread on the part but doesn't actually create the thread geometery.  I figured (perhaps incorrectly) that threads created this way would be ignored by the stress analysis function.  Is this assumption incorrect?

I did go a bit crazy with the meshing on the current model; it's arguably too fine.  I have a total of about 10 million nodes.  Sluggish behavior isn't too much of an issue for me; I'm fine with letting the computer do its thing over night.  When I go overboard with the mesh, I'll just have to pay the price and wait longer.  Software stalls/ hang ups are another much less acceptable story, though.  I'm hoping that there's an actual reason for the stalling - something that I can proactively address (in my models, software settings, or with teh actual hardware or operating system).  If there's a hard node/element limit - then I'll make a point of judiciously meshing my models so I stay inside the hard limits.  I fear, though, that there isn't a hard node or element limit.  The ambiguity (and fear) that results is frustrating.  Every time I run a simulation - I'll wonder if its too big and will cause a software hang-up.  Sadly, I won't know whether it will or not for several hours.  That's the rub.  If I could be told in 2 minutes, "Hey buddy - you got too many elements/nodes; the software is going to hang up on you six hours into this analysis" I might be able to make better choices and save wasted hours.  Anyhow, enough of my rambling.  Again, many thanks for your thoughts.

[GSE_Dan_A]

Inventor threads are acceptable! From my experience, those 2 features I mentioned are usually the culpret for me,  The mesh size does have a large factor in wait time.  If I recall, when using my 32 bit system, an analysis of no more than 50 parts could take anywhere from 20-30 minutes! Also, if you have constraints or contacts set in such a way that would cause problems, the analysis will hang up or take a lot longer in performing the FEA.  That was the real kicker for me! I would wait 30 minutes for the thing to tell me that the Deformation was too large... blah blah blah.  This was due to some improper constraint or contact.   When in doubt, try to simplify the model as much as possible! If it is symmetrical, do only half.  Try to eliminate any hardware (bolts, nuts, etc...) and use pre-tension forces.  There are threads that explain how to simulate a bolt in an assembly.  Sorry I am not much help.  Hopefully Hugh H can chime in (he's the FEA guru).

[blair]

If you can suppress any small holes and features that are not really required it will help with meshing and model analysis. Try starting with a coarse mesh and possibly refine the mesh where it's required.

If the model is symmetrical, cutting/splitting the model in either 1/2 or 1/4 will help as well. The use of symmetry constraints will be required.

[henderh]

Hi Dave,

  Those are excellent suggestions from Dan and Blair, I'm hoping one of those will alleviate the stalling issue.  Even if the suggestions improve the situation, please contact me via email to get a closer look at the models (if possible).

 The stalling during the meshing that you and Steve describe sounds like it could be an issue in our mesher, where we're somehow caught in an infinite loop, or we are making multiple attempts to generate a successful mesh using different approaches, and some are taking too long or failing.

  This could happen with geometry that isn't overly complex.  One of the best ways we can improve our mesher is to get a hold of these problematic models so that Development can investigate the cause and find a solution.  This will also help others in the future, that may face the same situation using different models.

Best regards, -Hugh

hughDOThendersonATautodeskDOTcom

[davef]

Hi Guys,  Thanks for your help.

An update; it seems that I'm finally past the problem.  I threw the kitchen sink at my model all at once (hardly a scientific approach) so I can't say with certianty exactly which of the changes I made solved the problem. 

I eased up on my crazy mesh and now have larger, more appropriate elements.  I also went through my contacts with a fine tooth comb and changed some contacts from bonded to other, more appropriate types.  I unsuppresed some contacts that I originally had supressed and asigned proper contact types to them. 

In addition to the mesh and contact changes, I also tweaked some local mesh controls and changed the constraint set between two parts. 

I suspect that there was something in my original set of contacts and/or constraints that couldn't be resolved and caused the software freeze.  A pop of message alerting me to the conflict would have been sweet - but I appreciate that the programming to trap 100% of these types of user errors before a freeze could just be impossible.  So I'm wiser now and when Inventor freezes in the future - I'll look hard at how I've set up contraints and contacts.

Again, thanks for your help.   

[tthoe]

Thanks for the follow up, Davef. This was helpful.