I am using Simulation Mechanical 2017 and am doing a "transient stress-modal superposition" analysis of a rubber mounted aluminum beam assembly with rotating eccentric masses. The main beam component is a square structural tube which I modeled as a surface part in Inventor. Some other parts that attach to the beam are modeled as surface parts and some as solids. The analysis runs fine and I can get graphs of displacements and Von Mises stresses of the beam tube just fine, but stresses do not show up for components attached to the beam, either solids or surfaces. One of the attached surface components runs the length of the beam and is attached to it at various spots so certainly should have some stresses in it. When I query a point on that part and ask for a graph of stresses at that point it shows essentially zero until right at the end of the run time when the stress jumps to some ridiculously high value. For the element definition of the surface components I enter the thickness and change the "element formulation" to reduced shear. I leave the material model to "isotropic" and "properties" as part based and all the normal points as zero. Can anyone tell me what might be going on. Thanks. Phil pjquenzi@gmail.com
Solved! Go to Solution.
Solved by pquenzi. Go to Solution.
Hi Phil,
Sorry, but I do not have any good ideas.
If I understand your description, all of the parts show displacement results, and a graph of displacement one of the parts that has "no stress" looks reasonable. Is that correct?
And some of the parts that do not show a stress are solid (brick elements) and some of plates. Correct?
Actually, what is the size of the stress result file (.ns2, located in the ".ds_data\design scenario" folder where the model is stored). Is it larger than 2 GB? (Or 4 GB?) If the size is large, maybe you can output fewer steps to reduce the size.
John,
The ds.ns2 file size was very large at 5.24 GB. I reduced the number of steps by about 2/3 and the file size went down to 1.57 GB and stresses then showed up for all parts. Is there some work around to this or do I need to keep my models simple enough and the runs short enough to keep within the 2GB file size? Actually, I'm pretty amazed that I can even do transient stress analysis on models of this complexity at all. Thanks for your help. Phil. pjquenzi@gmail.com
Hi Phil,
It sounds like a file size problem although the behavior is not entirely like I would expect. (But that just depends on how things are programmed.) I will check with a co-worker on Monday about the file size limit.
Other than using fewer output steps, I do not know of any easy work arounds. Conceivably, you could write a program to split the results into N files so that each one is below the limit. Although you would be able to view all of the results, you would have to check N models separately, and each model would show the steps going 1 to x and the time from 0 to T. (Some of the analysis types get the step number and time from a file, so you can display "anything you want" by editing the file. But transient stress does not appear to work this way.)
P.S. Your previous post mentioned some attached images. The attachments did not come through.
John,
Thanks again. As long as I know about the file size limit, I can work around it. Apparently the reason the attachments did not get posted was because I sent them as an attachment to the email notice I got regarding your reply rather than going to the discussion group web site and posting them there. I won't bother to post them anymore since I know what the problem is.
Phil pjquenzi@gmail.com