LE10 Benchmark - Solid Modeling Question

I took notice of the thread Greg had posted a while back of the NAFEMS LE10 benchmark and had also seen it in the Abaqus verification package and decided I would give it a try as well.

My trial on the problem I have been met with limited success. I get a solution but I am not sure I trust the answer because the problem is reporting a displacement in the Z of ~278,000 meters.

If someone has some time I would be interested to see what I might be missing that is causing the displacements to be so wildly off.

Thank you!
Clint

LE10 Benchamark.json (133.9 KB)

Hi Clint,

It is funny that you should ask this question, I have been working on this exact problem this week. I had the same results and found out that we currently aren’t enforcing edge based boundary conditions on solids. Which means your nodal_dva #4 isn’t being enforced and the whole part is free to move in the Z direction. Resolving this issue will be one of my main focuses these next few weeks so I hope to have more to tell you soon. In the meantime you can substitute the edge set for a node set, which will get you more reasonable results, but I can’t say that they will converge on the reported target because there aren’t any nodes on the desired midplane curve. Here is my version of that for your reference:
NAFEMS_LE10_Coarse.json (120.4 KB)
Let me know if you have any further questions, but again, I hope to have more on this in the near future.

Mitch

Hey Mitch,

I also just reached a similar conclusion. My current model is a coarse attempt, but, like Greg, I am seeing low error ~4% with the 16 elements on a 3 order basis. I hope to have a few more models of increasing mesh density once I get an appropriate card deck saved that will give me consistent results.

Another aspect I would like to see if there is an easier way to do this is selecting element sets in a 3D body. For 2D/shell parts, it seems like the sidesets in Trelis are automatically assigned as sets whereas the 3D sidesets are kept as sets upon import, but the actual content of the sets are lost.

I look forward to the fix, though it looks like it will be after the ‘trial’ period of the training course will have finished.