Inventor FEA results not matching

[Anonymous]

So I am having problems with getting my stress analysis results to match together.  I get that the same simulations (even if they are copied) may yield different stress results from mesh and convergence and all that.

My problem is that when I run a simulation in a part (.ipt) file a few times and average the fairly close results, and then place that part inside an assembly (.iam) by itself and run the same setup, I get a pretty different result.  My part is a pretty simple rod fixed constrained at one and and a force on the other.  The values for stress in the part file are ranging from 103-122 ksi, and the stress of the part in an assembly by itself are around 71-76 ksi.

If anybody has any idea why this is happening, some insight would me much appreciated.

Thanks.

[Anonymous]

One thing to note is that assemblies and parts carry out their mesh refinements in different ways in Inventor, regarding the "p" and "h" refinements.

[JDMather]

---

@Anonymous wrote:
IAM with only one part?

---

Yes.  I was thinking of adding a second instance as the "wall" connection part.  Fully Fix the faces of the "wall" and Bonded Contact and see what results I get, but haven't had time.

[Anonymous]

---

@Anonymous wrote:
Thanks, I IAM with only one part?

---

Yes.  As the OP mentioned, he/she was simply trying to determine why the models mesh differently when a single .ipt, vs that same .ipt in an assembly.

[blair]

Odd, Displacement is identical, yet different Stess. Looks like an issue in FEA. I thought there was an issue back in IV2014 that was simular to this and a patch fixed it. Looks like the same issue is back in IV2016.

I'll forward this on to ADSK.

[blair]

Works fine in Simulation Mechanical. I get the same results as a IPT file or IAM file.

[Anonymous]

---

@Anonymous wrote:

Works fine in Simulation Mechanical. I get the same results as a IPT file or IAM file.

---

Nice find.

That's a little unsettling though...

[blair]

I thought the same thing happened back on IV2014 and a patch came out fairly quickly. Looks like someone grabbed the same code for IV2016 release.

Since Stress is a secondary value as a result of Displacement. If the Material is the same and the Displacement is the same, then the Stress must be the same.

[Anonymous]

I'm able to get the same result in the part vs the assembly by using local mesh control on all faces in the part and the assembly, obviously using a common element size, such as 3".

You need to pay attention to the convergence rate though, as assemblies and parts have different mesh refinement processes.