Non-Linear Analysis in Inventor Nastran

[henriquedpires]

Hello, everyone,
Thanks in advance for reading and taking care of my problem,

I am currently using Autodesk Inventor Nastran to solve a sequence of press-fits/interference-fits in a plate for my internship project. I've already tried to reach out to Nastran Forum but got nothing back. What I'm asking here is another way/idealization/idea to model this up and solve the problem or a recommendation of another software that could help me out.

My geometry is a plate with 5 openings (hubs) in which 5 shafts are going to be press-fitted. There is no symmetry, and the openings have different sizes. The assembly of each shaft will produce a stress and displacement field in all other openings, which I must measure and analyze.
What I want to do is: assemble the first shaft and look at the displacement and stress field of the plate. Assemble the second shaft and reanalyze the displacement and stress field of the plate… Do the same procedure until the 5th shaft.

Using the Inventor Nastran I’m struggling to achieve good outcomes, especially for the regions of contact between the shafts and the openings. For an interference-fit in Nastran I must use a non-linear analysis and the results of the contact regions have significant perturbations that invalidate the stress field in this zone. Since interference-fit does not give me a good solution, I have tried to use a shell in the place of the shaft and apply a described radial displacement to the shell until the position I want. For this case I used a simple “separation contact” (do not allow interference) and even this way I get bad and non-physically possible results for the contact zone (with a smaller amplitude right now, but evenly relatively high). I tried using hydrostatic pressure to the shell, however, it is not realistic due to the displacement field of the plate (not uniform around the openings), so it turns out the described displacement is the best solution in my point of view (but please let me know yours).
The problem is that when I constrain a described displacement (forced movement) in Inventor Nastran, I must block the same direction of the displacement I want to impose. So, in my case, I must block the radial displacement of the shell to input an actual displacement of its nodes. To impose sequential loads (1 and 2, for example) in Inventor Nastran non-linear analysis I must apply one load in one “subcase”, and both loads in the second “subcase”. This means that to get the result of the assembly of 2 shafts, I must press-fit one shaft in subcase1 and both in subcase2.
This method turns out into a problem for the 2nd, 3rd, 4th and 5th shaft assemblies in which I want to measure the actual displacement and final stress field in all the openings especially in those that had already been assembled. Due to the method, for example, when I press-fit the 3rd shaft, the 1st and the 2nd shafts (shells) have the global position restricted by the constraints I must apply to apply the actual displacement and keep the load throughout the subcases for a sequential load.

In your experience/know-how have faced a close position/similar problem?
Any idea to help me out? Any other way you would use to model this problem?
I am open-minded to different solutions. If you know/use an alternative FEM software (Abaqus, ANSYS, …) that would help me to model my problem and would work fine, please let me know.

 

[onatirec]

In Ansys, you can add/subtract solutions; I've done this before for a fatigue analysis with very complex loading.

I'm not sure how well your displacement boundary conditions will play with each other in this situation. If necessary, you could redefine the displacements as pressures acting on the IDs of the holes. There are several sources out there for hand-calculating pressures resulting from interference fits. I believe Ansys sums the resulting individual stress fields, but I'm a little rusty on how it works under the hood.

https://www.youtube.com/watch?v=2oLs7ES0hJc

It's not entirely clear to me why your using pressures didn't work for you. It could be that the proximity of neighboring holes resulted in less force from the strain, in which case the non-circularity of the resulting geometry could be physically correct. It could also be a meshing issue.

One final thought: you could potentially simulate it thermally. Set the shaft ODs to be the same size as the hole IDs, set the plate thermal expansion coefficient to zero and adjust the shaft CTEs individually if necessary. Then just "heat it up". At the end of the day, I only trust simulations but so much.

GL

 

[henriquedpires]

Hello, @onatirec

Thank you very much! It helped a lot.
I'm going to try the thermal condition and come back here to present the news if it worked, but it seems it's a very good idea. Thanks for your perspective.

Regarding the pressure condition, that's exactly the case, the openings are very close to each other. Therefore, we have zones that strain more due to the displacement of the neighborhood boundaries, which I think isn't the case since the displacement due to a press-fit should be uniform around the hole, shouldn't it? Anyway, I'll give it a try to refine the mash in those zones.

Thank you once again,
I wish you a great day,
Best regards,