Contact for very large interference with compressible material.

[stuartf]

Hi Everyone

I am fairly new to Nastran and am struggling with a problem involving a very large interference fit with a compressible material. Due to IP restrictions at work I cant upload my model but have replicated the problem with a simpler model.

In this example I have a cylindrical part that is wrapped in a layer of compressible insulating material. It is inserted into a thin stainless steel sleeve so that the insulation is compressed and holds the inner cylinder inside the sleeve. The insulation is 8mm thick and needs to compress to 5mm. I have modeled the insulation at the uncompressed thickness and want to simulate the effect it has on the sleeve. Below is a picture of the model with a section cut away.

I am using a non linear study and have manually entered the material properties I have for the insulation material. I have set it to nonlinear elastic and entered the data I have for the stress strain curve. For the sleeve I have left it to behave in a linear fashion as I don't expect it to deform much and will certainly remain elastic.

For the study I am using a dummy temperature load and ignoring the inner cylinder. I have fixed the inner face of the insulation as I expect no movement here. To simplify things I have also cut the model along all 3 planes of symmetry and put symmetry constraints on these faces. I am also using shell elements for the sleeve although I have also tried solid elements without success.

The problem I have when I run the simulation is that the insulation doesn't "compress" inside the sleeve. See Image below.

I have read through Santino Keusemann' thread on pressfits (https://forums.autodesk.com/t5/nastran-in-cad-forum/press-fit/td-p/6496560) which helped a bit but doesnt solve my problem. I have set NCONTACTGEOMITER to 0 which doesnt help.

Any Ideas?

[stuartf]

As a test I have modeled a small flat piece of of the insulation material and done a nonlinear study by fixing the bottom and applying a forced displacement load to the top in order to compress it a known amount. The simulation behaves as expected and the reaction forces are what I would expect them to be considering the compression. This leads me to believe the problem lies with the way i have defined the contact between the sleeve and insulation.

[John_Holtz]

Hi @stuartf

I took a look at your test model. My suggestion is to model the steel sleeve outside of the insulation and use temperature to cause it to contract. In the process of contracting, it will compress the insulation. So the ID of the steel sleeve would equal the OD of the insulation, and the contraction will cause the steel sleeve to shrink 5 mm radially.

If the contact would work like you have it set up now (with the steel sleeve theoretically passing through the middle of the insulation), the insulation would have to compress "instantaneously". That will not be very good for convergence.

The alternative is to push the sleeve onto the insulation, but that is a lot more calculation to get to the final step.

Hopefully, you can ultimately do a similar setup in the real model.

[stuartf]

Hi John

Thanks for the feedback. The built in Inventor FEA has 2 contact types, "Shrink Fit/Sliding" and "Shrink Fit/No Sliding", that are able to do this however this is only for linear static analysis. This is why I had hoped to be able to do the same for Nastran and non linear analysis. Perhaps this is something I should add to the "Ideas" list.

Your idea for shrinking the sleeve using temperature is a good a one that I will investigate however the test model I uploaded is simplified from my actual problem. In my case the sleeve isnt round in cross section and the ends of the sleeve that isn't in contact with the insulation are "end formed" to a slightly different shape. Because of the non round shape the change in size due to temperature will not be equal on all sides and will need to be compensated for differently on all axis's. My aim is to analyse the effect the insulation has on the shape of the sleeve. What might be easier for me to do is include the inner part in the analysis. I can draw it undersized and use temperature to force it to expand.

I had also considered the idea of pressing the sleeve over the insulation. This is actually how they are assembled in real life but then I would need to include the tooling for compressing the insulation prior to pressing which I feel is beyond my capabilities in Nastran.

I will feedback attempts at using temperature to force a size change.