Shrink Fit Contact Force Issues

[Anonymous]

Anyone seen the below where for a symmetry model, the linear shrink fit contact produces high contact forces at the edges?

This throws off the overall contact force by a high degree.  The mesh is well-matched and I verified that there are separate nodes at those edges.  If I probe the axial force of one of the center nodes (in red) and divide by the area of a single element face, the pressure is spot-on with that from a hand calc.

I considered going the MES route, but initial testing revealed some considerable convergence challenges.

AstroJohn, I also saw something similar, but not to the same degree, in your archive file from an April 2017 thread (see bottom pic).

[Anonymous]

Moving to a full 3D model cures the issue and the total contact force matches hand calcs very closely.

So why the problem with symmetry?

[John_Holtz]

Hi @Anonymous

What direction is the axis of your cylinder parallel to? And in the full model, is there a symmetry constraint on either end face?

I did some tests with 1/8 symmetry cylinders. The contact force at nodes on the X-symmetry face and y-symmetry face were wrong, but the contact force at the nodes on the z-symmetry face were correct. I do not know why that would occur (and maybe it is just a coincidence in my model). In the correct results, the contact force at a corner node will be 1/4 the value of an interior node, and the contact force at a node on the edge will be 1/2 the value of an interior node (assuming the mesh is uniform over the contact area). As you point out, your model and my model are not showing this pattern. (Your full model is not showing those results either, but it depends on whether there is an edge effect that is causing a difference in the contact force.)

This will require a little more investigating to see if a pattern emerges, but it looks like a problem in the solver.

By the way, the summary file (DS.L) gives the reaction forces at the end of the file. The "Largest unfixed direction residuals" should be 0 when the results are valid. In my test, the X-Force and Y-Force unfixed direction residuals are quite large which indicates there is a problem in the solution. Viewing the "Reactions > Reaction Force (Negative) > Vector Plot" also shows an abnormal reaction force at those contact nodes.

[Anonymous]

X-axis.  For the full model, I extended the shaft and just constrained it.  I took a shot and left the hub constrained and it was OK.  Otherwise, I would split the outer face of the hub and add symmetry constaints to keep all displacements radial.

1/4 sym archive attached.

[Anonymous]

The contact forces from the 1/4 symmetry are 8 times higher than those of the full model and hand calcs.

[Anonymous]

Any updates here?  Should I file a bug on behalf or your client?

[John_Holtz]

Hi,

I do think this is a bug in the solver.

Strangely, the test models that I am creating now have a contact force of 0 on the symmetry faces. Before the contact force was higher that it should have been on the symmetry plane. (I hope I can find the old model. It would be nice to compare the two sets of results.)

I will document this behavior.