Re-entrant Corner Stresses

[nav]

I know re-entrant corners are a problem with FEA, however, I've found little information in the way of handling it, minus actually modeling radii, etc.  The pictures I've attached show the issue I'm having.

This area has already had the mesh refined.  The corners where the contact between plates is made only shows the high stress in the first elements and this is contained as the mesh size is decreased.

I've included a picture to show the larger mesh as well, this case had different loads so the values are different, but the distribution is the same and still shows the concentration I'm referring to. 

Currently the elements are modeled as 20 node bricks to help capture the effects of bending in the plates.

This is LSS model.

Basically, my question is, am I able to extrapolate stresses from surrounding nodes to achieve and an estimation of what types of stresses are occuring in this location?

[nav]

Another additional note is that the stress values are for fatigue so running the model as nonlinear is really out of the question.

[John_Holtz]

Hi nav,

I agree with your suggestion of using extrapolation to get a more accurate estimate of the stresses at that point -- minus the stress concentration effects of course.

I remember an issue with someone simulating a cantilever beam in order to understand the "stress anomaly" at the constraints. That was a similar situation. In this example, extrapolation gave the "exact" stress calculated by hand.

[nav]

Thanks for the reply John.

Do you have a good recommendation for stress concentration values in situations similar to those to those posted above?