Is this a singularity?

[Jieve]

I am modeling a 1/8" plate plug welded to 1/4" tabs that are welded to a relatively rigid frame, and am trying to determine the min plate thickness I can get away with based on the applied plate load. The tab to frame welds are modeled as fixed (overly stiff but approx), plate is modeled with 1/4 symmetry, tabs are connected to plate using 3/8" spot weld connectors, and the tab edges have been radiused. I'm using solid elements (had some issues with shell/solid bonding), direct sparse solver and split lines to mesh control in the areas of high stress. There is no penetration contact between the other contacting tab/plate faces. I ran an h-adaptive study, but it will not converge since the spot welds result in singularities (I've read that these stresses are nonsense anyway). So ignoring these stresses, I've been applying manual finer mesh controls to the critical regions, and thought i had relative convergence until running my last (6th) study, where suddenly the stress at the tab radii and plate backside changed from being stable (plate from 5800 to 7300 psi, tab edge dropped from 8200 to 5700 psi), and I'm not sure whether there is an issue with my geometry, or if I can trust the results. I ran with the FFEPlus solver and got significantly different tab stresses (on the order of 2.5x less), so I'm not sure what to believe.

In reality the fixed mounting of the tabs is not perfectly realistic so this is sort of a worst case scenario, but I'm more confused by the fact that the results changed in the way they did.

I'm attaching images of the setup and results. Would appreciate any input on this.

Thanks!

(pics are uploaded as zip file, the next post has images uploaded in a single jpeg)

 

[Jieve]

Here are the results in a single jpeg.