Concentration 2

[WARose]

I have a stress concentration (see attached). It's a steel plate assembly (welded up). What surprises me is (looking at the forces in the post-processing file) is how much shear from the applied load tries to go out at the point where that concentration is. (Although it probably shouldn't since the span of that plate is about equal to it's depth. It's sort of like a compression strut to the point of concentration.)

The issue I have is: how do I smooth that out? (Most of the texts I have address this for openings in a mesh.) How far can I average it (if at all)?

Thanks.

 

[rb1957]

this is a linear run, right ?

1) run NL.
2) do a neuber calc
3) wave hands (modelling issue, local yielding at the end of the web, blah, blah, blah …)
4) can you add a radius to the web, to soften the attmt to the other web, and remove the hard 90deg corner ?
5) make the angle from two flat pieces (and a bent flange) ?

another day in paradise, or is paradise one day closer ?

 

[FEA way]

What kind of elements do you use in this simulation ? Can you show the mesh (especially around the stress concentrations) ?

 

[WARose]

this is a linear run, right ?

Correct.

1) run NL.
2) do a neuber calc
3) wave hands (modelling issue, local yielding at the end of the web, blah, blah, blah …)
4) can you add a radius to the web, to soften the attmt to the other web, and remove the hard 90deg corner ?
5) make the angle from two flat pieces (and a bent flange) ?

Good suggestions.

What kind of elements do you use in this simulation ?

All quads.

 

[WARose]

rb, one thing I forgot to mention is: it's not really going into the plastic range. (In fact, it's just below yield.) But what worried me is: it was beyond the allowable stress I figured for that area.

 

[Erik Panos Kostson]

Try to refine the mesh a couple of times and see what happens with the stress there. Just to see if there is a mesh dependency on the results that needs to be addressed, as rb1957 mentioned

 

[FEA way]

Yes, mesh refinement is first thing to try (that's why I asked for a picture showing current mesh in that region) along with nonlinearity. Are these quads second order ? If that doesn't work try modeling this with solid elements. They are better at capturing such stress concentrations.

 

[rb1957]

I think mesh refinement will only show a bigger "problem" as the FEM tries to deal with the web instantly joining the other web. Creating a transition like a radius, would help both the FEA "problem" and the real world problem.

How did you calc an allowable for this detail ?

another day in paradise, or is paradise one day closer ?

 

[WARose]

How did you calc an allowable for this detail ?

It's A36 plates (in a smallesh welded up assembly).....so I used AISC.

 

[ThomasH]

What I normally do in such situations is to do a non-linear analysis with yield stress as limit. Then I can see the size of the yield zone and the effect of the deformations.

If the allowed stress is below yield I would use the allowed stress as yield stress. I guess that is a code issue because I often have yield stress ass allowed unless the issue is buckling.

Or, how is the stress calculated exactly? Is it in the gauss points or extrapolated from gauss points to nodes?
The "problem" could be more of a software question.

Thomas

 

[rb1957]

so the plates are welded together ?

adding a radius to the web is probably difficult. how about drilling a hole in the edge (so you get a semi-hole in the web, yes?) sort of like an under-cut as a way to soften the transition into the web ? maybe attached pic helps.

another day in paradise, or is paradise one day closer ?

 

[WARose]

so the plates are welded together ?

Yep. Full pen.

adding a radius to the web is probably difficult. how about drilling a hole in the edge (so you get a semi-hole in the web, yes?) sort of like an under-cut as a way to soften the transition into the web ? maybe attached pic helps.

Good suggestion.