Should I worry about this warning message in my FEA analysis (structural engineering)?

[rte4563]

I have three separate parts (shell element plates) coming together at the bottom where I get my warning message - here the bottom concrete deck, a rectangular and circular concrete part all meet and my FEA software doesnt seem to like it, can I still run this safely or should I try to model it some other way? I dont see anything off with forces and displacements when I run my load case.

Warning: Unexecutable local refine of finite element mesh. Very small angle on the region/subregion at this point. The local refine was ignored for region/subregion.

Seen from birds view

How the mesh looks, dont see anything too crazy?

 

[rb1957]

I suspect the warning is that the element in your base at the intersection of the two standing shells is probably pretty awful aspect ratio.
It's not good ... and people looking at the model later will say terrible things about your model (and probably also about you).

An idea ... just an idea. Maybe don't mesh the base to match the curved walls, not at least at this point. If the base has a rectangle element here (instead of, my assumption, of triangular) and the standing shell was diagonal acrooss the base element ... doesn't sound like The Worst idea.

"Hoffen wir mal, dass alles gut geht !"
General Paulus, Nov 1942, outside Stalingrad after the launch of Operation Uranus.

 

[rte4563]

The aspect ratio doesnt look too bad no?

Mesh which is automaticall generated, looks like the base is rectangular ish:

 

[GregLocock]

At a guess it is complaining about that very skinny triangular element

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[rb1957]

that's what I meant by the "element in your base at the intersection of the two standing shells".
and my suggestion was to replace this triangular element with a rectangular one (so that the shell attaches to diagonal nodes).

another thing might be the base modelled in 3D elements and the standing shells are 2D ... and maybe there's no loadpath for the rotational freedoms of the standing shells ?

"Hoffen wir mal, dass alles gut geht !"
General Paulus, Nov 1942, outside Stalingrad after the launch of Operation Uranus.

 

[SWComposites]

What FEM code is this?
Are you running some sort of auto mesh refinement option?
The long skinny triangle is the likely source of the warning.
And depending on the FEM code, triangular elements are rubbish or worse.
Further, the stress results at the intersections of the parts are likely incorrect, depending on the type of loading.
And note, just because a mesh is auto generated does not make it good.

 

[GregLocock]

OTOH, that is naturally a very stiff part of the structure so probably not too important in the scheme of things.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[SWComposites]

Maybe. Depends on what the real structure geometry is, and the loading, and the material. Could be a fatigue critical area.

 

[rb1957]

"rubbish or worse" ... hee, hee !!

TET10s are not too bad.

Do you guys see a problem with attaching to diagonal corners ?

"Hoffen wir mal, dass alles gut geht !"
General Paulus, Nov 1942, outside Stalingrad after the launch of Operation Uranus.

 

[ThomasH]

Based on how the GUI for the software I guess this is made in FEM-Design. Then the model may show a concrete slab with a circular wall meeting a straight wall. I don't use FEM-Design but I know people that does and I think the possibilities to work with the mesh is limited.

The walls seems to be modelled based on center lines. Then it is impossible to not get a small angle when the curve meets the straight line. But, in reality the two walls will be cast together, creating a short thicker wall. If you model with that approach I think it will work better.

But if my assumptions regarding what you are actually modelling is wrong, the idea may not work .