FEM - High Shear Force Concentrations

[lawrence_budzisz]

Hi all,

I have a question regarding shear force results obtained from my FE model. I’ve modelled a concrete slab of a uniform thickness of 500mm pin-supported at three sides. The slab has an area load applied of 7.5 kN/m^2 and three line loads around the perimeter of 600mm openings of 28.5 kN/m.

The attached screenshots represent a sample ULS result of shear in Qxx direction (local x direction is as per the global). The slab is developing very high shear force along the long (pinned and free) edges of the slab switching from positive to negative values as if the edges were acting as beams.

I have checked the actual reaction values along the edge and they are nowhere near the results shown in the maps.

I would appreciate any help in explaining the issue behind my problem and why the slab is behaving this way producing such high shear values.

Thank you in advance for any help on the matter.

Kind Regards,

Lawrence

[Rafacascudo]

can you send the model?

Zip it before attaching

[lawrence_budzisz]

Thank you for taking interest in my problem @Rafacascudo 

Model is attached

[Rafacascudo · Accepted solution]

I think the results are correct.

Yes the slab strip on the free edge side behaves almost like a beam and it also has the 3 holes to help isolate this strip . On the side close to the long pinned support the diagram is similar but it is inverted.

I think the long pinned edge reactions are better compared when you plot the shear diagram that is occuring perpendicular to this edge.

   The below picture  shows a panel cut that represents the shear perpendicular to the long pinned edge . On the same picture is also shown the vertical reaction on each node of this pinned edge. See how they agree perfectly.

[lawrence_budzisz]

Thank you for your prompt response and graphical illustration @Rafacascudo 

I do however wonder why is there a shear force in Qxx of such a high magnitude.

The mesh is quite dense as it's 100x100mm square mesh. The drop of force between the edge of the slab (over pinned support) and say 200mm away from the support is quite rapid. 

Looking at this from a completely theoretical point of view, would you be able to recommend a way of averaging a result like that? If I were to look at this as an integrated beam in the slab it is hard to see where the "realistic" shear value starts.  Over 200mm it drops from 749.9 kN to 301.63 kN.

 

Qxx Shear over the long pinned support

 

Thank you in advance for any help with this problem. 

[Rafacascudo]

What I normally do is set a cut in the lenght I find reasonable. For example , in your slab I set a 171mm cut and chose to show the Shear through the cut , Qxx. Then Robot gives the cut integral value , which happens to be the total shear through the cut . Then I have a value per meterwhich I can use as a base for design

[lawrence_budzisz]

Thank you very much for your response. I used local averaging before using map results and manual average calculation but the shortcut you've presented me using map cuts over certain distance looks like a great time saver. 

I guess the range over which the averaging should be done is really a judgment call. Would you know of any typical rules of thumb that could be used as a starting point? I usually fear getting results like this as they could easily be misinterpreted and I'd still be half tempted to try and design for maximum values even though in most cases it's simply impossible. 

Thank you a very much for looking into my problem and sharing your tips. 

[Rafacascudo]

When, for example, it is a the slab region around a supporting column(modelled as a bar) , I usually set the panel cut to be positioned on the "virtual" column faces . Alternatively you can set rigid links from the column end node to created nodes on column section perimeter and read the force values on this perimeter.

  So ,looking at your case , maybe you could set "that" cut to begin on the "virtual" face your support(the long wall or beam) and stop inwards wherever you judge to be enough.

Assuming a 10cm width for you wall or beam , you will have that panel cut 5cm from the slab edge  and overall Qxx value would be significantly smaller.

In your case ,  QXX map (paralel to the linear long edge support) seems to depend entirely on the reactions along the slab shorter edges.

Also , your model was giving warnings for isolated nodes tha couldn´t be detected and preventing to have different mesh sizes.

So I opened your file on repair mode and look what I found.

So I unsmeshed the panel and deleted ALL the model nodes. Model then worked fine.