Different results in MRz for panel cuts

[khattel]

Hello fellow ARSA-users!

I have come across a modelling issues, which statically are the same way of treating a problem for in-plane forces for walls (panels) and could use some explanation as to why this is.

I have created a test-model, which I will attach to enlighten the matter (I was not able to attach the model so here's a link instead: https://we.tl/t-77SOwuBLdb).

The problem is regarding the modelling of a UDL on the outside of the contour of a panel/wall. This could happen for at cantilivered part of a slab, where twisting of the slab is introduced. See the picture below:

In the picture above I have modelled the wall for four different types, where I would like to obtain the reduced results (here the MRz and TRy) to extract data for a plastic calculation of stabilizing walls in e.g. a spreadsheet.

The loading and in-plane geometry is the same from a simple equilibrium standpoint, however the 4 test-cases creates different results as shown down below:

Performing a hand caluclation of the external forces and what the reduced reaction forces should be I get:
NRx = 8 m * 10 kN/m = 80 kN

MRz = 10 kN/m * ((2m*3.5m)-(1m*3m)) = 40 kNm.

For panel 2: There is a value in the in-plane shear force TRy, even though there is no external horizontal forces applied? The moment calculation results in something different than the hand calculation.

For panel 4: The calculation seems nummerically acceptable. However is assume the bar and mesh overlap creates a slight over-stiffnes?

For panel 5: The same results as for panel however there is no horizontal shear force TRy = 0.

For panel 11: The same results as for panel 5.

Disregarding out of plane stress the external forces should equal to the reaction for the system to be in equilibrium (MRz, NRx and TRy). Why does the different cases create these results?

The test is done for a small model, however when looking at this matter in a global FE-model with partially cantilevered slabs, what is the correct approach to model this and how do I get the correct results to extract to e.g. a spreadsheet for a plastic caluclation of the wall-elements?

Hope you're able to help. Thank you in advance

Best Regards khattel

[Rafacascudo · Accepted solution]

Zip the model before attaching

Slab panel 12 in cantilever is unstable. Top supports of panel 11 should have RX blocked.

Instead of Deck slab(one way) use Shell calculation model.

Cantilever beans 1 and 3 on wall2 are also unstable as RSA Finite element type does not support torsion perpendicular to the panel plane on its nodes.

Workaround is extend the bar to the 2nd inner node or create a full rigidlink as the picture below

Test model attached

[khattel]

Hello Rafacascudo

One again thank you for your swift and detailed reply! Always very helpful. When I ran the model I only had an instability type 3, where I checked magnitude og rotational displacements and deemed the model stable based on this?

I can see the error with the panel calculation model should be changed to shell, however I don't understand the torsional moment of the beam not able to be fixed by the wall? A possible torsional moment in the cantilevered beam should just result in out of plane bening moment of the panel, which is allowed?

BR khattel

[Rafacascudo]

No , maybe I was not clear enough.

It is not the torsion moment of the beam that causes the instability.

It is the beam cantilever main moment that cannot be resisted by just one node of the wall because in the panel FE type used by RSA ,the nodes are free to rotate around local Z panel axis.

  That's why you need 2 nodes (binary) to resist this torsion

[khattel · Accepted solution]

Ok that makes sense. The reason for this little experiment is due to a problem with a much larger model, where I have TRy (in plane shear forces) at cuts near the base of wall supports for a case (case 1) where only vertical load is present. I have attached the model, where I do a check on the forces for reduced results on panel 478, where the TRy is not 0 even though not horizontal loads are applied as case 1 is simply the selfweight of the structure.

Note: I have modelled the walls with a small spacing horizontally of 50 mm to make sure they aren't transferring shear forces inbetween like a closed profile behavior. I know I probably could have used releases for this, but I wanted to be 100% sure.

From the instability warnings i only get the type 3 warning, but I can tell by the check values that the model is not entirely stable somehow - see attached. I watched Artur's webinar on how to deal with the different instability types, but still find it difficult to figure out why the program is acting up for larger models. I feel like I always end up trying to run down each element and their releases in a huge mazerun.

In much desperation - khattel

[Rafacascudo · Accepted solution]

You need to resolve the instability problems you have in the model . any Results are not reliable while you still have displacements like those shown below, even for self weight

The same 478 panel alone in a model with  a linear load applied to a finer mesh gives the expected results

You also have inchoerent mesh error in your original model which must be corrected

Panel 478 model attached

[khattel]

Thank you kindly Rafascudo 🙂

I'm curios to how you took out panel 478 from the model. Did you simply save it as a substructure and changed the mesh or did you create a whole new model?

[Rafacascudo]

Selected panel 478 and then with CTRL pressed ,I selected the whole structure.

Then , everything was selected except panel 478.

Hit " delete" and only panel 478 remained.

Then I deleted all load cases but the 1st and applied the linear load and deleted the SW entry on the load table.

Redone the mesh with a 5cm size and that's it