Self-weight of panels not included in notional loads

[PatrickEC]

The self-weight of the panels in my model is not converted to notional loads.

Attached is a sample model.

Load case "1 DL1" (self-weight of the whole structure): total FZ reaction: 42146 kN

Load case "12 Panels" (self-weight of the panels only): total FZ reaction: 15676 kN

Load case "9 Notional loads X+", generated from load case 1: total FX reaction: -264 kN (should be 0.01*42146 =421 )

Load case "10 Notional loads X-", generated from load case 12: total FX reaction: 0 kN

[tony.ridley]

Its because you have used panels with no finite elements and simplified method of transferring loads

Change your panel calculation model to finite element type, and load transfer to analytical and it will work. 

Tony

[Artur.Kosakowski · Accepted solution]

Patrick,

Please define self-weight in a horizontal direction (e.g. X-) manually in a separate load case and combine it with the corresponding notional load. I will create a development request for having this done automatically inside the notional load case.

If you find your post answered press the Accept as Solution button please. This will help other users to find solutions much faster. Thank you.

[PatrickEC]

Tony: Obviously I have a reason why I use simplfied load distribution and not shell elements.

Artur: Thanks for the workaround and I am glad to see this added as a development request.

[tony.ridley]

Patrick,

yeah yeah yeah OK but you didn't elaborate on your "obvious reasons".

Still haven't. 

So, in reply to your original post I gave my opinion on why your model is not working, and how I would fix it.   No point being cagey, it only a forum. 

Tony

[PatrickEC]

Well, using shell elements would add additional stiffness to the model. Or maybe you know how to avoid it? then I am taking back my "cageyness".

[Rafal.Gaweda]

Panel thickness = 1mm or less ?

[CristianLANCULESCU1367]

The 

---

@PatrickEC wrote:
Well, using shell elements would add additional stiffness to the model. Or maybe you know how to avoid it? then I am taking back my "cageyness".

---

Make a material (let's say concrete "C16/20_Eb1") with Young's modulus  (Eb) very small (0.1MPa). This way, the shell is very flexible, but you still have the self wight of a, let's say, 15cm concrete floor.

[CristianLANCULESCU1367]

---

@rg_Adsk wrote:

Panel thickness = 1mm or less ?

---

The problem with your solution is that the self weight must be calculated manually.

[PatrickEC]

Making shell element extremely flexible will cause huge displacements of these elements. Even leaving out the problems with displaying and interpreting results, I am sure this would cause problems in modal/seismic analysis. Or am I missing something?

[tony.ridley]

Well, using shell elements would add additional stiffness to the model.

I'm not sure I understand this.  Your panels were set to full stiffening (rigid body) meaning they are blocked in all directions.  Surely this is adding massive amount of stiffness to the model that is not really there?  The other problem with the way your model is set up is that pattern loading or a concentrated heavy load (in a small zone) is not properly distributed or dealt with.  i.e. if you place a small area of say 5kPa pn one side of your floor slab, the entire floor will rotate but stay plane in it's own right.  So the buildings global deformation is not correct. 

The beams under the heavy load will not deflect at all.  So actually I'm not sure using the full stiffening method is appropriate here. 

You could model as a shell with flexible stiffening, still get the diaphragm action laterally but account for the real stiffness in the vertical direction.  In some cases (especially like yours with very regular geometry) Robot is quicker using FE shells than with trapezoidal distribution method anyway. 

Additionally your modal / seismic run will also be more accurate when using a shell, as the applied loads are converted to masses at each node on the shell.  You will have more control over the size of the mesh and therefore more control over the accuracy of the results (especially if you use the "consistent" mass matrix method for the modal run. 

Tony 

[PatrickEC]

Tony:

The attached model is just a book example to show that the notional loads for panels does not work. In my actual model I do not use diaphragms.

I appreciate your comments though. You certainly make interesting points. I just do not see how you can make shell elements work in the situations where you have floor supported on beams but working independently. E.g. a concrete slab (non-composite - no studs) supported on steel beams. In this case you do not want the slab to contribute its stiffness to the beams, because these elements are not well connected.

[PatrickEC]

Artur:

Similarly to the conversion to notional loads, panel self-weight is not converted to mass when defined in "Analysis Type -> Load to Mass Conversion" - I just checed that. Looks like it also requires a fix. Any good workaround for this one?

[Artur.Kosakowski]

Could you send me the example file please? I can't see such problem.

[PatrickEC]

Artur:

Like before I am using panels with no finite elements and simplified method of transferring loads. You used a panel with finite elements, change it to what I use and you will see the same behaviour.

[Artur.Kosakowski]

Have you checked with SP1 installed?

http://forums.autodesk.com/t5/Autodesk-Robot-Structural/Curtain-walls-and-modal-analysis/td-p/332946... (message 😎

[PatrickEC]

Apparently it also works when a stiffening diaphragm is turned on, but without diaphragms it does not work. See the attached file and screen capture.

http://screencast.com/t/AyJXdz8p6

[PatrickEC]

---

@Artur.Kosakowski wrote:

Have you checked with SP1 installed?

---

I use the 2012 version with Service Pack 5.

[Artur.Kosakowski]

Good reason to upgrade

[PatrickEC]

Good reason to release a hotfix/patch/update.