Some nodes are not being included in the stiffening diaphragm of panels

[s.follador]

Hello,

I have a problem with nodes that should have a rigid link property because they belong to a panel with "partial stiffness" ticked on, but they don't seem to be included in the diaphragm. Not every node is affected, some have the property as expected. This results in odd modal shape.

Let's be more detailed:

1) I used panel objects to model horizontal floors (NO SHELL). These objects redistribuite forces according to the simplified method and provide the "stiffening diaphragm" as shown in the picture below:

2) I then designed the panels. I usually divide the floor in multiples panels, but the yellow lines show that just one diaphragm is created. NOTE: I don't use the grid when I draw bar elements or panels, I prepare the construction lines so that the snap points are not ambiguous.

3) Now the hilighted node (1) should have a rigid link property (2) because it belongs to the panel, same as the other node (3)(4). 

4) this cause the following behaviour in the modal analysis, which is not accetable:

The example node is not the only one, some other are scattered along the structure, without an apparent pattern. If I zoom deeply in the area around the node, I can see a gap, but I can't measure that. In other nodes where I can see the same gap, they have the rigid link property. I suppose it is just a graphical glitch.

What I already tried, without success:

• delete and redraw the panel, being careful to snap at the nodes
• untick the stiffening diaphragm in panel calculation method and manually assign the diaphragm link in the rigid link tab, i.e. selecting a master node and slaves node (this resulted in really wrong calculations: the modal mass was almost half of what it should be, like if floors weren't distributing the applied loads)

A fix would be to model the panel as a shell, so that eventual errors would be fixed by the others node of the finite elements, but I really don't want to do it because the computational time they require. I have a working model with shell elements, but it is too cumbersome to use, I have to wait several minutes in order to read the data and display results, even for just few combinations.

[t.sautierr]

Post your model quickly (I'am off in 2 hours :))

Have you same local axis tsystem for all the pannels?

[s.follador]

Nevermind, I found an (almost) satisfying solution.

Upon further investigation, the gap was indeed the culprit (it measured 0.37 mm). It's how it got created that is interesting.

In order to fix the gap, I bumped up the precision and of dimension display and changed units to mm. This way I was able to compare exactly nodes that should be on the same lavel (same Z coordinate) and it showed that some nodes had a sligthly different (0.37mm) value for the Z coordinate from the chosen floor height.
So I selected all the node belonging to the same floor and manually changed the Z coordinate to its proper value. I checked the rougue node to be sure, it got moved in the right position and the gap was gone. 

Then I generated the mode.

The gap has returned!

After the generation of the model, the node got slightly moved. I figured out it was because of the vertical cladding you can see in some pictures. I genuinely made a mistake drawing the cladding and indeed it was a little bit short of the full height of the floor. I deleted all the incorrect cladding panes and redrawn with more care. This fixed the issue of changing heights of the nodes: after generating the model, the nodes didn't move along the Z axis and indeed they appear to be part of the floor and gained the diaphragm property.

Solved that problem, another one showed up. This time some internal nodes didn't appear to belong to the diaphragm link, but their Z coordinates were fine.

I ended up introducing a manual diaphragm link between nodes of the same level AND keeping the diaphgram stiffening of the floor panels in panel calculation method. This way the modal shape I got, were the same of the model using shell elements for the floor panels. 

It's a workaround, but it works.

[Artur.Kosakowski]

Try to run the detailed correction to structural axes for your original model and check if this works for you. 

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.

[s.follador]

It's the first thing I tried, even before posting here. It didn't solved it.

I had some good results by going in the rigid link table tab and deleting all the rigid link's definitions.

Then I regenerated the model and most of the nodes were included correctly in the diaphragm link. I found just 4 nodes that should be included in the diaphragm link and didn't so I added them manually to definition.

The rigid link property showed in the manager tab, but after running calculation it disappeared again.

[Artur.Kosakowski]

Could you attach the model?

[s.follador]

Here the model. Some notes:

1. Just to be sure I opened my model with the latest version of robot I use for testing purpose (student version 2016, for work I instead use 2015) and some of the problems I solved in the work version came back;
2. The most noticeable is at floor level 2, where I get 2 different diaphragm in the rigid link table's tab because all nodes beloging to an axis got neither of diphragm link property. This is quite evident in the 4th modal shape (in the picture below):
3. I am unable to fix this in the 2016 version of robot, as I did in the work version;
4. you'll find compatible nodes and offset in the model (in the columns), but I had no problems with that. Indeed some of the nodes that are not included in the diaphragm have no compatibility or belong to bars with offset;
5. Some bars are inactive, but that's fine too;

The version attached is the one saved with Robot 2016, which has the most problems, but the model is the same I am working on the 2015 version.

[Artur.Kosakowski · Accepted solution]

It's the first thing I tried, even before posting here. It didn't solved it.

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.

[s.follador]

Thanks! I wasn't aware of such tool, I assumed you were referring to "quick" correction tool you can find in the toolbar