Coupled RC Core Wall with MVLEM-3D

[kolozvari]

Thank you for your reply. This worked for Y direction, but not for X direction where the analysis still fails...Any other suggestions? Can I implement somehow the resolution strategy?

[STKO Team]

Before I tried to "tweak" the solution strategy to make it converge. But it's not always possible.
In cases you cannot make it converge, you need to understand what is the strongest source of nonlinearity that gives the issues.

Looking at the curve for the Y direction, you can see that the overall behavior is almost bilinear with strain-hardening, so one may think that the problem is stable enough and should not have huge convergence issues.

But if you look carefully at what's going on between U=0.2 and U=0.5, you can see lots of small sharp load drops. They are due to the very brittle behavior you assigned to the tensile behavior of your concrete. When a material point is so brittle (the area under the descending part is comparable with the initial elastic energy ft*ft/E/2), the structural behavior will be even more brittle, because there are other parts in your model that remain elastic and store extra elastic energy. As a result, your global model may show small snap-backs that cannot be captured by the displacement control.

If you are lucky, you will see a vertical drop, and the analysis will continue, but this is not always true. When you have such a brittle concrete tensile response, you can simply remove it (i.e. set ft = 0).
You will see that it will converge must faster, and the overall behavior will be the same. You will simply not see anymore those small load drops

[kolozvari]

Thank you for your reply. I don't think the small drops are due to tension of concrete because these little drops are happening around yield. Concrete cracks way before the structure yields.

Is there a way to use some custom functions to add resolution strategy to switch to initial tangent when not converged? I think I saw this somewhere but can't remember where...

Thanks

[STKO Team]

The small drops are due to concrete. It's true that it starts cracking before, but the cracking that is causing issues to happen also during yielding.
This is what I have pushing in X with ft = 0 in both concretes. And it converges much faster

concrete_no_tension.png (54.58 KiB) Viewed 1191 times

Is there a way to use some custom functions to add resolution strategy to switch to initial tangent when not converged? I think I saw this somewhere but can't remember where...

Yes. Analysis Step -> misc -> customCommand

[kolozvari]

I see. Great, thanks.

I am aware of the custom commands, but do you have an example for this (I think I have seen a video, but cant remember which one)? Where do I add this in the analysis que? I know how to write these in OpenSees, but I am not sure what part of OpenSees code needs to be included in STKO.

Thank you!

[kolozvari]

FYI, there is some kind of a bug with MVLEM-3D elements. When I try to edit the element properties I receive the following error. Please advise.

Thanks

[STKO Team]

It was a small bug we already fixed for the next version.
Here are the updated files:

MVLEM_SFI_MVLEM_3D.zip

(5.14 KiB) Downloaded 11 times

Replace the original ones in: C:\Program Files\STKO\external_solvers\opensees\element_properties\beam_column_elements

Don't worry because that small bug was only affecting the automatic resizing of the vectors when you change the number of fibers. It's just automation to make sure all the vectors have the same size. It does not affect the analysis

[STKO Team]

I am aware of the custom commands, but do you have an example for this (I think I have seen a video, but cant remember which one)? Where do I add this in the analysis que? I know how to write these in OpenSees, but I am not sure what part of OpenSees code needs to be included in STKO.

Regarding this,
there are 2 ways of doing it.

• Just replace your last AnalysisComand Step with a customCommand step where you will insert your custom logic. In this way however you may loose the control over the monitor function, adaptive time step and other automation controlled by STKO
• Alternatively, STKO allows you to add custom Tcl procedures that can be called either before or after the "analyze 1" command inside the loop set by STKO, still using the analysis set by STKO. This feature has not be described yet in a webinar, so if you decide to go with this, we can explain to you how to do that

[kolozvari]

Hello again!

I am running a pushover analysis for the 26-story reinforced-concrete coupled wall building we discussed so far. The model has a slab modeled explicitly at all levels. I have noticed that when I use the ASDshellQ4, the out-of-plane stiffness of the slab affects greatly the building's overall stiffness and strength, and the results seem unreasonable (similar results with ShellMITC4). However, when I use ShellDKGQ, the results make more sense. I compare results from the same model made in Perform 3D (see results comparison below). I use ElasticShellMembraneSection with Ep_mod = 0.10 to account for concrete cracking.

What is the difference in model formulations of ASDshellQ4 versus ShellDKGQ elements that causes such a big difference in the building behavior? What element would you recommend for modeling slabs in reinforced concrete buildings? Based on my results, I would prefer to use ShellDKGQ, otherwise I have to use Ep_mod = 0.001 to get the building capacity correctly, which creates very large out-of-plane deformations under gravity load.

Please advise. Thank you.

[kolozvari]

Hello,

I just want to follow up on my previous message. Can you please provide a response?

Also, I tried using ShellDKGQ dynamic, but the model doesnt converge even at all. Ay ideas why is the model converging for static but not for dynamic analysis?

Thanks.