soil structure interaction

[mmc]

Thanks for your help, STKO Team.
The dynamic analysis works fine now. However, there are sometimes that a bug appears in the analysis... like super deformed beams and weird stuff. To solve it, I close the app, open it and run the analysis again (I also delete the outputs in the folder)... By doing this, the model works perfectly. Maybe it is something related to the capacity of my computer. I also added the updatestagematerial and the model worked perfectly. I added it by using a custom comand. Thing is that I had to check in the imput file the tag of the material... It would be nice to do that automatically. It did not take me a lot of time because I only have one material like that one. However, there might be some cases with a lot of different types of materials.

Also, linked to this, I just wanted to make a suggestion since you will release another version of the software. When I want to identify a node (ID node), I do it by using the query vertex option. However, there is not such an option for the elements like edges. I am used to write my records manually to obtain the forces, drifts and so and so to check them in MATLAB. It would be nice to have an option that can do that work... Because I have to do it manually (searching in the tcl file)...

Regarding the nonlinear static analysis, I havent been able to "make the soil move"... I have added a faceforce load in the base... But I dont think that I would work... I think I didnt understand your comment.

Regards.

[STKO Team]

By doing this, the model works perfectly. Maybe it is something related to the capacity of my computer.

When you have random strange results, probably yes, it is a problem with your machine.

Thing is that I had to check in the imput file the tag of the material... It would be nice to do that automatically. It did not take me a lot of time because I only have one material like that one. However, there might be some cases with a lot of different types of materials.

You are right, it would good to have it automatically, we put it on our to-do list.
However, in the meantime, it is not necessary to look into the generated TCL files to know the tag of the material: the tag will be equal to the Physical Property ID in the STKO's pre-processor Work Tree:

mat tag.png (42.19 KiB) Viewed 5855 times

When I want to identify a node (ID node), I do it by using the query vertex option. However, there is not such an option for the elements like edges.

If you have the latest STKO version (1.1.5) you have this feature. It's called LABEL:

labels.png (331.67 KiB) Viewed 5855 times

Regarding the nonlinear static analysis, I haven't been able to "make the soil move"... I have added a face force load in the base... But I don't think that I would work... I think I didn't understand your comment.

That was the fix to make the soil move in the dynamic analysis. I did not notice you were talking about a static pushover analysis.
The only static analysis you have in your model is the first gravity analysis. And with the file you sent me, I've been able to visualize displacements in the gravity step. What is exactly your problem?

[mmc]

Wow!!!! Thanks a lot. I did not know about the Label Option. It is going to save me a lot of time working...

That was the fix to make the soil move in the dynamic analysis. I did not notice you were talking about a static pushover analysis.
The only static analysis you have in your model is the first gravity analysis. And with the file you sent me, I've been able to visualize displacements in the gravity step. What is exactly your problem?

Yes, in the dynamic analysis the problem was solved by adding the faceforce...

I previously uploaded the file including the pushover analysis. If the soil is modelled as elastic, I can verify that it is moving (a little bit). However, when it is modelled as PressureIndependMultiYield Material I can not see that it is moving... Not even if it is not updated to be plastic...

[STKO Team]

I previously uploaded the file including the pushover analysis. If the soil is modelled as elastic, I can verify that it is moving (a little bit). However, when it is modelled as PressureIndependMultiYield Material I can not see that it is moving... Not even if it is not updated to be plastic...

Actually it is moving, (you can see it if you hide the structure and increase the deformation scale). However it looks like it is much stiffer than the structure.
Make sure you computed the material properties properly.

[mmc]

Hi again. I am still working on improving my models.
I am using the PressureIndependMultiYield Material to model the clays. I have found that if I use the suggested values for a soft and stiff clay from here https://opensees.berkeley.edu/wiki/inde ... d_Material... there is practically no difference in the response of the building. It is weird because the values are very different G 13,000-175,000 and B 65,000-750,000. Also, if I consider the values of my soil, the pushover curve is better with values more related to a softer soil (minvalues).
Do you think that this is okey or if the soil is behaving weird?

Also, I dont understand why the UpgradematerialStage command is working in the little model but not in the one with the whole structure. This appears in the command window.

WARNING: MaterialStageParameter::setDomain() - no effect with material tag 34
WARNING: MaterialStageParameter::setDomain() - no effect with material tag 35
WARNING: MaterialStageParameter::setDomain() - no effect with material tag 36
WARNING: MaterialStageParameter::setDomain() - no effect with material tag 37

And this other error regarding the zerolength element. I think that it is just a warning but nothing that affects the analysis...

WARNING ZeroLength::setDomain(): Element 14850 has L= 0.707107, which is greater than the tolerance

The pushover curve for the fixed model is very similar to the one of the model considering the soil. I am aware that the soil is very rigid.. but it is also very similar if I even consider the soil as soft...

It would also be very helpful if this kind of material could be tested.

[STKO Team]

there is practically no difference in the response of the building. It is weird because the values are very different G 13,000-175,000 and B 65,000-750,000. Also, if I consider the values of my soil, the pushover curve is better with values more related to a softer soil (minvalues).
Do you think that this is okey or if the soil is behaving weird?

In the large model, your soil mesh is extremely coarse (1 element subdivision for each solid geometry). Keep in mind that the more the element is distorted with respect to a perfect shape (in your case a perfect cube with all the sides equal), the worst its response is. Typically it gets stiffer.
In the mesh tab, use the EdgeSeed command (to assign a different seed to certain edges), select all the edges of the soil, and use a Unifrom By Size seed. Choose a length that is appropriate for your model. Then, repeat your tests and see what happens.

Also, I dont understand why the UpgradematerialStage command is working in the little model but not in the one with the whole structure. This appears in the command window.

There is nothing wrong with this warning in your specific example. In your small model, you used 4 partitions, and each partition has some elements from the soil domain, so the updateMaterialStage is successful in each partition.
In your large model, at least 1 partition does not have soil element (thus there is no element with material tag = 7). So in that partition the updateMaterialStage is doing nothing.

And this other error regarding the zerolength element. I think that it is just a warning but nothing that affects the analysis...

It does not affect the results. It is only warning you that a zero length element has a finite distance between the nodes. This warning is just telling you:
Hey, the zeroLength, does not couple displacement and rotation (like in Disp_1 = Disp_2 + distance * Rot_2), because it assumes a zero-length between the nodes. But it is fine, if you are trying to simulate a spring.
If instead you want to create a rigid body with displacement-rotation coupling, you cannot use an EDOF, you should use a rigidLink

[mmc]

Hi! Thanks for your answer.
I have divided the edges in 3 divisions so that the mesh is more discretized. I have run the analyses and if I consider the gravity analysis, the pushover analysis does not start... I have increased the number of iterations in the pushover analysis but I obtained no results.

If I only consider the pushover analysis, the model works correctly and the pushover curve is something like this... I am astonished with the decrease of the pushover curve. I am now worried about such bad behaviour

[Clarabella]

I think your results make sense. Does the soil become inelastic? The high flexibility is probably due to the soil inelastic behavior

[mmc]

I think your results make sense. Does the soil become inelastic? The high flexibility is probably due to the soil inelastic behavior

Yeah I know... it makes sense but... I could obtain the pushover curve only without the gravitational analysis... If I consider this analysis, the pushover analysis doesnt run...It looks like in the gravity analysis the structure is deforming a lot...

[Clarabella]

This is weird, I will look at it and get back to you