soil structure interaction

[mmc]

Hello! I'm new to STKO. I have read the manual and watch some of the videotutorials. They are very good and instructive. Thanks for doing and uploading them.

My problem has do with soil foundation structure interaction. I have modelled a simple 3D one-bay frame following one of the tutorials. Then, I added the foundation and soil as solids following another tutorial.

I have fixed the bottom surfaces of the soil in the Z direction. I also fixed the sides soil surfaces in the X and Y direction, respectively. However, an error apperared saying that those nodes were already fixed in the corresponding direction. I deleted this condition and the analysis started but I think they should be fixed, as seen in the SSI tutorial.

I added EqDOF in the soil and foundation solids in each direction. I also added them to consider the interaction between the soil and the foundation. I used the beamsolidcoupling constrain to link the base node of the columns and the top nodes of the foundations. I found this command in the STKO manual. Maybe I did not apply it correctly.

However, the analysis does not run and I can not find the error. Maybe, it is related to the DOFs restrictions or the material properties. I have modelled the struture in N/mm. Could you help me trying to guess what can be wrong?

Thanks.

PS: I want to properly study the interaction between the soil and the foundation. I thought that I can duplicate the surfaces between the soil and the foundations and apply the 3dcontactbeammaterial to them. But I am not sure how to use it...any thoughts?

[mmc]

Hi again! I have redone the model using SI units and I have not deleted the fixities of the soil slides.
Looking forward to hearing from the community.
Regards!

[akanon]

Dear mmc,

I hope you are doing well. I checked the second model that you have posted.

1) The warning messages at the beginning of the analysis about sp constraints not being applied mean that there are sp constraints (fixes) that have already been applied to those nodes at a previous stage. So, OpenSees is just saying that it can't apply fixes to those nodes again. I suggest checking your analysis flow to see if some condition is applied twice. However, this warning, by itself, should not cause any problem during the analysis.

2) I see that you have preferred to use the Transformation constraint handler. Unfortunately, this method might not work properly with models where more than one kinematic constraint (equal dof, rigid link etc.) has been assigned to a node or a set of nodes at the same time. For such problems, Penalty constraint handler (with a reasonable amount of penalty, being 1e+8 times higher than the biggest value in the stiffness matrix) would be more suitable.

3) Moreover, the Penalty constraint handler should be used with a convergence test that checks the displacement vector since the unbalance component contains the high forces resulting due to the applied penalty.

4) Finally, I would suggest using a sparse solver such as UmfPack instead of BandGeneral for computational performance.

In fact, just by applying these changes, I was able to run the analysis and have convergence. I hope this is helpful. In case you have further questions please do not hesitate to post.

Have a nice day,

[akanon]

Dear mmc,

Regarding the BeamContact3D element, this element is creating contact between a beam element (having 6 dofs) and a solid element (having 3 dofs) which is not compatible with this case if I understand your goal correctly. For connecting solids to solids (solid foundation to solid soil), ZeroLengthContact3D would be enough. It is possible to implement direct and frictional contact by using these zero-length elements. I hope this would be a solution for you.

Have a nice day,

[mmc]

Thank you very much for your answer akanon!!!
I think that I'm not an expert on defining the correct values to perform the analyses. Do you happen to know any manual in which this is explained? It would be very useful for me to understand how finite element analyses work. I put a max dU of 0.2 m but the curve does not reach this value. It stops earlier. (figure attached).

I have changed the values and performed the analyses. However, the pushover curve at the control node is too "elastic"... I think that the model in not behaving as nonlinear. I changed the material concrete02 to concrete04. And, I have found this post about the unit system. viewtopic.php?f=36&t=331. The support team say that "the only place in STKO where you find a unit system is the Beam Section Editor. However that unit is only necessary because we internally use a database written in [mm]." I have modelled the sections in m, and maybe I should have done it in mm... what do you think about that? I think that the problem is related to the material properties because the stress-strain response of concrete curve has a completely different slope...

I have attached the model in which I am now working following your suggestions.

Regards.

[akanon]

Dear mmc,

I was able to download and check your model today. First of all, in terms of beam section editor, it seems like your model is correct. If you have doubts about the unit of section sizes you can check this by switching to the Property tab where you can see the extruded beam sections. In case the sections are wrong, they will appear as extremely large or small here. Returning back to your questions,

1) I think your model is working. It is just that I have realized you have been using the optional (adaptive) tick box of the displacement control during the pushover analysis. OpenSees has this option, yes, but in fact, it's not that straightforward to use it. So, I have never used it, to be honest. I usually write a custom .tcl script to have an adaptive time step during the analysis. In your case, your model and material properties are okay but because of the alterations that the adaptive option does in the du, pushover is not reaching to 0.2 roof displacement in 200 steps. If you turn this off you can see that the pushover curve is fine. (Figure I)

Figure I

mmc pushover.png (34.03 KiB) Viewed 9282 times

2) Second, could you please elaborate more on the problem with the concrete model that you are expecting to occur. I could not understand exactly where you are coming from.


3) If you would like to learn more about FE analysis, there are really good books and online courses, some of which you might enjoy checking:

The Finite Element Method: Linear Static and Dynamic Finite Element Analysis
(Dover Civil and Mechanical Engineering) 1st Edition
by Thomas J. R. Hughes

The Finite Element Method for Problems in Physics
by University of Michigan (in Coursera)

Introduction to Finite Element Methods (ASEN 5007)
Department of Aerospace Engineering Sciences
University of Colorado at Boulder


I hope this is helpful.


Have a nice day,

[mmc]

Hi again! I have been busy these past few weeks. I have continued working on this model.
I have performed the pushover analysis and I finally got some results...However, I compared the pushover curves obtained for the model without (columns fixed) and with the soil and they are practically the same. However, it seems that the soil is moving (deformed shape captures), but not enough. I tried to increase the load factor but the results were the same.

Then, I tried to apply a SimpleContact3D element and corresponding ContactMaterial3D in the interaction between the soil and the foundation. And a different error appeared...maybe related to the mesh options? I have read that "The SimpleContact3D element is a three-dimensional node-to-surface contact " and that's why I can't use it for this type of interaction. Should I move to using Zerolength elements and the STKO SSI tutorial????

Regards.

[mmc]

Hi again! I have decided to model the foundings as they were a pile as in viewtopic.php?f=29&t=1623... I am just trying different solutions.. But again, a new error appears... It definetly has to do with the meshing but neither can I delete surfaces to redo them nor change the edge sides...

[STKO Team]

dear User,
model didn't work beacuse geometries were not solid but faces. With SSpbrick is necessary to use solid geometry and structured hexaedric mesh. If faces are merged they still are faces, but you can extrude them in order to obtain solid geometry.
With dispbeamcolumn is necessary to use the special purpose beamSectionProperty as physical property, as shown in the model attached.
BeamContact3D requires radius in input properties, not diameter, and their orientation has to be the same of the beam connected. If is not, it's necessary to apply local axes to interaction.
We hope to be helpfull. Please, let us know if you need any help.
Have a nice day.

[mmc]

Hello!
Thanks for your comments.
I have redone the model creating solid elements and hexaedrid mesh. I have added the beamSectionProperty as physical property in the edges that represent the foundation. Then, changed the radius of the BeamContact3D and the local axes orientation. Still, an error appears which I think it is related to the mesh properties.
Could you please indicate me what it is wrong with it, again?.
Thanks.
Regards.