soil-structure interaction continuum model

[begumemte1991]

Hello STKO team,

Please make a video tutorial on soil-pile interaction. I am facing convergence issues with whatever SSI model I build. Till now I have tried a number of models but none of them worked. Even I tried to build a model which was posted here (pile.zip) as it is but failed to get convergence. I will be attaching the .scd file herewith.

Also, a continuum model that was corrected by the team is giving strange results (figure attached below). I think there might be some problems with the interface of soil and the pile.

Thank You

[STKO Team]

Dear user,

I will have a look at the problem you posted. Convergence issues are expected with contact elements (expecially those using Lagrange multipliers).

The results look strange to you probably because you see those "lines" around the pile-soil interface moving abnormally. This is however perfectly fine:

The contact element you are using is based on the Lagrange Multipliers to enforce the non-penetration condition. They have a total of 4 nodes (1 solid (3 DOFs) + 2 beam (6 DOFs) + 1 fictitious nodes with 3 Lagrangian DOFs). In the Lagrangian node, the unknows (which are displacement in the solid) are the Lagrange multipliers, which in a contact problem are exacltly equal to the contact force. However STKO does not know this (because OpenSees saves those values in the Displacement vector)... I know it is quite confusing if you don't know much about the Lagrange Mutlipliers method.

To avoid their visualization you have to:

• create a Selection Set with all your model except the interaction
• in the analysis step, create a region with that selection set
• in the recorder, in the region option, select that region analysis step (this will tell the recorder to record only that region of the model)

Anyway I will have a look at your file to see what's going on

[begumemte1991]

Thank you for your revert. I know you all have a very busy schedule. If you are done with checking my model please let me know.

[yuweihwang]

Hello STKO,

May I ask you if now the dof of soil element is 4. Can we still use BeamContact3D? or is there any other choice?

Many thanks for your help.

[STKO Team]

The constrained (soil) node must have 3 DOFs.
So if your soil model is a U-P (4 DOFs) your best choice is to:

• create an auxiliary face coinciding with the face of the solid domain that will be in contact with the beam. You can do it by using the Extract command.
• do not assign any physical or element property to the auxiliary face. In this way its nodes will have 3 DOFs
• make sure the auxiliary face has the same mesh settings and mesh seed of the original face in the solid domain. In this way, their meshes will be compatible.
• create an EqualDOF 3D-U attaching all translational DOFs (Ux, Uy and Uz), between the original and the auxiliary face.
• now you can use the BeamContact3D between the beam and the auxiliary face

If you have problems following this procedure let us know.

[yuweihwang]

Hello STKO,

Many thanks for your quick response.
May I ask you another question? you mentioned that "do not assign any physical or element property to the auxiliary face". However, if I didn't assign any physical or element property of the auxiliary face, won't it show the convergence issues when I run the analysis?

Many thanks for your kind suggestions

[STKO Team]

When you do not assign any element property to the auxiliary face, STKO will not generate any OpenSees Element in the TCL files, but it will only write the nodes of the auxiliary face.

And this is exactly what you want: these "floating" nodes will have 3 DOFs (because they don't have a UP element) so that you can use them with the Contact element.
Of course, you need to "link" this face (and so its auxiliary nodes) with the real (4-DOFs soil nodes) with Equal DOFs (x,y,z). Of course, if you don't do this last step, your model will be singular.

[yuweihwang]

Hello STKO,
I have tried to follow your suggestions, but it doesn't work.
I have attached my file. Can you give me some suggestions?

Many thank for your help

[STKO Team]

Here is the fixed file:

pile_YW.zip

(241.1 KiB) Downloaded 239 times

You correctly understood the concept of the auxiliary surface. In fact, the model was correct. There was only one tiny mistake:
Make sure the auxiliary surface has the (possibly) same mesh as the source surface. In your case, I had to impose an EdgeSeed to the bottom lines (curves) to impose a uniform division by size = 0.4. In this way, you make sure the overlapping nodes are actually coinciding.

The error you had was due to some mistakes in the analysis:
You were using Newmark with gamma = 0 and beta = 0.51. Note that if you use 0 for gamma and/or beta, the Newmark integrator in opensees will issue an error.

Then there were some other convergence issues:
Contact algorithms need tiny increments to converge. However, you were imposing self-weight in the element property. In that case, the weight is applied instantaneously (no matter how many steps you use in the first stage).
So I applied the self-weight using the VolumeForce (STKO automation). using Fz = -rho*g.

[yuweihwang]

Hello STKO,

Many thanks for your help. Now I understand how it works.
May I ask another question. If I use body force to create the gravity weight of soil. Can it simulate the pore water pressure?
Because even I only fix the pore pressure at the soil surface, I can not obtain any pore pressure below the ground surface.
May I ask for your suggestions?

Many thanks for your help