asdeasoft.net

Dear Massimo,

In the original model I was trying to solve the following issues:

1. I have a contact2D in the interface between the backfill soil and the abutment walls. And in the old model the top of the structure seemed to detach from the soil, producing a strange deformed shape. I would like to obtain a model that evaluates correctly the tensions sigma11 on the wall of the structure (in the gravity analysis they should start from 0 at the top and then linearly increase, like in a retaining wall). Of course I want to obtain correct results both for the static and the dynamic analyses.

2. (related to question 1) how can I model the static (gravity) analysis properly? Note that at the moment I have only the self-weight of the Foundation, the structure (edgeforceselfweight) and the backfill. But I am interested especially in the effect of the backfill on the structure wall because I want to catch properly the sigma11.

3. The interface for the piles is equalDOF, but it seems that the model is not able to catch adequately the experimental response that I would like to match. So is it correct to model the piles in this way or there is something that I am not modelling properly?
The difference could also be due to the difference between the piles section in 3D and 2D, I am aware of this, I just want to be 100% sure that there is no error in the numerical model and that equalDOF are adequate to model the soil-pile interaction (all is elastic).

4. Just as a reminder, you have mentioned that it seems that the modified model doesn't "see" the seismic input.

I am attaching the model that you have modified.

Please let me know if you need further details.

Thank you very much

Gabriele

I found the error: The reason why you cannot use the Automatic ASD Absorbing Boundary is that the right-end side of your geometry is not precise (see the X coordinate, there's an error of about 1 mm in tolerance):
Then we switched to the manual ASD Absorbing Boundary. But I forgot to tell you that, if we use the manual ASD Absorbing Boundary, we need to use the standard SetParameter to change its "stage" parameter from 0 (static) to 1 (absorbing).
Instead, the simpler Automatic Activation works only with automatic ASD Absorbing Boundary.
Therefore I first created a selection set with the absorbing faces: Then I used this:

Dear Massimo,

Thank you very much for this. However, there are still some issues with the contact2D:

1a) The problem with the contact2D elements is still present. That is, both in the static and in the dynamic analysis the soil and the abutment wall are detaching (not in contact) thus producing negative sigma11. I can understand this for the dynamic analysis but in the static one the sigma11 should be all with the same sign. How can we solve this?

1b) At the beginning of the analysis, there are several warnings like this one:

"BeamContact2D::setResponse(const char **argv, int argc, OPS_Stream &eleInfo): localAxes unknown request"
There are similar warnings for LocalForce, Deformation and so on.

1c) In the results, the Contact2D seem to have a strange response, with unrealistic high values (please see attached image). I guess this is related to point 1b).
2) The transient analysis does not converge in some steps. I think this could be a minor error and maybe this is related to point 1.

Thank you

Gabriele

1a) The problem with the contact2D elements is still present. That is, both in the static and in the dynamic analysis the soil and the abutment wall are detaching (not in contact) thus producing negative sigma11. I can understand this for the dynamic analysis but in the static one the sigma11 should be all with the same sign. How can we solve this?

They are in contact. Here you see a gap due to the radius of the beam: But if you show the beam extrusion you will see that they are in contact: Of course, due to the deformation of the columns, only some portions of the column are in contact, while others (bending inwards) are not.
Pay attention to the sign!, if they are in contact, you will see NEGATIVE stress (compression) in the soil

1b) At the beginning of the analysis, there are several warnings like this one:

Don't warry about them

1c) In the results, the Contact2D seems to have a strange response, with unrealistic high values (please see attached image). I guess this is related to point 1b).

What high values? Keep in mind that they are based on Lagrange multipliers.

2) The transient analysis does not converge in some steps. I think this could be a minor error and maybe this is related to point 1.

Actually, we developed a new contact element (ZeroLengthContactASDimplex) which is more robust. We gave a webinar on that. Try to have a look at the webinar to see if it fits your needs

Dear Massimo,

Thank you very much for your help.

Of course, due to the deformation of the columns, only some portions of the column are in contact, while others (bending inwards) are not. Pay attention to the sign!, if they are in contact, you will see NEGATIVE stress (compression) in the soil

I understand your comment, so I have one last question on this. The Contact2D elements are behaving correctly, so we are on the right track for the solution.

Given that, in the real case in the gravity analysis if I have sand behind a wall there is always contact between the two elements and the stress is negative along all the length of the wall. How can I replicate this behaviour?

If I understood correctly you are referring to the fact that the sand is not actually a continuum, but a granular matter, and like a fluid it will fill what in your model is a gap... But this is not easy to model with FEM