New update in ASDEmbeddedElement for UP element

[BuuNguyen]

Dear STKO,

I am using ASDEmbeddedelement in Soil-structure Interaction.
My Model run according following sequence:
1/ Gravity application (elastic) for soil domain --> worked well
Update Stage 1
2/ Gravity application (plastic) for soil domain --> worked well
Initial state analysis OFF
Set parameter xPerm, yPerm, zPerm for each soil layer
3/ Structure application and its gravity --> didnot converge

I have 2 questions:
1/ For updating permeability parameters, there are 2 options: Set parameters (as in Fig1)and Parameter+Update parameters (as in Fig2+Fig3). Which one should I use to update permeability coefficient
2/ As you said in the Webinar, KP of ASDEmbeddedelement should be close to permeability coefficient, I used exactly KP=permeability coefficient, it did not converge. So, now I am trying to start 10^-3 as you said, and still waiting for the result. But the problem is in SSPBrickUP element, it does not provide the pressure in the result. Should I change to BrickUP element and try to get appropriate value of KP, then change back to SSPBrickUP element for further analysis?

Thank for your kind support,
Buu

[STKO Team]

You should also download the new version of the OpenSees-Solvers from prof. Tarque's website for this new feature

[BuuNguyen]

Yeah, I am looking forward to your example and next webinar about this topic.

Can you have a look at my model in attached file? I modified your example in webinar 25 with BrickUP elements (I havent tried SSPBrickUP element yet)
- It worked till Step 4 (Gravity analysis of structure). I checked that water pressure was continuous in the transition zone in step 4, but Im not sure It is correct or not
- But in step 5 (Relaxation analysis), it became non-convergence.

Thank for your support,
Buu

[STKO Team]

Ok, I had a look at your model and I think I understood why it was giving problems in stage 5:
The duration of the analysis was too high (1.0e6 seconds) for the constraint on the pressure DOF using the Penalty approach.

Last time I told you to use a large time value (1.0e6 seconds) for quasi-static analyses because you don't want spurious accelerations. It was fine until you used the ASDEmbeddedNode element to constraint the Pressure DOF (it internally uses the Penalty approach).

Here is the problem: In OpenSees the pore pressure is stored as the "velocity" associated with the 4th DOF of a node. But to constrain it, we actually constraint its integral wrt time (i.e. the 4th DOF in the "displacement" field). If the timeStep is too large, the 4th DOF of the displacement (integral of pressure wrt time) is extremely high, leading to numerical issues.

A solution that is a good compromise between a) using a large time for quasi-static analyses and b) not so large as to avoid numerical issues, is to use duration = 1000 seconds, and the KP factor approximately equal to the permeability.

It seems to give small accelerations (1.0e-8) and good convergence, and the Pressure DOF is properly constrained as you can see in the image: it is continuous.

PWP.png (493.7 KiB) Viewed 4404 times

I also made sure all your analyses are consistent. Some of them used the Penalty, some others the Transformation. And a few other differences.
Check the modified file:

sfsi_try2.zip

(1.77 MiB) Downloaded 291 times

[BuuNguyen]

Yeah, It worked in this example.

But it seems difficult to find out an appropriate KP. In my model, I tried many times to change KP from 10^-3 to 10^-12, and load duration, but not converge.

Can I you have a look at my model? can I send you via email?

Thank for your kind support.
Buu

[STKO Team]

You can post it here and we will have a look as soon as possible.
In the near future we will work on implementing an easier way to constrain the pressure DOF.

[BuuNguyen]

Please have a look at my model.

thank you for your kind support.
Buu

[STKO Team]

The problem does not seem to be the KP in the embedded element. In fact, the first stage (elastic) runs well.
I think there is a problem with your nonlinear material/contact parameters. You have to check them carefully, we cannot help you with that. Testing nonlinear material properties can be a long operation.

Also check the upper soil layer under the structure. There is no condition to "Tie" it at its vertical sides to the surrounding elements

[BuuNguyen]

Yeah, I checked and fixed the tie of upper part.

At the beginning, permeability = 1 to generate pressure in the soil domain. Then it will be update to real values of permeability for structure loading and dynamic analysis as following procedure:
Stage 0 (take into account elastic)
1/ Gravity Elastic (before excavation)
2/ Gravity Elastic (excavation)
Update Stage 1 (take into account plastic)
3/ Gravity plastic
Update permeability to real values
4/ Structure loading
5/ Relaxity
6/ Dynamics
In the model I sent you, I deleted some stages (1 and 3) to reduce calculation time to reach stage 4 (structure loading) soon.

But if I did not update parameter of permeability and did not use pressure constraint. It worked well for the whole analysis (6 stages). Thats why I am thinking that its due to the pressure constraint problem.
Let me send you the model without updating permeability.

I attached 2 files of models (update and without_update). hope you can spend some time to look at it.

Thank for your kind support.
Buu

[STKO Team]

Dear user we tried to identify the problem, but your model is too large for this kind of assistance. It seems that with KP = 1.0e-10 it runs the first analysis, but when you add the contact elements something goes wrong, and it takes a lot of time to converge.

You should try to add one component at a time, to identify the problem. Otherwise, you can redesign your soil to make a compatible mesh avoiding the use of the embedded constrain when you are using UP elements. Even though I think the issue is in the contact probably