High Displacements with Plastic Material Stage

[vokumus]

Hi;

I am working on an SSSI model. If I change soil material stage elastic to plastic, I got some high displacement on the edges of the soil. If I use elastic soil material there is no problem.

When I plot the displacement data of the corner of the soil, displacements are continuously increasing. Which makes me think this is a damping problem.

I know this is (probably) not a STKO problem but I have no one to ask about it. And I have been trying the solve it for 3 days with no luck. Any help is appreciated.

My model is attached. Best regards.

2B-NL-2M-20SN-20SN.rar

(579.93 KiB) Downloaded 329 times

2020-07-21 08_48_47-Window.jpg (209.89 KiB) Viewed 6759 times

[Clarabella]

Dear Vukkumus,

I have looked at your model. We think that maybe there is an error in the calibration of the absorbing boundaries (see the image attached with the strains). The deformation of the soil is not correct and the free field columns do not deform in pure shear, there is some bending which causes tensile stresses/strains and therefore the abnormal plastic behavior.

We also modified your model to run with the MP. We used 5 processors, you can optimized the number of processors. Bear in mind that even with one processor the MP is much faster than the sequential thanks to the MUMPS.

Best regards

Image

image.png (348.53 KiB) Viewed 6749 times

[vokumus]

Thank you! I guess I should work on the boundaries.

Best regards.

[vokumus]

I have messed it with one day more and still couldn't find the culprit. I have tried to increase the size of FF boundaries, tried different damper coefficients, different records, soil props... Nothing works

[Clarabella]

Dear Vokumus,
the topic is very interesting and challenging. We have experience with SSI and San Diego soils' model and we know it requires a lot of work to make the model run correctly so do not give up. We wish we could help more. The good news for all the expert users like you and everybody else, it is that we are planning to have meetings for OpenSees users. Round tables in which we will invite researchers and experts in structures, soil and SSI to discuss about research topics like the one you are showing or answer questions related to OpenSees. Our target is to help researchers and students with their research, we hope that this initiative is going to be useful for all OpenSees community and great students like you. What do you think of this idea?

[vokumus]

Sounds great actually. ASDEA is one of the greatest company I have ever seen not only because you have true professionals but also you have the desire to share your knowledge with us. Looking forward to hear you.

Thank you very much.

[STKO Team]

Dear Vokumus,

Here is a modified version of your file (note: the file is set up for parallel analysis with openseesMP with 5 processors):

2B-NL-2M-20SN-20SN - MOD2.zip

(837 KiB) Downloaded 344 times

There were some modeling and conceptual issues.
This is a list of the main changes:

• LK and EDOF interaction - make sure to select only edges, and not the boudary vertices! otherwise the lumping will not work
• Fixities are not correct, Uy = 0 only on bottom, while Ux = 0 on bottom and vertical edges. You need to fix in Ux also the FF, because in this first stage you don't use the EDOF
• Fixities in Ux and Uy + EDOF structure-foundation are applied at the beginning, while the EDOF of the FF are applied after removing the Ux fixities, otherwise the ForceFromReaction will also pick reaction forces coming fromt that EDOF!!
• If you use the initialStageAnalysis option, keep in mind that it works only for solid elements... so you should build your model in 3 different stages: soil+foundaiton (do gravity) -> superstructure (do gravity) -> absorbing boundary elements in vertical boundaries (do dynamic)
• You used eleLoad for beams... keep in mind that they work in local coordinates.. in your case some columns have the localX axis pointing upward!!! so you should use the reverse command to change their direction
• LK properties should be funciton of the shear modulus throught the formula for P-wave and S-wave velocities. I changed them. Then I also multiplied them by 5, because 5 is the thickness of your 2D model, and in the FF by 1000.
• FF and Soil should have the same material model

Note: in this file I commented the conversion from elastic to plastic. It was just to make sure the LK work propertly in linear analysis. Since they depend on G, if you switch on the conversion to plastic state, the results will still be good, but not as good as in the elastic case. This is a known drawback of keeping the dashpot coefficients constant. A better solution would be to update them step-by-step.

This is the main workflow of the analysis:

workflow.png (107.73 KiB) Viewed 6710 times

[Clarabella]

At the end the STKO team found the problem too interesting and decided to solve it...

[vokumus]

I don't know how to thank you. I'll examine the changes step by step. This was very important for me. Thanks again.

[vokumus]

Dear @STKO Team;

In the model you submitted there are some issues;

1. LK base coefficients are very high. The values I choose is Cx=9128, Cy=15811 for a base rock with Vs = 762 m/sec and 5.00 m element thickness. (for example, Cx = 762 x 2.4 x 5 = 9128 kNsec/m). But paramter of models are Cx = 456250 and Cy = 790500.

2. Also, I think the Edge force applied to the base should has a multiplier same as LK coefficient. But the model has 45625 which is 1/10 of the LK coefficient.

If I use the values you provided, results are good. But if I use my parameters result are bad again. My model is attached. Thank yo very much.

2B-NL-2M-20SN-20SN - MOD3-Vefa.rar

(570.84 KiB) Downloaded 301 times