Unstability of solution when thickness too high

[sebi_magi]

Hello there,

I was recreating your SSI Absorbing Boundaries Webinar but in a 2D environment

[url]https://www.youtube.com/watch?v=OGII3a0 ... TpJFi&[url]

I tried to do the same than you in the webinar but in 2D (e.g quad elements instead of bricks, Distributed LK2D, etc.), but I would obtain unstable results (1e+50 accelerations!). I tried a lot of things to solve this, and something that worked was reducing the element thickness from 1m to 5e-6 or low values. This, however, feels I´m cheating and I'm not sure why this works and why with 1m element thickness doesn't.

I feel there is something going on with the dashpots, or maybe the problem becomes unstable because of large thickness with respect to the time step.

Here is the model attached. If you change the quad element thickness back to 1m, you should replicate the error.

[sebi_magi]

I actually solved this by building it from scratch with ASDA Absorbing Boundaries and is just much easier. Just one question.


With LK dashpots I can simulate bedrock by specifyng the C parameter for the viscous material with bedrock shear wave velocity, then I input the time series in the bottom edge. To my understanding In ASDA Absorbing boundaries this is not the case, isn't it? So in order to model an upcoming wave from the bottom, I should include a bedrock layer with proper elastic properties or is not necessary?

[STKO Team]

With LK dashpots I can simulate bedrock by specifyng the C parameter for the viscous material with bedrock shear wave velocity, then I input the time series in the bottom edge. To my understanding In ASDA Absorbing boundaries this is not the case, isn't it? So in order to model an upcoming wave from the bottom, I should include a bedrock layer with proper elastic properties or is not necessary?

Actually, you can do the same, on the bottom edge you can assign an ASDAbsorbingBoundaryMaterial with the G,v, rho of the bedrock. But in this case, you are placing the absorbing boundary exactly at the interface between the bedrock and the soil. In this case, you should pay attention to the input you give there. It really depends on how you obtain the input motion. Is it an out-crop-motion or a within-motion?

Anyway, try it, and also try modeling a layer of bedrock and check the differences

[begumemte1991]

Hello team STKO,

Could you kindly clarify the significance of element thickness in plane-strain elements? I tried searching this forum, but there is not much information regarding this topic. I carried out a 2D pile-soil interaction analysis and tried to validate the model with an experiment. At first, I took the thickness of quad elements and absorbing boundaries as 1 (note that the units I used are kN, m, s). The ground surface acceleration was matching, but the pier head acceleration was not. I played with so many parameters, but all in vain. However, on changing the thickness to 0.1, the result matched for both ground surface acceleration and pier head acceleration.

Therefore, I would like to know how is the element thickness actually affecting the results. It is not just the magnitude that changes. I am attaching the figures for your reference.

[STKO Team]

The plane-strain is a 2D simplification of a 3D model.
When you have a uniform model (for example, only soil), you can use whatever thickness you want. You just need to be consistent with the load you apply. And the result will be the same, the only difference of course will be the magnitude of the reactions that scale with the thickness

In your case, when you have a pile, which has a physical cross-section size, a change of thickness in the plane-strain will result in changing the relative stiffness between the soil and the beam (whose cross section remains fixed).
So in this case you should use a thickness which is exactly the same size of you have in the experimental test

[begumemte1991]

I see. Thank you for the information.

[STKO Team]

You're welcome