Issues in 2D Simplification of 3D SSI Model: Unexpected Fluctuations in Acceleration Response

[liushun]

Hello STKO team,

I hope this message finds you well.

Recently, I studied the webinar on Soil-Structure Interaction (SSI), which employed a 3D model. I attempted to simplify it into a 2D model for analysis, aiming to simulate the theoretical solution where the ground surface acceleration is twice the input acceleration in an elastic material. However, the simulation was not successful, and the contour plots show unusual fluctuations. I have been unable to identify the specific cause. If you have time, I would greatly appreciate your assistance in resolving this issue.

I look forward to your guidance.

[liushun]

Hello STKO team,

I hope this message finds you well.

I would like to ask another question regarding the operation in my model of removing boundary constraints and applying "Force from reaction". What is the specific reason for this approach? I have never considered such a method in my previous modeling processes. Is it related to viscous boundary conditions? I noticed that the one-dimensional soil column example in the provided link(https://opensees.berkeley.edu/wiki/inde ... oil_Column) did not include this step of removing constraints and applying reaction forces. I would greatly appreciate your help in clarifying this.

Since I haven’t received a reply to my previous message yet, I tried a new approach based on the one-dimensional soil column example mentioned above. Instead of placing dampers on both sides, I used equalDOF to handle the lateral boundary conditions. However, after running the analysis, the contour plots show that the entire soil model exhibits rigid body motion, which did not occur in my previous attempts. I will attach my model at the end and would greatly appreciate your help in analyzing whether there might be errors in my model setup.

Thank you for your time and support.

Best regards,
shun

[liushun]

Hi again,

I’d like to ask another question about damping coefficients. At the moment, all the models I build are 2D. I set dashpots at the base. When I calculate the coefficient, I use rho · Vs · element thickness. DistributedLK2D will automatically calculate the distributed length of the dashpot, and the product of these two is the final damping coefficient. The input applied to the soil is a force time-history curve, which also involves a numerical factor. In this case, should that factor be equal to the damping coefficient, or twice the damping coefficient?

I mention the “twice” issue because I previously built an elastic material model where the target result was that the surface acceleration is twice the base acceleration. I used EqualDOF on the two sides and DistributedLK2D at the base. I found that I had to multiply the applied load factor by 2 × ( rho · Vs · element thickness) to obtain the desired result.

So far I’ve posted three separate messages here, but I think I should reorganize my questions more clearly. Here are the issues I’d like to consult about:

1. I want to simplify the SSI webinar model into a 2D model for analysis, but I have not been able to reproduce the theoretical solution, i.e., surface acceleration = 2 × base acceleration.

2. Is the step “Force from reaction” related to the viscous boundary? If so, why does this step not exist in the 1D soil column example in the link I provided?https://opensees.berkeley.edu/wiki/inde ... _Analysis)

3. Following the 1D soil column modeling approach: dashpots at the bottom, EqualDOF constraints on both sides. Different from the soil column, I did not set EqualDOF between the base nodes. However, during the analysis I observed an overall downward displacement of the model.

4. The damping coefficient is computed as ρ · Vs · element thickness. This value is assigned to DistributedLK2D, which automatically computes the effective distributed length and yields the final damping coefficient. Then for the load factor applied to the force time history: should it be multiplied by 2 (i.e., use 2 · rho · Vs · element thickness) or not? In the link I provided, the load factor is not multiplied by 2. However, in my elastic-material model I had to scale it by 2, so I’m a bit confused now. I will attach my elastic-material model here as well.https://opensees.berkeley.edu/wiki/inde ... _Analysis)

These are the confusions I’ve encountered recently. They seem to be connected, so I put them together here and hope to get your help. For the models related to Questions 1 , 3 and 4, I have re-attached them under this post.

I apologize for raising multiple questions at once. Because I have not been able to resolve the initial problem yet, and additional related issues have emerged during my attempts, I am currently in a difficult situation and would really appreciate your guidance.

Thank you again for your time and support. I look forward to your reply.

[STKO Team]

Please, have a look at the webinar using ASDAbsorbingBoundary elements. It's much easier to use them rather than the old LK dashpots