I am looking into the possibility of using OpenSees to model dynamic soil-foundation-structure interaction during a soil liquefaction scenario. I intend to attach soil blocks directly to foundation elements (probably considering only two-dimensional model).
I am interested in the response of the structure during ground motions that cause liquefaction, but I need soil elements that can capture correct liquefaction behavior. Though computationally taxing, I know that OpenSees is capable of this type analysis. I have experimented with some success with UC-San Diego soil elements (BrickUP and QuadUP) and have also looked into using UC-Davis soil elements (Brick_u_P_U).
I have run across some issues with degrees of freedom however, when adding a structure/foundation to soil blocks. Can I keep rotational degrees of freedom for the structure and still keep track of pore water pressure in the soil? Must I use u_P_U elements? Is the QuadUP element compatibile with both UC-SD and UC-D material models?
Any advice would be appreciated for this type of analysis. Before I invest more time experimenting with soil elements, I need to make sure I can feasibly peform this type of brute force analysis with either UC-SD or UC-D elements.
Soil Elements Used in Soil-Foundation-Structure Interaction
Moderators: silvia, selimgunay, Moderators
Liquefaction and SSI
It should not be done in 2D! The problem is inherently 3D, as there is flow of fluid and soil skeleton around the piles and other 3D effects...
Take a look at last few presentations that I have linked to my web site:
http://geomechanics.ucdavis.edu
for an idea what ca be done.
As for DOFs, you connect the solids to beams by considering only translational DOFs while you neglect the rotational once. This is OK if you create a model for a pile that sits in an opening (hole) and has short, stiff beams as connections to surounding soil (u-p-U) elements. You also need a set of very soft and lighweight elements with appropriate permeability of concrete to fill up that opening so that the flow of water in soil is right...
Boris
Take a look at last few presentations that I have linked to my web site:
http://geomechanics.ucdavis.edu
for an idea what ca be done.
As for DOFs, you connect the solids to beams by considering only translational DOFs while you neglect the rotational once. This is OK if you create a model for a pile that sits in an opening (hole) and has short, stiff beams as connections to surounding soil (u-p-U) elements. You also need a set of very soft and lighweight elements with appropriate permeability of concrete to fill up that opening so that the flow of water in soil is right...
Boris
-
srisri_iitk
- Posts: 43
- Joined: Wed Oct 25, 2006 9:29 pm
- Location: IIT KANPUR
Yes, I can see now that 2D would be an oversimplification for liquefaction, especially for detailed analysis. Fluid flow would be restricted in an unrealistic manner.
I have seen some of the large scale bridge models in 3D in some of your presentations (with DRM Soil Blocks beneath bridge piers). Ideally, I would be able to create a similar model, but I fear this may be too ambitious for someone with little experience in geomechanics and parallel computing, like myself. I am, after all, concerned primarily about the response of the superstructure. Any thoughts?
At this stage, I have been advised to avoid using Parallel Computing due to its inherent complexity, but I anticipate analysis times for a ground motion record causing liquefaction would be too great unless it is used. Without parallel processing, can I expect run times of days? months? years?
Thanks for your help.
I have seen some of the large scale bridge models in 3D in some of your presentations (with DRM Soil Blocks beneath bridge piers). Ideally, I would be able to create a similar model, but I fear this may be too ambitious for someone with little experience in geomechanics and parallel computing, like myself. I am, after all, concerned primarily about the response of the superstructure. Any thoughts?
At this stage, I have been advised to avoid using Parallel Computing due to its inherent complexity, but I anticipate analysis times for a ground motion record causing liquefaction would be too great unless it is used. Without parallel processing, can I expect run times of days? months? years?
Thanks for your help.