2D varying soil properties

Forum for OpenSees users to post questions, comments, etc. on the use of the OpenSees interpreter, OpenSees.exe

Moderators: silvia, selimgunay, Moderators

Post Reply
SebastianNunez
Posts: 2
Joined: Mon May 31, 2021 12:37 pm

2D varying soil properties

Post by SebastianNunez » Thu Aug 05, 2021 8:37 pm

I want to do a model soil with varying soil properties (for example, an elastic soil with different shear wave velocity (Vs) for every element in a 2D or 3D mesh). Geotechnical examples I have seen in the wiki model soil with layers, and that means all elements in the layer will have the same Vs. Is there and efficient way to approach what I want to do? Im just learning OpenSees and what i want to do doesn't seem the easiest task.

Thanks!

csj
Posts: 3
Joined: Mon Jul 20, 2020 8:47 am

Re: 2D varying soil properties

Post by csj » Thu Aug 05, 2021 11:39 pm

Hi Sebastian,

It seems to me that you want to deal with soil variability by modeling soil properties as a random field.

I think that the best way to start is by the simplest case, a 1D model. For example, one may assume that Vs varies linearly with depth. You can then replicate the same steps given in the examples to model the soil as continuum, but the key is to create "several thin layers". Basically the idea is to map the Vs(z) onto the center of each layer to reproduce this linear depth dependent Vs profile.

After that, you may start to look at the 2D and finally 3D problem. If you really want to represent as RdF the soil properties, you want to learn about Gaussiand RdF and RdF generators to perform simulations in other to create plausible representations over the space of the soil properties. But that's is another story and it's not related with OpenSees or any FE framework.

Gl

SebastianNunez
Posts: 2
Joined: Mon May 31, 2021 12:37 pm

Re: 2D varying soil properties

Post by SebastianNunez » Fri Aug 06, 2021 5:55 am

Thats pretty much what I want to do! I have some 1D data and I want to interpolate it as a random field and generate some 2D profiles from that. Later, I want to make more measurements of the soil and then the interpolation would be better.

What do you mean when you say that is not related to OpenSees framework? Is not possible to generate random fields and put them as inputs in OpenSees?

Thank you very much!

csj
Posts: 3
Joined: Mon Jul 20, 2020 8:47 am

Re: 2D varying soil properties

Post by csj » Mon Aug 09, 2021 12:45 am

First, generate a 2DRdF by interpolating 1D RdF simulations is not correct even if u assume orthotropic conditions. Simply u can not model/preserve the spatial structure following the horizontal direction with the approach that u proposed. Again, I advise you to start by 1D case to understand the basis that u're doing. For the 1D case, u can generate the RdF by Cholesky decomposition, the simplest approach.

Then, to my best knowledge OpenSees doesn't have implemented any module to simulate RdF. It has a reliability analysis module based on a classic Random variable modelling framework to deal with uncertainties. Therefore you have to generate yourself the RdF and then map to the grid. Literature in this field is extensive (i.e. midpoint method). I also advise to u work through the Python interpreter (OpenSeesPy) to further automate your workflow.

KR

Post Reply