Spatially variable soil layers in FLAC 3D

FLAC3D
1

I am working with a 3D model that has a spatially varying soil profile. What is the recommended method for assigning soil layer groups? Can soil layers be interpolated from a few 2D cross sections? Any example workflows would be helpful.

2

Hi Milads,

There are many ways of doing it in FLAC3D.
Which way to use will dependent a lot of the data source you have and the complexity of your spatial variation and how much you want to simplify it. It is also intrinsequally linked to how your are going to mesh your space and what problem you are trying to solve: for example, if the different horizon in your soil have fairly similar properties -or have a mostly elastic behaviour- then you don’t need to have a very smooth delimitation between the horizons with your grouping. If you expect a lot of deformation, very no-linear creep and large contrasts in properties: then you need to resolve the boundary very finely and the approach will be very different.

A lot of explanation are available here:
Grid Generation in FLAC3D — Itasca Software 9.1 documentation

Here are a few examples:

  • If the horizontal variation is very simple then just take some key points from your 2d sections and stitch together primitive forms. Each primitive form can be assigned a group at creation and then you can merge the ones you need to with

zone group

and apply the range on groups.

If the horizons have very different properties or very non-linear behaviour:

  1. If your boundaries of group are complex, but do not varies too much in vertical amplitude spatially (what too much is very much application dependent as the main risk is to have a bad mesh), the simpler way to do things is to mesh each group boundary as a surface using an external software (the easiest is to use a CAD software, QGis is also very good at that, but you can do it with some python scripts fairly easily) and save it to an stl that can be import by FLAC. Then follow the procedure described in the section " Surface Topography and Layering" from the documentation (it is in the link above). you can generate a group at each creation of mesh ensemble.

  2. If your soil structure is very complex (for example very irregular slops or large local vertical drops) then the best bet is to use Griddle but I doubt you will need to go that far.

If the horizons have fairly similar properties or not too non-linear behaviour
Then you mesh does not need to smoothly follow the horizons boundaries. In this case you can use an externally generated geometry (like describe above) as a grouping technic with the command:

zone group 'mygroup' range geometry-space 'mygeometry'

See details in section " Geometry-Based Densification: Octree Meshing" of the documentation (again link above will take you there).

You can also used the approach described in this post:
Assigning zone properties from boreholes like SPT, CPT, etc. for creating own 3D soil layers - FLAC3D - Itasca Software Forum

I hope this overview helps.