This is an issue I raised previously to Itasca Support, but apparently it is still there in the latest version of FLAC3D (i.e. 147). Below I am using a simple example to explain the issue. However, for my real problem, the issue is worse and make me slightly concerned on the robustness of the liner (and maybe shell as well) elements, though so far the calculated soil resistance seems to be reasonable.
This issue is exaplained via the example “EmbeddedRetainingWall” by modifying the datafile slightly by generating the grid using “zone create brick size 11,11,11 ratio 1.2 1.2 group ‘right’”. After running the program until achieving equilibrium for the initial conditions (no need to run the 2nd stage), then plotting the normal stress on the liner, you will see below.
The results will become worse if the mesh become more non-uniform. Then I tried to running the file without “struct liner initialize coupling”, the normal stress calculated by FLAC3D becomes even worse.
So I cannot find a solution for this.
At the bottom nodes (links) because the zones still share the corresponding gridpoints (this is unavoidable since the zones are separated to that depth only), the behaviors of the normal stress from the links at the bottom are thus different from other places.
This is what I am suspecting, too. However, the interface logic for the liner should be robust enough to account for such senario, i.e. the interation of liner with soil should only occur in the direction tangent to the liner surface.
Could you please elaborate “this is avoidable …”? Do you have a solution? Or Is the solution to delete the links between the nodes and gridpoints for the bottom nodes?
If the separation is not complete to zone base, it is unavoidable (sorry, should be unavoidable) that some gridpoints are shared. The links are by default at the nodes.
Personally I think it would be a good suggestion to delete the links at the bottom.
