Dam break simulation

Official adapters and tutorials
, ,
1

Hi everyone,

Last time I try the dambreak simulation in explicit scheme and it was not end well, so I follow the @ uekerman’s advice to develop the implicit scheme.

The develop is finished recently, and I try the dambreak simulation again. I found out that when ddtSchemes is in default CrankNicolson 0.9;, the simulation will crash about 0.5s, but in default Euler;, it went well, so anyone knows why is this happend?

Furthermore, I notice that the mesh on the top of the flap keep oscillating, it confuses me for a bit, although it seen fine in structure apart and it will not affect the stability of fluid part, but it still a problem.
图片

图片

Any suggestion will be appreciate!~

2

Well… numerics is complicated! :sweat_smile: Maybe @uekerman has a guess here.

What does “keep oscillating” mean? I could imagine that this could be originating from the structure solver configuration (material properties, rigidity of the material) or numerics (type of the finite elements, space integration scheme, …).

3

HI @Makis ,

Thanks for reply. “keep oscillating” means one point of mesh on the top of flap keep moving up and down, as the snapshot shows above. But in structure solver, the mesh seems fine as shown below.
图片

4

And how do the exported values look like? Is this an issue with the mapping, with the OpenFOAM adapter, or just with visualization?

5

I think it’s about the mapping, it can be seen from the exported value:
图片

The arrow marked by red circle is apparently too lower than the other

6

Please upload your preCICE configuration file.

7

Here is the preCICE configuration file.
precice-config.xml (2.8 KB)

8

Since there are only a few points in that mesh region, I can imagine that the RBF mapping can give such strange artifacts. In the perpendicular-flap tutorial, we restrict the support radius:

    <mapping:rbf direction="write" from="Fluid-Mesh" to="Solid-Mesh" constraint="conservative">
      <basis-function:compact-polynomial-c6 support-radius="1." />
    </mapping:rbf>
    <mapping:rbf direction="read" from="Solid-Mesh" to="Fluid-Mesh" constraint="consistent">
      <basis-function:compact-polynomial-c6 support-radius="1." />
    </mapping:rbf>

We should probably be doing the same in the breaking-dam-2d tutorial. Do you observe the same behavior there?

9

I have restricted the support radius, but the behavior still stay the same. I used NN mapping also, but this kind of behavior keep happening. So I’m so confused about it, because it will affect the stability of simulation.

10

I found that when I carried OF in serial, the mesh seem fine, but in parallel, this behaviour will happend

11

I made a test about different way of decomposing mesh, I tried
simpleCoeffs { n (1 4 1); delta 0.001; }
and the result is shown below, the boundary of different rank will appear the mapping problem
图片
The part with and without mesh visualization denotes different rank

12

To validate that the issue is indeed related to the mapping, one would need to visualize both, the exported mesh and values of the Fluid and the Solid on the same participant, e.g., Fluid-Mesh-Fluid and Solid-Mesh-Fluid, then compare the displacement values on the solid mesh and on the fluid mesh. I don’t think that this issue is related to the mapping configuration.

1 Like