Fluid-structure interactions of 2D flexible thin palte

Official adapters and tutorials
, ,
1

Hi everyone,

I have been trying preCICE (OpenFOAM & CalculiX) recently to modify the ‘Perpendicular Flap tutorial’ to simulate fluid-structure interactions of 2D flexible thin palte.

The only changes made to the original tutorial case were flap geometry and material properties. The thickness of the thin palte is 0.002m and the height is 0.05m. I simulate my example at different flow velocity. During the process I got the following a problem that may need your help.

A tough problem: for the large-velocity scenario (i.e., larger deformation), after a few time steps, the OpenFOAM solver ended with errors. Unfortunately, mesh refinement near the flap as well as timestep reduction do not help alleviate the situation. As shown below, the flap is stretched to produce large deformation in the initial several time steps and the grids are significantly squeezed and stretched, which makes me consider this as a grid-related problem.

Fig.1
Fig.1827×392 74.3 KB

Fig.2
Fig.2827×391 197 KB

Can preCICE handle cases with such a significant structural deformation? I want to know whether the dynamicMeshDict of OpenFOAM should be modified to solve this problem. If so, how should it be modified? Any idea with adaptive mesh? The settings for dynamicMeshDict are as follows.

Fig.3
Fig.3591×690 76.4 KB

Thank you in advance.

2

Hi @Rvarane,

as you probably already know, simulating FSI with thin structures and high velocities is complicated. Partitioned simulation has its limitations, but I am not sure if you have already reached that level.

The coupling and mapping settings of the perpendicular flap tutorial are already quite optimal, but you may need to further tune some numbers, such as the coupling time window size, the convergence measures, and more.

But before that, as you already identified, the main issues come from the meshes themselves, as well as from the mesh motion solver. You would definitely need some mesh refinement. You would also need a better mesh motion solver, such as RBFMeshMotionSolver: Mesh Solver OpenFOAM Error - #10 by Makis

Don’t forget to apply similar changes to the solid case, next to the changes you apply to the OpenFOAM case.

3

Hi @Makis,
Thanks for your reply！
I will try the advice you provide. I am a novice at this simulation, so I would like to ask if I can set up the overset mesh in OpenFOAM as shown below to improve this problem? This image shows a simulation using ANSYS Fluent and Transient Structural.

Fig.4
Fig.41881×941 177 KB

4

I know that @JulianSchl has used overset grids already, and he has uploaded a case here: FSI coupling between OpenFOAM and MBDyn of a cycloidal rotor (not sure if this includes overset). I have not myself tried that.

I understand that a tutorial or some documentation would be better to answer your question. Why currently don’t have that, but I opened an issue: Needed: Tutorial with OpenFOAM and overset mesh · Issue #335 · precice/tutorials · GitHub

Still, overset may not be the feature you are looking for. I mainly know of overset as a solution to prescribed motions. Looking at the picture, the important part is that the overset mesh is adapting around the solid (as you see, the shape of the overset region is changing). I don’t know if OpenFOAM can do that (but this does not mean that it cannot).