two perpendicular flap case

[carme-hp]

Motivation for case tutorials/FSI/two_flap_perp_2D is to test multi-coupling. Instead of one single flap, now we have two flaps, which means that the case involves three participants: Fluid, Solid1 and Solid2. For the first one we use OpenFOAM solver pimpleFoam and for the second we use the deal.II Adapter. In this link you can check the different possible approaches for multi-coupling: https://github.com/precice/precice/wiki/Multi-Coupling-Configuration
In this case we use a fully-implicit multi-coupling scheme.

The geometry of the flaps is defined directly in the deal.II Adapter solver. Because the flaps are located at different positions we need different executables for the solid participants. To obtain them we need to compile https://github.com/precice/dealii-adapter/blob/c332050b5770bf99459b8be596fd6e41f3401a26/linear_elasticity/linear_elasticity.cc#L208-L220 changing the parameters regarding the position of the flap.

• For Solid1:

    point_bottom =
    dim == 3 ? Point<dim>(-1.05, 0, 0) : Point<dim>(-1.05, 0);  
    point_tip = dim == 3 ? Point<dim>(-0.95, 1, 0.3) : Point<dim>(-0.95, 1); //flap1 
  

• For Solid2:

    point_bottom =
      dim == 3 ? Point<dim>(0.95, 0, 0) : Point<dim>(0.95, 0);  
    point_tip = dim == 3 ? Point<dim>(1.05, 1, 0.3) : Point<dim>(1.05, 1); // flap2
  

The executables are named linear_elasticity1 and linear_elasticity2 respectively.

In the precice-config.xml we define the communication between the solid participants and the Fluid participant. Here we can select the mapping (see https://github.com/precice/precice/wiki/Mapping-Configuration ).

Current status

• mapping:nearest-neighbor

• mapping:rbf-thin-plate-splines

• linear case

• non-linear case

• parallel run

The current priority is to find out why the solver crashes, which happens at t=0.55. If we use the rbf-thin-plate-splines mapping instead, it crashes immediately.

[MakisH]

In a good shape already! There are a few points that need fixing or implementation, this review may help.

And we need to make the case work, of course.

[MakisH]

Very good! A few more small points.

[MakisH]

Almost ready! (you can also change the status of the PR from "draft" to "ready")

I added a few (very picky) language and Markdown-related suggestions in the README, which already looks very good. 👍

[MakisH]

I ran the case for 15s and the results look very reasonable: both flaps oscillate back-and-forth, with different frequencies.

Video (unlisted for now): https://youtu.be/ClZNJC3qR6M

[MakisH]

This looks quite clean and nice now!

Dependent on precice/dealii-adapter#39, this runs nicely in serial for the linear case. However, in parallel (with (2 1 1) decomposition) I get this error: precice/precice#919 @carme-hp does it run in parallel for you with the default (2 2 1) and with (2 1 1)? I think it should be working with both.

The nonlinear case also runs nicely in serial.

[MakisH]

Everything looks good! Let's merge now and we will revisit this case in the context of the tutorials restructuring. For now it looks consistent with the rest.