I’m trying to replicate the Turek-Hron benchmark case for SU2-mbdyn.
I’m following the reference values proposed in “Proposal for Numerical Benchmarking of Fluid-Structure Interaction between an Elastic Object and Laminar Incompressible Flow” by Turek-Hron.
MBDyn alone gives similar results, but I’m struggling to obtain comparable Lift and Drag with SU2 alone.
Has anyone ever attempted to do this case with SU2? I’ve read Rusch thesis and one of the validation cases is FSI3, but I’m not sure how he accomplished it.
in my experimentation of FSI2 and 3 I used OpenFOAM, but if you like I could try to simulate with SU2 and compare the results.
What do you exactly mean with SU2 alone?
As to the results, afaik, there aren’t many studies on FSI2, but a lot about FSI3.
Turek, S., Hron, J., and Razzaq, M. in Numerical benchmarking of fluid-structure
interaction between elastic object and laminar incompressible flow state that “clear differences between the different approaches with regard to accuracy are visible. Particularly for the drag and lift values, which lead to differences of up to order 50%, and also for the displacement values which are
in the range of 10% errors”.
So it can be a tricky task, depending on the mesh, the fluid inlet (are you using a parabolic profile?) and other things. Are you getting a comparable flapping frequency?
I followed the tests proposed in the paper by Turek and Hron:
- CSM3: structural solver alone, gravity force on the beam where g = 2 m/s^2.
- CFD1: fluid solver alone, with a rigid beam.
The first one gave correct results, but I’m struggling with SU2.
In order to obtain a parabolic profile for the inlet velocity I followed Rusch thesis, where he used an extended offset between the inlet and the object. I checked and the parabolic profile is obtained.
The problem I’m facing is with Lift and Drag, which are very different from the paper values (1.119 and 14.29).
So the problem is within SU2 alone. Are you using a 2D or 3D case in SU2? If you made a 3D case, could it be something related to the thickness?
As far as I remember, SU2 let you measure Cl and Cd on a specific boundary, given some reference length. If you make a a case with the cylinder alone (no flap and no parabolic inlet profile), are you able to obtain a reasonable Cd?
If you don’t mind sharing the SU2 case I might make some experiments myself.
I’m using a 2D case and monitoring the CL and CD on the flap+cylinder.
I will try and run a case with the cylinder alone.
I’ll attach you the mesh and config file in an e-mail.
@alicezanella did you manage to solve the issue? If yes, how?
Sorry for the late reply. Unfortunately, I was unable to solve the case with SU2.
I am new to SU2 and was trying to implement this benchmark as well. So far my simulation keeps diverging and I suspect that the issue is on my boundary conditions for the inlet. Could you share your boundary conditions set up and how you implement the offset that Rusch mentions in his thesis even though you could not solve it?