[yhnhex]

Hi!


I am currently working with SU2 to define the flow around a spinning cylinder. I do not know how to define de boundary conditions (I made a mesh and defined the upper and lower walls as farfield, the inlet with the freestream p, T and speed and the outlet with the freestream pressure), but it isnt working as expected. I have seen some examples of SU2 and they define everything as farfield (no outlet or inlet) anybody knows why??


thank you!!

[naffrancois]

Hello,

With your current bc set up, what goes wrong ? Slow convergence ? At initial times your solid body generates waves which will interact with your left and right bc where you impose entering/exiting flow. These waves reflect back and forth throughout the domain before dampening enough hence slowing down convergence rate.

I do not know how farfield is specifically implemented in Open Foam, but usually this kind of bc can handle inlet/outlet subsonic/supersonic automatically based on local direction of the flow and mach number. Farfield, as its name suggests, should however be set at boundaries quite far away from any disturbance as a solid body for example. These bc are usually constructed from a kind of linearized Riemann Solver and as such cannot handle well high amplitude waves, that is why they should be set quite far from a solid body. However they will let low amplitude waves quit the domain quite easily and so convergence to steady state is better.

[yhnhex]

Hi,


I am trying to compile it without spinning first and the issue is that the result I get in post processing is not what I expect. I attach a picture of how I defined my domain. The conditions are Re=200, M=0.05 and T=500 as freestream and also at the inlet.



The result I get is steady flow in all the domain except for a tiny area surrounding the cylinder, and I was expecting some kind of recirculation behind that area.


Also, I am working with SU2, not openfoam.


Thank you!!



https://drive.google.com/file/d/11qj...ew?usp=sharing

[naffrancois]

At Re=200 the flow is unsteady still 2D, if you use laminar NS equations you should see von karman street given you use a sufficiently fine grid around the cylinder and a not too big time step size.

If not the case use su2 in unsteady mode, no steady tricks such as local time stepping, multigrid etc, what numerical schemes are you using ? The SU2 tutorial with Re=40 is indeed a case where you get a steady solution because there's no separation. If you increase Re, the flow becomes unsteady.

M=0.05 is also maybe too low, compressible solvers have difficulties with low mach numbers and preconditionning is sometimes necessary...

Have you started from the tutorial ?

https://su2code.github.io/tutorials/Laminar_Cylinder/

From there, increase Re number and switch to unsteady numerical scheme