Taylor Couette

[IRage]

Hey Everyone,
i am new to Code Saturne and i am trying to simulate a Taylor Couette Flow.
I am using Version 2.0.2, because it is used where i work and it is not possible to update it.
I already modeled the fluid mesh and defined there groups for the wall and the interaction part with the rotating cylinder.
I also modeled the mesh for the cylinder and defined groups for also the wall and the interaction area with the fluid.
Is there anyone, who did something similar and could help me with updating the user subroutines?
Thank you in advance!
Paul

[Yvan Fournier]

Hello,

You can do most of the setup using the GUI, which should make things much simpler.

Probably the only thing you cannot do with the GUI is to define the boundary conditions for the rotating cylinder, because the GUI for the interface only allows a fixed veclocity vector for sliding walls, where you need a vector whose directon is tangential to a cylinder...

The logic of GUI/user subroutine combinations is as follows:
- user subroutines take priority over the GUI, but do not overwrite things defined by the GUI and not touched by a given user subroutine (a possible exception is definition of head losses and source terms, where the user subroutines redefine everything).

So I would suggest preparing as much as possible using the GUI, and addding SRC/REFERENCE/base/usclim.f90 to SRC in your test case, then adapting the examples therein to your sliding wall condition (you can define the other boundary conditions with the GUI if you want).
You probably do not need any other user subroutines if you use mostly the GUI, so this keesp things relatively simple (you only need to examine one example).

Best regards,

Yvan

[IRage]

Thank you for your fast answer.
But the problem i have now is, that if i create a new case with CS i can add 2 meshes. But how can i give every mesh different properties (densities etc).
As i understood it (i.e. in the turbine modeling thread) i have to create a case in which i can couple these two meshes. But if i type this "code_saturne create -s Studyname --nsat 2" (i hope thats the way i create a case with coupled meshes), i go to data.1 and run the saturne gui, but i can not load a mesh, so that i could assign any properties.
Thank you
Paul

[Yvan Fournier]

Hello,

Depending on the level of detail of the geometry, you might not need 2 meshes (rotor/stator), but simply a sliding wall, as I mentioned above (much simpler).

If you do use rotor/stator modeling with "code_saturne create -s Studyname --nsat 2", you cannot use the "Calculation Environment" (Meshes Selection or Mesh quality criteria) parts of the GUI to associate meshes with each case (you need to edit the runcase for this), nor the "Calculation Management" part, but you can actually use all the rest of the GUI (with a separate XML file for each domain).

Regards,

Yvan

[IRage]

Hey everyone,
i managed to start the calculation after i configured everything without the gui.
but now i get an error.
i attached the listings. i hope someone can help me with this error message.
Thank you in advance
Paul

[Jacques Fontaine]

Hello,

You do not need to use Code_Saturne/Code_Saturne coupling to simulate Taylor-Couette flow.
You must impose sliding walls on the FLUID mesh (as Yvan has said before).

Regards,
JF

[IRage]

I already managed to do it with sliding walls, but i don't get constant vortexes in the whole fluid. Only (in my case i set the velocity in x and z direction, y is the rotation axis) in certain parts of the fluid.

[Jacques Fontaine]

Hello,
Could you post your case with sliding walls on the forum?
About your listings: the linear solver for velocity blow-up and this may be due to many things. Maybe we can find the source of problem if you post your dataset.
Regards,

JF

[IRage]

As soon as i can (probably Monday or Tuesday) i will post my cases.
Does anyone know where i can change from a steady to an unsteady flow in the user subroutines?
I need an unsteady flow, but no turbulence model. Or what would you choose for taylor couette vortices?