Hi all,
I am a newbie to Elmer, and have been trying to get the K-Epsilon turbulence model to converge. I have tried to apply modified boundary conditions from the step test case to 3D problems, as well as examples from these forums, with no success in converging the simulation. I have gone back to the basic problem of water in a 50mm diameter, 1.5 metre long straight pipe flowing at 1.8 m/s. Attached is the sif file of the case that comes closest to converging.
Is there something obvious that I’m doing wrong here?
K-Epsilon convergence issues
K-Epsilon convergence issues
- Attachments
-
- case.sif
- (3.3 KiB) Downloaded 438 times
Re: K-Epsilon convergence issues
Hi,
some ideas after a first look to your sif file:
- You must set the inlet values of the turbulent variables (Kinetic Energy and Kinetic Dissipation), not only its initial values.
- The boundary condition for the wall is wrong. You must use a wall law for the K-Epsilon turbulence model, as long as the K-Epsilon version in Elmer is only for hi Reynolds flows. So, delete the keyword "Noslip Wall BC" (keeping the wall law related keywords) , and ensure normal velocity = 0 with a dirichlet condition on the normal component.
- The convergence will probably improve doing just one nonlinear iteration of each solver per coupled system iteration ("Nonlinear System Max Iterations = 1" in both solvers).
Regards,
Cesar
some ideas after a first look to your sif file:
- You must set the inlet values of the turbulent variables (Kinetic Energy and Kinetic Dissipation), not only its initial values.
- The boundary condition for the wall is wrong. You must use a wall law for the K-Epsilon turbulence model, as long as the K-Epsilon version in Elmer is only for hi Reynolds flows. So, delete the keyword "Noslip Wall BC" (keeping the wall law related keywords) , and ensure normal velocity = 0 with a dirichlet condition on the normal component.
- The convergence will probably improve doing just one nonlinear iteration of each solver per coupled system iteration ("Nonlinear System Max Iterations = 1" in both solvers).
Regards,
Cesar