Restrictions on time steps
Hi there,
I am working with Incompressible, viscous and time dependent flows. In recent days, I am implementing a Low-Reynolds turbulence model in a explicit code. If every one to know what is the restrictions on time steps in turbelent flows, or related papers, please let me know. Thank you. |
Re: Restrictions on time steps
There are a couple of papers by Kunz and Lakshminariayana on the stability of explicit solvers using the Low-Re Chien model. Their conclusion was that the Chien Low-Re k-epsilon model does not lead to any additional requirements. My real-life experience is not the same though... I'd say that if you want to use low-Re models you should consider moving to an implicit solver. The explicit Runge-Kutta solvers are not very suitable to solve this type of source-term dominated equations, IMHO.
|
spelling correction
look for Kunz and Lakshminarayana (Jonas got the second name wrong, but who can blame him)
|
Re: Restrictions on time steps
Just a general observation, I don't have any particular references for you. The turbulence model introduces an eddy viscosity. If you already have a stability estimate based on viscosity considerations, that should work just fine with the combined eddy and molecular viscosity. The solver has no "knowledge" of any difference. This is how I have handled it before and it seemed to work just fine.
Also, a previous poster made reference to this being a "source term dominated" system. In what sense does \nu_t look like a source term? I had never conceptualized it as such. Bob |
Re: Restrictions on time steps
I meant that the k and epsilon equations are source term dominated. The effect on the other equations is only through \nu_t, as you say.
|
Re: spelling correction
Heh, my misstake... I even took out his excellent turbomachinery book to check the spelling and I still got it wrong. Btw, the papers are in J. Comput. Physics, vol 103, p 141 and AIAA J. vol 30, no 1 p 13.
|
All times are GMT -4. The time now is 18:42. |