hi

i'm coding unsteady rans channel flow(k-epsilon, Ret=180) but turbulence doesn't appear. it converges larminar flow..... i don't know why it does. i need your help.

Hi,

Here are two possibilities:

(1) your algorithm is overly dissipative (which is not a good situation);

(2) Keep in mind: larminar flow is a possible solution to the NS system. Add much larger random fluctuations to your initial field, and then let the field gradually converges to a steady state turbulent field.

Hi Thank you for your response.

It seems that my case is (1) among possibilities you give.

May I know the ratio nu(kinematic viscosity) to nut(eddy viscosity)? And I want to konw the initial condition of kinetic energy(k) and eddy dissipation ratio(epsilon).

Sorry I thought you were doing LES or DNS. But, in any case, do not give up easily. Since you are using k-epsilon model, I assume it is steady RANS. Check you k- and epsilon-equations and also the eddy viscosity model, and see if they are coded correctly. Check, e.g., the calculation of the strain rate tensor.

The case of Re_tau=180 is a very classical one. You can find the ratio between nu and nu_t in, e.g., ``Turbulent Flows'' by S. Pope. If I remember clearly, nu_t/nu is about 36 at the center of the channel.

Just my curiosity: if this is for basic training (e.g., course study), it is fine. But if this is for your PhD research, not good. You should be provided a basic code which can do this classical benchmark test case, at least. Otherwise,,,

Thank you for your advice~ Actually I have studied hybrid RANS-LES model. Unsteady problem is terrible thing itself..... T T Now turbulence is generated thank to your help but the result is not good. I want to know the periodic boundary condition for pp. When soving Poisson eq., how are coefficients, ae, aw(steamwise direction), ad, au (spanwise direction) treated for SIMPLE algorithm?