I meet such a problem during Fluent carrying a calculation: "Turbulent Viscosity Ratio Limited to 1e+5 in ¡Á¡Á¡Á¡Á¡Á¡Á cells" till the calculation achieves convergence.

1.I have not change the default Max limitation of Turbulent Viscosity Ratio in Fluent;

2.to the same model and the same settings, standard k-epsilon model and RNG k-epsilon model can achieve convergence, but RSM model can not;

3.the problem of grid can be eliminable;

4.I think the initial setting of turbulence parameters causes it.

My model is a simulation of wind tunnel test, incompressible, and I give the k, epsilon explicitly. The referring materials available are the turbulence intensity and average velocity of inlet, so I can get the k value from these two parameters. It is about 5.8. When I determine the value of epsilon, I consult a formula applied by Fluent help, epsilon=(delta_K/U*L), while delta_K is the approximate decay of k across the flow domain. And epsilon is about 0.02 from this formula. "When using this method to estimate epsilon, you should also check the resulting turbulent viscosity ratio to make sure that it is not too large¡±, and I check it. The resulting turbulent viscosity ratio is about e+7~8. It exceeds the default Max limitation of turbulent viscosity ratio in Fluent. So I think it is the reason that prompts appears during calculation.

I am a novice, and I don't understand the meaning of limiting turbulent viscosity ratio well. I have several questions:

1.Epsilon could be calculated based on turbulence length scale. Some material says the integral scale of turbulence is equal to the turbulence length scale. Is this right? I think these two are different.

2.To the simulation of flow around high buildings in the bottom of atmosphere B.L., which method is appropriate in determining turbulence parameters and what the approximate value of epsilon is?

3.To my problem, what the reasonable Turbulent Viscosity Ratio limitation is?

3X.

K seems very large, do use the a percentage or a fraction for the turbulence intensity (it should be a fraction) i.e. 1% = 0.01. A wind tunnel should have virtually zero turbulent viscosity.

If you use a length scale then I would use one related to the size of any turbulence screens in the tunnel rather than the integral scale.

The buildings are probably within the log-law so I fould use that for your inlet profile.

Thank you very much. The buildings are within the log-law zone, if I use turbulence screens as the parameters determining turbulence, how can I determine their relations? Could you give me some direction? 3X.

depending on the turbulence model I am also having this problem. in my case it only appears during the beginning of the solution process. ormally I just keep on itterating and after some hundred steps fluent finds its way to the right range for the turbulent viscosity. under my impression increasing the limit of turbulent viscosity (and thereby having less cells "limited") helps speeding up this process. however a value to great leads to divergence, due to the fact, that fluent calculates a really wierd flowfield with huge gradients of turbulent viscosity, temperatur, velocity etc.