Hi

I made a 1 meter block model calculation with k-e model, 5% turbulence at inlet and automatic length scale, it converged nicely. Then I tried to introduce more turbulence (25%)from inlet, I defined the turbulence and length scale approximat to the dimension the fluid domain to keep the turbulence reach the model. Then the convergence becomes very hard to go down to 1.E-4 and they(the curves) are just around 1.E-3 with up and down flucuation. What mechnisim behind this? why?

Thanks

Ahlo

Hi,

Has the turbulent viscosity decreased? If the effective fluid viscosity reduces you are more liable to get instabilities which can cause convergence problems.

Glenn Horrocks

Hi Glenn,

Thank you for your reply. It is true that the "eddy viscosity" (is it the same thing as the turbulence viscosity?) drops down rapidly rightway from the inlet, similar to the turbulence intensity with small length scale.

Is it necessary to keep the same viscosity to be more realistic? if any, how I can keep it for longer distance from inlet?

Thanks

Ahlo

Hi,

The turbulent viscosity is how the turbulence equations link back into the momentum equations in 2 equation turbulence models (k-eps, SST, k-omega etc). The turbulence viscosity the model predicts is a function of the turbulence energy and epsilon (or omega) so it is not something you explicitly "set". If the turbulent viscosity is too low it is because you have not correctly set the turbulence conditions up. You need to check your initial and boundary conditions to check they represent reality.

Flows with higher viscosity (whether it be molecular viscosity, turbulent viscosity or artificial viscosity from using an upwinding differencing scheme) are more stable because the viscosity damps out flow instabilities.

Glenn Horrocks