Nagano-Tagawa as low-Re k-epsilon model
 

Hi,

does someone know if it is possible to model an OpenFoam case with the Nagano-Tagawa low-Re k-epsilon turbulence model?
I still have looked the OF user doc, but it isn't listed there.
I still have read here in the forum that Launder-Sharma is available in OF. But I want to use Nagano-Tagawa because of its developement for heat-transfer.

Best Regards,
tH3f0rC3

Ok, it seems like there are "only" these turbulence models available.

Best Regards,
tH3f0rC3

Hi,

does someone know it if is right to set k and epsilon to zero when using a low-Re turbulence model, as it is described here?

Best Regards,
tH3f0rC3

tH3f0rC3,
Normally k is set to 0 at the wall for all k-e models that I know of with low Re functions (along with nu_t). For epsilon it depends on the particular version you are using and how it is applied by the solver. For launder-sharma the value of epsilon at the wall is the term 2*nu*(d(sqrt(k)/dy)^2 (sometimes also referred to as epsilon_wall though not the epsilon_wall of your link). This is at least the value used by the solvers in combination with epsilon_tilda where epsilon= epsilon_tilda + epsilon_wall. It is not clear to me which value you are setting in OF for specifying epsilon at the wall. If you are setting just the value of epsilon_tilda it should be 0 at the wall as appears to be indicated by your link as "epsilon_wall". I am not very familiar with the Nagano-Tagawa formulation, but it appears you specify 1/2 half the value of that term above as the value of epsilon at the wall. Essentially the purpose of the dampening terms is to prevent the numerical singularities that exist in the epsilon equation as y->0 from blowing up the simulation as k->0. Hope this helps, rather than muddles things for you.

Dave

The OpenFOAM implementation of the Launder-Sharma low-Re model solves for epsilonTilda: thus, the correct BC at the wall is epsilon = 0 (which practically means a very low value, such as for instance 10^-10)

V.