discretization scheme in recirculating flows
 

Let me explain a little the context first. I'm trying to do some simulations of a coaxial jet running at low and high velocity ratios. It's well known since ages that such jet have a self-sustained oscillating recirculation zone at potential core ending at high velocity ratios. Simulations were conducted with a RANS formulation and i was making some parametric tests on discretization schemes when a strange problem emerged. here are the parameters :

- Momentum equation is at 1st order discretizeation

- turbulence équations are at a) first order then at b) second order.

When no recirculation is involved, switching to case b) has no influence on the mean fields nor on the turbulent ones. But when recirculation takes place in the jet, switching to b) greatly enlarge the turbulence level in the recirculation zone (a factor 5 from a) without having any effect on the mean fields. This amplification of turbulence appears independently of the model used (ke, RNG, RSM) but is more pronounced for isotropic formulations (i.e. ke).

If someone has an idea on how second order discretisation of turbulence eq. can influence turbulence level in recirculating zones without changing the mean levels, he (or she) can leave me a message.

Thx in advance for your help.