Hi all

It possible for the RNG to by pass transition and it is formulated in such a manner that it can evaluate the Eddy viscosity, molecular viscosity depending on how close you are to the wall right down to through the viscous sublayer, which is fair enough.

Fluent will not let you switch off all the wall functions, at least one has to be on. Had anyone similar difficulty with Fluent?.

Regards oliver

The RNG k-eps model in Fluent is not intended to be used as a low-Re model all down through the viscous sublayer. You should use either their two-layer model which uses a one-equation Wolfstein-model in the inner parts or the boundary layer or one of the low-Re models that are avalable through the text-interface for the standard k-eps model. Some of the low-Re models available in Fluent have been used succesfully to predict by-pass transtition. I would recommend either the old Launder-Sharma model or the Yang-Shih model. However, predicting by-pass transtition with two-equations models is very difficult and in most applications impossible. You really need to validate your simulation approach for your specific application in order to be able to trust your results. For most applications you will predict a too early transition and a transition which incorrectly is independent of the free-stream turbulence level. If you want to predict by-pass transition you also need to have a very good grid, both in the normal direction to the wall (y+<0.5) and in in the streamwise direction. I haven't done any by-pass prediction simulations of this kind in Fluent som I'm not sure if Fluent's numerics are suitable for it - has anyone done this type of predictions with Fluent successfully?

hi Jonas

1 For 2D planar flow, such as flow over a blade, computational restrictions are near non-existent even for a PC of even low spec. The RNG theory is based on a statistical approach, which uses its own formulation to calculate the effective viscosity right down to the wall. Therefore the implementation of a wall function only serves to corrupt results, by breaking the link in the RNG modeller and replacing it with data obtained from the implementation of a sophisticated wall function which either uses a one equation or 2 equation low Re model.The question now arises what was the model formulated for?. 2 On a similar note an according to your logic the RSM could be used to bypass transition by the implementation of an "appropaite" wall function. Therefore if one ran a RSM and noted a dip in transition which is similar to the WSS obtained experimentally, then the RSM model predicted transition, even though it not formulated to do so. Can you explain this?. 3 Can all wall functions be switched off in fluent?.

Regards oliver

When the original RNG k-eps model was presented by Yakhot and Orzag people thought that it would be possible to use this model through the viscous sublayer down to wall. However, later experience shows that the RNG k-eps model also needs special low-Re tuning in order to work well down through the vicous sublayer. I don't think that Fluent has any low-Re RNG k-eps model implemented, hence you can't use RNG k-eps down to the wall in Fluent - you have to use a wall-function or a two-layer model. Fluent does have low-Re std k-eps models though, which you can use down to the wall without any wall-function. These models might be able to predict transition.

A wall-function approach can of course never predict by-pass transition since it assumes a turbulent boundary layer.

Hi Jonas

Thanks for the advice Jonas, It would be nice to see the omega formulation as optional closure models in Fluent.

Regards Oliver

Fluent is working on a UDF which implements the k-omega model. If you talk to support they might give you a copy of it. If you intend to try to predict by-pass transition this might be a good idea - you could implement Wilcox's special transition version of the k-omega model and perhaps even tune it for your special application.

Hi Jonas

What omega model are they working on. Is it a first order closure turbulence model or is it the more elaborate Wilcox Multi Scale (WMS) turbulence model.

Regards

Oliver

Hello Jonas,

just a question referring to your statement concerning the range of use of the RNG-model. I am a little bit confused by your answer that RNG with its "differential viscosity option" is not suitable for resolving the near wall domain. Because this is just that what FLUENT proposed the RNG-model to be able when it was introduced some years ago.

Hence what for is the "differential viscosity option", unless for getting better results with a resolved boundary layer?

Hi Volker

As with all models based upon epsilon formulation, the base RNG closure model is inherently weak. It cannot model the physics of the flow properly in the near wall region, therefore a Low Re variant of the RNG model is required which contains fudge factors (i.e. viscous damping functions) to ensure the viscous stress in the near wall region take over from the Reynolds stresses. However due to the strong coupling between all of the fluid flow regions, viscous sublayer, log law etc., which when numerically computed, are computed in tandem. If accurate results are obtained for the outer portion of the Boundary Layer, these results are bound to influence the weaker near wall results, hence overall producing a reasonably accurate "engineering" answer. Therefore it's logical to assume that the accuracy of the solution arises from the "differential viscosity" option.

Regards Oliver

[AdidaKK]

can i work with k-e RNG model ,in a combustion problem?( can combustor of a gas-turbine) .

best regards

[denizen]

i see you are necroposter

Yes you can use rng-ke in a combustion problem.

regards

[AdidaKK]

thanks very much XDD

best regards

[idrama]

Hey All,

does anybody know how the eddy-viscosity is calculated from RNG-k-e model?

Cheers