[jmf]

Hello dear all

I'm working on a 3D airfoil model.

I would like to impose a given transition location.
Practically speaking, how could I force the boundary layer to be laminar on the upwind side of transition, and turbulent after ?

Which RAS model and wall functions setting should I use (the size of the simulation requires a high Re layer model)

Thanks in advance for sharing your experience
Best wishes

J-Michel

[madad2005]

I suppose you could use an if else in the wall function code for nut and omega so that if x < x_trans (x being freestream flow dir), then nut = 0 and Pk = 0, else if x >= x_trans where you default to the normal wall function? x_trans would either be a user-defined value or hard-coded. I've never tried it myself.

Otherwise, the K-Omega SST has numerical transition albeit at a Reynolds number a couple of orders of magnitude too low. Check Turbulence Modelling for CFD by Wilcox. There is also a paper by Rumsey and Spalart on the k-w SST and S-A models where they discuss transition (http://pdf.aiaa.org/preview/CDReadyM...V2008_4403.pdf) behaviour.

[jmf]

Hello everybody

Thanks Madad for your answer, you perfectly understood my concern

Increasing turbulence as suggested by Rumsey gives better repeatability, but bias laminar airfoil performances.

I'm new to Openfoam, and not skilled in C++.

Did anybody already programmed in OF a domain condition for layer turbulence ? Any help would be appreciated

Best wishes

J-Michel

[jackpap]

Only switching the wall function will not be enough in your case I believe. For example, if you are using SA model, setting nut to 0 on the leading edge boundary of your airfoil (where you want your laminar region) does not impose laminar flow. It only imposes "laminar" flow in the first cell....
If your grid is fine enough, nut should always be 0 at the wall since you are in the laminar sublayer ! So what you really want to do is deactivate the turbulence model in the entirety of the boundary layer in the imposed laminar (or pre-transition) region.

I think one way of doing this is using "cell zones" you will then need to modify the turbulence model you are using to check for "laminar cell zones" and set your turb. quantities to 0 in those regions.

[jmf]

Hello

Thanks for these advices

Thus I have to select cells within a layer from a given patch and apply zero turbulence condition.
Could anybody give me details on how to proceed in OF ? (My mesh is imported from Salome)

Regards

J-Michel

[jmf]

Dear all

Could the first step be to select cells with insideCells function, and thus build a cellSet ? Where should I store the selection domain ?

After it is not clear for me : how to use this cellSet for laminar turbulence condition ?

Thanks in advance for your time

J-Michel

[madad2005]

Originally Posted by jackpap

Only switching the wall function will not be enough in your case I believe. For example, if you are using SA model, setting nut to 0 on the leading edge boundary of your airfoil (where you want your laminar region) does not impose laminar flow. It only imposes "laminar" flow in the first cell....
If your grid is fine enough, nut should always be 0 at the wall since you are in the laminar sublayer ! So what you really want to do is deactivate the turbulence model in the entirety of the boundary layer in the imposed laminar (or pre-transition) region.

I think one way of doing this is using "cell zones" you will then need to modify the turbulence model you are using to check for "laminar cell zones" and set your turb. quantities to 0 in those regions.

Jacques,

I also said to turn off turbulent production by setting Pk = 0 (and omega = 0 at the wall too). This should have the effect of reducing the production and transport of turbulence in the surrounding regions to a minimal level, which should produce a "laminar" region of flow away from the wall (ie nut ~ 0) even without turning the entire turbulence model off. These values are zero at the wall when using a refined near field mesh, because you are expecting the first cell to be residing in the laminar sub-layer. Actually hard-coding Pk = 0 in the cells themselves is a better way of doing it, but will require more effort on his part. It is also subjective.