Problem with NACA simulation

[Sébastien Thomas]

Hello everybody,
 
I realize different simulations with an airfoil of type NACA0012 with different angles. I post my results and two examples of simulations at the end of my message. In fact, I realized a fonctiun to compute the drag and the lift corfficient. This subroutine have correct results for cylinder but my results of drag coefficient for NACA are far removed from theory. You can see this problem on excel file at the end of the message. ANd I search a solution to know the momentum coefficient at 25% at the chord of my foil.
 
Have you got a solution for my two problems?
 
Thank you very much
 
Best regards
 
Sebastien

[James McNaughton]

Have you performed a mesh independence study - the mesh you provide has a large jump in cell size and large aspect ratios (especially at the trailing edge).
 
James

[Sébastien Thomas]

Hello,
Thanks for your answer,
I have performed a mesh independance study ans it is conformed to comput with FLuent.
I have realizes different tests to compute with different extrapolation and other variable and my results are better but they have always an important difference in drag coefficients.
I haven't relazied my problem on momentum coefficient.
Have you got a solution for my problems?
Thank you very much
 
Best regards
 
Sebastien
 

[James McNaughton]

Not sure about the drag then - it's always harder to get than the lift . What turbulence model do you use and are you using wall functions? I'd recommend trying the k-omega SST (iturb = 60) with y-plus around unity.
 
Moment coefficient is essentially the torque coefficient so:
M = x F y - y F x
Then:
C M = M / (0.5 rho u^2 A c)
(I think that's right!)
You need the origin to be at the aerodynamic centre point (25%) or make a transformation (something like:
M' = M + x 1 F y - y 2 F x
where x 2 and y 2 are the offset from the 25% point).
 
Hope that's of some help,
James

[Sébastien Thomas]

Hello
 
I use the model k-epsilon with an y-plus between 30 and 300, because the we work on different fluid environnement with bigger dimensions after this qualification and so quantity of memory and the time to compute with k-omega SST is too hight.
 
Thank you for your formul for the momentum coefficient but i search the function in Code Saturne to keep this coeffecient like the pressure coefficient or the skin-friction coefficient.
Best regards
 
Sebastien

[James McNaughton]

I don't think there is a function for it.. You must loop on your boundary face and calculate M:

FM = cdgfbo(1,ifac)*FY - cdgfbo(2,ifac)*FX

Where FX, FY are in the usproj you attach - the rest you have already done for your other coefficients.
Perhaps you could try SST on this smaller domain and establish the error in drag you get from a change in turbulence model. Then you can use this information when making a conclusion on your results for the bigger domain with the k-epsilon.
Also are you running steady solver, if so you could try the unsteady one as I've often had more success with this.
Cheers,
James