Hello Everyone ,
I am working on a 3D airfoil for quite a bit of time . I am comparing the results obtained from wind tunnel with that of CFD simulation .
Although , the results of my 2D simulations are quite match-able with the 2D version of wind tunnel data .
However , for 3D the results (values of Cl and Cd) I am obtaining from CFD simulations are 25-30 % lower than that of wind tunnel data .
Can you pinpoint where I might be lagging ?
Regards
Aamir
Difficult to say without having a close look on some mesh and convergence statistics of your simulation. Since 3d simulations are much more demanding than 2d simulations, here some questions:
1) Is your 3d mesh in the order of a y+~1.0 on the surface of the 3d airfoil?
2) Are you using SST turbulence model (please don't use any k-epsilon model, if possible)?
3) Is your simulation steady-state or transient + time-averaged?
4) If you try to converge in steady-state, is it really converging to very small residual level, or are the residuals stagnating and CD and CL coefficients are varying over the number of iterations in an oscillatory manner? The latter would be an indication of not properly converged steady-state solution and an inherent transient flow field around your airfoil.
Just some thoughts...
Regards, Th. Frank.
Thomas Frank Thank you so much for taking the time to answer my question . I really appreciate it .
1) Yes , I have taken care of the first cell height while generating mesh.I have taken the vale of Y+ =1 , and calculated the first cell height.
2) I am using Spalart Allmaras Turbulence Model ( 2D simulations gave good results for this model , so I am using it for 3D simulation )
3) At low angle of attacks I am running steady state analysis , at higher angle of attacks > 10 degrees I change it to transient time averaged .
4) Yes the results do converge for steady state analysis .
PS: I had cross checked the reference values too. But nothing seems to work .
I can share case file with you , if you could have a look on it , as I am trying to resolve this issue for more than 3 months but I am unable to get the desired results.
Ok. I see. Good care taken of all these points. SA model should be fine for aerodynamics flows. Nevertheless I would recommend to just take one or two angles of attack, where comparison to data is not that good and to try to predict the flow with SST. Just for curiosity (and because SST is predicting flows with flow separation quite well).
...three more questions:
5) Do you have in certain cross-section for this airfoil some C_p values, so that you can make a good judgment about possible flow separation on the upper side of the airfoil (e.g. good or not so good match of the point of flow separation)?
6) Is the prediction of CD and CL not as good for the whole range of angle of attack or is the deviation from the measurements just beginning from a certain angle of attack onwards? At least I have seen cases, where the comparison between data and CFD was quite good for angle of attack within approx. [-12 deg, 12 deg] and was rapidly deteriorating beyond this range.
7) Did you had a look on the flow field for higher angles of attack? Is there flow separation ongoing and is the used mesh capable of resolving any detaching vortices and potentially a forming wing tip vortex? If these detaching vortex systems are too fast dissipating because of a rapidly coarsened mesh in some distance from the airfoil, than this can affect correctness of CD, CL prediction as well, i.e. the accuracy of turbulence resolution in the airfoil wake.
Best regards,
Th. Frank.
- Badges
- Science topic