CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   turbolent models for wings (http://www.cfd-online.com/Forums/fluent/27785-turbolent-models-wings.html)

 Maria Teresa Milesi July 18, 2000 08:54

turbolent models for wings

Hi, I am studyng some NACA wings with Fluent 5.3, I've chosen k-e realizable models and the non equilibrium wall functions, I've found good results for Cl but I've some problems with Cd (the sperimental value is about 0.0055 and mine is 0.011). I tried also the RNG models with the differential viscosity models, I've found Cd=0.006 (a good value) but the distribution of velocity vectors is very strange (there is a distact in the end). Can you tell me which is the best turbulent models to use?

 John C. Chien July 18, 2000 09:40

Re: turbolent models for wings

(1). What I would do is to use classical engineering approach, that is to keep only two numbers behind the decimal point. (2). For Cd=0.0055, this becomes Cd=0.01. (3). For Cd=0.011, this becomes Cd=0.01. (4). For Cd=0.006, this becomes Cd=0.01. (5). So, the result is they are identical, that is Cd=0.01. (6). I think, it is possible to get more accurate results, but it will become a research project. And it will cost you a lot of time and money.

 Rüdiger Schwarze July 18, 2000 11:14

Re: turbolent models for wings

Modeling the turbulent flows near an air foil is a special field of research. There a many turbulence models, which are especially designed to describe the turbulence of a certain air foil. You have to study the literature in oder to find out which model is suitable for your purpose.

 Steve Collie July 19, 2000 19:48

Re: turbolent models for wings

What Re number are you using? For example the NACA data given in "Theory of wing sections" gives the drag coefficient of NACA0012 at Re=9*10^6 and zero angle of attack as 0.055, but at that reynolds number the flow remains laminar for ~30% of the distance along the aerofoil. And a laminar BL exerts considerably less skin friction. If you use a normal 2 eqn model then turbulent transition will be predicted way too early and the drag will be too much.

 John C. Chien July 23, 2000 10:25

Re: turbolent models for wings, the physics

(1). Yes, this is very important issue. That is the understanding of the test data and the physics of the flow. (2). First of all, a thorough understanding of the subject is essential. This can only be obtained through the study of experimental results. Although it is funny to say that the engineer using cfd approach must first study the experimental results, the fact is, cfd is only the flow simulation through solving equations. And if the equations do not include the correct model of the physics, then it will not produce the right answer. (3). Why would people sell commercial codes without proper physical models? That is because of the principle of "the customer is always right". (4). Well, the understanding of the physics of the flow alone is not adequate. When making comparison, the cfd simulation must also reproduce the exact experimental conditions. Whether the old test data were obtained under the controlled environment is sometimes questionable. Ideally, one should try to duplicate the test conditions, that is using the wind tunnel as the boundary.(if the test was performed in a wind tunnel) And was the free stream condition perfectly matched? (were the free stream velocity, temperature, Reynolds number, tke, etc... identical to the test condition? The free stream turbulence level can affect the laminar-to-turbulent flow transition on the airfoil.) (5). At least, one needs to run a laminar flow case (if it is possible to create the extremely fine mesh near the wall for high Re laminar flow condition), and determine the drag contribution in the nose portion of the airfoil. This can be used to substitute the turbulent prediction in the same region of the airfoil, to provide more realistic simulation.(without actually performing the laminar-to-turbulent flow calculations) (6). Without a thorough understanding of the physics of flow, and without proper matching of the boundary conditions to the test data, the cfd prediction is always on the "wrong " side of the answer.

 All times are GMT -4. The time now is 04:27.