Hi all,

I am involved in simulating flow around an airfoil for my project. I am using cfx5.4.1 in order to simulate this. For flows at -4 to 12 degrees, cfx will simulate a very good agreement in terms of flow modelling and the value of Cl. However when the angle increases to 14 and above , there are noticeable divergence from the lift curve slope of experimental results. I have tried many ways in order to improve this. I have tried to increase the turbulence intensity from .05 to .3. however the increase in Cl is almost negligible( in order of .01). also changed was eddy scale length but giving the same outcome.

Have anyone really successful in modelling such simple problem? If so to what extend ? how many elements? how about the inflated boundaries? what other parameters I should have look for ? how about turbulence model selection (I'm using Reynold Stress Model)? etc etc.

If there is any reference on the web or papers which shows their success please advise me.I hate to just simply suggest that the cfx could not model the problem accurately as my examiner will accuse me of not competent enough in cfx or CFD as a whole. So please help me .

Thank you

Yusoff

(1). So far, the problem is still a research problem. (2). There is no way to generate a solution to a research problem using existing methods or codes. (3). You are wasting your time.

hmm... okay but I could not put in my report that I am wasting my time. If possible can you point out any research paper that in favour of your argument? Just as reference in my report. Hope I dont waste your time though .

Thanks

Yusoff

I agree with John, the problem of predicting aerofoil performance around and past the stall position using the RANS method is still a research problem. The correct prediction of separation and hence stall is highly dependent on the turbulence model used. It is well known that turbulence models like the standard k-epsilon model are poor at predicting separation within regions with high-pressure gradients. At the end of the day you have to except the limitations of such turbulence models and endeavour to obtain solutions which are credible within these limits. With regard to your competence of using CFD, the sheer fact that you have managed to correctly set up and obtain a converged solution using a CFD package is testimony enough to your level of competence. Demonstrating that you fully understand the numerical modelling technique, its limitations and the importance validation plays in CFD, is by far the most important skills of all. Good luck with your CFD research.

Regards

James

Well thanks James for your praise . sure I need to read and learn more about this CFD thing. I just need guidance about the latest development in CFD .

Yusoff

(1). CFD is between the man and the nature. The center is the man. (2). Computer hardware and software are necessary tools to express his understanding of the nature, in this case the fluid dynamics (including heat transfer, chemical reactions, etc...) (3). The key activity of the man in CFD is research. And the use of commercial codes is only one way to explore the limitations of the existing models and knowledge about the nature. (4). Its immediate goal is to train more engineers in CFD, so that more will be known about his limitations in fluid dynamics. (5). So, the challenge in CFD is that there are many unsolved problems waiting to be answered. And the most important factor is the well-trained and experienced researcher, who can see beyond current limitations in modeling and his working enviroment. (tens of thousands of workers and engineer, researchers will be out of work because of cut back in auto and engine industries, as reported in the recent news. And that will affect the CFD enviroment right away.) (6). To make progress in CFD, the man must fully understand his current knowledge in modeling (mainly the limitations) and his working enviroment (through which to come up with new ideas and models).

Hi,

My speciality is in numerical analysis. As far as i know ,next to the airfoil the flow have avery thin boundary layer. And the higher the Reynold number is, the more steep the boundary layer is. The simulation of flows in boundary layers needs extremely advanced techniques. The standard finite element theory is not enough. If you use FEM, then the mesh next to the boundary layer must be very streched (thin, anisotropic mesh). But some element that are convergent in the standard FEM may not converge any more in the anisotropic FEM. I think that is why your simulation diverges after some higher parameter. I am a mathematician and i dont understand much about physical interpretations. But what i propose to you is to use element such as P2+ for the velocity and P1 for the pressure and to refine the mesh anisotropically next to the airfoil.