CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   problem is modeling and simulation of an air foil (

alnabhani November 29, 2010 04:41

problem is modeling and simulation of an air foil
2 Attachment(s)

i have to simulate one system which is the flow of air around an air foil. the speed from the inlet - the circle red - is around 160 m/s. the outlet is marked with the green circle. this image shows the model of the problem:
Attachment 5557

the following picture shows the mesh;
Attachment 5558

the problem is the pressure distribution i'm getting which is totally wrong. what i can do now ?

ghorrocks November 29, 2010 18:24

Why do you say the pressure distribution is totally wrong? What are you comparing it against? If you are comparing it against far-field values (ie in an infinite domain) then of course your result will be totally different as you have a very constrained flow with a upstream wall and strange outlet.

Josh November 29, 2010 19:44

Agreed with Glenn. Your domain is pretty tiny. You may be experiencing problems due to artificial truncation of the domain. Try making the domain larger in every direction.

alnabhani December 2, 2010 08:02

this model is of a real device and we need to inspect it and insure the work that it will do. this device is a separator , so the pressure up the wing should be smaller the aht under it by an enough amount . but this is not the case here !

what do you think about the provided mesh above ? is it good or not? because that is the best one i can obtain from ANSYS CFX until now!

Josh December 2, 2010 14:24

OK, I'll ignore the domain for now, but I can almost guarantee it's causing problems.

You can't properly judge a mesh from one picture or one simulation. You need to check its quality, refinement, and convergence. See:

For this type of simulation, I'd use a structured mesh, which the CFX built-in mesher cannot do. Use a program like ICEM. The advantages of a structured mesh are listed in the above link. It should be very easy to create a structured mesh for that domain due to its straight edges. At the very least, your boundary layer should be structured. Also, if the domain you're using represents a real device, then shouldn't the outside walls be viscous (and, therefore, have a refined boundary layer)? Whichever type of mesh you use, be sure to check the quality (e.g., aspect ratios, orthogonality... see link above).

You should create at least 3 meshes of varying density. Keep simulating until further mesh refinement does not significantly change the parameter(s) of interest (e.g., pressure distribution).

How's the convergence? What are your residuals like?

All times are GMT -4. The time now is 00:01.