CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   2-D Airfoil Simulation (http://www.cfd-online.com/Forums/fluent/108582-2-d-airfoil-simulation.html)

OSUStudent October 26, 2012 17:18

2-D Airfoil Simulation
 
Hello,

I am working on a senior capstone project and need to analyze several airfoil sections in Fluent. I am using ANSA to generate the meshes. The solution methods and solution controls are as follows:

-Methods-
Scheme>SIMPLE
Gradient>Gree-Gauss Node Based
Pressure>Standard
Density>Second Order Upwind
Momentum>Second Order Upwind
Energy>Second Order Upwind

-Controls-
Pressure>0.3
Density>1
Body Forces>1
Momentum>0.7
Energy>1

The far-field boundary is defined as a pressure far-field and the airfoil geometry is defined as a wall. Currently the solution is not converging, the residuals drop initially but then they slowly start to either level off or rise. Does anyone have any tips on setting Fluent parameters?

Thanks!

Rhyno466 October 26, 2012 19:54

Please post some good pictures of your mesh before we go any further.

stuart23 October 27, 2012 09:25

You can increase your URFs if its converging and then not doing much more. Also, plot your drag and lift coefficients, they will help you decide if your solution is well converged.

Convergence is a subjective thing. Just because the default setting says something is converged when the residuals get below 10^-4, doesn't mean a sloution is converged.


Stu

Far October 27, 2012 11:34

Quote:

Originally Posted by stuart23 (Post 388821)
You can increase your URFs if its converging and then not doing much more. Also, plot your drag and lift coefficients, they will help you decide if your solution is well converged.

Convergence is a subjective thing. Just because the default setting says something is converged when the residuals get below 10^-4, doesn't mean a sloution is converged.


Stu

Very true. May be for some one 1e-4 is converged solution and for other case you may require 1e-6. Check your main parameters with both convergence criteria and see if there is any change in solution.

vicarious October 28, 2012 16:25

What is your Mach number? If you determine the both inlet and outlet boundaries as pressure far field, then you can calculate the inlet and outlet static pressure based on the total pressure of your case and enter them in the panel. There will be no problem if you do so since I have done the similar case before.
P.S : Try first order schemes for initial calculations and capture Y+ of your grid. You may need too refine your mesh as well.

Hope it will help with your problem,
Best regards.

OSUStudent November 1, 2012 17:50

3 Attachment(s)
Thank you for the help!

Here are some pictures of the mesh I am using, I don't have that much experience using ANSA and am not really sure how to determine the quality of the mesh.

vicarious November 1, 2012 18:04

You better consider structured quadrilateral grid close to the wall to resolve the boundary layer more accurately. outside the boundary layer You can use either unstructured triangular or tetrahedral grid.

OSUStudent November 1, 2012 18:22

3 Attachment(s)
I know how to check the mesh quality in FLUENT where it gives you a value between 0 and 1 but is there a way to do it in ANSA?

vicarious November 1, 2012 18:32

I'm afraid I don't have information about ANSA. But since your quality does not exceed more than 0.9 the mesh is fine. If the skewness gets too high, FLUENT will notify you in command window. Your recent mesh is fine. You just need to keep Y+ low.

OSUStudent November 1, 2012 18:45

Ok, thank you. In reference to you earlier question my velocity is M=0.02035, I have read elsewhere that if the velocity gets too low then the calculations won't work out.

vangelis November 6, 2012 08:57

Just some tips for the use of ANSA

When you mesh with CFD meshing algorithm make sure
that you specify a Maximum Length at the Options window
at the bottom right of the ANSA GUI.
In this way the mesh will not grow in size inside the macro
above the value that you set.

You can find bad elements if you activate HIDDEN button
(near SHADOW at the bottom)

If you start ANSA in CFD mode then the default Fluent
skewness values are already used. ANSA will report elements
failing this limit as OFF and will display them in green or red color
if the fail Fluent equiarea or equiangle skewness respecitively.

Finally in order to have a nice spacing near the wall you could
make some cuts in the inner macro, and then use the functions
PERIMETERs>NUMBER and SPACING to aling the nodes and use
MAP QUAD mesh.

Vangelis

OSUStudent November 8, 2012 20:23

4 Attachment(s)
Thank you for the reply.

I was able to create a quad map mesh in the boundary layer and checked the mesh using the HIDDEN button like you described but I am still having convergence issues within fluent. My current mesh has about 650K points and fluent reports its quality as 0.60. At this point I think the problem might be something I'm doing (or not doing) in fluent. Do you have any suggestions for this problem?

Far November 9, 2012 00:03

did you check the scale of problem in Fluent? What about the boundary conditions?

cfd seeker November 9, 2012 02:20

Quote:

Originally Posted by OSUStudent (Post 391112)
Thank you for the reply.

I was able to create a quad map mesh in the boundary layer and checked the mesh using the HIDDEN button like you described but I am still having convergence issues within fluent. My current mesh has about 650K points and fluent reports its quality as 0.60. At this point I think the problem might be something I'm doing (or not doing) in fluent. Do you have any suggestions for this problem?

Strange behavior of residuals,there is something wrong in the setup of problem. From which zone you have initialized the problem?


All times are GMT -4. The time now is 15:58.