One hint would be to check if the reference area (or values) you used to obtain cl values in your simulation is consistent with the reference study or not.

If you are running a turbulent simulation, do think about the wall y+ requirements required by the turbulence model you are using.

Hi Awais Ali,

I made a slight adjustment to my question above as I had a typo in the original Cl value.

Thanks for your help.

The model that I am comparing against is similar to the one shown here: http://turbmodels.larc.nasa.gov/naca0012_val.html, except I used a domain size of 50 chord lengths and not 500. I am using an area of 1 for the airfoil. which is in keeping with the model.

Also, as I am using a Reynolds Number of 6million, and a Realizable k-e model with standard wall functions, with in a wind velocity of 87.64m/s at 8 degrees. In keeping with the study outlined here, http://www.academicjournals.org/article/article1379753908_Eleni%20et%20al.pdf, I used a y+ value of 0.2, which results in a first cell height of 8.9E-7.

The only other things that can influence your results would be the Numerical Schemes. Try comparing the coefficient of pressure (Cp) for the airfoil surface, if there is any data available.

Thanks Awais Ali for the advice. I'm not certain what you mean by the Numerical Schemes.

The following are the steps and conditions I have used to date:

- created the geometry in Ansys Geometry, imported to Ansys ICEM.

- The computational domain is similar to this: http://turbmodels.larc.nasa.gov/naca0012_val.html, except I used 50 chord lengths instead of 500.

- I used the following boundary conditions: Velocity Inlet, Pressure Outlet, and symmetry at the upper and lower boundaries.

- Exported the model to Ansys Fluent.

- In the "Models" tab, I used Realizable k-e model, with standard wall functions.

- Calculating Cl for an angle of attack of 8 degrees, a chord length of 1m, Reynolds number of 6 million. Therefore, in the "Boundary Conditions" tab, I used velocity magnitude 87.64 m/s, x-direction flow direction (cos alpha)=.9903 and y-direction flow direction (sin alpha) = 0.1391

- In "Reference Values" tab, I used area = 1 (m2), density of air=1.225 (kg/m3), and viscosity 1.7894e-5 (kg/m-s), temp 288.16K

- "Solution Methods" tab: Simple, and the remainder default.

- "Monitors" tab: Convergence of all residuals at 1e-6

- "Monitors" tab: Cd-1: Force Vector X = 0.9903, Force Vector Y = 0.1391

- "Monitors" tab: Cl-1: Force Vector X = -0.1391, Force Vector Y = 0.9903

- "Solution Initialisation" tab: Compute from Inlet -> Initialise

- Run Calculation

Also, in your experience, would you expect that a scheme like this would take a few hours or a few days to run until convergence? I've had the computer running over night without obtaining an answer, so I am just wondering if this is normal?

The best way to analyse is export the x,y,z, field data file to any other file and then perform the calculation from the data without relying on Fluent, unless you have properly assigned the reference length scales in FLUENT

Thank you both for your assistance to date. It has been a great help.

I deleted all my previous work and started again. I found that when I used a simpler mesh and reduced the number of nodes, similar to the one shown here (https://www.youtube.com/watch?v=EknKVAJGEJ8), I obtained a coefficient of lift of .54 instead of ;65. Whilst this isn't ideal, it is a lot closer to the value that I wanted. I did not use the "split block" option in ICEM Meshing to refine the mesh around the airfoil - each time I used this, I got much lower values of the Coefficient of Lift as I outlined above.

However, then I ran into the problem as I have outlined below. I started with a mesh convergence study, and everything was going to plan (after a few initial teething problems), until I increased the number of nodes in the mesh above 65,000.

When I used 55,000 and 60,000 nodes, I got similar answers (Coefficient of lift=0.54). Although this was slightly lower than I expected (0.65), I was trying to increase the mesh one more time to ensure that it had converged, when I ran into this problem

When I used 70,000 nodes, for example, the residual values start to vibrate excessively after about 6000 iterations. Does anyone have an idea why this is the case, and could you please suggest a solution?

Thanks a lot for any help.

Aidan