[Jansen]

Hi all,

I am trying to simulate air flow over a NACA0021 aerofoil using k-omega SST(chord = 70mm, span = 50mm, planform area = 0.0035 square metres) at different angles of attack as to try and match another experiment conducted in a 0.5m x 0.5m wind tunnel at a Reynolds number of roughly 120,000, free-stream velocity 25m/s and turbulence intensity 0.8%. There is quite reasonable agreement with regards to the drag coefficient of the aerofoil but the lift coefficient does not match the experimental values nearly as well, even when considering general error and other differences.

From viewing the flow in paraview it appears that the aerofoil stalls in the simulation well before the point that it stalls in the experiment. The residuals also seem to plateau at quite a high value too. I have tried changing many things around but to no avail.

I feel that the problem of the early onset of separation may still lie in the boundary conditions but I am unsure of which ones to change and what to change them to.

I have attached the graphs, a plot of the residuals and the relevant OpenFOAM files.

Thanks in advance.

Ben

[mechy]

Hi
your problem is related to your mesh.

1- for freestream boundary condition, your farfield should be 20 times of chord. in your mesh the domain of solution is very small

2-for sst k-w you should refine the mesh on the airfoil (y+<10)
I advise you to use the Spalart-Allmaras model for this airfoil with 30<y+<300

3-your mesh has very cell numbers and I advise you to use the 2D model

[Jansen]

Hey, thanks for your reply

I have a couple of questions regarding the mesh
1- Does the (y+<10) mean that I should refine the mesh so that there are more cells at the airfoil?
2- Regarding your third point; does my mesh have too few cells?

As for using a 2D model, the next stage in my project once I validate the baseline is to add a sinusoidal leading edge to the airfoil. Would it still be beneficial to use a 2D model for the baseline?

[mechy]

your mesh has very cell numbers and you should reduce it.
but you should use the inflation (grading) to refine the mesh on the airfoil.
if the sinusoidal leading edge is cover all of the span, I advise you to use the 2D model