[ashf9]

Hi,

I am doing a 3D RANS simulation on an airfoil under stall conditions. I've done a 2D simulation with the same boundary conditions and it converged to 1e-5. For the 3D simulation, I extruded the boundary by the span and set the walls to a periodic wall boundary condition to replicate endplates. However, my residuals are not converging past 1e-2, and I'm not too sure why? The turbulence model used is k-w and the Y+ along the airfoil is <1.

I understand unsteady RANS is usually required, but I'm trying to figure out how to get the residuals to converge, as this has been done before and worked with the 2D simulations.

Thanks

[fluid23]

My first question; is this a steady or unsteady analysis? If you are trying to model an airfoil under stall using a steady solver, you are wasting your time. Stall is the result of flow separation which typically results in a dynamic flow field.

[ashf9]

It is steady at the moment. So I presume for this simulation I would have to change the physics method to implicit unsteady.
I have seen some papers using steady RANS simulations before to capture stall cells, and I am trying to replicate this.

Originally Posted by fluid23

My first question; is this a steady or unsteady analysis? If you are trying to model an airfoil under stall using a steady solver, you are wasting your time. Stall is the result of flow separation which typically results in a dynamic flow field.

[fluid23]

That would be my recommendation, yes. However, if you are trying to duplicate data from a paper you should follow whatever methodology they have employed in said paper.

[ashf9]

So I am trying to follow that methodology but I am still getting relatively high residuals and difficulty in converging.

Originally Posted by fluid23

That would be my recommendation, yes. However, if you are trying to duplicate data from a paper you should follow whatever methodology they have employed in said paper.

[openfoam_aero]

Hi. So I am also doing the same thing. I have found that the spanwise spacing makes a huge difference. I am trying to run a case at 14 degree AoA at 400,000 Reynolds number with an AR (span/chord) or 2.5. I started with a span of 0.5 and increased it to 1 and 1.5 and then 2. I managed to get the number of stall cells = AR. The spanwise spacing is kind of trial and error so I am still working on that. I could share more details if you are interested. Or if you solved it, perhaps you could help me.
Btw, my residuals converged down to 1e-05