[KevinW]

Hi guys.
This is a 2D airfoil mesh I'm using in CFX. However, the simulation result is not so acccurate. And I wanna check if the mesh contributes to the error.
Here is it. I'm using k-omega SST, the Yplus equals 0.2, which is Ok. The orthogonal quality, determiant, angle, as you see, is fine.
Well, I have the following problems:
1) The aspect ratios. The ARs around the airfoil surface are all below 6000. But a large number of cells extending from the trailing edge have high ARs. It seems that this cannot be eliminated if you wanna set the distance from the first cell to the wall quite small, can it?
2)I saw someone say the rule of thumb is that aournd the airfoil ARs even could have a maximum value of 10000, but high ARs should be got rid of in the far field. Are they correct? So how about those high AR cells in my grid, do they impact on the result? Any improvement I can do?

Thank you in advance.

[KevinW]

Here is more.

[KevinW]

By the way, problem 3: What are those redundant cells outside the domain shown in picture 5?

[siw]

I assume you know that CFX cannot solve 2D simulations. Have you extruded the mesh so that it is 1 element thick?

[KevinW]

Originally Posted by siw

I assume you know that CFX cannot solve 2D simulations. Have you extruded the mesh so that it is 1 element thick?

Sure, the calculation converges, but the monitor point Cl,Cd is not accurate.
More specificly, I extrude it with one element whose spacing equals the same order of the smallest length scale of the elements, which in my case is 0.001mm.

[siw]

High aspect ratio elements extending downstream from the sharp trailing edge are an unfortunate outcome of a C-grid like this. What about changing to an O-grid instead (e.g. http://www.grc.nasa.gov/WWW/5810/rvc/grape/GK-xyz.png) if you don't like those high AR elements.

Did you see PSYMN's youtube videos for this geometry (https://www.youtube.com/watch?v=tYrbScUH9RE)?

High AR elements are ok in the boundary layer and the CFX guide (cannot remember where exactly) says something like 10^6 is ok.

What's the Reynolds number, Mach number and angle of attack? Are you running CFX in double precision mode? Are you using the High Resolution advection scheme?

[Far]

Originally Posted by KevinW

Sure, the calculation converges, but the monitor point Cl,Cd is not accurate.
More specificly, I extrude it with one element whose spacing equals the same order of the smallest length scale of the elements, which in my case is 0.001mm.

Export it as Fluent mesh and import into CFX. It will automatically extrude it by one element.

[Far]

Originally Posted by KevinW

By the way, problem 3: What are those redundant cells outside the domain shown in picture 5?

I guess these are the VORFN.

Originally Posted by KevinW

Hi guys.
2)I saw someone say the rule of thumb is that aournd the airfoil ARs even could have a maximum value of 10000, but high ARs should be got rid of in the far field. Are they correct? So how about those high AR cells in my grid, do they impact on the result? Any improvement I can do?

Thank you in advance.

Did I say that ever

You can open up mesh in the farfield.

[KevinW]

Originally Posted by siw

High aspect ratio elements extending downstream from the sharp trailing edge are an unfortunate outcome of a C-grid like this. What about changing to an O-grid instead (e.g. http://www.grc.nasa.gov/WWW/5810/rvc/grape/GK-xyz.png) if you don't like those high AR elements.

Did you see PSYMN's youtube videos for this geometry (https://www.youtube.com/watch?v=tYrbScUH9RE)?

High AR elements are ok in the boundary layer and the CFX guide (cannot remember where exactly) says something like 10^6 is ok.

What's the Reynolds number, Mach number and angle of attack? Are you running CFX in double precision mode? Are you using the High Resolution advection scheme?

Hi, thank you for you reply first.

I haven't tried a O-grid, maybe I can give it a shot. Since many literature claim they are successful in utilizing C-grid for this S809 airfoil.

Well, bad news is that Youtube is blocked here, good news is that here is not DPRK. If you refer to Mr.Simon Pereira's tutorial(3 parts) on airfoil meshing, I'm lucky enough to have gone it through. My grid is following his tutorial.

In my case, Re=2*10^6, AOA=1.02, I've tried both incompressible and compressible condition, the difference of the results are negligible, since the Mach number ranges from 0 to 0.199 in the compressible case, so I just regard the low-speed flow as incompressible. Yes, with double precision and high resolution advection scheme.

Well, my Cl=0.24<0.27,Cd=0.14>0.07, the latter are experimental results. I can show you the Cp by the way.

[KevinW]

Originally Posted by Far

I guess these are the VORFN.



Did I say that ever

You can open up mesh in the farfield.

Yes, they are VORFN blocks, how to get rid of them? Since they account for more high ARs which influences my judgement, and they are not good-looking.

Someone mentioned this in this forum. What do you mean by saying "You can open up mesh in the farfield" exactly? I don't understand.

Thank you.

[cesarcg]

Originally Posted by KevinW

Someone mentioned this in this forum. What do you mean by saying "You can open up mesh in the farfield" exactly? I don't understand.

He means to open up the edges of your blocking that project from the trailing edge to the far field. This could help you reduce the high aspect ratios that result in the far field.

By the way, I can see from picture 8 in the post #2 that you are not imposing a smooth change in cell size. I mean that it appears that the size from the cells at the trailing edge change too abruptly to a significant small cell immediately downstream.

I hope I explained myself well enough.

Regards,
César