failed validating Cd for 2D airfoil

Sereff · January 24, 2019, 10:52

Dear Foamers:

The case I was simulating is air flow around a 2D airfoil with Re = 1e6, with following:

Solver: pimpleFoam
Model: k-Omega SST
Meshing: snappHexMesh

I've been getting pretty covincing results for Cl, but not for Cd. I've read from some source that Cd is far more sensitive to mesh quality than Cl does. I assume that 'high mesh quality' means more refined mesh as in more cells and finer boundary layers?

And talking about boundary layers, I took a closer look at the yPlus value for my simulations on different mesh, please correct me if I was using wrong BCs,
case 1,30<y+<70, with boundary conditions on airfoil:

U: type noSlip;
turbulence properties: type xxWallFunction; value $internalField

case 2,0.09<y+<1.2, with boundary conditions on airfoil:

U: type noSlip;
turbulence properties: type fixedValue; value uniform 1e-15

But all my simulations are having trouble with Cd. Any idea?

Kind regards,

Sereff · January 24, 2019, 11:50

Attached are my Cd and Cl results for different aoa.

yambanshee · January 25, 2019, 02:41

Drag is a difficult property to get right in airfoil simulations. It is very dependent on the wall boundary layer, and so using a wall function will likely not work.

What is meant by a high mesh quality is details such as density, minimal grading near the walls, low non-orthogonality and skewness, and cells aligned to the direction of flow as much as possible.

Also, as the drag is so dependent on the boundary layer, accurate turbulence modelling at the boundary is also very important. Unfortunately, all RANS formulations are simply an approximation of turbulence effects, so this introduces enhanced difficulty in getting a correct boundary layer

Sereff · January 25, 2019, 03:44

Hi, thank you for your anwer.

So I assume with k-Omega-SST, it is expected to have a rather bad Cd prediction. But do you know what boundary condition should I use for velocity on the surface of airfoil? I've been told if wall function were to apply then I should use 'zeroGradient' for airfoil surface instead of 'noSlip'.

Kind regards,

yambanshee · January 25, 2019, 05:10

AFAIK if you are interested in the effect of the wall, a no slip condition is always needed

January 24, 2019, 10:52	failed validating Cd for 2D airfoil	#1
Sereff Member Sereff Join Date: Jan 2019 Posts: 49 Rep Power: 7	Dear Foamers: The case I was simulating is air flow around a 2D airfoil with Re = 1e6, with following: Solver: pimpleFoam Model: k-Omega SST Meshing: snappHexMesh I've been getting pretty covincing results for Cl, but not for Cd. I've read from some source that Cd is far more sensitive to mesh quality than Cl does. I assume that 'high mesh quality' means more refined mesh as in more cells and finer boundary layers? And talking about boundary layers, I took a closer look at the yPlus value for my simulations on different mesh, please correct me if I was using wrong BCs, case 1,30<y+<70, with boundary conditions on airfoil: U: type noSlip; turbulence properties: type xxWallFunction; value $internalField case 2,0.09<y+<1.2, with boundary conditions on airfoil: U: type noSlip; turbulence properties: type fixedValue; value uniform 1e-15 But all my simulations are having trouble with Cd. Any idea? Kind regards, Last edited by Sereff; January 24, 2019 at 15:04.

January 24, 2019, 11:50		#2
Sereff Member Sereff Join Date: Jan 2019 Posts: 49 Rep Power: 7	Attached are my Cd and Cl results for different aoa. Last edited by Sereff; January 28, 2019 at 05:09.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Foam::error::printStack(Foam::Ostream&) with simpleFoam -parallel	U.Golling	OpenFOAM Running, Solving & CFD	52	September 23, 2023 04:35
Initial conditions for uniform flow	andreas	OpenFOAM	5	November 16, 2012 16:00
[OpenFOAM] ParaView/Parafoam error when making animation	Disco_Caine	ParaView	6	September 28, 2010 10:54
user subroutine error	CFDUSER	CFX	2	December 9, 2006 07:31
user defined function	cfduser	CFX	0	April 29, 2006 11:58

January 25, 2019, 02:41		#3
yambanshee Senior Member Zander Meiring Join Date: Jul 2018 Posts: 125 Rep Power: 8	Drag is a difficult property to get right in airfoil simulations. It is very dependent on the wall boundary layer, and so using a wall function will likely not work. What is meant by a high mesh quality is details such as density, minimal grading near the walls, low non-orthogonality and skewness, and cells aligned to the direction of flow as much as possible. Also, as the drag is so dependent on the boundary layer, accurate turbulence modelling at the boundary is also very important. Unfortunately, all RANS formulations are simply an approximation of turbulence effects, so this introduces enhanced difficulty in getting a correct boundary layer

January 25, 2019, 03:44		#4
Sereff Member Sereff Join Date: Jan 2019 Posts: 49 Rep Power: 7	Hi, thank you for your anwer. So I assume with k-Omega-SST, it is expected to have a rather bad Cd prediction. But do you know what boundary condition should I use for velocity on the surface of airfoil? I've been told if wall function were to apply then I should use 'zeroGradient' for airfoil surface instead of 'noSlip'. Kind regards,

January 25, 2019, 05:10		#5
yambanshee Senior Member Zander Meiring Join Date: Jul 2018 Posts: 125 Rep Power: 8	AFAIK if you are interested in the effect of the wall, a no slip condition is always needed