SU2_EDU and Airfoils at High Angles of Attack

anon_k · November 12, 2015, 16:16

Hi,

I tried out SU2 and SU2_EDU. I liked that SU2_EDU automatically meshes airfoils. However, I had a question. It seems like the meshes are meant for normal angles of attack (+/- 12 degrees or so). I was wondering about very high angles of attack (0 to 90 degrees). Would there be a way to have SU2_EDU make meshes appropriate for those AOAs as well. Also, could you create examples for very high angles of attack, so that we can get the solver to converge on a solution.

I'm envisioning the aforementioned features as new additions to a future release of SU2_EDU. Implementing these features would be great projects for students. These features would benefit me and I am sure many others. I need the data for aircraft propeller calculations.

A few notes; I need to look at all altitudes covered by the US Standard Atmosphere of 1976, Mach Numbers from .1 to 1.3, and AOAs of 0 to 90 degrees. So some of these scenarios can be challenging to converge on a solution. I have been taking the approach of averaging the oscillation of the solution coefficients, however, it would be nice if experts could implement something appropriate. As an example; say you are at 60,000 feet, 90 degrees AOA, and Mach 1.25, have SU2_EDU automatically converge on the average Cl, Cd, and Cm.

anon_k · February 23, 2016, 01:13

Hi,

I like that SU2_EDU has automatic airfoil meshing. I don't have much of a CFD background. I do, however, have a lot of FEA experience. In FEA, you do a mesh density study to see if the solution is converging. I wonder if something like that could be applied to SU2_EDU. The reason I ask is I am not getting good agreement to test data (even at low angles of attack). I would like to experiment with the mesh but there is no easy way to do that. To that end, I would like to be able to select a C or O grid mesh for both the Euler and RANS solvers. I would also like it if there was a setting from 1 to 5 for the mesh density. 1 being very coarse and 5 being very fine with a linear distribution in between. I will leave it to the CFD experts to determine what is coarse and fine. I am mainly looking for an easy way to experiment with the mesh and see if the solution begins to match the test data. Or which mesh matches the best. Lastly, it seems like an O grid would make more sense if you want to go from AOAs of 0 to 90 degrees. I don't see how a C grid would work well at high AOAs. But that is just my intuition and not any experience.

hlk · February 23, 2016, 03:32

Thanks for your question.
While SU2 uses a custom mesh format, it can also read in CGNS grids which is a format that can be generated by most mesh generation software, including free meshing software.
For an example of using a CGNS mesh, one of the tutorials includes that feature:
https://github.com/su2code/SU2/wiki/...personic-Wedge

November 12, 2015, 16:16	SU2_EDU and Airfoils at High Angles of Attack	#1
anon_k Disabled Join Date: Jul 2015 Posts: 7 Rep Power: 11	Hi, I tried out SU2 and SU2_EDU. I liked that SU2_EDU automatically meshes airfoils. However, I had a question. It seems like the meshes are meant for normal angles of attack (+/- 12 degrees or so). I was wondering about very high angles of attack (0 to 90 degrees). Would there be a way to have SU2_EDU make meshes appropriate for those AOAs as well. Also, could you create examples for very high angles of attack, so that we can get the solver to converge on a solution. I'm envisioning the aforementioned features as new additions to a future release of SU2_EDU. Implementing these features would be great projects for students. These features would benefit me and I am sure many others. I need the data for aircraft propeller calculations. A few notes; I need to look at all altitudes covered by the US Standard Atmosphere of 1976, Mach Numbers from .1 to 1.3, and AOAs of 0 to 90 degrees. So some of these scenarios can be challenging to converge on a solution. I have been taking the approach of averaging the oscillation of the solution coefficients, however, it would be nice if experts could implement something appropriate. As an example; say you are at 60,000 feet, 90 degrees AOA, and Mach 1.25, have SU2_EDU automatically converge on the average Cl, Cd, and Cm. Last edited by anon_k; November 12, 2015 at 20:08.

February 23, 2016, 01:13	Meshing Options	#2
anon_k Disabled Join Date: Jul 2015 Posts: 7 Rep Power: 11	Hi, I like that SU2_EDU has automatic airfoil meshing. I don't have much of a CFD background. I do, however, have a lot of FEA experience. In FEA, you do a mesh density study to see if the solution is converging. I wonder if something like that could be applied to SU2_EDU. The reason I ask is I am not getting good agreement to test data (even at low angles of attack). I would like to experiment with the mesh but there is no easy way to do that. To that end, I would like to be able to select a C or O grid mesh for both the Euler and RANS solvers. I would also like it if there was a setting from 1 to 5 for the mesh density. 1 being very coarse and 5 being very fine with a linear distribution in between. I will leave it to the CFD experts to determine what is coarse and fine. I am mainly looking for an easy way to experiment with the mesh and see if the solution begins to match the test data. Or which mesh matches the best. Lastly, it seems like an O grid would make more sense if you want to go from AOAs of 0 to 90 degrees. I don't see how a C grid would work well at high AOAs. But that is just my intuition and not any experience.

February 23, 2016, 03:32		#3
hlk Senior Member Heather Kline Join Date: Jun 2013 Posts: 309 Rep Power: 14	Thanks for your question. While SU2 uses a custom mesh format, it can also read in CGNS grids which is a format that can be generated by most mesh generation software, including free meshing software. For an example of using a CGNS mesh, one of the tutorials includes that feature: https://github.com/su2code/SU2/wiki/...personic-Wedge