bad accuracy CD in Euler simulations

jespero@kth.se · November 10, 2023, 09:14

We are running tests of mesh generators on a wing-body for student use.

When given a mesh from the open software DWFsumo from larosterna.com
the SU2 blackbird results of alpha-sweep 0-10, M 0.3 compressible Euler
are good.

A similar mesh from gmsh gives good CL but very bad CD, some < 0.
Flow visualization with paraview shows nothing untoward to the
naked eye.

The same gmsh msh when run with the Swedish national CFD program
M-Edge (now owned by Saab) on license to Swedish universities gives
again "good" results. M-Edge is a time-stepping
compressible solver sharing some basics with SU2.

I tried to attach config file but could not figure how.
Hints would be much appreciated. Obviously SU2 is more
mesh-sensitive than M-Edge.

bigfootedrockmidget · November 10, 2023, 10:54

Can you say a bit more about the difference in the meshes? It would help to have 2 cases available, exactly the same configuration, except different meshes, one with good performance, the other with bad performance. Maybe you can create a small github repository for it and give a link?

bene_ · December 12, 2023, 11:47

Hi @bigfootedrockmidget, here you can find the repository https://github.com/GBenedett/Negative_CD_SU2

Thanks for your help

bigfootedrockmidget · December 12, 2023, 16:50

Hi,
Can you also share some figures from the validation study?

Edit: would be nice to know what the actual values should be and how the other codes compare.

bigfootedrockmidget · December 12, 2023, 18:08

Hi,
I can reproduce the negative Cd. It is also visible when integrating the pressure field in paraview. Do you also have the sumo mesh that gives good results?
I did notice that the mesh quality around the leading and trailing edge is very coarse, actually the entire mesh is very coarse. Does the problem disappear when you refine the mesh?

bene_ · December 14, 2023, 05:50

Thanks for your quick reply, I also tried to get the cd from paraview and I get the same result as well. As for the sumo mesh, you can find it in the github repository, it is called "labARstraight_sumo.zip". As for the validation cases, I uploaded them to the same repository instead, they are called "m0.3a3_gmsh.zip" and "m0.3a3_sumo.zip". The link is still the same https://github.com/GBenedett/Negative_CD_SU2

Yes, the mesh quality is not very good, but we did it on purpose, of course we also tried with fine mesh, but the results are the same.

Thank you very much.

November 10, 2023, 09:14	bad accuracy CD in Euler simulations	#1
jespero@kth.se New Member jesper oppelstrup Join Date: Nov 2023 Posts: 1 Rep Power: 0	We are running tests of mesh generators on a wing-body for student use. When given a mesh from the open software DWFsumo from larosterna.com the SU2 blackbird results of alpha-sweep 0-10, M 0.3 compressible Euler are good. A similar mesh from gmsh gives good CL but very bad CD, some < 0. Flow visualization with paraview shows nothing untoward to the naked eye. The same gmsh msh when run with the Swedish national CFD program M-Edge (now owned by Saab) on license to Swedish universities gives again "good" results. M-Edge is a time-stepping compressible solver sharing some basics with SU2. I tried to attach config file but could not figure how. Hints would be much appreciated. Obviously SU2 is more mesh-sensitive than M-Edge.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
A good shcheme for 2D euler equations on carthesian grid	ander	Main CFD Forum	4	February 1, 2022 04:13
steadyState - Euler - different endresults	PSander	OpenFOAM Running, Solving & CFD	0	January 30, 2021 06:18
bad results when compared with other simulations	Le Stanc	CFX	12	November 8, 2006 02:47
y+ for NACA0012 transonic simulations	Oscar Arias	Main CFD Forum	7	June 9, 2006 02:27
Problems of Duns Codes!	Martin J	Main CFD Forum	8	August 15, 2003 00:19

November 10, 2023, 10:54		#2
bigfootedrockmidget Senior Member bigfoot Join Date: Dec 2011 Location: Netherlands Posts: 676 Rep Power: 21	Can you say a bit more about the difference in the meshes? It would help to have 2 cases available, exactly the same configuration, except different meshes, one with good performance, the other with bad performance. Maybe you can create a small github repository for it and give a link?

December 12, 2023, 11:47		#3
bene_ New Member Giacomo Join Date: Feb 2023 Posts: 4 Rep Power: 3	Hi @bigfootedrockmidget, here you can find the repository https://github.com/GBenedett/Negative_CD_SU2 Thanks for your help

December 12, 2023, 16:50		#4
bigfootedrockmidget Senior Member bigfoot Join Date: Dec 2011 Location: Netherlands Posts: 676 Rep Power: 21	Hi, Can you also share some figures from the validation study? Edit: would be nice to know what the actual values should be and how the other codes compare.

December 12, 2023, 18:08		#5
bigfootedrockmidget Senior Member bigfoot Join Date: Dec 2011 Location: Netherlands Posts: 676 Rep Power: 21	Hi, I can reproduce the negative Cd. It is also visible when integrating the pressure field in paraview. Do you also have the sumo mesh that gives good results? I did notice that the mesh quality around the leading and trailing edge is very coarse, actually the entire mesh is very coarse. Does the problem disappear when you refine the mesh?

December 14, 2023, 05:50		#6
bene_ New Member Giacomo Join Date: Feb 2023 Posts: 4 Rep Power: 3	Thanks for your quick reply, I also tried to get the cd from paraview and I get the same result as well. As for the sumo mesh, you can find it in the github repository, it is called "labARstraight_sumo.zip". As for the validation cases, I uploaded them to the same repository instead, they are called "m0.3a3_gmsh.zip" and "m0.3a3_sumo.zip". The link is still the same https://github.com/GBenedett/Negative_CD_SU2 Yes, the mesh quality is not very good, but we did it on purpose, of course we also tried with fine mesh, but the results are the same. Thank you very much.