How to get su2 meshes from NASA's plot3D/cgns grids

OVS · December 7, 2015, 20:52

Hello,

I'm kinda new with CFD so please don't be too harsh on me. I have read a considerable amount of information on it and on SU2 though but I lack some practice.

For a student homework, I'm doing some simple validations concerning the SA turbulence model on SU2. I'm trying to reproduce NASA's NACA0012 airfoil validation case.

http://turbmodels.larc.nasa.gov/naca0012_val.html

I was thinking on doing my own grids on icem but then I saw they released readily available grids. The problem though is that

(1) the 2D grids are in a special format which I've never seen before and the 3D cgns grids are in the "wrong" coordinate system (plus they are on 3D).

(2) How do I add markers to these grids to identify the farfield and the airfoil? I tried importing a mesh into ICEM CFD but apparently you can only mark points...

Thanks for your help!

Oliver

guxin7005 · February 19, 2016, 14:13

https://confluence.cornell.edu/displ...arning+Modules

Here you can find a lot example about using those mesh to simulate.both 2d cases and 3d cases.

EMolina · February 23, 2016, 00:13

Did you tried the naca0012 grids provided by the SU2 team in github? They should perform pretty well...

guxin7005 · February 23, 2016, 09:30

Quote:

Originally Posted by EMolina

Did you tried the naca0012 grids provided by the SU2 team in github? They should perform pretty well...

That is awesome!!

by the way, do you know where can I find 3d naca0012 with wingtip mesh ?

OVS · February 29, 2016, 12:07

Yes I did. I found they had done similar validation cases.

I used the 897x257 meshes. Although I find their .cfg files have a VERY STRICT convergence criteria which increases the time calculation a lot. I based my .cfg on the one found in the source files (TestCases/rans/naca0012/turb_NACA0012_sa.cfg)

Their convergence criteria is Residual with 10 orders of magnitude reduction (or a 1e-12 minimum value) which is A LOT, asking around I was told industry would use 6 orders of magnitud at max. Calculation ran for ages before I had to stop it. I found out results weren't as good unless this restrictive convergence was used...

Using the TestCase, the convergence comes after 186636 iterations (95.5h on an 8 core machine)

CL and CD values for test cases were close enough at low angles of attack, but when I moved to 15 it didn't really fit all that well until harsh convergence criteria were used (CL = 1.55649 CD = 0.02237 ). Compared to the range proposed by NASA's study it is very close, even in range for the CL (1.5446 - 1.5642), but a tad bit high on the CD (0.02073 - 0.02159). Cd is more than 3% higher at 15 degrees than the maximum Cd given by NASA.

My guess is that it may have something to do with the Cf calculation near the walls...

December 7, 2015, 20:52	How to get su2 meshes from NASA's plot3D/cgns grids	#1
OVS New Member Oliver V Join Date: Dec 2015 Posts: 17 Rep Power: 10	Hello, I'm kinda new with CFD so please don't be too harsh on me. I have read a considerable amount of information on it and on SU2 though but I lack some practice. For a student homework, I'm doing some simple validations concerning the SA turbulence model on SU2. I'm trying to reproduce NASA's NACA0012 airfoil validation case. http://turbmodels.larc.nasa.gov/naca0012_val.html I was thinking on doing my own grids on icem but then I saw they released readily available grids. The problem though is that (1) the 2D grids are in a special format which I've never seen before and the 3D cgns grids are in the "wrong" coordinate system (plus they are on 3D). (2) How do I add markers to these grids to identify the farfield and the airfoil? I tried importing a mesh into ICEM CFD but apparently you can only mark points... Thanks for your help! Oliver

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Tool to convert OpenFoam mesh to SU2 mesh (2D and 3D)	Combas	SU2	44	March 12, 2024 00:35
[Other] plot3DToFoam: converting backstep grids from NASA's website	samiam1000	OpenFOAM Meshing & Mesh Conversion	2	January 2, 2015 16:51
Tool to download: SU2 post-processing	Combas	SU2	2	June 5, 2014 14:55
EnGrid makes SU2 grids	Speng	SU2	4	October 31, 2013 15:59
Welcome to the Stanford University Unstructured (SU2) forum!	economon	SU2	0	January 7, 2013 02:48

February 19, 2016, 14:13		#2
guxin7005 Member xin Join Date: Dec 2013 Posts: 33 Rep Power: 12	https://confluence.cornell.edu/displ...arning+Modules Here you can find a lot example about using those mesh to simulate.both 2d cases and 3d cases.

February 23, 2016, 00:13		#3
EMolina Member Eduardo Molina Join Date: Sep 2010 Location: Brazil Posts: 35 Rep Power: 15	Did you tried the naca0012 grids provided by the SU2 team in github? They should perform pretty well...

February 29, 2016, 12:07		#5
OVS New Member Oliver V Join Date: Dec 2015 Posts: 17 Rep Power: 10	Yes I did. I found they had done similar validation cases. I used the 897x257 meshes. Although I find their .cfg files have a VERY STRICT convergence criteria which increases the time calculation a lot. I based my .cfg on the one found in the source files (TestCases/rans/naca0012/turb_NACA0012_sa.cfg) Their convergence criteria is Residual with 10 orders of magnitude reduction (or a 1e-12 minimum value) which is A LOT, asking around I was told industry would use 6 orders of magnitud at max. Calculation ran for ages before I had to stop it. I found out results weren't as good unless this restrictive convergence was used... Using the TestCase, the convergence comes after 186636 iterations (95.5h on an 8 core machine) CL and CD values for test cases were close enough at low angles of attack, but when I moved to 15 it didn't really fit all that well until harsh convergence criteria were used (CL = 1.55649 CD = 0.02237 ). Compared to the range proposed by NASA's study it is very close, even in range for the CL (1.5446 - 1.5642), but a tad bit high on the CD (0.02073 - 0.02159). Cd is more than 3% higher at 15 degrees than the maximum Cd given by NASA. My guess is that it may have something to do with the Cf calculation near the walls...