[Doriris]

Hello,

My supervisors and I have decided we should use SU2 as the tool to use in my thesis. After gaining some experience in SU2, the end goal with SU2 for my thesis is to perform a hypersonic CFD calculation of a low angle of attack, high lift over drag reentry vehicle. But, before I get to that stage, I want and need to do a simple validation studies of the code.

What I would like to do is do a supersonic calculation of a sphere, attached to a cylinder (to overcome stability issies with a Karman wake). First I would like to perform an Euler calculation for first estimates and to see what effects the mesh has (i.e. capturing of the shock and refining the mesh there). Then, I would like to move on to a Navier-Stokes solution. The idea here is that this test case can be compared with literature.

Now, I've looked at the wedge tutorial in SU2, and thought of a similar problem, but then with a cylinder cone. Please view picture attached below:
http://imgur.com/bIe5yCW

The wedge problem uses several markers to indicate the supersonic inlet, the outlet and the Euler boundaries ("airfoil"). I think this is the part where I don't quite understand how SU2 works and where the code stops working. Below an image is attached of the marker definition:
http://imgur.com/QuFVWYc

Now, this will not work. SU2 gives an error as soon as it tries to calculate the first iteration:
http://imgur.com/guAcyqp

If you look at the "compute the surface curvature", you will see it produces a "NaN". Apparently, the curvature cannot be computed. Is that due to the faulty definition of markers? Has it got something to do with the orientaton of my geometry and mesh with respect to the XYZ?

Any help would largely be appreciated. It might be a simple thing, it might be a difficult thing. Either way, I'm eager to learn how to use SU2.

Thank you in advance for responding to this topic!

Jeffrey

[Doriris]

Update: The orientation of the mesh was faulty, so I fixed that. Now there probably remains the issue of setting the right markers as the solution blows up and errors arise at 34 iterations:

CSysSolve::dotProd(w[i+1],w[i+1]) - 0.0

[Doriris]

Silly me. Had a small definition error in my geometry file. All is good now. Silly silly me.

[fpalacios]

Originally Posted by Doriris

Silly me. Had a small definition error in my geometry file. All is good now. Silly silly me.

No problem at all,
Thanks a lot for your interest in SU2,
Best,
Francisco

[laurent_soucasse]

Hi Jeffrey,

I am very curious on how you solved your problem.
I always encounter the error:
"CSysSolve::modGramSchmidt: w[i+1] = NaN"
when I mesh a curved surface (I use the software gmsh).

How have to be specified the orientation of the mesh ?
thanks for your answer.
Laurent.

[Doriris]

Hello Laurent,

If you view the first picture I attached (http://imgur.com/bIe5yCW) you can see in the bottom right corner that the geometry is oriented in the XZ plane (it's 2D). The correct orientation would be to create the geometry in the YZ plane: I have simply exchanged the former Z location with the Y location. In the code block below you will find the corresponding GMSH *.GEO file contents:

Point(1) = {2,0,0,0.5};
Point(2) = {1.5,0,0,0.01};
Point(3) = {2,0.5,0,0.01};
Circle(1) = {2,1,3};


// Boundary
Point(101) = {0,0,0,0.5};
Point(102) = {3,0,0,0.5};
Point(103) = {3,3,0,0.5};
Point(104) = {3,0.5,0,0.01};
Circle(2) = {101,102,103};


Line(3) = {101,2};
Line(4) = {3,104};
Line(5) = {104,103};

Line Loop(13) = {2, -5, -4, -1, -3};
Plane Surface(14) = {13};


Physical Surface(12) = {14};

//Characteristic Length {101,1, 2, 3, 104, 103} = 1;

Physical Line("farfield") = {2, 5};
Physical Line("euler") = {4, 1};
Physical Line("symmetry") = {3};

Note that in this particular example I only use the farfield and euler boundary and not the supersonic inlet and the outlet markers!

I hope this helps.

Regards,

Jeffrey

[laurent_soucasse]

Thank you very much for your answer but I tried different orientations witout success.
I will send a general post and see if I can get an answer from SU2 staff.
Thank you again.
Laurent.