CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   SU2 (http://www.cfd-online.com/Forums/su2/)
-   -   edge vector vs. normal vector (http://www.cfd-online.com/Forums/su2/117990-edge-vector-vs-normal-vector.html)

Mojtaba M May 18, 2013 18:46

edge vector vs. normal vector
 
Dear SU2 developers
I am considering using SU2_CFD for some unstructured problems. However I happen to face some difficulties understanding some parts of the code. I would be very grateful if you could help me.
1-- From the paper AIAA 2013-0287 - fig 6 it seems that the PEBI grid is used (therefore edge unit vector is the same as face unit normal) however from the paragraph just above of the figure (about the definition of the dual grid) and also based on what I get from the code they are different. Based on my information there should be 2 sub-faces connecting each edge (in 2D) both of which have their own normal and the fluxes are integrated on some quadrature points on each of these sub-faces. However, from what I see in the code, normal vectors of these two faces are added to give one normal for the whole face (summation of 2 sub-faces) which should be different in the direction from edge vector connecting two vertices in the primal grid.
2-- On the other but not completely unrelated subject, I wonder what vector "s" is in eq (35) paper AIAA 2013-0287 as its definition is quite strange to me.

fpalacios May 27, 2013 21:23

Quote:

Originally Posted by Mojtaba M (Post 428482)
Dear SU2 developers
I am considering using SU2_CFD for some unstructured problems. However I happen to face some difficulties understanding some parts of the code. I would be very grateful if you could help me.
1-- From the paper AIAA 2013-0287 - fig 6 it seems that the PEBI grid is used (therefore edge unit vector is the same as face unit normal) however from the paragraph just above of the figure (about the definition of the dual grid) and also based on what I get from the code they are different. Based on my information there should be 2 sub-faces connecting each edge (in 2D) both of which have their own normal and the fluxes are integrated on some quadrature points on each of these sub-faces. However, from what I see in the code, normal vectors of these two faces are added to give one normal for the whole face (summation of 2 sub-faces) which should be different in the direction from edge vector connecting two vertices in the primal grid.
2-- On the other but not completely unrelated subject, I wonder what vector "s" is in eq (35) paper AIAA 2013-0287 as its definition is quite strange to me.

Hi,

In order to minimize the operations without sacrificing the accuracy of the method, SU2 adds the normal vector of the 2 sub-faces, otherwise you should compute the convective, and viscous fluxes in the middle of each sub-face. The vector s is the vector tangent to edge that connects two vertices, which is different from the normal vector of a face (summation of 2 sub-faces).

Cheers,
Francisco


All times are GMT -4. The time now is 02:18.