[Joachim]

Hey everyone!

I am trying to generate an O-grid using Pointwise, but I am having some orthogonality issues (see Fig. init).

I am using the elliptical solver, with 'current grid' for the spacing and 'orthogonal' for the angle (wall BC). However, when I run a few iterations, I end up with a highly stretched grid (see Fig. stretching). What should I do to correct the orthogonality of my grid, while maintaining reasonable stretching ratios?

Thank you very much!

Joachim

[tcarrigan]

Because you are creating an O-grid, is there a reason you've elected to generate by hand and use the elliptic solver to smooth rather than generate mesh via hyperbolic extrusion?

[Joachim]

I want to look at the vortex shedding behind the airfoil, so I need more points in the wake. On the other hand, I have a blunt trailing edge, and I need to resolve it so that I can't use a C-grid.

Actually, I am now trying to use hyperbolic extrusion for the boundary layer, and then transitions to the elliptic solver for the outer regions...

is there another way to do it?

Thanks!

[tcarrigan]

In fact, you can use a C-grid. A C-grid is a great way to get more points in the wake and can be generated via a single extrusion. The blunt tip requires just a little special treatment. You can see an image of a c-grid that's been created with a single hyperbolic extrusion here https://github.com/traviscarrigan/SU2-NACA2412-CGrid.

What you'll do is create a thin domain extending off the trailing edge that travels all the way to the outlet downstream. You then select the the airfoil surfaces and top and bottom connectors for that thin domain and perform a normal hyperbolic extrusion.

To get this working properly, ensure the boundary conditions for the extrusion are set so the downstream portion doesn't splay but rather is constrained to a constant coordinate. Start off with a low growth rate and a low value of volume smoothing. After a few layers you'll then be able to turn up the volume smoothing which will allow for the nice c-shape farfield.

If you prefer using the elliptic solver, I've had better luck setting the spacing to an explicit value (initial ds) rather than current grid. Do this and only run it a couple iterations. I find that if I must run the smoother more than 50 iterations, I should go back and adjust my overall topology.

[Joachim]

haha thanks for the advice!
However, I kept it simple in the post. In reality, I am trying to simulate an airfoil in reverse flow, so that the wake is actually before the geometric leading edge...That's why I am using an O-grid...