How is the Turbine grid points defined?

[prith-gs]

Hello,

I'm using 'turbine_grid' for my Turbines and I have a quick question.
What does the turbine grid look like?

The bounds of the rectilinear grid is [-R,R] or [-0.5R,0.5R]?

The figure on the left has bounds [-R,R] and the one on the right is [-0.5R,0.5R].

I exported x,y,z from the sequential_solver and looks like the bounds are [-0.5R,0.5R].
I want to make sure if that's case.

Best,
Pritheesh

[misi9170]

Hi @prith-gs ,

That's right, if the solver: type: field of the Floris input file is set as turbine_grid, the points are placed in a square grid from -0.5R to 0.5R, as you found. Here is the definition of the turbine grid points where the disc_area_ratio is defined here (and note that radius_ratio is set to 0.5, which results in the half-radius-sized grid).

Misha

[prith-gs]

Hi @misi9170

Thanks for the quick response.

Why not use [-R,R] or [- 0.707 * R, 0.707 * R]?

Pritheesh

[prith-gs]

@misi9170

one more follow-up question.

If I have changed the disc_area_ratio to 1 but I want to get REWS only based on points inside the Rotor area.
So, I want to do something like np.mean(u_points[y^2 + z^2 < R]).

How can I do this?

[misi9170]

Hi @prith-gs ,

Regarding your first question---as you suggest, the reason not to use $[-R,R]$ it to avoid collecting points that are actually outside of the disk. In terms of using $[-R/\sqrt{2}, R/\sqrt{2}]$, I think this would be OK, but it may put too much importance on points right on the rotor edge, which do not actually contribute as much to power production (being at the blade tips). Still, $-R/2$ to $R/2$ is perhaps somewhat conservative in it's focus on the rotor center. The "furthest" points from the rotor center, in this case, are at a distance of $R/\sqrt{2}$ from the hub.

As to your follow-up question: I highly recommend you look into using the "turbine_cubature_grid" rather than the default "turbine_grid". This uses a radial grid (rather than a square grid) to sample the velocities across the rotor, and is expected to give a more precise integral approximation. We don't have good documentation for this yet, but we hope in future to feature the "turbine_cubature_grid" more prominently. This feature was introduced in #649 and is also used by default in the recently-improved turboparkgauss model #907 .

Other models can also be run with the turbine_cubature_grid. Results will likely change somewhat, and we need to do a bit more work on our side to ensure that everything is fully compatible with the turbine_cubature_grid, but I think it may be exactly what you are looking for. See also further discussion in #883.

If you'd rather stick with the using the "turbine_grid" default for within the wake calculations, but want a higher resolution for the REWS only in post-processing (i.e., not in the actual wake calculation), you could look at defining a series of points across the rotors with the FlorisModel.sample_flow_at_points() method, supplying arguments x, y, and z to be the points across the rotor you are interested in.

Misha

[prith-gs]

Thanks very much @misi9170 !

I'll check out the cubature grid and looks like that's exactly what I need as well.
Appreciate the help!

Best,
Pritheesh