Add virtual turbine plotting method

[paulf81]

Feature or improvement description
Some wake models (cc, turbopark) are not yet set up to enable visualization of the horizontal flowfield. However this visualization can be useful in understanding, sanity checking etc. This pull request provides a function to calculate the flowfield, through a slower process, of sweeping an extra turbine around the domain in a nested for loop and extracting the u-velocity at hub height. As the turbine is moved throughout the domain, the calculated velocities at the turbine are stored in a local variable and the turbine's wake is not applied to the flow field. For the same resolution, this produces the same results as the FlorisInterface.calculate_xx_plane methods. However, this method is significantly slower since it does a full flow field calculation including iterating over each turbine for each visualization grid point.

Pull request comes with demo example to show the new method:
calculate_horizontal_plane_with_turbines

Which at at upper most level behaves much like calculate_horizontal_plane except height can not be freely specified. Results in plot:

Related issue, if one exists
Issue #355

[Bartdoekemeijer]

Hi @paulf81. I like this idea of having an artificial turbine in FLORIS function as a probe. I ran the example and saw it run very slowly on my machine. I think this is because of the overhead coming from reinitializing the FLORIS object for every measurement (in the example, we reinitialize FLORIS 600 times). Simply reinitializing floris 600 times on my laptop took 222.75 s.

Here's an idea based on yours, but circumventing the reinitialize issue. What if we add one turbine for every probe measurement? So if we want to calculate a grid of 40 x 15 wind speeds, we add 600 turbines to the farm. We make these turbines of the turbine_type=probe, give them a rotor diameter of -1, and a thrust curve of zeros. We can change that probe's location and hub height to specify the (x, y, z) coordinate of every measurement. Then, in solver.py, at the very start of the loop:

floris/floris/simulation/solver.py

Line 383 in a23241b

for i in range(grid.n_turbines):

we can add

if farm.rotor_diameters[i] < 0.001:
    continue

so that these probes have zero effect on the flow field. This would also enable users to handpick measurement locations, rather than always being forced to a grid. It would also resolve the issue with requiring a separate flowfield solver function, like full_flow_cc_solver.

@bayc @rafmudaf Feel free to let me know if this is a horrible idea. I understand it would go against some principles in how you designed the FLORIS model.

[paulf81]

I think that's really brilliant @Bartdoekemeijer , I really like it!

[Bartdoekemeijer]

Okay I just opened a pull request here: paulf81#12

It's a very primitive version of what the turbines-as-probes method that I was trying to explain. It's a little iffy but it mostly works. Try running the example code that I added and see what you think.

[rafmudaf]

TODO:

• Remove example
• Move plotting function to floris.tools.visualization

[rafmudaf]

This PR depends on #569