Bug with y cut plane

[RRiva]

y cut plane does not seem to follow the wind direction

The documentation shows how to visualize a y cut plane. I've tried changing the wind direction, but 90 deg produces the same result as 270, while I was expecting the wake to go from right to left. The horizontal plane looks fine instead.

Furthermore, the documentation of calculate_y_plane was copy-pasted from calculate_horizontal_plane. You should fix crossstream_dist.

How to reproduce

I copy here the documentation and just change the wind direction.

import matplotlib.pyplot as plt

from floris import FlorisModel
from floris.flow_visualization import visualize_cut_plane


fmodel = FlorisModel("../inputs/gch.yaml")

# Set a 3 turbine layout with wind direction along the row
fmodel.set(
    layout_x=[0, 500, 1000],
    layout_y=[0, 0, 0],
    wind_directions=[90],
    wind_speeds=[8],
    turbulence_intensities=[0.06],
)

# Collect the yplane
y_plane = fmodel.calculate_y_plane(x_resolution=200, z_resolution=100, crossstream_dist=0.0)

# Plot the flow field
fig, ax = plt.subplots(figsize=(10, 4))
visualize_cut_plane(
    y_plane, ax=ax, min_speed=3, max_speed=9, label_contours=True, title="Y Cut Plane"
)

plt.show()

Relevant output

Floris version

v4.4.2

System Information

[misi9170]

Hi @RRiva ,

Thanks for the note! This is just to do with how the cut planes are visualized. The flow always appears "from the left", regardless of the wind direction used. To see this, here is some modified code, where I have moved the "first" turbine vertically to distinguish it from the others:

from floris import FlorisModel
from floris.flow_visualization import visualize_cut_plane


fmodel = FlorisModel("../inputs/gch.yaml")

# Set a 3 turbine layout with wind direction along the row
wind_direction = 270
fmodel.set(
    layout_x=[0, 500, 1000],
    layout_y=[0, 0, 0],
    wind_directions=[wind_direction],
    wind_speeds=[8],
    turbulence_intensities=[0.06],
)
turbines = fmodel.core.farm.turbine_definitions
turbines[0]["hub_height"] = 180.0
turbines[0]["turbine_type"] = "nrel_5MW_lifted"
fmodel.set(turbine_type=turbines)

# Collect the yplane
y_plane = fmodel.calculate_y_plane(x_resolution=200, z_resolution=100, crossstream_dist=0.0)

# Plot the flow field
fig, ax = plt.subplots(figsize=(10, 4))
visualize_cut_plane(
    y_plane, ax=ax, min_speed=3, max_speed=9, label_contours=True, title="Y Cut Plane"
)

plt.show()

When wind_direction = 270, this produces

whereas when wind_direction = 90, the generated figure is

[misi9170]

Thanks for the note about the documentation! Adding that to #1105

[RRiva]

Thanks for checking Misha, but now I'm even more confused 🙈 not only calculate_y_plane is always for 270 deg, but now the turbine height depends on the wind direction and isn't coherent with the layout.

[misi9170]

Haha no, the turbine height stays with the location. I think the "bug" here is that the x axis shown in the y cut plane is not the global $x$ value. Here, the turbine at (global) $(x,y) = (0,0)$ has a hub height of 180 m. When the flow is at 270 degrees, this is the first turbine in the row. When the flow is at 90 degrees, this turbine is the last turbine in the row. The flows are shown "from the left" either way, with the layout rotated to match, and the x axis is simply marking the downstream distance (not the global $x$ location).

[misi9170]

Note that in previous versions of FLORIS (older than v3.4), top-down visualizations (calculate_horizontal_plane) also showed all flows from the left and simply rotated the layout. #578 introduced the change to make the flow appear from the direction that it is "actually" coming from, but it's not clear how this should be done for calculate_y_plane. It's clear if the flow is from 270 or 90 (simply flow from the left or from the right), but not clear (to me at least) for any other angle. So, for now, in calculate_y_plane the flow always "appears" from the left and the layout it rotated to match that, regardless of the wind direction. This keeps things consistent across all wind directions, rather than making 90 a special case.

[RRiva]

Thanks for the clarification. To make things clearer I suggest to do the visualization using two reference frames:

• Downwind, crosswind and down ⬅️ this one rotates with the wind direction.
• North, east and down ⬅️ global frame that does not rotate.

In both cases, you can define planes orthogonal to each axis and plot accordingly.

What do you think about this?

[misi9170]

Yes, having two versions of each of the visualizations is a good idea, I think. To be clear, FLORIS already uses both of the reference frames you mentioned, with some minor reordering:

• $(x, y, z)$ = (East, North, Up) is the more "user facing" frame of reference. Layouts are specified using this reference frame, and wind directions (Clockwise from North) are defined in this frame. These are stored as Grid.x_sorted_inertial_frame, Grid.y_sorted_inertial_frame, and Grid.z_sorted_inertial_frame here.
• $(x, y, z)$ = (Downwind, Crosswind, Up) is the more "fundamental" frame of reference that FLORIS actually uses for wake calculations. These are stored as Grid.x_sorted, Grid.y_sorted, Grid.z_sorted here.

Perhaps new flow visualization tools could also address #1072 .

[misi9170]

I've now fixed the docstrings in FlorisModel.calculate_y_plane() and FlorisModel.calculate_cross_plane(), and added to the docstring in flow_visualization.visualize_cut_plane() to explain how the flows are plotted in #1171 . This should address the immediate issue raised here, although I think the idea of having an option to generate planes in the "user facing" reference frame, so I plan to leave this issue open with the "ideas-list" label.