Bug: Setting the turbine type with a list causes power setpoints to be ignored

[paulf81]

Setting the turbine type with a list causes power setpoints to be ignored

I've noticed an issue where if the turbine type is provided via a list, power_setpoints applied will be ignored.

How to reproduce

Adding this a test file into the examples folder will show the issue

from floris import (
    FlorisModel,
    TimeSeries,
)
import numpy as np

# Create a FlorisModel instance with simple derating
fmodel = FlorisModel("inputs/emgauss.yaml")
fmodel.set_operation_model("simple-derating")

# make a time series of 2 wind speeds
time_series = TimeSeries(
    wind_speeds=np.array([8.0, 9.0]), wind_directions=270.0, turbulence_intensities=0.06
)

# Set to a 2 turbine layout
fmodel.set(layout_x=[0.0, 126 * 5], layout_y=[0.0, 0.0], wind_data=time_series)

# Set all power setpoints to 0
power_setpoints = np.zeros((2, 2))
fmodel.set(power_setpoints=power_setpoints)
fmodel.run()

# Correctly prints out a 2x2 matrix of 0s
print(fmodel.get_turbine_powers())

# Specify the turbine types using a list
fmodel.set(turbine_type=["nrel_5mw", "nrel_5mw"])
fmodel.set(power_setpoints=power_setpoints)
fmodel.run()
# Now ignores the power setpoints
print(fmodel.get_turbine_powers())

Relevant output

The first printout is (correct)

[[0. 0.]
 [0. 0.]]

The second (ignores the power setpoints)

[[1753954.45917917  678834.8317748 ]
 [2496427.86183577  987681.57314292]]

Floris version

Develop branch

System Information

macOS

[misi9170]

I think this is happening because the NREL-5MW, by default, uses the "cosine-loss" operation model. So, when fmodel.set(turbine_type=["nrel_5mw", "nrel_5mw"]) is executed, the operation model is set to "cosine-loss", which doesn't respond to a power setpoint.

It's a little unclear what the "correct" behavior is here. For now, I think running

# Specify the turbine types using a list
fmodel.set(turbine_type=["nrel_5mw", "nrel_5mw"])
fmodel.set_operation_model("simple-derating")
fmodel.set(power_setpoints=power_setpoints)
fmodel.run()

should work (i.e., specifying the "simple-derating" operation model after setting the turbine types).

[misi9170]

Alternatively, we could raise a warning if the existing operation model doesn't match the one that is set by setting turbine_type, and indicate that the user will need to set the operation model again if they still want the earlier one. We might need to be careful of a situation occurring similar to #1045 with warnings coming up in unexpected places if we do this though.

[misi9170]

@paulf81 , looking back through some old issues, do we consider this to be an issue still? Or just "user error", so to speak?

We could consider adding a warning, similar to that raised regarding the reference_wind_height, if a user sets a turbine_type that does not match the existing operation_model. I'm not sure this is necessary, and as I mentioned here, raising warnings can lead to a lot of warnings being printed unnecessarily.