"""Example: Approximate Model Parameters
This example demonstrates how to use the UncertainFlorisModel class to
analyze the impact of uncertain wind direction on power results.
"""
from time import perf_counter as timerpc
import matplotlib.pyplot as plt
import numpy as np
from floris import (
ApproxFlorisModel,
FlorisModel,
TimeSeries,
)
# Generate time series data using a random walk on wind speeds with constant wind direction
N = 5000
n_turbines = 25
# Random walk on wind speed with values between 5 and 20 m/s
ws = np.ones(N) * 10
for i in range(1, N):
ws[i] = ws[i - 1] + np.random.normal(0, 0.25)
if ws[i] < 5:
ws[i] = 5
if ws[i] > 20:
ws[i] = 20
time_series = TimeSeries(
wind_directions=270.,
wind_speeds=ws,
turbulence_intensities=0.06)
# Instantiate a FlorisModel and an ApproxFlorisModel
fmodel = FlorisModel("../inputs/gch.yaml")
afmodel = ApproxFlorisModel("../inputs/gch.yaml", ws_resolution=0.5)
# Set both models to an n_turbine layout and use the above time series
layout_x = np.array([i*500 for i in range(n_turbines)])
layout_y = np.zeros(n_turbines)
fmodel.set(layout_x=layout_x, layout_y=layout_y, wind_data=time_series)
afmodel.set(layout_x=layout_x, layout_y=layout_y, wind_data=time_series)
# Now time both runs to show the speedup from approximating the wind speed
start = timerpc()
fmodel.run()
end = timerpc()
print(f"FlorisModel run time: {end - start} s")
start = timerpc()
afmodel.run()
end = timerpc()
print(f"ApproxFlorisModel run time: {end - start} s")
# Plot the power output from both models
fig, ax = plt.subplots()
ax.plot(fmodel.get_farm_power(), label="FlorisModel")
ax.plot(afmodel.get_farm_power(), label="ApproxFlorisModel")
ax.set_xlabel("Time Step")
ax.set_ylabel("Farm Power [W]")
ax.legend()
ax.grid(True)
# Compare the expected power results
print(f"Expected power from FlorisModel: {fmodel.get_expected_farm_power()/1E6:0.2f} MW")
print(f"Expected power from ApproxFlorisModel: {afmodel.get_expected_farm_power()/1E6:0.2f} MW")
plt.show()
import warnings
warnings.filterwarnings('ignore')
