Multi-height wind data

In this example we explore the calculation of multi-height wind data, as for example obtained from WRF results or downloaded from the NEWA website at a single point.

Here we will use the static data file WRF-Timeseries-4464.csv.gz that is part of the foxes static data. It has the following data structure:

Time,WS-50,WS-75,...,WS-500,WD-50,WD-75,...,WD-500,TKE-50,TKE-75,...,TKE-500,RHO
2009-01-01 00:00:00,7.37214,7.42685,...,1.28838
...
2009-01-31 23:50:00,10.27767,10.36368,...,1.30095

The time stamp column marks one month in 10 minute steps, and the wind speed (WS), wind direction (WD) and turbulent kinetic energy (TKE) are provided at 8 heights between 50 and 500 m. The air density (RHO) does not have height dependency but varies with time.

The basic assumption of this example is that we can calculate our wind farm results based on this data, i.e., that the horizontal variation can be neglected (for completely heterogeneous inflow data, see the corresponding example).

These are the imports for this example:

In [1]:

%matplotlib inline
import matplotlib.pyplot as plt
from xarray import open_dataset

import foxes
import foxes.variables as FV
import foxes.constants as FC

First, we setup the model book and the wind farm. We choose 5 turbines in a row:

In [2]:

# create wind farm, a single row of turbines:
farm = foxes.WindFarm()
foxes.input.farm_layout.add_row(
    farm=farm,
    xy_base=[0.0, 0.0],
    xy_step=[600.0, 0.0],
    n_turbines=5,
    turbine_models=["NREL5MW"],
    H=200.,
    verbosity=0,
)

ax = foxes.output.FarmLayoutOutput(farm).get_figure(figsize=(5,3))
plt.show()
../_images/notebooks_multi_height_5_0.png

Note that we manually change the hub height from 90 m to 200 m here. Next, we create the states based on the static data file WRF-Timeseries-3000.nc:

In [3]:

# This is what the nc file looks like:
fpath = foxes.StaticData().get_file_path(foxes.STATES, "WRF-Timeseries-3000.nc")
open_dataset(fpath)

Out[3]:

<xarray.Dataset> Size: 528kB
Dimensions:  (Time: 3000, height: 8)
Coordinates:
  * Time     (Time) <U19 228kB '2009-01-01 00:00:00' ... '2009-01-21 19:50:00'
  * height   (height) float32 32B 50.0 75.0 90.0 100.0 150.0 200.0 250.0 500.0
Data variables:
    ws       (Time, height) float32 96kB ...
    wd       (Time, height) float32 96kB ...
    ti       (Time, height) float32 96kB ...
    rho      (Time) float32 12kB ...

Now let’s create the corresponding states object:

In [4]:

states = foxes.input.states.MultiHeightNCTimeseries(
    data_source="WRF-Timeseries-3000.nc",
    time_coord="Time",
    h_coord="height",
    output_vars=[FV.WS, FV.WD, FV.TI, FV.RHO],
    var2col={FV.WS: "ws", FV.WD: "wd", FV.TI: "ti", FV.RHO: "rho"},
)

o = foxes.output.StatesRosePlotOutput(states, point=[0., 0., 100.])
fig = o.get_figure(16, FV.AMB_WS, [0, 3.5, 6, 10, 15, 20], figsize=(6, 6))
plt.show()
../_images/notebooks_multi_height_9_0.png

Note how the var2col option offers a mapping from the expected to the actual column names, if needed.

Let’s next setup our algorithm. Notice that we include the z-sensitive rotor model level10, with 10 points on a vertical line (also the grid models would be an option). The partial wakes choice None represents default settings for all wake models. It is important that we do not select rotor_points together with the level10 rotor, since averaging over a vertical line of points does not make much sense.

In [5]:

algo = foxes.algorithms.Downwind(
    farm,
    states,
    rotor_model="level10",
    wake_models=["Bastankhah2014_linear_ka02"],
    partial_wakes=None,
    chunks={FC.STATE: 1000},
    verbosity=0,
)

We now visualize the vertical flow profile. Alternatively to the here shown use of the states_sel parameter in the plotting function, we could have reduced the whole states object by states.reset(states_loc=["2009-01-06 13:50:00"]).

In [6]:

farm_results = algo.calc_farm()

o = foxes.output.FlowPlots2D(algo, farm_results)
g = o.gen_states_fig_xz(FV.AMB_WS, resolution=10, x_direction=270,
        xmin=0., xmax=1000., zmin=50., zmax=500., figsize=(8,6), states_sel=["2009-01-06 13:50:00"])
fig = next(g)
plt.show()

States 'MultiHeightNCTimeseries': Reading file /home/jonas/gits/wakes/foxes/foxes/data/states/WRF-Timeseries-3000.nc
../_images/notebooks_multi_height_13_1.png

We can now calculate the full states results:

In [7]:

farm_results = algo.calc_farm()

fr = farm_results.to_dataframe()
print(fr[[FV.WD, FV.REWS, FV.P]])

o = foxes.output.FarmLayoutOutput(farm, farm_results)
o.get_figure(color_by="mean_REWS", title="Mean REWS [m/s]", s=150, annotate=0)
plt.show()

o = foxes.output.FarmResultsEval(farm_results)
P0 = o.calc_mean_farm_power(ambient=True)
P = o.calc_mean_farm_power()
print(f"\nFarm power        : {P/1000:.1f} MW")
print(f"Farm ambient power: {P0/1000:.1f} MW")
print(f"Farm efficiency   : {o.calc_farm_efficiency()*100:.2f} %")
print(f"Annual farm yield : {o.calc_farm_yield(algo=algo):.2f} GWh")

                                     WD      REWS            P
state               turbine
2009-01-01 00:00:00 0        340.144012  7.593225  1460.198707
                    1        340.144012  7.593225  1460.198707
                    2        340.144012  7.593225  1460.198707
                    3        340.144012  7.593225  1460.198707
                    4        340.144012  7.593225  1460.198707
...                                 ...       ...          ...
2009-01-21 19:50:00 0        201.682007  7.628299  1512.458987
                    1        201.682007  7.628299  1512.458987
                    2        201.682007  7.628299  1512.458987
                    3        201.682007  7.628299  1512.458987
                    4        201.682007  7.628299  1512.458987

[15000 rows x 3 columns]
../_images/notebooks_multi_height_15_1.png 

Farm power        : 14.8 MW
Farm ambient power: 15.2 MW
Farm efficiency   : 97.25 %
Annual farm yield : 129.91 GWh