Ensure that the used df_opt is in the correct ordering of wind direction, wind speed, and turbulence intensity.

Author: misi9170

tests/wake_steering_design_test.py

@@ -8,6 +8,7 @@
     apply_wind_speed_ramps,
     build_simple_wake_steering_lookup_table,
     build_uncertain_wake_steering_lookup_table,
+    check_df_opt_ordering,
     compute_hysteresis_zones,
     create_uniform_wind_rose,
     get_yaw_angles_interpolant,
@@ -281,3 +282,20 @@ def test_create_uniform_wind_rose():
     wind_rose = create_uniform_wind_rose()
     frequencies = wind_rose.unpack_freq()
     assert (frequencies == frequencies[0]).all()
+
+def test_check_df_opt_ordering():
+
+    # Pass tests
+    df_opt = generic_df_opt()
+    check_df_opt_ordering(df_opt)
+
+    # Remove a row so that not all data is present
+    with pytest.raises(ValueError):
+        check_df_opt_ordering(df_opt.drop(0))
+
+    # Artificially create bad ordering by swapping columns
+    df_opt_2 = df_opt.copy()
+    df_opt_2.wind_speed = df_opt_2.wind_direction
+    df_opt_2.wind_direction = df_opt.wind_speed
+    with pytest.raises(ValueError):
+        check_df_opt_ordering(df_opt_2)

whoc/design_tools/wake_steering_design.py

@@ -168,8 +168,8 @@ def apply_static_rate_limits(
     ws_step = ws_array[1] - ws_array[0]
     ti_step = ti_array[1] - ti_array[0] if len(ti_array) > 1 else 1
 
-    # 4D array, with dimensions: (turbine, wd, ws, ti)
-    # TODO: will this ordering always work? Or not?
+    check_df_opt_ordering(df_opt)
+    # 4D array, with dimensions: (wd, ws, ti, turbines)
     offsets_array = offsets_all.reshape(
         (len(wd_array), len(ws_array), len(ti_array), offsets_all.shape[-1])
     )
@@ -255,6 +255,7 @@ def compute_hysteresis_zones(
     """
 
     # Extract yaw offsets, wind directions
+    check_df_opt_ordering(df_opt)
     offsets_stacked = np.vstack(df_opt.yaw_angles_opt.to_numpy())
     wind_directions = np.unique(df_opt.wind_direction)
     offsets = offsets_stacked.reshape(
@@ -349,6 +350,7 @@ def apply_wind_speed_ramps(
         pd.DataFrame: A yaw offset lookup table for all wind speeds between ws_min and ws_max
             with wind speed ramps applied.
     """
+    check_df_opt_ordering(df_opt)
 
     # Check valid ordering of wind speeds
     if (ws_wake_steering_cut_in
@@ -455,6 +457,7 @@ def get_yaw_angles_interpolant(df_opt):
             dimensions, and returns the yaw angles for all turbines. This function
             incorporates the ramp-up and ramp-down regions.
     """
+    check_df_opt_ordering(df_opt)
 
     # Extract points and values
     wind_directions = np.unique(df_opt["wind_direction"])
@@ -465,8 +468,6 @@ def get_yaw_angles_interpolant(df_opt):
     # Store for possible use if no turbulence intensity is provided
     ti_ref = float(np.median(turbulence_intensities))
 
-    # TODO: check ordering in df_opt is correct
-
     # Reshape the yaw offsets to match the wind direction, wind speed, and turbulence intensity
     yaw_offsets = yaw_offsets.reshape(
         len(wind_directions),
@@ -565,4 +566,40 @@ def create_uniform_wind_rose(
         wind_speeds=wind_speeds,
         wind_directions=wind_directions,
         ti_table=ti,
-    )
+    )
+
+def check_df_opt_ordering(df_opt):
+    """
+    Check that the ordering of inputs is first wind direction, then wind speed,
+    then turbulence intensity.
+
+    Raises an error if this is found not to be the case.
+
+    Args:
+        df_opt (pd.DataFrame): A yaw offset lookup table.
+    """
+
+    inputs_all = df_opt[["wind_direction", "wind_speed", "turbulence_intensity"]].to_numpy()
+    wd_unique = np.unique(df_opt["wind_direction"])
+    ws_unique = np.unique(df_opt["wind_speed"])
+    ti_unique = np.unique(df_opt["turbulence_intensity"])
+
+    # Check full
+    if not inputs_all.shape[0] == len(wd_unique)*len(ws_unique)*len(ti_unique):
+        raise ValueError(
+            "All combinations of wind direction, wind speed, and turbulence intensity "
+            "must be specified."
+        )
+
+    # Check order is correct
+    wds_reshaped = inputs_all[:,0].reshape((len(wd_unique), len(ws_unique), len(ti_unique)))
+    wss_reshaped = inputs_all[:,1].reshape((len(wd_unique), len(ws_unique), len(ti_unique)))
+    tis_reshaped = inputs_all[:,2].reshape((len(wd_unique), len(ws_unique), len(ti_unique)))
+
+    if (not np.all(wds_reshaped == wd_unique[:,None,None])
+        or not np.all(wss_reshaped == ws_unique[None,:,None])
+        or not np.all(tis_reshaped == ti_unique[None,None,:])):
+        raise ValueError(
+            "df_opt must be ordered first by wind direction, then by wind speed,"
+            "then by turbulence intensity."
+        )