Feature/test nan

[paulf81]

PR #222 proposed to fill powers but per discussion this might not be expected behavior from users. Prefer that user's decide for themselves what do with NaNs. Still, user's may be unaware of NaNs in data and what Hercules should do with them when they are in certain key signals is ambiguous. For example, if one turbine in 100 has a NaN value, the total power for that time step is also NaN.

This PR address the issue by throwing an error:

1. Wind speeds input to WindFarm are NaN
2. Wind turbine powers input to WindFarmSCADAPower are NaN

A test for each new condition is added.

[Copilot]

Pull request overview

This PR changes WindFarm and WindFarmSCADAPower initialization behavior to fail fast when key input signals contain NaN values, avoiding ambiguous downstream NaN propagation in aggregate power calculations.

Changes:

• Add NaN validation for wind input files used by WindFarm and raise ValueError when required wind columns contain NaNs.
• Add NaN validation for SCADA power input files used by WindFarmSCADAPower and raise ValueError when required power columns contain NaNs.
• Add/extend pytest coverage to assert these new error conditions.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

File	Description
hercules/plant_components/wind_farm.py	Adds NaN checks on key wind input columns during initialization.
hercules/plant_components/wind_farm_scada_power.py	Adds NaN checks on key SCADA columns during initialization.
tests/wind_farm_direct_test.py	Adds a test asserting WindFarm raises when wind input contains NaNs.
tests/wind_farm_scada_power_test.py	Adds a test asserting WindFarmSCADAPower raises when power columns contain NaNs.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The NaN validation only checks ws_* columns, but this class also supports wind input files that provide wind speeds via ws_mean when per-turbine ws_### columns are absent. As written, NaNs in ws_mean will not be caught and can still propagate into the simulation. Consider including ws_mean in nan_check_cols (at least when ws_000 is not present) so the new behavior covers all supported wind-speed inputs.

Suggested change

	if df_wi[nan_check_cols].isna().any().any():
	raise ValueError(
	"wind input file contains NaN values in required columns (time_utc, wd_mean, ws_*)"
	# If no per-turbine wind speed columns are present, fall back to validating ws_mean
	if not ws_cols and "ws_mean" in df_wi.columns and "ws_mean" not in nan_check_cols:
	nan_check_cols.append("ws_mean")
	if df_wi[nan_check_cols].isna().any().any():
	raise ValueError(
	"wind input file contains NaN values in required columns (time_utc, wd_mean, ws_*, ws_mean)"

[paulf81]

added

[Copilot]

pow_cols can be empty (e.g., malformed SCADA file with no pow_* columns). In that case initialization later will fail with an IndexError when indexing self.power_columns[0] to infer rated power. Since this block is already validating the SCADA schema, consider explicitly erroring when no pow_* columns are present (and ideally mention which required columns are missing).

Suggested change

	# Check key columns for NaN values
	pow_cols = sorted(col for col in df_scada.columns if col.startswith("pow_"))
	# Identify power columns and check key columns for NaN values
	pow_cols = sorted(col for col in df_scada.columns if col.startswith("pow_"))
	if not pow_cols:
	raise ValueError(
	"SCADA file must contain at least one power column with name starting with 'pow_'."
	)

[paulf81]

Done

[Copilot]

This new NaN test covers NaNs in per-turbine ws_### columns, but WindFarm also supports the ws_mean-only input format (see existing tests in dynamic/precomputed modes). If the intent is to reject NaNs in wind speeds generally, add a test case where ws_mean is present (and ws_### columns are absent) and contains NaN, asserting the same ValueError is raised.

Suggested change

	os.unlink(temp_wind_file)
	os.unlink(temp_wind_file)
	def test_wind_farm_raises_on_nan_in_wind_input_ws_mean_only():
	"""Test that WindFarm raises ValueError when ws_mean-only wind input contains NaN values."""
	wind_data = {
	"time_utc": [
	"2018-05-10 12:31:00",
	"2018-05-10 12:31:01",
	"2018-05-10 12:31:02",
	"2018-05-10 12:31:03",
	"2018-05-10 12:31:04",
	"2018-05-10 12:31:05",
	"2018-05-10 12:31:06",
	"2018-05-10 12:31:07",
	"2018-05-10 12:31:08",
	"2018-05-10 12:31:09",
	"2018-05-10 12:31:10",
	],
	"wd_mean": [180.5, 185.2, 190.8, 175.3, 170.1, 165.7, 160.4, 155.9, 150.2, 145.6, 140.3],
	# ws_mean-only input format (no per-turbine ws_### columns), with a NaN value
	"ws_mean": [8.2, np.nan, 7.8, 6.5, 10.2, 11.5, 9.8, 8.7, 7.3, 6.9, 8.4],
	}
	with tempfile.NamedTemporaryFile(mode="w", suffix=".csv", delete=False) as f:
	pd.DataFrame(wind_data).to_csv(f.name, index=False)
	temp_wind_file = f.name
	try:
	test_h_dict = copy.deepcopy(h_dict_wind_direct)
	test_h_dict["wind_farm"]["wind_input_filename"] = temp_wind_file
	with pytest.raises(ValueError, match="wind input file contains NaN values"):
	WindFarm(test_h_dict)
	finally:
	if os.path.exists(temp_wind_file):
	os.unlink(temp_wind_file)

[paulf81]

I think it's ok as is