Uruguay: measured wind output and dispatch limits#

Read 169 native SCADA samples from Cerro Grande wind farm (ADME central 7). The complete real source file contains measured power and the operator’s generation limit as separate channels. Eight unavailable output readings remain missing; they are not filled with zeros or forecasts.

Source: ADME open-data instructions. Download the original SCADA SAS file and capture manifest. The gallery runs offline with the unchanged public source file.

import hashlib
import json
from pathlib import Path

import matplotlib.pyplot as plt

from energy_data_retrieval.sources.adme import parse_scada_sas

1. Verify a measured SCADA archive file#

ADME’s PRONOS application links separate forecast and SCADA directories. This connector selects only the SCADA directory. Fresh acquisition is bounded and includes directory/file provenance:

from datetime import date
from energy_data_retrieval.sources.adme import ADME_HOSTS, ADMEClient
from energy_data_retrieval.transport import HTTPClient
with HTTPClient(ADME_HOSTS) as http:
    result = ADMEClient(http).generation(date(2026, 10, 22), date(2026, 10, 27),
                                         central_id=7)

To find another plant’s ID, select it in the official PRONOS page and inspect its central URL parameter. Six-day query chunks are checked against the actual coverage of each published rolling seven-day SCADA file.

directory = Path(globals().get("__file__", Path.cwd() / "plot_generation_adme.py")).resolve().parent
data = directory / "data"
manifest = json.loads((data / "generation_adme_provenance.json").read_text())
raw = (data / manifest["fixture"]).read_bytes()
if hashlib.sha256(raw).hexdigest() != manifest["fixture_sha256"]:
    raise ValueError("ADME capture differs from its provenance")
frame = parse_scada_sas(raw.decode("utf-8"), central_id=7)
print("Source:", manifest["source_url"])
print("Captured:", manifest["retrieved_at_utc"])
Source: https://pronos.adme.com.uy/svg/pronos/cache_series_scada/scada_central_7_2026102121.sas
Captured: 2026-10-30T08:17:07.182140+00:00

2. Parse native timestamps, actual power and missing sentinels#

SAS is tab-separated text with a six-line header. Its Excel/LibreOffice serial dates carry no offset; this example explicitly interprets local labels as America/Montevideo (UTC-03 on these dates), converts them to UTC, and preserves the labels. The source’s below-10000 missing sentinels become NaN.

GETOTMEDP.AV is measured MW. CONSIGNAP is a dispatch limit, not output. Additional wind and control channels remain in the normalized frame.

print("Actual power channel:", frame["power_series"].iloc[0])
print("Sampling seconds:", frame["sampling_seconds"].iloc[0])
print("Reported measurements:", frame["value"].count(), "missing:", frame["value"].isna().sum())
Actual power channel: PCGRANDE.030GETOTMEDP.AV
Sampling seconds: 3600
Reported measurements: 161 missing: 8

3. Plot output alongside the control limit#

This is a comparison of measured power with an operating instruction. The gap between a limit and output is not lost energy: available wind power and the reason for each control instruction require separate evidence. No interval energy or interval boundary is inferred from these sample labels.

fig, ax = plt.subplots(figsize=(10, 4.8), layout="constrained")
ax.plot(frame["timestamp"], frame["value"], label="Measured output", linewidth=2)
ax.plot(
    frame["timestamp"],
    frame["dispatch_limit_mw"],
    label="Dispatch limit (CONSIGNAP)",
    linestyle="--",
    color="#a63d40",
    linewidth=1.4,
)
ax.set(
    title="ADME Cerro Grande · original 169-sample SCADA file",
    xlabel="Native sample label interpreted in Montevideo time and converted to UTC",
    ylabel="Power (MW)",
)
ax.grid(alpha=0.2)
ax.legend()
plt.show()
ADME Cerro Grande · original 169-sample SCADA file

Reuse#

Credit Administración del Mercado Eléctrico (ADME), retain the native file and clock interpretation, and preserve missing-value meaning. This plant capture is not a national generation total or a forecast of future wind output.

Total running time of the script: (0 minutes 0.108 seconds)

Gallery generated by Sphinx-Gallery