The data is public. Take it with you.

To take today

The counter’s snapshot is public at the same address the site consumes: nameplate.com.br/contador.json. It is a ~1 KB file regenerated twice a day, stable enough for a spreadsheet, a script or a story: the field contract is locked by tests on both sides of the pipeline.

snapshot_atstring · ISO 8601

When this snapshot was computed (Brasília time).

mwh_anonumber · MWh

Curtailed energy accumulated in the current year, wind + solar.

reais_anonumber · R$

The same total priced by the methodology’s assumption.

taxa_mw_atualnumber · MW

The system’s average curtailment over the last 24 hours of data.

ultimo_dado_onsstring · ISO 8601

Timestamp of the most recent ONS data point in the calculation.

The house publishes four more files in the same pattern, each read by its own page: saude.json, the ingestion picture, and cargas.csv, one line per load since 27 Aug 2026, which together feed the status page; previsao.json, the next three days of curtailment with each horizon's error inside, feeding the forecast and accuracy pages — which also read historico-acuracia.csv, the day-by-day forecast-vs-realized curve; and usinas.json, the year's picture by asset behind the plants page.

Since 31 Aug 2026, each day's curtailment in the current year is also in CSV: nameplate.com.br/serie-diaria.csv — columns dia, mwh_dia and corte_medio_mw, updated alongside the counter. Today's scope is the current year, with no extra call to ONS; full history back to October 2021 is the next delivery, once the base the forecast model already keeps justifies publishing it here too.

Since 4 Sep 2026, the same series exists per asset: nameplate.com.br/serie-diaria-usinas.csv — columns dia, id (the ONS id_ons, the same one in usinas.json), mwh_dia and corte_medio_mw, one row per plant or cluster per day, ~240 assets, ~1.5 MB for the year. A zero row is an asset ONS reported with no curtailment that day; a missing row is ONS silence. It is the series the alerts panel checks against each plant's threshold.

Data freshness

ONS publishes with a lag and the site extrapolates on top of the last snapshot: the methodology page opens that arithmetic. When this page was built, the picture was:

Last verified ONS data point

September 26, 2026 at 11:30 PM

Last pipeline snapshot

September 28, 2026 at 2:37 AM

Cadence

Two snapshots a day, by a scheduled job; no server stays up.

Lag at the edge

The ONS portal delivers the previous day in full on both daily loads: the newest data point lags between ~13 and ~43 hours — measured on 27 Aug 2026.

Revision window

Closed months keep being rewritten at the source for ~4 weeks: recent-month figures are preliminary until that window passes.

These dates are baked when the site is generated. The always-current values live in contador.json itself, in the ultimo_dado_ons and snapshot_at fields.

The raw data, at the source

Nameplate does not mirror the raw data: the canonical source is the ONS open-data portal, which publishes constrained-off per plant, in half-hour intervals, back to October 2021. The two datasets feeding the counter are restricao_coff_eolica_tm and restricao_coff_fotovoltaica_tm, monthly Parquet files, per source.

Licence and citation

• The source data belongs to ONS · Open Data, licensed CC-BY: use, redistribute and transform it, citing the source.
• What Nameplate publishes on top (the snapshot, treated series) follows the same rule: cite Nameplate (nameplate.com.br) and ONS as the origin of the raw data.
• Need a cut that is not here (another aggregation, another period, programmatic access)? Write to contato@viniciuslisboa.com.br: that is exactly the request that helps decide what ships next.

How to cite this work:

Nameplate. Wind and solar curtailment in Brazil. https://nameplate.com.br/en/data. Source data: ONS · Open Data, CC-BY. Accessed 2026-09-28.