The problem’s words, demystified.

#24–72h forecast

Nameplate’s forecasting horizon, live since August 2026: where and when curtailment will happen, one to three days ahead, with the uncertainty stated beside every number. It is the window that lets flexible load be scheduled into the surplus hours — auditable through the accuracy page’s public backtesting.

#Backtesting

Checking the forecast against what actually happened, day after day. That is how Nameplate’s model is measured: each horizon’s typical error over the last four weeks, published on the accuracy page — for anyone to audit the hit rate.

#CC-BY licence

The Creative Commons "attribution" licence: use, redistribute and transform, citing the source. It is the licence of ONS open data and the same one Nameplate applies to what it publishes on top (see /data).

#CMO

The marginal operating cost: what producing the system’s next MWh would cost. It moves with the PLD and is one of the signals the forecasting model will watch.

#Curtailment (constrained-off)

The generation cut ONS orders when the grid cannot carry or absorb all the energy being produced: the plant could generate, but is told to reduce. It is the phenomenon Nameplate measures; in the official records the term appears as "constrained-off".

#Data lag

The time between energy being curtailed and the record appearing on the ONS portal. Measured by the project on 27 Aug 2026: both daily loads deliver the previous day in full — the edge lags between ~13 and ~43 hours (the full note is on the research page).

#Dispatch

The moment-by-moment decision, made by ONS, of which plants generate how much. Curtailment is an act of dispatch: the order for a renewable plant to generate less than it could.

#Distributed generation

Small, scattered generation, such as rooftop panels, connected at the distribution level. It sits outside Nameplate’s numbers — the scope is utility-scale — but it feeds the phenomenon: it shrinks visible demand and pushes even more surplus into midday.

#Flexible load

Consumption that can wait for the right hour: computing jobs, pumping, cooling with slack. It is the "put to use" piece — a signal switches this load on when energy is left over, as in the school prototype (see the home page).

#Live estimate

The counter’s honest label: between one snapshot and the next, the number climbs by extrapolation (total + rate × elapsed time) and converges when the recomputed truth arrives. The movement is declared estimation; measurement lands twice a day.

#Load

The system’s consumption at an instant: everything switched on, summed. Generation and load must balance second by second — when renewable output exceeds the load the grid can serve, curtailment is born.

#MW × MWh

A megawatt is power (how much is flowing right now); a megawatt-hour is energy (that power sustained for an hour). ONS records are average power per half-hour, which is why the arithmetic multiplies by 0.5 h to get energy.

#Nameplate capacity

How much a plant is able to generate, stated on its identification plate, the nameplate. The distance between what the plate promises and what the grid lets it deliver is the project’s subject, and gave it its name.

#ONS

Brazil’s national grid operator: the body that coordinates the power system and decides, moment by moment, who generates how much. The curtailment order is theirs, and so is the open-data portal Nameplate reads everything from.

#Parquet

The file format ONS uses for its open data: columnar and compact, built for analysis. The constrained-off files are monthly, one per source (wind and solar).

#PLD

Brazil’s short-term energy settlement price. The counter uses a reference average PLD as the assumption for turning MWh into reais; the exact value and the reasoning are on the methodology page.

#Reference generation

How much the plant would have generated in that half-hour had there been no restriction: ONS publishes it alongside what was actually delivered. The difference between the two is the curtailment, and the first line of the counter’s arithmetic (see the methodology).

#Restriction reason

The reason ONS records for each cut: electrical reliability (the grid cannot take it) or energy balance (the system does not need the energy). The latter is the structural surplus — the one that grows, and the one Nameplate wants to put to work.

#Revision window

The period in which ONS still rewrites published data: a closed month keeps being revised for at least four weeks. Recent-month figures are preliminary until the window passes — the pipeline logs every load precisely to measure the size of these revisions.

#SIN

The National Interconnected System: the grid tying together generation and consumption across almost all of Brazil. It is because everything is connected that surplus energy in the Northeast becomes the whole system’s problem.

#SIN subsystems

The four electrical regions of the interconnected system — North, Northeast, Southeast/Centre-West and South. Curtailment is not uniform: it concentrates where renewable generation outgrew consumption, with the Northeast in front.

#Snapshot

The picture the pipeline takes twice a day: a ~1 KB public file (contador.json) with the year’s total, the recent rate and the date of the last verified data point. It is the truth the counter animates over; the full dictionary is at /data.

#Solar photovoltaic power

Electricity generated from sunlight by panels. Output peaks in the same hours at every plant at once, which deepens the midday surplus — the other half of what Nameplate measures.

#Transmission grid

The high-voltage lines carrying energy from plants to consumption centres. When lines are short, curtailment is a transport bottleneck; but ONS projections say most future curtailment is not that — it is structural surplus (see the home page).

#Utility-scale generation

Large plants connected directly to the transmission grid: the scope of Nameplate’s numbers. It does not include distributed generation, such as rooftop panels.

#Wind power

Electricity generated from wind. It is Brazil’s most curtailed source — the Northeast’s wind farms produce most in the hours the system can least absorb — and therefore half of what Nameplate measures.

Missing a term? That is exactly the kind of note that improves this page: write to the team via the e-mail in the footer.