Daily Electricity Load Shapes and the Duck Curve, 30 Day Window to 20 September 2026

• In CAISO (29 complete local days from 22 August to 19 September 2026), average net load, demand minus solar and wind, rose 14,507 MW between 16:00 to 19:00 local time, after a midday low of 14,383 MW at 12:00.
• In ERCOT (29 complete local days from 22 August to 19 September 2026), average net load, demand minus solar and wind, rose 15,366 MW between 16:00 to 19:00 local time, after a midday low of 31,087 MW at 09:00.
• Among United States grids with reported solar, the steepest evening ramp relative to average demand was IID: net load rose 59% of its average demand over three hours (29 complete local days from 22 August to 19 September 2026).
• Among Europe grids with reported solar, the steepest evening ramp relative to average demand was Italy Sardinia: net load rose 96% of its average demand over three hours (28 complete local days from 23 August to 19 September 2026).
• 100 grids have a usable profile: 53 in the United States and 47 in Europe, and 69 of them report enough solar to draw a net load curve.

A load shape is the average pattern of electricity demand over a day. Most grids have a low point in the small hours and a high point in the afternoon or evening, when people come home and cooling or lighting is on.

Net load is demand minus wind and solar output: what conventional plants, hydro, batteries and imports must cover. On a grid with a lot of solar, net load falls through the middle of the day and then climbs steeply as the sun sets and people are still using power. Plotted over 24 hours, that dip and climb look like the body and neck of a duck, which is where the name comes from. The steepness of the climb is what operators care about, because other resources must ramp up quickly to meet it.

The figures below are averages over recent days and mix weekdays and weekends. A single day can look very different, especially on a heat wave, a cloudy day or a windy night.

• Source: the last 30 days of hourly demand and generation by source that each operator publishes. US grids come from EIA and European bidding zones from ENTSO-E; the two are shown separately and never mixed with Ember statistics.
• Local time: hours use the IANA time zone of each grid, so a 17:00 peak is 17:00 on the clock in that grid. Daylight saving follows the local clock, so a clock change moves the pattern with it instead of smearing it. A local day is used only if it is whole inside the window (23, 24 or 25 hours) and at least 22 hours are complete. On a clock change day the repeated hour is averaged over both readings.
• Same days for every line: demand, net generation, solar and wind use the same days and hours, so net load is exactly demand minus solar minus wind.
• Solar and wind are counted only when the series is reported (values in at least half the hours, not flat zero, and not zero for the last 14 days or more). A grid whose wind or solar is left out is noted on its page. Small negative solar or wind readings are treated as zero.
• A grid needs at least 7 usable local days to appear. Averages mix weekdays and weekends. A 30 day window is a short, recent sample, not a seasonal norm.
• Figures are as published by the operators, who may revise them. See methodology and data quality for known gaps.