Cleanest Hour

How this is calculated

1. Hourly generation. For each hour of the last 12 months () we take your grid's electricity generation by source from the U.S. Energy Information Administration's Hourly Electric Grid Monitor (EIA-930), using EIA's adjusted values.
2. Emission factors. Each source gets a CO₂-equivalent rate from the EPA's eGRID2023 for your balancing authority: coal and gas use that region's own plant averages, oil uses the region's rate when it is plausible (otherwise the U.S. rate), other and unknown fuels and battery or pumped-storage output use the region's overall average, and nuclear, hydro, wind, solar and geothermal count as zero at the smokestack.
3. Hourly intensity. intensity = Σ(generation × factor) ÷ Σ generation, in grams CO₂e per kWh.
4. Your run. kg per run = energy per run × average intensity over the hours the run covers (energy spread evenly, partial hours weighted). We try every allowed start hour, average it over the matching days, and recommend the lowest.
5. Saving. (kg at your usual time − kg at the recommended time) × runs per week × 52.18 weeks. The track record replays the actual days: on how many of them the fixed recommended time beat your usual time. The recommendation is chosen from these same days, so this is a replay of the past year, not a test of how well it forecasts next year.

Factors for

Appliance assumptions

What this does and does not tell you

• Average, not marginal, emissions. We report the average intensity of the generation mix. When you move a load, the plant that actually ramps up or down (the marginal unit) can be dirtier or cleaner than the average, so the emissions you actually avoid may be smaller or larger than shown here. (A midday solar surplus that would otherwise be curtailed is the favorable case; shifting onto hours when coal plants are the ones ramping is the unfavorable one.) Marginal-emissions services such as WattTime, and NREL's Cambium data, estimate that effect.
• Typical history, not a forecast. The recommendation is a fixed habit based on the last 12 months. Weather changes the best hour day to day; the track record shows how often the fixed habit still won.
• Generation-based. Electricity imported from neighboring regions is not traced, which matters most for New York, New England and California.
• Smokestack (combustion) emissions only. No lifecycle or upstream emissions such as methane leaks or building solar panels. 2023 factors are applied to 2025–26 generation.
• Batteries. Battery and pumped-storage output is energy stored earlier from the same grid, so it gets the region's average rate (some grids, including California's, report batteries inside "other"). Charging is not counted. In reality batteries mostly charge from midday solar, so evening hours on battery-heavy grids may be somewhat cleaner than shown.
• Region and time zone. Hours are shown in each region's main time zone ( for this region). Grid borders don't follow state lines. Your utility bill names your provider.
• Scale. One household's saving is small. The point is that it costs nothing, repeats every week, and adds up across many households and devices with timers.

Data provenance