How Energy Trace finds the greenest time to use an appliance or charge a battery

Electricity isn't equally "clean" throughout the day. At some times the grid may be powered mostly by low-carbon sources such as wind and nuclear, while at other times it relies more on gas, which increases emissions. We use regional carbon-intensity data to recommend the time window when using electricity is likely to cause the least CO₂ emissions.

What data we use

The data is provided by the National Energy System Operator (NESO) in partnership with Environmental Defense Fund Europe, the University of Oxford Department of Computer Science, and WWF:

https://carbonintensity.org.uk/

Each data point includes:

  • A time slot (e.g. 18:00–18:30)
  • Carbon intensity in gCO₂/kWh (grams of CO₂ emitted to produce 1 kWh of electricity at that time)

How we estimate the cleanest time

We use the regional carbon-intensity forecast from Carbon Intensity API https://carbonintensity.org.uk/ for the next 48 hours at 30-minute resolution. We retrieve a 48-hour forecast to cover potetial long-running use (more than 12h) during calculations but limit recommendations to the next 24 hours to ensure relevance.

We then search across all possible consecutive time slots to find the block with the lowest expected emissions.

How this works in more detail:

Step 1. Retrieving the next 48 hours of forecast data

For your postcode, we retrieve the regional forecast for the next 48 hours at 30-minute resolution - 96 time slots. These values already reflect:

  • Electricity demand
  • Generation by fuel type
  • Regional imports and exports
  • Transmission and distribution losses

Step 2. Finding the best usage window

If an appliance runs for 4 hours, that means 8 consecutive half-hour periods must be evaluated (we assume usage is continuous once it starts).

We test every possible 4-hour window across the day and calculate the expected emissions for each one.

The window with the lowest average carbon intensity becomes the recommended time.

Step 3. Estimating emissions and savings

To estimate emissions, we multiply:

Energy used (kWh) × carbon intensity (gCO₂/kWh)

Example:

  • A typical EV top-up might use 15 kWh
  • If the best window averages 172 gCO₂/kWh

    → Emissions ≈ 15 × 172 = 2,580 gCO₂ (≈ 2.6 kg)

  • If the "dirtiest" window averages 263 gCO₂/kWh

    → Emissions ≈ 15 × 263 = 3,945 gCO₂ (≈ 3.9 kg)

Estimated saving ≈ 1.4 kgCO₂

We also show the percentage difference between the cleanest and highest-emission times.

About the carbon intensity forecast

The regional forecasts are produced by NESO using:

  1. Machine learning models to forecast demand and generation,

  2. A reduced GB network model to calculate power flows.

  3. Accounting for imports, losses, and regional electricity consumption.

Confidence is labelled as:

Default appliance assumptions

To calculate example savings, we use typical durations and energy use (actual values vary by model and usage):

  • EV charger: 4 hours, 15 kWh
  • Dishwasher: 1 hour 30 minutes, 1.5 kWh
  • Washing machine: 1 hour, 0.5 kWh
  • Tumble Dryer: 1 hour 30 minutes, 2.5 kWh
  • Heat pump: 4 hours, 4 kWh

In the future, users will be able to enter their specific appliance model to improve accuracy.

Limitations of the model

The recommendations depend on forecast values. If grid conditions change unexpectedly - for example due to sudden weather changes or generation outages - actual carbon intensity may differ from the forecast.

Forecasts are updated every 30 minutes, and recommendations are refreshed accordingly.