What Is a Load Forecast?

A load forecast predicts how much electricity a group of customers, a balancing group or a grid area will consume. Typical horizons range from intraday through day-ahead to months ahead, in 15-minute resolution for short-term use.

How a load forecast is created

Electricity demand follows recurring patterns, which a load forecast models explicitly:

• Calendar effects: time of day, weekday, public and school holidays, bridge days.
• Weather: temperature drives heating and cooling, irradiance affects lighting and, increasingly, behind-the-meter PV.
• Customer structure: households, commerce and industry behave very differently.

Historically, small customers in Germany were described by standard load profiles (SLP), fixed curves per customer group. With smart meters, heat pumps, electric cars and rooftop PV, real consumption deviates more and more from these profiles. Modern load forecasts therefore learn from measured load data: statistical and machine learning models combine calendar and weather features and are retrained as the portfolio changes. Clustering customers with similar behaviour helps where single meters are too noisy to forecast individually.

What load forecasts are used for

Utilities and suppliers use load forecasts to buy electricity on the futures and spot markets and to nominate their balancing group schedule. Every deviation between forecast and actual consumption has to be settled as balancing energy. Grid operators need load forecasts for grid planning and to estimate grid losses.

A good load forecast is therefore not an IT detail but a direct cost factor. You can find out how we build load forecasts for electricity, gas and heat on our load forecast page. The best way to see what they can do for your portfolio is a forecast benchmark with your own data.