• Elisavet Sergiadou

The art of accurate load forecasting for optimal grid management

Grid operators need to be able to predict with extreme accuracy where grid hotspots will occur, in order to identify infrastructure reinforcement needs, plan distribution, manage assets and ameliorate day-ahead or intraday planning.

A particularity that makes load forecasting extremely challenging is that energy consumption is extremely volatile, far more volatile than any other commodity. Another feature of electricity is that the spikes’ intensity is non-homogenous in time. Load spikes are mostly observed during specific time windows. Although these spikes usually have a very short duration, the impact on the system is extreme if no measures are taken. Load forecasting on an aggregated basis can help limit and manage these abnormalities. For the grid operator to deal with such spikes usually cost-inefficient actions are required, so prior knowledge will help for more effective asset management from his side.

Traditional load forecasting models take into account historical and weather data to predict the electricity demand. With the extensive renewable sources’ injection into the modern grid though, it’s much harder to make accurate load predictions.