Enhancing the resolution of energy forecasts: Application to the Spanish electricity market

Abstract

The operation of an electrical system requires continuous adjustment between energy demand and supply. The coordination between various market participants, such as generators, retailers, and the system operator, is essential to efficiently manage electricity offers and demands. Until a few years ago, most European countries performed this scheduling with hourly discretization. However, since 2021, the European Union has implemented a directive that reduced the time intervals to 15 min, replacing the traditional hourly interval.

The proper functioning of an electrical system depends on the instantaneous balance between energy production and demand. Both factors are subject to uncertainties caused by external and random elements, such as solar and wind energy production, and electricity demand itself. The system operator routinely uses predictive models for these variables, with hourly discretization for time horizons ranging from 1 to 15 days. For example, it is common to schedule generation one day in advance, which requires an accurate demand forecast for the 24 h of the following day. With the change to 15-min intervals, these forecasts must be made for the 96 15-min intervals of the next day.

In this article, we propose a highly efficient and accurate method to convert hourly predictions into 15-min predictions. The proposed techniques are analyzed in detail, employing real data from the Spanish electricity market and utilizing the predictive algorithm currently used by the Spanish Transmission System Operator.