LAUSR

Abstract One of the most promising ways of de-carbonising the energy sector is through increasing the amounts of variable renewable energy (VRE) generation in power systems. While the inherent uncertainty of VRE generation is a challenge, it can be mitigated through improved forecasting and energy system modelling. Typically, stochastic energy system studies have focused on the day-ahead horizon of 36 h ahead of time, while studies about hydro-thermal scheduling and expansion planning often neglect uncertainty of VRE generation entirely. In this work, the potential benefits of extending the horizon of VRE forecasts on the operation of power systems with significant amounts of hydropower was examined using a future Nordic system case study. 15-day ensemble weather forecasts were processed into realistic VRE and demand forecasts up to 348 h ahead of time, and their impact on power system operations was simulated using stochastic unit commitment and economic dispatch optimisation. Increasing the length of the modelled forecast horizon reduced the total yearly operational costs of the system by 0.18–0.41%, as well as the spillage of run-of-river hydropower and the curtailment of wind power by 0.42–0.47 and 0.05–0.07% points respectively.