Framework for multi-scale planning of the Swiss hydro-power operations: from forecast to grid integration

Abstract

Hydro-power is the backbone of the Swiss electricity system, accounting for approximately two-thirds of the country’s electricity production. Cleverly managing these plants is essential to maximizing profit while balancing demand and production at any time. However, that strategy is often compromised by its dependence on uncertain future inflows, electricity prices, and demand. To help operators in the decision-making process and to simulate the storage hydro-power behavior in the grid, we propose a generic model for the short-to long-term (up to 1 year) using Switzerland as an example. The model relies on a rolling-horizon optimization approach, continuously updating decisions as new forecasts become available. This framework aims to reproduce observed hydro-power operations under realistic conditions. In this study, we evaluate its ability to simulate the Swiss electricity system without using perfect foresight, but solely seasonal and sub-seasonal weather forecasts. Results show that the framework successfully captures both long- and short-term hydrological dynamics. Despite relying exclusively on seasonal and sub-seasonal forecasts, the model reproduces monthly turbine production with a Mean Absolute Percentage Error of 18% relative to actual observed production. While most existing models rely on perfect foresight or simplified representations of hydro-power, we show how such a management would work when only forecasts are available. Spillage only increases from 0.18% to 0.64% of total inflows without perfect foresight, while total hydro-power revenue decreases by 6%. This comparison with an hypothetic perfect situation highlights the impact of forecasting error on the short-and long-term planning of hydro-power operations.