Generation Expansion Planning under Uncertainty: An Application of Stochastic Methods to the German Electricity System

Abstract

Renewable energies are expected to be the main electricity generation source. However, the variability of renewable energy supply poses challenges to the generation expansion modelling as uncertainty of hourly generation need to be adequately taken into account. This paper analyzes the implications of different approaches to optimization under uncertainty, ranging from stochastic to robust optimization. We apply these specific approaches to the German electricity system in 2035 and compare them to a deterministic optimization for each realization of the uncertainty. We consider the availability of wind and solar generation as explicit uncertainties affecting the second-stage dispatch level. The deterministic generation expansion problem shows significant variations of optimal capacity mixes depending on the underlying assumptions on hourly renewable feed-in. Moreover, these capacitiy mixes are hardly robust to unexpected situations. Contrarily, stochastic as well as robust approaches provide a consistent and robust capacity mix at only slightly higher total costs.