An improved treatment of operating reserves in generation expansion planning models

Abstract

Energy system optimisation (ESOM) and generation expansion planning (GEP) models are often used to study energy transition pathways. These typically entail an increased penetration of variable renewable energy sources (VRES), which can lead to increased operating reserve requirements due to their associated forecast uncertainty. Representing this effect has previously been tackled using either stochastic programming techniques or deterministic GEPs which use heuristics to size reserves while ignoring their activation cost. In this paper, we propose a novel GEP formulation which determines operating reserve requirements using a second order cone (SOC) constraint. This formulation approximates the solution of a stochastic GEP by accounting for reserve activation costs without resorting to scenario based methods. A case study on the Belgian system indicates possible cost savings of 70 MAC(0.9%) and less bias towards installing peaking technologies to satisfy reserve requirements compared to a deterministic GEP. The sensitivity of the results to the assumption of normality of forecast errors and temporal detail is also investigated. Two final case studies on the value of emergency measures and improving forecast uncertainties illustrate the benefits of accounting for reserve activation costs and appropriate reserve sizing.