Affine and Exact Reformulations of Uncertainty-Aware Energy and Reserve Dispatch

In recent years, renewable energies have increased their share in the energy sector. Renewables are characterized by their intermittent and uncertain nature, which brings severe challenges to system operators. In that context, probabilistic optimization techniques have gained increased attention to better describe the uncertainty and make optimally-informed energy and reserve dispatch decisions in the day-ahead stage. In this paper, we explore Distributionally Robust Optimization (DRO) technique to formulate an Optimal Power Flow (OPF) problem. We model the second-stage decision variables, such as real-time activation of balancing energy, via affine decision rules. The probabilistic real-time operating constraints are reformulated using Conditional Value-at-Risk (CVaR) risk measures. The contribution of this paper is to compare its out-of-sample performance on a fair basis against standard probabilistic optimization techniques using different reformulations of recourse actions, either affine decision rules or exact recourse models. Results demonstrate that DRO can outperform traditional techniques on a fair basis. However, DRO using affine decision rules also shows limitations against simpler probabilistic modelling approaches using exact recourse actions.