Risk-Averse Markov Decision Processes: Applications to Electricity Grid and Reservoir Management

TL;DR

This paper introduces risk-averse Markov Decision Process models incorporating CVaR and new risk measures, enabling more reliable electricity grid and reservoir management under renewable energy uncertainties.

Contribution

It develops novel risk-averse MDP models with efficient solution methods, integrating financial and power reliability risk measures for system planning and operation.

Findings

Efficient algorithms for risk-averse MDPs with CVaR

New risk measures tailored for power system reliability

Improved planning strategies under renewable uncertainty

Abstract

This paper develops risk-averse models to support system operators in planning and operating the electricity grid under uncertainty from renewable power generation. We incorporate financial risk hedging using conditional value at risk (CVaR) within a Markov Decision Process (MDP) framework and propose efficient, exact solution methods for these models. In addition, we introduce a power reliability-oriented risk measure and present new, computationally efficient models for risk-averse grid planning and operations.