Risk-sensitive stochastic optimization for storage operation management

Abstract

This study concerns charging/discharging management of an energy storage device, integrated with a renewable energy supply to fully serve loads over a finite time horizon. The goal is to minimize the cost (or maximize the revenue) of exchanging energy with the electrical grid. The storage device is defined through conversion and dissipation losses, capacity limitations, and bounds on conversion rates. A general-purpose rolling horizon procedure is developed, with the goal of optimizing a risk functional of the cost distribution. Our computational results suggest that the general-purpose algorithm for our setting retains the qualitative behavior of the exact optimal control strategy, which is established under special assumptions. The value of the general-purpose algorithm is evaluated for both risk-neutral and risk-averse criteria, with notable improvements reported in the study.