Decomposition methods for stochastic optimal coordination of energy storage and generation

This paper investigates decomposition methods in stochastic optimization to facilitate parallel computing for the security constrained economic dispatch (SCED) problem in power systems with renewable energy and storage devices. A two-stage stochastic version of SCED based on stochastic model predictive control is formulated. The objective is to supply the demand under uncertainties in the most economic and safe manner, achieving optimal coordination between energy storage and dispatchable generators with respect to real power balancing. The uncertainties correspond to the power output of renewable generators and the demand at load buses. A dual decomposition scheme based on optimality condition decomposition (OCD) is applied to the considered stochastic SCED problem and compared with the Benders decomposition algorithm. The effectiveness of the OCD algorithm is demonstrated through the comparison between the OCD and Benders decomposition on a modified WECC 9-bus test system.