ABSTRACT
Power system operation necessitates monitoring of the prevailing system conditions by measuring different system signals such as voltage and power flows, to paint a clearer and updated picture of the system status on round-the-clock periodic basis. Based on such a picture, grid system operators (GSOs) take specific control actions corresponding to different conditions. These legacy hand-coded rule book actions are executed via technology enablers that invoke decision support tools at the GSO’s disposal. With the increasing proportion of variable renewable energy generation (REG), for instance, solar PV and wind, connected to the grid, new unprecedented operation scenarios are emerging with a myriad of technical challenges. The higher volatility in the net demand, due to the integration of renewables, requires larger grid support that would, in turn, instigate the adoption of new technologies and decision support tools to cope with the evolving grid system. Unlike conventional generation, REG resources cannot be forecasted precisely well beforehand (uncertainty). On the other hand, depending on meteorological conditions, they vary with time (variability), especially when dealing with a shorter time step problem. Deterministic optimization approaches have been applied for several decades, typically in power system operation utility practices, for various problems, which cover optimal power flow and unit commitment. Nonetheless, the exacerbated variability and uncertainties injected into the grid system have compelled energy planners and GSOs to revisit the use of deterministic operation system analysis tools to be replaced by stochastic ones. Stochastic optimization can include REG uncertainty as random variables, through the use of scenario generation or with an assumed probability distribution. This book chapter covers the state-of-the-art stochastic optimization approaches and their applications in power system operation, with particular emphasis to the unit commitment (UC) problem. Stochastic optimization principles are first presented. Stochastic optimization categories including robust optimization (RO) and chance-constrained programming (CCP) are further discussed. Case studies on a unified standard test system that has different proportions of REG are also demonstrated.
