Large-Scale Stochastic Mixed-Integer Programming Algorithms for Power Generation Scheduling

This chapter presents a stochastic unit commitment model for power systems and revisits parallel decomposition algorithms for these types of models. The model is a two-stage stochastic programming problem with first-stage binary variables and second-stage mixed-binary variables. The here-and-now decision is to find day-ahead schedules for slow thermal power generators. The wait-and-see decision consists of dispatching power and to schedule fast-start generators. We discuss advantages and limitations of different decomposition methods and provide an overview of available software packages. A large-scale numerical example is provided using a modified IEEE 118-bus system with uncertain wind power generation. 1 Stochastic Unit Commitment Model Unit commitment is a decision-making process that schedules power generation units and production levels over a planning horizon. This process is central to ensure efficiency and reliability of the power system operation. While fossil-fuel power plants produce 67% of the total electricity generation in US (according to 2014 data), renewable power continue to penetrate into the electricity market [16]. This trend has motivated the development of many unit commitment models that can mitigate uncertainty of renewable supply. In this chapter, we present a stochastic unit commitment model that determines on-off schedules of the generating units and Kibaek Kim Mathematics and Computer Science Division Argonne National Laboratory, 9700 South Cass Avenue, Argonne, IL 60439, USA e-mail: [email protected] Victor M. Zavala Mathematics and Computer Science Division Argonne National Laboratory, 9700 South Cass Avenue, Argonne, IL 60439, USA e-mail: [email protected]