Short-term generator commitment to establish a desired system reliability while minimizing system fuel costs

Abstract

A method is described for determining the most economical generating unit commitment policy and loading schedule for a day's operation of an electric utility system while maintaining a desired level of system reliability. Generating units are scheduled to supply the predicted system load for a day. As the load changes during the day it may be desirable to plan the shutdown and starting of one or more generating units. Dynamic programming is applied to determine which unit to shut down and start up at future hours such that system fuel costs including start-up costs are minimized. The startup and shutdown times of generating units are determined so as to maintain desired system reliability. The scheduling technique employs quantitative reliability criteria. The measure of reliability used is the probability that available generating capacity at any time is less than the system load at that time. It is assumed that at any hour the system resides in one of a set of mutually exclusive states where each state defines a particular level of generating capacity. The operation of generating units is modeled as a Markov process to calculate the existence probability of each of these states. The reliability figure-of-merit at any time t is the sum taken over all states of the product of: (1) the state existence probability at time t and (2) the probability that the generating capacity defined by that state is sufficient to supply the system load at time t. The proposed scheduling technique was programmed for an IBM 360/65 computer, and applied to a sample power system. The sample study required approximately 60 kilobytes of computer storage. The resulting schedule for the sample system was compared to a schedule which was obtained using more conventional scheduling techniques. The comparison shows that the application of the proposed technique results in a generating unit schedule which meets prescribed system reliability requirements and yields minimum or near-minimum fuel costs.