A simulation analysis of the operational issues in power generating system reliability

Abstract

The purpose of this dissertation is to extend and validate GENESIS (GENErator SImulator System), a power system simulation model, and to study the relative importance of a group of variables in the estimation of power generation system reliability. The variables considered have been classified into system characteristics and modeling issues. System characteristics relate to the power system configuration, and include the following variables: generator mix, system size, load cycle, and installed reserve. Modeling issues relate to specific system capabilities which could be included in the modeling effort; the variables included are: planned outage management, outage postponability, duty cycle, and start-up delay. To study the significance of system characteristics and modeling issues, as well as the interactions between the two variable classes, a factorial experiment was constructed. Three simulation responses were used to perform the statistical analysis, namely, the average number of negative margin hours per year, the average number of negative margin events per year, and the average unserved energy per hour. From the results it has been concluded that to model systems with various configurations appropriately, it is necessary to consider all the modeling issues, except planned outage management. The planned outage schedule for the various systems considered has been built by alloting the maintenance periods to the valleys of the load cycle curve. A better system reliability performance was observed by following precisely this schedule and not by allowing the simulation model to manage the planned outages. Before the experiments of the factorial design were performed, three preliminary tasks were accomplished. The first task involved the extension of the GENESIS model by including some possibly important operating realities such as minimum up-times and planned outage management. Second, the GENESIS model was validated using historical data from two power generating systems. The third task consisted of the development of a run-length control procedure to be used internally by the GENESIS model during the running of the cases in the experimental design. This alternative stopping rule was developed because of the excessive number of observations required by other available procedures.