Techniques for large scale power system simulation

Abstract

This research entails the study of the applicability of Monte Carlo simulation to the study of the reliability of large scale power system simulation. Reliability in this context refers to the ability of the system to meet the demand for electricity over time. As a part of this study a generalized program capable of modeling any pool of generators has been developed using a modified version of the GASP-IV simulation language. GASP-IV was modified to enhance the event filing and retrieval procedures in order to reduce the time required to execute a simulation program. Monte Carlo simulation was first applied to power system reliability modeling by an operations research team at Westinghouse in the late 1950's. This effort was abandoned in the mid-1960's due to high computation costs. The Generator Simulation System (GENESIS) developed during the present research has demonstrated that power systems can be simulated in an effective and computationally efficient manner, having execution times on the order of one minute per simulated year for a large pool of generators on an Amdahl 470/V6 and having results which compare reasonably with historical data. Much of the computational efficiency of GENESIS is due to enhancement of the GASP-IV simulation language, particularly in the improvement of the structure used for event filing (the synchronization procedure). A new synchronization procedure, known as the two-list procedure, was developed during this research. Implementation of the two-list procedure in GASP-IV reduced the execution time of GENESIS by about 50%. The procedure is simple to implement, and can easily be used with GASP, GPSS, or SIMSCRIPT, and should result in substantial saving in execution time for any simulation having a large list of pending events. This research has demonstrated that Monte Carlo simulation can be used to model power system reliability in an efficient and effective manner. A modified version of GASP has been created which is capable of executing such simulations in a small fraction of time required by unmodified GASP. Further, several areas in which future models may be improved have been identified and discussed.