Transient stability analysis examines the dynamic behavior of a power system for as much as several seconds following a power flow disturbance. Transient stability analysis is concerned with the electrical distribution network, electrical loads, and the electro-mechanical equations of motion of the interconnected generators [1,3,46]. Traditionally, power system transient stability analysis has been performed off-line to understand the system's ability to withstand specific disturbances and the system's response characteristics, such as damping of generator oscillations, as a system returns to normal operations. Such contingency studies were limited to the system design/upgrade phase in order to ensure robust network design and limited to operator training exercises in order to assist with robust power system operations [3].
To date, the computational complexity of transient stability stability problems have kept them from being run in real-time to support decision making at the time of a disturbance. If a transient stability program could run in real-time or faster-than-real-time, then power system control-room operators could be provided with a detailed view of the scope of cascading failures. This view of the unfolding situation could assist an operator in understanding the magnitude of the problem and its ramifications so that proactive measures could be taken to limit the extent of the incident. Faster transient stability simulation implementations may significantly improve power system reliability which in turn will directly or indirectly affect:
In addition to real-time analysis, there are other areas where transient stability analysis could become an integral part of daily power system operations:
It will be shown that computational requirements are a significant problem with transient stability simulations. The scope of real-time or faster-than-real-time transient stability analysis places this application in the category of being a grand computing challenge that could benefit from future teraflop ( trillion floating-point operations per second) supercomputers. Past research into techniques to enhance the performance of transient stability simulations has included both concurrent processing and better algorithms, however, there are still considerable areas for research into this problem.
This paper provides background on the power system transient stability problems, and discusses various techniques available to improve the computational tractability of the transient stability problem. The goal is to develop efficient, accurate implementations of power system simulations that are sufficiently fast to offer inputs to interactive operator support routines. Techniques are examined to solve differential-algebraic equations --- the computational heart of the transient stability problem. An examination of previous research reported in the literature has been used to form the basis of a discussion of relevant numerical research topics for transient stability analysis. The paper describes plans to develop a parallel transient stability analysis testbed at the Northeast Parallel Computing Center (NPAC) at Syracuse University to permit a consistent test environment to develop new algorithms and parallel computing applications as research is performed on this grand computing challenge. Lastly, estimates of the magnitude of potential speedup in power systems transient stability simulations are presented.