Cluster Computing: A Tool for Looking Before Leaping

Much power system behavior goes unstudied because it lies outside the conventional software environment of the industry. Large cascading outages, for example, are rare events and are not considered in reliability investigations. Experience in using a cluster to investigate cascading outages will be reported in which 160,000 major disturbances were produced for a 3000 bus system. Cluster applications for investigation of seams problems and examining the implication of different policies will be presented. The low cost ( $5.2 million) and the speed of Virginia Tech’s System X (the third fastest computer in the world as of November 2003) made up of 1100 Mac G5 computers suggests that we can now examine more versions of the future than previously believed possible.