A Parallel Algorithm for Transient Stability Analysis Using Sparse Vectors and Inverse Factors

Abstract

A new parallel algorithm for time domain transient stability analysis is proposed. It is built over a framework in which an interlaced partitioned scheme for the transient stability problem is assumed and the network equations are solved by LDU factorization and sparse vector techniques. The parallelization approach is based on the factorization path tree and its main feature is the switching from direct factor to inverse factors to best exploit the parallelism during the solution of the network equations. The task scheduling procedure assumes a computational model that exhibits both global and local memories. A tutorial example and application results showing the effectiveness of the algorithm are also presented.