Iterative solver techniques in large scale power system computation

Abstract

The solution of large sparse linear systems forms the core of power system computations whether it is power flow, state estimation or security assessment (static or dynamic). There is a continuous need to speed-up this process by improved numerical algorithms. Iterative solver techniques based on the Krylov subspace projection method offer an attractive alternative to the traditional LU factorization methods because of the ease of parallelization and/or vectorization. In this paper we review the efforts since the late 80's in applying these techniques to power system problems.