Large Scale Optimal Power Flow

Abstract

A new optimization method is applied to optimal power flow analysis. The method is shown to be well suited to large scale (500 buses or more) power systems in that it is computationally efficient and is particularly effective with infeasible starting points. The optimization approach is based on transforming the original problem to that of solving a sequence of linearly constrained subproblems using an augmented Lagrangian type objective function. A fundamental feature of this algorithm (developed by Murtagh and Saunders) is that the solution converges quadratically on the nonlinear power flow constraints, rather than being forced to satisfy the constraints throughout the iterative process. To demonstrate the performance of this algorithm, a set of descent directions, which includes quasi-Newton (variable metric), conjugate directions, and steepest descent, are compared on the basis of convergence and computational effort for a 118 bus and a 600 bus power system.