A comparison of numerical algorithms for solution of the transient stability problem

Abstract

A comparison is presented of five different numerical algorithms for the solution of the differential equations contained in the transient stability problem. A classical transient stability model is used, including the swing equation to characterize the machine's dynamic behavior.

The five algorithms studied are: fourth-order Runge-Kutta; state transition; the trapezoidal rule; Adams-Moulton predictor-corrector; and Hamming predictor-corrector. The five algorithms, using various step sizes, are applied to three test systems, including the IEEE l4-bus system, using a common main program to call each algorithm as an independent subroutine. The results obtained are compared with regard to accuracy, solution time, and ease of programming. The comparison of fourth-order Runge-Kutta to the other methods is given special emphasis.