Comparison of accuracy in load-flow programs
Ron Roebuck, C.H. Saylor, Yehuda Wallach · 2002
The equations for solving the load-flow problem by the Newton-Raphson method are developed in order to show that they are extremely sparse. By running LINPACK, optimally ordered factorization and parallel optimally ordered factorization routines, it is shown that the last two generate rather high roundoff errors, especially when close to a solution. The overall conclusion is that because of error accumulation a larger number of iterations is needed to solve the load-flow problem by the optimization methods than by the LINPACK routines. Accuracy was traded for memory space.>