Periodic steady state solution of power systems by selective transition matrix identification, LU decomposition and graphic processing units

This paper describes the application of parallel processing techniques based on graphic processing units (GPUs) to the fast steady state solution of electric power systems. A Newton technique based on a Numerical Differentiation (ND) procedure is used. Conventionally, this technique requires of a significant computational effort, measured in terms of computation time. Parallel processing technology based on GPUs is applied in this investigation to considerably reduce the computational time required to reach the steady state solution of the reported case studies. It is shown that the proposed parallel processing scheme is an effective and efficient alternative to obtain the steady state solution, in particular for larger-scale power grids.