CPU based parallel computation of electromagnetic transients for large power grids

This paper presents the implementation of a parallel sparse matrix solver for improving the computational speed of an electromagnetic transients (EMTs) simulation software. The new method is established on the KLU sparse matrix solver which is suitable for circuit based simulation methods The solver is programmed using parallelization through automatic detection of sparse matrix submatrices separated by the natural decoupling available in transmission line/cable models. The proposed approach is demonstrated in an EMT-type software that uses a fully iterative solution method for all nonlinear models. Furthermore, it is demonstrated for realistic large scale grids.