Accelerated and Localized Newton Schemes for Faster Dynamic Simulation of Large Power Systems

This paper proposes two methods to speed up the demanding time-domain simulations of large power system models. First, the sparse linear system to solve at each Newton iteration is decomposed according to its bordered block diagonal structure, in order to solve only those parts that need to be solved, and update only sub-matrices of the Jacobian that need to be updated. This brings computational savings without degradation of accuracy. Next, the Jacobian structure is further exploited to localize the system response, i.e. involve only the components identified as active, with an acceptable and controllable decrease in accuracy. The accuracy and computational savings are assessed on a large-scale test system.