Project Details
Recent rapid advances of variable energy resources and power electronics have begun to change the landscape of the power industry and give rise to a far more dynamically controlled grid. The objective of this research is to apply the latest computational algorithms and parallelization techniques to enable faster than real-time power system dynamic simulations. The current state-of-the-art in electric power system or grid computation mainly involves direct methods of stability analysis that do not explicitly solve differential equations and transient system trajectories. The power flow computation, being the fundamental tool used in most of the power system analysis has been the target of most of the development efforts so far. The parallel computing has been primarily used to model many independent power flow cases and not to speed up the individual calculations. This effort will evaluate existing and develop new fast methods for time-domain simulations of power system dynamics using numerical integration of the nonlinear transient differential equations.
The topics of the research include development of an adaptive multi-level parallelism: single-case in one dimension (time), independent multi-case parallelization in two dimensions (time-case), addition of multi-area spatial decomposition in three dimensions (time-case-space); combination with other parallelization approaches such as method for four dimensions (time-case-space-method) and introduce self-adaptive, dynamic partitioning in the multi-dimensional universe for optimal task allocation.
