Interaction variables for distributed numerical integration of nonlinear power system dynamics

Abstract

This paper describes two novel methods for distributed simulation of nonlinear power system dynamics, which are named two-level integration and interaction-level integration. An underlying assumption is that the simulation of dynamic response is distributed spatially, i.e. each power system component and/or each group of components perform numerical integration of their own states while occasionally communicating relevant information with their neighbors. The novelty of the proposed methods is in exploiting the structure of dynamic equations of typical power system components to shorten the total simulation time in this distributed setting. The proposed methods consist of a number of state transformations on the dynamical power system model which bring this model into a form favored for distributed numerical integration. These transformations, proposed first in [1] and [2] for control design, separate internal component dynamics from dynamics of its interactions with neighboring components. It is further shown on a simple example that it is sufficient to exchange the information between components at a rate determined by the dynamics of the interaction variables to successfully perform numerical integration of the entire power system model with a significant improvement in simulation time. Since the proposed methods are based on the state transformation of the power system model, this approach could be used together with other numerical and machine-level programming optimization techniques to further enhance performance of the power system dynamics simulator.