Systematic Approach for Finding the Linearizable Regions in Dynamic Power System Models

For dynamic stability assessment, TSOs are simulating large numbers of contingencies on a dynamic model that depicts their own network and beyond. This extension is needed for accurately reproducing the behavior of the real network, but intensifies in return the computational burden. In this paper, a systematic approach is proposed for finding the linearizable regions within the external area, which are suitable for model reduction. The overall idea consists in partitioning the external area into multiple clusters and assessing their level of nonlinearity individually. Upon the end of a temporary disturbance, the free motion of every cluster is confronted with the one of their linear approximation to determine a nonlinearity indicator. The responses are derived analytically to facilitate the integration into existing simulation software. The validity of the approach is assessed on the IEEE 118-bus system model for different fault locations.