Stochastic Sensitivity Methods Applied to Shipboard Power Systems

In the design of large-scale and dynamic power systems, sensitivity analysis is an effective tool for ranking the importance of variations in governing parameters, and quantifying their interactions. In this work, we propose new methods for stochastic sensitivity analysis, which extend the standard approach by Morris. We apply these to subsystems of an all-electric ship model having multi-rate dynamics. This model consists of an AC power distribution grid and a propulsion system. In a startup transient, we identify the most important parameters in the short-time regime, where the electrical components dominate, and in the long-time regime where the mechanical components dominate.