Propagating Uncertainty in Power System Dynamic Simulations Using Polynomial Chaos

Quantifying the uncertainty of the renewable energy generation units and loads is critical to ensure the dynamic security of next-generation power systems. To achieve that goal, the time-consuming Monte Carlo simulations are usually used, which is not suitable for online dynamic analysis of large-scale power systems. To circumvent this difficulty, two uncertainty quantification approaches using polynomial-chaos-based methods are proposed and investigated. The first approach is the generalized polynomial chaos method that is able to reduce the computing time by three orders of magnitude compared with Monte Carlo methods while achieving the same accuracy. We find that this approach is very useful for short-term power system dynamic simulations, but it may produce unreliable results for long-term simulations. To address the weakness of that approach, we present the second method, namely the multi-element generalized-polynomial-chaos method. It is seen that this method is more accurate and more numerically stable than the generalized polynomial chaos method. Since the uncertainties of the renewable energy generation units and loads can follow very different distributions, we extend the Stieltjes' recursive procedure that allows us to derive the orthogonal basis functions for any assumed probability distribution of the input random variables. Extensive simulations carried out on the WECC 3-machine 9-bus system and the New England 10-machine 39-bus system reveal that our proposed approaches are able to produce comparable accuracy as the Monte Carlo based method while achieving significantly improved computational efficiency for both stable and unstable power system operating conditions.