Confidence intervals for adequacy assessment using Monte Carlo sequential simulation

The most common way for probabilistic adequacy assessments of composite generation/transmission systems is based on Monte Carlo (MC) simulation. This simulation can be either non-sequential or sequential. Non-sequential MC simulation has a better efficiency, but sequential MC simulation is required to consider time-related phenomena, such as storage systems, or to compute interruption frequency and duration indices. A large number of independent 1-year sequential MC simulations are usually run to estimate confidence intervals based on classical formula for independent and identically distributed random variables. For real-size systems, this technique is hardly applicable due to the computing time needed to have a good estimation of the sample variance. The aim of this paper is to propose alternative techniques to estimate confidence intervals of reliability indices given by a single 1-year sequential MC simulation. These techniques are applied to two systems, the Reliability Test System and a model of the United Kingdom transmission network, and are compared to the usual approach.