Modeling Load Forecast Uncertainty in Power System Reliability Studies

A review of the literature on the topic indicates that the assumption of normality to model load forecast uncertainty in power system reliability studies is valid [1,2,3,4]. From a practical perspective, the use of normal curves models the possibility of extreme loads that have not been observed in history. These extreme loads represent rare events that have a very low probability of occurrence. In critical systems, such as the electric power network, the occurrence of these rare events can have a very high societal cost and thus, must be included in reliability analyses. (The total economic cost of the August 14, 2003 blackout across the eastern part of North America was estimated to be $7 to $10 billion.)[8] The practice of analyzing the impact of extreme events with low probabilities is common to many industries that provide a vital service such as aerospace, nuclear power, and commercial/industrial construction. Furthermore, as discussed below, the use of normal curves is consistent with a recommendation made by a consultant who reviewed PJM’s PRISM model in 2004.