Judgmental and statistical methods of peak electric load management

This paper presents the results of a study comparing two computerized peak-shaving systems for an electricity distributing utility company. A judgmentally-based system was designed to replicate an actual human expert in the field, while a statistically-based system was developed to incorporate statistically-based decision rules to replace some of the purely judgmental rules of the expert. These peak management systems were tested against each other, and then tested against the human expert for periods in which data on his actual decisions were available. The results show that when the objective is to correctly predict peaks, and the number of false alarms is not important, the human expert should be used. Alternatively, when the objective is to reduce system nervousness by limiting the number of false alarms, the statistically-based peak management system should be used.