Probability of Peak Analysis for Time-of-Use Rate Design
6 November 2008
By Amparo Nieto
NERA Senior Consultant Amparo Nieto delivered this presentation to utility rate and costing professionals at the Marginal Cost Working Group meeting in Boston on 6 October 2008. One of the key requirements for efficient Time-of-Use (TOU) rates is producing accurate hourly estimates of all the time-varying components of the marginal costs of service -- namely generation, transmission, and high-voltage distribution. The goal is building TOU periods that capture the expected variation in marginal costs across periods and seasons over the year (or years) when the rates are going to be in effect. In her presentation, Ms. Nieto provided a technical discussion of a "Probability of Peak" (PoP) statistical method that allocates annual marginal capacity costs to hours based on each hour's relative probability of being the peak of the year. The method may be adapted to estimate probabilities of seasonal or monthly peaks, depending on what drives the need for investment in each case. The PoP analysis takes into account historical load patterns in recent years and captures weather variability. Ms. Nieto emphasized the importance of testing for normality of the hourly data distributions and presented the results of a case study to discuss the impact of using data transformation techniques such as the Box-Cox method on the resulting probabilities of peak.
