Public Utilities Reports

PUR Guide 2012 Fully Updated Version

Available NOW!
PUR Guide

This comprehensive self-study certification course is designed to teach the novice or pro everything they need to understand and succeed in every phase of the public utilities business.

Order Now

Metering, Loads & Profiles: Let the Cherry-Picking Begin

Fortnightly Magazine - November 1 1997

rate. All four rates are uncontrolled, non-time-of-day rates. All rates have a strong seasonal price signal with winter unit charges approximately 1.5 times summer charges.

In each experiment, 10 to 20 percent of the customers identified by the strategy are removed from the sample. The load factors of the sample before and after removal, and the load factor of the removed customers taken as a "mini-class," are calculated. Load duration curves are then drawn. The strategies are implemented and selections for removal are made once, based on January data. Calculations are also done on July data so that the implications of the selection strategy over time can be evaluated.

The data is based on load research efforts by Central Vermont Public Service, a small, winter-peaking utility with a highly developed price structure that relies heavily on seasonal and time-of-day pricing. The reliance on this type of pricing allows for and encourages the penetration of many off-peak load management-type rates.

For large customers at Central Vermont, rates for primary service and transmission are virtually 100-percent interval metered. Thus, on the Central Vermont system, the sample and the population for this group of customers are identical. As such, any post-restructuring LSE targeting these large customers will know precisely what service has been provided. Importantly, the entity serving the remaining customers via a standard offer, or as a provider of last resort, also will know what service it provides.

For small customers, the situation is quite different. The patterns of use for these mid-size and small customers are estimated from a sampling designed to replicate the load shape of each rate class with a 95-percent confidence level. Stratification techniques are employed for some but not all of the class samples. About 400 sample points are used to replicate the loads of approximately 140,000 customers at Central Vermont. The fact that there is variation in the population sample should not surprise anybody. How much variation and the associated impact of using statistically estimated load profiles is the issue here.

Figure 1 provides an example of the type of output produced by the experiments. Three normalized load duration curves are drawn for January data for demand-metered general service (Rate GD). The line marked "pre" is the load duration curve before any customers are removed from the sample. The line marked "post" is the load duration curve of the sample after the customers with the highest load factors (Strategy 1) are removed. The load duration curve marked "delta" represents the "mini-class" curve of the cherry-picked customers. In this simulation, the LSE has identified and cherry-picked the highest load-factor customers out of the profiled class. The execution of this strategy degrades the load factor of the customers that remain in the load profile.

Figure 2 contains class peak-day load profiles for the pre-, post- and cherry-picked customers for Rate GD, Strategy 1. The cherry-picked customers, taken as a mini-class, have a much higher peak day load factor than the profiled class taken as a whole.

Figure 3 depicts peak-day load shapes for Rate GD for Strategy 3. This strategy