Using Five-Minute Data to Allocate Load-Following and Regulation Requirements among Individual Customers

B. Kirby and E. Hirst, Customer-Specific Metrics for the Regulation and Load-Following 1 Ancillary Services, ORNL/CON-474, Oak Ridge National Laboratory, Oak Ridge, Tenn., January 2000. In an earlier project, we analyzed data on total system load and the loads of eight large industrial customers in terms of system- and customer-specific requirements for the regulation and load-following ancillary services. We conducted these analyses using 12 days of data from February 1 1999 plus 12 days of data from August and September 1999. These analyses were conducted using data provided by the control area at the 30-s level, which we then aggregated to the 2-min level for subsequent analysis. The current project analyzes the feasibility of using 5-min revenue-meter (RM) data to allocate load-following and regulation requirements among retail customers. This project does not use the 5-min data to determine the actual ancillary services requirements for individual loads because these requirements depend strongly on the time-averaging period chosen for the load data. In particular, the amount of regulation required declines as the time-averaging period increases. Our earlier project showed that 2 min was a reasonable time-averaging period for this control area. The reason for examining 5-min data is that supervisory control and data acquisition (SCADA) data, which are available every 4 s, are collected for only eight of this control area's very largest industrial customers. On the other hand, this control area collects 5-min RM data for about 600 customers. Although these customers account for less than 0.2% of the retail customers, they account for 60% of total energy sales and more than 40% of total revenues. Thus, if the allocations obtained with 5-min data for these eight large customers agree well with the allocations obtained with the SCADA data for these customers, the method we developed to allocate regulation costs to individual customers can be applied to many more customers--that is, no additional metering or data collection costs are required to charge individual customers for the regulation service. In addition, RM data are more accurate than SCADA data. SCADA data are used primarily by system operators to monitor and control the bulk-power system and must therefore be provided frequently (at least once every few seconds). However, SCADA data need not be accurate enough for billing purposes as RM data must be. In addition, the RMs at this utility record demand down to 1 kW, while the SCADA meters can record data to only 100 kW. The data used in this project consist of 30-s SCADA data (aggregated to the 2-min level) and 5-min RM data for the same eight industrial customers for the same period in February 1999. In addition, we used the same SCADA data on total system load and nonindustrial load as we had used in the earlier project. The next section of this paper presents a comparison of the SCADA and RM data on an hourly basis for the eight customers. Section 3 compares the load-following results obtained with the two data sets. Section 4 compares the regulation-allocation results obtained with the two sets of data. The final section presents our conclusions from this analysis