Utility Customer Database Development Summary

A wide variety of survey, association and other data sources have been used to develop the Commercial, Industrial and Residential Customer Utility Customer and Hourly Loads Databases. The broad scope of the data required in this database (nearly 300 variables for commercial and over 500 for residential) requires that the databases consist of a sample of firms and households rather than a census. Construction of these databases was accomplished in three stages:

Develop Current State and Service Area Level Commercial, Residential and Industrial Population Characteristics. The Utility Customer Databases provide state and service area level detail with a sample of firms and households that statistically represent all customers in the state/service area. Knowledge of current population characteristics is required to select and properly weight samples to accurately represent customers and characteristics of customers within each state/service area. Data sources such as the American Hospital Association's listing of individual hospitals by bed-size category, county-level data on establishments by employee size category, household data from census surveys, construction data, and many other public and proprietary data sources were used to develop detailed characteristics of current commercial and residential customers in each state and service area.
Develop State/Service Area Customer Samples. Once current state/service population characteristics were determined, a sampling plan was developed to insure representative accuracy and to provide a breadth of information important in marketing and load research analysis. The sampling plan was applied to survey data from a variety of sources including publicly-available survey results, surveys conducted on customers in individual utility service areas and surveys of customers developed specifically by Jackson Associates to support MAISY database development.
In cases where an insufficient number of customers was available from the particular state/service area, customers in neighboring geographic areas were "borrowed" for the sample. The energy use characteristics of the borrowed customers were updated to reflect heating, air conditioning and ventilation in the state/service area. This adjustment process was developed and refined in our utility database development projects conducted over the past 20 years. Convenience stores in Georgia "look" just like convenience stores in North Carolina, after accounting for weather-sensitive end -use energy adjustments. This same borrowing technique is used to represent new buildings in the building stock; that is, a new construction building record is represented in the databases by borrowing a recently constructed building and updating its energy use to reflect new construction practices and trends in new equipment efficiencies.
Disaggregate Building Energy Use into End Use Hourly Energy Use. Energy is used for end-use services (heating, lighting, etc.); consequently end-use energy detail is important in energy marketing and load research. Building energy use in each survey record was disaggregated into up to 10 end uses with statistical/engineering model analysis developed over the last decade in our utility modeling applications. For example, an automated statistical/engineering model analysis is used to process 15 months of billing file data (monthly kWh and peak kW) for commercial customers along with site-specific 8760 hourly weather data and individual customer building, equipment and operating characteristics to develop and calibrate 8760 hourly loads consistent with actual billing month kWh and peak kW.