Improvement of Customer Class Load Schedules Utilizing AMI Measurements

Abstract

The customer class load schedules traditionally used by electric utility distribution management systems (DMS) inform system-level modeling and analysis, including distribution power flow, which in turn dictates decision making at the most foundational levels. These load schedules vary based on the customer's load category (e.g., residential, commercial, and industrial), season, and type of day (e.g., weekend or weekday). In the absence of detailed customer data, load schedules have conventionally been derived from heuristic techniques, assumptions, and examples, and in some cases have remained static as the modern power grid has evolved to contain more modern load types such as LED lighting fixtures, smart appliances, and household electric vehicle charging stations. Given the advent of more readily-available data due to advanced metering infrastructure (AMI), this work provides data-driven improved customer class load schedules that decrease average error across a particular load category. Additionally, the improved schedules will be shown to decrease error in the aggregate when viewed from the level of a distribution feeder.