Price Normalization to Facilitate Energy Efficiency— Algorithms and Issues

One of the key DR applications involves the use of smart appliances and devices within a facility (residential or small commercial buildings) that can be programmed and automated to respond to DR signals (that could be either pricing signals or other energy curtailment signals) arriving from a utility, wholesale market, or a 3 party energy service providers. Several technical, business, and policy challenges need to be overcome before such DR applications become ubiquitous and gain widespread consumer acceptance. In this paper we concentrate on the abstraction of dynamic prices to Simple Levels, and propose metrics for quality of mappings. We also address general issues with mapping including volatility, seasonal variation, and effective use of Levels where price (hence cost) is abstracted away. Several significant issues arise in performing price normalization. 1. The general range of prices changes over time—from summer to winter, from year to year. 2. The mapping between price ranges and QoS provided to consumers from various smart devices. Furthermore, these mappings need to consistent among different price providers, software, and/or curtailment requesters. 3. The normalization must express information needed to properly respond to actual price information, and must show differentiation across time In this paper we examine issues on evaluation of price normalization algorithms, and examine several classes of algorithms, including moving averages, statistical analysis of history, and hybrid approaches. We apply them to sample historical price streams where wholesale prices are used as proxies for retail prices, and compare to actual approaches used in pilot projects