Quantifying Changes in Building Electricity Use, With Application to Demand Response

Abstract

We present methods for analyzing commercial and industrial facility 15-minute-interval electric load data. These methods allow building managers to better understand their facility's electricity consumption over time and to compare it to other buildings, helping them to 'ask the right questions' to discover opportunities for demand response, energy efficiency, electricity waste elimination, and peak load management. We primarily focus on demand response. Methods discussed include graphical representations of electric load data, a regression-based electricity load model that uses a time-of-week indicator variable and a piecewise linear and continuous outdoor air temperature dependence, and the definition of various parameters that characterize facility electricity loads and demand response behavior. In the future, these methods could be translated into easy-to-use tools for building managers.

Uncertainty estimation improves energy measurement and verification procedures

“…With this knowledge, we selected a model structure that is limited to the "time of week" and the outdoor air temperature as predictor variables. A similar model is described in more detail in [2]. This development and demonstration of this research applies to any general forecast model, such as one that includes additional explanatory variables (e.g., humidity, occupancy).…”

“…Mathieu et al [2] provides a good summary of energy prediction methods. Coughlin et al [3] considers methods that average load profiles from the last several days.…”

Bayesian Energy Measurement and Verification Analysis

“…4. The forecasting accuracy of other machine learning (ML) methods can be higher than regression in some cases [31,32], although regression-based approaches such as time-of-week-and-temperature [33] still perform very well [32,34] and may be preferred for simplicity. Note that this is a limitation of regression, not the overall Bayesian paradigm, although regression is the way most M&V analysts would use Bayesian methods.…”

Review of Prior Commercial Building Energy Efficiency Retrofit Evaluation: A Report to Snohomish Public Utilities District

“…Figure 6 shows the load and temperature data from the Pre period, and predictions from the resulting model fit to these data. The model whose results are shown here is described in Mathieu et al (2011); it is a regression model in which the prediction at a given time depends on the time during the week -analogous to time of day, but counting sequentially through a 7-day period rather than a single day -and on the outdoor air temperature, where the dependence of load on outdoor air temperature is assumed to be piecewise-linear. This particular model is used here only to illustrate the point that a statistical model can capture at least some of the time-and temperature-dependence; we don't claim this is the best model that could be created for this building.…”