Quantifying flexibility of commercial and residential loads for demand response using setpoint changes. (1st September 2016)

• Record Type: Journal Article
• Authors: Yin, Rongxin; Kara, Emre C.; Li, Yaping; DeForest, Nicholas; Wang, Ke; Yong, Taiyou; Stadler, Michael
• Is Part Of: Applied energy. Volume 177(2016)
• Page Start: 149
• Page End: 164
• Publication Date: 2016-09-01
• Subjects: Demand response -- Thermostatically controlled loads -- Regression models -- Two-state model -- Simplified DR potential estimation; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2016.05.090
• Languages: English
• ISSNs: 0306-2619

Abstract

Highlights: Presents a novel demand response estimation framework for residential and commercial buildings. Applies a combination of EnergyPlus and two-state models for thermostatically controlled loads. Regression models are fit to each dataset for predicting DR potential quickly without any computational burden. Regression model equations are based on key inputs, including hour of day, set point change and outside air temperature. Models are validated for DR measurement in commercial buildings. Abstract: This paper presents a novel demand response estimation framework for residential and commercial buildings using a combination of EnergyPlus and two-state models for thermostatically controlled loads. Specifically, EnergyPlus models for commercial and multi-dwelling residential units are applied to construct exhaustive datasets (i.e., with more than 300M data points) that capture the detailed load response and complex thermodynamics of several building types. Subsequently, regression models are fit to each dataset to predict DR potential based on key inputs, including hour of day, set point change and outside air temperature. For single residential units, and residential thermostatically controlled loads (i.e. water heaters and refrigerators) a two-state model from the literature is applied. For commercial office building and Multiple Dwelling Units (MDUs) building, the fitted regression model can predict DR potential with 80–90% accuracy for more than 90% of data points. The coefficients of determination (i.e. R 2 value) range between 0.54 and 0.78 for the office buildings and 0.39–0.81 for MDUs, respectively. The proposed framework is then validated for commercial buildings through a comparison with a dataset composed of 11 buildings during 12 demand response events. In addition, the use of the proposed simplified DR estimation framework is presented in terms of two cases (1) peak load shed prediction in an individual building and (2) aggregated DR up/down capacity from a large-scale group of different buildings.