Analysis, modeling and optimization of residential energy use from smart meter data
MetadataShow full item record
Approximately 38% of electricity consumption within the United States can be attributed to residential buildings, a vast share of which is in heating, ventilation and cooling. The load placed on the grid by residential consumers is highly variable and strongly influenced by weather and human activity patterns. Meeting fluctuations in demand is challenging and expensive for electricity producers and grid operators. Reducing variability in residential energy use can contribute significantly to increasing the uniformity of energy demand on the grid and diminish reliance on inefficient, polluting “peaking” plants that are used to meet extremely high demands. Achieving this goal requires tight coordination between energy consumption and generation, as well as the means to store energy generated in periods of low demand for use during the time intervals when consumer demand peaks. There is a common perception that a single home has a minor impact on the entire grid. However, owing to the fact that consumption patterns of homes are similar, while a single home does not have a large impact on the grid, entire neighborhoods do. Motivated by the above, this work explores the interaction between residential energy consumption and the electric grid. An analysis, modeling and optimization framework on smart meter data is developed to anticipate and modulate energy usage of ensembles of residential homes in order to reduce peak power demand. Much of the data used in this work come from Pecan Street, Inc., a smart grid demonstration project in Austin, TX. First, a nonintrusive load monitoring algorithm is developed to isolate air-conditioning (A/C) energy use from whole-house energy consumption data. Subsequently, a simplified reduced-order model is derived from smart meter data and thermostat set-point data to predict A/C energy use. The models of an ensemble of homes are placed within a centralized model predictive control scheme to minimize peak community A/C energy use. Reductions in peak energy use are achieved by shifting the thermostat set-points of individual homes. The approach is further expanded by simultaneously scheduling the operation of time-shiftable appliances to further reduce the community peak load. This integrated operation reduces peak loads by an average of 25.5%. This work also considers the impact of control and optimization techniques on designing a micro-grid that operates near autonomously from the electric power grid. Lastly, this work presents a tool to compare energy demand patterns of houses from smart meter data and indicates that high-energy houses would benefit from energy audits to improve energy efficiency.