(59c) Proactive Strategies for Building Energy Management in the Smart Grid

Residential and commercial buildings account for 40% of U.S. primary energy consumption, and a significant portion of this energy supports heating, ventilation and air conditioning (HVAC) systems. Recently, utility companies have been adopting smart electric meters and advanced metering infrastructures (AMI) as a means to support demand response and distributed generation.  In 2011, over 20% of utility customers were equipped with smart meters (a 60% increase from 2010), and Texas has one of the highest smart meter penetrations, at over 50% [1].

Smart metering and forecasts of consumer energy use enable the application of advanced control techniques to synchronize the demand of energy with the operation of power facilities, resulting in significant reductions in peak electricity demand. However, this requires a paradigm shift from the standard “on-off” or simple linear controllers that are still the norm for temperature regulation in residential buildings. The inherent complexity of the building energy management process thus lends itself naturally to the use of an optimization-based solution. The practical implementation of such an approach is, however, hindered by the considerable number of equations and complexity of the mathematical models representing the time evolution of building temperatures and energy consumption; such models are not amenable to real-time, on-line computations.

Motivated by the above, we have developed a general model reduction procedure that yields a portable, first-principles, nonlinear model of the building dynamics. Using the reduced model, we have extended the Economic Nonlinear Model Predictive Control (E-NMPC) framework developed for chemical processes, to formulate and solve the predictive energy management problem for residential buildings. We have shown that significant operating cost and energy savings are possible by proactively accounting for disturbances (e.g., changes in cloud cover, outside temperature, building occupancy, time-of-day energy prices).

Existing E-NMPC theory (including stability properties) and the aforementioned applications are developed for systems with continuous states and manipulated inputs [4]. However, the operation of HVAC equipment (e.g., fan speed, or fuel flow to a heater) also includes modifying their operating state (i.e., turning them on or off). In this talk, we present results extending E-NMPC theory to include hybrid systems with binary manipulated inputs, and discuss several case studies emphasizing the relevance of these theoretical developments to energy management in residential buildings. Additionally, we present an example concerning mitigating demand fluctuation via the implementation of on-site high-capacity energy storage for large-scale electricity consumers (e.g., commercial buildings or residential neigborhoods), which improves the synchronization between consumer demand and energy production, and results in smoother grid load profile throughout the day.

[1] U.S. Energy Information Administration. “Smart Meter Deployments Continue to Rise”. Nov. 1, 2012. <http://www.eia.gov/todayinenergy/detail.cfm?id=8590>.

[2] C. Touretzky and M. Baldea, “Modeling and Model Predictive Control Strategies for Building Energy Management”. AICHE Conference Presentation, Nov 1. 2012.

[3] C. Touretzky and M. Baldea, “Model Reduction and Nonlinear MPC for Buildings”. American Control Conference 2013 (under review).

[4] M. Baldea and C. Touretzky, “Concurrent Nonlinear Predictive Control and Economic Management of Energy-Integrated Systems”. European Control Conference 2013 (under review).