CHEAPER: A novel, mixed integer, linear program to minimize commercial building electricity costs under real-time conditions

Abstract

Commercial buildings account for 35 percent of all U.S. retail electricity sales. The average commercial customer's electricity bill is $647 a month and utility costs, on average, represent 17 percent of a commercial building's annual operating budget. Technology that can save businesses money by automatically reducing or shifting building electricity use based on real-time pricing, weather and occupancy is highly desirable. This thesis presents a novel, mixed integer, linear program called Cooling and Heating Efficiently through Automated Planning of Energy Resources (CHEAPER). CHEAPER creates optimal operating schedules for one or more building systems that minimize the total periodic electricity cost of system operation. The program uses building designs, system schedules, and local weather forecasts to model indoor temperature change based on outdoor conditions and building activities. Occupant comfort is addressed through use of one or more user-specified constraints pertaining to acceptable indoor thermal and visual conditions. Real-time pricing accessed through a utility web portal provides the 5-minute, electricity, spot prices necessary for cost planning over a 24-hour time horizon. Due primarily to CHEAPER's size and RTP cost symmetries, the majority of problem instances do not solve fast enough to be practical for everyday use. To alleviate this issue, a relative optimality threshold, or gap, is used to relax the requirements for optimal CHEAPER schedules, which significantly decreases the program's runtime. With a 1.5 percent optimality gap, CHEAPER solutions are obtained, on average, within 45 seconds of program start. This gap size equates to an increase in daily electricity costs of $0.02 to $0.08. Under these conditions, application of CHEAPER to a prototypical small office building located in northeast Kansas demonstrated daily cost savings of two to 55 percent as compared to the same building and systems operating with standard control strategies. Average savings of 22 percent were achieved. Cost savings are a result of three control strategies: occupancy control, light-level dimming and load shifting. For the average customer, use of CHEAPER schedules could result in an average, annual cost savings of $1,025. CHEAPER also produced consistent monthly energy savings, which ranged from 11 to 33 percent as compared to the baseline model. The most important research need related to CHEAPER is the need for its demonstration in actual commercial buildings. The program must be tested in-situ to validate the approach and savings potential detailed in this thesis. In addition, CHEAPER currently includes a relatively small suite of control options and a single electricity-cost objective. Many other building features and optimization opportunities are possible such as expansion of the program to accommodate multiple building HVAC and lighting zones. Similarly, research to address the competing objectives of cost and carbon emissions reduction is needed to ensure CHEAPER can serve as a tool for meeting both energy and environmental goals.