|dc.description.abstract||Buildings account for approximately 40% of all U.S. energy usage and carbon emissions. Reducing energy usage and improving efficiency in buildings has the potential for significant environmental and economic impacts. To do so, reoccurring identification of hardware and operational opportunities is needed to maintain building efficiency. Additionally, the development of controls that continually operate building systems and equipment at energy optimal conditions is required.
This dissertation provides contributions to both of the aforementioned areas, which can be divided into two distinct portions. The first presents the framework for the development of an automated energy audit process, termed Autonomous Robotic Assessments of Energy (AuRAE). The automation of energy audits would decrease the cost of audits to customers, reduce the time auditors need to invest in an audit, and provide repeatable audit processes with enhanced data collection. In this framework of AuRAE, novel, audit-centric navigational strategies are presented that enable the complete exploration of a previously unknown space in a building while identifying and navigating to objects of interest in real-time as well as navigation around external building perimeters. Simulations of the navigational strategies show success in a variety of building layouts and size of objects of interest. Additionally, prototypes of robotic audit capabilities are demonstrated in the form
of a lighting identification and analysis package on a ground vehicle and an environmental baseline measurement package on an aerial vehicle.
The second portion presents the development and simulation of two advanced economic building energy controllers: one utilizes steady-state relationships for optimizing control setpoints while the other is an economic MPC method using dynamic models to optimize the same control setpoints. Both control methods balance the minimization of utility cost from energy usage with the cost of lost productivity due to occupant discomfort, differing from standard building optimal control that generally addresses occupant comfort through setpoint limits or comfort measure constraints. This is accomplished through the development of component-level economic objective functions for each subsystem in the modeled building. The results show that utility cost and the cost of occupant productivity from optimal comfort can be successfully balanced, and even improved over current control methods. The relative magnitude of the cost of lost productivity is shown to be significantly higher than the cost of utilities, suggesting that building operators, technicians, and researchers should make maintaining occupant comfort a top priority to achieve the greatest economic savings. Furthermore, the results demonstrate that by using steady-state predictions, the majority of the performance gains produced with a fully dynamic MPC solution can be recovered.||en