Development of Diagnostic and M&V Agents, and Implementation in an Occupied Office Environment

Author(s): Granderson, Jessica; Price, Phillip; Czarnecki, Stephen; Page, Janie; Brown, Rich; Piette, Mary Ann | Abstract: The Building Technologies Office (BTO) at the US Department of Energy is pursuing a framework concept to improve building energy efficiency and increase savings potential in commercial buildings. The transaction based framework enables market-based transactions within and between buildings, and between buildings and the electric grid. Over the past three years, BTO has funded the development of a Transactive Network to supports energy, operational and financial transactions. The initial scope focused on transactions between roof top units (RTUs), and those between RTUs and the electric power grid. Using an open architecture agent--based platform called Volttron [Haack 2013], a set of applications were developed that reside either on the equipment, on local building controllers, or in the Cloud. This report documents the development and testing of an extended set of applications, or agents, for the Tansactive Network, focusing on applications for the lighting end use. Although lighting controls can generally save five to forty percent in end use energy, they remain under-adopted in commercial buildings. Key barriers include high cost and complexity, improper configuration and commissioning that prevent optimal operations and savings, and lack of visibility into achieved savings. In response, this work focuses on two applications - automated fault detection and diagnostics (FDD), and automated measurement and verification (MaV) of energy savings. The agents are designed for schedule-based controls, and in the future, can be enhanced for application to more advanced strategies such as occupancy-based control, and daylighting. The contents of this report detail the programming of the MaV and FDD applications; specifically, the architecture of the Transactive Network lighting system and associated agents, agent inputs and outputs, and underlying logic. In addition, demonstration and testing results are presented, covering a description of the occupied office environment in which the agents were implemented, energy use characterization for the no-controls-automation base case, achieved energy savings, and results from executing the diagnostic agent. The ability to effectively and accurately quantify project specific savings was successfully demonstrated for the MaV agent. This is in contrast to demonstration studies that have focused on establishing general savings levels that can be expected from one control strategy versus another. For example, it is generally accepted that scheduling can save five to fifteen percent in energy use. However, given the atypical occupant behavior in the demonstration site, the measured and verified savings were on the order of sixty percent. These savings are significantly higher than those generally expected when implementing schedule-based controls. This was due to the fact that energy use in the base case was extremely high because of severe under-use of manual controls; and occupants commonly left the lights on all night and all weekend in both demonstration zones. The ability of the FDD agent to successfully identify scheduling faults was also demonstrated in the Transactive Network demonstration site, and with data from an additional set of buildings with schedule-based controls.