CH-18-015 -- Software-Based Fault Detection for Multicircuit Building Lighting Systems PDF

This paper is based on findings resulting from an ASHRAE 2017-2018 Graduate Student Grant-In-Aid award.

Lighting systems account for over 11% of total commercialbuilding energy use and represent a similar proportion ofenergy wasted due to faults (NRCan 2013). Many buildingmanagers use aggressive preventative maintenance to addresslighting faults and, as such, will often prematurely changebulbs on a schedule. With a typical bulb life of more than 20,000hours, premature maintenance can generate significant andpreventable waste. Power submonitoring has become popularover the last decade, providing researchers with additionalopportunities to monitor and manipulate live lighting systems.Floor-level power monitoring granularity is not an unreasonableexpectation in modern buildings. This paper uses floor-levelbuilding automation system (BAS) data to characterizethe lighting system, develop active and passive fault detectionapproaches, and present the data in a manner that is easy toassimilate. The active approach is based on systemic manipulationand monitoring of the lighting circuits, while thepassive approach was based on deductive observations duringruntime. It was found that both active and passive fault detectioncould identify faults within one hour or slightly over onestate change, respectively, and that even with only one powermeter, inferences could be made for multiple circuits.