Grant: Alliance Grants

We propose an international research collaboration to develop novel and hardware-ready algorithms to support the automatic fault detection and diagnosis (FDD) of fan coil units (FCUs) in heating, ventilation, and air conditioning (HVAC) systems. FCUs provide local heating and cooling to occupied spaces, thus their ineffective or faulty operation leads building's occupant discomfort and significant wastage of thermal energy. Existing data-driven solutions for FCU FDD are either limited to diagnosis only (unsupervised methods) or require a large labelled dataset (supervised methods) and tend to require access to large amount of data, resulting in high complexity computations that prohibit real-time or edge implementation. Conversely, existing model- or physics-based solutions have high computational cost and require fast-response hardware controllers for real-time implementation, limiting their applicability. Further, these tend to consider limited fault types (e.g. fan failure) or very specific unit configurations; generalized solutions for FDD for FCU operation are currently lacking. In this proposed research, we adopt a generalized hybrid rule-based and data-driven strategy to develop a novel FDD algorithm to detect and diagnose FCU faults. By collaborating with Dr. Tanougast at Université de Lorraine (UdeL), this algorithm will be specifically refined for edge device implementation to facilitate its deployment. This preliminary research is anticipated to develop not only a prototype algorithm but also set the stage for a longer-term collaboration between TMU and UdeL to develop a more comprehensive suite of edge devices to support Smart and Ongoing Commissioning, achieving improved occupant comfort and energy efficiency in buildings.