Conference Year





Previous research on automated fault detection and diagnostics (FDD) for HVAC systems has shown promising benefits like earlier detection and more accurate isolation of different faults. While most researchers, equipment manufacturers, and policymakers agree that HVAC system FDD is important and has the potential to reduce significant energy waste due to faulty system operation, widespread adoption of these tools has been slow. An automated fault detection and diagnosis system has been developed for packaged (rooftop) air conditioners based on the VOLTTRONTM monitoring and controls framework developed by the Department of Energy. The system implements a virtual-sensor-based FDD methodology capable of isolating common rooftop unit faults such as improper refrigerant charge level, heat exchanger fouling, liquid-line restrictions, and compressor valve leakage. A fault impact evaluation component has also been implemented in order to determine the relative impact that faults have on system performance. This is accomplished using virtual sensor outputs and manufacturers’ performance map reference models for performance indices such as cooling capacity and COP. This system has been implemented using low-cost electronics components and was be tested using a 5-ton RTU in a laboratory environment. In this work, a high-level overview of the automated rooftop unit (RTU) FDD system structure will be presented detailing how individual software agents interact along with a description of the computational and network requirements of the system. Alternative system architectures will also be discussed in comparison to the hybrid system presented. A review of the FDD algorithms is also presented that details the virtual sensors implementations along with the methodology to detect, diagnose, and evaluate different faults.  Finally, the performance of the FDD system will be demonstrated using laboratory test data collected for a 4-ton RTU with micro-channel condenser. The goal of this research is to produce a field ready FDD tool for RTUs that can be used to show the benefits of FDD in real systems. Ultimately, the software implementation (using Python) and hardware designs of all the systems components will be released under an open source license in an effort to reduce the engineering effort required by equipment manufacturers interested in a complete AFDD solution.