Development and validation of aggregation method for fault detection and diagnostics in HVAC systems

This paper describes the development, demonstration, and evaluation of a fault detection and diagnostics (FDD) system that integrates a fault aggregation methodology. Many FDD systems provide actionable information based on individual events, which sometimes results in misleading information going to the building operators. The primary aim of this work was to enhance diagnostics at the component and subsystem levels by leveraging statistical analysis to inform better decision-making in building operations. Although similar methods have been used in other fields, they have not been used in the buildings field. The proposed fault aggregation method uses rules from engineering principles, analyzing independent diagnostic results through the binomial probability distribution function to calculate detection probabilities with adjustable sensitivity thresholds. By aggregating fault detections over daily, weekly, or monthly periods, the system provides a comprehensive and user-friendly approach for building operators to manage real and false alarms effectively. This significantly reduces alarm overload and enhances confidence in FDD applications. The annual aggregation results of the economizer diagnostics for five rooftop units and 19 air-handling units (AHUs) with variable-air-volume boxes across seven different buildings showed 79% with one or more faults. The results showed 67% of AHUs having at least one fault and 58% having multiple airside faults. Furthermore, the paper suggests incorporating economic evaluation techniques to balance service costs with fault impacts, ultimately optimizing FDD systems for improved operational efficiency and economic returns. The findings underscore the potential for more robust FDD performance measurement beyond basic alarms or actionable information, highlighting areas for future research and development in FDD aggregation capabilities.