Quantifying Performance Degradation of HVAC Systems for Proactive Maintenance Using a Data-Driven Approach

Abstract

Poorly maintained and degraded Heating, Ventilating, Air Conditioning (HVAC) systems waste significant amount of energy. Current Facilities Management (FM) practice is mostly based on reactive and scheduled maintenance of HVAC systems instead of proactive maintenance, which aims at detecting anticipated failures before they occur, so that lower life cycle costs can be accomplished. Therefore, current FM practice needs approaches to detect anticipated failures, so that proactive measures can be taken. Building Automation Systems (BASs) in smart buildings provide historical data on HVAC operations, which can be leveraged for detecting performance degradation of HVAC systems. This study provides a data-driven methodology to quantify and visualize performance changes of HVAC systems over the years using historical BAS data. Our results on a case building demonstrated that there are statistically significant differences between the dataset over the years due to behavioral changes in the HVAC system when other factors (e.g., weather) are controlled. The contribution of this work is a computational approach to identify behavioral changes in HVAC equipment over time using custom selected algorithms for the HVAC domain.