Elsevier

Energy and Buildings

Volume 43, Issue 7, July 2011, Pages 1774-1783
Energy and Buildings

A robust fault detection and diagnosis strategy for pressure-independent VAV terminals of real office buildings

https://doi.org/10.1016/j.enbuild.2011.03.018Get rights and content

Abstract

A robust fault detection and diagnosis (FDD) strategy using a hybrid approach is presented for pressure-independent variable air volume (VAV) terminals in this paper. The residual-based cumulative sum (CUSUM) control charts are utilized to detect faults in VAV terminals. The residuals between the temperature error and its predication are generated using autoregressive time-series models. The standard CUSUM control charts are used to monitor the residuals which are statistically independent. If the CUSUM value exceeds the chart limits, it means the occurrence of fault or abnormity in the corresponding VAV terminal. The residual-based CUSUM control chart can improve the accuracy of fault detection through eliminating the effects of serial correlation on the performance of control charts. Also, the residual-based CUSUM control chart can enhance the robustness and reliability of fault detection through reducing the impacts of normal transient changes. A rule-based fault classifier consisting of expert rules and fault isolation algorithms is developed to isolate 15 fault sources. The FDD strategy was online tested and validated using in real time data collected from real VAV air-conditioning systems.

Highlights

► A robust fault detection and diagnosis strategy using a hybrid approach is developed for pressure-independent VAV terminals. ► The residual-based CUSUM control charts detect faults in VAV terminals. ► The standard CUSUM control charts monitor the residuals which are statistically independent. ► A rule-based fault classifier consisting of expert rules and fault isolation algorithms is developed to isolate 15 fault sources. ► The FDD strategy was online tested and validated using in real time data collected from real VAV air-conditioning systems.

Introduction

Variable air volume air-conditioning systems are widely developed and applied in office and commercial buildings to save more energy. VAV air-conditioning systems and their control strategies become more and more complex to meet the increasing demands on indoor environment quality and energy conservation. The complex VAV air-conditioning systems tend to fail to satisfy the performance expectations envisioned at design because of the problems caused by improper design, improper installation, inadequate maintenance, equipment failure, or control and sensor failure. If these faults cannot be detected, diagnosed and removed, they will bring about abnormal operations, which subsequently increase energy consumption of the air-conditioning system, deteriorate the indoor air environment and decrease the useful service life of air-conditioning equipment. Fault detection and diagnosis for VAV air-conditioning systems is an important technique to reduce energy consumption, to reduce maintenance costs and to improve comfort [1]. Therefore, it is significant to develop suitable fault detection and diagnosis methods that can be used in the VAV air-conditioning systems.

In recent years, many fault detection and diagnosis methods have been proposed for VAV air-conditioning systems. Schein et al. [2] used a set of expert rules derived from mass and energy to detect faults in the air handling units (AHU). Qin and Wang [3] presented an automatic fault detection and diagnosis strategy for VAV air-conditioning systems. The FDD strategy utilized a hybrid method consisting of expert rules, performance indexes and statistical process control models to detect and diagnose the faults in pressure-independent VAV terminals. Xiao and Wang [4] developed a robust sensor fault detection and diagnosis strategy based on the PCA method for air handling units. Sensor faults are detected using the Q-statistic and diagnosed using an isolation-enhanced PCA method that combines the Q-contribution plot and knowledge-based analysis. Schein et al. [5] used a small number of control charts to assess the performance of VAV terminals. Most recent studies of fault detection and diagnosis in VAV air-conditioning systems have focused on the major equipments such as chiller, air handling unit, fan, etc. However, study on fault detection and diagnosis of VAV terminals has been relatively less than other types of VAV air-conditioning equipments [6].

VAV terminals are common and key components in VAV air-conditioning systems. As VAV terminals serve the end users, their performances have significant effects on the environmental quality provided by heating, ventilation and air conditioning (HVAC) systems and the energy efficiency of buildings. There are some barriers to detect and diagnose faults in VAV terminals. First, VAV terminals are instrumented with the minimum number of sensors sufficient to implement basic local-loop and supervisory control strategies, lack of sensor information is a significant barrier to detect and diagnose faults in VAV terminals. Second, there is little effort to consolidate the information into a clear and coherent picture of VAV terminals status. The data that is collected overwhelms building operators. Third, building operators may not fully understand the control strategies implemented, they may overlook symptoms of a failure [7]. Fourth, the quantity of VAV terminals in a VAV system is huge, their disparate locations in false ceiling areas, they benefit from almost no preventive maintenance. As a result, study on fault detection and diagnosis for the VAV terminals is very insufficient.

A robust fault detection and diagnosis strategy using a hybrid method is presented for pressure-independent VAV terminals in this paper. The residual-based CUSUM control charts [8] are utilized to detect faults. A rule-based fault classifier is developed to find fault sources. The fault detection and diagnosis strategy is tested and evaluated using in real time data collected from real VAV air-conditioning systems. The FDD strategy can be conveniently implemented on real buildings as it relies only upon sensor data, design data, and control signals that are commonly available in energy management and control system (EMCS).

Section snippets

Building and air-conditioning system description

The fault detection and diagnosis strategy developed was tested and validated on VAV air-conditioning systems in a 36-story office building located in Hong Kong. The building is 166.5 m high. Each floor of the building has an area of 786 m2. The 2nd, 20th and roof floors service as mechanical floors. Each floor in the building is served by a single duct VAV air-conditioning system. Fresh air is transported to every floor by fresh air fan. Return air is sent to the AHU through the ceilings. Supply

Fault detection and diagnosis strategy for VAV terminals

The fault detection and diagnosis strategy of pressure-independent VAV terminals uses a hybrid method. The residual-based CUSUM control charts are used to detect faults in pressure-independent VAV terminals. A rule-based fault classifier consisting of expert rules and fault isolation algorithms is developed to isolate fault sources. Fig. 2 shows the flow chart of fault detection and diagnosis for pressure-independent VAV terminals.

Validation and discussions

The FDD strategy was tested and validated on real VAV air-conditioning systems in the office building described in Section 2.1. The air-conditioning systems operate from 8:00 to 18:00 every work day. The operation data of all AHU systems and VAV terminals was collected and stored in a SQL server at 5 min intervals. A computer program based on the FDD strategy was developed and connected to the SQL server. The computer program can access current operation data from the SQL server and in real time

Conclusions

In this study, a robust fault detection and diagnosis strategy and its online program are developed for pressure-independent VAV terminals. The program is employed to online detect and diagnose the faults of 1186 VAV terminals in a real office buildings. The FDD strategy was tested and validated using in real time data from real VAV air-conditioning systems. In situ tests, the results of fault detection and diagnosis matched the on-site verification results well. The results of tests show that

Acknowledgments

The research work presented in this paper is financially supported by Building Intelligent Control Research Fund (JRP0901) granted by Swire Properties Management Ltd. The authors would like to appreciate Honeywell International Inc. (Hong Kong) for the technical help to this research.

References (27)

  • H. Yoshida et al.

    Typical faults of air-conditioning systems and fault detection by ARX model and extended Kalman Filter

    ASHRAE Transactions

    (1996)
  • T.J. Harris et al.

    Statistical process control procedures for correlated observations

    Canadian Journal of Chemical Engineering

    (1991)
  • S.W. Cheng et al.

    Max-CUSUM chart for autocorrelated processes

    Statistica Sinica

    (1997)
  • Cited by (39)

    • A comprehensive review: Fault detection, diagnostics, prognostics, and fault modeling in HVAC systems

      2022, International Journal of Refrigeration
      Citation Excerpt :

      The second part mainly focuses on the applications of the FDD in the HVAC system. Numerous FDD methods have been emerged for the building energy systems such as chillers (Rueda et al., 2005; Beghi et al., 2016; Huang et al., 2018) air handling units (AHU) (Yu et al., 2011; Schein et al., 2006; Yang et al., 2008; Cho et al., 2005; House et al., 2001) variable air volume terminals (Wang et al., 2012; Schein and House, 2003; Seem et al., 2003; Wang et al., 2011) HVAC system level (Zhang and Hong, 2017; Schein and Bushby, 2006; ; Zhao et al., 2013a, 2013b; Wang and Haves, 2014; hak Song et al., 2008; Li et al., 2012) and economizers (Hjortland and Andrew, 2014; Hjortland and Braun, 2014). Review work by Shi and O'Brien (2019) reflects different approaches to feature generation, problem detection, and fault diagnosis.

    • A review of computing-based automated fault detection and diagnosis of heating, ventilation and air conditioning systems

      2022, Renewable and Sustainable Energy Reviews
      Citation Excerpt :

      This is rarely mentioned in existing literatures of data-driven approaches [202]. On the other hand, according to the collected literatures, the methods requiring only fault-free data in the FDD process are mainly composed of physics-based modeling [69,74], unsupervised learning [14,34,157], and regression or control chart based methods [191,194]. Although obtaining reliable fault-free data is also challenging in practice, requiring only fault-free operation data provides an advantage of relevant methods in FDD deployment.

    View all citing articles on Scopus
    View full text