Back to results list
Please use this identifier to cite or link to this item:
|Title:||Barriers to application of fault detection and diagnosis (FDD) techniques to air-conditioning systems in buildings in Hong Kong||Authors:||Lee, Sze-hung||Keywords:||Hong Kong Polytechnic University -- Dissertations
Air conditioning -- China -- Hong Kong
Fault location (Engineering)
|Issue Date:||2010||Publisher:||The Hong Kong Polytechnic University||Abstract:||Automatic fault detection and diagnosis (FDD) is already a well-developed and widely used technology in hi-tech industries, such as the aviation and nuclear power industries. The application of this technology in buildings is lagging behind despite a large amount of effort has been made on developing FDD models for detecting system faults and sensor faults in air-conditioning systems. Through analyses of operating records collected from existing air-conditioning plants and attempts to develop models and FDD algorithms for chillers in those plants, the barriers that hinder widespread application of FDD to buildings have been identified in the present research study. Whether a technology will penetrate the market hinges on the benefits of using the technology. The benefit of using FDD, however, is indirect; its use may not necessarily lead to energy savings but can help early detection of system and sensor faults and avoid energy being wasted due to malfunctioning of equipment or control systems. Among various building services systems, air-conditioning systems dominate the energy use in buildings and, therefore, are often the targets of FDD applications. In order to provide a picture of the achievable benefits of applying FDD to air-conditioning systems, the energy costs of several air-side system and chiller component faults were estimated by computer simulation. This work formed the first part of the PhD study. Since FDD relies on measured performance data to tell whether any faults have emerged, the availability of accurate and reliable performance measurements is a pre-requisite to successful application of FDD. For a study on whether typical chiller plants can fulfil this pre-requisite, operating records of an existing chiller plant were extracted from the building management system (BMS) that controls and monitors the performance of the plant. Attention was focused on measurements of the performance of chillers, the dominant energy consuming equipment in the plant. Many problems were encountered with the BMS records in that many missing data and corrupted data were found. The operation and maintenance (O&M) practice of the O&M staff on the way in which they handled system malfunctioning or component failures in the chiller plant was also studied, which unveiled several key barriers, such as faulty sensors.
In preparation for developing a FDD algorithm for chillers in the plant where twin-circuit chillers with two screw compressors per circuit were used, a mathematical model was developed for the chillers. The chiller model is semi-empirical model comprising a set of linked thermodynamic component models with coefficients that need to be evaluated based on measured chiller performance data. The chiller model includes a new evaporator model that can simulate heat transfer in the two separate compartments (one for each circuit) at the refrigerant side. Unfortunately, continuation with trial FDD implementation could not proceed further with that chiller plant due to problems with the chiller performance data measuring and recording functions of the BMS, which could not be resolved within a short period of time. Consequently, another chiller plant had to be found for continuation of the study. Nevertheless, the chiller model developed has been verified to be capable of predicting chiller performance that compared well with measured data and has the potential to be used for detecting and identifying chiller faults. Similar modelling approach was also used in developing models for other chillers in the later part of the study. The second chiller plant studied had been incorporated with a chilled water circuit design that allows efficient measurement of full- and part-load performance of chillers and thus was selected as the target chiller plant for further studies. The chillers used in this plant were simpler, water-cooled single-circuit screw chillers and a chiller model was developed and tested with the field data. FDD strategy was developed based upon measurements from available sensors and flow meters rather than requiring other sensors to be installed specifically for chiller FDD. The fault classifiers were determined based on the experimental results of the ASHRAE 1043-RP project. The characteristic parameters that can indicate whether a chiller is in healthy condition were identified. Monitoring the differences of the parameters evaluated from the measurements from the corresponding reference values will allow whether faulty conditions existed to be told. The strategy has been validated with field data and found to be effective for chiller FDD. Lastly, the barriers to adoption of FDD for chillers in an existing plant were reviewed and discussed in the light of the experience gained in the studies on the two chiller plants, and several recommendations were made to tackle the problems. Such barriers include unavailability of refrigerant measurements, faulty sensors and missing data. The minimum range of system variables that should be consistently measured and logged in a BMS was also proposed for chiller FDD. Recommendations have also been made on the required accuracy of sensors, the periods of inspection and calibration of sensors, and the provision of uninterruptible power supply for and sufficient data storage capacity in a BMS.
|Description:||xxx, 294 leaves : ill. ; 30 cm.
PolyU Library Call No.: [THS] LG51 .H577P BSE 2010 Lee
|URI:||http://hdl.handle.net/10397/4079||Rights:||All rights reserved.|
|Appears in Collections:||Thesis|
Show full item record
Files in This Item:
|b23930457_link.htm||For PolyU Users||162 B||HTML||View/Open|
|b23930457_ir.pdf||For All Users (Non-printable)||6.14 MB||Adobe PDF||View/Open|
Citations as of Jun 18, 2018
Citations as of Jun 18, 2018
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.