California Energy Commission Public Interest Energy Research Program LBNL No 50678 HPCBS No E5P2 3T3d HPCBS High Performance Commercial Building Systems Field Testing of Component Level Model Based Fault Detection Methods for Mixing Boxes and VAV Fan Systems Element 5 Project 2 3 Task 2 3 3 Peng Xu and Philip Haves Ernest Orlando Lawrence Berkeley National Laboratory LBNL 50678 CD 452 Presented at the ACEEE 2002 Summer Study on Energy Efficiency in Buildings August 18 23 2002 Asilomar Conference Center Pacific Grove California and published in the proceedings Field Testing of Component Level Model Based Fault Detection Methods for Mixing Boxes and VAV Fan Systems Peng Xu and Philip Haves Building Technologies Department Environmental Energy Technologies Division Ernest Orlando Lawrence Berkeley National Laboratory University of California 1 Cyclotron Road Berkeley California 94720 8134 USA May 2002 This work was supported by the California Energy Commission Public Interest Energy Research Program and by the Assistant Secretary for Energy Efficiency and Renewable Energy Office of Building Technology State and Community Programs Office of Building Research and Standards of the U S Department of Energy under Contract No DE AC03 76SF00098 Field Testing of Component Level Model Based Fault Detection Methods for Mixing Boxes and VAV Fan Systems Peng Xu and Philip Haves Building Technologies Department Environmental Energy Technologies Division Lawrence Berkeley National Laboratory Abstract An automated fault detection and diagnosis tool for HVAC systems is being developed based on an integrated lifecycle approach to commissioning and performance monitoring The tool uses component level HVAC equipment models implemented in the SPARK equation based simulation environment The models are configured using design information and component manufacturers data and then fine tuned to match the actual performance of the equipment by using data measured during functional tests of the sort using in commissioning This paper presents the results of field tests of mixing box and VAV fan system models in an experimental facility and a commercial office building The models were found to be capable of representing the performance of correctly operating mixing box and VAV fan systems and detecting several types of incorrect operation Introduction There is a growing consensus that most buildings do not perform as well as intended and that faults in HVAC systems are widespread in commercial buildings There is a lack of skilled people to commission buildings and commissioning is widely seen as too expensive and or unnecessary There is also a lack of skilled people and procedures to ensure that buildings continue to operate efficiently after commissioning One approach to these problems is to wholly or partly automate both commissioning and performance monitoring using computer based methods of fault detection and diagnosis FDD Component level FDD which is the subject of the work presented here uses a bottom up methodology to detect individual faults by analyzing the performance of each component in the HVAC system Hyvarinen 1997 LBNL 1999 Haves Khalsa 2000 Model based approaches to fault detection for different HVAC system or sub system have been proposed by various researchers Benouarets et al 1994 describe two model based schemes for detecting and diagnosing faults in airconditioning systems They examined their ability to detect water side fouling and valve leakage in the cooling coil subsystem of an air handling unit McIntosh et al 2000 developed a mechanical model for fault detection and diagnosis in chillers The model was calibrated using data from an operating system and was used in identifying operating faults Ahn et al 2001 present a model based method for the detection and diagnosis of faults in the cooling tower circuit of a central chilled water facility Faults are detected from deviations in the values of the characteristic quantities from the corresponding values for fault free operation The patterns of the deviations are different for each fault allowing rules to be developed that can be used to diagnose of the source of the fault For commissioning a baseline model of correct operations is normally first configured and adjusted using design information and manufacturers data Next the behavior of the equipment measured during functional testing is compared to the predictions of the model significant differences indicate the presence of one or more faults Once the faults have been fixed the model is fined tuned to match the actual performance observed during the functional tests performed to confirm correct operation The model is then used as part of a diagnostic tool to monitor performance monitoring diagnostic tool during routine operation In each case the reference model is used to predict the performance that would be expected in the absence of faults A comparator is used to determine the significance of any differences between the predicted and measured performance and hence the level of confidence that a fault has been detected The performance of a model based fault detection tool is critically dependent on the ability of the model to represent the performance of correctly operating equipment in the field The paper presents the results of tests to assess how well simple models can represent the performance of HVAC secondary systems air handling units and distribution systems The tests were performed at the Iowa Energy Center s Energy Resources Station and in a commercial office building in San Francisco California 1 Component Models The simulation program SPARK Simulation Problem Analysis and Research Kernel SPARK 2002 was used to develop and implement the component models SPARK is an object oriented software system that can be used to simulate physical systems that can be modeled using sets of differential and algebraic equations One advantage of SPARK is that a solution procedure for the model equations is generated automatically leaving the developer free to concentrate on specifying the equations that define the behavior of the component The models addressed in this paper are listed in Table 1 along with the measurements that are assumed to be available as inputs to the model A brief description of each model is presented below a more detailed description is presented in Xu and Haves 2001 The use of the models for functional testing and for performance monitoring is explained and
View Full Document
Unlocking...