463509 Fault Detection and Isolation and Optimal Parking for HVAC Systems

Wednesday, November 16, 2016: 1:06 PM
Carmel II (Hotel Nikko San Francisco)
Hadi Shahnazari1, Craig McDonald1, Prashant Mhaskar2, John House3 and Tim Salsbury3, (1)Department of Chemical Engineering, McMaster University, Hamilton, ON, Canada, (2)Chemical Engineering, McMaster University, Hamilton, ON, Canada, (3)Johnson Controls, Milwaukee, WI

Fault detection and isolation and optimal parking for HVAC systems

Hadi Shahnazari, Craig McDonald, Prashant Mhaskar*, John House and Tim Salsbury

Government regulations and initiatives have placed a large emphasis on the reduction of energy consumption and increase in energy efficiency. Heating, ventilation, and air-conditioning (HVAC) systems are responsible for 40-50% of total building energy consumption. In general, 15 to 20% per annum of energy consumption can be reduced by efficient and optimal operation of buildings. A major factor contributing to this inefficiency are undiagnosed faults (such as stuck dampers) being compensated through wasted, energy intensive effort (higher static pressures). In US alone, the fault detection and isolation (FDI) and fault tolerant control methods are estimated to be capable of saving 10-40% of HVAC energy consumption (see e.g., [1]).

These realizations have motivated significant research effort on devising FDI frameworks for HVAC systems (see e.g., [1], [2], [3] and [4]). In [2], a fault detection tool is proposed that uses a set of expert rules derived from mass and energy balances to detect faults in air handling units (AHUs). A subset of the expert rules which correspond to that mode of operation are then evaluated to determine whether a fault exists. In [3], a principal component analysis (PCA) based approach is presented to detect single sensor faults in heating, ventilation and air conditioning (HVAC) systems and faults are isolated using joint angle plot, which compares the new fault vector with known ones in the library. In [4], a PCA approach is used to extract the correlation of measured variables in heating/cooling billing system and reduce the dimension of measured data. Square prediction error (SPE) statistic is used to detect sensor faults in the system. Then, sensor validity index (SVI) is employed to identify faulty sensors and a reconstruction algorithm is presented to recover the correct data of faulty sensor in accordance with the correlations among system variables. In [5], a combination of PCA and wavelet transform has been used for FDI of HVAC systems. To this end, at first wavelet transform has been used to remove effect of normal weather conditions changes from data. Then PCA has been used as a data driven methodology for fault detection and isolation. The existing results, however, consider only isolation of single fault scenarios or at best multiple sensor faults or multiple actuator faults, and do not consider simultaneous actuator and sensor faults.

There also exist results on fault tolerant control (FTC) of HVAC systems. In [6], the control design compensates for the fault effect as much as possible by switching between different control modes available in the air handling unit design. In [7] and [8], single sensor faults are diagnosed via estimation of healthy value of sensors using PCA method and handled by utilizing the healthy values of sensors in the closed loop, upon fault isolation. In [9], the fault tolerant control design is based on real time estimation of the fault magnitude, and determining MPC constraints (input constraints) based on those values. In this way, MPC design explicitly accounts for fault effect. These fault-tolerant control approaches, however, are all predicated on the idea of maintaining nominal operation, which might simply be impossible, or very expensive, in case of certain faults. Recently, safe-parking based approaches for fault-tolerant control have been proposed (see e.g., [10] and [11]) that park the process at an appropriate operating point, instead of trying to maintain nominal operation. These ideas, however, have not been applied to HVAC systems. In summary there is lack of results in simultaneous actuator and sensor fault detection and isolation and optimal parking of HVAC systems to minimize energy expenditure.

Motivated by the above considerations, in this work, we design and implement an integrated framework for fault diagnosis and fault handling in HVAC system. The results include simulations as well as application to a room at McMaster University (room 207 of Hamilton Hall). To this end, first, we identify a linear model for the room HVAC system using the existing data. Then we design a model based FDI framework based on the methodology proposed in [12]. The key idea is to exploit analytical redundancy in the system through state observer design. We consider subset of actuator and sensor faults and design observers that only use information of inputs and outputs that are not subject to fault. Then we generate residuals that are only sensitive to a subset of fault. The fault is detected if the corresponding residuals breach their thresholds and isolated using a bank of residuals and a logic rule. We also compare the result obtained using model based approaches for fault detection and isolation (FDI) with a principal component analysis (PCA) based approach (see e.g., [13]). Finally, to minimize energy expenditure, we design and implement an optimal parking approach that minimizes energy expenditure in the presence of a fault.

References

[1] Jeffrey Schein, Steven T Bushby, Natascha S Castro, and John M House. A rule-based fault detection method for air handling units. Energy and Buildings, 38(12): 1485-1492, 2006

[2] John M House, Hossein Vaezi-Nejad, and J Michael Whitcomb. An expert rule set for fault detection in air-handling units. Transactions-American Society of Heating Refrigerating and Air Conditioning Engineers, 107(1):858-874, 2001.

[3] Zhimin Du and Xinqiao Jin. Detection and diagnosis for sensor fault in HVAC systems. Energy Conversion and Management, 48(3):693-702, 2007.

[4] Youming Chen and Lili Lan. Fault detection, diagnosis and data recovery for a real building heating/cooling billing system. Energy Conversion and Management, 51(5):1015-1024, 2010.

[5] Shun Li and Jin Wen. A model-based fault detection and diagnostic methodology based on PCA method and wavelet transform. Energy and Buildings, 68:63-71, 2014.

[6] John E Seem. Integrated control and fault detection of HVAC equipment, May 1 2001. US Patent 6,223,544.

[7] Xiaoli Hao, Guoqiang Zhang, and Youming Chen. Fault-tolerant control and data recovery in HVAC monitoring system. Energy and buildings, 37(2):175-180, 2005.

[8] Xinqiao Jin and Zhimin Du. Fault tolerant control of outdoor air and AHU supply air temperature in VAV air conditioning systems using PCA method. Applied Thermal Engineering, 26(11):1226-1237, 2006.

[9] Sorin C Bengea, Pengfei Li, Soumik Sarkar, Sergey Vichik, Veronica Adetola, Keunmo Kang, Teems Lovett, Francesco Leonardi, and Anthony D Kelman. Fault-tolerant optimal control of a building HVAC system. Science and Technology for the Built Environment, 21(6):734-751, 2015.

[10] Rahul Gandhi and Prashant Mhaskar. Safe-parking of nonlinear process systems. Computers & Chemical Engineering, 32(9):2113-2122, 2008.

[11] Miao Du and Prashant Mhaskar. A safe-parking and safe-switching framework for fault tolerant control of switched nonlinear systems. International Journal of Control, 84(1):9-23, 2011.

[12] Miao Du, James Scott, and Prashant Mhaskar. Actuator and sensor fault isolation of nonlinear process systems. Chemical Engineering Science, 104:294-303, 2013.

[13] Seongkyu Yoon and John F MacGregor. Statistical and causal model-based approaches to fault detection and isolation. AIChE Journal, 46(9):1813-1824, 2000.


Extended Abstract: File Not Uploaded
See more of this Session: Process Monitoring and Fault Detection
See more of this Group/Topical: Computing and Systems Technology Division