1、61.1CHAPTER 61 SMART BUILDING SYSTEMSAutomated Fault Detection and Diagnostics 61.1Sensing and Actuating Systems . 61.5Smart Grid Basics . 61.7MART building systems are building components that exhibitScharacteristics analogous to human intelligence. These charac-teristics include drawing conclusion
2、s from data or analyses of datarather than simply generating more data or plots of data, interpretinginformation or data to reach new conclusions, and making decisionsand/or taking action autonomously without being explicitly in-structed or programmed to take the specific action. These capabilitiesa
3、re usually associated with software, but they can also be possessedby hardware with embedded software code, or firmware. The linebetween systems that are “smart” and “not smart” is blurry, and, forpurposes of this chapter, does not need to be absolutely defined. Thepurpose of this chapter is to intr
4、oduce readers to emerging technolo-gies that possess some of these smart characteristics.Smart technologies offer opportunities to reduce energy use andcost while improving the performance of HVAC systems to providebetter indoor environmental quality (IEQ). This chapter covers smartsystems and techn
5、ologies in the fields of automated fault detectionand diagnostics, sensors and actuators, and the emerging modernizedelectric power grid and its relationship to buildings and facilities.1. AUTOMATED FAULT DETECTION AND DIAGNOSTICSMany buildings today use sophisticated building automation sys-tems (B
6、ASs) to manage a wide and varied range of building systems.Although the capabilities of BASs have increased over time, manybuildings still are not properly commissioned, operated, or main-tained, which leads to inefficient operation, excess expenditures onenergy, poor indoor conditions at times, and
7、 reduced lifetimes forequipment. These operation problems cause an estimated 15 to 30%of unnecessary energy use in commercial buildings (Katipamula andBrambley 2005a, 2005b). Much of this excess consumption could beprevented with widespread adoption of automated fault detectionand diagnostics (AFDD)
8、. In the long run, automation even offersthe potential for automatically correcting problems by reconfiguringcontrols or changing control algorithms dynamically (Brambley andKatipamula 2005; Fernandez et al. 2009, 2010; Katipamula andBrambley 2007; Katipamula et al. 2003a).AFDD is an automatic proce
9、ss by which faulty operation,degraded performance, and failed components are detected andunderstood. The primary objective is early detection of faults anddiagnosis of their causes, enabling correction of the faults beforeadditional damage to the system, loss of service, or excessive energyuse and c
10、ost result. This is accomplished by continuously monitoringthe operations of a system, using AFDD processes to detect and diag-nose abnormal conditions and the faults associated with them, thenevaluating the significance of the detected faults and deciding how torespond. For example, the temperature
11、 of the supply air provided byan air-handling unit (AHU) might be observed to be chronicallyhigher than its set point during hot weather. This conclusion might bedrawn by a trained analyst visually inspecting a time series plot of thesupply air temperature. Alternatively, a computer algorithm couldp
12、rocess these data continuously, reach this same conclusion, andreport the condition to operators or interact directly with a computer-based maintenance management system (CMMS) to automaticallyschedule maintenance or repair services.Automated diagnostics generally goes a step further than simplydete
13、cting for out-of-bounds conditions. In this air-handler example,an AFDD system that constantly monitors the temperature andhumidity of the outdoor, return, mixed, and supply air, as well as thestatus of the supply fan, hot-water valve, and chilled-water valve ofthe air handler, might conclude that t
14、he outdoor-air damper is stuckfully open. As a result, during hot weather, too much hot and humidoutdoor air is brought into the unit, increasing the mechanical cool-ing required and often exceeding the capacity of the mechanicalcooling system. As a result, the supply air temperature is chronicallyh
15、igh. This is an example of how an AFDD system can detect anddiagnose this fault.Over the past two decades, fault detection and diagnostics (FDD)has been an active area of research among the buildings andHVAC Fernandez et al.2009, 2010; Katipamula and Brambley 2007; Katipamula et al.2003a, 2003b).As
16、shown in Figure 1, the first functional step of an AFDD processis to monitor the building systems and detect abnormal (faulty)The preparation of this chapter is assigned to TC 7.5, Smart Building Sys-tems.Fig. 1 Generic Process for Using AFDD in Ongoing Operation and Maintenance of Building SystemsA
17、dapted from Katipamula and Brambley (2005a)61.2 2015 ASHRAE HandbookHVAC Applicationsconditions. This step is generally referred to as the fault detectionphase. If an abnormal condition is detected, then the fault diagnosisprocess identifies the cause. If the fault cannot be diagnosed usingpassive d
18、iagnostic techniques, proactive diagnostics techniques maybe required to isolate the fault (Katipamula et al. 2003a). Followingdiagnosis, fault evaluation assesses the impact (energy, cost, andavailability) on system performance. Finally, a decision is made onhow to react to the fault. In most cases
19、, detection of faults is easierthan diagnosing the cause or evaluating the effects of the fault. De-tailed descriptions of the four processes are provided in Katipamulaand Brambley (2005a, 2005b) and Katipamula et al. (2003a).Applications of AFDD in BuildingsAFDD has been successfully applied to cri
20、tical systems such asaerospace applications, nuclear power plants, automobiles, andprocess controls, in which early identification of malfunctions couldprevent loss of life, environmental damage, system failure, and/ordamage to equipment. In these applications, AFDD sensitivity, thelowest fault seve
21、rity level required to trigger the correct detectionand diagnosis of a fault, is a vital feature; false-alarm rate is therate at which faults are incorrectly indicated when no fault has actu-ally occurred. A high false-alarm rate could result in significant eco-nomic loss associated with investigati
22、on of nonexistent faults orunnecessary stoppage of equipment operation.The ability to detect faults in HVAC PECI and Battelle 2003). AFDDmethods applied during initial building start-up differ from thoseapplied later in a building lifetime. At start-up, no historical dataare available, whereas later
23、 in the life cycle, data from earlier oper-ation can be used. Selection of methods must consider these differ-ences; however, automated functional testing is likely to involveshort-term data collection, whether performed during initialbuilding commissioning or during routine operation later in thebu
24、ildings lifetime, and therefore, the same methods can be usedregardless of when the functional tests are performed. Such a shorttime period is generally required for functional testing to eliminatethe possibility that the system being tested changes (e.g., perfor-mance degrades) during the test itse
25、lf. Besides use in functionaltesting, AFDD methods could be used to verify the proper installa-tion of equipment without requiring visual inspection. Labor inten-sity could be minimized by only performing visual inspections toconfirm installation problems after they have been detected auto-matically
26、.During building operation, AFDD tools can detect and diag-nose performance degradation and faults, many of which go unde-tected for weeks or months in most commercial buildings. Manybuilding performance problems are automatically compensated bycontrollers so occupants experience no discomfort, but
27、energy con-sumption and operating costs often increase. For example, when thecapacity of a packaged rooftop air conditioner decreases because ofrefrigerant loss, the unit runs longer to meet the load, increasingenergy use and costs, and occupants experience no discomfort (untildesign conditions are
28、approached). AFDD tools can detect these, aswell as more obvious, faults.AFDD tools not only detect faults and alert building operationstaff to them, but also identify causes of faults so that maintenanceefforts can be targeted, ultimately lowering maintenance costs andimproving operation. By detect
29、ing performance degradation ratherthan just complete failure of physical components, AFDD tools canalso help prevent catastrophic failures by alerting building operationand maintenance staff to impending failures before failure occurs.This condition-based maintenance allows convenient scheduling ofm
30、aintenance, reduced downtime from unexpected faults and fail-ures, and more efficient use of maintenance staff time.AFDD MethodsAFDD tools use many different methods for detecting faults andsubsequently isolating or diagnosing their causes. Figure 2 shows acategorization of these methods (Katipamula
31、 and Brambley 2005a),in which fault detection and diagnostic methods are organized intothree primary categories based on (1) quantitative models, (2) qual-itative models, and (3) process history.Quantitative model methods use quantitative models of theunderlying equipment, relationships between type
32、s of equipment,and processes occurring in the equipment and its components. Setsof quantitative mathematical relationships capture the underlyingphysics of the processes. The quantitative results from applying themodels to actual driving conditions represent baseline performancewithout faults. Diffe
33、rences between measured performance and thebaseline performance from the models under identical driving con-ditions, known as residuals, are used to detect the occurrence offaults. Quantitative models can be based on detailed fundamentalphysical principles and engineering relationships or on simplif
34、iedmodels representing the physical processes. Analyses of residualscan also be used to distinguish among possible causes of a fault toprovide a fault diagnosis. Quantitative model-based methods areapplicable to information-rich systems, where satisfactory modelscan be built in an affordable way and
35、 sufficient sensors are availableto provide the data that are required. Methods described by Castro(2002), Dexter and Ngo (2001), Haves and Norford (1997), Li andBraun (2007a, 2007b, 2007c, 2007d, 2009a), Norford et al. (2002),Reddy (2007a), Seem and House (2009), Shaw et al. (2002), andSiegel and W
36、ray (2002) fall into this category.Qualitative model methods include qualitative physics-basedmethods and rule-based methods. Qualitative-physics-based meth-ods express the underlying physical relationships (equations) asqualitative expressions (De Kleer and Brown 1984) but have seenlimited use in A
37、FDD for HVAC rules derived from knowledge of the fundamentalSmart Building Systems 61.3physical processes occurring in HVACand alarms based simply on conditions exceeding prescribed upperand/or lower bounds for acceptable values of variables during oper-ation (e.g., an alarm triggered by duct static
38、 pressure exceeding itsupper limit). The techniques presented by Dexter and Ngo (2001),Gerasenko (2002), House et al. (2001, 2003), and Lo et al. (2007)are some examples.Process-history-based methods depend on the availability of alarge amount of historical data. These methods include black-box(inpu
39、t-output) models derived from the data and gray-box modelsthat use first principles or engineering knowledge to specify themathematical form of terms in the model but for which parame-ters (e.g., coefficients in the model) are determined from processdata. Black-box methods include statistically deri
40、ved models(e.g., regression), artificial neural networks (ANNs), and pattern-recognition techniques. Approaches based on process history pri-marily apply to large systems such as whole buildings, where it isdifficult to construct an analytical model that captures all importantphysical behaviors adeq
41、uately in a cost-effective way, but existinginstrumentation yields sufficient data for analysis. Methods used byBailey (1998), Choi et al. (2004), Li and Braun (2003), Reddy et al.(2003), Riemer et al. (2002), Rossi (2004), and Rossi and Braun(1997) can be classified in this category.For further det
42、ails of each of the basic modeling techniques andAFDD methods, any constraints that would limit the application ofeach technique, and to assess strengths and weaknesses of eachtechnique for application to fault detection and diagnostics, seeKatipamula and Brambley (2005a, 2005b).Benefits of Detectin
43、g and Diagnosing Equipment FaultsThe benefits of AFDD have been validated in part by studies thatdocumented common HVAC equipment operating faults and theireffects (Breuker and Braun 1998a; Breuker et al. 2000; Comstocket al. 2002; House et al. 2001, 2003; Jacobs 2003; Katipamula et al.1999; Proctor
44、 2004; Rossi 2004; Seem et al. 1999). Faults examinedincluded economizers not operating properly, incorrect refrigerantcharges, condenser and filter fouling, faulty sensors, electrical prob-lems, chillers with a variety of faults, air-handling units with too lit-tle or too much outdoor-air ventilati
45、on, stuck outdoor-air dampers,and other problems.Studies of the benefits of HVAC fault detection and correctionhave found positive savings. Rossis (2004) fault survey of unitaryequipment used measurements by service technicians to computefour performance indices from which unit efficiency was estima
46、tedand savings potential calculated. Half of the equipment was esti-mated to have a savings potential of at least $170/year, and 33% hada potential of at least $225/year. Li and Braun (2007e) investigatedthe following factors that affect the economics of air conditioning:(1) energy efficiency ratio
47、(EER) or coefficient of performance(COP), which quantifies the energy performance of the refrigerationcycle (lower scores equal greater operating costs); (2) cooling capac-ity Qcap, the degradation of which can affect comfort in the condi-tioned space, increase compressor run times, and reduce equip
48、mentlifetimes; and (3) sensible heat ratio (SHR), which can decrease withmany faults, leading to higher total equipment load and greaterenergy consumption for the same sensible building load. All threefactors can be combined in an overall economic performance deg-radation index (EPDI), which is defi
49、ned as the net increase in thetotal operating costs (Li and Braun 2007e) and is given by(1)whererSHR= 1 rSHR= degradation ratio of SHRrCOP1 rCOP= degradation ratio of COPrcap= = 1 rcap= degradation ratio of capacityrSHR= SHR/SHRnormal= SHR ratiorCOP=COP/COPnormal= COP ratiorcap= Qcap/Qcap,normal= capacity ratioSHR = actual sensible heat ratioCOP = average actual coefficient of performanceQcap= average actual equipment cooling capacity= average equipment price, $/kWh= = average normal cost of operation, $/h= average electricity price, $/kWh= power consumption of unit (including bo
