1、908 ASHRAE TransactionsABSTRACTThis paper presents a new approach for the developmentand use of benchmarking models in the context of ongoingcommissioning. Different techniques are explored to establishthe benchmarking models: (1) a static approach, which isbased on predefined training set size and
2、established differentmodels for weekdays and weekend and holidays, or (2) windowtechniques, which are either augmented or sliding. The perfor-mance of each approach is evaluated for two chillers installedin an existing central cooling and heating plant. Both chillershave identical capacity and perfo
3、rmance characteristics;however, they have quite different operating hours. For chillerCH-1, the models established with a 21-day data set for theaugmented window technique and the 14-day data set for thesliding window technique give the best testing results. Forchiller CH-2, the static model establi
4、shed with one week ofdata and the augmented window model with seven days of dataprovides the most accurate testing results. INTRODUCTIONImproving building energy performance is becoming acommon task in both newly constructed and existing build-ings. The terms commissioning, re-commissioning, retro-c
5、ommissioning, and ongoing commissioning are often used todescribe the actions undertaken to verify if the installed build-ing components or systems performed in compliance with thedesign specifications, current goals, and/or owners projectrequirements (OPR). For new buildings, commissioningnormally
6、takes place before occupancy, and ensures themechanical systems are checked for performance and systeminteroperability at part-load and design conditions (ASHRAE2005a). For existing facilities, re-commissioning or retro-commissioning are used to restore the facilitys performanceto its initial design
7、 specifications or to make the mechanicalsystems work efficiently (Abouzelof 2001). More recently, thenew concepts of continuous or ongoing commissioning havebeen proposed for existing buildings to ensure that the strate-gies implemented continue to meet the current or evolvingOPR throughout time (A
8、SHRAE 2005a). Ongoing commis-sioning is a comprehensive process used to help resolve oper-ation problems, improve comfort, optimize energy use, andidentify retrofits for existing commercial and institutionalbuildings and central plant facilities (Liu et al. 2002). Ongoingcommissioning is performed p
9、eriodically, usually every threeor fourth months, while a much smaller time step, usuallybetween ten minutes and one hour, is selected for ongoingcommissioning (Claridge et al. 2004). The continuous andongoing commissioning processes are a continuation of theinitial commissioning process. During con
10、tinuous or ongoing commissioning, the proj-ect intent is only considered as a reference, not as the perfor-mance target, realizing that (1) the building designer rarelyspecifies the optimal operation of the systems, and (2) thebuilding function and use have often changed significantlyfrom original e
11、xpectations (Liu et al. 2003). The processes ofcontinuous or ongoing commissioning are integratedapproaches that implement optimal schedules for operatingsetpoints, and ensure optimal operation of the systems andpersistence of the integrated changes. The objective is to main-tain optimum operation b
12、y performing ongoing monitoring ofthe systems, data analysis, and sensors calibration, andsystems tuned up, as needed. However, this process has yet tobe automated and only a limited number of applications havebeen developed, beyond the conventional control of HVACOngoing Commissioning Approach for
13、a Central Cooling and Heating PlantDanielle Monfet Radu Zmeureanu, Eng, PhDStudent Member ASHRAE Member ASHRAEDanielle Monfet is a PhD candidate in the Department of Building Engineering and Radu Zmeureanu is associate dean of Student AcademicServices and a professor in the Department of Building, C
14、ivil and Environmental Engineering, Concordia University, Montral, Qubec,Canada.LV-11-0252011. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions, Volume 117, Part 1. For personal use only. Additional reproduction, distr
15、ibution, or transmission in either print or digital form is not permitted without ASHRAES prior written permission.2011 ASHRAE 909systems, to automatically assess and maintain the perfor-mance of the mechanical systems.Continuous or ongoing commissioning include differenttasks to ensure the performa
16、nce of the systems is maintainedthroughout time, which are defined for different buildingsubsystems, such as the air-handling units (AHU), the centralcooling or heating plant, or the storage system. ASHRAE has published two guidelines on commission-ing: ASHRAE Guideline 0-2005, The Commissioning Pro
17、cess(ASHRAE 2005a), and ASHRAE Guideline 1.1-2007,HVAC itinforms the service technician of the existence of these condi-tions, the impact of these faults on energy consumption, andthe available savings potential. It also calculates the efficiencyindex (EI) and capacity index (CI) of the system. Data
18、 aretransmitted from the databases and accessed via the onlineWeb service. Reports are generated, including, for instance,the efficiency of the system, the priority of retro-commission-ing or tune-up measures, and the estimation of electricity usewhen the tune-up is completed. ENFORMA (2009) is an i
19、nternet-based application thatutilizes data from existing building automation systems tocontinuously and automatically identify energy inefficienciesof HVAC systems by using rule-based fault detection anddiagnostic techniques. The system displays the variation intime of selected parameters and evalu
20、ates the financial impactof faults. The Performance and Continuous Re-commissioningAnalysis Tool (PACRAT 2009) is a tool for monitored datamining to improve facility management. It diagnoses systemproblems and poor performance, manages and summarizesmonitored data, produces extensive reporting and v
21、isualiza-tion of system operational parameters, documents importantoperational parameters, provides interoperability for differentautomation systems, and summarizes/formats the data foreffective visualization. It combines historical data from vari-ous sources, such as data collected via EMCS and dat
22、a loggersinto one format. It is designed to complement the EMCS, notduplicate its functions. The tool includes modules for AHU aswell as chillers and hydronic components. Boiler and VAV boxmodules are under development (Santos et al. 2008). For eachmodule, standard or user-defined characterization m
23、oduleshelp in determining the cost of energy waste or anomalies. Although the tools presented above have useful featuresassessing the performance of HVAC systems and the detectionof eventual faults, they cannot be used directly in the ongoingcommissioning of central cooling and heating plant.Assessi
24、ng the Energy Performance of BuildingsOne approach to assess the monitored energy perfor-mance of buildings and their systems is to compare themeasured data to benchmark data. The benchmark informa-tion is often determined using baseline models. Since the1970s energy crisis, different tools, techniq
25、ues andapproaches have been proposed to develop baseline models.Although the extensive review of all those developments isbeyond the goal of this paper, a few examples related to thedevelopment of baseline models or benchmark data arepresented in this section. Spielvogel et al. (1977) developedsever
26、al energy indices by using statistical analysis of simula-tion and actual building data. Sud and Wiggins (1983) devel-oped simplified graphical procedure to estimate heating,cooling, and electricity energy usage. Extensive research wascarried out at Texas A Jarnagin 2009; Nall 2009). The Build-ing E
27、nergy Quotient (Building EQ) consists of two ratingcomponents: (1) an asset rating (as design) and (2) an opera-tional rating (in operation). The objective is to assist ownersand operators in understanding their building and identifyingpotential energy improvement measures. The Building EQ iscalcula
28、ted as the ratio between the Energy Use Index (EUI) forthe subject building, divided by the median source EUI for thattype of building in the same climate zone. For the operationalrating, the calculation of Building EQ is based on actual utilitybills over one full year. The asset rating is based on
29、simulationresults and shows the potential performance of the buildingsystems and subsystems under predetermined set of condi-tions. The difference between the asset rating and operationalrating shows the difference between the actual performanceand estimated performance based on design conditions. P
30、roposed ApproachContinuous and ongoing commissioning are complexapproaches for the monitoring and analysis of operationalparameters of HVAC systems and components in order to(1) detect faults and failures, (2) display warnings and recom-mend remedial actions and estimate the energy or cost impli-cat
31、ions of such measures, (3) compare the monitoredperformance with benchmarking data to detect the deteriora-tion of performance or abnormal operation conditions, and(4) present the relevant indicators of energy performance tohelp building operators and managers to become aware of thesystems performan
32、ce and, therefore, to undertake the required2011 ASHRAE 911actions for achieving high performance along the systemsuseful life. So far, no detailed or standard approach has beenproposed by the industry to establish benchmarking models atthe component and central plant level in the context of ongoing
33、commissioning. This paper proposes a new approach todevelop benchmarking models that characterize the equip-ment or system performance under normal operation and toidentify operation problems using benchmarking models. It isassumed that the systems have been commissioned prior toestablishing the ben
34、chmark models used to perform ongoingcommissioning, and that data preprocessing has beenperformed to verify the quality of the monitored data. The benchmarking models are established using previ-ously monitored data. The paper also presents a case study fordeveloping benchmark models using different
35、 training andretraining techniques to establish the models. The remediationtasks to be performed, once operation problems have beenidentified, are beyond the purpose of this paper. For specificaction that could be undertaken to address the identified prob-lems, refer to the Continuous Commissioning
36、(CC)SMGuide-book (Liu et al. 2002) and the Methods for Automated andContinuous Commissioning of Building Systems (ARTI 2003).This paper is a contribution to the development of theongoing commissioning approach, and focuses on discussingand presenting results for different training and retrainingappr
37、oaches to establish the benchmarking models. To illus-trate the proposed approach, a case study for developingbenchmarking models, in the context of ongoing commission-ing, is presented for the electric power input to chillers of anexisting central cooling and heating plant that serves severalbuildi
38、ngs.The proposed approach can later on be extended to theother equipment present in the central plant or to the wholecentral plant. The proposed approach can be used in buildingsof different sizes; however, it should be more cost effective inlarge buildings with complex systems and operating strate-
39、gies. The cost implications of the proposed approach are mini-mal for data collection when the data can be collected via theMonitoring and Data Acquisition System (MDAS), alreadyinstalled in a building. The additional cost is due to (1) thedevelopment and implementation of the application software,b
40、ased on the methodology proposed in this paper, and (2) theincrease in data storage capacity. ONGOING COMMISSIONING The proposed ongoing commissioning approach iscomposed of (1) the preliminary phase, in which data aremonitored and archived in a database, and benchmarkingmodels (inverse models) of t
41、he energy performance of thecentral cooling and heating plan are developed and tested,based on past normal operation conditions, and (2) the ongo-ing commissioning phase, in which the actual performance ofthe central plant is compared with the results from the bench-marking models; finally, reports
42、and warnings are sent to thebuilding operators (Figure 1). Preliminary Phase During the preliminary phase, the benchmarking modelsof the central plant energy performance under normal operat-ing conditions, without known problems, are established. Thebenchmarking models are developed using monitored
43、data,collected via the Monitoring and Data Acquisition Systems(MDAS), at the beginning of the ongoing commissioningprocess, and used as reference for future measurements.Normally, the data set used to establish the benchmarkingmodels is composed of data monitored at the beginning of theongoing commi
44、ssioning process, which is supposed to berepresentative of the equipment operating conditions;however, this is not always the case since the equipment oper-ation might be in a lower or higher range than normal opera-tion. Thus, the minimum amount of data required to establishaccurate benchmarking mo
45、dels and the frequency of retrain-ing the models should be evaluated. The topic of qualitycontrol of monitored data, available in the database, is notcovered in this paper; however, the quality of the monitoreddata should be verified prior to proceeding with the proposedapproach.In addition to the “
46、as-operated” benchmarking model, an“ideal” benchmarking model is established by simulating theoptimal operation of the systems using a calibrated model ofthe central plant. This approach mimics the two rating compo-nents of the Building EQ. These “ideal” benchmarking modelsare used as additional ref
47、erence/benchmark targets whenmajor changes have been made to the systems and, therefore,the design data can no longer be used to establish the models.A library of benchmarking models contains, for instance,about 10 correlation-based models and 12 artificial neuralnetwork (ANN) models for the electri
48、c power input (E, in kW)and coefficient of performance (COP). Correlation-basedmodels and ANN models predict the value of a dependent vari-able (e.g., electric power input to a chiller) in terms of knowninput independent variables (e.g., outdoor air temperature T).Figure 1 Overview of the proposed o
49、ngoing commissioningapproach.912 ASHRAE TransactionsThe correlation-based models can be as simple as E = a + b T,where the coefficients a and b are identified from the datameasured in the past (ASHRAE 2005b). The ANN models aremore complex, as they mimic the information transfer in thehuman brain. In order to select the most appropriate bench-marking model, a large number of benchmarking models needto be developed and tested using several samples of monitoreddata. The central plant manager or consultant does not need tobe familiar with the benchmarking models, with the detailsabout t
