1、2008 ASHRAE 273ABSTRACTPrevious studies indicated that the coupling of multizoneand CFD (Computational Fluid Dynamics) models canprovide a good compromise between the accuracy andrequired computation time. The results show that the coupledmodel predicts contaminant distribution more accurately thanm
2、ultizone model alone for the zones close to the contaminantsource location. For all other zones, the multizone modelsperformed similarly or slightly better than the coupled model.The computational time of the coupled model is lower whencompared to CFD alone and higher when compared to multi-zone alo
3、ne. These observations show tradeoffs between accu-racy and calculation speed. This paper presents results of on-site field experiments conducted to further validate the perfor-mance of the coupled model. In a real office space, contami-nant concentration, temperature, and HVAC supply airflowrate ar
4、e measured to validate the coupling method with a newlyproposed indirect validation method. This method is composedof an experimental validation for the CFD model, and a numer-ical validation of the coupled multizone and CFD model. Over-all, the conducted validation shows that the coupled multizonea
5、nd CFD model gives good results. Therefore, the developedindirect validation method can be applied to other studies toevaluate the performance of multizone or coupled multizoneand CFD models.INTRODUCTIONThe two most widely used types of computer methods forbuilding airflow and contaminant transport
6、simulations aremultizone and computational fluid dynamic (CFD) models.Multizone models usually treat a single zone (room) as a nodethat has connections to the other nodes by flow paths. Themodel calculates macro-scale bulk airflow and contaminanttransport in and between the zones. On the other hand,
7、 CFDmodels divide the domain of interest, usually a single room,into smaller control volumes and calculate detailed micro-scale velocity, temperature and concentration distributionwithin the domain (room). The two models are similar in theprinciples of mass conservation, but CFD also solves themomen
8、tum conservation equation. Furthermore, these twoairflow models use different transport equation solutionprocedures, discretization methods, and boundary conditionspecifications. Due to low computation demand, multizonemodels are widely used for bulk flow movement and contam-inant transport calculat
9、ions in entire buildings, while CFDmodels are typically used for calculations of microscopicairflow, temperature and contaminant distributions in a singlespace.With perfect air mixing in zones, multizone models areapplicable to each zone in a building. The perfect mixingassumption is acceptable in s
10、paces where no major contami-nant sources exist and the room air is completely mixed by theventilation airflow jets. Therefore, the concentration within asingle zone can be assumed to be uniform with the perfectmixing assumption. However, in the zones with contaminationsources or ventilation other t
11、han mixing, the assumption ofconcentration uniformity is crude and can possibly lead toerroneous overall calculations. If the contaminant transport atthe source is not correctly predicted, the distribution within thebuilding and personal exposure in different parts of the build-ing cannot be correct
12、ly calculated. To solve this problem,coupling methods have been proposed to combine thestrengths of the multizone and CFD models, while mitigatingtheir respective inherent weaknesses. In the present study, thecoupled model used commercial PHOENICS CFD softwareOn-Site Experimental Validation of a Cou
13、pled Multizone and CFD Model for Building Contaminant Transport SimulationsJelena Srebric, PhD J. Yuan Atila Novoselac, PhDMember ASHRAE Associate Member ASHRAE Jelena Srebric is an associate professor of Architectural Engineering, The Pennsylvania State University, University Park, PA. J. Yuan is a
14、research assistant, Massachusetts Institute of Technology, Cambridge, MA. Atila Novoselac is an assistant professor in the Department ofCivil, Architectural, and Environmental Engineering, University of Texas at Austin, Austin, TX.NY-08-0332008, American Society of Heating, Refrigerating and Air-Con
15、ditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions, Volume 114, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission.274 ASHRAE Transactions(CHAM 2005)
16、, and CONTAMW multizone program (Dolsand Walton 2002). Finally, this coupling method is experimen-tally validated using an indirect validation method.Based on experiences from the previous studies (Schaelinet al. 1993, Negrao 1995, and Musser 2001), Yuan and Srebric(2002, 2004) as well as He and Sre
17、bric (2004), this studyfurther developed and applied the idea of validating coupledmultizone and CFD model. The coupled model consists ofthree steps illustrated in Figure 1. First, a multizone flowmodel is applied to the entire building to establish airflow ratesand contaminant transport among the z
18、ones. Then, a detailedCFD model is applied only to the zones with the contamina-tion sources. In this step, the predicted non-uniform airflowand concentration distributions are calculated and transferredto the multizone model as fluxes for the final third calculationstep. At the interface surface be
19、tween CFD and multizonemodels, the averaging of CFD results is necessary if multiplecontrol volumes are adjacent to a single zone in the multizonemodel. The final step is a multizone model that excludesdomain simulated by CFD. The three steps can be a part of aniterative loop, but normally very few
20、iteration steps would beneeded for convergence of iterations (Yuan and Srebric 2004).In the present study, the convergence was achieved in only twoiterations.The coupling strategy, presented in Figure 1, has also apotential for computational time savings. With this method,most of the complex CFD sim
21、ulation steps are replaced bysimple multizone calculations and the coupled model isusually much faster than the CFD model alone. Furthermore,defining the simulation model is relatively simple becausespecification of complex boundary conditions such as walls,windows, inlets/outlets is largely simplif
22、ied by the use of amultizone model. However, the results of multizone modelvery much depend on boundary conditions, and, therefore,Furbringer et al. (1999) addressed the need for user-friendlytools and guidelines for the analysis of simulation output ofmultizone programs. There are several recent st
23、udies on thecoupling of different CFD and multizone programs (Clark2001, Musser 2001, Gao 2002), and many experimental vali-dations of multizone programs (Emmerich 2001). However,only a few experimental validation studies are available toexamine the performance of the coupled multizone and CFDmodels
24、. The first study to validate coupled multizone and CFDmodel compared in details the temperature profiles and near-wall heat transfer (Negrao 1995), but did not validate thecoupled model for contaminant concentration profiles.Another recent study included the validation of concentrationprofiles (Wan
25、g and Chen 2007), but used an environmentalchamber partitioned into four sections to represent multiplezones. Thus, the present study proposes an experimentalmethod conducted to validate the accuracy of a coupled multi-zone and CFD model for concentration profiles in real build-ing environments with
26、 on-site experiments.DEVELOPMENT OF ANINDIRECT VALIDATION METHODOLOGYAn evaluation of the accuracy and effectiveness of thecoupled multizone and CFD model is undertaken. The resultsobtained by multizone method, coupled method, and experi-mental values are compared within the calculation domain. Astr
27、aightforward approach is to compare data obtained by threemethods within each space, which we called direct validationas shown in Figure 2. This method is considered accuratebecause it gathers and compares data from three differentmethods directly. However, for this direct comparison, anaveraged con
28、centration from experimental data is need, whichmakes the measurement very complex and difficult. An exper-imental concentration measurement takes only point values atfixed locations, while the real building spaces are usually largewith non-uniform temperature and concentration distribu-tions. If an
29、 average value within a space is required, the spacehas to be divided into small cells (0.2 m or less). Within eachcell, temperatures and concentrations should be measured toobtain averaged values for the entire space or each zone withina building. This would drastically increase experimental time,d
30、ifficulties of equipment control, and the cost of the experi-ment. Therefore, an indirect validation method is used in thisstudy.To effectively validate the coupled method using typicalspace concentration sampling equipment, an indirect valida-tion approach is proposed for the comparison. The indire
31、ctvalidation decomposes the validation into two steps as shownin Figure 2. The first step is comparing point measuredtemperatures/concentrations with point temperatures/concen-trations from full-scale CFD at the same locations. The objec-tive of this step is to prove the full-scale CFD prediction of
32、 thetemperature and concentration distributions to be accurate.Once the accuracy of CFD simulation is evaluated, the secondstep is a comparison among averaged full-scale CFD, multi-zone, and coupled models. This comparison is valid only in thecase that the full-scale CFD is proven to be a valid subs
33、titutefor the experimental data.Figure 1 The tested coupling of multizone and CFDmodels.Figure 2 Two possible validation methods for a coupledmultizone and CFD model.ASHRAE Transactions 275ON-SITE MEASUREMENTSExperimental measurements were conducted in an inte-rior cubicle office with displacement v
34、entilation system asshown in Figure 3. Vertical partitions divided the lower part ofthis area into four large and two small cubicles. The large cubi-cles had two, and the small cubicles had one working stationsper cubicle including computers and lamps, which arepresented in Figure 4. During the meas
35、urements, occupantswere not present, and the major heat sources were computersand the lighting. Table 1 lists the objects and heat sourceslocated in the cubicles during all of the experiments.The concentration source was represented by a controlledflow rate of SF6(sulfur-hexafluoride) tracer gas. Th
36、e spacehad only one supply diffuser, so the tracer gas penetrated allcubicles. As shown in Figure 5, the SF6source was locatedthe floor level at a rate of 28.5 ml/min. The source was notlocated on the centerline of the diffuser. Instead, it had anoffset of 0.2 m to the east side of the diffuser to c
37、reate anasymmetric SF6distribution that is more challenging for vali-dation of the coupled multizone and CFD model. The concen-tration of SF6for the east area (zones 1, 2, and 3) and westarea (zones 4, 5, and 6) of this cubicle office was considerablydifferent. Concentration is measured in all of th
38、e six cubicles,three east and three west cubicles, and at two locations in thecorridor. Figure 5 marks the locations where the stands andthe sensors were positioned with the blue dots. Sensor standswere located in the middle of cubicles and each stand carriedsensors for temperature, velocity, and co
39、ncentration measure-ments at 0.8, 1.5, and 2.4 m from the ground. Totally, 25 SF6sampling tubes were located in the cubicles; 24 probes werelocated in the cubicle area, and 1 additional probe was locatedin the supply duct. In the cubicle area, velocity and tempera-ture sensor were placed in the same
40、 locations as the 24 tubesfor SF6sampling. Also, the temperature of supply air wasmeasured.Figure 3 The office photo with the on-site measurementinstrumentation.Figure 4 The cubicle office layout with displacementventilation diffuser and major heat sources.Table 1. The Objects and Heat Sources Locat
41、ed in the Cubicles During the ExperimentsObject NameNumber of Objects in the SpaceConvective Heat Source (each)Ceiling Lamps 14 40 WSide Lamp 1 160 WComputers 10 72 WOverhead lamps 10 25 WPartitions 10 Tables 10 Book shelves 10 Total Internal Convective Heat Sources 1690 WFigure 5 Locations of the s
42、even temperature and SF6measuring stands in the cubicles.276 ASHRAE TransactionsInstrumentationThe SF6tracer gas system used in validation wascomposed of three functional subsystems: gas releasingsystem, automatic sampling system, and gas analyzingsystem. The gas releasing system discharged a soluti
43、on of0.1% SF6in nitrogen with an accuracy of 3%. The samplingsystem consisted of a series of nylon tubes and air pumpingdevices. Tubes were fixed in the space at the sampling loca-tions, and were connected to the sampling device locatedoutside of the experimental area. The gas analyzing systemwas a
44、tracer gas monitor based on gas chromatography, and itmeasured the sample concentration with an accuracy of 3% inthe range of 100 ppt (parts per trillion) to 10 ppb (parts perbillion).For the air temperature and velocity measurements, 24low velocity omni-directional probes with temperaturesensors we
45、re distributed within the experimental area. Veloc-ity was measured with an accuracy of 0.02 m/s, while thetemperature probes had an accuracy of 0.2C. Also, tempera-tures of the floor and walls were measured with an accuracy of0.5C. In order to determine the flow rate of the supplydiffuser, a Pitot
46、tube was used for high air velocity measure-ment. Velocity was measured at 16 points in the rectangularduct cross-section, and the flow rate was calculated accordingto the measured velocity and the duct area.Measurement Procedure and ResultsTo obtain quasi steady state airflow, temperature, andconce
47、ntration distributions, the on-site measurements wereconducted during the night (2:30 to 5:00 AM), when the inter-nal heat gains and outdoor weather conditions are relativelystable. The night time experimental conditions enabled theexperiences of the validation studies conducted in environ-mental ch
48、ambers with tightly controlled environmental condi-tions to be transferred to the on-site experimentation, wherethe environmental conditions are influenced by air-condition-ing operation, floor plan, thermal loads, human activity andinfiltration. In our experiments, there were no human activi-ties,
49、infiltration was low, and the lighting plus equipment heatgains were dominant. Thus, during the measurement, the air-conditioning system was operating at relatively steady condi-tions. It is important that our experiments were conductedwithout adjusting any of the air-conditioning systems settingsand without sealing or controlling any of the air pathways,making the validation directly applicable to real building envi-ronments.Temperatures and local air velocities were continuouslyrecorded for the entire duration of the experiment. For concen-trations, three groups of repeated samplings
