ASHRAE NY-08-031-2008 Integration of Network Flow Modeling and Computational Fluid Dynamics to Simulate Contaminant Transport and Behavior in the Indoor Environment《整合网络流向建模和计算机流体动.pdf

《ASHRAE NY-08-031-2008 Integration of Network Flow Modeling and Computational Fluid Dynamics to Simulate Contaminant Transport and Behavior in the Indoor Environment《整合网络流向建模和计算机流体动.pdf》由会员分享，可在线阅读，更多相关《ASHRAE NY-08-031-2008 Integration of Network Flow Modeling and Computational Fluid Dynamics to Simulate Contaminant Transport and Behavior in the Indoor Environment《整合网络流向建模和计算机流体动.pdf（9页珍藏版）》请在麦多课文档分享上搜索。

1、250 2008 ASHRAE ABSTRACTThe flow of air from one room to another may be approx-imated by network flow models which consider the bulk flowof air. Such models can predict inter-zone air distributions butcannot predict intra-zone air flow conditions. Computationalfluid dynamics, on the other hand, can

2、be used to predict intra-room air flows with a high degree of accuracy provided suffi-cient care is taken in specification of boundary conditions,initial conditions and grid definition. Contaminant transportand behavior prediction models are supported by both model-ing techniques. To overcome shortc

3、omings of the individualtechniques, both methods are combined within an integratedmodeling framework. The methodology for prediction ofcontaminant concentration uses three solution procedures inaddition to CFD. These involve the setting up and solution ofcontaminant distribution and transport equati

4、ons (a sparselinear system), the setting up and solution of air flow equations(a non-linear system) and the setting up and solution of build-ing thermal equations (a sparse non-linear system). This paperpresents a method to integrate these approaches in order toaccurately predict both inter- and int

5、ra-room air flows andcontaminant distributions.INTRODUCTIONIn recent times there have been advances in a number ofcomputational methodologies for prediction of air flow andcontaminants. Notable among these are the network air flowprediction algorithm and computational fluid dynamics(CFD). The former

6、 deals with bulk air flow and contaminantstransport e.g. from one room in a building to another. The lattercan deal with micro-climatic distribution of air and contami-nants e.g. variations in contaminant concentration within aroom. Both methods have their advantages and disadvantages.This paper pre

7、sents an approach which combines the twomodels. This approach optimizes the advantages of the twomethods in order to arrive at a solution that simultaneouslysatisfies governing equations of the two methods and isobtained by iteratively solving both at each time step.The development of a comprehensiv

8、e contaminantprediction capability requires the integration of CFD and airflow network models. This is because contaminant predictiondepends on the air flow rates predicted from these models.Therefore the CFD and air flow solutions must first be inte-grated, and once this is achieved the contaminant

9、 concentra-tions can be predicted throughout the air flow network andCFD domains. This paper describes a robust algorithm thatsatisfactorily combines the two domains. The approach allowsappraisal of contaminant distributions in terms of speciesconcentration on two levels: on the coarser level of the

10、 air flownetwork where each air flow node has a contaminant concen-tration associated with it, and on a much finer CFD level. Thesolution procedure using this integrated approach allows thesimultaneous study of bulk contaminant flow patterns in thewhole building and detailed contaminant flow charact

11、eristicswithin one or more spaces within the building. The contami-nant predictions usually benefit from the linking of the CFDdomain with the building thermal model so that the predictedtemperature fields inform the flow predictions. However, theapproach adopted here is flexible in that this therma

12、l linkingis not a necessary condition. The concept of conflation of different domains is shownin Figure 1. Network air flow modeling is integrated withbuilding energy simulation in order to link the thermal and airflow domains. Furthermore network air flow modeling drivesIntegration of Network Flow

13、Modeling and Computational Fluid Dynamics to Simulate Contaminant Transport and Behavior in the Indoor EnvironmentAizaz Samuel, PhD Paul Strachan, PhDAizaz Samuel is Research Fellow at Energy Systems Research Unit and Paul Strachan is Depute Director of Energy Systems Research Unitand Senior Lecture

14、r at the Department of Mechanical Engineering, University of Strathclyde, Glascow, UK.NY-08-0312008, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions, Volume 114, Part 1. For personal use only. Additional reproduction,

15、 distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission.ASHRAE Transactions 251the contaminant prediction model which is represented by thearrow pointing from air flow modeling to contaminant model-ing. These three domains pass informa

16、tion to CFD in the formof boundary conditions for temperature (or heat flux), air flowand contaminant concentration. The boundary conditions areconfigured dynamically and also provide information back tothe respective domains. The arrows pointing from CFD tonetwork air flow modeling and building ene

17、rgy simulationrepresent two way conflation between CFD and the air flownetwork (Samuel 2005), and adaptive conflation betweenCFD and the building thermal domain (Beausoleil-Morrison2000). This process should arrive at a solution that is moreaccurate than results from individual solution of the perfo

18、r-mance prediction schemes.Whereas imposing realistic boundary conditions takenfrom prevailing climatic conditions onto a thermal model isrelatively straightforward, determining boundary conditionsfor a CFD study of a part of a building or associated energysystem requires more thought. The importanc

19、e of boundaryconditions is expounded upon in Versteeg and Malalasekera(1995) who state that flows inside the CFD solution domainare driven by the boundary conditions. In a sense, the processof solving a CFD problem is nothing more than the extrapo-lation of a set of data defined on a boundary contou

20、r or surfaceinto the domain interior. De Gids (1989) cites combining CFDand multi-zone models as one of the most pressing researchactivities. Armstrong et al (2001) state that integration of CFDwith building thermal and air flow simulation can yield infor-mation useful to new advances in building op

21、eration, such ascontinuous commissioning, optimal control, fault detectionand diagnosis, and other intelligent building functions. Zhaiand Chen (2003) showed that the solution set of a coupled ther-mal and CFD problem is real and unique. It is therefore impor-tant that the boundary conditions impose

22、d onto a CFDproblem are physically realistic and well posed, otherwisedifficulties may arise in obtaining an accurate solution.Buildings and related energy systems can easily have tensof principal parameters (insulation level, capacity position,ventilation rate, glazing area, glazing type, lighting

23、load, fueltype and so on) and the permutations available for the domainconfiguration are very large (SESG 1999). Within integratedmodeling environments a building comprises a collection ofinteracting technical domains, the properties of which and asso-ciated processes for which are well understood.

24、Each domain issolved by exploiting the specific nature of the underlying phys-ics and mathematical theories (Clarke and Tang 2004).It has been shown previously that domain integration isvery important when appraising a parameter that is inherentlydependent on other parameters that are computed using

25、 morethan one solution technique and boundary condition defini-tion. A good example of this is species concentration, whichcan be highly dependent on the temperature and air flow distri-bution within (and outside) the building (Samuel 2005). CFDis commonly used to determine detailed air flow, temper

26、ature,relative humidity and species concentration fields. A key prob-lem faced by a designer when setting up a CFD model is thedefinition of boundary conditions because for most flowconfigurations, if the boundary conditions have been wellspecified, it is probable that the results will be correct. C

27、oupling of domains can either be sequential (one way)or simultaneous (two way). In the sequential approach the firstmodule solves, results from it are input into the secondmodule, then the second module solves and simulationprogresses to the next time step. In the simultaneous approachthe two domain

28、s keep on passing information back and forthuntil some mutual convergence criterion is satisfied. Thesimultaneous approach can thus be much more computation-ally intensive, but it has been shown to be more accurate thanthe sequential approach for certain design configurations(Hensen 1995). The adopt

29、ed approach integrates contaminant predictionfacilities using the air flow network algorithm into mainstreambuilding thermal simulation. Predicted contaminant concen-trations can then be passed on to a CFD solver as the boundaryconditions of a CFD domain. There is dynamic (at run time)passage of inf

30、ormation between the different modules in orderto make the simulation more realistic in order to give an accu-rate depiction of reality. Two way conflation makes sure thatthe two conflated domains converge to a solution that isacceptable to all such domains before simulation progresses tothe next ti

31、me step.DOMAIN INTEGRATION WITHIN DYNAMIC THERMAL MODELLINGWithin integrated modeling tools the coupling approachis extensively used to impose communication between differ-ent building simulation domains (Clarke 2001). The differentinteracting technical domains are solved by exploiting indi-vidual s

32、pecific properties and nature of the underlying theoryand physical principles (e.g. linear / non-linear, sparse /compact etc.). Important couplings include: building thermalprocesses natural illuminance distribution; building thermal/ plant processes distributed fluid (usually air and / or water)flo

33、w; building thermal processes intra room air flow; elec-Figure 1 Schematic representation of domain integration.252 ASHRAE Transactionstrical demand profiles integrated embedded generation;building thermal processes moisture flow.Clarke and Tang (2004) define conflation of the differentdomains as th

34、e coordinated solution of the domain equationsunder control action that links certain model critical parame-ters (e.g. room air temperature to the mass flow rate inducedby a fan). All the couplings within the different domains ofintegrated simulation take place by passing this critical infor-mation

35、between themselves. Figure 2 summarizes the proce-dure which is based on the iteration of nested domains. Thebuilding side capacitances are generally greater than plant orair flow side capacitances, so it is possible to run plant and / orelectrical and/or building side flow networks at a higherfrequ

36、ency than building thermal simulation. If a plant sideflow network is active this runs at the same frequency as thesolution of the plant network. Contaminant flow, because of itsinherent dependence on flow parameters, thus needs to be runat a higher frequency as well.Conflation of building thermal a

37、nd air flow processeswithin the CFD domain is described by Negrao (1995 and1998) who worked on the interactions of building thermal andCFD domain interfaces, namely zone surfaces and zone open-ings. Thermal boundary conditions for the CFD domain aretaken from building thermal simulation results for

38、zonesurfaces, and zone openings are considered to be air flowboundary conditions with sources and sinks of thermal energyresulting from the air flow.In recent years there have been further advances in thecoupling of building thermal processes and CFD. Notably,Beausoleil-Morrison (2000, 2001 and 2002

39、) has described anadaptive coupling algorithm that intelligently configureswhich thermal boundary conditions to use at run time.CFD CONFLATION: A BRIEF OVERVIEWWith energy and mass prediction systems as complicatedas building thermal simulation and network air flow, manydifferent methods of integrat

40、ion exist. These methods havetheir own advantages and disadvantages. One possibility is touse CFD, with an appropriate radiation model, as a standalonesolver for building simulation problems. This is the conjugateheat transfer method and many applications are available(Holmes et al 1990, Chen et al

41、1995, Moser et al 1995 andSchild 1997). With an energy model for the HVAC systemsand plant, the CFD can include the function of building simu-lation. This method sounds powerful but it is computationallyexpensive (Chen et al 1995). The reason for this is twofold.First, when the CFD calculates the he

42、at transfer in solid mate-rials, the equation set becomes stiffer and the computing timegoes up dramatically (Thompson and Leaf 1988). CFD simu-lation must be performed over a long period for the thermalperformance of the building envelope, but it must use a smalltime-step to account for the room ai

43、r characteristics. Secondly,the computing time grows exponentially with building size.Hence, the conjugate heat transfer method is not practical forimmediate use in a design context with current computer capa-bilities and speed (Zhai et al 2002). One solution to this prob-lem lies within domain inte

44、gration.Zhai et al (2002) list a number of coupling approachesbetween energy simulation and CFD. Although the particularcoupling for their research was building thermal (temperatureand heat flux calculations) and CFD the concepts developed intheir work can be extended to understand air flow, contami

45、-nants and CFD conflation. In addition to these conflation approaches otherapproaches exist in which one or more parameters have beenchosen to be scheduled or approximated by functions gener-ated empirically. Srebric et al (2000) report using a CFD solverthermally coupled to a quasi steady state the

46、rmal solver. Heatinjections were taken from ASHRAE methods (ASHRAE1997) and time invariant air flow was assumed.Whereas work has been done on integration of thermalsimulation with CFD (Djunaedy et al 2003 and 2005, Beau-soleil-Morrison 2000, 2001 and 2002, Negrao 1998 and 1995)not much development f

47、or mass flow and CFD integrationFigure 2 Iterative solution of nested domains (Clarke andTang 2004).ASHRAE Transactions 253could be found. Wang and Chen (2005) have attempted to inte-grate stand alone network air flow and CFD. They report threemechanisms for it. These mechanisms essentially differ i

48、n theboundary conditions that are exchanged between the massflow solver and the CFD solver. The pressure pressure typeconflation exchanges only pressures between the domains.The mass pressure type conflations work with one domain(either CFD or air flow) passing mass flow rate and the otherdomain pas

49、sing pressure.CFD AND AIR FLOW NETWORK CONFLATION WITHIN BUILDING SIMULATIONThe present concept of network air flow and CFD confla-tion is built on the work of Negrao (1995). The air flownetwork is defined for the whole building (or part of the build-ing) and the CFD domain for a part of it e.g. one room. Thenetwork model representation of this room may be a singlenode (termed domain node). When applying the conflationalgorithm the domain node is replaced by the CFD domain, ascan be seen in the simple example in Figure 3. One air flownetwork node is then created for each conne