ASHRAE AB-10-023-2010 Evaluation of CFD for Simulating Air Pollutant Dispersion Around Buildings.pdf

1、2010 ASHRAE 597ABSTRACTThe distribution of air pollutant concentrations around buildings is a main concern of building and air-conditioning engineers that design the ventilation inlets and outlets on building facades or roofs. CFD is increasingly used to predict air flow and related processes around

2、 buildings. In this paper, the possibilities and limitations of CFD for simulating air pollutant dispersion around buildings are discussed. The focus is on dispersion around an isolated building, as the generic basic case for dispersion in the urban environment. The advan-tages and disadvantages of

3、RANS and LES are briefly described, and results from three different cases obtained in separate studies are compared and discussed. It is shown that even for the case of an isolated building, considerable diffi-culties exist and that CFD is not yet at a stage where it can be used as a stand-alone pr

4、actical engineering tool for pollutant dispersion modeling.INTRODUCTIONOutdoor air pollution is one of the major environmental problems today. It is associated with a broad spectrum of acute and chronic health effects (e.g. Brunekreef and Holgate 2002). The pollutants that are brought into the atmos

5、phere by various sources are dispersed (advected and diffused) over a wide range of horizontal length scales (L). Dispersion within the urban environment (L 5 km / 3.1 miles) is referred to as microscale dispersion. Important parameters for microscale dispersion are building geometry and environment

6、 topogra-phy, wind speed, wind direction, turbulence, stability, temper-ature, humidity and solar radiation.In the built environment, both the outdoor exposure of pedestrians and the indoor exposure of building inhabitants are of concern (Fig. 1). Outdoor and indoor air pollution are a main concern

7、of building and air-conditioning engineers that design the ventilation inlets and outlets on building facades or roofs (Drivas and Shair 1974, ASHRAE 1999). Indoor air pollution by outdoor air pollutants can be caused by the re-ingestion of the contaminated exhaust air by the same building or by the

8、 intake of exhaust from other sources such as nearby buildings, street traffic, vehicle parking lots and loading docks, emergency generators and cooling towers (Smeaton et al. 1991, Meroney 2008).The precise prediction of pollutant concentration distri-butions on and near buildings is important for

9、building design and evaluation. The prediction of such concentrations however is a difficult task, especially in the urban environ-ment. It does not only require the knowledge of air pollution meteorology and dispersion, it also requires knowledge of building aerodynamics because wind and buildings

10、can strongly affect plume behavior. Due to the complexity of microscale pollutant dispersion around buildings, much of the research in the past has focused on two generic basic situa-tions: the urban street canyon and the isolated building. While both situations are strong simplifications of reality

11、, the flow and dispersion processes involved are very complex and contain most of the salient features that are also present in the complex urban environment. In this paper, only dispersion around simple isolated buildings is considered.Different methods exist for the analysis of pollutant concentra

12、tions around buildings. Several field tests have been conducted in the past (Barad 1958, Wilson and Lamb 1994, Stathopoulos et al. 2002, 2004). These are very valuable because they are conducted in the real atmospheric boundary layer and provide information on the real complexity of the phenomenon.

13、Disadvantages however are the uncontrollable Evaluation of CFD for Simulating Air Pollutant Dispersion Around BuildingsBert Blocken, PhD Ted Stathopoulos, PhD, PEngBert Blocken is an associate professor in the Unit Building Physics and Systems, Eindhoven University of Technology, Eindhoven, The Neth

14、-erlands. Ted Stathopoulos is a professor in the Department of Building, Civil and Environmental Engineering, Concordia University, Montreal, Canada.AB-10-0232010, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions (2010

15、, Vol. 116, Part 2). For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission.598 ASHRAE Transactionsnature and variation of wind and weather conditions. More-over, it is impossible to meas

16、ure resulting pollutant distribu-tions if a new facility or a new building has not yet been constructed. As opposed to field tests, wind tunnel modeling allows controlled physical simulation of dispersion processes (Huber and Snyder 1982, Li and Meroney 1983, Stathopoulos et al. 2002, 2004). Drawbac

17、ks of wind tunnel tests are that they can be time-consuming and costly, that they are not applicable for light wind conditions, and that scaling similarity can be a difficult issue. Semi-empirical models, such as the Gaussian model (Turner 1970, Pasquill and Smith 1983) and the ASHRAE models (Wilson

18、 and Lamb 1994, ASHRAE 1999, 2003) are valuable because they are relatively simple and easy-to-use, at the expense however of reduced applicability and generally less accurate estimates. The Gaussian model, in its original form, is not applicable when there are obstacles between the emission source

19、and the receptor, and the ASHRAE models only evaluate the minimum dilution factor on the plume centerline. Numerical simulation with CFD could offer some advantages compared to other methods. As opposed to field and wind tunnel tests, it provides results of the flow features at every point in space

20、simultaneously. It is also not subjected to similarity requirements as numerical simula-tions can be performed at full scale. However, CFD requires specific care for the results to be reliable. Important parame-ters determining the accuracy of CFD simulations are the grid resolution and iterative co

21、nvergence (Li and Stathopoulos 1998), the choice of turbulence models and near-wall treat-ment (Franke et al. 2007), avoiding unintended streamwise gradients in the approach flow profiles of mean wind speed and turbulence quantities (Blocken et al. 2007a, 2007b, Franke et al. 2007), the order of the

22、 discretization schemes, etc. CFD simulations of turbulent flow based on the Reynolds-averaged Navier-Stokes (RANS) equations or with Large Eddy Simu-lation (LES) should at least always be validated by comparison with high-accuracy experimental data. While it is often said that CFD is less expensive

23、 than field and wind tunnel tests, the actions that are required to provide confidence in the CFD results, such as grid-sensitivity analysis and validation, are very time-consuming. The often mentioned statement that CFD is less expensive than wind tunnel modeling is therefore not necessarily true.

24、This paper provides an evaluation of CFD for modeling air pollutant dispersion around an isolated building, as opposed to street canyons. Previous studies such as those by Selvam and Huber (1995) provided a general overview of the status of pollutant dispersion modeling around buildings. Meroney (20

25、04) compiled a very comprehensive review of wind tunnel and CFD studies of microscale pollutant disper-sion in the larger framework of hybrid wind tunnel CFD modeling. Canepa (2004) reviewed a large number of the existing models for stack and building wake downwash. The present paper, on the other h

26、and, explicitly focuses on the physical aspects of flow and dispersion around a building and their connection to the possibilities and limitations of CFD for pollutant dispersion modeling. Its intention is to identify possibilities and limitations of CFD as a stand-alone practical engineering tool.

27、“Stand-alone” refers to the fact that no addi-tional wind tunnel or field measurements are used to support fine-tuning of CFD results, e.g. as can be done by tuning the turbulent Schmidt number. FLOW AND DISPERSION AROUND AN ISOLATED SHARP-EDGED BUILDINGWithout going in too much detail, a few specif

28、ic features of the flow pattern around an isolated sharp-edged building are briefly recalled to support the evaluation of CFD in the next sections. Fig. 2 is a schematic representation of the mean flow pattern downwind of a building, for wind perpendicular to the Figure 1 Exposure of building inhabi

29、tants and pedestrians by exhaust re-ingestion by the same building and by other building and by accumulation in streets (modified from Petersen et al. 2002).Figure 2 Schematic representation of the mean atmospheric boundary layer flow around an isolated sharp-edged building (modified from Hosker (19

30、84), Woo et al. (1977) and Hunt et al., (1978).2010, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions (2010, Vol. 116, Part 2). For personal use only. Additional reproduction, distribution, or transmission in either pr

31、int or digital form is not permitted without ASHRAEs prior written permission.2010 ASHRAE 599windward building facade. In reality, such flow patterns exhibit a strong degree of unsteadiness. While the separation points at the sharp upwind building edges are fixed by geom-etry, the other separation p

32、oints and the reattachment points are determined by the interaction of aerodynamic forces. As a result, the locations of these points fluctuate with perturba-tions of the overall flow field. In addition, separation bubbles can collapse intermittently, and the flow in the building wake is characteriz

33、ed by vortex shedding from the leeward building sides and roof. Pollutants that are released in the separation bubble on the roof or sides with a low momentum ratio will be trapped in this bubble and are advected upstream, yielding locally high concentrations near the upwind edge (Wilson 1976, Li an

34、d Meroney 1983). The intermittent collapse of the recirculation bubble causes a “wash-out” (Peterka et al. 1985) in which the pollutants are rapidly advected downstream. In addition, the lateral and vertical fluctuations in the bubble promote lateral and vertical mixing of the pollutants in this reg

35、ion (Thompson and Lombardi 1977). The unsteadiness of the flow over the roof, outside the separation bubble, can cause a strong increase in the vertical spread of the plume (Huber and Snyder 1982). In the leeward cavity, pollutant build-up occurs which is peri-odically swept away by the vortices she

36、d from the leeward edges, again followed by a pollutant build-up (Huber 1988). This vortex shedding is an important mechanism and provides strong lateral mixing of pollutants further downstream. CFD MODELING: STEADY RANS VERSUS LESA distinction can be made between different approaches: RANS modeling

37、, LES modeling and hybrid RANS/LES modeling. In this paper, only steady RANS and LES modeling are discussed, because these are the methods that have been most often used for pollutant dispersion studies. Steady RANS is often employed with first-order closure linear eddy-viscosity turbulence models s

38、uch as the standard k- model. Such models assume an isotropic eddy viscosity, while in reality a strong anisotropic behavior is present in the flow around a building. Second order closure can be achieved by Reynolds stress modeling (RSM), in which a separate transport equation is solved for each of

39、the Reynolds stresses, which fully allows taking Reynolds stress anisotropy into account. Independent of the type of turbulence model however, steady RANS is incapable of modeling unsteady behavior.In LES, the spatially filtered Navier-Stokes equations are solved, and the effect of the non-solved tu

40、rbulence scales is modeled by a subgrid-scale (SGS) model. The main advantage of LES over RANS is that the large-scale unsteadiness in the flow is resolved and that therefore more accurate information on turbulence is obtained. Various SGS models are available, including the conventional standard Sm

41、agorinsky model, the Dynamic Smagorinsky model and the Dynamic Mixed model (Murakami 1993, Tominaga et al. 1997). The dispersion of air pollution can be modeled with Lagrangian or Eulerian models. In Lagrangian models, indi-vidual pollutant particles are tracked within the flow field. In the Euleria

42、n approach, the variables (pollutant concentrations) are obtained at a fixed grid in space. The Eulerian approach is most widely used for dispersion modeling. In the RANS approach, modeling of the turbulent mass fluxes in the advec-tion-diffusion equation is almost exclusively performed by first ord

43、er closure (Franke et al. 2007) and by the assumption of the analogy between turbulent momentum and mass trans-fer. The turbulent mass flux is assumed to be proportional to the mean mass concentration gradient (gradient diffusion hypothesis). The proportionality factor, the mass diffusivity, is obta

44、ined from the momentum diffusivity by the turbulent Schmidt number. In the LES approach, an unsteady solution is obtained, and only SGS turbulent mass fluxes need to be modeled, by analogy between momentum and mass transfer and the specification of a turbulent SGS Schmidt number. For RANS modeling,

45、the often-made assumption of a constant turbulent Schmidt number, independent of the flow features, is an important limitation (He et al. 1999, Tominaga and Stathopoulos 2007b, Blocken et al. 2008). For dispersion modeling around buildings, constant turbulent Schmidt numbers of about 0.7 to 0.9 are

46、mostly used for the entire flow field, while it is known that these numbers can vary consider-ably within the flow field (He et al. 1999, Koeltszch 2000). As opposed to steady RANS, LES can model both the anisotropy and the unsteady behavior, and can therefore provide predictions of separation, reci

47、rculation and vortex shedding that are very close to wind tunnel measurements. LES is therefore most promising for dispersion modeling, at least in terms of accuracy. Two main disadvantages of LES however are the increased model complexity and the strongly increased computer requirements in terms of

48、 storage, memory and time. LES requires time and space resolved data as inlet boundary conditions, which have to be extracted from high-resolution experiments or calculated a priori by e.g. CFD simulations of atmospheric boundary layer flow over a numer-ical roughness fetch (e.g. Tamura 2008). For u

49、rban environ-ments, which typically include a large number of buildings in a large computational domain, the use of LES is considered impractical due to the limited conditions of computer resources currently available (Schatzmann et al. 1997, Yoshie et al. 2007).CFD MODELING EXAMPLESMost CFD studies of pollutant dispersion around build-ings were conducted for isolated buildings, placed in a neutral atmospheric boundary layer, with wind direction perpendicu-lar to one of the facades, and for non-buoyant and non-reactive gaseous exhaust. Most studies used the RANS approach with t