Environment International 89–90 (2016) 138–146

Contents lists available at ScienceDirect

Environment International

j ourna l homepage: www.e lsev ie r .com/ locate /env int

Quantifying the impact of traffic-related air pollution on the indoor airquality of a naturally ventilated building

Zheming Tong a,⁎, Yujiao Chen a, Ali Malkawi a, Gary Adamkiewicz b, John D. Spengler b

a Center for Green Buildings and Cities, Graduate School of Design, Harvard University, Cambridge, MA 02138, USAb Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, MA 02115, USA

⁎ Corresponding author.E-mail address: ztong@gsd.harvard.edu (Z. Tong).

http://dx.doi.org/10.1016/j.envint.2016.01.0160160-4120/© 2016 Elsevier Ltd. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 8 November 2015Received in revised form 19 January 2016Accepted 20 January 2016Available online 29 January 2016

Improper natural ventilation practices may deteriorate indoor air quality when in close proximity to roadways,although the intention is often to reduce energy consumption. In this study, we employed a CFD-based air qualitymodel to quantify the impact of traffic-related air pollution on the indoor air quality of a naturally ventilatedbuilding. Our study found that the building envelope restricts dispersion and dilution of particulate matter. Theindoor concentration in the baseline condition located 10 m away from the roadway is roughly 16–21% greaterthan that at the edge of the roadway. The indoorflow recirculation creates awell-mixed zonewith little variationin fine particle concentration (i.e., 253 nm). For ultrafine particles (b100 nm), a noticeable decrease in particleconcentrations indoors with increasing distance from the road is observed due to Brownian and turbulent diffu-sion. In addition, the indoor concentration strongly depends on the distance between the roadway and building,particle size, wind condition, and window size and location. A break-even point is observed at D′ ~ 2.1 (normal-izeddistance from the roadwayby thewidth of the road). The indoor particle concentration is greater than that atthe highway where D′ b 2.1, and vice versa. For new building planning, the distance from the roadway and theambient wind condition need to be considered at the early design stagewhereas the size and location of thewin-dow openings, the interior layout, and the placement of fresh air intakes are important to the indoor air quality ofexisting buildings adjacent to roadways.

© 2016 Elsevier Ltd. All rights reserved.

Keywords:Indoor air qualityNatural ventilationI/O ratioParticulate matterUltrafine particles (UFPs)Computational fluid dynamics (CFD)

1. Introduction

Natural ventilation utilizes natural outside air movement andpressure differences to both passively cool and ventilate a buildingwithout the use of any mechanical system. Previous studies haveshown that the cooling energy consumption of studied buildings canbe reduced by 40–50% using natural ventilation in European cities(Cardinale et al., 2003; Gratia and De Herde, 2004; Kolokotroni andAronis, 1999). While the intention is often to utilize natural ventilationfor energy saving, the possible adverse impact on indoor air quality isnot well addressed. Studies have shown that improper natural ventila-tion practices may deteriorate the indoor air quality (Chen and Zhao,2011; Leung, 2015; U.S. EPA, 2014). In urban environments, indoor airquality is greatly affected by traffic-related air pollution (Fischer et al.,2000; Janssen et al., 2001; Rijnders et al., 2001; Zhu et al., 2005). Epide-miologic studies have shown associations between exposure to traffic-related particulate matter and acute increases in morbidity and mortal-ity (Dockery et al., 1993; Samet et al., 1987; Spengler and Chen, 2000;Vedal, 1997; Zhu et al., 2002). As most people spend 85–90% of theirtime indoors, assessing human exposure to particulate matter in near-

road environments is essential to understanding its impact on humanhealth (Chen and Zhao, 2011; Jenkins et al., 1992; Klepeis et al., 2001;Zhang et al., 2014a).

To quantitatively evaluate the indoor air quality in near-roadenvironments, it is critical to understand how particulate matter istransported from outdoor emission source (e.g., traffic emissions) tothe indoor environment through natural ventilation. Studies in thepast focused on I/O ratio, i.e., the ratio between indoor and outdoorpollutant concentration (Blondeau et al., 2005; Brunekreef et al., 2005;K. Koponen et al., 2001; Liu et al., 2003; Meng et al., 2005; Monn et al.,1997; Pellizzari et al., 2000). However, the I/O ratios reported in theliterature vary within a wide range from 0 to 5 (Chen and Zhao, 2011;Li and Chen, 2003), due to differences in measurement condition,particle size, building envelope, and indoor emission sources, etc. Inaddition,many studies discovered strong spatial variations in air pollut-ant concentrations around buildings located in the near-road environ-ments (Di Sabatino et al., 2007; Gousseau et al., 2011a; Tong et al.,2012; Tong et al., 2015b; Wang et al., 2013; Zhang et al., 2014b). As aresult, simply using I/O ratios offers little help to understand therelationship between indoor and outdoor air quality especially for thedesign of effective natural ventilation strategies. On the other hand,computational fluid dynamics (CFD) captures the complexity of un-steady fluid and aerosol dynamics in urban microenvironments, and is

139Z. Tong et al. / Environment International 89–90 (2016) 138–146

capable of providing 3-D high-resolution pollutant concentration databoth indoors and outdoors. Due to the steady advances in computingcapacity, CFD has been increasingly used for indoor and outdoor airquality modeling (Blocken et al., 2008; Chen et al., 2006; Tominagaand Stathopoulos, 2013; Tong et al., 2016; van Hooff and Blocken,2010; Zhang and Chen, 2009; Zhao et al., 2004).

In this study, a CFD-based air quality model is employed to predictthe spatial relationship between indoor and outdoor air pollutants vianatural ventilation for a generalizable road-building scenario. Ourmain objective is to quantify the impact of different buildingparameters (e.g., distance from roadway, wind condition, and windowopening size and location) on the particle size-dependent (15 nm to253 nm) relationship of indoor/outdoor air quality, and the spatialvariation of particle concentration by using a CFD-based air qualitymodel. This paper is organized as follows. We first describe a CFD-based air quality model that links the indoor space to the outdoormicroenvironment within the same computational domain. The CFDmodel is then evaluated with Particle Image Velocimetry (PIV) mea-surements of flow over and through a cross-ventilated building modelin a boundary layer wind tunnel (Karava et al., 2011). In the secondpart of the paper, we evaluate the impact of various building parameterson indoor air quality through a series of simulations.

2. Model description

A full description of the CFD-based air quality model is presented inour previous work (Tong et al., 2012; Tong and Zhang, 2015; Wanget al., 2013; Wang and Zhang, 2009; Wang and Zhang, 2012). A shortdescription is presented here. Large Eddy Simulation (LES) is employedto resolve the unsteady turbulent flow field. In LES, a low-pass filteringoperation is performed so that the resulting velocity field ~ui can beresolved on a relatively coarse grid. A dynamic subgrid model is chosen,which allows the Smagorinsky constant to vary in space and time(Germano et al., 1991). The constant is dynamically computed andbased on the information provided by the resolved scales of motion.The particle phase is resolved by solving a scalar transport equationwith sink terms to represent the effect of indoor particle deposition.As for the modeling of cross ventilation, the outdoor and indoorenvironments are solved within the same computational domain andthe interaction at the opening is also explicitly resolved (Ramponi andBlocken, 2012a). This coupled approach, in contrast to a decoupledmethod, eliminates uncertainties associated with assumptions onexternal pressure distribution and the dissipation of turbulent kineticenergy on openings.

2.1. Fluid phase

LES with a dynamic subgrid model is employed for modelingturbulent flow field (Germano et al., 1991). It is a proper model forthe purpose of our analysis as it resolves the large-scale unsteadymotions and requires the modeling of only small-scale, unresolvableturbulent motion, which is less influenced by physical geometries. Inaddition, turbulence in urban environment is dominated by large-scale motion generated by scales comparable to the size of buildings(Xie and Castro, 2006).

In LES, filtered continuity and momentum equations are shownbelow:

∂ρ∂t

þ ∂ρ~uj

∂xj¼0 ð1Þ

∂ρ~ui

∂tþ ∂ρ~ui~uj

∂xj¼ −

∂~p∂xi

þ ∂τij∂xj

þ ∂σ ij

∂xj: ð2Þ

τij is the filtered stress tensor, and σij is the subgrid-scale Reynoldsstress, which is modeled with the Boussinesq hypothesis in Eq. (3):

σ ij−13δijσkk ¼ −2μ tSij: ð3Þ

Sij is the rate of strain tensor of the resolved scale. A well-establisheddynamic Smagorinsky model is employed to model the subgridviscosity μt in Eq. (4) (Germano et al., 1991). The dynamic modelremoves some issues associated with the constant coefficientSmagorinsky model by eliminating the need to prescribe a lengthscale and near-wall correction (Pope, 2000).

μ t ¼ ρ LSð Þ2 S�� �� ð4Þ

Ls is the mixing length scale that depends on the size of thecomputational cell and the dynamically computed Smagorinskyconstant.

2.2. Particle phase

Scalar transport equation (Eq. (5)) is employed to model particledispersion and deposition from on and near roadways (Tong et al.,2012; Tong and Zhang, 2015). Particle size distributions (PSD) aremodeled as discrete size bins.

∂ gNp DPð Þ∂t

þ d eujgNp DPð Þ

dxj¼ ∂

∂xjDt þ Dmð Þ ∂

gNp DPð Þ∂xj

" #þ Sd DPð Þ ð5Þ

gNpðDPÞ is the average particle concentration for each particle size. Dt

is the turbulent diffusivity, and Dm is the molecular diffusivity. Sd is thesink term to represent the effect of deposition, which is the function ofparticle size DP.

Sd Dp� � ¼ β Dp

� �Np Dp

� � ð6Þ

β(Dp) is the size-resolved particle loss coefficient, and it is given as:

β Dp� � ¼ vd Dp

� �A

Vð7Þ

where vd (Dp) is deposition velocity, A is the inner surface area, and V isthe volume of the room. In this study, we adopt a well-establisheddry deposition model by Lai and Nazaroff (2000), accounting forBrownian diffusion, turbulent diffusion, and gravitational settling.Only dispersion and deposition in the model are considered. Anyprocess that leads to particle transformation is not explicitly simulatedas a model limitation. However, by deriving the emission factors fromthe near-road measurement, we have equivalently incorporated theaerosol dynamics (Tong et al., 2016).

2.3. Model evaluation

2.3.1. ExperimentThe CFDmodel is evaluated with PIV measurement of flow over and

through a cross-ventilated building model in a boundary layer windtunnel (Karava et al., 2011). Building models with different openingswere built from 2 mm cast transparent polymethylmethacrylate sheetat a scale of 1:200. The modeled buildings have a dimension ofL × W × H = 100 × 100 × 80 mm3 at wind-tunnel scale and20 × 20 × 16 m3 at real scale. The thickness of the wall is 2 mm.Configurations with both openings at the center of the two oppositewalls and with wall porosities (the ratio between the opening areaand the area of the wall) of 5% and 10% are shown in Fig. 1a and b.The height of the opening is 18 mm for both configurations while thewidth is varied to provide the corresponding porosity.

Fig. 1. Schematics of the numerical setup for the CFD evaluation. a) Wall porosity = 5%, b) wall porosity = 10%, c) the dimensions of the building model in the wind tunnel.

140 Z. Tong et al. / Environment International 89–90 (2016) 138–146

The opening of the building model was placed perpendicular to theapproaching flow. The mean velocity and turbulence intensity profileswere measured with a hot-film probe in the test section. The PIVmeasurements were conducted in the vertical plane of symmetrythrough the openings.

2.3.2. Numerical setupThe schematic drawing for simulating the experiment is shown in

Fig. 2. The numerical setup of the boundary condition in this case is sim-ilar to existing studies and generally follows the CFD guidelines for sim-ulating flows in the urban environment (Blocken, 2015; Franke et al.,2011; Ramponi and Blocken, 2012b; Tominaga et al., 2008; Tominagaand Stathopoulos, 2010). The inlet boundary condition is obtainedbased on measured wind speed and turbulent intensity. A logarithmicinlet velocity profile was employed according to Eq. (8).

u zð Þ ¼ u�

κln

zþ z0z0

ð8Þ

where κ is von Karman constant (0.42), and z0 is aerodynamic rough-ness height (0.025 mm). u⁎ is calculated from the measured velocity uat its reference height. Turbulent kinetic energy k is estimated accordingto Eq. (9) based on previous study (Tominaga et al., 2008). Theturbulence dissipation rate is derived from Eq. (10).

k zð Þ ¼ I zð Þu zð Þð Þ2 ð9Þ

ϵ zð Þ ¼ u�3

κ zþ z0ð Þ ð10Þ

The time-dependent feature of the inlet turbulence profile is simu-lated by the vortex method, where random vortices in the inlet flow

Fig. 2. Top view of the numerical setup.

plane for the wall-normal components are generated, providing aspatial correlation (Kraichnan, 1970; Mathey et al., 2006; Tong andZhang, 2015). Symmetry boundary conditions are applied for the twosides as slip walls with zero shear. Outflow boundary condition isspecified at the end of the domain. The standard wall function with aroughness modification is employed (Cebeci and Bradshaw, 1977;Launder and Spalding, 1974). The sand-grain roughness height ks androughness constant Cs are estimated from the aerodynamic roughnessheight z0 based on the relationship derived by Blocken et al. (2007).The computational domain contains 2.1 million unstructured cells. Agrid independency study is conducted to ensure the results areindependent of the mesh resolution. The grids are more refined nearthe solid cube. Prism layer mesh is applied near the wall in order toreduce numerical diffusion. The horizontal homogeneity is evaluatedby creating a flow simulation in an empty domain of the same size assuggested by Blocken et al. (2007). The unintended differences betweenthe inlet and incident profiles are very minor. Figures that present thehorizontal homogeneity, grid topology, grid convergence study areincluded in the Supplementary Material.

2.3.3. Results of model evaluationFig. 3 depicts a comparison between predicted and measured time-

averaged streamwise flow velocity profiles. Overall, the performanceof the LES model is satisfactory. For the 5% porosity case, themodel was able to accurately capture the flow acceleration nearboth openings, as well as deceleration inside the building. The 10% po-rosity case presents a less accurate agreement. This is likely due to thereason that the size and position of the upstream vortex in the PIVexperiment was not able to be reproduced in the LES simulation. Dis-crepancies are observed at the inlet and outlet openings (x/D ~ 0) andfloor (x/D ~ 1.1). As discussed in the study by Karava et al. (2011), theexperimental data is not reliable in the immediate vicinity of the open-ings due to the effects of shadows or reflections that might haveproduced uncertainties in the PIV measurements at these positions.

2.4. Case study: Naturally ventilated building near a roadway

The motivation of this study is to evaluate the potential ofnatural ventilation in a near-road environment. In this section, themodeled road-building geometry (Fig. 4) is a representative andgeneralizable scenario in an urban setting where most people livein buildings that are in close vicinity of roadways (HEI, 2010). Wecreated a series of test scenarios to quantitatively evaluate the im-pact of traffic-related air pollutants on indoor air quality throughnatural ventilation.

Fig. 3. Comparison of experimental and numerical results for the normalized streamwise velocity component. The experimental data is from the study by Karava et al. (2011).

141Z. Tong et al. / Environment International 89–90 (2016) 138–146

2.4.1. Model setup of the case studyFig. 4 presents the modeled geometry where a naturally ventilated

building is located near a major roadway and test scenarios for variousbuilding parameters, i.e., distance (Fig. 4b), window size (Fig. 4c), win-dow opening location (Fig. 4d), and oncomingwind speed (Fig. 4e). Fig.4a shows the baseline scenario (D=10m, wall porosity=9%, windowopening location= Case 1, wind speed=u). With regard to the locationof window openings (Fig. 4d), three scenarios are evaluated. Case 1 rep-resents the condition where the interior doors are shut. Case 1′ has thesame configuration as Case 1 except the open interior doors. Case 2 rep-resents the scenario where the windows facing the roadway are closed.Case 3 shows the situation where both the side windows and windowsfacing the roadway are shut. A roadway is modeled as a well-mixedemission zone similar to our previous studies on near-road air quality(Tong et al., 2012; Wang et al., 2013; Wang and Zhang, 2009). The on-road size-resolved emission rate and vehicle-induced turbulence (VIT)in this study is obtained from previous study (Hagler et al., 2012; Tonget al., 2016). Themodeled road represents amajor roadwaywith Annu-al Average Daily Traffic (AADT) of 38,000 and a speed limit of 45mph. Asmall office building is created at a distance D from the roadway. Thebuilding geometry employed in this study is chosen from the U.S. De-partment of Energy (DOE) commercial reference building database

Fig. 4. Schematics of the baseline geometry (a) and test building parameters, i.e., distance

(Deru et al., 2011), which represents realistic building characteristicsand construction practices. The dimensions of the windows are1.8 × 0.8m2. There are six windows on the sides parallel to the roadwayand four on the perpendicular sides. For tested configurations, the com-putational domain has a dimension of 200 m × 200 m × 40 m dividedinto 3.3 to 4.5 million unstructured elements. A grid independencestudy is conducted to ensure the results are independent of the meshresolution. The height of the domain is more than 5 times the heightof the building, which avoids blocking effect and unphysical flow accel-eration (Tominaga et al., 2008). The boundary conditions in the casestudy are similar to Section 2.3.2. The ground and building surfacesare defined as a no-slipwall. The top and two sides of the computationaldomain are set as symmetry conditions with zero shear. The outflowboundary condition is specified at the outlet of the domain. A neutrallystratified wind profile similar to Eq. (8) is specified at the inlet perpen-dicular to the roadwaywith a referencewind speed uref=3m/s at z=3m. Any surrounding structures and topographies that may complicatethe near-road dispersion are not considered in this study. Traffic emis-sion is the only modeled source in the computational domain. Infiltra-tion is neglected in the model as the natural ventilation flow rates atthe window openings are much larger than those from infiltration(Deru and Burns, 2003).

(b), window size (c), window opening locations (d), and oncoming wind speed (e).

Fig. 5.Normalized particle concentration C′(i.e., 253 nm) contour plot of the baseline configuration: a) Case 1with interior doors closed; b) Case 1′with interior doors open. The top view istaken at the height of window openings, and the side view is taken at the domain centerline. A grayscale figure is presented in the Supplementary Material.

Fig. 6. Horizontal gradients of normalized particle concentrations. C′ is the particleconcentration normalized by the maximum on-road concentration of the particle at thesame size. Distance is normalized by the road width. Nine particle sizes are modeled,but only five are displayed for clarity. The floor plan of the building is overlaid on the plot.

142 Z. Tong et al. / Environment International 89–90 (2016) 138–146

3. Results and discussion

3.1. Baseline configuration

Fig. 5 displays the contour plots of the normalized particle concen-tration (253 nm) from the baseline configuration. Fig. 5a presents theconfiguration where the interior doors are closed (Case 1) whereasthe configuration with open interior doors (Case 1′) is presented inFig. 5b. To generalize the results, the roadside particle concentrations(C′) are normalized by the maximum on-road concentration and thedistance away from the road (D′) is normalized by the road width,which to a large extent eliminates the effects of uncertainties associatedwith road type and traffic condition (Tong et al., 2012). The presence ofthe roadside building decelerates the incident flow and greatly elevatesthe concentration near the windward wall of the building. Similar to asolid cube, which has been well studied in the past (Castro andRobins, 1977; Martinuzzi and Tropea, 1993; Murakami, 1993; Rodi,1997; Yakhot et al., 2006), the oncoming flow separates at the frontcorners of the building, and forms a horseshoe-type vortex that bendsaround the cube and extends downstream. A great portion of the trafficplume first spreads laterally, passing the front sharp edges and continu-ing downwind in the direction of the horseshoe vortex. The openings atthewindwardwall draw the plume into the office. The flow recirculatesinside the room and creates a well-mixed zone with little variation inparticle concentration as shown in Fig. 5. In Case 1, a sharp concentra-tion drop is observed at the partition between the open office space(Zone 1) and the private offices (Zones 2, 3, 4) along the oncomingwind direction (Fig. 5 and Fig. 6). This is due to the wall separationamong zones. As shown in the side view of Fig. 5, another portion ofthe traffic plume is forced to deflect upward when approaching thebuilding. Unlike the configuration with a source on the rooftop stackpresented in other studies (Blocken et al., 2008; Gousseau et al.,2011b; Li and Stathopoulos, 1997; Tominaga and Stathopoulos, 2013),the entrainment of pollutants into the leeward recirculation zone is

absent in the configuration with an upwind ground-level source pre-sented in this study. This evident difference in dispersion patterns isdue to both the lengthier distance from the upwind source, which cre-ates sufficient dilution before entering the rearwake zone, and thewin-dow openings in the windward wall that draw a great portion of theplume towards the indoor environment.

In Case 1′ (Fig. 5b), keeping the interior doors open on the separationwall causes an inflow of particles from the large office (Zone 1) to the

Fig. 8.Normalized average indoor concentration C′ as a function of distance D′ away fromroadway. Concentration is normalized by the maximum on-road concentration, anddistance is normalized by the width of the roadway. For clarity, only the relationshipsbased on 15 nm and 253 nm particles are presented.

143Z. Tong et al. / Environment International 89–90 (2016) 138–146

private rooms (Zones 2, 3, 4). As a result, the approximatelywell-mixedzone is expanded to all rooms in the building, and the large drop in par-ticle concentration occurs at the leeward external wall instead. In com-parison with Case 1, the average indoor concentration (i.e., 253 nm) is6% lower since opening the interior doors moderately enhances thecross-ventilation and particle dilution. The area downwind of the build-ing becomes slightly more polluted due to the plume outflow from theopenings at the leewardwall of the building,which is not seenwhen theinterior doors are shut.

The normalized horizontal gradients of particle concentration C′for the baseline are shown in Fig. 6. The gradients are particle size-dependent inside the office due to surface deposition. For the baselineconfiguration, the particle concentration is roughly 16–21% greater(particle size-dependent) indoors than at the edge of the roadway fortwo reasons: 1) the flow deceleration near the windward side; and2) the wake zone due to restricted dispersion inside the building. Bothinstances reduce the dilution in the near-road indoor and outdoorenvironments. Unlike the nearly constant concentration of fine particles(i.e., 253 nm), a noticeable decreasewith distance in particle concentra-tions indoors are observed for ultrafine particles (b100 nm)with great-er deposition velocity due to Brownian diffusion. In contrast with thesmall rooms (Zones 2, 3, 4), the decrease in particle concentrationwith distance is less in Zone 1 as the surface-to-volume ratio is smaller.

The particle size distribution (PSD) at the edge of the highway,indoors, and immediately downwind of the building are shown in Fig.7. The horizontal axis represents particle size on a logarithmic scale,while the vertical axis represents particle number concentration. Theparticle number concentration varies considerably with particle sizeand distance from the road. The peak of particle number concentrationis observed at the size of 15 nm due to the characteristic of on-roademission, and generally decreases with increasing particle size in theimmediate near-road environment. The particle number concentrationindoors is shown consistently greater than that at the edge of the high-way and a drop in particle number concentration is seen downwind ofthe building. In contrast with larger particles (dp N 100 nm), the numberconcentrations for smaller ones drop more sharply across the buildingdue to deposition effect.

3.2. Distance and wind speed

The average indoor particle concentration as a function of distanceaway from the roadway is displayed in Fig. 8. For clarity,we only present

Fig. 7. Particle size distribution (PSD) at the edge of roadway, inside the building, andimmediate downwind of the building.

the relationships based on particle sizes at the two ends of the particlesize distribution (i.e., 15 nm and 253 nm).The exponential curve fittingis based on concentrations atfive simulated distances as shown in Fig. 8.The fitting curve of 15 nm particles is nearly parallel to that of 253 nmparticles in the accumulation mode with lower concentration. This ismostly due to the removal of ultrafine particles by Brownian diffusion.At D' = 1, the indoor concentration is approximately 16–21% greaterthan the concentration at the edge of the highway upwind of theoffice building. A break-even point is observed atD' ~ 2.1where the par-ticle concentration declines to the same level as that at the edge of theroadway. Placing the building further beyond that point will reducethe indoor concentration relative to the roadway. At D'= 4, the indoorconcentration declines to approximately 82% of the roadsideconcentration.

The effect of oncomingwind speed on the average indoor concentra-tion is explored by varying the baseline wind speed to 0.5 u′, 1.5 u′, and2 u′ (u′ is the baseline wind speed). The results and curve fitting of thefour simulated scenarios are shown in Fig. 9. The average indoor particleconcentration is found to decline exponentially with increasing windspeed towards the building. The cross-ventilation and dispersion ofparticles between indoors and outdoors is improved by increasing thewind speed, which lowers indoor concentration as a result. The differ-ence in indoor concentration between 15 nm and 253 nm particles isgreatest at u′ = 0.5 and gradually decreases towards u′ = 2. This is aresult of reduced surface deposition (mostly due to Brownian diffusionfor UFPs) with decreasing particle residence time in the building.

Fig. 9.Normalized average indoor concentration as a function of oncoming wind speed u′.Concentration is normalized by the maximum on-road concentration. Wind speed isnormalized by the wind speed of the baseline case (uref = 3 m/s). For clarity, only therelationships based on 15 nm and 253 nm particles are presented.

Fig. 10. Horizontal gradients of normalized particle concentration C′: a) different wall porosity (or window opening size); b) different window opening location. Only the gradients of253 nm particle are presented for clarity. The floor plan of the building is overlaid on the plot.

144 Z. Tong et al. / Environment International 89–90 (2016) 138–146

3.3. Window opening size and location

The impact of the size and locations of the window openings isstudied in this section. In contrast with the baseline, doubling the sizeof a window opening decreases the indoor concentration as shown inFig. 10a. This is due to increased cross-ventilation and the air changerate, which enhances the indoor dilution. The plume that is drawninto Zone 1 is deflected sideways by the separation wall and flows outfromwindows at the sides. Enhanced lateralmomentum and dispersionlead to a greater concentration drop across the building, and thus alower concentration also downwind of the building. Unlike the zoneof nearly constant concentration observed in the baseline configuration(Case 1), a decreasing concentration occurs as distance increases insideZone 1when the size of window openings are doubled (Fig. 10a). This isa result of sufficient dilution by cross ventilation.

Fig. 10b presents the impact of a window opening location onthe decline of particle concentration with increasing distance from theroadway. Closing the windows facing the roadway (Case 2) noticeablyreduces the indoor concentration by blocking the direct pathway ofparticles to the indoor environment. A 22.5% reduction from the base-line configuration (Case 1) is found in Case 2. In addition to closingthewindows facing the roadway, closing the sidewindows (Case 3) fur-ther decreases the indoor concentration to 50% of the baseline condi-tion. In contrast with Cases 2 and 3, opening windows facing theroadway (Case 1) results in a lower concentration behind the buildingalthough the indoor concentration becomes much greater. This is be-cause more particles are drawn and suspended in the office, and lessparticles travel to the downwind side of the building.

4. Conclusion

The primary objective of this study is to quantify the impact ofvarious building parameters on the indoor air quality of a naturallyventilated building in a near-road environment. We first evaluatedthe model performance against the PIV measurement on a cross-ventilated building model, and a good agreement between simulationand experiment is observed. Our study found that the building enveloperestricts dispersion and dilution of particulate matter. For the baselineconfiguration where the building is 10 m away from the roadway, theindoor concentration is roughly 16–21% greater than that at the edgeof roadway. The indoor flow recirculation creates a well-mixed zone

with little variation in fine particle concentration. For ultrafine particles,the concentration indoors decreases as the distance from the roadincreases. This is because deposition is mostly driven by Browniandiffusion, which is more effective at these particle sizes.

The indoor air quality strongly depends on the distance between theroadway and building, particle size, wind condition, window size, andwindow opening location. The indoor particle concentration is foundto decline exponentially with the distance from the roadway. A break-even point is observed at D′ ~ 2.1 (normalized distance from the road-way by the width of the road). The concentration is greater indoorsthan at the highwaywhereD′ b 2.1, and vice versa. For buildings locatedin close vicinity to roadways, this relationship suggests the need forme-chanical ventilationwith fresh air intakes located on the leeward side ofthe building or at the rooftop level where the air is less polluted by traf-fic emission (Tong et al., 2015a; Wu et al., 2002). The impact of oncom-ing wind speed is also studied, finding that the average indoor particleconcentration declines exponentially with increasing wind speed to-wards the building. The difference in indoor concentration between15 nm and 253 nm particles is greatest at u′ = 0.5 and gradually de-creases towards u′ = 2. Increasing the window opening improvescross-ventilation and reduces the indoor particle concentration as a re-sult. Meanwhile, the increased lateral dispersion results in a lower con-centration behind the building. Our analysis also suggests that theindoor air quality can be considerably improved by naturally drawingair from less contaminated sides of the building. A 22.5% reduction inparticle concentration from the baseline configuration is achieved byclosing windows facing the roadway. Further closing the windows atthe side of the building decreases the indoor concentration to 50%from the baseline.

The findings concluded in this paper demonstrate a need foreffective ventilation strategies that reduce human exposure to traffic-related air pollution in near-road environments. For the planning ofnew buildings, the distance from the roadway and the ambient windcondition need to be deliberately evaluated at the early design stagein order to maximize the potential of natural ventilation with theconsideration of traffic-related air pollution. On the other hand, forexisting buildings located adjacent to roadways, utilizing effectiveventilation strategies that take into account the size and location ofwindow openings, the layout of interior space, and the placement offresh air intakes are alternatives that can improve the indoor air quality.The conclusions herein can provide guidelines to assist architects, urbanplanners and building operators in evaluating natural ventilation

145Z. Tong et al. / Environment International 89–90 (2016) 138–146

strategies with a heightened consideration for indoor air qualityespecially in developing countries. Given the complexity of the urbanenvironment, future implementation needs to take into account thesynergistic effects of other factors such as acoustics, road configuration,and surrounding microenvironment.

Acknowledgments

The authors would like to thank Dr. Stephen Ervin and Dr. Bin Yanfor their helpful comments.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.envint.2016.01.016.

References

Blocken, B., 2015. Computational fluid dynamics for urban physics: importance, scales,possibilities, limitations and ten tips and tricks towards accurate and reliablesimulations. Build. Environ. 91, 219–245.

Blocken, B., Stathopoulos, T., Carmeliet, J., 2007. CFD simulation of the atmosphericboundary layer: wall function problems. Atmos. Environ. 41, 238–252.

Blocken, B., Stathopoulos, T., Saathoff, P., Wang, X., 2008. Numerical evaluation of pollut-ant dispersion in the built environment: comparisons between models andexperiments. J. Wind Eng. Ind. Aerodyn. 96, 1817–1831.

Blondeau, P., Iordache, V., Poupard, O., Genin, D., Allard, F., 2005. Relationship betweenoutdoor and indoor air quality in eight French schools. Indoor Air 15, 2–12.

Brunekreef, B., Janssen, N., de Hartog, J.J., Oldenwening, M., Meliefste, K., Hoek, G., et al.,2005. Personal, indoor, and outdoor exposures to PM2. 5 and its components forgroups of cardiovascular patients in Amsterdam and Helsinki. Res. Rep. Health Eff.Inst. 1–70 (discussion 71–9).

Cardinale, N., Micucci, M., Ruggiero, F., 2003. Analysis of energy saving using naturalventilation in a traditional Italian building. Energy Build. 35, 153–159.

Castro, I., Robins, A., 1977. The flow around a surface-mounted cube in uniform andturbulent streams. J. Fluid Mech. 79, 307–335.

Cebeci T, Bradshaw P. Momentum transfer in boundary layers. Washington, DC,Hemisphere Publishing Corp.; New York, McGraw-Hill Book Co., 1977. 407p. 1977; 1.

Chen, C., Zhao, B., 2011. Review of relationship between indoor and outdoor particles: I/Oratio, infiltration factor and penetration factor. Atmos. Environ. 45, 275–288.

Chen, F., Yu, S.C.M., Lai, A.C.K., 2006. Modeling particle distribution and deposition inindoor environments with a new drift-flux model. Atmos. Environ. 40, 357–367.

Deru, M., Burns, P., 2003. Infiltration and Natural Ventilation Model for Whole BuildingEnergy Simulation of Residential Buildings.

Deru, M., Field, K., Studer, D., Benne, K., Griffith, B., Torcellini, P., et al., 2011. USDepartment of Energy Commercial Reference Building Models of the National Build-ing Stock.

Di Sabatino, S., Buccolieri, R., Pulvirenti, B., Britter, R., 2007. Simulations of pollutantdispersion within idealised urban-type geometries with CFD and integral models.Atmos. Environ. 41, 8316–8329.

Dockery, D.W., Pope, C.A., Xu, X., Spengler, J.D., Ware, J.H., Fay, M.E., et al., 1993. Anassociation between air pollution and mortality in six U.S. cities. N. Engl. J. Med.329, 1753–1759.

EPA, U.S., 2014. An Introduction to Indoor Air Quality (IAQ).Fischer, P.H., Hoek, G., van Reeuwijk, H., Briggs, D.J., Lebret, E., vanWijnen, J.H., et al., 2000.

Traffic-related differences in outdoor and indoor concentrations of particles andvolatile organic compounds in Amsterdam. Atmos. Environ. 34, 3713–3722.

Franke, J., Hellsten, A., Schlunzen, K.H., B., C., 2011. The COST 732 Best Practice Guidelinefor CFD simulation of flows in the urban environment: a summary. Int. J. Environ.Pollut. 44, 419–427.

Germano, M., Piomelli, U., Moin, P., Cabot, W.H., 1991. A dynamic subgrid-scale eddyviscosity model. Phys. Fluids A 3, 1760–1765.

Gousseau, P., Blocken, B., van Heijst, G.J.F., 2011b. CFD simulation of pollutant dispersionaround isolated buildings: on the role of convective and turbulent mass fluxes in theprediction accuracy. J. Hazard. Mater. 194, 422–434.

Gousseau, P., Blocken, B., Stathopoulos, T., van Heijst, G.J.F., 2011a. CFD simulation of near-field pollutant dispersion on a high-resolution grid: a case study by LES and RANS fora building group in downtown Montreal. Atmos. Environ. 45, 428–438.

Gratia, E., De Herde, A., 2004. Natural cooling strategies efficiency in an office buildingwith a double-skin façade. Energy Build. 36, 1139–1152.

Hagler, G.S.W., Lin, M.-Y., Khlystov, A., Baldauf, R.W., Isakov, V., Faircloth, J., et al., 2012.Field investigation of roadside vegetative and structural barrier impact on near-road ultrafine particle concentrations under a variety of wind conditions. Sci. TotalEnviron. 419, 7–15.

HEI, 2010. Traffic-related Air Pollution: A Critical Review of the Literature on Emissions,Exposure, and Health Effects. Health Effects Institute, Boston, MA.

van Hooff, T., Blocken, B., 2010. Coupled urban wind flow and indoor natural ventilationmodelling on a high-resolution grid: a case study for the Amsterdam ArenA stadium.Environ. Model Softw. 25, 51–65.

Janssen, N.A.H., van Vliet, P.H.N., Aarts, F., Harssema, H., Brunekreef, B., 2001. Assessmentof exposure to traffic related air pollution of children attending schools near motor-ways. Atmos. Environ. 35, 3875–3884.

Jenkins, P.L., Phillips, T.J., Mulberg, E.J., Hui, S.P., 1992. Fifth International Conference onIndoor Air Quality and Climate Indoor Air '90: characterization of indoor air activitypatterns of Californians: use of and proximity to indoor pollutant sources. Atmos. En-viron. Part A 26, 2141–2148.

Karava, P., Stathopoulos, T., Athienitis, A.K., 2011. Airflow assessment in cross-ventilatedbuildings with operable façade elements. Build. Environ. 46, 266–279.

Klepeis, N.E., Nelson, W.C., Ott, W.R., Robinson, J.P., Tsang, A.M., Switzer, P., et al., 2001.The National Human Activity Pattern Survey (NHAPS): a resource for assessingexposure to environmental pollutants. J. Expo. Anal. Environ. Epidemiol. 11, 231–252.

Kolokotroni, M., Aronis, A., 1999. Cooling-energy reduction in air-conditioned offices byusing night ventilation. Appl. Energy 63, 241–253.

Koponen, I., Asmi, A., Keronen, P., Puhto, K., Kulmala, M., 2001. Indoor air measurementcampaign in Helsinki, Finland 1999 — the effect of outdoor air pollution on indoorair. Atmos. Environ. 35, 1465–1477.

Kraichnan, R.H., 1970. Diffusion by a random velocity field. Phys. Fluids 13, 22–31.Lai, A.C., Nazaroff, W.W., 2000. Modeling indoor particle deposition from turbulent flow

onto smooth surfaces. J. Aerosol Sci. 31, 463–476.Launder, B.E., Spalding, D.B., 1974. The numerical computation of turbulent flows.

Comput. Methods Appl. Mech. Eng. 3, 269–289.Leung, D.Y.C., 2015. Outdoor–indoor air pollution in urban environment: challenges and

opportunity. Front. Environ. Sci. 2.Li, Y., Chen, Z., 2003. A balance-point method for assessing the effect of natural ventilation

on indoor particle concentrations. Atmos. Environ. 37, 4277–4285.Li, Y., Stathopoulos, T., 1997. Numerical evaluation of wind-induced dispersion of

pollutants around a building. J. Wind Eng. Ind. Aerodyn. 67–68, 757–766.Liu, L.S., Box, M., Kalman, D., Kaufman, J., Koenig, J., Larson, T., et al., 2003. Exposure

assessment of particulate matter for susceptible populations in Seattle. Environ.Health Perspect. 111, 909.

Martinuzzi, R., Tropea, C., 1993. The flow around surface-mounted, prismatic obstaclesplaced in a fully developed channel flow (data bank contribution). J. Fluids Eng.115, 85–92.

Mathey, F., Cokljat, D., Bertoglio, J.P., Sergent, E., 2006. Assessment of the vortexmethod for large eddy simulation inlet conditions. Prog. Comput. Fluid Dyn.Int. J. 6, 58–67.

Meng, Q.Y., Turpin, B.J., Korn, L.,Weisel, C.P., Morandi,M., Colome, S., et al., 2005. Influenceof ambient (outdoor) sources on residential indoor and personal PM2. 5 concentra-tions: analyses of RIOPA data. J. Expo. Sci. Environ. Epidemiol. 15, 17–28.

Monn, C., Fuchs, A., Högger, D., Junker, M., Kogelschatz, D., Roth, N., et al., 1997. Particulatematter less than 10 μm (PM10) and fine particles less than 2.5 μm (PM2.5):relationships between indoor, outdoor and personal concentrations. Sci. Total Envi-ron. 208, 15–21.

Murakami, S., 1993. Comparison of various turbulence models applied to a bluff body.J. Wind Eng. Ind. Aerodyn. 46, 21–36.

Pellizzari, E., Clayton, C., Rodes, C., Mason, R., Piper, L., Fort, B., et al., 2000. Particulatematter and manganese exposures in Indianapolis, Indiana. J. Expo. Anal. Environ.Epidemiol. 11, 423–440.

Pope, S.B., 2000. Turbulent Flows. Cambridge University Press.Ramponi, R., Blocken, B., 2012a. CFD simulation of cross-ventilation flow for different

isolated building configurations: validation with wind tunnel measurements andanalysis of physical and numerical diffusion effects. J. Wind Eng. Ind. Aerodyn.104–106, 408–418.

Ramponi, R., Blocken, B., 2012b. CFD simulation of cross-ventilation for a generic isolatedbuilding: impact of computational parameters. Build. Environ. 53, 34–48.

Rijnders, E., Janssen, N.A., van Vliet, P.H., Brunekreef, B., 2001. Personal and outdoornitrogen dioxide concentrations in relation to degree of urbanization and trafficdensity. Environ. Health Perspect. 109, 411–417.

Rodi, W., 1997. Comparison of LES and RANS calculations of the flow around bluff bodies.J. Wind Eng. Ind. Aerodyn. 69–71, 55–75.

Samet, J.M., Marbury, M.C., Spengler, J.D., 1987. Health effects and sources of indoor airpollution. Part I. Am. Rev. Respir. Dis. 136, 1486–1508.

Spengler, J.D., Chen, Q., 2000. Indoor air quality factors in designing a healthy building.Annu. Rev. Energy Environ. 25, 567–600.

Tominaga, Y., Stathopoulos, T., 2010. Numerical simulation of dispersion around an isolatedcubic building: model evaluation of RANS and LES. Build. Environ. 45, 2231–2239.

Tominaga, Y., Stathopoulos, T., 2013. CFD simulation of near-field pollutant dispersion inthe urban environment: a review of current modeling techniques. Atmos. Environ.79, 716–730.

Tominaga, Y., Mochida, A., Yoshie, R., Kataoka, H., Nozu, T., Yoshikawa, M., et al., 2008. AIJguidelines for practical applications of CFD to pedestrian wind environment aroundbuildings. J. Wind Eng. Ind. Aerodyn. 96, 1749–1761.

Tong, Z., Zhang, K.M., 2015. The near-source impacts of diesel backup generators in urbanenvironments. Atmos. Environ. 109, 262–271.

Tong, Z., Whitlow, T.H., Landers, A., Flanner, B., 2015a. A case study of air quality above anurban roof top vegetable farm. Environ. Pollut.

Tong, Z.,Whitlow, T.H., MacRae, P.F., Landers, A.J., Harada, Y., 2015b. Quantifying the effectof vegetation on near-road air quality using brief campaigns. Environ. Pollut. 201,141–149.

Tong, Z., Baldauf, R.W., Isakov, V., Deshmukh, P., Max, Z.K., 2016. Roadside vegetationbarrier designs to mitigate near-road air pollution impacts. Sci. Total Environ. 541,920–927.

Tong, Z., Wang, Y.J., Patel, M., Kinney, P., Chrillrud, S., Zhang, K.M., 2012. Modeling spatialvariations of black carbon particles in an urban highway-building environment.Environ. Sci. Technol. 46, 312–319.

146 Z. Tong et al. / Environment International 89–90 (2016) 138–146

Vedal, S., 1997. Ambient particles and health: lines that divide. J. Air Waste Manage.Assoc. 47, 551–581.

Wang, Y.J., Zhang, K.M., 2009. Modeling near-road air quality using a computational fluiddynamics model, CFD-VIT-RIT. Environ. Sci. Technol. 43, 7778–7783.

Wang, Y.J., Zhang, K.M., 2012. Coupled turbulence and aerosol dynamics modeling ofvehicle exhaust plumes using the CTAG model. Atmos. Environ. 59, 284–293.

Wang, Y.J., Nguyen, M.T., Steffens, J.T., Tong, Z., Wang, Y., Hopke, P.K., et al., 2013.Modeling multi-scale aerosol dynamics and micro-environmental air qualitynear a large highway intersection using the CTAG model. Sci. Total Environ.443, 375–386.

Wu, Y., Hao, J., Fu, L., Wang, Z., Tang, U., 2002. Vertical and horizontal profiles of airborneparticulate matter near major roads in Macao, China. Atmos. Environ. 36, 4907–4918.

Xie, Z., Castro, I., 2006. LES and RANS for turbulent flow over arrays of wall-mounted ob-stacles. Flow Turbul. Combust. 76, 291–312.

Yakhot, A., Liu, H., Nikitin, N., 2006. Turbulent flow around a wall-mounted cube: a directnumerical simulation. Int. J. Heat Fluid Flow 27, 994–1009.

Zhang, Z., Chen, Q., 2009. Prediction of particle deposition onto indoor surfaces by CFDwith a modified Lagrangian method. Atmos. Environ. 43, 319–328.

Zhang, S., Wu, Y., Liu, H., Huang, R., Yang, L., Li, Z., et al., 2014a. Real-world fuelconsumption and CO2 emissions of urban public buses in Beijing. Appl. Energy 113,1645–1655.

Zhang, S., Wu, Y., Hu, J., Huang, R., Zhou, Y., Bao, X., et al., 2014b. Can Euro V heavy-dutydiesel engines, diesel hybrid and alternative fuel technologies mitigate NOXemissions? New evidence from on-road tests of buses in China. Appl. Energy 132,118–126.

Zhao, B., Zhang, Y., Li, X., Yang, X., Huang, D., 2004. Comparison of indoor aerosol particleconcentration and deposition in different ventilated rooms by numerical method.Build. Environ. 39, 1–8.

Zhu, Y., Hinds, W.C., Kim, S., Shen, S., Sioutas, C., 2002. Study of ultrafine particles near amajor highway with heavy-duty diesel traffic. Atmos. Environ. 36, 4323–4335.

Zhu, Y., Hinds, W.C., Krudysz, M., Kuhn, T., Froines, J., Sioutas, C., 2005. Penetration offreeway ultrafine particles into indoor environments. J. Aerosol Sci. 36, 303–322.