ABSTRACT
ABSTRACT
In atmospheric environment, the layout difference of urban buildings has a powerful influence on accelerating or inhibiting the dispersion of particle matters (PM). In industrial cities, buildings of variable heights can obstruct the diffusion of PM from industrial stacks. In this study, PM dispersed within building groups was simulated by Reynolds-averaged Navier-Stokes equations coupled Lagrangian approach. Four typical street building arrangements were used: (a) a low-rise building block with Height/base H/b = 1 (b = 20 m); (b) step-up building layout (H/b = 1, 2, 3, 4); (c) step-down building layout (H/b = 4, 3, 2, 1); (d) high-rise building block (H/b = 5). Profiles of stream functions and turbulence intensity were used to examine the effect of various building layouts on atmospheric airflow. Here, concepts of particle suspension fraction and concentration distribution were used to evaluate the effect of wind speed on fine particle transport. These parameters showed that step-up building layouts accelerated top airflow and diffused more particles into street canyons, likely having adverse effects on resident health. In renewal old industry areas, the step-down building arrangement which can hinder PM dispersion from high-level stacks should be constructed preferentially. High turbulent intensity results in formation of a strong vortex that hinders particles into the street canyons. It is found that an increase in wind speed enhanced particle transport and reduced local particle concentrations, however, it did not affect the relative location of high particle concentration zones, which are related to building height and layout.
Implications: This study has demonstrated the height variation and layout of urban architecture affect the local concentration distribution of particulate matter (PM) in the atmosphere and for the first time that wind velocity has particular effects on PM transport in various building groups. The findings may have general implications in optimization the building layout based on particle transport characteristics during the renewal of industrial cities. For city planners, the results and conclusions are useful for improving the local air quality. The study method also can be used to calculate the explosion risk of industrial dust for people who live in industrial cities.
Introduction
With ongoing development of cities, the urban population is increasing and the scale of urbanization expanding. Typically, industrial districts originally on the urban fringes have been engulfed by urban growth, causing more people to be exposed to industrial pollutants. Particulate matter (PM) from industrial emissions is one of the most important sources of atmospheric haze. High concentrations of particles are associated with urban centers and industrial zones, where air quality is seriously affected by local and regional anthropogenic emissions (Pontiggia et al. 2010).
The dispersion and deposition characteristics of atmospheric particles are closely related to underlying surface topography or urban structure in a region. During the renewal of old cities, a large number of new high-rise buildings are often built near existing low-rise buildings. This causes a height disparity over the urban surface, which markedly changes the originally flat dispersion trajectories of atmospheric PM. Studies have shown that pollutant concentrations in cities with street canyons are several times higher than those without such obstacles (Kumar et al. 2009). Thus, the layout difference of urban buildings has a powerful influence on local airflow (Hang, Li, and Sandberg 2011), which is extremely important for accelerating or inhibiting the dispersion of pollutants in cities.
A two-dimensional (2D) idealized urban street is a simplification of actual urban geometry (Chan et al. 2002). Numerous studies have qualitatively focused on the impacts of building heights (Li, Liu, and Leung 2008), shape of building roofs (Huang, Hu, and Zeng 2009), and density of building groups (Scargiali et al. 2011) on the diffusion of pollutants (Addepalli and Pardyjak 2013; Gu et al. 2011; Hang et al. 2012; Huang, He, and Kim 2015; Scungio 2013). Hang et al. (2012) numerically studied a parallel approaching wind on arrays of uniform height, using a normalized pollutant transport rate to quantify the contribution of mean flow and turbulent diffusion to pollutant removal. They found that decreasing aspect ratios or increasing street lengths may enhance pollutant removal by turbulent diffusion across canopy roofs. Gu and Zhang et al. (2011) investigated airflow and pollutant dispersion in these nonuniform street canyons. They confirmed that at pedestrian level, the concentrations of simulated pollutants in such canyons are lower than those in uniform ones, suggesting that uneven building layouts can improve the dispersion of pollutants in an urban area. Other studies extended those models to predict pollutant concentrations and flow fields in actual urban areas with irregular arrays and noncubic buildings, such as in Macao, London, and Montreal (Gousseau et al. 2011; Liu et al. 2011; Xie and Castro 2009). However, all these researches have treated the pollutants as gas, not presenting the transportation of particulate matters within street canyons in the atmospheric environment.
Wind is the driving force of particle transport. Impelled by natural airflow, particles enter urban street canyons or disperse out of them. The question is whether wind speed has a positive or negative effect on pollutant removal. The answer is determined by the locations of pollutant sources and characteristics of the urban territory. Cheng et al. (Cheng, Liu, and Leung 2009) studied correlation between air and pollutant exchange for street canyons in combined wind/buoyancy-driven flow. Some investigators have examined the effect of wind on the dispersion of nucleation and accumulation-mode particles in urban street environments (Kumar, Fennell, and Britter 2008; Liu et al. 2011; Scungio et al. 2015; Zhang et al. 2011) and ultrafine particle concentrations in an incinerator plant (Scungio et al. 2015). Soulhac et al. (Soulhac, Perkins, and Salizzoni 2008) have developed an analytical model for the flow along a street canyon generated by an external wind blowing at any angle relative to the axis of the street. Considering the special case of a wind blowing parallel to the street, the flow within the street canyon depends only on the external wind and the distance to the closest solid boundary.
The influence of complex urban structures on pollutant dispersion has been simulated using computational fluid dynamics, based on the Navier-Stokes equations. Some researchers used large-eddy simulation (Li et al. 2010; Tominaga and Stathopoulos 2011, 2012; Xie and Castro 2009) and others a Reynolds time-averaged approach. However, in these simulations, air pollutants were modeled as gas dispersion, based on an Eulerian method (Nikolova et al. 2014; Wang and Max Zhang 2012; Wang et al. 2013), and particle dynamics were neglected. Particle deposition is the combined result of inertial impaction, turbulence-eddy impaction, interception, gravitational sedimentation, and so on. Therefore, more recently, a discrete-phase, particle trajectory model based on a Lagrangian reference frame (Saidi et al. 2014; Wang, Lin, and Chen 2011; Zhang and Chen 2009) has been used to predict particle transport concentration and suspension characteristics around complex buildings.
In developing China, some heavy industrial plants and stacks could not be moved out of the central city immediate, so that considerable numbers of individuals live in industrial areas and are exposed to industrial PM pollutants for prolonged periods. If industrial PM emissions exceed safety standards, long-term inhabitants in such areas are at high risk of adverse health (Johnson 2016; Scungio et al. 2016). Therefore, in order to research the transportation of fine particles from industrial stacks in the urban building groups, any other pollutant sources are not considered in the street canyons. Reflecting various stages of city development, four typical building arrangements were considered in this study: a low-rise building block (Figure 1a), step-up and step-down building layout (Figure 1b and c), and unified high-rise building block (Figure 1d). These layouts represent early developing cities and developed cities, respectively. For the first time, we explored and compared wind fields and fine particle dispersion within special building groups according to the stages of urban development, as a reference for planning regional layout when rebuilding older cities. We simulated the effects of wind speed on airflow and particle concentration distributions in these four urban scenarios as a reference for predicting the variation of pollutant distributions.
Published online:
24 May 2018Figure 1. Computational domains for particle transport within urban street canyons. (a) model a; (b) model b; (c) model c; (d) model d; (e) geometrical size. Label 1 is buildings; label I is a street canyon.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2018/uawm20.v068.i10/10962247.2018.1465487/20181001/images/medium/uawm_a_1465487_f0001_oc.jpg"})
Figure 1. Computational domains for particle transport within urban street canyons. (a) model a; (b) model b; (c) model c; (d) model d; (e) geometrical size. Label 1 is buildings; label I is a street canyon.
Methodology
Physical and computational domains of the numerical simulation
This study investigates airflow and particle transport properties of two-dimensional (2D) idealized street canyons, having various building heights. It is assumed that the length of the building is far longer than the width of the building, so the airflow character along the Z direction is not considered. The computational domain (Figure 1e) measures 600 (height) × 1720 (width) m and comprises four identical street canyons under a free surface layer. The widths of both the street canyon and buildings are b = 20 m, whereas building height is H0 = n × b (n = 1, 2, 3, 4, 5). A number of numerical experiments were carried out to test the effects of the domain length on the flow variables within the street canyon. The length extended behind the final building was set to Le = 1000 m to ensure fully developed turbulence at the outlet. Four typical street building layouts having various H/b ratios were used in this study: (a) a low-rise building block (H0/b = 1); (b) a step-up building layout (H0/b = 1, 2, 3, 4); (c) a step-down building layout (H0/b = 4, 3, 2, 1); and (d) a high-rise building block (H0/b = 5). These geometric models represent an undeveloped city with ranch-style houses, two developing cities with varying building heights, and a developed city with a cluster of high-rise buildings, respectively.
Computational model
Turbulence models provide the foundation of our numerical simulations. In this study, the incompressible Reynolds time-averaged Navier-Stokes (RANS) renormalization group (RNG) k-ε turbulence model (Yakhot and Orszag 1986) was used, because Cheng, Liu, and Leung (2008, 2009,) validated its applicability to evaluating street canyon flows. The RNG k-ε turbulence model is for a high Reynolds number (Re = u0H/γ = 1.12 × 107) dynamic flow in the computational domain; thus, a scalable wall function is used for areas near the ground and building walls.
For microparticles, a Lagrangian reference frame can be used to analyze particle trajectories within the atmosphere, related to effects of diverse forces, such as Brownian, Saffman, drag, and gravity forces. Because of the high particle-to-air density ratio, dilute particle suspensions, negligible Brownian motion, thermophoretic forces, drag, and gravity are considered the dominant forces away from the model walls (Kleinstreuer, Zhang, and Li 2008). The dynamic equations of particle transport can be expressed, as follows: 
(1)
(2)
(3)
(4)
where 
is particle velocity,
is the fluid density,
is the particle density,
is the particle diameter, S is the density ratio between the particle and adjacent fluid,
is the kinematic viscosity,
is the unit delta function,
is the particle relaxation time, and
is the deformation rate tensor. When the flow domain is relatively complicated, at each fluid flow time step, this code uses flow properties computed by the solver.
Boundary conditions
Velocity boundary conditions
The profile of the airstream velocity is prescribed as the power law at the inflow boundary to drive the prevalent flow in the atmosphere. Here, u0 (= 1, 2, 4 m/sec) is the prevalent wind speed, α (= 0.28) is the wind exponent, zf (= h) is the height, and hf (= 10 m) is the standard reference height. An intermediate turbulence intensity value (I = 0.05) is imposed at the inlet. A pressure outlet boundary is applied at the downstream outflow. The height of the computational domain is sufficient to develop a free-stream wind layer. The top plane of the computational domain has a free-slip boundary condition. No slip boundary conditions are applied to all of the street canyon facets. 
(5)
Pollutant boundary conditions
We discuss that PM from factory chimneys transport within the building groups; any other pollutant sources (e.g., vehicle emissions) are not considered in the street canyons. According to PM emission data of the Qingshan industrial zone in Wuhan, China, a source with a constant particle emission rate (E = 0.002 kg/sec) and particle mean size of 2.5 × 10−6 m is distributed equally with grid nodes at the inlet of the domain. All the street canyon facets are set as trap boundaries.
Mesh independence study and numerical analysis
This study generated a hex mesh by using commercial ICEM software (Ansys Inc, PA, USA). The number of mesh elements was set as 118126 for discretizing the computational domain. The mesh sensitivity analysis is presented in Appendix A. The SIMPLE scheme was adopted to couple the velocity and pressure governing equations, which were solved by the commercial computational fluid dynamics (CFD) code ANASYS Fluent (Ansys Inc. 2016) .
Model validation
Full-scale simulations were performed based on the modeling sensitivity tests performed for the idealized street canyon case, such as grid and domain sizes, convergence criteria, and so on. The wind field of the RANS RNG k-ε turbulence model used in the current study was validated using water channel experiments (Britter and Hanna 2003) and simulations (Cheng, Liu, and Leung 2008). The detailed validation is explained in Appendix B.
Results and discussion
Airflow over building groups
The inlet boundary of the computational domain is set as a wind speed profile, with reference velocity u0 = 1 m/sec. Because the dispersion and transport of particles within the urban environment are mainly influenced by the airflow vortex structure and mass exchange between the street canyon and its overlying air space, the stream lines are analyzed within this area. Figure 2 shows air streamlines for all four urban layouts, composed of various arrangements of buildings. Given our research focus, these figures are restricted to show the region 300 < X < 1200 m and 0 < Y < 200 m, which is denoted by dash-dot lines in Figure 1e.
Published online:
24 May 2018Figure 2. Airflow stream lines for all four building layouts for the region 300 < X < 1200 m, 0 < Y < 200 m. (a) model a; (b) model b; (c) model c; (d) model d.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2018/uawm20.v068.i10/10962247.2018.1465487/20181001/images/medium/uawm_a_1465487_f0002_b.gif"})
Figure 2. Airflow stream lines for all four building layouts for the region 300 < X < 1200 m, 0 < Y < 200 m. (a) model a; (b) model b; (c) model c; (d) model d.
Judging from the flow pattern, when the street canyon aspect ratio is H0/b = 1, the entrainment of air within the street canyon forms an obvious vortex, which is skim flow (Oke 1988) as shown in Figure 2a for model a. In model b, with buildings of increasing heights, the external airflow accelerates, and the entrainment effect on the air within the canyon is enhanced. However, a vortex does not appear to reach the bottom of the model, until the aspect ratio of the fourth canyon is 4. This indicates that the flow close to the bottom of model is weak.
In model c, the barrier of the front high-rise building causes the flow velocity above the canyons and the entrainment of air within the canyons to decrease. This causes two countervortices to form within the first two canyons, having aspect ratios of greater than 3. The upper vortex formed by the entrainment effect of incoming flow is so intense that it drives the air into the bottom of the model, forming a comparative weak vortex.
In model d, with an aspect ratio equal to 5, the upper airflow follows the wind stream, whereas airflow in the bottom of the canyon is extremely weak, having a velocity of almost zero. In addition, in models c and d, the blocks of high-rise buildings on the windward side create a stagnation corner at a height of less than H0/2 in front of the first building, where fluids cannot exit. Particles are accumulated by the vortex in this stagnation corner, producing a high-particle-concentration zone.
In these 2D idealized street canyon models, there are three main mechanisms of pollutant transport within the street canyons: (1) dispersal of pollutants across street canyons by circulating flow at street height; (2) trapping of pollutants within recirculation regions; and (3) exit of pollutants from the canyon top. As shown in Figure 2d, the mechanism (1) has an inhibiting effect on particle dispersion into the street canyon. In Figure 2b and c, at the upper half of the canyons, the mechanism (3) helps particles disperse out of the canyons, and at the bottom half of canyons, the mechanism (2) has inhibiting effect on PM migration.
Figure 3 illustrates turbulence intensities within the region 300 < X < 1200 m. Turbulence intensity is the ratio of the turbulent fluctuation velocity to mean velocity, which reflects the degree of turbulence. In model a, the turbulence intensity reaches a maximum value behind the building group. However, in models c and d, the maximum turbulence intensity occurs above the roof of the buildings. From the turbulence distributions within canyons of models a and b, the turbulence intensity on the windward side is stronger than on the leeward side. This illustrates that turbulence propagates from the windward side into the canyons. In street canyons of model c, turbulence intensity is less than 0.08, with no difference between windward and leeward sides. All of the building layouts show that turbulence within this region is weak and that particles do not follow the airflow from the roof region into the canyons. Furthermore, an air speed decrease at the roofs of the buildings causes flow in the canyons and above the buildings to separate, producing a strong vortex with opposite airflow into the canyons. This inhibits particle entrance into the canyons.
Published online:
24 May 2018Figure 3. Contours of turbulence intensity for all four building layouts within the region 300 < X < 1200 m, 0 < Y < 200 m. (a) model a; (b) model b; (c) model c; (d) model d.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2018/uawm20.v068.i10/10962247.2018.1465487/20181001/images/medium/uawm_a_1465487_f0003_oc.jpg"})
Figure 3. Contours of turbulence intensity for all four building layouts within the region 300 < X < 1200 m, 0 < Y < 200 m. (a) model a; (b) model b; (c) model c; (d) model d.
Particulate matter transport within building groups
Figure 4 shows the dimensionless concentration distribution of atmospheric PM. The dimensionless concentration is defined as in eq 6. 
(6)
Published online:
24 May 2018Figure 4. Contours of particulate matter dimensionless concentration for all four building layouts within the region 300 < X < 1200 m, 0 < Y < 200 m. (a) model a; (b) model b; (c) model c; (d) model d.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2018/uawm20.v068.i10/10962247.2018.1465487/20181001/images/medium/uawm_a_1465487_f0004_oc.jpg"})
Figure 4. Contours of particulate matter dimensionless concentration for all four building layouts within the region 300 < X < 1200 m, 0 < Y < 200 m. (a) model a; (b) model b; (c) model c; (d) model d.
where Q0 is the inlet concentration of PM from industrial emissions
, A is the area of the computational domain inlet (m2), and c is its absolute concentration value (kg/m3).
Because the ratio of particles to fluid is low, and the particle flow is passively driven by air, the particle transport trajectories and concentration distributions are determined from flow fields. When the location of particle injection is fixed at a constant height, the particle diameter is small enough to ensure that they are suspended within the atmosphere; in this case, they do not disperse upwards or deposit downwards. This behavior is confirmed by the notable high-concentration zone formed above the roof of the buildings at about (0.1–0.2)H0, as shown in Figure 4a and b.
Particles disperse across the canyon in circulating flow. Their concentrations appear as a discrete distribution, with dimensionless concentrations in the street canyons of less than 0.2. On the windward side of building groups, at a height of less than H0/2, there are contaminated zones. For example, Figure 4c and d show that local concentrations on the windward sides of model c and model d are 2 times higher than those of the emission source so that the heavy air pollutant may appear on the windward side of high buildings.
The suspension fraction can be used to compare the relative number of particles at different locations within the model. The particle suspension fraction f is defined as the ratio of the number of suspended particles in a region to the number of particles entering this region. Figure 5 shows the particle suspension fractions for the space above the buildings and the street canyons within the region 540 < X < 720 m, where the building groups are located. The PM concentration within street canyons of low-rise building groups (model a) is higher than that of high-rise building groups (model d). The step-up building layout (model b) has the largest suspended particle fraction within the street canyons. In contrast, the step-down building layout (model c) has a particle suspension fraction of almost zero. The reason is that in model b, the increasing building heights accelerated stream flow, which entrained and trapped more particles into the street canyons. Gu (2011) and Zhang et al. (2011) also confirmed that the concentrations of simulated pollutants in uneven canyons are lower than those in uniform ones. But this is dependent on pollution source location and stream flow direction. Clearly, increasing wind speed does not significantly change the suspended particulate fraction in a given area; this suggests that the fraction is determined by the height and layout of the buildings.
Published online:
24 May 2018Figure 5. Histograms showing PM suspension fractions in different areas within the region 540 < X < 720 m, 0 < Y < 200 m. (a) model a; (b) model b; (c) model c; (d) model d.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2018/uawm20.v068.i10/10962247.2018.1465487/20181001/images/medium/uawm_a_1465487_f0005_b.gif"})
Figure 5. Histograms showing PM suspension fractions in different areas within the region 540 < X < 720 m, 0 < Y < 200 m. (a) model a; (b) model b; (c) model c; (d) model d.
Given the above, in planning the layout of buildings in the industrial area, according to the location of industrial emissions, the step-down arrangement is recommended instead of the step-up one based on the PM distribution. Otherwise, front buildings on the windward side should be constructed preferentially in rebuilding old cities, since these buildings can obstruct PM migrating into street canyons.
Effects of the wind speed on the particle transport
To study the effects of wind speed on particle transport, we used reference velocities of 1, 2, and 4 m/sec. Under different wind speeds, vertical velocity, turbulence intensity, and the particle dimensionless concentration at the windward and leeward walls vary, as shown along the Y direction for the four building layouts in Figures 6–8, respectively. The windward wall is located at a position 0.15b in front of the first building, whereas the leeward wall is 0.15b behind the last building.
Published online:
24 May 2018Figure 6. Profiles of vertical velocity at the windward and leeward walls of the four building groups. (a) model a; (b) model b; (c) model c; (d) model d.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2018/uawm20.v068.i10/10962247.2018.1465487/20181001/images/medium/uawm_a_1465487_f0006_b.gif"})
Figure 6. Profiles of vertical velocity at the windward and leeward walls of the four building groups. (a) model a; (b) model b; (c) model c; (d) model d.
Published online:
24 May 2018Figure 7. Profiles of turbulence intensity at the windward and leeward walls of the four building groups. (a) model a; (b) model b; (c) model c; (d) model d.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2018/uawm20.v068.i10/10962247.2018.1465487/20181001/images/medium/uawm_a_1465487_f0007_b.gif"})
Figure 7. Profiles of turbulence intensity at the windward and leeward walls of the four building groups. (a) model a; (b) model b; (c) model c; (d) model d.
Published online:
24 May 2018Figure 8. Profiles of PM dimensionless concentrations at the windward and leeward walls of the building groups. (a) model a; (b) model b; (c) model c; (d) model d.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2018/uawm20.v068.i10/10962247.2018.1465487/20181001/images/medium/uawm_a_1465487_f0008_b.gif"})
Figure 8. Profiles of PM dimensionless concentrations at the windward and leeward walls of the building groups. (a) model a; (b) model b; (c) model c; (d) model d.
Profiles of vertical velocity along the Y direction at the windward and leeward walls of the four building layouts in Figure 6 show that there are different trends in vertical velocity with height of y ≤ 1 and y > 1. When y < 1, the velocity along the leeward side increases with height, reaching a maximum value of y = 1. However, when y > 1, the vertical velocity decreases with the height. In model d, when y < 0.5, the leeward wall velocity is close to zero, whereas the windward wall velocity is negative. This suggests that the buildings hinder air transmission, causing an eddy to form in front of the buildings. All variation laws are identical with wind speed increasing.
Profiles of turbulence intensity along the Y direction at both the windward and leeward walls for all four building layouts show that turbulence intensity increases as wind speed increases (Figure 7). The variation in turbulence intensity along the Y direction has three distinct trends. When y ≤ 1, the leeward wall turbulence decays towards the ground and the windward wall turbulence reaches a maximum. However, when 1 < y < 1.4, turbulence intensity increases as air transmission increases, whereas the leeward wall turbulence reaches a maximum. Finally, when y > 1.4, the airflow is not disturbed by the building group and turbulence reduces to zero.
Variation trends in particle concentrations at the windward and leeward walls for the four street models show that the relative relationship between windward and leeward wall particle concentrations in the same model is invariant to wind speed (Figure 8), since building layouts determine the flow structure in the computational domain. In model a, the region with y ≤ 2 defines the high-particle-concentration zone, with a maximum concentration at y = 1, coincident with the roof of the buildings. In models b, c, and d, the variation in particle concentrations along the Y direction also has two distinct trends. When y ≤ 1, the windward particle concentration oscillates and is larger than that of the leeward side. When y > 1, the windward particle concentration is less than the leeward one. The windward particle concentration reaches a maximum at roof height of the first building. Meanwhile, at y = 1.1~1.4, the leeward particle concentration reaches a maximum, which is less than that of the windward one.
It is also shown that the smaller the wind speed, the higher the amplitude in the particle concentration (Figure 8). This is because particle dispersion cannot be driven when the wind speed is low, causing particles to be suspended in the atmosphere, forming a discrete concentration distribution. As wind speed increases, particle transport is enhanced and concentrations decrease in all cases. In model c, when u0 = 4 m/sec, the particle concentration on the leeward side is reduced to zero at y ≤ 1. In model d, even when the wind speed is 1 m/sec, the leeward concentration is close to zero. These results indicate that high-rise buildings impede particle transport, regardless of wind speed.
Conclusion
Using gas-solid, two-phase flow theory and a Lagrangian approach, we simulated airflow and PM dispersion in two-dimensional (2D) idealized building group models on the microscale. Profiles of airflow and turbulence intensity were used to examine the effects of various building heights and layouts on airflow. Particle suspension fractions and concentration distributions were used to evaluate the effects of building layouts and wind speeds on fine particle transport, yielding the following conclusions.
The heights and layouts of urban buildings affect local airflow and PM dispersion in the atmosphere. The step-up building layouts accelerate top airflow and entrain more particles into street canyons, with associated adverse effects on residents’ health. The step-down building layout and high-rise building groups obstruct atmospheric transport so as to decrease the PM concentration in the residential area. During the renewal of old industrial cities, according to the location of the industrial emissions, the step-down building arrangement should be constructed preferentially.
The wind condition enhances particle transport, so the particle concentrations at both windward and leeward walls of the building groups reduce with wind speed increasing. Below building height, the windward particle concentration oscillates, but its average value is greater than the leeward one. Above roof height, windward particle concentration is less than the leeward one. Therefore, the particle concentration variation with heights at the windward side of buildings will be researched accurately in the future, which will become a guide for designing building heights in industrial areas.
The 2D idealized street canyon models have some limitations. For example, buildings and streets are assumed to have regular profiles and the natural wind at the inlet is considered to be perpendicular to the street centerline. However, realistic urban areas are three-dimensional (3D) and the building heights are irregular. The presence of street crossings enhances local turbulence intensities and produces momentum exchange between neighboring streets. When the approaching wind is not perpendicular to the street centerline, a helical flow is formed along the street. Therefore, deriving the relevant results by using irregular and realistic urban models is necessary in future studies.
| No. of elements | Airflow velocity (m/sec) | PM concentration (kg/m3) |
|---|---|---|
| 44,240 | 0.3303 | 4.039 × 10–5 |
| 77,980 | 2.1158 | 5.782 × 10−6 |
| 115,692 | 2.0046 | 5.109 × 10−6 |
| 118,126 | 2.0046 | 4.977 × 10−6 |
| 170,251 | 2.0020 | 4.944 × 10−6 |
Appendix A: Mesh independence study
This study generated a hex mesh by using commercial ICEM software (Ansys Inc. 2016). Prior to performing the CFD simulations, a grid independence study was conducted over five grid resolutions. In this study, with a fixed grid expansion ratio of 1.1, five types of first-layer grid thickness within a range of 0.005–0.5 m were selected. The grid discretization followed the Architectural Institute of Japan (AIJ) guidelines (Tominaga et al. 2008). Thus, five mesh types, from coarse to dense, were generated to ensure that the simulation results were sufficiently grid independent. As shown in Table A1, when the number of mesh elements was higher than 118,126, the airflow velocity and PM concentration C at the reference location became stable and remained nearly unchanged. Considering computational efficiency, the mesh of the first-layer grid thickness ∆x = ∆y = 0.05 in the canyons and total 118,126 mesh elements were used for further simulation.
Appendix B: Model validation
Full-scale simulations were performed based on the modeling sensitivity tests performed for the idealized street canyon case, such as grid and domain sizes, convergence criteria, and so on. The wind field of the RANS RNG k-ε turbulence model used in the current study was validated using water channel experiments (Baik and Kim 2002) and simulations (Cheng, Liu, and Leung 2008). For the water channel experiment, the street canyon was constructed using two building models measuring 0.1 m (length) × 0.1 m (width) × 0.4 m (height), which were positioned 0.1 m apart to form an isolated street canyon with h/b = 1. According to the model validation procedure of Baik and Kim (2002), the measured and simulated vertical velocity profiles on the upwind (0.15b measured from the leeward wall) and downwind (0.15b measured from the windward wall) sides inside a street canyon were compared (Figure 9a and b). Figure 9 also shows the numerical results of Cheng (2008) and the present study. The variation trends of the vertical wind profiles on the upwind and downwind sides of the four results were almost identical. Therefore, in the current study, the CFD results were considerably close to those of the water channel experiment.
Published online:
24 May 2018Figure 9. Mean vertical velocities at the upwind and downwind positions inside the center street canyon. (a) upwind; (b) downwind.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2018/uawm20.v068.i10/10962247.2018.1465487/20181001/images/medium/uawm_a_1465487_f0009_b.gif"})
Figure 9. Mean vertical velocities at the upwind and downwind positions inside the center street canyon. (a) upwind; (b) downwind.
Table A1. Mesh independence study.