ASSESSMENT OF AIR CHANGE RATE AND CONTRIBUTION RATIO IN IDEALIZED URBAN CANOPY LAYERS BY TRACER GAS SIMULATIONS

Topic B4: Ventilation ASSESSMENT OF AIR CHANGE RATE AND CONTRIBUTION RATIO IN IDEALIZED URBAN CANOPY LAYERS BY TRACER GAS SIMULATIONS Qun WANG 1, Mats SANDBERG 2, Jian HANG 1* 1 Department of Atmospheric Sciences, School of Environmental Science and Engineering of Sun Yat-Sen University, Guangzhou, P. R. China 2 Laboratory of Ventilation and Air Quality, University of Gävle, SE-80176 Gävle, Sweden *Corresponding e-mail: hangj3@mail.sysu.edu.cn Keywords: Air change rate per hour (ACH), Concentration decay method, Urban canopy layer (UCL), Contribution ratio, Computational fluid dynamics (CFD) simulations SUMMARY This paper first performed CFD simulations with the concentration decay method to investigate air change rate per hour in entire urban canopy layers (UCL) under neutral atmospheric condition (building packing density p =0.25, frontal area density f =0.25, aspect ratio H/W=1). Two approaches are included. One is the normal way (ACH 1 ), the other (ACH 2 ) fixes concentration out of the UCL volume to be zero in CFD, i.e. no tracer gas reentry. ACH 2 with a quicker concentration decay always exceeds ACH 1. With a parallel approaching wind (θ=0 o ), the larger urban size (Lx=390m to 2940m), the lower ACH is. For 5-row and 5-column UCL model (Lx=Ly=270m), θ=0 o attains greater ACH than θ=15 o, 30 o, 45 o. Both ACH 1 and ACH 2 are less than the sum of ACH induced by mean flows and turbulence across UCL boundaries. Then the contribution ratio was numerically predicted to quantify the transfer of pollutants between different urban spaces of a UCL. INTRODUCTION Besides pollutant source control, urban canopy layer (UCL) ventilation by rural/ marine/mountain wind can help reducing urban air pollution and urban heat island intensity. Tracer gas techniques have been widely used to measure ventilation indices in indoor environments(chen, 2009) to evaluate how external air enters a room and ventilates it. How to quantify UCL ventilation? In recent years, researchers have started to apply similar tracer gas simulations to estimate UCL ventilation, including volumetric flow rate and air change rate per hour (Liu et al., 2005; Hang and Li, 2010), pollutant retention time and purging flow rate (Bady et al., 2008), age of air and air exchange efficiency (Hang et al., 2009) etc. The application of these

ventilation indices assumes that the surrounding air is relatively clean and cool, which can be transported into UCL to help pollutant dilution and heat removal. Air change rate per hour (ACH) is one of the key ventilation indexes, which represents the volumetric air exchange rate of the entire UCL volume per hour. It was initially used for 2D street canyon ventilation (Liu et al.,2005), then was extended for 3D UCL ventilation (Hang and Li, 2010). These literatures confirmed that both mean flows and turbulence across UCL boundaries contribute to UCL ventilation, and ACH index induced by them are usually calculated separately. It is still unclear that, how to calculate ACH index produced by the net effect of both mean flows and turbulence. The concentration decay method has been widely used to measure air change rate in rooms(etheridge and Sandberg, 1996; Chen, 2009). Thus this paper aims to introduce this tracer gas technique into CFD simulations and numerically predict air change rates in various UCL models. Another question is how to quantify the pollutant transport and correlation of air pollution between different urban spaces within a UCL. Thus the contribution ratio between different urban spaces is analyzed by using CFD simulations, in which a uniform pollutant source is defined within a given urban space and the mass fraction of pollutants entering other spaces was analyzed. METHODOLOGIES We define the urban canopy layer (UCL) as the air volume surrounded by street openings and street roofs. In CFD simulations, full-scale UCL models are investigated in which building width (B), building height (H) and street width (W) are the same constant of 30m (see Fig. 1). Key urban parameters include building area density p =0.25, frontal area density f =0.25, street aspect ratio H/W=1. Two groups of UCL models are investigated. As summarized in Table 1, all test cases are named as Case "[row number-column number, wind direction θ]". The row and column numbers are referred to the numbers of the main and secondary streets. For Group I (see Table 1 and Fig. 1a), the approaching wind (θ=0 o ) is parallel to the main streets with a length of Lx and the column number is defined as "N", which denotes that the lateral urban size (Ly) is sufficiently large to disregard the influence of lateral boundaries of UCL models. For Group II (see Table 1 and Fig. 1b), the lateral UCL boundaries are taken into account under a neutral atmospheric condition with parallel (0 ) or non-parallel (θ=15,30,45 ) approaching winds. The CFD code FLUENT 6.3 was used to solve the steady-state isothermal flow field. The standard k-ε model was used to simulate the turbulence effects. All transport equations were discretized by the second order upwind scheme. The hexahedral cells of 1.0 to 2.9 million are produced, and the minimum of grid size is 0.5m near street ground and building surfaces with an expansion ratio of 1.15 toward the surrounding. The SIMPLE scheme was used for the pressure and velocity coupling. To quantify UCL ventilation performance, the concentration decay method is applied to calculate volumetric air exchange rate per hour (ACH) in the entire UCL, in which a uniform

Y(m) y (m) UCL initial pollutant concentration is defined at time t=0s (C(0)=10-5 kg/kg or 10 ppm, see Fig. 1), then its decay history (C(t)/C(0)) is analyzed. The time step is 1s. The decay duration of all test cases is nearly 400s. Two CFD setups are defined in the decay method. One is the normal way (ACH 1 ) to solve concentration field without the above assumption, thus tracer gas concentration is not zero outside UCL volume and tracer gas reentry into UCL volume occurs; The other (ACH 2 ) fixes tracer gas concentration out of the UCL volume to be always zero in CFD, assuming that tracer gas leaving UCL volume never come back (i.e. no tracer gas reentry). To quantify the correlation of pollutant transport between different urban units within a UCL(see Fig. 1c), a uniform tracer gas source with an emission rate of 10-5 kg m -3 s -1 is defined in only one unit (for example, unit B11), then the mass fraction of tracer gas in each unit is calculated as contribution ratio. ( ) 3.0( z U ) 0 z H 0.16 y z x H B W=B B H=W=B=30 m in full-scale CFD simulation of Case [7-N, 0] W=B Pollutant concentration within entire urban canopy layer is C(0)=10-7 kg/kg at initial state (t=0 s). Concentration decay method is used to calculate ACH 300 Stream-wise urban size Lx = 270 m 250 θ o (c) Figure 1. CFD model and definition of initial state in concentration decay method Table 1 Test cases investigated ( p = f =0.25, H/W=1) Case name (Group I) Number of rows /total length Lx Number of columns /total length in Ly [7-N,0] 7 row, Lx=390m N or Ly is sufficiently large [14-N, 0] 14 row, Lx=810m to neglect the lateral [21-N, 0] 21 row, Lx=1230m boundary effect as the [28-N,0] [35-N,0] 28 row, Lx=1650m 35 row, Lx=2070m approaching wind is parallel to main streets [42-N,0] [49-N,0] 42 row, Lx=2490m 49 row, Lx=2910m 0 30 60 90 120 150 180 210 240 270 300 X(m) x(m) Uniform source B11 (emission rate 10-5 kg/m 3 -s) [5-5,0] [5-5,15] 5 rows 5 columns 15 o [5-5,30] Lx=270m Ly=270m 30 o [5-5,45] 45 o θ o 200 150 100 50 0 B14 B24 B34 B13 B12 B11 B23 B22 B21 B33 B32 B31 B44 B43 B42 B41 Wind direction 0 o 0 o Lateral urban size Ly = 270 m

z / H z / H RESULTS AND DISCUSSION Validation of CFD Simulations using Wind Tunnel Data Brown et al. (2001) performed a detailed wind-tunnel experiment of 7-row and 11-column cube array with a parallel approaching wind (see Fig. 2a, B=H=W=15cm, Lx=13H, Ly=21H). The scale ratio to full-scale model is 1:200. Point Vi represents the centre point of the secondary street No i for the middle column (see Fig. 2a). Vertical profiles of mean stream-wise velocity uz, ( ) mean vertical velocity wz () at these points were measured. Hang and Li (2010) confirmed that the lateral urban boundaries hardly affect airflows in the middle column (see Fig. 2a) since the lateral size is sufficiently large (Ly=21H). Using similar technique, this paper uses these wind tunnel data to estimate the accuracy of CFD simulations. All CFD arrangements on domain sizes, grid sizes and boundary conditions fulfill the major requirements as recommended by Tominaga et al. (2008). Fig.2b shows uz ( ) and wz () at V1 as two examples. Obviously standard k model predicts mean flows better than RNG k model. Thus standard k model is selected for the following CFD simulations. 3.0 2.5 2.0 Vertical profiles at Point V1(x/H=1.5) Wind tunnel data CFD results using Standard k- model RNG k- model y W H=B=15cm B W=B 1.5 1.0 uz ( ) V1 V2 V3 V4 V5 V6 0.5 Wind V1 x Wind x/h 0.0 x/h=1.5-1 0 1 2 3 4 5 Stream-wise velocity (m/s) 3.0 Vertical profiles at Point V1(x/H=1.5) Wind tunnel data 2.5 CFD results using Standard k- model RNG k- model 2.0 1.5 Wind V0 V1 V2 V3 V4 V5 V6 1.0 0.5 wz () x/h=-0.5 x/h=1.5 x/h=3.5 x/h=5.5 x/h=7.5 x/h=9.5 x/h=11.5 x 0.0-1.5-1.0-0.5 0.0 0.5 1.0 Vertical velocity (m/s) Figure 2. Wind tunnel model description and vertical profiles of CFD results and experimental data at Point V1. Characteristics of flow pattern and concentration decay process The decaying concentration is associated with turbulent airflow pattern. To verify this correlation, Fig. 3 shows 3D streamline and concentration field at time of 38s and 138s in Case [5-5, 0]. The flow is channeled toward downstream regions, and 3D

vortices and helical flows exist behind each buildings. Across the lateral boundary of cube array (at y=135m), there are lateral helical airflows leaving or re-entering across lateral UCL boundary. Thus tracer gas concentration decreases more quickly in upstream regions than downstream regions, and that near lateral boundary than near urban centre regions. It shows that the downstream and urban centre regions experience a worse ventilation performance. Two concentration decay methods are compared, including the normal way (ACH 1, see Fig. 3a-3b) with tracer gas reentry, and the second method (ACH 2, see Fig. 3c-3d) assuming that tracer gas leaving UCL volume never comes back. Obviously the second one (see Fig. 3c-3d) attains a quicker concentration decay process than the first (see Fig. 3a-3b). (c) (d) (e) Figure 3. Concentration field at time of 38s,138s in Case [5-5, 0]: (a-b) the method for ACH 1, (c-d) the method for ACH 2 without tracer gas reentry, (e) 3D streamline.

Air change rate per hour (ACH) Analysis of air change rates by various methods Air change rate per hour (ACH) is used to quantify the volumetric air exchange rate of the entire UCL volume per hour. ACH was initially used for 2D street canyon ventilation (Liu et al.,2005), then was extended for 3D UCL ventilation (Hang and Li, 2010). Both mean flows and turbulence contribute to UCL ventilation. In Eq. (1), ACH calculated by total mean flow rates (Q T ), by effective flow rated induced by turbulence fluctuations across street roofs (Q roof (turb)) are defined as below. ACH T =3600Q T /Vol ACH turb =3600Q roof (turb)/vol (1a) (1b) where Vol is the entire UCL volume. Fig. 4a shows air change rates calculated by flow rates across UCL boundaries (ACH T, ACH turb ), and by the concentration decay methods (ACH 1, ACH 2 ) in Case [5-5, θ] (θ=0,15,30,45 ). Here we pay more concern about the difference with and without tracer gas reentry. Note that, for ACH 2, the tracer gas concentration out of UCL volume is set as 0 ppm without tracer gas reentry. ACH 2 is found a little greater than ACH 1 with tracer gas re-entry. To investigated the effect of urban size under a parallel approaching wind, Fig. 4b shows air change rates in test cases of Case [i-n,0] (i=7,14,21,28,35,49, urban size Lx=390 to 2910m) with a parallel wind. Obviously as urban size rises, ACH T, ACH 1, ACH 2 all decrease, but ACH turb change little. ACH 2 still exceeds ACH 1 with tracer gas re-entry. Finally both Fig. 4a and 4b confirm that, ACH 1 and ACH 2 are always less than the sum of ACH T and ACH turb. It verifies that there is an issue of ventilation efficiency by mean flows and turbulence. Air change rate per hour (ACH) 60 55 50 45 40 35 30 25 20 15 10 5 0 In Case [5-5, o ] 'air change rate per hour (ACH)' Calculated by flow rates across UCL boundaries ACH T ACH roof (turb) Calculated by concentration decay method ACH 1 (with tracer gas reentry) ACH 2 θ o Urban size in x direction Lx (N rows) Ly (M columns) 0 15 30 45 Wind direction of o ACH turb ACH T ACH 2 ACH 1 with re-entry 30 25 20 15 10 Total street length Lx ACH roof (turb) ACH2 ACH1 5 In Case [i-n,0] (i=7,14,21,28,35,49) ACH T ACH roof (turb) ACH T ACH 0 1 with tracer gas re-entry ACH 2 0 420 840 1260 1680 2100 2520 2940 3360 Total street length or urban size Lx (m) Figure 4. Air change rate in test cases of Case [5-5, θ], Case [i-n,0]. Analysis of Contribution Ratio in the Example Case As an example, Fig. 5a shows 3D streamline and concentration field in z=1.5m in Case [5-5, 15] as tracer gas source unit is B11. 3D vortices and helical flow also exist.

Thus turbulent airflows transport pollutant to the neighbor units and toward downstream. The concentration in source B11 unit and its neighbor units are much higher than that far downstream. Then Fig. 5b-5g display contribution ratio in Case [5-5, 15] and Case [5-5, 45] with source units of B11, B44. With the upstream source unit of B11(see Fig. 5b-5c), tracer gas concentration in B11 unit itself is relatively low (ratio of 26.97% and 23.71%) due to better ventilation in B11 unit, but tracer gas disperses to many other units with considerable contribution ratios. With a downstream source unit (B44), the contribution ratio is small in its upstream unit, and tracer gas is dispersed to a limited number of urban units, however contribution ratio in the source unit itself is large (79.84%-82.24%) because the downstream unit has a worse ventilation performance locally. Tracer gas concentration (ppm) in z=1.5 m in Case [5-5, 15] 15 Uniform source in B11 15 0.77%% 1.53%% 45 0.13% 1.20% 5.16% 6.46% 1.02%% 3.59%% 5.36%% 1.24% 6.39% 8.82% 4.69% 3.23% 8.13%% 7.9%% 7.89%% 9.13% 13.48% 6.01% 1.03% 26.97% 16.43%% 8.89%% 5.93%% 23.71% 9.04% 1.16%% 0.12% Source B11 (c) 0.19% 79.84% 0.38% 82.24% 0.74% 0.36% 15 45 (d) (e) Figure 5. 3D streamline and concentration in z=1.5m in Case [5-5, 15], (b-e) Contribution ratio with source units of B11, B44 in Case [5-5, 15] and Case [5-5, 45]. CONCLUSIONS This paper applied some tracer gas techniques into CFD simulations to predict air

change rates and contribution ratio in urban spaces. Results show that the approaches are effective tools to evaluate the effect of urban morphologies on overall ventilation capacity and to quantify the pollutant flux between different urban spaces within the UCL. The normal concentration decay method (ACH 1 ) attains smaller air change rates than the other (ACH 2 ) without tracer gas reentry which fixes concentration out of UCL volume to be zero. Moreover ACH calculated by concentration decay methods are always less than the sum of ACH calculated by flow rates induced by mean flows and turbulence across UCL boundaries. It verifies that further investigations are still required to analyze ventilation efficiency by mean flows and turbulence. With a parallel approaching wind (θ=0 o ), larger urban sizes experience smaller ACH (Lx= 390m to 2940m). For the 5-row and 5-column UCL model (Lx=Ly=270m), the parallel wind obtains greater ACH than non-parallel winds (θ=15 o, 30 o, 45 o ). Finally the contribution ratio approach is effective to quantify the transfer of pollutants between different urban spaces of a UCL, and it is worth being applied if urban planners select an urban site to place heavy pollutant sources. ACKNOWLEDGEMENT This work was funded by the National Natural Science Foundation of China (Grant No. 51108102). REFERENCES Bady M, Kato S, Huang H (2008) Towards the application of indoor ventilation efficiency indices to evaluate the air quality of urban areas. Building and Environment, 43,1991-2004. Brown MJ, Lawson RE, Decroix DS et al (2001) Comparison of centerline velocity measurements obtained around 2D and 3D buildings arrays in a wind tunnel, Report LA-UR-01-4138, Los Alamos National Laboratory, Los Alamos,pp.7. Chen Q (2009) Ventilation performance prediction for buildings: a method overview and recent applications. Building and Environment, 44, 848-858. Etheridge D and Sandberg M (1996) Building Ventilation Theory and Measurement. John Wiley & Sons, Chichester, UK. Hang J, Sandberg M, Li Y (2009) Age of air and air exchange efficiency in idealized city models. Building and Environment, 44(8),1714-1723. Hang J, Li Y (2010) Wind conditions in idealized building clusters macroscopic simulations using a porous turbulence model. Boundary-Layer Meteorology, 136(1),129-159. Liu CH, Leung DYC, Barth MC (2005) On the prediction of air and pollutant exchange rates in street canyons of different aspect ratio using large-eddy simulation. Atmospheric Environment, 39, 1567-1574. Tominaga Y, Mochida A, Yoshie R, et al (2008) AIJ guidelines for practical applications of CFD to pedestrian wind environment around buildings. Journal of Wind Engineering and Industrial Aerodynamics, 96, 1749-1761.