- Email: [email protected]
Effect of ventilation strategies on particle decay rates indoors: An experimental and modelling study
ARTICLE IN PRESS
Atmospheric Environment 39 (2005) 4885–4892 www.elsevier.com/locate/atmosenv
Effect of ventilation strategies on particle decay rates indoors: An experimental and modelling study Je´roˆme Bouilly, Karim Limam, Claudine Be´ghein, Francis Allard LEPTAB, University of La Rochelle, Av. M. Cre´peau, 17042 La Rochelle Cedex 01, France Received 22 August 2004; received in revised form 5 January 2005; accepted 30 April 2005
Abstract A cubic experimental chamber of 2.5 m 2.5 m 2.5 m was designed to study the impact of ventilation strategies on the indoor particle concentration. Particles of 0.3–15 mm aerodynamic diameter were used. The combined effects of the ventilation rate (0.5 and 1.0 ach) and the inlet and outlet locations (six different strategies) were tested. Results show that the ventilation acts differently according to the particle size. For small particles (particle diameter lower than 5 mm in diameter), deposition is increased by a factor 2 when the airflow was changed from the Top–Top to the Bottom–Top inlet/outlet configuration. Increasing the ventilation rate from 0.5 to 1.0 h1 does not modify deposition for the Top–Top configuration but decreases it by 2.8 for the Bottom–Top. The effect of the inlet and outlet locations is less notable for coarse particles. This experimental study reveals that the ventilation strategy has to be well adapted to the particle size in order to improve its effectiveness. We show that the locations of the inlet and the outlet can be a very important parameter and have to be taken into account to predict particle indoor air quality. In addition, a numerical model of particle dispersion was developed. The program calculates instantaneous distributions of air velocity, using the large eddy simulation (LES) method. Trajectories of particles are obtained by implementing a Lagrangian particle model into the LES program. We simulated the experimental conditions in the three-dimensional numerical model and results show that ventilation strategy influences particle deposition in the room. A comparison of numerical and experimental results is given for 5 and 10mm particles. Particle behaviour is well predicted and this model seems to be adapted to predict indoor particle air quality in buildings. More experimental results are needed for a better validation of the numerical model, essentially for small particles. r 2005 Elsevier Ltd. All rights reserved. Keywords: Air quality; Particle matter; Large eddy simulation; Lagrangian model; Building physics
1. Introduction Indoor air pollution has become a major subject over the past few decades. In the urban environment, outdoor air which is heavily polluted by industrial activities and vehicle emissions, penetrates inside building, and influCorresponding author. Tel.: +33 546458310; fax: +33 546458241. E-mail address: [email protected] (J. Bouilly).
ences the indoor air quality. In addition, indoor particle sources, such as tobacco smoke and cooking vapours can have a greater effect on personal exposure (Abt et al., 2000). Particle deposition on surfaces and adapted ventilation strategy can substantially reduce indoor particle concentrations, resulting in improving the indoor air quality. To predict particle pollution in buildings, sizeresolved deposition rate can be used. Reviews of experimental studies on the particle deposition process
1352-2310/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2005.04.033
ARTICLE IN PRESS 4886
J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
were reported by Hinds (1982), Wallace (1996) and recently by Lai (2002). Generally, these studies give large variability in deposition rate for each particle size. Size of the experimental room (Nazaroff et al., 1993), roughness of surfaces (Abadie et al., 2001), airflow rates (Roed and Goddard, 1990; Fogh et al., 1997; Nomura et al., 1997; Jamriska et al., 2000), inlet/outlet locations (Mundt, 2001; Zhao et al., 2004), furnished or unfurnished room (Thatcher et al., 2002), are parameters that influence the indoor particle deposition rate. In the present study, measurements of the particle concentration evolution in a mechanically ventilated room have been carried out to investigate the combined effects of ventilation strategies and air change rate on the size-resolved particle deposition rate. A numerical model, using large eddy simulation (LES) and Lagrangian particle model was developed. Particle deposition rates are obtained by calculating a large number of particle trajectories. Numerical results show that particle deposition velocity is influenced by the ventilation strategy.
2. Experiments 2.1. Experimental methodology Particle concentration measurements were performed in a cubic test-room with 2.5 m sides, covered with wood panels. The layout of the room is shown in Fig. 1. The test-room is equipped with a mechanical ventilation system. The airflow was adjusted via an electronic fan speed controller: two airflow rates, corresponding to 0.5 and 1.0 air change per hour (ach), were calibrated using the tracer decay method with sulphur hexafluoride (SF6). These air-exchange rates were chosen according to AIVC 44 (1994) that reports that in European buildings exchange rates range from 0.2 to 1.0 h1, with a mean value of 0.4 h1. The uncertainty of the calculated air change rates was 70.05 h1. Highefficiency filters were used to prevent incoming particles from outdoor (filter 1) and to avoid particle
Fig. 1. Schematic diagram of the experimental system (crosssection).
Table 1 Inlet and outlet centre locations (vertical axis: Y) Configuration
‘‘Bottom–Top’’ ‘‘Top–bottom’’ ‘‘Top–Top’’
Inlet location (m)
X ¼0
X ¼0
X ¼0
Y ¼ 0:3 Z ¼ 1:25
Y ¼ 2:2 Z ¼ 1:25
Y ¼ 2:2 Z ¼ 1:25
X ¼ 2:5
X ¼ 2:5
Y ¼ 0:3 Z ¼ 1:25
Y ¼ 2:2 Z ¼ 1:25
Outlet location X ¼ 2:5 (m) Y ¼ 2:2 Z ¼ 1:25
releases outside of the test-room (filter 2). The ventilation air enters the room through the inlet section (0.07 m 0.07 m) and exhausts through the outlet section (0.07 m 0.07 m). Locations of the inlet and the outlet are presented in Table 1. The two ventilation rates give Reynolds numbers based on the air velocity at inlet and inlet height from 2000 to 4000. All measurements were performed under isothermal condition (20 1C72 1C) and constant relative humidity (50%710%). Two dust monitors (optical particle counter, GRIMM 1.108) continuously measured the particle concentration in the range 0.3–25 mm. In the present study, we focused our analysis on particles lower than 15 mm in diameter because they represent the majority of indoor particle pollutant and a potential danger for human health. The first counter was located at the geometrical centre of the room; the second one was placed at the inlet to control the particle concentration of the incoming air. The experimental procedure was the following: the ventilation system was set to a predetermined airflow rate, the dust monitors were switched on. The particle concentration was measured every minute during 2 h. Fig. 2 presents typical evolutions of the particle concentration during the tests. Values during the first 15 min correspond to the background particle concentration. Polydispersed polyamide powders (0.3–15 mm; spherical shape; obtained by direct polymerization) were instantaneously injected at time t ¼ 15 min by the use of a compressed air device (as suggested by Hinds, 1982) at different locations within the whole volume. The ventilation airflow is enough to obtain concentration homogeneity as already reported by Thatcher et al. (2002). Maximum particle concentration was 107 particles dm3. At this level, coagulation is insignificant compared to the deposition process (Hinds, 1982). Based on this particle concentration, the time required to reach the Boltzmann equilibrium is 5–6 min (Reist, 1993). Note that preliminary tests have been carried out to assess the particle concentration homogeneity within the room for each configuration. Comparison of the
ARTICLE IN PRESS J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
4887
Fig. 2. Particle concentrations for selected particles size range during an experiment.
concentration levels measured by dust monitor 2 at the outlet, in the jet (at the centre of the room) and in one bottom corner with those measured by the dust monitor 1 (located at the centre of the room) shows that the overall concentration differences are 3% and 6% for small and coarse particles respectively, with a maximum value of 9% for the ‘‘Top–Top’’, 0.5 ach configuration. We did not notice any significant change of the concentration homogeneity by modifying the inlet/outlet location. Independently of the tested configuration, the room can be considered as a well-mixed zone.
2.2. Indoor particle pollution modelling To evaluate the particle concentration evolution within a room, under isothermal conditions, the mass balance of the pollutant is written as a function of the incoming polluted air, the particle deposition on the walls and the mass of pollutant leaving the zone. Considering that the room is a well-mixed ventilated zone (see previous section) and that there is no particle generation or coagulation in the room during the decay, the time-dependent particle concentration is given by dC i ðtÞ ¼ lv C 0 ðtÞ lv C i ðtÞ lde C i ðtÞ, dt
(1)
where t is the time (h), Ci(t) and C0(t) are, respectively, the indoor and the outdoor particle concentration (number of particles m3) at time t, lv is the air change rate (h1), and lde is the particle deposition loss rate coefficient (h1). In the case of low outdoor particle concentration (in comparison with indoor level) or ventilation system equipped with a particle high-efficiency filter, the particle infiltration from outdoor can be neglected and a direct analytical solution to Eq. (1) is given by C i ðtÞ ¼ C i ð0Þexpðlg tÞ,
(2)
where Ci(0) is the initial indoor concentration and lg ¼ lv lde represents the overall loss rate (h1). Linear decreases (in log scale) corresponding to particle concentration exponential decays were then measured. The overall loss rate lg of Eq. (2) was determined after 5–6 min from the injection (time required to reach the Boltzmann equilibrium), by regression of the decrease part of the curves. The correlation coefficient r2 was usually higher than 0.95. All experiments were reproduced five times. 2.3. Experimental results Table 2 summarises the particle deposition loss rate coefficient lde (with lde ¼ lg lv ) measured under all experimental conditions, i.e. ‘‘Bottom–Top’’, ‘‘Top– Bottom’’ and ‘‘Top–Top’’ configurations with a ventilation rate lv set to 0.5 and 1.0 ach. Taking into account the variability of the experimental results, we have arbitrary chosen that a 25% relative difference between two deposition loss rates is needed to consider that the ventilation strategy really influences the particle deposition. With this criteria, values obtained for the ‘‘Bottom–Top’’ configuration are close to those for the ‘‘Top–Bottom’’ configuration with a ventilation rate lv set to 0.5 ach, and close to those for the ‘‘Top–Top’’ configuration for lv ¼ 1:0 ach. For the two studied ventilation rates, the ‘‘Top–Bottom’’ ventilation strategy gives the highest deposition loss rates. The lowest values are given by the ‘‘Top– Top’’ configuration, except for particles greater than 7.5 mm. For these particles, the ventilation strategy has a negligible effect on particle deposition and the gravity forces are the most important phenomena. Besides, for the smallest particles, the airflow pattern has a great influence on the particle deposition. In order to analyse the effects of the ventilation strategy, we developed a numerical model using a Lagrangian model associated with LES for the airflow computation. The model is presented in the next section.
ARTICLE IN PRESS J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
4888
Table 2 Experimental deposition loss rates (r2 40:95) as a function of particle size and ventilation strategy Strategy
Bottom–Top
Ventilation 0.5 rate (h1) Deposition Mean Min loss rate (h1) Particle xize range 0.3–0.4 0.55 0.4–0.5 0.54 0.5–0.65 0.53 0.65–0.8 0.55 0.8–1.0 0.65 1.0–1.6 0.72 1.6–2.0 0.75 2.0–3.0 0.86 3.0–4.0 1.10 4.0–5.0 1.27 5.0–7.5 1.75 7.5–10.0 3.72 10.0–15.0 6.46
(mm) 0.54 0.50 0.52 0.45 0.53 0.50 0.50 0.65 0.92 1.11 1.39 3.24 6.27
Top–Bottom 1.0
Top–Top
0.5
1.0
0.5
1.0
Max Mean Min
Max Mean Min
Max Mean Min
Max Mean Min
Max Mean Min
Max
0.56 0.58 0.54 0.66 0.72 1.01 0.99 1.03 1.26 1.38 2.18 3.99 6.66
0.35 0.31 0.45 0.50 0.72 0.56 1.06 1.37 1.57 1.63 2.47 4.26 6.70
0.56 0.52 0.75 0.75 0.83 0.82 1.31 1.31 1.79 2.01 2.47 5.03 10.38
0.56 0.48 0.56 0.78 0.82 0.77 0.96 1.40 1.77 1.85 2.48 4.88 9.94
0.32 0.32 0.35 0.50 0.61 0.83 0.71 0.97 1.21 1.40 1.89 4.00 8.18
0.32 0.31 0.35 0.50 0.67 0.71 0.80 1.24 1.04 1.36 2.32 5.50 8.99
0.20 0.19 0.31 0.37 0.40 0.48 0.59 0.78 1.22 1.41 1.86 2.91 5.58
0.04 0.07 0.16 0.28 0.21 0.38 0.12 0.35 1.03 1.24 1.47 2.1 3.83
0.51 0.49 0.62 0.67 0.74 0.78 0.97 1.05 1.19 1.40 1.98 3.33 5.98
0.47 0.46 0.53 0.55 0.57 0.76 0.84 0.85 0.79 0.99 1.59 2.39 3.83
3. Numerical study 3.1. Lagrangian model In this study, a Lagrangian model is used to compute particle trajectories in the room. The interaction between the fluid (air) and the particles is treated as a one-way coupling, assuming that the effect of particles on the turbulent flow is negligible due to particle size and concentration levels. According to our experimental results, two particle sizes were chosen: 5 and 10 mm particle aerodynamic diameter. For 5 mm particles, deposition starts to depend on both airflow rates and inlet/outlet locations whereas 10 mm particle deposition is not affected by the ventilation strategy. Moreover, the required computational time is acceptable for these particle sizes. When a particle touches a solid surface (walls, floor and ceiling), it sticks to it without rebound. Since a dilute aerosol is modelled, there is no particle coagulation. Moreover, we focus on monodisperse aerosols of spherical solid particles. The Lagrangian method consists in computing the trajectory of each particle by solving the momentum equation based on Newton’s second law: ! dU i r 3 C D r ~ðui U i Þ þ ~ ¼ 1 gdi2 (3) uU dt rp 4 d p rp where dp is the particle’s diameter, rp is the particle’s density, r is the air density, Ui and ui are the particle and ~ and ~ fluid instantaneous velocities respectively, U u are
0.44 0.41 0.51 0.63 0.71 0.74 0.82 1.17 1.46 1.62 2.10 3.61 8.12
0.36 0.36 0.44 0.44 0.44 0.71 0.69 0.85 1.22 1.42 1.88 2.78 5.37
0.21 0.20 0.26 0.36 0.44 0.50 0.56 0.65 0.87 1.05 1.53 2.90 6.34
0.10 0.10 0.19 0.21 0.21 0.25 0.33 0.48 0.71 0.85 1.13 2.15 4.71
0.21 0.21 0.23 0.32 0.43 0.49 0.54 0.72 0.84 1.09 1.80 3.47 6.66
0.11 0.09 0.11 0.11 0.16 0.21 0.30 0.53 0.65 0.86 1.27 2.47 4.18
the particle and fluid instantaneous velocity vectors, g is the gravitational acceleration, di2 is the Kronecker symbol. CD is the drag coefficient, depending on the flow parameters such as Reynolds number, turbulence level (Hinds, 1982): CD ¼
24 ð1 þ 0:15Re0:687 Þ p Rep
C D ¼ 0:44 if
if
Rep p1000,
Rep 41000.
ð4Þ
Rep is the particle Reynolds number: ~ d p j~ u Uj , (5) n where n is the kinematic viscosity of air. Eq. (3) is solved at each time step for every particle and gives the instantaneous velocity of the particle. Then, the position xi of each particle is obtained using the following equation (trajectory equation): Rep ¼
dxi ¼ U i. (6) dt Eq. (3) requires the knowledge of the instantaneous fluid velocity at the particle’s location. For the weakly turbulent flow considered in this paper, the instantaneous fluid velocity is calculated using LES. 3.2. Large eddy simulation In LES, the turbulent flow is separated into large eddies and small eddies. The contribution of the large, energy carrying eddies to momentum and energy
ARTICLE IN PRESS J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
transfer is computed exactly, and the effect of the smallest eddies is modelled. The large eddy variables are calculated by applying a filter to the continuity and Navier–Stokes equations. A one-dimensional filtered variable can be obtained from: Z f ¼ Gðx; x0 Þ fðxÞ dx0 , (7) where G is the filter function and f is the variable to be filtered . In the real space, the tophat filter is employed: 8 D 1 > 0 > < if x x p ; D 2 Gðx; x0 Þ ¼ (8) D > 0 > : 0 if jx x jX ; 2 where D is the filter width. Applying the filtering operation to the continuity and Navier–Stokes equations of an isothermal incompressible flow yields: qui ¼ 0, qxi qtij qui q 1 qp q2 ui þ ðui uj Þ ¼ þn . r qxi qt qxj qxj qxj qxj
(9)
(10)
These equations give the instantaneous variables of the large eddies. In the momentum equation, one notices the subgrid Reynolds stresses that represent the effect of the small eddies on the large energy containing eddies: tij ¼ ui uj ui uj .
(11)
For most subgrid scale models, the subgrid stresses are modelled via an eddy viscosity hypothesis: tij ¼ 2nSGS Sij , where the resolved scale stresses are:
1 qui quj . Sij ¼ þ 2 qxj qxi
(12)
(13)
The eddy viscosity is defined as: nSGS ¼ CD2 ð2Sij S ij Þ1=2 ,
(14)
where the coefficient C varies according to time and location. In this study, the filtered dynamic subgrid scale model proposed by Zhang and Chen (2000) is used to compute C. This LES code coupled with the Lagrangian particle model was validated using the experimental results from Snyder and Lumley (1971) (Jiang et al., 2002). 3.3. Numerical procedure At each time step, the LES equations and particle’s equations are solved. The governing equations of LES coupled by the pressure are solved with the Simplified Marker and Cell Method (Harlow and Welch, 1965). The present study uses the finite difference method to
4889
discretize the LES equations and a staggered grid. The convection terms are discretized with a second-order central differencing scheme. The time term in the filtered momentum equations is discretized by a second-order explicit differencing scheme (explicit Adams–Bashforth scheme). The particle’s equations of motion are integrated with the fifth-order Runge–Kutta method (William, 1992). To integrate according to time the equations of flow field and particle motion, two different time steps could be used. The first time step is relative to the integration of the LES equations. The CFL condition requires that this time step satisfies: DtpDmin =umax ,
(15)
where Dmin is the minimum grid size and umax is the maximum velocity. The second time step is relative to the explicit integration of the particle’s equations. Since the particles studied are small and thus very sensitive to the fluctuating velocity field, this time step is much smaller than the time step relative to the LES equations. This time step is thus used to integrate both the LES and particle’s equations. The time step used for the computations presented in this paper is 0.0002 s. The experimental room presented in the first section was reproduced numerically. The room is divided into 45 47 48 cells (i.e. control volumes). A refinement mesh is established near all walls of the enclosure, and also in the inlet and outlet areas. The supply air velocity is uin ¼ 0:443 m s1 (for ach 0.5 h1), and uin ¼ 0:886 m s1 (for ach 1.0 h1). The particles are spherical and solid, with 5 or 10 mm aerodynamic diameter and a density rp ¼ 1000 kg m3. At the initial time, the particles are assumed to be uniformly distributed in the room. Statistical results are obtained by calculating a large number of particle trajectories (1000 particles). At each time step, the percentage of particles deposited on each surface of the room, the percentage of particles exhausted by the ventilation and the percentage of particles remaining in the air are calculated. 3.4. Numerical results The particle concentration in the room is relatively homogeneous during the simulation. Fig. 3 presents an example of the 10 mm particle dispersion at different times for the ‘‘Bottom–Top’’ configuration with 1.0 ach. In order to check the homogeneity, the volume of the room was divided in 3 3 3 cells and the concentration levels of each cell are recorded at each time step. The concentrations in these zones were roughly the same during the whole set of simulations. For example, the particle concentration of each cell was 64 particles. m3 11% at t ¼ 0 s and 22 particles. m 3 14% at t ¼ 600 s for the ‘‘Top–Bottom’’ configuration, 10 mm
ARTICLE IN PRESS 4890
J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
Fig. 3. Particle location in the median vertical plan (‘‘Bottom–Top’’, 10 mm particles, 1.0 ach; left side: t ¼ 0 s; right side: t ¼ 500 s).
the configuration ‘‘Top–Top’’, 0.5 ach. For 10 mm particles, effects of ventilation are negligible: all values are within 4.72 h179%. In this case, the influence of the gravity on the particles’ behaviour is higher than the strength of the ventilation. The highest value (5.49 h1) is obtained for the configuration ‘‘Top–Bottom’’, 1.0 ach. This configuration accelerates the particle deposition phenomenon. With the ‘‘Bottom–Top’’ configuration, the airflow isolates the floor from the core of the room and delays the particles to deposit onto this surface which is the preponderant particle deposition surface in rooms for these particle sizes.
4. Discussion Fig. 4. Evolution of the particle concentration obtained by the presented model (‘‘Top–Bottom’’, 10 mm particles, 1.0 ach).
particles, 1.0 ach. As a result, the one-zone model described by Eq. (1) can be used for each configuration. An example of the particle concentration decay is given in Fig. 4. By regression, the overall loss rate coefficient is obtained. The correlation coefficient r2 is usually higher than 0.97. Table 3 gives the particle deposition loss rate values for each ventilation strategy tested with the code. First of all, we can see that for 5 mm particles, the deposition loss rate ranges between 1.11 and 2.23 h1. This observation confirms that the ventilation strategy acts on the particle deposition. The lowest value is given by
Previous experimental studies report the particle deposition velocity Vde which is usually used to evaluate the strength of the particle deposition on surfaces. This parameter is linked to the deposition loss rate lde by the following equation: V de ¼
lde V , 3600 S
(16) 1
where Vde is the particle deposition velocity (m s ), lde represents the particle deposition loss rate (h1), V is the volume of the room (m3) and S is the total area of the room (m2). As we know the quantity of particles extracted through the outlet at each time step, the deposition loss rate lde can be deduced from simulation, and thus the deposition velocity can be determined according to the ventilation strategy.
ARTICLE IN PRESS J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
4891
Table 3 Numerical deposition loss rates ðr40:97Þ as a function of particle size and ventilation strategy Strategy
Bottom–Top 1
Top–Bottom
Top–Top
Ventilation rate (h )
0.5
1.0
0.5
1.0
0.5
1.0
Deposition loss rate (h1) (5 mm particles) Deposition loss rate (h1) (10 mm particles)
1.30 4.68
1.13 4.24
1.62 4.65
2.23 5.49
1.11 4.81
1.38 4.47
Fig. 5. Particle deposition velocity versus particle size and ventilation strategy (previous experimental results reported by Lai, 2002).
Fig. 5 shows both the previous studies values, as reported by Lai (2002), and our experimental and numerical results. Scattered values from the other studies are due to multiple experimental conditions (furnishing, room geometry, roughness of surfaces, ventilation rate, etc.). In the present study, we modelled our experimental conditions and obtained a good agreement, especially for the lowest ventilation rate. For 5 mm particles, choosing appropriate locations of the inlet and outlet can increase the deposition velocity by a factor 1.5 and 2 for ventilation rates of 0.5 and 1.0 h1 respectively. For 10 mm particles, these factors are close to 1. Thus, doubling the ventilation rate is not enough to increase the deposition velocity for 10 mm particles. The smaller the particles are, the more important the ventilation strategy is, on particle deposition velocity.
5. Conclusion By measuring the particle concentration evolution in a mechanically ventilated room, the present study brought
new experimental data that were needed to improve our knowledge of the influence of the ventilation strategy on the particle pollution. It would be necessary to increase the number of measurements for each case in order to lower the remaining deposition loss rate uncertainty. However, as confirmed by the numerical results, the particles removal from indoor air depends on the airflow rate but, it depends on the airflow pattern within the room as well. We found that the influence of the inlet/ outlet locations is stronger for fine particles (lower than 5 mm in diameter) than for coarse particles and that an increase of the ventilation rate does not necessarily lead to higher deposition. The choice of a ventilation strategy, i.e. the airflow rate as well as the inlet/outlet locations, has to be carefully taken into account in order to limit the particle pollution in rooms. The numerical model used in this study seems to be well adapted to predict indoor particle air quality in buildings but is time consuming for the small particles. Limits of the Lagrangian method have to be investigated according to particle size; since the gravity effect becomes negligible as the particle size decreases, an Eulerian model could be well adapted. The benefit of the numerical model described in this paper lies in the knowledge of the deposited particle and high concentration locations. Consequently, the particle deposition velocity can be calculated for each surface of the room (floor, ceiling and vertical walls etc.) which could help better understanding the effects of the ventilation strategy on the particle loss rate. This investigation will be detailed in a further paper.
Acknowledgements This study is part of the Ph.D. dissertation ‘‘Study of the impact of particle pollution on indoor air quality in urban sites’’ co-financed by the Agency of the Environment and the Energy Management (ADEME) and the Poitou-Charentes Region. The authors wish to express their thanks to the CINES (Centre Informatique National de l’Enseignement Supe´rieur, France); computations were carried out on the IBM computers of the CINES. The authors are grateful to Marc Abadie for the helpful editorial comments.
ARTICLE IN PRESS 4892
J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
References Abadie, M., Limam, K., Allard, F., 2001. Indoor particle pollution: effect of wall textures on particle deposition. Building and Environment 36, 821–827. Abt, E., Suh, H.H., Catalano, P., Koutrakis, P., 2000. Relative contribution of outdoor and indoor particle sources to indoor concentrations. Environmental Science and Technology 34, 3579–3587. AIVC 44, 1994. An analysis and data summary of the AIVC’s numerical database. Technical note AIVC (Air Infiltration and Ventilation Centre). Fogh, C.L., Byrne, M.A., Roed, J., Goddard, A.J.H., 1997. Size specific indoor aerosol deposition measurements and derived I/O concentration ratios. Atmospheric Environment 31, 2193–2203. Harlow, F.H., Welch, J.E., 1965. Numerical calculation of time dependent viscous incompressible flow of fluid with free surface. Physics of Fluids 8, 2182–2189. Hinds, W.C., 1982. Aerosols Technology, Proprieties, Behaviour and Measurement of Airborne Particles. WileyInterscience Publication, New York. Jamriska, M., Morawska, L., Clark, B.A., 2000. Effect of ventilation and filtration on submicrometer particles in an indoor environment. Indoor Air 10, 19–26. Jiang, Y., Be´ghein, C., Chen, Q., 2002. Study of particle dispersion in buildings with large eddy simulation. Proceedings of Indoor Air 2002, 30th June–5th July 2002. Monterey, CA. Lai, A.C.K., 2002. Particle deposition indoors: a review. Indoor Air 12, 211–214. Mundt, E., 2001. Non-buoyant pollutant sources and particles in displacement ventilation. Building and Environment 36 (1), 829–836.
Nazaroff, W.W., Gadgil; A.J., Weschler, C.J., 1993. Critique of the use of deposition velocity in modeling indoor air quality. Modeling Indoor Air Quality and Exposure, STP 1205, American Society for Testing and Materials, Philadelphia, PA, pp. 81–104. Nomura, Y., Hopke, P.K., Fitzgerald, B., Mesbah, B., 1997. Deposition of particles in a chamber as a function of ventilation rate. Aerosol Science and Technology 27, 62–72. Reist, P.C., 1993. Aerosol Science and Technology, second ed. McGraw-Hill, Inc., New York. Roed, J., Goddard, A.J.H., 1990. Ingress of radioactivity into dwellings. Proceedings of Methods and Codes for Assessing Off-site Consequences of Nuclear Accidents, Athens, 13013, pp. 433–450. Snyder, W.H., Lumley, J.L., 1971. Some measurements of particle velocity autocorrelation functions in a turbulent flow. Journal of Fluid Mechanics 48 (1), 41–71. Thatcher, T.L., Lai, A.C.K., Moreno-Jackson, R., 2002. Effects of room furnishings and air speed on particle deposition rates indoors. LBNL Report 48414. Atmospheric Environment 36, 1811–1819. Wallace, L.A., 1996. Indoor particles: a review. Journal of Air Waste Management Association 46, 98–126. William, H., 1992. Numerical Recipes in Fortran : the Art of Scientific Computing. Cambridge University Press, Cambridge, England. Zhang, W., Chen, Q., 2000. Large eddy simulation of indoor airflow with a filtered dynamic subgrid scale model. International Journal of Heat and Mass Transfer 43, 3219–3231. Zhao, B., Zhang, Y., Xianting, L., Xudong, Y., Dongtao, H., 2004. Comparison of indoor aerosol particle concentration and deposition in different ventilated rooms by numerical method. Building and Environment 39, 1–8.
Atmospheric Environment 39 (2005) 4885–4892 www.elsevier.com/locate/atmosenv
Effect of ventilation strategies on particle decay rates indoors: An experimental and modelling study Je´roˆme Bouilly, Karim Limam, Claudine Be´ghein, Francis Allard LEPTAB, University of La Rochelle, Av. M. Cre´peau, 17042 La Rochelle Cedex 01, France Received 22 August 2004; received in revised form 5 January 2005; accepted 30 April 2005
Abstract A cubic experimental chamber of 2.5 m 2.5 m 2.5 m was designed to study the impact of ventilation strategies on the indoor particle concentration. Particles of 0.3–15 mm aerodynamic diameter were used. The combined effects of the ventilation rate (0.5 and 1.0 ach) and the inlet and outlet locations (six different strategies) were tested. Results show that the ventilation acts differently according to the particle size. For small particles (particle diameter lower than 5 mm in diameter), deposition is increased by a factor 2 when the airflow was changed from the Top–Top to the Bottom–Top inlet/outlet configuration. Increasing the ventilation rate from 0.5 to 1.0 h1 does not modify deposition for the Top–Top configuration but decreases it by 2.8 for the Bottom–Top. The effect of the inlet and outlet locations is less notable for coarse particles. This experimental study reveals that the ventilation strategy has to be well adapted to the particle size in order to improve its effectiveness. We show that the locations of the inlet and the outlet can be a very important parameter and have to be taken into account to predict particle indoor air quality. In addition, a numerical model of particle dispersion was developed. The program calculates instantaneous distributions of air velocity, using the large eddy simulation (LES) method. Trajectories of particles are obtained by implementing a Lagrangian particle model into the LES program. We simulated the experimental conditions in the three-dimensional numerical model and results show that ventilation strategy influences particle deposition in the room. A comparison of numerical and experimental results is given for 5 and 10mm particles. Particle behaviour is well predicted and this model seems to be adapted to predict indoor particle air quality in buildings. More experimental results are needed for a better validation of the numerical model, essentially for small particles. r 2005 Elsevier Ltd. All rights reserved. Keywords: Air quality; Particle matter; Large eddy simulation; Lagrangian model; Building physics
1. Introduction Indoor air pollution has become a major subject over the past few decades. In the urban environment, outdoor air which is heavily polluted by industrial activities and vehicle emissions, penetrates inside building, and influCorresponding author. Tel.: +33 546458310; fax: +33 546458241. E-mail address: [email protected] (J. Bouilly).
ences the indoor air quality. In addition, indoor particle sources, such as tobacco smoke and cooking vapours can have a greater effect on personal exposure (Abt et al., 2000). Particle deposition on surfaces and adapted ventilation strategy can substantially reduce indoor particle concentrations, resulting in improving the indoor air quality. To predict particle pollution in buildings, sizeresolved deposition rate can be used. Reviews of experimental studies on the particle deposition process
1352-2310/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2005.04.033
ARTICLE IN PRESS 4886
J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
were reported by Hinds (1982), Wallace (1996) and recently by Lai (2002). Generally, these studies give large variability in deposition rate for each particle size. Size of the experimental room (Nazaroff et al., 1993), roughness of surfaces (Abadie et al., 2001), airflow rates (Roed and Goddard, 1990; Fogh et al., 1997; Nomura et al., 1997; Jamriska et al., 2000), inlet/outlet locations (Mundt, 2001; Zhao et al., 2004), furnished or unfurnished room (Thatcher et al., 2002), are parameters that influence the indoor particle deposition rate. In the present study, measurements of the particle concentration evolution in a mechanically ventilated room have been carried out to investigate the combined effects of ventilation strategies and air change rate on the size-resolved particle deposition rate. A numerical model, using large eddy simulation (LES) and Lagrangian particle model was developed. Particle deposition rates are obtained by calculating a large number of particle trajectories. Numerical results show that particle deposition velocity is influenced by the ventilation strategy.
2. Experiments 2.1. Experimental methodology Particle concentration measurements were performed in a cubic test-room with 2.5 m sides, covered with wood panels. The layout of the room is shown in Fig. 1. The test-room is equipped with a mechanical ventilation system. The airflow was adjusted via an electronic fan speed controller: two airflow rates, corresponding to 0.5 and 1.0 air change per hour (ach), were calibrated using the tracer decay method with sulphur hexafluoride (SF6). These air-exchange rates were chosen according to AIVC 44 (1994) that reports that in European buildings exchange rates range from 0.2 to 1.0 h1, with a mean value of 0.4 h1. The uncertainty of the calculated air change rates was 70.05 h1. Highefficiency filters were used to prevent incoming particles from outdoor (filter 1) and to avoid particle
Fig. 1. Schematic diagram of the experimental system (crosssection).
Table 1 Inlet and outlet centre locations (vertical axis: Y) Configuration
‘‘Bottom–Top’’ ‘‘Top–bottom’’ ‘‘Top–Top’’
Inlet location (m)
X ¼0
X ¼0
X ¼0
Y ¼ 0:3 Z ¼ 1:25
Y ¼ 2:2 Z ¼ 1:25
Y ¼ 2:2 Z ¼ 1:25
X ¼ 2:5
X ¼ 2:5
Y ¼ 0:3 Z ¼ 1:25
Y ¼ 2:2 Z ¼ 1:25
Outlet location X ¼ 2:5 (m) Y ¼ 2:2 Z ¼ 1:25
releases outside of the test-room (filter 2). The ventilation air enters the room through the inlet section (0.07 m 0.07 m) and exhausts through the outlet section (0.07 m 0.07 m). Locations of the inlet and the outlet are presented in Table 1. The two ventilation rates give Reynolds numbers based on the air velocity at inlet and inlet height from 2000 to 4000. All measurements were performed under isothermal condition (20 1C72 1C) and constant relative humidity (50%710%). Two dust monitors (optical particle counter, GRIMM 1.108) continuously measured the particle concentration in the range 0.3–25 mm. In the present study, we focused our analysis on particles lower than 15 mm in diameter because they represent the majority of indoor particle pollutant and a potential danger for human health. The first counter was located at the geometrical centre of the room; the second one was placed at the inlet to control the particle concentration of the incoming air. The experimental procedure was the following: the ventilation system was set to a predetermined airflow rate, the dust monitors were switched on. The particle concentration was measured every minute during 2 h. Fig. 2 presents typical evolutions of the particle concentration during the tests. Values during the first 15 min correspond to the background particle concentration. Polydispersed polyamide powders (0.3–15 mm; spherical shape; obtained by direct polymerization) were instantaneously injected at time t ¼ 15 min by the use of a compressed air device (as suggested by Hinds, 1982) at different locations within the whole volume. The ventilation airflow is enough to obtain concentration homogeneity as already reported by Thatcher et al. (2002). Maximum particle concentration was 107 particles dm3. At this level, coagulation is insignificant compared to the deposition process (Hinds, 1982). Based on this particle concentration, the time required to reach the Boltzmann equilibrium is 5–6 min (Reist, 1993). Note that preliminary tests have been carried out to assess the particle concentration homogeneity within the room for each configuration. Comparison of the
ARTICLE IN PRESS J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
4887
Fig. 2. Particle concentrations for selected particles size range during an experiment.
concentration levels measured by dust monitor 2 at the outlet, in the jet (at the centre of the room) and in one bottom corner with those measured by the dust monitor 1 (located at the centre of the room) shows that the overall concentration differences are 3% and 6% for small and coarse particles respectively, with a maximum value of 9% for the ‘‘Top–Top’’, 0.5 ach configuration. We did not notice any significant change of the concentration homogeneity by modifying the inlet/outlet location. Independently of the tested configuration, the room can be considered as a well-mixed zone.
2.2. Indoor particle pollution modelling To evaluate the particle concentration evolution within a room, under isothermal conditions, the mass balance of the pollutant is written as a function of the incoming polluted air, the particle deposition on the walls and the mass of pollutant leaving the zone. Considering that the room is a well-mixed ventilated zone (see previous section) and that there is no particle generation or coagulation in the room during the decay, the time-dependent particle concentration is given by dC i ðtÞ ¼ lv C 0 ðtÞ lv C i ðtÞ lde C i ðtÞ, dt
(1)
where t is the time (h), Ci(t) and C0(t) are, respectively, the indoor and the outdoor particle concentration (number of particles m3) at time t, lv is the air change rate (h1), and lde is the particle deposition loss rate coefficient (h1). In the case of low outdoor particle concentration (in comparison with indoor level) or ventilation system equipped with a particle high-efficiency filter, the particle infiltration from outdoor can be neglected and a direct analytical solution to Eq. (1) is given by C i ðtÞ ¼ C i ð0Þexpðlg tÞ,
(2)
where Ci(0) is the initial indoor concentration and lg ¼ lv lde represents the overall loss rate (h1). Linear decreases (in log scale) corresponding to particle concentration exponential decays were then measured. The overall loss rate lg of Eq. (2) was determined after 5–6 min from the injection (time required to reach the Boltzmann equilibrium), by regression of the decrease part of the curves. The correlation coefficient r2 was usually higher than 0.95. All experiments were reproduced five times. 2.3. Experimental results Table 2 summarises the particle deposition loss rate coefficient lde (with lde ¼ lg lv ) measured under all experimental conditions, i.e. ‘‘Bottom–Top’’, ‘‘Top– Bottom’’ and ‘‘Top–Top’’ configurations with a ventilation rate lv set to 0.5 and 1.0 ach. Taking into account the variability of the experimental results, we have arbitrary chosen that a 25% relative difference between two deposition loss rates is needed to consider that the ventilation strategy really influences the particle deposition. With this criteria, values obtained for the ‘‘Bottom–Top’’ configuration are close to those for the ‘‘Top–Bottom’’ configuration with a ventilation rate lv set to 0.5 ach, and close to those for the ‘‘Top–Top’’ configuration for lv ¼ 1:0 ach. For the two studied ventilation rates, the ‘‘Top–Bottom’’ ventilation strategy gives the highest deposition loss rates. The lowest values are given by the ‘‘Top– Top’’ configuration, except for particles greater than 7.5 mm. For these particles, the ventilation strategy has a negligible effect on particle deposition and the gravity forces are the most important phenomena. Besides, for the smallest particles, the airflow pattern has a great influence on the particle deposition. In order to analyse the effects of the ventilation strategy, we developed a numerical model using a Lagrangian model associated with LES for the airflow computation. The model is presented in the next section.
ARTICLE IN PRESS J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
4888
Table 2 Experimental deposition loss rates (r2 40:95) as a function of particle size and ventilation strategy Strategy
Bottom–Top
Ventilation 0.5 rate (h1) Deposition Mean Min loss rate (h1) Particle xize range 0.3–0.4 0.55 0.4–0.5 0.54 0.5–0.65 0.53 0.65–0.8 0.55 0.8–1.0 0.65 1.0–1.6 0.72 1.6–2.0 0.75 2.0–3.0 0.86 3.0–4.0 1.10 4.0–5.0 1.27 5.0–7.5 1.75 7.5–10.0 3.72 10.0–15.0 6.46
(mm) 0.54 0.50 0.52 0.45 0.53 0.50 0.50 0.65 0.92 1.11 1.39 3.24 6.27
Top–Bottom 1.0
Top–Top
0.5
1.0
0.5
1.0
Max Mean Min
Max Mean Min
Max Mean Min
Max Mean Min
Max Mean Min
Max
0.56 0.58 0.54 0.66 0.72 1.01 0.99 1.03 1.26 1.38 2.18 3.99 6.66
0.35 0.31 0.45 0.50 0.72 0.56 1.06 1.37 1.57 1.63 2.47 4.26 6.70
0.56 0.52 0.75 0.75 0.83 0.82 1.31 1.31 1.79 2.01 2.47 5.03 10.38
0.56 0.48 0.56 0.78 0.82 0.77 0.96 1.40 1.77 1.85 2.48 4.88 9.94
0.32 0.32 0.35 0.50 0.61 0.83 0.71 0.97 1.21 1.40 1.89 4.00 8.18
0.32 0.31 0.35 0.50 0.67 0.71 0.80 1.24 1.04 1.36 2.32 5.50 8.99
0.20 0.19 0.31 0.37 0.40 0.48 0.59 0.78 1.22 1.41 1.86 2.91 5.58
0.04 0.07 0.16 0.28 0.21 0.38 0.12 0.35 1.03 1.24 1.47 2.1 3.83
0.51 0.49 0.62 0.67 0.74 0.78 0.97 1.05 1.19 1.40 1.98 3.33 5.98
0.47 0.46 0.53 0.55 0.57 0.76 0.84 0.85 0.79 0.99 1.59 2.39 3.83
3. Numerical study 3.1. Lagrangian model In this study, a Lagrangian model is used to compute particle trajectories in the room. The interaction between the fluid (air) and the particles is treated as a one-way coupling, assuming that the effect of particles on the turbulent flow is negligible due to particle size and concentration levels. According to our experimental results, two particle sizes were chosen: 5 and 10 mm particle aerodynamic diameter. For 5 mm particles, deposition starts to depend on both airflow rates and inlet/outlet locations whereas 10 mm particle deposition is not affected by the ventilation strategy. Moreover, the required computational time is acceptable for these particle sizes. When a particle touches a solid surface (walls, floor and ceiling), it sticks to it without rebound. Since a dilute aerosol is modelled, there is no particle coagulation. Moreover, we focus on monodisperse aerosols of spherical solid particles. The Lagrangian method consists in computing the trajectory of each particle by solving the momentum equation based on Newton’s second law: ! dU i r 3 C D r ~ðui U i Þ þ ~ ¼ 1 gdi2 (3) uU dt rp 4 d p rp where dp is the particle’s diameter, rp is the particle’s density, r is the air density, Ui and ui are the particle and ~ and ~ fluid instantaneous velocities respectively, U u are
0.44 0.41 0.51 0.63 0.71 0.74 0.82 1.17 1.46 1.62 2.10 3.61 8.12
0.36 0.36 0.44 0.44 0.44 0.71 0.69 0.85 1.22 1.42 1.88 2.78 5.37
0.21 0.20 0.26 0.36 0.44 0.50 0.56 0.65 0.87 1.05 1.53 2.90 6.34
0.10 0.10 0.19 0.21 0.21 0.25 0.33 0.48 0.71 0.85 1.13 2.15 4.71
0.21 0.21 0.23 0.32 0.43 0.49 0.54 0.72 0.84 1.09 1.80 3.47 6.66
0.11 0.09 0.11 0.11 0.16 0.21 0.30 0.53 0.65 0.86 1.27 2.47 4.18
the particle and fluid instantaneous velocity vectors, g is the gravitational acceleration, di2 is the Kronecker symbol. CD is the drag coefficient, depending on the flow parameters such as Reynolds number, turbulence level (Hinds, 1982): CD ¼
24 ð1 þ 0:15Re0:687 Þ p Rep
C D ¼ 0:44 if
if
Rep p1000,
Rep 41000.
ð4Þ
Rep is the particle Reynolds number: ~ d p j~ u Uj , (5) n where n is the kinematic viscosity of air. Eq. (3) is solved at each time step for every particle and gives the instantaneous velocity of the particle. Then, the position xi of each particle is obtained using the following equation (trajectory equation): Rep ¼
dxi ¼ U i. (6) dt Eq. (3) requires the knowledge of the instantaneous fluid velocity at the particle’s location. For the weakly turbulent flow considered in this paper, the instantaneous fluid velocity is calculated using LES. 3.2. Large eddy simulation In LES, the turbulent flow is separated into large eddies and small eddies. The contribution of the large, energy carrying eddies to momentum and energy
ARTICLE IN PRESS J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
transfer is computed exactly, and the effect of the smallest eddies is modelled. The large eddy variables are calculated by applying a filter to the continuity and Navier–Stokes equations. A one-dimensional filtered variable can be obtained from: Z f ¼ Gðx; x0 Þ fðxÞ dx0 , (7) where G is the filter function and f is the variable to be filtered . In the real space, the tophat filter is employed: 8 D 1 > 0 > < if x x p ; D 2 Gðx; x0 Þ ¼ (8) D > 0 > : 0 if jx x jX ; 2 where D is the filter width. Applying the filtering operation to the continuity and Navier–Stokes equations of an isothermal incompressible flow yields: qui ¼ 0, qxi qtij qui q 1 qp q2 ui þ ðui uj Þ ¼ þn . r qxi qt qxj qxj qxj qxj
(9)
(10)
These equations give the instantaneous variables of the large eddies. In the momentum equation, one notices the subgrid Reynolds stresses that represent the effect of the small eddies on the large energy containing eddies: tij ¼ ui uj ui uj .
(11)
For most subgrid scale models, the subgrid stresses are modelled via an eddy viscosity hypothesis: tij ¼ 2nSGS Sij , where the resolved scale stresses are:
1 qui quj . Sij ¼ þ 2 qxj qxi
(12)
(13)
The eddy viscosity is defined as: nSGS ¼ CD2 ð2Sij S ij Þ1=2 ,
(14)
where the coefficient C varies according to time and location. In this study, the filtered dynamic subgrid scale model proposed by Zhang and Chen (2000) is used to compute C. This LES code coupled with the Lagrangian particle model was validated using the experimental results from Snyder and Lumley (1971) (Jiang et al., 2002). 3.3. Numerical procedure At each time step, the LES equations and particle’s equations are solved. The governing equations of LES coupled by the pressure are solved with the Simplified Marker and Cell Method (Harlow and Welch, 1965). The present study uses the finite difference method to
4889
discretize the LES equations and a staggered grid. The convection terms are discretized with a second-order central differencing scheme. The time term in the filtered momentum equations is discretized by a second-order explicit differencing scheme (explicit Adams–Bashforth scheme). The particle’s equations of motion are integrated with the fifth-order Runge–Kutta method (William, 1992). To integrate according to time the equations of flow field and particle motion, two different time steps could be used. The first time step is relative to the integration of the LES equations. The CFL condition requires that this time step satisfies: DtpDmin =umax ,
(15)
where Dmin is the minimum grid size and umax is the maximum velocity. The second time step is relative to the explicit integration of the particle’s equations. Since the particles studied are small and thus very sensitive to the fluctuating velocity field, this time step is much smaller than the time step relative to the LES equations. This time step is thus used to integrate both the LES and particle’s equations. The time step used for the computations presented in this paper is 0.0002 s. The experimental room presented in the first section was reproduced numerically. The room is divided into 45 47 48 cells (i.e. control volumes). A refinement mesh is established near all walls of the enclosure, and also in the inlet and outlet areas. The supply air velocity is uin ¼ 0:443 m s1 (for ach 0.5 h1), and uin ¼ 0:886 m s1 (for ach 1.0 h1). The particles are spherical and solid, with 5 or 10 mm aerodynamic diameter and a density rp ¼ 1000 kg m3. At the initial time, the particles are assumed to be uniformly distributed in the room. Statistical results are obtained by calculating a large number of particle trajectories (1000 particles). At each time step, the percentage of particles deposited on each surface of the room, the percentage of particles exhausted by the ventilation and the percentage of particles remaining in the air are calculated. 3.4. Numerical results The particle concentration in the room is relatively homogeneous during the simulation. Fig. 3 presents an example of the 10 mm particle dispersion at different times for the ‘‘Bottom–Top’’ configuration with 1.0 ach. In order to check the homogeneity, the volume of the room was divided in 3 3 3 cells and the concentration levels of each cell are recorded at each time step. The concentrations in these zones were roughly the same during the whole set of simulations. For example, the particle concentration of each cell was 64 particles. m3 11% at t ¼ 0 s and 22 particles. m 3 14% at t ¼ 600 s for the ‘‘Top–Bottom’’ configuration, 10 mm
ARTICLE IN PRESS 4890
J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
Fig. 3. Particle location in the median vertical plan (‘‘Bottom–Top’’, 10 mm particles, 1.0 ach; left side: t ¼ 0 s; right side: t ¼ 500 s).
the configuration ‘‘Top–Top’’, 0.5 ach. For 10 mm particles, effects of ventilation are negligible: all values are within 4.72 h179%. In this case, the influence of the gravity on the particles’ behaviour is higher than the strength of the ventilation. The highest value (5.49 h1) is obtained for the configuration ‘‘Top–Bottom’’, 1.0 ach. This configuration accelerates the particle deposition phenomenon. With the ‘‘Bottom–Top’’ configuration, the airflow isolates the floor from the core of the room and delays the particles to deposit onto this surface which is the preponderant particle deposition surface in rooms for these particle sizes.
4. Discussion Fig. 4. Evolution of the particle concentration obtained by the presented model (‘‘Top–Bottom’’, 10 mm particles, 1.0 ach).
particles, 1.0 ach. As a result, the one-zone model described by Eq. (1) can be used for each configuration. An example of the particle concentration decay is given in Fig. 4. By regression, the overall loss rate coefficient is obtained. The correlation coefficient r2 is usually higher than 0.97. Table 3 gives the particle deposition loss rate values for each ventilation strategy tested with the code. First of all, we can see that for 5 mm particles, the deposition loss rate ranges between 1.11 and 2.23 h1. This observation confirms that the ventilation strategy acts on the particle deposition. The lowest value is given by
Previous experimental studies report the particle deposition velocity Vde which is usually used to evaluate the strength of the particle deposition on surfaces. This parameter is linked to the deposition loss rate lde by the following equation: V de ¼
lde V , 3600 S
(16) 1
where Vde is the particle deposition velocity (m s ), lde represents the particle deposition loss rate (h1), V is the volume of the room (m3) and S is the total area of the room (m2). As we know the quantity of particles extracted through the outlet at each time step, the deposition loss rate lde can be deduced from simulation, and thus the deposition velocity can be determined according to the ventilation strategy.
ARTICLE IN PRESS J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
4891
Table 3 Numerical deposition loss rates ðr40:97Þ as a function of particle size and ventilation strategy Strategy
Bottom–Top 1
Top–Bottom
Top–Top
Ventilation rate (h )
0.5
1.0
0.5
1.0
0.5
1.0
Deposition loss rate (h1) (5 mm particles) Deposition loss rate (h1) (10 mm particles)
1.30 4.68
1.13 4.24
1.62 4.65
2.23 5.49
1.11 4.81
1.38 4.47
Fig. 5. Particle deposition velocity versus particle size and ventilation strategy (previous experimental results reported by Lai, 2002).
Fig. 5 shows both the previous studies values, as reported by Lai (2002), and our experimental and numerical results. Scattered values from the other studies are due to multiple experimental conditions (furnishing, room geometry, roughness of surfaces, ventilation rate, etc.). In the present study, we modelled our experimental conditions and obtained a good agreement, especially for the lowest ventilation rate. For 5 mm particles, choosing appropriate locations of the inlet and outlet can increase the deposition velocity by a factor 1.5 and 2 for ventilation rates of 0.5 and 1.0 h1 respectively. For 10 mm particles, these factors are close to 1. Thus, doubling the ventilation rate is not enough to increase the deposition velocity for 10 mm particles. The smaller the particles are, the more important the ventilation strategy is, on particle deposition velocity.
5. Conclusion By measuring the particle concentration evolution in a mechanically ventilated room, the present study brought
new experimental data that were needed to improve our knowledge of the influence of the ventilation strategy on the particle pollution. It would be necessary to increase the number of measurements for each case in order to lower the remaining deposition loss rate uncertainty. However, as confirmed by the numerical results, the particles removal from indoor air depends on the airflow rate but, it depends on the airflow pattern within the room as well. We found that the influence of the inlet/ outlet locations is stronger for fine particles (lower than 5 mm in diameter) than for coarse particles and that an increase of the ventilation rate does not necessarily lead to higher deposition. The choice of a ventilation strategy, i.e. the airflow rate as well as the inlet/outlet locations, has to be carefully taken into account in order to limit the particle pollution in rooms. The numerical model used in this study seems to be well adapted to predict indoor particle air quality in buildings but is time consuming for the small particles. Limits of the Lagrangian method have to be investigated according to particle size; since the gravity effect becomes negligible as the particle size decreases, an Eulerian model could be well adapted. The benefit of the numerical model described in this paper lies in the knowledge of the deposited particle and high concentration locations. Consequently, the particle deposition velocity can be calculated for each surface of the room (floor, ceiling and vertical walls etc.) which could help better understanding the effects of the ventilation strategy on the particle loss rate. This investigation will be detailed in a further paper.
Acknowledgements This study is part of the Ph.D. dissertation ‘‘Study of the impact of particle pollution on indoor air quality in urban sites’’ co-financed by the Agency of the Environment and the Energy Management (ADEME) and the Poitou-Charentes Region. The authors wish to express their thanks to the CINES (Centre Informatique National de l’Enseignement Supe´rieur, France); computations were carried out on the IBM computers of the CINES. The authors are grateful to Marc Abadie for the helpful editorial comments.
ARTICLE IN PRESS 4892
J. Bouilly et al. / Atmospheric Environment 39 (2005) 4885–4892
References Abadie, M., Limam, K., Allard, F., 2001. Indoor particle pollution: effect of wall textures on particle deposition. Building and Environment 36, 821–827. Abt, E., Suh, H.H., Catalano, P., Koutrakis, P., 2000. Relative contribution of outdoor and indoor particle sources to indoor concentrations. Environmental Science and Technology 34, 3579–3587. AIVC 44, 1994. An analysis and data summary of the AIVC’s numerical database. Technical note AIVC (Air Infiltration and Ventilation Centre). Fogh, C.L., Byrne, M.A., Roed, J., Goddard, A.J.H., 1997. Size specific indoor aerosol deposition measurements and derived I/O concentration ratios. Atmospheric Environment 31, 2193–2203. Harlow, F.H., Welch, J.E., 1965. Numerical calculation of time dependent viscous incompressible flow of fluid with free surface. Physics of Fluids 8, 2182–2189. Hinds, W.C., 1982. Aerosols Technology, Proprieties, Behaviour and Measurement of Airborne Particles. WileyInterscience Publication, New York. Jamriska, M., Morawska, L., Clark, B.A., 2000. Effect of ventilation and filtration on submicrometer particles in an indoor environment. Indoor Air 10, 19–26. Jiang, Y., Be´ghein, C., Chen, Q., 2002. Study of particle dispersion in buildings with large eddy simulation. Proceedings of Indoor Air 2002, 30th June–5th July 2002. Monterey, CA. Lai, A.C.K., 2002. Particle deposition indoors: a review. Indoor Air 12, 211–214. Mundt, E., 2001. Non-buoyant pollutant sources and particles in displacement ventilation. Building and Environment 36 (1), 829–836.
Nazaroff, W.W., Gadgil; A.J., Weschler, C.J., 1993. Critique of the use of deposition velocity in modeling indoor air quality. Modeling Indoor Air Quality and Exposure, STP 1205, American Society for Testing and Materials, Philadelphia, PA, pp. 81–104. Nomura, Y., Hopke, P.K., Fitzgerald, B., Mesbah, B., 1997. Deposition of particles in a chamber as a function of ventilation rate. Aerosol Science and Technology 27, 62–72. Reist, P.C., 1993. Aerosol Science and Technology, second ed. McGraw-Hill, Inc., New York. Roed, J., Goddard, A.J.H., 1990. Ingress of radioactivity into dwellings. Proceedings of Methods and Codes for Assessing Off-site Consequences of Nuclear Accidents, Athens, 13013, pp. 433–450. Snyder, W.H., Lumley, J.L., 1971. Some measurements of particle velocity autocorrelation functions in a turbulent flow. Journal of Fluid Mechanics 48 (1), 41–71. Thatcher, T.L., Lai, A.C.K., Moreno-Jackson, R., 2002. Effects of room furnishings and air speed on particle deposition rates indoors. LBNL Report 48414. Atmospheric Environment 36, 1811–1819. Wallace, L.A., 1996. Indoor particles: a review. Journal of Air Waste Management Association 46, 98–126. William, H., 1992. Numerical Recipes in Fortran : the Art of Scientific Computing. Cambridge University Press, Cambridge, England. Zhang, W., Chen, Q., 2000. Large eddy simulation of indoor airflow with a filtered dynamic subgrid scale model. International Journal of Heat and Mass Transfer 43, 3219–3231. Zhao, B., Zhang, Y., Xianting, L., Xudong, Y., Dongtao, H., 2004. Comparison of indoor aerosol particle concentration and deposition in different ventilated rooms by numerical method. Building and Environment 39, 1–8.