Professional Documents
Culture Documents
Building and Environment: Jinliang Wang, Tin-Tai Chow
Building and Environment: Jinliang Wang, Tin-Tai Chow
a r t i c l e i n f o a b s t r a c t
Article history: This paper introduces a numerical simulation model for investigating the influence of moving subjects
Received 7 February 2011 on the dispersion and deposition of expiratory droplets, rather than on the dispersion of surrogate
Received in revised form gaseous counterparts generally adopted in related research works. In our work, the Lagrangian discrete
4 April 2011
trajectory model is used for tracing the motion of droplets, the Eulerian RANS method is used for solving
Accepted 5 April 2011
the airflow field, and the dynamic mesh model for describing the human movement. The model vali-
dation was performed through result comparisons with published data from literatures. A case study on
Keywords:
the influence of human walking on the dispersion and deposition of expiratory droplets in an airborne
Droplet
Human movement
infection isolation room (AIIR) is then presented. Our findings show that the human walking disturbs the
Dynamic mesh local velocity field with wake formation. The increase of walking speed could effectively reduce the
Lagrangian model overall number of suspended droplets, which may have a positive impact on releasing the infection risk
Isolation room of health workers in AIIR.
Ó 2011 Elsevier Ltd. All rights reserved.
0360-1323/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.buildenv.2011.04.008
1994 J. Wang, T.-T. Chow / Building and Environment 46 (2011) 1993e2002
of droplets from coughing and speaking are 13.5 mm and 16.0 mm.
Meanwhile, it is also believed that droplets larger than 20 mm will
rapidly deposit onto surfaces [28], while droplets between 0.5 and
20 mm will stay in the air for a much longer time [29]. Considering
all these, four size groups of expiratory droplets i.e. 0.5 mm, 5 mm,
10 mm and 20 mm have been chosen in this study, as detailed in the
following sections. In addition, since droplets initially exhaled by
patients are composed of inner dry droplet nuclei and outer moist
coating and mucus, the evaporation effect of droplets may influ-
ence the dispersion and deposition of droplets. Nicas et al. [30]
estimated that the shrinkage time from the original droplets to
droplet nuclei is rapid and is in the order of 0.5 s. The experimental
work of Morawska et al. [26] also showed that non-equilibrium
droplet evaporation is not detectable for particles between 0.5
and 20 mm. Chen and Zhao [31] found that modeling the transient
process from a droplet to a droplet nucleus due to evaporation can
be neglected when the normalized evaporation time is less than
0.051, corresponding to the initial droplet diameter of 100 mm.
Therefore, for the four size groups examined in this study, the
droplet evaporation effect is deemed negligible. Initial velocities of
Fig. 1. Geometric configuration of AIIR.
exhaled droplets also depend on expiratory modes. Gupta et al.
[32,33] indicated that exhaled flow rate over time can be repre-
data [11]. The density of droplets is 600 kg/m3, according to Zhao
sented as a sinusoidal function for breathing, a constant for talking
et al. [10]. The coughing or sneezing lasting time is assumed 0.5 s.
and a combination of gamma-probability-distribution functions for
Four droplet sizes i.e. 0.5 mm, 5 mm, 10 mm and 20 mm were adopted,
coughing. Zhu et al. showed that the peak cough velocity varied
as explained in the Introduction section. The overall number of
from 6 to 22 m/s [11]. Chao et al. [27] reported that the average
droplets coughed or sneezed out is 10,000, with equal allocation to
expiration air velocity was 11.7 m/s for coughing and 3.9 m/s for
each size.
speaking. The differences of initial velocities reported by different
In our investigation, the walking man is initially positioned at
researchers can be mainly attributed to different measurement
0.75 m away from the wall with the air supply. His straight line
techniques. In fact, droplets are aerosols; some of their physical
walking along z direction is assumed to be back and forth for one
characteristics such as gravitational deposition and inertia are very
time, during which the overall walking distance is 5.0 m. He will
much different from those of the gaseous counterparts. Direct using
remain stationary. Three steady walking speeds of 0.25 m/s, 0.5 m/s
surrogate gaseous counterparts in studying the influence of indoor
and 1.0 m/s are respectively investigated in this study.
human walking on exhaled droplets may lead to findings much
deviated from the real situation.
Airborne infection isolation room (AIIR) is a specific place where
inpatients infected with highly airborne infectious diseases are 3. Mathematical model
treated. In order to better prevent the transmission of airborne
diseases in AIIR, a full understanding of the impact of human 3.1. Governing equations and numerical methods for airflow
walking on the dispersion and deposition of expiratory droplets is
very important. The following introduces a new computational Indoor airflow is generally turbulent. There can be three simulation
fluid dynamics (CFD) approach that models the dispersion and approaches, namely DNS (Direct Numerical Simulation), LES (Large
deposition of expiratory droplets in the presence of human Eddy Simulation) and RANS (ReynoldseAveraged NaviereStokes)
movement. The model accuracy was verified by result comparisons simulations. The application of DNS and LES to turbulent simulation
with published data from literatures. This numerical model is requires very demanding computer memory and calculation speed
subsequently used to explore the influence of human walking on [35]. The RANS method with modified standard k e turbulence
the dispersion and deposition of expiratory droplets in AIIR. model was then adopted in this study. The approach is similar to the
Table 1
2. The AIIR model
Physical dimensions of AIIR model.
The physical arrangements of the AIIR investigated in this study Name x-Direction y-Direction z-Direction
width (m) height(m) length (m)
are shown in Fig. 1. Our physical model is relatively the same as the
one in literature [19], except for the geometry of the lying patient. Room 4.0 2.5 4.0
Bed 0.9 0.2 2.0
In our study, this is represented by a rectangular block of 1.73 m
Bed-leg 0.1 0.85 0.1
(height) 0.2 m (thickness) 0.5 m (width), which is the same in Air supply 0.8 0.2 0
the work of Mazumdar et al. [21]. In AIIR the patient is covered Air outlet 0.8 0.2 0
either by sheet or by quilt most of the time. The rectangular Walking man
representation is deemed more realistic than the complex shape Body 0.3 0.675 0.2
used for modeling the walking man. Head 0.15 0.305 0.2
The detailed dimensions of the room, walking man and patient Left hand 0.1 0.575 0.2
Right hand 0.1 0.575 0.2
are given in Table 1. The opening size of the patient mouth [34]
Left leg 0.1 0.75 0.2
when coughing or sneezing is 0.01 0.02 m2. The patient head is Right leg 0.1 0.75 0.2
at 0.15 m from the lower edge of the air outlet. The coughing or
Lying patient
sneezing initial velocity of droplets is taken as 22 m/s, with Body 0.5 0.2 1.73
a vertical upward direction based on Zhu et al.’s PIV experimental
J. Wang, T.-T. Chow / Building and Environment 46 (2011) 1993e2002 1995
Table 2
Initial and boundary conditions for simulation.
work of Shih et al. [19] and Huang and Tsao [36] in AIIR airflow 3.2. Lagrangian discrete trajectory model for droplets
simulation.
The governing equations, including continuity, momentum, Based on the discussions in the Introduction section, the evap-
energy, turbulent kinetic energy k and turbulent dissipation rate e, oration effect from a droplet to a droplet nucleus can be neglected.
can be written in the general format as follows: In this numerical study, the evaporation effect of droplets is not
considered, and the droplets herein are treated as droplet nuclei for
vðrfÞ !
þ V$ðrf V Þ ¼ V$ Gf Vf þ Sf (1) simulation. The CFD tool FLUENT [37] was used in our simulation
vt work. In order to track the motion of droplet after being coughed or
! sneezing out, the Lagrangian discrete trajectory model is adopted.
where r is the air density, V is the velocity vector, f represents each
of the three velocity components (u, v, w), and H is the air enthalpy. The Lagrangian approach tracks particle phase separately through
When f ¼ 1, the equation becomes the continuity equation. Gf is the flow domain by solving the force balance equations of particle
the effective diffusion coefficient for each dependent variable. Sf is movement. In this study, the following assumptions/simplifications
the source term. have been used: (a) the evaporation process of droplets is not taken
In our model, the aforementioned conservation equations are into account; (b) heat and mass transfer between air and droplet is
discretized into algebraic equations by the finite volume method negligible; (c) no particle rebound and resuspension on solid
(FVM). The QUICK scheme is adopted for the discretization of the surfaces, such as walls, floor, ceiling and human surface; (d) coag-
equations and the SIMPLEC algorithm is used to couple pressure ulation effect of droplet is negligible according to the findings of
and velocity. The Boussinesq model is employed to consider the Hinds [38]; (e) all droplets are spherical solid. For a particle, the
buoyancy effect. As for the boundary conditions, the no-slip force balance in the i-direction can be written as:
condition is used for the velocity quantities at the wall; the air
change rate is assumed 12 h1. The air temperature and turbulent dupi ui upi gi $ rp r
¼ þ þ Fai (2)
intensity at the air supply are 295.15 K and 20% respectively; the dt s rp
pressure outlet boundary condition is specified at the air outlet
with a static pressure of 8 Pa and a turbulent intensity of 10%. where ui and upi are respectively the velocity of air and particles
Since the residence time of the walking man is relatively short, his (m/s), gi is the gravitational acceleration (m/s2), s is the particle
body surface is assumed to be adiabatic. All other boundary walls relaxation time [39], rp and r are the density of particle and air
are also assumed to be adiabatic. The only heat source is then the respectively (kg/m3), and Fai is the additional forces (besides drag
patient body with a heat flux of 46.52 W/m2 (corresponding to 0.8 force and gravitation) exerted on unit particle mass in the i-direc-
met). The standard wall function is employed to treat the turbulent tion (N/kg).
flow properties in the near wall regions. The additional force Fai may include pressure gradient force,
Basset force and virtual mass force caused by unsteady flow, Brow-
nian force, Saffman’s lift force due to shear, and thermophoretic
force due to temperature gradient. For room airflow involving fine
Fig. 2. Layout of mesh domain in AIIR. Fig. 3. Experimental case used for validation of Lagrangian discrete trajectory model.
1996 J. Wang, T.-T. Chow / Building and Environment 46 (2011) 1993e2002
Fig. 5. Comparison of transient static pressure distributions at y ¼ 1 m in this study with those in literature [19] (bold straight arrow denotes walking direction).
J. Wang, T.-T. Chow / Building and Environment 46 (2011) 1993e2002 1997
This method can keep the grid size constant in the dynamic zone. 3.4. Validation of mathematical model
The applicability requires all cells adjacent to the moving boundary
zone to be either wedges or hexahedra. In our work, hexahedral Our mathematical model was validated through two separate
cells are generated for the whole grid system. The whole mesh steps: firstly the Lagrangian discrete trajectory model validation
domain is split into two mesh zones: “static” and “dynamic”, as and then the dynamic mesh model validation.
shown in Fig. 2. The interface data exchange between the two is
realized by the grid interface principles for sliding mesh theory. 3.4.1. Lagrangian discrete trajectory model validation
A user defined function (UDF) is dynamically loaded to realize the The experimental case of Lu et al. [41] was employed for result
stipulated back and forth walking style. comparisons. As shown in Fig. 3, the overall dimensions of the two-
Fig. 6. Velocity vector distributions at y ¼ 1 m at different time instants and walking speeds.
Fig. 7. Velocity vector distributions at x ¼ 3 m at different time instants and walking speeds.
Fig. 8. Number of overall suspended droplets at different time instants and walking Fig. 9. Number of overall escaped droplets at different time instants and walking
speeds. speeds.
J. Wang, T.-T. Chow / Building and Environment 46 (2011) 1993e2002 1999
walking speed, the longer is the wake behind the human back.
However, as shown in Fig. 6, the influence of walking speed on the
velocity field at a distance from the walking site appears insignif-
icant; this is in line with literature [19]. This also leads to the
conclusions in literature [19] that the removal of gaseous contam-
inant exhaled by patient is not obviously affected by human moving
speed. Actually, depending on sizes of droplets and initial exhaled
velocity as well as expiratory direction, droplets can behave
differently from gaseous counterparts under the influence of
human walking. As illustrated in the following section, the walking
speed does have remarkable effects on the dispersion and deposi-
tion of expiratory droplets, which is different from the conclusions
given in literature [19].
Fig. 7(a)e(f) show the velocity distributions at x ¼ 3 m at
different time instants and walking speeds. It can be seen that
when the man walks towardtoward the wall with the air outlet,
there exist two recirculation zones: an upper zone close to the
upper back of the human head and a lower zone around the lower
part of the human back (as highlighted by the two curved arrows).
However, when the man walks back towards the wall with the air
inlet, at v ¼ 0.25 m/s the upper recirculation zone is tiny, and hence
this is not highlighted in Fig. 7(b). With the velocity increased to
0.5 m/s and 1.0 m/s respectively, the upper recirculation zone
becomes observable. Notwithstanding this, the lower recirculation
zone is not noticeable in all three cases in Fig. 7(b), (d) and (f).
Fig. 13. Number of escaped droplets with different diameters at different time instants
and walking speeds.
References
[1] Knight J. Researchers get to grips with cause of pneumonia epidemic. Nature
2003;422:547e8.
[2] Li Y, Leung GM, Tang JW, Yang X, Chao CYH, Lin JZ, et al. Role of ventilation in
airborne transmission of infectious agents in the built environmentea
multidisciplinary systematic review. Indoor Air 2007;17:2e18.
[3] Bjørn E, Nielsen PV. Dispersal of exhaled air and personal exposure in
displacement ventilated rooms. Indoor Air 2002;12:147e64.
[4] Micallef Alfred, Caldwell Julieann, Colls JJ. The influence of human activity on
the vertical distribution of airborne particle concentration in confined envi-
ronments: preliminary results. Indoor Air 1998;8:131e6.
[5] Mazumdar S, Chen Q. Impact of moving bodies on airflow and contaminant
transport inside aircraft cabins. In: Proceedings of the 10th International
Conference on Air Distribution in Rooms, (Roomvent 2007); Helsinki; 2007.
p. 165.
[6] Matsumoto H, Hai NL, Ohba Y. Influence of moving object on air distribution in
ventilated rooms. In: Proceedings of the 8th International conference on air
distribution in rooms (Roomvent 2002); Copenhagen; 2002. pp. 261e264.
[7] Matsumoto H, Hai NL, Ohba Y. CFD simulation of air distribution in
displacement ventilated room with a moving object. In: Proceeding of, the 9th
International conference on air distribution in rooms (Roomvent 2004);
Coimbra; 2004. pp. 5e8.
[8] Mattsson M, Sandberg M. Displacement ventilation-influence of physical
activity. In: Proceedings of the 4th International conference on air distribution
in rooms (Roomvent 1994) vol. 2; Krakow; 1994. pp. 78e92.
[9] Berrouk AS, Lai ACK, Cheung ACT, Wong SL. Experimental measurements and
large eddy simulation of expiratory droplet dispersion in a mechanically
ventilated enclosure with thermal effects. Building and Environment 2010;
45:371e9.
[10] Zhao Bin, Zhang Zhao, Li Xianting. Numerical study of the transport of
droplets or particles generated by respiratory system indoors. Building and
Environment 2005;40:1032e9.
[11] Zhu Shengwei, Kato Shinsuke, Yang Jeong-Hoon. Study on transport charac-
teristics of saliva droplets produced by coughing in a calm indoor environ-
ment. Building and Environment 2006;41:1691e702.
[12] Wei Sun, Jie Ji, Li Yuguo, Xie Xiaojian. Dispersion and settling characteristics
of evaporating droplets in ventilated room. Building and Environment 2007;
42:1011e7.
Fig. 14. Number of deposited droplets with different diameters at different time [13] Alani A, Barton IE, Seymour MJ, Wrobel LC. Application of Lagrangian particle
instants and walking speeds. transport model to tuberculosis (TB) bacteria UV dosing in a ventilated
2002 J. Wang, T.-T. Chow / Building and Environment 46 (2011) 1993e2002
isolation room. International Journal of Environmental Health Research 2001; [27] Chao CYH, Wan MP, Morawska L, Johnson GR, Ristovski ZD, Hargreaves M,
11:219e28. et al. Characterization of expiration air jets and droplet size distributions
[14] Qian H, Li Y, Nielsen PV, Hyldgaard CE, Wong TW, Chwang ATY. Dispersion of immediately at the mouth opening. Journal of Aerosol Science 2009;40:
exhaled droplet nuclei in a two-bed hospital ward with three different 122e33.
ventilation systems. Indoor Air 2006;16:111e28. [28] Gold E, Nankervis GA. Cytomegalovirus. In: Evans A, editor. Viral infections of
[15] Varol Y, Oztop HF, Firat M, Koca A. CFD modeling of heat transfer and fluid humans. New York: Plenum Medical Book Co.; 1989. p. 175e6.
flow inside a pent-roof type combustion chamber using dynamic model. [29] McCluskey R, Sandin R, Greene J. Detection of airborne cytomegalovirus in
International Communications in Heat and Mass Transfer 2010;37:1366e75. hospital rooms of immuncompromised patients. Journal of Virological
[16] Miao JM, Ho MH. Effect of flexure on aerodynamic propulsive efficiency of Methods 1996;56:115e8.
flapping flexible airfoil. Journal of Fluids and Structures 2006;22:401e19. [30] Nicas M, Nazaroff WW, Hubbard A. Toward understanding the risk of
[17] Mousavi SM, Zamankhan P, Jafari A. Computer simulations of sodium formate secondary airborne infection: emission of respirable pathogens. Journal of
solution in a mixing tank. Communications in Nonlinear Science and Occupational and Environmental Hygiene 2005;2:143e54.
Numerical Simulation 2008;13:380e99. [31] Chen C, Zhao B. Some questions on dispersion of human exhaled droplets in
[18] Lecrivain Gregory, Slaouti Arezki, Payton Carl, Kennedy Ian. Using reverse ventilation room: answers from numerical investigation. Indoor Air 2010;
engineering and computational fluid dynamics to investigate a lower arm 20:95e111.
amputee swimmer’s performance. Journal of Biomechanics 2008;41:2855e9. [32] Gupta JK, Lin CH, Chen Q. Characterizing exhaled airflow from breathing and
[19] Shih Yang-Cheng, Chiu Cheng-Chi, Wang Oscar. Dynamic airflow simulation talking. Indoor Air 2010;20:31e9.
within an isolation room. Building and Environment 2007;42:3194e209. [33] Gupta JK, Lin CH, Chen Q. Flow dynamics and characterization of a cough.
[20] Tung Yun-Chun, Shih Yang-Cheng, Hu Shih-Cheng. Numerical study on the Indoor Air 2009;19:517e25.
dispersion of airborne contaminants from an isolation room in the case of [34] He Qibin, Niu Jianlei, Gao Naiping, Zhu Tong, Wu Jiazheng. CFD study of exhaled
door opening. Applied Thermal Engineering 2009;29:1544e51. droplet transmission between occupants under different ventilation strategies
[21] Mazumdar S, Yin Y, Guity A, Marmion P, Gulick B, Chen Q. Impact of moving in a typical office room. Building and Environment 2010;46:397e408.
objects on contaminant concentration distributions in an inpatient room with [35] Tao WQ. Numerical heat transfer. 2nd ed. Xi’an: Xi’an Jiao Tong University
displacement ventilation. HVAC&R Research 2010;16(5):545e64. Press; 2001. pp. 333e335.
[22] Wells WF. On air-borne infection study II: droplets and droplet nuclei. [36] Huang J-M, Tsao S-M. The influence of air motion on bacteria removal in
American Journal of Hygiene 1934;20:611e8. negative pressure isolation rooms. HVAC&R Research 2005;11:563e85.
[23] Duguid JP. The size and the duration of air-carriage of respiratory droplets and [37] Fluent 6.1, 2003, User’s Guide. Centerra Research Park, 10 Cavendish Court,
droplet-nuclei. Journal of Hygiene 1946;44:471e9. Lebanon, NH 03766, USA: Fluent Inc.
[24] Loudon RG, Roberts RM. Droplet expulsion from the respiratory tract. Amer- [38] Hinds WL. Aerosol technology. 2nd ed. New York: Wiley-Interscience; 1999.
ican Review of Respiratory Disease 1967;95:435e42. [39] Lai ACK, Nazaroff WW. Modeling indoor particle deposition from turbulent
[25] Papineni RS, Rosenthal FS. The size distribution of droplets in the flow onto smooth surfaces. Journal of Aerosol Science 2000;31:463e76.
exhaled breath of healthy human subjects. Journal of Aerosol Medicine [40] Zhao Bin, Zhan Ying, Li Xianting, Yang Xudong, Huang Dongtao. Comparison
1997;10:105e61. of indoor aerosol particle concentration and deposition in different ventilated
[26] Morawska L, Johnson GR, Ristovski ZD, Hargreaves M, Mengersen K, Corbett S, rooms by numerical method. Building and Environment 2004;39:1e8.
et al. Size distribution and sites of origin of droplets expelled from the human [41] Lu W, Howarth AT, Adam N, Riffat S. Modeling and measurement of airflow
respiratory tract during expiratory activities. Journal of Aerosol Science 2009; and aerosol particle distribution in a ventilated two-zone chamber. Building
40:256e69. and Environment 1996;31:417e23.