Professional Documents
Culture Documents
Abstract
In this paper, the transport characteristics of saliva droplets produced by coughing are examined in a calm indoor environment.
Three subjects are studied, with results indicating that more than 6.7 mg of saliva is expelled at speeds of up to 22 m/s during each
individual cough, and that saliva droplets can travel further than 2 m. In addition, the dispersion processes of saliva droplets of
different diameters expelled during coughing are analyzed using the Lagrangian equation. The results indicate that the transport
characteristics of saliva droplets due to coughing change with size. The effects of gravity or inertia on droplets of 30 mm or less are
negligible due to their small sizes, and therefore their transport is mostly influenced by the indoor flow field. Droplets of 50–200 mm,
which are significantly affected by gravity, fall as the flow-field weakens. Droplets of 300 mm or more, which are affected more by
inertia than gravity, fall difficultly. Moreover, the analytical results also indicate that the droplets’ transport is greatly influenced by
the spatial relationship between the air-conditioner and the subjects. Finally, based on the experimental and analytical results,
droplet infection by saliva droplets due to coughing is examined.
r 2005 Elsevier Ltd. All rights reserved.
Keywords: Subject experiment; Computational fluid dynamics (CFD); Lagrangian equation; Particle image velocimetry; Saliva droplets
0360-1323/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.buildenv.2005.06.024
ARTICLE IN PRESS
1692 S.W. Zhu et al. / Building and Environment 41 (2006) 1691–1702
was observed. In the numerical analysis, we firstly boundary conditions for the following numerical simu-
analyzed the indoor flow field assuming coughing and lations.
respiration to be steady phenomena, and using the
maximum initial velocity of the coughed airflow 2.2.1. Outline of the PIV system
determined by the PIV experiment as the boundary The PIV system used in this study is outlined in Table
condition. We subsequently analyzed the transport 2. As shown in Fig. 1, a cross section of a flow field is
process for droplets with diameters of 30, 50, 100, 200, irradiated at very short intervals using a high-powered
300, and 500 mm using the Lagrangian equation to show pulsed laser, and the image frames of the particles are
how saliva droplets were expelled from the mouth by recorded as digital data by a CCD camera synchronized
coughing, dispersed in the air, and finally attached to with the laser. The wind velocity vectors are then
surfaces. Based on the results, the transport character- calculated using the pattern correlation method by
istics of saliva droplets produced by coughing are detecting the peak of the correlation coefficient, which
investigated in a calm indoor environment. corresponds to the amount of particles’ movement [7].
Three healthy male graduate students were used as 2.2.3. Outline of PIV experiment
subjects in the experiments. The physiological data of During the experiment, the measurement area of the
the subjects are listed in Table 1. PIV system was kept in a calm indoor environment,
while the subjects coughed through the opening into the
2.2. Measurement of initial velocity of coughed airflow measurement area. As shown in Table 3, the pulse
interval was set to 0.05 ms in accordance with the
The initial velocity of the coughed airflow was measurement area and maximum wind velocity. Addi-
measured using a PIV system. The result was used as tionally, in order to remove the erroneous vectors in the
resulting vector diagram, velocity-range validation,
Table 1 peak-height validation and moving average validation
Physiological data of subjects were carried out.
Subject Age Weight (Kg) Height (m) Lung capacity (L)
2.2.4. Measurement results
A 28 68 1.75 4.3 Because the PIV system used was incapable of
B 29 67 1.74 3.6
continuous measurement, the flow field around the mouth
C 30 70 1.68 3.5
of the subject was measured repeatedly at intervals of
ARTICLE IN PRESS
S.W. Zhu et al. / Building and Environment 41 (2006) 1691–1702 1693
Table 2
Make-up of PIV System
Table 4
Mass of coughed saliva
Fig. 6. Room model analyzed: (1) Case 1 and Case 2; (2) Case 3; (3) Case 4.
Table 5
Cases analyzed
1 Ordinary conference room Sitting+sitting Wall by the side of the coughing person
2 Wall opposite to the coughing person
3 Ordinary bedroom Supine+standing Wall behind the bed
4 Wall opposite to the standing person
(CFD), assuming the unsteady airflows produced by constantly. In case 2, the human model on the side
coughing and respiration to be steady. The resulting away from the air-conditioner was assumed to cough
dispersion process of the saliva droplets due to coughing consecutively, and the human model closer to the air-
was subsequently analyzed using the Lagrangian equa- conditioner was assumed to inhale constantly. In cases 1
tion in the flow field based on the steady models. and 2, only half the room area was simulated because of
Because the steady models assign a larger momentum to the spatial symmetry in the Y-direction.
the air than in reality, care should be exercised when In cases 3 and 4 (Figs. 6(2)–(3), Table 5), the room
comparing the analytical results with the actual situa- was assumed to be an ordinary bedroom, with a bed
tion. Moreover, to regenerate the warm ascending air placed in the center of the room adjacent to a wall, the
stream around the human body caused by metabolic human model lying on the bed was assumed to cough
heat, human models of complex shape (including consecutively, and another human model standing next
detailed features such as limbs, chests, and jaws), were to the bed was assumed to inhale constantly. An
used in the numerical analyses. air-conditioner is set on the wall behind the bed in
case 3 and on the wall facing the standing human model
3.1. Room model in case 4.
In the model of consecutive coughing, the maximum
Four cases were analyzed for an air-conditioned room velocity of the coughed airflow (22 m/s) obtained in the
(3.5 m 3.0 m 2.5 m) with an ordinary wall-mounted aforementioned PIV experiment was used. The steady
air-conditioner. The air-conditioner had an inlet open- inhalation model assumed a constant inhalation rate of
ing with an area of 0.035 m2 (0.05 m 0.7 m), from 0.0004 m3/s through the nose.3 Moreover, the area of the
which air flowed horizontally at 26 1C at a speed of mouth opening of the sitting and supine human models
0.4 m/s. was 3.5 cm2, and that of the nose opening of the sitting
In cases 1 and 2 (Fig. 6(1), Table 5), the room was and standing human models was 3.3 cm2 [9].
assumed to be an ordinary conference room, with a desk
placed in the center of the room, and two human models
seated on either side of the desk. The distance between 3
the mouths of the two human models was about 1.54 m. In this study, the respiration phenomenon was assumed to be
steady inhalation. The inhaled volume was determined from the
In case 1, the human model closer to the air-conditioner metabolic rate of 1.3 met [13]. Thus, as shown in Fig. 15, we assumed
was assumed to cough consecutively, and the human that a person would inhale constantly at 10 l/min (inhalation of 10
model on the other side was assumed to inhale times/min and a volume of 1 l per inhalation).
ARTICLE IN PRESS
1696 S.W. Zhu et al. / Building and Environment 41 (2006) 1691–1702
Table 6 Table 7
Numerical methods Boundary conditions
Turbulence model Standard k– e model Supply opening Velocity: 0.40 m/s, Temperature:
Algorithm SIMPLE 26 1C,
Space scheme Velocity terms: MARS; scalar Turbulence intensity: 10%,
terms: first order up wind Turbulence scale: 0.05 m
Model for particle simulation Lagrangian model Size: 0.05 m 0.7 m
3.3. Results
Fig. 9. Scalar velocity distrbution in Case 3 (m/s): (1) in region A(enlarged); (2) in section ABCD; (3) in section EFGH.
Fig. 10. Scalar velocity distrbution in Case 4 (m/s): (1) in region A(enlarged); (2) in section ABCD; (3) in section EFGH.
impacted on the face and neck of the person opposite, dispersion of saliva droplets of 300 mm in diameter was
while the remainder passed by and then dropped to the limited to the area above the pillow (Figs. 13(2)–(5) and
floor. Figs. 11(3) and 12(3) show that among the saliva 14(2)–(5)). For the droplets of 500 mm in diameter, their
droplets of 100 mm in diameter, more than 80% velocity did not decrease significantly after being
impacted on the lower neck area of the person opposite, expelled from the mouth, and they impacted directly
while the remainder dropped to the floor soon after with the ceiling (Fig. 13(6) and 14(6)). Moreover, in case
passing their body. Figs. 11(4) and 12(4) show that the 3, saliva droplets of 50–300 mm fell some way from the
saliva droplets of 200 mm in diameter started to fall standing person because of the effect of the warm
before reaching the person opposite, and nearly 100% of ascending airflow resulting from natural convection. On
these droplets impacted on the chest of the person the other hand, in case 4, droplets of 50–300 mm fell
opposite. However, as shown in Figs. 11(5) and 12(5), closer to the standing person because of the effect of the
the saliva droplets of 300 mm or more in diameter did air-conditioner’s airflow.
not descend significantly, and nearly 100% of them
impacted on the neck of the opposite person. Figs. 11(6) 3.4. Discussion on transport characteristics of saliva
and 12(6) show that most saliva droplets of 500 mm in droplets due to coughing
diameter impacted on the face of the opposite person
with little or no observable gravitational effect. Based on the above results, we examined the transport
The results of cases 3 and 4 are shown in Figs. 13 and characteristics of coughed saliva droplets after
14, respectively. As shown in cases 1 and 2, although the affirming the feasibility of the assumption of consecutive
saliva droplets of 30 mm in diameter were dispersed coughing. Moreover, the influence of the indoor
throughout the room and adhere randomly to the ventilation situation on the droplets’ transport and
surfaces (Figs. 13(1) and 14(1)), the extent of dispersion the droplet infection via coughed droplets will be
narrowed in droplets of 50 mm or more, and the investigated.
ARTICLE IN PRESS
S.W. Zhu et al. / Building and Environment 41 (2006) 1691–1702 1699
Fig. 11. Saliva droplets’ dispersion (simulation results of Case 1): (1) D ¼ 30 mm; (2) D ¼ 50 mm; (3) D ¼ 100 mm; (4) D ¼ 200 mm; (5) D ¼
300 mm; (6) D ¼ 500 mm.
Fig. 12. Saliva droplets’ dispersion (simulation results of Case 2): (1) D ¼ 30 mm; (2) D ¼ 50 mm; (3) D ¼ 100 mm; (4) D ¼ 200 mm; (5) D ¼
300 mm; (6) D ¼ 500 mm.
ARTICLE IN PRESS
1700 S.W. Zhu et al. / Building and Environment 41 (2006) 1691–1702
1l
2.5s 0.5s 2.5s 0.5s 2.5s 0.5s Time (s)
Table 8
Numbers of saliva droplets adhering to standing human body and outlet opening in cases 3 and 4
saliva droplets’ distribution limited to a small area or more. The results of the weight assessment of the
around the supine person’s head. Moreover, because the coughed saliva indicate that wearing a mask is a very
air-conditioner was on the wall facing the standing effective way of preventing droplet infection of
person in case 4, the coughed airflow rose closer to the infectious respiratory diseases via saliva droplets
standing person under the influence of the supplied due to coughing.
airflow of the air-conditioner, and saliva droplets of (2) We used numerical analyses to investigate the
50–200 mm in diameter fell closer to the standing person. transport characteristics of the saliva droplets due
Table 8 shows the numbers of the saliva droplets per to coughing of each size in a calm indoor environ-
size, which adhered to the standing person and the ment. Since droplets of 30 mm or less in diameter are
outlet opening of the air-conditioner, for cases 3 and 4. very small, the effect of gravity or inertia was
The results clearly indicate that the number of the saliva negligible, and their transport was mostly influenced
droplets that adhered to the standing person in case 4 by the indoor flow field. Saliva droplets of
was greater than that in case 3 for droplet diameters of 50–200 mm, which were significantly affected by
30–200 mm. The number of the saliva droplets that gravity, fell as the flow field weakened. The droplets
adhered to the outlet opening of the air-conditioner was of 300 mm or more were affected more by inertia
greater in case 3 than case 4 for droplet sizes of than gravity. The droplets of 500 mm traveled almost
50–300 mm because the air-conditioner was attached to straight and impacted on the first opposite object
the wall above and behind the supine person in case 3. It within their effective travel range.
is notable that more than 50% of the droplets of 300- (3) Analytical results of cases 1 and 2 indicate that there
mm-diameter were forced out from the outlet opening in was a high risk of droplet infection within short
case 3. The above results indicate that the spatial distances for infectious respiratory diseases, via the
relationship between the air-conditioner and the occu- saliva droplets produced by coughing. Analytical
pants greatly affected the transport of the saliva droplets results of cases 3 and 4 indicate that the spatial
due to coughing. It also implies that droplet infection of relationship between the air-conditioner and the
infectious respiratory diseases can be greatly affected by occupants greatly affects the transport of saliva
the position of the air-conditioner in the room. droplets due to coughing.
(4) In future studies, we plan to perform transient
simulation by modeling actual coughing phenomena
4. Conclusions and further clarify the nature of material transport
by coughing.
In this study, the transport characteristics of the saliva
droplets due to coughing were studied in a calm indoor
environment in detail through subject experiment and Acknowledgements
numerical analysis. Based on the experimental and
analytical results, droplet infection by saliva droplets This study contains part of the results obtained by a
due to coughing was discussed. Moreover, human research task force ‘‘Study on SARS Prevention in View
models of complex shape, including detailed features, of Sanitation in Buildings’’ (conducted by Susumu
were used for the numerical analysis. The primary Yoshizawa, Part-Time Lecturer of Aichi Shukutoku
conclusions of this study are as follows. University) in a Health Science Special Research Project
‘‘Urgent Research on SARS.’’ The present study was
(1) According to the results of the subject experiments, a also partly supported by Grant-in-Aid for Scientific
total of more than 6.7 mg of saliva was expelled on Research (Exploratory Research), Ministry of Educa-
aggregate at a maximum velocity of 22 m/s in each tion, Culture, Sports, Science and Technology (Shinsuke
cough, and the affected area extended as far as 2 m Kato, as a representative researcher). The authors would
ARTICLE IN PRESS
1702 S.W. Zhu et al. / Building and Environment 41 (2006) 1691–1702
like to express sincere thanks to all those who have [8] Hayashi T, Ishizu Y, Kato S, Murakami S. CFD analysis on
supported their study. characteristics of contaminated indoor air ventilation and its
application in the evaluation of the effects of contaminant
inhalation by a human occupant. Building and Environment
2002;37:219–30.
References [9] Zhu SW, Hayashi T, Kato S, Murakami S. Investigation of flow
field in human’s respiration area in a calm environment by
[1] Leith DE, Bulter JP, Sneddon SE, Brain JD, Cough M. In: visualization experiment and numerical analysis. Journal of
Handbook of Physiology. The Respiratory System. Mechanics of Environmental Engineering (Transactions of AIJ) 2004;
Breathing, vol. III. Bethesda, MD: American Physiological 583:37–42.
Society; 1986. p. 315–36 (Sec. 3,Chapter 20). [10] van Leer B. Towards the ultimate conservation difference scheme,
[2] Irwin RS, Curley FJ, French CL. Chronic cough. The spectrum V: a second-order sequel to Godunov’s method. Journal of
and frequency of causes, key components of the diagnostic Computational Physics 1979;32:101–36.
evaluation, and outcome of specific therapy. American Review of [11] Xu ZL. The theory of air cleaning technology. Tongji University
Respiratory Disease 1990;141:640–7. Press; 1998.
[3] Tibbling L, Gibellino FM, Johansson K-E. Is mis-swallowing or [12] Jiang Y, Li XF, Zhao B, et al. SARS and ventilation. In:
smoking a cause of respiratory symptoms in patients with gastro- Proceedings of the fourth international symposium on HVAC.
oesophageal reflux disease? Dysphagia 1995;10:113–6. vol. K04(1). 2003. p. 27–36
[4] Johnson WE, Hagen JA, et al. Outcome of respiratory symptoms [13] Nishi Y, et al. Mechanism of thermal comfortable environment—
after anti-reflux surgery on patients with gastro-oesophageal aim at a plentiful living space. Society of Heating, Air-
reflux disease. Archives of Surgery 1996;131:489–92. conditioning and Sanitary Engineers of Japan; 1997.
[5] Ekstrom T, Johansson K-E. Effects of anti-reflux surgery on [14] Yang JH, Kato S, Murakami S, Kasahara H, Hayashi T.
chronic cough and asthma in patients with gastro-oesophageal Numerical analysis on air quality of human’s inhaled air in room
reflux disease. Respiratory Medicine 2000;94:1166–70. with infiltration, E02-3. In: Proceedings of the 15th symposium on
[6] Knight J. Researchers get to grips with cause of pneumonia computational fluid dynamic. 2001
epidemic. Nature 2003;422:547–8. [15] Murakami S, Kato S, Zeng J. Combined simulation of airflow,
[7] Raffel M, Willert CE, Kompenhans J. Particle Image Velocime- radiation and moisture transport for heat release from human
try. Berlin, Heidelberg: Springer; 1998. body. ROOMVENT’98 1998;2:141–50.