See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/225558396

Study of cooling system with water mist sprayers: Fundamental examination

of particle size distribution and cooling effects

Article  in  Building Simulation · September 2008

DOI: 10.1007/s12273-008-8115-y

CITATIONS READS

21 2,982

4 authors, including:

Gyuyoung Yoon
Nagoya City University
29 PUBLICATIONS   97 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Gyuyoung Yoon on 21 April 2015.

The user has requested enhancement of the downloaded file.

Build Simul (2008) 1: 95 – 110
DOI 10.1007/s12273-008-8115-y
RESEARCH ARTICLE

Study of Cooling System with Water Mist Sprayers:

Fundamental Examination of Particle Size Distribution and Cooling Effects

Hideki Yamada1, Gyuyoug Yoon1( ), Masaya Okumiya1, Hiroyasu Okuyama2

1. Graduate School of Environmental Studies, Nagoya University, Furo-cho, Chikusa-ku, Nagoya-shi, Aichi 464-8603, Japan
2. Institute of Technology, Shimizu Corporation, 3-4-17, Etchujima, Koto-ku, Tokyo 135-8530, Japan

Received: 23 November 2007 / Revised: 25 April 2008 / Accepted: 16 May 2008

© Tsinghua Press and Springer-Verlag 2008

Abstract A cooling system that sprays micro water droplets could prove useful in mitigating
temperature increases in urban areas by using the heat of water evaporation, a process that consumes
only small amounts of water and energy. If water mist is sprayed in a semi-outdoor area, for example,
under a canopy, it could potentially improve conditions on hot days. However, there is little reference
data concerning the design or control of such systems. In order to propose a method for designing and
predicting the performance of a water mist system, we discuss differences in cooling effects in the
context of particle size distribution of water mist. The results of numerical fluid analysis showed there
is no significant difference in temperature reduction for different particle sizes. However, the water
particles remained in a lower position with larger particles.

Keywords water mist, evaporative cooling, urban heat island, numerical fluid analysis

List of symbols

ags solar absorptance at ground surface (—) m mass flow rate (kg/s)
A surface area (m2) M molecular weight (kg/mol)
c heat capacity (J/(kg·K)) n spread parameter of a Rosin-Rammler
d droplet diameter (m) distribution (—)
dL ligament diameter (m) N molar flux of vapor (mol/(m2·s))
dinj injector orifice diameter (m) Oh Ohnesorge number
d0 most probable droplet size (m) p vapor pressure (Pa)
dug depth (m) qcond heat diffusion to the underground (W/m2)
qconv heat convection and long-wavelength radiation
d size constant (m)
(W/m2)
fD drag force coefficient (s−1)
qsol transmissive solar radiation (W/m2)
fx additional force coefficient (m/s2)
R universal gas constant (Pa·m3/(mol·K))
F momentum (N·m/s)
t time (s)
h film thickness (m)
T temperature (K)
J horizontal solar radiation (W/m2) u velocity (m/s)
km mass transfer coefficient (m/s) v axial component of sheet velocity (m/s)
Kb the most unstable wavelength (s−1) V total sheet velocity (m/s)
l latent heat (J/kg) Xi local bulk mole fraction of species i (—)
Lb breakup length (m) Yd mass fraction of droplets with diameter greater
m mass (kg) than d (—)
E-mail: yoonq4@hotmail.com α heat transfer coefficient (W/(m2·K))
2 Build Simul

η wave amplitude (m) 2.1 Discrete phase model

θ spray angle measured from spray axis (rad)
λ thermal conductivity (W/(m·K)) We adopted the discrete phase model (DPM) that it allows
ρ density (kg/m3) to simulate a discrete second phase in a Lagrangian frame
τ solar transmittance (—) of reference in addition to solving transport equations for the
Ω growth of the most unstable wave (s−1) continuous phase. Also, DPM can compute the trajectories
of the discrete phase entities, as well as momentum transfer,
Subscripts
heat and mass transfer to/from them (Fig. 1).
0 at the injector exit
air air property
b sheet breaks up
g gas property
gs ground surface
l liquid property
L liquid ligament
i vapor species
p particle
rf roof surface
sat saturated condition
ug under ground

1 Introduction Fig. 1 Diagram of discrete phase model

In recent years, warming of urban areas in summer has Momentum exchange

become a problem known as the “urban heat island effect”.
One means of mitigating this effect is to spray micro water The momentum transfer from the continuous phase to the
droplets. This method suppresses the temperature rise in discrete phase is computed by examining the change in
urban areas by using the heat of water evaporation, while momentum of a particle as it passes through each control
using only small amounts of water and energy. If water volume. The trajectory of a discrete phase particle is
mist is sprayed in a semi-outdoor area, for example, under predicted by integrating the force balance on the particle,
a canopy, it could potentially improve conditions on hot which is written in a Lagrangian reference frame. This
days. Through a field experiment (Yamada et al. 2006), we momentum balance is given by
have verified the effectiveness of this method and confirmed
a temperature reduction under a canopy of up to about 3℃. du p ⎛ g ( ρ p − ρg ) ⎞
Such water mist systems are expected to be used increasingly Fp = mp Δt = ⎜ f D (ug − up ) + + f x ⎟ m p Δt , (1)
dt ⎜ ρp ⎟
in the future, yet there is little reference data concerning ⎝ ⎠
the design or control of such systems. Therefore, this here, f x is the additional acceleration (force/unit particle
study proposes a method for designing and predicting the
mass), f D (ug − up ) is the drag force per unit particle mass
performance of water mist systems. In this study, we
term.
conducted a numerical fluid analysis and then examined the
particle size distribution and the cooling effect of water mist. Heat and mass exchange

2 Simulation Outline The heat exchange according to convective and latent heat
transfer is given by
In this study, the simulations were conducted by using a dTp dmp
CFD (computational fluid dynamics) code Fluent. And, we mp cp = α Ap (Tg − Tp ) + l. (2)
used discrete phase model (DPM) for calculating momentum, dt dt
heat and mass exchange between the water mist and the air. The mass of droplet is reduced according to
Also, we adapted the pressure-swirl atomizer model in this
simulation. DPM and the pressure-swirl atomizer model dmp
= − Ni Ap M i , (3)
have been mounted in Fluent basically. In Section 2.1 and dt
2.2, we described these models by quoting the user’s manual
(Fluent 2006). And, we presented calculation conditions of where Ni is the molar flux of vaporize species, H2O in this
simulations in Section 2.3. study, given by
Build Simul (2008) 1: 95 – 110 3

⎛ psat (Tp ) pg ⎞ The total velocity V is determined by the fluid pressure

Ni = km ⎜ − Xi ⎟. (4) and the injector orifice diameter.
⎜ RT RTg ⎟⎠
⎝ p

Sheet breakup
The evaporation of water mist is computed by this mass
exchange and latent heat transfer. The wavy disturbances occur on the liquid film; the sheet
breaks up and ligaments will be formed at a length given by
2.2 Atomizer model and droplet size distribution
V ⎛η ⎞
Lb = ln ⎜ b ⎟ , (7)
Atomizer model Ω ⎝ η0 ⎠

In this study, the pressure-swirl atomizer is used. This where the quantity ln(ηb / η0 ) is an empirical sheet constant.
atomizer accelerates the liquid through nozzles known as The default value of 12 was obtained theoretically
swirl ports into a central swirl chamber. As shown in Fig. 2, (Dombrowski and Hooper 1962). Ω is the maximum
the swirling liquid pushes against the walls of the swirl growth rate of wave that occurred into the liquid film.
chamber and develops a hollow air core. It then emerges
Atomization
from the orifice as a thinning sheet, which is unstable,
breaking up into ligaments and droplets (Schmidt et al. If it is assumed that the ligaments are formed from tears in
1999). The most probable droplet size spout atomizer is the sheet twice per wavelength, the resulting diameter is
calculated by a series of following equations. given by
Film formation 8hb
dL = , (8)
The centrifugal motion of the liquid within the injector Kb
creates an air core surrounded by a liquid film. The thickness (dinj − h0 )h0
of this film, h0 , is given by hb = . (9)
⎛ ⎛θ ⎞⎞
⎜ dinj − h0 + 2 Lb sin ⎜ 2 ⎟ ⎟
1⎛ 4m l ⎞ ⎝ ⎝ ⎠⎠
h0 = ⎜ dinj − dinj − ⎟, (5)
⎜
2⎝ πρl v ⎟⎠ The most probable droplet size d 0 is calculated by
where dinj is the injector orifice diameter, and m l is the d 0 = 1.88d L (1 + 3Oh)1 6 , (10)
effective mass flow rate, v is the axial component of velocity
at the injector exit. The axial component of velocity at the here, Oh is the Ohnesorge number which is a combination
injector exit is calculated by of the Reynolds number and the Weber number. It can be
supposed d 0 corresponds to volumetric mean diameter of
v = V cos θ , (6) generated entire droplets.
where V is total velocity of axial and tangential velocity,
Droplet size distribution
θ is the spray angle.
The particle size distribution is determined by Rosin-
Rammler diameter distribution method when the pressure-
swirl atomizer model is used. According to Institute for Liquid
Atomization and Spray System — Japan (2001), the mass
fraction of droplets of diameter greater than d is given by

Yd = exp(−(d d ) n ), (11)

where d is the droplet diameter, d is the size constant, n

is the distribution parameter. When the pressure-swirl
atomizer model is used, n should be given by default value
of 3.5. This expression can be inverted by taking logs of
both sides and rearranging
d
d = 1
(12)
(− ln(Yd )) n
Fig. 2 Formation of liquid droplet for pressure-swirl atomizer model here, substituting d 0 for d, Yd is to be 0.92. Because d 0
4 Build Simul

corresponds to volumetric mean diameter, this equation is high as shown in Fig. 4, and was analyzed assuming that it
to be is being used as a waiting area at an event hall. The
boundary conditions in Table 1 were given. Air temperature
d = 2.03d0 (13) and humidity in the area were given with typical summer
day in Tokyo (air temperature of 33.4 ℃ , humidity of
Then, Eq. (11) can be rewritten as
58%RH). It was also assumed that there was airflow (wind
Yd = exp(−(d 2.03d 0 ) n ) (14) velocity of 0.1m/s) of outdoor wind. The roof was made of
PVC-coated glass-fiber and the ground was covered by
So, if the value of d 0 was determined, we can obtain the concrete; those material properties are shown in Table 2.
droplet size distribution as shown in Fig. 3. Sol-air temperature and thermal transmittance were given for
the roof surface. At the ground surface, the ground surface
2.3 Calculation area and boundary conditions temperature was given. The ground surface temperature was
estimated in consideration of heat transfer at the ground
The semi-outdoor area is 50m long, 15m wide and 4m surface as shown in Fig. 5.

Fig. 3 Example of particle diameter distribution: (a) cumulative size distribution of particles, (b) particle size distribution

Fig. 4 Diagram of calculation area

Table 1 Boundary conditions

Boundary Momentum condition Thermal condition
Roof surface (z = 4m) No-slip External air temperaturea: 38.2℃
External heat transfer coefficient: 23W/(m2·K)
Ground surface (z = 0) No-slip Surface temperature: 29.8℃
Inflow outside air (x = 0) Inflow velocity: 0.1m/s Outside temperature: 33.4℃
Outside humidity: 18.9g/kg(DA)
Outflow (x = 50m) Pressure: atmospheric pressure Outside temperature: 33.4℃
Outside humidity: 18.9g/kg(DA)
Side surface ( y = 0, y = 50m) Free-slip Adiabatic boundary
a
Sol-air temperature was given. It assumed that the horizontal solar radiation is 1.026kW/m2.
Build Simul (2008) 1: 95 – 110 5

Fig. 5 Heat balance at the ground surface

The heat balance at the ground surface is Table 3 Spraying conditions

Mass flow rate 0.83g/s
qconv + qcond + qsol = 0, (15)
Water temperature 33.4℃

qconv = α (Tair − Tgs ), (16) Injector orifice diameter 0.16mm

Spray cone angle 50°
λug
qcond = (Tug − Tgs ), (17) Atomizer dispersion angle 6°
d ug Injection pressure 6MPa

qsol = τ rf ags J , (18)

Table 4 Sheet constant and the representative droplet diameter
here, qconv is the heat convection and long-wavelength Case 1 Case 2 Case 3
radiation from the air to the ground surface, qcond is the
Sheet constant 24 12 6
heat diffusion to the underground, qsol is the solar radiation
Volumetric mean diameter (μm) 9.0 12.7 17.9
after penetrating the roof to the ground.
Sauter mean diameter (μm) 16.9 20.8 32.6
Table 3 lists the spraying conditions of the atomizer. For
this study, only the sheet constant was varied in order to
generate different particle size distributions under the same
spray conditions. Figure 6 and Table 4 show the particle size
distributions for the three cases considered in this analysis.
The CFD simulation in this study is based on conservation
equations of energy, mass, and momentum of incompressible
air. Also, this simulation adopted the standard k-ε turbulence
model. The grid distance in each axial set to be 0.5 − 1.0m
and the entire calculation domain is composed a total of
11,200 cells. We conducted unsteady simulation until the
fluctuation of average air temperature in the calculation
domain is smaller than 10−6 K.

Table 2 Material properties for the roof and ground surface Fig. 6 Particle size distributions
Roof surface (z = 4m) Material: PVC-coated glass-fiber plain-weave
Thickness: 0.53mm
3 Analysis result and discussion
Thermal conductivity: 0.11W/(m·K)
Solar absorptance: 10.8%
Solar transmittance: τrf = 13.7% 3.1 Single-nozzle: results and discussion
Ground surface (z = 0) Material: paved concrete
Thickness: dug = 5m Figures 7 to 9 show the vertical cross-section temperature
Thermal conductivity: λug = 1.4W/(m·K) contours at the spray location ( y = 7.5m) for the three
Solar absorptance: ags = 60.0% cases. Figures 10 and 11 show the temperature distribution
6 Build Simul

and the absolute humidity distribution at 1.5m ( y = 7.5m, to about 20.0g/kg (DA) at the spray location to only about
z = 1.5m) above the ground. For each case, the air 19.2 to 19.5g/kg (DA) at the position behind the spray
temperature decreased about 1.5℃ at the spray location location, representing an increase of 0.3 to 0.6g/kg (DA).
(x ≈ 15m). Also, about 2.5m behind the spray position, it Figure 13 shows the remaining particle mass distribution
was observed that the maximum air temperature increase at every height of the analysis area. The total mass of
was about 0.5℃. It was thought that this phenomenon is remaining particles was smaller for case 1 than for the
caused by the downward airflow from the mist spray other two cases, indicating a fast evaporation rate. In case 3,
position. The wind direction diagram shown in Fig. 12 also which was for comparatively large particles, there was a
confirms this phenomenon. Also, there was no significant large mass of particles remaining in the area. When the
temperature reduction at the position behind the spray mass of remaining particles was analyzed as a function of
location (x = 20 − 50m). The absolute humidity increased height, the particles remained even at a height of 1.4m.

Fig. 7 Case 1, air temperature contours at the spray location ( y = 7.5m)

Fig. 8 Case 2, air temperature contours at the spray location ( y = 7.5m)

Fig. 9 Case 3, air temperature contours at the spray location ( y = 7.5m)

Fig. 10 Air temperature distribution at the spray location of 1.5m height

Fig. 11 Absolute humidity distribution at the spray location of 1.5m height

Build Simul (2008) 1: 95 – 110 7

Fig. 12 Airflow direction at spray location ( y = 7.5m)

single-nozzle scenario. Three nozzles were used, and the

2nd and 3rd were installed at 2.5m intervals from the
single-nozzle spray position in order to reduce the
temperature increase behind the spray position. Only cases
2 and 3, which had large particles, were simulated. Also,
cases 2' and 3' had the same particle size distributions as
cases 2 and 3, respectively. Figures 14 and 15 show the
cross-sectional ( y = 7.5m) temperature contours for cases
2' and 3'. Figures 16 and 17 show the air temperature
distribution and the absolute humidity distribution at a
height of 1.5m ( y = 7.5m, z = 1.5m). Similarly to the case
of the single nozzle, there was no significant difference in
the temperature reduction for different particle size
distributions. The air temperature just below each nozzle
was about 31.3℃, which was approximately equal to that in
the case of a single nozzle. Compared with the absolute
humidity distribution, the humidity just below the 1st
nozzle was similar to the single-nozzle case. However, the
humidity just below the nozzle further behind was higher
Fig. 13 Remaining particle mass distribution at each height than that below the nozzle further ahead. The humidity of
the 3rd nozzle was about 0.2g/kg (DA) greater than that
These results revealed that the temperature reduction does for the corresponding case of the single nozzle.
not vary greatly with differences in particle size distribution. Figures 18 and 19 show the horizontal distribution of air
However, as the particle size increases, it takes longer time temperature at a height of 1.5m for cases 2 and 2'. In the
for the particles to completely evaporate, and particles cases where multiple nozzles were used, the air temperature
may remain even at low heights. Thus, the particle size is in front of the 1st nozzle was lower than with the single
an important parameter in mist design when considering nozzle. The cause of the difference of temperature
the spray height. distribution was considered to be that the downward flow
by water mist evaporation was greater than in the case of
3.2 Multiple nozzles: results and discussion the single nozzle, and that the heat of mist vaporization
was diffused more easily because of the downflow due to
This section discusses the effect of particle size distribution mist blocking the inflow of outdoor air. This phenomenon
when multiple nozzles were arranged at same intervals. may have occurred because the side boundary condition
The same calculation conditions were used as for the was a wall without new air inflow or outflow.

Fig. 14 Case 2', air temperature contours at the spray location ( y = 7.5m)

Fig. 15 Case 3', air temperature contours at the spray location ( y = 7.5m)
8 Build Simul

Fig. 16 Air temperature distribution at the spray location of 1.5m height

Fig. 17 Absolute humidity distribution at the spray location of 1.5m height

Fig. 18 Horizontal distribution of air temperature at a height of 1.5m with a single nozzle (case 2)

Fig. 19 Horizontal distribution of air temperature at a height of 1.5m with three nozzles (case 2')

Figure 20 shows the particle mass distribution at different few particles remained at a height of 1.8m or lower in both
heights. In comparing the cases of single and multiple cases 2 and 2', which had comparatively small particles.
nozzles, we see that the maximum mass of remaining Comparing cases 3 and 3', which had comparatively large
particles was found to be at almost the same height. As for particles, a few particles remained even at a height of
the minimum height at which particles remained, however, 1.0m for case 3'.
Build Simul (2008) 1: 95 – 110 9

it is expected to make a relatively larger effect on the

height of remaining particles. Also, mist particles remained
at lower heights when the number of nozzles was increased.
The above results demonstrate that spray height and particle
size need to be selected carefully while the height of
remaining particles is to be strictly controlled.
For those cases (spray particle size is Sauter mean
diameter 16.9μm, 20.8μm and 32.6μm), the air temperature
reduction was same with that of these field experiments
(Hayashi et al. 2005; Yamada et al. 2006). That is to say, the
simulation in this study could reproduce the evaporation
cooling effect of mist spray system. In the future, we will
confirm the validity of this simulation technique by
considering the temperature reduction and the minimum
height of remaining particles.

References

Fig. 20 Remaining particle mass distribution at each height Dombrowski N, Hooper PC (1962). The effect of ambient density or
drop formation in sprays. Chemical Engineering Science, 17:
291 − 305.
4 Conclusions Fluent Inc. (2006). Fluent 6.3 User’s Guide. http://www.ansys.com.
Hayashi A, Kodama N, Okuyama H, Tsujimoto M, Okumiya M,
In this study, we discussed the cooling effect in terms of Harada M, Ichinose S, Shindo Y (2005). Development of heat
island control system with water mist sprayer Part 3: Field
particle size distribution and the height distribution of
measurement and computational fluid dynamics calculation.
remaining particles. A significant difference in temperature Summaries of technical papers of the annual meeting of the
reduction for different particle size distributions was not Architectural Institute of Japan, Kinki. (in Japanese)
observed. As the particle size increased, however, the Institute for Liquid Atomization and Spray Systems — Japan (2001).
minimum height of remaining particles was lower. In the Atomization Technology. Morikita Publishing Co. (in Japanese)
case of single nozzle, there is no significant difference in Schmidt DP, Nouar I, Senecal PK, Hoffman J, Rutland CJ, Martin J,
the minimum height of remaining particles between case 1 Reitz RD (1999). Pressure-swirl atomization in the near field.
SAE technical paper, 1999-01-0496.
(Sauter mean diameter 16.9μm) and case 2 (20.8μm).
Yamada H, Okumiya M, Tsujimoto M, Harada M (2006). Study on
However, mist particles remained at lower heights in the cooling effects of water mist sprayer. Summaries of
case 3 (Sauter mean diameter of 32.6μm). Thus, when the technical papers of the annual meeting of the Architectural
different particle size was larger than a dozen micrometer, Institute of Japan, Kanto. (in Japanese)

View publication stats