Surface & Coatings Technology 203 (2009) 1733–1740

Contents lists available at ScienceDirect

Surface & Coatings Technology

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s u r f c o a t

Effects of operating conditions on droplet deposition onto surface of atomization

impinging spray
Jianzhong Lin a,b,⁎, Lijuan Qian a, Hongbin Xiong a, Tat Leung Chan c
a
State Key Laboratory of Fluid Power Transmission and Control, Zhejiang University, Hangzhou 310027, PR China
b
China Jiliang University, Hangzhou, 310018, PR China
c
Department of Mechanical Engineering, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong

a r t i c l e i n f o a b s t r a c t

Article history: A comprehensive model based on the Navier–Stokes equation and particle tracking method is used to study
Received 18 August 2008 the effervescent atomization impinging spray, and another model is used to establish the relationship
Accepted in revised form 6 January 2009 between the droplet velocity near the plate and the different operating conditions. The models and
Available online 15 January 2009
numerical code are validated by comparing the numerical results with the published experimental results.
The effects of air-to-liquid ratio, nozzle diameter, liquid mass flow rate, and the position of impinging plate
Keywords:
Atomization impinging spray
on the Weber number and K number as well as the droplet deposition onto the plate are discussed. The
Droplet deposition results show that the droplet velocity near the plate increases with increasing air-to-liquid ratio and liquid
Operating conditions mass flow rate, and with decreasing nozzle diameter and axial distance from the nozzle exit to the plate. The
Numerical simulation droplet diameter near the plate increases with increasing axial distance from the nozzle exit to the plate, and
with decreasing air-to-liquid ratio. As a function of the nozzle diameter and liquid mass flow rate, the
variation of droplet diameter is not monotonous and the effect of liquid mass flow rate on the droplet
diameter is insignificant. In the studied operating conditions, it is difficult for the droplet to rebound off the
plate when impinging on the plate but it is easier for the droplet to splash. In order to create a condition
which can benefit the droplet deposition when impinging on the plate, the suggested ways are to reduce the
air-to-liquid ratio and liquid mass flow rate, increase the nozzle diameter, and select a suitable range of axial
distance from the nozzle exit to the plate.
© 2009 Elsevier B.V. All rights reserved.

1. Introduction
Spray impingement onto a solid surface is an important funda-
mental process in a variety of applications. For impacting on a surface,
three main outcomes are expected, i.e., droplet rebound, splash or
deposition onto the surface. In the interest of obtaining better paste
intensity, for instance, coating the different polymeric membranes
onto the textile surface to produce a variety of functional chemical
textile materials, the atomized droplet impinging onto the plate is
expected to stick or deposit onto the surface without causing
rebounding or splashing.
There are some investigations devoted to the spray impingement
and deposition onto a solid surface. Van Steenkiste et al. [1] produced
coatings by entraining metal powders in an air flow which is
accelerated by a de Laval type of nozzle. A threshold behavior was
observed for coating deposition as a function of nozzle inlet air
temperature. Garbero et al. [2] investigated the paint deposition
process by spraying by means of computational fluid dynamics, and
⁎ Corresponding author. State Key Laboratory of Fluid Power Transmission and
Control, Zhejiang University, Hangzhou 310027, PR China. Tel.: +86 571 86836009.
E-mail address: mecjzlin@public.zju.edu.cn (J. Lin).
particular attention had been devoted to the prediction of the maxi-
mum diameter reached by an impinging droplet at the end of the
spreading phase. Han et al. [3] investigated the effects of spray
conditions on coating formation by the kinetic spray process. They
found that increasing particle temperature provides an additional
means that can effectively enhance the coating formation by the
kinetic spray process. Kang [4] developed a semi-empirical methodol-
ogy for predicting three-dimensional deposits by the plasma spray
process. The deposition volumes and thicknesses for different
substrate inclinations with good agreement to experimentally sprayed
deposits were predicted.
In order to confirm the outcomes of deposition, several transition
criteria are used to restrict the dimensionless numbers within a certain
range. Firstly, the rebound-deposition criterion: the rebound regime
occurs when the impinging droplet bounces off the plate when the
impinged energy is low. The transition criterion between deposition
and rebound is given as Weber number, We = ρldlvl2 / σl where ρl is the
droplet density, dl is the droplet diameter, vl is the droplet normal
incidence velocity, and σl is the surface tension. In the deposition
regime, Weber number is larger than 5 [5]. Secondly, the deposition-
splash criterion: the dimensionless number is able to specify the
transition between the deposition and splashing regimes for both the

0257-8972/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.surfcoat.2009.01.009
1734 J. Lin et al. / Surface & Coatings Technology 203 (2009) 1733–1740
smooth and the rough surfaces which is so-called K number, K =
µ−l 0.25ρ0.75
l σ−l 0.5d0.75
l vl1.25 (where µl is the droplet viscosity)[6]. Accord-
ing to the empirical correlation proposed by Mundo et al. [7], the value
of K is less than 57.7 which leads to complete deposition of the liquid.
From the definitions of Weber number and K number, the para-
meters affecting We and K are the material properties of the fluid (i.e.,
density, ρl, surface tension, σl and viscosity, µl), and geometric and
kinematic parameters (i.e., droplet diameter, dl and velocity, vl). For a
definite fluid, Weber number and K number are related to dl and vl.
While dl and vl are dependent on the spray process and operating
conditions (i.e., atomizing air-to-liquid ratio (ALR), nozzle diameter,
liquid mass flow rate, and position of impinging plate). However, there
is no study devoted to the effects of the above parameters on Weber
number and K number as well as the droplet deposition on the plate
up to now. Therefore, the objective of the present study is to explore
numerically the functions of ALR, nozzle diameter, liquid mass flow
rate and position of impinging plate by adjusting the operating
conditions in order to achieve a better result of droplet deposition
onto the plate.
2. Mathematical model
Because the droplet size–velocity correlation is insignificant [8],
two models are used to establish the relationship between droplet
diameter dl near the plate, normal incidence velocity vl and operating
conditions, respectively.
As shown in Fig. 1, the effervescent atomization is a form of
internal-mixing atomization in which the atomizing gas is bubbled
into a liquid and resulting in a two-phase mixture which is discharged
from the atomizer orifice. Leaving the nozzle exit, the rapidly
expanding gas phase will shatter the liquid into finer droplets. These
droplets are unstable in the turbulence spray, they will undergo a
series of events such as collision, breakup and coalescent. Finally the
gas entrained droplets impinge on the plate.
Spray droplet size is the parameter of greatest interest in most
applications, and there are many investigations devoted to it.
However, the research of the droplet velocity is seldom studied.
From the late 20th century, the research group of Purdue University,
USA has begun a systematic study on atomizer performance and the
fundamental mechanism involved in the effervescent atomization [9].
In their numerical simulations, not only a one-dimensional model of
predicting primary droplets close to the nozzle exit is established, but
also a model for predicting droplet velocity profile in dilute spray
region is developed. Unfortunately, the process of secondary atomiza-
tion which determines the final droplet characteristics before
impacting onto the plate is not mentioned. In the present study, a
comprehensive three-dimensional model of droplet-gas two-phase
flow is presented to predict the droplet diameter downstream of the
nozzle exit in an effervescent atomizer. Moreover, a one-dimensional
Fig. 1. A schematic diagram of effervescent atomization with an impinging plate.
Fig. 2. A schematic diagram of primary breakup process.
model developed by Panchagnula and Sojka [8] is also used to predict
the droplet velocity close to the impinging plate.
2.1. Relationship between operating conditions and droplet diameter
To predict the droplet diameter downstream of the nozzle exit in
an effervescent atomizer, a comprehensive three-dimensional model
involved in the primary breakup and secondary breakup is presented.
The primary breakup sub-model is a one-dimensional model used to
predict the initial droplet mean diameter near the nozzle exit, and the
initial droplet characteristics provides proper boundary conditions for
the secondary breakup. Then the sub-model of secondary breakup is
used to calculate the whole evolvement of droplet size downstream in
three-dimensional space.
2.1.1. Initial droplet mean diameter from the primary breakup
Based on the experimental results [10] of the flow near the nozzle
exit, it is assumed that the ligaments formed can be approximated by
the cylindrical jets. Three sub-processes occur during the primary
breakup as shown in Fig. 2. Firstly, the annular liquid sheath breaks
into a number of cylindrical ligaments. Secondly, the ligaments break
into ligament fragments. Finally, the ligament fragments are stabilized
to form the droplets. To model the primary breakup of droplets, a one-
dimensional primary breakup model [11] is used to estimate the
droplet mean size from the atomizing gas and liquid mass flow rates,
liquid physical properties, and atomizer exit geometry. It is assumed
that the annular liquid sheet breaks into several cylindrical ligaments
with almost the same diameter as the thickness of the annular sheet.
The ligaments further break into ligament fragments at the wave-
length of the most rapidly growing wave. Each fragment is assumed to
form one spherical droplet under the effect of surface tension.
2.1.1.1. Estimation of annular liquid sheet thickness. To estimate the
thickness of annular liquid sheet, it is assumed that the annular two-
phase flow within the discharge orifice is one-dimensional, inviscous
and isothermal with compressible ideal gas and small interface
velocity slip ratio. The momentum equation of the annular flow can
then be written as:
dp
+ vg dvg = 0 ð1Þ
ρg
where p is the gas pressure, ρg is the gas density, vg is the gas velocity.
Combined with the state equation, Eq. (1) can be integrated as:
: !2

 1 ml ALR
RT ln ρg RT + = constant; ð2Þ
2 ρg πrg2
where R is the gas constant, T is the temperature, rg is the radius of gas
flow, ṁl is the mass flow rate of liquid, and ALR is the air–liquid ratio
by mass. The radius of gas flow can be written in terms of orifice radius
pffiffiffiffi
ro and void fraction α, rg = α ro . The interface velocity slip ratio for
different flow rates can be expressed as [12]:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffi ffi
ρl α
sr = ð3Þ
ρg 1 + C ð1−α Þ
 J. Lin et al. / Surface & Coatings Technology 203 (2009) 1733–1740 1735


where ρl is the liquid density, C is the experimental coefficient. The A ρg e

interface velocity slip ratio and void fraction has the following + jd ρg eug = −pjd ug −jdq + ρg e + Q; ð10Þ
At
relationship:

ρg sr 1 where ug is the gas velocity vector, F is the momentum transfer

1+ = ð4Þ
ρl ALR α between the gas and discrete liquid phase, k is the turbulent kinetic
energy per unit mass, σs is the viscous shear stress tensor, q is the heat
By solving Eqs. (2)–(4), ρg, α and sr can be calculated for different flux vector, ɛ is the viscous dissipation rate, and Q is the heat transfer
operating conditions. The thickness of annular liquid sheet is then between the gas and liquid phase. In the computation, it is necessary
calculated from δ = ro − rg, which is also the diameter of the typical to sum over all parcels in the corresponding computational cell. Each
cylindrical ligament. parcel represents a certain number of droplets that have the same
properties of position, size, velocity and temperature. Therefore, F and
2.1.1.2. Mean droplet size after the primary breakup. After obtaining Q are defined as:
the thickness of annular liquid sheet, the length of a typical ligament
fragment (i.e., the breakup wavelength λ) where the disturbance 1
grows most rapidly is computed. Here the linear instability analysis F =− ∑ Np F p ; ð11Þ
α g Vcell p
theory [13] is used to determine λ:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 

pffiffiffi 3μ l 1
λ = 2πδ 1 + pffiffiffiffiffiffiffiffiffiffiffi ffi; ð5Þ Q= ∑ Np Qp + F p ug − ul ; ð12Þ
α g Vcell p
ρl σ l δ

where ρl is the liquid density, µl is the viscosity and σl is the surface where αg is the fraction of gas in the computational cell, Vcell is the cell
tension. The movement of the sheet is not considered in Eq. (5) which volume, Np is the number of the droplets in this parcel, and Fp is the
is approximately correct only in the case of long wave disturbances. force acting on each droplet, Qp is heat flux at each droplet surface,
Therefore, a more accurate solution [14] is used here. The growth rate and ul is the droplet velocity.
of surface disturbances for the sinusoidal mode is given by:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2.1.2.2. Droplet phase modeling. The liquid droplets are discretely
2
2μ k2 tanhðkhÞ 4μ 2l k4 tanh ðkhÞ−Q 2 V 2 k2 −½tanhðkhÞ + Q ð−QV 2 k2 + σ l k3 =ρl Þ treated in a Lagrangian manner. Computational droplets are generated
ω=− l +
tanhðkhÞ + Q tanhðkhÞ + Q at the points of injection and tracked throughout their flight. Assume
ð6Þ that only drag force and gravity affect the momentum transfer of
droplet significantly, the momentum transfer between the droplet and
where k is the wave number given by k = 2π / λ, h is the half-thickness atomization jet can be given by:
of sheet h = δ / 2, Q is the density ratio of gas/liquid, V is the velocity of
liquid sheet. Here, the value of the velocity of liquid sheet is assumed dul πrl2 ρl CD

to be the same as the relative velocity difference of the gas and the F p = ml = jug −ul j ug − ul g; ð13Þ
dt 2
liquid at the nozzle exit. The breakup wavelength λ is determined by
the wave number k where the maximum growth rate ω occurs in the dxl
curve calculated by Eq. (6). Assuming that each fragment is stabilized = ul ; ð14Þ
dt
to one droplet, the Sauter mean diameter (SMD) of the droplet size can
then be calculated from the conservation of mass:
where Fp is the force exerted on droplet, ml is the mass of droplet, rl is
 1=3 the radius of droplet, g is the gravity force, and CD is drag coefficient
3
SMD = δ3 λ : ð7Þ and can be expressed as:
2

24
 2ρg rl jug −ul j

Using this model, the mean diameter of droplet from the primary −1=2
CD = + 6 1 + Rel + 0:4; for Rel = b100 ð15Þ
breakup can be calculated for different operating conditions. Rel μg

2.1.2. Mean droplet size downstream from the secondary breakup where rl is the particle radius, ρg and µg are the mean density and
In the secondary breakup model, the Navier–Stokes equation for viscosity of gas phase around the particle, respectively. In the present
the gas phase coupled the standard k–ε turbulence model is solved study, Rel is less then 100, therefore, Eq. (15) can be used.
with interaction between the droplet and gas phase taken into
consideration. The dispersed droplet phase is modeled using the 2.1.2.3. Droplet distortion. To model the droplet secondary breakup,
Lagrangian method, which has taken into consideration the collision a cascade atomization and drop breakup model [15] is utilized. The
and coalescence, momentum and heat transfer processes. droplet distortion is described by a forced, damped and harmonic
oscillator where the forcing term is given by the aerodynamic droplet/
2.1.2.1. Gas phase modeling. The gas flow is turbulent, and the con- gas interaction, the damping is caused by liquid viscosity, and the
servation equations of mass, momentum and energy are written as restoring force is supplied by the surface tension.
follows: The droplet distortion is described by the deformation parameter,
ζ = 2x / r where x is the radial cross-section change from its equilibrium
Aρg
 position and r is the original radius of the droplet. The deformation
+ jd ρg ug = 0; ð8Þ equation in terms of the normalized distortion parameter, ζ is cal-
At
culated by:


A ρg ug
  
2 :: 5μ l : 8σ l 2ρg jUj2
+ jd ρg ug ug = −j p + ρg k + jd ½σ s  + F; ð9Þ f+ f+ f= ; ð16Þ
At 3 ρl a2 ρl a 3 3ρl a2
1736 J. Lin et al. / Surface & Coatings Technology 203 (2009) 1733–1740
where U is the relative droplet-gas velocity, a is the volume fraction of
gas. Droplet breakup occurs when the normalized droplet distortion,
ζ(t) exceeds the critical value of 1.
The creation of the product droplets is derived by using the
principles of the population dynamics. For each breakup event, it is
assumed that the number of product droplets is proportional to the
number of critical parent droplets, where the proportionality constant
depends on the drop breakup regime. The rate of droplet creation in
conjunction with the principle of mass conservation between the
parent and product droplets, leads to the basic cascade breakup law:
d sion for the velocity profile.
m ðt Þ = −3ωbu m ðt Þ; ð17Þ
dt
_ turbulent jet is used to obtain a two-parameter description of the
where m (t) is the mean mass of the product droplet distribution, and
the breakup frequency ωbu depends on the drop breakup regimes. As
suggested by Reitz [16], three breakup regimes are classified with
respect to increasing gas Weber number into the bag breakup, strip-
ping breakup regime and catastrophic breakup regime. In the present
study, the gas Weber number is mostly less than 80, which falls into
the bag breakup regime. Breakup frequency, ωbu = 0.05ωos as
suggested by O'Rourke and Amsden [17] is used. The droplet oscil-
lation frequency, ωos is given by:
8σ l 25μ 2l
ω2os = − ð18Þ
ρl r03 4ρ2l r04
where r0 is the radius of parent droplet. It should be noted that the
actual size distribution of the product droplet is not specified yet
except for the mean mass m(t). A uniform product droplet size dis-
tribution is assumed and Eq. (17) becomes:
r where ū is the mean axial velocity at a point in the jet, Umax is the
= e−ωbu tbu ; ð19Þ
r0
where r is the radius of the product droplet, and tbu is the breakup
time (i.e., the time taken until the normalized deformation ζ(t) from
the solution of Eq. (16) exceeds the value of 1).
2.1.2.4. Droplet collision modeling. Collision and possible coales-
cence among the droplets are modeled through a collision Weber
number. Collisions between the droplets can result in bouncing,
permanent coalescence or coalescence followed by separation. Most
studies have defined the collision parameters such as collision Weber
number, Wel, and impact parameter, χ/(r1 + r2) where χ is the distance
between the center of two neighboring droplets, and the relative
velocity vector, r1 and r2 which are the radii of small and large
droplets, respectively. Two droplets are assumed to collide if they
touch each other, and collision is treated as an instantaneous event.
One droplet replaces two colliding droplets in a coalescence event,
while the bouncing droplets simply change velocity. Separation-type
collisions result in an instantaneous replacement of the colliding
droplets with a collection of smaller satellite droplets. Such case
happens only when the relative velocity of droplets is very large,
which is not considered in the present study. The balance between the
surface, kinetic and internal energy determines the collision regime.
Coalescence occurs when the surface energy dominates, while
bouncing occurs when the kinetic energy dominates. The criteria,
χcr determines the transition points between the different regimes
(e.g., droplets coalescence when χ ≤ χcr and bounces when χ N χcr).
The criteria χcr can be expressed as follows:
 
f ðγ Þ
χ2cr 2
= ðr1 + r2 Þ min 1:0; 2:4 ; ð20Þ
Wel
_ _
where Wel = ρl|U|r_1 / σl(T d), σl(T d) is the surface tension at the average
temperature of T d = (r1T1 + r2T2) / (r31 + r32) where T1 and T2 are the
3 3
temperature of the two colliding droplets, and f(γ) is a function of the
radius ratio as f(γ) = γ2 − 2.4γ2 + 2.7γ with γ = r2 / r1 ≥ 1 [17].
2.2. Relationship between the operating conditions and droplet velocity
For predicting the droplet velocity profile in effervescent spray, a
one-dimensional model developed by Panchagnula and Sojka [8] is
used to determine the droplet velocity in terms of the operating
parameters (i.e., liquid mass flow-rate and ALR) and the density of the
liquid, atomizing gas, and ambient gas. They modeled the spray as a
variable-density and single-phase turbulent jet, and used an expres-
Firstly, an equation of velocity profile in the self-preserving
velocity distribution. Based on the theory of White [18], Gortler [19]
took the boundary layer approximation to the exact Navier–Stokes
equation with the appropriate boundary conditions to yield the
following similarity solution of the velocity profile:
 2
2
u = Umax ; ð21Þ
eη + e−η
x
η=β ; ð22Þ
y
sffiffiffiffiffi
J 1
Umax = b ; ð23Þ
ρe y
maximum centerline axial velocity, J is the jet momentum rate, ρe is
the entrained gas density, β and b are the empirical constants, x is the
radial distance measured from the centerline of the jet, and y is the
axial position measured downstream of the jet exit orifice. White [18]
compared the predictions based on Eqs. (21)–(23) with the measured
data of velocity profile given by Wygnanski and Fiedler [20] and found
excellent agreement. Eqs. (21)–(23) are therefore chosen for the
velocity profile in the present study. Because the solution of Eqs. (21)–
(23) is for a free jet injected into the quiescent air, Eq. (21) needs
modification for applying to a two-phase effervescent spray. This is
accomplished by introducing the two-phase momentum rate for J and
the entrained gas density ρe:
: :
J = ml ul + mg ug ; ð24Þ
where ṁl and ul are the liquid-phase mass flow rate and velocity,
respectively, ṁg and ug are the gas-phase mass flow rate and velocity,
respectively. By substituting Eq. (24) into Eq. (23) and substituting ALR
by ṁg/ṁl and sr for ug/ut, Eq. (23) can then be expressed as:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
:
ml ul ð1 + ALR×srÞ 1
Umax = b : ð25Þ
ρe y
From the liquid-phase continuity:
:
4 ml
ul = ð26Þ
πd20 ρl ð1−α Þ
where d0 is the orifice diameter of atomizer exit, and α can be expressed
as shown in Eq. (4). Substituting Eqs. (26) and (4) into Eq. (25) and
 J. Lin et al. / Surface & Coatings Technology 203 (2009) 1733–1740 1737

3. Results and discussions

3.1. Validations of the computational model and code

For validating the computational model and code, two numerical

results of Sauter mean diameter (SMD), downstream of the nozzle exit
based on the models of Weber and Senecal are compared with the
experimental data [21] as shown in Fig. 3, in which the injection
pressure is 0.38 MPa, the radius of nozzle exit is 0.5 mm and ALR
changes from 0.06 to 0.2. Obviously the Senecal's model is much
better than that of Weber's model in this case. The relative error of
both numerical results based on the Senecal's model and the
experimental data is less than 5% when the value of ALR is larger
than 0.04. However, the discrepancy is relatively large in the region of
small ALR. The reason is that less gas fraction leads to a bubbly flow
and thus the assumption of annular flow is not applicable. Whitlow
and Lefebvre [22] studied the flow transition experimentally and
showed that the upper limit of ALR for the operation condition within
Fig. 3. Comparison of numerical and experimental results of Sauter mean diameter
(SMD). the bubbly flow is 4.6 ρg/ρl, which is approximately 0.04 for air and
water. Fig. 3 also shows that SMD decreases with increasing ALR. Since
ALR is one of major factors which would affect the droplet
making simplification, the predictive equation for Umax can be expressed atomization. Under the circumstance of a large ALR, the mean droplet
as: diameter becomes smaller and atomized effect can be improved.
rffiffiffi : rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffisffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Fig. 4(a) shows the numerical normalized mean droplet velocity
4 ml ρl ρ ×ALR versus x/y (i.e., x and y are radial and axial coordinates, respectively). The
Umax = b ð1 + ALR×srÞ 1 + l : ð27Þ
π ρl d0 y ρe ρg ×sr experimental results [8] measured in a representative spray with the
Staley high-fructose corn syrup as the atomized liquid for the operating
The empirical constants β and b in the equations can be calculated conditions of ALR= 0.1 and ṁl = 30 g/s. The mean velocity profile exhibits
from experimental results. Mahesh et al. [8] presented the expression a self similarity. Fig. 4(b) shows the comparison of the numerical results
of β and b based on the experimental data, where b was found to be of the number averaged droplet velocity and the experimental data for
related to ALR linearly, while β is as follows: the same operating conditions as that in Fig. 4(a). The numerical results
are basically consistent with the experimental ones.
β = ψ × Recl × ALRd ð28Þ
3.2. Effect of air-to-liquid ratio on the droplet deposition onto the plate
where the constants are ψ = 40.0, c = −0.37, d = 0.15 and the superficial
liquid Reynolds number can be defined as:
Air-to-liquid ratio (ALR) in mass basis is one of the most important
: factors affecting the effervescent atomization spray as well as other
4 ml
Rel = : ð29Þ output parameters. Fig. 5 shows the variations of mean droplet velocity
πd0 μ l
and SMD near the plate, Weber number and K number as functions of
K = 40.0, c = −0.37, d = 0.15. This velocity profile can provide a descrip- ALR. It can be seen from Fig. 5(a) that the mean velocity increases
tion of the spatial velocity distribution in effervescent sprays. gradually while SMD decreases with increasing ALR. The increase of ALR

Fig. 4. Comparison of numerical and experimental results of mean droplet velocity profile.
1738 J. Lin et al. / Surface & Coatings Technology 203 (2009) 1733–1740

Fig. 5. Effect of air–liquid flow ratio (ALR) on the mean droplet velocity, SMD, Weber number and K number. (Atomized liquid: water; Liquid flow rate: 90 kg/h; Nozzle diameter:
0.3 cm; Axial distance of the impinging plate: 150 mm).

means the increase of gas flow rate when the liquid mass flow rate
remains constant, which results in an increase in the momentum of the
spray, consequently enhances the drag force of the gas on the liquid. On
one hand, this force increases the droplet velocity obviously. On the
other hand, it accelerates the droplet distortion and breakup. As the
velocity of air flow increases in the impacting spray, the droplet
distribution becomes more scattered, and the probability of droplet
collision and coalescence decreases, so the droplet mean diameter
becomes smaller. Fig. 5(b) shows that We number and K number
increase with the increase of ALR. When ALR is too low (e.g., ALR = 0.02)
and We number is less than 5, the outcome of droplet impacting on the
plate would be rebound. When ALR is too high (e.g., ALR N 0.06), the
droplet impacting on the plate will probably splash.
3.3. Effect of nozzle diameter on the droplet deposition onto the plate
The effect of nozzle diameter on the mean droplet velocity, SMD,
Weber number and K number is shown in Fig. 6. With the increase of
nozzle diameter, the mean droplet velocity decreases sharply at first,
then tends to be steady. In the numerical simulation, the nozzle diameter
has influence on the primary atomization. Larger the nozzle diameter,
larger the droplet mean diameter from the primary atomization. While
this effect is deeply weakened by the secondary atomization at the
downstream of the spray. Therefore the droplet mean diameter at the
downstream of the spray is quite similar and does not change with the
nozzle diameter monotonously, which conforms to the characteristics of
effervescent atomization nozzle. In Fig. 6(b), We number and K number

Fig. 6. Effect of nozzle diameter on the mean droplet velocity, SMD, Weber number and K number. (Atomized liquid: water; Liquid flow rate: 100 kg/h; ALR: 0.08; Axial distance of the
impinging palte:150mm).
 J. Lin et al. / Surface & Coatings Technology 203 (2009) 1733–1740 1739

Fig. 7. Effect of liquid mass flow rate on the mean droplet velocity, SMD, Weber number and K number. (Atomized liquid: water; ALR: 0.08; Nozzle diameter: 0.3 cm; Axial distance of
the impinging palte:150mm).

decrease with the increase of nozzle diameter. In these cases, all We
numbers are larger than 5. When nozzle diameter is larger than 0.35 and
K number is less than 57.7, the droplet deposition criterion is complied.
3.4. Effect of liquid mass flow rate on the droplet deposition onto the plate
Fig. 7 shows that the mean droplet velocity increases steadily with
increasing liquid mass flow rate. When ALR remains constant, in-
creasing liquid mass flow rate means the gas mass flow rate also
increases, which results in the enhancement of the drag force of the
gas on the liquid. Although the increase of gas mass flow rate will
make the droplet size smaller, the increase of liquid mass flow rate will
lead to the enhancement of the probability of droplet collision and
coalescence and cause the mean droplet diameter becoming larger.
These two forces are offset. Therefore, the effect of liquid mass flow
rate on the mean droplet diameter is not obvious when ALR remains
constant. As shown in Fig. 7(b), the value of We number is larger than
5, and K number is less than 57.7 when the liquid mass flow rate is less
than 80, which means that the liquid mass flow rate should be less
that 80 in order to meet the droplet deposition condition.
3.5. Effect of axial distance from the nozzle exit to the impinging plate on
the droplet deposition
Fig. 8 shows the relationship between the axial distance from the
nozzle exit to the impinging plate and mean droplet velocity, SMD,
Weber number and K number. It can be seen from Fig. 8(a) that the
mean droplet velocity decreases with an increase of axial distance
from the nozzle exit to the impinging plate, while the case for SMD is
on the contrary. The reason is that the gas momentum decreases, and
the probability of droplet collision and coalescence increase as the
axial distance from the nozzle exit to the impinging plate increases, so

Fig. 8. Effect of axial distance from the nozzle exit to the impinging plate on the mean droplet velocity, SMD, Weber number and K number. (Atomized liquid: water; Liquid flow rate is
140 kg/h; ALR: 0.15; Nozzle diameter: 0.3 cm).
1740 J. Lin et al. / Surface & Coatings Technology 203 (2009) 1733–1740
the droplet velocity decreases and droplet diameter increases. Fig. 8
(b) shows that both We number and K number decrease as the axial
distance from the nozzle exit to the impinging plate increases. Values
of all We number are larger than 5 for all axial distances. K number is
less than 57.7 when the axial distances are beyond of 150 mm where
the deposition criterion is satisfied.
4. Conclusions
A comprehensive three-dimensional model based on the Navier–
Stokes equation and particle tracking method is used to study the
effervescent atomization spray with an impinging plate. Another one-
dimensional model is used to establish the relationship between the
droplet velocity near the plate, and the different operating conditions
(i.e., atomizing air-to-liquid ratio, nozzle diameter, liquid mass flow
rate, and position of impinging plate). The numerical results are
validated by comparing them with the published experimental result.
The effects of air-to-liquid ratio, nozzle diameter, liquid mass flow rate,
and position of impinging plate on Weber number and K number as
well as the droplet deposition onto the plate are discussed.
The results show that the droplet velocity near the plate increases
with increasing of air-to-liquid ratio and liquid mass flow rate, and
with decreasing of nozzle diameter and axial distance from exit to
plate. The droplet diameter near the plate increases with increasing
axial distance from the nozzle exit to the plate, and with decreasing
air-to-liquid ratio. As function of nozzle diameter and liquid mass flow
rate, the variation of droplet diameter is not monotonous and the
effect of liquid mass flow rate on the droplet diameter is insignificant.
In all studied operating conditions, when the value of Weber number
is larger than 5, then the rebound-deposition criterion is easily met. In
other words, it is difficult for the droplet to rebound off the plate when
impinging the plate. The value of K number is less than 57.7 in the
following operating conditions: air-to-liquid ratio is less than 0.06,
nozzle diameter is larger than 0.35, liquid mass flow rate is less than
80, axial distances from the nozzle exit to the plate are beyond
150 mm. It is easier for the droplet to splash than rebound when
impinging the plate. In order to create a condition which can benefit
the droplet deposition onto the plate when impinging on the plate, the
suggested ways are to reduce the air-to-liquid ratio and liquid mass
flow rate, increase the nozzle diameter, and select a suitable range of
axial distance from the nozzle exit to the plate.
Acknowledgement
This work was supported by the Major Program of the National
Natural Science Foundation of China with Grant No 10632070, and the
National Natural Science Foundation of China with Grant No. 10602052.
References
[1] T.H. Van Steenkiste, J.R. Smith, R.E. Teets, J.J. Moleski, D.W. Gorkiewicz, R.P. Tison, D.R.
Marantz, K.A. Kowalsky, W.L. Riggs, P.H. Zajchowski, B. Pilsner, R.C. McCune, K.J.
Barnett, Surf. Coat. Technol. 111 (1999) 62.
[2] M. Garbero, M. Vanni, G. Baldi, Macromol. Symp. 187 (2002) 719.
[3] T.Y. Han, Z.B. Zhao, B.A. Gillispie, J.R. Smith, J. Therm. Spray Technol. 14 (2005) 373.
[4] C.W. Kang, H.W. Ng, S.C.M. Yu, J. Therm. Spray Technol. 16 (2007) 261.
[5] C. Bai, A.D. Gosman, SAE 950283 (1995).
[6] G.E. Cossali, M. Marengo, M. Santini, At. Sprays 15 (2005) 699.
[7] Chr. Mundo, M. Sommerfeld, C. Tropea, Int. J. Multiph. Flow 21 (1995) 151.
[8] M.V. Panchagnula, P.E. Sojka, Fuel 78 (1999) 729.
[9] S.D. Sovani, P.E. Sojka, A.H. Lefebvre, Prog. Energy Comb. Sci. 27 (2001) 483.
[10] L.S. Liu, Ph.D. Thesis, Tianjing university, China, 2001.
[11] M.T. Lund, P.E. Sojka, A.H. Lefebvre, P.G. Gosselin, At. Sprays 3 (1993) 77.
[12] M.T. Lund, C.Q. Jian, J. Fluids Eng. 120 (1998) 750.
[13] C. Weber, Z. Angew, Math. Mech. 11 (1931) 136.
[14] P.K. Senecal, D.P. Schmidt, I. Nouar, C.J. Rutland, R.D. Reitz, M.L. Corradini, Int. J.
Multiph. Flow 25 (1999) 1073.
[15] F.X. Tanner, At. Sprays 14 (2004) 211.
[16] R.D. Reitz, At. Spray Technol. 3 (1987) 309.
[17] P.J. O'Rourke, A.A. Amsden, SAE Paper 872089 (1987).
[18] F.M. White, Viscous Fluid Flow, McGraw-Hill, New York, 1992.
[19] G.H. Agnew, Math. Mech. (1942) 222.
[20] I. Wygnanski, H.E. Fiedler, J. Fluid Mech. 38 (1969) 577.
[21] L.S. Liu, J.X. Wu, J. Therm. Sci. Technol. 1 (2002) 128.
[22] J.D. Whitlow, A.H. Lefebvre, At. Sprays 3 (1993) 137.