International Journal of Dynamics of Fluids

ISSN 0973-1784 Volume 6, Number 1 (2010), pp. 25–40

© Research India Publications
http://www.ripublication.com/ijdf.htm

Modelling of Externally Mixed Air Blast Atomizer

*Suresh Lall, A. Kushari2, J.C Kapoor3 and S. Maji4

1,3
Centre for Fire, Explosive & Environment Safety,
2
Indian Institute of Technology Kanpur, 4Delhi College of Engineering
*Corresponding Author: E-mail: *[email protected]

Abstract

This paper presents the result of experimental and theoretical studies of an

siphon fed externally mixed air blast atomizer. In this type of the atomizer,
atomization is attained by injecting a air stream at tip of the water injection
port. The water is siphon fed because of the pressure difference. The result
presented in this paper indicate that flow becomes supersonic beyond the
supply pressure of 0.8 bar. This paper also describes the prediction of
mathematical model that depics the two phase flow in externally mixed
atomizer. The model uses compressible flow relationships to predict the air
flow rate, a neural network based model to predict the water flow rate and a
novel droplet formation model to predict the droplet SMD. A statistically
distributed Weber number criterion is used to estimate the sizes of the droplets
produced in the atomization process. The liquid flow rate, air flow rate and
sauter mean diameter predicted by model are compared with the experimental
data and they are found to be in fair agreement with each other, suggesting
that model properly describes the physics of two- phase flow within the
atomizer.

Introduction
The process of atomization is one in which liquid is disintegrated into droplets by the
action of internal and/or external forces. In the absence of such forces, surface tension
tends to pull the liquid molecules together to form liquid jets or sheets. According to
Lefebvre[1], “Atomization can be considered as a disruption of the consolidating
influence of surface tension by the action of internal and external forces”.
The literature on the theory of droplet formation [2-5] is quite extensive, but it
deals with fairly low velocity and low Reynolds’s number flows, which are not very
important in practice. Currently, there is no known model of the disintegration of a
high velocity liquid discharge. This is due to the highly complex, turbulent and
26 Suresh Lal, A. Kushari, J.C Kapoor and S. Maji

random flow phenomena occurring during disintegration, which is not well

understood.
Existing theory of spray formation are based on instability of liquid jets and sheets
and are generally based on method of small disturbances. It is believed that the
disintegration of liquid jet or sheet is caused by the waves that form on the surfaces of
the jets and sheets of liquids. During instability, these waves rapidly grow into finite
amplitude waves that eventually break the liquid jets or sheets into a multitude of
small droplets. Reitz and Bracco [6] have classified the atomization process into
Rayleigh breakup, first wind induced, second wind induced and atomization regimes.
The Rayleigh regime is governed solely by capillary instabilities, while in the first
wind-induced regime; the capillary instabilities are enhanced by the aerodynamic
interaction with the surrounding gas. For the second wind-induced and atomization
regimes, the breakup is characterized by the formation of drops significantly smaller
than jet diameter. It is unfortunate that the atomization regime, which is most
important for practical applications, is the least understood one and no theory that
adequately describes the processes occurring in this regime exists. Table 1.1 shows a
classification of jet break up regimes:

Table 1.1: Classification of jet breakup regimes (Source Lefebvre).

Regime Description Predominant drop formation

mechanism
1 Rayleigh breakup Surface tension force, capillary waves
2 First wind-induced Surface tension force; dynamic pressure
breakup of ambient air
3 Second wind-induced Dynamic pressure of ambient air opposed
breakup by surface tension force initially
4 Atomization Unknown

The theoretical basis for the capillary instability was first investigated by
Rayleigh[7], who conducted a linear, inviscid analysis neglecting the effects of gas
pressure variations on jet distortion. He obtained the following relationship between
the initial jet diameter and the droplet diameter:
D = 1.89d (1)
Thus, for the Rayleigh breakup mechanism, the average drop size is nearly twice
the diameter of undisturbed jet.
Weber [8] developed a linear theory similar to Rayleigh’s that included the effects
of both the liquid viscosity and the pressure of the surrounding gas on the jet
behavior. He assumed that any disturbance causes rotationally symmetric oscillations
of the jet. If the wavelength of the initial disturbance is less than a particular
wavelength, the surface forces tend to damp out the disturbance. However, if λ is
greater than λmin, the surface tension forces tend to increase the disturbances, which
Modelling of Externally Mixed Air Blast Atomizer 27

eventually lead to disintegration of the jet. There is, however, one particular
wavelength, λopt, that is most favorable for drop formation. For viscous liquids
λmin = πd (2)
0.5
⎛ 3μl ⎞⎟
λopt = 2πd ⎜1 + (3)
⎜ ρ σ d ⎟
⎝ l ⎠
After breakup, a cylinder of length λopt and diameter d forms a spherical drop of
diameter D, so that
π π
d 2λopt = D3 (4)
4 6

No simple expressions, relating droplet size to other parameters, exist for the
wind-induced regimes and atomization regime. To predict the behavior of these
regimes, basic fluid dynamics conservation equations for both the liquid and air flows
are numerically solved, along with matching boundary conditions at their interfaces
[9-11]. Small perturbations or disturbances are imposed on the surface of the liquid
jets or sheets and the optimum wavelengths of disturbances that cause maximum
instabilities are estimated numerically, which are then correlated to the droplet sizes.
Although these theories can predict the conditions suitable for the initiation of jet or
sheet disintegration, they cannot accurately predict the size spectrum of the generated
droplets. These theories have successfully predicted the formation of droplets in well-
controlled laboratory conditions, as in the case of capillary droplet formation, but they
have not yet succeeded in accurately modeling the actual droplet formation in a
practical system. These fundamental studies have identified, however, an important
criterion that governs the formation and breakup of droplets. Specifically, the above
discussed research points to the existence of a critical value of the Weber number,
which is the ratio of inertial and surface forces acting on the surface of a liquid jet or
sheet, above which the liquid flow is very unstable and disintegrates into droplets.
The Weber number is defined as
ρ V 2D
We = a r (5)
σ
Therefore,
We * σ
D= (6)
ρ aV r2

Experimental studies of the disintegration of liquid jet and formation of droplets

also suggest the existence of a critical Weber number. Griffin and Muraszew [12]
suggested that the magnitude of the critical Weber number between 7 and 12,
depending on the flow conditions.
Sutherland, Sojka and Plesniak [13] have developed a model that describes the
performance of a ligament-controlled effervescent atomizer. Their model describes
the spray formation process as the breakup of individual cylindrical ligaments subject
to a gas stream and solves for the fastest growing disturbance, which is then
28 Suresh Lal, A. Kushari, J.C Kapoor and S. Maji

correlated to the droplet diameter. However, their model relies upon empirical
relationships to estimate the ligament diameters and the slip velocity between the gas
and liquid streams. Therefore, their model cannot be used to model the atomization
process in an airblast atomizer. Therefore, a new model has been developed to study
the behavior of the investigated atomizer.
The atomization modeling effort in the present study was primarily focused on the
modeling of a design newly developed air blast atomizer. This atomizer is a multi-
hole, siphon fed, air-blast atomizer.

Geometry of Atomizer
A sectional drawing of the atomizer assembly is shown in Fig.1. Liquid enters the
water chamber (2) through water inlet port (1). The liquid is sucked through three
hypodermic needles (3) because of the pressure difference created at the needle tip by
air pressure. The air is supplied through air inlet port (4) and injected through three
holes (5) at the upstream of water entry point. Two streams, air and water, interact at
the exit of the atomizer and droplets are discharged through outer chamber (6).

2 5
3
4
6

Figure 1: Schematic of the Atomizer [(1) Liquid inlet port, (2) Water chamber, (3)
Hypodermic needle for water injection, (4) Air inlet port, (5) Air discharge port, (6)
Outer chamber for discharge of droplets].

The major parameters required for the modeling effort are the number and size of
air injection holes, the number and size of water supply holes and the air pressure
used in the study [14]. Although the information regarding the water head would have
been beneficial in estimating the mist generation rate, in the absence of that
information a physical model for the pressure drop across the water head was
developed to predict the water flow rate. The model used fundamental physical
relationships to predict the atomizer performance and a novel model, based on energy,
force and mass balance was developed to predict the droplet SMD. Some physical
relaxation variables were introduced in the model to compensate for the real flow
effects.
The basic geometric parameters of the investigated atomizer used in this study are:
No. of air injection holes = 3
Modelling of Externally Mixed Air Blast Atomizer 29

No. of water ports = 3

Diameter of air injection holes = 0.7 mm
Diameter of water injection holes = 0.6 mm
Total area of air orifice = n(π/4)*d2 = 1.154 X 10-6 m2
Total area of water orifice = 8.48 X 10-7 m2

Experimental Set Up and Methods

Figure 2 shows the schematic of the experimental setup used for this study. The water
was siphon fed to the atomizer through water bottle. The water feed to the atomizer is
a function of air flow rate which was controlled using a control valve. The flow rate
of the atomizing air was controlled and measured using the air pressure regulating
valve and the calibrated rotameter respectively. For the sake of corrections in density
variation, the supply pressure of the atomizing air was closely monitored using a
pressure gauge.

4
1

6
5
7 8

Figure 2: Schematic of the mist generation setup [ (1) Externally mixing air atomizer,
(2) Pressure gauge, , (3) Air rotameter, (4) Control Valve, (5) Compressed air line (6)
Water Bottle, (7) Solid cone water spray, (8) Stand for holding atomizer, water bottle
and air rotameter,].

Figure 3 indicate atomizer characterization facility for droplet size measurement.

As indicated in this the atomizer was fixed on atomizer traversing system for two axis
(1 meter horizontal and 1 meter vertical) traverses of the atomizer. The atomizer was
connected with water and air feed system using flexible tubing for ease of the
traverse. The traversing system was computer controlled using microcontrollers. The
droplet size analyzer was fixed on the stand which was fixed on floor to avoid any
30 Suresh Lal, A. Kushari, J.C Kapoor and S. Maji

vibration during the measurements. The droplet size distribution measurements were
taken at a fixed location i.e. at a distance of 1 meter from the atomizer.

Air Line

Atomizer
Air Rotameter
Laser Sheet Water Bottle Pressure Gauge
Pressure
Laser Source Regulator
Atomizer
Traversing From
System Compressor
CCD Camera

PC

Figure 3: Schematic of the experimental setup.

An Ensemble Particle Concentration and Sizing apparatus (EPCS,

Insitec/Malvern®, Worcestershire, U.K.) was used to measure the average droplet
diameter of the spray. The operating principle of this apparatus is based on low angle
laser light scattering. The dynamic measurement range for this system was between 2
microns to 850 microns. To determine droplet diameter from the input signals, EPCS
uses a computer program (RTSizer, Insitec/Malvern®, Worcestershire, U.K.), which
is based on the Mie theory of light-particle interactions. The instrument uses 670
nanometer He-Ne laser with a beam diameter of 10 mm. The accuracy of the
instrument was ± 3% of full scale (specified by the manufacturer) and it could
measure the size distribution of sprays with obscurations up to 95%. The
measurement rate of equipment was one measurement every 400 microseconds.
Measurements were taken along the centerline of the spray cone at a distance of 1
meters below the exit of the atomizer. Sauter mean diameter (SMD) was used as a
meaningful quantity to represent the average of droplet diameter. SMD represents an
average droplet diameter with the same volume to surface area ratio as that of the
entire spray. It should be noted that for fuel combustion applications, emphasizes is
given to the use of this average diameter criteria, since it is least susceptible to a large
spread in the droplet diameter distribution.
The spray was required to be separated from the electronic noise and the
background. In order to filter out the electronic noise, a sample snap (keeping the lens
of the camera covered) was captured. This image was then subtracted from the spray
images in order to deduct the effect of the electronic components. To filter the
background optical noise, an image, containing only the background (keeping the
Modelling of Externally Mixed Air Blast Atomizer 31

spray off), was taken and the pixel wise intensity distribution was estimated. Once the
spray was turned on, the zone containing the spray got brighter than the background
resulting in an increase in the corresponding pixel intensities in the image matrix.
Thus, a pixel, in an image containing the spray, with an intensity value higher than the
corresponding values in the image containing solely the background was considered
to be the one representing the spray.
Figure 4 shows images of the spray produced by the presented atomizer at
different air flow rates. It is observed that with an increase in air flow rate, the kinetic
energy of the flow keeps on increasing causing a increase in spray cone angle. It was
observed that spray forms the solid cone at all the operating conditions.

(a) (b) (c) (d)

Figure 4: Spray Images at different ALR [(a) ALR = 1.147., (b) ALR = 1.117., (c)
ALR =1.057., (d) ALR = 1.004].

In the present study, the liquid supply pressure was kept constant and the air flow
rate through the atomizer was varied over a range of air supply pressure to obtain the
variation in ALR. The spray solidity was studied by taking pictures of the spray at
different liquid air supply pressures. The liquid flow rate corresponding to a particular
air flow rate (i.e .ALR) was measured using a calibrated rotameter. The liquid supply
pressure was then varied and the entire procedure was repeated for different values of
air supply pressure and a performance map was obtained.

Computational Modeling
For the computational modeling of the atomizer, three sub-models were developed.
These sub-models are:
1. Model for predicting the air flow rate
32 Suresh Lal, A. Kushari, J.C Kapoor and S. Maji

2. Model for predicting the water flow rate

3. Model for predicting the droplet SMD
Out of these three models, the model for predicting the air flow rate is most
critical as it provides the required air velocity and pressure that is used to model the
mist generation rate and the droplet SMD.

Prediction the air flow rate

The first attempt to predict the air flow rate was by using standard Bernoulli’s
equation and continuity equation, which assumed the flow to be subsonic and hence
incompressible. The equations used were:
m& a = ρAv
1 2
ρv = ΔP (7)
2
⇒ m& a = A 2 * ρ * ΔP
The density of the incoming air was estimated using perfect gas equation of state:
P
ρ= (8)
RT
The value of temperature was taken to be 300 K (270 C)
The compressible flow equations need to be used to model the flow properly. In
order to use the compressible flow relations, first we needed to estimate the exit Mach
Number of air. This was done by using the isentropic relationship between the supply
pressure (P0) and ambient pressure (P). The relationship used was
γ
P0 ⎛ γ − 1 2 ⎞ γ −1
= ⎜1 + M ⎟
P ⎝ 2 ⎠
⎛ γ −1 ⎞ (9)
⎜
2 ⎛ P0 ⎞ γ ⎟
M = ⎜⎜ ⎟ − 1⎟
γ −1⎜⎝ P ⎠ ⎟
⎝ ⎠
Value of γ was taken to be equal to 1.4. From the physics of compressible flows
it is well known that the flow cannot become supersonic without going through a
throat area (the minimum area) and the Mach number at the throat must be equal to
1.0. Therefore, following condition was used for the rest of the model
M a = M i KM i < 1
M a = 1.0K M i > 1
Where Ma is the actual Mach number and Mi is the isentropic Mach number.
Next, we estimate the static temperature (assuming total temp. T0 = 300 K), the
local speed of sound and the air velocity using the relations
Modelling of Externally Mixed Air Blast Atomizer 33

T0 γ −1 2
= 1+ M
T 2
a = γRT
v = Ma
Then, the exit air density is estimated using
1
ρ 0 ⎛ γ − 1 2 ⎞ γ −1
= ⎜1 + M ⎟ (10)
ρ ⎝ 2 ⎠
Next, the flow rate of air was calculated using the equation
m& a = C f ρAv (11)
Where, Cf is a constant to take care of the real gas and other un-modeled effects.
The value of Cf was chosen to be 2.35 for this study.

Prediction of water flow rate

Since the water was sucked into the flow domain by siphon effect, the information
regarding the water head was crucial to model the process. However, that data was not
available. Therefore, the head was assumed to be equal to zero. The air coming out of
the air injection whole expands rapidly to the available larger area at the point of
interaction between the water and air. This expansion leads to a pressure drop at that
point. Therefore, the pressure differential between the water supply (which is at
ambient pressure) and the lowered pressure at the point of interaction forces the water
into the flow domain. The given water flow rate data was used to form a neural
network based model to estimate the pressure drop due to the flow. The pressure at
the exit air orifice was considered to be equal to ambient pressure. The actual pressure
ratio at the point of interaction was estimated using the relationship
P0 ⎛ P0 ⎞
= ⎜ ⎟* N f
Pi ⎜⎝ Pa ⎟⎠
⎛ P ⎞
N f = ⎜⎜ 0.0034 * 0 + 0.9868 ⎟⎟ (12)
⎝ Pa ⎠

Where Nf is the factor estimated in the neurons by training the system with the
data available. After estimating the pressure ratio across the water supply, the pressure
drop was estimated using the relation
ΔP = Pa − Pi (13)
Then, the water flow rate is calculated using the continuity equation, i.e.,
m& w = C d Aw 2 * ΔP * ρ w (14)
The value of Cd used in this study was equal to 1.0.

Prediction of droplet SMD

The size of the droplet produced by an atomizer depends on various factors like
atomizer design, atomization mechanism, supply pressure, liquid properties (viscosity
34 Suresh Lal, A. Kushari, J.C Kapoor and S. Maji

and surface tension) etc. Therefore, it is not possible to propose a unified theory for
droplet formation. Most of the droplet size correlations available in the literature are
empirical relations that work over a limited range of operating conditions and for
specific atomizer designs. Keeping this in mind, a new phenomenological model to
predict the droplet SMD for the atomizer being investigated was developed. The
model uses basic mass and energy balance as well as the Weber number to predict the
droplet SMD. The details of the developed model are given in this section.
Let
Surface Energy of the droplets = E s = σS
Total number of drops = N
Surface area of a single drop (assuming spherical drops) = Si = πDi2
Where the subscript ‘I’ represents individual drops
NN
Therefore, Total Surface Area = ∑ ∑
S = S i = πDi = π Di2
2
∑ (15)
i i =1 i =1
N
Total mass of the atomized liquid = M = ∑ M i (16)
i =1
Where, mass of individual droplets = M i = ρVi (17)
4 π
Here, volume of a drop = Vi = πri3 = Di3 (18)
3 6
π
∴ M i = ρ lVi = ρ l Di3
6
N N (19)
π
⇒ M = ∑Mi = ρl ∑ Di3
i =1
6 i =1
N N
σ ∑ Si σπ ∑ Di2
E s σS
Surface energy per unit mass = = = i =1 = i =1
(20)
N N
M M π
∑ M i 6 ρ l ∑ Di3
i =1 i =1
N
∑ Di3
By definition, SMD = D32 = i =1 (21)
N
∑ Di2
i =1
Es 6σ
Therefore, = (22)
M ρ l D32
Now, Total energy of the liquid per unit mass = surface energy of liquid per unit
mass + kinetic energy of liquid per unit mass
E E E
⇒ T = s + k (23)
M M M
Modelling of Externally Mixed Air Blast Atomizer 35

Kinetic energy per unit volume of liquid can be given as

N N
1 1
Ek
∑ 2 M i vi2 ∑ 2 (ρ lVi )vi2
i =1
= = i =1 (24)
V ∑Vi ∑Vi
i i
Assuming vi2 = v 2 = average droplet velocity
Ek v 2
= (25)
M 2
Again, by definition
E
We = k
Es
⇒ E k = We * E s
6σ
⇒ ET = E s + E k = (1 + We )E s = (1 + We ) M (26)
ρ l D32
M
= ∑ Vi = V
ρl i
6σV
⇒ ET = (1 + We )
D32
Therefore, total energy per unit time of the droplets =
dET (1 + We )6σ dV 6σ &
= = (1 + We ) Ql (27)
dt D32 dt D32
Now, the energy of the droplets is primarily coming from the atomizing air blast.
Therefore, from energy conservation one can write
η
1
m& a v a2 = (1 + We ) 6 σ Q& l (28)
2 D 32
Where η = fraction of air energy transferred to the atomized liquid
Now,
6σ 6σ
1
2
(
η m& a v a2 = ρ l Q& l (1 + We ) ) ρ l D 32
= m& l (1 + We )
ρ l D 32
12 (1 + We )σ
⇒ η ( ALR ) v a2 =
ρ l D 32
m& a
ALR =
m& l
Therefore,
36 Suresh Lal, A. Kushari, J.C Kapoor and S. Maji

12 (1 + We )σ
D 32 =
η ( ALR )v a2 ρ l (29)

In order to use this model to estimate the droplet size, one has to specify the We as
well as the efficiency (η). In order to estimate the We, the air velocity and density
estimated in the first section of the model as used along with the measured droplet
size. The relationship used was

ρ a va2 D
We = (30)
σ

Results and Discussion

(1) The measured and predicted values of air flow rate are shown in Fig. 5. The data
in Fig. 5 shows that Bernoulli’s equation not only under predicts the flow rate
(which can be very easily corrected using a correction factor of 1.5) but it is also
unable to capture the proper slope of the curve. As can be seen from the measured
data, there is a change in the slope of the curve after a pressure of about 0.8 bar.
This deviation in slope is primarily due to the compressibility of air, which brings
in a change in air density and velocity.
The estimated exit Mach number is plotted in Fig. 6 with respect to the air
supply pressure. The results in Fig. 6 show that the flow becomes supersonic
beyond a supply pressure of 0.8 Bar (the same point were the air flow rate slope
changes).
The estimated air flow rate was converted to LPM within proper pressure
correction and the results are plotted in Fig. 7. The data in Fig. 7 shows excellent
matching between the experimental values and the model predictions capturing all
the details of the flow rate variation.
(2) The estimated water flow rate (kg/s) was converted into LPM and the results are
shown in Fig. 8. The data presented in figure shows remarkable matching
between the measured value and the model prediction (except for higher values of
Psupply, the reason for which is explained earlier).
It should be point out that the last three points in the given data shows a dip in
the water flow rate with an increase in air supply pressure. This is physically
impossible. This dip can be attributed to an appreciable change in the water head
during the course of experimentation. Therefore, those data points were excluded
while training the model. It should be pointed out that if the neurons were trained
using the full data set available, then the model will reproduce the exact
experimental data.
m&
The predicted ALR ( = a ) was estimated and is presented in Fig. 9 along
m& w
with the measured values. The deviations at higher air supply pressures are due to
the variation in water head.
Modelling of Externally Mixed Air Blast Atomizer 37

(3) The average We for the entire data set was calculated and that was found to be
equal to 43. Therefore, that value was used in the model. Similarly, the average
value of η was estimated to be equal to 0.013 and that was used in the model.
Using these numbers and the value of ALR estimated in the previous section, the
droplet SMD was calculated and the predicted droplet diameter were compared
with the measured values. The model predictions, along with the measured values
are shown in Fig. 10.
The data presented in Fig. 10 shows very good agreement between the model
predictions and the measured SMD. It should be noted that the model inputs were
only the atomizer exit geometry and the air supply pressure and the model
estimated the air flow rate, the water flow rate, the ALR as well as the droplet
SMD. The deviation of the model prediction from the measured data at higher
values of ALR was again due to the change in water head, which brings about an
error in the prediction of ALR. It should be pointed out that the model prediction
will be better if the atomized liquid is also pressurized and the liquid head
information is made available.

0.0035
Measured
Predicted (Bernoulli)
0.003
Corrected (Bernoulli)

0.0025
Air Flow Rate (Kg/s)

0.002

0.0015

0.001

0.0005

0
0 0.5 1 1.5 2 2.5 3 3.5
Air Supply Pressure (bar)

Figure 5: First estimate of air flow rate.

38 Suresh Lal, A. Kushari, J.C Kapoor and S. Maji

80
1.8

1.6 70

1.4 60
E x it A ir M a c h N u m b e r

Air Flow Rate (LPM)

Predicted
1.2 SUPERSONIC FLOW 50 Measured
1
40
0.8
30
0.6
20
0.4 SUBSONIC FLOW
0.2 10

0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 4
Air Supply Pressure (Bar) Air Supply Pressure (Bar)

Figure 6: Estimated exit Mach Number. Figure 7: Measured and predicted

variation of air flow rate with supply
pressure.

90 3.5
Predicted
80
Measured
70 3
Water Flow Rate (ml/min)

60
2.5
50
ALR

40
2
30
Predicted
20 Measured
1.5
10

0 1
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
Air Supply Pressure (Bar) Air Supply Pressure (bar)

Figure 8: Measured and predicted water Figure 9: ALR vs. Air Supply Pressure.
flow rate.

70

60 Predicted
Measured
50
Droplet SMD (μm)

40

30

20

10

0
0 0.5 1 1.5 2 2.5 3 3.5 4
Air Supply Pressure (Bar)

Figure 10: Predicted and measured droplet SMD.

Modelling of Externally Mixed Air Blast Atomizer 39

Conclusions
A phenomenological model is developed and used to predict the performance of a
siphon fed airblast atomizer. The model uses compressible flow relationships to
predict the air flow rate, a neural network based model to predict the water flow rate
and a novel droplet formation model to predict the droplet SMD. The model
predictions are found to be in good agreement with the measured data.

List of symbols and abbreviations

ALR Air liquid mass ratio
Cd Discharge Coefficient
D32 Sauter Mean Diameter (μm)
ma Mass flow rate of air (kg/s)
mw Mass flow rate of water (kg/s)
Pa Air supply pressure (Pa)
Pw Liquid supply pressure (Pa)
SMD Sauter Mean Diameter (μm)
Va Velocity of air (m/s)
Vw Velocity of liquid (m/s)
We Weber number of liquid ligament
ρa Density of air (kg/m3)
ρw Density of water = 1000 kg/m3
σ Surface tension of water = 0.0717 kg/s2
μw Viscosity of water = 0.00085 kg/ms

References
[1] Lefebvre, A. H., 1989. [1], “Atomization and Spray”., Hemisphere Publishing
Corporation.
[2] Chigier, N., 1993, “Spray Science and Technology”, Fluid Mechanics and
Heat Transfer in Sprays, ASME Fluid Eng. Div. Publ. FED, v 178, pp. 1-18,
ASME New York, NY.
[3] Lefebvre, A. H., 1983, “Gas Turbine Combustion”, McGraw Hill, New York.
[4] Yule, A, J., Dunkley, J, J., 1994, “Atomization of Melts”, Oxford University
Press Inc., New York,.
[5] Bayvel, L., Orzechowski, Z., 1993, “Liquid Atomization”, Taylor & Francis,.
[6] Reitz, R, D., Bracco, F, V., 1982. [7], “Mechanism of atomization of a liquid
jet”, Phys. Fluids, 25, 1730,
[7] Rayleigh, Lord., 1945, “The Theory of sound”, Dover Publications, NY.
[8] Weber, C., 1931. [9], "Disintegration of liquid jets", Z. Angew. Math. Mech.,
vol. 11, no. 2, pp. 136-159,
40 Suresh Lal, A. Kushari, J.C Kapoor and S. Maji

[9] Kushari, A., Neumeier, Y., Zinn, B, T., 2000, “ A Theoretical Investigation of
the Performance of an Internally Mixed Liquid Atomizer”, AIAA 2000-1021,
38th AIAA Aerospace Sciences Meeting and Exhibit, Jan, Reno, NV.
[10] Kushari, A., Neumeier, Y., Lubarsky, E., B. T. Zinn, “Heuristic Modeling of
Two-Phase Gas-Liquid Flow in an Internally Mixed Liquid Atomizer”, AIAA
2000-3493, 36th AIAA/ASME/SAE/ASEE Joint Propulsion Conference, July
17-19, 2000, Huntsville, Alabama.
[11] Kushari, A., Neumeier, Israeli, Y,O., Peled A., Zinn, B, T., 1999, "An
internally mixed injector for active control of atomization process in liquid
fueled engines", AIAA 99-0329, 37th AIAA Aerospace Sciences Meeting and
Exhibit, Jan. 11-14, Reno, NV.
[12] Giffen, E., Muraszew, A., 1953, “The Atomization of Liquid Fuels”, Chapman
& Hall Ltd., London,.
[13] Sutherland, J, I., Sojka, P, E., Plesniak, M, W., 1997, “Ligament-controlled
effervescent atomization”, Atomization and Sprays, vol. 7, no. 4, 99. 383-406,
[14] 14. Lal, Suresh., Kushari, A., Kapoor, J.C., Maji, S., 2006, “Characterization
of Externally Mixed Air- Assited atomizer,” ICLASS International conference
paper ID ICLASS 06- 265.