United Slates EPA"600/R"98-089
Environmental Protection
Agency Julv 1998
<&ERA Research and
Development
DEVELOPMENT OF
AN INNOVATIVE SPRAY DISPENSER
TO REDUCE INDOOR AIR EMISSIONS
FROM AEROSOL CONSUMER PRODUCTS
Prepared for
Office of Prevention, Pesticides, and
Toxic Substances
Prepared by
National Risk Management
Research Laboratory
Research Triangle Park, NC 27711
-------
FOREWORD
The U. S. Environmental Protection Agency is charged by Congress with pro-
tecting the Nation's land, air, and water resources. Under a mandate of national
environmental laws, the Agency strives to formulate and implement actions lead-
ing to a compatible balance between human activities and the ability of natural
systems to support and nurture life. To meet this mandate, EPA's research
program is providing data and technical support for solving environmental pro-
blems today and building a science knowledge base necessary to manage our eco-
logical resources wisely, understand how pollutants affect our health, and pre-
vent or reduce environmental risks in the future.
The National Risk Management Research Laboratory is the Agency's center for
investigation of technological and management approaches for reducing risks
from threats to human health and the environment. The focus of the Laboratory's
research program is on methods for the prevention and control of pollution to air,
land, water, and subsurface resources; protection of water quality in public water
systems; remediation of contaminated sites and groundwater; and prevention and
control of indoor air pollution. The goal of this research effort is to catalyze
development and implementation of innovative, cost-effective environmental
technologies; develop scientific and engineering information needed by EPA to
support regulatory and policy decisions; and provide technical support and infor-
mation transfer to ensure effective implementation of environmental regulations
and strategies.
This publication has been produced as part of the Laboratory's strategic long-
term research plan. It is published and made available by EPA's Office of Re-
search and Development to assist the user community and to link researchers
with their clients.
E. Timothy Oppelt, Director
National Risk Management Research Laboratory
EPA REVIEW NOTICE
This report has been peer and administratively reviewed by the U.S. Environmental
Protection Agency, and approved for publication. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Information
Service, Springfield, Virginia 22161.
-------
E PA-600/R-98-089
July 1998
Development of an Innovative Spray
Dispenser to Reduce Indoor Air Emissions
from Aerosol Consumer Products
by
P.E. Sojka and M.W. Plesniak
Thermal Sciences and Propulsion Center
School of Mechanical Engineering
Purdue University
West Lafayette, IN 47907-1003
EPA Cooperative Agreement Number: CR822618
Project Officer:
Kelly W. Leovic
Air Pollution Prevention and Control Division
National Risk Management Research Laboratory
U.S. Environmental Protection Agency-
Research Triangle Park, NC 27711
Prepared for:
U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
-------
Abstract
The operating principles and performance of a new type of spray nozzle are presented.
This nozzle, termed a "ligament-controlled effervescent atomizer," was developed to allow
consumer product manufacturers to replace volatile organic compound (VOC) solvents with
water and hydrocarbon (HC) propellants with air, while meeting the following restrictions: that
the spray mean drop size (reported here as Sauter mean diameter, or SMD) remain below 70 jam.
that the atomizing air consumption be less than 0.009, and that atomizer performance be
uncompromised by the increase in surface tension or by changes in viscosity. The current
atomizer differs from previous effervescent designs through inclusion of a porous disc located
immediately upstream of the nozzle exit orifice. The purpose of this disc is to control the
diameter of ligaments formed at the injector exit plane.
First, steady-state atomizer performance is reported in terms of the spray SMD. Drop
size data were analyzed using a model developed from first principles. The model describes the
spray formation process as the breakup of individual cylindrical ligaments subject to a gas
stream. Ligament breakup length is obtained using the expression of Sterling and Sleicher
(1975). Ligament diameter is estimated from manufacturer supplied pore size data for the porous
disc. The model correctly predicts the experimentally observed relationship between SMD and
air-to-liquid ratio by mass, liquid surface tension, and liquid viscosity.
Transient atomizer performance is reported second, again in terms of spray SMD. Issues
associated with timing of atomizing air and liquid product mass flow rates are considered.
Entrainment of ambient air into these sprays is reported last. Entrainment data were
obtained using a device similar to that described by Ricou and Spalding (1961). Entrainment
data were analyzed using the model of Bush and Sojka (1994), in concert with measured
momentum rate data that were acquired as part of this study. The analysis shows that
entrainment by sprays produced using this type of atomizer is predicted to within about 35% by
the expression E = —,mp , where E is the experimentally determined entrainment number
whose value is 0.15± 0.056 (2a), me is the entrained gas mass flow rate, x is the distance along
the spray axis measured from the dispenser exit orifice, pe is the density of the entrained air, and
M0 is the spray momentum rate at the exit orifice. Approaches for controlling entrainment (i.e.,
modification of the dispenser exit orifice geometry) are also introduced, and their utility
discussed in terms of their entrainment number values.
This report was submitted in fulfillment of CR822618 by Purdue Research Foundation
under the sponsorship of the U.S. Environmental Protection Agency's National Risk
Management Research Laboratory. This report covers a period from February 1994 to
September 1997 and was completed as of 9/22/97.
ii
-------
Contents
Abstract ii
List of Figures iv
List of Tables vii
Symbols viii
Chapter 1 Introduction 1
Drop Sizes 1
Entrainment 4
Entrainment Control 8
Chapter 2 Experimental Apparatus 10
Chapter 3 QA/QC 22
Effervescent atomizer and its supply system 22
Drop size measurements 22
Entrainment data 22
Momentum rate probe 22
Chapter 4 Results and Discussion 25
Steady-state Drop Size Measurements 25
Steady-state Drop Size Modeling 37
Transient Drop Size Measurements 41
Entrainment Measurements 43
Entrainment Control 53
Chapter 5 Summary and Conclusions 59
Summary 59
Conclusions 60
References 62
iii
-------
List of Figures
Figure 1. Conventional effervescent atomizer
Figure 2. Ligament-controlled effervescent atomizer
Figure 3. Experimental apparatus for transient flow rate control and drop size
analysis
Figure 4. Entrainment device
Figure 5. Entrainment data for a 1.0 g/s turbulent air jet, obtained using the
entrainment device of Figure 4
Figure 6. Radial droplet velocity profile for a ligament-controlled effervescent
atomizer produced spray formed from a 0.080 Pa-s viscosity, 0.030 Pa-m surface
tension fluid
Figure 7. Number averaged droplet diameter (Djq) versus radial position for a
ligament-controlled effervescent atomizer produced spray formed from a 0.080 Pa-s
viscosity, 0.030 Pa-m surface tension fluid
Figure 8. Momentum rate probe of Bush et al. (1996)
Figure 9. Calibration data for momentum rate probe of Figure 8
Figure 10. Nozzle geometries investigated during this study: inclined exit (one-point
crown), stepped exit, four-point crown, and two-point crown
Figure 11. SMD versus ALR for three fluids having viscosities of 0.020, 0.040, and
0.080 kg/m-s and a common surface tension of 0.030 kg/s2. Error bars represent 1
standard deviation
Figure 12. SMD versus ALR for three fluids having viscosities of 0.020,0.040, and
0.080 kg/m-s and a common surface tension of 0.067 kg/s2. Error bars represent 1
standard deviation
Figure 13. SMD versus ALR for two fluids having surface tensions of 0.030 and
0.067 kg/s2, a common viscosity of 0.020 kg/m-s, and operating at two liquid mass
flow rates. Error bars represent 1 standard deviation
Figure 14. Atomizer supply pressure versus ALR for three liquids having viscosities
of 0.020,0.040 and 0.080 kg/m-s, a common surface tension of 0.030 kg/s2, and a
common liquid mass flow rate of 0.6 g/s
Figure 15. Atomizer supply pressure versus ALR for three liquids having viscosities
of 0.020,0.040 and 0.080 kg/m-s, a common surface tension of 0.067 kg/s2, and a
common liquid mass flow rate of 0.6 g/s
2
5
12
14
16
17
18
19
20
21
26
27
28
29
30
iv
-------
List of Figures (Cont .)
Figure 16. Atomizer supply pressure versus ALR for two liquids having surface 31
tensions of 0.030 and 0.067 kg/s2, a common viscosity of 0.020 kg/m-s, and liquid
mass flow rates of 0.6 and 0.8 g/s
Figure 17. Near-nozzle structure for a 1.0 g/s water spray at an ALR of 0.015 32
produced by a conventional effervescent atomizer (i.e., without a porous insert)
Figure 18. Near-nozzle structure for a 1.0 g/s water spray at an ALR of 0.015 33
produced by a ligament-controlled effervescent atomizer (i.e., with a porous insert)
Figure 19. Near-nozzle structure at an ALR of 0.01, mass flow rate of 0.6 g/s, 34
viscosity of 0.020 kg/m-s, and surface tension of 0.067 kg/s2
Figure 20. Near-nozzle structure at an ALR of 0.0075, mass flow rate of 0.6 g/s, 35
viscosity of 0.020 kg/m-s, and surface tension of 0.067 kg/s2
Figure 21. Near-nozzle structure at an ALR of0.005, mass flow rate of 0.6 g/s, 36
viscosity of 0.020 kg/m-s, and surface tension of 0.067 kg/s2
Figure 22. Artist's rendition of near-nozzle hologram 37
Figure 23. Experimental data and predicted SMDs, based on an average pore 40
diameter of 37 ^m, for three fluids having viscosities of 0.020,0.040, and 0.080
kg/m-s, and a common surface tension of 0.030 kg/s2. Error bars represent 1 standard
deviation
Figure 24. Experimental data and predicted SMDs. based on an average pore 41
diameter of 37 jim, for three fluids having viscosities of 0.020,0.040, and 0.080
kg/m-s. and a common surface tension of 0.067 kg/s2. Error bars represent 1 standard
deviation
Figure 25. Experimental data and predicted SMDs, based on an average pore 42
diameter of 25 (am, for three fluids having viscosities of 0.020,0.040, and 0.080
kg/m-s, and a common surface tension of 0.030 kg/s . Error bars represent 1 standard
deviation
Figure 26. Experimental data and predicted SMDs, based on an average pore 43
diameter of 25 jim, for three fluids having viscosities of 0.020, 0.040, and 0.080
kg/m-s, and a common surface tension of 0.067 kg/s . Error bars represent 1 standard
deviation
Figure 27. Transient performance of a ligament-controlled effervescent atomizer 44
Figure 28. Normalized entrainment versus normalized axial distance for water being 45
sprayed at a mass flow rate of 1.0 g/s using a conventional effervescent atomizer
Figure 29. Normalized entrainment versus normalized axial distance for water being 46
sprayed at a mass flow rate of 1.0 g/s using a ligament-controlled effervescent
atomizer
v
-------
List of Figures (Cont.)
Figure 30. Normalized entrainment versus normalized axial distance for a fluid 48
(Fluid 5) having a viscosity of 0.040 Pa-s, a surface tension of 0.030 Pa-m, and
operating at liquid mass flow rates of 0.5 and 0.6 g/s
Figure 31. Momentum rate versus ALR for water 49
Figure 32. Momentum rate versus ALR, at two mass flow rates, for a fluid (Fluid 5) 50
having a viscosity of 0.040 Pa-s and a surface tension of 0.030 Pa-m
Figure 33. Entrainment number versus ALR for a fluid (Fluid 5) having a viscosity 52
of 0.040 Pa-s and a surface tension of 0.030 Pa-m, operating at two liquid mass flow
rates
Figure 34. Entrainment number versus ALR for all fluids 54
Figure 35. Normalized entrainment versus normalized axial distance for the inclined 55
(one-point crown) exit orifice
Figure 36. Normalized entrainment versus normalized axial distance for the first 56
style of two-point crown exit orifice (thinner porous insert)
Figure 37. Normalized entrainment versus normalized axial distance for the second 57
style of two-point crown exit orifice (thicker porous insert)
Figure 38. Normalized entrainment versus normalized axial distance for three 58
different styles of exit orifice
vi
-------
List of Tables
Table 1. Composition and Physical Properties of Spray Fluids (at room conditions). 11
Table 2. Data Quality Indicators. 23
Table 3. Data Quality. 23
Table 4. Instrumental Accuracy and Precision. 24
Table 5. Parameters for Experimental Investigation. 25
Table 6. Relative Velocities (in m/s) Between Atomizing Gas and Liquid for Five 39
Fluids at Four ALRs.
Table 7. Slopes and Coefficients of Determination for Least Squares Linear Fits to 47
Normalized Entrainment Versus Normalized Axial Distance Data for Liquids
Sprayed at 0.6 g/s.
Table 8. Slopes, y-Intercepts and Coefficients of Determination for Least Squares 50
Linear Fits to Momentum Rate Versus ALR Data.
Table 9. Average Entrainment Numbers. 51
vii
-------
Symbols
Roman Symbol
Description and Units
a
Ligament radius, m
a
Annular gap between spray and exit mask, m
A
Area, m2
ALR
Air/liquid ratio by mass, dimensionless
d
Diameter, m
E
Entrainment number, dimensionless
g
gravitational acceleration, m/s2
K
Modified Bessel function
m
Mass flow rate, kg/s
M0
Momentum rate, N
AP
Pressure drop, Pa
z
r
Coefficient of determination
SMD
Sauter mean diameter, jam
sr
Velocity slip ratio, dimensionless
U
Relative velocity, m/s
V
Entrained air velocity, m/s
X
Axial distance from the nozzle, m
Az
Exit mask thickness, m
Greek Symbol
a
Void fraction, dimensionless
P
Growth rate, dimensionless
^opt
Weber's optimum breakup wavelength, ^m
Viscosity, kg/m-s
P
Density, kg/m3
a
Surface tension, kg/s2
%
Instability wavenumber, dimensionless
e
S opt
Critical wavenumber, dimensionless
Subscript/Superscript
e
Entrained air
err
Error, or uncertainty
ex
Exit mask
8
Atomizing gas
I
Liquid
L
Ligament
0
At the nozzle exit
A
Air
E
Exit
viii
-------
Chapter 1
Introduction
The Pollution Prevention Act of 1990 states that, if possible, pollution should be
prevented at the source. This is often difficult to achieve due to the potential for transferring
emissions from one medium to another. The optimum approach would be to eliminate the
sources of pollution, which for some consumer product aerosol sprays are volatile organic
compound (VOC) solvents and hydrocarbon (HC) propellants. Simply removing VOCs and HCs
affects the quality of the spray delivered by current twin-fluid and pressure-swirl atomizers.
Consequently, an atomizer whose performance is independent of solvent and propellant type
would be very useful. The best situation would be an atomizer that would allow water to be
substituted for the VOC solvent, air to be substituted for the HC propellant, and performance to
remain uncompromised. These goals provide the motivation for this study.
Achieving these goals for twin-fluid atomization requires a substantial reduction in
propellant consumption for three reasons. First, deceptive packaging guidelines suggest that at
least 60% of a spray container be filled with product. Second, U.S. Department of
Transportation container charging restrictions limit package pressures to less than 1 MPa (147
psig) for systems employing non-liquefied propellants. Finally, a minimum propellant pressure
is required to supply the last of the product, so not all of the propellant mass can be used to form
sprays. The result is an upper bound on atomizing air consumption of less than 0.01 of the liquid
product to be dispensed.
An air/liquid ratio (ALR) by mass of 0.01 or less is outside the range of conventional
twin-fluid nozzles. The only design that comes close to meeting this criterion is the effervescent
atomizer. As will be demonstrated below, a new type of effervescent atomizer can achieve the
goals stated above.
Effervescent atomization is characterized by actively introducing gas bubbles into a
liquid flow immediately upstream of the exit orifice, thereby forming a two-phase flow. This
allows an efficient transfer of energy between the atomizing gas and the liquid so a high quality
spray may be produced at ALRs lower than those required by most conventional twin-fluid
atomizers. A typical "conventional" effervescent atomizer is shown in Figure 1.
Drop Sizes
A number of investigators have studied effervescent atomizer-produced sprays. Early
work includes: Lefebvre et al (1989), who demonstrated very good atomization with mean drop
sizes comparable to those obtained with air-assist atomizers operating at much higher ALRs;
Wang et al (1989), who showed that orifice diameter and gas injector geometry had little effect
on the quality of atomization; Roesler and Lefebvre (1988), whose photographic results showed
that bubble explosions were an important mechanism in the atomization process at low ALRs
and that bubble spacing influenced droplet size at these conditions; and Whitlow and Lefebvre
(1993) whose most significant result was the observation that acceptable atomization could be
achieved when using an orifice geometry that turns the two-phase flow through an angle just
prior to ejection from the nozzle.
1
-------
Other investigators have been concerned with the influence of fluid rheology on atomizer
performance. Most notable are the studies of Buckner and Sojka (1993), who investigated the
effects of viscosity and non-Newtonian fluid rheology on mean drop size and concluded that
viscoelasticity controlled spray formation, and Geckler and Sojka (1995), who developed an
analytical model that successfully predicted the influence of viscoelasticity on mean drop size.
Air
Brass Top
Plate
Brass
Containment
Tube-
I
' \V\XN\^X
\\ \ \
\v\\NN\NX\V^."vV\\
^ vVX vV
\ -V \
\\\ \>
-------
Studies most pertinent to the current work are those of Santangelo and Sojka (1995) and
Lund et al. (1993), all of whom performed investigations of the near-nozzle breakup regions of
effervescent sprays. Santangelo and Sojka (1995) employed a focused image holography system,
while Lund et al. (1993) used high speed photography to obtain images of this region.
Santangelo and Sojka (1995) constructed holograms for sprays formed from fluids having
three different viscosities and two different surface tensions in order to study how fluid physical
properties affect the near-nozzle structure and, ultimately, Sauter mean diameter (SMD). Their
holograms revealed that the jump in SMD associated with operation at low ALRs was the result
of a transition in near-nozzle structure from liquid breakup dominated by single bubble
explosions to formation of an annular ring of smaller diameter ligaments. Breakup of these
smaller ligaments resulted in a decrease in SMD.
Lund et al. (1993) utilized near-nozzle images of the breakup structure in the
development of an analytical model to predict SMD. Their model is based on the ligament
breakup analysis of Weber (1931), whose analytical expression for the hydrodynamic instability
mode having the maximum growth rate is:
where A.opt is Weber's (1931) optimum ligament breakup length, dL is the ligament diameter, and
|ii, ph and CT| are the liquid viscosity, density, and surface tension. Lund et al. (1993) assumed
that each ligament forms a single spherical drop with a diameter equal to the SMD. Initial
conditions for their model were determined using the velocity slip ratio expression of Ishii
[where sr is the velocity slip ratio (quotient of gas and liquid velocities), pg is the gas density,
and a is the void fraction (quotient of gas to gas-plus-liquid volumes)] and conservation of mass
for the air and liquid streams,
(1)
(1977),
,r= IE j ^
pg \;l + 75(l-a)
(2)
a =
(3)
After some manipulation, Lund et al. (1993) showed that
(4)
3
-------
This model accurately predicts the viscosity and surface tension scaling observed in their
experimental data.
The experimental work of Lund et al. (1993) demonstrated that a sub-70 Sauter mean
drop size (SMD) spray was attainable only at ALRs above 0.02. The work of Santangelo and
Sojka (1995) attributed this minimum ALR to a transition in near-nozzle breakup occurring at
about this ALR. Therefore, further reduction in SMD at low ALRs is not possible without a
change in the breakup structure at the exit. Ligament-controlled effervescent atomizers were
designed to avoid the transition in breakup structure that occurs, allowing their use in
applications that require low air consumption; i.e. consumer product sprays.
As further evidence of the potential for consumer product spray formation via ligament-
controlled effervescent atomization, the work of Whitlow and Lefebvre (1993), in conjunction
with that of Roesler and Lefebvre (1989), Lefebvre et al. (1988) and Wang et al. (1989)
demonstrates that effervescent atomizers are capable of achieving large cone angles with little
sensitivity to exit orifice diameter, over the range of 1 to 3 mm. These findings indicate that a
single design can be employed for a wide variety of products, thereby reducing unit costs.
Collectively, this information demonstrates that ligament-controlled effervescent atomizers
represent a viable alternative to current twin-fluid atomizer designs used in consumer product
applications.
The observations of Lund et al. (1993) and Santangelo and Sojka (1995) were used to
design the ligament-controlled effervescent atomizer shown in Figure 2. It differs from earlier
effervescent atomizers because a porous plug was inserted into the exit orifice. The plug, made
of sintered plastic with pores nominally 25 jam in diameter, allows sprays to be formed at ALRs
as low as 0.0075. Spray performance, in terms of SMD, is described in Chapter 4.
Entrainment
While SMD is an important aspect of spray performance, it is not the only indicator. One
topic of spray research which has not been studied in detail is entrainment. Entrainment, for the
purposes of this discussion, is defined as the quantity of ambient gas which is drawn in through
the interface of a spray as it expands downstream of the nozzle. It is the result of momentum
transfer, from both liquid drops and any atomizing gas used, to the ambient air surrounding the
spray.
Entrainment has important implications in many engineering applications. In consumer
product sprays, the entrained mass flow rate has a cause and effect relationship on carrier liquid
evaporation. In combustion applications such as gas turbines and diesel engines, entrainment has
a significant effect on the local equivalence ratio and, therefore, a direct impact on nitrogen
oxides (NOx) formation. In furnaces, entrainment has a large impact on droplet residence time
since higher entrainment rates lead to greater rates of spray deceleration and, therefore, slower
moving drops. Entrainment is also important in spray drying, where the goal is to remove as
much liquid as possible from a droplet during the drying process. Finally, entrainment can
influence the finish quality of painted or coated surfaces through both transfer efficiency and
solvent evaporation.
4
-------
A key to understanding entrainment is the ability to measure it. Three methods are
common: experimentally measuring the velocity profile of the entrained air and then integrating
it over a control volume that encompasses the spray, experimentally measuring the global
entrainment rate, and numerically simulating the two-phase spray field. Studies that have used
these techniques, and that are relevant to the current work, are summarized below.
Air
Brass Top
Plate
Liquid
Brass
Containment
Tube
Brass
Bottom
Plate
i
¦ >.Vv
\ \ \\ N
\\ \ \
\ \\x \ \
\ N s\\W
s\\X\\\\\ \ V* \u\\
^ \\\ \ N
"
N\vw
//////
/y///
K'JZ/,
y / / // / ,
' y / "
""'t/sttAf!
""/'/"/A// ' /f//'/
, , '/,/ / / / / / y/ /
/ / / / /%/
my
/yym
/ ''/*/*" * ''f *// '
y/// / / ///'/ , / '
?ws/*jzay)
'/ 'v/'//// 'MM
v///,y/,y/y/y/y/y.
''/y/''/ i/'v/
\v\
\ Vs \ XV ^
•xN^NNS \\Xw V
\ \
/?//, ///'/(yy/'M'/
/ / / / /
¦ / y// ' / /y y
yus/vyr* ?(*>
y;uyj'<
////4M
// J /,yy//y/
yjy//yyy
"// ///// /
,, ' '' ///' / mw
'/ss&m//''*'
\ s .
0^
Exit Orifice-7 L Porous Insert
Figure 2. Ligament-controlled effervescent atomizer.
Liquid
Brass
Aerator
Tube
Air
Injection
Holes
5
-------
Binark and Ranz (1958) measured velocities outside the spray using a constant-resistance
hot wire anemometer and inside the spray using an impact-static probe. Data were used to
develop an expression that yielded induced air velocities for a set of homologous nozzles.
Unfortunately, their expression is of limited use for estimating entrainment into sprays because
atomizer-specific experimental data are necessary in order to apply their model to other types of
atomizers. Binark and Ranz (1958) also developed a theoretical model that yielded results within
an order of magnitude of their experimental data. However, it required the assumption that the
velocity of the drops was constant and equal to the injection velocity. If this assumption were
correct, no transfer of momentum would occur between the drops and the surrounding air and,
thus, no air would be entrained.
Rasbash and Stark (1962) correlated spray entrainment in terms of "reaction at the
nozzle," or the force exerted on a flat obstacle oriented normal to the spray axis. The fact that
Rasbash and Stark (1962) present their results as an empirical correlation suggests that their
expression is applicable only to sprays formed using their type of atomizer.
Briffa and Dombrowski (1966) measured entrained air velocity by seeding the air in and
around a flat spray with lycopodium powder and taking double-flash (microsecond duration)
photographs. Air velocities inside the spray were measured assuming the air velocity was
tracked by the 15 nm diameter droplets. A model was developed which demonstrated a linear
relationship between entrained mass flow rate and axial distance. However, a non-linear
relationship was observed experimentally
Benatt and Eisenklam (1969) used the Ricou and Spalding (1961) method to measure
global entrainment and compared results obtained to predictions from a model based on their
knowledge of the liquid sheet breakup process, spray dynamics, momentum loss calculations,
and induced gas flow observations. The expected linear scaling of entrainment with axial
distance was observed, and experimental results confirmed their theory. However, the theoretical
proportionality constant was lower than the experimental value. Benatt and Eisenklam's (1969)
model is restricted to pressure-swirl atomizers, since it relies on calculations based on a
particular breakup process.
Tishkoff (1985) studied air entrainment into two different pressure-swirl atomizers by
visualizing the entrained air flow patterns using helium jet seeding and measuring velocities
using a constant-temperature hot-wire anemometer. He developed a correlation for the
entrainment number, which is limited to pressure-swirl atomizer produced sprays.
MacGregor (1991) developed a simple model based on jet momentum rate and the drag
on spherical droplets and then used a Ricou and Spalding (1961) device to test it. Experimental
results obtained from two nozzles show a linear relationship with inlet mass flow rate. However,
the gradient of the experimental data implies that there exists a critical mass flow rate below
which little or no entrainment occurs, suggesting little or no jet breakup is occurring under these
conditions. In addition, predictions did not agree with experimental results. Nonetheless,
MacGregor (1991) used his model to demonstrate the effects of drop size distribution parameters
and noted that entrainment rates were a linear function of axial distance for drops having
diameters greater than 1000 Jim, that entrainment rates fell off at large axial distances for smaller
drops, and that entrainment rates were relatively insensitive to the width of the drop size
distribution.
6
-------
Boysan and Binark (1979) and Rothe and Block (1977) performed numerical simulations
in order to predict induced air flows into sprays. Boysan and Binark (1979) assumed a Rosin-
Rammler drop size distribution and numerically solved the partial differential equations for
stream function and vorticity transport coupled with the ordinary differential equations of spray
motion. Good agreement was observed upon comparison of predicted and measured air
velocities. However, this agreement is misleading due to the incorporation of adjustable
parameters into their model. These parameters would have to be determined for each new type
of spray.
Rothe and Block (1977) were interested in the cone angle contraction due to entrained gas
flow that was noted earlier by Binark and Ranz (1958). They numerically solved equations for
drag, axial momentum, and drop trajectory in order to calculate the induced air velocity and
amount of spray contraction. Comparisons of model predictions with experimental data of
previous researchers were presented. Again, the level of agreement between model predictions
and experimental results is misleading due to the use of fitting parameters in their analysis.
Ruff et al. (1989) were among the first to consider entrainment in the dense region of the
spray; i.e. where the liquid is breaking up. Their analysis was based on the assumption of a
locally homogenous flow (LHF). Mean streamwise and radial entrainment velocities were
obtained using laser Doppler anemometry near the edge of the spray. These velocities were then
integrated to provide experimental entrainment rates. Predicted values were greater than
measured values in all cases. This is because entrainment rates are strongly related to flow
properties near the edge of the spray. Unfortunately, this is also a dilute region of the spray
where the LHF model over-predicts quantities such as the entrainment rate.
Until recently, no studies addressing entrainment into effervescent sprays had appeared in
the literature. Bush and Sojka (1994) were the first to study entrainment in such sprays, using an
injector similar to that shown in Figure 1. As such, they were the first to study entrainment in
two-phase jets characterized by velocity slip between the two phases at the nozzle exit. Bush and
Sojka (1994) developed a two-phase model analogous to that of Ricou and Spalding (1961) that
was based on dimensional analysis and the conservation of momentum. A Ricou and Spalding
(1961) device was then used to measure entrainment rates. They found that "conventional"
effervescent sprays entrain air similar to other sprays and single-phase gas jets, in that the
normalized entrainment rate scales linearly with the dimensionless axial distance. However,
their results also show that "conventional" effervescent sprays are not characterized by a single
entrainment number, as are their single-phase gas jet counterparts, instead requiring scaling by
liquid density and nozzle diameter. Recent work by Luong (1996) suggests this lack of a single
value for entrainment number is due to unsteadiness inherent in conventional effervescent sprays.
This review of the spray entrainment literature demonstrates that previous work cannot be
directly applied to ligament-controlled effervescent atomizer produced sprays. Consequently, the
current study extended previous work to this important class of atomizer by answering the
practical question "At what rate does a ligament-controlled effervescent atomizer produced spray
entrain surrounding air?" The current study also answered a question of more fundamental
importance; i.e. "Can entrainment by sprays having initial inter-phase velocity slip be modeled
using the momentum-rate approach first suggested by Ricou and Spalding (1961) for gas jets?"
7
-------
Entrainment Control
Entrainment per se was not the only topic of interest during this study. Early work
indicated that entrainment by ligament-controlled effervescent atomizer produced sprays was less
than that produced by conventional effervescent atomizer produced sprays. As such, entrainment
control, or modification, was deemed worthy of consideration.
While entrainment control in sprays has received scant attention, gas flow entrainment
has been the subject of a number of studies, with local entrainment enhancement and suppression
well documented in the literature for a number of free shear flows, including jets. Both passive
and active control strategies have been implemented. Passive control may be thought of as the
capacity to modify jet development jet by directing energy into selected scales of turbulence, or
by modifying coherent structural flow features. The augmentation of large scale turbulent
motions promotes mixing and entrainment. Variation of the nozzle geometry constitutes the
primary passive means of controlling the structure and entrainment of the flow. Active control,
on the other hand, requires an energy input to the flow, for example by acoustic excitation. It is
necessarily more complex and costly than passive control, and thus not as well suited to
consumer sprays.
Based on practical considerations gleaned from the gas flow literature, we focused on
passive entrainment control for sprays. The following summary of the gas flow entrainment
literature provides rationale for the types of passive control schemes considered.
In planar gas mixing layers, the introduction of streamwise vorticity has been found to
enhance entrainment in the near field (Bell and Mehta, 1993). A number of passive devices have
been used to generate the streamwise vorticity, including half-delta-wing "vortex generators,"
and cylindrical pegs distributed across the span of the splitter plate. Modifications to the splitter
plate trailing edge have also been utilized. For example, serrated or corrugated (wavy) trailing
edges give rise to streamwise vorticity and alter the near-field entrainment. The application of
this concept in "lobed mixers" for combustion applications is well known (Eckerle et a!., 1992).
Jets are formed by the merger of mixing layers present around their periphery. Changes
in nozzle geometry are known to have a profound effect on jet spreading and hence entrainment
in a wide number of flows, including subsonic and supersonic single-phase jets and particle-
laden jets. We shall briefly review some of the rich literature on this topic.
Ho and Gutmark (1987) studied elliptic jets with a 2:1 aspect ratio, while Krothapalli et
al. (1981) studied rectangular jets, and Schadow et al. (1988) studied triangular jets. These
asymmetric nozzle configurations modify the shear layer turbulence. In the elliptical jet, the
shear layer vortices became distorted azimuthally due to self-induction causing greatly increased
entrainment near the minor axis in the near-field. The total entrainment was several times higher
than that in a round jet.
W'lezien and Kibbens (1986) studied a set of nozzles with "indeterminate origins." These
were initially circular jets with inclined and stepped exit geometries (see Figure 1). In general,
the jet spreading (hence entrainment) increased in the nozzle planes of symmetry and decreased
in the perpendicular planes. They found that the degree of control effected is dependent not only
on the nozzle geometry, but also on the ratio of the shear layer instability wavelength to the
8
-------
nozzle scale. Relatively thick shear layers were necessary to achieve significant flowfield
modification.
Longmire et al (1992) used a series of crown-shaped nozzles in a particle-laden jet to
control the jet characteristics. To varying degree, these shapes impacted local entrainment in an
asymmetric manner. Because of their geometries, they were termed nozzles of "indeterminate
origin."
In addition to these indeterminate origin experiments, various investigators have added
tabs or vortex generators to alter jet spreading characteristics, cf. Krishnappa and Csnady (1969),
Bradbury and Khadem (1975), Zaman et al. (1994), and Zhang and Schneider (1995). Zhang and
Schneider (1995) used various numbers (2 and 4) of small rectangular tabs, which protruded into
a round jet, to enhance mixing and found the tabs induce more rapid velocity decay and jet
spreading. Mixing of jet and ambient fluid increased up to 45%.
9
-------
Chapter 2
Experimental Apparatus
The atomizer, air and liquid supply systems, rheology instrumentation, drop size
instrumentation, imaging systems, entrainment apparatus, and momentum rate probe used to
acquire the data presented here are described in this chapter.
Figure 2 illustrates the ligament-controlled effervescent atomizer used in this study. It is
made entirely from brass and consists of a top plate, a containment tube, an aerator tube, and an
exit orifice (bottom) plate. Liquid is fed into the side and flows downward through an annular
gap between the containment and aerator tubes. Air, supplied from the control panel, is injected
into the liquid through two holes located at the bottom end of the aerator tube, creating a two-
phase flow. The flow then passes downward through a porous medium before leaving through
the exit orifice.
The aerator tube is 133 mm long, has an outside diameter of 3.2 mm, and passes through
a Cajon "Ultra-torr" vacuum fitting, threaded into the atomizer top plate, which allows for fine
adjustment of the aerator tube position relative to the exit orifice. The containment tube has an
outside diameter of 50.8 mm and an inside diameter of 3.7 mm. The gap between the aerator and
containment tubes was sized at 0.3 mm in order to create a downward liquid velocity sufficient to
counteract the buoyancy of the air bubbles. The exit orifice plate has a diameter of 50.8 mm and
a thickness of 3.2 mm. A 4.1 mm diameter blind hole with a depth of 2.95 mm is used to hold
the porous medium in place, just upstream of the exit orifice. The exit orifice diameter is 0.38
mm and its length is 0.25 mm. A very short exit length was used in order to minimize
coalescence of either bubbles generated inside the atomizer or liquid ligaments formed in the
porous medium.
The porous medium was obtained from Porex Technologies. It is a polyvinylidene
fluoride (PVDF) disc 4.1 mm in diameter and 1.0 mm thick. The pore diameter and porosity (a
ratio of the volume of the void space to the total volume of the medium) will later be shown to be
important inputs to our spray formation model. Values were measured by the manufacturer
using mercury intrusion porosimetry and reported to be 37 ^m and 49%, respectively (Reed,
1996.).
Air was used as the atomizing gas and to pressurize the free surface of the liquid. Air
flow rate was monitored using a Matheson 602 rotameter with a stainless steel float and
regulated using a Nupro B-SS2-D needle metering valve. A Nupro C-series check valve was
placed immediately upstream of the aerator tube to prevent backflow of liquid into the air
system. Calibration was effected by timing and measuring the volume of gas passing through the
rotameter using a dry test meter. Rotameter pressure was kept constant by monitoring a gauge.
Laboratory temperature was thermostatically controlled to within ± 2 C. A straight line fit to the
calibration data gave a coefficient of determination (r2) of 0.991.
The liquid was supplied from a steel sphere whose free surface was pressurized. The
liquid mass flow rates for the low viscosity fluids were monitored using a Matheson 604
rotameter with a stainless steel float, while flow rates for the higher viscosity fluids were
measured using a Matheson 605 rotameter with a stainless steel float. Flow rates were regulated
using a Whitey SS-1RS4 integral needle valve. A Nupro 15 |im in-line filter was placed
10
-------
immediately upstream of the atomizer liquid inlet port in order to prevent clogging of the
atomizer. The liquid rotameter was calibrated by collecting, timing, and measuring the volume
of fluid passing through the rotameter at several different settings. Straight lines were fitted to
the data for all liquids, resulting in coefficients of determination (r~) greater than 0.95 for all
cases.
The nozzle was suspended over an exhaust system for all experiments. The exhaust
system contained a sump to remove liquid that collected in the bottom of the duct, and a blower
to exhaust the spray and remove any airborne particles.
Since a major objective of this study was to develop an atomizer whose performance was
not compromised by the viscosity and surface tension of the fluid being sprayed, accurate
methods for measuring these properties were necessary. Fluid viscosity was measured using a
Haake falling ball viscometer. Viscometer accuracy was confirmed using calibration oils of 9
and 98 mPa-s with values measured to within 5% of published data. The manufacturer states that
the instrument accuracy is 1% in this range. Surface tensions were measured using a CSC model
70535 du-Nuoy ring tensiometer. This instrument was calibrated by placing a known weight on
the ring and measuring the resulting force. Fluid densities were calculated from the quotient of a
known volume of fluid and its measured weight. Weights were measured using a Mettler model
P1200N electronic balance (having an accuracy of 0.01 g) with volumes measured using a
graduated cylinder (accuracy 1%). Fluid physical properties are summarized in Table 1.
Table 1. Composition and Physical Properties of Spray Fluids (at room conditions).
Fluid
Composition, weight-%
Viscosity,
Surface
Density,
Number
Pa-s
Tension, Pa-m
kg/m3
1
63/37 Glycerine/Water
0.020
0.067
1170
2
72/28 Glycerine/Water
0.040
0.067
1197
3
80/20 Glycerine/Water
0.080
0.067
1217
4
75/25 SNO-100/BennzoiI
0.020
0.030
840
5
90/10 SNO-lOO/SNO-320
0.040
0.030
847
6
Water
0.001
0.072
998
Drop size distribution data were obtained using a Malvern 2600 Particle Size Analyzer
fitted with a 300 mm focal length receiving lens. All drop size measurements were taken with
the laser beam passing through the center of the spray at a location 15 cm downstream of the exit
orifice. Each measurement consists of 3000 samples. A minimum of three measurements were
obtained at each operating condition.
Qualitative information about breakup mechanisms leading to drop formation was
obtained using high-speed photography and focused-image holography. Magnified images of
near-nozzle spray structures were obtained using high-speed black-and-white photography.
Images of approximately lOx magnification were obtained using a conventional Nikon 35 mm
single-lens reflex (SLR) camera, a 55 mm focal length lens, and a bellows extension. The light
source was a 500 ns duration pulse generated using an EG&G Microflash. The camera was set to
11
-------
an f-stop of 1.8, with the exposure time fixed by the flash duration. Images were captured on
Kodak TMAX ISO 400 black-and-white film.
Holographic images of the near-nozzle structure were obtained using the focused-image
holographic system of Santangelo and Sojka (1994). A general overview of focused-image
holography as a spray diagnostic tool is provided by these authors (1993). The recording
medium was Agfa HD 8E75 NAH holographic plates. The holographic plate developing system
used was supplied by H.I. Bjelkhagen of Northwestern University and E. Wesley of Lake Forest
College. Specific details are provided by Santangelo (1993).
The effervescent atomizer used in the steady-state SMD measurements was used for the
transient drop size measurements, although some minor modifications were performed. In
particular, special care was taken to minimize the distance that the liquid traveled from its
injection at the side of the containment tube, down through the annular gap between the
containment and aerator tubes, to the point where it formed a two-phase flow (immediately
above the porous insert). The addition of rubber "biscuits," to minimize the void space in the
atomizer, decreased the "dead" time which occurred between atomizing air and liquid product
valve actuation and spray formation. Decreased dead time improved spray quality during
transient operation.
The transient spray control system is shown in Figure 3. It consists of the following main
components: Pneutronics Voltage Sensitive Orifice (VSO) proportional control solenoid valves
to control the spray pulse(s); Alicat flowmeters; a liquid storage reservoir; a personal computer
(PC) equipped with LabVIEW software and I/) board for on-line data acquisition, valve control,
and drop size instrumentation triggering; and an interface box for signal conditioning.
Air Inlet
Liquid Fill Inlet
0
Stop Valve
Regulating Valve
0
Pressure Guagc
0
Solenoid Valve
0
Filter
Relief Valve
r-1
Flow Meter
©
©
Liquid Storage
Reservoir
Atomizer
M Malvern
Figure 3. Experimental apparatus for transient flow rate control and drop size analysis.
12
-------
The valves were VSO Model #1 (0.254 mm diameter orifice) for the atomizing air side
and Model #3 (0.762 mm) for the liquid product side. They offer independent control of the
initiation, ramp up, ramp down, and cessation for each stream. Both valves were computer
controlled using LabVIEW.
The flowmeters were custom made by Alicat. The atomizing air side unit had a 10 msec
response time and could operate at flow rates between 0 and 1 SLPM. The liquid product side
unit also had a 10 msec response time and could also operate at flow rates between 0 and 1 LPM.
Both flowmeters were sampled by the PC using the LabVIEW software.
The liquid storage reservoir was constructed of commercially available schedule 80 PVC
pipe. It was filled with liquid product prior to a series of runs. Liquid was forced into the
atomizer by pressurizing the free surface inside the reservoir using dry, high pressure air.
The same Malvern Spray Analyzer that was used to make steady-state drop size
measurements was used when making transient drop size measurements. The Malvern laser
beam was again located 15 cm downstream of the dispenser exit orifice. Time resolved drop size
data was sampled in 10 usee windows.
The PC was a 33 MHz, 486 manufactured by Gateway. It was fitted with National
Instruments Model CIO-DAS08-PGH digital and analog I/O board, which was used for on-line
data acquisition (flowmeters), valve control, and Malvern triggering.
A custom built interface box was used for signal conditioning. It linked the CIO-DAS08-
PGH board with the atomizing air and liquid product solenoid valves, with the atomizing air and
liquid product flowmeters, and with the Malvern.
Pulsed spray operation was performed by first specifying the time duration of the spray —
a value of two seconds was arbitrarily selected for this study. A "recovery" time, i.e. the time
between successive spray events was chosen next. A value of five seconds was arbitrarily
selected for this study. The sampling window for drop size measurements by the Malvern was
then specified. This window is 10 jisec long, and can be placed anywhere within the 2 sec spray.
It was "scanned" from spray initiation to spray cessation during tests. Sufficient spray samples
must be accumulated to ensure that the resulting time history is statistically significant. As such,
100 spray events were sampled at each time window. All transient drop sizes reported are the
result of averaging over 100 spray events at each instant in time. Finally, ALR values are
derived from averaged flow rate measurements, as calculated by LabVIEW software running on
the PC.
Global entrainment rates were measured using a device similar the one developed by
Ricou and Spalding (1961). Their device consisted of a cylindrical housing that enclosed a
nozzle mounted to a back plate. Entrained gas was injected into the cylinder and forced to pass
through a porous inner cylinder before mixing with the jet. The porous boundary created a
uniform radial velocity profile for the entrained gas. A mask of the same diameter as the jet was
placed at the cylinder exit under zero pressure drop conditions. This mask helped to prevent any
inflow or outflow of gases that might affect the entrainment measurement. By supplying
entrained gas such that the pressure differential across the exit mask is zero, the entrained mass
flow rate may be measured. Since there are no pressure gradients in an ambient environment, the
flow conditions of a jet spraying into ambient air are duplicated when the pressure differential
across the mask is zero.
13
-------
The entrainment device discussed above was successfully adapted by Bush (1994) to
measure entrainment rates of effervescent sprays. A schematic of the device is shown in Figure
4, while exact design specifications may be found in Bush (1994).
Atomizing
Air
Atomizing
Liquid
Entrainment
Air
Atomizer
Porous
Cylinder
Spray
Velocity
Profile
Micromanometer
Figure 4. Entrainment device.
A key to adapting the device to sprays was proper sizing of the exit mask. To accomplish
this, 12 square Plexiglas plates were constructed, each with a hole bored in its center. Hole
14
-------
diameters range from 13 to 152 mm, which allowed rapid fitting of the proper mask size at each
spray operating condition. A proper size is achieved when the exit mask is large enough to allow
the spray to pass through (i.e., no droplets striking the mask), but not so large as to leave a gap
between the edge of the spray and the mask. Once the proper mask is installed, data acquisition
consists of adjusting the entrained air mass flow rate such that there is no pressure gradient
across the exit mask and then recording the corresponding rotameter setting. The entrained air
mass flow rate is calculated from this reading.
The pressure differentials across the exit mask were monitored using an MKS model
229HD differential pressure transducer with a Newport 4-20 rnA, 3.5 digit process indicator.
This enabled measurements of ± 1 torr (135 Pa) full scale with 1 mtorr (135 mPa) resolution.
The entrained air mass flow rates were measured using an Omega Engineering FL-1503A
rotameter and controlled by a needle valve. The rotameter was calibrated using an American
Meter DTM-115 volumetric gas flow meter and a MicroMotion Model D25 electronic mass flow
meter. A straight line was fit to the data, yielding a coefficient of determination (r ) of 0.999.
Entrainment by a 1 g/s air jet was measured at several axial positions and compared to
results obtained by Ricou and Spalding (1961) in order to check the performance of the
entrainment device. The data were reduced to the form reported in Ricou and Spalding (1961):
where me is the entrained gas mass flow rate, x is the axial distance from the nozzle, pe is the
entrained gas density, and M0 is the spray exit momentum rate. E is the dimensionless
entrainment number, as defined by Ricou and Spalding (1961), and determined by them to be
0.282 ±0.015 for high Reynolds number air jets (Re(j>25,000).
Data taken using the entrainment device described above are shown in Figure 5. Each +
represents one data point.
The results are presented as normalized entrainment versus normalized axial distance.
Normalized entrainment is defined as the entrained gas mass flow rate, me, divided by the liquid
mass flow rate at the atomizer exit, ih|, while normalized axial distance is defined as the distance
along the spray axis, x, divided by the atomizer exit orifice diameter, d0. The linear fit for the
data shown in this plot has a coefficient of determination (r^) of 0.978 and results in an
entrainment number of 0.232 ± 0.010, a value within 18% of that reported by Ricou and
Spalding (1961).
In order to obtain entrained air flow measurements characteristic of a spray operating in
an ambient environment, it is critical that the entrainment device does not interfere with the
structure of the spray. To confirm this, an Aerometrics Phase/Doppler Particle Analyzer
(P/DPA) was used to collect radial velocity and drop size profiles for the case where the nozzle
was operating inside the entrainment device and for the corresponding case where the nozzle was
operating in the open environment. These two scenarios were compared for all liquids sprayed in
this study at several liquid mass flow rates, ALRs, and axial positions.
15
-------
Figure 6 shows the mean velocity profile for a 0.080 Pa-s viscosity, 0.030 Pa-m surface
tension spray operating at a liquid flow rate of 0.8 g/s and an ALR of 0.10 (each symbol
represents one data point). It is representative of data obtained for all fluids and operating
conditions. Note that the average velocities for the atomizer operating with the entrainment
device are in excellent agreement with the average velocities for the atomizer operating without
the device.
c
O
£
c
c
u:
*o
N
1-2= 0.997
0
10
20
30
40
Normalized Axial Distance
Figure 5. Entrainment data for a 1.0 g/s turbulent air jet, obtained using the entrainment
device of Figure 4.
Figure 7 shows the number averaged drop size (Djq) profiles for the same spray used in
the collection of data for Figure 6. This figure is also representative of data obtained for all
fluids and operating conditions. Note that the average drop sizes for the atomizer operating with
and without the entrainment device are in excellent agreement. This level of agreement in the
drop velocity and size profiles was observed for all cases studied, indicating that the entrainment
device does not significantly alter the structure of the effervescent spray. We therefore conclude
16
-------
that entrainment data obtained using the device are representative of entrainment in the ambient
environment when the differential pressure boundary condition at the device exit is satisfied.
A number of previous spray entrainment studies have demonstrated the importance of jet
momentum rate on entrained gas flow rates. Bush et al. (1996) constructed a device for
determining the axial momentum of two-phase jets based on the design of Deichsel and Winter
(1990). That probe was used in this study. A summary of its principles and operation is
provided below.
- <$> - Without Instrument
—0— With Instrument
E
o
4—>
5 —
-2.0
1.5
1.0
-0.5
0.0
0.5
1.0
1.5
2.0
Radial Position (cm)
Figure 6. Radial droplet velocity profile for a ligament-controlled effervescent atomizer
produced spray formed from a 0.080 Pa-s viscosity , 0.030 Pa-m surface tension fluid.
The momentum rate of a spray is measured by converting the axial flow to a radial flow
by spraying against a deflection cone whose contour is obtained from the equation of streamlines
for an incompressible, axisymmetric, stagnation point flow (White, 1991). It is easily shown that
the amount of reaction force needed to hold the cone in place is equal to the momentum rate of
the spray.
17
-------
For this study, the deflection cone was mounted on a cantilevered beam. When the spray
is directed at the cone, the beam is deflected, resulting in a strain at the base. This strain can be
measured with strain gages and appropriate signal conditioning hardware. Bush et al. (1996)
describe the design details of the deflection cone, the strain gage beam, and the signal
conditioner. A schematic of the momentum rate probe is shown in Figure 8.
20
f
\
- o -
Baseline
_Q_
With Instrument
V
>
15
V
1u
E
nj
Q
c
ca
a
2
o
tt5
E
.c
10
0
-0.5 0.0 0.5
Radial Position (cm)
Figure 7. Number averaged droplet diameter (Djq) versus radial position for a ligament-
controlled effervescent atomizer produced spray formed from a 0.080 Pa-s viscosity, 0.030
Pa-m surface tension fluid.
A liquid jet with a known velocity profile, whose momentum rate is easily calculated,
was used to calibrate the momentum rate probe. By operating the atomizer with only liquid at
several mass flow rates, a calibration curve of momentum rate versus probe voltage output was
produced. The momentum rate was calculated based on either a fully developed or slug flow-
velocity profile, depending on the viscosity of the fluid. A sample calibration curve is shown in
Figure 9, where each + repres^pts one data point. A straight line was fit to the data, yielding a
coefficient of determination (r ) of 0.995.
18
-------
Sprays were formed using a total of six separate fluids, including water. Fluids (other
than water) were formed from either mixtures of water and commercially available glycerine, or
commercially available refined hydrocarbons (Texaco solvent neutral oils, SNO-lOO and SNO-
320, or Bennzoil Universal Calibration Fluid, UCF-1). Their compositions and physical
properties are presented in Table 1. Note that five of the fluids differ in viscosity (0.020. 0.040
or 0.080 Pa-s) and surface tension (either 0.030 Pa-m for 0.020 and 0.040 Pa-s or 0.067 Pa-m for
all three viscosities). The sixth fluid was water. These viscosities and surface tensions were
chosen to span the range of current alcohol-based consumer products and their (projected) water-
based counterparts.
Strain Gages
Canti levered
Beam
Deflection
Cone
Damper
Microcomputer
Atomizer
Signal
Conditioner
Figure 8. Momentum rate probe of Bush et al. (1996).
The uncertainty in determining the entrainment number, E, was calculated using the
expression developed by Bush (1994). Error in entrained mass flow rate measurements arises
from the limited resolution of the micromanometer and is calculated as follows:
d)-(d, -2a)2]j2p(AP) + 2p1g(Az) + j(de-2of
AP | pgAz
V V
(6)
where merr is the error in the measured entrained air mass flow rate, de is the exit mask
diameter, a is the annular gap between the edge of the exit mask and the edge of the spray drop
sheath, p is the entrained air density, AP is the resolution limit of the micromanometer, g is the
acceleration due to gravity, Az is the vertical thickness of the exit mask, and v is the entrained air
velocity.
The mass flow rate uncertainty calculated using this equation is as high as 30%. The only
other significant source of uncertainty is associated with the momentum rate probe. This is
19
-------
approximately 7%, as noted by Bush et al. (1996). Assuming uncorrelated uncertainties, a value
of 31% was calculated as the maximum overall uncertainty for the entrained mass flow rates.
Entrainment control experiments were performed using the same apparatus as was used
for the original (baseline) entrainment measurements described above, with the exception of
modifications to the dispenser itself. Gas flow entrainment results lead us to modify the original
dispenser exit orifice (a flat plate) to asymmetric nozzle exits and nozzles with indeterminate
origins. Figure 10 shows the four geometries considered—stepped, inclined, two-point and four-
point. Each geometry was carefully machined into a separate exit orifice plate with an exit
orifice diameter of 0.38 mm.
0.03
0.02 —
a
OL
E
3
C
V
E
o
2
0.01
0.00
0.995
0
1
2 3 4 5
A V (volts)
Figure 9. Calibration data for momentum rate probe of Figure 8.
Data obtained using the modified geometry' dispensers is compared to each other and to
the base case in Chapter 4. Results are provided in terms of the dimensionless entrainment
number, E, as was done for the baseline study.
20
-------
Figure 10. Nozzle geometries investigated during this study: inclined exit (one-point
crown), stepped exit, four-point crown, and two-point crown.
21
-------
Chapter 3
QA/QC
This Quality Assurance Chapter of this report is organized into four subsections. The
first subsection discusses control systems for the effervescent atomizer itself, thereby indicating
how air/liquid ratio (ALR) will be determined. Each subsequent subsection deals with a separate
measurement technique. Results are summarized in three tables at the end of this Chapter.
Effervescent atomizer and its supply system
The data quality indicator chosen for ALR calculations is the coefficient of variation.
Acceptable ALR data is defined as a sample having a coefficient of variation less than or equal to
12%. This value is based on the criterion that acceptable data are obtained when the liquid and
gas mass flowrates are measured to within their maximum fractional uncertainties. The
coefficient of variation was calculated at discrete measured values that cover the applicable range
of ALRs.
Data bias was determined by calculating values at discrete flow rates encompassing the
applicable ranges of air and liquid mass flow rates. Individual values were determined by
comparing rotameter supplied air and liquid mass flow rates with values obtained using both a
Micromotion mass flowmeter and a dry test meter. Agreement to within 5% was obtained in all
cases.
Drop size measurements
The data quality indicator for this measurement has also been chosen to be the coefficient
of variation. A total of three drop size measurements were made at each value of ALR with
acceptable data defined as sets of three having coefficient of variations less than or equal to 10%.
The coefficient of variation was calculated at discrete measured values that cover the applicable
range of SMDs.
Data bias is addressed by checking instrument calibration using a reticule. Bias was
evaluated at only a single SMD value as only one calibration reticle is available.
Entrainment data
The data quality indicator for these measurements has been chosen to be the coefficient of
variation. A minimum of three trials will be made for each measurement set. They will be
deemed acceptable if their coefficient of variation is less than 15%. The coefficient of variation
was calculated at discrete measured values that cover the applicable range of entrainment mass
flow rates.
Bias was addressed when demonstrating how data obtained with this instrument compare
with similar data reported in the scientific literature. Bias values were calculated at discrete
values of entrained gas mass flow rate that encompass the range of interest.
Momentum rate probe
The data quality indicator for these measurements has been chosen to be the coefficient of
variation. A minimum of three trials will be made for each measurement set. They will be
22
-------
deemed acceptable if their coefficient of variation is less than 15%. The coefficient of variation
was calculated at discrete measured values that cover the applicable range of spray momenta.
Bias has been addressed when discussing how the instrument is calibrated. Bias values
were calculated at discrete values of spray momentum rate that encompass the range of interest.
Tables 2 and 3 summarize Data Quality for this investigation. Table 2 presents Data
Quality Indicators and Bias levels for each of the four key measurements. Table 3 provides
maximum, minimum, and average DQI values for each of the four key measurements, plus the
percentage of time that the DQI goal was met.
Table 2: Data Quality Indicators.
Measurement
Data Quality Indicator
Bias
Air-liquid ratio by mass (ALR)
coefficient of variation < 12% for
each sample point
<5%
Sauter mean diameter (SMD)
coefficient of variation < 10% for
each sample of at least three data
points
<5%
Entrainment
coefficient of variation < 15% for
each sample of at least three data
points
< 18%
Momentum Rate
coefficient of variation < 15% for
each sample of at least three data
points
<31%
Table 3: Data Quality.
Measurement
DQI Values
% of tests DQI acceptable
Air-liquid ratio by mass (ALR)
max = 7.2%
min = 0%
average = 2.3%
162/162 = 100%
Sauter mean diameter (SMD)
max = 11.4%
min = 0.6%
average = 4.3%
32/33 = 97%
Entrainment
max = 14.3%
min = 1.5%
average = 4.4%
97/97 = 100% if only two
trials allowed
87/97 = 90% if three
trials required
Momentum Rate
max = 3.5%
min = 0.2%
average = 1.3%
32/32 = 100 %
23
-------
Tabic 4 presents accuracy and precision values for laboratory instruments used during
this study. Accuracy values were provided by the manufacturers. Precision was obtained by
comparing measured data with calibration standard values. Instrument calibration was
performed on an approximately semi-annual basis.
Table 4: Instrumental Accuracy and Precision.
Measurement/instrument
Accuracy
Precision
Dynamic Viscosity/
Haake Falling Ball Viscometer
manufacturer states < 1%
within 5% of
calibration
standard
Surface Tension/
CSC Model 70535 du-Nuoy Ring
Tensiometer
manufacturer states < 2%
within 3% of
calibration
standard
Fluid Volume/graduated cylinder
manufacturer states <1%
not checked
Fluid Mass/Mettler Model P1200N
centigram balance
manufacturer states 0.01 g
<1% for masses
considered here
24
-------
Chapter 4
Results and Discussion
Results are presented for steady state drop size, followed by development of a model to
describe those results. Transient drop size data are then introduced. Supply pressure data are
included for the sake of completeness.
The drop size results are followed by entrainment data. This chapter closes with
entrainment control data and a discussion of its implications.
Steady-state Drop Size Measurements
Experimental results describing the drop size performance of ligament-controlled
effervescent atomizers operating at low ALRs are presented and discussed first in this chapter.
Drop size performance was determined for a number of operating conditions and spray fluids.
The parameters that were varied, and the ranges over which data were collected, are shown in
Table 5.
Table 5. Parameters for Experimental Investigation.
Parameter
Range Studied
Units
Viscosity (n)
0.001 to 0.080
kg/m-s
Surface Tension (a)
0.030 to 0.067
kg/s2
ALR
0.005 to 0.04
Dimensionless
•
Liquid Mass Flow Rate (m,)
0.5 to 1.0
g/s
The influences of fluid physical properties and operating conditions on SMD are
considered. All drop size measurements were obtained using a Malvern Particle Size Analyzer,
with the probe volume passing through the center of the spray at a position 15 cm downstream of
the exit orifice. High-speed photographs and a summary of 3-dimensional holographic images of
the near-nozzle breakup structure of the spray are also presented. These were instrumental in the
development of a model to predict drop sizes for ligament-controlled effervescent sprays.
The operating conditions expected to affect nozzle performance are ALR and liquid mass
flow rate. The ALR was varied from 0.005 to 0.01. Liquid mass flow rate was varied from 0.5
to 1.0 g/s.
Figure 11 contains drop size data for the three fluids with a common surface tension
(0.030 kg/s2) that is representative of current VOC solvent-based consumer products. This figure
illustrates the influence of ALR on SMD - an increase in ALR above 0.0075 reduces SMD
slightly, while a reduction in ALR below 0.0075 results in an increase in SMD. Error bars
representing 1 standard deviation are included. The data demonstrate that SMDs of less than 70
jam (within experimental uncertainty, defined here as the standard uncertainty or experimental
standard deviation) are obtained for ALRs as low as 0.0075.
Figure 12 contains drop size data for the three fluids with a common surface tension
(0.067 kg/s2) that is representative of water-based consumer products. Again, error bars
25
-------
representing 1 standard deviation are included. The data demonstrate that SMDs are less than 70
jim (within experimental uncertainty, again defined as the standard uncertainty) for ALRs as low
as 0.0075. As with the lower surface tension fluids, a marked increase in SMD is observed at
ALRs below 0.0075. This behavior will be discussed when the high-speed photographs of the
near-nozzle breakup structure are presented.
Figure 13 illustrates the influence of liquid mass flow rate on SMD. Data were obtained
by spraying two fluids with a common viscosity (0.020 kg/m-s), but different surface tensions
(0.030 and 0.067 kg/s2) at two different mass flow rates (0.6 and 0.8 g/s). Figure 13 also
demonstrates that no clear conclusions can be drawn concerning the effect of liquid mass flow
rate on SMD. For the fluid having a surface tension of 0.067 kg/s2, an increase in liquid mass
flow rate resulted in a slight (<10%) decrease in SMD. However, an increase in liquid mass flow
rate for a fluid having a surface tension of 0.020 kg/s2 resulted in an increase in SMD of
approximately 20%.
250
200 -- - -
150
E
a.
t/1
100 -4-
f
~N
A
n = 0.020 kg/m-s
O
(i = 0.040 kg/m-s
~
n ¦= 0.080 kg/m-s
V
[]
£
50 -- - -
0.000
0.005
0.010
0.015
ALR
Figure 11. SMD versus ALR for three fluids having viscosities of 0.020,0.040, and 0.080
kg/m-s and a common surface tension of 0.030 kg/s . Error bars represent 1 standard
deviation.
The most important conclusion to be drawn from Figure 13 is that sub-70 jim SMD
sprays were achieved at ALRs less than 0.01 for all fluids tested. This achieves the stated goals
of this study. In addition, the flat slope between ALRs of 0.0075 and 0.01 indicates that
increasing ALR above 0.01 has little benefit toward reducing SMD.
26
-------
Figures 11 and 12 may also be used to demonstrate the influence of fluid physical
properties on SMD. Both figures indicate that the performance of ligament-controlled
effervescent atomizers is relatively insensitive to the viscosity of the liquid being sprayed: any
pair of SMDs lie within the sum of their standard deviations. This is an improvement over the
small, although still noticeable, increase in SMD with an increase in viscosity reported by Lund
et al. (1993).
Comparison of the data presented in Figure 11 with those of Figure 12 reveals little
scaling of SMD with surface tension; a slight increase (<10%) in SMD was observed upon
changing the surface tension from 0.030 to 0.067 kg/s2. This is opposite to the trend observed by
Lund et al. (1993), who noted a decrease in SMD when surface tension increased by the same
amount.
250
200
ISO
f
\
A
= 0.020 kg/m-s
<0
» 0.040 kg/m-s
~
H ¦= 0.080 kg/m-s
J
£
a
a
S
100 -4- - -
50
0.000
0.005
T~
0.010
0.015
ALR
Figure 12. SMD versus ALR for three fluids having viscosities of 0.020,0.040, and 0.080
kg/m-s and a common surface tension of 0.067 kg/s . Error bars represent 1 standard
deviation.
Since U.S. Department of Transportation container charging restrictions limit consumer
product package pressures to less than 1 MPa, it is important to determine the supply pressures
required to achieve the SMDs presented in Figures 11 through 13. This information is provided
in Figures 14 through 16, where the influence of ALR, fluid physical properties, and liquid mass
flow rate are considered.
27
-------
Supply pressures required when spraying fluids having a surface tension of 0.030 kg/s"
are presented in Figure 14. For the 0.6 g/s liquid flow rate data presented here, 440 kPa is
required for the 0.020 kg/m-s liquid, 610 kPa for the 0.040 kg/m-s liquid, and 650 kPa for the
0.080 kg/m-s liquid. Supply pressures required when spraying fluids having a surface tension of
0.067 kg/s2 are presented in Figure 15. For the 0.6 g/s liquid flow rate data presented here, only
290 kPa is required for the 0.020 kg/m-s liquid, while the 0.040 and 0.080 kg/m-s liquids require
630 and 780 kPa, respectively. [These values compare favorably with the 240 to 515 kPa supply
pressures used by Lund et al. (1993), in their study.] Note that in both cases, the supply
pressures are well below the 1 MPa limit. Furthermore, data in both figures demonstrate that
ALR has little effect on supply pressure over the range of fluid properties, mass flow rates, and
ALR's considered during this study.
250
200
150 --
E
C
s
100 --
50
A
~
O
~
a = 0.030 kg/s2, m = 0.6 g/s
a = 0.030 kg/s2, ra = 0.8 g/s
c = 0.067 kg/s2, ra = 0.6 g/s
a = 0.067 kg/s2, m = 0.8 g/s
1 1
1 1
1 1
I 1
1 1
1 1
1 b*H '
i i
t i
- - i -
I
X
2
1
i
0.000
0.005
0.010
0.015
ALR
Figure 13. SMD versus ALR for two fluids having surface tensions of 0.030 and 0.067
kg/s2, a common viscosity of 0.020 kg/m-s, and operating at two liquid mass flow rates.
Error bars represent 1 standard deviation.
28
-------
800
700
600
«T
a.
f 500
t-
s
*/>
an
% 400
*S.
Q.
3
X.
j 300
>
<
200
100
0
0.000 0.005 0.010 0.015
ALR
Figure 14. Atomizer supply pressure versus ALR for three liquids having viscosities of
0.020,0.040, and 0.080 kg/m-s, a common surface tension of 0.030 kg/s2, and a common
liquid mass flow rate of 0.6 g/s.
Figures 14 and 15 also show that a change in liquid surface tension has a mixed effect on
supply pressure. Increasing the surface tension in the low viscosity case (0.020 kg/m-s) reduces
the supply pressure from about 450 to less than 300 kPa. Increasing the surface tension in the
high viscosity case (0.080 kg/m-s) increases the supply pressure from about 650 to
approximately 780 kPa. Increasing the surface tension in the intermediate viscosity case (0.040
kg/m-s) has little effect on supply pressure. There is no known explanation for this behavior.
Finally, the data of Figures 14 and 15 illustrate the expected increase in supply pressure
with an increase in liquid viscosity. This relationship is characteristic of flow through porous
media.
The data in Figures 14 and 15 were obtained at a common liquid mass flow rate of 0.6
g/s. Figure 16 illustrates the influence of liquid mass flow rate on supply pressure. The data
exhibit the expected increase in supply pressure with mass flow rate, a phenomenon common to
flow through porous media.
A physical explanation for the observed SMD behavior is provided in Figures 17 through
21. These figures were obtained using the high-speed photographic apparatus discussed earlier.
~
o
~
o
A (i ¦= 0.020 kg/m-s
^ H = 0.040 kg/m-s
I I m = 0.080 kg/m-s
29
-------
800
700
600
ct
Cl
X 500
u.
3
M
Vt
«»
% 400
a.
o.
B
I 300
*>
>
<
200
100
0
0.000 0.005 0.010 0.015
ALR
Figure 15. Atomizer supply pressure versus ALR for three liquids having viscosities of
0.020,0.040, and 0.080 kg/m-s, a common surface tension of 0.067 kg/s2, and a common
liquid mass flow rate of 0.6 g/s.
Figures 17 and 18 compare sprays produced using the Lund et al. (1993) atomizer to
those produced using the ligament-controlled effervescent atomizer. The spray shown in Figure
17 was produced using the Lund et al. (1993) atomizer spraying water at a rate of 1.0 g/s with an
ALR of 0.015. Figure 18 shows a spray produced using a ligament-controlled effervescent
atomizer operating under the same conditions. It is obvious from the photographs that the
inclusion of a porous disc results in better spray quality at low ALRs; when compared to the
Lund et al. (1993) nozzle, the ligament-controlled effervescent atomizer produces a larger
number of smaller diameter ligaments. It is this decrease in ligament diameter that leads directly
to the decreased drop size obtained when using the ligament-controlled effervescent atomizer.
Figures 19 through 21 illustrate how the ligament formation and breakup processes vary
as ALR is reduced from 0.01 to 0.005. It is clear that the number of ligaments is reduced as ALR
goes down, and that their diameters increase. Furthermore, spray quality is observed to
deteriorate markedly as ALR drops from 0.0075 to 0.005. The presence of a pronounced central
liquid jet is the cause of this deterioration, since only a few large diameter ligaments are present
at this low ALR operating condition, resulting in a large value of SMD. This observation is
consistent with the drop size results discussed earlier.
-
~ n
~
-
o o o o
o
-
A A
A
/ \
A n" 0.020 kg/m-s
ji = 0.040 kg/m-s
~ H = 0.080 kg/m-s
- - -
-
V.
30
-------
&
£
a.
"5.
o.
9
C/3
Ci
bfi
C5
V.
CI
>
<
800
700
600
500 --
400 --
300 —
200
100
A a = 0.030 kg/s2, m = 0.6 g/s
^ o = 0.030 kg/s2, m = 0.8 g/s
^ a = 0.067 kg/s5, m = 0.6 g/s
^ cr = 0.067 kg/s2, m = 0.8 g/s
0.000
0.005
—1—
0.010
0.015
ALR
Figure 16. Atomizer supply pressure versus ALR for two liquids having surface tensions of
0.030 and 0.067 kg/s2, a common viscosity of 0.020 kg/m-s, and liquid mass flow rates of 0.6
and 0.8 g/s.
High speed photography was very useful in obtaining qualitative information about how
spray quality is affected by decreasing ALR. However, one limitation of the photographs
presented in Figures 17 through 21 is their inability to accurately portray the 3-dimensional
processes occurring at the nozzle exit plane. For that reason, focused-image holograms were
obtained using the system of Santangelo and Sojka (1994).
Figure 22 is an artist's rendition of the near-nozzle breakup regime as seen in various
holograms. As Figure 22 shows, the presence of the porous disc does not completely modify the
two-phase flow structure, as ligaments are still preferentially formed in an annular band that
surrounds a gas core. However, the porous medium does limit the diameter of ligaments formed
at the nozzle exit plane for ALRs of 0.0075 and above, leading to smaller droplets.
Consequently, the single bubble expansion regime that leads to the sharp rise in SMD as ALR
falls below about 0.03 in the Lund et al. (1993) design is delayed until ALR falls below 0.0075
when using the design introduced here. The holographic images also support the conclusion that
only a limited number of large diameter ligaments are formed as ALR approaches 0.005.
31
-------
Figure 17. Near-nozzle structure for a 1.0 g/s water spray at an ALR of 0.015 produced by
a conventional effervescent atomizer (i.e., without a porous insert).
32
-------
Figure 18. Near-nozzle structure for a 1.0 g/s water spray at an ALR of 0.015 produced by
a iigamcnt-controlled effervescent atomizer (i.e., with a porous insert).
33
-------
Figure 19. Near-nozzle structure at an ALR of 0.01, mass flow rate of 0.6 g/s, viscosity of
0.020 kg/m-s, and surface tension of 0.067 kg/s2.
34
-------
Figure 20. Near-nozzle structure at an ALR of 0.0075, mass flow rate of 0.6 g/s, viscosity of
0.020 kg/m-s, and surface tension of 0.067 kg/s*.
35
-------
Figure 21. Near-nozzle structure at an ALR of 0.005, mass flow rate of 0.6 g/s, viscosity of
0.020 kg/m-s, and surface tension of 0.067 kg/s .
36
-------
Figure 22. Artist's rendition of near-nozzle hologram.
Steady-state Drop Size Modeling
This section describes a model that has been developed to understand the process, and to
predict which variables influence the performance, of ligament-controlled effervescent atomizers.
It was shown previously that the Lund et al. (1993) model is successful in predicting
SMD for low mass flow rate effervescent atomizer-produced sprays. However, their model does
have limitations. Most notably, it does not incorporate the effects of the relative velocity that
exists between the two phases. The model developed during this investigation addresses this
shortcoming. The geometric portion of this model is based on 3-dimensional holographic
images, which clearly indicate that liquid breakup proceeds through the formation of an annular
band of ligaments whose individual diameters are on the order of the size of the pores in the
porous medium. The analytical portion of this model consists of determining the breakup
wavelength of these ligaments. If it is further assumed that no secondary atomization takes place
and that each ligament collapses into a sphere whose diameter is equal to the SMD, an
expression for spray mean drop size is obtained.
37
-------
The length of the ligaments was determined using the relationship developed by Sterling
and Sleicher (1975). Their expression predicts the wavelength of the fastest growing disturbance
in a capillary jet and accounts for the aerodynamic interaction between the jet and the
surrounding medium:
pl+l4P -
pa 2pa x ' 2a p£j(4)
Here p is the dimensionless growth rate of the disturbance, % is the dimensionless wavenumber,
\i, p and a are the liquid viscosity, density, and surface tension, p is the gas density, a is the
radius of the jet, U is the relative velocity between the liquid and gas phases, and Kg and K| are
modified Bessel functions of the second kind.
If the jet radius and relative velocity are known, Equation (7) can be solved numerically
for the dimensionless wavenumber, 4. that results in the largest dimensionless disturbance
growth rate, p. Note that when the relative velocity between the two phases is assumed to be
zero, this expression can be manipulated to obtain the Weber (1931) expression for the critical
wavelength.
We assume in this investigation that the ligaments can be modeled as cylindrical jets and
that their diameter, dL, is controlled by the pore size of the porous medium. Therefore, only the
relative velocity between the two phases is needed to solve Equation (7). Determining the
relative velocity was the biggest challenge in modeling ligament-controlled effervescent
atomizers.
Due to the difficulty in analytically solving for the velocities of the two phases as they
pass through the porous medium, measured momentum rates were used to experimentally
determine the liquid and gas velocities. This was accomplished using the equation derived by
Deichsel and Winter (1990) for determining the velocity slip ratio between the liquid and gas
phases:
2
sr +sr
mgPl ^ m; AgMppi
• • I •
m
iPg
mg mgmi
+ Pl_ = o
pg
(8)
Here sr is the velocity slip ratio, mg is the gas mass flow rate, p/ is the liquid density, m, is the
liquid mass flow rate, p^ is the gas density, AE is the area of the atomizer exit orifice, and Ma
is the momentum rate at the nozzle exit. Relative velocities between the atomizing gas and
liquid, as determined using Equation (8), are presented in Table 6.
The momentum rate of the spray exiting the atomizer was measured using the apparatus
of Bush et al. (1996). Their device transforms the axially flowing spray into a radial flow
through use of a deflection cone that is suspended from a cantilevered beam. When the spray
impacts the cone, the beam is deflected and a strain is imposed on the base of the beam. This
38
-------
strain is measured using precision strain gages and appropriate signal conditioning electronics.
Bush et al. (1996) describe the design details of the deflection cone, strain gage beam, and signal
conditioner.
Table 6. Relative Velocities (in m/s) Between Atomizing Gas and Liquid
for Five Fluids at Four ALRs.
Fluid
ALR=0.0075
ALR=0.010
ALR=0.015
ALR=0.020
0.020 kg/m-s, 0.030 kg/s1
32
43
66
88
0.040 kg/m-s, 0.030 kg/s'
36
47
72
96
0.080 kg/m-s, 0.030 kg/s2
15
23
42
56
0.020 kg/m-s, 0.067 kg/s2
30
42
66
90
0.040 kg/m-s, 0.0670 kg/s'
38
49
70
93
With the velocities of both the liquid and gas phases known, Equation (7) is solved
numerically for the dimensionless critical wavenumber, lop,, that results in the fastest growing
disturbance. The critical wavenumber can then be used in a predictive equation for the mean
diameter of the spray:
hndf
SMD = 3 1 (9)
v w
Predicted SMD for sprays of varying viscosities, and surface tensions of 0.030 and 0.067
kg/s2, are plotted in Figures 23 and 24, respectively. Experimental results are also plotted for
comparison. A pore size of 37 p.m was used in predicting the SMDs shown in these two figures.
Spray SMDs are slightly over predicted for all cases investigated using this average pore size.
Figures 25 and 26 illustrate the influence of porous medium pore size on SMD by
presenting predicted drop sizes assuming an average pore size of 25 ^m for sprays of varying
viscosities and surface tensions of 0.030 and 0.067 kg/s2, respectively. Note that the model
predictions exhibit the expected decrease in SMD with a decrease in porous medium pore size
(or, equivalently, ligament diameter). Also note that the SMD does not decrease linearly with a
reduction in porous medium pore size.
Figures 23 and 24, as well as Figures 25 and 26, may be used to demonstrate the
influence of fluid physical properties on the predicted values of the spray mean diameter. The
experimental results presented in both figures indicate that the performance of ligament-
controlled effervescent atomizers is relatively insensitive to the viscosity of the liquid being
sprayed: any pair of SMDs lie within the sum of their standard deviations. The predicted values
of SMD obtained from the model, however, show a slight increase in SMD with an increase in
viscosity. From a physical standpoint, viscosity dampens instabilities, leading to longer breakup
lengths at higher viscosity. This leads to a larger spray SMD, as predicted by the model.
However, the increase predicted by the model is slight and it can be concluded that the model
39
-------
correctly predicts the performance of ligament-controlled effervescent atomizers to be relatively
insensitive to the viscosity of the fluid being sprayed.
The effects of surface tension on the predicted SMD of the spray can be determined by
comparison of Figure 23 with Figure 24 (or Figure 25 with Figure 26). Experimental results
show a slight increase in SMD upon changing from a surface tension of 0.030 to 0.067 kg/s2.
This is opposite to the scaling noted by Lund et al. (1993). However, predictions obtained from
the model do demonstrate a slight increase in spray mean diameter with an increase in surface
tension, which agrees with the experimental results.
250
200
150 -----
E
a
2
en
100
50 --
f
N
A
= 0.020 kg/m-s
O
(i = 0.040 kg/m-s
~
H - 0.080 kg/m-s
~
Model, n = 0.020 kg/m-s
~
Model, n = 0.040 kg/m-s
V
0.000
0.005
0.010
0.015
0.020
0.025
ALR
Figure 23. Experimental data and predicted SMDs, based on an average pore diameter of
37 fxm, for three fluids having viscosities of 0.020,0.040, and 0.080 kg/m-s, and a common
surface tension of 0.030 kg/s2. Error bars represent 1 standard deviation.
One of the major benefits of the Lund et al. (1993) style effervescent atomizer was that
switching from an alcohol-based to a water-based carrier compound resulted in a decrease in
spray SMD. Although the ligament-controlled effervescent atomizer does not behave in the
same manner, the increase in drop size due to an increase in the surface tension is less than 10%.
Furthermore, acceptable SMDs were obtained using the ligament-controlled effervescent
atomizer notwithstanding the increase of 10%.
The operating conditions expected to affect the predicted spray SMD are air/liquid ratio
(ALR) and liquid mass flow rate. Figures 23 through 26 show the influence of ALR on the
40
-------
predicted SMD. For the two pore sizes considered, an SMD of approximately 70 |im is obtained
for ALRs less than 0.01, but greater than 0.0075. The model accurately predicts the scaling for
ALRs above 0.075, but not for those below. This is due to a shift in breakup structure of the
spray that is not accounted for in the model. At very low ALRs (ALR < 0.0075), a large increase
in the measured SMD of the spray is noted. This is due to coalescence of the ligaments. The
model developed in this study does not account for ligament coalescence and thus is unable to
predict the large jump in spray mean diameter that was observed at very low ALRs.
250
-
c
\
"
i
A
(i = 0.020 kg/m-s
O
H = 0.040 kg/m-s
200 -
~
M = 0.080 kg/m-s
-
I
]
~
Model, (i = 0.020 kg/m-s
-
<
>
~
Model, n = 0.040 kg/m-s
150 -
_ _ _
- -
I
¦
Model, n = 0.080 kg/m-s
-
V
y
c
5
(ft
100 --
¦
T
50 -- - - -
0.000
0.005
0.010
0.015
0.020
0.025
ALR
Figure 24. Experimental data and predicted SMDs, based on an average pore diameter of
37 jam, for three fluids having viscosities of 0.020,0.040, and 0.080 kg/m-s, and a common
surface tension of 0.067 kg/s2. Error bars represent 1 standard deviation.
In summary, a model to predict the SMD of sprays considered in this investigation was
developed, based on the work of Sterling and Sleicher (1975), images of the near-nozzle
structure, and momentum rate measurements. The model correctly predicts SMD. In addition,
the scaling due to surface tension, viscosity, and ALR was accurately predicted.
Transient Drop Size Measurements
Transient drop size data was obtained using the system described in Chapter 2. The goal
of this portion of the study was to demonstrate that ligament-controlled effervescent atomizers
41
-------
can produce acceptable sprays when operating under conditions typical of a consumer product
application. To that end, drop size was measured as a function of time for a representative duty
cycle. Conventional video movies of the spray were also obtained to assist in interpretation of
the drop size data.
250
200
150
\
A
(i = 0.020 kg/m-s
0
jx ~ 0.040 kg/m-s
~
H = 0.080 kg/m-s
~
Model, n - 0.020 kg/m-s
~
Model, |i = 0.040 kg/m-s
V.
>
£
£3
100
50
0.000
0.00S
0.010 0.015
ALR
0.020
0.025
Figure 25. Experimental data and predicted SMDs, based on an average pore diameter of
25 nm, for three fluids having viscosities of 0.020,0.040, and 0.080 kg/m-s, and a common
surface tension of 0.030 kg/s2. Error bars represent 1 standard deviation.
Figure 27 shows measured SMD as a function of time. Data taking was initiated when
the first liquid exited the dispenser, and continued until reasonable steady behavior was
demonstrated (usually after about 100 msec). As can be seen, there is an initial (transient)
increase in SMD associated with the startup of liquid and atomizing air flows, followed by a
second peak. However, these transients die out quickly and drop size does not fluctuate
significantly from that point onward.
Companion video movies taken of the spray indicate the presence of a "burst" of liquid at
the beginning of the spray process, followed by a period of steady operation. The burst is
associated with the increase in SMD seen early in the spray process. It is likely to result from a
temporary decrease in ALR due to the liquid and gas flow rates ramping up at different rates.
42
-------
The transient measurements do indicate that acceptable spray performance can be
achieved when operating under transient conditions. The time-averaged SMD is below the 70
|im target and, as Figure 27 shows, the portion of time for which the spray SMD is above 70 (im
is small. Finally, the peak SMD is not excessive, being only about 130 jam.
250
200
150
..
\
A
H = 0.020 kg/m-s
o
H = 0.040 kg/m-s
~
H = 0.080 kg/m-s
A
Model, n = 0.020 kg/m-s
~
Model, fi = 0.040 kg/m-s
¦
Model, n = 0.080 kg/m-s
\
y
Q
5
V}
50 ----- -
i i 1 r-
0.000
0.005
0.020
0.025
0.010 0.015
ALR
Figure 26. Experimental data and predicted SMDs, based on an average pore diameter of
25 |im, for three fluids having viscosities of 0.020,0.040, and 0.080 kg/m-s, and a common
surface tension of 0.067 kg/s2. Error bars represent 1 standard deviation.
Entrainment Measurements
Normalized entrainment and momentum rate measurements were obtained using the
entrainment device and momentum rate probe described in Chapter 2 (see Figures 4 and 8).
Entrainment rate data are presented as normalized entrainment rate versus dimensionless axial
• « •
distance. Normalized entrainment rate is defined as metmt, where me is the entrained air mass
flow rate and mi is the liquid mass flow rate. Dimensionless axial distance is defined as x/d0.
where x is the axial distance and d0 is the atomizer exit orifice diameter. Momentum rate results
are plotted as momentum rate versus ALR. Six fluids, including water, were sprayed in order to
determine the influence of operating conditions and fluid physical properties on the entrainment
43
-------
behavior and momentum rates of ligament-controlled effervescent atomizer produced sprays.
See Table 1 for a list of liquid physical properties.
150 —1
100 —
E
a.
O
S
Xfl
50 —
I
0
0.00 0.05 0.10 0.15 0.2
Time, sec
Figure 27. Transient performance of a ligament-controlled effervescent atomizer.
Bush (1994) reported normalized entrainment results for the Lund et al (1993)
"conventional" effervescent atomizer. His data showed that normalized entrainment
measurements scaled linearly with dimensionless axial position, as was predicted by the
dimensional analysis of Ricou and Spalding (1961). Normalized entrainment rates were also
reported to increase with increasing ALR. Normalized entrainment rates, plotted against
dimensionless axial distance for a 1.0 g/s water spray produced using a Lund et al. (1993) style
atomizer, are shown in Figure 28.
Figure 29 shows normalized entrainment rates versus dimensionless axial distance for a
ligament-controlled effervescent atomizer spraying water under the same conditions as for Figure
28. Again, normalized entrainment rates are fotpd to scale linearly with dimensionless axial
distance and to increase with increasing ALR (r >0.984 for all cases). These features are
common to all normalized entrainment measurements for all liquids sprayed in this investigation.
The ALR scaling, in particular, is not surprising since increasing ALR increases the exit
momentum rate of the spray. In addition, increasing ALR results in a small, but noticeable
(Sutherland, 1996). decrease in SMD which should result in more effective momentum transfer
44
-------
from the spray to the surrounding air. The more effective momentum transfer is expected to
increase entrainment.
10
8 --
+
~
ALR = 0.01 0^= 0.999)
ALR = 0.02 (r^= 0.997)
ALR = 0.03 (r2^ 0.997)
ALR = 0.0375 (r= 0.993)
c
O)
E
c
e
U)
"O
41
.N
o
2
6 --
0
0
100
400
500
200 300
Normalized Axial Distance
Figure 28. Normalized entrainment versus normalized axial distance for water being
sprayed at a mass flow rate of 1.0 g/s using a conventional effervescent atomizer.
When comparing the magnitudes of the normalized entrainment rates, values decreased
upon changing from the Lund et al. (1993) atomizer to the ligament-controlled effervescent
atomizer. This is due to the dependence of entrainment behavior on the exit momentum rate of
the spray. Unlike the Lund et al. (1993) atomizer, the ligament-controlled effervescent atomizer
requires the two-phase flow to pass through a porous insert before exiting. This alters the near
nozzle breakup structure and thus the exit momentum rate of the spray, thereby causing a change
in the entrainment behavior. As will be demonstrated, the end result is a decrease in the
dimensionless entrainment number, E.
Entrainment data for ligament-controlled effervescent atomizer produced sprays using
working fluids other than water are presented in Table 7. A representative plot is included as
45
-------
Figure 30. Plots for the remaining fluids are reported by Sutherland (1996). The five fluids
sprayed have surface tensions of 0.030 or 0.067 Pa-m and viscosities varying from 0.020 to
0.080 Pa-s. In each case, the data exhibit a linear relationship between normalized entrainment
and normalized axial distance; the correlation coefficients range from 0.989 to 0.998. In
addition, the slope of the normalized entrainment versus normalized axial distance lines increases
with ALR in all cases. This behavior is anticipated since an increase in ALR leads directly to an
increase in exit orifice momentum rate.
10 -i
ALR = 0.01 (i*= 0.984)
ALR = 0.02 (i3® 0.998)
ALR = 0.03 (r2= 0.997)
ALR = 0.0375 (r^ 0.997)
8
6
4
2
0
0 100 200 300 400 500
Normalized Axial Distance
Figure 29. Normalized entrainment versus normalized axial distance for water being
sprayed at a mass flow rate of 1.0 g/s using a ligament-controlled effervescent atomizer.
The effect of viscosity on normalized entrainment rate is observed through comparison of
data for fluids having similar surface tensions. Comparison of data from Figure 29 with that of
Fluids 1 through 3 (at equal ALRs) indicates an increase in normalized entrainment of 25 to 35%
as viscosity is increased from 0.001 to 0.020 Pa-s, but a smaller effect on the normalized
entrainment rates when the viscosity is 0.020 Pa-s or higher (normalized entrainment rates were
within 16% of the mean for each ALR for viscosities between 0.020 and 0.080 Pa-s). Similar
46
-------
viscosity scaling is obtained from sprays having a common surface tension of 0.030 Pa-m; i.e.
Fluids 4 and 5 — the data show that viscosity has little effect on the normalized entrainment rates
for viscosities of 0.020 and 0.040 Pa-s, with variations being less 7% for each ALR.
Table 7. Slopes and Coefficients of Determination for Least Squares Linear Fits
to Normalized Entrainment Versus Normalized Axial Distance Data
for Fluids Sprayed at 0.6 g/s.
iFluid / ALR->
0.075
0.01
0.015
0.02
1
p.0082,*
r =0.998**
(j.0097,
r =0.994
|).012,
r =0.998
J).015,
r =0.998
2
(J.0083,
r =0.998
0.990 for all cases), as well as an increase in the slope of these lines with an
increase in ALR. Most importantly, the data for both mass flow rates collapse to a single line for
any particular ALR value.
Momentum rate data corresponding to the entrainment rate data of Figures 29 and 30 and
Table 7 are presented in Figures 31 and 32 and Table 8, respectively. Figures 31 and 32 are
representative of all momentum rate data obtained during this study and show that momentum
rate is linearly proportional to ALR. Coefficients of determination (r2) are above 0.962 in all
cases.
47
-------
10
ALR = 0.0075 (r2= 0.997)
ALR = 0.01 (r= 0.997)
ALR = 0.015 (1^= 0.996)
ALR = 0.02 (1^= 0.991)
4J
E
c
c
UJ
-o
u
N
c
6 --
0
0
100
400
500
200 300
Normalized Axial Distance
Figure 30. Normalized entrainment versus normalized axial distance for a fluid (Fluid 5)
having a viscosity of 0.040 Pa-s, a surface tension of 0.030 Pa-m, and operating at liquid
mass flow rates of 0.5 and 0.6 g/s.
Momentum rate for a water spray is plotted versus ALR and shown in Figure 31. When
compared to data obtained by Bush (1994) at the same liquid mass flow rate (i.e., 1.0 g/s), the
momentum rates for ligament-controlled effervescent atomizer produced sprays are lower over
the entire range of air-to-liquid mass flow rate ratios. This supports the hypothesis that the
addition of a porous insert does indeed alter the exit momentum of the spray and, ultimately, the
entrainment behavior.
The influence of surface tension on momentum rate can be determined by comparing the
Fluid 1 and 4 or 2 and 5 entries in Table 8. In both cases, momentum rates decrease when the
surface tension is increased. This same trend was reported by Bush (1994) and was attributed to
differences in liquid density, not surface tension. Lower liquid density results in a lower void
fraction at the atomizer exit and, thus, higher interphase velocity slip. Higher slip results in a
higher momentum rate.
48
-------
0.050
0.040
5, 0.030
u
i2- 0.992
E
3
C
0>
E
| 0.020
0.010
0.000
0.00 0.01 0.02 0.03 0.04
ALR
Figure 31. Momentum rate versus ALR for water.
The influence of viscosity on momentum rate is unclear. The Fluid 1 through 3 entries in
Table 8 show momentum rates for liquids having viscosities of 0.020,0.040 and 0.080 Pa-s,
respectively, and a common surface tension of 0.067 Pa-m: An increase in viscosity from 0.020
to 0.040 Pa-s results in a decrease in momentum rate (slope), while a further increase in viscosity
from 0.040 to 0.080 Pa-s results in an increase in momentum rate (slope). The relative changes
in momentum rate are small (-20%) and could be due to experimental uncertainty.
When examining the influence of viscosity on momentum rates for fluids with a common
surface tension of 0.030 Pa-m, the effects are less obvious. The Fluid 4 and 5 entries in Table 8
contain data for mixtures having viscosities of 0.020 and 0.040 Pa-s, respectively, and operating
at two liquid mass flow rates. These data show that momentum rate (slope) increases with
viscosity at the low mass flow rate, yet exhibits only a minor variation with viscosity for the
higher mass flow rate.
49
-------
0.020
0.015
0.010
c
u
E
o
2
Liquid mass flow rate
+ 0.6 g/s (1^=0.992)
o 0.5 g/s (^=0.979)
0.005
0.000
0.00 0.01 0.02 0.03
ALR
Figure 32. Momentum rate versus ALR, at two mass flow rates, for a fluid (Fluid 5) having
a viscosity of 0.040 Pa-s and a surface tension of 0.030 Pa-m.
Table 8. Slopes, y-Intercepts and Coefficients of Determination
for Least Squares Linear Fits
to Momentum Rate Versus ALR Data.
Fluid @ liquid mass flow rate
Slope, y-Intercept; r^
1 @ 0.6 g/s
0.39,0.0060; r^=0.988
2 @ 0.6 g/s
0.32,0.0057; rM>.993
3 @ 0.6 g/s
0.37,0.0065; r*=0.962
4 @ 0.5 g/s
0.17,0.0057; rM.992
4 @ 0.6 g/s
0.37,0.0086; r*=0.989
5 @ 0.5 g/s
0.28,0.0053; r*=0.992
5 @ 0.6 g/s
0.38,0.0054; r^=0.979
50
-------
The effects of liquid mass flow rate on momentum rate may also be inferred from Table
8. In each case, the momentum rate (slope) increases with an increase in liquid mass flow rate,
from 0.17 to 0.37 for Fluid 4 and from 0.28 to 0.38 for Fluid 5. This is expected, since the
momentum rate is expected to increase proportionally to the liquid mass flow rate.
The entrainment rate data were combined with the momentum rate to calculate
entrainment numbers using the model developed by Bush (1994). Bush's (1994) model is based
on the work of Ricou and Spalding (1961), who employed a Buckingham Pi analysis using the
exit momentum rate, entrained gas density, and axial distance as the normalizing parameters to
obtain Equation (5).
Entrainment numbers were calculated for five different fluids with varying viscosities and
surface tensions by correlating experimental data via Equation (5). They are listed in Table 9. A
representative sample of the data is included as Figure 33, while Figure 34 is a composite of all
the data acquired in this study (Fluids 1 through 5, plus water).
Table 9. Average Entrainment Numbers.
Fluid @ mass flow rate
E±2a
1 @ 0.6 g/s
0.15810.058
2 @ 0.6 g/s
0.153±0.053
3 @ 0.6 g/s
0.175±0.065
4 @ 0.5 g/s
0.124+0.040
4 @ 0.6 g/s
0.153±0.056
5 @ 0.5 g/s
0.150±0.046
5 @ 0.6 g/s
0.148±0.063
Figure 33 shows entrainment number results for a water spray. A noticeable increase in
entrainment number with air-to-liquid mass flow rate ratio is observed. This behavior occurred
for all fluids considered during this investigation.
Inspection of Figure 34 leads to the conclusion that air-to-liquid mass flow rate ratio has
some influence on the value of the entrainment number. In all cases, the value of the entrainment
number increases with increasing ALR. It might be argued that this variation is due to
experimental uncertainty, since the entrainment number values differ from the mean by less than
40% in all cases, which is approximately the range of uncertainty involved in the calculation.
However, consideration of the limiting cases of pure gas and pure liquid jets demonstrates some
ALR scaling should be expected. For the pure gas jet, the entrainment number has been shown
by Ricou and Spalding (1961) to be approximately 0.28. This is nearly twice the value reported
here for ligament-controlled effervescent atomizer-produced sprays. For the pure liquid jet, the
entrainment number would be zero, if defined to be equal to the amount of surrounding gas
crossing the spray boundary, as it is in this study. If the entrainment number is instead defined to
be proportional to the amount of surrounding gas set in motion by the liquid jet, a simple
boundary layer analysis indicates that E would be approximately 0.02. This is less than 8% of
51
-------
the Ricou and Spalding (1961) value and only 14% of the value reported here. Regardless of the
approach taken to estimate the zero ALR limit for E, the conclusion is that E should increase
with ALR.
0.25
0.20
E
s
Z
c
V
e
c
c
W
0.15
0.10 —
0.05 —
0.00
Liquid mass flow rate
0.00
0.01
0.02
0.03
ALR
Figure 33. Entrainment number versus ALR for a fluid (Fluid 5) having a viscosity of
0.040 Pa-s and a surface tension of 0.030 Pa-m, operating at two liquid mass flow rates.
Table 9 may be used to demonstrate the influence of the fluid physical properties on the
entrainment number. A comparison of data from the Fluid 1 through 3 entries, characteristic of
water-based mixtures, indicates that entrainment number is relatively insensitive to fluid
viscosity — average entrainment numbers differ by less than 15%. The lack of an influence of
viscosity is also observed when comparing data from equal mass flow rate Fluid 4 and 5 entries,
which contain results characteristic of hydrocarbon-based mixtures - average entrainment
numbers differ by less than 21%. This leads to the conclusion that fluid viscosity has a
negligible impact on entrainment number throughout the range of conditions considered in this
study.
52
-------
The effects of surface tension on the entrainment number are demonstrated when
comparing data from Table 9 entries for Fluid 1 with Fluid 4 and for Fluid 2 with Fluid 5, all at
equal mass flow rates. As can be observed, variations in mean entrainment number due to
changes in surface tension are less than 4% and therefore well within experimental uncertainty
(i.e., standard uncertainty) of the mean value. This indicates that surface tension has little effect
on entrainment number.
Figure 34 reinforces the conclusion that the entrainment number is insensitive to fluid
physical properties, and also demonstrates that ALR does have an effect. Considering the data as
a whole, the resulting value for E is 0.15 ± 0.056 (2cr). Note that 2a corresponds closely to the
standard experimental uncertainty in these estimates (±30%). Also note that E values for all
fluids sprayed under all combinations of liquid mass flow rates and ALR are within 40% of the
mean value. Finally, since normalized entrainment scales linearly with normalized axial distance
(to within our experimental accuracy), we conclude that the model accurately estimates
entrainment in low air-to-liquid mass flow rate ratio, ligament-controlled effervescent atomizer-
produced sprays.
The single E value reported here, for a wide range of conditions, is in contrast to the
earlier results of Bush (1994), who noted that entrainment number depends on the liquid density
and the diameter of the exit orifice for sprays produced by conventional effervescent atomizers.
Bush (1994) attributed the observed scaling to a possible shift in the flow structure at the nozzle
exit or a significant shift in SMD. These phenomena were not observed in this investigation and
their absence may explain the similarity in the entrainment number for all fluids sprayed. The
single value of E may also indicate that the current atomizer provides relatively steady sprays, in
contrast to those produced by conventional effervescent atomizers (Luong, 1996).
Entrainment Control
Normalized entrainment data were obtained with the entrainment device described in
Chapter 2. Four nozzle exit geometries were considered-stepped, inclined, two-point and four-
point. See Figure 10. Each geometry was carefully machined into a separate exit orifice plate
with an exit orifice diameter of 0.38 mm. These geometries were tested at two liquid mass flow
rates (0.5 and 0.6 g/s), four air-to-liquid ratios (0.0075 < ALR < 0.02) and four axial distances
(38 < z < 171 mm). Data were acquired for only a single fluid—water—because previous results
showed that liquid physical properties have little impact on entrainment into effervescent
atomizer-produced sprays (Sutherland et al., 1997). Results are presented in Figures 35 through
38 as normalized entrainment rate versus dimensionless axial distance. Normalized entrainment
rate is defined as me/m], where me is the entrained air mass flow rate and ih| is the liquid mass
flow rate exiting the dispenser. Dimensionless axial distance is defined as z/do, where z is the
axial distance along the spray and d0 is the atomizer exit orifice diameter. In each case, the
uncertainty bars correspond to one standard deviation.
The stepped exit geometry provided sprays of such poor quality (i.e., pronounced central
liquid jets, sprays not exiting perpendicular to the plane of the exit orifice, and large SMDs) that
its defects were apparent to the naked eye. The poor spray quality is believed to be due to the
53
-------
short "stub" that was present in the initial design. As a result, entrainment data for this geometry
is not presented here.
Figure 35 contains data obtained for the inclined exit (or one-point crown) orifice.
Figures 36 and 37 contain data for two versions of the two-point crown exit orifice geometry-
Figure 36 data were obtained using a thinner porous insert than for Figure 37 data. Figure 38
compares the results of Figures 36 and 37 with results obtained for the baseline case by
Sutherland et al. (1997).
0.50
0.40
D
XI
E
c
l>
E
c
c
w
0.30
0.20
0.10 --
0.00
+ H = 0.001 Pa-s, o = 0.072 Pa-m
^ H = 0.020 Pa-s, a = 0.067 Pa-m
^ H = 0.040 Pa-s, o = 0.067 Pa-m
X n = 0.080 Pa-s, o = 0.067 Pa-m
I I = 0.020 Pa-s, o = 0.030 Pa-m
A n = 0.040 Pa-s, 0 = 0.030 Pa-m
Typical
uncertainty
0.00
0.01
0.02
0.03
0.04
0.05
ALR
Figure 34. Entrainment number versus ALR for all fluids.
Several features are apparent when considering these figures. First, the expected linear
relationship between normalized entrainment and normalized axial distance is observed,
regardless of exit orifice type and ALR (see Figures 35 through 37). Note that the correlation
coefficients are above 0.95 in all cases, and routinely greater than 0.99. Second, the slope of the
normalized entrainment versus normalized axial distance lines show the expected increase with
ALR in all cases (see Figures 35 through 37). Finally, by comparing the results of Figure 35
with those of Figure 37, the data demonstrate that entrainment can be controlled through
54
-------
variations in dispenser exit orifice geometry. The extent of control available is indicated in
Figure 38. In each case, the straight line fit to the data has been extrapolated back to zero so the
slopes in the legends (m values) can be used to compare entrainment by the various geometries.
Based on this comparison, we see that entrainment can be increased over the baseline case by-
over 50%.
5.00
4.00
E
.5 3.00
C3
U
e
¦©
O
N
One-point crown
A ALR=0.5%
O ALR-0.75%
~ ALR=I.0%
0.5V., m-0.0058, r2-0.959
0.75%, m-0.0075, r2-0.989
1.0%, m-C.0086, r2=*>.992
at
T,"
E
L.
C
Z.
2.00 —
1.00
0.00
100
500
200 300 400
Normalized axial distance
Figure 35. Normalized entrainment versus normalized axial distance for the inclined (one-
point crown) exit orifice.
Comparison of data from Figures 36 and 37 shows that normalized entrainment rates are
higher for the two-point crown orifice geometry with a thicker porous insert (style #2, and Figure
37) than for the same geometry with a thinner insert (style #1, and Figure 36). The increase is
related to the higher pressure drop experienced by the two-phase flow crossing the thicker insert,
which results in the atomizing gas having a higher velocity at the nozzle exit plane. Higher gas
velocities are associated with higher momentum rates. Since entrainment depends on the exit
momentum rate of the spray (Bush and Sojka, 1994; Sutherland et ah, 1997), increasing the exit
momentum rate should increase entrainment.
The variation in entrainment with nozzle exit orifice geometry is thought to result from
changes in fluid structures within the spray. Changes in these structures in turn alter
entrainment.
For instance, the stepped exit geometry produces "sprays" having pronounced central jets
and exceptionally large SMDs. These sprays have narrow cone angles and, as a result, do not
spread significantly as they propagate away from the nozzle exit orifice. Since the drops do not
55
-------
move far off the centerline, they do not interact with the surrounding air. That lack of interaction
leads to a marked reduction in entrainment.
In the one-point and two-point crown cases, the physical phenomena are different.
Entrainment by the one-point crowns sprays is not significantly different than that by the
baseline case. This suggests that the fluid structures will be the same in both cases. Entrainment
by the two-point crown sprays is significantly different from that by the baseline case sprays. At
this time, we hypothesize that the changes in fluid structures leading to this variation in
entrainment are similar to those found in gas jets.
5.00
E
js
*5
L.
-w
C
¦g
15
E
u
o
Z
4.00
3.00 —
2.00 —
1.00 —
0.00
2-paint crown. ltyk #1
A ALR-0.5%
O ALR-0.75%
0 ALR=1.0%
0.5V., m-0.00«5, r 2-0.983
- 0.75%, bi-O.OOSI. r2-fl.99S
- - 1.8%, m-0.0090, r2*0.993
J*
I
100
500
200 300 400
Normalized axia! distance
Figure 36. Normalized entrainment versus normalized axial distance for the first style of
two-point crown exit orifice (thinner porous insert).
56
-------
8.00
6.00
&>
E
2-point crown, style #2
A ALR=0.5%
O ALR»0.75%
~ ALR=I.0%
0.5%, m =0.0090, r2«€995
0.75*/., m-0.010, r2-0.995
- - 1 .0%, m«O,012, r2«0.995
£
E
u
©
Z
4.00 —
2.00 —
0.00 —i 1 1 1 1 1 j 1 1
100 200 300 400 S00
Normalized axial distance
Figure 37. Normalized entrainment versus normalized axial distance for the second style of
two-point crown exit orifice (thicker porous insert).
57
-------
8.00
6.00
e
o>
£
c
-------
Chapter 5
Summary and Conclusions
Summary
The results of steady-state drop size performance portion of this study can be summarized
as follows: (1) Sub-70 ^m SMD sprays were obtained at and below the target air/liquid ratio
(ALR) of 0.009; (2) SMD was observed to increase markedly at ALRs below 0.0075, but to
decrease only slightly for ALRs from 0.01 to 0.02; (3) SMD showed only a minor (< about 10%)
increase when liquid viscosity or surface tension was increased throughout the range considered
during this investigation (0.001 to 0.080 kg/m-s and 0.030 to 0.072 kg/s2, respectively).
Three-dimensional holographic and high-speed photographic images indicate why mean
diameter increased markedly as ALR was reduced below 0.075; the near-nozzle breakup
structure undergoes a transition from an annular band of small diameter ligaments surrounding a
gas core to a much smaller number of larger diameter ligaments.
An analytical model was developed to describe atomizer performance. It is based on the
expression of Sterling and Sleicher (1975) for the instability of capillary liquid jets exposed to a
moving air stream, and requires knowledge of the pore size of the porous medium and the
relative velocity between the air and liquid at the nozzle exit plane. The model correctly predicts
the influence of ALR, liquid surface tension, and liquid viscosity on SMD.
Transient performance was shown to be acceptable. Although an initial (transient)
increase in SMD was observed, time averaged SMDs were below the target value of 70 (am and
the fraction of time for which drop sizes were above that value was small. Finally, the SMD
always remained reasonable, not climbing much above 100 fim.
The results of the entrainment portion of this study are summarized as follows: (1)
Normalized entrainment by ligament-controlled effervescent atomizer produced sprays is linearly
proportional to normalized axial distance, as has been reported for gas jets and for some other
types of sprays; (2) Normalized entrainment by these sprays increases with atomizing air-to-
liquid mass flow rate ratio (ALR), as would be expected since an increase in ALR increases the
initial momentum rate of the spray; entrainment data obtained at two liquid mass flowrates
collapsed onto a common line; (3) Liquid surface tension has a negligible effect on normalized
entrainment; increasing liquid viscosity from 0.001 to 0.020 Pa-s increased normalized
entrainment by approximately 30 to 50%, with further increases in liquid viscosity having a
much smaller impact (less than 12%); (4) The initial spray momentum rate is proportional to
ALR and to liquid mass flow rate, as expected; (5) Liquid surface tension had a negligible effect
on momentum rate; the influence of viscosity was mixed with momentum increasing with an
increase in viscosity in some cases and decreasing with an increase in viscosity in others; (6) The
entrainment number is relatively insensitive to liquid physical properties-all variations were
within experimental uncertainty, defined as the standard uncertainty or experimental standard
deviation; (7) The entrainment number increases with ALR, as would be expected by considering
the limiting cases of pure liquid (ALR=0) and pure gas (ALR=oo) jets-the value of entrainment
number determined here, 0.15±0.056 (2ct), lies between the gas jet value, 0.282, reported by
Ricou and Spalding (1961) and the liquid jet (issuing into air) value ofO.
59
-------
Entrainment control was demonstrated. Entrainment was varied by changes in the nozzle
exit orifice geometry, with both increases (of up to 50%) and decreases (to effectively zero
entrainment) relative to a flat plate baseline nozzle being realized. In addition, the linear
relationship between normalized entrainment and normalized axial distance, and between
normalized entrainment and ALR, observed in the baseline case was retained for the single-point
and two-point crown configurations.
Conclusions
Two conclusions were drawn from the steady-state drop size results of this study:
• Ligament-controlled effervescent atomizers are an effective means to achieve sub-70 |im
SMD consumer product sprays, within the supply pressure and ALR constraints imposed by
Department of Transportation regulations and deceptive packaging guidelines. They
facilitate replacement of volatile organic compound (VOC) solvents and hydrocarbon (HC)
propellants with environmentally benign water and air.
• The model given by Equation (9) successfully predicts the SMD of ligament-controlled
effervescent atomizer produced sprays.
We also conclude that transient effervescent atomizer performance is acceptable. The
transient increase in SMD observed near the start of the spray process is attributed to a "burst" of
liquid that occurs upon initiation of the gas and liquid flows.
The following conclusions can be drawn, based on the results of the entrainment
portion of this study :
• Entrainment into steady two-phase jets where the gas and liquid streams exhibit inter-
phase velocity slip can be modeled using the momentum rate approach of Ricou and
Spalding (1961), although their entrainment number value is no longer applicable. The
experimentally determined entrainment number for the two-phase jets studied here
increases with ALR.
• The appropriate experimentally determined entrainment number, E, for ligament-
controlled effervescent atomizer produced sprays is 0.15±0.056 (2a). This value of E
predicts entrainment to within 40% for sprays considered in this study.
• Entrainment depends on the structure of the spray present at the atomizer exit, as shown
by a comparison of results from this study with those of Bush (1994).
Data obtained during the entrainment control portion of this study lead us to conclude
that entrainment into ligament-controlled effervescent atomizer-produced sprays can be
modified by correct selection of the dispenser exit orifice geometry. We hypothesize that
changes in entrainment are due to changes in fluid structures near the dispenser exit plane, and
that in some cases the mechanisms are similar to those found in gas jets.
Interested manufacturers now have the tools necessary to design effervescent
atomizers for consumer product dispensation. The design procedure proceeds as follows:
• Determine product physical properties density, surface tension and viscosity.
• Compute ALR, by mass, that is consistent with the desired package.
• Choose a porous medium, thereby dictating the pore size.
• Use Equations (7) and (9) to calculate the resulting mean drop size. Note that use of
Equation (7) requires knowledge of the relative velocity that exists between the atomizing
60
-------
gas and product to be sprayed. An estimate for the relative velocity can be obtained from
Table 6, or from momentum rate measurements and Equation (8). The latter method is
recommended.
61
-------
References
ASTM Standard E 799-87. Standard Practice for Determining Data Criteria and Processing for
Liquid Drop Analysis. American Society for Testing and Materials: Philadelphia.
Beer, F. P. and Johnston, E. R. 1981 Mechanics of Materials. McGraw-Hill, New York.
Bell, J.H. and Mehta, R.D. 1993 Effects of imposed spanwise perturbations on plane mixing-
layer structure. Journal of Fluid Mechanics, 257, 33-63.
Benatt, F.G.S. and Eisenklam, P. 1969 Gaseous Entrainment into Axisymmetric Liquid Sprays.
Journal of the Institute of Fuel, August, 309-315.
Binark, H. and Ranz, W.E. 1958 Induced Air Flows in Fuel Sprays. ASME Paper No. 58-A-284 (1958
Annual Meeting of the ASME).
Boysan, F. and Binark, H. 1979 Predictions of Induced Air Flows in Hollow Cone Sprays. Journal of
Fluids Engineering, 101, 313-318.
Bradbury, L.J.S. and Khadem, A.H. 1975 The distortion of a jet by tabs. Journal of Fluid
Mechanics, 70, 801-813.
Briffa, F.E.J, and Dombrowski, N. 1966 Entrainment of Air into a Liquid Spray. AlChE Journal,
12(4), 708-717.
Buckner, H.N. and Sojka, P.E. 1993 Effervescent Atomization of High Viscosity Fluids. Part
II: Non-Newtonian Liquids. Atomization and Sprays, 3(2), 157-170.
Bush, S.G. 1994 Entrainment by Effervescent Sprays at Low Mass Flow Rates. MSME Thesis.
Purdue University.
Bush, S.G. and Sojka, P.E. 1994 Entrainment by Effervescent Sprays at Low Mass Flow Rates,
presented at the 6th ICLASS, Rouen, France.
Bush, S.G., Bennett, J.B., Sojka, P.E., Panchagnula, M.V., and Plesniak, M.W. 1996 A
Momentum Rate Probe For Use With Two-Phase Flows. Review of Scientific Instruments, 67(5),
1878-1885.
Deichsel, M. and Winter, E. R. F. 1990 Adiabatic Two-Phase Flow of Air-Water Mixtures
Under Critical Flow Conditions. International Journal of Multiphase Flow, 16(3), 391-406.
Eckerle, W.A., Sheibani, H., and Awad, J. 1992 Experimental Measurements of the Vortex
Development Downstream of a Lobed Mixer. Trans ASME J. Engr. Gas Turbines & Power,
114, 63-71.
Geckler, S.C. and Sojka, P.E. 1995 Effervescent Atomization of Viscoelastic Liquids. Under
review for Atomization and Sprays.
Ho, C. M. and Gutmark, E. 1987 Vortex Induction and Mass Entrainment in a Small-Aspect-
Ratio Elliptic Jet. Journal of Fluid Mechanics, 179(2), 383-405
62
-------
Ishii, M. 1977 One Dimensional Drift -Flux Model and Constitutive Equations for Relative
Motion Between Phases in Various Two-Phase Flow Regimes, Argonne National Laboratory
Report, 47-77.
Krishnappa, G. and Csnady, G.T. 1969 An experimental investigation of the composition of jet
noise. Journal of Fluid Mechanics, 37,149.
Krothapalli, A., Baganoff, D., and Karamcheti, K. 1981 On the Mixing of a Rectangular Jet.
Journal of Fluid Mechanics, 107, 201-220.
Lefebvre, A.H., Wang, X.F., and Martin, C.A. 1989 Spray Characteristics of Aerated-Liquid
Pressure Atomizers. A1AA J Propulsion & Power, 4,293-298.
Longmire, E.K., Eaton, J.K., and Elkins, C.J. 1992 Control of Jest Structure by Crown-Shaped
Nozzles. AIAA Journal, 30(2) 505-512.
Lund, M.T. 1992 Effervescent Atomization at Low Flow Rates, MSME Thesis, Purdue
University.
Lund, M.T., Sojka, P.E., Lefebvre, A.H., and Gosselin, P.G. 1993 Effervescent Atomization at
Low Mass Flow Rates. Part I: The Influence of Surface Tension. Atomization and Sprays, 3(1),
77-89.
Luong, J.T.K. 1996 Unsteadiness in Low Mass Flow Rate Effervescent Atomizer Produced
Sprays, MSME Thesis, Purdue University.
MacGregor, S. A. 1991 Air Entrainment in Spray Jets. International Journal of Heat and Fluid Flow,
12(3), 279-283.
Rasbash, D.J. and Stark, G. 1962 Some Aerodynamic Properties of Sprays. The Chemical
Engineer, December, A83-A88.
Reed, D. 1986 Porex Technologies, Fairburn, GA, private communication.
Ricou, F. P and Spalding, D. B. 1961 Measurements of Entrainment by Axisymmetric
Turbulent Jets. Journal of Fluid Mechanics, 11,21-32.
Roesler, T.C. and Lefebvre, A.H. 1989 Studies on Aerated-Liquid Atomization. lnt'lJTurbo&
Jet Engines, 6,221-239.
Rothe, P.H. and Block, J. A. 1977 Aerodynamic Behavior of Liquid Sprays. International Journal of
Multiphase Flow, 3, 263-272.
Ruff, G.A., Sagar, A.D., and Faeth, G.M. 1989 Structure and Mixing Properties of Pressure-Atomized
Sprays. A1AA Journal, 27(7), 901-908.
Santangelo, P.J. 1993 A Holographic Investigation of the Near Nozzle Structure of an
Effervescent Spray, Master of Science Thesis, Purdue University.
Santangelo, P.J. and P.E. Sojka, P.E. 1993 Holographic Particle Diagnostics. Progress in
Energy and Combustion Science, 19, 587-603.
63
-------
Santangelo, P.J. and Sojka, P.E. 1994 Focused-Image Holography as a Dense Spray Diagnostic,
Applied Optics, 33(19), 4132-4136.
Santangelo, P.J. and Sojka, P.E. 1995 A Holographic Investigation of the Near-Nozzle Structure
of an Effervescent Atomizer-Produced Spray. Atomization and Sprays, 5(2), 137-155.
Schadow, K.C., Gutmark, E., Parr, D.M. and Wilson, K. 1988 Selective Control of Flow
Coherence in Triangular Jets. Experiments in Fluids, 6, 129-136.
Sterling, A. and Sleicher, C. 1975 The Instability of Capillary Jets, Journal of Fluid Mechanics,
68(3), 477-495.
Sutherland, J.J..1996: Ligament-controlled Effervescent Atomization. MSME Thesis, Purdue
University.
Sutherland, J.J., Sojka, P.E., and Plesniak, M.W. 1997 Entrainment by Ligament-Controlled
Effervescent Atomizer-Produced Sprays. International Journal of Multiphase Flow, 23(5), 885-
897.
Tishkoff, J. 1985: Air Entrainment Into Sprays From Swirl Chamber Atomizers. Proceedings
3rd ICLASS, London, UK.
Wang, X.F., Chin, J.S., and Lefebvre, A.H. 1989 Influence of Gas Injector Geometry on
Atomization Performance of Aerated-Liquid Nozzles. International Journal of Turbo & Jet
Engines, 6, 271-279.
Weber, C. 1931 Disintegration of Liquid Jets, Z. Angew. Math. Mech., 11, 136-159.
White, F.M. 1991 Viscous Fluid Flow, Second Edition. McGraw-Hill, New York.
Whitlow, J.D. and Lefebvre, A.H. 1993 Effervescent Atomizer Operation and Spray
Characteristics. Atomization and Sprays, 3(2), 137-155.
Wlezien, R.W. and Kibbens, V. 1986 Passive Control of Jets with Indeterminate Origins. AIAA
Journal, 24(8), 1263-1270.
Zaman, K.B.M.Q., Reeder, M.K., and Samimy, M. 1994 Control of an axisymmetric jet using
vorteX generators. Physics of Fluids, 6, 778.
Zhang, S. and Schneider, S.P. 1995 Quantitative molecular-mixing measurements in a round jet
with tabs. Physics of Fluids, 7(5), 1063-1070.
64
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completingj
1. REPORT NO. 2.
EPA-600/R-98-089
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Development of an Innovative Spray Dispenser to
Reduce Indoor Air Emissions from Aerosol
Consumer Products
5. REPORT DATE
July 1998
6. PERFORMING ORGANIZATION CODE
7. AUTHORIS)
P.E. Sojka and M. W. Plesniak
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Purdue University
School of Mechanical Engineering
West LaFayette, Indiana 47907
10. PROGRAM ELEMEN1 NO.
11. CONTRACT/GRANT NO.
CR822618
12. SPONSORING AGENCY NAME ANO ADDRESS
EPA, Office of Research and Development
Air Pollution Prevention and Control Division
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final; 2/94 - 9/97
14. SPONSORING AGENCY CODE
EPA/600/13
15.supplementary notes APPCD project officer is Kelly W. Leovic, Mail Drop 54. 919/
541-7717.
16. abstractreport presents the operating principles and performance of a new type
of spray nozzle. This nozzle, termed a "ligament-controlled effervescent atomizer,'
was developed to allow consumer product manufacturers to replace volatile organic
compound (VOC) solvents with water and hydrocarbon (HC) propellants with air,
while meeting the following restrictions: that the spray mean drop size (reported
here as Sauter mean diameter, or SMD) remain below 70 micrometers, that the
atomizing air consumption be less that 0.009, and that atomizer performance be
uncompromised by the increase in surface tension or by changes in viscosity. The
current atomizer differs from previous effervescent designs by inclusion of a por-
ous disk immediately upstream of the nozzle exit orifice. This disk controls the
diameter of ligaments formed at the injector exit plane. First, steady-state atom-
izer performance is reported in terms of the spray SMD. Transient atomizer per-
formance is reported second, again in terms of spray SMD. Entrainment of am-
bient air into these sprays is reported last. Approaches for controlling entrainment
(i. e., modification of the dispenser exit orifice geometry) are also introduced, and
their utility discussed in terms of their entrainment number values.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COS ATI Field/Group
Pollution Atomizing
Sprayers
Emission
Aerosols
Atomizers
Spray Nozzles
Pollution Control
Stationary Sources
Indoor Air Quality
Consumer Products
Effervescence
13 B 13 H
07 A, 131
14G
07D
13K
13 G
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. N^gOF PAGES
20. SCCURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
------- |