Available via license: CC BY-NC 3.0
Content may be subject to copyright.
Original Research Article
A novel spray generator
for low-energy oil burners
M Etzold, Y Han and F Durst
Abstract
Outcomes of development work that resulted in a spray generator which yields spray droplets of small Sauter diameters
at very low flow rates through the spray-generating nozzle are reported. This nozzle produces a liquid jet that operates
in two modes to switch on or switch off the spray production. To achieve this, a liquid jet impinging on a flat surface is
employed. The jet is operated at Ohnesorge–Reynolds numbers that do not yield a spray. The conditions for this state of
operation are given. The spray production is, in a second mode of operation, initiated by triggering the Rayleigh instability
of this jet so that the impingement on the flat plate occurs in the form of drops. As the drops are initiated, an
intermittent spray is produced by the flat surface, that is, each droplet results in a spray leaving the flat surface. This
is demonstrated experimentally and measurement results are presented to show that droplets of small Sauter mean
diameters are produced in this way. The influence of the jet-generating nozzle diameter on the Sauter diameter is also
reported. Comparisons of the droplet-generated sprays are carried out for various other forms of spray generators,
keeping the nozzle diameter and the flow rate the same as those for the droplet sprays.
Keywords
Spray, spray generator, droplet sprays, atomization at low flow rates
Date received: 222; accepted: 222
Introduction
Motivation and field of application
In the market for low-energy burners, gas burners dom-
inate the field of applications because there is no real
lower limitation on the gas fuel supply. This is not the
case when oils are used as fuels. In this case, the fuel for
low-energy burners needs to be prepared for the com-
bustion process, that is, a spray generator needs to be
employed. Suitable generators for low-energy burners
require properties that are not readily available with
existing spray generators, for example, such burners
should operate well at the lower power end of an oil
burner and also at its upper end, preferably providing a
power range of 1–25. With such a dynamic power
range, burners could be designed to operate in the
power range from 300 W–7.5 kW. Burners of this kind
are currently not available and it was the aim of this
work to provide a spray generator for heating oil that
permits low-energy oil burners to be constructed with a
power modulation range of 1–25. The research and
development work described in this paper utilizes the
differences in the spray production properties between
liquid jets and liquid drops to obtain the low power and
the operating range needed for low-energy oil burners.
The realization of this novel spray generator is
described in Section 3.
Summary of spray generation principles
Sprays are extensively used in various fields of engin-
eering science and medicine to provide swarms of drop-
lets for coating purposes, injection of fuels into internal
combustion engines, homogenization of fluid emul-
sions, powder production through spray drying, etc.
For these different applications, different spray gener-
ators have been developed and applied for spray
FMP Technology GmbH, Germany
Corresponding author:
M Etzold, FMP Technology GmbH, Am Weichselgarten 34, 91058
Erlangen, Germany.
Email: m.etzold@fmp-technology.com
International Journal of Spray and
Combustion Dynamics
0(0) 1–12
! The Author(s) 2016
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1756827715627065
scd.sagepub.com
Creative Commons CC-BY-NC: This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 3.0 License (http://www.
creativecommons.org/licenses/by-nc/3.0/) which permits non-commercial use, reproduction and distribution of the work without further permission provided the
original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage).
formation in the above fields. The applied spray gener-
ation methods can be grouped as follows:
. spray generation by turbulence-caused disintegra-
tion of liquid jets;
. formation of liquid lamellas or sheets and triggering
of their instabilities;
. air-assisted liquid jet disintegrations by shear and
elongational forces in gas streams.
For the technical realization of these methods, various
spray generators have been proposed. Summaries of
these different spray generators and of their employ-
ment in various fields were given by, for example,
Lefebvre
1
and Walzel.
2
The spray generator described in this paper uses a
single fluid nozzle to provide the resultant spray.
Therefore, in this short literature survey, only sprays
produced by single fluid nozzles are considered. Their
properties are dependent on the nozzle diameter D and
the supply pressure P applied to drive the spray-pro-
ducing fluid through the nozzle. These two properties
define the liquid mass flow rate
_
m of the produced spray
_
m ¼
4
D
2
ffiffiffiffiffiffiffiffiffiffi
2P
s
ð1Þ
where is the liquid density, D the nozzle diameter, P
the pressure difference between nozzle inlet and nozzle
outlet and the discharge coefficient, which depends on
the nozzle geometry and liquid properties.
A corresponding relationship for of hollow-cone
nozzles was given by Musemic and Walzel
3
.
Hollow cone nozzles form the best sprays for liquids
with low viscosities. In general they can produce very
fine sprays. A relationship for the Sauter mean diam-
eter d
32
of a hollow-cone nozzle spray generator was
given by, for example, Lefebvre and Wang
4
d
32
¼ 4:52
2
A
P
0:25
½h cos
0:75
þ 0:39
A
P
0:25
½h cos
0:25
ð2Þ
In the upper equation, liquid properties such as the
viscosity , the surface tension and the density are
included, in addition to nozzle/flow properties such as
the half spray cone angle and the thickness of the
liquid film h. Another equation was given by Walzel
5
for the Sauter mean diameter of hollow-cone nozzle
spray generators
d
32
D
¼ 1:6
1
We
13
0:4
1 þ 5OhðÞ
1:5
ð3Þ
The dimensionless quantities Weber (We) and
Ohnesorge (Oh) numbers can be computed by the fol-
lowing equations
We ¼
DU
2
¼
2 PD
ð4Þ
Oh ¼
ffiffiffiffiffiffiffiffiffiffi
D
p
ð5Þ
From both equations (2) and (3) one can see that the
Sauter mean diameter d
32
of a hollow-cone nozzle
depends also on the pressure difference P . By refor-
mulating the above equations, one can show that the
Sauter mean diameter is a function of the flow rate and
of the pressure difference: d
32
ð
_
mðPÞÞ. A decrease in
mass flow rate yields an increase in Sauter mean diam-
eter. However, one must realize that the mass flow rate
cannot be decreased to an arbitrary low value. If the
mass flow rate and therefore the supply pressure to the
spray generator are too low, there is no or only very
poor atomization. This means that there exists a min-
imum Weber number for spray formation. Such a low
limit of the Weber number was given by Walzel:
5
We ¼1700. Furthermore, with common single fluid
nozzles it is not possible to modulate the aerosol mass
flow rate without affecting its drop size distribution.
This is another reason why existing spray generators
do not possess the essential properties required for
applications in low-energy oil combustion.
Approaches to increase the modulation range of pres-
sure-swirl nozzles have been described; see Lefebvre and
Wang.
4
One is the duplex atomizer, which has two tan-
gential ports in the swirl chamber. One port is used for
low flow rates. At a certain pressure, a spring-loaded
valve opens the second port and both ports work
together. Another spray system with such properties is
the dual-orifice atomizer, which has two inlets: one for
low and the other for high flow rates. It consists out of
two simplex nozzles. By combining these two nozzles, a
wide range of liquid flow rates can be achieved with good
atomization properties. Spill return atomizers are simplex
nozzles with a spill in the swirl chamber. With the help of
a valve, the amount of flow through the spill can be con-
trolled. Therefore, it is possible to run the nozzle at high
pressures and low flow rates by changing the flow through
the spill. With these techniques, the modulation range of
single fluid atomizers can be increased to 20.
An atomizer that fulfills the requirement of constant
spray properties for all flow rates and has, at the same
time, the mass flow modulation range required for low-
power burners is not available. This was a major motiv-
ation for the authors to develop an atomizer of this
kind, and this aim has been achieved, as described in
the following sections.
2 International Journal of Spray and Combustion Dynamics 0(0)
The formation of sprays by jets and drops
To produce liquid jets through small orifices several
hundred micrometer in diameter, a certain mass flow
rate is required. This flow rate is reached at the jetting
point, which was given, for example, by Walzel
6
as
We ¼ 8 ð6Þ
This relation is represented in an Ohnesorge (Oh)–
Reynolds (Re) diagram (see Figure 1) by the line (a).
Line (b) in Figure 1 distinguishes between the two
regions, with and without atomization, of a liquid jet
impinging on a plate. The corresponding condition
depends on the jet velocity, the shape and roughness
of the solid body and obviously on the liquid proper-
ties. If a liquid jet impinges on a solid flat plate, formed
by a pin of cylindrical shape with sharp edges, it tends
to form a lamella which itself disintegrates into droplets
at very low velocities (see Figure 2). Since the aim of the
present study was a stable jet without any atomization
at high velocities, the edges of the border of the flat
surface on top of the cylindrical pin were rounded or
the pin top was hemispherical shaped (see Figure 10
later). Owing to these shapes, the liquid film is stabi-
lized by the so-called Coanda effect and the liquid flows
down the pin. In comparison with a flat plate with
sharp edges, higher jet velocities can be obtained with-
out atomization.
The attraction of a liquid or a gaseous jet to a solid
body is called the Coanda effect. Figure 3 illustrates the
basic principle of this effect. A flat jet is attached to the
surface of a curved body. It separates at an angle that
is very much higher than expected. The reason for
this late separation is the specific velocity and pressure
distribution of the near-wall flow. The smaller the
thickness of the jet in relation to the radius of the
solid body t/R , the larger the angle before separation
occurs. To achieve a high flow separation angle for
given jet velocities and thicknesses, large radii R are
necessary. A body with sharp edges leads to flow sep-
aration since the radius tends towards zero (see Figure
2). Hence, for the stabilization of the liquid jet, a pin
with rounded edges was used in this work. For high jet
velocities and Reynolds numbers and small jet thick-
nesses t/R, a flow separation angle of 240
has been
reported (Truckenbrodt
7
). Further information regard-
ing the Coanda effect is available, for example, from
Wille and Fernholz
8
and Truckenbrodt
7
.
In Section 4, the critical Weber number for flow sep-
aration is determined experimentally for a jet impinging
on a cylindrical body with a hemispherical tip. For a
dimensionless sphere diameter D
S
¼ D
S
=D ¼ 10, the
following criterion could be obtained from the experi-
ments reported in Section 4
We
crit
5000 ð7Þ
Below the Weber number given in equation (7), the
liquid film formed from the jet flows down the cylin-
drical pin.
In Figure 1, the different above-mentioned regions
are shown in one diagram. The region of liquid drip-
ping (no jet formation) is limited by the jetting criterion
given by equation (6). The other two regions feature jet
Figure 2. A liquid jet impinging on a cylindrical pin with sharp
edges leading to flow separation and to the formation of a lamella
(We ¼695; D
pin
/D ¼6.7).
Figure 1. Ohnesorge–Reynolds diagram to show the spray
behavior of jets impinging on plates: (a) jetting criterion, equation
(6); (b) criterion for flow separation/atomization of an impinging
jet flow on a solid surface, equation (7).
Etzold et al. 3
formation with and without atomization by impinging
on a hemispherical-shaped pin marked off by
equation (7).
For drops impinging on a solid flat surface, the
range of spray and no spray formation can also be
expressed in terms of Reynolds and Ohnesorge num-
bers. The so-called splashing behavior of drops was
investigated, for example, by Mundo et al.
9
They
found an empirical relation for the splashing limit
of impinging drops depending on Reynolds and
Ohnesorge numbers. Their findings were given by the
relation
Oh
d
Re
5=4
d
4 57:7 ð8Þ
The dimensionless numbers are computed in the above
relation with the drop diameter. Assuming that the
drops are generated by Rayleigh breakup of laminar
jets from a nozzle with diameter D, the approximate
drop diameter is d ¼2D. Then equation (8) can be
reformulated
2
3=4
OhRe
5=4
4 57:7 ð9Þ
Figure 4 shows an Ohnesorge–Reynolds diagram for
the atomization behavior of impinging drops. The
regions of spray and no spray formation are described
by equation (9).
The generation of drops by Rayleigh breakup
requires a liquid jet with a laminar velocity distribution.
To achieve a laminar jet, its velocity has to stay below a
certain value. Above this value, a transition from lam-
inar to turbulent flow takes place. A criterion for this
effect was given by, for example, Van de Sande and
Smith
10
Re ¼ 12000
l
0
D
0:3
ð10Þ
where l
0
is the channel length of the nozzle. This limit-
ing line is also shown in Figure 4.
By combining the above mentioned atomization
properties of jets and drops impinging on a plate,
a single Ohnesorge–Reynolds diagram arises; see
Figure 5. All the plotted lines refer to equations of
dimensionless quantities, which are calculated with
the same liquid properties and with the same character-
istic length scale, namely the nozzle diameter D.
Therefore, it is possible to represent all the criteria of
the Figures 1 and 4 together in one diagram.
From Figure 5 it can be seen that there is a region
(highlighted gray) where laminar liquid jets are pro-
duced and are stable when they impinge on a plate,
that is, there is no spray generation. At the same
time, drops with a diameter assumed to be twice the
jet diameter (generated, for example, by the Rayleigh
breakup) with the same velocity as a jet, form sprays
after impinging on a plate. In order to obtain a large
region of no jet splashing but drop splashing, it is
advantageous to use the Coanda effect to stabilize the
liquid film formed by the impinging jet. This leads to a
shifting of curve (b) in Figure 5 and to an increase in
the gray highlighted area. Therefore, a solid body with
rounded edges or a hemispherical shape is necessary to
produce a spray when droplets impinge on the pin and
no spray exists when a liquid jet impinges. Figure 6
shows a sketch which illustrates this above-mentioned
Figure 3. The Coanda effect: attraction of a jet to a solid body
of round shape (see Wille and Fernholz
8
).
Figure 4. Ohnesorge–Reynolds diagram to show spray behav-
ior of drops impinging on plates: (c) splashing criterion for
impinging drops on a solid surface, equation(9); (d) transition
from laminar to turbulent jet flow (for l
0
D ¼ 0:5), equation(10).
4 International Journal of Spray and Combustion Dynamics 0(0)
approach of spray formation. It can be seen that there
is a radial axisymmetric direction of spray propagation.
The above-mentioned finding regarding a flow
region without jet splashing but with drop splashing
under the same flow conditions was used to develop
the new spray generator. Its working principle is
described in the section below.
Realization of the spray generator
The main working principle of the developed spray gen-
erator is the disintegration of liquid drops impinging on
a plate. The inventive approach towards a new spray
generator is that, under certain conditions, a liquid jet
does not splash after impinging on a plate but imping-
ing drops are sprayed (at the same impingement vel-
ocity). When no atomization occurs, the whole flow
can be returned (back flow
_
V
B
). Figure 6 shows two
sketches to illustrate the operating principle. If a jet
impinges on the plate and there is no atomization, the
total flow rate
_
V
T
can be returned (
_
V
B
¼
_
V
B
max
¼
_
V
T
)to
the container from which the fluid is pumped. The
spray flow rate
_
V
S
is then zero (left side of Figure 6).
An impinging drop chain with the same total flow rate
produces a spray (
_
V
S
¼
_
V
S
max
), but there is still a
small back flow
_
V
B
¼
_
V
B
min
¼
_
V
T
_
V
S
max
(right side of
Figure 6).
By switching between these two states, the aerosol
mass flow rate can be controlled, since atomization
occurs only when drops are impinging on the top of
the cylindrical pin. The switching operation can be rea-
lized by a pulse-width modulated signal with period T,
so that during the duty cycle, drops are produced. The
longer the duty cycle time t chosen, the more spray is
produced. The mean spray flow rate
_
V
S
avr
can then be
calculated with
_
V
S
avr
¼
t
T
_
V
S
max
¼
t
T
_
V
T
_
V
B
min
ð11Þ
In Figure 7, the different flow rates obtained by pulse-
width modulation of the drop chain production are
shown.
The production of a liquid jet was realized by a
nozzle with diameter D, which is the same as the jet
diameter. To achieve the above-mentioned spray
effect, single drops were produced from the jet. This
was realized by triggering the Rayleigh instability of
the jet with the help of a piezo oscillator.
The Rayleigh breakup acts on every laminar jet. It
was first discovered by Lord Rayleigh.
11,12
The natural
breakup leads to a polydispersed spray with drop diam-
eters approximately twice the nozzle diameter (d & 2D).
By triggering the Rayleigh instability with a piezo actu-
ator, the generation of monodisperse sprays is possible.
Then the disintegration into single drops occurs with
the oscillating frequency of the piezo driving signal. The
size of the generated drops depends on the frequency of
the signal f, the nozzle diameter D, the jet velocity U
and the liquid properties. The achievable diameters are
in the range [1.74 D 2.50 D]. For the work described
in this paper, a commercial monodisperse droplet gen-
erator of FMP Technology GmbH was used. Further
information about this device and the controlled
breakup of laminar jets was given by Brenn et al.
13,14
The spray-generating principle is the atomization
due to collisions of drops with a plate. Many research-
ers have investigated this kind of spray generation
mechanism. Schmidt and Knauss
15
investigated the
Figure 5. Ohnesorge–Reynolds diagram to show the spray
behavior of jets and drops: (a) jetting point, equation (6); (b)
criterion for flow separation/atomization of an impinging jet flow
on a solid surface, equation (7); (c) splashing criterion for
impinging drops on a solid surface, equation (9); (d) transition
from laminar to turbulent jet flow (for l
0
D ¼ 0:5), equation (10).
Gray highlighted region: range of drop splashing and no jet
splashing at the same flow conditions (d ¼2D).
Figure 6. Impinging jet on a plate with no spray production
(left) and impinging drop chain with spray production (right).
Etzold et al. 5
drop size distribution of the generated spray by imping-
ing water and mercury drops on smooth surfaces at
relatively low velocities. They obtained mean diameter
values in the range d
mean
=d ¼ 0:2; 1½. Walzel
16
investigated the splashing behavior of drops impinging
on wet and dry surfaces. Mundo et al.
9
derived an
empirical relation for the splashing behaviour of
impinging drops that depends on Reynolds and
Ohnesorge numbers. Mundo et al.
9
also performed
measurements of the drop diameters in sprays produced
by drop breakup.
In Figure 8 two photographs are shown. The left one
shows a jet (D ¼100 mm) impinging on a hemispherical-
shaped pin with a velocity of U ¼24m/s. No atomiza-
tion occurs. On the right image, the piezo oscillator was
switched on, leading to controlled disintegration of the
jet into single, monodisperse drops and these drops
formed a spray by impinging on a cylindrical plate.
The system pressure and therefore the velocity are the
same in both photographs. Hence, these pictures point
out that the above-mentioned considerations of spray
and non-spray forming conditions are applicable in
reality.
Coanda effect for liquid jet stabilization
and corresponding measurements
Aim of the experiment
As mentioned in the section above, the spray generator
presented in this paper operates by switching between
two flow regimes. One of these flow regimes is a smooth
liquid jet that impinges on a flat surface or a rounded
pin, and the jet flow is stabilized by the Coanda effect
(see Section 2). In this jet mode, the flow is adjusted to
produce no spray. When the droplet generation is
switched on, the spray formation sets in.
Figure 8. A jet (left) and a droplet chain (right) impinging on a pin running under same conditions (We ¼348; D
S
¼ 13:3). The jet
flow does not atomize whereas the droplet chain disintegrates into a spray.
Figure 7. Operating principle of the novel spray generator.
6 International Journal of Spray and Combustion Dynamics 0(0)
In order to achieve a spray with small droplet diam-
eters, it is advantageous to increase the Weber number,
that is, the jet velocity (see Figure 13 in Section 5.2).
With increasing velocity, it is intuitively clear that a
liquid jet also tends to form a spray. This conflict of
goals between good atomization with a droplet chain
and no atomization with a jet needed to be solved by a
suitable design of the impinging plate. Therefore,
experiments with impinging liquid jets on a cylindrical
pin with a hemispherical shape at its tip were carried
out. The aim of the experiments was to determine the
critical Weber number We
crit
where the liquid film,
which is attached to the pin by the Coanda effect, sep-
arates from the pin. This is the initial state of spray
formation by the jet flow also.
Experimental set-up
In Figure 9, the experimental set-up is shown. The
liquid jet was formed by a nozzle, which was set up
to study the spray formation by an impinging jet. To
produce the jet, the nozzle was supplied with water
stored in a vessel with a maximum pressure of 30 bar.
The jet impinged on a cylindrical pin with a hemispher-
ical shape at the tip and the behavior of the flow was
observed by a charge-coupled device (CCD) camera.
The diameter of the hemisphere was D
S
¼2 mm.
Nozzle diameters D in the range 50–500 m were used.
By increasing the water pressure and therefore the jet
velocity, the critical point at which flow separation
occurred could be determined. The mass flow rate
_
m
was measured by measuring the mass of water con-
veyed through the nozzle in 1 min. The critical Weber
number We
crit
was calculated with the nozzle diameter
D and with the jet velocity U, which could be obtained
by the measured mass flow rate and the nozzle diameter
(U ¼ 4
_
mD
2
). Figure 10 shows camera shots at two
different Weber numbers with a jet diameter of 150 m.
In the right image the Weber number was below the
critical value, hence no flow separation is visible. The
left image was obtained at a jet velocity above the crit-
ical velocity. Here the liquid film is partly separated
from the pin.
Results
Figure 11 shows the results for the critical Weber
number at different dimensionless hemisphere diameters
(D
S
¼ D
S
=D). The displayed range of D
S
extends only to
Figure 10. Impinging jet on a pin with a hemispherical tip; D
S
¼ 13:3. Left, spray formation due to flow separation, We ¼5380; right,
no spray formation, We ¼2649
Figure 9. Experimental set-up for the investigation of the sta-
bilization of the liquid jet by the Coanda effect.
Etzold et al. 7
13.3. The reason is that at higher hemisphere diameters,
the critical jet velocity could not be reached with the
maximum vessel pressure of 30 bar. The graph illustrates
that with increasing sphere diameter, the critical Weber
number also increases to D
S
¼ 10. Above this value
there is no major change in the critical Weber number,
which is nearly constant at We
crit
& 5000. The results
show that with a pin with a hemispherical shape and a
diameter 10 times larger than the jet diameter, Weber
numbers up to 5000 can be adjusted and no flow sep-
aration occurs. In dimensional notation, We
crit
& 5000
corresponds to a water jet with a diameter of 100 m
with a velocity of 60 m/s or a pressure of 18 bar.
Measurements of the drop-size
distribution
Experimental set-up
It was the aim of a first set of experiments to produce a
chain of droplets with fluid and flow properties such
that all droplets would spray when impinging on flat
surfaces. As mentioned in Section 3, an FMP
Technology droplet generator was employed for the
production of the required droplet chain. The piezo
oscillator of this generator was driven by an electrical
drive followed by an amplifier. To control the mono-
disperse droplet production, a CCD camera and a
stroboscope LED were used. The camera was con-
nected to a PC on which the acquired picture of the
monodisperse droplet chain could be observed. The
camera and the LED frequency were synchronized
with the frequency of the piezo drive, that is, the fre-
quency of drop production.
The plate on which the drop chain impinged was a
flat pin with a diameter of 1 mm. The distance between
the pin and the nozzle outlet could be adjusted and
varied between 50 and 100 mm during the experiment
in order to achieve a monodisperse droplet chain.
The liquid used in all experiments was water, sup-
plied by a vessel allowing a maximum fluid driving
pressure of 30 bar. A filter was used to avoid any
nozzle blockage. The mass flow rate of the fluid to
the nozzle could be adjusted by the air pressure in the
vessel. To achieve repeatable results that are independ-
ent of the fluid supply system, the mass flow rate was
measured by weighing the amount of water flowing out
in a definite time period. In this way, the flow rate was
calculated as a function of the supply pressure.
The measurement of the droplet size distribution was
carried out by the laser diffraction method. Figure 12
shows a sketch of the applied experimental set-up. The
device used was a Malvern Mastersizer X, which uti-
lized a HeNe laser with a wavelength of 633 nm and a
beam diameter of 18 mm in the measuring region. For
the droplet size measurements, the light of this beam
was diffracted by the droplets whereupon the magni-
tude of diffraction depended on the drop diameter.
Each of the 33 photo detectors of the Malvern
Mastersizer X (Malvern Instruments Limited,
Malvern Worcestershire, UK), mounted at different
radial positions, measured the light intensity. From
these values the drop-size distribution was obtained.
The lenses used in the experiments had focal lengths
of 100 and 300 mm, leading to drop size measurements
in the range 0.5–180 and 1.2–600 m, respectively. The
spray was measured 10 times with 2000 droplet signals
per measurement with a frequency of 445Hz to obtain a
mean distribution with a low residual. For each of these
mean distributions the spray characteristics (introduced
in the next section) were calculated. All spray charac-
teristics given in this paper are averaged values of these
10 measurements. The sample area is located in the
plane of the spray origin (the surface on the top of
the pin) at radial distances s ranging from 50 to 100
mm (see also Figure 12).
Results
In Figure 13, the experimental results for the normal-
ized Sauter mean diameter
d
32
=
D
are plotted with respect
to Weber number. The latter was computed using the
jet velocity at the nozzle outlet deduced from the con-
tinuity equation; see Section 4.2. As Figure 13 shows,
the Sauter mean diameter decreased with increasing
Weber number for all nozzles. The highest Weber
number measured with a nozzle diameter of 50 mmis
approximately 1700 for the maximum driving pressure
of 30 bar employed in this study. For Weber numbers
above 2000, the change in the dimensionless Sauter
mean diameter with increasing Weber number is very
low for all data. As Figure 13 suggests, for higher
Figure 11. Experimental results of measurement of the critical
Weber number at different dimensionless hemisphere diameters.
8 International Journal of Spray and Combustion Dynamics 0(0)
pressures the 100 mm nozzle produces smaller drops
than the 150 mm nozzle. The reason might be that for
high jet velocities, the diameter of the pin had an influ-
ence on the atomization behavior. The pin was the
same for all nozzles investigated, leading to different
ratios of pin to jet diameter.
In Figure 13, the function
d
32
D
¼ We
0:3
ð12Þ
is also plotted. With this empirical relation, the experi-
mental results can be approximated in the Weber
number range 200–4600.
The spray characteristics D
10
, D
50
and D
90
are
shown in Figure 14. About 10% of the total mass of
the spray has a drop diameter below D
10
. For D
50
and
D
90
, the masses of the spray below the limiting values
are 50% and 90%. All three spray characteristics
decrease with increasing Weber number.
For each tested operating point, 10 drop size dis-
tributions were obtained. In Figure 15 two exem-
plary distributions are shown for a 100 mm nozzle
at different supply pressures, hence different Weber
numbers. The ordinate represents the volume frac-
tion with respect to the drop diameter. The spray
characteristics are given in the caption. Both distri-
butions possess two peaks at different drop diam-
eters. One major peak is located at a big drop
diameter and a minor peak indicates smaller drop
diameters of the spray. It can be seen that, with
increasing Weber number and accordingly increasing
momentum the major peak shrinks while the minor
peak grows and additionally shifts to lower drop
diameters. The location of the major peak on the
abscissa is nearly unaffected by the momentum
change. Nevertheless, the Sauter mean diameter is
reduced by a factor of approximately two, which is
unapparent from both distributions since the
abscissa has logarithmic scaling.
Obviously there are two atomization mechanisms
producing different drop diameters. By changing the
conditions of the spray (e.g. supply pressure) the ratio
of the proportion of drops produced by each mechan-
ism also changes. Consequently the averaged spray
characteristics vary, too.
Comparisons with other single fluid
nozzle generated sprays
The obtained spray properties of the presented
atomizer can be compared with those of other
Figure 12. Experimental set-up for the measurement of the
drop-size distribution for the spray generated by an impinging
droplet chain.
Figure 14. Normalized spray characteristics (D
10
, D
50
and D
90
)
with respect to Weber number for atomization of an impinging
droplet chain with a drop diameter of d ¼2D.
Figure 13. Normalized Sauter mean diameter with respect to
Weber number for atomization of an impinging droplet chain
with a drop diameter of d ¼2D.
Etzold et al. 9
available spray generators such as the hollow-cone
nozzle, the twin-jet nozzle and the plain-orifice atom-
izer. The hollow-cone nozzle was chosen since it is a
very common spray generator for the production of
very fine sprays. The twin-jet nozzle produces a
spray by the impingement of two liquid jets
and is used as an alternative injector to the
single-jet injector, which is also taken into account
for the comparison with the proposed new spray
generator.
Since the presented spray generator is built for low
flow rates, the lower limits of the different atomizing
systems are shown together in one Ohnesorge–
Reynolds diagram, see Figure 16. The lower limit for
the atomization with a single-jet injector was given by
Borman and Ragland.
17
Sazhin et al.
18
gave an equa-
tion to approximate the limiting values in terms of
Ohnesorge and Reynolds numbers
log Oh 4 3 1:2 log Re ð13Þ
The limiting value for the hollow-cone nozzle was given
by Walzel
5
We 4 1700 ð14Þ
Two impinging jets always form a spray. Hence the
limiting value for twin-jet atomization was the jetting
criterion, which is given by, for example, Walzel
6
(see
equation (6)). For the investigated spray-generating
principle in this paper, drops impinging on plates,
Mundo et al.
9
gave a corresponding relation (see equa-
tion (9)).
Figure 16 illustrates that the single-jet injector is not
suitable for low flow rates. The twin-jet nozzle starts
the production of a spray at lower Reynolds and
Ohnesorge numbers. The hollow-cone nozzle produces
a good spray at low Reynolds and Ohnesorge numbers.
However, the atomization of an impinging droplet
chain on a flat plate starts at very low velocities and
at low flow rates.
The big advantage of the novel spray generator
described in this paper is that it is able to provide rela-
tively good atomization properties and low flow rates
can be obtained at the same time. This becomes clearer
through the information provided by Figure 17, where
the Sauter mean diameter is plotted with respect to
volume flow rate of the different spray systems. The
nozzle diameter for this comparison was chosen as
D ¼100 m for every atomizer and the liquid proper-
ties of water were used to compare the Ohnesorge num-
bers (Oh ¼0.012). The Sauter mean diameter of a
(a) (b)
Figure 15. Drop size distributions for a nozzle diameter of 100 m at two operating points:
a) We ¼592, d
32
¼15 m, D
10
¼10 m, D
50
¼19 m, D
90
¼33 m;
b) We ¼3154, d
32
¼7 m, D
10
¼5 m, D
50
¼20 m, D
90
¼39 m.
Figure 16. Lower limits of different atomizers: (a) drop
impinging on plate, equation (9); (b) hollow-cone nozzle, equa-
tion (14); (c) twin-jet nozzle, equation (6); (d) plain-orifice nozzle,
equation (13).
10 International Journal of Spray and Combustion Dynamics 0(0)
single-jet injector can be approximated with an equa-
tion given by Elkotb
19
d
32
¼ 3:08
0:385
0:737
0:737
P
0:54
ð15Þ
For the computation of the Sauter mean diameter for
hollow-cone nozzles, equation 3 of Walzel
5
was used
with a constant lamella number of ¼0.27. The flow
rate was computed with equation 1 with set to 0.85.
The normalized Sauter mean diameter of the twin-jet
spray was calculated using an equation given by
Han et al.
20
d
32
D
¼ 23:2Oh
0:1
We
0:535
K
d
2
1 þ 3Oh½
1
6
ð16Þ
The sheet thickness parameter K was set to 2.75 10
8
,
which represents a nozzle diameter of D ¼100 m and
an impinging angle of the two jets of 40
. The
Ohnesorge number was set to Oh ¼0.012 for both the
hollow-cone and the twin-jet nozzle, since a 100 m
nozzle with water as liquid was considered. For the
Sauter mean diameter of the novel spray generator,
the correlation d
32
D ¼ We
0:3
was used. The ratio of
total nozzle flow rate to aerosol flow rate
_
V
S
avr
=
_
V
T
was set to 0.5, which is a rough value measured in
experiments.
In Figure 17, it can be seen that the single-jet and
twin-jet injectors produce the largest drop sizes. The
hollow-cone nozzle is able to produce a very fine
spray but its lowest flow rate is approximately 14 ml/min.
The atomization of impinging drops produces drops at
lower flow rates. This principle also has a limiting value
at approximately 3 ml/min with large drops (17 m). By
using the novel approach described this paper, the
switching between an impinging jet and an impinging
droplet chain, the operating point with a Sauter mean
diameter of 11 m can be used for varying the flow rate
from 0 to 9 ml/min, since at these flow rates no atom-
ization in the case of an impinging jet occurs. If the
aerosol flow has to be further increased, the alternating
mode is not necessary. The nozzle flow rate can be
adjusted to the desired value.
Conclusions, final remarks and outlook
In this paper, a new spray generator, suitable for low
flow rates, is introduced. Its main working principle is
the atomization due to the impingement of drops on a
plate. The drops are produced by Rayleigh breakup of
a liquid jet, yielding monodisperse drops. By switching
between the states of drop and jet production, the spray
flow rate can be controlled and modulated, since there
is no atomization in the state of jet production.
Measurements of a liquid jet impinging on a hemi-
sphere-shaped pin have shown which ratio of hemi-
sphere to jet diameter is applicable for the working
principle of the proposed spray generator.
Drop-size measurements were carried out. The
results revealed that the spray-generating principle
chosen for the work described in this paper produces
smaller drops than single-jet and twin-jet injectors, but
slightly larger drops than hollow-cone nozzles. The
advantages of the proposed spray generator, compared
with other spray nozzles, are its high volume flow
modulation range and its ability to produce fine
sprays at very low aerosol flow rates.
The shape of the plate on which the drops impinge
has to be further investigated. On the one hand, the
back flow rate
_
V
B
has to be minimized to achieve a
convenient efficiency regarding the ratio of total
nozzle flow rate to spray flow rate
_
V
T
=
_
V
S
. On the
other hand, the jet stabilization due to the Coanda
effect should be possible at high flow rates in order to
produce small drop diameters. This issue has to be
resolved in the future by experimental and theoretical
considerations.
The practical application of the spray generator also
has to be investigated. Since the spray propagation is in
a radial direction, it will be very difficult to use it for a
diffusion flame. Nevertheless, the spray generator could
be mounted within a chamber with an additional air
supply producing a mixture of air and fuel. The back
flow can be spilled back to the oil reservoir and the air
Figure 17. Sauter mean diameter with respect to volume flow
rate of different atomizer systems (for atomization of water with
D ¼100 m, Oh ¼0.012): (a) plain-orifice nozzle, equation (15);
(b) twin-jet nozzle, equation (16); (c) hollow-cone nozzle,
equation (3); (d) drops impinging on plate (dashed line);
(e) novel spray generator presented in this paper (solid line),
equation (12).
Etzold et al. 11
fuel mixture can be conveyed to a combustion zone.
This application of the novel spray generator could be
combined with a porous medium burner, since these
burners provide a high volume flow modulation rate,
like the invented spray generator.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship and/or publication of this
article.
Funding
The author(s) received no financial support for the research,
authorship, and/or publication of this article.
References
1. Lefebvre AH. Atomization and sprays. Boca Raton, FL:
CRC press, 1989.
2. Walzel P. Spraying and atomizing of liquids. Ullmann’s
encyclopedia of industrial chemistry. Weinheim, Germany:
Wiley-VCH Verlag GmbH & Co., 2012.
3. Musemic E and Walzel P. Durchsatzverhalten von
Hohlkegeldu
¨
sen. Chem Ing Tech 2011; 83: 237–246.
4. Lefebvre AH and Wang XF. Mean drop sizes from pres-
sure-swirl nozzles. J Propul Power 1987; 3: 11–18.
5. Walzel P. Auslegung von Einstoff-Druckdu
¨
sen. Chem Ing
Tech 1982; 54: 313–328.
6. Walzel P. Zersta
¨
uben von Flu
¨
ssigkeiten. Chem Ing Tech
1990; 62: 983–994.
7. Truckenbrodt E. Fluidmechanik: Bd. 2: Elementare
Stro
¨
mungsvorga
¨
nge dichtvera
¨
nderlicher Fluide sowie
Potential-und Grenzschichtstro
¨
mungen. Berlin: Springer
Verlag, 1980.
8. Wille R and Fernholz H. Report on the first European
Mechanics Colloquium, on the Coanda effect. J Fluid
Mech 1965; 23: 801–819.
9. Mundo CHR, Sommerfeld M and Tropea C. Droplet-
wall collisions: experimental studies of the deformation
and breakup process. Int J Multiphase Flow 1995; 21:
151–173.
10. Van de Sande E and Smith JM. Jet break-up and air
entrainment by low velocity turbulent water jets. Chem
Eng Sci 1976; 31: 219–224.
11. Rayleigh L. On the capillary phenomena of jets. Proc R
Soc London 1879; 29: 71–97.
12. Strutt JW and Rayleigh L. On the instability of jets. Proc
London Math Soc 1879; 10: 4–13.
13. Brenn G, Durst F and Tropea C. Monodisperse sprays
for various purposes – their production and characteris-
tics. Part Part Syst Char 1996; 13: 179–185.
14. Brenn G, Helpio
¨
T and Durst F. A new apparatus for the
production of monodisperse sprays at high flow rates.
Chem Eng Sci 1997; 52: 237–244.
15. Schmidt P and Knauss G. Prallzersta
¨
ubung von
Flu
¨
ssigkeiten bei Nichtbenetzung. Chem Ing Tech 1976;
48: 659–659.
16. Walzel P. Zerteilgrenze beim Tropfenprall. Chem Ing
Tech 1980; 52: 338–339.
17. Borman GL and Ragland KW. Combustion engineering.
New York: McGraw-Hill, 1998.
18. Sazhin SS, Feng G and Heikal MR. A model for fuel
spray penetration. Fuel 2001; 80: 2171–2180.
19. Elkotb MM. Fuel atomization for spray modelling.
Prog
Energy Combust Sci 1982; 8: 61–91.
20. Han Y, Durst F and Zeilmann M. High-pressure-driven
twin-jet sprays and their properties. Atomization Sprays
2014; 24: 375–401.
12 International Journal of Spray and Combustion Dynamics 0(0)
