International Journal of Multiphase Flow 28 (2002) 773789

www.elsevier.com/locate/ijmulow

Liquid lm disintegration regimes and proposed correlations

a,*
I.S. Carvalho , M.V. Heitoyr b, D. Santos c

b
Department of Mechanical Engineering, Instituto Superior de Engenharia de Lisboa,
Rua Conselheiro Emidio Navarro 1, 1949-014 Lisbon, Portugal
b
Department of Mechanical Engineering, Instituto Superior T
ecnico, Technical University of Lisbon,
1049-001 Lisbon, Portugal
c
PROETProjectos, Engenharia e Tecnologia S.A, Grupo EDP, 1749-061 Lisbon, Portugal
Received 7 March 2000; received in revised form 30 November 2001

Abstract
The process of liquid sheet disintegration and break-up into a spray is studied making use of a at liquid
lm surrounded by two air streams. The work considers a 2-D liquid lm with an aspect ratio of 114 and a
liquid velocity between 0.7 and 6.4 m/s (543 6 ReL 6 4963) and an air velocity up to 39 m/s (ReG 6 17 647),
and allows to study the basic phenomena typical of airblast atomisation. The results include the quanti-
cation of instability amplitudes, break-up lengths and frequencies for the above-mentioned range of liquid
(water) and air ow velocities. Three dierent experimental techniques are used: back light and laser light
sheet illumination, to measure the instability amplitudes, which leads to the liquid lm disintegration and
the spray formation; strobe light illumination, for the quantication of either break-up lengths and fre-
quencies, and; a laser attenuation technique for frequency measurements and comparison with the visu-
alisation technique. The results quantify the range of application of the dierent techniques, and include an
analysis of the liquid disintegration process as function of the liquid velocity when exposed to dierent air
ow velocities, along with the quantication of break-up lengths, disintegration frequencies and spray
angles, and show that the disintegration of the liquid sheet is associated with a periodic process, which is
mainly dependent on the absolute air velocity and the airliquid momentum ratio. Correlations are pro-
posed for liquid sheet break-up length and frequency based on the air-to-liquid momentum ratio. 2002
Elsevier Science Ltd. All rights reserved.

Keywords: Liquid sheet disintegration; Visualisation; Two-phase ows; Laser attenuation technique; Frequency
analysis; Break-up lengths

*
Corresponding author. Tel.: +351-1831-7013; fax: +351-1831-7213.
E-mail address: [email protected] (I.S. Carvalho).

0301-9322/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 3 0 1 - 9 3 2 2 ( 0 1 ) 0 0 0 8 8 - X
774 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789

1. Introduction

The basic mechanisms involved in airblast atomisation have been studied for a number of years
making use of planar liquid lms and major achievements have been reported since Fraser et al.
(1963) and Dombrowski and Johns (1963), namely by Arai and Hashimoto (1985), Mansour and
Chigier (1990, 1991), Eroglu and Chigier (1991), Stapper et al. (1992), Lozano et al. (1996),
Carentz et al. (1998) and Hardalupas et al. (1998), among others. These studies were conducted in
quiescent ambient and/or between two co-owing air streams. A summary of some of the most
relevant studies performed in 2-D liquid lms is given in Table 1. The main contribution/
advantage of the present conguration is related with the possibility of on-line visual control of
both liquid and air streams. This is of great importance, because it allows the visualisation of the
liquid ow inside the nozzle (which is transparent), and air bubble formation can be visualised and
appropriate ow regulation measurements can be performed to eliminate them. This fact also
allows the minimisation of other overlapping phenomena (due to air bubble formation in the
liquid) that can be misleading in the interpretation of the results for the process under study.
Analysis of the liquid disintegration process has been based on the aerodynamic interaction
between liquid and air, which gives rise to the formation of waves (dilatational or/and sinusoidal)
on the liquid surface, that grow up to a critical point, as illustrated in Fig. 1. Then, ligaments and/
or liquid clusters separate from the liquid sheet, which subsequently fragment into droplets
(Dombrowski and Johns, 1963). Those waves are characterised by a wave length, a wave am-
plitude rate and a frequency, which depend on the working conditions (Eroglu and Chigier, 1991;
Chigier and Dumouchel, 1996). Fig. 1 illustrates these mechanisms, based upon the theoretical
and experimental study of Hagerty and Shea (1955). These authors show that only two kind of
perturbations may exist that can lead to the disintegration of the liquid sheet, namely: (i) a si-
nusoidal or non-symmetric perturbation (see Fig. 1a), where the two liquidair interfaces oscillate
in phase, and; (ii) a dilational or symmetric perturbation, characterised by not in phase oscilla-
tions. Hagerty and Shea (1955) also observed that the growth rate of the sinusoidal instabilities is
always higher in comparison to that of the dilational instabilities. In consequence, more eort
have been put into the study of characteristics of the sinusoidal perturbations (e.g., Chigier and
Dumouchel, 1996).
Arai and Hashimoto (1985) studied a planar liquid lm and report similar results for the break-
up frequency, making use of strobe light illumination, and the liquid lm oscillation frequency,
which was measured with a laser dispersion technique. The oscillating frequency of a planar liquid
lm was later analysed by Mansour and Chigier (1991) and Lozano et al. (1996), for dierent
geometries and working conditions. All these studies showed that the disintegration frequency
slightly increases with the liquid velocity, but strongly increases with the air velocity. The resulting
wave amplitudes and spray angles were particularly studied by Eroglu and Chigier (1991),
Mansour and Chigier (1991) and Lozano et al. (1996) using several visualisation techniques,
making comparisons dicult to achieve, due to the dierent geometrical congurations used (see
Table 1). Moreover, streamwise vorticity was shown to considerably inuence the break-up
process (e.g., Lozano et al., 1996; Chigier and Dumouchel, 1996), but its eect on the disinte-
gration process was not quantied in a systematic way.
Although several studies have recognised the importance of non-dimensional groups, such as
the Reynolds and Weber numbers, ow phenomena inside the nozzle, turbulence of the liquid
 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789 775

Table 1
Summary of the main ow congurations studied
Reference L (mm) tL (mm) UL (m/s) mL (g/s) tG (mm) UG (m/s) mG (g/s) Comments



Rizk and 50 0.089 15 4.522.7 5.505.65 55120 18.6 Studied thickness
Lefebvre (1980) 0.400 40.68


Arai and 200 0.41.0 0.52.0 39.9

2367 Estimated values
Hashimoto (1985) 399.2



Not available
Mansour and 30.6 0.254
1.58.0 1163 1.450
0145 07.72

Studied thickness
Chigier (1990) 1.450 2.210
Mansour and 30.6 0.254
0.81 6.31 1.450
17.73 1.89

Studied thickness
Chigier (1991) 1.450 16.23 126.18 2.210 119.62 12.73



Beck et al. (1991) 0.102; 0.66
4.9 1025

1360 3.0
13.8

Estimated values



0.203 e Similar to Rizk
0.305 and Lefebvre (1980)
Eroglu and 30.48 0.254 1.616.1 12.5125 1.45 17.774.7 Similar to Mansour
Chigier (1991) and Chigier (1991)
experimental set up
Stapper et al. 47 0.508 15 23.9 3.7
060 0126



Estimated values



(1992) 119.4 Maximum possi-
ble value
Lozano et al. 80 0.95 0.22.45 15186 10 865 862.4
(1996)
The present case 80 0.7 0.66.4 30.6 7 040 036 Transparent nozzle
359.0 Independent air ve-
locity control
L: liquid lm length, UL : liquid lm velocity at injector tip, UG : air stream mean velocity, tL : liquid lm thickness, tG : air
lm thickness, mL : liquid mass ow rate, mG : air mass ow rate.

lm, the balance of axial and radial components of the liquid velocity, existence of primary
disturbances and/or order of magnitude of relative velocities (Dombrowski and Johns, 1963; Lee,
1985; Mansour and Chigier, 1991; Stapper et al., 1992), there is no clear explanation of the
dominant cause for liquid disintegration. It is, nevertheless, recognised that all these mechanisms
are connected with wave formation, wave development and wave disintegration, the break-up
length being mainly dependent on the growth rate of the wave itself. The wave amplitude increases
with the distance from the nozzle, but the spatial amplitude growth rate decreases with liquid mass
ow rate and increases with the air velocity (Eroglu and Chigier, 1991; Chigier and Dumouchel,
1996). High spatial amplitude growth rates occur for low liquid mass ow rates. The disturbances
are known to be amplied faster when the aerodynamic force increases at the air/liquid interface
due to higher relative velocities.
Although a range of tests in axisymmetric geometries have been reported in literature (Adzic
et al., 1994; Carvalho and Heitor, 1997; Karl et al., 1996), the particular complexity of the airblast
geometry requires fundamental studies on planar liquid sheets. In this context, the present paper is
aimed to improve on understanding of the periodic nature of the liquid break-up process and
assess the use of optical diagnostics for this purpose.
The present work involves the analysis of the ow downstream a 2-D planar liquid lm sur-
rounded by two independently controlled air ows, and follows the previous reported work
776 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789

Fig. 1. Instability types that can exist in a liquid lm disintegration process.

(Carvalho et al., 1997a,b, 1998). The ultimate objective is to improve knowledge of the liquid
break-up mechanisms and their relation with the atomisation quality in airblast nozzles of
practical relevance. In this context, this paper also aims to discuss and compare two dierent
experimental techniques for the frequency analysis of the disintegration process, and two visu-
alisation techniques for wave amplitude and spray angle measurements.
Section 2 describes the ow conguration and the experimental techniques used throughout
the work. Section 3 presents and discusses the results and the main ndings are summarised in
Section 4.

2. Experimental method: liquid lm generator and instrumentation

Fig. 2 shows a schematic diagram of the liquid lm generator along with the instrumentation
used. The 2-D liquid lm generator consists of: (i) an inner liquid ow, with an exit thickness of
 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789 777

Fig. 2. Schematic diagram of the at liquid lm generator and experimental techniques conguration.

tL 0:7 mm, and an aspect ratio of L=tL 114, that is, 80 mm wide; and, (ii) two co-current air
ows, with a thickness tG 7 mm, which are passed along both sides of the liquid lm to produce
a shear force at the airliquid interface. Air can be fed through the left and right channels sep-
arately, thus allowing the air velocity ratio to be varied between 1 and 4, with absolute average
velocities up to 39 m/s (ReG qG UG tG =lG 6 17 647). The impingement angle of the air ows
towards the liquid lm is 30. Water was used as the test liquid, and the results reported here are
for a liquid velocity between 0.7 and 6.4 m/s (543 6 ReL qL UL tL =lL 6 4963).
Dierent visualisation techniques were used throughout the work, which include laser light
sheet illumination obtained by spreading a laser beam with a cylindrical lens, with individual
images acquired by a 35 mm camera operating with 400 ASA lms, and exposure time varying
between 4 and 66 ms. In addition, strobe light visualisation was used and ow images were ac-
quired by a CCD video camera, each frame corresponding to an individual image of the ow, with
an exposure time of 1.2 ls. These techniques allowed the quantication of spray angles, wave
amplitudes, and break-up lengths and frequencies. The break-up frequency of the liquid sheet was
analysed by matching the strobe light frequency, fs , with that of the ow, following the work of
Arai and Hashimoto (1985). These authors studied the disintegration of a liquid lm, measuring
break-up lengths and wave amplitudes by means of photography, and break-up frequency and
oscillation frequency using strobe light matching and laser light attenuation techniques, respec-
tively. Their study made available two empirical correlations: one for the break-up length and
another for the break-up frequency.
778 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789

Frequency measurements were also performed by means of a laser light beam (HeNe laser,
30 mW), positioned parallel to the ow in a 180 arrangement with a photodiode which generates
a current proportional to the incident light intensity, which varies as the liquid lm crosses the
laser light beam, and causes its attenuation (Arai and Hashimoto, 1985; Mansour and Chigier,
1991). For the results presented here, the system is located in the break-up region, previously
assessed by the strobe light visualisation technique. The current of the photodiode is converted
into voltage, which was in turn fed to a digital oscilloscope to visualise the waveform, and si-
multaneously transferred to a PC, and a Fourier analysis was performed to determine the
dominant frequencies. In each point 10 series of 16 304 points were acquired, with an acquisition
rate of 20 kHz, and individually submitted to a Fourier analysis (Fernandes and Heitor, 1997),
which gives us the mean power spectra, where dominant frequencies (higher energy) plus its
harmonics can be identied. As previously reported by Arai and Hashimoto (1985), the break-up
frequency was seen to be coincident with the oscillating frequency of the liquid lm, prior to
disintegration.
Frequency measurements show that the frequency is not dependent on either the axial or radial
measurement locations. In the presence of the ow, the SNR is good. The control volume position
inside the ow does not aect the dominant frequency, as previously observed by Mansour and
Chigier (1991). The SNR is dependent on the control volume position, in a way that it is minimum
in the plane of the liquid lm (higher attenuation) and maximum in the outer regions (1 mm o
axis).

3. Results and discussion

The results of the work discussed in this paper are presented in three parts, namely: (i) spray
angles; (ii) break-up frequencies; and, (iii) break-up lengths of the liquid sheet. First, for a better
overall evaluation of the referred ows, extended visualisation results will be shown, both front
and side views, for dierent experimental techniques, namely, strobe light illumination and laser
light sheet visualisation.

3.1. Visualisation results

Fig. 3 presents sample results of two dierent views of the liquid lm for dierent air mass ow
rates, and UL 0:7, 1.3, 1.6, 1.8, 2.1, 2.5, 2.9, 3.4, 4.3 and 6.4 m/s. The air velocity ratio is kept
constant and equal to unity, and the absolute air velocities are UG 0, 15, 20, 30 and 39 m/s.
Fig. 3a presents front views of the at liquid lm obtained in the absence of air (UG 0), and
shows a convergent liquid sheet bounded by thick rims that are drawn together by surface tension
forces, as rst described by Mansour and Chigier (1990). The convergence length, Lc , increases
with the liquid lm velocity, UL . Although the present results must be extrapolated with care, due
to their dependence on the geometry of the sheet generator, they show a clear linear increase of the
convergence length as the liquid ow rate increases, which is in qualitative agreement with the
results of Mansour and Chigier (1990). For the present case it is veried that Lc 78:6UL 11:5,
with a correlation coecient of 0.999. Representing the results in a non-dimensional form, using
Lc =tL and ReL tL UL =mL as dimensionless groups for Lc and UL the results of both studies follow
 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789 779

the curve Lc =tL 0:15ReL 6:72, with a correlation coecient of 0.996. The central part of the
lm keeps a smooth appearance for low liquid velocities, UL . Small ripples start to appear in the
centre of the liquid lm for UL P 2:1 m/s (ReL 1628). As the liquid mass ow rate increases, this
eect is more pronounced and at UL 2:9 m/s (ReL 2249) the liquid surface becomes clearly
disturbed over all its extension, following again Mansour and Chigier (1990). These disturbances
are damped on the downstream region, because turbulent energy production ceases at the nozzle
exit and the liquid sheet tends to laminarize as a result of viscous dissipation of turbulent energy.
In Fig. 3b and c (which provide front and side views), the liquid lm behaviour, when sur-
rounded by two air ows, can be observed. The characteristic of these ows resembles those

Fig. 3. Visualisation of the at liquid lm for dierent liquid and air velocities: (a) front view of the liquid lm in the
absence of air, using strobe light illumination, (b) front view and side view of the liquid lm for UG 15 m/s and
UG 39 m/s, using strobe light illumination, (c) side view of the liquid lm for 15 6 UG m=s <39, using laser light
sheet illumination.
780 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789

Fig. 3. (continued)
 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789 781

obtained by Mansour and Chigier (1991) and Lozano et al. (1996), and we can see the cellular type
of structures (well organised), and the streamwise ligaments which are formed very close to the
nozzle exit. All the results are in agreement with previously obtained ones, although attention
should be drawn to the fact that the geometrical conguration plays an important role in this type
of phenomena, aecting the mixing process between the two uids, as in the present conguration
the atomising air has a radial velocity component towards the liquid lm, which as a consequence,
enhances the interaction between the two uids, as previously referred by Lefebvre (1992). This is
one of the reasons why comparison and extrapolation of the present results and the ones obtained
in previous studies reported in literature, must be carefully performed.
In the disintegration of the liquid lm, the formation of a surface wave is one of the recognised
important factors contributing to the break-up phenomena. The wave formation may be caused
by KelvinHelmholtz instability due to the liquid/air interaction (Farag o and Chigier, 1992).
Disintegration occurs when the wave amplitude reaches a critical value, as previously described by
Dombrowski and Johns (1963), and fragments of the liquid are torn o from the outer layers of
the liquid lm.
As the liquid velocity decreases, the sheet gives rise to streamwise ligaments, which are formed
very close to the nozzle. The analysis shows that a spanwise wave is superimposed on top of the
streamwise ligaments, which grow together and, as a result, give rise to the type of distorted waves
described by Mansour and Chigier (1991) downstream of the initial sinusoidal wave. The present
liquid lm behaviour is consistent with the results from Arai and Hashimoto (1985), Lee and
Wang (1986) and Eroglu and Chigier (1991).
The presently studied atomisation process is mainly driven by the energy transfer from the air
streams to the liquid lm. The present work shows that the amount of energy transferred from the
air to the liquid has a critical value for which the atomisation quality suers an abrupt increase
(referred to as transition points). A parameter that allows the characterisation and prediction of
the transition points of the liquid lm disintegration mode is the air-to-liquid momentum ratio,
dened as MR qG UG2 =qL UL2 .
For low air-to-liquid momentum ratio the dilatational wave dominates the liquid lm disin-
tegration mode. The atomisation quality is rather poor and a narrow spray angle is obtained (e.g.,
UG 15 m/s and UL P 1:8 m/s). For higher air-to-liquid momentum ratios, the sinusoidal waves
dominate, and the atomisation quality is considerable improved, as the spray angle increases
signicantly (e.g. UG 39 m/s and UL 6 3:9 m/s). The above-mentioned working conditions dene
the transition ows, and the air-to-liquid momentum ratio value that characterises this tran-
sition is about 0.5. For considerably higher air-to-liquid momentum ratios, namely with low
liquid velocities or high air velocities (e.g., UG 15 m/s and UL 6 0:7 m/s; or UG 39 m/s and
UL 6 2:9 m/s), the liquid lm emerging from the nozzle is rapidly torn into small fragments by the
immediate interaction between the liquid and the impinging air streams, as already suggested by
Lefebvre (1992). For considerably low air-to-liquid momentum ratios, namely for high liquid
velocities or low air velocities (e.g., UG 15 m/s and UL P 4:3 m/s) no wave was seen to be present
in the liquid surface.
The analysis of all the visualisation results, plus the quantication of the transition disin-
tegration processes, is also observed by the quantication of spray angles, frequencies and break-
up lengths, as discussed in the following section, which gives further insight into some of the
studied ows, and provides empirical correlations for frequency and break-up lengths.
782 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789

3.2. Spray angles, frequencies and break-up lengths measurements

3.2.1. Instability amplitudes and spray angles

The spray angle measurement is a form of evaluating the amplitude of the liquid sheet insta-
bility, as the former is closely dependent on the latter. The critical amplitude of the instabilities for
which the break-up occurs determines the angle of the spray and several techniques can be used to
measure these characteristics (e.g., Mansour and Chigier, 1990, 1991; Lozano et al., 1996). In the
present work two dierent visualisation techniques are used: laser light sheet illumination and
strobe light illumination.
Fig. 4a shows the results obtained for UG 15 and 30 m/s with laser light sheet and back light
illumination, with a strobe light, for UL 1:3 m/s. In the case of the laser light sheet illumination,
the spray angle is measured making use of a visually observed average value of contrast between
the white spray and the black background. For the back light illumination the criteria was to
follow the crests of the major disturbances. The surrounding droplet mist was not generally
considered. In both cases the uncertainty of the measurements is estimated to be 5%.
Fig. 4b shows the results obtained with both techniques for UG 30 m/s. It is observed that the
results obtained with the laser light sheet are 40% higher, when compared with the back light
illumination technique. This is related with the exposure time used, which was about 33 ms for the
laser light sheet, and between 1.2 and 0:8  103 ms for the back light illumination. In addition,
the dierences observed are also related to the droplet mist that surrounds the wave crests (see
Fig. 4a1 and a3), which is not considered for spray angle measurements in the back light illu-
mination, and is obviously accounted for with the laser light sheet technique. When comparing
both techniques, it is observed that the values obtained with the laser light sheet a more accurate
quantication of the spray angle, and that the back light illumination technique is more adequate
for instability amplitude measurements. This is due to the fact that, spray angle measurements
require some level of temporal integration, as those provided by the laser light sheet technique.
Fig. 4c shows measured spray angles making use of the laser light sheet technique. For all
operating conditions, regardless the air velocity, a region of maximum spray angle occurs, fol-
lowed by a sharp decrease for higher values of liquid velocity. The maximum values of the spray
angle (e.g., for UG 15 m/s the maximum angle is observed at UL 1:3 m/s) are displaced for
higher values of liquid velocity as the air ow rate increases (e.g., for UG 30 m/s the maximum
angle is observed at UL 1:8 m/s), the maximum value decreasing with air ow rate, as easily seen
in Fig. 3c and d. For example, it was observed that for UG 15 m/s the region corresponding to
the sharp decrease of the spray angle (i.e., the transition ow) appears for a specic liquid
velocity value, as illustrated in Fig. 2d. Compare, for example, in Fig. 3d, at UG 15 m/s the
characteristics of the ows with UL 1:3 and 2.1 m/s. The transition ow is characterised by a
spray angle oscillation in a range of about 1520. For a small increase in the liquid velocity, very
poor atomisation is obtained and the spray angle assumes very small values. Also, for higher air
ow velocities the spray angle decrease becomes less sharp, when the liquid velocity increases (see,
e.g., sequences for UG 15 and 30 m/s in Fig. 3c and d).

3.2.2. Break-up frequencies

For break-up frequency measurements two dierent techniques are used: strobe light with
variable frequency illumination and laser light attenuation. Fig. 5 shows results for the two
 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789 783

Fig. 4. Spray angles as function of liquid velocity: (a1a4) Side views of the liquid lm using strobe light and laser light
sheet illumination, (b) spray angles as function of liquid velocity for UG 30 m/s and for the two dierent techniques,
(c) spray angles as function of liquid lm for 15 6 UG m=s < 39, using laser light sheet visualisation.

techniques for UG 15 m/s, along with three typical power spectrum for dierent liquid velocities
(see Fig. 5bd), together with the electronic noise, which can be observed in the loglinear rep-
resentation, in Fig. 5c. A xed noise frequency at 286 Hz, due to the acquisition board itself, is
always present. The peak corresponding to that frequency does not aect the frequency analysis
presented here. In Fig. 5a the power spectrum for UG 15 m/s and UL 0:7 m/s is shown in a
linear scale. This gure clearly shows the fundamental frequency at f 119 Hz, by its maximum
power. In Fig. 5c the fundamental frequency for UL 1:5 m/s can be identied at f 138 Hz.
Additionally, the loglinear representation of the power spectrum allows the identication of the
784 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789

Fig. 5. Break-up frequency of the liquid sheet: (a) as a function of the liquid velocity for UG 15 m/s, (b) corre-
sponding power spectrum obtained for UL 0:7 m/s, (c) corresponding power spectrum obtained for UL 1:5 m/s, (d)
corresponding power spectrum obtained for UL 2:2 m/s.

harmonic frequencies fn nf f2 276 Hz, f3 409 Hz and f4 555 Hz. The frequencies below
the fundamental one correspond to the electrical net noise at 50 Hz and to a subharmonic.
The axial position of the control volume determines the meaning of the measured frequency.
If it is positioned upstream of the break-up region, an oscillating frequency is measured, while the
results of the break-up frequency are obtained when the control volume is located downstream of
that region. As the fundamental frequency does not depend on the axial position, it is assumed
that the break-up and the oscillating frequencies are equal. This conclusion is also supported by
the comparison between the results obtained with the dierent frequency measuring techniques, as
shown in Fig. 5a. As referred before, the strobe light technique allows the freezing of the break-
up phenomena for dierent frequency values. Three dierent measurements were always recorded
for each operating condition, and three curves were obtained as shown in Fig. 5a. On the other
hand, the laser attenuation technique allowed the identication of the fundamental frequency, and
the results show maximum relative deviations between the values measured with two techniques
below 3%.
Based on the strobe light measurements, the wave frequencies, f , for several air velocities are
plotted in Fig. 6a, as a function of the liquid velocity. The dashed lines plotted here, represent
dierent regions dened in previous works performed by Chigier and co-workers, and correspond
to dierent disintegration modes (A, B and C). An overall appreciation of the results shows that
the wave frequency increases with liquid ow rate and also with air velocity. At low liquid ow
rates (region A) the energy transferred by shear is sucient to lead to sinusoidal mode oscillations,
with long wavelengths or low frequencies. In region B, we have a combined mode oscillation:
sinusoidal and dilational (Mansour and Chigier, 1991). As the liquid ow rate increases, the di-
lational mode becomes more and more important and the frequency still increases with UL . The
 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789 785

Fig. 6. Break-up frequency of the liquid sheet: (a) as a function of the liquid velocity, (b) as function of air velocity.

plateau is observed for all ows with the onset of the maximum spray angle, that is, after that
point the sinusoidal mode is still present, but the dilational mode begins to have a stronger in-
uence. The transition to region C corresponds to the onset of a dominant dilational mode
(Mansour and Chigier, 1991), always present for high values of UL , where not enough energy is
transferred to the liquid lm due to air interaction, and corresponds to the onset conditions of the
transition ow. In this region, no dominant frequency was identied. With the strobe light it is
no longer possible to freeze the ow, and the power spectrum obtained with the laser atten-
uation technique does not allow the identication of a fundamental frequency due to the total
widening of the bandwidth of frequency peaks, as shown in Fig. 5d. The peak with higher energy
in this spectrum corresponds to the electrical net noise at 50 Hz followed by the electronic noise of
the acquisition board at 286 Hz.
Also shown in Fig. 6b is the inuence of the air velocity on the liquid lm disintegration fre-
quency, for UL 0:8, 1.0 and 1.5 m/s. The oscillation frequency is strongly dependent on UG , with
f / UG0:73 , f / UG0:70 and f / UG0:88 , for UL 0:8, 1.0 and 1.5 m/s, respectively.

3.2.3. Break-up lengths

The break-up length, dened as the distance between the injector tip and the horizontal plan
were the lm break-up rst occurs, was measured by means of the images obtained with back light
illumination.
Fig. 7a shows sample front views of the ows, with UG 15 m/s and dierent liquid velocities,
allowing a better insight into the break-up process, as well as the way the break-up lengths are
dened for dierent operating conditions. Fig. 7b shows the break-up length, Lb , as a function
of the liquid velocity, UL , for dierent air velocities. As the coowing air streams result in an
786 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789

Fig. 7. Break-up length of the liquid sheet: (a) representation of the break-up length measurement, (b) break-up length
as function of the liquid velocity, (c) break-up length as function of the air velocity.

acceleration of the liquid lm layers, the break-up length increases with the liquid velocity and
decreases with the air ow rate, independently of its magnitude.
Also shown in Fig. 7c is the inuence of the air velocity on the liquid lm break-up length, for
UL 0:73, 1.79 and 2.85 m/s. The break-up length is strongly dependent on UG , with Lb / UG1:39 ,
Lb / UG1:85 and Lb / UG2:51 , for UL 0:73, 1.79 and 2.85 m/s, respectively.
As visualisation results have been widely represented and analysed in detail, it is now necessary
to move forward to a joint analysis of the above presented results, mainly break-up frequencies
and break-up lengths, in order to be able to present the proposed correlations, for the dierent
parameters under study.

3.2.4. Joint analysis and proposed correlations

Although the importance of certain non-dimensional groups, as the Reynolds and Weber
numbers, is well established (Arai and Hashimoto, 1985; Eroglu et al., 1991; Chigier and Reitz,
1996), the relative weight of the dierent non-dimensional groups was evaluated, and the im-
portance of the air-to-liquid momentum ratio, MR, was noticeable. As it is, a better correlation
 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789 787

can be obtained between all the studied ow conditions, if the momentum ratio group, MR, is
considered.
A non-dimensional break-up length, Lb =tL , was chosen (Stapper et al., 1992; Lozano et al.,
1996). Although it is common to nd, in the literature, the break-up frequency represented as
function of the liquid velocity (Mansour and Chigier, 1991; Eroglu and Chigier, 1991; Lozano
et al., 1996) a non-dimensional frequency, is dened here as ftL =UL , which represents the ratio
between the transversal and the axial liquid velocity (as well as in accordance to the Strouhal
number).
Fig. 8 shows the non-dimensional break-up length, Lb =tL , and frequency, ftL =UL , as a function
of, MR qG UG2 =qL UL2 , the non-dimensional group that is found to control the disintegration
process. The non-dimensional group, MR, provides a good correlation for the dierent ow
conditions. Non-log representation of these results shows that, there is a sharp decrease of the
break-up length until MR 0:5, and for higher values of MR, Lb tends to be less dependent of
this parameter. For all the studied air velocities the break-up length is related with the air-
to-liquid momentum ratio by Lb =tL 6:51MR0:68 , with a correlation coecient of 0.804. As for
the non-dimensional frequencies, shown in Fig. 8b, the air-to-liquid ratio is seen to be an important
governing parameter, and is related with the break-up frequency by ft=UL 0:13MR0:38 , with a
correlation coecient of 0.965. The lower correlation coecient obtained for the break-up length
correlation is probably due to the higher uncertainty in the break-up length measurements, but
can also be due to the need of considering additional parameters in future studies and analysis.

Fig. 8. Non-dimensional break-up length and frequency as function of the air-to-liquid momentum ratio: (a)
Lb =t versus MR, (b) ft=UL versus MR.
788 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789

4. Conclusions

The process of liquid lm disintegration into a spray is studied, making use of a 2-D liquid lm
surrounded by two air streams, using two dierent visualisation techniques and through the
quantication of instability amplitudes, break-up lengths and break-up frequencies.
The visualisation results, extensively presented in this paper, compare fairly well with the ones
previously presented in literature and allowed a good description of the disintegration process
of the 2-D liquid lm studied.
Despite the traditionally used non-dimensional groups in this type of studies, namely the liquid
Reynolds number and gas Weber number, an extended analysis of the non-dimensional breakup
lengths and break-up frequencies showed the importance of the air-to-liquid momentum ratio
group, MR, as the parameter that allows to correlate the present results. The MR is an always
present parameter independently of the operating ow conditions. A better correlation is ob-
tained for break-up frequencies (ft=UL 0:13MR0:38 ), when compared with the one obtained for
the break-up length (Lb =t 6:51MR0:68 ). This is probably due to the higher uncertainty in the
break-up length measurements, but can also be due to the need of considering additional pa-
rameters in future studies and analysis.
Worth mentioning is the fact that the present nozzle conguration was made transparent in
order to minimise other overlapping phenomena (due to air bubble formation in the liquid) that
can be misleading in the interpretation of the results for the process under study.

Acknowledgements

Financial support has been provided through the BRITE/EURAM programme of the Euro-
pean Commission under the contract AERO-CT95/0122. The authors are pleased to acknowledge
the research grant of Dina Santos from the PRAXIS XXI program of the Portuguese Ministry of
Science and Technology.

References

Adzic, M., Carvalho, I.S., Heitor, M.V., 1994. Laser shadowgraphy of break-up of a liquid sheet downstream of a
model prelming airblast atomizer. 7th International Symposium on Applications of Laser Anemometry to Fluid
Mechanics, 1114 July, Lisbon, Portugal.
Arai, T., Hashimoto, H., 1985. Disintegration of a thin liquid sheet in a cocurrent gas stream. ICLASS-85, Paper VIB/1,
910 July, London, UK.
Beck, J.E., Lefebvre, A.H., Koblish, T.R., 1991. Airblast atomization at conditions of low air velocity. J. Propulsion,
207212.
Carentz, H., Berthoumieu, P., Lavergne, G., 1998. Study of planar liquid sheet disintegration. ILASS-Europe98, 68
July, Manchester, England, pp. 1317.
Carvalho, I.S., Heitor, M.V., 1997. Liquid lm break-up in a model of a prelming airblast nozzle. Exp. Fluids 24, 408415.
Carvalho, I.S., Heitor, M.V., Santos, D., 1998. Liquid lm disintegration mechanisms. Third International Conference
on Multiphase Flow, ICMF98, 812 June, Lyon, France.
Carvalho, I.S., Heitor, M.V., Santos, D., 1997a. Disintegration mechanisms in at and annular liquid lms. 13th
Annual Conference on Liquid Atomization and Spray Systems, 911 July, Florence, Italy.
 I.S. Carvalho et al. / International Journal of Multiphase Flow 28 (2002) 773789 789

Carvalho, I.S., Heitor, M.V., Santos, D., 1997b. On the analysis of shear-driven liquid break-up processes. Proceedings
of the 11th Symposium on Turbulent Shear Flows, 811 September, Grenoble, France, pp. 27.627.11.
Chigier, N., Dumouchel, C., 1996. Atomization of liquid sheets. In: Recent Advances in Spray Combustion: Spray
Atomization and Drop Burning Phenomena, Vol. I. Prog. Astronaut. Aeronaut. 166, 241259.
Chigier, N., Reitz, R.D., 1996. Regimes of jet breakup and breakup mechanisms (physical aspects). In: Recent
Advances in Spray Combustion: Spray Atomization and Drop Burning Phenomena, Vol. I. Prog. Astronaut.
Aeronaut. 166, 109136.
Dombrowski, N., Johns, W.R., 1963. The aerodynamic instability and disintegration of viscous liquid sheets. Chem.
Eng. Sci. 18, 203214.
Eroglu, H., Chigier, N., 1991. Liquid sheet instability in coowing air stream. ICLASS-91, 1518 July, Gaithersburg,
MD, USA, pp. 679686.
Eroglu, H., Chigier, N., Farag o, Z., 1991. Coaxial atomizer liquid intact lengths. Phys. Fluids A 3, 303308.
Farago, Z., Chigier, N., 1992. Morphological classication of disintegration of round liquid jets in a coaxial air stream.
Atomizat. Sprays 2, 137153.
Fernandes, E.C., Heitor, M.V., 1997. Simultaneous measurements of velocity, pressure, temperature and heat release in
an oscillating ame. 90th Symposium of Propulsion and Energetics on the Advanced Non Intrusive Instrumentation
for Propulsion Engines. AGARD 2024 October, Bruxelles, Belgium.
Fraser, R.P., Dombrowski, N., Routley, J.H., 1963. The atomization of a liquid sheet by an impinging air stream.
Chem. Eng. Sci. 18, 339353.
Hagerty, W.W., Shea, J.F., 1955. A study of stability of plane uid sheets. J. Appl. Mech. 22, 509514.
Hardalupas, Y., Tsai, R.F., Whitelaw, J.H., 1998. Primary breakup of coaxial airblast atomisers. ILASS-Europe98,
68 July, Manchester, England, pp. 4247.
Karl, J., Huilier, D., Burnage, H., 1996. Mean behavior of a coaxial airblast atomized spray in a co-owing air stream.
Atomizat. Sprays 6, 409433.
Lee, S.Y., 1985. Eect of condensation on the breakup of liquid sheet-experimental observations. ICLASS-85, Paper
VIB/2, 910 July, London, UK.
Lefebvre, A.H., 1992. Energy considerations in twin-uid atomization. Trans. ASME, J. Eng. Gas Turbines Power 114,
8996.
Lee, C.P., Wang, T.G., 1986. A theoretical model for the annular jet instabillity. Phys. Fluids A 29, 20762085.
Lozano, A., Call, C.J., Dopazo, C., Garcia-Olivares, A., 1996. Experimental study of the atomization of a planar liquid
sheet. Atomizat. Sprays 6, 7794.
Mansour, A., Chigier, N., 1990. Disintegration of liquid sheets. Phys. Fluids A 3, 29712980.
Mansour, A., Chigier, N., 1991. Dynamic behavior of liquid sheets. Phys. Fluids A 2, 706719.
Rizk, N.K., Lefebvre, A.H., 1980. The inuence of liquid lm thickness on airblast atomization. Trans. ASME, J. Eng.
Power 102, 706710.
Stapper, B.E., Sowa, W.A., Samuelsen, G.S., 1992. An experimental study of the eects of liquid properties on the
breakup of a two-dimensional liquid sheet. Trans. ASME, Eng. Gas Turbine Power 114, 3945.