International Journal of Heat and Fluid Flow 20 (1999) 513±519

www.elsevier.com/locate/ijh€

In¯uence of liquid ¯ow conditions on spray characteristics of

internal-mixing twin-¯uid atomizers
A. Ku€erath *, B. Wende, W. Leuckel
Engler-Bunte-Institut, Bereich Feuerungstechnik, Universit
at Karlsruhe (TH), D-76128 Karlsruhe, Germany

Abstract
For an internal-mixing twin-¯uid atomizer with coaxial liquid feed the dependence of the spray characteristics on the liquid ¯ow
conditions, i.e. laminar, turbulent or cavitating, has been investigated. Results obtained with Phase Doppler Analyzer, high speed
photography and pressure drop measurements were correlated with the nozzle geometry and the ¯ow rates of liquid and gas. The
investigated nozzle allowed variation of ®ve geometric parameters, the most important ones being the liquid inlet port diameter, the
nozzle outlet port diameter and the length of the outlet port. The volume ¯ow rate of the liquid V_L ranged from 5 to 100 L/h, and of
the compressed air V_A from 1 to 20 m3n /h. For visualizing the internal ¯ow, an optically accessible planar nozzle of comparable
geometry has been used. It was observed that ¯ow conditions of the liquid jet leaving the inlet port have a strong in¯uence on the
distributions of the D32 droplet diameter along radial traverses across the spray. For short outlet ports (lo /do [ 2), the maximal D32
for laminar liquid jet is always found on the spray axis, whereas a turbulent liquid jet leads to nearly even pro®les. For long outlet
ports (lo /do J 4) even pro®les are found in either case. Ó 1999 Elsevier Science Inc. All rights reserved.

Keywords: Twin-¯uid atomization; Primary breakup; Breakup modes; Laminar±turbulent transition; Flow visualization

Notation 1. Introduction

ALR air-to-liquid mass ratio (±) In many industrial applications such as spray-drying, ex-
d inlet/outlet port diameter (mm) haust cleaning and combustion, well de®ned spray character-
D32 Sauter Mean Diameter (lm) istics and a wide ¯ow rate operation range are required. In
l inlet/outlet port length (mm) most cases air-assisted atomizers are able to ful®l these re-
m_ liquid mass ¯ux density (kg/(m2 s)) quirements and are used, therefore, despite their additional
p pressure beyond atmospheric (bar) energy consumption. As a group, air-assisted atomizers can be
r radius (mm) divided into internal and external mixing types, depending on
u axial velocity (m/s) where the ®rst contact between the liquid and gas phase takes
V_ volume ¯ow rate (L/h, m3n /h) place. If abrasion and fouling are unimportant, internal-mix-
z distance downstream of the nozzle (mm) ing atomizers are preferred due to their more ecient energy
Re liquid ¯ow Reynolds number Re ˆ uL di qL =gL transfer from gas to liquid. From this category, the Y-jet at-
(±) omizer, the mixing chamber atomizer, the e€ervescent atom-
g dynamic viscosity (Pa s) izer, and the internal-mixing atomizer with coaxial liquid feed
q density (kg/m3 ) are the most common ones in chemical engineering and com-
r surface tension (N/m) bustion applications. The last of these will be the subject of this
paper.
Subscripts In the past, the performance of air-assisted internal-mixing
A air (±) atomizers with coaxial liquid feed has been investigated in
i inlet port for the liquid (±) order to clarify the e€ects of nozzle design, load, liquid and gas
L liquid (±) properties on spray characteristics (Nukiyama and Tanasawa,
mean mean (±) 1938±1939 Lorenzetto and Lefebvre, 1977). Results of these
n standard temperature and pressure (±) investigations are semi-empirical equations which yield a
o outlet port of the nozzle (±) characteristic droplet diameter as a function of those param-
eters. In the corresponding relationships, the governing prop-
erty is the velocity slip between the liquid jet and the
surrounding gas. Quite often a second term is added to ac-
*
Corresponding author. E-mail: andreas.ku€erath@ciw.uni- count for the dependence on the liquid viscosity, which leads to
karlsruhe.de an increase of the resulting droplet diameter for liquids of

0142-727X/99/$ ± see front matter Ó 1999 Elsevier Science Inc. All rights reserved.
PII: S 0 1 4 2 - 7 2 7 X ( 9 9 ) 0 0 0 4 0 - 5
514 A. Ku€erath et al. / Int. J. Heat and Fluid Flow 20 (1999) 513±519

higher viscosity. These equations are of a similar type as the tures of the liquid jet surface leaving the duct of the liquid
attempts to correlate a diameter with the Weber and Oh- insert. In the laminar case the surface is smooth whereas, for
nesorge number. They are only applicable for any particular the turbulent liquid jet, distortions on the surface are present.
atomizer within the investigated ranges. Extrapolations seem They provide the target for aerodynamic shear forces and,
impossible, since the physical basis of the equations is weak as therefore, will lead to improved disintegration. In addition, the
the disintegration process is not yet fully understood (Chigier, occurrence of cavitation can lead to a drastic change of the
1991). The process consists of many successive and simulta- ¯ow pattern of the liquid jet. Both, transition from laminar to
neous steps, i.e. the rearrangement of the liquid velocity pro®le turbulent and onset of cavitation, may lead to a major change
after exiting from the liquid insert (Schweitzer, 1937), the in spray characteristics.
generation of small scale disturbances on the liquid surface, the
onset of breakup (Wu and Faeth, 1993), the disintegration of
the jet (Farago and Chigier, 1992), and ®nally the breakup of 2. Experimental
ligaments and droplets (Gelfand, 1996). Detailed knowledge of
these single steps provides the possibility of interpreting a more The design of the nozzle used for most of the parametric
complex disintegration process of the liquid. studies described in this paper is shown in Fig. 1. The pres-
For the internal-mixing twin-¯uid atomizer with coaxial surized air is fed through the annular slot (b), and the liquid is
liquid feed, the in¯uence of liquid ¯ow conditions on spray fed through the inlet port (a). After the ®rst contact of air and
characteristics has not yet been the subject of detailed pub- liquid, a two-phase ¯ow develops in the outlet port (c). Due to
lished research. Investigations focussed mainly on the in¯uence the ¯exible construction, the length lo (0 6 lo 6 16 mm) and
of the relative velocity between the liquid jet and the sur- the diameter do (1.1 6 do 6 4.8) of the outlet port and the
rounding air. As a result, the liquid is usually injected with low length li and the diameter di (0.4 6 di 6 2.0 mm) of the inlet
velocity in order to maximize the relative velocity between the port can be varied independently. The length to diameter ratio
liquid ¯ow and the atomizing air. The important in¯uence of li /di for the inlet port varied between 5 and 20. For the visu-
liquid turbulence for an improved and accelerated breakup of alization of the processes which take place inside the nozzle a
the liquid jet has even been doubted (Lefebvre, 1989). In planar, optically accessible nozzle with a square cross section
contrast to this assumption, Sato et al. 1988 presented results (3 ´ 3 mm) outlet port has been used. For optical access, two
of a similar atomizer type for the atomization of coal±water sides of this planar nozzle were made out of acrylic glass.
mixtures. In this case an ecient atomization has been Details of the design and the geometric parameters of this
achieved by increasing the interfacial perturbation between nozzle were published in Ku€erath and Leuckel (1997). The
liquid and air in the exit port of the atomizer and promoting transferability of the results obtained with the optically ac-
the turbulence of the liquid supplied to the atomizer. cessible planar nozzle for the rotationally symmetric nozzle
On the other hand it is also well known from investigations was checked up beforehand.
dealing with injecting a liquid jet into quiescent air, that the
turbulence of the liquid jet has an important in¯uence on jet
breakup. It must be noted for those experiments that the rel-
ative velocity between air and liquid is only inverted but
should have the same e€ect, namely to propagate an initial
disturbance amplitude (Mansour and Chigier, 1994). In com-
bination with the dynamic pressure of the air any initial dis-
turbance starts an enhanced breakup. Concerning this aspect,
the importance of the turbulent boundary layer for the
breakup of the liquid jet was investigated by Wu et al., 1995.
They stripped o€ the turbulent boundary layer of a not fully
developed turbulent liquid ¯ow with a cutter mechanism. With
this removal of the boundary layer it was possible to generate
liquid jets with a Reynolds number greater than 105 with a
smooth surface. The in¯uence of air ¯ow turbulence on the
breakup mechanism is virtually unexplained, though recently
this in¯uence became a subject of research (Shavit and Chigier,
1994). Investigations concerning the in¯uence of cavitation on
jet breakup were undertaken by many research facilities, but
for the atomizing system presented here cavitation is of minor
importance. Summarizing, one may assume that, for the in-
vestigated internal-mixing twin-¯uid with coaxial liquid feed,
the e€ect of liquid ¯ow conditions on the break up process
cannot be neglected.
As mentioned above, the spray characteristics of an inter-
nal-mixing air-assisted atomizer strongly depend on the in-
teraction of gas and liquid inside the nozzle, where the
di€erence in velocity of the two phases is suciently high to
disintegrate the liquid jet. In this zone, the major part of lig-
ament formation and further droplet breakup takes place. This
study focusses on the in¯uence of liquid ¯ow conditions upon
the atomization process and the resulting spray characteristics.
The ¯ow condition of the liquid when entering the mixing
chamber is determined by the volumetric ¯ow rate, the vis-
cosity, and the geometry of the inlet port. The two main liquid
¯ow regimes, laminar and turbulent, result in di€erent struc- Fig. 1. Investigated internal mixing nozzle system.
 A. Ku€erath et al. / Int. J. Heat and Fluid Flow 20 (1999) 513±519 515

Most experiments were conducted using a glycol±water nomenological description the main liquid ¯ow regimes: lam-
mixture at 25°C as a test liquid (q ˆ 1084 kg/m3 , g ˆ 5 ´ 10ÿ3 inar, turbulent and cavitating, are illustrated. In the second
Pas, r ˆ 48 ´ 10ÿ3 N/m). The spray is locally characterized by part, the separate variation of the governing parameters for the
droplet diameter, droplet velocity and volume ¯ux density. laminar±turbulent transition of the liquid ¯ow is being inves-
This information is obtained by spatially resolving phase tigated. The onset of cavitation and its e€ect on spray char-
doppler analyzer (PDA) measurements 200 mm downstream acteristics is the subject of the third part. Finally, other
of the nozzle exit where no further droplet breakup occurs and in¯uences apart from liquid ¯ow conditions, such as the gas
the spray density allows reliable PDA measurements over a ¯ow rate, the geometry of the outlet port and the length of the
wide range of ¯ow rates. Photos of the ¯ow structures and the mixing chamber, on transition between the di€erent ¯ow re-
break up process in-and outside the nozzle were taken using a gimes of the liquid phase and on the resulting spray charac-
nanolite spark in combination with a synchronized re¯ex teristics are presented.
camera. The spark has a duration time of approximately 20 ns
which is suciently short to freeze the motion of the liquid 3.1. Phenomenological description
phase. Due to the diculties in determining accurately the
local mass ¯ux density with the PDA, an isokinetic sampling of In Fig. 2, upper part, the pressure drop on the air side is
the liquid fraction was used. The accuracy of the balance when plotted against the liquid volume ¯ow rate or the liquid Rey-
integrating the local mass ¯ux densities over the entire cross nolds number (ReL ), respectively. The values are speci®c for
section was better than 90%, even in the nozzle near ®eld at a constant gas ¯ow rate and ®xed nozzle geometry. High speed
distance 25 mm downstream of the nozzle. A detailed photos of the disintegration process inside and outside the
description of the complete test rig is published elsewhere optically accessible nozzle ± corresponding to the four most
(Ku€erath et al., 1996). important regimes ± are assigned to the diagram.
For very low liquid volume ¯ow rate (I), disintegration of
the liquid jet, is completed before reaching the outlet port. The
3. Results and discussion bulk of the liquid impacts on the inner wall surface of the
nozzle and approaches the outlet as a wall ®lm. Therefore,
First, examples of the qualitative in¯uence of liquid ¯ow satisfactory atomization cannot be achieved. With increasing
conditions on spray characteristics are shown. In the phe- liquid volume ¯ow rate (II) stationary atomization starts. In

Fig. 2. Air pressure drop DpA as a function of liquid volume ¯ow rate V_L and ¯ow characteristics inside the nozzle.
516 A. Ku€erath et al. / Int. J. Heat and Fluid Flow 20 (1999) 513±519

this regime the pressure drop weakly depends on the liquid eter of the inlet port, and the liquid viscosity. The inlet port
volume ¯ow rate. Disintegration of the laminar liquid jet starts diameter di has been varied between 0.6 and 1.95 mm (denoted
just above the outlet port where the di€erences in velocity as 1.9 mm) for constant gas ¯ow rates, outlet port lengths and
between the slowly ¯owing liquid jet and the fast ¯owing air diameters. Fig. 3 shows the pressure drop of the air as a
are high enough to generate the ®rst visible disturbances on the function of the Reynolds number for four inlet port diameters
surface. At the end of the outlet port, marked by the horizontal (li /di ˆ 10). There is a sudden increase in the pressure drop for
lower edge of the acrylic glass window, the cross-section is not Re J 3000, except for the smallest diameter investigated. In
homogeneously ®lled with two-phase ¯ow. Therefore, part of this case a gradual pressure increase could be observed. For
the air passes the nozzle as an annular co¯ow without really di ˆ 0.9 and 1.3 mm a transition region from Re  2000 to
contributing to the atomization process. Further increase of Re  3000 occurs, whereas for di ˆ 0.6 mm no transition region
the liquid volume ¯ow rate (III) causes a strong increase of could be found. These di€erences in the appearance of tran-
pressure drop. This regime is characterized by sudden changes sition can be explained by the formation of disturbances. The
of spray characteristics and pressure drop, the atomization smaller the diameter, the more important are external in¯u-
process becomes non-stationary, accompanied by the emission ences such as manufacturing imperfections. For the smallest
of high-pitch sound. The frequency of pressure ¯uctuations diameter it was even impossible to generate a jet with laminar
and changes in spray characteristics is about 1 Hz. When ap- appearance. This jet is transparent, but has a large scale helix
plying the PDA for the measurement of time resolved mea- structure. Therefore, these smallest inlet ports could not be
surements of droplet arrival statistics in this transition region investigated in detail.
(Re ˆ 3250), these sudden changes in pressure drop could well As mentioned before, transition between laminar and tur-
be correlated with the changes in spray characteristics. At the bulent liquid ¯ow is related to a drastic change of spray
spray axis the laminar state is characterized by an increase in characteristics. Fig. 4 shows spatial D32 distributions for two
the frequency of the occurrence of bigger droplets. As a sta-
tionary atomization process cannot be achieved there, this
transition between the laminar (II) and the turbulent (IV) ¯ow
regime of the liquid excludes technical applications. As in the
laminar case, for the turbulent liquid ¯ow (IV) the pressure
drop again increases slowly with the liquid volume ¯ow rate.
The atomization process is now steady again. Disintegration
starts shortly after the inlet port. The cross-section at the end
of the outlet port is nearly completely ®lled with two-phase
¯ow. Therefore, better energy transfer from the gas to the
liquid than in the laminar case is being obtained. For still
higher liquid volume ¯ow rate, an unsteady transition (V) to
cavitation (VI) takes place, accompanied by another increase
of pressure drop. It must be noted, however, that this transi-
tion is not correlated with Reynolds number and, conse-
quently, does not necessarily start in the turbulent region. The
disintegration process in the cavitation regime is similar to the
turbulent case. The improvement of the disintegration process
and droplet formation is not remarkable. For higher volume
¯ow rate (VII) cavitation becomes unstable. The liquid jet is
de¯ected o€ the nozzle axis. The pressure drop may decrease
drastically because the cross-section of the outlet port is now Fig. 3. Air pressure drop DpA as a function of liquid ¯ow Reynolds
partly released. A great deal of air leaves the outlet port number and inlet port diameter di .
without contributing to the atomization process. As a result,
the atomization quality deteriorates and a strong imbalance of
the droplet diameter distribution and the mass ¯ux density
over the cross-section can be observed. For very high liquid
¯ow rates and moderate air pressures in the mixing chamber
also a hydraulic ¯ip (Yule et al., 1998) of the liquid jet was
observed. In the absence of distortions of a suciently high
amplitude, the liquid jet after the hydraulic ¯ip showed a de-
layed primary breakup, resulting in very poor atomization.
When investigating a liquid jet in quiescent air, the regions
of transition from laminar to turbulent are comparable. It can
be concluded, therefore, that transition (III) between two at-
omization regimes is caused by liquid ¯ow properties which are
correlated with the liquid Reynolds number. On the other
hand, the onset of cavitation strongly depends on the pressure
inside the mixing chamber, liquid inlet geometry (Ramamurthi
and Nandakumar, 1994) and manufacturing imperfections
(Ohrn et al., 1991) and, hence, cannot be described by liquid
¯ow properties alone.

3.2. Governing parameters for the laminar±turbulent transition

Since transition can be correlated with the liquid Reynolds Fig. 4. Radial pro®les of local D32 as a function of inlet port diameter
number, it depends on the liquid volume ¯ow rate, the diam- di and liquid volume ¯ow rate V_L .
 A. Ku€erath et al. / Int. J. Heat and Fluid Flow 20 (1999) 513±519 517

inlet port diameters, related to both ¯ow regimes (li /di ˆ 10). In
both laminar cases the maximum of the D32 could be observed
at the spray axis. At the spray boundary the D32 mean droplet
diameter is far smaller. As in Fig. 2-II, air leaving the nozzle
with high velocity led to disintegration of the liquid into ®ne
droplets just in the outer zones of the spray. A complete dis-
integration of the ligaments near the spray axis could not be
achieved, however. In contrast to the laminar case, no liga-
ments can be seen at the bottom of the photograph for the
turbulent case (Fig. 2-IV). At the nozzle exit, liquid phase
disintegration is nearly complete. This corresponds to the re-
sult that in both turbulent cases a nearly even pro®le of the D32
was obtainable.
According to Fig. 5, for constant liquid volume ¯ow rate,
the laminar (di ˆ 1.95 mm) and the turbulent (di ˆ 1.3 mm)
liquid ¯ow regime show drastically di€erent distributions of
mass ¯ux density. In contrast for the laminar case at a distance
of 25 mm downstream the nozzle, a distinct peak of the mass
¯ux density exists at the spray axis, for the turbulent jet the Fig. 6. Air pressure drop DpA as a function of Reynolds number for
mass ¯ux density is about half an order of magnitude smaller. di€erent viscosities of the liquid phase, V_A ˆ 5:9 m3n /h, do ˆ 2.4 mm,
These di€erences are con®rmed by light sheet photographs di ˆ 1.3 mm, lo ˆ 2 mm, li /di ˆ 20.
taken at the nozzle exit. Even 200 mm downstream a distinct
maximum of the mass ¯ux density is apparent for the laminar
liquid jet. It should be mentioned that this great change in
spray characteristics also leads to an increase in the total en- constant, the gas density and the pressure drop must increase
trainment rate of the expanding two phase ¯ow of about 50 % for decreasing sonic velocity. For the spatial distribution of the
when exceeding the laminar±turbulent transition. This is ex- D32 a similar behavior to that with the variation of the inlet-
tremely important for all processes for which rapid mixing of port diameter could be observed. For all laminar cases, the
the surrounding gas with the two phase ¯ow is desired. maximum of the D32 could be detected at the spray axis, in-
In order to investigate the in¯uence of liquid viscosity, this dependent of the individual liquid viscosity, whereas for tur-
was varied by varying the temperature of the liquid and by bulent cases an even pro®le with a similar average D32 has been
using di€erent liquids. In Fig. 6 the pressure drop at the air achieved.
side is shown for four viscosities at di€erent temperatures of For technical applications the inlet duct is usually shorter
the same liquid. As in Fig. 3 there is a strong increase of the than the length required for fully developed ¯ow; the required
pressure drop at about Re ˆ 3000. Along the laminar regime, length for laminar ¯ow is 130 di and for turbulent ¯ow 50 di .
the pressure drop at constant Reynolds number is equal for all The appearance of transition from laminar to turbulent ¯ow
viscosities although the liquid volume ¯ow rate is di€erent. In depends on the state of development of the ¯ow. Fig. 7 shows
this region, the pressure drop is mainly governed by the an- the pressure drop of the air for three di€erent length-to-
nular air ¯ow in the nozzle exit. In the turbulent regime, the diameter ratios (li /di ) of the inlet port. The ratio li /di ˆ 20
pressure drop for constant Reynolds number strongly depends shows the most stable laminar ¯ow (late transition). The
on the liquid volume ¯ow rate because the nozzle exit is ®lled transition is restricted to a small region at about Re ˆ 3000.
with a two-phase ¯ow. For two-phase ¯ow, the sonic velocity For smaller ratios of li /di the distortion of the liquid surface
decreases with increasing liquid volume fraction and, hence,
with the liquid volume ¯ow rate. Since the gas ¯ow is kept

Fig. 5. Radial pro®les of mass ¯ux density m_ L for laminar and tur-
bulent liquid ¯ow regime, V_L ˆ 70 L/h, V_A ˆ 5:9 m3n /h, do ˆ 2.4 mm, Fig. 7. Air pressure drop DpA as a function of liquid volume ¯ow rate
lo ˆ 2 mm, li /di ˆ 10. for di€erent ratios of length to diameter (li /di ) for the liquid inlet.
518 A. Ku€erath et al. / Int. J. Heat and Fluid Flow 20 (1999) 513±519

starts earlier and a transition region with ¯uctuations in

pressure occurs. Fig. 8 shows D32 distributions corresponding
to Re ˆ 3250. The ratio of li /di ˆ 20 still shows a laminar
pro®le, whereas the shorter ratios of li /di show a nearly even
pro®le which is characteristic for turbulent liquid ¯ow. The
di€erences in the average D32 between li /di ˆ 5 and li /di ˆ 10
may be explained by di€erent exit conditions, i.e. the degree of
turbulence of the liquid jet when leaving the inlet port.

3.3. Governing parameters for transition to cavitation

Cavitation is caused by an increase in the dynamic pressure

at the ¯ow restriction inside the inlet port (see Fig. 1). As a
consequence, the static pressure reaches the magnitude of the
vapor pressure. The vapor bubbles formed collapse after a
short time downstream of the inlet port, generating additional
disturbances on the liquid jet surface and tearing droplets from Fig. 9. Radial pro®les of local D32 and umean for turbulent and cavi-
the jet surface. These droplets can be seen as cone-shaped tating liquid ¯ow, V_A ˆ 5:9 m3n /h, do ˆ 2.4 mm, lo ˆ 2 mm, li /di ˆ 10.
deposits on the window (Fig. 2-VI). In contrast to the non-
cavitating jet (IV) the ®rst visible part of the liquid jet is
opaque due to dispersed vapor bubbles. According to Fig. 9,
the improvement of atomization quality due to the onset of
cavitation is negligible, and a slight deterioration of the droplet
diameter symmetry may be observed. The mean velocity of the
droplets is also nearly una€ected by transition to cavitation
(Fig. 9). Further increase in liquid volume ¯ow rate is followed
by increased cavitation. As a result, a ¯ow structure similar to
Fig. 2 VII appears with drastic asymmetry of the spray. It is
worth mentioning that, at this ¯ow condition, a slight increase
in the air mass ¯ow rate, coupled with an increase in pressure
inside the mixing chamber, could reconvert this state of ¯ow to
the stable cavitating ¯ow condition marked in Fig. 2 with VI.

3.4. In¯uences other than liquid ¯ow characteristics

According to Fig. 10, with increase of the gas ¯ow rate the
transition region moves slightly to smaller Reynolds numbers,
which can be attributed to stronger aerodynamic forces.
However, the general appearance of the D32 pro®les remains
the same. For all air ¯ow rates with a laminar liquid ¯ow, a
distinct peak was observed at the spray axis, whereas for the
turbulent liquid jet an nearly even pro®le has been obtained.
However, as expected, the D32 decreases for both ¯ow condi-
tions over the entire cross section with increasing gas ¯ow rate. Fig. 10. Air pressure drop DpA as a function of the liquid ¯ow rate or
liquid ¯ow Reynolds number, do ˆ 2.4 mm, lo ˆ 2 mm, li /di ˆ 10.

In addition, the length to diameter ratio (lo /do ) of the outlet

port has a strong in¯uence on the spray characteristics (Ku-
€erath et al., 1996). For high ratios of lo /do ( J 4) the in¯uence
of the inlet characteristics of the liquid jet disappears. Now, for
laminar liquid ¯ow also an even D32 -pro®le has been obtained,
because the residence time of the two-phase ¯ow in the outlet
port is suciently long for a nearly complete liquid-phase
disintegration, and therefore one ends up with the two phase
¯ow covering the whole cross-section of the outlet port. In this
case, the pro®les for the laminar and turbulent cases are nearly
the same. Nevertheless, it must be considered that the length of
the outlet port has a strong in¯uence on the spray character-
istics (Ku€erath et al., 1996). The distance between the inlet
port for the liquid and the beginning of the outlet port of the
nozzle (mixing chamber length) is of minor importance within
the investigated range from 5 to 15 mm. However, for a mixing
chamber length shorter than 5 mm in the turbulent and the
laminar case, the time is too short for sucient distortion of
Fig. 8. Radial pro®les of local D32 for di€erent length to diameter the liquid jet. Therefore a slight increase of the D32 diameter
ratios (li /di ) for the liquid inlet. distribution was observed over the entire cross-section.
 A. Ku€erath et al. / Int. J. Heat and Fluid Flow 20 (1999) 513±519 519

4. Conclusions Ku€erath, A., Leuckel, W., 1997. Experimental investigation of ¯ow

conditions inside an air-assisted internal-mixing nozzle and their
The ¯ow characteristics of a liquid jet leaving the inlet port correlation with spray data. In: Proc. ICLASS 97, Seoul, South
have a strong in¯uence on the radial distributions of the D32 Korea, pp. 262±269.
droplet diameter and the mass ¯ux density in the spray. For Lefebvre, H., 1989. Properties of Sprays. Particle and Particle Systems
short outlet ports (lo /do [ 2), the maximal D32 mean droplet Characterization 6, 176±186.
diameter for laminar liquid jet is found on the spray axis, Lorenzetto, G.E., Lefebvre, A., 1977. Measurements of drop size on a
whereas turbulent liquid jets lead to a nearly even radial pro- plain-jet airblast atomizer. AIAA Journal 15 (7), 1006±1010.
®le. The turbulent liquid jet leads to a much broader distri- Mansour, A., Chigier, N., 1994. E€ect of turbulence on the stability of
bution of the dispersed liquid phase downstream of the nozzle. liquid jets and the resulting droplet size distribution. Atomization
For long outlet ports (lo /do J 4), however, even pro®les are and Sprays 4, 583±604.
found in either case. Nukiyama, S., Tanasawa, Y., 1938. An experiment on the atomization
Transition between both regimes of spray characteristics of liquid by means of an air stream. Trans. Soc. Mech. Engrs.
turned out to be really a function only of the liquid ¯ow (Japan) 4/5, 13±17.
Reynolds number. The transition region should be avoided in Ohrn, T.R., Senser, D.W., Lefebvre, A.H., 1991. Geometric e€ects on
technical applications due to an unsteady behavior of the discharge coecients for plain-ori®ce atomizers. Atomization and
spray. Furthermore, the volume ¯ow rate for the liquid should Sprays 1, 137±153.
be limited below the onset of cavitation, because its e€ect on Ramamurthi, K., Nandakumar, K., 1994. Disintegration of liquid jets
the disintegration mechanism cannot be well predicted and from sharp-edged nozzles. Atomization and Sprays 4, 551±564.
may lead to spray jet asymmetries. Sato, K., Okiura, K., Shoji, K., Akiyama, I., Takahashi, Y., 1988.
Experimental study on the atomization process of slurry fuels. In:
Proc. ICLASS 88, Sedai, Japan, pp. 73±80.
References Schweitzer, P.H., 1937. Mechanism of disintegration of liquid jets.
J. Appl. Phy. 8, 513±521.
Chigier, N.A., 1991. The Physics of Atomization, In: Proc. ICLASS- Shavit, U., Chigier, N., 1994. Turbulence ¯ow characteristics in a
91, Gaithersburg, MD, USA., pub NIST 813, pp. 1±15. special atomizer. In: Proc. ICLASS 94, Rouen, France pp. 205±212.
Farago, Z., Chigier, N.A., 1992. Morphological classi®cation of Wu, P.K., Faeth, G.M., 1993. Aerodynamic e€ects on primary
disintegration of round liquid jets in a coaxial air stream. breakup of turbulent liquids. Atomization and Sprays 3, 265±289.
Atomization and Spray Tech. 2, 137±153. Wu, P.K., Miranda, R.F, Faeth, G.M., 1995. E€ect of initial ¯ow
Gelfand, B.E., 1996. Droplet breakup phenomena in ¯ows with conditions on primary breakup of nonturbulent and turbulent
velocity lag. Progress Energy Combustion Science 22, 201±265. round liquid jets. Atomization and Sprays 5, 175±196.
Ku€erath, A., Heyse, C., Leuckel, W., 1996. In¯uence of outlet port Yule, A.J., Dalli, A.M., Yeong, K.B., 1998. Transient cavitation and
length on liquid atomization by an air-assisted internal-mixing separation in a scaled-up model of a VCO ori®ce. In: Proc. ILASS
nozzle. In: Proc. ILASS 96, Lund, Sweden, pp. 179±184. 98, Manchester, UK, pp. 230±235.