Detailed Numerical Simulations

of the Primary Atomization of a

Turbulent Liquid Jet in Crossflow
Marcus Herrmann
Department of Mechanical and Aerospace
Engineering,
Arizona State University,
Tempe, AZ 85287
e-mail: marcus.herrmann@asu.edu

This paper presents numerical simulation results of the primary atomization of a turbulent liquid jet injected into a gaseous crossflow. Simulations are performed using the
balanced force refined level set grid method. The phase interface during the initial
breakup phase is tracked by a level set method on a separate refined grid. A balanced
force finite volume algorithm together with an interface projected curvature evaluation is
used to ensure the stable and accurate treatment of surface tension forces even on small
scales. Broken off, small scale nearly spherical drops are transferred into a Lagrangian
point particle description allowing for full two-way coupling and continued secondary
atomization. The numerical method is applied to the simulation of the primary atomization region of a turbulent liquid jet q 6.6, We 330, Re 14, 000 injected into a
gaseous crossflow Re 570, 000, analyzed experimentally by Brown and McDonell
(2006, Near Field Behavior of a Liquid Jet in a Crossflow, ILASS Americas, 19th
Annual Conference on Liquid Atomization and Spray Systems). The simulations take the
actual geometry of the injector into account. Grid converged simulation results of the jet
penetration agree well with experimentally obtained correlations. Both column/bag
breakup and shear/ligament breakup modes can be observed on the liquid jet. A grid
refinement study shows that on the finest employed grids (flow solver 64 points per
injector diameter, level set solver 128 points per injector diameter), grid converged drop
sizes are achieved for drops as small as one-hundredth the size of the injector diameter.
DOI: 10.1115/1.4000148

Introduction

The atomization of turbulent liquid jets injected into fast moving subsonic gaseous crossflows is an important application for
example in gas turbines, ramjets, and augmentors. It is a highly
complex process that has been extensively studied experimentally
over the past decades. Early studies of the atomization of nonturbulent liquid jets in crossflows have recently been reviewed in
Ref. 1, whereas newer studies of this case include the work
reported in Refs. 18. Most experimental work has focused on
jet penetration, including both the column trajectory and the resulting spray penetration. Some recent jet penetration correlations
can be found in Refs. 4,8. Unlike turbulent liquid jets atomized
either by injection into still air or by a coflowing fast moving gas
stream, experimental access to the primary atomization region of
liquid jets injected into crossflows is relatively straightforward.
This is due to the fact that at least the windward side of the liquid
jet is typically not surrounded by many drops that could otherwise
conceal the jet phase interface geometry. This is not, however,
typically the case for the leeward side. Still, experimental studies
of nonturbulent liquid jets injected into subsonic crossflows have
concluded that depending on the momentum flux ratio and the
crossflow Weber number, essentially two different breakup modes
can be observed 8. For low momentum flux ratios and crossflow
Weber numbers, the liquid jet breaks up as a whole some distance
downstream of the injector. For high momentum flux ratios and
crossflow Weber numbers, on the other hand, surface breakup or
stripping at the sides of the liquid jets is observed prior to column
breakup 8. Unlike nonturbulent jets, turbulent liquid jets do not
exhibit this clear separation of breakup modes 9. Stripping at the
sides of the liquid column occurs even at low Weber numbers,
Manuscript received May 5, 2009; final manuscript received May 6, 2009; published online March 30, 2010. Review conducted by Dilip R. Ballal.

where nonturbulent jets exhibit only the column-breakup mode.

Turbulence initiates the generation of ligaments that then break up
due to the Raleigh mechanism 9.
Even though at least some level of consensus exists concerning
the dominant breakup modes, modeling attempts of the atomization process have had mixed results. While correlations for jet
penetration derived from experimental data give good agreement
for parameter ranges and configurations for which they were developed, numerical simulations predicting the jet penetration, drop
sizes, and liquid volume fluxes have had mixed success 1,10,11.
This is in part due to the fact that under most operating conditions,
turbulence interaction is strong and several different atomization
mechanisms occur on the jets surface at the same time. Detailed
numerical simulations can help study these simultaneously occurring mechanisms, even in regions of the liquid jet where traditional experimental methods cannot observe the phase interface
dynamics.
Although detailed simulations solve the NavierStokes equations directly, it is incumbent on any numerical simulation to demonstrate that spatial and temporal discretization errors are not unduly impacting the obtained results. It is this required grid
independence of the numerical results that has yet to be demonstrated in numerical simulations of the primary atomization of
high speed liquid jets 1216. Besides the enormous resolution
requirement, detailed numerical simulations are challenging because one not only has to track the position of the liquid/gas
interface and handle a large number of topology changes, but one
also has to account for the fact that the phase interface is a discontinuity and the surface tension force represents a singular force
17. Treating the surface tension force numerically in a stable and
accurate manner is of crucial importance, since breakup by definition involves small scales where capillary forces are dominant.
The outline of this paper is as follows: after summarizing the
governing equations, the numerical methods employed to solve
them are briefly outlined. Finally detailed simulation results of the

Journal of Engineering for Gas Turbines and Power

Copyright 2010 by ASME

JUNE 2010, Vol. 132 / 061506-1

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 03/25/2014 Terms of Use: http://asme.org/terms

ghost cells

boundary cells

active supergrid block

active
G-grid cells

hG

Fig. 1 RLSG grid structure

primary atomization of a turbulent liquid jet q = 6.6, We

= 330, Re= 14, 000 injected into a gaseous crossflow Re
= 570, 000 analyzed experimentally in Ref. 2 will be presented
and discussed.

Governing Equations

The equations governing the motion of an unsteady incompressible immiscible two-fluid system are the NavierStokes
equations

u
1
1
1
+ u u = p + u + Tu + T
t

where u is the velocity, is the density, p is the pressure, is the

dynamic viscosity, and T is the surface tension force, which is
nonzero only at the location of the phase interface x f . Furthermore, the continuity equation results in a divergence-free constraint on the velocity field, u = 0. The phase interface location
x f between the two fluids is described by a level set scalar G, with
Gx f ,t = 0

at the interface, Gx , t 0 in fluid 1, and Gx , t 0 in fluid 2.

Differentiating Eq. 2 with respect to time yields

G
+ u G = 0
t

Assuming and are constant within each fluid, density and

viscosity at any point x can be calculated from

x = HG1 + 1 HG2

x = HG1 + 1 HG2

where indices 1 and 2 denote values in fluid 1, respectively, 2, and

H is the Heaviside function. From Eq. 2 it follows that

x x f = GG

with the Dirac delta function. Furthermore, the interface normal

vector n and the interface curvature can be expressed in terms of
the level set scalar as
n=

G
,
G

=n

Using Eqs. 6 and 7, the surface tension force T can thus be

expressed as
Tx = x x f n = GGn
with the surface tension coefficient.
061506-2 / Vol. 132, JUNE 2010

Numerical Methods

In this section, we first briefly summarize the refined level set

grid RLSG method used to track the phase interface during primary atomization. Then, the level set-based balanced force algorithm is reviewed that allows for the accurate and stable treatment
of surface tension forces. Finally the coupling procedure of the
RLSG tracked phase interface to the Lagrangian point particle
method is described.

Refined Level Set Grid Method

In the RLSG method, all level set-related equations are evaluated on a separate equidistant Cartesian grid using a dual-narrow
band methodology for efficiency. This so-called G-grid is overlaid
onto the flow solver grid on which the NavierStokes equations
are solved and can be independently refined, providing high resolution of the tracked phase interface geometry, see Fig. 1. Details
of the method, i.e., narrow band generation, level set transport,
re-initialization, curvature evaluation, and its performance compared with other interface tracking methods in generic advection
test cases, can be found in Ref. 18.
In the current simulations, refinement of the G-grid is limited to
a factor two in each spatial direction as compared with the flow
solver grid. While higher refinements are feasible from an efficiency standpoint, they would require the use of a subflow solver
grid model to correctly capture the otherwise nonresolved phase
interface dynamics on the G-grid scale 19. Instead, here, the
refined G-grid simply serves to increase the accuracy of the level
set-based phase interface tracking scheme.

Balanced Force Algorithm

In the NavierStokes equations, the position of the phase interface influences two different terms. The first term is due to Eqs.
4 and 5, since HG is a function of the position of the phase
interface. For finite volume formulations, the volume fraction cv
of a control volume is defined as

cv = 1/Vcv

HGdV

Vcv

with Vcv the volume of the control volume cv. In the RLSG
method, the above integral is evaluated using the high-resolution
G-grid, see Refs. 18,20 for a detailed description. Then both
control volume density and viscosity are simply

cv = cv1 + 1 cv2

10

cv = cv1 + 1 cv2

11

Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 03/25/2014 Terms of Use: http://asme.org/terms

The second term that is a function of the interface position is

the surface tension force term, Eq. 8. Here it is critical for stability and accuracy that the surface tension force can be balanced
by the pressure gradient jump across the phase interface exactly
on the discrete level. This is ensured by the balanced force approach 18,21 based on the continuum surface force CSF model
22. Then the surface tension force at the control volume face f is
T f = f f

12

with the phase interface curvature at the control volume face

f =

cvcv + nbrnbr
cv + nbr

13

where nbr is the control volume sharing the face with cv, and

cv =

1 : 0 cv 1
0 : otherwise

14

The control volume curvature cv is calculated from the phase

interface geometry on the high-resolution G-grid using a secondorder accurate interface projected curvature calculation method
18. A detailed description of the balanced force algorithm for the
RLSG level set method and its performance compared with alternative numerical methods in a range of test cases involving capillary forces can be found in Ref. 18.

Coupling to Lagrangian Spray Model

Atomization typically produces a vast number of both large and

small scale drops. Resolving the geometry by tracking the phase
interface associated with each of the resulting drops quickly becomes prohibitively expensive, such that a different numerical description has to be employed. An alternative approach is to introduce simplifying assumptions concerning the drop shape and treat
all drops smaller than a cut-off length scale in a point particle,
Lagrangian frame. One of the typical prerequisite of such standard
spray models is that the drop size be smaller that the flow solver
grid size. Note that the RLSG approach can resolve and track
subflow solver sized liquid structures. Drop transfer is initiated if
a separated liquid structure has a liquid volume
VD Vtrans

15

and its shape is nearly spherical

rmax

3
VD
4

1/3

16

with typically = 2 and rmax the maximum distance of the liquid

structures surface to its center of mass. This second criterion
ensures that small scale detached ligament structures that would
fulfill the first criterion are not transferred and replaced by spherical drops, since these ligaments often continue to break up by
capillary instabilities producing a range of small scale drops. Not
prematurely transferring stretched out structures thus allows this
breakup process to be simulated directly instead of relying on
secondary atomization models for the Lagrangian description.
Further details concerning the application of the Lagrangian spray
model coupling procedure can be found in Refs. 23,24.
In the Lagrangian description, full two-way momentum coupling between the drop and continuous phase is used including a
stochastic secondary atomization model 25. However, the cell
volume occupied by the Lagrangian drops is not explicitly taken
into account and neither drop/drop nor drop/tracked phase interface collisions are modeled. But as long as liquid structures have
not been transferred into the Lagrangian description, i.e., they are
still tracked by the level set scalar, secondary breakup, cell volume effects, and all collisions are fully captured.

Employed Solvers

In this work we use the flow solver CDP/CHARLES that solves the
incompressible two-phase NavierStokes equations on unstrucJournal of Engineering for Gas Turbines and Power

Table 1 Operating conditions and characteristic numbers

Jet exit diameter D mm

Crossflow density c kg/ m3
Jet density j kg/ m3
Crossflow velocity uc m/s
Jet velocity u j m/s
Crossflow viscosity c kg/ms
Jet viscosity j kg/ms
Surface tension coeff. N/m
Momentum flux ratio q
Crossflow Weber number Wec
Jet Weber number We j
Crossflow Reynolds number Rec
Jet Reynolds number Re j

Experiment

Simulation

1.3
1.225
1000
120.4
10.83
1.82 105
1.0 103
0.07
6.6
330
2178
5.7 105
14,079

1.3
1.225
12.25
120.4
97.84
1.82 105
1.11 104
0.07
6.6
330
2178
5.7 105
14,079

tured grids using the finite volume balanced force algorithm 18.
In the single phase regions, the employed scheme conserves the
kinetic energy discretely. Turbulence in the single phase regions
of the flow is modeled using a dynamic Smagorinsky large eddy
simulations LES model, however, none of the terms arising from
filtering the phase interface are modeled. The approach instead
relies on resolving all relevant scales at the phase interface, thus,
ideally reverting to a direct numerical simulation DNS at the
phase interface. As such, the current simulation approach is a
combination of LES in the single phase regions and ideally DNS
at the phase interface. A Lagrangian particle/parcel technique is
employed to model the small scale liquid drops of the atomizing
liquid spray 26.
The liquid/gas phase interface during primary atomization is
tracked by the interface tracking software LIT, using the RLSG
method 18. The solver uses a fifth-order WENO scheme 27 in
conjunction with a third order TVD RungeKutta time discretization 28. The phase interfaces curvature on the level set grid is
evaluated using a second-order accurate interface projection
method 18.
The flow solver CDP/CHARLES and the interface tracking software LIT are coupled using the parallel multicode coupling library
CHIMPS 18,29. In order to couple the level set equation to the
NavierStokes equation, u in Eq. 3 is calculated from the flow
solver velocity by trilinear interpolation. To achieve overall
second-order accuracy in time, the level set equation is solved
staggered in time with respect to the NavierStokes equations.

Computational Domain and Operating Conditions

The case analyzed in this paper is one studied experimentally

by Brown and McDonell 2. Table 1 summarizes the operating
conditions and resulting characteristic numbers. Note that although the density ratio is artificially reduced from the experimental values, all relevant characteristic numbers, i.e., momentum flux
ratio q, crossflow Weber number Wec, jet Weber number We j,
crossflow Reynolds number Rec, and jet Reynolds number Re j,
are the same as in the experiment.
Figure 2 depicts the computational domain and the used boundary conditions, as well as a zoom into the near-injector region to
show the mesh detail used in the simulations. The chosen computational domain 25D . . . 50D 0 . . . 25D 10D . . . 10D is
smaller than the channel used in the experiment 77D . . . 127D
0 . . . 54D 27D . . . 27D. However, simulations using the full
experimental channel geometry were conducted to verify that the
reduced computational domain does not impact the reported results.
The injector geometry used in the experiments consists of a
long initial pipe section of diameter 7.49 mm, followed by a 138
deg angled taper section, followed by a short pipe section of diameter D with L / D = 4, whose exit is mounted flush with the
JUNE 2010, Vol. 132 / 061506-3

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 03/25/2014 Terms of Use: http://asme.org/terms

y+: slip
wall

x-: channel
inflow

z+: sl
ip

inje

cto

y
z

z-: slip
wall

wall

x+: channel
outflow

y-: w
all

Fig. 2 Computational domain and boundary conditions left and mesh detail near the
injector right

lower channel wall. The experiments conducted for this injector

geometry suggest that the specifics of the liquid velocity distribution in the injector exit plane can have a large impact on the
resulting atomization process of the liquid jet 2,3. Both the discharge coefficient of the nozzle and the geometry of the taper
region are critical. To account for their effect in the atomization
simulations, detailed single phase large eddy simulations of the
pretaper pipe, the taper, the post-taper pipe, and the crossflow
channel in the vicinity of the injector exit were performed using a
dynamic Smagorinsky model. The channel section was included
in these simulations to capture the effect of the crossflow on the
injector exit plane velocity distribution. The simulations were performed using the correct experimental momentum flux ratio and
Reynolds number, see Table 1. Inflow boundary conditions for the
pretaper pipe were taken from a precomputed LES pipe flow
simulation database at the appropriate Reynolds number. The injector exit plane velocity distributions were then stored as a time
sequence in a database to be used in the subsequent two-phase
atomization simulations.

Figure 3 shows part of the injector simulation depicting the

axial velocity distribution in the post-taper pipe section of the
injector. All velocities are made dimensionless with the corresponding bulk jet velocity u j. The effect of the sharp corners at the
end of the taper are clearly visible, generating both recirculation
regions in the pipe immediately following the corners and significant velocity fluctuations further downstream.
Figure 4 shows an example of the velocity distributions in the
injector exit plane that are stored in the database for use as inflow
boundary conditions in the crossflow atomization simulations. The
effect of the channel crossflow is clearly visible in the shift of the
axial velocity distribution in the downstream channel direction.
The two-phase atomization simulations were performed using
three different grids of increasing resolution in order to address
the important question of how much the employed grid resolution
impacts the atomization results. Table 2 summarizes the meshes
employed in the three cases c01, c12, and c23. The flow solver
grid consists of hexahedra of edge length D / 4, which are isotropically refined in layers near the injector and lower channel walls,
such that the spatial region where the phase interface is tracked is
completely filled with equidistant grid cells of the minimum cell
size x reported in Table 2, see also Fig. 2. Note that in each case,
the G-grid is a factor 2 finer than the flow solver grid in order to
enhance numerical accuracy of the interface tracking scheme.
Separate simulations were conducted to ensure that this does not
negatively impact the resulting drop size distributions, i.e., using a
flow solver minimum grid size equal to the respective G-grid size
yielded virtually identical drop size distributions.
At t = 0, the liquid jet is initialized in the computational domain
by a small cylindrical section of length D capped by a half-sphere
protruding into the crossflow channel.

Fig. 3 Zoom of the instantaneous axial velocity distribution in

the injector midplane

Results and Discussion

Figure 5 shows the temporal evolution of the level set tracked

liquid mass for the three different grid resolutions. All cases show
an initial linear increase. This is due to the fact that at this early
stage, almost no small scale drops are generated that would be
transferred out of the tracked and into the Lagrangian representa-

Fig. 4 Instantaneous injector exit plane velocity distributions. From left to right: crossflow direction, axial direction, and
transverse direction.

061506-4 / Vol. 132, JUNE 2010

Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 03/25/2014 Terms of Use: http://asme.org/terms

Table 2 Grid resolution and mesh sizes

Case
Flow solver min x
RLSG xG
Flow solver mesh size 1 106
RLSG max active cells 1 106

c01

c12

c23

D / 16
D / 32
8.7
3.5

D / 32
D / 64
21
13

D / 64
D / 128
110
66

tion. Then at around t = 5, significant numbers of small scale drops

that fulfill the transfer criterion start to be generated resulting in a
slower increase in the tracked mass until at around t = 20 a statistically steady state is reached. From this point on, a balance exists
between the injected liquid mass and the atomized liquid mass
that is transferred into the Lagrangian description. It is interesting
to note that the coarse grid simulation c01 tends to a larger value
of the tracked liquid mass, whereas the two higher resolution
cases result in slightly smaller but similar values. This appears to
be due to the fact that for the coarse grid simulations, small scale
unresolved turbulent eddies cannot initiate atomization, thus, resulting in a smaller transfer rate per phase interface area. The
higher resolution cases, on the other hand, do resolve significantly
more small scale turbulent eddies that can corrugate the interface
and thus initiate atomization quicker. This conjecture is supported
by analyzing the instantaneous phase interface geometries shown
in Figs. 79 that show significantly more small scale structures
with increasing grid resolution. Since Fig. 5 indicates that a steady
state is reached for t 20, all statistics presented in the following
are evaluated only for t 20 for a total of 43 time units for case
c01, 35 time units for case c12, and 8.1 time units for case c23.
Figure 6 shows averaged side views of the atomizing liquid jets
evaluated for t 20. The jet penetration is compared with the
predictions of two common correlations for the penetration of the
upper edge of the liquid jet derived by fitting experimental data,
namely

x
y
= 1.37 q
D
D

1/2

17

due to work of Wu et al. 8 valid in the near-injector region only

shown in upper curve and

x
y
= 2.63q0.442
D
D

0.39

We0.088

l
H2O

0.027

18

due to the work of Stenzler et al. 4 shown in lower curve.

Recent experimental observations reported in Ref. 2 show better
agreement with the latter correlation even in the near-injector region. This is the case in the simulation results as well. Figure 6
shows better agreement with the correlation of Stenzler et al. 4
and thus also to the experimental results 2. Note that there is

Fig. 6 Impact of grid resolution on averaged side view of the

liquid jet; case c01 left, c12 center, and c23 right. Jet penetration is compared with correlations due to Wu et al. 8 upper curve and Stenzler et al. 4 lower curve.

virtually no difference between the three different resolution

cases, indicating that even for coarse grids, good jet penetration
results can be obtained.
Figures 79 show snapshots of the atomizing liquid jets at different times and different grid resolutions viewed from the side,
the front, and the top. Two main simultaneous atomization mechanisms can be observed. All cases show instability waves being
generated, predominantly visible on the windward side. These instabilities generate role-ups and continue to grow along the jet
axis until they form bag like structures that rupture resulting in a
broad range of drop sizes, not unlike the column-breakup mode. It
is speculated that this instability mode is due to a Kelvin
Helmholtz instability, however, this conjecture is still to be verified using the generated simulation data. The larger scale instability mode is most clearly visible in the lowest resolution case c01.
It is equally present in the higher resolution cases, but there it is
overlaid by significant turbulence induced surface corrugations
and thus not as visibly pronounced.
In addition to the conjectured column-breakup mode, ligaments
are formed at the sides of the liquid jets near the injector exit that
stretch out and then break up forming a range of drop sizes. At the
current stage, the above description of the breakup mechanism is
speculative and still requires a detailed, yet to be performed, quantitative study of the generated time dependent simulation data.
Note that due to the turbulent chaotic nature of the flow, it
cannot be expected that the instantaneous phase interface geometry shown in Figs. 79 converges under grid refinement. Only
statistical quantities, such as the mean jet penetration or the drop
size distribution, should converge. However, it can be seen that
many of the larger scale structures on the liquid jet appear to be
similar for all three grid resolutions, indicating that the large scale
instability mode generated on the upstream pointing side of the
liquid jet might be deterministic and sufficiently resolved even on
the coarser grids. However, the higher resolution cases show significantly more fine scale structure on the liquid jet and maintain
the thin liquid sheets associated with the bag breakup mode in the
upper half of the jet slightly longer. This is due to the fact that
unlike on the coarse grid, the artificial grid induced topology

50

50

50

40

40

40

30

30

30

20

20

20

10

10

10

0
0

10

20

30

40

50

60

0
0

10

time

20

30

40

50

60

10

time

20

30

40

50

60

time

Fig. 5 Temporal evolution of tracked liquid mass for different grid resolutions: case c01 left, c12 center, and c23 right

Journal of Engineering for Gas Turbines and Power

JUNE 2010, Vol. 132 / 061506-5

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 03/25/2014 Terms of Use: http://asme.org/terms

Fig. 7 Side view snapshots of jet in crossflow atomization at t = 5, 10, 15, 20, 25, and 28.1 time units top to
bottom; case c01 left, c12 center, and c23 right

change event is slightly delayed because the level set inherent

breakup length scale being equal to the grid size is reduced.
Moreover, the higher resolution cases generate significantly more
small scale drops.
061506-6 / Vol. 132, JUNE 2010

Besides mean jet penetration and observed breakup modes, the

key quantity to ascertain the quality of a numerical simulation is
the grid dependency/independency of the resulting atomized drop
size distribution. Figure 10 depicts the calculated drop size distriTransactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 03/25/2014 Terms of Use: http://asme.org/terms

Fig. 8 Front view snapshots of jet in crossflow atomization at t = 5, 10, 15, 20, 25, and 28.1 time units top to
bottom; case c01 left, c12 center, and c23 right

butions obtained from the primary atomization at t 20 for the

three different grid resolutions analyzed in this study. Shown are
only those drops that are generated directly from the liquid jet,
i.e., drops that are transferred from the level set tracked phase
interface representation to the Lagrangian point particle descripJournal of Engineering for Gas Turbines and Power

tion. Note that these drops can continue to atomize via secondary
atomization mechanisms further downstream; however, within the
computational domain, only a limited number of these secondary
atomization events occur.
The drop size distributions in Fig. 10 were generated by binJUNE 2010, Vol. 132 / 061506-7

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 03/25/2014 Terms of Use: http://asme.org/terms

Fig. 9 Top view snapshots of jet in crossflow atomization at t = 5, 10, 15, 20, 25, and 28.1 time units top to
bottom; case c01 left, c12 center, and c23 right

ning the data into 20 bins of equal size in terms of logD. The
approximate total number of drops used to calculate each distribution was 40,000 c01, 176,000 c12, and 193,000 c23. To be
able to compare the three probability density functions pdfs, the
061506-8 / Vol. 132, JUNE 2010

coarser grid pdfs were normalized using a single bin of the finest
grid pdf. Also shown in Fig. 10 is a log-normal fit to the larger
drop sizes of the finest grid results. All three cases show a similar
behavior. Larger drop sizes collapse well to the log-normal fit,
Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 03/25/2014 Terms of Use: http://asme.org/terms

c23

c12
c01

d
Fig. 10 Grid convergence study of pfd of drop diameters d
generated directly by primary atomization: case c01 open
circles, c12 open triangles, and c23 solid squares. Solid
line is log-normal fit to large drop sizes on finest grid.

however, for each grid there is a distinct departure point from the
fit from which point on smaller drop sizes do not collapse and fail
to match the log-normal fit. This departure point d p decreases in
value proportional to the employed grid size, with d p = 1.6xG for
the coarse and medium grids and d p = 1.3xG for the fine grid. The
key result of Fig. 10 is the fact that larger drops, i.e., those resolved by at least 1.6 G-grid cells, collapse to a single pdf, in this
case a log-normal distribution independent of the employed grid
resolution. These drop sizes can thus be considered grid independent. Drops smaller than about d = 1.6xG, however, show a
strong grid dependency. The reason for this behavior is the following: fixed grid methods used to track interfaces, like the level
set method used here, have an inherent topology change length
scale that is proportional to the local grid size. As soon as two
front segments enter the same grid cell, a topology change event is
automatically triggered. The exact moment of breakup is thus always a function of the employed grid size. However, for larger
drops, the error introduced by missing the exact moment of
breakup due to the inherent grid size dependency is small compared with the generated drop size. Thus larger drop sizes are grid
independent, as demonstrated in Fig. 10. For smaller drops, the
exact moment of breakup is however the dominant source of error; thus, smaller drops are dominated by the employed grid resolution. The cut-off appears to occur at drops resolved by about
1.6 G-grid cells. This number is surprisingly low and appears to
decrease with increasing grid resolution. This seems to indicate
that the numerically induced topology change starts to mimic the
physical breakup mechanism on small scales.

10

Conclusion

Detailed simulation results of the primary atomization of a turbulent liquid jet injected into a subsonic gaseous crossflow previously analyzed experimentally by Brown and McDonell 2 have
been presented. The simulations match all key nondimensional
numbers, except for the density ratio, which is artificially reduced
to study its impact on jet penetration and atomization.
A preliminary qualitative analysis of the simulation results indicates that breakup of the liquid jet occurs via two main simultaneous atomization mechanisms. In the first, instability waves on
the liquid column, predominantly visible on the windward side,
generate role-ups and continue to grow along the jet axis until
they form baglike structures that rupture resulting in a broad range
of drop sizes, not unlike the column-breakup mode. It is speculated that this instability mode is due to a KelvinHelmholtz instability. In the second mode, corrugations on the liquid jet surface
are stretched out into ligaments at the sides of the liquid jet near
Journal of Engineering for Gas Turbines and Power

the injector exit. The stretched out ligaments then break up and
form a range of drop sizes. At the current stage, the above description of the breakup mechanisms is speculative and still requires a
detailed, yet to be performed, quantitative study of the generated
time dependent simulation data.
Grid independent predictions of the jet penetration in agreement
with experimental observations can be achieved on relatively
coarse grids D / x = 16. Moreover, the large scale instability
mode conjectured to be responsible for the first atomization
mechanism is sufficiently resolved even on these coarse grids.
Analysis of the temporal evolution of the pre-atomization tracked
mass indicates that it is important to resolve turbulent eddies at
least up to the maximum flow solver resolution considered
D / x = 64, since these turbulent eddies can generate significant
surface corrugations to initiate the second atomization mechanism. The grid independent prediction of atomized drop sizes requires significant computational resources. Grid independent drop
sizes on the finest grids have been achieved for drops as small as
13 m, or one-hundredth of the injector exit diameter.
While the reduced density ratio analyzed in this paper appears
to have no significant impact on the jet penetration determined
from averaged side views of the jet geometry, it is conceivable
that the atomization mechanism and resulting drop sizes are a
function of the chosen density ratio. Currently ongoing simulations employing higher density ratios will focus on this question.

Acknowledgment
This work was supported in part by CASCADE Technologies
Inc. under the NavAir SBIR Project No. N07-046. The author
would like to thank S. Hajiloo and Y. Khalighi for valuable help in
generating the fluid solver meshes, and F. Ham for many valuable
discussions concerning the flow solver CDP/CHARLES.

Nomenclature
d
dp
D
G
H
n
p
q
r
Re
u
t
T
V
VD
We
xf
x
x
xG
y

Greek Symbols






Subscripts

drop diameter
minimum grid independent drop diameter
injector diameter
level set scalar
Heaviside function
phase interface normal vector
pressure
momentum flux ratio
radius
Reynolds number
velocity
time
surface tension force
volume
drop volume
Weber number
location of phase interface
crossflow direction
flow solver grid size
level set grid size
jet direction
Dirac delta function
phase interface curvature
dynamic viscosity
volume fraction
density
surface tension coefficient

1 fluid 1, i.e., liquid

2 fluid 2, i.e., gas
c crossflow
JUNE 2010, Vol. 132 / 061506-9

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 03/25/2014 Terms of Use: http://asme.org/terms

cv control volume
f control volume face
j jet

References
1 Aalburg, C., van Leer, B., Faeth, G. M., and Sallam, K. A., 2005, Properties
of Nonturbulent Round Liquid Jets in Uniform Gaseous Cross Flows, Atomization Sprays, 153, pp. 271294.
2 Brown, C. T., and McDonell, V. G., 2006, Near Field Behavior of a Liquid Jet
in a Crossflow, ILASS Americas 19th Annual Conference on Liquid Atomization and Spray Systems.
3 Brown, C. T., Mondragon, U. M., and McDonell, V. G., 2007, Investigation
of the Effect of Injector Discharge Coefficient on Penetration of a Plain Liquid
Jet Into a Subsonic Crossflow, ILASS Americas 20th Annual Conference on
Liquid Atomization and Spray Systems, ILASS Americas, Chicago, IL.
4 Stenzler, J. N., Lee, J. G., Santavicca, D. A., and Lee, W., 2006, Penetration
of Liquid Jets in a Cross-Flow, Atomization Sprays, 16, pp. 887906.
5 Mazallon, J., Dai, Z., and Faeth, G. M., 1999, Primary Breakup of Nonturbulent Round Liquid Jets in Gas Crossflows, Atomization Sprays, 93, pp.
291312.
6 Sallam, K. A., Aalburg, C., and Faeth, G. M., 2004, Breakup of Round
Nonturbulent Liquid Jets in Gaseous Crossflow, AIAA J., 4212, pp. 2529
2540.
7 Fuller, R. P., Wu, P.-K., Kirkendall, K. A., and Nejad, A. S., 2000, Effects of
Injection Angle on Atomization of Liquid Jets in Transverse Ai, AIAA J.,
381, pp. 6472.
8 Wu, P. K., Kirkendall, K. A., Fuller, R. P., and Nejad, A. S., 1997, Breakup
Processes of Liquid Jets in Subsonic Crossflows, J. Propul. Power, 131, pp.
6473.
9 Lee, K., Aalburg, C., Diez, F. J., Faeth, G. M., and Sallam, K. A., 2007,
Primary Breakup of Turbulent Round Liquid Jets in Uniform Crossflows,
AIAA J., 458, pp. 19071916.
10 Madabhushi, R. K., 2003, A Model for Numerical Simulation of Breakup of
a Liquid Jet in Crossflow, Atomization Sprays, 134, pp. 413424.
11 Arienti, M., Madabhusi, R. K., Slooten, P. R. V., and Soteriou, M. C., 2005,
Numerical Simulation of Liquid Jet Characteristics in a Gaseous Crossflow,
ILASS Americas 18th Annual Conference on Liquid Atomization and Spray
Systems.
12 de Villiers, E., Gosman, A. D., and Weller, H. G., 2004, Large Eddy Simulation of Primary Diesel Spray Atomization, SAE Technical Report No. 200401-0100.
13 Menard, T., Beau, P. A., Tanguy, S., Demoulin, F. X., and Berlemont, A.,
2006, Primary Break Up Modeling, Part A: Dns, a Tool to Explore Primary
Break Up, ICLASS-2006, pp. 17, Paper No. ICLASS06-034.
14 Menard, T., Tanguy, S., and Berlemont, A., 2007, Coupling Level Set/Vof/

061506-10 / Vol. 132, JUNE 2010

15
16
17
18
19

20
21

22
23
24
25
26
27
28
29

Ghost Fluid Methods: Validation and Application to 3d Simulation of the Primary Break-Up of a Liquid Jet, Int. J. Multiph. Flow, 335, pp. 510524.
Bianchi, G. M., Pelloni, P., Toninel, S., Scardovelli, R., Leboissetier, A., and
Zaleski, S., 2007, 3d Large Scale Simulation of the High-Speed Liquid Jet
Atomization, SAE Technical Paper No. 2007-01-0244.
Desjardins, O., Moureau, V., and Pitsch, H., 2008, An Accurate Conservative
Level Set/Ghost Fluid Method for Simulating Turbulent Atomization, J. Comput. Phys., 22718, pp. 83958416.
Gorokhovski, M., and Herrmann, M., 2008, Modeling Primary Atomization,
Annu. Rev. Fluid Mech., 401, pp. 343366.
Herrmann, M., 2008, A Balanced Force Refined Level Set Grid Method for
Two-Phase Flows on Unstructured Flow Solver Grids, J. Comput. Phys.,
2274, pp. 26742706.
Herrmann, M., and Gorokhovski, M., 2008, An Outline of a LES Subgrid
Model for Liquid/Gas Phase Interface Dynamics, Proceedings of the 2008
CTR Summer Program, Center for Turbulence Research, Stanford University,
pp. 171181.
Herrmann, M., 2007, On Simulating Primary Atomization Using the Refined
Level Set Grid Method, ILASS Americas 20th Annual Conference on Liquid
Atomization and Spray Systems, ILASS Americas, Chicago, IL.
Francois, M. M., Cummins, S. J., Dendy, E. D., Kothe, D. B., Sicilian, J. M.,
and Williams, M. W., 2006, A Balanced-Force Algorithm for Continuous and
Sharp Interfacial Surface Tension Models Within a Volume Tracking Framework, J. Comput. Phys., 213, pp. 141173.
Brackbill, J. U., Kothe, D. B., and Zemach, C., 1992, A Continuum Method
for Modeling Surface Tension, J. Comput. Phys., 100, pp. 335354.
Herrmann, M., 2006, Simulating Two-Phase Flows Using the Refined Level
Set Grid Method, ILASS Americas 19th Annual Conference on Liquid Atomization and Spray Systems, ILASS Americas, Toronto, Canada.
Kim, D., Desjardins, O., Herrmann, M., and Moin, P., 2007, The Primary
Breakup of a Round Liquid Jet by a Coaxial Flow of Gas, ILASS Americas
20th Annual Conference on Liquid Atomization and Spray Systems.
Apte, S. V., Gorokhovski, M., and Moin, P., 2003, Les of Atomizing Spray
With Stochastic Modeling of Secondary Breakup, Int. J. Multiphase Flow,
299, pp. 15031522.
Moin, P., and Apte, S. V., 2006, Large-Eddy Simulation of Realistic Gas
Turbine Combustors, AIAA J., 444, pp. 698708.
Jiang, G.-S., and Peng, D., 2000, Weighted ENO Schemes for HamiltonJacobi Equations, SIAM J. Sci. Comput. USA, 216, pp. 21262143.
Shu, C. W., 1988, Total-Variation-Diminishing Time Discretization, SIAM
Soc. Ind. Appl. Math. J. Sci. Stat. Comput., 96, pp. 10731084.
Alonso, J. J., Hahn, S., Ham, F., Herrmann, M., Iaccarino, G., Kalitzin, G.,
LeGresley, P., Mattsson, K., Medic, G., Moin, P., Pitsch, H., Schluter, J.,
Svard, M., der Weide, E. V., You, D., and Wu, X., 2006, CHIMPS: A HighPerformance Scalable Module for Multi-Physics Simulation, 42nd AIAA/
ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, AIAA Paper No.
2006-5274.

Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 03/25/2014 Terms of Use: http://asme.org/terms