MCS 7 Chia Laguna, Cagliari, Sardinia, Italy, September 11-15, 2011

ON THE BOUNDARY CONDITIONS AND SPATIAL STRUCTURE

OF TURBULENT JET FLAMES OF DILUTE SPRAYS
A. R. Masri*, J. Gounder**, W. O’Loughlin* and L. DeFina*
[Link]@[Link]
*School of Aerospace, Mechanical and Mechatronic Engineering,
The University of Sydney, NSW 2006 Australia
**Duetsches Zentrum Für Luft und Raumfahrt, Institut für Verbrennungstechnik
70569 Stuttgart Germany

Abstract
This paper discusses the spatial structure of reaction zones in turbulent flames of dilute sprays
of methanol and ethanol fuels. Additionally, the paper highlights the importance of specifying
the boundary conditions in experimental jets of dilute sprays. Measurements of axial velocity
fields at the jet exit plane show bimodal distributions for large droplets near the inner wall of
sufficiently long tubes. High speed images of Mie scattering from droplets reveal that such
bimodality is due to droplet shedding from the boundary layer that develops on the inner wall
of the pipe. This paper also reports images Laser induced fluorescence (LIF) of OH and
acetone in flames of acetone sprays as well as images of LIF-OH and formaldehyde in flames
of ethanol sprays. The information reveals an interesting evolution of the reaction zones
changing from thin to thick regions depending on the level of partial premixing that takes
place. The paper discusses the effects of increasing the jet velocity or the spray loading on the
structure of the reaction zones.

Introduction
Advancing current capabilities to compute the structure of spray flames requires the correct
representation of many physical sub-processes such as primary and secondary atomization,
coalescence and dispersion, droplet evaporation and interaction with the surrounding
turbulence and heat release, as well as droplet combustion [1-11]. With dilute sprays, which
are the subject of this paper, the latter two processes are more dominant and the question of
how reaction zones evolve within the droplet fields becomes one of prime importance [11-
16]. Whether combustion occurs around single droplets or droplets clouds depends on many
factors related to the properties of the fuel as well as the local flow conditions. The Group
Combustion number is one parameter devised to characterize such demarcation [17-20]. It
should also be noted that the local flame structure is directly affected by the stratification or
partial premixing which in turn is controlled by the physical properties of the evaporating
droplets as well as the local turbulence.

Another key issue that controls the accuracy of calculating the structure of spray flows is the
adoption of correct boundary conditions. This is an essential requirement for model burners
in general if these are to be useful as benchmarks for the development of physical sub-models
and the validation of computational approaches. The relevance of this condition has been
clearly demonstrated and emphasized in the burners employed by the turbulent non-premixed
workshops on measurements and computations of turbulent non-premixed flames (TNF) [21,
22]. With spray burners, satisfying the requirement of well-known boundary conditions is
more complex than in gaseous flows. Injecting sprays directly into the flow, while more
convenient is clearly not preferred by modelers due to uncertainties in representing the droplet
fluxes and size distributions. Forming the spray early and guiding it into the experiment
provides better control of these parameters at the initial condition of the experiment.
However, this mode of injection causes some concern [23, 24] particularly if the droplets are
guided through a pipe as is the case with the burner studied here.

This paper sheds light on both issues discussed above, namely combustion of dilute sprays
and the related boundary conditions. Planar imaging of selected scalars using Laser Induced
Fluorescence (LIF) is applied to a range of dilute spray flames of acetone and ethanol fuels.
The imaged scalars, albeit qualitative, are assumed to be indicative of the spatial structure of
the reaction zones and these are presented under different conditions of changing carrier
velocity and spray loading. A feature of this burner is that the spray is formed upstream and is
carried to the exit plane of the pipe using a co-flowing stream of air (or nitrogen). The
interaction of the droplets with the pipe during transit causes an issue at the exit plane near the
pipe wall where the measured velocity of larger droplets tends to be bimodal and hence,
resulting in apparent high rms fluctuations. This anomaly is further scrutinized in this paper
using a variety of pipe lengths and diameters to guide the spray-air mixture.

Experimental
Burner and flames:
Full details of the piloted spray burner can be found elsewhere [24] and only a brief
description is given here. Spray is generated using a Sono-Tek ultrasonic nebulizer located
215 mm upstream and is advected to the burner’s exit plane with a co-flowing stream of
carrier air. The central fuel tube, with an inner diameter of 10.5 mm is surrounded by an
annular pilot flame holder, which is 25mm in diameter and has 72 holes concentrically
aligned at 7.0 mm, 9.0 mm and 11 mm from the centre. Each row contains 24 holes with
diameters 0.9 mm, 1.0 mm and 1.1 mm respectively. A co-flow of diameter 104 mm
surrounds the burner and the co-flow/burner assembly is mounted in a vertical wind tunnel.
The tunnel exit has a cross section of 290 x 290 mm. The exit plane of the co-flow and nozzle
are located 59.0 mm downstream of the exit plane of the wind tunnel. The co-flowing air
velocity is fixed at 4.5 m/s, and the pilot unburnt velocity is fixed at 1.5 m/s with an
equivalence ratio of 1.0 so that the main controlling parameters for the flames are the mass
flow rates of liquid fuel and the bulk velocity of the carrier fluid.

For each of the acetone and ethanol fuels (referred to as AcF and EtF, respectively), six cases
are chosen to study the effects of increasing the carrier velocity at a fixed liquid flow rate
(referred to as sequence 1, 2, 5 and 7) or the effects of increasing the liquid fuel flow rate for a
fixed carrier velocity (referred to as sequence 4, 3, and 1). To facilitate comparison, cases
with the same numerical reference (such as AcF1 and EtF1) have similar mass flow rates for
both carrier gas and liquid fuel. Three liquid fuel mass flow rates are used, namely 23.4, 45
and 75 g/min and these are referred as “Low”, “Mid” and “High” respectively. Also, four
carrier velocities of 24, 36, 48 and 60 m/s (corresponding respectively to carrier mass flow
rates of 150, 225, 301 and 376 g/min) are employed as shown in Tables 1 and 2. It should be
noted that air has been used as carrier for all the cases studied in this paper.

Joint LIF-OH-acetone is performed in acetone flames. The Q1(6) line of the (0,1) vibrational
band of the X 2П-A2Σ+ transition at 283.9 nm is used as a single laser source to
simultaneously excite LIF-OH and acetone. The laser pulse energy was measured to be 4 mJ
and the laser beam was focused into a 150 micron thick sheet using a 300 mm focal length
cylindrical lens. Joint LIF-OH-CH2O and Mie scattering was imaged in ethanol flames. The
set-up for LIF-OH remains unchanged as described above for acetone. Formaldehyde (CH2O)
was excited in 410 vibrational transition in the Ã1A2- A1 band near 353.17 nm and this same
line was used to collect Mie scattering from ethanol droplets at 353.17 nm. LIF-OH-acetone
and LIF-OH-CH2O was collected on two intensified Flowmaster CCD cameras located on
opposite sides of the flame. The LIF-OH signal is filtered using a high pass WG-295 filter and
an interference filter centred at 310 nm with a 10 nm bandwidth while broadband LIF-acetone
is filtered using a Schott glass filter GG-400 and low pass filter SPF-450. The broadband LIF-
CH2O signal is filtered using a Schott glass filter GG-400 and low pass filter SPF-450. The
droplet Mie signal is not intensified and passes through an interference filter centred at 355
nm with a bandwidth of 10 nm.

Table 1 Initial conditions for the acetone flames.

Reacting Spray Cases - Acetone AcF 1 AcF 2 AcF 3 AcF 4 AcF 5 AcF 7
Bulk Jet Velocity Ujet (m/s) 24 36 24 24 48 60
Carrier mass flow rate (g/min) 150 225 150 150 301 376
Liquid fuel injection rate (g/min) 75 75 45 23.4 75 75
Measured liquid flow at exit (g/min) 18.0 23.9 15.9 7.2 27.8 31.1
Vapor fuel flow at jet exit (g/min) 57.0 51.1 29.1 16.2 47.2 43.9
Overall equivalence ratio, Φoverall 4.7 3.2 2.9 1.5 2.4 1.9
Equivalence ratio at jet exit, Φexit 3.6 2.2 1.8 1.0 1.5 1.1

Table 2 Initial conditions for the ethanol flames.

Reacting Spray Cases - Ethanol EtF 1 EtF 2 EtF 3 EtF 4 EtF 5 EtF 7
Bulk Jet Velocity Ujet (m/s) 24 36 24 24 48 60
Carrier mass flow rate (g/min) 150 225 150 150 301 376
Liquid fuel injection rate (g/min) 75 75 45 23.4 75 75
Measured liquid flow at exit (g/min) 45.7 66.6 30.7 14.5 70.1 73.0
Vapor fuel flow at jet exit (g/min) 29.3 8.4 14.3 8.9 4.9 2.0
Overall equivalence ratio, Φoverall 4.7 3.2 2.9 1.5 2.4 1.9
Equivalence ratio at jet exit, Φexit 1.75 0.34 0.85 0.53 0.15 0.05

While it is known that interference from PAH molecules affects LIF-CH2O particularly in
fuel rich conditions, the question here is whether such effects are significant in the regions of
heat release so that the product CH2OxOH is corrupted by PAH. Simple laminar flame
calculations using a detailed mechanism that includes some PAH species confirm that there is
no overlap between the mass fraction profiles of the representative PAH molecules and the
rate of heat release represented by the product of CH2OxOH. Hence, even though PAH
interferes with CH2O in rich mixtures (ξ > 0.1), the region where heat release is significant as
marked by CH2OxOH, is generally free from such interferences.

Non-reacting studies of pipe effects:

In order to study the effects of the fuel pipe guiding the spray to the burner’s tip, six different
pipe configurations are investigated using non-reacting sprays of mineral turpentine. Three
configurations use the pipe length of 196 mm (which is adopted for the reacting cases) but
with decreasing inner diameters of 10.1 mm, 7 mm and 3.5 mm and three other configurations
for the 10.1mm diameter tube but with decreasing length of 196 mm, 96 mm and 6 mm. Two
different carrier velocities are tested, namely 24 m/s and 60 m/s as well as various different
levels of fuel loading. Table 3 shows details of the conditions studied.

The position of the ultrasonic spray atomizer with respect to the tube inlet is unchanged
throughout these tests. The atomizer is designed to produce a droplet distribution with a
Sauter mean diameter of approximately 40 μm when used with water although this may vary
with the specific fuel used. The nebuliser produces droplets that have a near zero initial
momentum such that the velocity of the droplets at the exit plane of the tube is dictated
entirely by the carrier velocity. Measurements of velocity across the exit plane of each
configuration were performed using the standard LDV/PDA method.

Table 3: Flow conditions for the various cases and pipe lengths studied.
Case A Case B Case C

Tube Diameter (mm) 10.1 7.0 3.5 10.1 7.0 3.5 10.1 7.0 3.5

Carrier air velocity

24 24 24 24 24 24 60 60 60
(m/s)
Carrier air Reynolds
15 700 10900 5400 15 700 10900 5400 39 300 27200 13600
number (ReA)
Carrier mass flow
138.8 66.7 16.7 138.8 66.7 16.7 347.9 167.1 41.8
rate (g/min)
Mass ratio of fuel/air
0.16 0.16 0.16 0.50 0.50 0.50 0.20 0.20 0.20
flow
Fuel injection rate
21.7 10.4 2.6 69.4 33.3 8.3 69.4 33.3 8.3
(g/min)

High speed imaging (5kHz) of Mie scattering from droplets was performed at the exit plane
of the six pipes studied here using a dual head Edgewave, Nd:YAG diode pumped laser. The
second harmonic output of the laser (532 nm) is formed into a vertical sheet of height 13 mm
and the two heads are configured to produce two separate overlapping beams (overlapped in
space but not time) that are passed over the exit plane through the centerline of the spray tube.
The time interval between the beams is set at 30 μs. A high speed CMOS camera (HSS6) run
in double frame mode coupled with a Komura 85mm, F#1.4 lens is used to capture the Mie
scattering from the droplets. A filter with a FWHM of 10 nm and centered at 532 nm is used
to filter out unwanted signals.

Results and Discussion

Flame structure:
Figure 1 shows samples of representative simultaneous LIF-OH-acetone images measured at
various axial stations in the five acetone flames listed in Table 1. The measurements are
presented here at upstream locations of x/D=5, 10 and 15, where x is the axial distance and D
is the spray jet diameter. The trends shown by flames AcF4, AcF3 and AcF1 represent the
effects of increasing fuel flow rate (at a fixed air carrier velocity) while the trends shown by
flames AcF1, AcF5 and AcF7 represent the effects of increasing carrier velocity (at a fixed
fuel loading). The top and middle frames of each image set represent LIF-OH and LIF-
acetone respectively and the third frame contains the superimposed image of LIF-acetone-
OH. The individual frame size is 29x15 mm. On the intensity color scale, black represents 0
and red maximum intensity, while white represents a saturated signal. LIF-acetone was
collected on a intensified camera setup in order to obtain a high signal to noise ratio from the
vapor and thus LIF intensity from the large liquid acetone droplets are saturated and are
marked by white spots on the LIF image. The droplet size on the images is considerably
larger than the actual droplet and this is due to intensifier blooming effect. Close to the jet exit
plane, all the fuel is confined to the center of the jet and is shrouded by the pilot flame. Large
amounts of fuel evaporates inside the burner and as the fuel loading increases the amount of
acetone vapor also increases creating a non-combustible rich mixture of fuel and air exiting
the nozzle. This is confirmed by the large values of Φexit reported for these flames in Table 1.

AcF 4 AcF 3 AcF 1 AcF 5 AcF 7

OH

x/D = 15
Acetone

OH
+
Acetone

OH

x/D = 10
Acetone

OH
+
Acetone

OH

x/D = 5
Acetone

OH
+
Acetone

Figure 1: Simultaneous and instantaneous LIF images of OH, acetone and acetone/OH image
combined together (OH+Acetone) measured at x/D=5, 10 and 15 in flames AcF4, AcF3,
AcF1 AcF5 and AcF7.

At x/D=5, flame AcF4 shows broad OH fields wrapping around the outer edges of the acetone
fuel almost without any separation. This is different from flames AcF3 and AcF1 which show
thin smooth OH zones with clear separation between the fuel and the reaction zone (marked
tentatively by the presence of OH). It is worth noting that increasing the carrier velocity in
flames AcF5 and AcF7 reduces this separation again such that the structure for flames AcF4
and AcF7 are similar. Further downstream at x/D=10, the concentration of liquid acetone in
the centre of the jet reduces significantly from flames AcF4 and AcF7 while the OH zones are
even broader which is more akin to premixed flames rather than non-premixed particularly at
such upstream locations. Interestingly, flames AcF3 and AcF1 show a totally opposite trend
where the fuel zones are even wider and the OH fields are thinner and pushed further away
from the jet centerline as expected for diffusion flames. At x/D = 15, the OH fields cover the
entire image for flames AcF4 and AcF7 with scattered patches of OH even on the centerline
wrapping around pockets of cold unreacted fuel that may contain droplets of acetone. Flames
AcF3 and AcF1 exhibit very little change in the overall structure at this location and show,
along with flame AcF5, thin OH zones surrounding a wide field of rich mixture of fuel and
carrier air in the center of the jet.

The pattern observed in Fig. 1 for acetone flames is interesting and can be summarized as
follows: at sufficiently low liquid fuel loading and carrier air velocity (AcF4), the gas
emerging from the jet exit plane appears to be sufficiently premixed to form broad OH
profiles and hence broad reaction zones surrounding the central fuel core which reduces in
concentration further downstream leading to further broadening in the reaction zones. As the
liquid fuel loading increases for the same carrier velocity (AcF3 and AcF1), a diffusion flame
structure is gradually resumed with relatively thin outer reaction zones clearly separated from
the central spray core. This structure persists from x/D=5 to 15. With increasing the carrier
velocity for the same fuel loading (as for AcF1, AcF5 and AcF7) the trend is reversed and a
structure not unlike that of AcF4 is observed for flame AcF7 with AcF5 being a transitional
flame from diffusion to premixed-like. It is worth noting that this picture is consistent with the
values reported in Table 1 for Φexit where the richest mixture at the jet exit plane is obtained
for flame AcF1 which takes a diffusion-like structure.

Representative, simultaneous LIF images of OH, formaldehyde (CH2O), droplet Mie

scattering and heat release, HR as determined by the product (LIF-OH x LIF-CH2O) are
presented in Fig. 2 for three axial locations in the five ethanol flames listed in Table 2. The
trends shown by flames EtF4, EtF3 and EtF1 represent the effects of increasing fuel flow rate
(at a fixed air carrier velocity) while the trends shown by flames EtF1, EtF5 and EtF7
represent the effects of increasing carrier velocity (at a fixed fuel loading). The physical size
of the individual image is 34x14 mm. The bottom image of each set (labeled HR+Droplet)
shows the droplet positions with respect to heat release regions by superimposing the HR
image on that of droplets (Mie scattering). The HR+Droplet images were regenerated, for
clarity, to have only three color codes where yellow marks the HR region, pale blue marks the
droplets and red marks the regions of overlap between HR and droplets. Mie scattering signal
from liquid droplets are saturated in order to show small as well as large droplets. Some of the
droplets are artificially large due to intensifier blooming effect. The spatial resolution of the
droplet image is 25 microns so droplets smaller than 25μm are not resolved.

At x/D=5, flame EtF4 shows a relatively broad reaction zone as represented by the OH image
which surrounds the LIF-CH2O image. The heat release zone lies on the inner edge of this OH
region as indicated by the images of HR which in turn surround the inner droplet as shown in
the image of HR+Droplet. Increasing the droplet loading for the same carrier velocity (EtF3
and EtF1 leads to the appearance of pockets of OH on the centerline of the jet and these are
separated from the outer OH profile which becomes thinner and diffusion like in flame EtF1.
This double reaction zone structure is confirmed in the images of HR for flame EtF1 at
x/D=5. It is also interesting that the HR+Droplet images show some overlap between HR and
the droplets which is not simple to interpret. One possible explanation is that droplets are
much smaller than the width of the imaging laser sheets and hence become partly embedded
within. Increasing the carrier velocity for the same droplet loading (cases EtF1, EtF5, EtF7)
causes the otherwise double reaction zones to mesh together as seen at x/D=5 in flame EtF5.
 EtF 4 EtF 3 EtF 1 EtF 5 EtF 7

OH

CH2O

x/D = 15
Droplet

HR

HR
+
Droplet

OH

CH2O

x/D = 10
Droplet

HR

HR
+
Droplet

OH

CH2O
x/D = 5

Droplet

HR

HR
+
Droplet

Figure 2: Simultaneous images of LIF OH, LIF CH2O, droplet Mie scattering Droplet), heat
release (HR=OH x CH2O) and superimposed image of HR onto droplet image (HR+Droplet)
collected at x/D=5, 10 and 15 in flames EtF4, EtF3, EtF1, EtF5 and EtF7.
This is further confirmed in flame EtF7 where a single, broad layer for heat release is
obtained at all axial locations. The similarities between the structure of flames EtF4 and EtF7
are also noted here.

Further downstream at x/D=10, the transition from a single broad region in flame EtF4 to a
double reaction structure in flames EtF3 and EtF1 is now much clearer as reflected both in the
OH as well as the HR images. Flame EtF1 shows thinner OH zones than those of flames EtF4
and EtF3. Increasing the carrier air velocity for the same liquid loading as flame EtF1 (EtF5
and EtF7) broadens and contorts the OH profiles and the double reaction zones are brought
closer together but not fully merged even in flame EtF7. It is not until x/D=15 that these
double reaction zones merge for flame EtF7 but not EtF5. It is also interesting that the
transition from single broad OH profiles to a double reaction zone occurs also at x/D=15 as
the liquid fuel loading increases from flame EtF3 to EtF1.

Evaporation of ethanol droplets inside the burner in not as significant as acetone, this is
evident from the values of Φexit shown in Table 2. The equivalence ratio at the jet exit plane
increases as the fuel loading is increased to a maximum of 1.75 for EtF1 and then decreases
with increasing carrier air velocity. Locally flammable mixtures may be formed at the jet exit
plane of flame EtF1 and this is evident from the double reaction structure imaged here.
Conversely, flames EtF4 and EtF7 have values of Φexit that are lean on average and hence
double reactions in these flames are not expected. In most flames, the regions of heat release
form on the periphery of the fuel vapor/droplet cloud but there are also instances where broad
HR regions form in the center of the jet. The latter feature is more common in the lower
velocity flames EtF1 and EtF3.

Effects of pipe on spray pattern at exit plane:

Earlier measurements of axial velocity conditioned with droplet size [24] have shown that at
the exit plane of the jet, close to the pipe wall, the rms fluctuations of large droplets
(30<d<40microns) are higher than those of small droplets (less than 10 microns). This aspect
is further studied here by providing measurements of velocity at the exit plane of tubes of
different diameter and different length (different distances from the spray injection point).
Additionally, high speed images of Mie scattering from droplets at the exit plane of selected
pipes are also presented.

Figure 3 shows scatter plots of axial velocity versus droplet diameter measured at the jet exit
and near the wall of pipes of various lengths and diameters. Also shown next to each scatter
plot is a histogram of the axial velocity for a small range of droplet diameters located within
the vertical dashed lines shown on the adjacent scatter plots. Plots on the LHS correspond to
the large pipe diameter of 10.1 mm while those on the RHS correspond to a narrower tube of
7 mm in diameter. The top row of scatter plots and histograms are taken at the exit plane of
the 196 mm long tube while the middle and lower rows correspond to shorter tubes of 96 mm
and 6 mm, respectively (note that the middle illustration shows distances along a single pipe
but physically refers to pipes of different lengths and diameters). The results shown in Fig. 3
correspond to Case B (shown in Table 3) which has the highest mass ratio of liquid fuel to
carrier air.

It is evident from both the scatter plots and the histograms shown in Fig. 3 that the measured
velocities near the walls of both the 7 mm and the 10.1 mm diameter tubes exhibit significant
scatter over the entire range of droplet diameters. For the long tubes (96 mm and 196 mm),
the scatter is extreme and the distributions are bimodal for larger droplet diameters. This
bimodality is clearly seen in the histograms and it explains the high rms fluctuations measured
at these locations. For the shorter tube of 6 mm (and up to a length of 30 mm but not shown),
the bimodality is lost and the distributions are uniform.

Figure 3: Effect of tube length on bimodality of the spray at the exit plane. Diameter
histograms and diameter-velocity plots are shown for three different pipe lengths and two
different tube diameters (10.1 mm and 7.0 mm).

Figure 4: Effect of tube diameter on bimodality of the spray at the exit plane. Diameter
histograms and diameter-velocity plots are shown for three different pipe diameters at three
different radial locations on the exit plane.
It is believed that the cause of the bi-modality is a result of interactions between the spray and
the tube wall. As liquid builds up on the wall it is eventually shed off the wall hence
producing large, relatively slow moving droplets that contribute to the lower leg of the
bimodal distribution. As the tube length is reduced, less of the spray has had a chance to
impact the wall of the jet. For this reason, the amount of droplet shedding from the wall of the
jet is reduced and therefore the bimodal nature of the spray is decreased. The 6 mm tube is so
short that the spray is completely confined to the region near the center of the jet while the
wall is completely dry. This explains the mono-modal velocity distribution close to the wall
of the 6mm tube.

Figure 4 shows the effects of changing the tube diameter on the mode of the spray at the exit
plane. Scatter plots of the measured axial velocity as well as histograms for the data restricted
to within the vertical bars on the scatter plots are shown for three tube diameters at three
radial locations on the jet exit plane, namely r/R=0, r/R=0.5 and r/R=0.9 (where R is the tube
radius) from left to right. The top row shows results for the 10.1 mm diameter tube. At the
centreline of the jet (r/R=0) a perfectly monomodal distribution of droplet diameter is
observed. Also, at r/R=0.5 the distribution is also monomodal, however a strongly bi-modal
distribution is found near the jet wall, at r/R=0.9. The second row shows results for the 7.0
mm diameter tube clearly featuring a bi-modal distribution at both r/R=0.5 and r/R=0.9.
Finally, the bottom row is for the 3.5 mm tube which also shows bimodality at r/R=0.5 and
r/R=0.9. A feature for the 3.5 mm tube is that a very weak bimodality is also obtained on the
jet centreline. These results are consistent with the concept of a liquid boundary layer
developing on the inner walls leading to droplet shedding. The thickness of this layer depends
on the length of the pipe but not its diameter so when the diameter narrows the boundary layer
can extend to the pipe’s centreline. The concept of droplet shedding is further confirmed in
the high-speed imaging presented next.

Figure 5 shows representative Mie scattering images of the spray at the exit plane of the jet
for several tube lengths. For the longer tube of 196 mm, the occurrence of strings of droplet
filaments shed from the inner wall of the jet as shown in Fig. 5 is frequent. These filaments of
droplets were found to have a relatively low velocity and are almost certainly the cause of the
lower velocity leg in the bimodal scatter plots shown in Figs. 3 and [Link] frequency of the
droplet shedding was found to decrease with tube length as is evident from the images shown
for the 96 mm and 36 mm long tubes with the shortest tube of 6 mm showing no droplet
shedding. It was estimated through observation of a large number of images that the
frequency of shedding for the 196 mm tube was ~10% and this decreases to zero for the 6 mm
tube. These findings provide an interpretation of the bimodality reported in Figs. 3-4 which in
turn explains the reasons for the higher rms of fluctuations in the axial velocity measured for
large droplets near the inner walls at the exit plane of these spray jets.

Conclusions
The spatial structure of reaction zones is reported here in a range of diluted sprays of acetone
and ethanol fuels. The effects of increasing the spray loading or the jet velocity of the carrier
are examined. It is found that in acetone flames and at low spray loading the reaction zones
are broad and appear to be largely premixed. Increasing the liquid fuel loading for the same
carrier velocity promotes a diffusion flame structure with thin outer reaction zones clearly
separated from the central spray core while increasing the carrier velocity supports premixing
and a return to broad reaction zones. In ethanol flames, the trends are different due to the
different evaporation rate of the fuel. Increasing the spray loading leads to the formation of
double reaction zones with the inner zones indicating significant premixing close to the jet
centreline. This double reaction zone structure gradually merges into one as the air carrier
velocity increases.

The anomaly of the high rms of fluctuations measured for large droplets at the jet exit plane
and near the inner wall of the pipe is explained as follows: For sufficiently long pipes, a liquid
boundary layer develops on the inner wall and filaments from this layer are intermittently
shed at the exit plane. These fragment into large, slow droplets and lead to the bimodal
distribution of axial velocity measured at these locations and hence the high rms fluctuations.

L=196 mm L=36 mm

L=96 mm L=6 mm
Figure 5: High speed images showing droplet Mie Scattering at the jet exit plane for different
tube lengths of 10.1mm diameter. Note the droplet shedding for the 196 mm and 96mm tubes.

Acknowledgment
This research is supported by the Australian Research Council.

References
[1] Abani, N.,Reitz, R., Diesel emgine emissions and combustion predictions using advanced mixing
models applicable to fuel sprays. Combust. Theory Modelling, 2010. 14: p. 715-746.
[2] Bachalo, W.D., Injection, dispersion, and combustion of liquid fuels, in Proc. Combust. Inst. 1994,
The Combustion Institute.: California. p. 333-344.
[3] Bazile, R.,Stepowski, D., Measurements of vaporized and liquid fuel concentration fields in a
burning spray jet of acetone using planar laser induced fluorescence. Experiments in Fluids, 1995.
20(1): p. 1-9.
[4] Bossard, J.A.,Peck, R.E., Droplet size distribution effects in spray combustion, in Proc. Combust.
Inst. 1996, The Combustion Institute: Napoli. p. 1671-1677.
[5] DePaola, G., Mastorakos, E., Wright, Y.M.,Boulouchos, K., Diesel engine simulations with multi-
dimensional conditional moment closure. Combust. Sci. Technol., 2008. 180: p. 833-899.
[6] Faeth, G.M., Spray combustion phenomena, in Proc. Combust. Inst. 1996, The Combustion
Institute: Napoli. p. 1593-1612.
[7] Faeth, G.M., Hsiang, L.P.,Wu, P.K., Structure and breakup properties of sprays. Int. J. Multiphase
Flow, 1995. 21: p. 99–127.
[8] Jiang, X., Siamas, G.A., Jagus, K.,T.G., K., Physical modelling and advanced simulations of gas
liquid two-phase jet flows in atomization and sprays. Prog. Energy Combust. Sci., 2010. 36: p. 131-
167.
[9] Nijdam, J., Stårner, S.,Langrish, T., An experimental investigation of droplet evaporation and
coalescence in a simple jet flow. Experiments in Fluids, 2004. 37(4): p. 504-517.
[10] Raffaele Ragucci, Alessandro Bellofiore,Cavaliere, A., Breakup and breakdown of bent kerosene
jets in gas turbine conditions, in Proc. Combust. Inst. 2007, The Combustion Institute. p. 2231-2238.
[11] Sornek, R.J., Dobashi, R.,Hirano, T., Effects of turbulence on dispersion and vaporization of
droplets in spray combustion, in Proc. Combust. Inst. 2000, The Combustion Institute: Scotland. p.
1055-1062.
[12] Beck, C.H., Koch, R.,Bauer, H.J., Identification of droplet burning modes in lean, partially
prevaporized swirl-stabilized spray flames, in Proc. Combust. Inst. 2009, The Combustion Institute. p.
2195-2203.
[13] Chen, G.,Gomez, A., Dilute laminar spray diffusion flames near the transition from group
combustion to individual droplet burning. Combust. Flame, 1997. 110(3): p. 392-404.
[14] Chen, Y.-C., Starner, S.H.,Masri, A.R., A detailed experimental investigation of well-defined,
turbulent evaporating spray jets of acetone. Int. J. Multiphase Flow., 2006. 32(4): p. 389-412.
[15] Masri, A.R.,Gounder, J.D., Turbulent Spray Flames of Acetone and Ethanol Approaching
Extinction. Combustion Science and Technology, 2010. 182(4): p. 702 - 715.
[16] Nakabe, K., Mizutani, Y., Hirao, T.,Fujioka, H., An experimental study on detailed flame
structure of liquid fuel sprays with and without gaseous fuel. Combust. Flame, 1991. 84(1-2): p. 3-14.
[17] Akamatsu, F., Mizutani, Y., Katsuki, M., Tsushima, S., Cho, Y.D.,Nakabe, K., Group combustion
behavior of droplets in a premixed-spray flame. Atomization and Sprays, 1997. 7(2): p. 199-218.
[18] Akamatsu, F., Mizutani, Y., Matsuki, M., Tsushima, S.,Cho, Y.D., Group combustion behavior of
premixed spray streams and group combustion number. Nippon Kikai Gakkai Ronbunshu, B-hen,
1996. 62(596): p. 1622-1628.
[19] Chiu, H.H., Kim, H.Y.,Croke, E.J., Internal group combustion of liquid droplets, in Proc.
Combust. Inst. 1982, The Combustion Institute: Haifa. p. 971-980.
[20] Chiu, H.H.,Liu, T.M., Group combustion of liquid droplets. Combust. Sci. Technol., 1977. 17(3-
4): p. 127-42.
[21] The International Workshop on Measurement and Computation of Turbulent Nonpremixed
Flames,. Available from: [Link]
[22] Masri, A.R., Dibble, R.W.,Barlow, R.S., The Structure of Turbulent Nonpremixed Flames
Revealed by Raman-Rayleigh-LIF Measurements. Prog. Energy Combust. Sci., 1996. 22(4): p. 307-
362.
[23] Gounder, J.D.,Masri, A.R. Flow Field and Mass Flux Measurements Near the Exit Plane of Spray
Jets. in 11th Triennial International Annual Conference on Liquid Atomization and Spray Systems.
2009. Vail, Colorado, USA.
[24] Masri, A.R.,Gounder, J.D., Details and Complexities of Boundary Conditions in Turbulent
Piloted Dilute Spray Jets and Flames, in ERCOFTEC Chapter (details to be completed). 2011.