**3.3 Locating of the jet breakup position**

Liquid column breakups were investigated using the same pair of the injectors (sharp and round edge) shown on the Figure 7. For this purpose injectors were modified to allow installation of the fiber optic connector coaxially with the injector orifice to provide capabilities for application of the light guiding technique. Measurements were conducted at the room and elevated temperature of the crossing air flow (T=300K and 555K respectively). Two liquids (Jet-A and water) were used to extend range of possible correlation of the jet location versus non-dimensional parameter.

Figure 14 schematically shows liquid jet light guiding technique that was used for locating the column breakup point (CBP) by letting the liquid jet act as an optical fiber and transmit light through it.

Fig. 14. Experimental schematic for the liquid jet light guiding technique

Pulsed laser light was introduced from the back of the injector using an optical fiber to illuminate the liquid jet. The laser light propagates through the liquid in the injector and reaches the liquid jet column. Light coming out of the orifice undergoes total internal reflection and is guided by the liquid jet like in optical fibers. This effect is based on the fact that the critical angle for total internal reflection for the interface between the Jet-A and air is 43°. In other words, if the liquid jet column bends by over 430 abruptly, a ray of light entering the liquid jet parallel to the injector will also be refracted out of the liquid jet column in addition to being reflected. No such abrupt bends were observed in this study. This ensures that the attenuation of light intensity in the liquid jet column due to refraction is not significant enough to completely terminate the light propagating through the jet.

Slightly different jet illumination techniques were used in this study for the Jet–A and water. When the liquid used was Jet A, Metalaser Technology MTS-20 pulsed Copper Vapor laser with tunable pulse frequency (in the range of 5 kHz – 8 kHz) and a power of about 5mJ per pulse was used for illuminating the liquid jet. When water was used as the liquid for creating the spray, a Nd:YAG laser with a frequency of 10 Hz and a power of about 50mJ per pulse was used for illumination. To make the entire mass of the liquid through which light is passing visible both liquids were seeded with a fluorescent dye. The dyes used were Pyrromethene 567 with Jet A and Fluorescein with water. Both these dyes absorb the laser light and fluoresce in the yellow region. An optical filter was used to cut off the scattered light. The farthest visible point from the center of the orifice in the image is considered to be the CBP.

Figure 15-a shows a typical image of a jet in cross flow obtained by employing the liquid jet light guiding technique. This raw image was eventually inverted into a binary field shown 70 Advanced Fluid Dynamics

Liquid column breakups were investigated using the same pair of the injectors (sharp and round edge) shown on the Figure 7. For this purpose injectors were modified to allow installation of the fiber optic connector coaxially with the injector orifice to provide capabilities for application of the light guiding technique. Measurements were conducted at the room and elevated temperature of the crossing air flow (T=300K and 555K respectively). Two liquids (Jet-A and water) were used to extend range of possible correlation of the jet

Figure 14 schematically shows liquid jet light guiding technique that was used for locating the column breakup point (CBP) by letting the liquid jet act as an optical fiber and transmit

Pulsed laser light was introduced from the back of the injector using an optical fiber to illuminate the liquid jet. The laser light propagates through the liquid in the injector and reaches the liquid jet column. Light coming out of the orifice undergoes total internal reflection and is guided by the liquid jet like in optical fibers. This effect is based on the fact that the critical angle for total internal reflection for the interface between the Jet-A and air is 43°. In other words, if the liquid jet column bends by over 430 abruptly, a ray of light entering the liquid jet parallel to the injector will also be refracted out of the liquid jet column in addition to being reflected. No such abrupt bends were observed in this study. This ensures that the attenuation of light intensity in the liquid jet column due to refraction is not significant enough to completely terminate the light propagating through the jet. Slightly different jet illumination techniques were used in this study for the Jet–A and water. When the liquid used was Jet A, Metalaser Technology MTS-20 pulsed Copper Vapor laser with tunable pulse frequency (in the range of 5 kHz – 8 kHz) and a power of about 5mJ per pulse was used for illuminating the liquid jet. When water was used as the liquid for creating the spray, a Nd:YAG laser with a frequency of 10 Hz and a power of about 50mJ per pulse was used for illumination. To make the entire mass of the liquid through which light is passing visible both liquids were seeded with a fluorescent dye. The dyes used were Pyrromethene 567 with Jet A and Fluorescein with water. Both these dyes absorb the laser light and fluoresce in the yellow region. An optical filter was used to cut off the scattered light. The farthest visible point from the center of the orifice in the image

Figure 15-a shows a typical image of a jet in cross flow obtained by employing the liquid jet light guiding technique. This raw image was eventually inverted into a binary field shown

Fig. 14. Experimental schematic for the liquid jet light guiding technique

**3.3 Locating of the jet breakup position** 

location versus non-dimensional parameter.

light through it.

is considered to be the CBP.

in Figure 15-b by application of the threshold that was set to the intensity of the image which corresponds to the sharp fall in intensity of the liquid jet. The edge of this binary field was tracked to obtain the complete boundary of the liquid jet (see Figure 15-c). The farthest point on this boundary from the center of the orifice is defined as the CBP in this study. This CBP position was averaged over 150 images. Figure 15-d shows the averaged image of the liquid jet obtained using this technique with crosses indicating individual CBPs and circle indicating the average CBP location for the investigated operating conditions.

Figures 16-a and -b show the coordinates of the mean location of the CBP in the direction of fuel injection (*X*) and airflow (*Z*) downstream of the orifice respectively. Data of all four experimental series demonstrate the same effect of the CBP approximation to the orifice with the growth of momentum flux ratio (*q*). Two competing factors control position of the CBP: (1) Increase of the liquid jet velocity with the growth of *q* and (2) acceleration of the jet disintegration with the growth of the liquid velocity and thus its internal turbulence. This competition is clearly indicated by the maximum on the graph, which shows *X/d* coordinate of CBP on the Figure 16-a. This effect is much stronger for the sharp edged orifice at higher temperature of the crossing air flow. This fact supports hypothesis of the influence of internal turbulence of liquid jet upon the location of CBP because of possibility of cavitation at increased temperature of the injector internal surfaces caused by the high temperature of the crossing air.

(c) Boundary with indicated CBP (d) Averaged image and CBP location

Fig. 15. Methodology for locating the column breakup point (CBP)

Fuel Jet in Cross Flow – Experimental Study of Spray Characteristics 73

Finally, the entire set of CBP obtained in this study for various values of airflow velocities (66 – 140 m/s) and velocity of the liquid jet (19 – 40 m/s) for two liquids (Jet-A and water) at two different cross flow air temperatures was summarized in the form of non-dimensional breakup time (*tcb*, defined in equation 1), which was calculated from the experimental data with the assumption that velocity of the jet in the X direction does not change until the column breaks up. *tcb* was obtained by dividing the X distance of the column breakup point from the orifice by the jet exit velocity. Dependence of the *tcb* upon the liquid jet Reynolds number (*Rej*) is shown in the Figure 18. Non-dimensional breakup time (*tcb* ) is chosen as a parameter that is commonly used in computational models of spray formation (Wu et al., 1995). Choice of the *Re* number is self explained by the fact that only one injector diameter was used in the current study and any variations in the Weber number (*We*) and momentum flux ratio (*q*) led to strong variation of velocity of the liquid jet (19 – 40 m/s) and thus of the *Re* number. This correlation is described by Equation 2 and as shown on the Figure 17 to be

\* 9.98 0.908ln(Re ) *cb*

0 10000 20000 30000 40000 50000

**Liquid Jet Reynolds Number**

Fig. 18. Non-dimensional breakup time dependence upon the Reynolds number of liquid jet

Measurements of spray penetration were obtained using NAC GX-1 high speed camera that captured shadowgraph high definition images of the spray at the rate of 24,000fps at a resolution ~8.5 pixel/mm with a record length of about 20,000 frames. Illumination of the spray was achieved by the copper-vapor laser flashes (30ns) synchronized with the shutter openings. Laser light was introduced into the test section through the 1mm diam. quartz fiber from the laser. Collimator lens and diffusing glass plate created a uniform light beam that illuminated spray from one side through the window in the pressure vessel. Camera that was installed on the other side of the pressure vessel captured shadowgraph images of the spray. Each of several thousands images (see example on the Figure 19-a) that compose a high speed movie of the fluctuating spray was processed individually in order to characterize the outer border of the spray pattern. For this purpose the following procedure was applied:

*t*

*j*

*<sup>t</sup>* . (2)

Jet A with crossflow air at 300K

Jet A with crossflow air at 555K

Water with crossflow air at 300K

Water with crossflow air at 555K Curve Fit

valid in the Rej range of 2,700 – 45,000.

0.1

**3.4 Results of the spray penetration measurements** 

1

10

Fig. 16. Location of the column breakup point (CBP)

Figure 17 shows position of the CBP as a function of Weber (*We*) number. In fact CBP location was determined to be at about 1-4 diameters downstream of the orifice. This distance is reduced with increase of *We s*imilar to the dependence upon the momentum flux ratio in Figure 16. This occurs because an increase of We causes an increase of the fuel flow rate and thus velocity of liquid which in turn enlarge the scale of structures (see Figure 11) in the jet boundary. Presumably these larger structures accelerate process of jet disintegration by aerodynamic shearing.

Fig. 17. Typical dependence of the CBP location upon the Weber number for the round edge orifice

It is worth to note that distances at which fuel jet disintegrates in this study are much shorter compared to prior studies (Wu et al., 1997; Sallam et al., 2004) that reported the CBP to lie at a distance of 8 diameters downstream of the orifice for most of the investigated cases. This discrepancy can be attributed to the difference of operating conditions and measuring techniques used for the CBP locating.

72 Advanced Fluid Dynamics

(a) In the direction of fuel injection (b) In the direction of crossing air flow

Figure 17 shows position of the CBP as a function of Weber (*We*) number. In fact CBP location was determined to be at about 1-4 diameters downstream of the orifice. This distance is reduced with increase of *We s*imilar to the dependence upon the momentum flux ratio in Figure 16. This occurs because an increase of We causes an increase of the fuel flow rate and thus velocity of liquid which in turn enlarge the scale of structures (see Figure 11) in the jet boundary. Presumably these larger structures accelerate process of jet

> 0 500 1000 1500 **Weber Number**

Fig. 17. Typical dependence of the CBP location upon the Weber number for the round edge

It is worth to note that distances at which fuel jet disintegrates in this study are much shorter compared to prior studies (Wu et al., 1997; Sallam et al., 2004) that reported the CBP to lie at a distance of 8 diameters downstream of the orifice for most of the investigated cases. This discrepancy can be attributed to the difference of operating conditions and

0

1

1 10 100 1000

Round edged orifice (300K) Sharp edged orifice (300K) Round edged orifice (555K) Sharp edged orifice (555K) Wu et al

**Mom en tum ratio**

q = 20 with Jet A

q = 20 with Water

Sallam et al

10

**Z /d**

100

Round edged orifice (303k) Sharp edged orifice (303k) Round edged orifice (555k) Sharp edged orifice (555k) Wu et al

1

orifice

1 10 100 1000 **Mom en tum ratio**

Fig. 16. Location of the column breakup point (CBP)

disintegration by aerodynamic shearing.

10

1

measuring techniques used for the CBP locating.

10

**X /d**

100

Finally, the entire set of CBP obtained in this study for various values of airflow velocities (66 – 140 m/s) and velocity of the liquid jet (19 – 40 m/s) for two liquids (Jet-A and water) at two different cross flow air temperatures was summarized in the form of non-dimensional breakup time (*tcb*, defined in equation 1), which was calculated from the experimental data with the assumption that velocity of the jet in the X direction does not change until the column breaks up. *tcb* was obtained by dividing the X distance of the column breakup point from the orifice by the jet exit velocity. Dependence of the *tcb* upon the liquid jet Reynolds number (*Rej*) is shown in the Figure 18. Non-dimensional breakup time (*tcb* ) is chosen as a parameter that is commonly used in computational models of spray formation (Wu et al., 1995). Choice of the *Re* number is self explained by the fact that only one injector diameter was used in the current study and any variations in the Weber number (*We*) and momentum flux ratio (*q*) led to strong variation of velocity of the liquid jet (19 – 40 m/s) and thus of the *Re* number. This correlation is described by Equation 2 and as shown on the Figure 17 to be valid in the Rej range of 2,700 – 45,000.

$$\frac{t\_{cb}}{t\_{\!\!\!\!\!}} = 9.98 - 0.908 \ln(\text{Re}\_{\slash}) \,. \tag{2}$$

Fig. 18. Non-dimensional breakup time dependence upon the Reynolds number of liquid jet

#### **3.4 Results of the spray penetration measurements**

Measurements of spray penetration were obtained using NAC GX-1 high speed camera that captured shadowgraph high definition images of the spray at the rate of 24,000fps at a resolution ~8.5 pixel/mm with a record length of about 20,000 frames. Illumination of the spray was achieved by the copper-vapor laser flashes (30ns) synchronized with the shutter openings. Laser light was introduced into the test section through the 1mm diam. quartz fiber from the laser. Collimator lens and diffusing glass plate created a uniform light beam that illuminated spray from one side through the window in the pressure vessel. Camera that was installed on the other side of the pressure vessel captured shadowgraph images of the spray. Each of several thousands images (see example on the Figure 19-a) that compose a high speed movie of the fluctuating spray was processed individually in order to characterize the outer border of the spray pattern. For this purpose the following procedure was applied:

Fuel Jet in Cross Flow – Experimental Study of Spray Characteristics 75

Series of curves each representing the mean position of the spray outer border at a certain momentum flux ratio (*q-sweep*) are shown on the Fig. 21 for the sharp- and round–edged orifices. Graphs reveal strong dependence of the spray border upon the momentum flux ratio. Both series of curves follow the same trend. At the same time they indicate greater spray penetration into the cross flow (~12%) for the sharp edge orifice comparing to the

**Mean(X) , mm**

(a) Sharp edged injector (b) Round edged injector Fig. 21. Mean spray penetration into the cross flow of air at different momentum flux ratios

This difference can be attributed to the larger droplets size created by the sharp edge orifice shown on the Figures 12 and 13 and to the difference in the fuel velocity profiles reflected by the difference in flow coefficients *Cd* of the two tested injectors (see curves on the Fig. 8). Both factors are working towards higher spray penetration. In spite of the fact that the average fuel velocity discharged from the sharp edge orifice is lower than from the round edge orifice because of hydraulic losses, velocity in the center of the jet may be higher and at least some droplets will have higher momentum exclusively because of velocity difference. It is worth to note that the spray border curves obtained for both orifices converge significantly while being normalized by the *Cd*, (i.e., by the maximum velocity) and by the

Curves on the Fig. 22 were obtained by normalizing the jet penetration into the cross flow by square root of the momentum flux ratio value, *q*. All the curves obtained in a wide range of *q=*5…100 and previously shown on the Figure 21 collapsed here in one line. This fact provides a good opportunity for the approximations of the spray penetration *X* using self

**Mean(X)/sqrt.(q)**

(a) Sharp edged injector (b) Round edged injector

Fig. 22. Normalized values of the mean spray penetration into the cross flow of air at

**q = 5 q = 10 q = 20 q = 40 q = 60 q = 80**

**q = 100**

**0 5 10 15 20 25 30 35**

**Axial distance from the orifice, mm**

**0 5 10 15 20 25 30 35**

**Axial distance from the orifice, mm**

**q = 5 q = 10 q = 20 q = 40 q = 60 q = 80**

**q = 5 q = 10 q = 20 q = 40 q = 60 q = 80**

**q = 100**

**q = 100**

**q = 5 q = 10 q = 20 q = 40 q = 60 q = 80**

**q = 100**

round edged orifice.

**0 5 10 15 20 25 30 35**

**Axial distance from the orifice, mm**

explained physical dependence *X~sqrt(q) ~ Ul*.

**0 5 10 15 20 25 30 35**

**Axial distance from the orifice, mm**

different momentum flux ratios (q)

(*q=*5…100). Note: *We=1000=const.*

diameter (*D32*) of droplets.

**Mean(X)/sqrt.(q)**

**Mean(X), mm**


In the final stage of processing, standard algorithms for calculating mean and maximum values and RMS were applied to the spray border lines.

(a) Raw image (b) Binary field

All together 58 high speed movies of the spray were captured at different flow conditions that are divided into two series. In the first one (so called *We-sweep*) fuel to air momentum flux ratio was kept constant equal to *q= 20* while Weber number was changed from movie to movie. Spray movies at *We=400, 600, 800, 1000, 1200, 1400,* and *1600* were captured.

In the other series of experiments (so called *q-sweep*) Weber number was kept constant (*We=*1000) while momentum flux ratio was varied from movie to movie. In the *q-sweep* momentum ratios of *q=*5, 10, 20, 40, 60, 80, 100 were examined. *We-*sweep *and q-*sweep were performed for both sharp and round edged injectors.

Typical results of the *We-*sweep are presented on the Fig. 20 in the form of the mean positions of the spray outer boarders at different Weber numbers (see Figure 20-a) and their RMS values (Figure 20-b). It is clearly seen that the position of the spray outer edge and its RMS are practically independent of *We* number. RMS value increases almost linearly with axial position downstream the injection point. Similar result (luck of dependence on the Weber number) was obtained in the *We-sweep* performed with the round edged injector. Luck of dependence of the spray outer border on the Weber number allows significant simplification of the correlation function.

Fig. 20. Spray penetration (*X*) into the cross-flow of air at different Weber numbers (*We=400 ...1600*) for sharp edge injector

74 Advanced Fluid Dynamics

 Each image was corrected by subtraction of the averaged background. Images of the background were captured before any fuel was injected at each flow condition and then

 Dynamic range of each image was adjusted to eliminate possible influence of laser pulse intensity fluctuations (i.e. to avoid affecting the overall brightness of the image). Threshold was applied to all images in the series to equalize pixel intensity value in the spray region to unity and background region pixels to zero. The result of this conversion to a binary field is shown on the Figure 19-b. Line that divided white and

In the final stage of processing, standard algorithms for calculating mean and maximum

(a) Raw image (b) Binary field

All together 58 high speed movies of the spray were captured at different flow conditions that are divided into two series. In the first one (so called *We-sweep*) fuel to air momentum flux ratio was kept constant equal to *q= 20* while Weber number was changed from movie to

In the other series of experiments (so called *q-sweep*) Weber number was kept constant (*We=*1000) while momentum flux ratio was varied from movie to movie. In the *q-sweep* momentum ratios of *q=*5, 10, 20, 40, 60, 80, 100 were examined. *We-*sweep *and q-*sweep

Typical results of the *We-*sweep are presented on the Fig. 20 in the form of the mean positions of the spray outer boarders at different Weber numbers (see Figure 20-a) and their RMS values (Figure 20-b). It is clearly seen that the position of the spray outer edge and its RMS are practically independent of *We* number. RMS value increases almost linearly with axial position downstream the injection point. Similar result (luck of dependence on the Weber number) was obtained in the *We-sweep* performed with the round edged injector. Luck of dependence of the spray outer border on the Weber number allows significant

**RMS(X), mm**

(a) Mean values (b) RMS Fig. 20. Spray penetration (*X*) into the cross-flow of air at different Weber numbers (*We=400* 

**0**

**1**

**2**

**3**

**Axial distance from the orifice, mm**

**0 10 20 30 40 50**

movie. Spray movies at *We=400, 600, 800, 1000, 1200, 1400,* and *1600* were captured.

averaged for the experimental series to be processed.

values and RMS were applied to the spray border lines.

**45 mm Cross-flow**

black zones on the image represented outer border of the spray.

**20 mm**

Fig. 19. Procedure for characterization of the outer border of the spray

were performed for both sharp and round edged injectors.

simplification of the correlation function.

**0 10 20 30 40 50**

**Axial distance from the orifice, mm**

**ShE / We= 400 ShE / We= 600 ShE / We= 800 ShE / We= 1000 ShE / We= 1200 ShE / We= 1400 ShE / We= 1600**

**Mean(X), mm**

*...1600*) for sharp edge injector

Series of curves each representing the mean position of the spray outer border at a certain momentum flux ratio (*q-sweep*) are shown on the Fig. 21 for the sharp- and round–edged orifices. Graphs reveal strong dependence of the spray border upon the momentum flux ratio. Both series of curves follow the same trend. At the same time they indicate greater spray penetration into the cross flow (~12%) for the sharp edge orifice comparing to the round edged orifice.

Fig. 21. Mean spray penetration into the cross flow of air at different momentum flux ratios (*q=*5…100). Note: *We=1000=const.*

This difference can be attributed to the larger droplets size created by the sharp edge orifice shown on the Figures 12 and 13 and to the difference in the fuel velocity profiles reflected by the difference in flow coefficients *Cd* of the two tested injectors (see curves on the Fig. 8). Both factors are working towards higher spray penetration. In spite of the fact that the average fuel velocity discharged from the sharp edge orifice is lower than from the round edge orifice because of hydraulic losses, velocity in the center of the jet may be higher and at least some droplets will have higher momentum exclusively because of velocity difference. It is worth to note that the spray border curves obtained for both orifices converge significantly while being normalized by the *Cd*, (i.e., by the maximum velocity) and by the diameter (*D32*) of droplets.

Curves on the Fig. 22 were obtained by normalizing the jet penetration into the cross flow by square root of the momentum flux ratio value, *q*. All the curves obtained in a wide range of *q=*5…100 and previously shown on the Figure 21 collapsed here in one line. This fact provides a good opportunity for the approximations of the spray penetration *X* using self explained physical dependence *X~sqrt(q) ~ Ul*.

Fig. 22. Normalized values of the mean spray penetration into the cross flow of air at different momentum flux ratios (q)

Fuel Jet in Cross Flow – Experimental Study of Spray Characteristics 77

) d <sup>z</sup> (1 1

<sup>z</sup> ln(1

2

 

<sup>q</sup> <sup>d</sup> x

Sharp Edge *1.2181* 

into the cross flow.

**0**

penetration *X*

**0 5 10 15 20**

**Mean(X)Exper***, mm*

**5**

**10**

**Mean(X)Correl***, mm*

**15**

**20**

the sharp edged injector.

 ) <sup>d</sup>

2 1 *a a*

Penetration Average Maximum

Injector Type *a1 a2 a1 a2*

Table 1. Empirical correlation coefficients for the average and maximum spray penetration

Comparison of the experimentally measured and correlated spray penetrations *X* are presented on the Fig. 24 for the average and maximum penetration of the spray created by

**0**

**0 5 10 15 20 25**

**Max(X)Exper.***, mm*

**ShE / q = 5 ShE / q = 10 ShE / q = 20 ShE / q = 40 ShE / q = 60 ShE / q = 80 ShE / q = 100**

**5**

**10**

**Max(X)**

(a) Average penetration (b) Maximum penetration

Fig. 24. Comparison between the correlated and experimentally measured values of spray

**Correl.***, mm*

**15**

**20**

**25**

*1.8806* 

The other coefficient (*a2*) only shaped the spray border described by the logarithmic function and was independent of the injector design. Thus average and maximum spray

penetrations were correlated using coefficients *a1* and *a2* presented in the table 1.

Round Edge *1.0724 1.8641* 

**ShE / q =5 ShE / q = 10 ShE / q = 20 ShE / q = 40 ShE / q = 60 ShE / q = 80 ShE / q = 100**  

*a* (3)

*1.9866* 

*0.7403* 

Measurements of the spray border obtained in the current study using high speed imaging technique were compared with the spray border data obtained using Phase Doppler method. For this purpose the data rate measured with the PDPA is used as a metric to locate the edge of the spray. The edge of the spray is assumed to be around a region showing 10% of the maximum data rate as shown in Fig. 23-b. Figure 23-a demonstrates a good agreement between the spray trajectories obtained using statistically relevant high speed imaging technique and borders of the spray measured by the processing of the PDPA data rate. It is clearly seen that the maximum spray penetration determined as *X\*=Xmean+ 2.8RMS* is equal to the border determined at the level of 10% threshold of the PDPA data rate curve maximum.

Fig. 23. Comparison of the maximum spray penetration (i,e *X\*=Xmean+ 2.8RMS*) at *q=*20 measured by the high speed (HS) imaging technique and by the PDPA
