Time Function DualPDA Study of Spray Growth and Droplet Size-Velocity Profiles of Chemically Modified Tapioca Starch

This study investigated the time based spray evolution of chemically modified tapioca starch through three full cone nozzles namely FC-2, FC-3 and FC-3.5. The objective was to study the effect of temperature, viscosity and nozzle orifice diameter on spatio-temporal droplet size-velocity profiles of the sprayed material. Owing to high viscosity, the unheated spraying medium did not exhibit any breakup over time. However, pre-spray heating of the medium at 80°C caused immediate sheet breakup near nozzle exit point at injection pressure of 5 bar. The spray growth from FC-2 was relatively faster than other tested nozzles; the unstable sheet disintegrated into fully developed spray patterns after 250 ms of the injection time. Unlike FC-2, the jet from FC-3 and FC-3.5 exhibited steady breakup at early and late injection stages, however, relatively fast breakup was seen in the middle part of the spray stream. Phase Doppler Anemometry (DualPDA) data showed a linear increase in Sauter Mean Diameter (SMD) with orifice size. At 140 mm downstream, the smallest SMD of 59 μm was obtained with FC-2 nozzle followed by FC-3 (65 μm) and FC-3.5 (83 μm). The overall SMD was decreased by 44.7% with an increase in orifice diameter from 1.19 to 1.59 mm. The droplet velocity in axial direction dropped sharply with time in the range of 100 to 300 ms, thereafter showed negligible decrease over time. Unlike this, the droplet velocity in the radial direction was dropping appreciably even after 300 ms of injection time.


INTRODUCTION
Although the jet flow and atomization of Newtonian and non-Newtonian liquids have been extensively studied during last few decades, it would be extremely difficult to search sufficient literature on the time based flow and spray evolution of native or modified tapioca starch (Suzuki et al., 2005;Jin et al., 2012).Starches (amylum) are natural carbohydrate polymers and have many advantages as a raw material in the production of adhesives including biodegradability, renewability, low cost, abundance and stability.The starches are also capable of demonstrating great potential as coating materials for slow release urea coating applications (Azeem et al., 2014).However, native starches are poor in terms of their viscosity, tacking ability and workability necessary for compact, water retardant and fracture free coatings.Nevertheless, to meet the slow release standards and to qualify as a good coating material, the physico-chemical properties of the native starches can be significantly altered by reacting them with other chemicals/ stabilizers (Jin et al., 2012).
In the given case, coating properties of the tapioca starch were improved by chemically modifying it with urea in the presence of borax as cross-linker and catalyst.In response, the viscosity of the cross-linked starch was increased from 2035 to 3330 cP.The enhanced viscosity can adversely influence the fluid flow and spray coating mechanism, specially, when the sprays are being generated via airless/hydraulic nozzles.When these nozzles are used to spray complex non-Newtonian liquids containing many different chemicals (Mei and Chen, 2008;Naz et al., 2015), the resulting combination of coating medium and spraying system often fails to meet the manufacturer's specifications and the standards set for slow release coated fertilizers.Most probably, the situation arises when high viscosity non-Newtonian solutions are pushed through the spraying systems designed for Newtonian fluids.In relation to the fluidized bed height of the material to be coated; the spray parameters including the injection delay, cone angle, cone length, breakup time and droplet sizevelocity distributions should be completely understood and optimized before starting the coating process.To the author's knowledge, there are no reports on the time function spray evolution and droplet size-velocity distributions of native and modified starches, in particular the tapioca starch.Therefore extensive experimental investigations are needed to gain insight into spatio-temporal spray growth and atomization of high viscosity non-Newtonian starchy solutions (Chen et al., 2002;Naz et al., 2013).
The droplet sizes and velocities are of special interest in spray coating processes (Naz et al., 2013).Normally, the spray patterns consist of a range of droplet sizes.The analysis of such dispersed flows as a spray requires information on droplet size and velocity vector fields.The non-intrusive PDA (which is an extension of Laser Doppler Anemometry) is an effective tool for measurement of size and velocity vector of the droplets passing through a small measurement volume at a fixed point (Xie et al., 2014).Although, it is a robust measurement technique but care must be taken in both the measurement and data processing stages, especially, when it is applied to partially atomized sprays, optically dense sprays and when different modes of atomization are contributing to the size-velocity fields.When applied simultaneously, the complementary nature of high speed imaging and PDA techniques may provide a much deeper insight into the spray jet breakup.
The spray jet in early injection/pre-swirl and in developing phase is partially atomized; it contains spray elements in the form of sheets, ligaments, filaments and the droplets produced by prompt atomization (Naz et al., 2014).Therefore, PDA measurements alone only shed light on the jet disintegration process rather than directly quantifying the jet breakup and atomization processes themselves.As a consequence, there could be lack of agreement among the researchers on the jet disintegration mechanics and consequently various breakup and atomization mechanisms have been postulated in the past.In this detailed note, the slow release coating properties of tapioca starch were improved by reacting it with urea in the presence of borax catalyst for future urea coatings.The modified starch was atomized into bottom-up full cone pulsing sprays with pulse duration of 100, 200, 300 and 400 ms.The time function spray growth was visualized by using a high speed camera, whereas a DualPDA system was employed to obtain the spatio-temporal droplet sizevelocity profiles along and across the spray centerline.

Preparation of Modified Starch
Food grade tapioca starch was chemically modified with urea in the presence of borax (Na 2 B 4 O 7 .10H 2 O) catalyst.The starch was commercially available, whereas the modifiers were supplied by R and M Chemicals.The native starch was dried at 110°C for removal of all moisture contents or until no further weight changes were happened over time.Herein, 1000 ml of de-ionized water was taken in a metal container and heated at 80°C.Once the desired temperature reached, 50 g of fully dried tapioca starch was then poured into the container.For complete gelation, the solution was reacted further for 30 minutes at a heating temperature of 80°C.The solution was stirred at 600 rpm for uniform mixing.After complete gelation, the known masses of the urea (15 g) and borax (4.5 g) were added into the starch dispersion.The solution was then heated for another 3 hours at same conditions.The final solution was left to cool at room temperature.Before conducting the spray experiments, the viscosity, density and surface tension of the modified starch were evaluated as shown in Table 1.

Experimental Setup
Once the physical properties of the modified starch completely understood, it was atomized into bottom-up full cone spray patterns and characterized by using non-intrusive high speed imaging and PDA techniques.Schematic of the experimental setup used for spray generation and characterization is shown in Fig. 1.Three axi-symmetric full cone nozzles namely FC-2, FC-3 and FC-3.5 with orifice diameters 1.19, 1.39 and 1.59 mm, respectively, were used to atomize the starchy solution at elevated temperatures (20-80°C) and fixed injection pressure of 5 bar.These hydraulic nozzles were operated in airless spray mode.The spray on-off operation was directly controlled by a solenoid valve and a programmable digital time relay (SIGMA, PTC-15).The desired temperature within the feed tank and spray feed line to spray point was attained by using computer controlled liquid immersion heater and heat tracing cables.A horizontal multistage chemical pump capable of withstanding the higher temperatures was used to drive the solution through nozzles.The liquid pressure at three different localized points in the main supply line was monitored by using spring type pressure gauges.

Imaging and DualPDA Study of Spray Growth
Spatio-temporal spray growth was studied by using high speed imaging and DualPDA techniques.The spray sheet breakup was visualized with a Phantom Miro M110 colored camera.This one megapixel camera with maximum frame rate of 400,000 fps @ of 64 × 8 pixels resolution and 2 µs exposure suites well to the larger field of view applications.In the present case, this camera was grabbing 1,600 frames-per-second at full resolution of 1280 × 800 pixels and 1.6 gigapixels/second throughput.It was simultaneously transferring the captured cine files to an image grabber, where A 60 GB CineFlash and Docking station were included in the package.Different aspects of the developing sprays were photographed and studied through frame by frame processing in Phantom Camera Control (PCC) software.
A Dantec Dynamics dualPDA system was used to measure the spatio-temporal droplet size-velocity profiles in the  main spray stream.A CW Argon Ion Laser generator and optical splitters were used to generate green (U1) and blue (U2) beams with an inter-beam spacing of 60 mm.The wavelength of U1 and U2 was 514.5 nm and 488 nm, respectively.The beam diameter was set to 1.35 mm with an expander ratio of 1.A shift frequency of 40 MHz was produced by passing one of the beams from laser generator through a Bragg cell.Both beams then passed through fiber optics and transmitted to the point of measurement where they cross formed a probe volume through an interferometry fringe pattern.The fringe spacing of U1 and U2 was 6.87 µm and 6.51 µm, respectively.Focal length of the beam transmitter was 500 mm with beam intersection angle of 4.3°.
The noise and reflected light contributions were minimized by fixing the beam receiver at 30° to the forward scattered direction.The focal length of the receiving optic was also 500 mm.The transmitting and receiving optics were mounted on a computer controlled 3D traversing system and the measurements were performed at different points along the axial and radial directions downstream of the nozzle exit.
In these studies, the jet injection time was set to 100, 200, 300 and 400 ms and the jet relaxation time was fixed to 400 ms.For each injection time, PDA measurements were performed between 0 to 140 mm along the axial direction and between -39 to +39 mm along the radial direction from the stream centerline.The signals from the receiving optics were processed with a Burst Spectrum Analyzer (BSA) software and the obtained data was transferred to a data acquisition system.The acquisition time for each measurement was set to 30 seconds with maximum sample size of 5000.The radial position at which the data rate was below 10 droplets per 30 seconds was considered as the spray edge point.These edge points were used to find out the spray boundary; no PDA measurements were performed beyond these points.These settings helped estimating the accurate droplet size and velocity statistics.In the given data, PDA validation rate remained in the range of 77-83%, whereas the spherical validation rate remained in the range of 79-85%.The standard deviation calculated from the selected data was about 5% in SMD and 4% in velocity measurements.

Time Based Spray Evolution
Visual study on the developing sprays of the modified starch was conducted through a non-intrusive imaging technique.A high speed camera was used to visualize the spray evolution from the tested nozzles.Fig. 2 shows the time based spray development of unheated solution at 5 bar injection pressure.No jet breakup was seen over time from all tested nozzles.However, when solution temperature was raised to 80°C, the spray jet started to break immediately after leaving the nozzle.The spray growth from FC-2 was relatively faster than that through larger orifices.The spray sheet was changed into fully developed spray patterns after 300 ms from start of the injection time as shown in Fig. 3.The complete breakup of the spray sheet was attributed to the pre-spray heating of the spraying medium.The unheated solution had a viscosity of 3330 cP, which was then reduced to 630 cP at 80°C.At this temperature, the inertial forces dominated the viscous and surface tension forces and eased the sheet breakup process.For less viscous fluids, the surface instabilities grow very quickly and ease the jet breakup process (Yao et al., 2012).These instabilities are normally caused by local vorticities in the flow.The process of spraying the hot liquids is one in which the spray jet is disintegrated by thermal and kinetic energies of partially evaporated liquids (Naz et al., 2013).The insertion of thermal energy into the spraying system stimulates partial evaporation of the spraying medium.When the liquid is discharged into the surrounding environment, a phase inversion takes place due to spray disintegration and the vapor phase inside the nozzle raises the level of disintegration.In comparison to highly pressurized atomization, the droplet size distributions shift to smaller diameters in thermally energized atomization even at very low injection pressures.It combines the benefits of pressure atomization and multiphase atomization, which make the jet disintegration possible form hydraulic nozzles even at moderate injection pressures (Naz and Sulaiman, 2014;Naz et al., 2015).
At solution temperature of 80°C, the surface waves caused by rotational motion in the swirl chamber quickly propagated and demonstrated both spatial and temporal growth.The amplitude of the wave oscillations on the sheet surface also started to increase and the sheet became very thin.At this point, the liquid sheet started to disintegrate into smaller droplets.Therefore this temperature was selected to study the spray breakup parameters including the injection delay, tip penetration, spray cone angle and droplet sizevelocity distributions (Dobermann, 2005;Naz et al., 2014).
The tip penetration as a function of injection time is shown in Fig. 4. It is defined as the distance of the leading edge of the spray jet from the nozzle exit for a particular time interval (Naz, 2013).It was observed that the tip penetration increased linearly with start of injection.However, the tip penetration did not vary appreciable after 300 ms from start of injection.The longest tip penetration of 489 mm was obtained with FC-3.5 nozzle followed by FC-3 (474 mm) and FC-2 (441 mm).It appears from this evidence that after 300 ms of jet injection, the variations in tip penetration were only associated with an increase in orifice diameter.The error analysis of the generated data showed overall 4.8% error in measurements made for tip penetration.
The different Weber and Reynolds numbers associated with different orifice diameters were also contributing to the variations in tip penetration (Dobermann, 2005;Naz et al., 2014).The highest Weber and Reynolds numbers were obtained for FC-2 nozzle (smallest orifice diameter) followed by FC-3 and FC-3.5 as shown in Table 2. Any increase in these parameters corresponds to a decrease in viscous and surface tension forces; the inertial forces start dominating the viscous and surface tensions forces.As a result of this, the growth rate of the instabilities on the jet surface also increases and the respective jet penetration decreases (Dumouchel, 2008).
The high values of dimensionless parameters from the smaller orifices also promote the spreading of the jet and oscillation damping during the droplets relaxation stage (Dumouchel, 2008).It reveals that the spray spreading or cone angle will increase with a decrease in orifice diameter.The spray cone angle as a function of injection time for different orifice diameters is reported in Fig. 5.It was seen that the cone angle decreased with an increase in orifice diameter, whilst increased with jet injection time.A concomitant increase in spray conge angle was reported with an increase in jet injection time from 0 to 300 ms, thereafter it exhibited no change over time.At the injection time of 300 ms, the largest spray cone angle of 64° was obtained with FC-2 nozzle followed by FC-3 (59.5°) and FC-3.5 (55.5°).The larger cone angle associated with smaller orifice reveals complete disintegration of liquid sheet into fine droplets spray with narrow size distribution.
Throughout the spray evolution, it was observed that the main spray streams diverged in the boundary ranging from55.5° to 64°.Therefore if the spray stream shifts its parameters out of this range then it would be the sign of malfunctioning or nozzle damaging.This situation was avoided by performing the spray visualization and data analysis atleast 5 times for each flow condition.The repeated measurements assured high accuracy in the results generated for the spray cone angle.The error analysis of the generated data showed overall 3.3% error in measurements made for spray cone angle.

SMD Distribution
For droplet size and velocity measurements, the pulsating sprays from the tested nozzles were scanned with PDA probe.The investigated spray area was 140 mm along the spray centerline and 39 mm towards the spray boundary on both sides of the centerline.The information on PDA measurement points and the step size used in this study can be depicted from half spray image in Fig. 6.The PDA measurement points were mapped with horizontal and vertical gridlines.The half spray image frames and corresponding droplet sizevelocity field plots of developing spray were used to interpret the spatio-temporal characteristics of the time based spray evolution from different orifices.The droplet size-velocity field plot in Fig. 6 reveals the spray growth qualitatively together with covariance Laser Doppler Anemometry data for droplet velocity vector profiles and PDA data for SMD profiles (Naz et al., 2015).The circles and vectors in sizevelocity plot reveal that the droplet diameters were centered at the measurement points on the mesh.SMD of the droplets in spray was measured using the equations: The time dependent axial SMD profiles for the sprays are shown in Fig. 7.The axial distribution reveals that SMD decreased appreciably with an increase in distance downstream of the nozzle.Near the nozzle exit point, the largest SMD of 157 µm was measured with FC-3.5 nozzle, which was reduced to 83 µm when measured at 140 mm downstream.Similarly, SMD for sprays using FC-3 and FC-2 nozzles revealed a decrease in size from 105 to 65 µm and 94 to 59 µm, respectively (Loebker and Empie, 1997;Ferreira et al., 2001).The results were in line with the previous investigations conducted by Ferreira et al. (2001) and Loebker and Empie (1997).They measured the droplet size at different positions on the spray centerline and away from the centerline towards the spray boundary.It was predicted that the droplet size would decrease along the centerline.Nevertheless, their investigations on droplet size towards the spray boundary were unlikely predicting different trend from the given work.
It was noted that the breakup regimes for sprays generated from three nozzles, under current operating conditions, were within the atomization regime specified in the Ohnesorge diagram (Deshmukh et al., 2012).In near nozzle regions, the Weber and Reynolds numbers were very high; therefore, most of the breakup was taking place close to nozzle exit.Further downstream, these dimensionless parameters decreased due to smaller droplet diameters and consequently lower relative velocities were predicted in the regions away from the nozzle exit (Payri et al., 2008;Xie et al., 2014).As a result of this, fewer breakups were noticed in the regions further downstream within the spray.Although these findings were in line with those of Deshmukh et al. (2012), a noticeable difference was noted when unheated solutions were pushed through the nozzles, where image visualization showed no jet breakup at all (Lefebvre, 1989).
The above discussed findings also confirmed a direct relationship between droplet size and orifice diameter.The SMD was increased with an increase in nozzle orifice diameter.At 140 mm downstream, the smallest SMD of 59 µm was obtained with FC-2 nozzle followed by FC-3 (65 µm) and FC-3.5 (83 µm).The overall SMD was decreased by 44.7% with an increase in orifice diameter from 1.19 to  1.59 mm.This trend was attributed to a decrease in Weber and Reynolds numbers with an increase in orifice diameter.In addition, the increased flow rates from the bigger orifices were also contributing to increase in the axial SMD.
Fig. 8 shows the variation of axial SMD with time from start of injection.The SMD is shown to drop from its initial high values to constant diameters after 300 ms from start of the injection time.Lowest viscosity and highest Weber and Reynolds numbers achieved at high temperature were assisting the fast growth of surface waves over time (Bröckel and Hahn, 2004;Naz et al., 2014).The inertial forces were dominating the viscous and surface tensions forces.The surface waves caused by rotation motion of the jet quickly propagated and started to grow in both axial and radial directions (Nienow, 1995;Bröckel and Hahn, 2004).The amplitude of the wave oscillations on the jet surface was also increased by making the jet sheet very thin and unstable.At this point, the sheet started to disintegrate into larger spray droplets (Katoshevski, 2006).These larger droplets disintegrated further into smaller droplets and finally into fine spray droplets at injection time of 300 ms.FC-2 nozzle was exhibiting relatively shorter breakup times (250-300 ms) among all the tested nozzles.
The shorter breakup times associated with FC-2 nozzle reveals fast growth of the surface waves and consequently the jet breakup at early injection stages.Differently, the jets from FC-3 and FC-3.5 were exhibiting a steady breakup at early and late injection stages.Nevertheless, relatively fast breakup was seen in the middle part of the spray streams.This behavior was attributed to a decrease in Weber and Reynolds numbers, consequently the slow growth of the surface waves (Naz et al., 2014).The viscous and surface tension forces were impeding the breakup process; the surface waves were taking time to grow and destabilize the liquid sheet (Stabile et al., 2013).In response, the liquid sheet was not thin enough to initiate fast breakup.The overall axial SMD from FC-2, FC-3 and FC-3.5 was decreased from 96.7 to 62 µm, 127 to 68 µm and 157 to 88 µm, respectively with an increase in injection time from 100 to 400 ms.Fig. 9 reports the radial SMD distribution on both sides of the spray centerline at 60, 100 and 140 mm downstream.The fast reduction in droplet size was noted in the spray from FC-2 nozzle followed by FC-3 and FC-3.5.For fixed injection time of 400 ms, the radial SMD distribution plots of the tested nozzles revealed that the overall SMD was decreased from 112 to 71 µm when measured at 60 mm downstream, from 102 to 64 µm at 100 mm downstream and from 85 to 61 µm at 140 mm downstream.In addition,   same decreasing trends were evident on both sides of the spray centerline.It rarely happens when radial SMD follows the same variational trend on both sides of the centerline (Nuyttens et al., 2009).This finding was attributed to the continuous growth of the surface waves and uniform spreading of the spray sheet (Kolybaba et al., 2003).It was a kind of evidence that the nozzles responded well to the spraying medium and worked in accordance with the system specifications.
It was also observed that SMD exhibited a decreasing trend towards the spray boundary with an increase in injection time from 100 to 400 ms.Fig. 10 reports the results of the radial SMD measurements made at different injection times.Sharp reduction in radial SMD with injection time was reported in these investigations.The size profiles of FC-2 nozzle showed a considerable reduction in the droplet diameter at early injection stages and almost constant values at later times.SMD at spray boundary was reduced from 68 to 58 µm with an increase in injection time from 100 to 400 ms.However, after 200 ms, the injection time was not significantly effecting the radial SMD distribution.For this injection time, the SMD was measured 59.8 µm which was slightly 1.8 µm after elevating the injection time from 200 to 400 ms.
The SMD from FC-3 was dropped linearly with an increase in time from 100 to 300 ms, thereafter reached a steady state.Overall SMD from FC-3 nozzle was decreased from 87 to 67 µm with an increase in time from 100 to 400 ms.However, unlike FC-2 and FC-3 nozzles, the FC-3.5 nozzle was not exhibiting constant radial SMD at any stage of the jet injection.A considerable decrease in SMD was noticed even after the longest injection time of 400 ms.The SMD was dropped from its initial high value of 103 to 82 µm after 400 ms of injection time.The reason of such decreasing trend in radial SMD was twofold (Dumouchel, 2008;Ejim et al., 2010).First, the jet breakup length increased and cone angle decreased with an increase in orifice diameter.This trend contributed to an increase in breakup time and droplet size from the bigger orifices (Katoshevski, 2006).Second, at optimized temperature and pressure conditions, the highest Weber and Reynolds numbers associated with smaller orifice (FC-2) were assisting the fast growth of the surface waves over time and consequently the complete jet disintegration within shorter time intervals (Ejim et al., 2010;Naz et al., 2014).

Droplet Velocity Distribution
The droplet velocity at different axial and radial positions on the spray mesh is plotted in Figs.11 and 12.The data for each injection time and location was plotted on the same scale so that the relative magnitudes of the droplet velocity are evident.The measured velocity was decreased sharply downstream of the nozzle exit, whereas a steady decrease was seen towards the spray boundary, on both sides of the spray centerline.
For shorter injection time periods (100 and 200 ms), the velocity downstream was not following the general jet flow (Ejim et al., 2010).However, the velocity along the radial direction was exactly following the trend predicted by (Ejim et al., 2010).Along the spray centerline, the velocity was decreased sharply even at far points from the nozzle exit (Ferreira et al., 2001).It was difficult to identify the core region, transition region and fully developed region in the flow.It reveals that the jet breakup was not completed yet; further disintegration was taking place over time and position along downstream of the nozzle exit point.
For longer time periods (300 and 400 ms), a steady decrease in velocity was seen in near nozzle (core region) and far nozzle regions (fully developed spray region), whereas the axial velocity was progressively decreasing along the centerline between these two regions (Ejim et al., 2010).A comparison of the velocity profiles from three nozzles at 400 ms of injection and different axial stations downstream is provided in Fig. 11.Highest droplet velocities were achieved with FC-2 nozzle followed by FC-3 and FC-3.5.
The overall effect of the injection time on spray breakup and velocity of the droplets from the tested nozzles is summarized in Table 3.The given study revealed that the spray breakup from FC-2 nozzle was entering the fully developed region at shorter injection time periods (Grant, 2005).The highest droplet velocities (150-76 m s -1 ) were measured in sprays generated with FC-2 nozzle followed by FC-3 (146-70 m s -1 ) and FC-3.5 (143-66 m s -1 ).
The results of the time based velocity at 140 mm downstream of the tested nozzles are compared in Fig. 12.The velocity along the spray centerline was dropping sharply with an increase in the injection time from 100 to 300 ms, thereafter showed negligible decrease over time.Along the radial direction, the velocity was dropping linearly even at 400 ms.This trend was attributed to a decrease in radial SMD outside the bed column (Grant, 2005;Ejim et al., 2010;Naz et al., 2014).The larger droplets near the spray centerline were exhibiting the higher momentums, therefore  Exist between 100-140 mm with velocity 72.6-66 m s-1 they were least influenced by the air entrainment and turbulence.They steady lose their momentum over time and consequently the velocity.In addition, the droplets motion along the spray centerline was almost unidirectional, which was changed to bidirectional by moving away from the centerline (Frye, 2005;Xie et al., 2014).This shift in direction was depressing the velocity along the radial direction.

CONCLUSIONS
In this detailed note, coating properties of the tapioca starch were improved by reacting it with urea in the presence of borax catalyst for future urea coatings.The native starch solution exhibited maximum viscosity of 2035 cP, which was increased to 3330 cP for chemically modified starch.However, the surface tension and density of the modified starch did not increase appreciably.A negligible increase in the surface tension from 68 to 70 mN m -1 and the density from 119 to 1022 kg m -3 was noted in these investigations.
The modified starch was atomized into bottom-up full cone pulsing sprays with pulse duration of 100, 200, 300 and 400 ms.The time function spray growth from the tested nozzles was visualized by using a high speed camera, whereas a DualPDA system was employed to obtain the spatio-temporal droplet size-velocity profiles along and across the spray centerline.
The imaging study of unheated solution revealed no jet breakup over time from all tested nozzles.The unheated solution exhibited 3330 cP viscosity, which was reduced to 630 cP at 80°C.At this temperature, the spray jet started to break immediately after leaving the nozzle.The spray growth from FC-2 was relatively faster than the growth from bigger orifices.The spray sheet changed into fully developed spray after 250 ms from start of injection time.FC-3 and FC-3.5 nozzles were taking relatively longer time for complete breakup.It was observed that the tip penetration increases linearly with an increase in injection time from 0 to 300 ms, thereafter it does not vary appreciable over time.The smallest tip penetration of 441 mm was associated with FC-2.After 300 ms of jet injection, the variation in tip penetration was only a function of increasing orifice diameter.Unlikely, the smallest spray cone angle of 55.5° was obtained FC-3.5 revealing relatively coarse sprays from bigger orifices.
PDA measurements showed a direct relation between droplet size and orifice diameter.The SMD was increased with an increase in nozzle orifice size.At 140 mm downstream, the smallest SMD of 59 µm was obtained with FC-2 nozzle followed by FC-3 (65 µm) and FC-3.5 (83 µm).The spray jets from FC-3 and FC-3.5 were exhibiting a steady breakup at early and late injection stages, and relatively fast breakup was seen in the middle part of the spray streams.The overall SMD was decreased by 44.7% with an increase in orifice diameter from 1.19 to 1.59 mm.The time based SMD from FC-2, FC-3 and FC-3.5 was decreased from 96.7 to 62 µm, 127 to 68 µm and 157 to 88 µm, respectively with an increase in injection time from 0 to 300 ms, thereafter reached almost constant diameters.Finally, based on the imaging and PDA studies, FC-2 nozzle was regarded as the most suitable atomizer among all the tested nozzles for spray coating applications under the given operating conditions.

Fig. 1 .
Fig. 1.Schematic of the experimental setup used for generation and characterization of hydraulic sprays.

Fig. 2 .
Fig. 2. Typical images of the spray evolution at solution temperature of 20°C.

Fig. 3 .
Fig. 3. Images of the spray evolution at solution temperature of 80ºC.

Fig. 8 .
Fig. 8. Axial SMD distribution as a function of injection time.

Fig. 9 .
Fig. 9. Comparison of radial SMD profiles of the tested nozzles at different axial stations.

Fig. 10 .
Fig. 10.Comparison of radial SMD profiles of the tested nozzles at 100, 200, 300 and 400 ms injection times and 140 mm axial distance.

Fig. 11 .
Fig. 11.Comparison of velocity profiles from the tested nozzles at 400 ms injection and different axial stations.

Fig. 12 .
Fig. 12.Comparison of the velocity profiles from the tested nozzles at 100, 200, 300 and 400 ms injection times and 140 mm axial distance.

Table 1 .
Physical parameters of modified and unmodified tapioca starch.