Liquid Column Deformation and Particle Size Distribution in Gas Atomization

A water-gas flow injected by a close coupled atomizer was studied via High Speed Photography and Phase Doppler Anemometry. The formation of a wave disturbance on the surface of the water column was confirmed. The flow converged within an area approximately 3 mm in diameter, independent of atomization conditions. The particle size distribution across the spray suggested a trend of decreasing particle sizes and particle velocities with increasing distance from the spray axis of symmetry.


Introduction
During close coupled atomization, a liquid column or sheet is perturbed by a high velocity gas flow and is broken up into droplets, in a two stage process.In the first stage, that of primary atomization [1], the surface of the melt is disturbed by a sinusoidal oscillation [2] and is subsequently broken up into large drops or unstable bodies, the ligaments [3].During the subsequent stage of secondary atomization, the drops/ligaments may further disintegrate in flight, either via a low-turbulence mechanism [4] or in a more chaotic high-turbulence stripping fashion [5].The principle of gas atomization is shown in Figure 1.In spray forming, atomization of a molten metal or alloy causes rapid solidification of the drops in flight.The spray's subsequent impingement on a substrate produces a spray cast of varying microstructure.It is in fact the localized size distribution of particle diameters inside the spray, which dictates the spray cast microstructure and mechanical properties.In turn, local size distributions depend on the break up mechanisms.The latter, have received considerable attention in earlier phenomenological studies [4][5][6][7][8][9][10][11][12][13][14] in respect to atomization parameters -such as nature of the gas and melt phase, gas injection pressures and melt superheat.More recently, experimental treatises of atomizing geometries have been presented [15][16].Liquid break up phenomena, however -although described in the macro scale early on (e.g.[17][18][19][20]) -have not been reflected on rigorous modeling implementations.Modern atomization modeling ap- pears to be focusing on CPU-intensive stochastic simulation of the liquid jet and primary atomization in terms of Reynolds-averaged Navier-Stokes mixing (e.g.[21]).Recently, the more realistic cases of turbulent atomization conditions have been addressed -e.g. by CFD (see [22][23][24]) and integrated models [15] have been proposed.
The current study investigates the initial stage of turbulent mixing in a close coupled atomizer, which is assumed to take place within a finite convergence region; this region constitutes a crucial subtlety of a flexible mathematical model for the atomization of liquid metals already presented elsewhere [25].The model -covering both primary and secondary atomization -is applicable to any liquid//gas system and is based on the formation of sinusoidal traveling waves along the surface of a liquid [26,27].Estimation of the convergence region diameter is of great importance to modeling of the gas flow [28], as it determines the Mach number, static temperature and sonic velocities of the gas inducing break up of the liquid column.

Experimental Procedure
A cross section of the close coupled assembly used in this study is shown in Figure 2. The atomizer consisted of 20 gas jets arranged in a ring configuration.Each jet outlet diameter was 0.75 mm and its each inclination from the vertical direction was 20˚.

High Speed Photography
The behavior of a water column perturbed by Nitrogen and Helium gases was studied.The choice of water as the atomized medium was due to its low viscosity, which in turn was expected to lead to the formation of larger surface wave amplitudes for a given gas velocity, as outlined in [25].An Imacon 790 high speed camera fitted with a Nikon micro-Nikkor 55 mm lens was used; the camera was capable of speeds ranging between 104 and 107 frames per second.An intermediate tube of 21 mm between the lens and the aperture offered a fixed magnification of × 1.5.The diameter of the water column was either 2 or 3 mm.The experiments were conducted in ambient pressure (0.1 MPa) and temperature (17˚C).The high speed frames presented in this study are based on original photographs in which the contrast between the actual water column and the background has been enhanced by means of response curve filtering.The experimental assembly used in the high speed photography studies is shown in Figure 3.

Phase Doppler Anemometry (PDA)
The dynamic history of moving water particles during atomization was studied by a Dantec Particle Dynamics Analyser.The system is based on the Phase Doppler principle for non-intrusive real time measurements of a wide range of particle sizes.An Ar-ion laser with a maximum output of 5 W was employed, capable of measuring particles in the range of 1-1000 μm over 1.2 m away from the source with an error of 4%.The maximum measurable velocity was 500 m/s with an error of 1%.The output included the mean and turbulent components of particle velocities in the downstream and radial direction of the flow, the mass flux inside the measurement volume and a number of characteristic mean diameters such as the D 32 (Sauter) particle size.The disintegration of a water column 3 mm in diameter atomized by Nitrogen, Argon and Helium gases at a pressure of 100 psi (0.68 MPa) was studied.A fixed 70˚ angle was main-  tained between the laser source and the detector.A fixed horizontal spacing of 600 mm was kept between the detector and the point of convergence of the individual laser beams.The Phase Doppler apparatus used in this study is shown in Figure 4.

5(a) and (b)
give supporting evidence.In the case of water issuing from the delivery tube at a velocity of 7 m/s atomized by Nitrogen at 20 psi (0.14 MPa), the conical spray jet is formed at the tip of the tube -see Figure 5(a) -while at a higher water velocity of 13 m/s -e.g. Figure 5(b) -there is an unbroken core of water 6 mm in length before the formation of the spray cone.At sufficiently high melt velocities and at relatively low gas pressures the column exhibited a tendency for sinusoidal antisymmetric oscillations, as shown in Figure 5(c).

High Speed Photography
The flow was studied between the tip of the melt tube and approximately 12 mm below the melt tube.
Superimposed on the antisymmetric mode, oscillations of the symmetric type, as shown in Figure 5(d) gave rise to the formation of crests which normally led to the detachment of fragments from the disturbed column surface, shown in Figure 5(e).The symmetric instability amplitudes were an order of magnitude smaller than those of the antisymmetric type.An increase in the diameter of the water column caused a reduction in the wavelengths disturbing the surface of the water column.This in turn led to the disintegration of the column further downstream from the point of initial atomization.
Atomization and complete disintegration of the water column was found to depend strongly on the velocities of the two phases.With reducing initial water velocity, break up of the column was more complete further upstream.A low water velocity amounts to a high relative velocity between the melt and gas phase and a correspondingly high growth rate of the surface disturbance [25].Higher growth rates also mean that the time required for the instability to acquire sufficiently large amplitudes is relatively large and as a result, the break up length of the jet is correspondingly decreased.Figures  Formation of a conical spray jet a certain distance below the point of convergence of the gas jets was always a predominant feature.This seemingly uniform spray jet initiated approximately 5 mm downstream from the tip of the water delivery tube for any set of experimental parameters.The angle of the jet was constant and roughly equal to 20˚ as long as turbulent conditions for the gas phase were satisfied -see Figure 5(f).In general, no crest observed reached amplitude greater than the water column radius.The diameter of the convergence region, taken to be the point at which the spray jet appeared to have the smallest diameter, was also measured on every photograph and was found to be equal to a constant value of 3 mm.This diameter was found to be independent of the radius of the water column, the type of atomizing gas and the injection pressure.

Phase Doppler Anemometry
In the water sprays examined by PDA, the radial distribution of drops 40 mm downstream from the tip of the melt tube was always found to be irregular.Figures 6(a 1. Variation of the D 32 particle size and the mean downstream particle velocity as a function of distance from the point of initial atomization are shown in Figures 7(a) and (b), for Nitrogen and Argon respectively.Comparison of the two plots suggests that the spray in its infancy contained globules of diameter 550 μm in the case of Nitrogen and 600 μm in the case of Argon.The nature of the gas did not substantially influence the products of primary atomization at this pressure, since the initial drop diameters for both flows were quite similar.It is possible that the primary particles formed during the disintegration of the melt column were even larger in diameter.This is suggested by the fact that the PDA technique cannot accurately measure drop sizes upstream a 50 mm distance from the tip of the melt tube, as Figure 7(a) indicates.Break up of the water particles was complete within 150 mm downstream of the point of initial atomization, resulting in a spray that consisted of particles 100 μm in diameter.In the case of Argon, completion of break up as a slight change in the D 32 slope could be distinguished at approximately 200 mm below the die.The overall reduction in diameters for the Nitrogen flow was 80% while in the case of Argon it was 65%.These figures, however, are by no means indicative of the atomization efficiency of the configuration, since they only serve as a comparison between the fragments of primary and secondary break up.In general, the velocity followed the inverse trend of the particle size, i.e. in the early atomization stages fragments decreased in size whilst gaining in velocity.After completion of the break up  process (e.g.200 mm in the case of water/Nitrogen, see Figure 7(a) the velocities remained constant within a limited distance and started decaying from that point downstream.Numerical data underlying to Figure 7 are presented in, Table 2.
The effect of the atomizing pressure on the particle size distribution inside the flow, in the case of Argon, is shown in Figure 8(a).D 32 decreased in all regions of the spray with increasing injection pressure of the gas phase.The primary fragments (in the centre of the spray) were not substantially affected by the change in injection pressure, while the fragments on the flow edge decreased in size.At 50 psi (0.34 MPa) the majority of the spray, lying in the outer 16o of the spray cone, was made up by particles in the region of 650 μm.The inner region of the cone, within an angle of 4o from the centre axis, contained particles approximately 50% smaller compared to the rest of the spray.At 75 psi (0.52 MPa) there was a wide variety of sizes along the radial direction, ranging from primary fragments in the centre of the flow, to the finer by 80% particles on the spray edge, the latter being finer than the ones at the same point produced at a pressure of 50 psi (0.34 MPa).At an injection pressure of 100 psi (0.69 MPa) the diameter of the larger particles in the centre of the flow was reduced while the size of the finest particles was not affected.The mean particle size and the distribution of diameters in the spray were greatly dependent on the type of the atomizing gas.A comparison of the diameters produced by Nitrogen, Argon and Helium, is shown in Figure 8(b), where the injection pressure of the gas was 50 psi (0.34 MPa).All types of gases produced similar primary fragments that covered most of the spray area.Nitrogen and Argon produced the largest particles whilst Helium yielded 25% finer particles along the central axis of the spray, compared to Nitrogen and Argon.Numerical data underlying to Figure 7 are presented in, Table 3.
The effect of the injection pressure on the distribution of particle velocities is shown in Figure 9(a), in the case of Argon.The mean components of the downstream velocities were normalized by the component measured on the theoretical centre axis of the spray for the flow generated by Argon at 100 psi (0.69 MPa).The general trend suggests that the particle velocities increase with decreasing distance from the real axis of symmetry of the spray and with increasing atomization pressure.Every 50 psi (0.34 MPa) increase in injection pressure seems to result in a 20% increase in the maximum velocity of the distribution.Figure 9(b) indicates that the type of atomizing gas does not affect the particle velocities substantially.Numerical data underlying to Figure 9 are presented in, Table 4.

Conclusions
High Speed Photography studies of the area of the spray  close to the tip of the melt delivery tube on a close-coupled atomizer for a water-gas spray, indicated that at sufficiently high melt exit velocities and at relatively low atomization gas pressures the water column was deformed by sinusoidal antisymmetric oscillations.Symmetric oscillations that were superimposed on the antisymmetric mode had amplitudes about an order of magnitude smaller than those of the antisymmetric type.No crest, formed on the surface of the water column during the process of primary break up, was observed to reach an amplitude greater than roughly half the diameter of the water column.An increase of the diameter of the water column seemed to cause a reduction in the wavelengths disturbing its surface and the subsequent breakdown of the column.The use of Helium as the atomizing medium was found to cause the disintegration of the water column further upstream compared to Nitrogen.In addition, what looked like an unbroken core of water covered by dense spray, initiating from the tip of the melt delivery tube, was always shorter for the Helium than for the Nitrogen atomization runs.The diameter of the convergence region was found to be equal to 3 mm and was not affected by the gas injection pressure or the melt flow rate.PDA measurements of the particle size and velocity in the water/gas jet indicated that the particle size decreases with increasing distance from the centre axis.Measurements of drop sizes along the centre axis of the flow indicated that break up was complete at a downstream distance of 150 to 200 mm from the die.Helium produced the finest particles and the highest particle velocity compared to Nitrogen and Argon.The radial distribution of particle size was sensitive to changes in the injection pressure of the gas but was not affected by the type of gas.As a general rule the velocities of the particles in the flow were not sensitive to the gas pressure or the nature

Figure 2 .
Figure 2. Geometry of the close coupled atomizer.

Figure 6 .
Figure 6.Radial distribution of particle size, velocity and volume flux for water atomized by: (a) N 2 at 50 psi (0.34 MPa); (b) Ar at 50 psi; (c) He at 50 psi.

Figure 7 .
Figure 7. Variation of particle velocities along the center axis for water atomized by: (a) N 2 at 100 psi (0.68 MPa); (b) Ar at 100 psi.

Figure 8 .
Figure 8.(a) Effect of the injection pressure of Ar on the radial variation of the D 32 size; (b) Effect of type of atomizing gas on the radial variation of the D 32 size.

Figure 9 .
Figure 9. (a) Effect of the injection pressure of Ar on the radial variation of particle velocity; (b) Effect of type of atomizing gas on the radial vari tion of particle velocity.a