| OCR Text |
Show Presented at: the Fifth InlernaJionai Symposium on the Applications of Laser Techniques to Fluid Mechanics. Lisbon. Portugal. July 1990 droplets 10 J.Lm or less, the droplets are being carried radially inward. This suggests that a shear-layer-like structure exists between the central recirculation zone (region 5 of Fig. 2) and the adjacent spray jet (regions 1 and 2). Small droplets follow the gas flow in this shear zone, while larger droplets fly through in a ballistic fashion. These radial data help to illustrate the utility of the phase Ooppler technique as a spray diagnostic: without the ability to discriminate velocities according to droplet size, one would conclude that the spray had a very broad distribution of radial velocities-the essential physics of the droplet transport process would not be exposed. Unlike the radial velocity, the tangential velocity data show a negative correlation with droplet size. Although the gas velocity is not known, this negative correlation implies that the gas phase has a higher tangential component than the spray and that the spray is lagging in its attempt to acquire that velocity. Although differences between adjacent size classes are not discemable, the difference between a 2 or 5 J.Lm droplet and a 40 J.Lm droplet is apparent. At z· = 25 nun, the axial velocity has begun to spread out between the size classes. Larger droplets appear to maintain about the same axial velocity as at z = 10 nun, but the smaller droplets have decelerated to the gas-phase velocity. Despite the correlation of size and axial velocity, the RMS width of the axial velocity distribution docs not change with size class. This is not true, however, for the radial and tangential velocity components. Each of these show both a correlation of mean velocity with size class (positive for radial, negative for tangential), and a systematic decrease in the width of the distribution as size increases. These trends are as one might expect-the larger droplets broaden the gas-phase velocity by virtue of the momentum they deposit in the flow, but are themselves unaffected by fluctuations in the gas-phase velocity. It is also interesting to note that, at this location. droplets up to 10 J.Lm in diameter track the mean flow fairly accurately, although the RMS is somewhat lower than that of the gas phase. This suggests that for estimation of the gas-phase velocity by LOV, seed particles as large as 10 J.I.ITl could be expected to reproduce the gas-phase mean velocities (however, the RMS velocity would be somewhat biased). At z = 50 mm, where 80% evaporation has occurred, intermittency effects become evident in the axial and radial velocity data. The small droplet velocity no longer coincides with the time-mean gas velocity, (except at the center of the spray sheath). Time-resolved phase Ooppler and LOV data show that the flow is alternating in time (intermittent) between the fast moving spray and the gas phase. Only at r = 30 nun is the spray sufficiently dense and steady that the small droplet and gas-phase statistics converge. This trend is less evident in the radial and tangential velocity components, occurring only at one or the other edges of the spray. It is not surprising that intermittency should be most evident in the axial data since the shear layer formed between the external recirculation zone and spray jet is aligned axially. As such, movement from one side of the mixing layer to the other corresponds to a large change in axial velocity with little difference in the transverse components. While some differences persist between the axial and radial spray velocities and the gas phase, by z = 50 mm the tangential velocity shows essentially no difference between phases. As noted previously at Z = 25 nun, the RMS width of the radial and tangential velocity distribution decreases with droplet size. At z = 75 and 100 nun, intermittency effects dominate the axial and radial velocity data, while the tangential velocity remains equilibrated between phases. It is interesting to note that at both axial locations the 2 J.Lm tangential velocity data show some scauer from the rest of the spray and the gas phase. Inasmuch as the statistical quality of the data are low (large confidence intervals) and these small droplets are most susceptible to experimental error (lowest signal-to-noise ratio), we believe Lhat Lhese data are not significant but are experimental artifacts. Figure 5 presents the arithmetic mean diameter (010) at each plane. As previously noted, the values of 010 reported may be overstated due to an inability to accurately account for the smallest droplets in the spray. The 10% and 90% diameters (dashed lines) of the 010 distribution are also shown, as are the locations of the internal recirculation zone (solid), 30· spray cone (dot-dot-dashed), and momentum boundary of the spray (dot-dashed). Oata reported at z = 10 mm is included for completeness, but is not considered reliable at r < 6 nun due to the high number density in this region. Three features are apparent in these data. First. there is an increase in mean droplet size at the outer edge of the spray sheath near the noule (up to z = 50 nun). This is consistent with the size-classified radial velocities in this region which showed that the large droplets were moving, on average, with higher radial velocities. Second, at the center of the spray cone (r = 0 mm), the mean droplet size becomes small, and the 010 distribution becomes narrow. The reason for this is again apparent in the size classified radial velocity data; small droplets are selectively carried toward the internal recirculation zone by the gas-phase structures in this region. The third feature of the data is that at the uppermost planes, where the spray has become dilute, the mean droplet diameter decreases with radius. This is in contrast to what has usually been reported in the literature, that mean droplet size increases with radius. Only Cameron et. al. (1988) show this same trend. The reason for this is a radial temperature gradient set up by the presence of a hot internal recirculation zone. This temperature gradient. in combination with the d2-law nature of evaporation, causes the mean droplet diameter to increase most rapidly where the gases are hottest. (Although the spray is also evaporating more quickly.) Both the present flowfield and that of Cameron and co-workers have a hot central recirculation zone. In those flows reported in the literature where this decrease of diameter with radius is not observed, it is either apparent that an internal recirculation zone does not exist (or does not extend down into the spray sheath) or the data presented are not sufficient to determine whether or not such a zone exists. The droplet size distributions of Fig. 6 help to illustrate how evaporation and transport shape the mean size characteristics of the spray. Part (a) shows the development of the distribution along the nominal 30· half-angle of the spray. The data show a steady progression from a narrow, small-droplet dominated distribution to a broad, larger-droplet distribution. Note that the apparent lack of large droplets in the size distribution near the nozzle does not mean that they do not exist, but merely that they are too infrequent relative to the smaller droplets to be apparent on this scale. Once the number of small droplets has been depleted by evaporation, the larger droplets become evident in the distribution. Part (b) of the figure shows development of the distribution as a function of radius at z = 50 nun. At the outer two radii, the distributions reflect the typical characteristics of the spray. Moving farther in however, the effect of the gas-phase in separating the droplets by size class becomes apparent. At r = 25 nun, the distribution appears similar to that from the next larger radius but with a part of the small droplet population removed. At r = 12.5 mm, this segment of the population reappears-this time as an augmented small droplet population. Again this is consistent wi th the radial droplet velocity characteristics observed at z = 10 nun in Fig. 4; the smaller droplets are selectively transported from the center of the spray sheath toward the central recirculation zone. Further evidence of this is shown by the size distribution at r = 0 mm, only the small droplets are convected irito this region. Figure 7 shows the volume flux distributions used to obtain the volume flow curve of Fig. 3. Repeatability of these profiles is about 20%. Note how rapidly evaporation occurs and that the distributions of any liquid which does persist into Lhe upper regions remam centered along the nominal 30· half-angle of the spray core. |
