| OCR Text |
Show Presented aJ: the Fifth InternaJional Symposium on the Applications of Laser Techniques to Fluid Mechanics, Lisbon, Portugal, July 1990 TABLE I Furnace and phase Doppler operating parameters. rumace Stabilization: rucl: Total Air: Atomizing air: Main Air Reynolds Number: Aerodynamic, Swirl number = 0.99 Kerosene, mass flow rate = 1.400 +1- .005 g/s Mass flow rate = 30.5 +1- .2 gls (50% excess) Mass flow rate = 1.400 +1- .002 g/s Re = 25,000 Re = 20,000 T = 300 K Atomiz.ing Air Reynolds Number: Reactant Temperature: Phase Doppler Anemometer Transmitter: Variable shift, 6 )lm radial fringe spacing, 13 J.Lrn axial fringe spacing 20· off-forward, 500 mrn focal length 10000-20000 samples per location, or 100 seconds f, = 1.43 + i 0.00 187 J.Lrn cj) (lIe ) x 1 mrn length Receiver: Sampling: Refractive Index: Probe Volume: The 'phase Doppler anemometer used in this study was a twocomponent. dual-beam, Phase Doppler Particle Analyzer (PDPA). The principles of phase Doppler anemometry and their implementation in this instrument have been reJX)rted previously by B achalo and Houser (1984). The operating parameters used for this study are summarized in Table I. Measurements were made across five axial planes, 25 mm apart. starting 10 mm from the nozzle. For each of these planes, data were collected as a function of radius at a spacing of either 2 or 2.5 mm. Throughout most of the spray, 10,000 data points were collected for each location. Exceptions were made where the data rate was exceedingly high (20,000 samples were taken), and in the upper rcgions of the spray where the data rate was very low. In these laner regions. the flow was sampled for a minimwn of 100 seconds. In the densest regions of the spray (z = 10 and 25 mm), a 50 J..lm slit was employed in the receiver optics (instead of the usual 100 J..lm slit) in order to reduce the effective probe volume by a factor of two. RESULTS Before viewing the details of the gas- and condensed-phase strucrurc. it is useful to have some orientation to the mean strucrure of the flame under study. Figure 2 (Edwards and Rudoff. 1990) provides a summary of the mean structure of this flame. The left side of the figure shows the mean characteristics as evident to the observer-the dense (1) and dilute (2) spray regions and location of the flame boundary (3). The right side provides a sectioned view. illustrating the internal structure of the flame. Evident in this view are key regions which have been identified: the dense and dilute spray regions. the main air jet (4). and the internal (5) and external (6) recirculation zones. Each of the recirculation zones have been hatched to remind the reader that these are regions composed of hot vitiated gas products-regions which act as sources of both enthalpy and species upon entrainment Time-mean aerodynamics of the two phases are illustrated by the velocity vectors of the spray (dashed) and gas (solid). Each is presented alone in regions where dominance of one phase is clear. and together in regions where both con tributions are significant. The mechanism of flame stabilization is apparent in the figure. Hot. v itiated. products are drawn from downstream to the inner edge of the conical spray sheath. Entrainment into the spray provides the enthalpy required both to vaporize the spray and initiate combustion. As penetration of these entrained gases proceeds, reaction begins to occur within the (then dilute) spray region. and finall y. a gaseous diffusion flame (8) is established along the air/fuel mixing boWldary outside of the fuel jet. Inasmuch as the product gases from the name are likely to contain substantial oxygen (recall that 50% excess was supplied). an internal flame zone (7) is also thought to exist. although proof or quantification of this has not been obtained. Figure 3 shows the integrated liquid volume flowrate as a function of axial distance from the nozzle. This curve was obtained by integrating the volwne flux profLles which will be shown later (Fig. 10). but is introduced now so that the reader can have some sense as to the density of the liquid phase at each of the measurement planes. It is apparent that much of the evaporation (-2!3) occurs within 25 mm of the nozzle. and that the spray is completely evaporated by 100 mm. As apparent in the figure. most of the measurements were taken after a significant amoWlt of evaporation had occurred. Measurements were also made at z = 10 mm. but with the exception of size-classified velocity. were found to be unreliable due to the high number density in Lhis region. The aerodynamics of both phases of the flow are summarized in Fig. 4. This composite figure shows the mean and RMS of all threc velocity components as functions of radius across each of the five measurement planes. The data for each component and location has been plotted on the same scale so that the relative magnitudes of the velocities are evident. Also. the scale used for the RMS velocity fluctuation is the same as that used for the mean velocity so that. again. a direct visual comparison may be made. Data for the condensed phase has been size-classified in six bins (2. 5, 10, 20, 30, and 40 J.U11). each 1.4 J..lm in width. The number mean spray velocities are also shown. as are the gas velocities. The gas velocities reported here were determined by one of three methods. In the non-spray region (no droplets). the gas velocity was measurcd by LDY using fixed-interval sampling at intenncdiate seeding rates (-1 kHz). At the edges of the spray (dilute spray region). timeresolved LDV at high seeding rate (> 10kHz) was used to reconstruct the gas-phase velocity signal as a function of time. Spray droplets which were ballistic relative to the gas flow were easily detected by virtue of their deviation from an otherwise continuous baseline. These droplets were removed by an "acceleration" fLlter. and the resulting droplet-free baseline was integrated to obtain the time-mean velocity and RMS. In the densest parts of the spray. the 2 J..lm droplets were used to estimate the gasphase velocity. using a procedure similar to that of McDonell and Samuelson (1989). This technique was only applied where intermittency of the small droplet signal was negligible (as determined by time-resolved PDPA). Since the number of droplets contained within each bin used to compute the size-classified mean velocities is not constant. the confidence level associated with the mean velocities is not uniform either across size classes or as a function of radius within a partiCUlar size class. For this reason the width of the 90% confidence interval about the mean is shown in the lowest part of each figure to advise the reader as to the statistical significance of each point. Size-classified velocity statistics could be obtained at z = 10 mm despite a high droplet rejection rate. Unlike volume flux and size distribution data, rejection rate does not affect the size-classified velocity data as long as a statistically significant number of droplets are sampled in each class. Due to the high spray number density and close proximity to the furnace floor. gas-phase velocities were not measured at this location. These data (z = 10 mm) show that even this close to the nozzle. thc spray has assumed a conical fonn and that the aerodynamic center of the spray is slightly offset from the nozzle centerline. The axial velocity is almost unifonn across size classes while the radial velocity shows a strong positive correlation with increasing size. Note that the slight bulge at the outer edge of the spray in the axial data corresponds to large droplets which have a significantly higher radial velocity component than the mean. As apparent from comparison of the size classified RMS and mean velocitics. the differences in radial velocity between the various size classes are statistically significant. These data show that while the largest droplets are moving radially outward at high vclocity. the smallcr droplets are moving with reduced velocity and. in the case of |
