OCR Text |
Show The system was operated using the green line of an argon-ion laser in TEM-OO mode at 1 watt output power. A 5 MHz net frequency shift was used to resolve directional ambiguities. Fringe spacing was set at 6.5 J..Lm and 2: 1 beam expansion was used to achieve a measured beam waist diameter of 130 J..U11 (1/e2 points). Scattered light was collected at 20- off forward using a 500 mm focal length, f/3.3 lens. The scattered light was imaged at 1: 1 onto a 550 J..Lm aperture, and through a 1 nm (FWHM) interference filter to a photomultiplier tube. In order to make measurements of the flow velocity at the entrance plane of the furnace (z = 0 mm), the transmitting and receiving optics assemblies were tilted at an angle of 5- from the horizontal, projecting the beams down to the entrance plane. Since this tilting of the beams does not effect measurement of the radial or tangential components of velocity, this configuration was used for measurement of these components at all axial locations. For axial velocity measurement, however, an error is introduced since the measured velocity vector (V zm) is tilted by 5- with respect to the true axis. For this reason, only the measurements at z = 0 mm were made with this tilted-beam configuration, all others were made with the beams on a horizontal centerline. Axial data taken at z = 0 mm with the tilted beam configuration have been corrected for this via the procedure discussed in Appendix A. The processing electronics used in this study consisted of a Pacific Instruments preamplifier ' (100: 1 gain) and TSI counter-type processor. The processor was operated using eight fringes for validation, in single measurement per burst mode, with a 1 % fringe comparison and amplitude limit set at three volts. This limit was used to reject signals of sufficient amplitude so as to cause saturation of the Pacific Instruments preamplifier. In order to obtain statistically independent sampling of the flow and minimize both velocity and number density bias effects, the LDV processor was operated as a "controlled processor" (Edwards, 1987) with a delay period of 150 ms between samples. This period was chosen in keeping with the suggestion of Tennekes and Lumley (1972) that the interval for statistical independence of the flow is approximately twice the integral time scale. Detailed axial velocity vs. time records at high seeding rate (> 10kHz) were collected throughout the flow and the integral time scale computed from autocorrelation of the data. The longest integral time scale found was 75 ms, as such, a control period of 150 ms was chosen for these measurements. The main air supply was seeded with aluminum oxide such that the data validation rate was between 1 and 3 kHz on the average. This rate was chosen as a compromise between the need to provide a high seeding rate for minimization of velocity and number density bias effects (Durao and Whitelaw, 1975), and the need to keep the furnace windows clean long enough to complete the required measurements. The atomizing-air stream was not seeded for these measurements since it was found that the alumina would agglomerate and plug the passages in the nozzle, resulting in unstable performance. Tests conducted in both the combusting flow and with just the airflow (no spray or flame) showed no difference in the mean velocity or standard deviation across the plane at z = 25 mm. It was concluded that entrainment of the seed from the main air supply was sufficiently rapid that any potential bias effect of not seeding the atomizing-air stream was negligible by the time the flow reached the first axial measurement plane. The alumina seed was supplied by a cyclone pickup and separation system. Prior to use, the seed was baked for a minimum of three hours to remove any moisture. An effective seed diameter of less than 1.2 J..U11 was measured using the Insitec Particle Counting Sizing Velocimeter (PCSV) system. - 3 - |