A central theme of our prior experimental and computational work on a transonic self-sustaining pulsatile three-stream coaxial airblast injector involved obtaining spectral content from compressible 2-D models and preliminary droplet size distributions from incompressible 3-D models. The three streams entail an inner low-speed gas, and outer high-speed gas, and an annular liquid sheet. Local Mach numbers in the pre-filming region exceed unity due to gas flow blockage by the liquid. Liquid bridging at somewhat regular intervals creates resonance in the feed streams. The effects of numerical decisions and geometry permutations were elucidated. The focus now shifts to compressible 3-D computational models so that geometric parameters, modeled domain size, and non-Newtonian slurry viscosity can be more elaborately explored. While companion studies considered circumferential angles less than 45°, specific attention in this work is given to the circumferential angles larger than 45°, the slurry annular dimension, and how this annular dimension interacts with inner nozzle retraction (pre-filming distance). Additional metrics, including velocity point spectral analyses, are investigated. Two-stream experimental studies are also computationally studied.
Multiple conclusions were drawn. Narrower annular slurry passageways yielded a thinner slurry sheet and increased injector throughput, but the resulting droplets were actually larger. Unfortunately the effect of slurry sheet thickness could not be decoupled from another important geometric permutation; injector geometry physical constraints mandated that, in order to thin the slurry sheet, the thickness of the lip which separates the inner gas and slurry had to be increased accordingly. Increased lip thickness reduced the interfacial shear and increased the thickness of the gas boundary layer immediately adjacent to the slurry sheet. This suppressed the sheet instability and reduced the resulting liquid breakup. Lastly, velocity point correlations revealed that an inertial subrange was difficult to find in any of the model permutations and that droplet length scales correlate with radial velocities.
As anticipated, a higher viscosity resulted in larger droplets. Both the incremental impact of viscosity and the computed slurry length scale matched open literature values. Additionally, the employment of a full 360° computational domain produced a qualitatively different spray pattern. Partial azimuthal models exhibited a neatly circumferentially repeating outer sheath of pulsing spray ligaments, while full domain models showed a highly randomized and broken outer band of ligaments. The resulting quantitate results were similar especially farther from the injector; therefore, wedge models can be used for screening exercises. Lastly, droplet size and turbulence scale predictions for two external literature cases are presented.