High-Speed Local Particle Injection for Particle Image Velocimetry

Abstract

High-speed local seeding particle injection for particle image velocimetry has been developed and demonstrated by measuring Mach 3 and Mach 10 turbulent boundary layers. The seeding injectors accelerate the particle aerosol through a converging/diverging supersonic nozzle that exits tangentially to the wall. Computational fluid dynamics codes using the multiphase application program interface particle tracking simulation were used to optimize the nozzle geometry in order to minimize impact to the flow and maximize seeding particle concentration. Normalized particle image velocimetry particle concentration profiles have excellent agreement with computational fluid dynamics, although the dimensional concentrations are higher by factors of about three and two for Mach 3 and Mach 10, respectively. This difference may be due to uncertainty in the measured laser sheet volume. Mach 3 experiments duplicated with global seeding indicate minimal impact of the local injection in the majority of the boundary layer. The boundary-layer thickness increased by less than 2.5%, and the relative errors in the mean and fluctuating velocities are within measurement uncertainty. The characteristic particle distribution of local injection is shown to lead the phenomenon of particle biasing. There is a high number of low-momentum particles ejected higher in the boundary layer but little to no high-momentum freestream particles swept back into the boundary layer. This creates an unrepairable detrimental bias to measurements above the point where the particle concentration diminishes: $y/\delta \ge 0.6–0.9$ . Particle biasing is shown to affect mean velocity and turbulence statistics profiles.