Experimental and computational analysis of evaporative spray cooling for gas turbine exhaust ejectors

This research studied the effects of evaporative spray cooling on air-air ejector performance. Experimental data was collected for the purpose of validating computational simulations. This was done by modifying an existing air-air ejector to accommodate four spray flow nozzles which were used to atomize cooling water. The only parameter that was varied throughout the study was the mass flow rate of cooling water. One single phase (air) case and four spray flow cases where performed and analyzed. The purpose of the single phase experiment was to have a baseline for the air-air ejector performance and isolate the sources of experimental error contributed by spray flow. Three specialized multiphase flow instruments were designed and fabricated by the author to measure, gas phase temperatures, spray mass flow rates, and mixture total pressures.;Comparison of experimental and computational cases produced mixed results. It was found that the experimental gas temperature instrument performed poorly for the local droplet fluxes encountered during testing. The spray sampling probe showed more promising results with two integrated mass flows agreeing within 6% of computational simulations. The total pressure probe solved the issue of pressure port clogging, but measurements were representative of mixture density which made an inferred velocity calculation difficult. It was found that evaporation of spray flow before the nozzle exit plane caused a reduction in dynamic pressure and a reduction in back pressure.;A full scale simulation was performed to determine the effects of scaling on evaporative spray cooling performance. It was found that for the geometrically similar full scale model, the total droplet surface area and particle residence times scaled up with the model which increased cooling performance.;A computational study was performed using the collected experimental data for inlet continuous phase and spray mass flow as boundary conditions for equivalent simulations. A temperature gradient modified turbulence model was written by the author to better predict the mixing rates found experimentally which was used for the duration of this research. Secondary droplet breakup was modeled by the author using empirical correlations following preliminary simulations recognizing the deficiencies of commercially available breakup models.