An Experimental and Numerical Investigation of Evaporative Spray Cooling for a 45 degree Bend near a Gas Turbine Exhaust

Thumbnail Image
Armitage, Grant
Computational vs Experimental , spray cooling , exhaust bend , gas turbine exhaust , evaporative cooling , CFD
The research performed in this work investigated evaporative spray cooling systems using water near a 45 degree bends in gas turbine exhaust piping systems. Both experimental data and numerical data were generated with the goal of evaluating the ability of Fluent 6.3.26 to predict the performance of these systems for the purpose of design using only modest computational resources. Three cases were investigated in this research: single phase exhaust flow with no water injection, injecting water before the bend and injecting water after the bend. Various probes were used to measure dry bulb temperature, total pressure and water mass flux of the two phase flow at the exit of the pipe. Seven hole probes and pitot static probes were used to measure single phase flow properties. Numerical simulations were performed using mass flow boundary conditions which were generated from experimental results. A turbulence model was selected for the simulations based on comparisons of single phase simulations with experimental data and convergence ability. Using Fluent’s discrete phase model, different wall boundary conditions for the discrete phase were used in order to find the model which would best match the evaporation rates of the experimental data. Mass flux values through the exit plane of the pipe were found to be the most reliable of all the two phase data collected. Results from numerical simulations revealed the shortcomings of the available discrete phase wall boundary conditions to accurately predict the interaction of the liquid phase with the wall. Experimental results for both cases showed extensive areas of the wall which had liquid film layers running down the length of the pipe. Simulations resulted in particles either failing to impact the wall and create a liquid film, or creating a liquid film which was much smaller than the film present in experimental results. This led to 8% and 15% discrepancy in evaporation amounts between numerical and experimental results for water injection upstream and downstream of the bend respectively. Under-prediction of areas wetted with a wall film in the simulations also led to gross over predictions of wall temperature in numerical results.
External DOI