Date of Award

Summer 8-2015

Author's Department

Mechanical Engineering & Materials Science

Degree Name

Master of Science (MS)

Degree Type



The aim of this thesis is to investigate the detailed flow field inside the supersonic ejector using numerical methods and to optimize the ejector’s mixing chamber wall shape to obtain a maximum entrainment ratio (ER) in order to obtain the highest possible efficiency that can be attained by the ejector. A steam ejector applied in the cooling industry is first studied to determine the most accurate turbulence model for its supersonic jet flow field simulation with mixing with the entrained steam in the mixing chamber. A commercial Computational Fluid Dynamics (CFD) package FLUENT 14.5 along with the meshing tool ICEM 14.5 is utilized to conduct the modeling and simulation to examine the ejector performance using two different turbulence models: k-ε realizable and k-ω SST. Velocity contours, pressure plots and entrainment ratio plots obtained from FLUENT are studied to investigate the effects of several ejector operating conditions as well as to verify the turbulence model accuracy by comparing the numerical results with experimental data. Simulations for three different supersonic ejectors (ejectors for refrigeration and desalination application with different working fluids namely the steam or compressed air) are conducted to further validate the numerical solution accuracy. The turbulence model producing more accurate results is applied to all three cases. In second part of the thesis, a single objective genetic algorithm (SOGA) is employed to optimize the mixing chamber wall shape for steam ejector for refrigeration to achieve the maximum entrainment ratio. Bezier Curves are used to generate the new wall shapes. The whole shape generation-meshing-simulation-SOGA process is repeated until the ER converges to a maximum value based on the specified convergence criteria for SOGA.


English (en)


Ramesh Agarwal

Committee Members

Kenneth Jerina, David Peters


Permanent URL:

Included in

Engineering Commons