(151d) Optimization of a Supersonic Ejector Coupled with a CFD Analysis

ABSTRACT

The present work studied
the behavior of ejectors which are widely used in industry as substitutes for
mechanical compressors or even for vacuum producers in distillation towers.
They drag a secondary fluid through the difference of pressure generated by the
entrance of a high pressure motive fluid on the main nozzle which has a
combination of converging/diverging ducts. A preliminary study using
Computational Fluid Dynamics (CFD) was validated against experimental data.
Then the effects of operating conditions and geometries on ejector performance
were investigated at different pressure and temperature conditions. The CFD
model was capable to predict well the experimental data (Figure 1). For
adequate phenomenon representation, a density based solver was used with a
steady-state approach for time prediction, ideal gas treatment for the working
fluid and a two-dimensional (2D) approximation. High velocities were achieved
after the primary nozzle entrance and a supersonic flow took place. For better
use of computational resources, an independency test for the flow pattern was
carried out by varying mesh refinement. As a second part of this study, an
optimization analysis was developed. In this step, the goal was to achieve the
best geometric configuration for an ejector by varying three different
geometrical parameters simultaneously with a fixed operating condition and
maximizing efficiency of the device. The parameters varied were: primary nozzle
inlet diameter where the high pressure motive fluid enters, the length throat
which is responsible for flow stabilization and finally the inlet diameter of
the mixture chamber which is the entrance of the secondary fluid. The device
efficiency was evaluated by the entrainment ratio which is the rate between the
entrance of secondary fluid and of motive fluid. In order to create an initial
population for the problem, the Uniform Latin Hypercube model was used as a
design of experiments. Moreover for the optimization itself the Non-dominated
Sorting Genetic Algorithm II (NSGA-II) was employed. The results for the
optimization cases show a direct effect on the ejector performance of all
geometrical parameters tested. Increases on the selected parameters provide
larger entrainment ratios under the same operating condition. The same analysis
also shows that highest sensibility on the device efficiency was achieved by
variations on the primary nozzle diameter. Small increases on this diameter
generated a much higher drag of secondary fluid.

Figure 1. Pressure profile along the wall of
the ejector.