Genetic Algorithm in Optimum Shape Design

Authors:

Abstract

The shape optimization method in Computational Fluid Dynamics applications using Genetic Algorithm and commercial turbulent flow modeling software FLUENT is presented. Two test cases are solved: the optimum nozzle design and the optimum Francis turbine spiral casing design. The shape of the nozzle and turbine spiral casing are parameterized by cubic Bezier splines, the coordinates of control points being optimizing parameters. In the nozzle case, the energy losses, defined as the difference of total mechanical energy flux through inflow and outflow cross-sections, are minimized. The objective functional for spiral casing shape optimization is defined as the surface averaged value of the magnitude of the difference between calculated and desired velocities at the outlet, so as to give as homogenized spiral casing outflow as possible.

Copyright 2013 - Department of Fluid Mechanics and Computational Engineering