A Design Method For Turbine Blades Using Genetic Algorithms On Parallel Computers

The design of 2-D turbine blades using tar- get pressure considerations is described. It is based on the optimization of a set of free-parameters through Ge- netic Algorithms (GAs). Profile shapes are modeled using combinations of circular arcs and Bezier functions. The number of free-parameters may vary, depending on the particular design requirements; they practically consist of the radii and orientations of the circular arcs and the control-points of the Bezier functions, but other parame- ters may be optimized as well. The exploitation capabili- ties of the GA are improved using adaptive ranges for the free-parameters. Close to the optimum solution, the GA switches to an iterative hillclimbing (HC) method, which undertakes the final refinement. The method application is demonstrated in a steam-turbine blade design problem.