A hierarchical optimisation approach for automatic turbomachinery blade design

The problem of turbine cascade design is considered using automatic optimisation strategies. Genetic algorithms (GA) have proved, in previous applications, to be able to deal with optimisation problems with many independent variables and to be extremely easy to use into a general computational system. On the other hand GA require a high number of fitness evaluation in order to get the best individual and this aspect is particularly negative when the fitness is obtained from Navier-Stokes computations. In the paper an optimisation strategy to reduce the overall computational time is proposed. It consists of an alternate use of fine and coarse grids for the evaluation of the profile loss with a Navier-Stokes solver developed by the authors. © 2001 by the American Institute of Aeronautics and Astronautics, Inc.