Resumo:
This dissertation presents the development of a methodology for inverse airfoil design
of axial turbomachine cascades by applying optimization algorithms based on Controlled
Random Search (CRS algorithms).
The airfoil geometry is parameterized with Bezier polynomial curves. The flow
calculation is made by means of a panel method with constant vortex distributions and by
applying the Dirichlet boundary condition (null velocities inside the airfoil).
A new version of the CRS algorithm is presented by using Variability Based Reflections
(CRS-VBR). This version aims to use information about the objective function variation
around the best point of the current population, in order to automatically balance local and
global searches, improving in this way the general search efficiency.
Three test cases of inverse airfoil cascade design are presented for evaluating the
methodology effectiveness. The proposed CRS-VBR algorithm has outperformed other
referred CRS algorithms, both in terms of precision and convergence rate.