U-Bend Optimization on the RBF4AERO Platform
Dimitrios Kapsoulis, E Papoutsis Kiachagias, Asouti, Kyriakos C. Giannakoglou, Emiliano Costa, Me Biancolini · Cineca Institutional Research Information System (Tor Vergata University) · 2016
This work is based on both stochastic and gradient–based optimization methods used to solve industrial optimization problems. The optimization is carried out through the RBF4AERO platform, developed in the framework of the EU–funded project. The stochastic optimization tool of the platform uses Evolutionary Algorithms (EAs) assisted by off–line trained surrogate models, based on the appropriate sampling of the design space. The continuous adjoint developed on the OpenFOAM Toolbox that provides the sensitivity derivatives for gradient–based methods is the second optimization tool available on the same platform. In either method, the design variables stand for the coordinates of points controlling the deformation of the shape to be optimized along with the computational mesh. Shape and mesh morphing is based on Radial Basis Functions (RBFs). Herein, the aforementioned platform is used for the shape optimization of a U–bend for minimum total pressure losses.