Global optimization based on metamodel construction app lied to design axial turbomachinery cascades using CFD

Abstract

It presents a methodology for global optimization with constraints of expensive functions using response surfaces models for aerodynamic cascade representing the turbomachine axial with profiles of family NACA65. For the calculation of flow, is used Fluent CFD software, which is on a local and global variations in the flow field. It has been verified that small geometric on the stagger angle, format airfoil and the spacing between the blades, can lead to changes in the efficiency of the blade. Accordingly, we intend to integrate the solution flow through CFD optimization programs based on the construction of metamodels, aiming to obtain considerable gains in computational time. Integration with the optimization programs is necessary to build "script" command to automatically generate the mesh, where the design variables that define the geometry of the blade cascade as stagger angle, pitch to chord and the camber be modified among pre-established limits based on optimization algorithms, in order to achieve an objective function pre-defined, how to obtain the maximum ratio of Cl/ Cd (lift/drag). This methodology for global optimization based on the construction of metamodels together with the random search algorithm controlled (CRSA) is based on iterative construction of response surfaces with radial basis functions (multiquadric) and the application of heuristic criteria to update the database during the optimization process. Cyclical patterns of search are iteratively used to determine the candidate points to be included in the database.