Aero-Structural Optimization of an Axial Turbine Stage in Three-Dimensional Flow

Abstract

This paper presents a simple, effective and practical shape optimization approach for axial turbine stages so as to minimize the three-dimensional flow losses and simultaneously improve the turbine structural properties. The main objectives of the optimization are to maximize the stage efficiency and simultaneously minimize the von Mises stress while constraining the design mass flow rate and the blade first natural frequency. The stacking curve, which controls three-dimensional flow effects and the spanwise stress distribution is parametrically represented by a quadratic rational Bezier curve (QRBC). The parameters of this QRBC are related to the design variables namely the blade lean, sweep and bow. The optimization method combines a Multi-Objective Genetic Algorithm (MOGA), with a Response Surface Approximation (RSA) of the Artificial Neural Network (ANN) type. During the optimization process, each objective function and constraint is approximated by an individual ANN, which is trained and tested using an aerodynamic as well as a structure database composed of a few high fidelity flow simulations (CFD) and structure analysis (CSD) that are obtained using AN-SYS Workbench 2.0. This methodology was then applied to the aero-structural optimization of the E/TU-3 turbine stage at design conditions and proved quite successful, flexible and practical, and resulted in an 0.8% improvement in stage efficiency and about 50% reduction in the maximum von Mises stresses. This improvement was accomplished with as low as five design variables, which is remarkable considering the problem complexity.