Process Integration and Automated Numerical Design Optimization of an Eigenfrequency Analysis of a Compressor Blade

Modern gas turbines for applications in power plants have to fulfill more and more demands defined by customer and grid requirements. These requirements address for example reduced time for run up and increased power output while providing maximum single or combined cycle efficiency. The demanding market requirements increase the pressure to further improve the design process of gas turbine parts by reducing the overall development time and simultaneously improving the quality of the design. This paper describes the implementation of an automated optimization process for the mechanical assessment of compressor blades applied during the preliminary design process. Previous work from Fedorov, Szwedowicz, et al [1], has shown that it is important to apply 3D FE methods for the accurate prediction of the dynamic behavior of compressor blades already in the early stage of the design phase. These key ideas were picked up in the present work while the FE model from [1] was extended to a complete 3D model of the compressor blade including airfoil and blade root geometry. The new approach completely automates the 3D FE analysis of compressor blades including CAD model generation, FE pre-processing, FE analyses, FE post-processing and takes it to the further level by integrating the FE analysis procedure into an automated design loop using the commercial optimization software iSight FD. The target of this optimization loop is to drive the frequency of critical mode shapes into allowed ranges by modifying airfoil parameters such as airfoil thickness and chord length. A scalar optimization technique is applied solving the design problem using penalty functions for excitation sources, mode shapes and eigenfrequencies. In order to achieve a smooth distribution of airfoil parameters Bezier-Spline approximations are used to parameterize the design space. The implementation of the mechanical analysis for compressor blades into a standardized and automated process was one of the main achievements of the presented work. The process was completely implemented in Abaqus CAE including 3D FE model preparation and post-processing. It was key to a successful integration into an overall optimization loop, which helped to substantially reduce the amount of manual work required to perform the design task.