Dimensional reduction and design optimization of gas turbine engine casings for tip clearance studies

Abstract

The objective of this research is to develop a design process that can optimize an engine casing assembly to reduce tip clearance losses. Performing design optimization on the casings that form a gas turbine engine's external structure is a very tedious and cumbersome process. The design process involves the conceptual, the preliminary and the detailed design stages. The redesign costs involved are high when changes are made to the design of a part in the detailed design stage. Normally a 2D configuration is envisaged by the design team in the conceptual design stage. Engine thrust, mass flow, operating temperature, materials and manufacturing processes available at the time of design, mass of the engine, loads and assembly conditions are a few of the many important variables that are taken into consideration when designing an aerospace component. The linking together of this information into the design process to achieve an optimal design using a quick robust method is still a daunting task. In this thesis, we present techniques to extract midsurfaces of complex 3D axisymmetric and non-axisymmetric geometries based on medial axis transforms. We use the proposed FE modeling technique for optimizing the geometry by designing a sequential workflow consisting of CAD, FE analysis and optimization algorithms within an integrated system. An existing commercial code was first used to create a midsurface shell model and the results showed that such models could replace 3D models for defection studies. These softwares being black box codes could not be customized. Such limitations restrict their use in batch mode and development for research purposes. We recognized an immediate need to develop a bespoke code that could be used to extract midsurfaces for FE modeling. Two codes, Mantle-2D and Mantle-3D have been developed using Matlab to handle 3D axisymmetric and non-axisymmetric geometries respectively. Mantle-2D is designed to work with 2D cross-section geometry as an input while Mantle-3D deals with complex 3D geometries. The Pareto front (PF) of 2000 designs of the shell based optimization problem when superimposed on the PF of the solid based optimization, has provided promising results. A DoE study consisting of 200 designs was also conducted and results showed that the shell model differs in mass and defection by <1% and <5.0% respectively. The time taken to build/solve a solid model varied between 45-75 minutes while the equivalent midsurface based shell model built using Mantle-2D required only 3-4 minutes. The Mantle-3D based dimensional reduction process for a complex non-axisymmetric solid model has also been demonstrated with encouraging results. This code has been used to extract and mesh the midsurface of a non-axisymmetric geometry with shell elements for use in finite element analysis. 101 design points were studied and the results compared with the corresponding solid model. The first 10 natural frequencies of the resulting shell model deviates from the solid model by <4.0% for the baseline design, while the mass and defection errors were <3.5% and <9.0% for all 101 design points.

More information

Identifiers

ORCID for A.J. Keane: orcid.org/0000-0001-7993-1569

Catalogue record

Export record