Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.436940
Title: Robust design methodologies : application to compressor blades
Author: Kumar, Apurva
Awarding Body: University of Southampton
Current Institution: University of Southampton
Date of Award: 2006
Availability of Full Text:
Access through EThOS:
Access through Institution:
Abstract:
Compressor blades are subtle aerodynamic shapes designed after years of research and insight. They inevitably show deviations from their desired shapes due to manufacturing errors, erosion or foreign object damage. In the present study we focus on seeking compressor blade geometries, that are robust in performance in the presence of geometric uncertainty. Sophisticated tools for representing and propagating uncertainty are employed. Novel method for modeling eroded blade geometry and simulating manufacturing variations with process capability data are presented. These are combined with an automatic meshing routine and a high fidelity viscous flow solver for performance analysis. A combination of Design of Experiment techniques and Gaussian Process emulators are employed to develop efficient surrogate models for uncertainty analysis and exploring the design space. Efficient multiobjective optimization based robust design methodologies are presented. The robust design methods in conjunction with the surrogate model are used to seek blades that have less variation in performance in the presence of erosion and manufacturing variations. Main effects and sensitivity analysis are also performed to understand the effect of each noise variable on the performance. The performance of the robust blades obtained are compared to that of deterministic optimal blades in the presence of the uncertainties. The robust optimal blades exhibit considerably less variability and mean shift in performance as compared to the optimal blades. Finally, a probabilistic framework is developed to deal with randomness in objectives during multiobjective optimization and is applied in conjunction with Gaussian Process emulators for robust design.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.436940  DOI: Not available
Keywords: TS Manufactures ; TL Motor vehicles. Aeronautics. Astronautics
Share: