Multi-Objective Aerodynamic Design Optimisation

Year: 2006
Author: Kipouros, Timoleon
Supervisor: Parks, Geoff
Institution: Department of EngineeringUniversity of Cambridge
Page(s): 152

Abstract

Ths thesis opens with an introduction to computational aerodynamic design highlighting automatic integrated design optimisation systems. The optimisation tool is serously considered as an indispensable part of the modern industrial automated design cycle for these purposes. The characteristics that any integrated design optimisation system should feature are illustrated through a review of early and current developments in industry and academia. In particular, multiobjective and multi-disciplinary approaches are essential in the design process of real-world applications. In turbomachinery design these approaches give insight into the design space and identity the trade-offs between the competing performance measures. An approach is proposed to assist and support the modern automatic aerodynamic design optimisation process and is demonstrated through the application of a novel multi-objective variant of the Tabu Search optimisation of turbomachinery blades. The aim is to improve the performance of a specific stage and eventually of the whole engine. The integrated system developed for this purpose is described. It combines the optimiser with an existing geometry parametrisation scheme and a well-established CFD package. Its performance is illustrated through three case studies in which the flow characteristics most impaortant of the overall performance of turbo-machinery blades are optimised. The system can readily be run on parallel computers, substantially reducing wall-clock run times - a significant benefit when tackling computationally demanding design problems. The results of a test case minimising the blockage and the entropy rate generation demonstrate that the multi-objective integrated turbomachinery design optimisation system can successfully tackle realistic real-world problems, negotiating the highly constrained, nonlinear search space, and presenting the designer with a range of designs showing the trade-offs between the objectives under consideration, giving insight into the nature of the design space and suggesting innovative designs for further consideration. In this case the performance of the designs found match or exceed the performance of the optimised blade identified in an earlier single-objective study. The design resulting from that test case achieved a measured rise in efficiency of over 3% when tested experimentally. Comparing the single-objective optimised blade and the blade for lowest blockage form the multi-objective test case it is anticipated that new optimal designs will attain similar performance. The factors influencing the efficiency of turbomachinery blades and the trade-offs between them are extremely complex and therefore to explore these interactions additional loss objectives are defined and modeled. Thus, new objective functions evaluate individually the profile losses and the secondary losses in order to improve understanding of the trade-offs between them in design. These investigations require the tackling of a three-objective problem and the effectiveness of the multi-objective Tabu Search variant is demonstrated on this higher dimension problem. A second variant of the optimiser is presented with enhanced local search capabilities, incorporating and intellingent design variable selection scheme based on the idea of path relinking. It is demonstrated that the computational time therefore the efficiency of the integrated system is increased.
Finnaly, avenues for future research in this field are identified and discussed.