Multi-Objective Aerodynamic Design Optimisation

Research Theme: Computational Design

The optimisation of airfoil designs is a challenging, computationally expensive, highly constrained, non-linear problem.

Any consideration of robust design – the retention of performance over a range of operating conditions – must inevitably entail multiple objectives.

To develop an efficient optimisation tool, each stage of the process (representation, evaluation and search) must be carefully implemented.

Motivation

As with most real-world problems, there are multiple (usually conflicting) performance metrics that an engineer might seek to improve in optimising, for example, the design of turbomachinery blades, wings or other aerodynamic surfaces.

Objectives

  • To develop and validate an integrated multi-objective design optimisationsystem for aerodynamic applications
  • To examine the trade-offs between the most important flow parameters affectingthe performance of turbomachiney blades

Method

This system is a combination of a number of individually established computational codes working together without human intervention. Each code performs a specific part of the whole design optimisation process: Parameterisation-representation, evaluation (using a well-established Computational Fluid Dynamics code) and search space exploration (using a novel multi-objective variant of the Tabu Search optimisation algorithm).

Findings

The output from the system is a selection of different designs ranged across the trade-off surface (Pareto front) for the problem under consideration. This gives the designer considerable insight into the possibilities available.

The system is suitable for aerodynamic applications involving multiple parameters that affect performance. It is, thus, ideally suited for application to robust design problems.

The system is inherently parallel and can be run on multi-processor systems, considerably reducing wall-clock run times. It offers great potential because optimal aerodynamic designs are then achievable automatically in a comparatively short time.

Details

Problem definition:

  • To design the blade geometry of a gas turbine compressor guide vane that efficiently gives a good pressure rise at a particular flow coefficient.
  • The first objective function describes the overall losses (entropy generation rate), while the second considers the span-averaged blockage (a throughflow performance measure).
  • Constraints on the flow conditions and geometry constraints (clearance and leading edge sharpness) are included as penalty terms in the objective function formulations.
Key Developments
  • Intelligent local search exploration
  • Sophisticated system for selecting the design variables
  • Improved intensification/diversification strategy
  • Improved restart strategy

Acknowledgements

  • Support for this project was provided by the Embiricos Foundation and the Cambridge European Trust.

Selected Publications

  • KIPOUROS, T., PARKS, G.T., SAVILL, A.M. and JAEGGI, D.M. (2004) 'Multi-objectiveaerodynamic design optimisation' in ERCOFTAC Design Optimization: Methodsand Applications. International Conference and Advanced Course Program, Athens,Greece, (CD-ROM. ERCODO2004_239).
  • KIPOUROS, T., JAEGGI, D.M., DAWES, W.N., PARKS, G.T. and SAVILL, A.M. (2005)'Multi-objective optimisation of turbomachinery blades using tabu search'in The 3rd International Conference of Evolutionary Multi-Criterion Optimisation,Guanajuato, Mexico, 3410, 897-910.
  • KIPOUROS, T., JAEGGI, D.M., DAWES, W.N., PARKS, G.T. and SAVILL, A.M. (2005)'Multi-criteria optimisation of turbomachinery blades: investigating the trade-offsurface' in The 41st AIAA/ASME/SAE/ASEE Joint Propulsion Conference andExhibit, Tuscon, Arizona, AIAA-2005-4023.
  • KIPOUROS, T., JAEGGI, D.M., DAWES, W.N., PARKS, G.T and SAVILL, M. (2005)'Multi-objective aerodynamic design optimisation for axial compressors', inERCOFTAC Bulletin, 66 (Sept), 21-24.