The complexity of modern turbomachines has led to their design being sub-divided across engineering discipline boundaries and thus handled by separate teams of expert designers, increasingly often by different sub-contractors. Thus very early in the design process it is necessary to specify the performance required, and key attributes, of the various modules and sub-assemblies into which the turbomachinery has been decomposed. Combined with the mixed integer and non-linear nature of the turbomachinery aerodynamic design space, and the wide range of fidelity of the analysis tools used, this presents significant optimization challenges, particularly should the use of higher fidelity tools late in the design process reveal that the early choices have placed a module in a sub-optimal position. Gradient-based optimization methods are relatively fast and computationally cheap, yet easily become trapped in local optima. Presented here is a method by which the multi-fidelity nature of the design space is exploited (rather than considered an obstacle to be overcome) such that simple, relatively smooth functions 'carry' higher fidelity models over various 'cliff-edges' in the design space. This technique has been embodied in MIDGE (Multi-fidelity Integrated Design Gradient-based-optimizing Environment), a new optimizer computer code, and illustrated by its successful application to a commercial steam turbine design.


    Access

    Access via TIB

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Multi-fidelity gradient-based optimization in turbomachinery aerodynamic design


    Contributors:


    Publication date :

    2006


    Size :

    8 Seiten, 3 Quellen



    Type of media :

    Conference paper


    Type of material :

    Print


    Language :

    English





    AIAA-2006-1907 Multi-Fidelity Gradient-Based Optimization in Turbomachinery Aerodynamic Design

    Jarrett, J. / American Institute of Aeronautics and Astronautics | British Library Conference Proceedings | 2006




    Robust Multi-Fidelity Aerodynamic Design Optimization Using Surrogate Models

    Jaeggi, D. / Parks, G. / Dawes, W. et al. | British Library Conference Proceedings | 2008