Applications and enhancements of aircraft design optimization techniques

Created by W.Langdon from gp-bibliography.bib Revision:1.3973

@PhdThesis{Powell:thesis,
  author =       "Stephen R. Powell",
  title =        "Applications and enhancements of aircraft design
                 optimization techniques",
  year =         "2012",
  school =       "Faculty of Engineering and the Environment, University
                 of Southampton",
  address =      "UK",
  month =        jun,
  keywords =     "genetic algorithms, genetic programming, TL Motor
                 vehicles. Aeronautics. Astronautics",
  URL =          "http://eprints.soton.ac.uk/348869/",
  URL =          "http://eprints.soton.ac.uk/348869/1/finalThesis2.pdf",
  bibsource =    "OAI-PMH server at eprints.soton.ac.uk",
  oai =          "oai:eprints.soton.ac.uk:348869",
  size =         "167 pages",
  abstract =     "The aircraft industry has been at the forefront in
                 developing design optimisation strategies ever since
                 the advent of high performance computing. Thanks to the
                 large computational resources now available, many new
                 as well as more mature optimisation methods have become
                 well established. However, the same cannot be said for
                 other stages along the optimisation process - chiefly,
                 and this is where the present thesis seeks to make its
                 first main contribution, at the geometry
                 parametrisation stage. The first major part of the
                 thesis is dedicated to the goal of reducing the size of
                 the search space by reducing the dimensionality of
                 existing parametrisation schemes, thus improving the
                 effectiveness of search strategies based upon them.
                 Specifically, a refinement to the Kulfan
                 parametrisation method is presented, based on using
                 Genetic Programming and a local search within a
                 Baldwinian learning strategy to evolve a set of
                 analytical expressions to replace the standard class
                 function at the basis of the Kulfan method. The method
                 is shown to significantly reduce the number of
                 parameters and improves optimisation performance - this
                 is demonstrated using a simple aerodynamic design case
                 study. The second part describes an industrial level
                 case study, combining sophisticated, high fidelity, as
                 well as fast, low fidelity numerical analysis with a
                 complex physical experiment. The objective is the
                 analysis of a topical design question relating to
                 reducing the environmental impact of aviation: what is
                 the optimum layout of an over-the-wing turbofan engine
                 installation designed to enable the airframe to shield
                 near-airport communities on the ground from fan noise.
                 An experiment in an anechoic chamber reveals that a
                 simple half-barrier noise model can be used as a first
                 order approximation to the change of inlet broadband
                 noise shielding by the airframe with engine position,
                 which can be used within design activities. Moreover,
                 the experimental results are condensed into an acoustic
                 shielding performance metric to be used in a
                 Multidisciplinary Design Optimisation study, together
                 with drag and engine performance values acquired
                 through CFD. By using surrogate models of these three
                 performance metrics we are able to find a set of
                 non-dominated engine positions comprising a Pareto
                 Front of these objectives. This may give designers of
                 future aircraft an insight into an appropriate engine
                 position above a wing, as well as a template for
                 blending multiple levels of computational analysis with
                 physical experiments into a multidisciplinary design
                 optimisation framework.",
}

Genetic Programming entries for Stephen R Powell

Citations