HOME



GARTEUR (Group for Aeronautical Research and Technology in Europe) is a multinational organisation that performs high quality, collaborative research in the field of aeronautics to improve the technological competence of the European Aerospace Industry. GARTEUR research activities cover the fields of Aerodynamics, Flight Mechanics, Systems and Integration, Helicopters, and Structures and Materials. For more details about GARTEUR visit the official web page www.garteur.org.

GARTEUR AD/AG52 (2013-2015)


The Action Group number 52 belongs to the GARTEUR-Aerodynamic (AD) research field and aims to explore surrogate-based global optimization methods for aerodynamic shape design. Its main objectives are: 
  • Assess surrogate modeling techniques for fast computation of the fitness function
  • Assess surrogate-based global optimization strategies for shape design
  • Evaluate DoE techniques given a certain geometry parameterization. Analysis of the cost/accuracy trade-off for different DoE plans and strategies
  • Demonstrate the applicability and CPU time savings by the use of surrogate-based global optimization
  • Provision of “best practice” guidelines for the industrial use of SBGO methods in shape optimization
  • Simplification of the use of SBGO methods in aeronautic industries


The main challenges to be faced within this Action Group are:

  • Deal with the “curse of dimensionality”. Assessment of surrogate modelling techniques when dealing with high dimensionality problems (n>>10)
  • Model validation strategies, assessment of the overall computational time required to build and train the surrogate
  • Assessment of the proper error metrics for comparison (RMSE, ME ....)
  • Handling constraints with surrogates, possible formulations of constraints
  • Accuracy of the model on large database (more than 200 samples)
  • Improvement of surrogate accuracy at fixed computational budget (VFM ....)
  • Efficient DoE techniques, adaptive DoE strategies for “optimal” selection of training points towards validation error mitigation
  • Reduction of the design space (parameterization sensitivity, ranking of design parameters, gradient information, etc...)

For more information about the AD/AG52 download the technical proposal here.