Flying through design spaces: efficient evolutionary optimisation of aircraft wings
Wilhelmus J Vankan, E. Kesseler, Robert Maas · 2007
Today’s high-tech products, such as civil aircraft wings, are designed by multidisciplinary teams of experts. Dedicated modeling and simulation tools are used to assess the behavior of the design for each relevant discipline. The required consistency among the different single discipline models is achieved by using an integrated design model, which includes a (large) set of design parameters on which each of the discipline models is based. In order to find the best design, the application of optimization algorithms in combination with the modeling and simulation tools is common practice nowadays. However, for products that require complex models and extensive simulations to assess their behavior, like aircraft wings, such design optimizations may become infeasible due to complicated computational sequences or excessive computational cost. To alleviate such complications, the products’ behavior should be assessed more efficiently. This paper presents a meta-modeling approach, applied to aircraft wing design where aircraft range and fuel consumption are optimized. This approach allows to quickly and conveniently evaluate the wing behavior, and to virtually fly through the considered wing design space. Extensive optimizations, exploiting thousands of metamodel evaluations, are performed using multi-objective genetic algorithms, yielding sets of Pareto optimal wing design points. These points represent those wing designs that have the best feasible fuel consumption for each value of range, and hence directly provide the designer with the most relevant design information.