Metamodel Assisted Multi-Objective Global Optimisation of Natural Laminar Flow Aerofoils
Lee Cameron, Juliana M. Early, Richard McRoberts · 29th AIAA Applied Aerodynamics Conference · 2011
A methodology for the multi-objective optimisation of natural laminar flow (NLF) aerofoils using a Kriging assisted multi-objective genetic algorithm is proposed. Kriging surrogates are built to model the variation of drag coefficient at two different flight conditions and are dynamically refined in promising regions of the design space as the optimisation progresses using an adaptive sampling technique. The optimisation is subject to constraints on moment coefficient and maximum lift coefficient; the latter of which is also estimated using a surrogate model. The maximum lift coefficient metamodel is dynamically refined when the error bounds on the Kriging prediction cross the constraint threshold. True objective function values are calculated at sample points by an interactive boundary layer solver in which prediction of transition onset location is achieved via the full linear stability theory and e N method. An adaption of the class-shape transformation technique is utilised to parameterise the aerofoil geometry and define the design space. The design methodology was found to be capable of identifying a set of Pareto optimal wing sections that exhibit low drag and extended regions of laminar flow at both flight conditions.