A multi-objective optimiZation method is demonstrated using an evolutionary genetic algorithm. The applicability of this method to preliminary nacelle design is demonstrated by coupling it with a response surface model of a wide range of nacelle designs. These designs were modelled using computational fluid dynamics and a Kriging interpolation was carried out on the results. The NSGA-II algorithm was tested and verified on established multi-dimensional problems. Optimisation on the nacelle model provided 3-dimensional Pareto surfaces of optimal designs at both cruise and off-design conditions. In setting up this methodology several adaptations to the basic NSGA-II algorithm were tested including constraint handling, weighted objective functions and initial sample size. The influence of these operators is demonstrated in terms of the hyper volume of the determined Pareto set.
展开▼
