Assessment of Optimization Algorithms for Winglet Design

Ubaldo Cella, Diego Giuseppe Romano (University of Naples Federico II, Piaggio Aero Industries)

A design problem can be regarded as a creative process of searching for the “optimal” compromise solution within an iterative loop. The more complex and thorough the designer’s knowledge and experiences are, the higher the design quality will be. Thanks to the improved reliability of modern numerical analysis tools and the exponential growth of computational power in recent years, numerical optimization methods are in great demand in nearly all industrial engineering areas. Today, numerical optimization is a leading design methodology in the aerospace sector, in both the industrial and research fields, and a key factor for competitiveness.










The aim of this work is to describe an industrial application of an optimization process and to investigate the performance of several algorithms for a mono and multi-objective case. The optimization of winglets suitable for a business aviation class aircraft is discussed in the first part of the article. In the second part, the results of the assessment of several optimization algorithms applied to the aerodynamic design of a 2D airfoil are detailed. In a mono-objective optimization case, the SIMPLEX method developed by Nelder & Mead is compared to the performances of a Genetic Algorithm using a smart elitism operator. In a multiobjective optimization, four types of Genetic Algorithms and three Advanced Models are compared.

In the first part of this work, winglets for a business class aircraft have been designed and optimized in cruising speed conditions with the following constraints:

• The wing root bending moment increase should not be higher than 5%;

• No degradation of the wing characteristics (stall path and shock generation) is allowed







The aim of the second part of this paper is to assess some of the optimization algorithms of the modeFRONTIER optimization environment. The actual test case is the optimization of an airfoil in transonic conditions. The geometry is described by 28 design variables, which define the control points of a Bézier polynomial  A coupled Euler/Boundary Layer 2D solver was used to evaluate the aerodynamic performance of the airfoil and each evaluation required about 20 seconds.