Winglet design for sailplanes using a multi-objective evolutionary algorithm

Due to the complex flight envelope of sailplanes, the addition of properly designed winglets to the aircraft allows for the possibility of significant performance benefits at several flight conditions. This is manifested in the observation that the majority of new sailplanes leave the factory with winglets. While this is advantageous for buyers of these new aircraft, owners of conventional sailplanes designed without winglets are left asking how they can retro-fit winglets, thus improving their aircrafts performance, without resorting to hire an experienced engineer to design winglets for them. This question has lead to the motivation to develop a free software tool which has the ability to design a winglet for an existing sailplane. In this work, winglets are designed for the Janus B sailplane, a 1978 aircraft design without winglets, through the employment of single-objective and multi-objective evolutionary algorithm optimization methods coupled with a high-order potential flow solver. The use of the single-objective optimizer, Covariance Matrix Adaptation Evolutionary Strategy (CMA-ES), serves as a stepping stone to employing a multi-objective optimizer, epsilon-dominance Mutli-Objective Evolutionary Algorithm (ε-MOEA), to solve this problem. During the multi-objective optimization study, a total of three objectives are addressed: i) minimizing total aircraft drag at high-speed cruise, ii) minimizing total aircraft drag during thermalling flight (high-lift coefficient flight condition), and iii) minimizing the root bending moment addition due to the winglet. The results from this paper show that these three objectives are needed to properly address the sailplane winglet design problem, which is inherently a multi-objective design problem. Ultimately, the employment of the multi-objective evolutionary algorithm proved to be successful by designing several winglets whose change in performance was advantageous for all three objectives. For example, one winglet generated in the three-objective study was able to achieve a 0.35% cruise drag reduction, a 3.4% thermal drag reduction, while only increasing the root bending moment by 0.16%. While this method is far from being fully matured, the work completed to date provides a solid foundation to build off of and refine further.