COB-2019-1386

Winglet Design Optimization Using A Multi-Objective Genetic Algorithm

Lucas André Colombo Gewehr
Bruno Silva de Sousa
Federal University of Itajubá, Av. BPS 1303, 37500-903, Itajubá-MG
[email protected]
[email protected]

Abstract. This project proposes a method for the design of a Blended Winglet configuration using a Multi-Objective
Genetic Algorithm aided with AVL’s VLM aerodynamic evaluations and takeoff estimations using an iterative algorithm.
The results obtained with the evaluated method of design were very satisfactory and close with the theoretical estimations,
showing a 2.96% of total cruise drag reduction.

Keywords: winglet, multi-objective optimization, aerodynamics, drag reduction

1. INTRODUCTION

The idea of drag reduction and fuel saving is the main focus of all modern aircraft projects. For this purpose, since the
early 20th century studies began on devices made to reduce lift-induced drag on wingtips, based on the already recognized
fact that nonplanar wing systems have less induced drag than planar ones. However, the first projects of a vertical plate
mounted to the wingtip did not live up to the theoretical estimations due to a big increase in profile and interference drag.
Later on, (Whitcomb, 1976) realized that with a proper design, these devices could effectively reduce the total drag of the
airplane.
Since Whitcomb’s findings there has been several changes to the methods of design and mainly the shapes of wingtip
devices. Today, the most used device is the Blended Winglet which was designed to significantly reduce the interference
and profile drag problems. In modern aircraft those devices can be responsible for up to 3.2% of total drag reduction
(McLean, 2005), which in a long distance flight can represent large fuel savings. Even with the new technologies of the
modern winglets, the design process is based on very old concepts, with reference values for some basic parameters and
a general geometry already defined, wich can lead to a useful but non-optimal design.
In this paper a method is presented for a blended winglet design optimization, with a theoretical review and results of
a winglet designed for a SAE Brasil Aerodesign competition aircraft.

2. THEORETICAL BACKGROUND

One of the main consequences of lift generation is the spanwise flow, Fig. 1. At the trailing edge, the inward flow
from the lower surface (high pressure) and outward flow from the upper surface (low pressure) are merged and generate a
certain vorticity which is the origin of induced drag and the vortex sheet, Fig. 2. This phenomena can also be seen as a
downwash angle generated by the trailing-vortex system.
Lucas André Colombo Gewehr, Bruno Silva de Sousa
Winglet Design Optimization Using A Multi-Objective Genetic Algorithm

Figure 1. Spanwise flow due to pressure imbalance

Figure 2. Vortex Sheet behind a lifting surface

The Vortex Lattice Method (VLM) is one of the most used low computational cost method for aerodynamic calcula-
tions. This method models the wing with a horseshoe vortex distribution along the wing’s span and chord and is mainly
based on four classical theories: Biot-Savart Law, Kutta-Joukowski theorem, Hermann von Helmholtz theory and Prandtl
Lifting-Line Theory. (Budziak, 2015) Another important consideration about VLM is the usage of a Trefftz Plane for in-
duced drag calculations, this is what makes VLM methods suited for calculating a large variety of configurations including
nonplanar wing sistems.

3. OPTIMIZER

Considering all presented above, and specially the validity of the VLM method for the non-planar induced drag
calculation, a ModeFRONTIER R
workflow was built, with all aerodynamics coefficients evaluated with AVL R
and
takeoff parameters estimated through an iterative algorithm.
A large study was conducted to ensure that AVL R
was suited for winglet calculations and the importance of Clα and
Cdp corrections for the airfoil (done with CLAF and CDCL functions) was observed.
The ModeFRONTIER R
algorithm used was the NSGA-II, with the controlled elitism configuration, for a better pop-
ulation spread and more precise convergence.
All the geometrical parameters that were varied are presented in the Tab. 1.

Table 1. Geometrical Parameters

Parameter Variation Interval
Initial Span [0,95; 1,125]m
Final Span 1,25m
Height [0,1; 0,3]m
Tip Chord [0,02; 0,16]m
Geometric Twist [-2; 6]◦
Airfoil [1; 9]
Sweep [0; 40]◦
 25th ABCM International Congress of Mechanical Engineering (COBEM 2019)
October 20-25, 2019, Uberlândia, MG, Brazil

Table 2. Winglet Airfoils

Reference
Airfoil
Number
1 PSU 94-097
2 DAE-51
3 FX 63-100
4 GOE 596
5 GOE 602 MOD
6 MH 38
7 MH 116
8 RAF 32
9 S4062-095-87

The intervals were determined according to previous studies and project limitations. Considering the general goal of
the wingtip use and also the competition requirement, the main objective was set to be the total drag coefficient reduction
in a cruise configuration, with an auxiliary constraint that helped maximize takeoff weight, since the competition points
were based on it.

4. RESULTS

After the optimizer convergence, a blended winglet was chosen. The geometrical parameters are presented in Tab. 3

Table 3. Winglet Characteristics

Parameter Value
Initial Span 1.125m
Final Span 1.25m
Height 0.22m
Tip Chord 0.06m
Geometric Twist 6◦
Airfoil 3 (FX 63-100)
Sweep 21,26◦

A CFD study was conducted using StarCCM+ R

in a RANS configuration and the k-ω SST γ-Reθ turbulence model
with a prism layer adjusted for y+<1 as required for the model (Siemens, 2016).
After the CFD evaluation of the chosen model, an early boundary layer detachment was seen, this problem was fixed
varying the winglet’s sweep, with almost no increase in drag when compared with the VLM estimation. Figures 3, 4 and
5 show respectively both winglet geometries, the velocity over each winglet, and the comparison of the turbulent kinetic
energy behind the wing with (right) and without (left) the final winglet.
The designed winglet configuration presented a cruise total drag reduction of 2.96%.

Figure 3. Winglet Geometries

Lucas André Colombo Gewehr, Bruno Silva de Sousa
Winglet Design Optimization Using A Multi-Objective Genetic Algorithm

Figure 4. Velocity Distribution on Winglet Surfaces

Figure 5. Turbulent Kinetic Energy

5. FUTURE WORK

Considering the good results of this optimization, in a close future the method will integrate cruise parameters, stability
estimations and specially preliminary structural calculations, due to the increase in weight and bending moment.

6. REFERENCES

Budziak, K., 2015. “Aerodynamic analysis with athena vortex lattice”. Hamburg University of Applied Sciences.
McLean, D., 2005. “Wingtip devices: What they do and how they do it”. Boeing Performance and Flight Operations
Engineering Conference.
Siemens, 2016. “Star-ccm+ user guide”.
Whitcomb, R.T., 1976. “A design approach and selected wind tunnel results at high subsonic speeds for wing-tip mounted
winglets”. NASA Technical Notes, Vol. D-8260.

7. RESPONSIBILITY NOTICE

The authors are the only responsible for the printed material included in this paper.