See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/333581540

A computational fluid dynamics investigation of subsonic wing designs for

unmanned aerial vehicle application

Article in Proceedings of the Institution of Mechanical Engineers Part G Journal of Aerospace Engineering · June 2019
DOI: 10.1177/0954410019852553

CITATIONS READS

3 3,108

2 authors, including:

Zaid Feroz Siddiqi

University of Arkansas at Little Rock
4 PUBLICATIONS 5 CITATIONS

SEE PROFILE

All content following this page was uploaded by Zaid Feroz Siddiqi on 22 August 2019.

The user has requested enhancement of the downloaded file.

Original Article

Proc IMechE Part G:

J Aerospace Engineering
A computational fluid dynamics 0(0) 1–10
! IMechE 2019

investigation of subsonic wing designs Article reuse guidelines:

sagepub.com/journals-permissions

for unmanned aerial vehicle application DOI: 10.1177/0954410019852553

journals.sagepub.com/home/pig

Z Siddiqi and JW Lee

Abstract
The wing of an unmanned aerial vehicle, RQ-7 Shadow, is modified to study the changes in the aerodynamics of the wing.
The main focus is to investigate the effects of changing the components of wing design when the aircraft climbs and
accelerates. These component modifications included changing the airfoil, planform, aspect ratio, and adding a winglet.
Another objective is to study the efficacy of employing high-lift airfoils like the EPPLER 559 for subsonic unmanned aerial
vehicle applications. For this, five wing designs are considered in this paper. Computational fluid dynamics simulations
using ANSYS FLUENTÕ are conducted for each wing design. The CL/CD ratios for all the wings are calculated at increasing
angles of attack (simulating Climbing) and increasing speed (simulating Acceleration). Compared to the NACA 4415
airfoil, which is utilized by the RQ-7 Shadow, the EPPLER 559 provides an increase in lift at the low angles of attack, but
yields less of these benefits as the angle of attack increases. The tapered planform significantly reduces the high drag
associated with the EPPLER 559 airfoil. The generation of higher lift forces with lower drag is further achieved by
increasing the aspect ratio and through the addition of a winglet. When compared to the NACA 4415 airfoil, it is
concluded that the EPPLER 559 airfoil is a viable candidate for subsonic unmanned aerial vehicle applications only
when the components of wing design are altered. The performance of the wings that employ the EPPLER 559 airfoil
improves when the planform is changed from rectangular to tapered, when the aspect ratio is increased and when a
winglet is added.

Keywords
Unmanned aerial vehicle, tapered wing, aerodynamics, lift, drag, winglet

Date received: 28 November 2018; accepted: 2 May 2019

can be increased. The more lift the UAV generates,

Introduction
the higher its payload capacity is. The less drag the
The role of unmanned aerial vehicles (UAVs) in both UAV creates, the further it can fly on the same fuel
civilian and military applications has witnessed an capacity. Changing the airfoil shape, using a different
unprecedented growth during the last two decades. planform, increasing or decreasing the aspect ratio
Many UAVs have been built to address a wide variety (AR) and the addition of a winglet are some of the
of mission requirements such as logistics, reconnais- design parameters considered for enhancing the over-
sance, combat, etc. By eradicating the presence of a all performance. Many approaches have been taken to
human pilot on board the aircraft, UAVs can operate optimize the aerodynamic capabilities of a UAV’s
in environments that are associated with high risks for wing design. The concept of genetic algorithm has
humans such as active combat zones, environments been introduced in order to optimize the aerodynamic
with biological hazards or long endurance flights. characteristics of a general airfoil for specific condi-
The RQ-7 is a fixed wing UAV that was originally tions.2 Genetic algorithms can also be coupled with
designed and manufactured by Textron Systems of
the AAI Corporation. The relevant performance and Department of Systems Engineering, University of Arkansas at Little
general dimensions of this UAV are shown in Table 1.1 Rock, Little Rock, AR, USA
UAVs are designed to be a cost-effective alternative
Corresponding author:
to manned aircrafts. Over the course of its service life,
JW Lee, Department of Systems Engineering, University of Arkansas
a UAV undergoes many modifications for the purpose at Little Rock, 2801 South University Avenue, Little Rock,
of enhancing its capabilities. By modifying the aero- AR 72204-1000, USA.
dynamics, the UAVs payload capacity and its range Email: [email protected]
2 Proc IMechE Part G: J Aerospace Engineering 0(0)

Table 1. RQ-7 specifications. There is a temptation to put in as many elements as

possible for highly accurate results. However, this
Length 3.4 m
would result in an excessive load on the hardware
Gross weight 170 kg available. For example, a 2D mesh size of 15,000
Wingspan 4.3 m elements when translated to 3D resulted in 45 million
Maximum speed 118 knots (135 mph or 218 km/h) elements. Since there are a total of sixteen 3D simu-
Cruise speed 90 knots (103 mph or 166 km/h) lations required for this study, it would have taken an
unnecessarily large amount of time to completely
simulate them all. After the mesh size is determined,
CFD solvers like ARC2D for 2D and KTRAN for the appropriate turbulence model is selected. There
3D problems in order to minimize drag acting on the are three turbulence models that are considered
airfoil and wing respectively.3 Genetic algorithms can namely: Spalart–Allmaras, k–" realizable and k–!
also be used to solve multiobjective problems. shear stress transport (SST) models. The coefficients
A multi-objective genetic algorithm (MOGA) opti- of lift (CL) and drag (CD) at angles of attack (AOA)
mization procedure has been applied to minimize ranging from 0 to 16 are determined for each tur-
the drag on a transonic wing coupled with a simultan- bulence model. They are then compared with the CL
eous minimization of the structural mass.4 To find the vs AOA and CD vs AOA data from XFOIL and the
optimum high-aspect ratio wing shape for minimizing wind tunnel tests conducted by the Ohio State
the drag, fluid–structure interaction (FSI) optimiza- University. This serves as a validation for the 3D
tion along with the design of experiment and results by making sure that the values attained are
Kriging model are applied.5 The Kriging aerodynamic as close to reality as possible.
model has also been used to solve multiobjective wing Then 3D full wing simulations are performed to
design problems which lead to improved performance estimate the aerodynamic coefficients. The simula-
at different flight conditions.6 An integrated approach tions are performed under two conditions. The first
to subsonic wing optimization is achieved based on condition aims to simulate the effects of aircraft when
targeted overall aircraft aerodynamic features.7 it climbs (gains altitude). In order to simulate this
There exist various optimization techniques for effect, the lift and drag coefficients are recorded at a
the winglet design. A gradient-based optimizer fixed velocity (105mph) with AOA ranging from 0 to
and a quantum-behaved particle swarm optimization 16 . The coefficients are also recorded at every second
algorithm are some of the optimization techniques AOA interval (0, 2, 4, 6. . . .). The second condition
applied to improving the aerodynamics of a winglet.8,9 aims to simulate the effects of aircraft under acceler-
Hence the application of multiobjective optimiza- ation. To simulate this effect, the lift and drag coeffi-
tion techniques would aid in the development of an cients are recorded at fixed AOA of 6 with different
ideal wing design. This paper aims to recognize freestream velocities from 55 mph to 135 mph. The
the patterns and the effects of changing the wing com- coefficients are also recorded with a 10 mph increment
ponents from a computational fluid dynamics (CFD) (55, 65, 75. . . . . .etc.). The fixed AOA of 6 in this case
perspective. is arbitrarily selected. This is because there are many
factors that determine the true AOA during straight
and level flight such as payload capacity, variable
Methodology
density of air at different altitudes, etc. The CL/CD
The objective of this study is to observe the impact of ratios are then plotted along the different AOAs
various wing design changes from the base wing (Climbing) and air speeds (Acceleration) in order to
design of the RQ-7 during two conditions namely: compare the performance of each wing design.
Climbing and Acceleration. The reason why these A commercially available engineering simulation soft-
two conditions are chosen is because UAVs like the ware package ANSYS FLUENTÕ is employed to
RQ-7 fly at high altitudes for prolonged periods of simulate the aerodynamic effects of various wing
time. During that time the UAV increases its altitude designs under these two conditions.
to provide more surveillance coverage and is also sup-
posed to fly to different locations as quickly as pos-
sible. The methodology starts with 2D CFD airfoil
Simulation assumptions
analysis which provides the foundations for the 3D Certain assumptions are made before proceeding to
full wing simulations. From the 2D simulations the both 2D and 3D simulations.
appropriate mesh size and the turbulence model are
determined, which are then translated for the 3D 1. Since the maximum cruise speed of UAV is well
simulations. Before extracting any data from the 2D below Mach 0.3, the flow is considered to be
simulations, it is essential to conduct a mesh depend- incompressible, which implies constant air density.
ency study (MDS). This helps in determining the mesh 2. The energy equation for all the simulations was
size that generates accurate and consistent results that disabled.
requires the least amount of computational time. 3. The air properties at standard condition are used.
Siddiqi and Lee 3

4. The flow is purely bi-directional i.e. the air flow adopted by using the following expression for a yþ
from the inlet has only x and y directional com- value of 100
ponents for the 3D simulations. pffiffiffiffiffi
5. In General Aviation and transport aircraft (the y ¼ L  yþ 74:Re13=14
RQ-7 has a similar configuration), only as much where L is the chord length of the airfoil (0.5 m) and
as 5% of the lift is produced by the fuselage. The Re is the Reynold’s number (1,000,000).16 For a yþ
primary functions of a tail surface are to provide value of 100, the estimated first layer thickness from
longitudinal trim, stability and control to the air- the above expression is 1.15  103 m. This value is
craft. Thus the tail surface utilizes a fraction of its then used as a reference for creating the inflation
ability to generate lift. As a results the effect of the layers. It is found that this first layer thickness gener-
fuselage and the tail surfaces are neglected.10 ated the values of yþ between 15 and 40 along the
6. To be consistent with the wind tunnel data, the length of the airfoil. Figure 2 shows the distribution
Reynold’s number for the 2D simulations is fixed of y+ values along the length of the airfoil.
at 1  106. The corresponding air speed at a length
scale of 0.5 m, which is the chord length, is
Turbulence model selection
29.22 m/s.
7. The solution setup for FLUENTÕ simulations Three different turbulence models are used for this
involves using the COUPLED pressure–velocity investigation: Spalart–Allmaras, k–" realizable, and
coupling scheme and Green–Gauss cell-based gra- k–! shear stress transport (SST). The k–! SST model
dient option. has shown to perform reliably for the wall-driven tur-
bulent flows especially.17 The k–" model has been
observed to predict the flow fields involving large
flow separations more accurately. The k–" realizable
2D airfoil simulations
model is the latest development in enhancing the per-
According to the airfoil database provided by the formance for boundary layers under strong adverse
University of Illinois at Urbana-Champaign, the RQ- pressure gradients, separations, and recirculation.18
7 employs the NACA 4415 airfoil.11 The airfoil coord- The lift and drag coefficient values using the three tur-
inates are downloaded from Airfoil Tools, which is an bulence models are compared with the wind tunnel
open source database.12 These coordinates are then data and the XFOIL data. The XFOIL software is a
imported to SolidWorksÕ to create the 2D geometry. program developed by Drela19 that uses a panel
method in the design and analysis of subsonic airfoils.
The wind tunnel data for the NACA 4415 airfoil was
Mesh generation compiled by Ohio State University (OSU).20 The OSU
CFD simulation results are normally dictated by the experiment involved testing a 457 mm (18 in.) NACA
mesh resolution. The mesh resolution refers to the 4415 airfoil in a 3  5 subsonic wind tunnel under
number of cells in the computational domain. There steady and unsteady conditions. The purpose of the
comes a point when increasing the mesh density OSU experiment was to determine the effect of leading
beyond a particular number of elements does not sig- edge grit roughness (LEGR) on the airfoil perform-
nificantly affect the results. In this paper, a grid ance. This meant that the besides being tested at
dependency study is performed with increasing mesh steady and unsteady states, to determine the effect of
resolutions while monitoring the lift coefficients. The LEGR on airfoil performance, a clean airfoil data was
grid independency is achieved at 7500 elements in a compared with a rough one. All the tests were per-
2D simulation with the C-mesh as shown in Figure 1. formed at Reynolds numbers of 0.75, 1, 1.25, and
The next step is to assign fine meshes entitled 1.5 million. The data were acquired and processed
‘‘inflation layers’’ that capture the flow near the airfoil from 60 surface pressure taps, four individual tunnel
wall region. Ferziger and Perić13 explained that at pressure transducers, an angle of attack potentiometer,
high Reynold’s number, the viscous sublayer of the a wake position potentiometer, and a tunnel thermo-
boundary layer was so thin that it was difficult to couple. The data were then reduced by processing the
assign enough grid points to resolve it. Instead, stand- data scans at each steady-state datum point and were
ard wall functions are employed to resolve this. averaged to provide one data set. For the purpose of
S. Mohammad et al. reported that for accurate calcu- this study, the data from a steady state and clean airfoil
lations, the yþ values should be between 30 and 100 was used at a Reynold’s number of 1,000,000 with
for k–" turbulence model.14 This ensured that the first AOA ranging from 0 to 16 .
cell center remained in the log-law region. The accur- The critical angle of attack is the angle at which the
acy of k–! and Spalart–Allmaras models highly lift coefficient value is highest for a given airfoil. Both
depends on the mesh resolution near the wall.15 It is the XFOIL and OSU data confirm that the critical
required to check whether the first layer thickness AOA for the NACA 4415 airfoil is 14 . Thus, the
within the low-law region is small enough for these 2D airfoil simulations are performed with the AOA
two models. The thickness of the first inflation layer is ranging from 0 to 16 at Reynolds number of
4 Proc IMechE Part G: J Aerospace Engineering 0(0)

Figure 1. (a) C-mesh fluid domain; (b) body of influence sizing function centered around the airfoil.

Figure 2. Wall Y plus (yþ) values plotted along the airfoil.

1,000,000 so that they were consistent with the OSU where FL and FD are the lift and drag forces respect-
and XFOIL results. The expressions for the coeffi- ively,  is the density, V is the air flow speed, and A is
cients of lift and drag are as follows a reference area. The reference area for 2D airfoil is
the chord length of the airfoil (0.5 m). The reference
2FL area for the 3D wing is the planform area.21
CL ¼
V2 A Figure 3 shows that the Spalart–Allmaras model
comes the closest to replicating the lift coefficient
2FD curve of the wind tunnel test. It also calculates the
CD ¼
V2 A critical AOA at 13 , which is fairly close to the
actual value of 14 . The k–! SST model comes in
Siddiqi and Lee 5

1.8

1.6

1.4
Spalart-Allmaras

Coefficient of li (CL)

1.2
K-Epsilon Realizable
1

0.8 K-Omega SST

0.6
X-FOIL
0.4
Wind Tunnel
0.2

0
0 5 10 15 20
Angle of aack (°)

Figure 3. CL vs AOA plot for Spalart–Allmaras, k–" realizable, and k–! shear stress transport (SST) models alongside XFOIL and
experimental wind tunnel results.

0.16

0.14

0.12
Spalart-Allmaras
Coefficient of drag (CD)

0.1
K-Epsilon Realizable
0.08

0.06 K-Omega SST

0.04 X-FOIL

0.02
Wind Tunnel

0
0 5 10 15 20
-0.02
Angle of aack (°)

Figure 4. CD vs AOA plot for Spalart–Allmaras, k–" realizable, and k–! shear stress transport (SST) models alongside XFOIL and
experimental wind tunnel results.

the second place. Although it overpredicts the lift simulations. The simulation a 2D NACA 0012 airfoil
coefficients of both the XFOIL and the wind tunnel also showed that the k–! SST gave the best results.22
data at higher AOA, it does manage to capture the
critical AOA relatively well (between 12 and 13 ).
3D wing simulations
The k–" realizable model performs the worst as it
not only overpredicts the lift coefficients but also pro-
vides the critical AOA of 15 . The mesh generation techniques and the FLUENTÕ
Figure 4 presents an opposite trend. The Spalart– setup procedure from the 2D airfoil analysis are then
Allmaras shows the largest deviation from both the translated into the 3D computational domains for the
XFOIL and wind tunnel data. The k–" realizable full wing simulations. The five wing designs considered
model comes in the second with a drag coefficient profile in this paper are created SolidWorksÕ . Depending on
that matches that of the wind tunnel at high AOAs. The the geometry of the wings, the number of elements in
k–! SST model predicts the drag profile more accur- the 3D computational domains ranges from 1.8 million
ately than k–" realizable model does. At high AOA, the to 2.4 million. The 3D fluid domain is a bullet-type
k–! SST yields almost the same drag coefficient values fluid domain. A symmetrical boundary condition is
as that from the wind tunnel. From the 2D studies, the used to model only one wing out of the two symmetric
k–! SST turbulence model is chosen for 3D full wings, which drastically reduces the size of the mesh.
6 Proc IMechE Part G: J Aerospace Engineering 0(0)

The 3D mesh of one of the wing designs is shown in Table 2. Airfoil data (source: www.airfoildb.com).
Figure 5.
Maximum
Thickness Maximum lift lift-to-drag Stall
Modeling the wings Name (%) coefficient (L/D) angle ( )

The original RQ-7 wing is denoted as ‘‘base wing’’. The NACA 4415 15.0 1.64 55.4 14.0
airfoil type for the base wing is the NACA 4415 with a EPPLER 559 16.0 1.72 69.7 14.0
rectangular planform and a chord length of 0.5 m. The EPPLER 560 16.1 1.83 76.0 14.5
wingspan of the base wing is 4.3m, which yields an AR EPPLER 856 18.1 1.76 62.0 14.5
of 8.6. The performance of the other wings is then GOE 510 13.7 1.77 68.1 14.0
compared with the base wing. The coefficients of lift GOE 528 12.4 1.82 57.9 14.0
and drag for a wing should not be confused with those
from the 2D airfoil analysis. Anderson uses Prandtl’s
classical lifting line theory to explain the difference Addition of winglet. A winglet is a small nearly vertical
between the aerodynamic characteristics of an airfoil lifting surface that is typically attached to the tip of a
and a wing.21 This difference is due to the generation of wing. The generation of vortices at the wing tip
a trailing vortices caused by the component of flow reduces the overall lift generated along the local
along the span wise direction of the wing. edges of the wing. Winglets are used to redirect
these vortices and generate additional lift.23–28 The
Changing the airfoil. The second wing design incorpor- fifth wing design incorporates a winglet which has a
ates the EPPLER 559 airfoil and is denoted as the height of 130 mm and is termed as ‘‘winglet wing’’.
‘‘Eppler Wing’’. The purpose of this wing is to inves- The details of all of the wing designs are summarized
tigate the effects of changing the airfoil shape. Some in Table 3. An isometric view of the five wings
EPPLER series airfoils (EPPLER 559, EPPLER 560, designed in SolidWorks is shown in Figure 6.
EPPLER 856) and GOE series airfoils (GOE 510,
GOE 528) have similar stall characteristics and
higher lift/drag ratios when compared to the NACA
Simulation results
4415 airfoil.12 The EPPLER 559 airfoil is selected to As mentioned previously the 3D wing simulations are
determine the efficacy of employing such airfoils for conducted under two different conditions. The CL/CD
UAV applications. The characteristics of these airfoils ratios are plotted as functions of AOA and the free-
are summarized in Table 2. The Thickness of the airfoil stream velocities, each of which represents one of the
refers to the distance between the upper and lower sur- two conditions.
faces, also measured perpendicular to the chord line.21 Climbing: Fixed air speed of 105 mph with various
AOAs ranging from 0 to 16 . The calculated
Changing the planform. Besides the rectangular plan- Reynold’s number is 1,606,698. CL/CD curve for this
form, another popular choice for subsonic flight appli- condition is plotted in Figure 7.
cations is tapered planform. There are many Acceleration: Fixed AOA of 6 with the air speed
advantages offered by the tapered planform such as varying from 55 mph to 135 mph. The corresponding
improved lift distribution, better lateral control, Reynold’s numbers are 841,603 and 2,065,755,
reduced weight, and a reduction in the overall drag. respectively. CL/CD curve for this condition is plotted
The downside is that they are more expensive from a in Figure 8.
manufacturing point of view than the rectangular plan-
forms.23,24 This wing is referred to as ‘‘tapered wing’’ Effect of changing the airfoil. During climbing, the Eppler
and has a taper angle of 3.5 . The AR of a tapered wing initially generates more lift forces than the base
planform is calculated from the following expression. wing does. However, it experiences a sudden drop in
the CL/CD ratio. This indicates even though the
S2 Eppler wing generates more lift, it also generates
AR ¼ Cr þCt more drag. The CL/CD curve during climbing shows
2 S
a sudden drop starting at 2 . At 4 , the ratio is the
where S is the wingspan, Cr is the root chord length, same as that of the base wing after which it keeps on
and Ct is the tip chord length of the wing.25,26 decreasing. The acceleration plot also shows a similar
trend that as the velocity increases, the benefit offered
Changing the aspect ratio. In general, the tapered plan- by the Eppler wing also diminishes.
form creates less drag than the rectangular one but at
the same time it also generates less lift forces. The AR Effect of changing the planform. A significant change in
of the tapered wing is increased to 12.4 (0.2 m incre- the aerodynamic characteristic has been observed
ment to the wingspan) in order to compensate for the when switching from the rectangular to the tapered
loss in lift.27 This wing is denoted as ‘‘extended taper planform. Despite the fact that the tapered wing uses
wing’’. the EPPLER 559 airfoil, it does not experience the
Siddiqi and Lee 7

Table 3. Wing design summary.

Design Airfoil Wingspan Aspect ratio Planform

Base wing NACA 4415 4.30 m 8.60 Rectangular

Eppler wing EPPLER 559 4.30 m 8.60 Rectangular
Tapered wing EPPLER 559 4.30 m 11.66 Tapered
Extended taper wing EPPLER 559 4.50 m 12.41 Tapered
Winglet wing EPPLER 559 4.63 m 14.25 Tapered þ Winglet

Figure 5. 3D mesh of tapered wing.

Figure 6. Isometric view of all five wing designs in SolidWorksÕ .

sudden drop in lift as the Eppler wing does during increases with a substantial gap between the tapered
climbing. It is also worth mentioning that even and the base wing. The tapered wing curve also indi-
though the CL/CD curve of the tapered wing is cates that the dip at 2 by the Eppler wing with
higher than the base wing, this does not mean that increasing velocity is also eliminated to some extent.
the wing generates more lift forces (FL) than the base Even though the lift forces generated are less, the
wing does. From 0 to 5 AOAs the tapered wing does tapered wing has an advantage associated with a
indeed generate more lift but beyond 6 it is unable to light overall weight. The extent of the weight reduc-
do so. The drag forces (FD) on the other hand are tion remains unknown due to the classified nature of
drastically reduced thereby yielding higher CL/CD the UAVs wing design composition. A surface area
ratios. As the AOA increases, the tapered wing reduction of 26.5% and a volumetric reduction of
curve starts to come closer to the base wing but 46.3% from the base wing could give a general
does not dip below it. During acceleration, there is a estimation of how much weight reduction there
steady rise in the coefficient ratio as the velocity might be.
8 Proc IMechE Part G: J Aerospace Engineering 0(0)

21

19

17 Base Wing

15
Eppler Wing

CL / CD 13
Tapered Wing

11
Extended Taper
Wing
9
Winglet Wing
7

5
0 2 4 6 8 10 12 14 16
Angle of aack (°)

Figure 7. CL/CD plot for climbing.

10.5

10.3

10.1 Base Wing

9.9
Eppler Wing
CL / CD 9.7
Tapered Wing
9.5

9.3 Extended Taper

Wing

9.1
Winglet Wing

8.9
55 65 75 85 95 105 115 125 135
Speed (mph)

Figure 8. CL/CD plot for acceleration.

Effect of increasing the AR. The percentage reduction on than the base wing up to 12 . Both the climbing and
the area and volume for the extended taper wing is acceleration plots depict that the CL/CD curve of this
24.5% and 45.3%, respectively. Increasing the AR by wing are the highest amongst the rest.
6.41% yields some benefits with respect to lift. As the
aircraft gains altitude, the extended taper wing has a
Conclusion
slightly higher CL/CD ratio than the tapered wing
does. It follows the same pattern as the tapered The EPPLER 559 airfoil proves to be a viable airfoil
wing and starts converging at higher AOAs. The lift candidate for subsonic UAV applications. However,
generated by this wing is greater than the base it is worth noting that in the rectangular planform
wing up to 8 AOA. The acceleration plot also shows configuration, the Eppler wing (which uses the
that as velocity increases, the trend of the wing is EPPLER 559 airfoil) performs poorly against the
observed to be similar to the tapered wing albeit base wing (which uses the NACA 4415 airfoil). This
with higher CL/CD values. is due to the aerodynamic nature of the EPPLER 559
airfoil. As mentioned previously, airfoils that generate
Effect of winglet. With regard to the CL/CD curve, high lift also generate high drag as well. Thus, the
adding a winglet provides a similar magnitude of Eppler wing yields higher lift coefficients than the
improvement like the one observed for the extended base wing and also generates higher drag coefficients
taper wing. The winglet wing generates greater lift as well. In the climbing condition, the CL/CD curves
Siddiqi and Lee 9

of these two wings show that the Eppler wing begins Funding
to lose the benefit of employing the EPPLER 559 air- The author(s) received no financial support for the research,
foil from an AOA of 6 and onwards. The same effect authorship, and/or publication of this article.
of performance loss is also indicated during the accel-
eration condition, as the slope of the Eppler wing ORCID iD
starts to decrease when compared to the base wing. JW Lee https://orcid.org/0000-0002-1092-6662
This loss in performance, however, is corrected
when the different wing design components are chan- References
ged. The drag forces are reduced substantially when the 1. Jeff Calvert - US Air Force -FAA Liaison Jean-
planform is modified from rectangular to tapered. In Christophe M. RQ-7B Shadow Achieved Performance
both the climb and acceleration conditions, the tapered Model VerificationFaa, Philip Maloney, http://www.tc.
wing does not experience the same sharp decline in faa.gov/its/worldpac/techrpt/tctn11-8.pdf (2011).
performance that the Eppler wing does after an AOA 2. Mukesh R, Lingadurai K and Selvakumar U. Airfoil
of 6 . The benefits with regard to increased lift are shape optimization using non-traditional optimization
observed as the AR is increased (extended taper technique and its validation. J King Saud Univ - Eng Sci
wing) and when a winglet is added (winglet wing) 2014; 26.
albeit at a higher magnitude. The winglet wing shows 3. Zhang F, Chen S and Khalid M. Optimizations of air-
that it able to generate more lift than the base wing up foil and wing using genetic algorithm. In: 23rd congress
of international council of the aeronautical sciences,
to 12 AOAs while having less drag acting on it. The
Toronto, Canada, 2002.
interesting observation is that the decline in perform- 4. Holst TL. Genetic algorithms applied to multi-objective
ance experienced by the Eppler wing from 6 AOA and aerospace shape optimization. J Aerosp Comput Inform
onwards is not present in the tapered wing, extended Commun 2005; 2: 217–235.
taper wing, and the winglet wing despite the fact that 5. Son SH, Choi BL, Won WJ, et al. Wing design opti-
they employ the same airfoil. mization for a long-endurance UAV using FSI analysis
It is observed that the wings that employ the and the Kriging method. Int J Aeronaut Sp Sci 2016; 17:
EPPLER 559 airfoil have higher CL/CD values at 423–431.
low AOA (0 to 6 ) during climbing. The increasing 6. Liang Y, Cheng XQ, Li ZN, et al. Robust multi-
CL/CD values are also consistent during acceleration objective wing design optimization via CFD approxi-
mation model. Eng Appl Comput Fluid Mech 2011; 5:
as well. In the rectangular planform configuration
286–300.
(when no wing component changes are incorporated),
7. Esfandiari N and Alinezhad B. Subsonic aircraft wing
the Eppler wing performs poorly against the base conceptual design synthesis and analysis. Int J Sci Basic
wing. This, however, does not hold true for the Appl Res 2014; 13: 303–316.
tapered wing, extended taper wing, and the winglet 8. Khosravi S and Zingg DW. A numerical optimization
wing. These three wings offer better overall perform- study on winglets. In: 15th AIAA/ISSMO multidiscip-
ance when compared to the base wing during climbing linary analysis and optimization conference, Atlanta,
and acceleration conditions. Our investigation shows GA, USA, 16–20 June 2014, pp.1–14.
that airfoils with high lift coefficients like the 9. Wei Z and Meijian S. Design optimization of aero-
EPPLER 559 can indeed enhance the performance dynamic shapes of a wing and its winglet using modified
of subsonic UAVs only when paired with certain plan- quantum-behaved particle swarm optimization algo-
rithm. Proc IMechE, Part G: J Aerospace Engineering
forms, aspect ratios, or winglets. There exists plenty of
2014; 228: 1638–1647.
room for optimizing each of these wing designs. For
10. Sadraey MH. Aircraft design: a systems engineering
instance, the high drag induced by the airfoil can be approach. New York: Wiley, 2012.
reduced by an increase in the taper angle. High taper 11. University of Illinois at Urbana-Champaign. UIUC
angles can reduce the amount of induced drag but it Airfoil Data Site, http://m-selig.ae.illinois.edu/ads/
comes with the caveat of reducing the lift generated by coord_database.html (accessed 17 August 2018).
the wing. Many other airfoil designs exist that have 12. Airfoil Tools. http://airfoiltools.com/index (2018,
similar (if not better) stall characteristics compared to accessed 18 August 2018).
the NACA 4415 airfoil. For future work, the intro- 13. Ferziger JH and Peric M. Computational methods for
duction of additional wing design parameters such as fluid dynamics. 3rd ed. New York: Springer, 2002.
the geometric twist, wing incidence, and dihedral 14. Saryazdi SME and Boroushaki M. 2D numerical simu-
lation and sensitive analysis of H-Darrieus wind tur-
angle will be introduced. This will eventually lead to
bine. Int J Renew Energy Dev 2018; 7: 23–34.
the application of multiobjective optimization tech-
15. Salim SM and Cheah SC. Wall y þ Strategy for dealing
niques that will further improve the wing performance with wall-bounded turbulent flows. In: International
for subsonic UAV applications. multiconference on engineers and computer scientists,
Hong Kong, 18–19 March 2009, pp.1–6.
Declaration of Conflicting Interests 16. White FM. Viscous fluid flow. 2nd ed. New York:
The author(s) declared no potential conflicts of interest with McGraw-Hill, 1991.
respect to the research, authorship, and/or publication of 17. Benim A, Diederich M and Pfeiffelmann B.
this article. Aerodynamic optimization of airfoil profiles for
10 Proc IMechE Part G: J Aerospace Engineering 0(0)

small horizontal axis wind turbines. Computation 2018; 23. Sadraey MH. Aircraft design. New York: Wiley, 2013.
6: 34. 24. Romeo G, Frulla G and Cestino E. Design of a high-
18. ANSYS FLUENT 12.0 theory guide - 4.4.3 realizable - altitude long-endurance solar-powered unmanned air
model, http://www.afs.enea.it/project/neptunius/docs/ vehicle for multi-payload and operations. Proc
fluent/html/th/node60.htm (accessed 19 August 2018). IMechE, Part G: J Aerospace Engineering 2007; 221:
19. Drela M. XFOIL: An Analysis and Design System for 199–216.
Low Reynolds Number Airfoils. In: Mueller TJ (eds) 25. Wing Area, https://www.grc.nasa.gov/WWW/k-12/
Low Reynolds Number Aerodynamics. Lecture Notes in VirtualAero/BottleRocket/airplane/area.html (accessed
Engineering, vol 54. Berlin, Heidelberg: Springer, 1989. 13 November 2018).
20. Hoffmann MJ, Reuss Ramsay R and Gregorek GM. 26. Geometry Definitions, https://www.grc.nasa.gov/www/
Effects of grit roughness and pitch oscillations on the k-12/airplane/geom.html (accessed 13 November 2018).
NACA 4415 airfoil. Golden, CO: National Renewable 27. Gandhi K, Singh N and Gupta D. Aerodynamic char-
Energy Laboratory, 1996. acteristics calculation of rectangular, tapered and swept
21. Anderson JD Jr. Fundamentals of aerodynamics. New wing using CATIA and ANSYS. Int J Aerosp Mech Eng
York: McGraw-Hill, 1985. 2018; 4.
22. Eleni DC. Evaluation of the turbulence models for the 28. Kasolang SI, Azlin CF, Taib MCF, et al. CFD analysis
simulation of the flow over a National Advisory of winglets at low subsonic flow. In: Proceedings of the
Committee for Aeronautics (NACA) 0012 airfoil. world congress on engineering, Vol. 1, 6–8 July 2011,
J Mech Eng Res 2012; 4: 100–111. London, UK.

View publication stats