You are on page 1of 12

Aerospace Science and Technology 29 (2013) 339–350

Contents lists available at SciVerse ScienceDirect

Aerospace Science and Technology


www.elsevier.com/locate/aescte

Design, performance evaluation and optimization of a UAV


Spyridon G. Kontogiannis a,∗,1 , John A. Ekaterinaris a,b,2
a
School of Mechanical & Aerospace Engineering, University of Patras, 26500 Patras, Greece
b
FORTH/IACM Heraklion, Crete, Greece

a r t i c l e i n f o a b s t r a c t

Article history: Following current trends towards UAV innovative designs, a small size light UAV was designed,
Received 30 January 2013 constructed and tested in flight. The purpose of this light UAV aircraft is to serve as a reconnaissance
Received in revised form 12 April 2013 plane capable of carrying state-of-the-art photography and video equipment. The UAV is also equipped
Accepted 16 April 2013
with first person view (FPV) systems for easier and more accurate control by a remote user. For the initial
Available online 22 April 2013
configuration, apart from classical conceptual design procedures, some new relations were developed and
Keywords: used. For the preliminary design, linear aerodynamic performance and stability analysis was carried out
UAV first. Next, the aerodynamic characteristics and efficiency of the airfoil section, the wing, and the full
Design configuration were evaluated using CFD. Finally, further improvements of the aerodynamic efficiency of
CFD the full configuration were carried out through computational optimization.
Optimization © 2013 Elsevier Masson SAS. All rights reserved.
Aerodynamics
Flight mechanics
FPV systems
Surveillance
Winglets
Aerial photography

1. Introduction ing, in a specifically formed fuselage the necessary electronic and


video equipment. In the fuselage, a high definition camera (typi-
Significant efforts invested in unmanned air vehicle (UAV) tech- cally weighing under 2.5 kg), a video transmitter with its batteries,
nology led to a wide variety of new applications such as aero- and the FPV systems are placed. The resulting empty weight of
photography and surveillance. Parallel advances in avionics and the proposed configuration is approximately 1.9 kg. The maximum
electronics applied to modern UAV technology combined with payload can reach up to three times the UAV empty weight. An ad-
rapid developments in video and photographic equipment resulted ditional stringent design requirement is to keep the UAV’s stalling
in significant weight reduction, enhanced efficiency, and quality speed down to very low levels, under 12 m/s, in order to ensure
improvements. All these elements enabled modern UAV designs to precise and high quality videos and photos. Apart from that, the
cover a wider spectrum of applications from large monument and design allows an easy and practical assembly process by keeping
building photographs, to city maps and road traffic information. all parts rather small and easy to store.
Until now, common solutions were helicopters, causing however In the following sections, a brief summary of the UAV concep-
considerably high costs due to large fuel consumption. This paper tual and preliminary design [14], aerodynamic analysis [15], flight
suggests a more practical and economical way of dealing with the mechanics, linear stability analysis and performance [10,2] is pre-
matter, by presenting the design, construction, optimization, and sented. After the preliminary design, due to lack of a relatively
in flight testing of a light, electric powered UAV prototype. An- large wind tunnel facility a CFD analysis was conducted. For val-
other advantage of this concept, apart from its low cost, is that idation purposes, initially the E387 airfoil section that is similar
it can be operated from a distant room using FPV (First Person to the one selected for the present UAV was analyzed using well
View) systems. Our UAV meets the above specifications by carry- established methods [4,17] for low Reynolds number airfoil flows.
The computed results assisted to evaluate the level of confidence
and existing limitations of CFD numerical models to effectively
*
Corresponding author. simulate low Reynolds number transitional flows [9]. The commer-
E-mail addresses: spkontogiannis@gmail.com (S.G. Kontogiannis),
cial software FLUENT [5] and the Stanford University open source
ekaterin@iacm.forth.gr (J.A. Ekaterinaris).
1
Student.
code SU [19] were used to compute airfoil and wing flows. For
2
Professor. Currently at Aerospace Engineering, Embry-Riddle Aeronautical Uni- the full UAV configuration, the commercial software FLUENT was
versity (ERAU), Daytona Beach, FL 32114. Associate Fellow AIAA. mostly used and the open source SU code was used only for few

1270-9638/$ – see front matter © 2013 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ast.2013.04.005
340 S.G. Kontogiannis, J.A. Ekaterinaris / Aerospace Science and Technology 29 (2013) 339–350

Table 1
UAV specifications.

Maximum W E 2 kg
Maximum T.O. runway 60 m
Motor/power Electric – P < 500 Watts
Operating payload W PL 6 kg
V stall Under 12 m/s
Specific design to include video equipment and relative systems

selected cases. A more improved and efficient configuration design


has been achieved by computational experimentation and evalu-
ation of possible changes [12,6] and add-ons for wing and full
aircraft configuration, such as winglets [11,8] and fairings, keep-
ing the cost of the development down to low level.
The work presented in this paper was later used as a basis
for developing a similar UAV prototype for a student competi-
tion, http://www.rcgroups.com/forums/showthread.php?t=1347952
Air Cargo Challenge 2011, and was presented as a conference pa- Fig. 1. Calculation procedure of acceptable aero-surfaces.
per [6]. The limitations set by the competition were (i) UAV weight
1.8 kg (ii) takeoff and landing distances 60 m and 120 m, respec- aircraft found in the literature do not apply in such a small UAV.
tively (iii) demonstration flight with maneuvers. The rest of this As a result, new relations had to be developed which would enable
paper is organized as follows. In Section 1, the conceptual design to estimate with precision the weight of all aerodynamic surfaces
is briefly presented. In Section 2, the preliminary design is shown. with respect to their area. The empty weight of the aircraft is given
Flight mechanics calculations for the preliminary design based on by:
linear theory are presented in Section 3. The aerodynamic analy-
sis using CFD and optimization is presented in Section 4. Summary W E = W motor + W batteries + W electronics + W propeller + W L.G.
and conclusions are given in Section 5.
+ W wing + W V + W H + W boom + W fuselage (1)
2. Conceptual design The weight of the first four components was readily obtained
with precision since the electrical and propulsion equipment was
2.1. Design specifications and limitations accessible giving a total weight of 0.7 kg. The Landing Gear (LG)
weight, constructed from aluminum, was also accurately estimated
The conceptual design process started directly from the new from CAD to be 0.95 kg. As far as the boom is concerned, with
specifications required for the UAV. In the absence of design lim- a first estimation of its length resulting in 1.1 m, satisfying both
itations set by the industry or a client, an extensive research in functionality and easy transportation of the aircraft, and carbon
possible photographic UAV’s was carried out and the results that fiber as a material (with a density of 0.95 kg/m) its weight will be
conform to our specifications are summarized in Table 1. An ultra- about 0.105 kg. The camera fairing and housing will be made out
light UAV design was selected and the stalling speed was kept of balsa wood and carbon fiber resulting in a weight of 0.08 kg.
down to low levels in order to ensure good quality photos and From Eq. (1), the weight of the wing, horizontal and vertical sta-
filming, while at the same time allowing an easy transportation bilizer should not exceed 1.02 kg. In order to satisfy this require-
from the user. ment, the iterative procedure shown in Fig. 1 was followed. This
Flexibility in use, especially in a city environment, demands uses the weight statistical relations, as well as other newly devel-
short takeoff runways. Therefore, the takeoff distance was set to oped equations that relate empennage surface, wing surface and
a maximum of 60 m. An electric power plant for the propeller dimensionless tail volumes V H and V V .
was selected since it offers low operating costs, simplicity in in-
stallation, use, and because it is eco-friendly. The electronic and 3. Preliminary design
video equipment that the aircraft should typically carry does not
exceed a total weight of 3 kg. However, a 6 kg payload was set as The first objective of the preliminary design is to determine
a target for the present UAV configuration because a design for a the optimum combination of the wing lift coefficient, surface and
larger payload allows possible changes to carry more sophisticated planform geometry. Major constraints in the design are the stalling
photographic equipment in the future while at the same time the speed that should not exceed 12 m/s and the takeoff runway that
stalling velocity V stall ∼ ( W / S )/(C L )max is kept at low levels, e.g. should not exceed 60 m. These two constraints dictated an iter-
sufficiently high (C L )max is required. ative design process that resulted in a wing surface of 0.9 m2
(leading to no overweight) and a C L = 1.3 for a stalling speed of
2.2. Weight estimations 10 m/s. For weight reduction, simplicity of construction, and low
cost considerations no high lift devices (slats or flaps) were used
The procedure used for UAV weight estimation differs in two since it was found that a high lift, low Reynolds number airfoil
aspects from the typical procedures suggested in the literature for section could satisfy both constraints. The X-Foil code [4] was used
general aircraft. Firstly, while the payload was estimated the empty during the selection process of the airfoil section. It was concluded
weight will not be estimated but is set as a specification. For the that the airfoil that best matches our demands is the E420 high
present design, it is of great importance to obtain accurate weight lift airfoil.
estimations of the components and their distribution so that the Depending on camera weight, the stalling speed could vary
weight of the aircraft itself does not exceed 2 kg. Secondly, com- from 7.5 m/s to 12 m/s resulting in a Reynolds number based on
pared to the typical methods, this weight calculation procedure the wing’s mean aerodynamic chord, Rec , for takeoff and landing
varies in terms of the way used to estimate the weight of the in the transitional flow range 1.3 × 105 < Rec < 2.4 × 105 . As the
components. We considered that traditional statistical relations for airfoil is characterized by high camber and high pressure gradient
S.G. Kontogiannis, J.A. Ekaterinaris / Aerospace Science and Technology 29 (2013) 339–350 341

Fig. 2. Wing design proposals.

Table 2
Wing characteristics.

Angle of sweep Λc/4 0◦


Root airfoil thickness (t /c )c 0.1429
Tip airfoil thickness (t /c )t 0.1429
Taper ratio λw 0.4
Fig. 3. Fuselage and camera bay design.
Root chord length c r (m) 0.375
Tip chord length c t (m) 0.15
Wing planform area S W (m2 ) 0.9
Aspect ratio AR 10

along its chord, the flow at these low Reynolds numbers is con-
sidered to be transitional. The design of the wing continues with
the selection of the most suitable aspect ratio (AR) for the UAV
wing. It is important to target a high AR value so that our aircraft
can reach the best performance possible, keeping drag and power
losses to a minimum. Structural dynamic calculations (not shown
here) determined that a maximum AR of 10 could be reached for
the low cost materials selected for construction. Note that higher
AR values could be achieved but would not enable sustained use
from an inexperienced UAV user. Fig. 4. CAD of the baseline aircraft configuration.
The wing planform design concluded with the estimation of the
optimum taper ratio that would result in an elliptical wing ap- used and the horizontal stabilizer rotates along its axis to produce
proximation. This was found to be 0.4 for the chosen AR = 10, pitch up or pitch down. The vertical stabilizer uses a symmetrical
combined with the semi-tapered planform (proposal 2) shown in NACA 0012 airfoil, and its surface area is 0.056 m2 as calculated
Fig. 2. The semi-tapered planform offers the possibility of an easy in the conceptual design. Rudder design is further described in the
and practical transportation as it consists of 3 retractable parts of following sections dealing with directional stability. Finally, pitch-
1 m span each. ing moment analysis [10,2] around the center of gravity (CoG) in
Enhanced lateral stability necessary for the aircraft in order to cruise showed that the optimum boom length is 0.8 m.
ensure video and photo quality is achieved by a high cantilevered
configuration so that no dihedral angle Γ W is needed. In this ba- 4. Flight mechanics
sic configuration, there is also no twist angle due to manufacture
constraints, but the twist angle will be examined in the optimiza- Based on the conceptual and preliminary design as well as the
tion process. The angle of incidence of the wing is 5.5 deg where aerodynamic analysis of the previous sections, the aircraft configu-
the aerodynamic efficiency of the airfoil is maximized, while at the ration has taken its complete form as shown in Fig. 4. Notice that
same time a necessary wing lift coefficient of C L = 1.3 is obtained. in the trapezoid sections of the wing, a small sweep angle appears.
The characteristics of the wing of the basic configuration are sum- This was dictated by structural integrity calculations demanding a
marized in Table 2. zero sweep angle in the (t /c )max line. Drag components and total
drag were calculated to estimate power consumption and thrust
3.1. Fuselage design/video equipment weight required during each phase of flight and to obtain the optimum
propeller choice. To ensure stability and control, longitudinal, lat-
Use of conventional fuselage shapes for the present UAV is not eral and directional stability were considered and the horizontal
required. Therefore the fuselage is designed in such a way that it stabilizer, ailerons and rudder were calculated. Apart from takeoff
would allow practical storage of the equipment and the camera, as analysis, flight performance including maneuvering, flight envelope
shown in Fig. 3. A maximum of six batteries can offer a flight time and climb and descent phase were also calculated.
of 1.5 hrs. In order to reduce drag, the aerodynamic shape PARSONS
F2-49 [13] was selected. 4.1. Drag coefficients calculation

3.2. Empennage design In this linear analysis, the wing, the empennage, the landing
gear, and the camera bay drag components were considered and
Among three examined configurations (conventional, T -tail, overestimated by 10% to include phenomena such as interference
V -tail), the conventional T -tail configuration was selected and drag and propeller induced flow.
connected to the wing via a boom (carbon tube). For the hori-
C D total = 1.1(C D wing + C D emp + C D fus + C D L.G. ) (2)
zontal stabilizer, the initial surface area predicted from conceptual
design relations, was S H = 0.082 m2 , however it was redesigned so Analyzing the drag component of the wing and the tail as function
that S H = 0.099 m2 when calculated in the pitching moment anal- of lift, we obtain the drag polar:
ysis. The airfoil chosen was a NACA5412 that being inverted can
produce sufficient down force to ensure longitudinal stability at
C D = C D 0 + kC L2 (3)
low AoA. For the sake of simplicity of construction no elevator was where C D 0 = 0.010015 and k = 1/(π A Re) = 0.03233.
342 S.G. Kontogiannis, J.A. Ekaterinaris / Aerospace Science and Technology 29 (2013) 339–350

Fig. 5. Takeoff distance required as a function of payload and runway friction coef-
Fig. 7. Flight envelope diagram.
ficient.

friction coefficient case of Cr = 0.02, the runway specification is


met and a 9 kg payload can be lifted. Thrust requirement calcu-
lations for the climb phase shown in Fig. 6 were also necessary
for the appropriate selection of the propulsion system. From the
calculations, the required thrust as a function of the payload and
the climb angle is extracted. Since the propulsion system can pro-
duce a static thrust of 25 N, the results ensure that motor power
is sufficient for this type of aircraft. The flight envelope shown in
Fig. 7 provides information about safe flight conditions regarding
stall and structural damage due to short turn radius. It is noted
that the aircraft can withstand side forces up to 2g.

4.4. Longitudinal stability

Longitudinal stability analysis ensures balance in every situation


that moves aircraft from the trim position. From the analysis, the
characteristics of the horizontal stabilizer were calculated, as well
as the neutral point position and the static margin.
 
dC m f dC L W hcg hac
Fig. 6. Thrust required versus payload for various climb angles. C mcg = C mof + af + aw −
da f da c c
 
4.2. Propeller analysis/propulsion system dC Lt dC Lt
− V Hn at − (4)
dat dat
For the selection of the most efficient propeller for the air-
craft, both computational and experimental tools were used. For The final dimensions of the horizontal stabilizer are shown in Ta-
the initial phase of the design, the JAVA prop. software was used. ble 3. Longitudinal stability conditions are satisfied, e.g.
Propellers of various diameter and pitch were tested with JAVA C m,α |α =0 > 0 and ∂ C m,α /∂ α < 0.
prop. software. Three of them that were found superior in every re-
spect, were experimentally tested in a simple custom made device. 4.5. Lateral stability/control
After the necessary tests, the 13 × 7 propeller (13 inches diam-
eter, 7 inches pitch) was chosen because it produced more thrust Lateral stability analysis is used to calculate the necessary
than the other two. In addition, it was found that this propeller re- aileron surface and deflection to cancel the torque of the motor.
tained high thrust level for a wider range of flight velocities. Drag As lateral and directional stability interact with each other, rolling
analysis has shown that the power of the motor to drive this pro- tendencies induced from yaw and moment created during sideslip
peller does not need to exceed 450 W. As a result, the AXI GOLD (skid) had to be taken into account as well. The calculations were
2826/10 motor was chosen, since it is one of the best available in guided by the above conditions and the resulting dimensions of
the market. each aileron are also shown in Table 3.

4.3. Aircraft performance and flight characteristics 4.6. Directional stability/control

To verify that the propulsion system selected is sufficient for Vertical stabilizers should be able to cancel roll induced yaw
the UAV needs and a takeoff runway distance under 60 m is moments (adverse yaw moments), and counteract intense cross-
achieved, a takeoff analysis had to be conducted. The analysis con- wind conditions. Another crucial factor, which is difficult also to
siders various camera weight scenarios as well as various friction predict even with sophisticated CFD analysis, is the flow of the
cases. Results shown in Fig. 5 demonstrate that for a typical rolling propeller which could create unpredictable destabilizing moments
S.G. Kontogiannis, J.A. Ekaterinaris / Aerospace Science and Technology 29 (2013) 339–350 343

Table 3
Characteristics of aircraft control surfaces.

Horizontal stabilizer Vertical stabilizer Ailerons


Airfoil NACA 5412 Airfoil NACA 0012 Planform area S A (m2 ) 0.054
Aspect ratio ARH 5.2 Aspect ratio ARV 2.07 Aspect ratio ARA 4.28
Span bH (m) 0.72 Span bV (m) 0.37 Span bA (m) 0.48
Root chord c rH (m) 0.198 Root chord c rV (m) 0.198 Root chord c rA (m) 0.16
Root chord c tH (m) 0.08 Root chord c tV (m) 0.158 Root chord c tA (m) 0.04
I H (deg) 4 Root rudder chord c rR (m) 0.108 SA/SW 0.12
Root rudder chord c tR (m) 0.071

that are difficult to handle. Extensive analysis and experimentation


led to the specifications summarized in Table 3.

5. Aerodynamic analysis with CFD and optimization

After the preliminary aerodynamic analysis and design were


conducted, a CATIA model was constructed and the necessary
structural analysis was carried out. Prior to construction the UAV
design analysis was completed with computational fluid dynamic
(CFD) tools. A CFD analysis of the full UAV configuration de-
scribed in Sections 1–3 was also performed. Furthermore, compar-
isons with analytical/semi-analytical methods and computational
results were carried out. With the numerical models validated and
trimmed, the redesign and optimization followed. The optimiza-
tion process focuses on the maximization of the aerodynamic effi-
ciency, L / D, with simultaneous lift increase and drag decrease so
that power consumption is minimized, endurance is increased and
flight performance is improved. The optimization process is limited Fig. 8. Mesh comparison for the E420 airfoil.
by structural integrity parameters not discussed in the present pa-
per.

5.1. Numerical approach

The first step in the computational analysis process is to as-


sess the numerical requirements (mesh topology and mesh density
characteristics, numerical schemes, turbulence models, etc.) using
experimental data for validation. Appropriate grids [3] were gen-
erated for the airfoils, the wing and the full configuration. Con-
sidering that transitional [20,16,7] low Rec flow is expected in all
flight conditions of the UAV special attention was given to the
near wall mesh quality. For the airfoils, intermediate and fine 2D
meshes were generated with the latter using approximately 68.000
cells. A grid independence study (see Figs. 8–9) was conducted. It
was found that the intermediate size mesh provided sufficient res-
olution for the needs of the simulation. A distance y f of the first
layer of cells away from the wall of y f = 10−5 c was set. Using
these meshes the resulting y + obtained for the computed solu-
tions never exceeded the value of 1, most of the times taking
values of y + = 0.2–0.4.
Fig. 9. Mesh comparison for the E420 airfoil.
In a similar manner, 3D meshes for the wing and the full
aircraft were generated as shown in Figs. 10–11. For the wing,
approximately two million elements were used. For the full UAV
configuration, approximately 4.5 × 106 elements were used for
symmetric (half body) numerical solutions. The same wall spac-
ing restrictions as the 2D meshes were applied for the full aircraft
as well, and the maximum computed value of y + approached 1.

5.2. CFD analysis

The CFD analysis for the initial configuration, which served also
as a validation study, was obtained in order to ensure that real-
istic results for the optimization and the final configuration could
be computed. The validation procedure begins with basic 2D air-
foil flows. This allows us to calibrate the numerical models with Fig. 10. Wing surface mesh and near wall layers.
344 S.G. Kontogiannis, J.A. Ekaterinaris / Aerospace Science and Technology 29 (2013) 339–350

Fig. 11. Full configuration volume mesh.

Fig. 14. E420 airfoil results.

Fig. 12. Surface pressure coefficient distribution, Cp, for the E387 airfoil at Rec =
2 × 105 .

Fig. 15. Baseline wing results comparison.

in Fig. 12 in order to determine the turbulence model that best


describes the transitional flow field. Turbulence modeling in tran-
sitional flows is not an easy task and in this case optimum results
are obtained using the k–ω SST transitional model. This particular
model appears to be the most suitable and will be used for the rest
of the analysis and optimization. However comparisons with nu-
merical solutions obtained with the S–A model will be also shown.
The 2D simulations conclude with E420 calculations. Having vali-
dated the numerical models with the E387 section, the results of
the E420 shown in Figs. 13–14 are considered to be accurate. They
can now provide a basis for the analytical calculations of the wing
as well as for the redesign process that was entirely based on CFD
analysis.
Three dimensional computational results were obtained for the
baseline wing to determine its aerodynamic characteristics, as well
Fig. 13. E420 airfoil results. as to be compared with the analytical calculations that used the
classical Prandtl–Lanchester theory [1]. This comparison is shown
low computational cost problems using the existing detailed ex- in Figs. 15–16. Clearly, there is good agreement in the linear region
perimental information. Unfortunately, low Reynolds number ex- between theory and computation, while the initial specification
perimental results of the E420 airfoil that was used for the present of C L = 1.3 is satisfied. The computational results of the baseline
UAV configuration were not available. Therefore, another very sim- wing shown in Figs. 17–18 are indispensable data for the develop-
ilar airfoil shape had to be used for validation and calibration ment of the final design as well as a reference value to evaluate
purposes. The selected section was the E387 airfoil, for which low the efficiency of every new wing design.
Reynolds number transitional flow, experimental measurements The final step in the CFD analysis is the simulation of the full
with partial flow separation and presence of a separation bubble configuration which serves as a comparison study in a similar way
starting at approximately mid-chord distance are available. The ex- to the wing mentioned above. Therefore, baseline full configura-
perimental results for Rec = 105 are compared with computations tion results are directly presented in comparison with results of
S.G. Kontogiannis, J.A. Ekaterinaris / Aerospace Science and Technology 29 (2013) 339–350 345

Fig. 19. Planform results comparison.

Fig. 16. Baseline wing results comparison.


These are mainly the region behind the propeller, the fuselage and
boom connection, and the wing tips. By closely examining the flow
structure in these regions and identifying where separations occur,
we were able to redesign these sections of the aircraft and reduce
the extent or to completely eliminate regions of separated flow.
For example, it was found that in the propeller shaft, a bounded
vortex occurs across its perimeter. This vortex could be eliminated
with the use of a specific curvature in the start of the shaft, be-
hind the propeller. To decrease the intensity of the interference
and separation vortices between the wing, the boom and its con-
nection, a fairing was added. The situation is a bit more complex
near the wingtips. To reduce wingtip vortices, winglets were used
as it is shown in the following section, while successive CFD sim-
ulations of the wing provided the necessary feedback to guide the
final design process.

5.3. Optimization

Fig. 17. Baseline wing results.


The optimization following CFD analysis aims at reducing
power requirements and increasing endurance by maximizing the
L / D ratio without compromising high lift at low speeds. An ob-
stacle towards this goal is the structural integrity of the wing.
As a result, thresholds were set for the optimization process so
that the AR and wingspan do not increase. Therefore AR is held
fixed and optimization at this stage only focuses on maximizing
span efficiency and simultaneously achieving drag reduction. Op-
timization results can be applied for a higher AR wing and/or for
higher wing loading W / S once the structural integrity consider-
ations would allow it. The current wing optimization procedure
focused on designing, examining and reevaluating possibilities of
improved wing planform, geometric twist, and use of winglets. The
final UAV configuration consisted of the optimized wing configu-
ration and aerodynamic fairings behind the propeller, fuselage and
empennage to reduce the flow separation and vortex strength ob-
served in CFD analysis.
i. Planform. As AR and the span, b, of the wing cannot change
nor redesigned, effort is focused in optimizing the planform shape,
e.g. to increase Oswald efficiency factor, e, and span efficiency [18].
Fig. 18. Baseline wing results.
Manufacturing and cost considerations exclude the use of an ellip-
tic planform and the leading edge (LE) or trailing edge (TE) lines
the final full configuration at the end of the present paper. The are set to be straight. Because the surface area and the wingspan
primary goal of the CFD analysis however, was to provide infor- are fixed, the taper ratio also remains fixed on its optimum for
mation regarding the flow field near the baseline aircraft so that this AR value. As a result, the planform problem reduces to finding
efficient and effective redesign and optimization could be made the optimum leading edge line angle. Apart from the initial design,
possible. Close examination of the computed results showed that two more configurations were considered: Zero sweep LE line and
in certain regions strong flow separations and wake vortices occur. Schumann planform. Figs. 19–20 show that the lift coefficient C L
346 S.G. Kontogiannis, J.A. Ekaterinaris / Aerospace Science and Technology 29 (2013) 339–350

Fig. 23. Winglet configurations.

Fig. 20. Planform results comparison.

Fig. 24. Wing with winglets comparison.

has not been changed in any position of the wing, as a result the
twist is only geometrical. The appropriate washout distribution is
calculated in two ways, Glauert’s method and using direct calcu-
lation. In Glauert’s method, the C L constraint set ensures that the
Fig. 21. Twist results comparison. twisted wing meets the specifications set in the conceptual de-
sign. In direct calculation of twist, angle of incidence distribution
was calculated directly from the elliptical lift distribution equa-
tion. A comparison of these two configurations and the initial wing
design is shown in Figs. 21–22. The AoA of the twisted wing is
measured from the root. As expected due to twist, lift is reduced
for a constant AoA. However, a constant lift coefficient is followed
by a considerable increase in aerodynamic efficiency.
iii. Winglets. Span efficiency increase and drag reduction is also
achieved by reducing tip vortices via specially designed wingtip
devices ramified in three directions: Endplates, Hoerner tips and
Winglets were employed. It is important that the winglet designed
does not interact with the wing producing interference vortices.
A total of eleven (11) configurations were examined, each one us-
ing feedback from previous configurations. From all configurations
examined, the four configurations shown in Fig. 23 were dominant
over the rest in every respect simultaneously considering (high L,
low D, and high L / D). These four designs produced simultane-
ously more lift increase and drag decrease compared to the others.
Fig. 22. Twist results comparison. From Figs. 24–26 and Table 4 that summarizes all winglets’ re-
sults, it is clearly seen that Winglet Nr. 4 is the optimum design.
is not sensitive to planform in the linear region but zero sweep This winglet combines all the positive characteristics encountered
reduces C D in the same region. in other winglets as it reduces the most strength of wingtip vor-
ii. Twist. As the wing planform approaches the elliptical shape, tices while at the same time does not appear to generate inter-
the elliptical lift distribution can be achieved with the appropriate ference vortices due to its specially designed curves for a wide
twist. Manufacturing and aerodynamic considerations set a limit range of flight conditions examined. The effect is clearly displayed
in twist angle variations (as flow separation and stall could occur in Figs. 27–28 which provide a comparison of the flow field of the
if a critical AoA is exceeded). High lift is crucial, and airfoil E420 baseline and the final wingtip region. It is obvious that the wingtip
S.G. Kontogiannis, J.A. Ekaterinaris / Aerospace Science and Technology 29 (2013) 339–350 347

Fig. 25. Wing with winglets comparison.


Fig. 27. Flow field and vortices near the final wingtip.

Fig. 26. Wing with winglets comparison.

vortices strength is greatly reduced by the use of the specifically


designed winglet. Fig. 28. Flow field and vortices near the final wingtip.

5.4. Final configuration

Geometric twist when combined with optimum planform and


winglets did not show significant improvements. As a result, twist
has not been used in the final configuration of the wing. Apart
from the new wing, three aerodynamic fairings were tested. It
was found that diminished vortex strength behind the fuselage
and the empennage could be achieved. It also became clear from
the computed results that the aerodynamic efficiency of the full
configuration shown in Figs. 29–32 could be enhanced. It is im-
portant to note that simulations of the full configuration modeled
the propeller both as an interior and as a fan, simulating no power
descent-gliding and full throttle conditions respectively. Figs. 33–
34, apart from showing the propeller flow in full throttle condi-
tions, provide a comparison of the baseline and the final aircraft
configuration, as well as an illustration of the beneficial effect of
the fairing behind the wing. It is evident that the positive flow ve-
locity observed in that region in Fig. 33 no longer exists in Fig. 34,
with the vortex formed been eliminated. A display of the pres-
sure distribution on the surface of each configuration for the full
throttle conditions simulation is shown in Figs. 35–36. After a close Fig. 29. Baseline–final aircraft comparison.
348 S.G. Kontogiannis, J.A. Ekaterinaris / Aerospace Science and Technology 29 (2013) 339–350

Table 4
Winglet characteristics.

Wingtip C L@L / D max C D@L / D max C L /C D max C L@6 deg C D@6 deg C L /C D@6 deg Increase (%)
Winglet Nr. 1 1.052 0.0844 12.46 1.396 0.1190 11.73 12.91
Winglet Nr. 2 1.029 0.08262 12.45 1.364 0.1170 11.665 12.43
Winglet Nr. 3 1.015 0.0834 12.17 1.353 0.1190 11.369 10.15
Winglet Nr. 4 1.064 0.0851 12.50 1.413 0.1185 11.928 14.36

Fig. 30. Baseline–final aircraft comparison. Fig. 32. Baseline–final aircraft comparison.

Fig. 31. Baseline–final aircraft comparison.


Fig. 33. X velocity contours in the symmetry plane – baseline configuration.

inspection, it was found that the pressure gradients caused by the


an angle of 6.5 deg. This was expected due to the new planform
flow separation behind the propeller have been diminished with
and winglets both reducing the wing’s pitching moment. The new
the use of the specially designed fairing. The effect of the propeller
planform brings the wing’s aerodynamic center (AC) and plane’s
on the pressure distribution of the wing is quite obvious for both
neutral point (NP) closer to the CoG, reducing the aircraft’s static
configurations. The leading edge near the wing root is affected by
margin.
the fan, and as a result its stagnation pressure is increased. In the
suction side respectively, the pressure drop shows the same behav- Simulations including the propeller clearly show that it en-
ior, as pressure losses are more intense. hances the tail’s performance, producing pitching up moment. This
The most significant effect of the propeller flow however, is the is evident in Figs. 33–34. The flow behind the propeller is of in-
alteration of the aircraft’s stability characteristics which is evident creased velocity and reaches the horizontal stabilizer at higher AoA
in Fig. 31. In the initial configuration, analytical linear stability cal- significantly increasing the produced lift. As a result the angle of
culations (not including fan) and the “No Fan” simulations, show incidence for which the aircraft is trimmed is increased as well,
an agreement, with the aircraft longitudinally trimmed in the de- and this finding was in agreement with test flight results. These
signed angle of incidence of the wing (5.5 deg). The final optimized flight tests were performed both during the prototype construction
aircraft is statically stable as well, but trim position changed to and in preparation for the Air Cargo Challenge 2011 competition.
S.G. Kontogiannis, J.A. Ekaterinaris / Aerospace Science and Technology 29 (2013) 339–350 349

Fig. 34. X velocity contours in the symmetry plane – final configuration. Fig. 35. Baseline configuration Cp distribution.

Table 5
Baseline–final configuration comparison in full throttle conditions.

Fan C L@6 deg C D@6 deg C L /C D@6 deg Increase (%)


Full configuration base 1.197 0.1585 7.55 -
Full configuration final 1.192 0.1458 8.18 7.70

The flight tests used a single flight profile consisted of takeoff,


a circular turn of a radius of about 150 m, gliding descent and
landing. The aim of the flight tests was to certify the ability of
the UAV to satisfy the initial design requirements, to ensure the
validity of the analytical predictions for takeoff and landing dis-
tances, to verify the efficiency of the designed control systems,
and to validate the analytical and computational results regarding
longitudinal stability. During these tests in order to simulate the
photographic equipment load, a lead plate of 2.5 kg was inserted
into the camera bay. For these flight tests the baseline configura-
tion was used. These flight tests proved that: (i) the predictions
regarding takeoff and landing runway distance had less than 10%
divergence from the actual test flights, (ii) the control surfaces Fig. 36. Final configuration Cp distribution.
provided an aircraft response exactly as calculated in flight me-
chanics analysis, (iii) the flight test results of the baseline aircraft 6. Conclusions
were also in agreement with the analytical and computational re-
sults regarding longitudinal stability and trim position shown in To satisfy the increasing needs of aviation in light UAV air-
Fig. 31. During the gliding descent phase (throttle at idle) no pitch- craft and the needs for aerial photography, city maps, traffic in-
ing up was observed. The aircraft was trimmed in the designed formation with FPV systems, a light UAV aircraft was designed,
AoA, exactly as predicted using linear stability approach and the constructed and successfully tested in flight. Linear aerodynamic
CFD simulations. Longitudinal stability for the full throttle ascent analysis as well as linear stability analysis was performed, con-
phase could not be calculated using the linear stability approach cluding the aircraft design. The propulsion system was selected
due to the propeller wake. Therefore, only CFD calculations could after an extensive survey of electric motors, batteries, etc. The
provide longitudinal stability characteristics for the ascent phase. propeller performance was initially evaluated using program JAVA
The flight tests proved that during ascent, the horizontal stabilizer prop., followed by experimental tests. The ability of the CFD tools
produced an increased pitching up moment due to the interference to accurately compute transitional flows was evaluated via com-
parison with available measurements. Then preliminary design was
of the propeller wake. This observation was in agreement with the
followed by an extensive CFD analysis that served as the founda-
CFD computations, shown in Fig. 33. As a result, during the full
tion for the optimization process that was conducted next. CFD
throttle ascent phase the aircraft was not trimmed in the desirable
analysis showed good agreement with the analytical results, and
AoA, exactly as shown in Fig. 31. aerodynamic optimization was the next step. Span efficiency and
Static margin is also reduced, but it is obvious that the changes induced drag were optimized by planform redesign, twist and
in the configuration improved the aerodynamic efficiency of the winglet design. The final wing design was the result of several
aircraft significantly. A summary of the aerodynamic performance CFD simulations that clearly demonstrated that the aerodynamic
comparing the baseline and the final configuration for full throttle characteristics of the wing could be improved after certain modi-
conditions is shown in Table 5. fications. To reduce form drag, specifically designed fairings were
350 S.G. Kontogiannis, J.A. Ekaterinaris / Aerospace Science and Technology 29 (2013) 339–350

successfully tested in the final optimized configuration. This paper [7] R. Langtry, F.R. Menter, Transition modeling for general CFD applications in
focused only on aerodynamic efficiency optimization constrained aeronautics, AIAA-2005-522, Reno, NV, January 2005.
[8] M.D. Maughmer, Design of winglets for high performance sailplanes, AIAA pa-
by wingspan. Once structural integrity obstacles are overcome, the
per 2001-2406, Journal of Aircraft 40 (6) (2003) 1099–1106.
aerodynamic performance can be further improved. The combina- [9] J. McArthur, Aerodynamics of wings at low Reynolds number, PhD dissertation,
tion of improvements proposed and the increase of wingspan and University of Southern California, Aerospace and Mechanical Engineering, 2007.
AR, would lead to a greater range, an endurance increase and to [10] B.W. McCormick, Aerodynamics Aeronautics and Flight Mechanics, 2nd edition,
a more economical and eco-friendly aircraft. It is expected that John Wiley & Sons, Inc., New York, 1995.
[11] P. Mosak, Winglet design for sailplanes, Free flight 2/92.
further aerodynamic improvements using CFD analysis, combined [12] W.F. Philips, Lifting line analysis for twisted wings and washout-optimized
with structural optimization could allow a similar design to fly us- wings, Journal of Aircraft 41 (1) (2004) 128–136.
ing solar power. [13] W.E. Pinebrook, C. Dalton, Drag minimization on a body of revolution through
evolution, in: Computer Methods in Applied Mechanics and Engineering, 1983.
[14] D.P. Raymer, Aircraft design: A conceptual approach, AIAA Education Series,
References 1992.
[15] J. Roskam, L. Edward, Airplane Aerodynamics and Performance, Design Analysis
[1] I.H. Abbot, A.E. von Doenhoff, Theory of Wing Sections, 2nd edition, McGraw– and Research, DAR Corporation, 1997.
Hill Book Company, New York, 1949; also: Dover Publications, Inc., New York, [16] C.L. Rumsey, P.R. Spalart, Turbulence model behavior in low Reynolds number
1959. regions of aerodynamic flow fields, AIAA-2008-4403, pp. 1–14.
[2] J.D. Anderson Jr., Introduction to Flight, 3rd edition, Aerospace Science Series, [17] M.S. Selig, B.D. McGranahan, Wind tunnel aerodynamic tests of six airfoils for
International Edition, McGraw–Hill Book Company, 1989. use on small wind turbines, University of Illinois at Urbana-Champaign, Ur-
[3] BETA CAE Systems S.A., ANSA user’s guide, May 2011. bana, IL.
[4] M. Drela, XFOIL, an Analysis and Design System for Low Reynolds Airfoils, MIT [18] G.R. Spedding, J. McArthur, Span efficiencies of wings at low Reynolds num-
Department of Aeronautics and Astronautics, Cambridge, MA, 2013. bers, Journal of Aircraft 47 (1) (2010) 120–128.
[5] Fluent Inc., FLUEN T 6.3 user’s guide, Sept. 2006. [19] SU 2 v.1.0 user’s guide documentation, Department of Aeronautics and Astro-
[6] S. Kontogiannis, J.A. Ekaterinaris, Design performance evaluation and optimiza- nautics Stanford University.
tion of a UAV, AIAA paper 2013-0375, 51st ASM, January 2013. [20] D.C. Wilcox, Turbulence Modeling for CFD, DCW Industries, 1994.

You might also like