Professional Documents
Culture Documents
2011 - Design and Wind Tunnel Tests of A Tiltwing UAV
2011 - Design and Wind Tunnel Tests of A Tiltwing UAV
DOI 10.1007/s13272-011-0026-4
ORIGINAL PAPER
Received: 3 April 2011 / Revised: 3 August 2011 / Accepted: 3 August 2011 / Published online: 24 August 2011
Ó Deutsches Zentrum für Luft- und Raumfahrt e.V. 2011
123
70 J. Holsten et al.
123
Design and wind tunnel tests 71
Table 1 Basic flight platform parameters motor and rotor efficiency factors gMotor = 0.8, gProp =
Parameter Values
0.7. Using the power required for hovering
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Configuration VTOL W3
Pel;0 ¼ gProp gMotor ; ð1Þ
Type High wing 2 q AR
Height above ground 0–300 m
the hovering time as well as the energy density of lithium-
Maximum velocity 40 m/s
polymer accumulators the weight of needed accumulators
Design speed 15 m/s
could be estimated. Along with the maximum payload
Wing span 2m
weight and estimates for the structure and motor weight
Maximum takeoff weight 10 kg
feasible combinations of maximum takeoff weight, hov-
Payload 1.5 kg
ering time and propeller diameter were obtained. The
Endurance 60 min
maximum propeller diameter is limited by the maximum
Maximum range 45 km
lateral dimension of 2 m. To achieve the maximum
endurance given in Table 1, a mission with 80% horizontal
flight at design speed is considered also ensuring the
were considered and compared. The tailsitter has disad- maximum range required. The power required for hori-
vantages in transition control, emergency procedure and zontal flight was estimated using the drag coefficients
payload volume/handling, since a similar orientation of the described in the following sections.
payload camera in all flight states was favored. Due to its A two-bladed propeller was specifically designed from
advantages with respect to the energy efficiency in forward Schuebeler Composites to accommodate the different
flight, a tiltwing configuration was selected. A more requirements on the propulsion system for hovering and
detailed insight into the design decision including further forward flight. Main focus is on optimizing it for the design
requirements is given by Ostermann et al. [10]. Basic speed in horizontal flight while still ensuring enough power
parameters of the AVIGLE tiltwing UAV are summarized to rise vertically. For this reason, the propellers are
in Table 1. equipped with variable blade pitch, which also allows au-
Within the specification of the tiltwing, a design air- toration in case of engine failure.
speed of 15 m/s in horizontal flight configuration was A CFD analysis confirmed the performance of the pro-
selected as a tradeoff between mission requirements from peller design in the different flight states. To reduce aero-
the payloads and endurance, range. Additionally, it had to dynamic forces and moments during hovering induced by
be ensured that all flight states including hovering, VTOL the slipstream of the propellers, the angle between pro-
and the transition between hovering and horizontal flight pulsion system and wing was set to an angle ensuring zero
remain possible. lift a0 = -4.5°. The propellers are counter rotating to
level the motor moments.
123
72 J. Holsten et al.
XFLR5 [11] providing lifting line theory and vortex lattice account for the corresponding Reynolds’ numbers. All
methods, three airfoil types were selected. Subsequently, airfoils were investigated concerning their stall character-
the wing area was modified and a stability analysis per- istic and maximum lift behavior at these Reynolds num-
formed before the final airfoil was selected. bers. The NACA 6412, SG 6050, Clark Y and GOE 398
Simultaneously, preliminary wind tunnel tests to eval- proved most suitable and were further analyzed for dif-
uate the effect of the large propeller, covering 78% of the ferent wing dimensions and overall configurations. After
wing, on the aerodynamic coefficients were carried out. this analysis, the GOE 398 proved most suitable for the
Different parameters such as distance of the propeller to the desired application.
wing, leading edge sweep and airfoil were varied with Due to the design requirements of a purely aerodynamic
thrust settings. The wingspan was limited by the wind flight at V = 15 m/s with maximum weight of 10 kg, an
tunnel to 1 m and the experiments were scaled accordingly. optimization regarding wing dimensions and aspect ratio
In Fig. 1, lift coefficients for three different thrust levels was performed. Since the wingspan is limited to 2 m, an
and one without thrust are depicted. In the measurements, decrease of the outer chord also reduces the wing area.
the propeller pitch was kept constant and the thrust was Therefore, the reduction of the induced drag, by increasing
controlled over the rotation speed. As a result, the total lift the aspect ratio through decreasing the outer chord, was
coefficient increases with the propeller thrust. The slope of limited by the required lift. Calculations showed that an
the lift coefficient though is slightly reduced, whereas the outer chord of co = 0.3 m still ensures a purely aerody-
stall conditions are pushed towards higher angles of attack. namic flight at V = 15 m/s with a wing area of
These effects can be explained by the acceleration of the AW = 0.707 m2, and satisfactory values for the aspect ratio
freestream by the propeller induced slipstream and the AR = 5.7 and wing loading W/AW = 138.755 N/m2.
deflection of the incident flow to lower angles of attack by Without leading edge sweep, this leads to a trailing edge
the slipstream. The effects were most pronounced without sweep of -6.1°. An overview of the final specification is
leading edge sweep and the shortest distance between given in Fig. 2.
propeller and leading edge.
For the airfoil design, different airfoils were analyzed 3.3 Tailplane design and stability analysis
for a 2 9 0.4 m rectangular wing in XFLR5. Different flap
angles were investigated to ensure high performance in The tailplane and fin design as well as the dimensioning of
trimmed conditions. Design parameters were the lift to drag the lever arms and the center of gravity was done regarding
ratio CL/CD, zero lift angle, camber, thickness and the design aspects for horizontal flight as well as for transition
pitching moment at zero lift angle. The pitching moment at and vertical flight. Due to the tiltwing approach, a high
zero lift angle for example is relevant for trimming during wing configuration was chosen. To reduce propeller slip-
hovering and vertical flight. All relevant combinations stream interferences, this led to a T-shaped empennage.
were calculated for a freestream velocity of V? = 15 m/s For horizontal flight, a first dimensioning using empir-
(design speed) and V? = 40 m/s (maximum speed) to ical data for the tailplane volume VH defined as
AH rH
VH ¼ 0:75 ð2Þ
AW ll
1.2
T = 210 N
T = 160 N and for the fin volume VV defined analog as
1 T = 100 N
w/o thrust AV r V
VV ¼ 0:095 ð3Þ
0.8 AW b=2
0.6
was done. Important parameters influencing these factors
are the horizontal and vertical lever arm rH, rV, mean
C
123
Design and wind tunnel tests 73
123
74 J. Holsten et al.
T
LT
ι
Lw
Vw
Dw
zR M0
Dz D σ xf
DX M
V∞
w
u
W xR
zf
60
steering during hovering, the chosen wing and airfoil were
40
built in an 1:2 ratio and mounted on an six-component
20
balance. Different thrust levels representing propeller
0
15
induced slipstreams of V = [5, 10, 15, 20, 23] m/s were
10
80
100 set and the deflection angle of the slipstream flaps was
5
40
60 varied. The final control device sizes are depicted in Fig. 2.
0
u [m/s] 20
−5 0
σ [°]
3.4 Fuselage design
Fig. 4 Needed thrust T for trimmed flight states during transition
over forward velocity u and incidence angle r The fuselage was designed with a focus on aerodynamic
aspects and an easy integration of the different payloads.
Two different payload types, an RGB camera and a mobile
velocity u, the incidence angle r and the thrust T. Since network basis station module based on the new long-term
discrete values of u, r and T were used for calculation an evolution (LTE) standard, had to be regarded. The size of
interval of X; Z; M e ^ X; Z; M was accepted as the payload compartment was specified according to the
trimmed condition. Figure 4 illustrates the corridor with a size of the LTE module. The camera can be mounted on
margin chosen for visualization of e ¼ 10 N; 10 Nm: three different directions depending on the mission.
Further parameters influencing the corridor depicted in Regarding the payload it had to be ensured that in no
Fig. 4 are the geometric distances depicted in Fig. 3. flight state the wing obscures the image of the camera.
Especially, the position of the tilt axis xR and zR was varied Furthermore, an easy integration and change of payloads
to enable trimming the pitching moment at low airspeeds should be ensured. This led to an rectangular-shaped
with the slipstream flaps. Out of these calculations and fuselage in contrast to an oval one. For autonomous
constructive considerations, the tilt axis was set at behavior and collision avoidance alternatively a radar
approximately 50% of the wing root chord. The calcula- sensor or a 3D-time of flight (ToF) camera will be on
tions also showed the adaptability of the previously board. Beside standard sensor equipment, such as e.g. GPS,
described aerodynamic components and gave a first esti- IMU, an additional embedded PC for communication and
mate for the slipstream flap size specified in Fig. 2. mission planning is integrated. All components were
123
Design and wind tunnel tests 75
(a)
Fig. 6 Wind tunnel model
AVIGLE Flightplatform (Hardware)
Antenna 1
Magneto-
GPS
Temp.- Air Data
RC Modul
Landing The wind tunnel used is a closed circuit tunnel with an
LTE UE WiFi meter Sensor boom Altimeter
Antenna 2 open measurement area of 1.5 m diameter. To avoid
communication sensors
effects, such as e.g. turbulence interactions, the maximum
size measurable is limited to 1 m. This size was determined
USB
I2C
USART
2 x I2C
2 x A/D
A/D
PPM
USART
direct
? Radar CAN
in earlier measurements of the open test section of the used
Flight Control Board
Embedded PC Ethernet
ToF
Camera
RS485
(NC+FC)
IMU wind tunnel.
RS232 RS232 Due to the size of the wind tunnel, a 1:2 scaled model of
Ethernet
2x PWM
6x PWM
2x PWM
A/D
A/D
A/D
A/D
Battery Battery
Power
Supply
Servo
Servos
Rudders
Blade pitch
control
and transition were measured separately using a balance
Patch-Antenna
payload engine/steering with strain gauges to measure forces and moments in all
three dimensions. Transition and horizontal flight condi-
(b)
tions were measured with and without thrust. For a thor-
Fig. 5 Components to be regarded within fuselage design. a Placing ough investigation, different control factors of the sub scale
within fuselage. b Architecture of hardware components model such as thrust, control devices, incidence angle of
the wing, engine pod and elevator were designed as
positioned according to their requirements and in consid- adjustable elements.
eration of the overall center of gravity. An overview of all Beside the scaling of the model, the propulsion system
components and their integration into the fuselage is given including propellers and motor had to be scaled as well.
in Fig. 5a and b. For correct measurement of the slipstream flap coefficients
and derivates, the propeller size was scaled with the model
in an 1:2 ratio. This also ensures the same overlap between
4 Design verification in wind tunnel tests wing and propellers in the scaled model as in the original
model. Scaling the propeller size geometrically and dou-
For verification of the design, extensive wind tunnel tests bling the slipstream velocity to ensure Reynolds’ similarity
of the tiltwing in all relevant flight states were conducted leads to the same amount of thrust for the scaled and ori-
and compared with computed data in XFLR5 and Digital ginal model when applying momentum theory [16]. This
DATCOM [15]. In XFLR5, vortex lattice method was allows to use the same propulsion system for horizontal,
used. XFLR5 provides two versions of the vortex lattice vertical and transition measurements. Especially in transi-
method. In the first method, each horseshoe vortex over the tion, the thrust and scaling of the slipstream velocity are
chord extends to infinity. In the second, method only the important to accurately reproduce the overlap between
trailing vortices extend to infinity. Since the two methods forces due to propulsion and due to aerodynamic effects.
produce very similar results, the first method is used for Figure 6 and Table 2 illustrate the measurement setup and
validation. The obtained aerodynamic coefficients are used varied parameters.
in a six degrees of freedom simulation model for design of For a first validation, the aerodynamic coefficients as
the flight control system. given in Eqs. 4–8 were measured without a propulsion
123
76 J. Holsten et al.
Table 2 Control device deflections and parameters set during the 1.8
XFLR5
wind tunnel tests Wind tunnel
1.6
DATCOM
Horizontal Transition Vertical
1.4
V [m/s] 30 [10; 30] 0 1.2
r [°] 0 [0; 90] [85; 100]
1
a [°] [-5; 20] [-2; 12] 0
CL
0.8
b [°] [-20; 20] 0 0
j [°] [-6; 6] [-20; 20] [-20; 20] 0.6
g [°] [-20; 20] [-20; 20] 0 0.4
n [°] [-20; 20] [-20; 20] [-20; 20]
0.2
f [°] [-20; 20] 0 0
0
−0.2
−5 0 5 10 15 20
system. The values thus obtained were then compared to α
the computed data mentioned above, XFLR5 and Digital (a)
DATCOM. Figures 7, 8, and 9 illustrate the lift and drag
0.4
coefficient as well as the influence of the control devices XFLR5
and the roll coefficient. Wind tunnel
0.35 DATCOM
Due to the software constraints, the computed results
comprise the fuselage only as a cylinder. The engine pod, 0.3
spinner and landing gear seen in Fig. 6 are not considered
0.25
in the computation. The landing gear is expected to
CD
influence the overall drag but not the lift. The spinner and 0.2
engine pod disrupt the wing and the flow over the wing
0.15
area behind the engine pod. This is assumed to reduce the
lift and increase the drag compared to the computed 0.1
results. These expected results can be seen in Fig. 7a
and b. The measured lift in Fig. 7a is smaller than com- 0.05
puted. The drag in Fig. 7b is significantly increased owing 0
to the landing gear, engine pod, and not cylindrical −5 0 5 10 15 20
α
shaped fuselage.
The effect of an elevator deflection on the pitching (b)
moment coefficient and the flaps on the lift coefficient is Fig. 7 Lift (a) and drag (b) coefficient of the wind tunnel measure-
shown in Fig. 8. The effect of the flaps matches well with ments compared with computed data
the with DATCOM computed data. The effect of the ele-
vator pitch measured is smaller than computed, although
the slope for small angles matches well. The trend of the and all measurements were repeated once to ensure
computed data does not correspond well to the fact that the reproducibility. In addition to these measurements, a ref-
tailplane airfoil is not symmetrical and, therefore, positive erence measurement without propellers was conducted.
and negative deflection should differ in their effect, which Figure 10a depicts the difference in lift coefficient for all
can be seen in the measurements in Fig. 8a. The achieved four cases at zero angle of attack and its change due to
pitching moment still satisfies the dimensioning criteria. slipstream flap deflection. The different thrust levels were
The lateral forces and moments correspond better with set to compensate for the drag and to generate a small
the computed data. Figure 9a and b illustrates the roll side acceleration and deceleration in xf-direction.
slip angle derivative and the change in roll moment due to As can be seen, the different thrust settings lead to
aileron deflection. The differences between the computed differences in lift, which confirm the results described in
and the measured data are marginal and can be explained Sect. 3.2 and can be explained by the acceleration of the
by real effects in the measurements. free stream through the propeller. It was expected that this
Beside the comparison with computed data, the mea- effect also influences the flap efficiency, thus changing the
surements were conducted with different levels of thrust to slope of the curves. This effect cannot be seen in the
account for the large overlap of propeller area and wing. depicted measurements. A flap deflection leads to almost
For horizontal flight, three different thrust levels were set the same amount of change in lift.
123
Design and wind tunnel tests 77
0.5 0
XFLR5 DATCOM
0.4 Wind tunnel Wind tunnel
DATCOM −0.05
0.3
0.2 −0.1
0.1 −0.15
Δ Cm
C lβ
0
−0.2
−0.1
−0.2 −0.25
−0.3
−0.3
−0.4
−0.5 −0.35
−20 −15 −10 −5 0 5 10 15 20 −5 0 5 10 15 20
η α
(a) (a)
0.15
0.25 XFLR5
XFLR5 Wind tunnel
0.2 Wind tunnel 0.1 DATCOM
DATCOM
0.15
0.05
0.1
0.05 0
Δ Cl
Δ CL
0
−0.05
−0.05
−0.1 −0.1
−0.15
−0.15
−0.2
−0.2
−0.25 −20 −15 −10 −5 0 5 10 15 20
−20 −15 −10 −5 0 5 10 15 20
κ ξ
(b) (b)
Fig. 8 Change in pitch and lift coefficient of the wind tunnel Fig. 9 a Illustrates the roll side slip angle derivative. b The change in
measurements compared with computed data due to elevator deflec- roll moment due to aileron deflection
tion g (a) and symmetrical aileron deflection j (b)
To validate the steering efficiency and capability in controllers operated in full throttle which led to a high heat
vertical flight, different thrust settings and control device development resulting in a variation between the thrust
deflections were set without free stream. Different inci- setting measured beforehand and the thrust achieved during
dence angles of r = [85;90;95]° were set. Due to the the actual measurements. This imprecise setting of the
incidence angle of the propulsion system of i ¼ 4:5 an thrust is one explanation for the asymmetric slope in
incidence angle of r = 85° with the according thrust cor- Fig. 11b. Further measurements using a different motor
responds to hovering. Figure 11 illustrates the effect of the and thrust controller combination are necessary to validate
different control devices. For longitudinal control a sym- this assumption. It is expected that the actual slope is
metrical aileron deflection j, as illustrated in Fig. 10b, can entirely symmetrical.
be used to accelerate in xf-direction and to trim the pitching Relevant combinations of incidence angle r and free
moment. Lateral control is obtained by the ailerons stream velocity were selected from the corridor for stable
(Fig. 11a) and a difference in thrust between the two transition, introduced in Sect. 3.3 The combinations are
engines (Fig. 11b). The difference in thrust was set by depicted in Fig. 12 on the x- and y-axis. For each combi-
adjusting the thrust controller signals while measuring nation, different levels of thrust were controlled to repro-
without flap deflection. Afterwards, all signals and duce the trimmed conditions where X & 0 and Z &
flap deflections were set and the acting forces measured. -100N. As the calculated results visualized in Fig. 4 show,
During the vertical measurements, the motors and thrust the thrust was increased with increasing incidence angle,
123
78 J. Holsten et al.
0.45 3
w/o propulsion σ = 85°
X<0N
0.4 X≈0N 2
X>0N
0.35 1
N [Nm]
0.3 0
CL
0.25 −1
0.2 −2
0.15
−3
0.1
−6 −4 −2 0 2 4 6 −4
κ −20 −15 −10 −5 0 5 10 15 20
(a) ξ [°]
(a)
10
σ = 85° 4
8 σ = 85°
6 3
4
2
2
X [N]
0 1
L [Nm]
−2
0
−4
−6 −1
−8
−2
−10
−20 −15 −10 −5 0 5 10 15 20
κ [°] −3
−20 −15 −10 −5 0 5 10 15 20
(b) Δ T [%]
Fig. 10 Change in lift coefficient for different thrust settings (a) in
(b)
horizontal flight and additional longitudinal force X during hovering
Fig. 11 Additional yaw moment (a) and roll moment (b) due to the
due to symmetrical slipstream flap deflection (b)
corresponding steering control in vertical flight
while the free stream velocity was reduced. In contrast to a corridor of trim conditions during transition as calculated
the horizontal measurements, actual forces and not the in Sect. 3.3
coefficients were considered in transition measurements
since the influence of the propulsion on the aerodynamic
coefficients cannot be neglected. Especially lift, drag and 5 Conclusion and future work
control device efficiency are influenced by the incidence
angle and the thrust level. Figure 12 shows the results for In this paper, the development and wind tunnel tests of a
horizontal and vertical forces for the different measurement tiltwing platform for multiple purposes has been described.
points. Even though the thrust was controlled to set Z & The complete process including wing, tailplane, control
-100 N and X & 0 variations from these values can be device and geometry design as well as fuselage design was
seen in Fig. 12. illustrated. The design point of the tiltwing configuration
The variations result from different angles of attack set was set to V? = 15 m/s in horizontal flight. The main
and also illustrate the effectiveness of the different control challenge in design—supported by extensive numerical
device deflections, especially the slipstream flap deflection. flow analysis and wind tunnel tests—was the consideration
The control device deflections set are given in Table 2. of transition states. A calculation determining trim points
Despite some variation, the results confirm the existence of in the longitudinal motion during transition between
123
Design and wind tunnel tests 79
0
0 References
−10
1. Rohde, S., Goddemeier, N., Wietfeld, C., Steinicke, F., Hinrichs,
−50 −20 K., Ostermann, T., Holsten, J., Moormann, D.: AVIGLE: a sys-
30 tem of systems Avionic Digital Service Platform based on micro
100 −30 unmanned aerial vehicles. In: IEEE International Conference
20
10 50 −40
Systems, Man, and Cybernetics (SMC), Instanbul, Turkey (2010)
2. Pegg, R.J.: Summary of Flight-Test Results of the VZ-2 Tilt-
V [m/s] 0 0 σ [°] Wing Aircraft. Technical Report on NASA TN D-989, NASA
(1962)
(a) 3. Ransone, R.: XC-142A V/STOL Transport Tri-Service Limited
Category 1 Evaluation. Technical report on air force flight test
center EDWARDS AFB, CA (1966)
−20 4. Fay, C.B.: A cursory analysis of the VTOL tilt-wing performance
and control problems. Ann. N. Y. Acad. Sci. 107, 102–146 (1963)
0
−40 5. McKinney, M.O., Kirby, R.H., Newsom, W.A.: Aerodynamic
factors to be considered in the design of tilt-wing V/STOL air-
−50
Z [N]
123