Sie sind auf Seite 1von 4

L3 Applications Exercise

Group 9

Academic responsible: Dr. Mirko Kovac


Department: Aeronautics
Academic year: 2016/2017

Student Name: CID:


Mihai Cristian Botea 00851598
James Calford 00959568
Pierre-Olivier Donnat 00950508
Andrew Gooding 00930324
Maria Atienza Menendez 00965790
Ben Ng 00842739
Arthur Pasquier 00951894
Conor Wilson 00869286

Department of Aeronautics
South Kensington Campus
Imperial College London
London SW7 2AZ
U.K.
I Introduction & Design Process range. The optimization process started with
choosing an appropriate motor and gear ratio.
The design task consisted of producing a self- The 5:1 High Power Micro motor was found to
driving unmanned aerial vehicle, under 70g within maximise the ramp exit velocity with an optimal
a sphere of 50cm, aiming to maximize the lon- gear ratio of 2:3. Front and rear wheels were cho-
gitudinal distance when launched off a 1.24m sen to have diameter of 32mm and 40mm respec-
high, 1.8m long ramp [1]. To achieve optimal tively. This set of parameters predicted a ramp
performance, the design was driven by attempt- exit velocity for our initial model of 4.4 m/s.
ing to maximize the end of ramp velocity and
glide trajectory, while maintaining stability, creat- Aerodynamics & Glide Phase
ing a structure strong enough to withstand multi- To accurately model the glide trajectory of the
ple launches, and simultaneously minimizing the robot, a standard form of glider equations of mo-
weight. First priority was placed on researching tions was taken and simplified to only model the
similar applications and modelling all aspects of longitudinal components as detailed in [6], [7].
the glider [2], [3], [4], [5]. With this information Similar to the ramp equations of motion, these
initial ideas were drafted, discussed within the were solved using a 4th order Runge-Kutta time-
group, and tested for their validity, leading to our marching scheme in Matlab, taking the end of the
first concept. This concept was made manufac- ramp phase as the initial conditions for the glide.
turable, produced, and assessed. After modifica- With the equations of motion found, the next
tions addressing the issues of the first design, a step was to accurately model the lift, drag, and
final design was created and tested, with the re- pitching moment of the glider. Knowing the robot
sults presented here. would be flying at very low speeds (3 m/s) with
Reynolds numbers typically less than 50,000,
II Performance Modelling the effects of compressibility were considered
Accurately predicting the elements of robot per- negligible. The wing consisted of a thin airfoil op-
formance was fundamentally tied to the quality of erating at low angles of attack with no flow sepa-
the models used. The modelling was split into ration further allowing for the assumption that the
three primary categories: the drivetrain and ramp viscous and turbulent effects would be small. In
phase, the aerodynamics and glide phase, and turn, the flow was idealized as potential flow and
structural simulations for landing loads. use of a vortex lattice method was justified.
To this end, Tornado was incorporated into the
Drivetrain & Ramp phase model to solve for the overall lift, drag, and pitch-
For the ramp phase, motion was assumed to be ing moment at each time step [8]. This involved
strictly parallel to the direction of the ramp. As- writing Matlab functions to automate the con-
suming a linear relationship between the torque struction of a wing based on defined parameters,
and the rotational speed of the motor and negligi- generate a lattice and solve for forces based on
ble mechanical losses, the equation of motion in the current glide state and output the final values
the horizontal direction and the torque equation to the equations of motion. Ground effect was
from motor parameters is as follows [1]: modelled by testing the given wing, and its re-
flection across the ground plane, at a set angle
d2 x T Tmax of attack and velocity, across a variety of heights
m = D ; T = Tmax (1) to generate a function to scale the lift and drag
dt2 r max values as the glider approached the ground.
In order to prevent slippage, the traction force Structural Simulations
should be less that of the static friction experi- Using the worst case impact velocity and an-
enced by the traction wheels. Therefore, the fol- gle from the glide phase, the structure was run
lowing relation must hold true: through Creo Simulate to test its response and
T determine the necessary thickness and number
N (2)
r of cross struts required to withstand landing.
A friction coefficient value was found using a III Mechanical Implementation
force-meter to be 1.33 between the rubber tyres
and the 400 grit sandpaper surface of the ramp. Wing
This value was used to determine the maximum The model allowed the group to iterate through
torque prior to slippage. designs (assuming a fixed structural center of
Equation 1 was solved in Matlab using a 4th gravity, and altering the positions, size, and shape
order Runge-Kutta time-marching scheme and of the wing and tail) testing their stability and fea-
was used in conjunction with Equation 2 to anal- sibility. It was determined that the benefits of a
yse the performance of various configurations to cambered wing (testing with a NACA 4408 air-
determine the factors that maximize the robots foil) were outweighed by the increase in mass

i
and structural complications. Therefore, taking of the wing due to the tail structure. Although
the optimum design from the model, the wing a sub-optimal design decision according to the
was constructed as a 3mm foam flat plate with model, to allow for greater wing movement the tail
a 3030cm inverse zimmerman planform shape. and the corresponding structure was removed.
A small tail was added for improved stability. The This decreased the weight allowing for the use
wing position was made variable through the use of a larger wing (4040cm) to generate greater
of carbon rods connecting to the undercarriage, amount of lift, whilst simultaneously shifting the
while the tail was fixed in place by the structure. models centre of gravity forward helping alleviate
Undercarriage Structure the pitching issues. Further amendments made
were the inclusion of re-designed cross stiffen-
The undercarriage was designed to be minimal- ers, as illustrated in Figure 2, to improve the struc-
istic and lightweight with the top edge set at a tures torsional rigidity and impact resistance, so
5o angle of attack for the main wing. The con- as to retain straight motion in the ramp phase af-
struction was made from 2D laser cut parts of ac- ter the first run.
etal, with the CAD parts sized to take into account Final Design
the 0.2mm tolerance of the laser cutter. To min-
imize the use of epoxy and save weight, push fit The final model did not perform as expected,
designs and interference fits were used wherever travelling a distance of just 1.6m compared to a
possible. A platform was designed to attach the 1.9m predicted for the final configuration. Testing
motor, circuit and battery. within the flight arena brought to light new issues;
most significantly, in all but one test the model
Drivetrain, Wheels, and Circuitry demonstrated a severe lateral bias towards the
The front and rear wheels were purchased rather port side. This was likely a skewed centre of grav-
than manufactured to ensure perfect in plane ro- ity caused by parts of the circuitry extruding from
tation. A 3mm hole was drilled through their cen- the main chassis creating a weight imbalance,
tral axis to provide an interference fit for the car- compounded by the greater moment arm of the
bon rod axles. Bearings were also utilized to min- larger wing. The model also displayed wheelie
imize the friction with the rear axle. An Adafruit behaviour as the motor was switched on; the wing
Trinket circuit board was connected to the motor, was already designed to be at a 5o angle of at-
a 3.7V battery, a Mosfet transistor, and an infrared tack so this pitching behaviour left the model in an
demodulator, with all the components soldered as off-design regime. The actual trajectory was also
neatly as possible [1]. The circuit would receive a poor because the robot started part way down the
PWN signal from a remote control to engage the ramp to mitigate the left bias resulting in an exit
motor, gradually increasing the voltage supplied velocity about 1.3 m/s rather than 3.7 m/s in the
so as to avoid any wheel slip. predicted model.
Next Steps
IV Performance Review
The lateral stability issues could be rectified by re-
Prototype Testing and Modifications fining the circuitry to fit within the chassis or con-
A series of preliminary tests highlighted two pri- structing a cage specifically for housing the cir-
mary issues in the initial prototype. Firstly, the cuit. A vertical stabilizer could also be added, with
gears responsible for transferring torque from the all these solutions coming at a weight penalty.
motor to the rear axle had a tendency to become The wheelie problem could be solved through
un-meshed which resulted in zero power transfer a more gradual increase in applied torque, or by
from the motor and subsequently a severe loss in switching to front-wheel-drive with the added ben-
velocity by the end of the ramp. Going forward efit of moving the center of gravity forward. Fur-
it was decided that a wider set of gears would thermore, the model needs to be revised. As
be used as, although heavier, they would provide shown in Figure 3, there are significant curvature
a greater contact area between gears and sub- differences in the second half of the actual and
stantially reduce the risk of becoming un-meshed. model trajectories for the same exit ramp veloci-
Furthermore, it was decided that a motor bracket ties. This is due to the fact that the model predicts
should be used to ensure accurate integration of significant stability problems for tailless configu-
the motor into the structure providing further im- rations, as illustrated by Figure 4, likely to do with
provements to the motor/axle connection. how the pitching moment, and possibly ground ef-
Secondly, it was observed that in cases where fect, were modelled. Moving forward, more em-
the gears remained in contact, the model exhib- pirical data should be collected with the model
ited a pitching up behaviour on departure from corrected with this information. Greater emphasis
the ramp. This resulted in an early onset of stall should be placed on the manufacturing and me-
significantly reducing the distance travelled. This chanical design of components. The listed mod-
was due to the inflexibility in moving the center ifications should be implemented and a tail and
of gravity of the robot, and restricted movement smaller wing should also be re-evaluated.

ii
V Figures and Tables
Final Distance 1.6m
Ramp Exit Velocity 1.3m/s
Final Mass 60.88g
Configuration 3 wheels, RWD
Engine 5:1 High Power Micro
Gear Ratio 2:3
Front Wheel 32mm
Rear Wheels 40mm
Wing Planform Inverse Zimmerman
Figure 1: Glider equations of motion free body diagram [6]. Wing Max Dimensions 4040cm
Wing Aspect Ratio 1.27
Wing Area 1256.64cm2
Wing Angle of Attack 5o

Table 1: The values characterizing the final robot.

References
[1] Kovac M. L3 Applications 2016/2017. 2016.
Figure 2: Comparing the initial prototype and final design [2] Laitone E.V. Wind tunnel tests of wings at
undercarriage and structure. Reynolds numbers below 70000. In: Exper-
iments in Fluids 23.405-409 (1997).
[3] Mizoguchi M. and Itoh H. Effect of As-
pect Ratio on Aerodynamic Characteristics
at Low Reynolds Numbers. In: AIAA 51.7
(2013).
[4] Nonomura T. Oyama A. Anyoji M. Liu T. and
Fujii K. Effect of Wing Planform on Aero-
dynamic Characteristics at Low Reynolds
Numbers using a Low Density Wind Tunnel.
In: AIAA (2013).
[5] Okamoto M. and Azuma A. Aerodynamic
Characteristics at Low Reynolds Numbers
Figure 3: Comparing the predicted trajectories of the final for Wings of Various Planforms. In: AIAA
model with the actual trajectories with the same center of 49.6 (2011).
gravity positions.
[6] Floreano D. Vidyasagar A. Zufferey JC. and
Kovac M. Performance analysis of jump-
gliding locomotion for miniature robotics.
Bristol, United Kingdom.: IOP Publishing Ltd,
2015.
[7] Moerder D. Beeler S. and Cox D. A Flight Dy-
namics Model for a Small Glider in Ambient
Winds. Langley Research Center, Virginia,
USA.: NASA, 2003.
[8] Users guide and reference manual for Tor-
nado. Online. http://tornado.redhammer.se/.
2012.

Figure 4: Comparing the predicted trajectories of three con-


Finally, please see l3appgroup9.weebly.com for
figurations. With a corrected model, future designs should more information, including videos of our interim
re-consider adding a tail to improve performance. and final designs, as well as the robot in action.

iii

Das könnte Ihnen auch gefallen