Вы находитесь на странице: 1из 3

Group 6 L3 Applications Exercise

Abdul Alam, Ion Berasaluce, Haobo Dong, Kieran Downie, Matthew Gillie, James Marchant, Zhehong Zhang

Abstract
Micro Autonomous Vehicle (MAV) design is a constantly developing field, and the study of biomime-
tics identifies many cases in nature where animals have the ability to cover large distances expending
as little energy as possible. A miniature glider prototype of maximum dimension of 50cm and subject
to multiple constraints was successfully designed, manufactured and tested. The design process began
sizing the main components of the glider and developing numerous mathematical models. Once ma-
nufactured, improvements to the original prototype were identified during testing. The last iteration
was then subject to a final test and the results correlate with the analytical predictions made during
the conceptual design stage.
Introduction animals extend their bodies as much as possi-
The objective of the project was to design a micro- ble while gliding to increase the lifting surface. A
glider capable of achieving the longest glide di- Zimmerman planform[1] was used to maximise the
stance possible, given a runway of 1.8m in length wing area, as it performs well at very low aspect
at an altitude of 1.2m. The design must have a ma- ratios, as demonstrated in the research performed
ximum dimension not exceeding 50cm and weigh by Notre Dame University. Specifically, an inverse
less than 70g. zimmerman was selected to increase lift and drag
Initially, extensive research on common gliders in performance at high angles of attack. An aspect
both nature and industry was conducted and ap- ratio of 1.5 was selected to provide a compromise
plied these concepts to the design. between maximising wing area and obtaining the
best possible lift-to-drag ratio. Lift and drag cha-
Design & Modelling racteristics were resolved from polars produced
Drivetrain Design by the panel solver XFLR5 for the 3 dimensio-
The acceleration phase was considered the most nal wing. With this data the in-flight equations
important of the design process as flight was ex- of motion [2] were then solved to give our model
pected to mainly display projectile motion. Hence predictions.
the velocity at the end of the ramp was to be ma- " #
S
"
2 gcos()
#
d 2m CL v
ximised. To model this phase of the task, the kine- = S (2)
matic equations of motion 1 were used, where drag
dt gsin() 2m CD v 2
was assumed to be that of a bluff body considering Tailplane Design
the low Reynolds number of the flow. Initial simulations run indicated that the glider
1 1 would be very sensitive to the offset between the
a= (Tmax (Tmax /max )) CD AV 2 (1)
mr 2 centre of mass and aerodynamic centre - even 1
From equation 1, it is clear that the most import- mm offset caused a significant phugoid oscillati-
ant factor to be taken into account when optimi- on. Due to the potential error during manufacture,
sing the acceleration is the wheel radius r. The and the change in the position of the aerodynamic
optimum ensures fast acceleration without wheel centre characteristic of a zimmerman platform, a
slip. The plot seen in Figure 1 shows that ramp tail was included to ensure longitudinal stability.
velocity increases as wheel radius decreases, ho- The tailplane, as compared to the wing, produ-
wever a small wheel is likely to slip. Using the ces very little lift. Therefore, for the purposes of
weight and torque estimates, the minimum wheel design and modeling, these effects were neglected
radius to guarantee sufficient traction was found and it was sized purely for stability. It was desi-
to be 11.8 mm. A wheel of radius 12.5 mm was gned to have an adjustable angle of attack such
available from the online retailer TechnoBots. that it could be adjusted easily between tests to
The chassis consisted of a tricycle design with 2 account for the unpredictability in the flight con-
drive wheels and one for stability. Previous tricy- ditions.
cle designs had directional stability issues, which Mechanical Implementation
was accounted for by limiting the lateral move-
Drivetrain Chassis
ment of the rear wheel. Lastly a 1:1 gear ratio was
The chassis and wheel supports were laser cut
selected, ensuring the correct acceleration profile.
from 2mm thick acetal sheets. Acetal provided
Wing Design enough stiffness to withstand the high impact
The wing design was inspired by natural gliders loads of the test landings. Furthermore a adjust-
such as Draco Lizard and Flying Squirrel. Both able motor mount was devised to ensure correct

1
Group 6 L3 Applications Exercise
Abdul Alam, Ion Berasaluce, Haobo Dong, Kieran Downie, Matthew Gillie, James Marchant, Zhehong Zhang

gear meshing. The micro-controller and the rest it was constrained to 10mm discrete intervals and
of the circuitry was soldered to the baseplate in a perhaps did not provide sufficient accuracy to en-
compact design. The wing mounts were then glued sure a stable configuration. It was not possible
on to the chassis with acetal glue to ensure a light- to adjust this during the test however due to the
weight strong bond. The axles were made from over-interference fit. If a next iteration were to be
carbon fibre rods. These are lightweight and stiff undertaken, a more continuous method of moving
enough to withstand the centrifugal forces they the wing relative to the chassis is recommended,
are exposed to without deflection, without incur- as with the tailplane.
ring a massive weight penalty. The tailplane did not work as efficiently as expec-
Wing & Tailplane ted, as due to the size constraints it had to be
positioned immediately in the wake of the main
The wing and the tail were laser cut from 3mm wing, resulting in its exposure to dirty airflow. It
Depron foam. This ensured a smooth leading, al- is expected that the tail would have been more ef-
lowing for a much better aerodynamic performan- ficient if it had been moved away out of the wake,
ce. The leading edges were then rounded to further perhaps further back or upwards, above the main
improve this. The tailplane was mounted on sliders wing. This configuration would have resulted in a
on two 2mm diameter carbon fibre rods allowing more complex mounting system however and in-
the tailplane to be positioned along a continuous curred a large weight penalty.
domain. The maximum velocity reached by the glider du-
Weight Estimations ring the ramp phase was 3.3ms1 , which was slight-
The predicted weight of our prototype according ly lower than the 3.7ms1 that was predicted by
to our CAD model was of 68g. However, this esti- the model. This may have been largely due to that
mation didnt include the baseplate, glue, wiring fact that at each point during the ramp simulati-
or manufacturing defects that could have happe- on, the model assumed that the torque deliver-
ned. This resulted in our final prototype being ed to the wheels was optimum, producing a non-
74g (4g over the weight limit). Nevertheless, ex- linear ramping function that was not possible to
cess material in the chassis could be removed and implement in the Arduino software. The resulting
a further optimisation in the structure, using a fi- linear approximation may not have been sufficient
nite element method, could further reduce weight to reach maximum velocity with slip and hence
and ensure that the glider is within the specified affecting the final velocity. Moreover mechanical
weight limit. defects in the bearings and gear meshing as well
Performance possible manufacturing defects could have also be-
During the final test, the glider achieved a maxi- en present and were not taken into account by the
mum distance of 3.99m from the start of the ramp. mathematical models produced.
The variance in the distance travelled was 1.2 ran- During the final tests, it was observed that the
ging from a minimum distance 2.79m. This shorter glider had a tendency to veer left during the ac-
distance occurred in early testing as the tail angle celeration phase. This was deemed likely due to
was not optimum, and was caused by a severe stall a slight bias in the center of mass location to the
early in the flight phase. Indeed stall was present left, or a lateral imbalance in mechanical resistan-
in each flight, although occurred much later as the ce, perhaps in the bearings. Upon further inspec-
tail setting angle was altered. tion, it was discovered that during operation, the
It is expected that moving the wing backward on rear wheel angled to one side and created a slight
the chassis would have further delayed stall, po- force to the left. This wheel was made in-house
tentially eliminating it completely if the aerody- and was of significantly lower quality than those
namic centre converged sufficiently on the centre purchased from the TechnoBots website. In ano-
of gravity. Although there was some adjustabili- ther iteration more consideration should be given
ty in the interface between the chassis and wing, to ensuring that this phenomenon is eliminated.

Conclusion
Overall, this was deemed to be a successful design. It is believed that very small changes to the current
iteration could yield significant increase in the performance of the glider. Most notably, the mechanism
for positioning the wing relative to the chassis should be refined for more accuracy. To improve upon
the design process, more emphasis must be given to the testing stage, to provide more data on the
current configuration to use to find an optimum for final operation .

2
References
[1] Aerodynamics of Low Aspect Ratio Wings at Low Reynolds Numbers With Applications to Mi-
cro Air Vehicle Design and Optimisation; Thomas J. Mueller & Gabriel E. Torres. Department
of Aerospace & Mechanical Engineering; University of Notre Dame. November 2001

Figures

Figure 1: Wheel diameter vs Velocity plot (left), Trajectory model prediction vs actual result (right)

Figure 2: XFLR5 Polar Plots

Figure 3: Wing Planform

Вам также может понравиться