You are on page 1of 4

I MPERIAL C OLLEGE L ONDON

AE3-418 L3 Application Report


Airframe design of a jump-glider

Department: Aeronautics
Course: AE3-418 L3 Application
Academic Supervisors: Dr. Mirko Kovac

Group: 15
Website: http://l3glider.weebly.com/
Students: Marc Ewenz 00834234
Zhi Han Gary Chew 00840486
Thilo Braun 00812244
Antoine Lubrano 00820239
Clemence Rubiella 00865125
Cyrus Hessabi 00834165
Gabriel Taconet 00866568
Guillaume Fontan 00851703
Date: 24 / 02 / 2017
Academic Year: 2016 - 2017
L3 Application Group 15

1 Introduction evolved by nature over millions if not billions of


years. Inspiration was taken from nature such as fly-
ing squirrel or even seeds but in short, Simple is
This report aims at explaining the conception, de-
beautiful.
sign, building and testing process of Group 15s long
distance jump-gliding robot while respecting the im-
posed constraints detailed in [1]. Gliders perfor- 3 Performance Modelling
mance is then analysed and compared to our math-
ematical initial prediction.
While evaluating the wing planform, various plan-
forms including rectangular, elliptical, delta, and
2 Design Process Zimmerman configurations were considered. A con-
sultation of various experiments [2] revealed that for
comparable low Reynolds number flight, an Inverse
One of the key aspects of any glider aircraft is its Zimmerman configuration yields the highest L/D ra-
lift generating surface the wing. Our design pro- tio, while having favourable stall performance and a
cess was driven by the tradeoff between weight, aero- high lifting area given the dimensional restrictions for
dynamic performance, and manufacturing complex- the glider. Additionally, biplane and tail configura-
ity. On the one hand, to reduce the required lift, the tions were considered. However, given the low flight
weight needs to be minimized in order to reduce the velocity, the additional lift of a second wing would
amount of lift required. At the same time, the amount be marginal while the drag and weight are increased
of lift generated should be maximized while ensuring significantly. Similarly, given the short flight distance
maximum stability and minimum drag. Lastly, the and large wing area, the additional stability from hav-
complexity of manufacture needs to be evaluated to ing a tail would be minimal while the weight increase
ensure that the design is both feasible for the given is significant. Lastly, other modifications for addi-
time frame, and minimizes potential for manufactur- tional lateral stability and higher lift such as dihe-
ing issues. dral and camber were considered. However, the addi-
tional manufacturing difficulties and potential break-
Another important aspect is structural integrity and
ing points on crash landing outweigh their respective
some lessons were learnt from the past: most robots
advantages. Subsequently, a flat plate tailless design
are unable to obtain optimal performance because of
was chosen for the initial configuration.
ill-placement in the centre of gravity and structure
breaks apart upon impact, wasting precious time in For testing, two different planform areas were de-
fixing it and adding extra cost. Our philosophy was signed: one with a maximum chord of 50 cm and
the following: (1) vehicle must accelerate off the an aspect ratio of 0.95, and another with a maximum
ramp with the highest possible velocity, generating span of 50 cm and an aspect ratio of 2.1. While the
the most lift and acquire the most inertia to travel the first of these designs maximizes the available area,
furthest possible, (2) vehicle chassis should be of high our predictions showed that the larger span design
structural efficiency - Maintain structural integrity would allow for more stability and a longer glide.
of the robot after repeated crashes with a minimum Since the cost and workload involved in the individual
weight design, (3) structural components are easily wing manufacture is quite low, we decided to manu-
manufacturable and line-replaceable - Each structural facture both, allowing for an experimental compari-
part used in the chassis can be swapped out easily son.
with minimal effort or tools, (4) built-in design with
the ability to alter the axial component of the robots Interestingly, the 50 cm chord variant showed better
centre of gravity as required. This provides us with aerodynamic performance and longer glide. Contrary
a margin of safety to account for design and manu- to our expectations, it was also laterally stable. Lon-
facturing uncertainties without impacting the perfor- gitudinal stability was achieved by moving the un-
mance of the robot and (5) overall robot is mass dercarriage into the right position, such that the CG
balanced/symmetric along the roll axis (i.e. stabil- and AC were at the same point. Four polymer pins
ity along the roll axis to travel straight on and off the were glued on the wing near the predicted position
ramp) of the AC. Figure C.1d shows the mechanism which
allowed to attach and detach the wing from the rest
Throughout the design process, we consulted some of the structure and hence find its optimum position.
of the most efficient glider designs ever built: those Special care was taken to mount the motor and the

1
L3 Application Group 15

rest of the circuit centrally for a laterally symmetric ability to take the structural components apart and put
load distribution. Hence no lateral nor longitudinal them together without special tools. Ability to al-
moments were induced, which allowed a steady flight ter the centre of gravity as desired is made possible
as seen during testing. Since the slightly larger area of through the attachment lugs of the wing and holes on
the 50 cm chord variant additionally increased aero- the side frame, connected by carbon rods. Concept
dynamic performance, it was chosen for testing. was further refined by using smaller and thinner lugs
when we realised the bond between the foam wing
and acetyl lugs was stronger than expected. Overall
4 Mechanical Implementation structural mass is also reduced in the process. At-
tempted to drive the rear wheels with gears placed
perpendicularly which turned out to problematic in
Trajectory: Firstly, considering the ramp phase, we
terms of the gear coupling. The gear are easily sepa-
have tried to ascertain the configuration that gives
rated. Trade-off is made to use straight spur gear that
the maximum velocity at the end of the ramp. For
couple easily and giving up mass symmetry along roll
this mathematical model , we have considered drag,
axis.
weight and calculated reaction forces. Design param-
eters have been approximated via educated compar-
isons such as the Cd which has been assumed equal
5 Robot Evaluation
to that of a modern saloon car. The motor parame-
ters considered were those of HP Pololu motors with
scaled performances and the wheels, offered on the The robot performed as expected, with excellent lon-
Technobot website. Solving the determined differen- gitudinal stability, which was reflected in a stable
tial equations of motion for displacement and veloc- pitch. The flight path was generally straight, while
ity. The optimal wheel radius and maximum velocity veering off course slightly towards the end of the
were then found. To further approximate the speed glide. It was observed that the direction of the veer-
of our vehicle more precisely and to determine the ing was always to the left, indicating a slight asym-
range, the given mathematical model was used. The metry of the robot. This could have arisen from
code has been modified accordingly to coincide with the slightly offset CG of the very asymmetric circuit
our tailess design, with the weight distribution and board. Another error source could have been the sty-
lateral stability having been carefully considered by rofoam wing, which may have deformed during a pre-
implementing a movable wing position to adapt the vious test flight and induced a slight rolling moment.
CG correctly for an optimum pitch with no stall. Tak- The longest glide length was 3.2965 m.
ing into account aerodynamic forces and moments,
The robot left the ramp slightly slower than desired.
a complete set of differential equations was deter-
While investigating this issue we realised that the rear
mined. Solving these, we determined the range of our
bar was not fixed as firmly as it should have been.
vehicle configuration, and then varying further the de-
Its movement moved the gear attached to the motor
sign parameters in the code we found the optimal con-
apart from the gear attached to the rear shaft and lead
figuration. The optimal trajectory for our last iteration
to imperfect power transmission. In future iterations
can be seen in Figure A.1, for a found optimal angle
this could be prevented by gluing the rear bar to the
of attack of 4 degrees.
side bars, hence fixing the position of both gears with
Structure: Investigation into the mathematical model respect to each other.
(ramp portion) tells us, we need to use the smallest
possible wheels and the most powerful motor to ac-
quire the greatest off ramp velocity. Power from mo- 6 Conclusion and Improvements
tors must be transmitted to the wheels in the most ef-
ficient manner, minimising losses. Off the ramp, the
In conclusion, a simple design was efficient enough
vehicle will acquire a nose-up attitude or positive an-
to respect all the imposed tasks for this application.
gle of attack, eventually landing on the rear wheels.
Practical results bears resemblance to the theoretical
This is the region with the highest impact stresses
model and thus validates our design process.
hence the need to structurally reinforce this region.
Use of mild interference fits between the base plates To improve the characteristics of the glider, a stiffer
and the side frames to keep the structure together wing frame may be used in order to reduce deforma-
Use of mild interference push-together fits implies the tion.

2
L3 Application Group 15

References C Glider Manufacturing

[1] Dr. Kovac, L3 Introduction Presentation, Impe-


rial College London, October 2016.
[2] Mueller, Thomas J., and Gabriel E. Torres,
Aerodynamics of low aspect ratio wings at low
Reynolds numbers with applications to micro air
vehicle design and optimization, No. Undas-Fr-
2015, Notre Dame Univ in Office of Research,
2001.

A Glide Trajectory
(a) Robot glider final assembly side view

(b) Bottom view

Figure A.1: Trajectory comparison


between predicted model and testing

B Initial CAD Modelling and Modifications

(c) Back view of motor transmission to the bach-


wheels

Figure B.1: Initial assembly

(a) Front base support (b) Lug (d) Side view of wings attachement mechanism

Figure B.2: Modified elements after Figure C.1: Glider robot once
first assembly assembled

You might also like