Professional Documents
Culture Documents
BACHELOR THESIS IN
AERONAUTICAL ENGINEERING
15 CREDITS, BASIC LEVEL 300
Abstract
To control an un-swept flying wing is problematic in some ways. One of the problems is that
when the wing experiences a disturbance in yaw, it does not, since it has no tail, generate any
torque in the opposite direction as a plane with a vertical stabilizer does. This thesis is
foremost aimed at exploring one particular solution to this problem.
One approach to this problem is to place the motors out on the wing and differentiate the
thrust, to achieve the same torque as splitted elevons or a vertical stabilizer does. This is what
NASA used on the flying unmanned wing HELIOS. Reducing the thrust on the right set of
engines, and increasing the thrust on the left side can mean that the combined thrust is
unchanged. And thus more fuel efficient, and increases endurance.
This projects main goal has been to construct a half scale model of the school project flying
wing Solaris, and to configure a control system for the differentiated thrust as used on Helios.
Thereafter conduct flight testing and evaluate the controllability of the wing in a number of
flight conditions, this to get a sense of the wings characteristics and which parameters one
should adjust to get the best controllability as possible.
After numerous adjustments and test flights it was concluded that it is possible to construct
and fly a wing in this configuration, with relatively simple means, with satisfactory results.
That the torsional rigidity has great influence on the controllability were evident after the test
flights. After redistribution of the components on the wing the conclusion could be made that
the dihedral could be held within the structural limit of the wing.
The results of this thesis will contribute to the project Solaris at Mlardalens University in
Vsters, Sweden. The project was carried out at Mlardalens University. The test flights
were conducted at the former Air Force base F-15 Flygstaden and Mohed in Sderhamn,
Hlsingland.
Sammanfattning
Att styra en flygande vinge som inte har ngon pilform medfr vissa problem i form av att en
sdan konfiguration ej ger ngot terbrdande moment vid en strning i gir-led. Det finns
flera stt att lsa detta. Denna rapport r framfrallt inriktad p att tillmpa en lsning p detta
problem.
En lsning finns p Aerovironment's /NASA:s flygande vinge Helios, dr dragkraften kan
differentieras p hger och vnster sida. Detta gr att dragkraftsvektorn, i teorin, kan vara
konstant och samtidigt ge ett terbrdande moment vid en strning i gir-led.
Detta projekt har gtt ut p att konstruera ett halvskala av projektet och flygande vingen
Solaris, samt konstruera ett styrsystem som fungerar p ett liknande stt som Helios. Drefter
provflyga och utvrdera systemet i olika flygfrhllanden. Detta fr att f en uppfattning om
hur systemet fungerar i luften och vilka parametrar som kan justeras fr att f systemet att
fungera s optimalt som mjligt.
Efter ett flertal justeringar och provflygningar kunde slutsatsen dras att det r mjligt att f en
flygande vinge i denna konfiguration, med relativt enkla medel, att fungera tillfredstllande.
Att vingens vridstyvhet har stor inverkan p kontrollerbarheten kunde konstateras.
Efter komponentplaceringen p vingen justerats kunde ven slutsatsen dras att uppbjningen
av vingen kunde hllas inom ramarna fr vingens strukturella styrka.
Projektet har utfrts vid Mlardalens Hgskola, och resultatet av detta arbete kommer ing i
projekt Solaris p Mlardalens Hgskola. Testflygningarna utfrdes vid F-15 Flygstaden och
Mohed i Sderhamn.
ii
iii
iv
Preface
The Bachelors Program in Aeronautics, at Mlardalen University, last task for the student is
to write a thesis about a subject that the student has chosen.
To build a half-scale of the wing Solaris and conduct flight testing is what I chose and this
thesis is about this.
Testing of a half-scale model was necessary before the full model was to be built.
Table of contents
Contents .............................................................................................................................................1
1 Introduction .....................................................................................................................................1
1.1 Background ...............................................................................................................................1
1.2 Purpose .....................................................................................................................................2
1.3 Problem formulation .................................................................................................................3
1.4 Scope of work ...........................................................................................................................4
2 Method ............................................................................................................................................5
2.1 Design and selection of components .............................................................................................5
2.1.1 Cost estimate .........................................................................................................................5
2.1.2 Design/construction ...............................................................................................................7
2.1.3 Control System .....................................................................................................................10
3 Flight Testing Procedure.................................................................................................................13
3.1 Cooper-Harper Scale ...............................................................................................................13
3.2 Center of gravity......................................................................................................................14
3.3 Flight Testing Schedule ............................................................................................................14
3.4 Results/Discussion...................................................................................................................15
Flight test stage 1 of schedule ...................................................................................................15
Flight test stage 2 of schedule ...................................................................................................16
Flight test stage 3 of schedule ...................................................................................................17
Flight test stage 4-6 of schedule ................................................................................................19
Conclusions/ Recommendations for future work ..............................................................................21
References ........................................................................................................................................22
Acknowledgements ..........................................................................................................................23
Appendices .......................................................................................................................................24
vi
Contents
1 Introduction
1.1 Background
It is possible to control a flying wing in the z-axis in a number of ways. The Northrop B2 on
figure 1 uses splitted elevons, where the elevons can act as air brakes separately out on the
wings. This generates torque to turn the aircraft. A downside of this configuration is that the
splitted elevons cause extra drag. This means that the engines must throttle up to compensate
for the increased drag, and hence burn more fuel. In short, it is not optimized in an endurance
viewpoint.
1.2 Purpose
The main purpose of this project was to investigate and test if it was possible to fly and
control a span loading flying wing with zero wing sweep, with only differentiated thrust as
means of control about the z-axis (yaw), with a limited budget. Other interesting parameters
that were to be investigated were how controllable the wing would be in pitch.
The objective was not to gain exact flight data, but to discover its flight characteristics and
how these characteristics are a function of different parameters.
The results from this project are aimed at gaining knowledge of how the wing will perform to
have a safer starting point for the full-scale version of Solaris, as well as to get an
understanding of how responsive the control system is and how well it will work.
2 Method
2.1 Design and selection of components
2.1.1 Cost estimate
To be able to make a cost estimate, I needed to know what electric components that was
suitable for the wing. As the wing was to be flown in a five section wing configuration with
motors placed on a pylon in the junctions between the wing sections, four motors was needed.
The motors was to be brushless due to its better efficiency, hence four brushless speed
controllers were needed as well. A brushless speed controller uses back EMF signals from the
motor to know the position of the rotor on the motor, this is why it was necessary to use one
speed controller per motor, otherwise the speed controller would not know from which motor
the signals are coming from. At this stage the wing was projected to have one battery pack on
each wing pylon. To go further in selecting a suitable motor package, I a rough total weight
estimate of the wing, with taking this information in consideration, MotoCalc was used to get
an indicator what kind of motor/propeller and battery that would suit the prototype.
I knew that I was going to use five 9 gram HXT900 servos, four 2-3 cell lipo packs, four
motors, four speed controllers, at least one gyro, 2.4 GHz Assan receiver, 5-7 meters of
servo/ESC wire, and the build material for the wing main frame.
The foam to be used was cut and weighed, the pylons made of 2mm thick light plywood. The
wood of which the plywood was composed of was not specified from Hobbytr. But generally
light plywood is composed of the woods Poppel, Okoum or Ceiba.
The motor MotoCalc calculated to suit the wing was the PP-28-26-1200 (Figure 3), this motor
is effective with a 8x4 propeller and a 3-cell lipo battery. The calculated input wattage per
motor was about 110 Watts at cruise. MotoCalc also calculated the amperage to peak at 16.
Item
HXM2730-1300
BA-G2J1
BV01
TGY8x38SF
HXT900
T500.3S.20
AM1002A
007-00301x10
OR006-01002
OR006-00806
Twill-18g
Carbon tubes
Shipping EMS
8,07
28
4,03
1,2
2,69
5,79
1,98
0,99
1,64
2,3
15
12
47,17
Total
(USD)
32,28
28
8,06
9,6
16,14
28,95
1,98
0,99
1,64
2,3
60
48
47,17
Description
Motors
Gyro
V-Tail Mix
Propellers
Servos
Lipos
Gold Connector
Pin Horn
Wheels, small
Main Wheels
Glassfiber
4mm tubes
Shipping
Grand total
(USD)
285,11
2.1.2 Design/construction
When designing the half-scale model of Solaris, some focus was put on keeping the wing
design as simple as possible. The wing structure was not anyway the focus of the model and
we also wanted to reduce complexibility and keeping a low number of build technique
parameters that can result in a twisted wing. Also to shorten the build time, in case of a
serious crash it would be relatively easy to make a new wing panel to be able to continue
flight testing as fast as possible.
A balsa-plywood built-up wing was deemed too complex and work-intense. Thus, a foam-cut
wing would be a good option to be used, figure 4 show comparison. Still, the wing would then
need to be covered with something that would enhance its stiffness and to get a surface that
will not scratch and break too easy. The covering material also needed to be able to resist
compression on the top of the wing and resist tension on the bottom. Covering the wing with
thin fiberglass proved to satisfy these requirements.
Figure 4, Typical sheeted foam core wing (left) and built up wing (right)
The airfoil to be used has previously been decided to be one called Phoenix (figure 5) and is
the result of another students thesis (Reference [1]). Also, a hot-wire saw was built with
constant wire tension to be able to cut out the raw foam and get a good repeatable result so
that all five wing panels could be built identical.
Cavities were cut out under the wing to accommodate the servos, HXT 900 servos were fitted
to the wing sections with hot melt adhesive, then each wing panel was put upside down and
the control rod and elevon control horn was glued in place to get a near perfectly centered
elevon when the servos are in neutral.
The motors were fitted on a 4 mm hollow carbon fiber tube. The tubes were fitted to the
pylons with zip ties at a angle of 2 degrees down measured from the wings mean chord line,
this to get the motors pointing straight forward when the wing was flying un accelerated (with
a 2 degree angle of attack). The tubes that the motors were fitted to had a length of 500 mm,
this to enable fitting an empennage during the initial test flights. The excess tube also made it
possible to adjust the cg. 500 mm long 4 mm diameter carbon tubes were also fitted to the top
of the pylons, these tubes were slid into holes in the wing sections and connect the pylons to
each of the wing sections. To prevent each pylon to rotate about this carbon tube, one
toothpick, as shown on figure 6, on each pylon was fitted close to the trailing side of the wing.
A toothpick was chosen because in case of a crash it would break, and therefore let the wing
section and pylon connected rotate. This will also absorb energy, and hopefully prevent
extensive damage. Figure 7 shows the dimensions of the Solaris prototype.
10
11
Since the wing was to be flown in a five section configuration, there would have to be a mixer
to make the control of the elevons symmetrical. Therefore a mixer was programmed in the
transmitter, the two outer sets of trailing edge flaps got elevon programming and the center
section was only used as elevator as seen on figure 10.
12
13
14
3.4 Results/Discussion
Flight test stage 1 of schedule
A taxi runway at F-15 Sderhamn was used for this test.
The first test stage was to test the wing`s ability to keep a straight line when taxing.
Throttle was applied and the wing yawed a bit to the right, counter stick was applied and the
yaw stopped but was over compensated to the left instead. Yaw compensation to the right was
then applied and the wing yawed back to the right again. It was very hard to avoid PIO in
yaw. And the test run was aborted. At this stage yaw control was rated a 9-10 on the CHs.
The EPA was reduced to about 25% on yaw and the gyro gain was increased to 60% and the
test was conducted again
At this stage, there was a smaller need to compensate since the wing accelerated in a
straighter line. Even though some PIO was evident when yaw was being corrected, the gyro
control was more noticeable now and the wing was able to taxi on the taxiway. The CHs
rating at this stage was between 4 and 5.
The gyro gain was further increased to 65% and EPA on yaw reduced to 19% and the test was
once again done.
At these settings the wing responded well to yaw input and corrected itself adequate,
especially at higher speeds. The system response is not as good as a non-span loader with a
high tail volume, this is due to the artificial stability systems need to first sense the yaw, and
then utilize the differentiated thrust to compensate the yaw. The time this takes and the time
for the motors to rev up and down are greater than the immediate change in lift a vertical
stabilizer generates when the wing or plane experience yaw. But still controllable and the fact
that the wing was getting more stable at higher speeds allowed the testing to move on the next
stage of the flight schedule.
The final CHs rating was 3.
Relevant settings:
Initial gyro gain was set at 50% and later changed to 65%. Initial EPA on the yaw-channel
was initially set at 40% but later adjusted to 19%.
15
16
17
I was suspecting that they had an impact on the wings control in roll since there were two
empennages that were not connected to each other.
A new test was conducted; as the wing lifted the pitch control was still good, but more
sensitive with the horizontal stabilizer removed. Some control in roll was observed for a while
but then it started to roll to the left again. Full left aileron was applied but only moderate
response was experienced. Right rudder was applied and the wing corrected itself for a brief
moment until it stalled and went for the ground. A large amount of elevator was applied just
before impact and the wing stalled close to the ground. The only thing that broke was a
toothpick which was easily replaced at the field.
At this point I rated the performance of the wing in roll-control at a 9-10 on the CooperHarper scale.
I increased the servo throw on aileron to 100% to 140% and reduced the elevator throw from
100% to 75%. Exponential was also added to the elevator channel, the amount exponential
added was -25%.
The test was conducted a third time, and this time the wing was more responsive than before
in aileron correction maneuvers, I also realized that the previous flights had been performed
with too much throttle applied. As I reduced the throttle to about 20-30% airspeed went down
considerably and the wing responded in a much better manner than before, which can indicate
that the sensitivity to over speed are related to a aeroelastical problem. This was later
confirmed when examining a picture taken during the test flight, see figure 14. The pitch
sensitiveness was also reduced and improved
At this stage I had gained the knowledge of some of the wings characteristics and in which
speed range it works best. When the wing was flying at this lower speed I rated this test stage
as a 2-3 on the Cooper-Harper scale.
Relevant settings:
The initial aileron and elevator EPA were set to 100%, and the elevator exponential was set to
0%. At the end of this test stage the aileron EPA was set to 140%, the elevator EPA set to
75% and the elevator exponential -25%
18
19
At this point I felt confident to take away the vertical stabilizers as well and conduct a full
flight (test stage 4-6) with the wing in its final configuration.
The vertical stabilizers were removed, and the crosswind had picked up to 5 m/s.
The wing was accelerated and rotated when flight speed had been attained, this flight was
filmed from behind at take-off. The crosswind made the wing want to yaw left in to the wind,
but after 2-3 seconds the control system for the motors had fully corrected the yaw and kept
the wing flying in a straight line without pilot input. A 180 degree right turn was performed
with no control issues what so ever. Now the wing was flying in crosswind from the right but
this was not noticed much from a pilots perspective. Mild corrective input was given due to
some turbulence when fling in a straight line, I performed another right turn and flew the wing
back to the airstrip. Every turn needed little to no rudder input, I turned right again and lined
up the wing for landing. Throttle was reduced and the wing landed successfully.
I rated the wings overall controllability to a 2-3 on the Cooper-Harper scale, which must be
seen as a great success. Figure 15 shows the wing at runway 29 after the completed flight
testing. Some test flights will be available on YouTube from the user aerodynmag.
20
21
References
[1] Mikis Tsagarakis, Project Solaris - Analysis of airfoil for solar powered flying wing UAV.
http://mdh.diva-portal.org/smash/record.jsf?pid=diva2:450812
[2] Cooper-Harper Scale, wikipedia. http://en.wikipedia.org/wiki/Cooper-Harper_rating_scale
22
Acknowledgements
I would like to thank Gustaf Enebog for proposing a very interesting project.
Thank you Aragorn AALL and Niklas Sjstrm for the support with camera equipment,
filming and photographing most of the test flights.
Thank you Tsagarakis for evaluating airfoil for Solaris and presented the winner to me so I
could build the Solaris half scale with the correct airfoil.
23
Appendices
Airfoil Phoenix coordinates.
Phoenix
1.00000
0.99000
0.98000
0.97000
0.95000
0.92500
0.90000
0.87500
0.85000
0.82500
0.80000
0.77500
0.75000
0.72500
0.70000
0.67500
0.65000
0.60000
0.55000
0.50000
0.45000
0.40000
0.35000
0.30000
0.27500
0.25000
0.22500
0.20000
0.17500
0.15000
0.12500
0.10000
0.07500
0.05000
0.02500
0.02000
0.01500
0.01250
0.01000
0.00750
0.00500
0.00250
0.00100
0.00000
0.00100
0.00250
0.00500
0.00750
0.01000
0.01250
0.00091
0.00146
0.00203
0.00261
0.00388
0.00571
0.00779
0.01005
0.01245
0.01498
0.01760
0.02029
0.02304
0.02586
0.02872
0.03163
0.03458
0.04051
0.04639
0.05205
0.05733
0.06200
0.06577
0.06814
0.06863
0.06855
0.06781
0.06631
0.06396
0.06068
0.05638
0.05093
0.04415
0.03557
0.02402
0.02109
0.01779
0.01595
0.01395
0.01174
0.00921
0.00611
0.00362
0.00010
-0.00259
-0.00373
-0.00471
-0.00532
-0.00576
-0.00610
24
0.01500
0.02000
0.02500
0.03500
0.05000
0.07500
0.10000
0.12500
0.15000
0.17500
0.20000
0.22500
0.25000
0.27500
0.30000
0.35000
0.40000
0.45000
0.50000
0.55000
0.60000
0.65000
0.67500
0.70000
0.72500
0.75000
0.77500
0.80000
0.82500
0.85000
0.87500
0.90000
0.92500
0.95000
0.97000
0.98000
0.99000
1.00000
-0.00639
-0.00685
-0.00721
-0.00777
-0.00835
-0.00907
-0.00971
-0.01032
-0.01089
-0.01143
-0.01195
-0.01243
-0.01288
-0.01331
-0.01371
-0.01443
-0.01506
-0.01561
-0.01607
-0.01641
-0.01654
-0.01638
-0.01617
-0.01585
-0.01542
-0.01487
-0.01420
-0.01340
-0.01247
-0.01139
-0.01017
-0.00880
-0.00726
-0.00554
-0.00402
-0.00323
-0.00242
-0.00161
25