Detailed design of a forest surveillance UAV

Detailed design of a forest surveillance UAV
Oleh Tkachuk
Thesis to obtain the Master of Science Degree in
Aerospace Engineering
Supervisors: Prof. Fernando José Parracho Lau

Eng. José Renato Martins dos Santos Machado
Examination Committee
Chairperson: Prof. Filipe Szolnoky Ramos Pinto Cunha
Supervisor: Prof. Fernando José Parracho Lau
Member of the Committee: Prof. Afzal Suleman
November 2018
ii
Acknowledgments
Firstly, I would like to acknowledge my supervisors, Prof. Fernando Lau, Dr. Frederico Afonso, Eng.
Renato Machado and Eng. Alexander Costa for all the support, knowledge shared with me, feedback
and guidance.
I want to thank to CEiiA for the opportunity to work in one of their most ambitious projects. To all of
the team, thank you for the help and constant motivation. A special thanks to engineers Tiago Nunes,
Inês Martins and Inês Furtado for their unwavering support.
It is with great gratitude that I thank Inês Silva for her unconditional patience, support and for always
being by my side.
Finally, I must dedicate this work to my parents and my sister, the foundation that allowed me to pur-
sue a higher education level. Thank you for always encouraging me to follow my dreams and respecting
my choices.
iii
iv
Resumo
A presente tese tem como objetivo desenvolver um Veı́culo Aéreo Não Tripulado (VANT) para vigilância
florestal, uma plataforma de baixo custo adequada a missões de alto risco. O projeto chama-se
DeltaSpotter, com um primeiro protótipo funcional já existente. Este primeiro protótipo em conjunto
com a análise de mercado permitiu perceber as caracterı́sticas a melhorar numa nova versão, de forma
a obter uma aeronave de grande autonomia. Após uma análise das possı́veis configurações, decidiu-
se manter a configuração da asa voadora, visto que é a configuração mais adequada aos requisitos.
Através de uma pesquisa bibliográfica analisou-se materiais inovadores, técnicas de construção e sis-
temas propulsivos a utilizar de forma a reduzir o custo sem comprometer o desempenho.
Com as tomadas de decisão finalizadas, procedeu-se ao desenho da asa, desde a sua forma ao
seu tamanho. Terminado o desenho conceptual, seguiu-se a análise de estabilidade, a construção do
envelope de voo e, com recurso a programas de desenho assistido por computador finalizou-se a asa
e a sua estrutura interna. A estes elementos foi ainda realizada uma análise estrutural com recurso
a elementos finitos. Com o modelo estrutural finalizado, estudou-se o desempenho aerodinâmico da
aeronave, permitindo assim obter o valor teórico da autonomia. Como resultado deste trabalho, obteve-
se um aumento em autonomia de cerca de duas horas quando comparado com o primeiro protótipo.
Por último, executou-se uma simulação da missão tendo sido descritos vários fatores associados à
realização da missão.
Palavras-chave: MA Plástico, Perfis com Núcleos em Espuma, Missões de Alto Risco, Baixo
Custo, Projeto do UAV.
v
vi
Abstract
The aim of this thesis is to develop an Unmanned Aerial Vehicle (UAV) for fire surveillance by using an
adequate, low-cost platform for high risk missions. This project, named DeltaSpotter, had a previously
developed first working prototype, which in combination with market research allowed for a better un-
derstanding of the characteristics necessary for an efficient long endurance UAV design. After detailed
analysis of several possible configurations, the flying wing design was kept, as it was the most adequate
for the mission requirements. Extensive research in regard to materials, methods of manufacturing and
propulsion systems was completed, resulting in lower costs without compromising performance.
In addition, both size and shape of the UAV were defined. With the conceptual design completed, the
stability analysis and flight envelop of the design ensues. This was followed by geometry modeling (by
means of CAD software) and structural design (using a finite element software). To wrap up the design,
the aerodynamic performance was studied, using CFD, which allowed for the calculation of a theoretical
endurance value. The resulting DeltaSpotter aircraft has a two hour increase in endurance compared
with the previous version. To finish the development process, several aspects of the drone’s mission
were described in detail.
Keywords: AM Plastic, Foam Core Airfoils, High Risk Missions, Low-Cost, UAV Design.
vii
viii
Contents
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
Resumo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii
1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 State of the art 7

2.1 Applications of UAVs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1 UAS Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.2 UAV Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.3 Market Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 Materials and Manufacturing Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.1 Wood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.2 Foams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.3 Polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3 Propulsive System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3.1 IC Engines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3.2 Electric Propulsion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3 Preliminary Design 21
3.1 Design Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.2 First Prototype: DeltaSpotter V1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.3 Requirements and Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.4 Mission Profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.5 Airframe Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
ix
3.6 Propulsion System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.7 Conceptual Design of DeltaSpotter V2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.7.1 Initial Sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.7.2 Endurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.7.3 Aerodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.7.4 Electric Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.7.5 Circuit Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.7.6 Stability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.7.7 Flight Envelope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4 Structural Design and Analysis 55

4.1 Geometry and Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.1.1 Material Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.1.2 Geometry Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.2 CFD Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.3 Structural Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.3.1 Computational Structural Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.4 Mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5 Conclusions 79
5.1 Achievements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Bibliography 83
A Market Analysis of Military UAVs 91

A.1 Technical Weights, Dimensions and Prices of the Components . . . . . . . . . . . . . . . 93
B Aerodynamic Analysis 94
C Sizing of Control Actuators 96
D Stability 97
D.1 Static Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
D.2 Dynamic Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
E Structural design 98
F Computational Dynamic Analysis 99
G Computational Structural Analysis 100
x
List of Tables
1.1 Decision matrix, comparison between different surveillance methods . . . . . . . . . . . . 4
2.1 Characteristics of each foam [31–33] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.2 Description of AM technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1 Data for cruise speed of 12 m/s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.2 Data related with battery and weight. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.3 Decision matrix of the UAV configuration, . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.4 Weight of the components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.5 Data of platform V2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.6 Reference values used in aerodynamic analysis. . . . . . . . . . . . . . . . . . . . . . . . 41
3.7 Performance parameters of DeltaSpotter V2 from XFLR5. . . . . . . . . . . . . . . . . . . 42
3.8 Battery features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.9 Static stability analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.10 Longitudinal and lateral analysis of DeltaSpotter V2. . . . . . . . . . . . . . . . . . . . . . 49
3.11 Data related with battery and weight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.12 Speed specifications of the V-n diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.13 Maximum load factors for each velocity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.1 Summary of relevant parameters for 3D printing. . . . . . . . . . . . . . . . . . . . . . . . 58

4.2 Resume of the material proprieties. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.3 Lift and drag coefficients from Fluent. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.4 Comparison between results obtained from XFLR5 and Fluent. . . . . . . . . . . . . . . . 64
4.5 Summary of the analytical results obtained. . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.6 Summary of the obtained maximum values. . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.7 Options for flight altitude and camera parameters and their influence in the distances
between lines of passage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
A.1 Military UAVs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

A.2 Weight and dimensions of the electric components . . . . . . . . . . . . . . . . . . . . . . 93
D.1 Longitudinal derivatives of DeltaSpotter V2 . . . . . . . . . . . . . . . . . . . . . . . . . . 97

D.2 Lateral derivatives of DeltaSpotter V2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
xi
D.3 Longitudinal and lateral analysis of DeltaSpotter V1 . . . . . . . . . . . . . . . . . . . . . 97
xii
List of Figures
1.1 Burned areas in continental Portugal where particular regions are routinely most affected
by fires [2]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Causes of ignitions in Portugal between 2006 and 2013 (adopted from [6]) . . . . . . . . 2
1.3 The first prototype: DeltaSpotter V1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1 Categories of UAVs [13] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.2 Penguin BE Electric Platform [16] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3 Litus [17] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Freya [18] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.5 UX5 AG [19] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.6 DT 18 HD [20] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.7 eBee SQ [21] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.8 AgEagle RX48 [22] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.9 Aeromapper EV2 [23] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.10 4 axis hot wire cutting the foam [30] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.11 IC engine - 3W-28i CS [42] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.12 Electric motor - Hacker A50-12S V4 [40] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.1 General Design Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

3.2 Weight vs endurance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3 Weight vs wingspan. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.4 Weight vs average cruise speed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.5 Endurance vs price. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.6 Mission profile (adapted from [47]). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.7 Propulsive systems [48] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.8 Conceptual Design Sequence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.9 Cl over Cd for Reynolds number of 430000. . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.10 Cd as a function of angle of attack (α) [deg] for Reynolds number of 430000. . . . . . . . 35
3.11 Lift coefficient as a function of the angle of attack for 185000 Reynolds number. . . . . . . 36
3.12 Representation of the wing dimensions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.13 Comparison of different airfoils in the platforms. . . . . . . . . . . . . . . . . . . . . . . . . 38
xiii
3.14 Wing root airfoil for DeltaSpotter V2, MH60-MOD, t/c=14% . . . . . . . . . . . . . . . . . . 39
3.15 Wing tip airfoil for DeltaSpotter V2, MH60-MOD, t/c=12% . . . . . . . . . . . . . . . . . . 40
3.16 Differences between DeltaSpotter V1 and V2 . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.17 Kit Autopilot + GPS + Radio Wireless Telemetry + Power module [72] . . . . . . . . . . . 43
3.18 Battery ZIPPY Flightmax 4S 5Ah [73] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.19 Servo [76] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.20 Propulsion system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.21 ESC 60A [79] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.22 Circuit scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.23 Modal response of the Dutch roll of both versions, red line represents the first version and
blue the second. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.24 Modal response of the Dutch roll for both versions, where the red line represents the V1
and the blue line the V2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.25 Maneuver V-n diagram of the DeltaSpotter V2 for standard sea level. . . . . . . . . . . . . 52
3.26 Combination of maneuver V-n diagram with the load factors induced by the gusts. . . . . 53
4.1 Flowchart with the various steps of structural design and aerodynamic analysis performed
during wing design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2 Scheme of the coupling between Fluent and Structural analysis. . . . . . . . . . . . . . . 56
4.3 Cantilever beam bending test. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.4 Load over deflection, where the blue points represent elastic mode, and after 100 mm of
deflection, the beam entered in the plastic mode (orange points). . . . . . . . . . . . . . . 57
4.5 Isometric view of the internal structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.6 Isometric view of the wing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.7 Representation of the dimensions of the fluid domain. . . . . . . . . . . . . . . . . . . . . 61
4.8 Examples of three different types of meshes [94]. . . . . . . . . . . . . . . . . . . . . . . . 61
4.9 Representation of the grid. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.10 Results analysis in function of the number of elements. . . . . . . . . . . . . . . . . . . . 63
4.11 Pressure distribution around the wing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.12 Representation of the velocity around the wing. . . . . . . . . . . . . . . . . . . . . . . . 65
4.13 Approximation of the wing to the beam with constant cross section at mean chord (z =
0.75m). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.14 Representation of the beam’s lift force distribution. . . . . . . . . . . . . . . . . . . . . . . 66
4.15 Representation of the element SOLID 186 [107]. . . . . . . . . . . . . . . . . . . . . . . . 68
4.16 Representation of mesh with 5840 elements (26709 nodes). . . . . . . . . . . . . . . . . . 68
4.17 Results analysis in function of the number of elements. . . . . . . . . . . . . . . . . . . . 68
4.18 Equivalent Von Mises Stress for cruise speed. . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.19 Wing deformation for cruise speed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.20 Element SOLID 187 [107]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
xiv
4.21 Representation of mesh with 16517 elements (31989 nodes). . . . . . . . . . . . . . . . . 71
4.22 Representation of the wing deformation with the reinforcement, for cruise speed. . . . . . 71
4.23 Representation of Von-Mises stress distribution, for cruise speed. . . . . . . . . . . . . . . 72
4.24 Representation of Von-Mises stress distribution in the wing. . . . . . . . . . . . . . . . . . 73
4.25 Representation of Von-Mises stress distribution in the reinforcement. . . . . . . . . . . . . 73
4.26 Representation of the internal structure with reinforcements. . . . . . . . . . . . . . . . . . 74
4.27 Representation of the camera position. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.28 Representation of the first simulation (400 m). . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.29 Representation of the second simulation (700 m). . . . . . . . . . . . . . . . . . . . . . . . 77
A.1 Wingspan vs endurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

A.2 Weight vs wing span . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
A.3 Weight vs endurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
B.1 Cl vs. Cd for the Reynolds number of 285000. . . . . . . . . . . . . . . . . . . . . . . . . . 94

B.2 Cl over Cd for Reynolds number of 570000. . . . . . . . . . . . . . . . . . . . . . . . . . . 95
B.3 CL /CD with five different relative thickness. The green line corresponds to 10% of the
relative tip thickness, the pink line corresponds to 11%, dark line to 12%, red line to 13%
and the blue line to 14%. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
C.1 Calculation of maximum required torque using PredimRC. . . . . . . . . . . . . . . . . . . 96
E.1 Top view of the internal structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

E.2 Lateral view of the internal structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
F.1 Mesh convergence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
G.1 Maximum Principal Stress in order of the number of elements . . . . . . . . . . . . . . . . 100

G.2 Minimum Principal Stress in order of the number of elements . . . . . . . . . . . . . . . . 100
xv
xvi
Nomenclature
Greek symbols
α Angle of attack.
ηP Propulsive efficiency.
ηmotor Motor efficiency.
ηpropeller Propeller efficiency.
Λ Sweep angle.
µ Atmospheric dynamic viscosity.
ωn Natural frequency.
ρ Density.
σ Stress.
σyield Plasticity yield stress.
ϕ Climb angle.
ζ Damping ratio.
Roman symbols
û Gust velocity.
c Mean aerodynamic chord.
AR Aspect Ratio.
b Wingspan.
c Chord.
CD Three-dimensional drag coefficient.
CL Three-dimensional lift coefficient.
Cl Two-dimensional lift coefficient.
xvii
CM Coefficient of moment.
cP Specific Fuel Consumption.
CD0 Three-dimensional zero lift drag coefficient.
CDi Three-dimensional lift induced drag coefficient.
CLα Three-dimensional lift vs angle of attack curve slope.
D Aircraft drag.
E Young’s modulus.
e Osvald factor.
ET Total energy stored.
h Altitude.
I Moment of inertia.
i Height.
L Aircraft lift.
l Length.
n Load factor.
P Power.
Q Load.
q Lift force distribution.
Qbattery Battery charge.
Re Reynolds Number.
S Wing area.
T Thrust.
t Time.
t/c Airfoil to chord ratio.
TE Endurance.
UM O Maximum operational speed.
Vbattery Battery voltage.
W Aircraft Weight.
xviii
w Width.
Wf Final Aircraft Weight.
Wi Initial Aircraft Weight.
y Deformation.
p Pressure.
U Velocity vector.
Subscripts
climb Climb conditions.
max Maximum conditions.
n Normal component.
r Root.
s Stall.
t Tip.
x, y, z Cartesian components.
xix
xx
Glossary
AM Additive Manufacturing.
CAD Computer Aided Design.
CEiiA Centre of Engineering and Product Development
CFD Computational Fluid Dynamics.
CG Center of Gravity.
CICLOP Sistema Integrado de Videovigilância Florestal.
CTOL Conventional Take-Off and Landing.
EPS Expanded Polystyrene.
EPP Expanded Polypropylene.
ESC Electronic Speed Controller.
FDM Fused Deposition Modeling.
FEM Finite Element Method.
FOV Field of View
GNR Guarda Nacional Republicana.
GPS Global Position System.
HALE High Altitude Long Endurance.
HTOL Horizontal Take-off and Landing UAV.
IMU Inertial Measurement Unit.
LiPo Lithium Polymer battery.
LLT Lifting Line Theory.
MALE Medium Altitude Long Endurance.
MTOW Maximum Take Of Weight.
MUAV Micro & Mini UAV.
RANS Reynolds Averaged Navier-Stokes.
RC Radio Control.
SSL Standard Sea Level.
TUAV Tactical Unmanned Aerial Vehicle.
UAS Unmanned Aerial System.
UAV Unmanned Aerial Vehicle.
VANT Veı́culo Aéreo Não Tripulado
xxi
VTOL Vertical Take-off and Landing UAV.
XPS Extruded Polystyrene.
xxii
Chapter 1
Introduction
1.1 Motivation
In Continental Portugal 39% of the territory is forest terrain, which translates into around 3,481 thousand
hectares [1]. In the last decade alone, hotter and drier summers have been setting off more forest fires,
which are accelerating a decades-old migration from rural areas to larger coastal cities. This leaves rural
lands unattended, which in turn fuel new and more intense fires that spread and burn even faster [2].
This vicious cycle culminated in 2017, the most catastrophic and deadly year on record with more than
100 mortal victims, hundreds more displaced and 400 thousand hectares of burned land, around 11.5%
of total forest area burned [3].
(a) Burned area in 2017 (b) Burned area 1975-2008
Figure 1.1: Burned areas in continental Portugal where particular regions are routinely most affected by
fires [2].
1
Figure 1.1(a) represents the burned area in 2017 and Figure 1.1(b) represents the burned area
between 1975 and 2008. According to these figures, it becomes evident that the most affected areas are
located in the center and north of Portugal. It is also observable that the affected areas are similar every
year, showing a clear flaw in the prevention and surveillance methodology that needs to be addressed.
The Pedrogão Grande report and other available governmental data shows that the surveillance costs
are only a fraction of the suppression costs and that there is a stagnant spending in surveillance, further
validating the need for improvements [4].
In order to better understand the surveillance methods currently used and the requirements that an
Unmanned Aerial Vehicle (UAV) should have in order to be a powerful surveillance tool, four meetings
were held with the entities responsible for surveillance and firefighting: Guarda Nacional Republicana
(GNR), Firefighters and Civil Protection. In these meetings, the main (forest) surveillance techniques
were thoroughly discussed, covering their implementation, advantages and shortcomings.
The bulk of these surveillance efforts were said to rely substantially on monitoring from strategically
placed watchtowers (usually on elevated ground). However, these towers have a limited range of vigi-
lance. In a given year, the first 30% of towers begin working on the 15th of May with two people on them
(phase Charlie). The next phase, called Bravo, is run from the 1st of July until the end of September, in
which all towers are functional. The most strategic towers perform surveillance 24-7, requiring four peo-
ple per tower to ensure the shifts of 8 hours (and days off). These workers are hired during 3 to 4 months,
meaning that they are not trained specialists. All the aforementioned may contribute to a success rate of
1
fire detection of roughly 10% in disadvantaged areas and 30% to 60% in the best conditions [5], thus
making the towers an expensive, and often inefficient, approach to surveillance. In order to complement
the monitoring by the control towers, GNR patrols are deployed to areas which are out of reach, even
intervening in an initial phase of fire [5]. This type of surveillance also works as a method of deterrence,
which is fundamental since most fires are thought to be man-made. In fact, around 30 to 40% of ignitions
are due to negligence and around 20% are found to be caused intentionally, as presented in Figure 1.2
[6].
Figure 1.2: Causes of ignitions in Portugal between 2006 and 2013 (adopted from [6])
1 Areas with less towers and some obstruction to visibility.
2
Another standard surveillance and fire confirmation procedure is performed using helicopters. In fact,
according to Major Luı́s Lourenço, between 15 of June and 15 of September of 2017 there were 55 daily
helicopter departures, on average. This method is widely used as a confirmation method due to its high
efficacy, although it is very expensive. Additionally, some technological alternatives are also being used
for surveillance. One is called CICLOPE (Sistema Integrado de Videovigilância Florestal, in portuguese),
a system that uses fixed cameras with different spectra (visible 2 , infrared 3
with Light Detection and
4
Ranging (LIDAR)) to detect smoke and fires. This fixed system must be installed at elevated points
in the terrain, often requiring the construction of new roads for the installation and maintenance of the
system. Besides, they require energy coverage and good communication systems, which are not always
available in forest terrains [5]. During these meetings, it was also discovered that some of these entities
already possess highly technological UAVs, which were expensive acquisitions. These UAVs require
trained pilots and have high maintenance costs, so the surveillance/firefighting entities do not use them
often, due to the fear of losing the equipment (and consequently the initial investment).
It is readily evident that new and more efficient solutions are needed to help preventing and min-
imizing the devastating power of forest fires. In order to understand specifically how the surveillance
methodology could be improved, meetings were held with 2nd Commander João Curto from Regimento
de Sapadores Bombeiros - Quartel Santo Amaro (Lisboa) and Major Luı́s Lourenço from GNR. Both
agreed that the most currently used methods are inefficient and that new technologies should be intro-
duced. This led to a discussion of which features were most fundamental to the process of surveillance:
• Reliability of the system, in order to ensure that the developed system is adequate for its surveil-
lance role;
• Maintenance (frequency and cost), which directly affects the time that the system is not available
and the cost of operation/maintenance;
• Human resources, which is a limited and expensive factor in the surveillance methodology;
• Versatility of missions, since it is expected that the developed system could be a powerful aid for
other types of missions besides surveillance making it a more cost-effective investment;
• Overview of the area, since it would permit better route identification in order to combat the fire;
• Precise identification of the fire, as it minimizes false alarms (thus saving resources) while permit-
ting an efficient and fast localization and intervention;
• Performance on rough terrain, as most of the Portuguese forest areas are located in rough terrain
making it crucial that the system can easily overcome this issue;
• Adequate performance during the night, since the most dangerous fires start during that time,
usually intentionally, representing around 30% of the total fires. This is due to the fact that during
the night they go undetected for longer, causing the fire to grow in size;
2 For the detection of smoke columns during the day and flames during the night.
3 For the detection of hot spots. However it only works in line of sight.
4 It calculates the distance until the objects.
3
• Method of deterrence.
During the meetings, an UAV was presented as a possible surveillance tool. This solution was
welcomed by both Major Luı́s Lourenço and 2nd Commander João Curto. The 2nd Commander was
particular enthusiastic about the idea and shared his view of what UAV could assist with: ”fires are the
major form of terrorism in our country! In my opinion 90% of the fires are human-made. Thus, the UAV
would not only be a deterrent but could even allow the identification of the arsonist”. In both meetings,
the current surveillance methods were compared with the capability of a theoretical UAV, allowing for a
productive discussion about the benefits and drawbacks of each and a clear idea of what features the
UAV ought to have in order to be an effective tool of surveillance. With the Major and Commander’s
expertise, it was possible to carry out a decision matrix5 (presented in Table 1.1) as a way to understand
whether the development of an UAV is justified. The ”Total” values below each column result in the
sum of every item’s performance criteria multiplied by the corresponding ”Importance” score. As an
example, in the ”Control Tower” column the ”Performance during the night” score of 4 is multiplied by the
”Importance” score, 9. The resulting 36 is added to the Total value.
Table 1.1: Decision matrix, comparison between different surveillance methods
Importance (1-10) Control tower CICLOPE Patroling UAV
Reliability of the system 9 4 4 5 6
Low maintenance 6 6 6 5 5
Low human resources 5 2 7 3 5
Safe for human life 10 4 7 4 7
Versatility of missions 8 4 4 7 7
Overview of the area 9 5 4 3 7
Method of deterrence 7 4 4 7 6
Precise identification of the fire 9 4 4 6 6
Performance independent of terrain 8 4 4 4 7
Performance during the night 9 4 5 5 7
Total 331 386 393 513
From Table 1.1 it is observable that UAVs could be a very powerful tool for surveillance, particularly
due to its capacity to reach any point independently of the condition of the terrain. To guarantee that
the UAV is useful, it is also fundamental to understand how it would be integrated in the already existing
surveillance and who would be responsible for its missions. According to both the 2nd Commander João
Curto and Major Luı́s Lourenço this project must be integrated in the CDOS 6 . These are distributed
5 The decision matrix combines the importance of each criteria with the performance of the surveillance method. The impor-
tance of each criteria is evaluated between 1 and 10, where 1 is unimportant and 10 is very important. The performance of each
criteria is between 1 and 7, where 1 is very low performance and 7 is high performance.
6 Comando Distrital de Operações de Socorro, in Portuguese, which stands for regional command for health rescue missions
4
across all regions in the country, meaning that each region would have its own fleet of UAVs and would
be responsible for its management and maintenance. Additionally, during the meetings it became clear
that, in order to be effective, the UAV would require a wind resistance around 30 to 40 km/h.
Responding to this necessity, Centre of Engineering and Product Development (CEiiA) is developing
an UAV (or drone), to survey forests and other high-risk areas. This system ought to have the capability
of live streaming and image processing at the ground station7 , which would alert an operator in case
either fire or smoke are detected. Additionally, the system ought to be fully autonomous, meaning
that the operator in the ground station would only have to define the intended area for surveillance to
generate the UAV’s mission. Last but not least, the UAV should be user friendly and be low cost, to
avoid fear of losing the system and subsequent investment. The project was named DeltaSpotter, with
a first working prototype previously developed8 , Figure 1.3. This prototype was assembled as a proof of
concept for flight and software testing, with fairly positive results. Both positive and mixed results of the
testing will be thoroughly explained in Section 3.2. After testing, CEiiA gave the green light to develop
a long endurance version, which should also correct and upgrade the first prototype. In addition to the
development of a long endurance version, special attention will be given to the materials used and the
manufacturing techniques to ensure the cost of production is low.
Figure 1.3: The first prototype: DeltaSpotter V1
1.2 Objectives
The challenge is the development of the DeltaSpotter version 2 aiming to maximize the endurance (over
an hour). According to the Portuguese authorities this would greatly improve the chances of using
disposable UAVs for firefighting surveillance and assistance. During the conceptual design, several con-
strains must be taken into account: the UAV needs to be low cost, easy to carry and must take off by
hand launching, which facilitates its use in routine surveillance procedures. Another aim is to design a
new internal structure, that could be used in both short and long endurance versions of DeltaSpotter. In
an early stage of development, solutions already available in the market will be studied and assessed,
as a way to understand which specifications are desirable in a long endurance UAV and how to possibly
improve the first existing prototype. The conceptual design is expected to follow an iterative process,
in which aerodynamics, flight stability, structures and propulsion will be evaluated to satisfy the require-
ments. Notably, the materials and techniques used to manufacture the UAV are crucial to keep the UAV
structural integrity, inexpensive and easy/quick to assemble. The conceptual validation is expected to
happen by the end of this work with the prototype’s first flight test.
7 Where the ground station is a portable computer.
8 This was developed during a summer internship, by the author of this thesis and the aeronautical team of CEiiA.
5
1.3 Thesis Outline
The work presented in this thesis is divided into five chapters. This first chapter shows the motivation
behind this work and its main goals.
Chapter 2 presents an overall description of the growing applicability of UAVs in worldwide society,
their classifications and configurations with several examples. Moreover, there is a brief introduction of
the materials, manufacturing techniques, and propulsion systems currently used in the development of
UAVs.
In Chapter 3 the airframe configurations, wing size, shape and dimensions are studied and decided.
An introduction of on-board main systems, stability analysis and flight envelop is also included.
Chapter 4 covers the characterization of the manufacturing materials and both CFD and structural
analysis. In addition, a mission outline is shown, with the theoretical endurance value taken into consid-
eration.
Finally, Chapter 5 summarizes the achievements of this work and gives recommendations for future
work.
6
Chapter 2
State of the art

The design process of an UAV depends on its purpose and mission requirements. Since the UAV does
not require a person inside, it opens up new possibilities for the use of the design space and allows
for more freedom regarding the size, construction techniques, materials and systems. This also allows
for the simplification of some components and the removal of systemic redundancies. In addition, the
materials used in manned aircrafts need to have their properties and behavior well described, because
of their function as a method of transportation for human life, making them a very expensive part of the
airplane development. While they require great strength, stiffness, minimal weight and development in
state of the art laboratories, for UAVs the process of selecting materials is far easier. The materials
themselves may be simpler, lighter and cheaper than those typically used in manned vehicles, which
also allows for the test and implementation of new types of materials (i.e. different foams, polymers,
woods) and new types of component manufacturing, like additive manufacturing. It is also possible to
simplify the components of command and control, akin to the ones already used in radio controlled
aircraft models.
The technological development in terms of materials, manufacturing techniques, aerodynamic, telecom-
munications and propulsion lead to a reduction of manufacturing costs without compromising the perfor-
mance of the UAV [7]. This, coupled with the possibility to carry out high-risk missions with lower risk of
life has caused the available number of UAVs to skyrocket.
2.1 Applications of UAVs

The first attempt to create unmanned flying crafts was centuries ago, in China, with hot air balloons
[7]. In the late 1890, Nikola Tesla fomented the idea of a remotely piloted aircraft way before the first
manned flight in 1903 by the Wright brothers. Fifteen years later, in 1918, the first unmanned system was
successfully launched, the Curtis N-9 Aerial Torpedo. During the twentieth century, this kind of system
was extensively developed for military applications, with the goal of reducing human life endangerment
[7].
Nowadays, however, drones are no longer just used for military applications. In fact, the domestic
market has been flooded by small drones often used to capture images of sporting events, concerts and
cityscapes. Nevertheless companies across industries have realized that drones have diverse commer-
7
cial applications, some of which go way beyond photography, videos and life-streaming [7, 8]:
• Agriculture: where drones are used to increase production efficiency. This is done by the analysis
of data collected in the farm terrain (i.e. crop health, stand counts of the crops, water distribution).
• Surveillance and emergency response: usually associated with military or governmental applica-
tions like tactical operations support, search and rescue or communications relay in case of some
disaster, which is crucial due to the potential to save human lives.
• Aerial mapping 2D and 3D: this process generates a map from the aerial imagery. By correcting
and joining these images it is possible to obtain a 2D map. When taking into account the camera
distortion and angle of view it is possible to obtain 3D shape of the terrain. After image processing
with advanced software a 3D model of the terrain is generated.
• Surveying: usually this application is used for topography, measuring the physical or geometric
characteristics of the earth (i.e. geodetic, hydrographic, land, mine, geologic).
• Social media: many photographers, movie makers or reporters use UAVs for their work. This
simplifies the process of obtaining new scenes and gives the opportunity to obtain the big picture
from event.
• Industrial inspection: small UAVs can be used for detailed analysis. It is possible to detect defects
like cracks or bending of the structure (i.e. monitoring of road conditions, inspection of bridges,
wind turbines, lines and structures of electric power industry).
For fire detection systems in particular, the earliest application of an UAV to gather information on
forest fires can be traced back to 1961 by the United States Forest Services Laboratory [9]. Fast forward
to 2006, where NASA’s UAVs “Altair” and “Ikhana” prove their capability of supporting near-real-time
forest fire imaging missions in the western United States [10]. Meanwhile, in Europe, several projects
have validated the use of UAVs as a method to detect forest fires in Spain and in the Netherlands [11].
In addition to the practical testing of UAV systems, there has been a significant amount of simulated
research on fire monitoring and detection abilities, from research of the practicality of using a group
of low-altitude, low-endurance UAVs [12] to the necessary features to turn an UAV into an effective
fire surveillance tool. Nevertheless, it is clear that UAVs are an increasingly realistic option for fire
surveillance, potentially providing rapid, mobile, and low-cost alternatives for monitoring and detecting
forest fires.
2.1.1 UAS Classification

Unmanned Aerial Systems (UAS) present a wide variety of characteristics, in terms of range, flight
altitude, endurance, and Maximum Take-Off Weight (MTOW)/payload, as shown in Figure 2.1. Several
broad categories are used to classify these systems, such as TUAV (Tactical Unmanned Aerial Vehicle),
MUAV (Micro & Mini UAV), VTOL (Vertical Take-off and Landing) UAV, MALE (Medium Altitude Long
Endurance) and HALE (High Altitude Long Endurance) [13]. The first is exclusively used in military
8
applications while the other four categories are mostly for civilian applications. MUAVs are compact
systems, with a MTOW lower than 30kg, that are easy to transport. VTOL systems have vertical take-off
and landing, allowing them to take-off and land anywhere without a runway. Moreover, this type of UAVs
is usually associated with rotary wings, although it could also have fixed wings in some particular cases.
MALE and HALE are UAVs designed with specific missions in mind. Both are heavy, with a MTOW
between 1500-7000kg for MALE and between 4500-15000kg for HALE. This last category, HALE, flies
at high altitude and is characterized by its long endurance (24-48h). The characterization in different
categories is not absolute so they do overlap in some cases [13].
Figure 2.1: Categories of UAVs [13]
2.1.2 UAV Configuration
The configuration of the drone depends on its intended mission. When monitoring a small area, to get
detailed imagery, it is better to use a vertical takeoff and landing aircraft (VTOL) configuration, with the
capacity to hover. These systems are very maneuverable, stable and easy to control. [7]. On the other
hand, if it is necessary to survey or monitor large areas, an UAV with fixed wings (Conventional Take-
Off and Landing, CTOL) would be more effective since these platforms have the advantage of offering
operator’s long endurance and the ability to conduct fights at much higher altitudes. Additionally, they
are also better equipped to carry substantial payloads for longer periods of time as compared to other
configurations [7]. As such, this type of vehicle is generally better suited to perform surveillance of large
areas and will be the configuration chosen to move forward in this aircraft development.
These vehicles are mainly divided in four major configurations [7, 14, 15]:
• Conventional configurations with tailplane-aft, such as Aeromapper EV2 in Figure 2.9;
• Canard configuration with a tailplane forward, such as Litus in Figure 2.3;
• Tail-aft on booms, such as Penguin BE Electric Platform in Figure 2.2;
• Flying wing, such as Freya in Figure 2.4.
These will be discussed and evaluated further along in this thesis.
9
2.1.3 Market Research
As previously stated, the conceptualization of this UAV involves the research and evaluation of solutions
already existent in the market. The aircraft will be developed with a fixed wing configuration, which will
be reflected in the analyzed options. To ensure that this step is useful to the development, two other
characteristics are essential: the capacity to insert surveillance systems in the drone; and medium to
long endurance rates. Subsequently, various drones are shown and analyzed, as seen in Figures 2.2
through 2.9:
Penguin BE Electric Platform

Endurance: 110 min
Wing Span: 3.3 m
Cruise speed: 22 m/s
Empty weight: 14.9 kg
Propulsion Type: electric
Max Payload: 6.6 kg
Take off: Catapult/Runway/car top launch
Figure 2.2: Penguin BE Electric Platform [16]
Price: unknown
This vehicle has a tail-aft with boom configuration. It has the advantage of being able to fly with large
payloads and it also allows the use of inverted V tail avoiding the adverse yaw-roll coupling and reducing
spiral tendencies. However it is very long and heavy [7, 16].
Litus
Endurance: 120 min
Wing Span: 6 m
Empty weight: 50 kg
Max Payload: 15 kg
Take off: Runway
Figure 2.3: Litus [17]
Price: unknown
As the previous UAV, Litus is also designed to carry large payloads. The canard configuration permits
the designer to move the center of gravity relatively far aft1 , which may be beneficial in pusher designs.
Consequentially, the UAV has greater flexibility in payload location [7, 17]
1 Meaning at, near, or towards the stern of a tail of an aircraft.
10
Freya
Endurance: 50 to 100 min
Wing Span: 1.2 m
Max cruise speed: 16 m/s
Weight: 1.3 kg
Take off: Hand launch
Price: $ 20, 269.00 2
Figure 2.4: Freya [18]
Freya has an efficient design, adequate to adverse conditions. It supports temperatures between -20
o
C and +40 o C and withstands winds of up to 13 m/s. This UAV has multispectral sensors: RGB and
thermal. Its main applications include agriculture, mining and forestry monitoring [18].
Aeromapper Talon
Endurance: 90 min
Wing Span: 2 m
Range: N.A.
Weight: 3.5 kg
Figure 2.5: UX5 AG [19]
Price: $9, 980.00
This UAV is the best seller of Aeromao. It is made of EPO foam and reinforced with carbon fiber, so
it is easy to repair. It is fully autonomous and it has a good performance in confined areas (able to climb
at 45 degrees). It can land with a parachute or in auto stabilization mode [19].
DT 18 HD
Endurance: Up to 120 min
Wing Span: 1.8 m
Range: 100 km
Weight: 2 kg
Figure 2.6: DT 18 HD [20]
Take off: Hand launch/catapult
Price: unknown
DT 18 HD is a hand lanched lightweight UAV. Its features include long endurance, long range, and
an integrated global shutter sensor which allows live streaming. This UAV has several applications
2 The original price is 16,350.00 e, corresponding to 20,269.00 $ on 18/04/2018.
11
like large-scale monitoring, corridor mapping, topographic surveys, vegetation monitoring, infrastructure
inspection and anomaly detection [20].
eBee SQ
Endurance: 55 min
Wing Span: 1.1 m
Cruise speed: between 11 m/s and 30 m/s
Range: 41 km
Weight: 1.1 kg
Figure 2.7: eBee SQ [21] Take off: Hand launch

Price: $ 10, 490.00
This UAV was developed for agriculture applications, to analyze the health of the crops. As such, it
has a multispectral and RBG sensor integrated. However, it has a small endurance [21].
AgEagle RX48
Endurance: 45 to 50 min
Wing Span: 1.2 m
Cruise speed: 14.7 m/s
Range: N.A
Weight: 2 kg
Price: $17, 950.00
Figure 2.8: AgEagle RX48 [22]
AgEagle RX48 is designed for agricultural purposes with a good camera resolution and exception-
ally precise sensors. It is very light and small, but the endurance of the system is not adequate for
surveillance use [22].
Aeromapper EV2
Endurance: 60 min
Wing Span: 2 m
Range: 30 km
Weight: 4.5 kg
Price: $ 10, 500.00
Figure 2.9: Aeromapper EV2 [23]

12
Aeromapper EV2 has a good combination between performance, size and weight. It was designed
for medium to large survey projects, with the advantage of being fully autonomous. It has a camera
with 24 megapixels incorporated with optional multispectral sensors. Nonetheless, the landing has to be
done with the aid of a parachute which may require larger areas to land and cause damage to the UAV
(since the landing with a parachute is usually more violent).
The civilian UAVs displayed so far were developed with specific missions/applications in mind, with an
average endurance between 1 to 2 hours and a wingspan below 2 m. Most are autonomous. Some have
the capacity to load payload, which comes with the drawback of making them more expensive. In order
to analyze longer endurance UAVs some military systems were also researched and studied (Appendix
A). Unfortunately, these UAVs have little information available, which complicates their use for market
research. Their missions are also widely different from civilian UAVs (i.e. most of military UAVs are
designed for high altitude flight and to be able to carry large loads). Nevertheless, these alternatives are
important to consider and analyze, as they allow estimates and correlations between (long) endurance,
wingspan and weight. From analyzing the different systems, in Figures A.1 and A.2, it becomes evident
that to increase the endurance it is fundamental to increase the wingspan and or the total weight, usually
by using batteries with more capacity. It is important to consider that in the case of military UAVs the
air-frames are usually made of carbon, glass and/or Kevlar fibers, which makes them light weight and
efficient but quite expensive.
When looking for lower cost models available in the market it is also possible to find non-autonomous
models, which are usually used for hobby activities. These models are usually made of simple and
light materials: polymer-based bodies and foam wings. In addition, since their design was not thought
for specific missions they are not well prepared for payloads and their endurance is quite small. Even
though payloads could be added to many of the models it is likely that the overall performance of the
UAV would suffer, making these systems inadequate for this project. However, it is possible to use some
solutions/ techniques present in these models to lower the cost of an UAV without compromising the
overall performance.
2.2 Materials and Manufacturing Methods

The development of an UAV requires the analysis of different components (i.e. structural, aerodynamics
and propulsion) [24]. A key factor in keeping its cost contained is the choice of materials and manu-
facturing techniques used. Overall, its structure should be lightweight to maximize endurance while stiff
enough to bear the required loads [7].
Historically, in the aeronautical industry, the most common material used was aluminum, mostly
due to its high stiffness, low density and the fact that it is relatively resistant to corrosion. However,
with the development of composite materials, aluminum has slowly been replaced as the industries
preferred choice. In fact, the introduction of Boeing 787 Dreamliner, made of approximately 50% of
composites, proved that by using these materials, one can reduce the aircraft’s weight and maintenance
costs, respectively around 20% of global weight and 30% of maintenance costs [25]. As mentioned
13
in Subsection 2.1.3, the use of materials like carbon, glass or Kevlar fiber in composites is becoming
more common, particularly in military grade drones. Nevertheless, they are far more expensive and not
particularly eco-friendly. Taking all this into account, for this product development, the focus will turn to
cheaper and eco-friendlier alternatives such as polymers, woods and foams.
2.2.1 Wood
Wood is one of the most commonly used structural materials [26]. It is widely abundant, biodegradable,
light and relatively low cost. In applications where low weight is a top priority, balsa is a type of wood
with the best stiffness to weight ratio [27]. Like most woods, the mechanical properties of balsa vary
significantly with density, from 40 to 380 kg/m3 . It also has an axial direction much stiffer and stronger
than the perpendicular direction. On average, the elastic modulus of balsa in tension (axial) test is
1.68 GP a and the ultimate strength is 10.12 M P a3 . In the tangential direction the elastic modulus is
48.27 MPa and the ultimate strength is 0.82 MPa4 [27]. However, balsa requires a lot of maintenance
and is less durable than other materials such as polymers [28].
2.2.2 Foams
Foams are a combination of liquid or solid substances with air bubbles. The most important properties of
foams are the volume fraction of the gas and the bubble size, which defines the density and consequently
their potential applications. Due to their low density, foams have a wide range of applications. They can
be used as thermal insulators and their capacity to absorb blunt impacts makes them ideal for packing
materials [29]. Recently, solid foams have been used as construction materials in UAVs due to their
many advantageous characteristics. They are low cost, easy to shape and repair, while also being
lightweight and having a high impact absorption. The manufacture process when working with foams is
mainly done with a hot wire foam cutter, such as the one shown in Figure 2.10. This cutter consists of
a thin wire in which a certain current is conducted, meaning the wire itself works like a resistance. The
current is chosen in order to make the wire reach temperatures higher than the fusion point of the foam
to be cut. This technique allows the manufacture process to be fast, precise and increasingly cheap.
Figure 2.10: 4 axis hot wire cutting the foam [30]
3 The tests were done with specimens of Balsa with density of 90.98kg/m3 and 5.5 ± 0.20 mm in thickness, 25.40 ± 0 mm in
length and 12.60 ± 0.25 mm in width.

4 The specimens had 4.4 ± 0.1 mm in thickness, 36.00 ± 0.5 mm in length and 18.20 ± 0.25 mm in width.
14
The most common foams are Expanded polystyrene (EPS), Extruded polystyrene (XPS) and Ex-
panded polypropylene (EPP), which main features are given in Table 2.1.
Table 2.1: Characteristics of each foam [31–33]
EPS XPS EPP

Densities kg/m3 12-48 27-45 20-200
Qualities Lightest Highest compression strength High impact resistance
2.2.3 Polymers
Polymers are large molecules composed of many repeated sub-units. They may occur naturally, but
most are formulated by chemists for a specific purpose. Due to their broad range of properties, which
are defined by their long molecular chain, these materials have an important role in everyday life. Some
useful properties are higher strength to weight ratios (light weight but comparatively stiff and strong),
toughness, resilience, resistance to corrosion, lack of conductivity (thermal and electrical), color, trans-
parency and processing. They are easy to soften or melt when heated, and become firm when cooled,
making them ideal for heat dependent manufacturing techniques. In general, polymers have low prices
due to being chemically mass produced from petroleum. Despite this, there have been significant invest-
ments in producing polymers from renewable sources (i.e. biofermentation of corn starch) to decrease
the consumption and dependence of fossil fuels. The most common used polymer filaments include
Acrylonitrile Butadiene Styrene (ABS), Polylactide (PLA), Nylon, Polycarbonate (PC) and Polyvinyl Al-
cohol (PVA) [34].
There are several ways to classify polymers. One such way is by their molecular structure, which
gives two types of subsets, thermoplastics and thermosets. While thermoplastics have only secondary
bonds between chains, thermosets also have primary bonds between them.
Thermoplastics polymers are divided in amorphous (with no regular molecular structure) and semi-
crystalline, which are often more denser. These polymers are usually associated with properties of
hardness, friction and wear, less creep or time dependent behavior and corrosion resistance. Some
examples of thermoplastic polymers are [34]:
Semi-crystalline: Amorphous:
• Polypropylene (PP) • Polyvinyl Chloride (PVC)
• Acetals • Polystyrene (PS)
• Polyesters • Polycarbonate (PC)
• Polytetrafluoroethylene (PTFE) • Polymethyl Methacrylate (PMMA)
• Polyetheretherketone (PEEK) • Acrylonitrile-butadiene-styrene (ABS)
• Polyethersulphone
15
Additionally there are polymers that can be both amorphous and semi-crystalline depending of the
required properties, i.e. Polyamide (Nylon) and Polylactic acid (PLA).
Thermosetting Polymers are generally used when features like high thermal and dimensional stability
are required. These are usually applied for adhesives, electrical and thermal insulation materials, high
performance composites and especially where high strength and modulus are required. Some examples
of thermosets are aminos, polyurethanes, epoxides, phenolic and polyesters.
Additive Manufacturing
Additive Manufacturing (AM) refers to a process by which digital 3D design data is used to build up a
component in layers by depositing material. It builds up components layer by layer using materials, such
as metals, plastics and composite materials [35, 36]. This kind of manufacturing allows for the production
of objects with great detail that otherwise would be complex to manufacture (i.e. hollow structures).
In contrast to traditional manufacturing, AM does not require molds and high production rates, which
enables the manufacturing of low quantities or even just sporadic prototypes. In a way, AM redefines
value at scale, by allowing the production of various customized items per printer [37]. Recently, GE
has 3D printed a jet nozzle, which was installed in an engine. The simplified jet nozzle was five times
more durable, and 25% lighter than conventionally manufactured parts [38]. Following GE’s lead, several
aerospace/aeronautical companies are now heavily investing in AM. Not only in aerospace but in diverse
areas, such as medicine 3D printing is making its mark, as it is now used to produce biodegradable
airway splints. These are transplanted in babies with severe cases of tracheobronchomalacia providing
support and protection for airway growth [39]. As seen from previous examples AM provides:
• Fast prototyping;
• Production of complex geometries, without complicated assemblies;
• Enhanced performance (i.e. providing similar strength at lighter weights);
• Production of low number of items, opening the road to personalized manufacturing.
Depending on the applications and requirements needed, different AM technologies can be chosen.
Each technique is described in detail in Table 2.2.
When considering 3D printing, one also has to examine the implications of such decision. The rate
of production is considered to be around 0.01 to 1 kg/h, which is minuscule when compared to injection
mold. In addition, an increased resolution in the printer causes it to be slower and far more expensive.
Lower resolutions require improved post processing and may be associated with lower quality in the
features of the piece. In addition, depending on the material used, AM might have a cost of 0.1 to 10$
per gram to print, which is a much higher scale than other manufacturing processes. One must also take
into account that the additional post processing, to perfect the pieces, may be a significant chunk of the
cost of the whole process.
16
Table 2.2: Description of AM technologies
Technology Technique Typical materials Advantages Disadvantages
Exposure from top (SLA) Acrylates/epoxides Excellent surface quality Limited mechanical properties
and precision
Vat Photopolymerization CLIP Acrylates High build speed Low-viscosity resins required
Exposure from bottom Acrylates/epoxides Low initial vat volume, bet- Limited mechanical properties
ter surface quality
Multiphoton lithography Acrylates Very high resolution Low build speed, lim-
ited materials
Powder Bed Fusion Polymer SLS PA12, PEEK Best mechanical proper- Rough surfaces, poor
ties, less anisotropy reusability of unsintered
powder
Polyjet Acrylates Fast, allows multimaterial AM Low viscosity ink required
Material and Binder Jetting Aerosol jet printing Conductive High resolution, low temp Low viscosity ink required
inks/dielectrics process
3D printing (binder jetting) Starch, PLA, ceramics Fast, allows multimaterial Limited strength of
17
AM, low temperature parts, rough surfaces
Sheet Lamination Laminated object man- PVC, paper Compact desktop 3D printer Limited materials, low res-
ufacturing olution, high anisotropy
FDM ABS, PLA, PC, HIPS Inexpensive machines Rough surfaces, high tem-
Material Extrusion and materials perature process
3D dispensing Thermo-plastics, com- Broad range of materials Rough surfaces; narrow
posites, photoresins, viscosity process window
hydrogels, biomaterials
When considering the production rate of this UAV, at an initial stage, one could speculate that a few
units would be bought per month, making AM particularly interesting for a multifaceted company such
as CEiiA.
2.3 Propulsive System

All aircrafts require powerful, lightweight and efficient propulsion systems (excluding gliders). These
systems provide the necessary power to propel the aircraft forward, (horizontally and/or vertically). The
power from the propulsion system must be enough to overcome the drag force and allow the aircraft to
operate at the desired velocity. Moreover it must exceed the drag and the weight of the UAV for it to climb
[40]. The propulsive system works with the transformation/conversion of an energy source (i.e. gasoline,
diesel fuel, liquid hydrogen, solar energy) into heat/pressure or electric current. Afterwards, the energy
is converted into motion (shaft rotation) which will be used to generate movement in a vehicle. This
could be done by a propeller (converts motion into airflow) or a turbine (converts high pressure into high
velocity exhaust) [41]. There are several types of propulsion, but only two are commonly used in UAVs,
the Internal Combustion (IC) engines and Electric Propulsion [7].
2.3.1 IC Engines
Figure 2.11: IC engine - 3W-28i CS [42]
For small and medium-size UAVs the IC engines (liquid-fueled) are either four-stroke or two-stroke and
use gasoline, methanol or diesel as fuel [40]. This system converts the potential energy stored in the fuel
into shaft motion. It is one of the most common propulsive systems used in UAVs and it is usually used
in long-endurance aircrafts [40]. One such example is presented in Figure 2.11. Its main merits are its
power and efficiency per weight and complexity. However, small engines have numerous deficiencies:
they are noisy; polluting; with a high thermal signature; and produce a lot of vibration, which requires
reinforcements in the structure [43]. Plus, it requires cooling systems, which adds weight to the engine.
Due to its complexity, meaning a bunch of moving pieces, the engine also requires frequent maintenance
[41].
18
2.3.2 Electric Propulsion
This system creates a rotational motion from electric power. In order to produce thrust, it is coupled with
the propeller. The rotational speed of an electric motor (Figure 2.12) is proportional to the voltage applied
to it, and consequently the torque is proportional to the current [41]. Recently, with the improvements of
Lithium-Polymer batteries and development of lightweight electric motors (brushless) the use of electric
propulsion from storage batteries became feasible. In fact, the most common power source for electric
motors is the battery, although they may be supplemented by solar-powered photovoltaic cells in the
near future. This type of engine minimizes the unfortunate event of crashing due to motor failure or
shutdown [24]. Currently only micro and mini UAVs (with a total weight of less than 30 kg) are powered
by batteries and electric motors due to an endurance limitation. However, as battery design and capacity
improves, the electric propulsion is expected to become more useful in longer endurance systems [40].
Taking into account that an electric motor practically does not have any movable parts when compared
to an internal combustion engine, it allows an increase of the engine revolutions. Overall it also requires
lower maintenance. Moreover, an electric system’s thrust can be controlled more precisely and responds
faster to throttle input due to the high torque available. In addition, it is known to be quiet, without gas
emissions, with low vibrations and does not require a cooling system, which contributes to a lower weight
and reduction of overall complexity [41]. While the system has many advantages, it does require large
currents, creates electromagnetic interference, and has a very limited endurance, (due to the batteries’
current capacity). Furthermore, the recharge rate is usually low and when using a fast recharge system,
there might be overheating associated.
Figure 2.12: Electric motor - Hacker A50-12S V4 [40]
19
20
Chapter 3
Preliminary Design
3.1 Design Process
In order to conceptualize an unmanned aerial vehicle, an iterative process is defined, which is presented
in Figure 3.1.
The starting point of this iterative process was the prototype DeltaSpotter V1, which was developed
in cooperation with the aeronautical team at CEiiA during a Summer Internship Program. The author of
this thesis accompanied the entire development process and participated in the decision making.
A new conceptual design was implemented to improve the first prototype and better adjust to meet
market requirements. After of this stage, in decision point D1, the results were analyzed in light of the
requirements: if considered accurate enough, the process proceeds to the aerodynamic and structural
design, otherwise it returns to the requirements/constraints step.
A comprehensive design of the structure was carried out alongside the aerodynamic analyses. After
completion, the resulting structural design performance was evaluated (decision point D2): if judged
accurate, the fabrication step initiates, otherwise the structure and geometry designs are revised (to
improve aerodynamics).
After the DeltaSpotter V2 prototype is built, flight tests are performed.
21
Figure 3.1: General Design Process
3.2 First Prototype: DeltaSpotter V1
As mentioned in Chapter 1, the first working prototype built was the DeltaSpotter V1. It was designed
as a proof of concept, in order to understand if it is feasible to develop a low-cost, fully autonomous
platform to use as a fire detection system. Accordingly, the system was developed with only open source
software (Mission Planner) and hardware (Autopilot 2.8). The main goal for this stage was to understand
the features required and to test the performance of Autopilot and Mission Planner. In addition, during
this stage, there was an effort to analyze the market and present the project to possible customers. CFD
and structural analyses were not performed, as this phase was still a proof of concept. The developed
platform had the following specifications:
• Wingspan: 2 m
• Endurance: 1 h
22
• Propulsion: electric
• Cruise speed: 12 m/s
• Weight: 1.5 kg
• Takeoff: hand launch
• Manufacture cost: around 3001 e
The main data related with DeltaSpotter V1 is presented in Tables 3.1 and 3.2.
Table 3.1: Data for cruise speed of 12 m/s. Table 3.2: Data related with battery and weight.
Data Value Data Value

CL 0.28 Vbattery [V ] 11.1
CL /CD 17.3 Qbattery [Ah] 3.3
CD 0.016 W [N ] 14.7
CDO 0.0122
S [m2 ] 0.7
AR 5.7
In order to have a low cost, lightweight UAV, the wings were made using XPS foam, which has a high
capacity to absorb impact, while the spar and central frame were built out of plywood.
After the manufacturing process was complete, flying tests were performed. The system was tested
in both radio control flight and autonomous flight. The autonomous flight was programmed using Mission
Planner [44] where the overfly area was defined and the waypoints were generated. Both tests were
successfully completed, which led to the conclusion that both hardware and navigating software were
adequate for the UAVs function. However, some issues were reported, such as problems with the
autopilot leveling which was thought to be caused by the central frame (due to the material used being
plywood) and a limited endurance time. With the first prototype finished and tested, the aim was now
to develop an improved long endurance version. Additionally, to fix the issue reported above, a new
internal structure should be developed that could be used in both short and long endurance versions of
the aircraft, per request of the company.
To further improve the new DeltaSpotter prototype, DeltaSpotter V1 characteristics were compared
to the characteristics of commercially available aircraft mentioned on section 2.1.3.
1 This price only takes into account the construction materials.
23
Figure 3.2: Weight vs endurance.
Figure 3.2 displays the relation between weight and endurance. The trend line shows that the en-
durance stays approximately constant with increased weight. This is most likely due to careful design,
as the manufacturers of each drone carefully optimized the endurance to be adequate for a respective
weight in order to maximize efficiency. When looking to the highest contributors of the total weight of
the drone, there are two main culprits: the wingspan and the batteries. For simplicity sake, the pay-
loads added to the drone will be excluded. Logically, to improve the endurance, batteries with a larger
charge or larger (ideally lighter) wingspans must be used, which will impact the total weight. Therefore,
a balance between these characteristics must be achieved to avoid compromising the UAVs efficiency.
Figure 3.3: Weight vs wingspan.
Figure 3.3 shows the relation between aircraft weight and wingspan. As expected, when the wingspan
increases, the weight will increase as well. Taking this into account DeltaSpotter’s weight could be further
decreased in order to match other commercially available UAVs.
24
Figure 3.4: Weight vs average cruise speed.
Figure 3.4 shows an average cruise speed between 15m/s and 20m/s for the commercially avail-
able UAVs Theoretically, the speed of DeltaSpotter should be around 10m/s [45], to guarantee a long
endurance flight. However, in order to survey a larger area during smaller amounts of time, it would
be ideal for the new prototype to have a cruise speed that at least matched the commercially available
drones. As a compromise to keep a long endurance and an improved speed the new prototype should
have a cruising speed of 15m/s.
Figure 3.5: Endurance vs price.
Figure 3.5 presents the relation between endurance and price. As the endurance of the aircraft
grows, the tendency is for the price to sharply increase. DeltaSpotter is located below the line of ten-
dency, which means the prototype has a similar endurance to other commercially available aircraft while
having a much lower cost. It is important to refer that the presented price is related with the produc-
tion cost, materials and components, and does not consider the manufacturing labor. Therefore, the
performed comparison is not completely realistic, due to the fact that the other models have a selling
price.
Nevertheless, DeltaSpotter V1 may be considered to have a good endurance for its price range,
thus accomplishing one of its imposed aims. This seems to be mostly due to its relatively low weight.
Additionally, it is shown that, by changing either the batteries (the component with a larger impact on
the weight) or the wingspan, there will be a large impact in the performance of the aircraft. Therefore,
by analyzing this initial prototype, it becomes clear which parameters require further development and
25
which are good baselines to import into a second prototype. In particular the number of batteries, the
size and weight of the wingspan and the aerodynamic performance may be optimized.
3.3 Requirements and Constraints

After acknowledging and analyzing the triumphs and shortcomings of the first prototype, and taking into
consideration the feedback from future operators, the aim is to develop a new UAV based on Deltaspotter
V1 but with a longer endurance, which translates into a longer flight time without recharging batteries.
The requirements for the 2nd version are similar to the requirements for the 1st version: it must takeoff
by hand launching, be low cost, easy to assemble and transport. For transportation and according to
the input from the Portuguese entities the wingspan should not exceed 3 m. Regarding payload, the V2
must be prepared to carry a minimum weight equal to the one used in V1 (around 250g) with the ability
to increase this value to 500g. Another imposed requirement (due to the type of missions that it will
perform), is high gust resistance with a minimum of 30/40 km/h 2 . Finally, the last prerequisite is related
with the internal structure as it should be adequate to use in both V1 and V2.
3.4 Mission Profile

To determine the flight conditions, it is fundamental to first define the mission profile, which in this case
starts with the takeoff followed by climb until it reaches cruise altitude. Cruise is the longest phase, in
which the UAV will survey a previously defined area. As cruise ends, the UAV starts to descend until it
lands, as is shown in Figure 3.6 [46].
Figure 3.6: Mission profile (adapted from [47]).
3.5 Airframe Configuration

As mentioned in Subsection 2.1.2, within the field of fixed wings there are four options: canard, tail on
fuselage (conventional configuration), tail on booms, and flying wing. Each configuration is analyzed
and compared in Table 3.3. Subsequently, it is possible to recognize that flying wing is the most ap-
propriate configuration for this application. This specific configuration is simpler in terms of the number
2 Corresponding to 8.3 and 11.1 m/s.
26
of components, manufacture process and set up. It also has a solid and strong build, which reduces
the production cost and the need for maintenance due to harsh landings. Moreover, it is easier to carry
(tailless), which is convenient for transportation [7].
Table 3.3: Decision matrix of the UAV configuration,
Importance Tail-aft on fuselage Tail-aft

Canard Flying-Wing
(1-10) (conventional configuration) on booms
Long endurance
10 6 7 7 6
(efficiency)
Low manufacture
8 5 5 5 7
price
Easy to setting up 8 5 5 5 7
Low maintenance 7 5 5 6 7
Easy to carry 6 4 4 4 6
Maneuverability 5 7 6 6 5
Total 234 239 246 282
Table 3.3 combines the importance of each criteria with the performance for each configuration.
The importance of each criteria is evaluated between 1 and 10, where 1 is unimportant and 10 is very
important. The performance of each configuration is evaluated between 1 and 7, where 1 is very low
performance and 7 is high performance. As in Table 1.1, the ”Total” values below each column result of
the sum of every item’s performance criteria multiplied by the corresponding importance score.
3.6 Propulsion System

Another feature to be defined is the type of propulsion, which will be chosen according to the system with
lower maintenance, purchase price and weight, without compromising efficiency and reliability. When
deciding between the IC and the electric propulsion, one important factor is that the internal combus-
tion engine not only requires a fuel deposit but also requires a battery for the electric system, overall
increasing the number of components and consequently the weight, as shown in Figures 3.7(a) and
3.7(b).
(a) Electric system (b) IC engine system
Figure 3.7: Propulsive systems [48]
27
The capacity of flight for longer periods of time, due to the high energy density of fossil fuels, is the
main advantage of an IC engine [40]. However, these fuels are highly flammable. Since one of the
applications of the UAV is forest surveillance, the aspect of carrying fuel could be dangerous in case
of system failure. In this case, carrying rechargeable batteries is preferable to inflammable fuels as it
reduces the chances of causing a fire. From an economical point of view, the electric propulsion is less
expensive. Moreover, these propulsion systems are smooth, powerful, and highly reliable devices since
they involve so few moving parts [40]. According to all this, the electric propulsion system seems to be
more advantageous and will be used in DeltaSpotter V2.
Now it is necessary to define the equations for a propelled aircraft. The operating power of the aircraft
corresponds to the product of the thrust with the flying velocity, P = T U . For the cruise, the required
thrust is equal to the drag value, while for climb the thrust is determined using a constant rate of climb
(dh/dt), Equation 3.1 [49]:
dh T −D
= Uclimb sin(ϕ) = Uclimb (3.1)
dt W
where T is the thrust, W weight and ϕ is the climb angle. The required thrust for climbing is calculated
as Tclimb = W sin(ϕ) + D. Moreover the appropriate velocity for climb (Uclimb ) can be calculated as a
function of the stall speed (Us ), Equation 3.2:
s
W 2
Uclimb = 1.2 Us = 1.2 (3.2)
S ρ CLmax
where CLmax is the maximum 3D lift coefficient for the aircraft, W is the weight, S is the wing area and
ρ is the air density. Thus the required net power for climb (Pclimbnet ) is given by Equation 3.3:
s
1 W 2
Pclimbnet = Tclimb Uclimb = (W sin(ϕ) + CD ρ U 2 S) 1.2 (3.3)
2 S ρ CLmax
Since the required power calculated above is net (Pclimbnet ), the efficiencies of the motor (ηmotor ) and
propeller (ηpropeller ) need to be taken into consideration in order to calculate the raw power, Equation
3.4. This value must be contemplated when choosing a propulsive system [49, 50].
Pnet
Praw = (3.4)
ηmotor ηpropeller
28
3.7 Conceptual Design of DeltaSpotter V2
Figure 3.8: Conceptual Design Sequence.
This section is related with the conceptual design of the platform which is an iterative process com-
posed by 5 steps (initial weight estimation, airfoil selection, wing configuration, stability analysis and
flight envelope) and 2 decision points (D3 and D4) as illustrated in Figure 3.8. D3 comes after wing
configuration, by defining the size and shape of the aircraft with a large influence on weight. If the final
weight estimation is different from the initial, the selection of the airfoil and the wing configuration must
be revised.
D4 follows the stability analysis, where there are several conditions which have to be met in order
to have a platform with good response to perturbations. If these conditions are not verified, the wing
configuration is redesigned (i.e. the vertical stabilizer’s size and shape are changed).
Since this product is developed for surveillance, whenever it would be necessary, agents would
launch the UAV by hand (without any additional tool i.e. catapult or runway). Afterwards, they would
keep it in the car and move it to other locations. This naturally constrains the drone’s wingspan and
weight, as it needs to be light and small enough to be hand launched. During meetings with governmen-
tal agencies sizes between 2 to 3 m were discussed. Increasing the wingspan and the aspect ratio (AR)
reduces the induced drag which causes an increase in endurance. Thus, the wing size was defined as
3 m of wingspan. This decision will be further studied in Section 3.7.3 were its impact will be thoroughly
examined. Another essential characteristic is for the UAV to be inexpensive, while still useful and resis-
tant. With that in mind, the wing will be made of XPS foam, as it is easy to shape, highly resistance to
impact and easy to maintain (or replace). The manufacturing process of the wing will consist in a 4 axis
29
hot wire cut technique, which is relatively cheap to implement and does not require molds (the same
machine can be used to produce different shapes or components). This technique fits with the wing size
constraint mentioned above as it is sufficient to build it in one setting, thus avoiding reinforcements. This
is essential to keep the weight down and maximize efficiency.
Additionally, the central frame and internal structure which incorporates all of the electrical compo-
nents must be easy to change as a function of the payload (as the payload may differ from mission to
mission). Thus, the technique most adequate for these parts will be additive manufacturing. This method
is widely used in the construction of UAVs [51], as it is fast and ideal for small production rates. It also
allows building structures with virtually any shape and with precise dimensions, which is fundamental for
the autopilot, as it must be lined up with the x-axis of the UAV in order to get correct flight data.
CEiiA, a company with various projects in diverse areas already owns a Fuse Deposition Modeling
(FDM) printer, which is adequate to print the desired pieces using polymers. In doing so, it avoids new
investments, which ultimately keeps the cost of production down. However, the maximum size of the
components is limited to the size of the build platform (280 x 270 x 180 mm) which in this case it should
not be of particular concern. This FDM printer, Leapfrog Creatr HS, is able to print using a broad range of
polymers including ABS, PLA, Nylon and others [52]. Ultimately, PLA, classified as a thermoplastic, was
chosen as the manufacturing material, for its high tensile strength. Plus, it is biodegradable, recyclable
and sourced from renewable sources, lowering the UAV’s carbon foot print [53].
3.7.1 Initial Sizing
The first step is to apply sizing methodology, which consists in getting an estimation of initial weight, as
a way to estimate the size of the platform [47].
WT O = Wairf rame + Wpayload + Wengine + Wbattery. (3.5)
The wing loading, and its fractions, from each of the components, are obtained by dividing all signifi-
cant weights by the wing area (S) in equation 3.5:
WT O Wairf rame Wpayload Wengine Wbattery

= + + + (3.6)
S S S S S
Table 3.4 presents the weight estimation of each component and the detailed description is presented
in Appendix A.1.
30
Table 3.4: Weight of the components
Weight [g]
Wairf rame 1100
Wpayload 250-380
Wengine 140
Wbatteries 1400
WT O 2790-3000
So the initial take-off weight estimation WT O is around 30 N . The weight estimation of engine and
payload was based on the first prototype, since the payload will be similar in both versions adding 50%
of the margin for V2. The engine should be identical in terms of weight.
As expected, the weight of the battery represents a significant part of the total weight. For a long
endurance UAVs, it could reach values of 50% to 60% of total weight [45, 54].
The weight of the airframe was roughly estimated, based on the weight of V1 multiplied by a factor
of 1.5. The total weight of the electric components was done in Table A.2 (Appendix A.1).
3.7.2 Endurance
The ultimate goal of this project is for the UAV to be able to cover large areas in one flight. As such,
the endurance was of special concern. Endurance may be defined as the maximum time that an aircraft
can fly, under a certain flight condition, without being refueled. To understand the theoretical limits
of endurance, the Breguet method was used (Equation 3.7). This is for combustion propeller-driven
airplanes and it is expressed in the imperial unit system [49].
3
! " #
Wi Wi
C L /2
Z Z
dt ηP dW ηP p 1 1
TE = dW = −550 = 550 2ρS p −√ (3.7)
Wf dW cP Wf DW cP CD Wf Wi
where TE is the endurance in hours, ηP is the propulsive efficiency, cP is the specific fuel consumption,
CL is the lift coefficient, CD is the drag coefficient, ρ is the air density, Wi is the initial aircraft weight, Wf
is the final aircraft weight and S is the wing area.
When analyzing the equation, it is clear that, in order to have higher endurance, the aerodynamic
3
CL /2
parameter CD should be maximized [49, 55]. However, since the propulsion of the UAV will be done
with an electric motor, the endurance equation will consequently be different (as the mass of the UAV
will be constant during the flight). Thus, the endurance equation will be the same as the one in an
electric airplane and will be deviated by the total energy stored on board of the airplane (ET ) to the rate
of change of energy with time ( dE
dt ) [56].
ET ηP
TE = dE
(3.8)
dt
31
The rate of the change of energy with time is directly related to the aerodynamics of the airplane.
Knowing that:
ET ηP
TE = (3.9)
P
P = DU (3.10)
1 2
D= ρU SCD (3.11)
2
ηP = ηmotor ηpropeller (3.12)
where U is the speed of the UAV, CDo is the zero lift drag, AR is the aspect ratio, e is the osvald factor
and W is the weight of the UAV.
1
CD = CDo + C2 (3.13)
πARe L
L
CL = 1 2
(3.14)
2 ρU S
Applying Equations 3.9-3.14 the endurance equation for the electric airplane during cruise is ob-
tained3 :
ET ηP
TE = 1 1 W
(3.15)
2 ρ U 3 S(CDo + π AR 2
e ( 1 ρ U2 S ) )
2
From this equation it is seen that in order to obtain the maximum endurance, the total energy stored
aboard the UAV (ET ) should be maximized. Additionally, the weight of the UAV and the drag coefficient
should also be minimized [56, 57].
The total energy storage is calculated by multiplying the amount of charge stored [Ampere second]
with the voltage of the battery [Volt], Equation 3.16.
ET = Vbattery Qbattery (3.16)
The efficiency of the propeller for electric aircraft is around 80% and for an electric motor around 90%
[58]. Thus, the efficiency of the propulsion system is 72%.
The endurance will be obtained using the Equation 3.15 and the data of the aircraft obtained in the
following chapters. This value considers that the whole flight is done at the same altitude, at cruise
speed, without considering the takeoff, climb, descent and landing. It also considers that the workman-
ship of the UAV and weather conditions are perfect and that all the energy stored is consumed, which
does not happen since it is necessary to keep some charge to avoid damaging the battery. As such, this
obtained value is an ideal number, that serves as an estimation.
3 The aerodynamic coefficients, wing area and the aspect ratio will be analyzed in the next chapter.
32
3.7.3 Aerodynamics
In order to do an aerodynamic analysis, an open source program named XFLR5 was used. Developed
by Mark Drela, this software uses a two-dimensional panel method code, with the integral boundary layer
theory for analysis of airfoils in viscous or inviscid flow fields [59]. XFLR5 is widely used as an analysis
tool for airfoils, wings and planes at low Reynolds Number. In the case of airfoils, the software uses a 2D
direct analysis and for 3D wing analyses it uses Lifting Line Theory (LLT), Vortex Lattice Method (VLM)
and 3D Panel Method [60]. Since the UAV has a low aspect ratio, with a sweep/dihedral that cannot be
neglected, a LLT analysis could not be used. Additionally, a 3D panel method could be useful if there
was an interest in the Cp distribution or if the wing had a body, which is not the case. Consequently, the
Vortex Lattice Method, VLM, was the method selected. This type of analysis can be used for any wing
geometry (i.e. low aspect ratio, sweep, including winglets). VLM has the limitation of assuming small
angles of attack which is not an issue as this UAV will fly in cruise speed with a low angle of attack [60].
Airfoil
Once the configuration of the UAV is defined, the next stage consists in choosing the airfoil, which
corresponds to the shape of the wing. A more efficient airfoil has a higher Cl /CD ratio.
Considering the average speed of the UAV (15 m/s) and that the air conditions are standard sea level
the selected airfoils will be studied with a low Reynolds Numbers (5, 000<Re<1, 000, 000), calculated by
the Equation 3.17 [61],
ρ U x,
Re = (3.17)
µ
where ρ is the air density, U is the average speed of the fluid with respect to the wing, x is a characteristic
linear dimension and µ is the dynamic viscosity of the fluid.
Since the most important feature of an airfoil in a flying wing is a moment coefficient close to zero [62],
only some airfoils may be used. Based on published data, various airfoils were selected and analyzed
using the software XFLR5 (XFOIL feature) to determine the most efficient one [62–64]. They are:
• S 5010 (airfoil used at the DeltaSpotter V1) • HS 520
• S 5020 • HS 522
• MH 45 • EH 3.0/12
• MH 60 • EPPLER 339
• MH 83 • LA2573A
Since the construction of the UAV will involve foam with hot wire cutting, the trailing edge should
have 5 mm of thickness to guarantee structural stiffness. This modification in the airfoil will affect the
33
aerodynamic properties. As for the relative thickness of the airfoil, it should be at least 14% of the chord.
This is due to the space required for all electronic components in the center of the wing and it is a widely
used requirement when designing an airfoil, as seen in [64].
With these constraints in mind the values for the lift and drag coefficients were obtained using
XFOIL4 , as seen in Figures 3.9 and 3.10. It is important to refer that all airfoils, although designated
by their original name, were altered to accommodate the constraints described in the previous para-
graph. Figure 3.9 shows that the chosen airfoil depends on the Cl required. Moreover, in order to have
an efficient aircraft, the chosen airfoil should have low drag. This reduces the required power to maintain
the aircraft in flight. Accordingly, Figure 3.10 shows the drag coefficient as a function of the angle of
attack. Since the UAV is designed to fly in cruise conditions, an estimation5 of the lift coefficient for
cruise speed is obtained, using Equation 3.14. The speed values used for the calculation were obtained
through the research of similar UAVs in Section 3, (Figure 3.4) and the other parameters related with
size were based on the first prototype by applying a multiplicative factor.
It is important to refer that the airfoil lift coefficient Cl is not equal to the aircraft lift coefficient CL .
However, as an approximation these will have the same value, with a Cl around 0.45 for a speed of
10m/s and 0.11 for 20m/s. As such, Figure 3.9 should be analyzed for values of Cl between 0.1 and
0.6.
(a) Overview of Cl over Cd (b) Detailed analysis in the range values of Cl = 0.1 to 0.6
Figure 3.9: Cl over Cd for Reynolds number of 430000.
4 This analysis was done for a constant Reynolds number at cruise speed, for standard sea level (SSL) and at mean aerody-
namic chord (MAC).

5 This estimation was made with approximated values of wing area, S, and U
cruise (between 10 and 20 m/s). The wing area
value was obtained through the wingspan (3m), the root chord (500mm) and with the estimation of the tip chord (200mm).
34
(a) Overview of Cd as a function of angle of attack (α) [deg] (b) Detailed analysis of Cd for small angles of attack
Figure 3.10: Cd as a function of angle of attack (α) [deg] for Reynolds number of 430000.
Figure 3.9 and 3.10 show results for a Reynolds of 430000. In order to verify if the results change
for lower or higher Reynolds number, several analyses were performed (Reynolds of 285000, 350000,
500000 and 570000). The obtained curves are similar for the following five simulations, which the airfoils
with the best results for the Cl range of values and the minimum drag coefficient 6 , (MOD means that
the airfoil was modified):
• HS 522 - MOD
• MH 45 - MOD
• MH 60 - MOD
• MH 61 - MOD
• S5010 - MOD
In addition, the selected airfoils were studied in terms of stall behavior, which is a function of the
angle of attack, rather than airspeed. This analysis was done for a low Reynolds number, Re=185000,
which is the most critical part (near of stall speed). Figure 3.11 presents the stall behavior as a function
of angle of attack.
6 In this case the best results are considered to be the airfoils with the lowest drag for the range of values of C and for different
l
values of angle of attack.
35
Figure 3.11: Lift coefficient as a function of the angle of attack for 185000 Reynolds number.
The analysis indicates that, for most cases, the stall occurs smoothly. S5010-MOD is the airfoil
with the smoothest positive stall behavior; however, it does not have the same performance in the
negative part. The airfoils MH60-MOD and MH45-MOD have similar behavior, however MH60-MOD
has a slightly smoother performance, since the abrupt stall occurs later, at around 18o , while for MH45-
MOD this behavior happens at 17o . The last two airfoils (MH 61-MOD and S5010-MDO) show the worst
performance in terms of positive angle of attack. Therefore, the airfoils that show better behavior in
terms of stall are MH45-MOD and MH60-MOD.
Wing Configuration
In order to obtain the airfoil with the highest efficiency when incorporated into the wing platform, five
equal 3D models of the wing with different airfoils were created in XFLR5. Apart from changes in the
airfoil, the wing dimensions were maintained for all the simulations (Table 3.5).
One of the project constraints is that the internal structure should be able to be used in both a normal
and long endurance version. In order to do so, the root chord and thickness of the wing will be considered
equal to those used in DeltaSpotter V1.
In order to complete the shape of the wing, characteristics like taper ratio, aspect ratio, dihedral and
sweep angles must also be defined. This may be done by considering aerodynamic aspects, which
translates into all of these degrees of freedom directly affecting the performance of the UAV.
To minimize the amount of lift induced drag, the elliptical wing would be an ideal platform [49]. How-
ever, it is structurally difficult to produce. A good compromise may be the trapezoidal wing as it the
closest to the ideal elliptical platform. The trapezoidal construction is simple to manufacture and has a
relatively lower root bending moment. For this type of wing, the taper ratio should be 40% (λ = 0.4)
[49, 65]. Another parameter essential to define is the sweep angle (Λ). The increment of this angle
contributes for the directional stability [24]. However, a large value of the sweep angle reduces the wing
36
efficiency (CL /CD ). Based on Ref. [66] an appropriate value for the sweep angle for tailless UAV is 15o .
Tailless vehicles often exhibit poor lateral dynamic stability. In order to improve this parameter, sweep
angle and taper ratio were analyzed. Another characteristic that improves lateral dynamic stability,
specifically roll stability, is the dihedral angle [67]. For this prototype, the selected angle was 3o . Apart
from this feature the dihedral angle also has a role during landing, as it causes the wing tip to stay above
ground. This allows the UAV to perform softer landings.
The next key attribute to consider is the aspect ratio. A high aspect ratio is globally more aerody-
namically efficient due to the reduction of induced drag [40]. This parameter cannot be overly high since
the greater the aspect ratio, the bigger the bending moment will be (which would require a heavier struc-
tural reinforcement). Consequently, in order to have a good aerodynamic performance and avoid high
structural reinforcements, a sensible range to consider for the aspect ratio is between 6 and 9, where
the aspect ratio commonly used in flying wings is 9 [40, 68]. In order to obtain the exact aspect ratio,
two parameters are required, the wingspan and the mean chord or the wing area. Due to the project
constrains explained in Section 3.3, the wingspan may have a maximum of 3 meters. The value of the
mean chord is 0.35 m, thus with Equation 3.18 it is feasible to calculate the aspect ratio.
b2 b
AR = = (3.18)
S cmean
The value of the aspect ratio (AR) is 8.57, which is very close to the recommended aspect ratio
(AR = 9 corresponding to b = 3.15 m). Therefore, from an aerodynamic point of view, the constraint
of the wingspan b = 3 m does not adversely affect the performance. Table 3.5 summarizes the data
of the new platform DeltaSpotter V2, with the parameters previously defined and the estimated weight
discussed in Subsection 3.7.1.
Table 3.5: Data of platform V2

Data Value
Wingspan (b) [mm] 3000
Root chord (cr ) [mm] 500
Tip chord (ct ) [mm] 200
Sweep angle (Λ) [o ] 15
o
Dihedral angle [ ] 3
Weight [N] 30
With all parameters defined, the schematics of the wing and respective dimensions are presented in
Figure 3.12. Additionally, using this configuration, five different airfoils were modeled and compared with
a sixth model, with the configurations of DeltaSpotter V1. This analysis is shown in Figure 3.13.
37
Figure 3.12: Representation of the wing dimensions.
(a) CL /CD vs. U x. (b) CL vs. Alpha.
(c) Cm vs. U x. (d) Alpha vs. U x.
Figure 3.13: Comparison of different airfoils in the platforms.
38
Figure 3.13(a) represents the CL /CD as a function of U x (The speed is in km/h). Figure 3.13(b)
provides the lift coefficient CL in relation to the angle of attack (Alpha). Figure 3.13(c) represents the
moment coefficient (Cm) as a function of speed (U x). Figure 3.13(d) represents the angle of attack as
a function of speed (U x).
The higher the value of CL /CD the better the aerodynamic efficiency will be [69]. The lift coefficient
increases with the angle of attack with a linear growth. From Figure 3.13(b) the lift coefficient during the
cruise is obtained, coresponding to when the angle of attack is equal to zero (considering the weight of
the aircraft equal to the previous estimation).
Figure 3.13(c) shows the moment coefficient as a function of speed. For cruise speed the curve
should cross the horizontal axis, meaning the moment coefficient is zero. This can be achieved in two
ways, the first is related with the airfoil reflex which in Figure 3.13(c) has already been changed7 . The
changes in reflex camber are performed in the last 20% of the airfoil’s chord, increasing the angle of
reflex [70]. The second way is to do some small adjustments to the center of gravity CG. Since the
curve moves with the modification of the center of gravity, moving the CG backwards will move the
curve up in the vertical axis. For the initial estimation, the CG was calculated with the ”Flying wing CG
calculator” (277 mm), using a static margin of 0.1 which will be further analyzed in detail in Subsection
3.7.6 [71].
Hereafter, the value of the center of gravity value will always be adjusted using this figure.
Figure 3.13(d) is used to obtain and adjust the incidence angle. When adjusting the incidence angle,
the curve should intersect the horizontal axis in the cruise speed value. In Figure 3.13(d) the incidence
angle has already been adjusted for cruise speed8 .
With the analysis complete it is now necessary to choose one of the five airfoil options. This will be
the airfoil with the best performance of CL /CD , for a cruise speed of around 15 m/s. Consequently, the
best airfoil is clearly the MH60-MOD, (the root airfoil) presented in Figure 3.14. Globally, it also has an
impressive performance for various speeds, making it a more adaptable choice.
Figure 3.14: Wing root airfoil for DeltaSpotter V2, MH60-MOD, t/c=14%
After choosing the airfoil it is also necessary to study the tip thickness to chord ratio in order to
complete the shape of the model. The simplest option is to maintain the same value, which is favorable
from a manufacturing viewpoint. Since the material used is a solid foam, a thicker airfoil would allow for a
7 The airfoils HS522-MOD, MH45-MOD, MH60-MOD had an increase of reflex camber of 2o , MH61-MOD had 1o and S5010-
MOD 3o .
8 The incidence angle used for the models is between 2.3o and 2.6o depending of the airfoil. In case of the model with airfoil
MH60-MOD the value of incidence angle is 2.5o .
39
stiffer structure, avoiding reinforcements in this region. From an aerodynamic point of view, the thickness
to chord ratio which allows for a higher CL /CD will be 12%, as shown in Figure 3.15. This may be seen
in Figure B.3 (presented in Appendix B) where five similar models with equal configuration/airfoils and
different relative tip thickness were compared.9
Figure 3.15: Wing tip airfoil for DeltaSpotter V2, MH60-MOD, t/c=12%
The next step is the improvement of the platform, which relies on the adjustment of the center of
gravity (CG) and subsequent stability analysis. As mentioned before, in order to obtain a leveled flight,
the moment coefficient should be equal to zero. Since the moment coefficient is dependent on the center
of gravity, this parameter should be adjusted. Figure 3.13(c) shows the center of gravity adjusted to the
cruise speed, with a value about CG = 280 mm. This is a reference value obtained from XFLR5, as the
final CG distance will be only obtained after the design of the internal structure.
Figure 3.16 displays the differences between the previous DeltaSpotter prototype and the new ver-
sion, V2. From Figure 3.16(a) it is noticeable a 27.5% CL /CDmax improvement in comparison to the first
version. Both prototypes were developed to fly with α = 0 during cruise speed (10 m/s for V1 and 15m/s
for V2). It is also worth referring that DeltaSpotter V1 has an incidence angle of 4.5o while the second
prototype has an incidence angle of 2.5o .
Figure 3.16(b) shows the evolution of the lift coefficient as a function of angle of the attack. The
second prototype shows a lower lift coefficient for α = 0 due to increased cruise speed and wing area.
The Figure 3.16(c) evidences a reduction of 30% of CD for the second version for each respective cruise
condition.
9 t/c = 10, t/c = 11, t/c = 12, t/c = 13 and t/c = 14.
40
(a) CL /CD as a function of Ux [km/h] (b) CL as a function of angle of attack (α) [deg]
(c) CD as a function of angle of attack (α) [deg]
Figure 3.16: Differences between DeltaSpotter V1 and V2 .
The aerodynamic analyses were made with the reference values shown in Table 3.6.
Table 3.6: Reference values used in aerodynamic analysis.
Value
Air density (ρ) [km/m3 ] 1.225
Dynamic Viscosity (µ) [m2 /s] 1.5 · 10−5
Cruise speed (Ucruise ) [m/s] 15
Wing area (S) [m2 ] 1.05
The aerodynamic performance parameters of the second prototype are presented on Table 3.7.
41
Table 3.7: Performance parameters of DeltaSpotter V2 from XFLR5.
Data Value
Maximum lift coefficient CLmax 1.08
Maximum lift angle of attack [o ] αCLmax 10.5
Maximum lift-to-drag ratio (CL /CD )max 23.7
3/2 3/2
Maximum lift -to-drag ratio (CL /CD )cruise 18.6
Lift VS angle of attack curve slope [deg −1 ] CLα 0.078
Zero lift drag coefficient for cruise CDOcruise 0.0097
Drag coefficient for cruise CDcruise 0.011
Lift coefficient for cruise CLcruise 0.20
Lift-to-drag ratio for cruise (CL /CD )cruise 17.82
3/2
Lift3/2 -to-drag ratio for cruise (CL /CD )cruise 7.9
The performance parameters from Table 3.7 and the Equation 3.19 were used to calculate the stall
speed (Ustall ), with an obtained value of 6.5 m/s for V2.
s
2W
Ustall = (3.19)
S ρ CLmax
3.7.4 Electric Components
Autopilot + GPS + Radio Wireless Telemetry + Power module
The system autopilot APM 2.8 allows the performance of autonomous missions by using the Mission
Planner software. In the software, the mission is defined and several waypoints are generated. The
autopilot controls the flight and ensures that the UAV passes over the waypoints. In order to perform
autonomous flight, the autopilot must have a GPS, compass, barometer and inertial measurement unit
(IMU). Additionally, the autopilot could have other sensors like a sonar (in order to perform higher pre-
cision landings), or a pressure sensor (which is useful for slow flight and landing), Figure 3.17 presents
the kit of autopilot with the rest of the communication system.
42
Figure 3.17: Kit Autopilot + GPS + Radio Wireless Telemetry + Power module [72]
Kit Autopilot example: [72]

Price: 70 e
Battery
The battery is usually one of the heaviest components in electric UAVs. It provides energy for all electric
systems, ensuring that the motor has power during all flight phases. The batteries are classified by
their respective specific energy. Instinctively, we would choose the battery with higher specific energy.
However, one needs to also take into account the price, modes of failure and life cycles. Currently the
batteries that are most used in UAVs and radio controller aircraft models are called LiPo, Figure 3.18.
These batteries have the benefit of being lightweight, low cost and be made in almost any size and
shape [56]. Voltage (V) and the amount of charge (Q) are the parameters that should be studied in
order to guarantee that the energy storage is enough for the entire flight. According to this, Table 3.8
presents several batteries with the respective parameters and ratios between amount of charge/weight
and amount of charge/price. Most of the presented batteries are Z.F. (ZIPPY Flightmax) and Z.C. (ZIPPY
Compact) due to their low price. Therefore, it is required to choose a battery preferably with both highest
ratios.
Table 3.8: Battery features

Capacity Voltage Amount of Weight Electric Price Electric
[Ah] [V] charge [Ws] [g] power/weight [Ws/g] [e] power/price [W/e]
Z.F. 4S 5 14.8 266400 460 579.13 26.51 10049
Turnigy 6S 5 22.2 399600 765 522.35 56.26 7103
Z.F. 5S 8 18.5 532800 915 582.30 75.95 7015
Z.C. 6S 4.5 22.2 359640 696 516.72 59.37 6058
Z.F. 5S 5 18.5 333000 550 605.45 40.07 8311
Z.C. 6S 5 22.2 399600 645 619.53 57.36 6967
Z.F. 4S 8 14.8 426240 845 504.43 57.66 7392
Z.F. 6S 5 22.2 399600 730 547.40 56.19 7112
43
Comparing the battery models, the best ones are the first (Z.F. 4S) and the fifth (Z.F. 5S). Both show
the highest rate of electric power per euro. The ratio of electric power over weight is also high. A good
compromise between capacity, weight and price is the Z.F. 4S with a capacity of 5Ah will be used, Figure
3.18.
Figure 3.18: Battery ZIPPY Flightmax 4S 5Ah [73]
Cost: 26.51 e (15-08-2018)
Capacity: 5 Ah
Voltage 18.8 V
Discharge: 20 C
Battery example: [73]
Control Actuators
The actuators are responsible for the movement of the flight control surface. This type of actuators are
used in ailerons, flaps, elevons and rudders. Since this UAV has a flying wing configuration, it is required
two control actuators for the elevons. One of the main features of each control actuator is the stall torque,
which should be chosen in function of the length and chord of flight control surfaces, speed of the aircraft
and the maximum deflection. In order to size this component, the PredimRC Exel
c
software is used.
Developed by Franck Aguerre [74], this software allows to estimate the maximum required torque. A
typical value for elevon chord (celevon ) is about 0.2 of the wing chord, meaning that celevon = 70 mm10
[75]. The length of each elevon is 1200 mm, with a gap of 150 mm between the elevon and tip/root. The
maximum torque is required for high speeds. For cruise speed using a maximum deflection of 15o the
maximum required torque is 2.97 kg.cm as presented in the Figure C.1, in Appendix C. According to this,
the actuator used will be the HS-425BB, presented in Figure 3.19, with the following data sheet:
Voltage Range: 4.8V-6.0V
Weight: 45g
Stall Torque (4.8V): 3.3kg.cm
Stall Torque (6.0V): 4.1kg.cm
10 The value is calculated considering the mean chord.
44
Figure 3.19: Servo [76]
Actuator example: [76]

Cost: 13 e (04-07-2018)
Electric Motor + Propeller
In agreement with the performance analysis of climb and cruise, Section 3.6, the electric motor and
respective propeller were selected. Figure 3.20(a) shows a Gt2820/07 motor and a propeller 12x6 (12 in
of diameter and 6 in of pitch). Gt2820/07 is a light brushless motor, that does not require a cooling
system. The propeller is light, strong and with a high performance due to its ultra-thin and stiff carbon
built. Using Equations 3.4 and 3.3 it is concluded that this motor+propeller has sufficient power for
DeltaSpotter V2.
Electric Motor: Propeller:
RPM/V: 850KV Weight: 26g
Max thrust: 2300g (using a 12x6 propeller) Cost: 4.62 e (15-07-2018)
Max RPM: 8700 Propeller example: [78]
Input Voltage: 11.1-14.8 V (3S/4S)
Max current: 41 A
Weight: 140g
Cost: 28.28 e (15-07-2018)
Motor example: [77]
45
(a) Electric motor [77] (b) Propeller 12x6 [78]
Figure 3.20: Propulsion system
Electronic Speed Controller (ESC)
Considering the motor specifications, the 50A ESC would be suitable. However, considering a safety
margin, the 60A ESC was selected, Figure 3.21.
ESC Max. Current: 60A
ESC weight: 61 g
Figure 3.21: ESC 60A [79]
Cost: 18.04 e
ESC example: [79]
3.7.5 Circuit Overview
Figure 3.22 presents the circuit scheme of the avionic components, with the typical layout of the circuit
commonly used in these types of vehicles.
46
Figure 3.22: Circuit scheme
3.7.6 Stability Analysis
The stability is a property of an equilibrium state. As such, for an aircraft to keep in a steady uniform flight
the resultant forces and moments in relation to the center of gravity must be zero. When this condition
is satisfied, the aircraft is in a state of equilibrium, which may be called trimmed flight. Stability can then
be considered as the capacity of the aircraft to keep the trimmed flight condition and restore it when it
is affected by perturbations like force or moment [40, 67]. The stability of the aircraft can be divided in
static and dynamic. These analyses are made using a tool of XFLR5 software.
Static Stability
The static stability has three different categories: statically stable, unstable and neutral stable, which
corresponds to the different reactions of the aircraft when it suffers a perturbation (force or moment or
both). The aircraft is statically stable when it tends to restore its original position (orientation and speed).
It is statically unstable when it takes the vehicle away from the original position and it is neutral if the
aircraft does not have any tendencies. In order to evaluate the static stability, the longitudinal and lateral
derivatives are presented in Appendix D.1 (Table D.1 and D.2) [67, 80]. These derivatives are related to
the cruise condition, meaning that it is for cruise speed and 0o of angle of attack.
Longitudinal Motion
The static stability of the longitudinal motion is analyzed by the longitudinal properties of the aircraft.
The explanation of each stability derivatives can be found in the references [67, 81].
In order to have a static stability of the longitudinal motion some conditions are required:
• For longitudinal static stability (pitching moment) the derivative Mw < 0.
• For forward speed stability the derivative Xu < 0.
• And for vertical speed stability the derivative Zw < 0.
47
Lateral Motion
For the lateral motion, lateral properties of the aircraft are evaluated. In order to have static stability of
the lateral motion some conditions must be satisfied [67]:
• For sideslip stability (restoring force that opposes the sideslip motion) the derivative Yβ < 0.
• For directional stability (opposition to the yaw moment) the derivative Nβ > 0.
• For lateral stability (maintain wings in level equilibrium) the derivative Lβ < 0.
Combining the data from Tables D.1, D.2 with the conditions for static stability, the resume Table 3.9
is obtained [67], where it was concluded that all conditions are satisfied and that the UAV is longitudinally
and laterally statically stable.
Table 3.9: Static stability analysis
Requirement Derivative Result

Mw < 0 -0.81078 Stable
Xu < 0 -0.18573 Stable
Zw < 0 -51.175 Stable
Yβ < 0 -1.6524 Stable
Nβ < 0 0.35057 Stable
Lβ < 0 -2.9225 Stable
Dynamic Stability Analysis
The dynamic stability is based on the time history of the aircraft motion after some perturbation. An
aircraft can be statically stable but dynamically unstable. Thus it is important to do a brief analysis of
dynamic stability. It is considered that the aircraft is dynamically stable whenever the amplitude of any
oscillatory motion (induced by some perturbation) decreases to zero relatively to the steady state flight
condition. Thus, dynamic stability is the reduction of the disturbance in time, which means that there is
energy dissipation, called positive damping (works against perturbation).
In order to study dynamic stability, it is necessary to analyze the differential equations of the aircraft
motion. These differential equations can be decomposed into longitudinal and lateral directions, when
considering small perturbations. Here the equations will not be presented (a detailed description may
be found in [67, 82]). Nevertheless, the eigenvalues, the respective natural frequency and the damping
ratio are shown in Table 3.10. From the eigenvalues, it is concluded that excluding the spiral mode, all
others are stable. This is unsurprising as the spiral mode is frequently unstable. Nonetheless, it is slow
and characterized by T2 (the time that is required to duplicate the amplitude).
Comparing the values from Table 3.10 with the flying qualities required by MIL-F-8785C11 it was
concluded that the flying qualities are clearly adequate for the mission’s flight phase12 . Additionally,
11 The data that was used is designated for piloted airplanes.
12 The flight phases that was analyzed is non-terminal flight phases that are normally accomplished using gradual maneuvers
(i.e. climb, cruise, loiter and descent).
48
there are several differences between lightweight piloted airplanes and UAVs. The absence of pilots or
passengers on board removes some of the constraints which means that for UAVs these flight qualities
are more permissive.
A brief comparison was also made between DeltaSpotter V113 (data of the first version is presented in
Appendix D.2, Table D.3) and DeltaSpotter V2, comparing both the stability of flying modes [67, 82, 83].
From that analysis (Appendix D.2) it is observable that all flying modes have better damping ratios and
T2 in the second version of DeltaSpotter. The Dutch roll has a very similar value of damping ratio, since
it is particularly influenced by the size and shape of the vertical stabilizer.
Table 3.10: Longitudinal and lateral analysis of DeltaSpotter V2.
Eigenvalue Damping ratio ζ Natural Frequency ωn [Hz] T2 [s]

Longitudinal modes
Phugoid -0.0295±0.5723i 0.051 0.091
Short-period -23.439±16.913i 0.811 4.600
Lateral modes
Dutch roll -0.3046±3.2923i 0.092 0.526
Roll -37.081 0.027
Spiral 0.0384 18.053
For DeltaSpotter V2, the vertical stabilizer was designed to be as small as possible14 and at the
same time allow the stabilization of the wing. With a careful tip and taper design, it is frequently possible
to reduce the induced drag of the wing by 20%. The design and the refinement of the shape of the
vertical stabilizer is commonly done with Computational Fluid Dynamics (CFD) and the wind tunnel [40].
In order to save time and resources the design of the vertical stabilizer was based on the first version,
although the size was increased (in order to improve a response in the time domain). Figure 3.23
presents the time response of the Dutch roll for the perturbation of 1km/h15 (in a perpendicular direction
of a movement for the UAV in the Y axis). In this case, the roll and yaw rates are presented in order to
compare with the first and second version.
13 This UAV was developed, tested and approved at CEiiA so if the second version has similar or better dynamic response, it will
be considered dynamically stable.

14 The size of the vertical stabilizer influences the drag of the UAV.
15 Corresponding to 0.28m/s.
49
(a) Roll rate (b) Yaw rate
Figure 3.23: Modal response of the Dutch roll of both versions, red line represents the first version and
blue the second.
Another comparison made is between both versions as a function of the settling time (Ts ) is the time
required for the response curve to achieve and stay inside the 5% of the initial value of the perturbation,
Equation 3.20.
−ln(0.05)
Ts = (3.20)
ωn ∗ ζ
Figure 3.24: Modal response of the Dutch roll for both versions, where the red line represents the V1
and the blue line the V2.
Figure 3.24 shows that both versions have almost the same modal response for the same perturba-
tion (1km/h in the Y axis). The settling time is around 10 seconds, which means that after this period
50
of time the perturbance corresponds to 5% of the initial value. Both versions have a good opposition to
the perturbance in regard to the time domain. Therefore, it is conceivable to assume that the vertical
stabilizers are well sized.
General Characteristics
With all aerodynamic analyzes and optimization complete the necessary data for the production of
DeltaSpotter V2 is now summarized in Tables 3.7 and 3.11.
Table 3.11: Data related with battery and weight
Data Value
Vbattery [V] 14.8
Qbattery [Ah] 5 (x3)
W [N] 30
Using the data from the tables above and the Equations 3.16 and 3.15, the theoretical estimation of
endurance is around 6 hours and 20 minutes. Nevertheless, it is fundamental to keep in mind that the
presented value does not take into account the atmospheric conditions, the expended energy for takeoff
and the energy requirements necessary for the UAV to land with some energy left.
3.7.7 Flight Envelope
In order to define the configuration of the UAV and the reinforcements it should have, all loads during
the operation life would ideally need to be well described. Since this is not feasible, only the most critical
cases are analyzed, i.e. aerodynamic limit given by the maximum lift coefficient CLmax , the aeroelastic
limit given by the maximum speed Umax and the structural limit given by the maximum load factor nmax
[50, 84]. In order to define the maximum load factors that can be expected for an UAV, STANAG 4703
norm [85] is used. This is destined for certification of fixed-wing light UAS’ with a maximum take-off
weight not greater than 150 kg and with a lower limit of 66 N m of impact energy.
According to the STANAG 4703 norm, for symmetric limit maneuvering, the load factor should be
greater than 3.8 for positives loads and less than -1.5 for negative loads [85].
With the maximum and minimum loads defined, it is now possible to define the aerodynamic and
aeroelastic limits. The aerodynamic limit is related with the flight altitude with respect to stall conditions,
and it obeys the Equation 3.21. The aeroelastic limit is related with the restriction on the maximum
speed and consequently the dynamic pressure. The maximum dynamic pressure is produced in a dive,
with the default dive speed given by UD = 1.5 · Ucruise .
1
ρ U 2 CLmax S
n=±2 (3.21)
W
Considering these three limitations a V-n diagram is obtained, as seen in Figure 3.25. The diagram
shows maximum expected loads that the aircraft will experience during a maneuver. The data related
51
with the V-n diagram is presented in Table 3.12 [50, 84].
Figure 3.25: Maneuver V-n diagram of the DeltaSpotter V2 for standard sea level.
Table 3.12: Speed specifications of the V-n diagram
Value [m/s]
Us 6.5
UA 12.7
Ucruise 15
UD 22.5
Taking into account that the atmosphere is considered a dynamic system, gusts are to be expected
during flight. When they occur, the angle of attack changes, which means that the lift force is affected
and the load factor as well. To simplify, the gusts will be considered symmetrical and vertical in order to
obtain the increment of the gust load (Equation 3.22).
ρ u U CLα
∆n = ± (3.22)
2W/S
The peak load factor is the mean load factor at cruise speed (which is n=1) and the increment due to
the gust.
npeak = n ± ∆n (3.23)
Gusts occur in a large range of frequencies. Therefore, the effect on the UAV depends of factors that
affect the frequency response, as is the case with the equivalent mass ratio, µ, Equation 3.24.
2W/S
µ= (3.24)
ρ g c CLα
The mass ratio is a parameter in a response coefficient, K, defined in Equation 3.25 (subsonic flight).
0.88µ
K= (3.25)
5.3 + µ
52
The normal component of the gust velocity (u) could be calculated as the product of overall gusts (û)
with the response coefficient, Equation 3.26
u = K û. (3.26)
Typically, the gust velocity û is obtained from statistical flight data, considering the altitude range
and the flight conditions. According to STANAG 4703 norm, between sea level and 6096 m, the positive
and negative gust values to consider should be 15.2 m/s at cruise speed and 7.6 m/s at maximum
operational speed. This should be no more than 90% of dive speed, thus VM O = 20.25 m/s [85]. These
requirements are more limiting than the ones imposed in the Section 3.3. Using the Equations 3.22 to
3.26 and the gust data, the load factors for both velocities are calculated, resulting in n = +8.1 and
n− = −6.1 for cruise speed and n+ = +5.8 and n− = −3.8 for maximum operating speed.
The gust values given by STANAG 4703 norm, are highly conservative. However, due to the fact
that this UAV is designed for high risk missions, this vehicle should be prepared for adverse weather
conditions.
The gust envelope is drawn and combined with the flight envelop, as seen in Figure 3.26 [49, 50].
Figure 3.26: Combination of maneuver V-n diagram with the load factors induced by the gusts.
Figure 3.26 shows that for high values of gust, the gust envelop has a higher load factor than the
flight envelop. In this scenario, the aircraft stalls. However, this will not occur, since the autopilot will
avoid stall conditions by changing the angle of attack and the velocity.
This type of figure is important to determine the conditions that produce the highest load factors,
which are presented in Table 3.13. The largest values will be used in the structural design.
53
Table 3.13: Maximum load factors for each velocity.
Velocity [m/s] n+ n-
US = 6.5 +3.1 -2.1
Ucruise = 15 +8.1 -6.1
UM O = 20.25 +5.8 -3.8
UD = 22.5 +3.8 0
54
Chapter 4
Structural Design and Analysis
Figure 4.1: Flowchart with the various steps of structural design and aerodynamic analysis performed
during wing design.
As shown in the flowchart above (Figure 4.1) the first step of structural design is the characterization
of the materials and the creation of a 3D Computer Aided Design (CAD) model, using the software
SolidWorks c
2016. In this stage, the wing and its internal structure are modelated. With the CAD
model finished, it is then imported to ANSYS
R
, where it will be used for Computational Fluid Dynamics
(CFD) and Structural analyses. The CFD analysis is performed first, using ANSYS Fluent
R
, in order
to obtain the loads required, for Structural analysis. ANSYS has a feature within Workbench that allows
the coupling of both systems, CFD and Structural, as illustrated in Figure 4.2. Thus, the aerodynamic
forces obtained in CFD are transferred for structural analysis, allowing to have a wing load close to the
real value. Finally, a comparison is made between the computational and analytical values, which allows
further verification of any mistake in the used methods.
55
Figure 4.2: Scheme of the coupling between Fluent and Structural analysis.
4.1 Geometry and Materials
This section has the main goal of characterizing the materials used in the aircraft, and in the design of
the internal structure.
4.1.1 Material Analysis
The manufacturing material has a large influence in the structural analysis and in the wing reinforce-
ments, ensuring that the UAV can withstand the load factors while keeping it lightweight. As mentioned
in Section 3.7, the wing will be made of XPS foam. This foam lacks a universal data sheet due to batch
variability, which means that there are some technical characteristics missing. Therefore, in order to ob-
tain the values of the Young’s Modulus and the Tensile Yield Stress, a cantilever beam bending test was
performed. In order to calculate the density, the beam (705 × 100 × 38 mm3 ) was weighed in a precision
balance1 three times with an average value of 91.88 g.
ρXP S = mass/volume (4.1)
Using the Equation 4.1 and calculating the beam volume, the XPS density, ρXP S = 34.296 kg/m3 ,
is obtained. A beam of XPS, with constant cross section, was used for the experiment. Firstly, the
beam was carefully clamped at one end and the other end was left free to apply normal loading. As
recommended, the clamped hight was around twice the cantilever beam width (Figure 4.3). [86, 87].
1 The Ohaus Pioneer balance was used for this measurement, with a precision of 0.01 g.
56
Figure 4.3: Cantilever beam bending test.
During this test, calibrated loads were always applied at the same application point. After each
increment of load, the deflection in the free end was registered2 in order to obtain the curve of load
over deflection. The process was repeated three times, the average of the obtained values3 is shown in
Figure 4.4.
Figure 4.4: Load over deflection, where the blue points represent elastic mode, and after 100 mm of
deflection, the beam entered in the plastic mode (orange points).
According to the coefficient of determination (R2 ), it may be concluded that the linear regression is
suitable for the test, confirming that, until 100 mm of deflection, the beam was in elastic deformation.
After this value, it is observable that the growth is no longer linear (entrees the plastic deformation stage).
With the obtained values, it is now possible to calculate the Young’s Modulus, E [88].
w i3
I= (4.2)
12
2 The measurement was done with a ruler with a precision of 0.5mm.

3 The obtained values did not fluctuate more than 10%.
57
Q l3
y= (4.3)
3E I
where, I is the moment of inertia, w is the width, i is the height, Q is the applied load, l is the length of
the unclamped beam, and E is the Young’s Modulus [88].
By combining the obtained results with Equations 4.2 and 4.3, the value of Young’s Modulus is
obtained, E = 16.958 M P a. In order to define the maximum allowed stress before the material begins
to deform plastically (yield stress σyield ), the Equation 4.4 is used [88].
M 6 l Fyield
σyield = 1 = (4.4)
6 wi2 w i2
where, M is the moment applied, and Fyield is the maximum force that is possible to apply to the beam
before it starts to deform permanently (yield). Using Equation 4.4 and the data obtained experimentally,
the Yield Tensile Stress is calculated, σyield = 0.361M P a.
Most FDM printed parts are not printed solid, since that would require large amounts of material and
a longer printing time, resulting in higher costs. In fact, most parts are printed with solid shells and filled
with infill. In order to print a geometry, several parameters related to PLA need to be adjusted, such as:
the infill density, extrusion temperature, raster angle, and layer thickness. In order to do so published
data will be used [53, 89]. These parameters were studied in respect to their infill densities by Fernandes
[53].
Table 4.1: Summary of relevant parameters for 3D printing.
Infill density [%] 20 40 60

o
Extrusion temperature [ C] 200 220 220
Raster angle [o ] 0/90 0/90 0/90
Layer thickness [mm] 0.1 0.2 0.1
Young’s Modulus [GP a] 1.04 1.19 1.32
Yield Tensile Strength [M P a] 17.27 20.07 21.31
Three different infill densities, 20%, 40% and 60% were characterized in regard to two extrusion
temperatures, 200o C and 220o C, two raster angles 0/90o and −45/+45o , and two layer thickness 0.1 mm
and 0.2 mm. Table 4.1 shows the best combination of parameters to maximize Yield Tensile Strength
for each infill density studied, all of the tests were made using a linear infill. By analyzing the table, it is
apparent that between an infill density of 40% and 60%, the variation of Yield Tensile Strength is low.
Moreover the 40% of infill density uses the layer thickness of 0.2 mm instead of 0.1 mm, which reduces
manufacturing time. This makes it the most efficient choice for the internal structure as it reduces weight,
manufacturing time while maintaining a high yield resistance.
58
Table 4.2: Resume of the material proprieties.
XPS PLA
Young’s Modulus [GP a] 0.01663 1.19
Yield Tensile Strength [M P a] 0.361 20.07
The data obtained in this section will contribute to more accurate calculations regarding the structural
analysis. Table 4.2 presents the resume of the proprieties of XPS and PLA.
4.1.2 Geometry Modeling
This section describes the internal structure of the UAV, which is the physical support for the motor and
all of the electronic components. In addition, because of its configuration (flying wing) it should also
facilitate the UAV launch (by using the central frame). As was mentioned in Section 3.7, this will be
manufactured with a FDM 3D printer using PLA.
Figure 4.5 shows the isometric view of the designed structure. This includes the main components:
batteries (yellow), control actuators (pink), autopilot (bright green) and motor (black).
Figure 4.5: Isometric view of the internal structure.
This structure was designed to consider the dimensions of the electronic components, although some
simplifications were made in order to facilitate the printing process and reduce the supports4 required.
For example, the structure of the control actuators (number 1 in Figure 4.5) has that shape to allow its
printing without physical supports. The structure is divided in two parts, both manufactured separately:
the box, and the central frame, with a thickness of 10 mm. The base of the box is flat, and the first
to be manufactured. It serves as a support for the remaining walls, and shapes like number 2, which
are used to fix the batteries. The base, walls and the supports for the battery all have a thickness of
4mm. All structure edges have a fillet of 1mm. Number 3 and 4 are supports for the autopilot, and
4 These supports are used to print overhanging parts.
59
the motor. These were designed in order to be aligned with the flight direction (taking into account the
angle of incidence). The camera should be installed at the bottom of the structure. Detailed figures are
presented in Appendix E (Figures E.1 and E.2), which show the structures and their main dimensions.
The central frame helps unite the two parts of the wing, while also facilitating the launch and landing.
Figure 4.6 shows a model of the full wing, with the vertical stabilizers, internal structure, and the main
electric components.
Figure 4.6: Isometric view of the wing.
4.2 CFD Analysis
The main goal of this section is to obtain the aerodynamic loads applied to the wing through a CFD
analysis, to then use them for structural analysis. Furthermore, this analysis allows a comparison be-
tween the results of lift and drag obtained from XFLR5. The software used for the CFD analysis is
ANSYS Fluent R
18.2 student version, with a limitation of 5.12 · 105 nodes/cell [90]. As such, to reduce
the computational time, only half of the wing is analyzed, thus taking advantage of the wing symmetry.
To further simplify it, the vertical stabilizers will be excluded from the analysis. To delimit the shape of
the interface between object and fluid, a CAD model of the wing is used. The next step is to model the
control volume (fluid domain), which should be large enough to minimize the numerical error, due to the
boundary conditions in the proximity of the geometry. The domain and its dimensions are presented in
Figure 4.7, where c is the mean aerodynamic chord [91].
60
Figure 4.7: Representation of the dimensions of the fluid domain.
The flow domain depends on the type of flow. In this case, the flow is subsonic, meaning that the
domain needs to be further out since the waves travel in all directions [92, 93].
After characterizing the domain model, the mesh generation of the flow ensues. The generation of the
mesh depends essentially on the geometry with three categories of mesh: structured, block-structured,
and unstructured, Figure 4.8.
(a) Structured mesh. (b) Block-structured mesh. (c) Unstructured mesh.
Figure 4.8: Examples of three different types of meshes [94].
Structured and block-structured meshes are usually applied in simple geometries. Unstructured
meshes are very common due to their flexibility, as they can fit an arbitrary solution domain boundary.
An unstructured grid is made of triangles or quadrilaterals in 2D, whereas tetrahedra or hexahedra in
3D.
In order to solve the analysis, it is crucial to choose a turbulence model, which will predict its evo-
lution. The velocity and pressure of a fluid flow is governed by Navier-Stokes equations, where the
instantaneous quantity is decomposed into a mean part and a fluctuating part. Due to the fact that there
are no exact solutions for these equations, the Reynolds-Averaged Navier-Stokes (RANS) equations
are used, giving approximate time-averaged solutions to the Navier-Stokes equations. Depending on
61
the time related behavior of the flow, there are two possible approaches, the first one being for statisti-
cally steady flows, and assuming that every variable can be written as the sum of the time mean value,
and a fluctuation about that value [95]. However, when the flow has unsteady behavior, time averag-
ing cannot be used, and the ensemble averaging approach has to be used. This process is known as
Reynolds averaging when applied to the Navier-Stokes equations. In order to close the equation system,
a turbulent model is required. There are several turbulent models available. ANSYS, in particular, uses:
Spalart-Allmaras Model, Standard k–, RNG k–, Realizable (RKE) k–, Standard k–w, Shear Stress
Transport (SST) k–w, and Reynolds Stress. The choice of the model depends on the application, level
of accuracy required, available computational resources, and the amount of time available for the simu-
lation. For this simulation, in particular the chosen model was the Standard k − , a two-equation model
which solves two transport equations, and models the Reynolds Stresses using the Eddy Viscosity ap-
proach. This is the most widely-used engineering turbulence model for industrial applications in CFD
analysis. It is also known to be robust, economic, and reasonably accurate for a wide range of turbulent
flows [90]. Moreover, it is used in similar wing and UAV design and modelling to obtain the lift and drag
forces [96, 97]. The model takes into account the effects of convection and diffusion in turbulent flows.
The first equation computes the turbulent kinetic energy of the flow, k, whereas the second equation
computes the rate of dissipation of turbulent kinetic energy per mass unit, [96, 98]. The k − model is
described with more detail in [94].
In order to better predict fluid behavior, in the boundary layers near the solid wall, prism layers are
used [95]. The parameter y + is used in order to define the thickness of the first prism layer, also defining
a growth rate for the remaining layers [99]. This parameter is used to define how coarse or fine a mesh is,
in a first layer. For turbulent analysis, using a k − models, it is recommended values of 30 < y + < 300
[99]. Thus, in this analysis, y + = 50 is used. The main constraint of using a very fine mesh is the
number of elements required. When considering other parameters like velocity, air density, air viscosity,
mean chord and flow type (all previously defined), the first layer thickness is obtained, 1.75 mm. This
parameter is used in the inflation layer in order to accurately capture the boundary effects of the flow
around the wing [91, 99].
The next step of the analysis is to define the flow boundary conditions. The inlet velocity is imposed
with 15m/s. The outlet wall was defined as the pressure-far-field. The other side of the flow was defined
with symmetry [100]. After choosing a model, defining the y + value and the boundary conditions, the
grid generation ensues. A hybrid grid is used, where the main elements are tetrahedral and in the solid
wall, prisms are used. Additionally a rectangular block around the wing and the wake is used, as it is
shown in Figure 4.9, in order to obtain better resolution and, achieving more accurate results.
62
(a) Overview of the grid. (b) A detailed view of the grid.
Figure 4.9: Representation of the grid.
Creating a high-quality mesh is one of the most critical factors in order to obtain accurate results and
fast convergence, meaning the results should not change with mesh refinement. With this in mind, a
convergence analysis is performed, by presenting the lift and drag forces in function of the number of
elements, as shown in Figures 4.10. Between each iteration a reduction of the element size is calculated,
increasing the number of elements.
(a) Lift force. (b) Drag force.
Figure 4.10: Results analysis in function of the number of elements.
Figure 4.10(a) shows that from the 3.7e5 element, the lift force stabilizes. The same happens with
the drag force (Figure 4.10(b)). Consequently, 3.7e5 numbers of elements are enough to obtain accurate
results with minimum computational effort. After solving all iterations and converging the residual sum,
the values of energy, k, and are analyzed (using the criteria defined by ANSYS Fluent User’s Guide in
[101]). This requires the scaled residuals to be below 10−3 , except for the energy that should be below
10−6 . According to these analyses, the energy value has an order of magnitude of 10−8 and 10−6 for k
and , guarantying that the requirements are satisfied, as it is possible to verify in Figure F.1, presented
in Appendix F. Moreover, another parameter that is used in order to analyze the mesh is the Orthogonal
Quality. This should be between 0 and 1, where worst cells will have an orthogonal quality closer to 0,
and the best cells closer to 1. The minimum value is required to be above 0.01. The performed analysis
has a minimum orthogonal quality of 0.1226 [102].
Table 4.3 presents the lift and drag coefficients obtained from ANSYS Fluent:
63
Table 4.3: Lift and drag coefficients from Fluent.
Data Value
Lift coefficient for cruise CLcruise 0.203
Drag coefficient for cruise CDcruise 0.0225
Zero lift drag coefficient for cruise CD0cruise 0.0102
Induced drag coefficient for cruise CDicruise 0.0124
In order to compare the results between both used software, Table 4.4 is presented, where it is also
shown the error in function of the values obtained from Fluent.
Table 4.4: Comparison between results obtained from XFLR5 and Fluent.
XFLR5 Fluent %error

Lift [N] 29.375 29.379 0.01
Drag [N] 1.628 3.301 50.68
Endurance [h] 6.4 3.23 115.5
It is possible to observe that the values presented in Table 4.4 are similar in terms of lift force and
different in terms of drag force, as expected. According to the literature, XFLR5 gives a good estimation
of lift, but tends to underestimate the drag (likely in the viscous part) [60, 103]. Consequently, since one
of the variables for the endurance calculation is the drag force, the endurance time calculated using data
from XFLR5 was overestimated, due to the fact that these parameters have an inverse relation.
Figures 4.11 and 4.12 present the pressure and velocity distribution around the wing.
Figure 4.11: Pressure distribution around the wing.
64
Figure 4.12: Representation of the velocity around the wing.
In the second Figure 4.12 it is noticeable that there is a thin boundary layer in the leading edge of the
wing and that it thickens in the trailing edge. The increment of the boundary layer thickness is justified
through the transition of laminar to turbulent flow.
This variation of drag force between software has been previously described and studied by Or
Dantsker and Moiz Vahora in [97]. In this work, an UAV with similar cruise speed, but different con-
figuration was used, making it a good model for comparative analysis of the results. This work shows
an error in relation to values obtained from Fluent, of 88.5% for the zero lift drag coefficient. It is also
shows Fluent’s zero lift drag coefficient has an error of 18% while XFLR5 has an error of 87%, when
compared to flight tests. Accordingly, the drag force coefficient to re-calculate the endurance will be the
one from Fluent. With the previous results verified, the forces obtained in this section can now be used
for structural analysis.
4.3 Structural Analysis

Prior to the computational analysis, analytical calculations should be performed in order to have an idea
about the order of magnitude of deflection and maximum stress. This would allow the comparison be-
tween the values obtained from both methods, and verification of possible errors in the models. In order
to simplify the system, the wing is approximated to a cantilever beam with a constant cross section (rect-
angular shaped), whose equivalent area is equal to the mean chord area. The estimation is executed
with the integration of the cross section with a SolidWorks tool (at z = 750 mm) and this area is divided
by the mean chord. Thus, the length of the semi-span (1500 mm) and cross section of 350 × 30.82 mm2
of the beam are obtained as shown in Figure 4.13.
65
Figure 4.13: Approximation of the wing to the beam with constant cross section at mean chord (z =
0.75m).
In order to simplify the system, the lift force was approximated as a constant distributed force, as
presented in Figure 4.14.
Figure 4.14: Representation of the beam’s lift force distribution.
Taking into account that the maximum deflection value occurs in the free end, this value can be
calculated by Equation 4.5 [88],
q l4
ymax = − (4.5)
8E I
This analysis is made for cruise conditions. However, the maximum deflection varies linearly with
the rate of loading applied, which simplifies the calculations when load factors are applied. For the lift
force distribution (q), the value from ANSYS Fluent, q = 9.793 N/m, is used (total lift for semi-span is
14.690N). According to this, and by combining the Equations 4.2 and 4.5 with the Young’s modulus
obtained in Section 4.1.1, the value of the maximum deflection in the free end is 427.988 mm
According to the pure bending theory [88], it is expected, that at the cantilever beam, the maximum
moment value and consequently the maximum stress value is at root face, x = 0. Therefore, its value is
calculated obtaining M = 11.018 N m.
My
σmax = ; y = i/2 (4.6)
I
Due to the fact that the maximum stress is in x = 0, it is necessary to obtain a new moment of inertia
for this cross section. The process to approximate the wing to the rectangular cross section is the same
as the one used above. The approximated cross section is 500 mm of the width and 45.1mm of the
height. By combining Equations 4.6 and 4.2, the maximum stress of σmax = 65003 P a is calculated.
Table 4.5 presents the summary of the analytical values obtained.
66
Table 4.5: Summary of the analytical results obtained.
Value
Maximum Principal Stress [kP a] 65.003
Total deformation [mm] 427.990
As was mentioned before, the calculated values give an order of magnitude of the maximum stress
and deflection. Regarding the deflection, this value seems to be too high (16.5o of deflection). It is also
important to refer that the weight of the structure is low, so it is not taken into account. However, when
considering that the used cross section is an approximation, the real value could be much lower than
428 mm. With the analytical analysis finished, the computational analysis ensues.
4.3.1 Computational Structural Analysis
Since the analytical value was calculated using simplified shapes, differences in the results are to be
expected. However, both should have the same order of magnitude.
This analysis uses the Finite Element Method (FEM), where the geometrically complex domain of the
structural problem is systematically represented by a large, but finite, collection of simpler sub-domains,
called finite elements. The polynomials are used in order to approximate the displacement field of
each element. These are interpolated with respect to the pre-selected nodes and are referred to as
interpolations, where different methods can be applied in order to determine the unknown nodal values
(displacements). Stresses and strains are determined in function of these nodal values [104].
The structure is expected to always be under elastic deformation, which means a linear analysis is
enough to characterize it [104]. A linear static analysis means that load conditions do not vary in time,
and that inertia and damping effects are ignored. It also assumes that the small deflection theory is
taken into consideration, and that the materials have linear elastic behavior [105]
Firstly, it is essential to define the material properties of the foam wing, previously calculated in
Section 4.1.1. The definition of the boundary conditions follows, in order to constrain the wing, where a
fixed support is used on the root face. This restrains all of the six degrees of freedom, translation, and
rotations in x, y and z directions [102].
The forces that are applied on the wing are descendant from the coupling system of the fluid-solid
interface. This system starts by analyzing both meshes, dividing the faces into nodes, in order to overlap
the areas, and to obtain the mapping weights of each node. The mapping weights algorithm is used for
transferring quantities like forces, mass, and momentum for structural analysis [102, 106].
Due to the fact that the wing does not have a hollow structure 2D elements are not appropriat for this
analysis and a solid 3D element should be used. These are three-dimensional finite elements that can
model solid bodies and structures without any a priori geometric simplification. The selected element is
SOLID 186, presented in Figure 4.15, is a second order element defined by 20 nodes with three degrees
of freedom per node: translations in the x, y, and z directions.
67
Figure 4.15: Representation of the element SOLID 186 [107].
After defining the element type, the mesh is now generated, as shown in Figure 4.16. Taking into
account that the used element is from second order, the number of nodes is much higher that the number
of elements.
Figure 4.16: Representation of mesh with 5840 elements (26709 nodes).
As it happens in CFD, during structural analysis, mesh convergence is fundamental, to ensure that
there are no variations in the results. Figure 4.17(a) presents the maximum Von Misses stress, and
Figure 4.17(b) presents the maximum deflection, both in order of the number of elements. In the stress
analysis, the maximum and minimum principal stresses are also analyzed, both shown in order of the
number of elements in Appendix G, Figures G.1 and G.2.
(a) Maximum Von-Misses stress. (b) Maximum wing deformation.
Figure 4.17: Results analysis in function of the number of elements.
68
From the four convergence graphs (Figures 4.17(a), 4.17(b), G.1 and G.2) it is possible to conclude
that the results converge after 3780 number of elements, which corresponds to 20511 nodes. Conse-
quently, the limitation of the number of elements and nodes does not compromise the accuracy of the
results. After guaranteeing mesh convergence, and with the assurance that the results have a minimal
numerical error induced by the mesh, the analysis of the results ensues.
Figure 4.18: Equivalent Von Mises Stress for cruise speed.
Figure 4.19: Wing deformation for cruise speed.
Figure 4.18 presents the Equivalent Von Mises Stress, where the critical area is observed in the
upper root of the wing, (red area). Adjacently to the red area there are large stress concentrations
(orange and yellow areas), which means that if the wing yields, it may crack in this area. Additionally,
Figure 4.19 shows wing deformation, with a maximum value of 106.7 mm (in Y axis) in the tip of the wing.
Table 4.6 presents a summary with the maximum stress and deformation obtained.
69
Table 4.6: Summary of the obtained maximum values.
Value
Von Mises Stress [kP a] 63.459
Maximum Principal Stress [kP a] 51.797
Minimum Principal Stress [kP a] -63.563
Total deformation [mm] 106.7
The comparison between both analytical and computational values shows that these values have the
same order of magnitude, which gives confidence in the validity of the obtained results. With the max-
imum stress obtained for the cruise speed, it is now possible to verify whether the wing can withstand
the aerodynamic loads caused by gusts. As mentioned in Section 3.7.7, the maximum load factor is
nmax = 8.1 for cruise speed. Therefore, it is only necessary to apply the load factor to the maximum ob-
tained value of stress and compare the result with the maximum yield stress of the XPS foam, Equation
4.7. This consideration is made, due to the fact that the structure is in elastic deformation. Moreover, this
is a simplified method making it a conservative approach. In this case, the maximum value of the three
methods is used in order to also be conservative. From Table 4.6 it is conceivable that the maximum
absolute value corresponds to the Minimum Principal Stress, which is | − 63.563| kP a.
σmax × nmax = 514.860 kP a > σyieldXP S = 361.0 kP a (4.7)
Since the maximum stress is clearly higher than the maximum yield stress, Equation 4.7 the wing
requires a small reinforcement. Consequently, a spar will be created using the same technique, FDM,
and the same material, PLA, of the internal structure. This reinforcement will be attached to the back
side of the frame in a perpendicular direction to the root face, in order to support the most critical areas. It
has a rectangular cross section with the height of the reinforcement, from the upper to the lower surface
of the wing, and a length of 250 mm, encompassing the area with higher values of stress.
As in the previous analysis, the imposed boundary condition is a fixed support on the root face
of the wing, including the reinforcement. Taking into account that there are two structures, the wing
and the reinforcement, it is essential to define the type of connection created and the surfaces that
are affected. In this case, three surfaces are used to connect both structures, and the contact type is
bonded, preventing sliding or separation between faces [108]. To do so, two types of elements are used,
TARGE170 and CONTA174. The first one is used to represent a 3D target surface for associated contact
elements (CONTA174). This overlays the solid elements, describing the boundary of a deformable body
that is potentially in contact with the rigid target surface [107]. These elements have 8 nodes, and
allow imposing any translational or rotational displacement, temperature, voltage, and magnetic potential
[107].
The reinforcement is meshed with element SOLID186 and the wing is meshed with element SOLID187.
This is a 10-node element, suited to model irregular meshes, with three translational degrees of freedom
at each node (x, y, and z direction) (Figure 4.20) [107]. Due to the geometry and the constraint of the
number of nodes/elements, it was not possible to use element SOLID186 for the wing.
70
Figure 4.20: Element SOLID 187 [107].
The generated mesh is presented in Figure 4.21. Although globally the wing’s mesh is coarse, in the
reinforcement the mesh is acceptable. This is due to the stiffness difference between both materials as
the stress concentration will now occur in the reinforcement.
Figure 4.21: Representation of mesh with 16517 elements (31989 nodes).
After defining the boundary conditions and meshes, the maximum deformation for cruise speed is
simulated, and it is presented in Figure 4.22.
Figure 4.22: Representation of the wing deformation with the reinforcement, for cruise speed.
71
Comparing both Figure 4.22 and 4.19, it is possible to verify that the maximum wing deformation
was reduced from 107 mm to 26 mm for cruise speed. This was expected since PLA has an increased
stiffness, and therefore, a higher value of Young’s module than XPS. Beyond the deformation, the stress
is also analyzed in Figure 4.23, which presents the Von-Mises stress distribution. This shows that with
the reinforcement, the average stress in the wing is low (around 20 kP a), and that most of the loads are
supported by the reinforcement. The scale in Figure 4.23 was adjusted in order to have an orange color
associated with the maximum allowed yield stress σallowed of XPS. This value is calculated taking into
account the load factor nmaxcruise = 8.1 (σallowed = σyield /nmax ), and it indicates which areas exceed
the maximum yield stress when applying a maximum load factor (red regions).
Figure 4.23: Representation of Von-Mises stress distribution, for cruise speed.
Figure 4.24 and 4.25 present the Von-Mises stress distribution in wing and reinforcement, respec-
tively. As in the previous figure, the scale was adjusted, using the same method, in order to have an
orange color for maximum allowed yield stress (for XPS in Figure 4.24 and PLA in Figure 4.25). Thus,
the regions that have red color exceed the maximum yield stress of each material when a maximum load
factor is applied. From Figure 4.24 it is observable that there are three regions with a red color, which
are called stress concentration regions. These areas are near to abrupt changes in geometry which
disrupt the smooth flow of stress through the structure. This induces large stress gradients, where the
maximum stress greatly exceeds the average and normal values [104].
In order to fix the stress concentration, the recommended procedure is to smooth the edges. This
can be made with the increment of the radius of the edge [104]. Due to the fact that the red zones are
stress concentration, induced by the shape, these values should be ignored for the structural analysis.
Nevertheless special attention should be given to them during the manufacture process, by performing
a fillet in these edges.
72
Figure 4.24: Representation of Von-Mises stress distribution in the wing.
Figure 4.25: Representation of Von-Mises stress distribution in the reinforcement.
All previous analyses were made in cruise speed with a maximum load factor of n = 8.1. The same
process was also made for the maximum operation speed, UM O = 20.25 m/s, which corresponds to the
maximum load factor of n = 5.8. However, it was concluded that due to the higher value of load factor,
the cruise speed is more critical, with higher values of stress in critical conditions (n = 8.1). Therefore,
the designed structure (Figure 4.26) allows the wing to withstand various aerodynamic loads, including
gusts.
To conclude, in this chapter, the internal structural design was completed and shown to be adequate
for both prototypes (long and short endurance). The wing performance was validated and if necessary,
reinforcements could be used, as it happens in V2, Figure 4.26. In addition, the design of the aircraft
respects STANAG 4703, which is crucial for its functionality.
73
Figure 4.26: Representation of the internal structure with reinforcements.
74
4.4 Mission
This section has the main goal of giving an overview of the necessary steps to put and keep the aircraft
in flight, the missions that the UAV can perform and the area that it can survey without changing the
batteries. Before describing the launching process, it is crucial to keep in mind that the main ambition
of this project is to have a fully autonomous system, which allows to reduce costs with operator training.
In this regard, the training involved would only be related to the mission planning and the launching of
the platform. For the accomplishment of a mission it is necessary to have two operators, one to give the
order to start the mission (in the software), and another to perform the launching.
The first step is to assemble the UAV before flight, which would require the connection of the batteries
to the system. The mission generation ensues, in the ground station with the Mission Planner software,
where the area to survey is delimited and automatically generated in function of parameters such as
the camera used, flight altitude and cruise speed. Currently the camera is not chosen yet, due to the
fact that it may change in function of the mission and the requirements for image processing5 used for
automatic fire and smoke detection. With the mission defined, the launching of the platform ensues. As
mentioned before, the area that the aircraft can survey is directly affected by several parameters. Table
4.7 shows the influence of the flight altitude and of the Field of View (FOV) of the camera (depicted
in Figure 4.27(b)) in the distance between each line of passage, which are presented in Figure 4.28.
In order to increase the view angle of the camera, it was decided to impose an inclination of 30o as
presented in Figure 4.27(a).
(a) Lateral view, impositions of inclination angle. (b) Front view, representation of FOV angle.
Figure 4.27: Representation of the camera position.
Table 4.7 shows the distance between passages. As an example, with a flight altitude of 100 m and
an inclination angle of 30o , a 115 m distance to the ground is obtained. Applying the FOV of the camera
(120o ), the distance of width in the ground that the camera can capture is 200 m. To better understand
these considerations, Figures 4.27(a) and 4.27(b) are presented.
5 Taking into account that the image processing software is not currently developed, the requirements of the camera can change
in the future.
75
Table 4.7: Options for flight altitude and camera parameters and their influence in the distances between
lines of passage.
FOV of camera
Flight altitude Increment +30o of inclination 120o 145o 160o
100 m 115.5 m 200 m 366.2 m 654.9 m
200 m 231.0 m 400 m 732.5 m 1309.7 m
300 m 346.4 m 600 m 1098.7 m 1964.6 m
400 m 461.9 m 800 m 1464.9 m 2619.5 m
500 m 577.6 m 1000 m 1831.1 m 3274.3 m
Taking all this into account, it is possible to simulate a mission, and consequently get an estimation
of the area that the aircraft can survey and the time that this would require. For these estimations, it was
imposed a flight speed of 15 m/s. Regarding the distance between the lines of passage, a distance of
400 m was used for the first simulation and 700 m for the second, both with 10% of overlapping in each
passage. Figures 4.28 and 4.29 present these two simulations, with the first one showing the mission in
detail and the second an overview of the mission and the map.
As mentioned before, Figure 4.28 was obtained considering that the distance between each passage
line is 400 m. In these conditions, it is possible to survey an area of 4, 523 ha, which would require around
two and a half hours.
Figure 4.28: Representation of the first simulation (400 m).
76
Regarding the second simulation, a distance of 700 m was used, obtaining a surveillance capability
of 7, 966 ha and also requiring around two and a half hours.
Figure 4.29: Representation of the second simulation (700 m).
Consequently, the higher the cruise altitude and the larger the FOV angle, the greater is the surveil-
lance capacity. However, these parameters must be chosen considering the image processing capability
and the camera resolution, which is outside the scope of this thesis.
77
78
Chapter 5
Conclusions
This chapter summarizes the key aspects of DeltaSpotter’s development. It also examines and evaluates
the steps and decisions made during this product development, in accordance to the proposed goals
mentioned in the Introduction. Furthermore, it highlights the final characteristics and configuration of the
UAV before finishing with a discussion on further improvements in the next iteration of this project.
5.1 Achievements
The main objective of this thesis was to design a new prototype of DeltaSpotter, with a longer flight
period. Due to its fire-surveillance purpose this project had several constraints in order to be useful
on daily, basic surveillance: it should be easy to carry; have a fast assembly time; and be low cost to
manufacture. In addition, it should also have a takeoff by hand launching, to facilitate frequent use by
non-specialized personnel.
After defining both the constraints and desired characteristics, available solutions with different con-
figurations were analyzed during market research. In addition, several materials, manufacturing pro-
cesses and propulsive systems were researched, presented and compared. Furthermore, the previous
prototype developed by CEiiA was described and studied to better understand the shortcomings and
necessary improvements.
Afterwards, several product features were decided, including the propulsion system, materials and
manufacturing techniques. In order to keep it easy to assemble and transport, a flying wing configuration
was shown to be more adequate, particularly when associated to an electric propulsion system. To keep
the manufacturing cost low, XPS foam with a 4 axis hot wire cut was selected for wing construction, while
for the internal structure an adequate choice was shown to be PLA, with a FDM printer. Both of these
materials were good compromises as they were shock resistant while being light-weight and relatively
cheap. In addition, since CEiiA already owned the hot cutter and the FDM printer, both manufacturing
steps are considered exceptionally inexpensive.
With key aspects of the product development decided it was time to move to initial size and weight
estimation, which was accomplished by using data from the DeltaSpotter V1. Several aerodynamic
79
analyses were made with the VLM method of XFLR5, thus allowing the selection of an airfoil with a higher
efficiency. Since the UAV has a flying wing configuration, the choice of airfoil was limited. Furthermore,
because of the wings’ manufacturing technique, the selected airfoil thickness and other characteristics
had to be modified. After the wing configuration was defined, the characterization of all degrees of
freedom ensued. In specific, the size of the wing was limited to a maximum wingspan of 3 m, which
naturally restricted the aspect ratio that could be used. To ensure that the expected poor lateral dynamic
stability of a tailless aircraft would not compromise the design, a stability analysis was performed. During
which, the vertical stabilizers were adopted from the V1 prototype, in order to minimize induced drag
and guarantee a stable aircraft. To finish the analysis, the flight and gust envelopes were designed in
accordance to the STANAG 4703 norm, which is highly conservative. In this stage, gust values were
considered similar to the cruise speed (15 m/s).
To execute the structural analysis of the aircraft several steps had to be completed first, from the
various materials characterization to the geometry modeling and CFD analysis. To guarantee that the
structural analysis would be close to the real world the Young’s Modulus and Yield Tensile Strength were
researched and selected carefully. In addition, to maximize efficiency the PLA density was carefully
chosen to not compromise resistance with a lighter weight. The framework of the geometry modeling
was divided in two parts, the internal structure that accommodates all electric components and the wing.
The internal structure should facilitate the launching and landing and it should be usable in both short
and long endurance versions. To keep with the theme of developing a low cost and lightweight aircraft
the manufacturing process selected is ideal to build complex geometries. With the geometric modeling
and material characterization complete, the CFD and Structural analyses ensues. ANSYS, the software
used, permits the coupling of both analyses thus simplifying the interaction between disciplines. The
CFD analysis is suitable to obtain the wing loads (for structural analysis), plus the lift and drag forces.
This enables a comparison between these values and the ones previously obtained through XFLR5,
which warrants that the value computed through ANSYS Fluent is preferable to estimate the endurance
(as it provides more accurate results). Last but not least, the structural analysis was implemented
to verify that the structure could withstand the aerodynamic loads imposed by STANAG 4703. This
analysis resulted in the implementation of a reinforcement coupled to the internal structure.
The theoretical endurance of the UAV was estimated to be between two hours and a half and three
hours 1 , an improvement of 1.5 to 2 hours in regard to the previous prototype. The manufacturing cost
was calculated by adding the cost of the materials (XPS and PLA, 30 e and 15 e respectively) to the
electric components shown in Table A.2. Consequently, the total cost of materials and components adds
up to 300 e, well below the price of most surveillance UAVs available in the market.
As a final note, it is worth to remember that these analyses were done with a computational limitations
in term of the number of cells/elements. This clearly constrains the analysis, since in this simulation the
vertical stabilizer was not included in the CFD and the structural analysis. In addition, the number of
elements restricted the reinforced wing analysis. Nevertheless, the simulation was completed with fully
defined wing and internal structure that follow the requirements and the norm STANAG 4703, which
1 These values were obtained considering a margin of 10% and 20% of total battery charge for takeoff, climb and remaining
battery charge.
80
itself is quite restrictive.
Chapter 4 is finalized with an overview of the missions that the aircraft can perform, the procedure
required for launching, and the area that it can survey in function of different parameters.
For reasons, unrelated to the author of this thesis the prototype has not yet been manufactured,
although completion is expected to be done until the end of October, where several flight tests are
expected to be performed, to guarantee structural integrity and obtain real endurance times.
5.2 Future Work

The main focus of any future work will be the aftermath of the flying tests, as the endurance, stability
and structural stiffness need to be confirmed and tested. To do so, the tests performed need to be done
in both normal and adverse conditions. In addition, several cruise speeds should be tested, and the
corresponding endurance values calculated to better understand how speed and flight conditions impact
the overall endurance. Additionally, in order further to simplify the transportation, it could be interesting
to put the vertical stabilizers horizontally.
In regard to the optimization procedure, it would be valuable to repeat the CFD and Structural analy-
ses by including the vertical stabilizers and increasing the number of elements. This would require much
more computational effort and likely a paid version of the program. Another gripping upgrade would be
an optimization of the internal structure reinforcement’s size.
Another interesting future perspective is related with the carbon footprint of this aircraft, by finding
biodegradable alternatives to XPS foam and developing strategies to recycle all components and lower-
ing missions and energy costs. This could be done with renewable energy, like solar panels, although
that would be a thesis by itself.
It is necessary to develop an image processing software, which will permit choosing a camera ad-
equate for the surveillance purpose. This was taken into consideration during development, and the
weight of the camera could be up to 300 g without negatively affecting the aircraft performance.
To conclude, it is conceptually possible to develop a low-cost drone for a high risk mission, as is the
case of fire surveillance. This is clearly a necessary step in the upgrade of the countries fire surveil-
lance strategy, although there are many steps that need to be tackled before this project is ready to be
commercialized.
81
82
Bibliography
[1] Direção Geral do Território - Cartografia de Uso e Ocupação do Solo. URL http:
//www.dgterritorio.pt/cartografia_e_geodesia/cartografia/cartografia_tematica/
cartografia_de_uso_e_ocupacao_do_solo__cos_clc_e_copernicus_/. Accessed: 2018-04-
03.
[2] N. Calado. Climate change and forest fire management, January 2018. URL
http://www.europeanlandowners.org/images/Intergroup/Mr._CALADO_UNAC_EP_BRUSSELS_
23_01_2018_vfinal2.pdf.
[3] C. C. Silva. 2017 foi o ano em que mais ardeu nos últimos dez anos — quatro vezes mais que o
habitual, November 2017. URL https://goo.gl/LSmiF9.
[4] Análise e apuramento dos factos relativos aos incêndios que ocorreram em Pedrogão Grande,
Castanheira de Pera, Ansião, Alvaiázere, Figueiró dos Vinhos, Arganil, Góis, Penela, Pampilhosa
da Serra, Oleiros e Sertã, entre 17 e 24 de junho de 2017. Technical report, 2018.
[5] J. D. A. Viana. Rede Nacional de Postos De Vigia – Tendências para o futuro. Master’s thesis,
Academia militar Direcção de ensino, August 2010.
[6] M. J. P. de Freitas. análise da problemática dos incêndios florestais – relatório do grupo de

trabalho – volume i. Technical report.
[7] D. Marshall, R. Barnhart, E. Shappee, and M. Most. Introduction to Unmanned Aircraft Systems,
Second Edition. CRC Press, 2016.
[8] SKYbrary. URL https://www.skybrary.aero/index.php/Unmanned_Aerial_Systems_(UAS)

#The_civil_applications_of_UAS. Accessed: 2018-03-07.
[9] C. C. Wilson and J. B. Davis. Forest Fire Laboratory at Riverside and fire research in California:
past, present, and future. USDA Forest Service, 1988.
[10] V. G. Ambrosia, S. Wegener, T. Zajkowski, D. V. Sullivan, S. Buechel, F. Enomoto, B. Lobitz,

S. Johan, J. Brass, and E. Hinkley. The Ikhana unmanned airborne system (UAS) western states
fire imaging missions: from concept to reality (2006–2010). Geocarto International, 26:85–101,
2011.
83
[11] M. van Persie, A. Oostdijk, J. Fix, M. C. van Sijl, and L. Edgardh. Real-Time Uav Based Geospatial
Video Integrated Into the Fire Brigades Crisis Management Gis System. ISPRS - International
Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, pages 173–
175, 2011.
[12] D. W. Casbeer, D. B. Kingston, R. W. Beard, and T. W. McLain. Cooperative forest fire surveillance
using a team of small unmanned air vehicles. International Journal of Systems Science, 37:
351–360, 2006.
[13] European Commission. Study analysing the current activities in the field of uav.
2007. URL https://ec.europa.eu/home-affairs/sites/homeaffairs/files/e-library/
documents/policies/security/pdf/uav_study_element_2_en.pdf.
[14] Botlink. URL https://goo.gl/Vsvp8T. Accessed: 2018-03-07.
[15] M. Hassanalian and A. Abdelkefi. Classifications, applications, and design challenges of drones:
A review. Progress in Aerospace Sciences, 91(September):99–131, 2017.
[16] UAV Factory, . URL http://www.uavfactory.com/product/69. Accessed: 2018-03-05.
[17] R. Roo, J. Verbeke, and I. Becuwe. Development of UAS for scientific monitoring, September 2013.
URL http://www.fzt.haw-hamburg.de/pers/Scholz/ewade/2013/EWADE2013_DeRoo.pdf.
[18] SmartPlanes, . URL http://smartplanes.com/our-offering/. Accessed: 2018-04-16.
[19] Aeromao. URL http://www.aeromao.com/aeromapper_talon_uav/. Accessed: 2018-04-16.
[20] DT18 HD Long-Range Mini UAV Survey, Monitor and Inspect. Delair-Tech SAS, October 2017.
URL http://delair.aero/wp-content/uploads/2018/01/Delair-datasheet-DT18-HD-WEB_
Oct2017vA.pdf.
[21] Parrot SA. URL https://www.sensefly.com/drone/ebee-sq-agriculture-drone/. Accessed:

2018-03-05.
[22] AgEagle. URL http://ageagle.com/rx48-2/. Accessed: 2018-04-02.
[23] Aeromao Inc. URL http://www.aeromao.com/aeromapper_uav/. Accessed: 2018-03-20.
[24] R. Austin. Unmanned Aircraft Systems. Wiley, 2010.
[25] J. Hale. Boeing 787 from the ground up. Boeing Aero Magazine, 4, January 2006.
[26] A. Balachandran, D. Karelia, and J. Challa. Material Selection for Unmanned Aerial Vehicle.
International Journal of Mechanical Enginering and Technology (IJMET), 5(8):34–40, 2014.
[27] G. Newaz, M. Mayeed, and A. Rasul. Characterization of balsa wood mechanical proprieties re-
quired for continuum damage mechanics analysis. Materials: Design and applications, November
2014.
84
[28] A. Balachandran, D. Karelia, and J. Challa. Material Selection for Unmanned Aerial Vehicle.
International Journal of Mechanical Engineering & Technology (IJMET), 4:34–40, August 2014.
[29] A. Bureiko, A. Trybala, N. Kovalchuk, and V. Starov. Current applications of foams formed from
mixed surfactant-polymer solutions. Advances in Colloid and Interface Science, 222:670–677,
2015.
[30] Flying Lines. URL http://flyinglines.org/buzz.foamcutter.4.html. Accessed: 2018-03-21.
[31] FlyingFoam, . URL https://flyingfoam.com/knowledgebase/eps-vs-epp-vs-xps/. Accessed:

2018-03-20.
[32] British Plastics Federation, . URL http://www.bpf.co.uk/plastipedia/polymers/Expanded_

Polypropylene_EPP.aspx. Accessed: 2018-03-20.
[33] Briotherm, . URL http://www.giasxps.ro/index.php/en/electronic-library-polystyrene/

77-xps-eps-comparison. Accessed: 2018-03-20.
[34] H. F. Brinson and L. C. Brinson. Polymer Engineering Science and Viscoelasticity. Springer US,
2008.
[35] Additive Manufacturing - Benefits and Functional Principle, . URL https://www.eos.info/

additive_manufacturing/for_technology_interested. Accessed: 2018-09-05.
[36] Additive Manufacturing Technology Standards, . URL https://www.astm.org/Standards/

additive-manufacturing-technology-standards.html. Accessed: 2018-09-05.
[37] B. P. Conner, G. P. Manogharan, A. N. Martof, L. M. Rodomsky, C. M. Rodomsky, D. C. Jordan,

and J. W. Limperos. Making sense of 3-D printing: Creating a map of additive manufacturing
products and services. Additive Manufacturing, 1-4:64–76, October 2014.
[38] Printing The Future: What Additive Manufacturing Means For Innovation, . URL https://www.ge.
com/reports/printingthefuture/. Accessed: 2018-09-05.
[39] R. J. Morrison, S. J. Hollister, M. F. Niedner, M. G. Mahani, A. H. Park, D. K. Mehta, R. G. Ohye,

and G. E. Green. Mitigation of Tracheobronchomalacia with 3D-Printed Personalized Medical
Devices in Pediatric Patients. Science transnational medicine, 7:285–264, 2015.
[40] A. J. Keane, A. Sóbester, and J. P. Scanlan. Small Unmanned Fixed-wing Aircraft Design: A
Practical Approach. Wiley, 2017.
[41] C. Griffis, T. Wilson, J. Schneider, and P. Pierpont. Unmanned Aircraft System Propulsion Systems
Technology Survey. 9, 2009.
[42] Motoren und Flugmodelle. URL https://3w-modellmotoren.de/product/3w-28i-cs/?lang=en.

Accessed: 2018-08-15.
85
[43] P. Osenar, J. Sisco, and Catharine Read. Advanced Propulsion for Small Unmanned Aerial Ve-
hicles. The Role of Fuel Cell-Based Energy Systems for Commercial UAVs. Unmanned Systems
Technology, 2017.
[44] Mission Planner. URL http://ardupilot.org/planner/docs/mission-planner-overview.

html. Accessed: 2018-09-15.
[45] L. W. Traub. Range and Endurance Estimates for Battery-Powered Aircraft. Journal of Aircraft, 48
(2):703–707, 2011.
[46] J. T. K. Ping, A. E. Ling, T. J. Quan, and C. Y. Dat. Generic Unmanned Aerial Vehicle (UAV) for civil-
ian application. Proceedings of the IEEE Conference on Sustainable Utilization and Development
in Engineering and Technology (STUDENT), pages 289–294, October 2012.
[47] G. J. V. Kimon P. Valavanis. Handbook of Unmanned Aerial Vehicles. Springer, 2015.
[48] The UAV Guide. URL http://wiki.theuavguide.com/wiki/Propulsion_System. Accessed:

2018-08-15.
[49] T. C. Corke. Design of Aircraft. Prentice Hal, 2003.
[50] L. M. A. Parada. Conceptual and Preliminary Design of a Long Endurance Electric UAV. Master’s
thesis, UL - Instituto Superior Técnico, 2016.
[51] S. K. Moon, Y. E. Tan, J. Hwang, and Y. J. Yoon. Application of 3D printing technology for de-
signing light-weight unmanned aerial vehicle wing structures. International Journal of Precision
Engineering and Manufacturing - Green Technology, 1(3):223–228, 2014.
[52] Manual Creatr HS. Leapfrog. URL https://www.lpfrg.com/ftp/support/creatrhs/manuals/

Leapfrog-3D-Printers_Creatr-HS-Manual.pdf. Accessed: 2018-09-15.
[53] J. F. M. Fernandes. Estudo da Influência de Parâmetros de Impressão 3D nas Propriedades

Mecânicas do PLA. Master’s thesis, UL - Instituto Superior Técnico, October 2016.
[54] T. Chang and H. Yu. Improving Electric Powered UAVs’ Endurance by Incorporating Battery Dump-
ing Concept. Procedia Engineering, 99:168–179, 2015.
[55] J. Roskam and C.-T. E. Lan. Airplane Aerodynamics and Performance. DARcorporation, 1997.
[56] L. A. Costello. State of the art of piloted electric airplanes, NASA’s centennial challenge data and
fundamental design implications. Master’s thesis, Embry-Riddle Aeronautical University - Daytona
Beach, 2011.
[57] H. Martin. Electric Flight - Potential and Limitations. pages 1–30, October 2012.
[58] O. Gur and A. Rosen. Optimizing Electric Propulsion Systems for Unmanned Aerial Vehicles.
Journal of Aircraft, 46(4):1340–1353, 2009.
86
[59] M. Drela. Xfoil: An analysis and design system for low Reynolds number airfoils. MIT Dept. of
Aeronautics and Astronautics, Cambridge, Massachusetts, 1989.
[60] XFLR5 Analysis of foils and wings operating at low Reynolds numbers, October 2009.
[61] T. J. Mueller and J. D. DeLaurier. Aerodynamics of small vehicles. Annual Review of Fluid Me-
chanics, 35:89–111, 2003.
[62] P. Edi, N. Yusoff, and A. A. Yazid. Airfoil Design for Flying Wing UAV (Unmanned Aerial Vehicle).
4th WSEAS International Conference on APPLIED and THEORETICAL MECHANICS (MECHAN-
ICS ’08) Airfoil, pages 106–111, 2008.
[63] Aerodesign, . URL http://www.aerodesign.de/english/profile/profile_s.htm. Accessed:

2018-05-20.
[64] A. A. Alsahlan and T. Rahulan. Aerofoil design for unmanned high-altitude Aft-swept flying wings.
Journal of Aerospace Technology and Management, 9(3):335–345, 2017.
[65] A. Lennon. Basics of R C Model Aircraft Design Practical Techniques for Building Better Models,
1996.
[66] J. Hall, K. Mohseni, and D. Lawrence. Investigation of variable wing-sweep for applications in
micro air vehicles. AIAA Journal, (September):26–29, 2005.
[67] B. M. S. Cunha. Development of Control Strategies for the Joined-Wing Aircraft. Master’s thesis,
UL - Instituto Superior Técnico, June 2011.
[68] H. Karakas, E. Koyuncu, and G. Inalhan. ITU Tailless UAV Design. pages 131–146, 2013.
[69] T. Benson. Lift to Drag Ratio . URL https://wright.nasa.gov/airplane/ldrat.html. Ac-

cessed: 2018-08-15.
[70] A. Lefebvre and C. Gables. Co-Flow Jet Airfoil Trade Study Part I : Energy Consumption and
Aerodynamic Efficiency. AIAA Journal, pages 1–20, 2014.
[71] Flying wing CG calculator. URL https://fwcg.3dzone.dk/. Accessed: 2018-06-5.
[72] Tomtop. URL https://bit.ly/2lnuUDw. Accessed: 2018-06-5.
[73] HobbyKing, . URL https://hobbyking.com/pt_pt/zippy-flightmax-5000mah-4s1p-20c.html.

Accessed: 2018-06-05.
[74] Jivaro Models. URL http://rcaerolab.eklablog.com/predimrc-p1144024.
[75] R. Carr, D. Chipman, G. Larson, N. Hopkins, D. F. Hunsaker, J. Boyce, R. Carr, D. Chipman,

G. Larson, N. Hopkins, D. F. Hunsaker, and J. Bowman. Design of ’ Iris ’, a Small Autonomous
Surveillance UAV. 2006.
87
[76] ServoCity. URL https://www.servocity.com/hs-425bb#258=24&259=25. Accessed: 2018-06-
05.
[77] Ebay. URL https://goo.gl/FoH1f2. Accessed: 2018-07-15.
[78] HobbyKing. URL https://goo.gl/s9xzRt. Accessed: 2018-07-15.
[79] HobbyKing, . URL https://hobbyking.com/pt_pt/hobby-king-60a-esc-4a-ubec.html. Ac-

cessed: 2018-07-15.
[80] A. Deperrois. About stability analysis using XFLR5, November 2010. URL http://www.xflr5.
com/docs/XFLR5_and_Stability_analysis.pdf.
[81] J. R. Azinheira. Controlo de Voo MEAero,. 2016.
[82] T. M. Foster. Dynamic Stability and Handling Qualities of Small Unmanned-Aerial-Vehicles. Mas-
ter’s thesis, 2013.
[83] R. Lykins. Unsteady Aerodynamic and Dynamic Analysis of the Meridian UAS in a Rolling-Yawing
Motion. Master’s thesis, 2013.
[84] P. F. Structural design, manufacturing and testing of a new wing for the CSIR’s Modular UAS in
composite materials. Master’s thesis, 2012.
[85] North Atlantic Treaty Organization. NATO Standard AEP-83. Light Unmanned Aircraft Systems
Airworthiness Requirements, STANAG 4703. NATO Standard AEP-83, 2014.
[86] M. T. Restivo, J. C. Marques, T. F. Andrade, and B. Santos, October 2014. URL https:
//remotelab.fe.up.pt/remote_exp/EPCBBT.pdf.
[87] B. W. Inc, 2010. URL https://materion.com/-/media/files/alloy/newsletters/

technical-tidbits/issue-no-21--cantilever-beams---part-2-analysis.pdf.
[88] F. P. Beer, J. E. Russell Johnston, J. T. DeWolf, and D. F. Mazurek. Mechanics of Materials. Mc

Graw Hill, 2015.
[89] T. S. Martins. Influência dos parâmetros de fabrico nas propriedades mecânicas de peças obti-
das por impressão 3D com um único material. Master’s thesis, UL - Instituto Superior Técnico,
November 2017.
[90] ANSYS Inc. ANSYS Fluent: CFD Simulation. URL http://www.ansys.com/Products/Fluids/

ANSYS-Fluent. Accessed: 2018-07-15.
[91] G. R. Center, . URL https://www.grc.nasa.gov/www/wind/TFAWS2007/CFD-Process.pdf. Ac-

cessed: 2018-08-10.
[92] N. I. Ismail, A. H. Zulkifli, R. J. Talib, H. Yusoff, and M. A. Tasin. Vortex structure on twist-morphing
micro air vehicle wings. International Journal of Micro Air Vehicles, 8(3):194–205, 2016.
88
[93] G. R. Center. CFD Analysis Process, . URL https://www.grc.nasa.gov/www/wind/TFAWS2007/
CFD-Process.pdf. Accessed: 2018-09-15.
[94] H. Versteeg and W. Malalasekera. An Introduction to Computational Fluid Dynamics: The Finite
Volume Method. Pearson Education Limited, 2007.
[95] J. H. Ferziger and M. Peric. Computational Methods for Fluid Dynamics. Springer, 2002.
[96] A. Prabhakar and A. Ohri. CFD Analysis on MAV NACA 2412 Wing in High Lift Take-Off Config-
uration for Enhanced Lift Generation. Journal of Aeronautics & Aerospace Engineering, 02(05),
2013.
[97] O. D. Dantsker and M. Vahora. Comparison of aerodynamic characterization methods for design
of unmanned aerial vehicles. 2018.
[98] B. Launder and D. Spalding. The numerical computation of turbulent flows. Computer Methods in
Applied Mechanics and Engineering, 3(2):269–289, 1974.
[99] M. Ariff, S. Salim, and S. Cheah. Wall Y + Approach for Dealing With Turbulent Flow Over a
Surface Mounted Cube : Part 2 – High Reynolds Number. 7th International Conference on CFD
in the Minerals and Process Industries, (December):1–6, 2009.
[100] M. U. Sohail and A. Islam. Verification and Validation of Flow Over a 3D ONERA Wing using CFD
Approach. 7(1):39–43, 2017.
[101] ANSYS Inc. ANSYS FLUENT Getting Started Guide. Knowledge Creation Diffusion Utilization,
15317(October):724–746, 2012.
[102] A. Inc. ANSYS Help Viewer 18.2. Accessed: 2018-09-10.
[103] About XFLR5 calculations and experimental measurements, October 2009. URL http://www.
xflr5.com/docs/Results_vs_Prediction.pdf. Accessed: 2018-08-05.
[104] W. C. Young and R. G. Budynas. Roark’s Formulas for Stress and Strain. Mc Graw Hill, 2002.
[105] J. N. Reddy. An Introduction to the Finite Element Method. Mc Graw Hill, 2nd edition edition,
1993.
[106] R. S. Raja. Coupled fluid structure interaction analysis on a cylinder exposed to ocean wave
loading. Master’s thesis, Chalmers University of Technology, 2012.
[107] Glenn research center. URL https://www.sharcnet.ca/Software/Ansys/17.0/en-us/help/

ans_elem/Hlp_E_CH3_Classifications.html#elsel_contact. Accessed: 2018-09-02.
[108] URL https://www.sharcnet.ca/Software/Ansys/16.2.3/en-us/help/wb_sim/ds_Contact_

Definition.html. Accessed: 2018-09-05.
[109] Israel Aerospace Industries Ltd, . URL http://www.iai.co.il/2013/36943-34720-en/Bird_

Eye_Family.aspx. Accessed: 2018-05-10.
89
[110] Israel Aerospace Industries Ltd, . URL http://www.iai.co.il/2013/36943-39739-en/Bird_
Eye_Family.aspx. Accessed: 2018-05-10.
[111] Airforce Technology. URL https://www.airforce-technology.com/projects/

tracker-mini-unmanned-aerial-system-uas/. Accessed: 2018-05-10.
[112] Skylark I-LE High Performance Man-portable Mini UAS. URL http://elbitsystems.com/media/
Skylark_I_LE_2016.pdf. Accessed: 2018-05-10.
[113] Israel-Weapon Skylite B. URL http://www.israeli-weapons.com/weapons/aircraft/uav/

skylite_b/skylite_b.htm. Accessed: 2018-05-10.
90
Appendix A
Market Analysis of Military UAVs
As mentioned in Section 2.1.3, most of the civilian UAVs have endurance values up to two hours. There-
fore, in order to have an overview of long endurance UAVs, some military UAVs are analyzed and
compared (Table A.1 and Figures A.1, A.2 and A.3).
Table A.1: Military UAVs
Endurance [min] Wing span [m] Weight [kg]

IAI Malat Bird Eye 400 [109] 90 2.2 5.6
IAI Malat Bird Eye 650 [110] 240 3 11
EADS Tracker [111] 90 3.6 8.5
Elbit Skylark l-LE [112] 180 3 7.5
Rafael Skylite B [113] 90 2.4 6
Figure A.1: Wingspan vs endurance
Figure A.1 presents the relation between wingspan and endurance, from which there is a clear in-
crease of endurance when the wingspan increases.
91
Figure A.2: Weight vs wing span
Figure A.2 relates weight with wing span. As expected when the wing span increases, the weight will
increase as well.
Figure A.3: Weight vs endurance
Figure A.3 presents the relation between weight and endurance. From the trend line we can see the
endurance stays nearly constant as the weight is increases.
92
A.1 Technical Weights, Dimensions and Prices of the Components
Table A.2: Weight and dimensions of the electric components
Component Dimensions [mm] Weight [g] Price [e]

Battery 143x51x33 460 26.51 × 3
Motor - 140 28.48
ESC 54x26x11 32 18.04
Propeller (12x6) 26 4.62
Servo 40.39 x 19.56 x 36.58 45 13.00 × 2
APM 2.8 77x45x15 28 70.00
Power module 152x22x12 19 -
GPS 38x38x8.5 28 -
Radio Telemetry - 14 -
93
Appendix B
Aerodynamic Analysis
Figures B.1 and B.3 show Cl over Cd for a Reynolds number of 285000 and 570000, respectively.
(a) Overview of Cl vs. Cd . (b) Detailed analysis in the range values of Cl = 0.1 to 0.6
Figure B.1: Cl vs. Cd for the Reynolds number of 285000.
94
(a) Overview of Cl vs. Cd . (b) Detailed analysis in the range values of Cl = 0.1 to 0.6
Figure B.2: Cl over Cd for Reynolds number of 570000.
From an aerodynamic point of view, the thickness to chord ratio that allows higher CL /CD is t/c=10%,
for speeds up to 11.1m/s (40km/h) is t/c=10%. For higher speeds the best performance is obtained with
t/c=12%. However the difference of CL /CD between thickness of 10%, 11% and 12% is very small, as
seen in Figure B.3. Thus the thickness to chord ratio that should be used is t/c=12%, as a thicker airfoil
allows for a stiffer wing.
Figure B.3: CL /CD with five different relative thickness. The green line corresponds to 10% of the
relative tip thickness, the pink line corresponds to 11%, dark line to 12%, red line to 13% and the blue
line to 14%.
95
Appendix C
Sizing of Control Actuators
Figure C.1 presents the PredimRC Exel software needed to calculate the maximum required torque for
elevon. Due to the fact that this software is not prepared for flying wing configurations, the calculations
are made for aileron/elevator.
Figure C.1: Calculation of maximum required torque using PredimRC.
96
Appendix D
Stability
D.1 Static Stability
Tables D.1 and D.2 present the stability derivatives for DeltaSpotter V2.
Table D.1: Longitudinal derivatives of DeltaSpotter Table D.2: Lateral derivatives of DeltaSpotter V2
V2 Stability Stability
Value Value
Stability Stability derivative derivative
Value Value
derivative derivative Yβ -1.652 C Yb -0.158
Xu -0.186 CXu -0.0177 Yp -2.766 CYp -0.176
Xw 1.124 CXa 0.107 Yr 0.822 C Yr 0.0523
Zu -3.621 CZu -0.0006 Lβ -2.923 CLb -0.0930
Zw -51.175 CLa 4.886 Lp -24.648 CLp -0.523
Zq -9.149 CLq 4.704 Lr 2.732 CLr 0.0580
Mu -0.00349 CMu -0.00089 Nβ 0.351 CNb 0.0112
Mw -0.811 CMa -0.208 Np -0.579 CNp -0.0123
Mq -0.980 CMq -1.356 Nr -0.0751 CNr -0.00159
D.2 Dynamic Stability
Table D.3: Longitudinal and lateral analysis of DeltaSpotter V1

Eigenvalue Damping ratio ζ Natural Frequency ωn [Hz] T2 [s]
Longitudinal modes
Phugoid -0.0293 ± 1.020i 0.029 0.162
Short-period -19.776±17.385i 0.751 4.191
Lateral modes
Dutch roll -0.3269±3.614i 0.090 0.578
Roll -27.500 0.036
Spiral 0.0645 10.753
97
Appendix E
Structural design
Figure E.1: Top view of the internal structure.
Figure E.2: Lateral view of the internal structure.
98
Appendix F
Computational Dynamic Analysis
Figure F.1: Mesh convergence.
99
Appendix G
Computational Structural Analysis
Figures G.1 and G.2 show the evolution of the Maximum Principal Stress and Minimum Principal Stress
in order of the number of elements.
Figure G.1: Maximum Principal Stress in order of the number of elements
Figure G.2: Minimum Principal Stress in order of the number of elements
100

Detailed design of a forest surveillance UAV

Enviado por

Dados do documento

Descrição original:

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Detailed design of a forest surveillance UAV

Enviado por

Direitos autorais:

Formatos disponíveis

Detailed design of a forest surveillance UAV

Thesis to obtain the Master of Science Degree in

Supervisors: Prof. Fernando José Parracho Lau

2 State of the art 7

4 Structural Design and Analysis 55

A Market Analysis of Military UAVs 91

C Sizing of Control Actuators 96

F Computational Dynamic Analysis 99

G Computational Structural Analysis 100

1.1 Decision matrix, comparison between different surveillance methods . . . . . . . . . . . . 4

2.1 Characteristics of each foam [31–33] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3.1 Data for cruise speed of 12 m/s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

4.1 Summary of relevant parameters for 3D printing. . . . . . . . . . . . . . . . . . . . . . . . 58

A.1 Military UAVs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

D.1 Longitudinal derivatives of DeltaSpotter V2 . . . . . . . . . . . . . . . . . . . . . . . . . . 97

2.1 Categories of UAVs [13] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3.1 General Design Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

A.1 Wingspan vs endurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

B.1 Cl vs. Cd for the Reynolds number of 285000. . . . . . . . . . . . . . . . . . . . . . . . . . 94

C.1 Calculation of maximum required torque using PredimRC. . . . . . . . . . . . . . . . . . . 96

E.1 Top view of the internal structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

F.1 Mesh convergence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

G.1 Maximum Principal Stress in order of the number of elements . . . . . . . . . . . . . . . . 100

ηmotor Motor efficiency.

ηpropeller Propeller efficiency.

µ Atmospheric dynamic viscosity.

σyield Plasticity yield stress.

c Mean aerodynamic chord.

CD Three-dimensional drag coefficient.

CL Three-dimensional lift coefficient.

Cl Two-dimensional lift coefficient.

cP Specific Fuel Consumption.

CD0 Three-dimensional zero lift drag coefficient.

CDi Three-dimensional lift induced drag coefficient.

CLα Three-dimensional lift vs angle of attack curve slope.

ET Total energy stored.

q Lift force distribution.

Qbattery Battery charge.

t/c Airfoil to chord ratio.

UM O Maximum operational speed.

Vbattery Battery voltage.

Wf Final Aircraft Weight.

Wi Initial Aircraft Weight.

climb Climb conditions.

max Maximum conditions.

(a) Burned area in 2017 (b) Burned area 1975-2008

1 Areas with less towers and some obstruction to visibility.

Table 1.1: Decision matrix, comparison between different surveillance methods

Importance (1-10) Control tower CICLOPE Patroling UAV

Reliability of the system 9 4 4 5 6

Low human resources 5 2 7 3 5

Safe for human life 10 4 7 4 7

Overview of the area 9 5 4 3 7

Precise identification of the fire 9 4 4 6 6

Performance independent of terrain 8 4 4 4 7

Performance during the night 9 4 5 5 7

Total 331 386 393 513

Figure 1.3: The first prototype: DeltaSpotter V1

State of the art

2.1 Applications of UAVs

2.1.1 UAS Classification

Figure 2.1: Categories of UAVs [13]