Design and Development of a Hybrid Electric PropulsionSystem for Unmanned Aerial Vehicles
Leonardo Miguel Gonçalves Machado
Thesis to obtain the Master of Science Degree in
Aerospace Engineering
Supervisor: Prof. Afzal Suleman
Examination Committee
Chairperson: Prof. Fernando José Parracho LauSupervisor: Prof. Afzal Suleman
Member of the Committee: Dr. Frederico José Prata Rente Reis Afonso
May 2019
ii
Dedicated to my family
iii
iv
Acknowledgments
I would like to express my sincere gratitude to Prof. Afzal Suleman for providing me the opportunity
to work on this project. I cannot imagine a better way of finishing my degree than by putting my knowl-
edge to the test in a multidisciplinary project such as this one. Furthermore, I would like to thank the
University of Victoria Center for Aerospace Research (UVIC-CfAR) for letting me use their test facilities
and providing funding to this research project.
A very special note goes to Jay Matlock for his brilliant skills, positive spirit and valuable contribution
to this project and to Cameron Pettit for his input on the early steps of this project. To John, a endless
source of knowledge and initiative, for the time spent at the UVIC-CfAR shop setting up the test bench
and all the valuable lessons learned in the process. To the UVIC-CfAR electrical department team,
Kieran, Pablo, Babak, Ali and Grant for their patience and fellowship as they introduced me to the perks
of soldering and helped me dealing with all the complex electronic related problems I faced along the
way. Moreover, I keep a special debt of gratitude to Maxym, for the lessons on machining and mechanical
design and for helping us solve the many mechanical problems this project insisted on raising. A very
special thanks also goes to Roy and Stephen, that despite their tight schedule, always tried to provide
support when things insisted in not working. I would also like to thank all the UVIC-CfAR team members
for their friendship and amazing work environment.
Also, I could not forget all the amazing friends I have made throughout these years. You truly made
this journey a pleasureful one and I am honoured to have shared these years with you.
Finally, to you, my beloved family. Thank you for supporting my choices and keeping me going when
the enthusiasm seemed to lack. Thank you for teaching me values which were instrumental in achieving
my life goals. To my sister Raquel, for being my advisor and best friend, for setting an example of hard
work, perseverance and ambition. Finally, to my parents Elia and Fernando, for the ever lasting support
and love, and for encouraging me to overcome and succeed. This is for you.
v
vi
Resumo
Num contexto de aumento do preco dos combustıveis e de uma consciencia ambiental cada vez
maior, a industria da aviacao procura encontrar alternativas aos sistemas convencionais de propulsao
movidos a combustıveis fosseis. Como a propulsao totalmente eletrica ainda e insuficiente no que diz
respeito a autonomia, os sistemas de propulsao hıbrida-eletrica apresentam-se como uma alternativa
interessante.
Estes oferecem vantagens em relacao aos sistemas convencionais, nomeadamente uma maior
eficiencia energetica, maior autonomia, reducao de assinaturas termicas e sonoras indesejaveis, alem
de abrir portas para inovadoras arquiteturas atraves de propulsao distribuıda. Alem disso, a sua multi-
plicidade de modos de funcionamento oferece ao sistema uma maior versatilidade e redundancia.
No entanto, esta e uma area de investigacao relativamente recente no sector da aviacao, com um
amplo universo de possibilidades ainda por explorar, principalmente no domınio dos metodos de con-
trolo. Nesse sentido, esta tese pretende abordar essas possibilidades, concebendo, construindo e
testando um prototipo de um sistema de propulsao hıbrida-eletrica, em configuracao paralela, para um
pequeno veıculo aereo nao tripulado. Como e usual nestes projetos, os estudos desenvolvidos com
modelos de escala reduzida podem ser considerados como uma plataforma que permite extrapolacoes
para aplicacoes de maior dimensao.
Este documento aborda o projecto mecanico e a instrumentacao de uma bancada de testes, in-
cluindo o desenvolvimento de um sistema automatizado de aquisicao de dados e controlo dos testes, e
o projeto de um controlador do tipo rule-based baseado na linha de operacao ideal. O sistema desen-
volvido possui um motor de combustao interna de 2,3kW (35cc a dois tempos movido a gasolina) e um
motor eletrico de corrente contınua de 1kW. Estes componentes foram escolhidos de forma a serem
representativos daqueles tipicamente encontrados num pequeno veıculo aereo nao tripulado.
Cada modo de funcionamento do sistema hıbrido-eletrico foi testado e caracterizado. Adicional-
mente, o desempenho do controlador hıbrido foi comparado com sua alternativa movida a gasolina, sob
condicoes identicas.
Palavras-chave: propulsao hıbrida-electrica, veıculos aereos nao tripulados, controlador do
tipo rule-based, configuracao paralela, propulsao energeticamente eficiente
vii
viii
Abstract
Against a background of rising fuel prices and an ever-increasing environmental concern, the aviation
industry urges to find alternatives to conventional fossil fuelled powered propulsion systems. As fully-
electric propulsion still falls short as far as operating range is concerned, hybrid-electric propulsion
systems arise as an interesting alternative.
This concept offers advantages over these conventional systems such as increased fuel efficiency,
longer endurance, a reduction in undesirable thermal and noise signatures as well as opening the door
to innovative architectures through distributed propulsion. Furthermore, its variety of operating modes
offer the vehicle increased versatility and redundancy.
However, this is a relatively new research area in the aviation industry with a wide range of possibili-
ties yet to be discovered, mainly on the control methods domain. This way, this thesis strives to explore
those possibilities by fully designing, building and testing a prototype parallel hybrid-electric propulsion
system for a small Unmanned Aerial Vehicle (UAV). As is common in these projects, small scale studies
can be regarded as a stepping stone to large aircraft applications.
This document addresses the thorough mechanical design and instrumentation of a test bench, in-
cluding the development of an automated data acquisition and test control system, using National Instru-
ments LabVIEW, and the design of a rule-based controller based on the ideal operating line concept for
the control of the powerplant. It features a 2.3kW internal combustion engine (35cc gasoline two-stroke)
and 1kW brushless direct current electric motor. These components were chosen to be representative
of those typically found in a small UAV.
Each operational mode of the hybrid-electric system was tested and characterized. Additionally, the
hybrid controller performance was compared to its gasoline-powered alternative through a sample UAV
mission.
Keywords: hybrid-electric propulsion, unmanned aerial vehicles, rule-based controller, parallel
configuration, energy efficient propulsion
ix
x
Contents
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Resumo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix
Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxi
1 Introduction 1
1.1 Motivation and topic overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Literature Review 5
2.1 Application of Hybrid-electric Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Hybrid Electric Propulsion System architectures . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2.1 Series Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2.2 Parallel Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Hybrid-Electric Vehicle Operating Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Controller theories for Hybrid Electric Vehicles . . . . . . . . . . . . . . . . . . . . . . . . 10
3 Concept Development 13
3.1 Performance Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2 Hybrid electric propulsion system components . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2.1 Internal Combustion engine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2.2 Electric motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.2.3 Propeller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2.4 Mechanical coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3 Controller Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3.1 Design Philosophy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3.2 Rule based Controller open loop control strategy . . . . . . . . . . . . . . . . . . . 30
xi
3.3.3 System working points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.4 Controller states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4 Experimental Setup 35
4.1 Test bench . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.1.1 Assembly description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.1.2 Sensors and actuators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2 Graphical User Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.3 Controller implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5 Results 51
5.1 System characterisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.1.1 Electric motor only testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.1.2 Internal Combustion Engine testing . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.1.3 Dash testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.1.4 Regenerative braking testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.2 Controller performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.2.1 Single EM operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.2.2 Single ICE operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.2.3 Hybrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
6 Conclusions 67
6.1 Achievements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.2 Recommendations and future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Bibliography 73
A Hybrid controller code 77
B Technical Datasheets 85
B.1 Desert Aircraft DA35 manufacturer test results . . . . . . . . . . . . . . . . . . . . . . . . 85
B.2 AXi 4130/20 GOLD LINE V2 manufacturer test results . . . . . . . . . . . . . . . . . . . . 88
C Standard Operating Procedures 91
D LabVIEW block diagram 94
xii
List of Tables
3.1 Desert Aircraft DA35 Test results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2 ESC parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3 Axi 4130/20 testing results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.1 Hybrid controller error codes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.1 Electric motor testing results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.2 Internal combustion engine testing results . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.3 DASH mode testing results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.4 Detailed results for DASH mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.5 DASH mode and Electric Motor (EM) ONLY mode comparison . . . . . . . . . . . . . . . 58
5.6 REGEN mode testing results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.7 Detailed results for REGEN mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.8 Fuel consumption results for ICE Single Operation Mission . . . . . . . . . . . . . . . . . 64
5.9 Fuel consumption results for Hybrid Mission . . . . . . . . . . . . . . . . . . . . . . . . . . 66
xiii
xiv
List of Figures
1.1 Energy Density and Specific energy for different fuels. Courtesy of [4] . . . . . . . . . . . 2
2.1 Hybrid electric series configuration flow diagram . . . . . . . . . . . . . . . . . . . . . . . 6
2.2 Launchpoint’s proposed hybrid bus architecture for a VTOL tilt-wing aircraft. Courtesy of
[15] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Launchpoint’s 6kW Gen set. Courtesy of [15] . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.4 Hybrid-electric parallel configuration flow diagram . . . . . . . . . . . . . . . . . . . . . . . 7
2.5 Project Condor’s parallel hybrid electric system. Courtesy of [19] . . . . . . . . . . . . . . 8
2.6 Glassock’s Hybrid-electric prototype [25] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.7 Energy paths on a Hybrid Electric parallel configuration . . . . . . . . . . . . . . . . . . . 10
3.1 Desert Aircraft DA35 and its specifications [30] . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2 Two stroke engines operating cycle. Courtesy of [2] . . . . . . . . . . . . . . . . . . . . . 14
3.3 DA35 Walbro Carburettor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.4 Ignition system. Courtesy of [30] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.5 Ignition system battery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.6 BSFC as a function of Throttle and RPM for the DA35 . . . . . . . . . . . . . . . . . . . . 17
3.7 Fuel flow as a function of Throttle and RPM for the DA35 . . . . . . . . . . . . . . . . . . . 17
3.8 Torque as a function of throttle and RPM for the DA35 . . . . . . . . . . . . . . . . . . . . 17
3.9 Power as a function of Throttle and RPM for the DA35 . . . . . . . . . . . . . . . . . . . . 17
3.10 Desert Aircraft DA35 test setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.11 Fuel flow as a function of RPM for the DA35 . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.12 BSFC as a function of Torque and RPM for the DA35 . . . . . . . . . . . . . . . . . . . . . 20
3.13 Power as a function of Torque and RPM for the DA35 . . . . . . . . . . . . . . . . . . . . . 20
3.14 Ideal Operating Line of the DA35 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.15 BSFC at the Ideal Operating Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.16 AXi 4130/20 GOLD LINE V2 and its specifications [36] . . . . . . . . . . . . . . . . . . . . 21
3.17 Simplified ESC circuit diagram. Courtesy of [37] . . . . . . . . . . . . . . . . . . . . . . . 21
3.18 BLDC motor current waveform simplified schematic. Courtesy of [37] . . . . . . . . . . . 21
3.19 Enertion boards FOCBOX. Courtesy of [40] . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.20 6 cell 8000mAh batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
xv
3.21 Discharge curves of Lithium-polymer battery for various discharge currents. Courtesy of
[38] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.22 Current signal measured with an oscilloscope . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.23 Castle Creations Capacitor pack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.24 DC electric motor equivalent circuit [7] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.25 AXi 4130/20 test setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.26 AXi 4130/20 current versus torque characteristic . . . . . . . . . . . . . . . . . . . . . . . 27
3.27 AXi 4130/20 efficiency test results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.28 Direct current motor torque characteristics. Courtesy of [25] . . . . . . . . . . . . . . . . . 27
3.29 19′′ × 10 propeller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.30 Torque versus RPM characteristic for the 19′′ × 10 fixed pitch propeller . . . . . . . . . . . 29
3.31 Clutch/one-way bearing installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.32 Decision tree (Adapted from [23]) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.33 Hybrid-electric propulsion system controller block diagram (Adapted from [22]) . . . . . . 31
3.34 System operating points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.1 Test bench schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.2 Parallel hybrid electric test bench CAD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.3 Detailed view of the parallel hybrid electric test bench CAD . . . . . . . . . . . . . . . . . 36
4.4 Flywheel design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.5 Propeller shaft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.6 Custom one-way bearing assembly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.7 Parallel hybrid electric test bench . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.8 1:1 PLA bevel gear design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.9 Failed 2:1 PLA bevel gear design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.10 Second design for the starter mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.11 Failure of starter timing belt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.12 Starting setup - final design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.13 Data acquisition printed circuit board . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.14 Voltage divider electric circuit schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.15 Current shunt used for voltage and current measurements . . . . . . . . . . . . . . . . . . 43
4.16 Hall effect circuit schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.17 DA35 installed RPM sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.18 EM RPM sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.19 Load cell installed underneath the fuel tank . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.20 LabVIEW Graphical User Interface of the parallel hybrid-electric test bench . . . . . . . . 47
4.21 Controller block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.1 RPM results for the EM ONLY test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.2 Current results for the EM ONLY test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
xvi
5.3 Voltage results for the EM ONLY test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.4 Efficiency results for the EM ONLY test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.5 Torque as a function of throttle for the AXi 4130/20 . . . . . . . . . . . . . . . . . . . . . . 54
5.6 Torque as a function of Throttle for the ICE ONLY test . . . . . . . . . . . . . . . . . . . . 55
5.7 Power as a function of Throttle for the ICE ONLY test . . . . . . . . . . . . . . . . . . . . . 55
5.8 Fuel flow as a function of Throttle for the ICE ONLY test . . . . . . . . . . . . . . . . . . . 55
5.9 BSFC as a function of Throttle for the ICE ONLY test . . . . . . . . . . . . . . . . . . . . . 55
5.10 Torque as a function of throttle for the ICE . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.11 RPM results for the DASH test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.12 Current results for the DASH test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.13 RPM results for the REGEN test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.14 Current results for the REGEN test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.15 Single EM operation mission profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.16 Propeller torque during Single EM operation mission . . . . . . . . . . . . . . . . . . . . . 61
5.17 Throttle variation during Single EM operation mission . . . . . . . . . . . . . . . . . . . . 61
5.18 RPM results for Single EM operation mission with indication of standard deviation . . . . 62
5.19 Average RPM results for Single EM operation mission . . . . . . . . . . . . . . . . . . . . 62
5.20 Current results for Single EM Operation mission with indication of standard deviation . . . 62
5.21 Average current results for Single EM Operation mission . . . . . . . . . . . . . . . . . . . 62
5.22 Single ICE Operation mission profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.23 Throttle variation for ICE Single operation mission . . . . . . . . . . . . . . . . . . . . . . 63
5.24 Propeller torque during ICE Single operation mission . . . . . . . . . . . . . . . . . . . . . 63
5.25 RPM results for ICE Single operation mission . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.26 Throttle variation for hybrid mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.27 Propeller torque hybrid mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.28 RPM results for hybrid mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.29 Battery Current during hybrid mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.30 Electric Motor Torque during hybrid mission . . . . . . . . . . . . . . . . . . . . . . . . . . 66
6.1 QT1 aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
6.2 Hybrid-electric propulsion system implemented in the QT1 aircraft . . . . . . . . . . . . . 71
xvii
xviii
Nomenclature
Greek symbols
ωm Motor rotational speed.
ρ Density.
τ Torque.
Roman symbols
BSFC Brake specific fuel consumption.
I Current.
I0 Motor no load current.
J Advance ratio.
kQ Torque coefficient.
Kv Motor speed constant.
L Cost function.
M Mach number.
mfuel Fuel consumption rate.
MEP Mean effective Pressure.
nR Number of revolutions per cycle.
P Power.
Rm Motor internal resistance.
Re Reynolds number.
RPM Rotations per Minute.
V Voltage.
Vd Displaced volume.
xix
Vm Motor back electromotive force.
Subscripts
bat Battery.
EM Electric motor.
prop Propeller.
thr Throttle.
tip Propeller tip.
xx
Acronyms
BLDC Brushless Direct Current.
BSFC Brake Specific Fuel Consumption.
CD Charge Depleting.
CDI Capacitor Discharge Ignition.
CMAC Cerebellar Model Arithmetic Computer.
COTS Commercial Off The Shelf.
CS Charge Sustaining.
CVT Continuously Variable Transmission.
DC Direct Current.
DP Dynamic Programming.
EM Electric Motor.
EMF Eletromotive force.
ESC Electronic Speed Controller.
FLC Fuzzy Logic Controller.
GUI Graphical User Interface.
HEPS Hybrid-Electric Propulsion System.
HEUAV Hybrid- Electric Unmanned Aerial Vehicle.
HEV Hybrid Electric Vehicle.
ICE Internal Combustion Engine.
IOL Ideal Operating Line.
xxi
IPM Intelligent Power Management.
ISR Intelligence, Surveillance and Reconnaissance.
MPM Manual Power Management.
NI National Instruments.
PCB Printed Circuit Board.
PID Proportional-Integral-Derivative.
PWM Pulse Width Modulation.
RPM Rotations per Minute.
SOC State Of Charge.
SOP Standard Operating Procedure.
TDC Top Dead Centre.
UAV Unmanned Aerial Vehicle.
UVIC-CfAR University of Victoria Center for Aerospace Research.
VI Virtual Instrument.
xxii
Chapter 1
Introduction
In this chapter, a brief summary over the hybrid-electric propulsion topic is presented alongside a
description of the main objectives of this research project. Additionally, the proposed methodology and
an overview over the thesis layout are also addressed.
1.1 Motivation and topic overview
An UAV can be defined as a remotely piloted or self-piloted aircraft with no onboard crew or passen-
gers capable of performing a given mission [1]. Their lower costs and risk compared to manned aircraft,
and availability in a variety of sizes and capabilities have contributed to an increase in its demand in the
civil and military markets.
In the early steps of powered flight, reciprocating internal combustion engines, or simply, Internal
Combustion Engine (ICE), were the sole propulsion source for aircraft. However, with the advent of
technologies such as the turbo-prop, the turbo-jet and turbo-fan, the ICEs were relegated to power
mostly smaller and slower aircraft, including small UAVs (up to 50kg take-off weight) [2]. Conventional
ICE powered vehicles provide good power-to-weight ratios and long operating range due to the high en-
ergy density of liquid hydrocarbon fuels. However, these vehicles have the disadvantages of a poor fuel
economy, high thermal and noise signatures and environmental pollution. In fact, the thermodynamic
cycle which the ICE operates under has thermal efficiency 1 limited to approximately 60% in modern
engines [3]. This is the thermodynamic limit for a perfect engine with no friction or heat losses. In prac-
tice, engines suffer significant losses and their efficiency also depend on the rotating speed and load.
Tipically, efficiencies between 5% and 25% are common [3]. Moreover, the central issue in designing
a conventional aircraft propulsion system is the sizing of the powerplant to the maximum power phase:
take-off and climb. During the cruise and loiter phases, which usually occupy a much longer portion of
the mission time, the power demand is lower and the powerplant runs in partial power. This goes along
with energy losses and unsatisfactory efficiency for most systems.
A popular alternative is the EM, which is capable of operating with a relatively high efficiency (com-
1Thermal efficiency is defined as the ratio of the net work output over the energy added to the system by heat
1
monly between 50% and 90%) throughout a large range of rotational speeds. Furthermore, the EM
offers a lower vibration and noise operation as well as a high torque across the rotational speed operat-
ing range. Additionally, unlike the combustion engine, the performance of the motor is not as sensitive to
environmental variables such as air density, temperature, etc [4]. However, these benefits are negated
by the low energy density and specific energy of the power storage system, in most cases a battery.
Indeed, in this regard, commercially available batteries are not competitive with fossil fuels as shown
in Figure 1.1. This induces severe weight penalties for electric powered vehicles resulting in a short
operating range and far less competitive than ICE powered vehicles for long range applications.
(a) Specific Energy (b) Energy Density
Figure 1.1: Energy Density and Specific energy for different fuels. Courtesy of [4]
A way of overcoming the shortcomings of both powerplants is to integrate an ICE with an EM to form
a Hybrid-Electric Propulsion System (HEPS). In the majority of these propulsion systems, the primary
production of power comes from a fossil fuel powered engine unit that is supplemented by some form of
electric energy through either batteries, fuel cells or solar panels. Within the framework of this research,
a HEPS is defined as a combination of battery-powered EM and an ICE. The design philosophy behind
a hybrid-electric propulsion system is to improve the efficiency of the overall propulsion system when
compared to traditional fossil fuel powered solutions while providing a longer operating range than the
alternative fully electric vehicles. Furthermore, they offer several advantages over these systems includ-
ing lower emissions than ICE powered vehicles, higher versatility in operating modes and redundancy in
case of failure.
It accomplishes it through three main drives. First, through the dual use of the ICE and the EM
during high power demanding manoeuvres such as take off and climb phases. This permits a better
balanced torque share between the two power units, which in turn enables each one to work closer
to their highest efficiency operating regions. Moreover, during low power demanding flight phases, the
excess power from the engine may be utilised to charge the onboard batteries. Second, allows the sizing
of the combustion engine for its longer flight phase, cruise, and thus maximizing its efficiency. Finally,
shorter use of the ICE since portions of the missions may be run with only the motor.
However, the aforementioned benefits come with an increased complexity in the powertrain design
2
together with the need of proper coordination between the different operating modes. This results in
the need for an overall vehicle system control strategy that is significantly more sophisticated than in an
ordinary vehicle. This an intricate task due to the interaction of electrical, mechanical, thermodynamic,
and electrochemical devices on the same system. Furthermore, it is important to keep in mind that a
hybrid powerplant may be heavier than its equivalent gasoline propulsion system due to the added mass
of the batteries. In fact, there is a turning point where the increase in efficiency is negated by its overall
higher weight as outlined by Matlock et al. in [5].
From a research and development perspective, small size UAVs can be regarded as stepping-stone,
enabling experimental scaling studies for larger aircraft applications. In fact, as air traffic growth is
expected to continue to increase at a rate up to 5% per year [6], using only fossil fuel powered aircrafts
would lead to an increase in emissions and depletion of fossil fuel reserves.
1.2 Objectives
This research work aims to evaluate the major challenges as well as advantages of a parallel HEPS.
This way, a prototype powerplant was designed and then implemented on a test bench in order to fully
test it. Primarily, it aims to demonstrate a proof-of-concept of the proposed propulsion system. Secondly,
the designed test bench intends to be built in a modular way, where the different HEPS components can
be switched, allowing a fully versatile test platform.
Overall, the following research questions are raised in this research project:
1. Can a HEPS yield overall effectiveness advantages for small UAVs?
2. Can a HEPS yield energy efficiency advantages for small UAVs?
3. What are the main challenges in implementing a functional hybrid system?
It is important to define the performance metrics used to evaluate effectiveness and energy efficiency
in this context. In the scope of this research project, effectiveness can be understood as reliability, safety
of operation as well as the adaptability and flexibility that the different operating modes offer. The energy
efficiency metrics used, on the other hand, are constrained to the overall fuel consumption of the system.
1.3 Methodology
The procedures presented hereafter are intended to result in a possible approach on how to design,
build and test a HEPS.
Firstly, the conceptual design was addressed. This began with outlining the powertrain layout and
the overall definition of the test bench sensor suite. In order to communicate with the hardware, con-
trol, monitor and log data from the tests, an automated data acquisition system was programmed using
National Instruments LabVIEW software. At this point, a HEPS controller was also designed and imple-
mented into the LabVIEW 2018 block diagram.
3
Afterwards, the mechanical design of the test bench was developed. This included the thorough
design of the powertrain, along with the full instrumentation of the test bench. Large amounts of redesign
and iteration were conducted in order reach the design presented in this document.
Lastly, a test campaign was planned. At first, component and level testing were conducted, where the
different components of the propulsion system were tested individually, comparing the data retrieved with
the one provided by the respective manufacturers. Then, prototype HEPS performance was assessed,
by testing every operational mode and then using the developed controller.
1.4 Thesis Outline
As per the organizational structure, this dissertation is divided into the following chapters:
Chapter 1 - Introduction: In this section, the main motivating factors for the research are presented
alongside a brief overview on the topic. Following such, the objectives intended to be achieved on this
project are defined and the proposed methodology discussed. In the end, a brief outline of the document
is presented.
Chapter 2 - Literature review: This chapter aims to present a comprehensive study over the current
state of the art of hybrid-electric vehicles, with, nonetheless, a special focus on UAV implementation. It
addresses the different application of hybrid-electric power, possible architectures, operating strategies
and controller theories.
Chapter 3 - Concept Development: The chapter has its goal on the development of the prototype
HEPS. It begins by first outlining the performance requirements for such a system, followed by a descrip-
tion and testing of each individual component of the system. It concludes by discussing the controller
design.
Chapter 4 - Experimental Setup: This presents the experimental apparatus developed. It goes
through the mechanical design and sensor integration of the test bench. It also describes the Graph-
ical User Interface (GUI) programmed using LabVIEW and concludes with the details of the controller
implementation.
Chapter 5 - Results: The results obtained during the test campaigns are presented. This comprises
two main sections. The first one addresses the system characterisation where each operational mode
of the HEPS is fully mapped. The second one focuses on the controller performance.
Chapter 6 - Conclusion: Considering the results obtained, this chapter summarizes the achieve-
ments of this research effort and addresses the research questions raised in Chapter 1. In the end,
detailed recommendations for future work are presented.
4
Chapter 2
Literature Review
This chapter is intended to clearly expose and investigate the current state of the art of Hybrid-
Electric Unmanned Aerial Vehicle (HEUAV). This ranges from different applications of hybrid-electric
power, hybrid-electric architectures, powertrain operating strategies and controller theories.
2.1 Application of Hybrid-electric Power
The concept of a HEPS is not new. In fact, early designs of a Hybrid Electric Vehicle (HEV) date
back to 1899 when two hybrid vehicles were presented at the Paris Salon [7]. However, it was not unit
1997 that HEVs became widely available to the general public with the release of the Toyota Prius and
later on, in 1999, with the Honda Insight. Both of these automobiles demonstrated that hybrid vehicles
were feasible and, more importantly, operate more efficiently than its gasoline powered brethren. These
results sparked an interest in pursuing the same approach in different industries.
Nowadays, it is possible to see ongoing research and application of hybrid-electric technologies to,
for instance, city buses, trucks ([8]), locomotives ([9]) and vessels ([10]). Its implementation on the large
scale aviation industry, however, is still taking its first steps with a few ongoing experimental projects.
An example is the E-fan X project, a partnership between Rolls-Royce, Airbus and Siemens that aims
to work on a hybrid-electric technology flight demonstrator that should take the skies in 2020 [11]. More
recently, in October 2018, Diamond Aircraft Industries in collaboration with Siemens AG announced the
first flight of a jointly developed multi-engine aircraft [12].
Regarding UAV applications, the remaining of this chapter provides insight on the technology state
of the art on this industry.
2.2 Hybrid Electric Propulsion System architectures
There are several possibilities of combining an ICE and an EM to form a HEPS in the framework of
UAV integration.
For instance, Hiserote [13], suggests using two propellers in a centerline-thrust configuration. This
5
architecture featured an ICE powered tractor propeller installed in the front of the aircraft while a second
one would be powered by an EM and featured in the back. In this case, the battery pack would be
charged by wind-milling the second propeller when cruising to turn the motor/generator. This however,
showed a low recharging efficiency.
Nevertheless, the majority of the hybrid systems architectures, can be grouped in two main configu-
rations: series and parallel.
2.2.1 Series Configuration
In a series configuration, the ICE and the EM are not mechanically coupled. Instead, the engine
is attached to a generator that can either directly power the motor or charge the onboard batteries as
illustrated in Figure 2.1.
Figure 2.1: Hybrid electric series configuration flow diagram
This means the ICE can be left to operate at its optimum torque and speed range, regardless of
the driving conditions, in the execution of its role supplying power to the electric grid from which the
EM draws power to propel the aircraft. Furthermore, a possible use of a series configuration is through
distributed propulsion. In this innovative concept, a single combustion engine and generator, located
anywhere in the architecture, can be used to drive multiple propeller motors independently controlled.
This cutting-edge concept offers promising solutions to further enhance the aerodynamic, propulsive
and structural efficiency of an aircraft [14]. In fact, the American company Launchpoint Technologies
is developing a new concept which they call ”Propulsion-by-wire”, in a clear analogy to the Fly-by-wire
system [15]. Their idea, illustrated in Figure 2.2 is for power to flow through wires from the energy
source to the propulsion units instead of through mechanical connections - pneumatics, pistons, and
shafts. An example of a genset developed by Launchpoint is shown in Figure 3.18. This unit features
2-stroke gasoline engine and alternator making up a total weight of 4.12kg outputting continuously 6kW
at 7500RPM .However, the series configuration requires bigger and thus heavier electrical machines to accommo-
date the peak power demands. This represents a significant weight penalty. Additionally, since a series
hybrid has multiple stages of energy conversion, it suffers from substantial losses [16]. The mechanical
energy available in the engine is converted to electrical energy through the generator and then back to
mechanical energy by the electric motor to propel the aircraft. In fact, a study by Harmats and Weihs
concluded that series HEPSs were not effective for use in UAVs due to large power losses [17].
Nonetheless, this configuration is mostly used in high-torque, low speed, large vehicles such as
buses, commercial trucks and locomotives [18].
6
Figure 2.2: Launchpoint’s proposed hybrid bus architecture fora VTOL tilt-wing aircraft. Courtesy of [15]
Figure 2.3: Launchpoint’s 6kWGen set. Courtesy of [15]
2.2.2 Parallel Configuration
Parallel hybrid electric configurations couple the ICE and EM together through some form of me-
chanical coupling. This way, the power requirement can be full fulfilled by either the sole action of the
combustion engine, the electric motor, or a combination of the two as illustrated in Figure 2.4.
Figure 2.4: Hybrid-electric parallel configuration flow diagram
Since both the ICE and EM provide power to the drivetrain, the size of each component is reduced
as neither needs to provide the maximum power required by the vehicle [19].
Also there is requirement for only one electric machine (that can act either as a motor or generator) in
contrast with the series architecture where at least two electric machines were required. This results in
an overall lighter solution. In fact, in a study conducted by Harmon, Frank and Joshi it was estimated that
a parallel configuration on a small UAV would be approximately 8% lighter than a series configuration
[20]. For this reason, parallel hybrid was initially preferred for the aircraft field.
In this configuration, however, the ICE may potentially not be operated at its most efficient point (since
it is directly coupled to the propeller through a transmission, therefore limiting the energy efficiency).
In fact, for an optimal propeller driven aircraft, the propeller and its power source should operate at
their peak efficiencies, which do not usually coincide. In order to mitigate this problem, Rotramel, [7],
7
suggests the use of gearing to better align the shaft’s rotational speed with that due to the torque demand
of the propeller. A study conducted by Hung [1] also implements a Continuously Variable Transmission
(CVT) in order to minimize this issue. Furthermore, work from Brace et al., [21], suggests the use of
electronic engine control together with a CVT to allow the pilot’s power demand to be implemented in
the most advantageous manner. This means the optimiser would determine the instantaneous ideal
engine speed and ideal engine torque to deliver the power demanded by the driver. In another study
from Koster et al., a standard planetary system was used to address this question and combine power
from the ICE and the EM to a single propeller shaft. This introduces complexity to the system controller
strategy, as well as more weight, more cost, and possibly less reliability.
One of the first major parallel hybrid-electric propulsion UAV projects began in 2010 with the Air Force
Institute of Technology’s ’Project Condor’ [22], [19], [13], [23]. This project investigated different config-
urations and control theories, managing to successfully implementing, and ground and flight testing a
parallel hybrid-electric UAV (see Figure 2.5).
(a) Hybrid system on dynamometer (b) Assembled Hybrid system on airframe
Figure 2.5: Project Condor’s parallel hybrid electric system. Courtesy of [19]
Furthermore, the University of Colorado has shown success with their project HELIOS which also
made use of a parallel hybrid architecture in their UAV with a projected 15% increase in fuel efficiency
over traditional gas powered UAVs of similar size [24].
Glassock at Queensland University constructed a prototype system with a 10cc metahnol two-stroke
ICE and a 600W Brushless Direct Current (BLDC) motor (see Figure 2.6). The system was tested on a
dynamometer in a wind-tunnel and proved an improvement in overall propulsive efficiency of 17% when
compared to a non-hybrid powerplant [25].
Possible working modes
Ehsani et all [26], and Ausserer [19] define four working modes for a parallel HEV (refer to Figure
2.7).
ICE ONLY mode is represented by path 1. On this mode, the engine is the sole source of power to
8
Figure 2.6: Glassock’s Hybrid-electric prototype [25]
the aircraft and is used whenever requested power is inside the region of higher efficiency of the ICE.
Regen mode is depicted by path 2 and represents the regenerative mode where the ICE is used
for both providing power to the propeller and also charge the on board batteries packs through the
generator/motor.
Another possible mode is EM ONLY, pictured by path 3, and is the pure electric propelling mode, in
which the engine is shut off or placed idling. In most cars this allows for low-speed operation, while in
aircraft this can differ. This mode offers the possibility of low noise and thermal signature flight. This
can be useful for Intelligence, Surveillance and Reconnaissance (ISR) portions of a mission, where
a stealth characteristic might be necessary. Another interesting opportunity arises in the use of this
mode for takeoff, minimising noise pollution near airports. However, this would require a sizing of the
EM for this high power demanding mission segment, what could lead to an increased weight of the
aircraft. Additionally, a way to restart the ICE during flight might be of interest. Conceptually, this
would completely eliminate thermal or noise signatures of the engine. However, Greiser, [23] suggests
idling the engine instead of shutting it down due to difficulties with in-flight restart. The aforementioned
author adds that an excess of fluid in the cylinder may impede the spark from actually igniting the fuel.
Additionally, as far as noise is concerned, Rotramel defends that the noise produce by idling the ICE is
far less than the noise that a 18inch propeller makes [7].
Finally, DASH mode is represented by the sum of path 1 and 3, where both power sources provide
power to the propeller and maybe used in high power demanding flight phases such as take-off or climb.
In this mode, the ICE may be left operating at its most efficient operating point while the EM provides
the rest of the requested power, allowing for a fuel efficient flight. Or, on the other hand, provide a rapid
increase in power when the EM aids the ICE in its normal operation.
9
Figure 2.7: Energy paths on a Hybrid Electric parallel configuration
2.3 Hybrid-Electric Vehicle Operating Strategies
In addition to the two primary HEV configurations, Harmon et all [20], define two possible operating
strategies for energy management of an HEV: Charge Sustaining (CS) and Charge Depleting (CD). Both
strategies can be used in a series or parallel configurations.
The CS strategy attempts to maintain the battery State Of Charge (SOC) at a certain level. In con-
trast, the CD strategy allows the SOC to decrease maximising the energy use from off-board charging
[20].This strategy is commonly found on electric scooters or bicycles used in the cities where, after each
use, they are placed on a charging station. This strategy would fit perfectly for a UAV mission, where the
user would have to plug in the aircraft to an external power source in between flights. Despite that, this
strategy would require a relatively large battery pack and thus, due to weight limitations, Harmon et al.
argues a purely CD strategy would not be feasible for a HEUAV [20].
Nonetheless, a completely CS strategy would limit the EM ONLY mode duration (as it would not
allow for the SOC to decrease below a certain level) and thus a combination of both these strategies is
suggested. For instance, a CS strategy could be implemented for the first half of the mission, and then
CD could be used until the battery SOC drops to a pre-determined level.
Finally, it is important to note that the most suitable operating strategy is highly dependent on the
mission profile and the aircraft design goals.
2.4 Controller theories for Hybrid Electric Vehicles
The choice and design of a control strategy plays a crucial role in optimising HEV technologies. HEVs
have two (or possibly more) power sources and complex power paths. Because of these complexities,
it is necessary to use a high level controller, the so-called vehicle supervisory controller [27]. The main
role of this supervisory controller is to determine the torque or power demand of the engine, the motor,
and the regenerative braking according to the pilot or autopilot throttle input, in order to maximise the
propulsion system efficiency.
Much research has been conducted on HEV control strategies, heavily focusing on automotive ap-
plications but not many have been applied to HEUAV.
10
Conceptually, the simplest control strategy to implement is a rule based control strategy. This con-
troller is, at its core, simply composed by a set of rules that establish criteria for switching between
different operational states. Due to its simple and easy design, it is able to rapidly make calculations
and decide on an action. For this type of controllers an attempt is made to operate the ICE and the
EM in their most efficient regions. However, since these regions often do not coincide, trade-offs are
imperative. In fact, the Ideal Operating Line concept would find its ideal application on such a control
strategy as outlined by Hung and Gonzalez [1]. The Ideal Operating Line, also called ”economy line”, is
a line made up of all the points which represent torque and speed combinations where the Brake Spe-
cific Fuel Consumption (BSFC) is minimal on different power lines for steady-state conditions. A power
plant operated on the Ideal Operating Line (IOL) will, theoretically, enable the best performance while
consuming the least amount of fuel possible. Their research, included the simulation of a parallel HEPS
for a small fixed-wing UAV by incorporating an IOL control strategy and a CVT as a mechanical coupler
between the two powerplants. The results showed fuel savings up to 6.5% compared to the engine-only
configuration [1].
Further simulation work from Harmon in [22], implemented a rule-based controller based on ideal
operating line concept and proved a 54% and 22% energy reduction in an one-hour and three-hour ISR
mission for a 13.4kg HEUAV when compared to a four-stroke gasoline powered UAV.
An alternative to rule-based controllers are Fuzzy Logic Controllers. In this case, the logical variables
takes values between 0 and 1 instead of just 0 or 1 (as in the case of rule based controllers). Salman et
al., [28], describe Fuzzy Logic Controller (FLC) as very suitable method for realising an optimal trade-off
between the efficiencies of the HEV components as well as robust, because of its tolerance to imprecise
measurements and component variability. A research conducted by Karunarathne et al., [29] included a
FLC for a fuel cell/battery system within the framework of UAV hybrid propulsion.
However, most of the research conducted thus far on these controllers is focused on the instanta-
neous optimal operation, not taking into account future mission possibilities of charging or discharging
the batteries to assist the engine, and thus failing to fully explore the HEV capabilities.
In this aspect, optimal control fills this gap by finding the global optimal operation for a certain tra-
jectory or mission profile. It finds the global optimal route of energy consumption minimisation but it is
hard to be realised in real-time as it requires the future state and heavy calculation load. However, it
shows the maximum benefit of the HEV, and can thus be used as an index to compare the performance
of other controllers. Indeed, on his work, Harmon, [20], uses an offline optimisation routine. This routine
uses the demanded torque, the rotational speed and the battery SOC as inputs to an algorithm that aims
to minimize a cost function, which, in this case, takes the form of the total power consumption:
L = PICE + αPEM + βPEMrecharge(2.1)
where PICE is the power consumption of the ICE to rotate the propeller (33.44kWh/gal of gasoline),
PEM is the electrical power consumption of the EM and PEMrechargeis the power consumption equivalent
for the ICE to operate the EM as a generator to recharge the batteries. The weighting factors, α and
11
β penalise the amount of electricity use and the amount of recharging, respectively, and are mission
dependent. The nonlinear control surface generated by this algorithm was then approximated using a
Cerebellar Model Arithmetic Computer (CMAC) neural network. Simulation results show that a HEUAV
with the CMAC controller used 37.8% less energy than a two-stroke gasoline powered UAV during a
three hour ISR mission.
In another attempt to tackle the disadvantages of global optimal control Cho, [27], suggests the use
of a predictive control strategy, based on offline data. It consisted of algorithms based on the database
gathered from real flight data. Therefore, the performance of these controllers is only as good as the
quality of the data base.
On a research conducted by Lin, Peng and Grizzle, [8], a near-optimal power management strategy
was implemented. The cost function consisted on minimising fuel consumption and selected emission
species over a driving cycle. Dynamic Programming (DP) was used to find the optimal power split
between the engine and motor whilst subject to the battery SOC-sustaining constraint. Through the
analysis of the behaviour of DP control actions, near-optimal rules were extracted, which, unlike DP
control signals, are implementable.
12
Chapter 3
Concept Development
This chapter addresses the development of the prototype parallel HEPS. This way, the performance
requirements are first outlined followed by a discussion of each individual component that makes the
hybrid system. This chapter concludes by detailing the controller design.
3.1 Performance Requirements
This proof-of-concept system aims to depict, as close as possible, a system able to be implemented
into an aircraft. This way, the author defined qualitative and conceptual performance requirements based
on typical UAV and aircraft missions.
A typical mission profile would consist of take-off and climb without any form of auxiliary launch
equipment using power provided by the ICE or the EM. The second phase would include cruise to a
location of interest using the engine. A loiter and land segment would follow using power provided by
the motor. This way, it is expected that the system to be able to operate under each one of these modes,
while striving to optimize its efficiency.
Additionally, another requirement is the use of Commercial Off The Shelf (COTS) components when
building the system. COTS parts are faster and less expensive to obtain than their custom manufactured
counterparts, making them preferable for a 6 months research effort.
3.2 Hybrid electric propulsion system components
The main components which comprise a HEPS are:
• Internal Combustion Engine
• Electric motor
• Propeller
• Mechanical coupling
13
This section describes in detail each of the components used as well as evaluating its performance.
The HEPS developed in this project was designed and sized taking into account the typical power
requirements of small UAV (under 50kg take-off weight). This way, each component was selected to be
representative of what would typically be found on an usual UAV. A thorough sizing is, however, out of
the scope of this research effort.
Despite an ever growing interest in small UAVs, these propulsion system find its most use in the
hobby market. This way, all the equipment chosen is available as a COTS commodity.
3.2.1 Internal Combustion engine
The ICE used in this project was a Desert Aircraft DA35 spark-ignited 35cc two stroke single cylinder
engine. Its specification can be found in Figure 3.1.
Maximum power 2344W (3.14hp)Maximum Torque 2.99N ·m @ 5500RPM
Displacement 35ccWeight 0.953kgLength 161mm
Minimum RPM 1500Maximum RPM 8200
Type combustion cycle OttoNumber of cylinders Single
Figure 3.1: Desert Aircraft DA35 and its specifications [30]
A two stroke engine allows for a higher power-to-weight ratio and simpler valve design compared to
the more common four-stroke engine and is thus an usual choice in the small UAV market. In this design,
ports in the cylinder liner, opened and closed by the piston motion, control the exhaust and inlet flows
as shown in Figure 3.2. This results in a complex gas exchange process into and out of the cylinder. As
a consequence, it is rather difficult to completely fill the displaced volume with fresh air, and even some
of the fresh mixture flows directly out of the cylinder. This in turn, results in a lower efficiency compared
to four-stroke engines.
Figure 3.2: Two stroke engines operating cycle. Courtesy of [2]
14
Additionally, this engine features a carburetor, shown in Figure 3.3. These systems are simpler,
cheaper, require less maintenance and last longer when compared to fuel injected ones. However, the
amount of fuel that flows into the cylinder is only dependent on the amount of air that can be pulled into
the carburetor venture, controlled by the throttle and choke valves. A key aspect with this design is that
it does not monitor the air to fuel ratio going into the cylinder. Therefore, the ratio which offers the best
performance is only approximated. This in turn results in a lower power output, poorer fuel efficiency
and higher emissions when compared to fuel injection systems.
This carburetor also includes tuning pins: a ”High RPM” and a ”Low RPM” adjust needles. This
helps adjusting the richness of the air/fuel mixture. The recommended configuration according to the
manufacturer (see [31]) was set as a starting point followed by small 14 turn adjustments until a good
performance was reached. This consisted in obtaining a smooth idle and a reliable transition to high
throttle, without leaning the mixture anymore than necessary to avoid overheating of the engine. Running
the engine too rich however, reduces power and creates other problems such as pre-mature carbon build
up, excessive exhaust residue, fouled spark plugs and overall rough running. A fair amount of experience
and trial-and-error attempts were needed until a satisfactory point was reached.
Figure 3.3: DA35 Walbro Carburettor
Another important aspect to take into account is the spark plug gap. This was continuously monitored
to assure it set between the range defined by the manufacturer, between 0.38mm to 0.5mm. This
is important because a gap that is too small may cause the engine to start ignition too early on the
compression stroke whereas a gap that is too wide may cause the engine to skip firing.
Fuel and lubrification
The fuel used was 94 octane gasoline following the instructions of the manufacturer. The gas was
mixed with 2-stroke engine oil in a 50 to 1 mix ratio so as to protect the engine from high heat and high
RPMs that lead to piston deposits, ring sticking and scuffing of the cylinder walls.
15
ICE Set up
Given the high vibrations of this engine, and the dangers associated with it, all the engine mounting
bolts were assembled using LOCTITE R© high strength threadlocker. Additionally, their tightness was
checked prior to every test.
An important aspect of the ICE is its starting settings. First, the fuel line was primed, and, as a
consequence the carburettor filled with fuel, assuring that there were no air bubbles in the line. This was
done by spinning the crankshaft with an auxiliary electric starter motor while having the throttle and the
choke valves completely closed. Afterwards, the choke valve was fully opened, the throttle set to 15%
and the starter ran. When the engine was warm, the author verified that it would start with relative ease
using the aforementioned parameters. When the engine was cold, on the other hand, a fair amount of
experience is involved in starting it and the process revealed to be more difficult and hard to standardise.
Once the engine shaft is continuously rotating, a higher throttle value, up to 35% was needed for the
engine to fire and start its self sustaining cycle. When it finally was continuously and consistently firing,
the throttle was set to idle, and left running for about 1min before advancing the throttle, so it would warm
up. Another important consideration concerns the restart of the engine. Indeed, after the stoppage of the
engine, the author verified that the cylinder would be filled with gas. As the rotational speed decreases
and the engine stops firing, the cylinder pressure still sucks fuel from the carburettor, filling the cylinder
with unburnt fuel. This excess fluid would prevent the spark plug from igniting. This way, a interval of
5 to 10 minutes between tests were deemed necessary for a sucessful engine start. Therefore, and as
previously outlined in section 2.2.2, in-flight restart faces severe obstacles.
The ignition system features Capacitor Discharge Ignition (CDI) which includes a microprocessor
that calculates the shaft speed and position accurately through multiple magnets mounted on the engine
hub and a bipolar hall sensor (Figure 3.4). This ignition system is powered by a 4 cell nickel cadminum
(NiCad) battery (Figure 3.5). A safety switch was placed in between the battery and the ignition system
to allow emergency stoppage of the spark plug, and thus, the engine.
Figure 3.4: Ignition system. Courtesy of[30] Figure 3.5: Ignition system battery
16
DA35 performance
Some important metrics when analyzing the performance of a ICE are the output Power (P ) in [W ]
defined as:
P =τ · 2π ·RPM
60(3.1)
where the τ is engine output torque measured in [N · m], and the BSFC, commonly evaluated in
[g/(kW · hr)], defined as:
BSFC =mfuel
P(3.2)
where mfuel is the fuel consumption rate in [g/hr] (but commonly evaluated in [g/min]) and P the
output Power in [kW ]
Presented below, from Figure 3.6 to Figure 3.9, is the manufacturer supplied performance data. In
Section B.1 can be found the raw data.
Figure 3.6: BSFC as a function of Throttle andRPM for the DA35
Figure 3.7: Fuel flow as a function of Throttleand RPM for the DA35
Figure 3.8: Torque as a function of throttle andRPM for the DA35
Figure 3.9: Power as a function of Throttle andRPM for the DA35
The torque and hence power of this type of engines depend on the ambient air density, therefore the
17
data shown is only valid for a specified ambient condition.
As depicted in Figure 3.6, the engine finds its most efficient point (lowest BSFC) at high throttle
percentages and high rotational speeds. This characteristic will be further developed in Section 3.2.1.
As expected and common among ICE, the DA35 is most inefficient at low throttle settings and low
speeds. From Figure 3.7, it can be verified that fuel flow increases with both the Rotations per Minute
(RPM) and throttle position, which goes in line with the expected behaviour of carburetor engines.
As it can be seen from Figure 3.8, as we sweep the RPM at a constant throttle position the curve as a
maximum of between 4500RPM and 6000RPM where the maximum torque is achieved. The maximum
torque takes the value of 2.99N ·m and occurs at full open throttle when the engine is running at around
5500RPM . Moreover, as we reach lower rotational speeds, the maximum torque available decreases
substantially. This behaviour is characteristic of combustion engines and is one of the reasons auto-
mobiles are featured with gearboxes. This represents an important difference to electric motors, which
have a more constant available torque throughout the rotational speed operating range, as previously
discussed in section 1.1.
There are several challenges when accurately modelling and controlling ICEs. As with any COTS
component, manufacturer data might be inaccurate for the particular unit in hands. Without compre-
hensive testing to provide an accurate map of the engine capabilities, the controller must do its best to
estimate the amount of torque and power the engine provides.
Torque estimation is comparatively easy on larger hybrid-electric systems such as cars and buses.
However, the methods commonly applied require multiple sensors that are not feasible for small UAVs.
An alternative method described by Khiar in [32] uses the rotational speed and the Top Dead Centre
(TDC) positioning, which are readily available in a car. While the assessment of the rotational speed is
feasible on a small UAV, the measurement of the TDC position raises severe difficulties as outlined by
Wilson et al. [33]. Moreover, this method requires the a priori knowledge of the engine inertia, value that
is not usually provided by small engine manufacturers. However Optrand, [34], developed a sensor for
measuring cylinder pressure that might be helpful. As Cadou [35] suggests, a torque estimation would
be possible through the use of this sensor along with a basic equation for mean effective pressure:
τ = MEP × Vd2× nR × π
(3.3)
where τ is the engine torque, MEP is the mean effective pressure, defined as the work per cycle
per unit displaced, Vd is the displaced volume of the cylinder of the engine and nR is the number of
revolutions per cycle (two on a two-stroke engine).
Despite the data provided by the manufacturer, the author considered important to address and
evaluate the performance of this particular unit. The test was conducted using a 22′′ × 10 propeller
and its setup shown in Figure 3.10. So as to obtain accurate fuel consumption values, the ICE was
left running for 5min at a constant throttle setting. The values can be compared to the ones obtained
by the manufacturer available in Appendix B.1. The main goal of these tests was to obtain a baseline
results trends in the engine behaviour, allowing a comparison with the manufacturer data and later tests
18
conducted on the hybrid powertrain.
Throttle RPM Fuel Flow [g/min] Manufacturer data [g/min] Relative difference [%]
10% 3323.0 4.91 - -20% 4029.9 12.97 3.64 71.92
30% 4293.9 13.19 5.67 56.94
40% 5033.5 19.83 8.88 55.21
50% 5698.7 24.57 13.20 46.26
60% 5800.3 21.26 14.68 30.92
70% 5828.1 22.59 15.51 31.30
Table 3.1: Desert Aircraft DA35 Test results
Figure 3.10: Desert Aircraft DA35 test setup
Figure 3.11: Fuel flow as a function of RPM for theDA35
As it is possible to see from Table 3.1 and Figure 3.11, the manufacturer data underestimates the
fuel consumption of the engine. Nevertheless, both curves present the same trends with a constant
increase in fuel flow with both the RPM and Throttle percentage. Overall, in all working points, it was
verified that the engine ran smoothly, holding a relatively constant speed. Additionally, the engine was
able to work with throttle percentages as low as 10%. Below this point, however, the rotational speed
rapidly decreased and the engine stalled. A steady increase in RPM was verified as the throttle vale was
progressively opened until about 50%. After this point, the increase in speed was minimal and no major
differences were verified as the throttle increase past 70%.
Ideal Operating Line analysis of the DA35
The manufacturer data presented in Section 3.2.1 provided the base to develop the IOL concept
implemented on this controller.
A method of calculating the IOL consists in plotting lines of constant power (Figure 3.13) on the BSFC
map (Figure 3.12)and then search along each iso power line for the lowest BSFC. Doing this in steps of
a few Watt, it is possible to connect the dots and trace an operating line that delivers the lowest BSFC
for any given power. [21]
19
Often the IOL points do not form a smooth line (such was the case here) mainly due to the limited
number of data points available in the engine map as well as the very small fluctuations of fuel con-
sumption in a rather large area. In order to obtain a workable IOL, a smooth line was fitted through
these points. In this case, a 3rd order polynomial function was created for this line. The obtained IOL is
shown in Figure 3.14.
Figure 3.12: BSFC as a function of Torque andRPM for the DA35
Figure 3.13: Power as a function of Torque andRPM for the DA35
Figure 3.14: Ideal Operating Line of the DA35 Figure 3.15: BSFC at the Ideal Operating Line
As per shown in Figure 3.15, the engine operates most efficiently at high RPM and high torque
values (lower BSFC values). It presents a clear decrease in BSFC as the RPM increases which shows
that the engine is most efficient at high RPM. For this engine in particular, the lowest BSFC and highest
efficiency region is located between a rotational speed of roughly 4000RPM to 6000RPM and a torque
from approximately 1.7N ·m to 2N ·m as shown in Figure 3.12. It is also important to note that the ICE
is most inefficient at low speeds in general and also at high speeds but providing a low torque.
20
3.2.2 Electric motor
The electric motor used was an AXi 4130/20 GOLD LINE V2 outrunner brushless direct current
motor. Its specifications can be seen in Figure 3.16.
Max. power 1650W (2.2hp)Number of poles 14
Weight 410gKv 305RPM/V or 31.94rad/(V · s)
Max. efficiency 90%Max. motor current 55A
Max. generator current 40AInternal resistance, Rm 0.99Ω
No load current 1.1ANo. of cells 6-8 Li-Poly
Figure 3.16: AXi 4130/20 GOLD LINE V2 and its specifications [36]
This type of motor does not require brushes for commutation like classic Direct Current (DC) motors
making them very reliable. They instead rely on switching power electronics, usually MOSFET transis-
tors arranged in half bridge or full bridge configuration, to enable digital commutation, converting the
direct current input into an approximated sinusoidal three-phase current using transistors as depicted in
Figures 3.17 and 3.18. The device responsible for this commutation is commonly know as an Electronic
Speed Controller (ESC). The basic function of an ESC is to change the amount of power to the electric
motor based on the pulse length of the Pulse Width Modulation (PWM) signal. By switching the power
from the batteries between ON/OFF at a rate around 20KHz and controlling the ONOFF time ratio with a
PWM signal, it is able to vary the voltage applied to the motor. The information on the rotor’s position is
paramount to the correct coordination between the three phases. It is obtained using sensors measuring
the back-Eletromotive force (EMF). This EMF is induced into the coil by the change of magnetic flux.
Figure 3.17: Simplified ESC circuit diagram. Courtesy of [37]
Figure 3.18: BLDC motorcurrent waveform simplifiedschematic. Courtesy of [37]
There are two main types of BLDC motors distinguished by their rotor location. In an outrunner
arrangement, the rotor runs around the stator, whereas in an inrunner arrangement it runs within the
stator. The bigger diameter of the rotor allows outrunner motors to build up bigger torque, whereas
the higher inertia leads to lower rotational velocities. Inrunner motors on the other hand create lower
torque at higher rotational velocities [38]. The outrunner configuration is favoured over the latter in model
21
aviation due to the need for lower energy magnets, reduced copper losses, reduced production costs,
and greater rotor inertia.
Enertion boards FOCBOX
The ESC used is a Enertion Boards FOCBOX Speed Controller. This device is based upon the
VESC R© Open Source Project [39]. This unit was particularly interesting as it offers a PWM control
interface, an easy setup through the Vesc Tool software, vast online documentation, and a built-in current
control feature.
In fact, the current control feature is not usually seen in hobby ESCs, the majority use use voltage
control. By varying the voltage applied to the motor, they control its speed. However, in this project it
is more important to control the torque output of the motor, and this is accomplished by controlling the
current applied to the motor coils as later depicted in Equation 3.7. This way, it is possible to make a
throttle input into a torque setting. This control mode also allows to recover energy by applying a counter
torque to the motor, called regenerative braking (regen) mode. This allows the hybrid system to recharge
the batteries using the combustion engine.
Figure 3.19: Enertion boards FOCBOX. Courtesy of [40]
Batteries
The batteries used to power the motor were Lithium Polymer 6 cell 8000mAh, presented in Figure
3.20.
These batteries are rated up to 15C of discharge rate, while it is only advisable to go up to 2C when
charging them. This means, according to equation 3.4, the maximum discharge current is 15 × 8A =
120A and a maximum charge current of 2× 8A = 16A
Max current draw = Capacity × C rating (3.4)
22
Since each cell has a maximum, nominal and minimum voltage of 4.2V , 3.7V and 3.0V , these 6 cells
batteries have a maximum, nominal and minimum voltage of 25.2V , 22.2V and 18.6V . These batteries
were chosen as they met the electrical requirements for the AXI4130/20 EM, as well as because of their
availability as a COTS commodity.
Figure 3.20: 6 cell 8000mAh batteries
The disadvantages of Lithium-polymer batteries are mostly related to safety and may be controlled
by proper handling. This type of batteries are extremely sensitive to heat and overcharging so precise
monitoring of the pack is necessary. The primary indicator of charge status, besides using complicated
techniques to estimate power draw, is to watch the pack voltage over time. Overcharging, over temper-
ature charging, charging too quickly, and charging a pack that has been depleted past a safe point are
all dangerous and can cause fire.
In Figure 3.21 are shown the discharge characteristics of a Lithium-polymer battery. It can be seen,
that there is a voltage drop with increasing capacity utilisation and current drawn. The lower the drawn
current, the smaller the voltage dip. Also, it is important to note the voltage drop follows a close to linear
trend up to the point where they are almost depleted, where there is a steep voltage reduction.
Figure 3.21: Discharge curves of Lithium-polymer battery for various discharge currents. Courtesy of[38]
Due to this fact, and given the high current draws the aforementioned batteries might be subject to,
the author decided to connected two batteries in parallel so as to minimise voltage drop.
23
Setting up the ESC
The setup was performed using the software VESC R©Tool available at [39]. The first step consists
on configuring the BLDC sensorless motor. This includes setting limits for the maximum current flowing
in or out of the motor/generator. In the next step, the same limits are set for the battery to ensure a safe
operation.
The following step includes the control interface setup. This was performed using the ”Current No
Reverse with Brake” control type, with a PWM signal input. The settings are summarised in Table 3.2.
Parameter Value
Motor type BLDCMotor Current Max 55A
Motor Current Max Brake −15A
Battery Current Max 80A
Battery Current Max Regen −15A
Battery Voltage cutoff start 20.4V
Battery Voltage cutoff end 18.6V
Control type Current no Reverse with BrakePWM frequency 50Hz
PWM Upper limit 2ms
PWM Lower limit 1ms
Table 3.2: ESC parameters
Additionally, since the way the ESC controls the EM is by switching ON/OFF MOSFET transistors,
as previously described, the current going into the ESC has several spikes and is an overall noisy
signal as seen in Figure 3.22. This can potentially harm the batteries as they are exposed to unsteady,
high frequency and high amplitude current draws. This way, a COTS Castle Creations capacitor pack,
shown in Figure 3.23, composed of four in parallel 220µF capacitors, totalling 880µF , was placed in
parallel between the battery and the ESC, following the recommendations of the manufacturer. These
capacitors act as a low pass filter, reducing voltage ripples and allowing a much smoother current flowing
from the battery, extending the battery’s life and, at the same time, reducing the load on the ESC on-
board capacitors.
AXi 4130/20 performance
Although there are different types of DC motors, their operation is the same [41]. Figure 3.24 shows
a DC EM equivalent circuit. The inefficiencies present in the motor can be attributed to its internal
resistance, represented by Rm, and the no load current, represented by I0.
The motor’s internal back-EMF (Vm) is given by the different between the supplied voltage and the
voltage drop across its internal resistance
Vm = V − IRm (3.5)
24
Figure 3.22: Current signal measured with an oscil-loscope
Figure 3.23: Castle Creations Capacitor pack
Figure 3.24: DC electric motor equivalent circuit [7]
The shaft rotation speed is the product of Vm by the motor speed constant (Kv)
ωm = VmKv (3.6)
The torque output of the motor can be calculated as follows:
τout =I − I0Kv
(3.7)
For this particular motor, I0 = 1.1A and Kv = 31.94 and so, τout = I−1.131.94 . This is an important metric
when evaluating the motor performance
The electric power supplied to the EM, PE , is equal to the product of the supplied voltage and the
current:
PE = IV (3.8)
while the output power, Pout, can be calculated by the product of the output torque and the rotational
speed:
Pout = τoutωm (3.9)
Presented in Section B.2 is the provided manufacturer data for a 20V and 25V supply voltage.
Along with the provided data by the manufacturer, available in Section B.2 a number of tests were
performed in order to fully map the performance of the particular unit in hands. These tests consisted of
running the EM with a 19′′ × 10 fixed pitch propeller, allowing to fully map the performance of these two
25
components. The tests were conducted at the Thrust Test Rig available at UVIC-CfAR, and included the
measurement of different test variables such as torque, thrust, rotational speed, current and voltage. The
torque and thrust were measured using a FUTEK MBA500 Torque and Thrust bi-axial sensor [42]. This
sensor is rated up to 200lb (890N ) of Thrust force and 200in− lb (23Nm) of Torque. The Voltage, Current
and RPM sensing was performed using the Castle EM Data logger. Due to its data logging capability,
these tests were run using the Castle Creation Phoenix Edge ESC. However, the ESC manufacturer
was not able to provide a value for the accuracy of the measurements. For further information about
these tests, refer to [43].
(a) Thrust Test Rig (b) LabView GUI
Figure 3.25: AXi 4130/20 test setup
EMthr RPM Torque [N ·m] Powerprop [W ] Current [A] V oltage [V ] Powerbat [W ] η [%]
0 0 5.3225× 10−05 0 0.034 25.157 0.870 −10 756.254 0.0329 2.611 0.294 25.147 7.394 35
20 1352.158 0.1004 14.217 0.960 25.118 24.118 59
30 1981.435 0.2082 43.201 2.404 25.055 60.243 72
40 2595.522 0.3512 95.471 4.897 24.942 122.159 78
50 3163.111 0.5212 172.657 8.620 24.769 213.519 81
60 3686.619 0.7099 274.066 13.567 24.547 333.050 82
70 4155.084 0.8988 391.116 19.725 24.190 477.167 82
80 4584.160 1.0970 526.663 27.314 23.731 648.222 81
90 4944.854 1.2811 663.414 35.927 23.132 831.108 80
100 5161.947 1.4024 758.129 42.611 22.471 957.526 79
Table 3.3: Axi 4130/20 testing results
In Figure 3.26, a close to linear relation between Torque and Current, as shown in Equation 3.7, was
expected. Instead, in the lower values of current, up to roughly 15A, this behaviour is not verified. This
can be attributed to the poor efficiency of the motor in this regions. As the current is increased, it is
possible to see a relation closer to the linear, as expected.
In Figure 3.27 we observe a maximum efficiency of 82.89% at 3686.619RPM , with a decreasing
trend past that point. This is an important fact because the HEPS under development will have the
motor running at the same speed as the engine, meaning it will run at speeds up to 7000RPM , where it
still should be able to provide the requested torque to assist the engine. This will be further explored in
Chapter 5. The obtained value is also close to the one provided by the manufacturer (Section B.2 Test
26
Figure 3.26: AXi 4130/20 current versus torquecharacteristic Figure 3.27: AXi 4130/20 efficiency test results
6). It is also visible a clear voltage drop as we reach the high end of the throttle. This behaviour goes
in line to the discussion previously carried out and illustrated in Figure 3.21. This is important because
as emphasized in Section 3.2.2, care is due to not over-discharge the batteries. However, the values
presented here were collected with the battery under a load. A using a higher capacity battery pack or
adding batteries in parallel might help to mitigate this problem.
An important aspect concerning electric motors performance is the maximum torque available as a
function on rotational speed. This function is depicted in Figure 3.28. In fact, the maximum torque is
available at 0 RPM with an ever decreasing maximum torque as the RPM increases. All DC motors,
exhibit this same characteristic linear slope [25].
Figure 3.28: Direct current motor torque characteristics. Courtesy of [25]
3.2.3 Propeller
The propeller used in this project was a 19′′ × 10 fixed pitch wooden propeller. Several fixed pitch
propellers were mapped during the electric motor test campaign. They were then plotted against the
BSFC engine maps to try to foresee the working points of the system. The choice of propeller, and thus
the load applied to the system, intended to provide a way of effectively testing the system. This is, it
27
aimed to present a Torque-RPM curve within the normal operating points of the engine. Nonetheless,
when implemented into the hybrid powertrain, a lot of losses through friction are added, and the load on
the engine was substantially higher than initially predicted.
Aerodynamically, a propeller blade can be seen as a rotating wing. This blade generates lift that
depending on the angle of attack and the twist angle will contribute to the propeller thrust. The drag
force created by this airfoil defines the necessary torque to rotate the propeller [38].
Given the high rotational speeds, it is important to guarantee the balance of the propeller to prevent
vibrations and thus, possible damage or failure of the propulsion system. This way, the selected propeller
was placed on a propeller balance. In case of unbalance, a sandpaper was used to thoroughly remove
material from the edge of the propeller while keeping its aerodynamic shape. The results are shown in
Figure 3.29.
From the theory we can estimate important performance parameters. For the scope of this research,
the most important parameter to evaluate in this propeller is its Torque versus versus RPM relation as
will be made clear in Section 4.1.2. The torque, τ , may be given by:
τ = kQρD5(RPM
60)2 (3.10)
where kQ is the torque coefficient and in general is a function of propeller design, the Reynolds
Number,Re, the Mach number at blade tip Mtip and the Advance ratio, J , defined as J = VnD where D is
the propeller diameter and V is the flight velocity [25]. It is important to note that this equation considers
an incoming air flow in a velocity V , whereas the conducted tests were static.
Additionally, propeller tests were conducted using the AXi 4130/20 EM at CfAR Thrust test rig to fully
map its performance, as described in Section 3.2.2 and in Figure 3.25.
(a) Before balancing (b) After balancing
Figure 3.29: 19′′ × 10 propeller
Figure 3.30, shows the Torque versus RPM curve of this propeller. The experimental data (blue dots)
was fitted with a second order polynomial curve (in red), with the goodness-of-fit statistics presented in
the table on the right. As it is clear, both from the graph on the left and the statistics on the right, torque
follows a parabolic dependency with RPM, as stated in Equation 3.10.
28
Goodness-of-fit statistics Value
Sum of squares due to error 0.0025R-squared 0.9998
Root mean square error 0.0103
Figure 3.30: Torque versus RPM characteristic for the 19′′ × 10 fixed pitch propeller
3.2.4 Mechanical coupling
A key component of a parallel HEPS is the mechanical clutch required to combine the power of the
ICE and EM. There are two primary requirements imposed on such a system. The first is to allow the
electric motor to power the propeller with the ICE disengaged, to enable EM ONLY operation. On the
other hand, it should let the engine run the motor so as to recharge the onboard batteries. There are
two potential clutch mechanisms for a small HEUAV: an electromagnetic clutch, and a one-way bearing.
In Figure 3.31 a simple schematic of the installation of the clutch or one-way bearing into the parallel
hybrid architecture is presented.
An electromagnetic clutch uses an induced magnetic field to transmit mechanical work. When a
current is flown through its coil, generates a magnetic field that attracts the armature into the rotor,
transmitting torque.
The other option is to link the two shafts with a one-way bearing. There are many types of one-way
bearings but they all follow the same working principle. When rotating in one direction, the shaft rolls
freely over the pins. When the inner shaft turns in the opposite direction, the rollers press against the
springs which bind them in place, stopping the inner shaft and transmitting torque to the shaft attached
to the outside of the bearing. The use of this bearing, however, prevents the EM from being able to start
the ICE which then has to be started using an exterior mechanism. Nevertheless, this option is cheaper,
simpler, more reliable and lighter than the electromagnetic clutch what made it a good choice for the this
proof-of-concept prototype HEPS. This was also the choice of different authors such as Koster [24] and
Greiser [23].
An Align RC one-way bearing assembly was selected due to its availability, and previous use in other
projects conducted at UVIC-CfAR.
29
Figure 3.31: Clutch/one-way bearing installation
3.3 Controller Design
Hybrid-Electric propulsion systems performance is highly dependent on the energy management
system. The presence of an additional degree of freedom for satisfying the drive power demands implies
that the performance of an HEV system strongly depends on the control of the power split. In order for
these components to work in an organised manner, a controller is required.
The basic components over which the controller has direct authority are:
1. Engine throttle servo (0% to 100% engine throttle)
2. Motor power (0% to 100% motor throttle)
3. Generator power (0% to −100% motor throttle)
3.3.1 Design Philosophy
The design philosophy behind a hybrid electric propulsion system is to improve the efficiency of the
overall system. For this, each component should operate at or close to its maximum efficiency region.
However, this is a complex problem given the diversity of components and energy sources. This way,
in this research, efforts are mainly focused in optimising the engine operation. This was deemed as
a good approach for this prototype system because it drastically reduces complexity of the controller
design and, in fact, the engine is the main source of inefficiency and poor fuel economy.
As described in Section 2.4, a rule based open loop control strategy might be ideal for the type of
controller intended.
3.3.2 Rule based Controller open loop control strategy
A rule-based controller was selected for use in controlling the prototype HEPS due to its inherent
simplicity, highly predictive behaviour, and simple debugging. Since this was the first iteration of this
propulsion system, simplicity was deemed paramount.
In this research, and given the experimental and proof-of-concept nature of this project, a previ-
ously developed controller by Harmon and described in [20] was used. This way, this research project
represents a validation of the results presented in [20] and [23].
30
The inputs to the particular system in question are a torque request and the system rotational speed.
The controller then decides which power mode to use and the correspondent power split, according to
the flowchart below (Figure 3.32). In short, the engine torque is commanded with increasing requested
torque value up to the engine IOL. Then, the motor torque is added up to the maximum motor torque,
after which point, the remaining engine torque is used.
The controller developed here can be understood under the framework of a charge depleting strategy
as it does not include any charging of the batteries.
Figure 3.32: Decision tree (Adapted from [23])
In an open-loop controller, the error between the obtained value and the commanded value are
unknown to the controller. In a closed-loop controller, the reverse is true and it uses these values to
make adjustments to the input. The rule-based controller presented here is not a closed-loop controller,
but it can have pseudo closed-loop behaviour. The piece that closes the loop is the aircraft operator (pilot
or autopilot) as shown in Figure 3.33. It has knowledge of the aircraft speed and flight characteristics
and adjusts the throttle accordingly. The controller interprets that signal and then decides how to split the
torque between the power units in the aircraft. This key factor is used throughout the code to establish
an efficient open-loop state machine and goes in line with the work carried out by Griser, [23].
Figure 3.33: Hybrid-electric propulsion system controller block diagram (Adapted from [22])
31
3.3.3 System working points
Figure 3.34 pictures the possible working points of the system. Note that operating points can occur
whenever a torque availability can equal a torque demand. Reducing the engine or motor throttle setting
will reduce the respective available torque and the propeller RPM will settle at a lower level. This way,
since we are using a fixed pitch 19′′×10 propeller, we expect the system to always operate on this curve.
Thus, conceptually and disregarding losses, we expect the system to only use the engine up to roughly
6000RPM , where the propeller curve intersects the IOL. From that point onwards, an increase in torque
request leads the engine to remain operating at its most efficient point while the motor makes up the rest
of the requested torque.
Figure 3.34: System operating points
3.3.4 Controller states
The implemented controller has different states it can operate under which are selected by the user.
These embody the versatility of a hybrid system and offer the user different possibilities of operation
depending on the mission requirements.
Single ICE operation
While a HEPS features both an EM and an ICE, the ICE is still the main focus of the propulsion
system. In fact, the engine should provide enough power to fly at cruise speed. Additionally, in case
of motor failure, the ICE should be able to fly the aircraft to a safe condition. This mode represents
a conventional gasoline power propulsion system and serves as a comparison to the Hybrid controller
developed.
The controller reads the torque request from the user and moves the ICE throttle valve accordingly,
so as to meet the torque value. A proper mapping of the torque-throttle curve is then imperative. In the
work developed by Greiser [23] a linear relation is established following the equation:
32
Throttle position =Requested Torque
Maximum Torque available(3.11)
Early iterations of the controlled used a Torque − Throttle − RPM look up table provided by the
manufacturer and available in Appendix B. This way, based on the current rotational speed and torque
request, the controller would select the throttle accordingly. However as it will be later made clear in
Section 5.1, the disparate engine behaviour when implemented on the hybrid powertrain prevented the
feasibility of this approach. This way, ICE ONLY mode tests (which results are later presented in Section
5.1.2) were conducted to map different engine working points allowing to establish a Torque versus
Throttle curve. This curve serves as a look-up table for the controller and allows it to update the throttle
value based on the requested torque.
This was performed in order to obtain better estimate of the relation between these two values and
works well as a first estimate for the controller. This was possible because the same fixed pitch propeller
was used throughout the whole validation and test campaign. This fact, couples the RPM and Torque
variables and thus makes this approximation feasible. Nonetheless, this introduces clear limitations to
the versatility and adaptability of the system. Despite that, this was deemed a reasonable option so as to
reduce the complexity of the controller, also taking into account the fact that it is its first iteration. Further
improvements to mitigate the discussed problems were left as recommendations for further versions of
the system.
Single EM operation
This mode offers a quiet and zero emission flight using only the EM while the ICE is set to idle
position. In this state, the one-way bearing should disengage given the lower rotational velocity of the
engine.
In a similar fashion the the Single ICE Operation, this mode reads the user input and sets the EM
throttle accordingly. In line with the discussions carried out above, a map of the Torque versus Throttle
position was used to feed into the controller. This simple approach provided a first estimate on the
dependence of these two variables, and was deemed necessary due to time constraints.
Hybrid
In the Hybrid mode, the controller optimises the operation of the ICE as described in Section 3.3.2. It
accesses look-up tables with the IOL information to continuously update the engine and motor operating
points based on the requested torque and rotational speed. In order to access the Torque-Throttle
correspondence, it uses the sames curves obtained in the ICE single operation and the EM single
operation. This way, an overlaying assumption is that the system works in tandem as it would separately.
This is a common assumption as, for instance, seen in the simulation work by Hung, [1], Harmon [20]
and Greiser [23]. This assumption will be put to the test, to analyze its ability to depict actual system
behaviour.
33
Conceptually, this mode promises to offer a lower fuel consumption when compared to the Single
ICE Operation and is thus the main focus of this research project.
Reset
This is the default mode which the controller jumps to whenever it is activated. The sole operation
of this mode is to place the ICE at idle and to turn off the EM. This is an important safety feature as it
allows for a safe and predictable behaviour when the controller is engaged.
34
Chapter 4
Experimental Setup
Once conceptually outlined, the prototype hybrid system was in need of a method of testing and
validation. In this chapter, the experimental apparatus developed is presented. This comprehends a
thorough discussion over the test bench design as well as important considerations when conceiving
a powertrain. It concludes with a description of the sensor suite, the test control and data acquisition
system developed, the Printed Circuit Board (PCB) electrical connections, and the controller implemen-
tation.
4.1 Test bench
The parallel hybrid-electric test bench was object of several iterations with many interactive phases.
The first step was outlining key design goals of the proposed system. Firstly, it should present a modular
platform. This means, every component of the hybrid system should be able to be switched with relative
ease, allowing to implement any component combination and test any parallel HEPS for small UAV
applications. This also permits sensitivity tests to analyze how changing parameters affect the overall
system performance. Secondly, it should be properly instrumented, allowing to gather all the important
data to fully map the hybrid system performance. Finally, it should provide a safe, reliable, and consistent
test base, where tests can be conducted and repeated without any major damage to the system and
risk to the involved personnel. In fact, given the system’s high rotational speeds, inherent vibrations and
possible resonance frequencies within the operational range, safety is paramount.
A schematic of the test bench mechanical design and sensor integration is presented in Figure 4.1.
Section 4.1.1 describes the mechanical considerations whereas section 4.1.2 addresses the sensor
suite.
4.1.1 Assembly description
The final mechanical design can be seen in Figure 4.2 and a simplified and easier to understand
version in Figure 4.3.
35
Figure 4.1: Test bench schematic
Figure 4.2: Parallel hybrid electric test bench CAD Figure 4.3: Detailed view of the parallel hybrid elec-tric test bench CAD
Starting from the engine, attached to the its output shaft there is a flywheel. In fact, an important
consideration when operating these engines is that they are designed to always work with a propeller
attached to its crankshaft. Among other things, this propeller acts as a flywheel, or mechanical capacitor.
Energy is transferred to the flywheel during the engine’s expansion stroke by the application of torque
through the crankshaft, evening out any torque spikes due to its large inertia. The flywheel releases
its stored energy by applying torque to the crankshaft during the compression stroke. This plays a
very important role since the energy coming from an ICE has an intermittent nature. On this set up,
however, the one way bearing prevents any reverse power transmission and thus the engine is deprived
of the flywheel effect of the propeller. This way, a flywheel had to be engineered and installed between
the engine and the one-way bearing. A precise measurement of the necessary inertia is difficult to
achieve though. The solution found was estimating the inertia of a 20 × 10 wooden propeller (size
recommended by the manufacturer) using SolidWorks. Thereafter a 0.25′′ (or 6.35mm) thick aluminium
36
disk was designed to equal the inertia of such a propeller. Its dimensions dimensions can be seen in
Figure 4.4. The disk was then thoroughly machined to assure concentricity to the engine output shaft.
A tight fit between the inner bore of the flywheel and the ICE output shaft, was used as a locating
feature preventing any eccentric rotations. The flywheel was secured using four bolts that go through
the flywheel into the the engine housing.
This addition had a paramount importance and, overall, enables the system to actually work. In fact,
prior to this installation, the engine was able to go through its expansion stroke but then it did not have
enough inertia to go through the compression one. What followed is that the spark plug would fire and as
the piston was far from top dead centre, the engine would try to rotate in the other direction, damaging
the engine.
Figure 4.4: Flywheel designFigure 4.5: Propeller shaft
Another important aspect is the engine mount. This needs to be stiff enough to prevent any major
vibrations, while not overstressing the engine structure. The engine was secured from the back using an
L bracket that connects to the engine crank case through standoffs (this setup can be observed in Figure
4.2). However, the force of the piston during its expansion stroke has a high momentum arm, leading to
high bending stresses on the bracket and eventual failure. Moreover, a resonance was verified close to
3500RPM where the amplitude of the vibrations would increase drastically, and then decrease as the
system went past this rotational speed. Therefore, additional mounts, consisting of the two aluminium
bars (one on each side), were placed supporting the crank case and on the direct line of the application
of the piston force. This reduced the amplitude of the vibrations across the whole operating range and,
as the stiffness of the system was increased, the resonance frequency previously mentioned shifted.
These mounts are visible in Figure 4.7.
The engine’s output shaft was connected to the one-way bearing shaft through a flexible shaft cou-
pling that absorbs vibrations from the engine. The latter shaft was supported by a pillow block bearing
which locates the shaft axially. The former end of that shaft was inserted into the inner race of the one
way bearing assembly and locked by a metal pin (refer to Figure 4.3).
The one-way bearing assembly includes an Align RC one-way bearing fitted inside a one-way bearing
37
housing, in between two ball bearings as depicted in Figure 4.6 b). A precisely machined hollow shaft,
goes through the housing, assuring a proper connection with the one-way bearing. A pulley was then
machined to be embedded into the one-way bearing assembly as shown in Figure 4.6 a). Since this
assembly connects two different shafts, it is essential to emphasize the concentricity requirement when
machining this particular sub assembly.
In the early designs of the test bench, this pulley would be connected to a timing belt. However, this
led to excessive radial loads on the ball bearings and eventually failure. Thus, a redesign iteration was
conducted using another another pulley, but the one-way bearing assembly design was maintained.
(a) CAD render (b) Detailed view of one-way bearing housing
Figure 4.6: Custom one-way bearing assembly
Connected to the pulley of the this assembly is a shaft that links directly to the propeller, depicted
in Figure 4.5. This shaft also features a pulley that connects to another pulley, present on the electric
motor side, through a timing belt (see Figure 4.3). Alignment between these two is important because
otherwise may lead to excessive wear or failure of the timing belt. The two pulleys have exactly the same
outer diameter so the gear ratio is 1 : 1. In the end of this shaft, a small step was machined to allow the
propeller to rest against this face. A thread was also machined to tight a nut that presses against the
propeller and holds it in place. This shaft is supported on both ends by pillow blocks.
On the EM side, the motor is attached to the a shaft through a flexible shaft coupling. That shaft is
supported by two pillow blocks. The implemented test bench can be see in Figure 4.7
Powertrain design considerations
Some of the important metrics used to evaluate the powertrain quality were alignment, friction of
the system as a whole, frequency and amplitude of vibrations and wear of any components such as
bearings, couplings, etc. The eccentricity in rotation was measured with a depth gauge, and was a
valuable tool in troubleshooting.
For such a setup, alignment plays a crucial role as otherwise may lead to excessive vibration, wear
of components and even failure. As a matter of fact, height misalignment on the pillow blocks supporting
the one-way bearing shaft, lead to a malfunction of the one-way bearing. The misalignment caused
that shaft to excessively rub against the one-way bearing, creating enough friction force to turn the
ICE crankshaft, allowing reverse power transmission, going against the working principle of a one-way
38
Figure 4.7: Parallel hybrid electric test bench
bearing. After identifying the problematic bearing block, which had a slightly lower height than the rest
of the system, a 0.004′′ steel shim disk was placed underneath to mitigate the problem. Moreover,
improper machining of the one-way bearing assembly led to misalignment between the two connecting
shafts which caused eccentric rotation of the propeller and thus high amplitude vibrations. As a results,
the assembly had to be remachined.
Another important consideration is friction. In fact, early designs of this test bench featured sealed
ball bearings. These added a considerable amount of friction to the system, preventing it from working
properly. The solution found was to remove the seals and all the grease inside the bearings replacing it
with oil to prevent excessive wear. Although this reduces the life span of the bearings, and makes them
more vulnerable to dust and contaminants, that ultimately damage the bearings, it drastically reduced
the resistance of the system.
Internal Combustion Engine Starting Mechanism
With the incorporation of the one-way bearing, conventional spinner-starting of the ICE is not com-
patible. At first, two PLA gears with a gear ratio 1:1 were printed and integrated on both the starter
and the ICE, as shown in Figure 4.8. However, these gears turned out not strong enough to handle
the associated torque as can be seen in Figures 4.9. Options using stronger 3D printed materials were
considered but too expensive.
Another design consisted on a placing a one-way bearing sprocket on the ICE crankshaft, powered
by an external starter electric motor through a chain, as seen in Figure 4.10. However, due to large
loads, both the one-way bearing and the rubber timing belt inside the starter failed as shown in Figure
4.11.
Thereby, a new and simpler solution was engineered. Using flywheel installed on the ICE crankshaft,
the starter was featured with a rubber disk on the outside that would hold against the flywheel as shown
in Figure 4.12. Friction force would transmit the torque and given the large radius of the flywheel, the
39
Figure 4.8: 1:1 PLA bevel gear design Figure 4.9: Failed 2:1 PLA bevel gear design
Figure 4.10: Second design for the starter mecha-nism
Figure 4.11: Failure of starter timing belt
load on the starter was largely reduced. The starter used was a Turnigy Lipoly Belt Drive Starter that
features a reduction belt system managing to provide 4.41N ·m of torque at its output shaft. This proved
to be a reliable setup.
This discussion emphasises the difficulties in starting an ICE when installed in a non conventional
architecture. Some other options are pull-start mechanisms, as used by Megistu in [2], or installing an
auxiliary motor attached to the crankshaft supported by some sort of bracketing.
Figure 4.12: Starting setup - final design
40
4.1.2 Sensors and actuators
Common HEV feature a number of sensors including position, speed, temperature, pressure, voltage
and current sensors [44]. However, while most of this sensors are implementable in automobiles, its
feasibility in small UAVs is limited. A HEPS bench testing campaign conducted by Gresier, [23], included
a total of 6 sensors: rotational speed of each power unit, temperature for the EM, ICE and battery pack,
current and voltage measurement for the battery, and cylinder pressure for the ICE.
On this research, the sensor suite aimed to meet the data gathering needs of both the HEPS con-
troller operation and the evaluation of performance metrics previously defined. This way, the variables
monitored with this test bench are:
1. Voltage and Current
2. Rotational Speed
3. Fuel Mass
4. Torque
The sensors were chosen considering precision, reliability and easy installation into an aircraft. It
is also a simple design which allows for ease of trouble shooting and monitoring its operation. Its
implementation on the test bench is outlined in Figure 4.1.
The testing and simulation environment was programmed using National Instruments (NI) LabVIEW.
This software allows to interface the hardware, control and acquire data from the tests. The data ac-
quisition procedure was carried out using the NI cDAQ-9188 outfitted with input and output modules
that interface with LabVIEW. The data acquisition process is made synchronous through the use of an
external clock available at the cDAQ station: ”Ctr0InternalOutput”. A Producer—Consumer loop is used
for reading and writing data into a .lvm file every at a rate of 5Hz. This is an important consideration
when designing a vast LabVIEW routine acquiring multiple sets of data to be processed. In fact, since
producing the data is much faster than writing it into a file (consuming), these two procedures should be
decoupled. Data queues are used to communicate data between these two loops that run at different
rates. This enhances data sharing between them and enable the whole system to run smoothly.
All the sensors were connected to a PCB to make the interface easier. Connected to PCB is a
harness that links to the cDAQ. Figure 4.13 shows the used board, featuring the position where each
individual channel should be connected, selected according to the PCB and harness electrical wiring.
Voltage and Current
The electric power was evaluated measuring the voltage and current flow from the battery, according
to equation 3.8. A current shunt was placed in line between the ESC and the battery. By measuring the
voltage drop between its two terminals (see Figure 4.15), and knowing the precise value of its resistance,
it is possible to calculate the battery current through Ohm’s law: I = VR . The battery voltage was directly
measured at the output terminals of the battery.
41
Figure 4.13: Data acquisition printed circuit board
So as to communicate their reading with a computer, the NI 9205 module was used. However, given
its the ±10V input range, a voltage divider was place in between the sensor and the module. This
voltage divider steps down the voltage by a factor of 4. A schematic of the electrical circuit is shown in
Figure 4.14.
Figure 4.14: Voltage divider electrical circuit schematic
These measurements were configured using the differential terminal configuration option available
in LabVIEW. It measures the difference between the positive and negative inputs and provides overall
more accurate measurements, good rejection of common-mode voltage and noise. This configuration is
especially recommended for battery devices. Two Analog Input (AI) channels were used to measure the
voltage raw signal coming from the sensor.
The data is treated using the sample compression Virtual Instrument (VI) to remove spikes and
unsteady readings. This is an important procedure as emphasised in section 3.2.2 and illustrated in
Figure 3.22.
The current and voltage measurements with the current shunt were compared the ones obtained
42
Figure 4.15: Current shunt used for voltage and current measurements
with a Multimeter, where the correspondent readings presented a negligible1 difference, validating this
setup.
Rotational speed
Hall effect sensors were used to measure RPM due to their no-contact, wear free operation and low
maintenance design. They were installed on the ICE and EM2 shafts. The sensor used was a ”Eagle
Tree Systems RPM Sensor with Magnets” and, in its essence, is a hall effect switch.
This sensor has three connections, Signal, Power and Ground. When in the presence of a strong
enough magnetic field with the correct polarity, this sensor closes the circuit between the Signal and
Ground wires. This way, when over a magnet the sensor is OFF and has the same voltage has the
Ground reference, which we can define as 0V . A resistor of 2.2MΩ was connected in between the
Power and Signal lines as shown in Figure 4.16. This way, when the sensor is far from the magnetic
field, it is turned ON , and displays the same voltage as the power supply, 5V .
The value of the resistance chosen is very important since the Hall Effect sensor cannot handle
much current (in the order of ≈ 10−3A). By choosing the aforementioned resistor, the current through
the sensor is I = VR = 5
2.2×106 = 2.27× 10−6A, three orders of magnitude below the referred limit.
Figure 4.16: Hall effect circuit schematic
1the small differences verified were attributed to a different tare2it is important to note that given the test bench design previously discussed, the EM will always rotate at the same speed as
the propeller
43
This sensor was then mounted onto the bearing block with a 3D printed casing. A magnet was
glued into a 3D printed ring that was press fitted around the shaft as seen in Figure 4.18, where the
implementation onto the EM shaft is shown. The DA35, however, came with a built-in Hall effect sensor
used for the ignition system, and is shown in Figure 4.17.
The voltage readings from both these sensors were interpreted using two Analog Input (AI) channels
in the NI 9205 module, configured to Referenced Single-Ended (RSE) terminal configuration (where the
signal measurements are referenced to the NI 9205 module ground).
Figure 4.17: DA35 installed RPM sensor Figure 4.18: EM RPM sensor
When the shaft is rotating, the signal is similar to a square wave: periods when the sensor is ON
followed by short periods, or pulses, where the sensor is OFF (when a crossing is detected). A sub
block in the LabVIEW routine counts the time between pulses and calculates the frequency in RPM. This
sensor gathered 5000 samples at a sample rate of 20kHz. This means that at a frequency of 4Hz, the
system goes through the calculations and outputs a RPM value. Furthermore, the minimum time inter-
val the system can measure is 0.05ms. The measurements using this setup were compared to the ones
obtained with a COTS tachometer, assessing the speed of an electric motor. The maximum difference
between values was approximately 2.5% and both presented a near constant value (low noise).
Fuel Mass
So as to evaluate the fuel consumption of the engine, a load cell was placed under the fuel tank. This
device is in its essence a transducer that creates an electric signal proportional to the force applied to it.
The load cell signal was connected to the NI 9237 module, which was programmed into a Full bridge
configuration. On the LabVIEW routine, an Analog Input Force Bridge (Two-point linear) channel was
created with a sample frequency of 10kHz and a number of samples of 1000.
The load cell was calibrated assuming a linear dependency between strain of the strain gauge and
the voltage from the sensor, which is a good assumption [42]. Given the load cell high sensitivity, it
should be place in a platform free from vibration to avoid unreliable measurements.
The measurements obtained with this setup were tested using know weights and demonstrated
an error of up to 1.5%. However, it was verified that the load cell signal was highly noisy. A sample
compression VI was used to attempt to mitigate this problem but, despite reducing some of the noise,
could not eliminate it.
44
Figure 4.19: Load cell installed underneath the fuel tank
Torque
In order to validate the controller developed, it is important to measure the torque output of the
system.
There are different options as far as Torque sensors are concerned. A method explored by various
researchers features a type of torque sensor using a momentum arm that deflects as the engine applied
torque, [35]. Conner et al. , [45], developed a low cost dynamometer system able to measure torque,
thrust, rotational speed, voltage, current and airspeed of the system. The featured torque sensor utilized
a design that transferred the rotational torque produced by the motor into linear displacement, which
was in turn measured by a load cell.
Furthermore, there are a few COTS torque sensors that present compact and reliable solutions.
Essentially, they act on the same principle as load cells, described in the previous section. Nonetheless,
they can be grouped into two types: reaction torque and rotary torque sensors.
The reaction torque sensors utilize the principle that the torque applied to the load by the power unit
is the same to the torque applied to the power unit by the load. These sensors place a load cell on the
desired measurement axis.
Other widely accepted method for measuring torque are called rotary torque sensors and rely on
mounting transducers in the machine train or on the rotating shaft. With few exceptions, these methods
use strain gauges.
However, these devices can be extremely expensive, and may required a meticulous mechanical
design to integrate them. For instance, the use of an rotary torque sensor requires extra shaft couplings,
inducing losses to the system.
Given time and budget constrains, the author chose a simpler approach. Using a fixed pitch propeller
as a mechanical load on static tests, enables to couple the system rotational speed to its total output
torque. By mapping the propeller Torque versus RPM curve, it is possible to get an estimate of the
output torque of the system by simply measuring its rotational speed. This technique uses the principle
that the torque is balanced and so, the propulsion system output torque is equal to the torque required
for the propeller to run at a that specific speed. Being the simplest and cheapest option, this was the
45
selected method for this prototype. The tests conducted in Section 3.2.3 allowed to build a function
which receives a rotational speed and outputs the corresponding torque value.
Furthermore, the statistics presented in Figure 3.30 indicate a small error between the fitted curve
and the experimental data. This way, given a precise RPM measurement, the curve should be able to
correctly estimate the torque output of the system. However, the torque measurements are only as good
as the RPM sensing and thus have limited precision and accuracy. Furthermore, in combustion engines,
RPM is a very noisy signal, hindering the torque calculation. Finally, this method disregards changes in
test conditions and although able to provide major trends, it lacks precision.
ICE and EM throttle
The control of both the ICE and EM throttles is done by a PWM signal. Those were generated
using the NI 9401 module through two counter output (CO) channels that generate digital pulses whose
frequency and duty cycle are user defined.
To move the engine throttle, a 50Hz servo motor was used. The servo must be calibrated so it
matches the carburettor’s range of motion. Typically, servos operate between a pulse width of 1ms to
2ms. However, to match the servo and carburettor, that range was adjusted to 1.05ms and 2.05ms.
On the motor side, the ESC enables the user to tune the parameters. This way, the PWM frequency
was set to 50Hz with a pulse length upper and lower limit of 2ms and 1ms, respectively.
4.2 Graphical User Interface
In order to provide a user-friendly interface with the test bench, a GUI was developed. It aimed to
allow the test operator to easily control and monitor the test. The developed LabVIEW2018 screen can
is shown in Figure 4.20.
All the engine related gauges are to the left, while on the right are the motor related ones.
On the top left the Hall effect signal is presented. This permits an easy tracking of the RPM signal.
This is an important debugging feature and allows identify a sensor malfunction. The gauge below it
indicates the engine RPM . The last gauge on the left side shows the fuel mass measurement. Its ”fuel
tank” design, allows the test operator to quickly understand the fuel level preventing the engine to run out
of fuel (situation that should be avoided to prevent air bubbles on the fuel system). Lastly, on the bottom
left side, the engine setup window is displayed. It enables the user to change the PWM characteristics
according to the servomotor, as well as setting the throttle position at which the engine idles.
Similarly to the engine, on the top right the hall effect signal and the RPM gauge are displayed.
Below it, there is an indication of the battery voltage, current and power. This way, user can easily
monitor the battery status, so as to avoid any situations that may damage the battery. Nonetheless, this
verification is also made automatically by the software where some safety features were programmed.
These automatically stop the system if any overcurrent, over or undervoltage situation is verified. The
limits can be changed and activated by the user on the bottom centre window of the GUI. The bottom
right box features the ESC PWM configuration.
46
Figure 4.20: LabVIEW Graphical User Interface of the parallel hybrid-electric test bench
On the top centre of the LabVIEW screen is a button to start recording data of the system and to the
the left of it, a LED that indicates whether this process is working correctly, saving data into a .lvm file.
The user can also keep track of the overall testing time through the indicators on the right.
Below it, the control panel is shown. On the top of this window, five LED lights inform the user the
current operating state of the propulsion system: ICE ONLY, EM ONLY, REGEN, DASH or RESET. The
last mode covers any other combination of ICE and EM throttle commands that are not predicted by any
other mode and are, in theory, not possible when running the system.
In the top right of the control panel there are two other small windows. The bottom one allows the
operator to switch between the Manual Power Management (MPM) mode, which allows the user to
manually control the throttle, and the Intelligent Power Management (IPM) that features the supervisory
controller described in section 3.3.4. To the right of this toggle switch, there is another LED that informs
the user if any error has occurred while running the controller with indication of the correspondent code
underneath. Over that box there are four toggle switches. These allow the user to switch between
operating control modes: Reset, Single ICE operation, Single EM operation, or Hybrid as discussed
in Section 3.3.4. The top central knob on the control panel defines the torque request of the system
and is only active when the control is set to IPM. Finally, the centre bottom left green bottom stops the
execution of the system in nominal conditions while the emergency button also abruptly shuts down the
engine and the motor.
The LabVIEW block diagram is presented in Appendix D.
47
4.3 Controller implementation
The propulsion system controller (depicted in Figures 3.32 and 3.33) was programmed using Math-
works MATLAB and featured in the LabVIEW2018 block diagram as shown in Figure 4.21. In this setup,
changing to a different controller is straight forward as the user would only need to change the MATLAB
script which LabVIEW2018 calls when executing this block.
Once the IPM is engaged, this block becomes active, running the controller script every 1s. The script
receives as input the desired operating mode, represented by MODE IN variable, the engine throttle
idle position, the torque request and the engine’s rotational speed. So as to provide robustness to errors
in the rotational speed measurement, the engine RPM signal is averaged out. The controller block
outputs the engine and motor throttle values as well as an error code indicating a non critical system
malfunction, predicted by the controller. Furthermore, if a major error occurs while running the MATLAB
script and LabVIEW is not able to execute it, a separate case structure was included as a safety feature.
It sets the engine to idle, turns off the motor, switches the control to MPM and informs the user through
a pop up message.
The controller code is shown in Appendix A. Firstly it uploads the engine and motor information stored
in .mat files. This includes information on the IOL operation and engine and motor throttle mapping look
up tables. Subsequently, a switch statement reads themode in variable and executes the corresponding
case. The controller continuously checks for invalid inputs or code mal-function and communicates these
errors through the ERROR CODE window in the GUI. The description of the possible errors is shown
in Table 4.1.
Error code Description
0 No error
1 The user selected ICE Single Operation, but the Torque Request value is too low. That wouldlead the engine to stall. The engine was placed idling
2 The user selected ICE Single Operation, but the Torque Request value is higher than themaximum torque the engine is able to provide. The throttle was set to its maximum, 100%.
3 The user selected EM Single Operation, but the Torque Request value is higher than themaximum torque the motor is able to provide. The throttle was set to its maximum, 100%.
4 The user selected HYBRID operation, but the Torque Request value is higher than the onethe hybrid system can provide.
5 The user selected HYBRID operation, but the engine rotational speed is too low. This meansthe look-up tables do not have information for that operating point.
6 The controller jumped to an unexpected state. This indicates a malfunction of the controller.The user is advised to check the code.
Table 4.1: Hybrid controller error codes
48
Figu
re4.
21:
Con
trolle
rblo
ck
49
50
Chapter 5
Results
With the implementation of the system complete, the focus now shifts to testing and validation. The
first section of this chapter focuses on testing and fully mapping each mode of the HEPS. The second
presents the controller performance results.
For each test, Standard Operating Procedure (SOP) where developed to assure consistent, correct
and safe testing procedures. These are presented in Appendix C. The measured variables included
rotational speed of the system, RPM , battery current draw, Current ([A]), battery voltage, V oltage ([V ])
and fuel mass, [g]. The propeller torque, Torqueprop ([N ·m]), the propeller power, Powerprop ([W ]), the
battery electrical power, Powerbat ([W ]), the fuel flow, [g/min], the BSFC [g/(kW · hr)] and the motor
torque τEM , ([N ·m]), were obtained through post-processing. Before each test, the batteries were fully
charged, the current and fuel measurements tared and the engine left operating at idle for a few minutes
so it would warm up.
5.1 System characterisation
In this section we explore in detail each operational mode of the system. Whenever possible, com-
parisons are established to the component level testing carried through in Section 3. This comparison is
paramount to understand the implications of featuring common COTS components into a non conven-
tional configuration such as a hybrid powertrain.
This test campaign is composed of four main tests:
1. Electric motor only
2. Internal combustion engine only
3. Dash
4. Regenerative braking (regen)
51
5.1.1 Electric motor only testing
This test aimed to map the performance of the electric motor on the hybrid powertrain. The obtained
maps were then used to update the controller as described in Section 3.3.4.
The test consisted on sweeping the EM throttle from 0% to 90% in steps of 5%, keeping the ICE
turned off. This test was repeated three times in order to mitigate possible bias.
Presented in Table 5.1 are the average of the results obtained, while Figures 5.1 through 5.4 present
also present the standard deviation1 of the measured quantities.
EMthr RPM Torqueprop τEM = I−I0Kv
Powerprop Current V oltage Powerbat η [%]
0 0 0 − 0 −0.001 25.139 −0.039 −10 1077.490 0.0637 − 7.189 0.863 25.101 21.684 33.154
15 1637.598 0.1444 0.03 24.775 2.006 25.046 50.258 49.295
20 2079.918 0.2307 0.074 50.259 3.468 24.971 86.617 58.025
25 2481.822 0.3263 0.130 84.824 5.262 24.878 130.933 64.784
30 2794.675 0.4121 0.192 120.623 7.245 24.775 179.507 67.196
35 3033.0155 0.4841 0.252 153.786 9.171 24.672 226.282 67.962
40 3296.136 0.5704 0.324 196.885 11.461 24.548 281.358 69.976
45 3531.591 0.6535 0.407 241.685 14.117 24.402 344.519 70.151
50 3774.494 0.7451 0.501 294.528 17.099 24.240 414.490 71.057
55 3979.335 0.8270 0.596 344.654 20.140 24.070 484.805 71.091
60 4168.709 0.9066 0.703 395.776 23.546 23.878 562.257 70.390
65 4317.672 0.9717 0.794 439.363 26.477 23.698 627.469 70.021
70 4485.517 1.0478 0.901 492.178 29.896 23.496 702.460 70.064
75 4656.808 1.1283 1.033 550.272 34.095 23.262 793.139 69.379
80 4672.875 1.1361 1.046 555.944 34.529 23.176 800.264 69.470
85 4679.337 1.1392 1.056 558.236 34.827 23.121 805.264 69.323
90 4794.725 1.1954 1.148 600.235 37.767 22.972 867.633 69.180
Table 5.1: Electric motor testing results
It is possible to verify a difference between Torqueprop and τEM , when, in theory, they should present
the same value. However, the difference is small and, since both these quantities were obtained through
post processing, may be due to imprecise measurements of RPM and Current
In Figures 5.1 and 5.2 a constant increase in both rotational speed and current draw is observed
as the throttle is swept. Despite that, an unexpected behaviour occurs at 75%. It is verified that the
current increases only slightly and is only when the throttle is increase to 90% that it is possible to see
reasonable increase in current. This behaviour is highly dependent on the programming of the ESC, a
possible malfunction might be present. However, a possible reason might be the considerable voltage
dip, shown in Figure 5.3, near the high end of the throttle making the ESC incapable of drawing a higher
current. In fact, a change in slope is observed. Using a higher capacity battery or increasing the number
of batteries in parallel might help to mitigate this problem. Despite that, is is important to note that the
battery voltage always remained over its the nominal value (22.2V ).
1The standard deviation, σ, defined as σ =√∑
(xi−µ)N
, where µ represents the mean, N the number of data points, and xithe data point.
52
Figure 5.1: RPM results for the EM ONLY test Figure 5.2: Current results for the EM ONLY test
Figure 5.3: Voltage results for the EM ONLY test Figure 5.4: Efficiency results for the EM ONLY test
Another important aspect is the low standard deviation observed between tests. This depicts the
consistent behaviour of the EM. The relatively higher value presented in Figure 5.3 is due to different
initial SOC of the battery between tests.
Considering Figure 5.4, it is possible to verify a maximum efficiency of 71% close to 3900RPM .
Indeed, this goes in line with the component level experiments described in Section 3.2.2, where the
maximum efficiency was verified at approximately 3600RPM . However, the values differ approximately
by 10%. This can be attributed to powertrain losses such as friction. The overall shape of the curve is
similar in both tests.
Figure 5.5 presents the Torque-Throttle curve where we can see a clear linear trend from 15% to 75%
which facilitates the control process. This was used as a look up table for the controller (as explained in
Section 3.3.4).
5.1.2 Internal Combustion Engine testing
Performance of the ICE can be analysed by monitoring the ICE’s fuel consumption, RPM and Torque
output through several power demand sweeps. The test consisted on sweeping the engine throttle from
40% (idle) to 65%. At each throttle position, the engine was run for 5 minutes so as to permit a more
accurate fuel flow calculation. Presented in Table 5.2 are the main results of this test campaign. Figure
53
Figure 5.5: Torque as a function of throttle for the AXi 4130/20
5.7 through 5.8, shown the data on a graphical form where the predicted manufacturer data for the same
working point was overlaid. The main objective of this test was to map the Torque − throttle curve for
this engine in order to update the estimation of the controller as well as serve as a baseline comparison
for the other working modes.
Throttle RPM Torqueprop [N ·m] Powerprop Fuel flow BSFC η a [%]
40% 4760.656 1.1787 587.622 19.155 1955.846 4
45% 5742.415 1.7084 1027.336 22.152 1293.753 6
50% 6097.704 1.9242 1228.698 24.812 1211.623 7
55% 6620.926 2.2654 1570.696 22.222 848.874 10
60% 6691.156 2.3133 1620.920 23.222 859.585 9
65% 6727.616 2.3383 1647.365 21.232 773.307 11
Table 5.2: Internal combustion engine testing resultsaA energy density of 44.4MJ/kg was used for the gasoline [46]
The first consideration is that when installed on the hybrid powertrain, the engine was only able to
idle at 40% throttle. In fact, it quickly stalled when given a throttle percentage lower than 40%. In contrast,
during the tests conducted with a the conventional setup, presented in Section 3.2.1, the ICE was able
to idle at 10% while operating a larger propeller (22×10). Only near higher rotational speeds and throttle
inputs was the engine able to consistently and smoothly operate, presenting what was considered a
nominal behaviour. This will be later illustrated in Section 5.2.2. This might indicate that the engine
is not powerful enough to cope with the friction losses of the hybrid powertrain. Indeed, these losses
had a great impact in the overall engine performance. Moreover, while the manufacturer recommends a
22inch× 10 propeller, it was verified that the engine presented severe difficulties when operating such a
propeller. This led to the use of a smaller propeller (lower load) and hence the choice of a 19inch × 10
propeller.
Referring to Table 5.2, it is possible to see the low fuel efficiency of the engine. In fact, this efficiency
reaches a higher values at higher rotational speed but never goes above 11%, a way lower than the
efficiencies presented by the electric motor. This goes in line with the discussion of Chapter 1.
54
Figure 5.6: Torque as a function of Throttle for theICE ONLY test
Figure 5.7: Power as a function of Throttle for theICE ONLY test
Summarised in Figure 5.6 are the obtained torque results. Both curves indicate a clear growing trend
as we increase the throttle and RPM up to roughly 55%. After that point, the slope decreases and the
engine is not able to provide more torque. In fact, further throttle increases past 65% did not induce
a significant difference in rotational speed. For this reason, the tests were only conducted up to that
throttle value. Additionally, there is a significant difference between the obtained values and the ones
predicted by the manufacturer especially in the low end throttle were the difference can reach 46%. As
throttle is increased, the difference reduces to roughly 15%. The calculated power output, shown in
Figure 5.7, presents a similar trend.
Figure 5.8: Fuel flow as a function of Throttle for theICE ONLY test
Figure 5.9: BSFC as a function of Throttle for theICE ONLY test
Furthermore, the fuel flow results were against the expected. These are summarised in Figure 5.8.
In fact, with an increase in throttle and rotational speed, an increasing fuel consumption was expected
as shown by the manufacturer data presented in orange. Instead, they do not exhibit a clear trend.
Nevertheless, due to the lack of statistical information (the tests were only run once), it becomes difficult
to evaluate and narrow down the possible causes. Despite the relatively long run time, the author
believes this test should be repeated several times (at least 5) in order to obtain trustworthy results.
Despite that, the results indicate an underestimation of the fuel flow by the manufacturer.
The BSFC results are shown in Figure 5.9. These present a decreasing trend, meaning the engine
55
efficiency increases with increasing RPM and throttle. In contrast, the manufacturer data presents an
opposite behaviour. It indicates an overall slight increase in BSFC with a minimum at 55% throttle
and 6620.926RPM . From a global perspective, the difference between the obtained results and the
manufacturer data is significant. This can be attributed to the cumulative difference of the power and
fuel flow values.
Figure 5.10: Torque as a function of throttle for the ICE
In Figure 5.10 the Torque-Throttle curve is depicted. It is possible to observe that a close to linear
behaviour is verified up to 55%. After this point, however, with increasing throttle the engine is not able to
provide a higher torque value. Given the second order polynomial characteristic of the propeller curve,
the engine might not be able to steadily operate at higher RPMs than approximately 6700RPM since it
would result in a considerable increasing in torque demand from the engine. In fact, according to the
manufacturer, the maximum output torque for this engine is 2.99N ·m at 5500RPM with full open throttle.
5.1.3 Dash testing
This test aims to investigate the dash working mode and understand how the two power sources
interact with each other when operating in tandem.
The test consisted of keeping a constant ICE throttle, 40%, while sweeping the EM throttle from 0%
to 60% in steps of 10%. This test was repeated twice.
ICEthr EMthr RPM Torqueprop Powerprop Current V oltage Powerbat
40%
0% 4625.237 1.1305 551.824 −0.002 24.932 −0.06
10% 4661.617 1.1307 551.966 2.937 24.842 72.979
20% 4815.543 1.2057 608.013 7.260 24.682 179.203
30% 5151.732 1.7800 743.414 12.553 24.448 306.912
40% 5568.715 1.6076 937.479 17.488 24.197 423.172
50% 5728.217 1.7000 1019.757 19.876 24.032 477.674
60% 5741.626 1.7079 1026.894 19.069 24.003 457.743
Table 5.3: DASH mode testing results
56
It is possible to see through the change in RPM and battery current that this mode works. Indeed,
the motor is able to assist the engine in its operation. Furthermore, the clutch system utilized in this
project successfully combines the power of these two power sources despite the intricate mechanical
interaction. This test also indicated that the EM never manages to completely overrun the ICE during
dash mode. Therefore, if the ICE remains at idle during electric only mode operation, it will still provide
power to the system.
An important aspect is that at 40% ICEthr, there is a slight change in RPM compared to the one
verified in Section 5.1.2. This behaviour was common throughout the whole test campaign, where the
engine would present slight differences in behaviour in different test runs. This is also visible by the
large standard deviation in the rotational speed measurement at 0% EMthr presented in Figure 5.11.
This further emphasizes the need of repeated test runs. Nevertheless, the main focus of this test was to
analyze the impacts of the tandem action of the ICE and EM and hence this disparity does not invalidate
the obtained results.
Figure 5.11: RPM results for the DASH test Figure 5.12: Current results for the DASH test
Figure 5.12 shows the current measurements. The low standard deviation value indicates consis-
tency on the current flows between tests. As the throttle is increased, the current also increases up
to 50% EMthr. For throttle percentages over this value, the current draw decreases. Furthermore, as
shown in Figure 5.11, the RPM presents a decreasing slope on the high end of the throttle. This was
an unexpected behaviour and possibly is attributed to the high rotational speeds where the motor is no
longer able to operate effectively. This way, the tests were only conducted up to that throttle percentage
to avoid damaging the motor. The following analysis excludes that operating point. A possible solution to
this problem might be to install a different gear ratio between the motor and engine pulleys. This would
allow to step down the rotational speed perceived by the motor, enabling it to work at a lower speed, and
thus more effectively.
Table 5.4 highlights the influence of the EM action on the system performance, where Powerbat was
repeated for ease of visualization. The variables ∆Torque [N · m] and ∆Powerprop [W ] indicate the
change in torque and power, respectively, relative to 0% EMthr, as we increase the throttle value. The
electric motor torque, represented by τEM , follows an increasing trend but does not directly correspond
to the change in Torque output. This might be due to the change in torque output of the engine pro-
57
ICEthr EMthr ∆Torqueprop τEM = I−I0Kv
∆Powerprop Powerbat∆PowerpropPowerbat
[%]∆PowerpropPowerprop
[%]
40%
10% 0.0002 0.05 0.141 72.979 0.194 0.02
20% 0.0752 0.193 56.188 179.203 31.354 10.17
30% 0.2475 0.358 191.589 306.912 62.424 31.51
40% 0.4771 0.513 385.654 423.172 91.134 51.87
50% 0.5695 0.588 467.932 477.674 97.960 49.91
Table 5.4: Detailed results for DASH mode
vided the difference in RPM. In fact, the analysis carried through in Section 3.2.1 proves exactly that.
Nevertheless, the change in rotational speed is relatively low (approximately 1000RPM ), and therefore,
the torque output of the engine is not expected to change drastically. It is also important to note that
the system total output torque measurement method applied in this research is not sensitive to small
changes in torque.
Nonetheless, assuming a constant engine output torque, the results of ∆PowerpropPowerbat
suggest that the
higher the throttle value, the more efficiently the motor can provide power to the system and assist the
engine. This is also indicated by the increasing slope in Figure 5.11. Furthermore, ∆PowerpropPowerprop
, indicate
a higher power hybridization with increasing EMthr.
ICEthr EMthr ∆Torqueprop τEM = I−I0Kv
EM ONLY Torqueprop EM ONLY τEM
40%
10% 0.0002 0.05 0.0637 −20% 0.0752 0.193 0.2307 0.074
30% 0.2475 0.358 0.4121 0.192
40% 0.4771 0.513 0.5704 0.324
50% 0.5695 0.588 0.7451 0.501
Table 5.5: DASH mode and EM ONLY mode comparison
In Table 5.5 is presented a comparison between the obtained ∆Torque results and expected torque
obtained during the EM testing described in Section 5.1.1, repeated here for ease of visualization. There
is difference between the two values, indicating a clear difference in behaviour of the motor when oper-
ating in this mode. The disparity in τEM result in a different current flow. Despite that, is is important to
emphasize that the data was gathered in two very different tests. In fact, as discussed in Section 3.2.2
and depicted in Figure 3.28, the maximum torque of a motor decreases with increasing RPM .
Overall, it was verified that for the same EMthr percentage, the motor was able to draw more current
the lower the RPM.
5.1.4 Regenerative braking testing
This test aimed to investigate the regenerative braking (regen) working mode and understand how
the system behaves when the motor acts as a generator and withdraws power from the engine to charge
the batteries.
The test consisted of keeping a constant ICE throttle, 60%, while sweeping the EM throttle from 0%
58
to −90% in steps of 10%. This test was repeated three times.
Figures 5.13 and 5.14 present the rotational speed and current measurements respectively. The
negative sign in the measured current indicate that it is flowing into the batteries, charging them. It is
possible to observe that the values of standard deviation are relatively low, with the clear exception of
the RPM for −70% EMthr. This means, the tests were consistent between eachother.
Furthermore, there is an increasing negative slope of the RPM with a sharp decrease near −70%.
This can be explained by the increasing load on the engine that lead the engine to set at a lower RPM.
The current presents an approximately linear and constant decrease, reaching a maximum value of
−12.848A. Higher current draws can be achieved by changing the ESC parameters described in Section
3.2.2.
Figure 5.13: RPM results for the REGEN test Figure 5.14: Current results for the REGEN test
Table 5.6 summarizes the results obtained. It is possible from the increase in current flow and
voltage of the batteries that this mode clearly works. The current is negative throughout the whole range
meaning the batteries were always being charged. Despite that, there is an increase in RPM is verified
from 0% to −10%. Nevertheless, the difference is small and can be attributed to the unsteady nature of
the engine RPM.
ICEthr EMthr RPM Torqueprop Powerprop Current V oltage Powerbat
60%
0% 6629.834 2.2714 1576.975 −0.003 24.178 −0.091
−10% 6666.666 2.2965 1603.259 −0.915 24.187 −22.151
−20% 6629.834 2.2714 1576.975 −2.499 24.252 −60.628
−30% 6575.391 2.2346 1542.901 −4.336 24.317 −105.447
−40% 6451.612 2.1519 1453.848 −6.568 24.404 −160.293
−50% 6315.789 2.0630 1364.443 −8.458 24.483 −207.088
−60% 6217.616 1.9990 1302.149 −9.435 24.545 −231.603
−70% 5769.230 1.7242 1041.679 −10.554 24.609 −259.726
−80% 5581.395 1.6148 943.822 −11.717 24.674 −289.128
−90% 5555.555 1.6000 930.842 −12.848 24.746 −317.956
Table 5.6: REGEN mode testing results
Table 5.7 exposes the decrease in the supplied torque to the propeller. By comparing it to τEM we
59
ICEthr EMthr ∆Torqueprop [N ·m] τEM = I−I0Kv
∆Powerprop [W ] Powerbat [W ]
60%
−10% +0.0251 −0.06 +26.284 −0.091
−20% 0 −0.113 0 −22.151
−30% −0.0368 −0.170 −34.074 −60.628
−40% −0.1195 −0.240 −123.127 −105.447
−50% −0.2084 −0.299 −212.532 −160.293
−60% −0.2724 −0.330 −274.826 −207.088
−70% −0.5472 −0.365 −535.296 −231.603
−80% −0.6566 −0.401 −633.153 −259.726
−90% −0.6714 −0.437 −646.133 −317.956
Table 5.7: Detailed results for REGEN mode
verify that the values are close, clearly indicating that the power withdraw by the motor. However, the
change in propeller power and the battery power differ especially in the high end throttle values. This
may indicate a lower engine torque output at this rotational speed. Finally, it was verified that for the
same negative EMthr, the motor was able to draw more current the higher the RPM (the higher the
available power).
5.2 Controller performance
In this section, the developed controller is tested. The test campaign involved evaluating the system
response to several step functions as well as investigating its dynamic response to changing inputs.
Thus, each mode was tested separately through a sample mission. The latter consisted of three dif-
ferent portions with different Torque Request values. The test campaign was composed of three main
milestones:
1. Single EM operation
2. Single ICE operation
3. Hybrid
All tests were run while trying to minimize changing variables to mitigate possible bias. This way, the
same standard procedures described in Section 5.1 were followed.
5.2.1 Single EM operation
Mission profile
Figure 5.15 depicts the mission profile for the Single EM test. The Torque Request values were
chosen according to those found in the system’s normal operation. The test was run three times.
60
Figure 5.15: Single EM operation mission profile
Results
Referring to Figure 5.16, it is possible to observe the controller is able to follow the Torque Request
closely. It shows the highly consistent motor performance. This makes the controller design easier and
more straight forward. In fact, it indicates that the mapping of the Torque−Throttle curve closely depicts
the system behaviour, and the open control loop, in this case, is successful.
In Figure 5.17 we see the throttle commands of the controller throughout the mission. As discussed
earlier, its inputs are only based on the requested torque value, ans thus, a constant value throughout
each mission leg it verified.
Figure 5.16: Propeller torque during Single EM op-eration mission
Figure 5.17: Throttle variation during Single EM op-eration mission
Figure 5.18 presents the RPM measurements, along with an indication of the standard deviation,
σ, for each point. A relatively low σ value is observed, further corroborating the consistent motor per-
formance. The average results are displayed in Figure 5.19 and clearly show a near constant speed
operation, with a number of negligible spikes.
The current measurements are shown in Figure 5.20 and 5.21. These present a negligible standard
61
Figure 5.18: RPM results for Single EM operationmission with indication of standard deviation
Figure 5.19: Average RPM results for Single EM op-eration mission
deviation on the the first two portions of the mission while a considerable value in its last leg. Further-
more, a sudden jump in current is verified despite the constant motor throttle command. This induces
an increase in the torque output of the motor and thus an increase in rotational speed is verified. This
is an unexpected behaviour and no possible cause could be advanced at the time this document was
written.
Figure 5.20: Current results for Single EM Operationmission with indication of standard deviation
Figure 5.21: Average current results for Single EMOperation mission
5.2.2 Single ICE operation
Mission profile
The mission profile is presented in Figure 5.22. It aimed to test the engine in different operating
points, also allowing to analyzing its dynamic response.
Results
Referring to Figure 5.23, the throttle input calculation is only based on the requested torque and
thus, it remained constant throughout each mission leg. In Figure 5.24 it is possible to observe the
62
Figure 5.22: Single ICE Operation mission profile
torque results for this test. We verify that for the first portion of the mission, the measured torque is
close to the requested one, meaning a successful mapping of the Torque-Throttle curve for this specific
operating point. The same is not verified for the last two portions of the mission where the values
drastically differ. This illustrates the engine’s inconsistent behaviour and the difficulties in accurately
mapping a Torque-Throttle curve.
Figure 5.23: Throttle variation for ICE Single opera-tion mission
Figure 5.24: Propeller torque during ICE Single op-eration mission
Figure 5.25 shows the rotational speed measurements. The overall signal is highly noisy in a clear
contrast to the electric motor operation as discussed earlier. This behaviour is a result of the torque
spikes during the engine operation. It is possible to see that the irregularities and spikes increase in
amplitude as the rotational speed is reduced. In fact, on the transition from the first to the second leg
of the mission, a decrease in throttle leads to a quick drop in rotational speed. This indicates a rapid
engine response to changing throttle commands. The engine then sets at that operating point with a
relatively constant rotational speed. However, near the end of the second portion of the mission, large
variations in rotational speed are verified with a severe drop in RPM when the system transitions to the
last mission leg. Despite that initial sharp decrease in rotational speed, the system accelerates again
but this time does not set at a defined RPM. In contrast, it presents a highly unstable behaviour where
no clear trend cannot be extracted. This illustrates the discussion carried through in Section 5.1.2 and
the difficulties in controlling such a power unit, where it is difficult to predict its torque output, in addition
63
to the challenges raised by its unsteady nature.
Figure 5.25: RPM results for ICE Single operation mission
In Table 5.8 is possible to observe the different fuel consumption rates in each section of the mission
as well as the overall fuel consumption. It shows a rather similar fuel consumption in the first two sections
of the mission, while presenting a clear reduction in fuel flow in the last portion, due to the lower rotational
speed and throttle value.
Section Fuel Flow [g/min] Fuel [g]
1 24.50 61.15
2 24.62 61.45
3 22.62 56.46
Total 24.11 180.74
Table 5.8: Fuel consumption results for ICE Single Operation Mission
Finally, it is important to note that at the time of writing of this document, no repeated runs were
conducted. This is of paramount importance to assess if the unexpected behaviours verified here are
consistent, to correctly evaluate the Torque-throttle estimation method, and accurately assess the per-
formance of the controller.
5.2.3 Hybrid
Mission profile
In order to directly compare the controller performance between the ICE Single Operation and the
Hybrid, the same mission profile was used for this test and is represented in Figure as 5.22.
Results
In Figure 5.26 is presented the throttle percentages of both the EM and the ICE. It is possible to
see that the controller behaves as expected and, tries to position the system in a state where it would
output the requested torque, according to its look-up tables. When the controller is engaged, it reads
the rotational speed and quickly calculates what should be the torque output of the engine, in order to
operate it at its most efficient point. It then consults the Torque-Throttle curves previously mapped in
64
Section 5.2.2 to set a throttle command, which in this initial moments is approximately 48.6%. The electric
motor is responsible for making up the difference between the engine torque and the requested one.
This way, the controller accesses the Torque-Throttle curves mapped in Section 5.2.1 and sets a throttle
command, in this case, approximately 32%. From Figure 5.29, in this initial segment, a positive current
flow is visible meaning the motor is successfully providing torque to the system, which is presented
in Figure 5.30. After an initial period where the rotational speed remains approximately constant at
6000RPM , the system drastically accelerates to 6600RPM as depicted in Figure 5.28.
Figure 5.26: Throttle variation for hybrid mission Figure 5.27: Propeller torque hybrid mission
As the system accelerates, the controller again calculates the optimum torque output for the engine
and sets the correspondent throttle position, this time at approximately 49.5% for the ICE and 28% for
the motor.
However, for this rotational speed, the motor is no longer able to operate effectively and a continuous
change in current value is verified in Figure 5.29 with an average close to 0A, despite the relatively
constant EMthr value. Comparing this to Figure 5.21, illustrate the important role an unsteady rotational
speed plays in the performance of the motor.
Given the interaction between mechanical, electric and power electronics components in this system,
it is difficult to assess the reasons of such behaviour. However, a possible cause might be advanced.
Given the high rotational speeds, the back EMF generated might surpass the difference in potential the
ESC sets at that specific throttle positions and thus the current starts flowing in the opposite direction.
This prevents a correct control of the system. Furthermore, as emphasized in the aforementioned sec-
tion, the system behaviour when working in tandem and separately can be significantly different. This
inhibits a successful implementation of the method used to estimate the motor throttle, based on the
single EM operation as also illustrated in Table 5.5. This behaviour was previously identified in Section
5.1.3.
During the remaining of the mission, the system shows an insensitivity to the input Torque Request,
despite the change in EMthr, and continues to operate at roughly the same rotational speed. The
correspondent torque is always higher than the requested one (refer to figure 5.27). It is important to
note that, at this rotational speed, the motor is no longer working effectively, as shown by the close to
zero current values in Figure 5.29.
65
Figure 5.28: RPM results for hybrid mission
Table 5.9 shows the fuel consumption data. It indicates a constant fuel consumption rate, due to
the near constant rotational speed and engine throttle value. There is a 3% reduction in fuel consump-
tion relative to the ICE Single Operation mission. Nevertheless, it is important to emphasize that the
observed torque output values were different.
Section Fuel Flow [g/min] Fuel [g]
1 23.24 58.01
2 23.35 58.19
3 23.26 58.07
Total 23.38 175.26
Table 5.9: Fuel consumption results for Hybrid Mission
Finally, similarly to the previous test, at the time of writing this document no additional test runs were
conducted. Nonetheless, they have a paramount importance in obtaining trustworthy conclusions on the
system’s performance.
Figure 5.29: Battery Current during hybrid mission Figure 5.30: Electric Motor Torque during hybridmission
66
Chapter 6
Conclusions
This chapter presents the final conclusions for this thesis, its major achievements as well as recom-
mendations for future work.
6.1 Achievements
The goal of this research was to design and implement a HEPS for a small UAV, while addressing
important research questions outlined in Chapter 1.
Firstly, it proved that HEPS can indeed provide advantages in effectiveness for a small UAV. A com-
prehensive test campaign was conducted that allowed to fully map each of the four operating modes.
The results of Chapter 5 proved that the system is reliable, safe and is able to enhance the versatility of
an UAV by providing it with a range of different operating modes. In particular, the system was able to
successfuly charge the battery packs using the internal combustion engine when on regenerative brak-
ing mode, where a maximum of approximately 13A was registered. Moreover, when working together,
the electric motor was able to assist the engine, increasing the overall system rotational speed, while
drawing a current of up to 19A.
In order to achieve these goals, a prototype system was conceived. This began with component
selection, where each one was fully tested and its performance analyzed, bearing in mind its future
implementation into the propulsion system. Thereafter, a test platform was needed to assess the per-
formance of the developed system. This led to the design of a test bench. A thorough design process
was conducted to achieve a fully versatile and robust solution. The final design featured the possibility
to change any component of a hybrid-electric propulsion system with relative ease, allowing to test any
parallel hybrid-electric propulsion system, while representing a safe and reliable test platform. The test
bench includes a sensor suite to evaluate the system performance and provide real-time measurements
to a control strategy implemented. The data was gathered using the NI cDAQ-9188.
To properly interface with the test bench, a PCB was used to provide a user-friendly operation where
the test operator only needs to connect the sensor to its indicated slot on the PCB, with no required
knowledge of the system electrical design. A harness was then used to connect the PCB to the cDAQ.
67
Finally, NI LabVIEW was used to setup a GUI to enable the test operator to control, monitor and ac-
quire data from the tests. Furthermore, it allows a full customisation of every possible variable in the
system, while featuring debugging capabilities to quickly identify a possible sensor malfunction. This
LabVIEW routine features the possibility to perform component and mode level testing in Manual Power
Management mode or assess the full system performance and validate a control strategy by switching to
Intelligent Power Management. Again, the ease of implementation was deemed paramount and so, the
software was programmed to allow a change in control strategy by simply switching a MATLAB script.
Failure cases are predicted and the software automatically informs the user. Finally, given the high power
of these propulsion systems, the LabVIEW routine includes safety features that automatically stop the
test if a dangerous situation is verified, shutting down the power units immediately.
Secondly, this research addressed the energy efficiency of such as system by developing and suc-
cessfully implementing supervisory controller that automatically drives the system based on the current
operating mode, the system’s rotational speed and the user’s torque request. The controller offers three
different operating modes, ICE Single Operation, EM Single Operation and Hybrid. On the last one, a
rule-based control strategy was implemented based on the ideal operating line. It aimed to improve the
efficiency of the overall system by operating the engine at its most efficient point while the motor assists
it whenever necessary. This controller allowed to evaluate the system dynamic behaviour to changing
inputs as well as its open-loop response to a step function. So as to accomplish it, sample missions were
created to characterise the controller and permit a trustworthy comparison between modes, namely on
the fuel consumption regard.
Throughout this process and discussed in the appropriate sections, the main challenges of designing,
implementing and controlling HEPS were assessed. In fact, this research provided insight on how the
two power units interact with each other when working together. It also pointed out the invalidity of some
assumptions used in simulation work and provided guidelines for further enhancing them. An example of
this is the motor inability to assist the engine at high rotational speeds. This means the effectiveness of
the dual mode is limited in RPM. Another problem encountered was the highly inconsistent performance
of the small two-stroke ICE, which raises difficulties when controlling the system. The electric motor,
on the other hand, proved a reliable and stable behaviour with repeated runs showing a low standard
deviation value.
Finally, this research offered an overview over the hybrid-electric propulsion topic from a global per-
spective by addressing each step of the development process. From conceptually outlining how such
a system could work, to selecting and testing its individual components, to the design of a powertrain
to couple them, and finally to the development of a controller strategy to totally explore the system
capabilities. This culminated with a test campaign.
6.2 Recommendations and future work
The work presented here is part of an ongoing research effort at UVIC-CfAR to investigate hybrid-
electric propulsion in the UAV industry. It represented the first ever realization of a a HEPS implementa-
68
tion and comprehensive experimentation at the center.
The developed test bench opens the door to a wide range of possibilities. In fact, it can be used
to validate the results of a simulation software developed to depict the system behaviour. On the other
hand, the data retrieved on the test bench could be used to upgrade its models, providing more accurate
results.
Another possibility is testing different controller strategies. After a control strategy is successful in
simulations, bench testing it follows. As described in Chapter 2, a lot of research has been conducted
on hybrid automobiles control strategies, but only a few to UAVs. A valuable contribution can thus be
made by simulating and bench testing different control approaches. Moreover, the gathered data, can
even be used tun the controller and improve its performance. A simulated flight performance can also
be assessed by providing the system real flight data.
Furthermore, given the versatility of the test bench, it is possible to test different hybrid components.
Indeed, as highlighted in Chapter 5, changing the gear ratio of the system might allow the motor to
operate at a more efficient speed. Another suggestion would be to use a more powerful engine or
featuring it with a fuel injection system to improve performance.
More system capabilities can be explored through the use of a clutch mechanism. This would allow
the EM to start the ICE, without needing any external power source. In addition, by disengaging the
clutch, the system could work solely with the EM and completely disengage the ICE, which is not possible
with the one-way bearing. Other interesting possibility is implementing gearing between the propeller
and the propulsion system, to enable them to work at their maximum efficiency speed.
Despite the success, this research also faced severe obstacles. The main one was the lack of an
accurate output torque measurement. The method applied was able to qualitatively describe the system
performance but it failed whenever an in depth analysis was pursued. This prevented a more thoroughly
analysis of dash and regenerative braking modes for instance. In fact, unexpected behaviours were
verified on both these modes, where tracking the engine and total system torque output would be very
useful. This way, a rotary torque sensor should be feature to assess this variable. In addition, pressure
sensors could be implemented on the ICE to better estimate its torque.
Another limitation of the test bench regards its fuel consumption measurements. Currently based on
the difference in fuel mass over a time period, this method requires long engine run times and precludes
continuous and real time tracking of the fuel rate, an important measure when optimizing the system.
This way, a flowmeter might be of interest. However, these units can be expensive but would be an
important addition for future iterations to the sensor suite.
An important aspect of the test bench is its friction resistance. This was specially visible in the engine
performance. A change to high speed, low resistance bearing blocks used is, hence, suggested.
Furthermore, repeated runs of the same tests could not be done due to time constraints and are left
as a recommendation for future test campaigns. This is an important process to gain statistical accuracy
and validate the results.
Regarding the controller developed, a closed control loop operation is recommended to allow each
power unit to follow a defined requested torque, where the sensors would continuously feed back the
69
error value. A Proportional-Integral-Derivative (PID) controller might be ideal for a first iteration of this
system.
Further investigation should also focus on understanding the ESC. In fact, it was operated as a
”Black Box”, and in dash and regenerative braking operational modes, a more careful analysis to the
power electronics installed should be pursued. The ESC Enertion Boards FOCBOX is based on an
open source software and therefore there is comprehensive documentation available online.
To conclude, the ultimate validation tier in the hybrid-electric propulsion project includes ground and
flight testing the HEPS. At this point, the system will be sized and optimized for the QT1 aircraft. The
QT1 aircraft, shown in Figure 6.1, is a fully-electric, highly versatile flight test vehicle developed by
UVIC-CfAR. Due to its accurately mapped performance characteristics and high payload mass, the QT1
aircraft represents a natural choice for testing a new propulsion system. However, the added weight
and volume of the additional components demand a thorough redesign study. A CAD render of how the
system could be implemented is shown in Figure 6.2.
Figure 6.1: QT1 aircraft
70
(a) Aircraft integration (b) Detailed view
Figure 6.2: Hybrid-electric propulsion system implemented in the QT1 aircraft
71
72
Bibliography
[1] J. Hung and L. Gonzalez. On parallel-electric propulsion system for unmanned aerial vehicles.
Progress in Aerospace Sciences, (51):1–17, 2012. https://doi.org/10.1016/j.paerosci.2011.12.001.
[2] I. Mengistu. Small internal combustion engine testing for a hybrid-electric remotely-piloted aircraft.
Master’s thesis, Air Force Institute of Technology, Wright-Patterson Air Force Base, Ohio, March
2011.
[3] M. Moran, H. Shapiro, D. Boettner, and M. Bailey. Fundamentals of Engineering Thermodynamics.
John Wiley and Sons, Inc., United States of America, seventh edition, 2011.
[4] L. A. Costello. State of the art of piloted electric airplane, NASA’s centennial challenge and fun-
damental design implications. Master’s thesis, Embry-Riddle Aeronautical University - Daytona
Beach, 2011.
[5] J. Matlock, S. Warwick, P. Sharikov, J. Richards, and A. Suleman. Evaluation of energy efficient
propulsion technologies for unmanned aerial vehicles. In Transactions of The Canadian Society for
Mechanical Engineering International Congress 2018, Victoria, Canada, October 2018.
[6] IATA. Annual growth in global air traffic passenger demand from 2006 to 2019. IATA Industry
Forecast, 2018. Accessed on the 14th January 2019.
[7] T. A. Rotramel. Optimization of hybrid-electric populsion systems for small remotely-piloted aircraft.
Master’s thesis, Air Force Institute of Technology, March 2011.
[8] C. Lin and H. P. snd J. Grizzle. Power management strategy for parallel hybrid electric truck. IEEE
TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, 11(03):839–849, November 2003.
[9] B. S. C.R. Akli, X. Roboam and A. Jeunesse. Energy management and sizing of a hybrid loco-
motive. In 12th International Conference on Power Electronics and Applications, editors, American
Control Conference, Aalborg, Danemark, September 2007.
[10] B. Manouchehrinia. Modeling, Optimization and Environmental Assessment of Electrified Marine
Vessels. PhD thesis, University of Victoria, 2018.
[11] Rolls-Royce. The e-fan x programme. https://www.rolls-royce.com/media/our-stories/
insights/2018/paul-stein-talks-about-e-fan-x.aspx, 2019. Accessed on the 8th May 2019.
73
[12] A. Lentsch. Diamond aircraft 1st flight multiengine hybrid electric aircraft.
http://www.diamond-air.at/en/media-center/press-releases/news/article/
diamond-aircraft-1st-flight-multi-engine-hybrid-electric-aircraft/, November 2018.
Accessed on the 14th January 2019.
[13] R. Hiserote. Analysis of hybrid-electric propulsion system designs for small unmanned aircraft
systems. Master’s thesis, Air Force Institute of Technology, March 2010.
[14] H. D. Kim. Distributed propulsion vehicles. International Journal of Computational Fluid Dynamics,
30(9–10):307–327, Oct. 2007. doi:10.1080/10618560701678647.
[15] Launchpoint. Hybrid uav systems. http://www.launchpnt.com/portfolio/aerospace/
propulsion-by-wire-technology, 2017. Accessed on the 4th May 2019.
[16] S. L. T. K. C. Kim, E. NamGoong and H. Kim. Fuel economy optimization for parallel hybrid vehicles
with cvt. Technical report, SAE Technical Report, 1999.
[17] M. Harmats and D. Weihs. Hybrid-propulsion high altitude long-endurance remotely piloted vehicle.
Journal of Aircraft, 36(2):321–331, 1999.
[18] J. Miller. Propulsion Systems for Hybrid Vehicles. IET, London, 2004.
[19] J. K. Ausserer. Integration, testing, and validation of a small hybrid-electric remotely-piloted aircraft.
Master’s thesis, Air Force Institute of Technology, March 2012.
[20] F. Harmon, A. Frank, and S. Joshi. Application of a cmac neural network to the control of a par-
allel hybrid-electric propulsion system for a small unmanned aerial vehicle. In International Joint
Conference on Neural Networks, Montreal, Canada, July 2005.
[21] C. J. B. et all. An operating point optimizer for the design and calibration of an integrated diesel/-
continuously variable transmission powertrain. In Proceedings of the Institution of Mechanical En-
gineers, Part D: Journal of Automobile Engineering, volume 213, pages 215–226, March 1999.
https://doi.org/10.1243/0954407991526810.
[22] F. Harmon, A. Frank, and J. Chattot. Conceptual design and simulation of a small hybrid electric
unmanned aerial vehicle. Journal of Aircraft, 43(5):1490–1498, September-October 2006.
[23] C. M. Greiser. Implementation of a rule-based open-loop control strategy for a hybrid-electric
propulsion system on a small rpa. Master’s thesis, Air Force Institute of Technology, March 2011.
[24] J. Koster, C. Humbargar, E. Serani, A. Velazo, and D. Hillery. Hybrid electric intergated optimized
system (HELIOS), design of a hybrid propulsion system for aircraft. In 49th AIAA Aerospace Sci-
ences Meeting including the New Horizons Forum and Aerospace Exposition, volume 1011, Or-
lando, Florida, January 2011.
74
[25] R. Glassock. Design, modelling and measurement of hybrid powerplant for unmanned aerial vehi-
cles (uavs). Master’s thesis, School of Electrical Engineering and Computer Science - Queensland
University of Technology, November 2012.
[26] M. Ehsani, Y. Gao, and A. Emadi. Modern Electric, Hybrid Electric, and Fuel Cell Vehicles. CRC
Press, 2010.
[27] B. Cho. Control of a Hybrid Electric Vehcile with Predictive Journey Estimation. PhD thesis, Cran-
field University School of Engineering, 2008.
[28] N. J. S. M. Salman and N. A. Kheir. Control strategies for parallel hybrid vehicles. In American
Control Conference, Chicago, June 2000.
[29] J. L. Karunarathne and K. Knowles. Fuzzy logic control strategy for fuel cell/battery aerospace
propulsion system. In IEEE Vehicle Power and Propulsion Conference (VPCC), Harbin, China,
September 2008.
[30] D. Aircraft. Desert aircraft da35. https://www.desertaircraft.com/products/da-35. Accessed
on the 11th March 2019.
[31] DA35 Owner’s manual. Desert Aircraft, Tuscon, Arizona, USA.
[32] D. Khiar, J. Lauber, T. Floquet, and T. M. Guerra. An observer design for the instantaneous torque
estimation of an ic engine. In 2005 IEEE Vehicle Power and Propulsion Conference, pages 5 pp.–,
Aalborg, Danemark, September 2005. doi:10.1109/VPPC.2005.1554587.
[33] C. Wilson. Performance of a small internal combustion engine using n-heptane and iso-octane.
Master’s thesis, Air Force Institute of Technology, March 2010.
[34] Optrand. Optrand introduces a combustion analyzer for its cylinder pressure sensors. http://www.
optrand.com, 2017. Accessed on the 9th May 2019.
[35] C. Cadou, N. Moulton, M. Aluko, T. Sookdeo, and T. Leach. Performance Measurement and Scaling
in Small Internal Combustion Engines. doi: 10.2514/6.2003-671. URL https://arc.aiaa.org/
doi/abs/10.2514/6.2003-671.
[36] A. model motors. Axi 4130/20. https://www.modelmotors.cz/product/detail/276/. Accessed
on the 11th March 2019.
[37] Dejan. How brushless motor and esc work. https://howtomechatronics.com/how-it-works/
how-brushless-motor-and-esc-work/, 2019. Accessed on the 4th May 2019.
[38] J. Schomann. Hybrid-electric propulsion systems for small unmanned aircraft. Master’s thesis,
Technische Universitat Munchen, October 2014.
[39] B. Vedder. Vesc project. https://vesc-project.com/vesc_tool, 2019. Accessed on the 4th May
2019.
75
[40] Enertion. Focbox motor controller. https://www.enertionboards.com/
FOCBOX-foc-motor-speed-controller.html, 2019. Accessed on the 4th May 2019.
[41] T. Ishikawa, T. Tsuji, S. Hashimoto, and N. Kurita. A simple equivalent circuit for efficiency cal-
culation of brushless dc motors. Journal of International Conference on Electrical Machines and
Systems, 3(1):54–60, May 2014. http://dx.doi.org/10.11142/jicems.2014.3.1.54.
[42] Futek. Load cells. http://www.futek.com, 2018. Accessed on the 9th May 2019.
[43] L. Machado and J. Matlock. Electric motor and propeller testing. Technical report, University of
Victoria, January 2019.
[44] A. Basu, S. Tatiya, and S. Bhattacharya. Overview of Electric Vehicles (EVs) and EV Sensors.
Springer Singapore, January 2019. ISBN 978-981-13-3290-6. doi: 10.1007/978-981-13-3290-6 7.
[45] J. P. Conner and A. S. Arena. Advanced dynamometer designed to fully characterize the propulsion
system for a uav. In SAE Technical Paper. SAE International, 11 2002. doi: 10.4271/2002-01-2921.
URL https://doi.org/10.4271/2002-01-2921.
[46] G. Thomas. Overview of storage development doe hydrogen program. US DOE Hydrogen Program
2000 Annual Review, page 2009, May .
76
Appendix A
Hybrid controller code
1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2 %%Implementat ion o f a r u l e based c o n t r o l l e r
3 % Describe the r u l e based c o n t r o l l e r
4 % Author : Leonardo Machado
5 % Modi f ied : Leonardo Machado 12 May 2019
6 % Revis ion 1.0
7 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
8
9
10 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
11 % MODE = 0 : Reset
12 % MODE = 1 : ICE Sing le Operat ion
13 % MODE = 2 : EM Sing le Operat ion
14 % MODE = 3 : Hybr id
15 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
16
17
18
19 %% FUNCTION STARTS HERE
20 f u n c t i o n [ i c e c o n t r o l , em contro l , ERROR CODE] = h y b r i d c o n t r o l l e r ( torque ,
ice rpm , t h r o t t l e i d d l e , mode in )
21
22 %% Load data
23
24 % I n t e r n a l Combustion Engine
25 load ( ’ DA35 data ’ ) ;
26 load ( ’ I O L l i n e ’ ) ;
77
27 load ( ’ i c e t h r o t t l e ’ )
28 m a x t h r o t t l e =65;
29 % E l e c t r i c Motor
30 load ( ’ AXi4130 data ’ ) ;
31
32
33 %% Pre processing
34 %Max torque provided by the EM
35 em. maxtorque = em. Torque Thro t t l e (100) ;
36
37
38 swi tch mode in
39 % RESET
40 case 0
41 em cont ro l = 0 ;
42 i c e c o n t r o l = t h r o t t l e i d d l e ;
43 ERROR CODE = 0;
44
45
46 % ICE Sing le Operat ion mode
47 case 1
48
49 %Max torque provided by the ICE at t h a t speed
50 i ce max torque = ice cu rve . t o r q u e t h r o t t l e ( m a x t h r o t t l e ) ; % evaluate
the maximum torque a v a i l a b l e a t the speed
51 em cont ro l =0; % i t i s
impor tan t to note t h a t the maximum torque always occurs a t f u l l
open t h r o t t l e va lve
52
53 i f isnan ( ice max torque )
54 i c e c o n t r o l = t h r o t t l e i d d l e ;
55 ERROR CODE=6;
56 disp ( ’ERROR − No data f o r t h a t RPM ( e i t h e r i s too low or too
high ) . T h r o t t l e was set to i d d l e ’ ) ;
57 r e t u r n
58 end
59
60 i f torque <i ce max torque
61 i c e c o n t r o l = i ce cu rve . t h r o t t l e t o r q u e ( torque ) ;
78
62 ERROR CODE = 0;
63 e l s e i f torque >= ice max torque
64 i c e c o n t r o l = m a x t h r o t t l e ;
65 ERROR CODE = 2;
66 disp ( ’ERROR − ICE i s not able to prov ide t h a t power alone !
Ac t i va te HYBRID MODE or lower torque request ’ )
67 else
68 i c e c o n t r o l = t h r o t t l e i d d l e ;
69 ERROR CODE=5;
70 disp ( ’ERROR − Something went wrong . . . Please check your code ’ ) ;
71 r e t u r n
72 end
73
74
75 %Saturate the i c e c o n t r o l s i g n a l
76 i f i c e c o n t r o l > m a x t h r o t t l e
77 i c e c o n t r o l = m a x t h r o t t l e ;
78 ERROR CODE = 2;
79 e l s e i f i c e c o n t r o l < t h r o t t l e i d d l e | | isnan ( i c e c o n t r o l )
80 i c e c o n t r o l = t h r o t t l e i d d l e ;
81 ERROR CODE = 1;
82 disp ( ’ERROR − That torque request would lead the ICE to STOP.
T h r o t t l e was set to i d d l e ’ ) ;
83 end
84
85
86
87
88
89
90 %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
91 %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
92 %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
79
93
94 % EM Sing le Operat ion mode
95
96 case 2
97 i c e c o n t r o l = t h r o t t l e i d d l e ;
98
99 i f torque < em. maxtorque
100 em cont ro l = em. Thro t t l e Torque ( torque ) ;
101 ERROR CODE = 0;
102 e l s e i f torque >= em. maxtorque
103 em cont ro l =100;
104 ERROR CODE = 3; %ERROR − EM i s not able to prov ide t h a t power
alone
105 disp ( ’ERROR − EM i s not able to prov ide t h a t power alone ! !
Ac t i va te HYBRID MODE or lower torque request ’ )
106 else
107 em cont ro l =0;
108 ERROR CODE=5;
109 disp ( ’ERROR − Something went wrong . . . Please check your code ’ ) ;
110 r e t u r n
111 end
112
113
114 %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
115 %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
116 %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
117
118 % Hybrid mode
119
120 case 3
121
122
123 %Idea l ICE Torque at t h i s RPM
80
124 i c e i d e a l t o r q u e = IOL . i n t e r p o l a t i o n s . IOL ( ice rpm ) ;
125
126 %Maximum ICE Torque at t h i s RPM
127 i ce max torque = ice cu rve . t o r q u e t h r o t t l e ( m a x t h r o t t l e ) ;
128
129 i f torque < i c e i d e a l t o r q u e
130 i c e c o n t r o l = i ce cu rve . t h r o t t l e t o r q u e ( torque ) ; % ICE ONLY MODE
131 em cont ro l = 0 ;
132 ERROR CODE = 0;
133 e l s e i f torque >= i c e i d e a l t o r q u e % i f the requested torque i s above
the engine IOL Torque , a lso use the e l e c t r i c motor torque
134 em torque=torque− i c e i d e a l t o r q u e ; % The torque needed from the
EM
135 i f em torque<em. maxtorque
136 em cont ro l = em. Thro t t l e Torque ( em torque ) ;
137 i c e c o n t r o l = i ce cu rve . t h r o t t l e t o r q u e ( i c e i d e a l t o r q u e ) ; %
ICE i s set a t i t s most i d e a l po i n t a t t h i s RPM
138 ERROR CODE = 0;
139 e l s e i f em torque >= em. maxtorque
140 i f torque < em. maxtorque+ ice max torque
141 em cont ro l = 100; % We w i l l se t the EM to
prov ide the maximum torque i t can and then the ICE w i l l
leave the IOL and prov ide the r e s t o f the torque
142 i ce rema in ing to rque =torque−em. maxtorque ;
143 i c e c o n t r o l = i ce cu rve . t h r o t t l e t o r q u e ( i ce rema in ing to rque ) ;
144 ERROR CODE = 0;
145 e l s e i f torque >= em. maxtorque+ ice max torque
146 em cont ro l = 100;
147 i c e c o n t r o l = m a x t h r o t t l e ;
148 ERROR CODE=4;
149 disp ( ’ERROR − HYBRID MODE i s not able to prov ide t h a t torque !
Please lower torque request ’ ) ;
150 end
151 end
152 e l s e i f isnan ( ice max torque )
153 i c e c o n t r o l = t h r o t t l e i d d l e ;
154 em cont ro l =0;
155 ERROR CODE=6;
156 disp ( ’ERROR − No data f o r t h a t RPM ( e i t h e r i s too low or too high ) .
81
T h r o t t l e was set to i d d l e ’ ) ;
157 r e t u r n
158 else
159 i c e c o n t r o l = t h r o t t l e i d d l e ;
160 em cont ro l =0;
161 ERROR CODE=5;
162 disp ( ’ERROR − Something went wrong . . . Please check your code ’ ) ;
163 r e t u r n
164 end
165
166
167
168 %Saturate the em cont ro l s i g n a l
169 i f em cont ro l > 100
170 em cont ro l = 100;
171 e l s e i f em cont ro l < 0
172 em cont ro l = 0 ;
173 end
174
175
176 %Saturate the i c e c o n t r o l s i g n a l
177 i f i c e c o n t r o l > m a x t h r o t t l e
178 i c e c o n t r o l = m a x t h r o t t l e ;
179 e l s e i f i c e c o n t r o l < t h r o t t l e i d d l e | | isnan ( i c e c o n t r o l )
180 i c e c o n t r o l = t h r o t t l e i d d l e ;
181 ERROR CODE = 1;
182 disp ( ’ERROR − That torque request would lead the ICE to STOP.
T h r o t t l e was set to i d d l e ’ ) ;
183 end
184
185
186
187
188
189 %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
190 %%END OF SWITCH
191 end
82
192 %% END OF FUNCTION
193 r e t u r n
83
84
Appendix B
Technical Datasheets
B.1 Desert Aircraft DA35 manufacturer test results
85
Performance Map
J1349 Power[W]
Throttle\RPM 1000 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 8000
0 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A
4.7 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A
9.8 #N/A -8 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A
14.9 #N/A #N/A 31 -20 47 68 81 87 74 69 71 0 #N/A
20 #N/A #N/A 100 173 158 192 245 292 333 344 341 #N/A #N/A
24.9 #N/A #N/A 261 320 375 363 424 504 579 632 638 #N/A #N/A
29.8 #N/A #N/A 394 486 552 602 652 751 815 792 856 859 904
34.9 #N/A #N/A 449 561 656 761 889 999 1116 1104 1142 1236 1193
40 #N/A #N/A 473 606 717 840 1014 1136 1271 1307 1393 1475 1526
49.8 #N/A #N/A 516 664 801 947 1179 1313 1476 1552 1671 1777 1934
59.6 #N/A #N/A 535 690 834 1013 1258 1424 1564 1652 1798 1938 2112
80 #N/A #N/A 546 696 858 1041 1332 1512 1658 1765 1952 2075 2302
100 #N/A #N/A 547 699 861 1048 1336 1521 1682 1791 1947 2099 2344
BSFC[g/kW-hr]
Throttle\RPM 1000 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 8000
0 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A
4.7 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A
9.8 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A
14.9 #N/A #N/A #N/A #N/A #N/A 2351 2208 1907 2677 2664 2933 #N/A #N/A
20 #N/A #N/A 1606 1138 1287 1109 1022 830 835 800 767 #N/A #N/A
24.9 #N/A #N/A 843 703 696 648 735 633 595 586 592 #N/A #N/A
29.8 #N/A #N/A 667 564 536 523 523 504 486 521 501 540 539
34.9 #N/A #N/A 669 552 530 498 453 452 457 474 453 472 563
40 #N/A #N/A 717 626 544 504 452 453 459 460 465 468 476
49.8 #N/A #N/A 762 639 607 540 485 465 503 525 494 483 481
59.6 #N/A #N/A 749 662 656 676 524 509 522 534 514 515 517
80 #N/A #N/A 834 705 655 704 598 578 540 571 565 563 576
100 #N/A #N/A 815 805 702 649 625 607 597 635 620 619 634
Raw Power[W]
Throttle\RPM 1000 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 8000
0 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A
4.7 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A
9.8 #N/A -8 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A
14.9 #N/A #N/A 32 -20 48 69 83 90 76 71 73 0 #N/A
20 #N/A #N/A 102 178 162 196 251 300 341 352 350 #N/A #N/A
24.9 #N/A #N/A 268 328 384 372 435 517 593 648 654 #N/A #N/A
29.8 #N/A #N/A 404 499 566 617 669 770 836 812 878 881 928
34.9 #N/A #N/A 460 575 673 780 912 1025 1145 1132 1172 1267 1224
40 #N/A #N/A 485 622 736 862 1040 1166 1304 1341 1429 1513 1565
49.8 #N/A #N/A 529 681 821 972 1209 1347 1514 1592 1714 1823 1984
59.6 #N/A #N/A 549 708 855 1039 1291 1461 1604 1695 1844 1988 2166
86
80 #N/A #N/A 560 714 880 1068 1367 1551 1701 1810 2003 2128 2361
100 #N/A #N/A 561 717 883 1075 1371 1560 1726 1837 1997 2153 2405
Fuel Flow[g/min]
Throttle\RPM 1000 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 8000
0 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A
4.7 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A
9.8 #N/A 1.4 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A
14.9 #N/A #N/A 2.2 2.4 2.6 2.7 3.1 2.9 3.4 3.2 3.6 4.6 #N/A
20 #N/A #N/A 2.7 3.4 3.5 3.6 4.3 4.1 4.7 4.7 4.5 #N/A #N/A
24.9 #N/A #N/A 3.8 3.8 4.5 4 5.3 5.5 5.9 6.3 6.5 #N/A #N/A
29.8 #N/A #N/A 4.5 4.7 5.1 5.4 5.8 6.5 6.8 7.1 7.3 7.9 8.3
34.9 #N/A #N/A 5.1 5.3 5.9 6.5 6.9 7.7 8.7 8.9 8.8 10 11.5
40 #N/A #N/A 5.8 6.5 6.7 7.2 7.8 8.8 10 10.3 11.1 11.8 12.4
49.8 #N/A #N/A 6.7 7.2 8.3 8.8 9.8 10.4 12.7 13.9 14.1 14.7 15.9
59.6 #N/A #N/A 6.9 7.8 9.3 11.7 11.3 12.4 14 15.1 15.8 17.1 18.7
80 #N/A #N/A 7.8 8.4 9.6 12.5 13.6 14.9 15.3 17.2 18.8 20 22.7
100 #N/A #N/A 7.6 9.6 10.3 11.6 14.3 15.8 17.2 19.4 20.7 22.2 25.4
87
B.2 AXi 4130/20 GOLD LINE V2 manufacturer test results
88
0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 2.750
250
500
750
1000
1250
1500
1750
2000
2250
2500
Pin
[W],
Pout
[W]
0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 2.750
10
20
30
40
50
60
70
80
90
100
Torque [Nm]
Eta%
, Cur
rent
[A],
Volta
ge [V
], R
PM/1
00
AXI 4130−20 GOLD − 20V (Test #7) Date: 11/29/2008
Controller DUALSKY XC9036HV Startup Normal, Timing Medium (10°)
IPR Silvagni, 2009
Legend:
Efficiency
Power In
Power Out
Rpm
Voltage
Current
Contact:
SILVAGNI MarioVia Repubblica, 1813897−Occhieppo I. (BI) − ITALYE−mail:[email protected]
89
0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 2.750
250
500
750
1000
1250
1500
1750
2000
2250
2500
Pin
[W],
Pout
[W]
0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 2.750
10
20
30
40
50
60
70
80
90
100
Torque [Nm]
Eta%
, Cur
rent
[A],
Volta
ge [V
], R
PM/1
00
AXI 4130−20 GOLD − 25V (Test #6) Date: 11/29/2008
Controller DUALSKY XC9036HV Startup Normal, Timing Medium (10°)
IPR Silvagni, 2009
Legend:
Efficiency
Power In
Power Out
Rpm
Voltage
Current
Contact:
SILVAGNI MarioVia Repubblica, 1813897−Occhieppo I. (BI) − ITALYE−mail:[email protected]
90
Appendix C
Standard Operating Procedures
91
STANDARD OPERATING PROCEDURES (SOP) FOR ICE TESTING
1. Check Headsets are working
2. Plug in DAQ Cards (modules) to DAQ Chassis
3. Connect Power to Test Power Station
4. Connect Power to DAQ Chassis
5. Connect Ethernet cable to DAQ Chassis
6. Connect ICE RPM sensor
7. Connect Fuel Mass sensor
8. Connect ICE Throttle
9. Bring Battery and Starter Motor
10. Tape everything down
11. Bring Fire Extinguisher
12. Prime Fuel Line
13. Turn on Power
14. Check DAQ station is connected to NI MAX
15. Assure Throttle is set to 0%
16. Establish Safety Perimeter
17. Connect Hall Sensor Battery
18. Run LabVIEW
19. Check Fuel Mass measurement
20. Check RPM Measurement
21. Check Servo Throttle working
22. Tare Fuel Cell
23. Move to test positions
24. Plug in Spark Plug
25. Set Throttle to 15%
26. Start Logging Data
27. Run the Starter motor
92
STANDARD OPERATING PROCEDURES (SOP) FOR EM TESTING
1. Check everything is tighten (Propeller, Electric motor and supports)
2. Safety glasses ON
3. Check Headsets are working
4. Plug in DAQ Cards (modules) to DAQ Chassis
5. Connect Power to Test Power Station
6. Connect Power to DAQ Chassis
7. Connect Ethernet cable to DAQ Chassis
8. Connect EM RPM sensor
9. Connect Battery Current sensor
10. Connect Battery Voltage sensor
11. Connect EM Throttle
12. Tape everything down
13. Bring Fire Extinguisher
14. Assure STOP button is pushed
15. Turn on Power
16. Check DAQ station is connected to NI MAX
17. Establish Safety Perimeter
18. Assure Throttle is set to 0%
19. Pull STOP button
20. Move to test positions
21. Run LabVIEW
22. Tare Current
23. Tare Load Cell
24. Start Logging Data
25. Run the EM motor
93
Appendix D
LabVIEW block diagram
94
AI Force Bridge (Two-Point Linear)
cDAQ9188-AMod2/ai0
kgf
5
second physical value
0
first physical value
mV/V
electrical units
1
second electrical value
0
first electrical value
kgf
physical units
Full Bridge
bridge configuration
External
voltage excitation source
5
voltage excitation value
1000
nominal bridge resistance
-1
100
Sample Clock
CO Pulse Freq Implicit
reserve
Continuous Samples
Low
CO Pulse Freq Implicit
Start
Digital Edge
/cDAQ9188-A/Ctr0InternalOutput
Rising
10000
0
Signals In
Signals Out
Signal Index
Set Dynamic
Data Attributes
WAIT
FUEL MASS (g)
EM THROTTLE
ICE THROTTLE
ICE RPM
EM RPM
Current (A)
Voltage (V)
Power (W)
CONTROLLER
MODE IN
TORQUE REQUEST (N.m)
MISSION
Stop
EMERGENCY
250
WAIT
Saving Data
Stop
EMERGENCY
We placed a Voltage divider. It steps down the
voltage by a factor of 10.
Thus, to have the correct value, you should
multiply the value read by LabView by 10
PWM EM Frequency (Hz)
PWM ICE Frequency (Hz)
Reset
Cruise
Endurance
Hybrid
EM THROTTLE0
ICE THROTTLE0
CONTROLLER
When the code tuns for the first
time, initialize these variables with
these values
MODE IN0
cDAQ9188-AMod1/ctr0
cDAQ9188-AMod1/ctr1
The maximum RPM we are going to
measure is 8000RPM, which corresponds
to 133.33 Hz. This way, the low cutoff
frequency of the filter was set to 140Hz to
eliminate any noise
Producer/Consumer loop
i_old
0
i_old 2
0
EM RPM
ICE RPM
0
0
95
S Load Cell
Analog Wfm
1Chan NSamp
statusStop
FUEL MASS (g)
1000 FUEL LOAD CELL TARE
EMERGENCY
Y
-6653448.48235
447.710548
LOG AVERAGE RATE2
1
LOG AVERAGE RATE3
ICE Throttle Servo
Counter Freq
1Chan 1Samp
status Stop
ICE THROTTLE
PWM ICE Frequency (Hz)
Lower Limit - No ICE Throttle (s)
Upper Limit - Full ICE Throttle (s)
100
False
EMERGENCY
EMERGENCY
Pulse length ICE (ms)
PWM ICE Frequency (Hz)
1000
EM Throttle Servo
Counter Freq
1Chan 1Samp
status Stop
Lower Limit - Full EM Brake (s)
Upper Limit - Full EM Throttle (s)
2
PWM EM Frequency (Hz)
True
EMERGENCY
EMERGENCY
Pulse length EM (ms)
PWM EM Frequency (Hz)
1000
Current (A)
chains or links together
array values (the battery
current) into an n-
dimensional array.
6determines
the array size
This while loop checks the value
of the battery current
This is the battery
current value
returning from the last
loop.
0
0
0
0
0
0
0
We placed a Voltage divider. It steps down the
Thus, to have the correct value, you
should multiply the value read by LabView by
Current, Voltage and RPM
data
DAQ Assistant
0
CURRENT TARE
1Voltage (V)
Current (A)
LOG AVERAGE RATE
Power (W)
status
Stop
EMERGENCY
1
2
EM RPM Graph
ICE RPM Graph
3
20000
5000
Y
dt
Y
dt
count
0
1
i_old60
i_old
i_old
i_old
True
EM RPM
count1
True
EM RPM0
True
count
0
count2
0
1
i_old 2
60
i_old 2
i_old 2i_old 2
True
count21
True
ICE RPM
ICE RPM0
True
0
count2
dt
00
The maximum RPM we are going to
measure is 8000RPM, which corresponds
to 133.33 Hz. This way, the low cutoff
frequency of the filter was set to 140Hz to
96
OK message + warnings
Continuous Current Safety cutoff
0
1
2
3
4
5
Max Continuous Current (A)
EM THROTTLE0
ICE THROTTLE
When battery current is greater
than continuous current 6x
continuously, cut the throttle.
Or else, do nothing.
4000
True
0
Deletes the oldest
(n=0) element of the
array.
True
500
True
error out
Elapsed Time (s)
1
1000
Stop
Measure run time
0
97
Stop
EMERGENCY
Signals
Write To
Measurement File
LOGGING DATA
File Path Control
ICE ONLY MODE
REGEN MODE EM ONLY MODE
HYBRID MODE
RESET MODE
True
EM THROTTLE
ICE THROTTLE
0
ICE ONLY MODE
HYBRID MODE
EM ONLY MODE
REGEN MODE
RESET MODE
True
ICE THROTTLE
EM THROTTLE
0
ICE ONLY MODE
EM ONLY MODE
HYBRID MODE
REGEN MODE
RESET MODE
True
EM THROTTLE
0
ICE ONLY MODE
EM ONLY MODE
HYBRID MODE
REGEN MODE
RESET MODE
True
Stop
EMERGENCY
Throttle Iddle
Throttle Iddle
CONTROLLER
Intelligent Power Management
Manual Power Management
TORQUE REQUEST (N.m)
path(path,'R:\0076-Parallel Hybrid Research\DESIGN\SOFTWARE\VI v1.0\TESTING\Tests\Hybrid Test\matlab_codes\Hybrid controller\')
[ice_control, em_control, error_code]=hybrid_controller(torque, ice_rpm, throttle_iddle,mode_in);
error_code
mode_in
throttle_iddle
ice_rpm
ice_controltorque
em_control
MATLAB script
ICE RPM
Throttle Iddle
MODE IN
ERROR CODE
0
ERROR!!
status
CONTROLLER
There was a problem running your controller. The control
was set to Manual. The ICE was set to Iddle and the EM turned
off. Please re-start LabVIEW.
Manual Power Management
Intelligent Power Management
ICE THROTTLEThrottle Iddle
EM THROTTLE0
True
EM THROTTLE
ICE THROTTLE
True
Stop
EMERGENCY
1000
To be changed in a later iteration (i used this just to confirm the controller is doing what it is supposed to)
NewVal
OldVal
CtlRef
Time
Type
Source
Endurance
Hybrid
Reset
MODE IN1
True
Cruise
Hybrid
Reset
MODE IN2
True
Cruise
Endurance
Reset
MODE IN3
True
Endurance
Hybrid
Cruise
Reset
Cruise
Endurance
Hybrid
0
MODE IN
True
[0] "Cruise", "Endurance", "Hybrid", "Reset": Value Change
If no Mode is selected, The HEPS will automatically
set itself to RESET mode
False
False False False False
98
Voltage (V)
Max Continuous Voltage (V)
0
1
2
3
4
5
When the array gets to a
size of n > 6 the structure executes
6
0
0
0
0
0
0
0
Voltage (V)
0
1
2
3
4
5
Min Continuous Voltage (V)
When the array gets to a
size of n > 6 the structure executes
6
0
0
0
0
0
0
0
Enable safety features
To be changed in a later iteration (i used this just to confirm the controller is doing what it is supposed to)
TORQUE REQUEST (N.m)3
150000
MISSION OVER
TORQUE REQUEST (N.m)2.8
150000
TORQUE REQUEST (N.m)2.4
150000
TORQUE REQUEST (N.m)0.5
5000
MISSION OVER
MISSION
True
MISSION
Stop
EMERGENCY
SAMPLE MISSION
False False False False False
99
ICE THROTTLE
Stop
EMERGENCY
Maximum Voltage Safety cutoff
Stop
EMERGENCY
EM THROTTLE0
ICE THROTTLE
4000
True
0
True
500
Min Voltage Safety cutoff
Stop
EMERGENCY
EM THROTTLE0
ICE THROTTLE
4000
True
0
True
500
Stop
EMERGENCY
Recording data Time (s)
1
1000
Stop
EMERGENCY
LOGGING DATA
0
True
LOGGING DATA
Stop
EMERGENCY
Recording data Time (s)0
Elapsed Time (s)0
EM RPM
False False
ICE RPM
False False False False False
100