Universitat Politécnica de Valencia

School of Aerospace Engineering and Industrial Design

Multidisciplinary Aircraft and

Trajectory Optimization for Novel
Regional Aircraft Architectures

Pablo Norczyk Simon

Thesis Advisors
Jorge García Tíscar
Rauno Cavallaro

Submitted in partial fulfillment of the master’s

degree in Aeronautical Engineering
Whenever you are asked if you can do a job,
tell ‘em ‘Certainly I can!’ Then get busy and
find out how to do it

Theodore Roosevelt
 Dedication
Dedicated to my friends, those old ones from Murcia who, though we only see each other months
apart, I can always rely on; and to those other new close friends I was able to meet and share good times
with, especially my friends and battle-buddies Nestor and Pablo, who have accompanied me through
these last months of the master’s program. Dedicated also to my former colleagues from UC3M, who,
while I have always been a ghost, have treated me as one more member of the lab. Also dedicated to my
tutors: first Rauno, who trusted in me to work in his lab, even from far away, and then Jorge, who, while
I have not been able to materialize some ideas, has responded to my Teams messages (even late at night).
Then, while friends are important, this could not have been possible without my family. Dedicated
to my parents, who give me unconditional love and always try to support me no matter what; to my
mother, from whom I inherited a great passion and love for the things in life; dedicated to my father,
who showed me how to be strong and firm throughout life, and how to transmit calm and security even
in tough times. To my sister, the future next Dr. in the family, who has shown me that things work out
in the end if one is stubborn enough.
And finally, dedicated to my partner, my companion, my other little half, my "Laury," my "Laurutka,"
my "Laurusha," who is a gift in life and, even now in the distance, still shows me unwavering love, care,
and trust.
Abstract

Novel aircraft configurations, such as hybrid-electric and hydrogen-powered

designs, offer strong potential to reduce dependence on conventional fuels and
contribute to the decarbonization of aviation. However, their efficient design
remains challenging, as conventional methods often fail to capture the complex
trade-offs between performance metrics, propulsion architectures, and mission
requirements. This research addresses this gap through OPALO (Overall
Powerplant and Aircraft synthesis with Layout-consistent Optimization), a
GEMSEO-based multidisciplinary analysis and optimization framework for
the concurrent synthesis of aircraft configurations and mission trajectories.
OPALO combines conventional handbook methods with dedicated physics-
based modules and surrogate models, capturing all relevant aircraft disciplines
and their interactions, to resolve trade-offs and achieve a truly holistic aircraft
design while maintaining computational efficiency.
The framework is demonstrated on four regional aircraft propulsion archi-
tectures: conventional kerosene, kerosene parallel hybrid, hydrogen parallel
hybrid, and hydrogen fuel-cell, evaluated under both fuel-minimization and
energy-minimization objectives. Results show that battery integration can
reduce emissions in kerosene-based architectures, while in hydrogen systems,
batteries primarily enable downsizing of fuel-cell installations. The study also
identifies technology performance thresholds that act as enablers for each
architecture and reveals the pivotal coupling between sustainable in-flight
decarbonization and clean, ground-based energy production, underscoring
the need for coordinated advancements in both aviation technology and
renewable energy infrastructure.

Keywords: Hybrid-electric aircraft, hydrogen propulsion, multidisciplinary

design optimization, MDAO, GEMSEO, aircraft design, sustainable aviation

v
vi

Resumen

Las aeronaves de nueva generación, como los diseños híbrido-eléctricos y

propulsados por hidrógeno, ofrecen un gran potencial para reducir la depen-
dencia de los combustibles convencionales y contribuir a la descarbonización
de los cielos. Sin embargo, su diseño eficiente sigue siendo un desafío, ya
que los métodos convencionales, en muchos casos, no logran capturar las dis-
tintas decisiones de compromiso entre disciplinas, rendimiento y trayectoria.
Este estudio aborda esta problemática mediante el uso de OPALO (Overall
Powerplant and Aircraft synthesis with Layout-consistent Optimization), una
herramienta de análisis y optimización multidisciplinar basada en GEMSEO
para la síntesis simultánea de aeronaves y sus trayectorias. OPALO combina
métodos tradicionales basados en la literatura existente con módulos físicos
dedicados y modelos subrogados, abarcando todas las disciplinas que confor-
man la aeronave y sus interacciones para resolver los compromisos y alcanzar
un diseño verdaderamente holístico.
La herramienta se aplica a cuatro arquitecturas de propulsión para aeronaves
regionales: queroseno convencional, híbrido paralelo con queroseno, híbrido
paralelo con hidrógeno y pila de combustible de hidrógeno; cada una
optimizada tanto para la minimización de combustible como de energía.
Los resultados muestran que la integración de baterías puede reducir las
emisiones en arquitecturas basadas en queroseno, mientras que en sistemas
de hidrógeno las baterías permiten principalmente reducir el tamaño de
las pilas de combustible necesarias. El estudio también identifica umbrales
de rendimiento tecnológico que actúan como habilitadores clave de cada
arquitectura y revela el acoplamiento entre la descarbonización sostenible en
vuelo y la producción limpia de energía en tierra, subrayando la necesidad de
avances coordinados en tecnología aeronáutica e infraestructura energética
renovable.

Palabras clave: Aeronaves híbrido-eléctricas, propulsión por hidrógeno,

optimización de diseño multidisciplinar, MDAO, GEMSEO, diseño de
aeronaves, aviación sostenible
 vii

Resum

Les aeronaus de nova generació, com els dissenys híbrid-elèctrics i propulsats

per hidrogen, oferixen un gran potencial per a reduir la dependència dels
combustibles convencionals i contribuir a la descarbonització dels cels. No
obstant això, el seu disseny eficient continua sent un desafiament, ja que els
mètodes convencionals, en molts casos, no aconseguixen capturar les diferents
decisions de compromís entre disciplines, rendiment i trajectòria. Este es-
tudi aborda esta problemàtica mitjançant l’ús de OPALO (Overall Powerplant
and Aircraft synthesis with Layout-consistent Optimization), una ferramenta
d’anàlisi i optimització multidisciplinària basada en GEMSEO per a la sín-
tesi simultània d’aeronaus i les seues trajectòries. OPALO combina mètodes
tradicionals basats en la literatura existent amb mòduls físics dedicats i mod-
els subrogats, abastant totes les disciplines que conformen l’aeronau i les
seues interaccions per a resoldre els compromisos i aconseguir un disseny
verdaderament holístic.
La ferramenta s’aplica a quatre arquitectures de propulsió per a aeronaus
regionals: querosé convencional, híbrid paral·lel amb querosé, híbrid paral·lel
amb hidrogen i pila de combustible d’hidrogen; cada una optimitzada tant per
a la minimització de combustible com d’energia. Els resultats mostren que
la integració de bateries pot reduir les emissions en arquitectures basades
en querosé, mentres que en sistemes d’hidrogen les bateries permeten
principalment reduir la grandària de les piles de combustible necessàries.
L’estudi també identifica llindars de rendiment tecnològic que actuen
com a habilitadors clau de cada arquitectura i revela l’acoblament entre la
descarbonització sostenible en vol i la producció neta d’energia en terra,
subratllant la necessitat d’avanços coordinats en tecnologia aeronàutica i
infraestructura energètica renovable.

Paraules clau: Aeronaus híbrid-elèctriques, propulsió per hidrogen, optim-

ització de disseny multidisciplinari, MDAO, GEMSEO, disseny d’aeronaus,
aviació sostenible
NOMENCLATURE
Acronyms

AZEA Alliance for Zero Emissions Aviation

EAS Equivalent Airspeed

LCA Life Cycle Assessment

MDA Multi Disciplinary Analysis

MDAO Multidisciplinary Design Analysis and Optimization

MIL Minimal Induced Losses

MLI Multi Layer Insulation

NPSS Numerical Propulsion System Simulation

NTU Number of Transfer Units

PH Parallel Hybrid

PSFC Power Specific Fuel Consumption

SAF Sustainable Aviation Fuels

SMR Steam Methane Reforming

TMS Thermal Management System

Symbols

α Angle of attack rad

δA Throttle of component A −

δe Horizontal tail deflection rad

ṁA Mass-flow of substance A kg/s

Q̇ Heat transfer rate J/s

ϵ Downwash angle rad

ix
x NOMENCLATURE

ϵ0 Downwash angle at zero alpha rad

ϵα Evolution of downwash angle with alpha −

η Component efficiency −

γ Flight path angle rad

κ Gas isentropic coefficient −

λ Lifting surface tapper ratio −

ΛLH2 Lower heating value of hydrogen MJ/kg

µ Ground friction coefficient −

ρ Air density kg/m3

θ Wing twist rad

accel Aircraft acceleration m/s2

AR Lifting surface aspect ratio −

b Wing aspect ratio −

c Lifting surface chord m

CP Coefficient of power −

CT Coefficient of thrust −

CD Drag coefficient −

CD0 First coefficient for coefficient of drag polar −

CDα Second coefficient for coefficient of drag polar −

CDαα Third coefficient for coefficient of drag polar −

CL Lift coefficient −

CL0 First coefficient for coefficient of lift polar −

CLα Second coefficient for coefficient of lift polar −

CM Pitching moment coefficient −

CM0 Evolution of the coefficient of pitch with apha −

CM0 Zero alpha coefficient of pitch −

cpA Specific heat capacity of component A J/kg/K

NOMENCLATURE xi

D Drag force N

dA Diameter of component A m

J Advance ratio −

L Lift force N

lA Length of component A m

lA→B Distance from point A to point B m

M Pitching moment N·m

N Normal ground force N

PA,out Output power of component A W

PAin Input power of component A W

PA Input power of component A W

q Dynamic pressure kg/m/s2

Sref Reference area m2

SM Stating margin m

T Thrust force N

TA Temperature of component A K

T∞ Freestream temperature K

V∞ Free-stream velocity m/s

W Aircraft weight N

WA Mass of component A kg

wA Width of component A m

xcg Position of the center of gravity m

xnp Position of the neutral point m

CONTENTS

1 Introduction and objectives 1

1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Multidisciplinary optimisation: a necessary tool. . . . . . . . . . . . . . . . . . . . 2
1.2.1 Previous studies and limitations . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Thesis Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 State of the Art and Reference Aircraft 9

2.1 A View on Sustainable Aviation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.1 SAFs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.2 The electric and the hybrids . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.3 Hydrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 The environmental cost of energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 The regional turboprop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

3 Modelling methodology 23
3.1 Aerodynamic modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1.1 Coefficient estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1.2 Balance and static stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2 Powerplant modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.1 Propeller model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.2 Gas Turbine Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2.3 Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2.4 Electric motors and inverters . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.2.5 Fuel cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.6 Heat exchangers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.2.7 Hydrogen tanks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3 Weight estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.4 Trajectory definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.5 Aicraft control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4 Aicraft architectures 45
4.1 Conventional Kerosene turboprop . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2 Kerosene Parallel Hybrid architecture . . . . . . . . . . . . . . . . . . . . . . . . . . 47

xiii
xiv CONTENTS

4.3 Hydrogen Parallel Hybrid Architecture . . . . . . . . . . . . . . . . . . . . . . . . . 50

4.4 Hydrogen Fuel Cell architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

5 Aircraft multidisciplinary analysis 57

5.1 A simplified introduction to multidisciplinary analysis . . . . . . . . . . . . . . . . 57
5.1.1 Gradient computation strategies . . . . . . . . . . . . . . . . . . . . . . . . 58
5.1.2 Jacobians of a coupled system . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.2 Aircraft multidisciplinary analysis within OPALO . . . . . . . . . . . . . . . . . . . 61
5.2.1 Pre-processor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.2.2 Steady flight multidisciplinary analysis . . . . . . . . . . . . . . . . . . . . . 62
5.2.3 Takeoff multidisciplinary analysis . . . . . . . . . . . . . . . . . . . . . . . . 63
5.2.4 Overall Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

6 Validation results 67
6.1 Flight Profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.2 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
6.3 Optimization Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

7 Kerosene Turboprop Results 75

7.1 Is Lighter Better? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.1.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
7.1.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
7.2 Sensitivity study on range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
7.2.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
7.2.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
7.3 Sensitivity study on passenger count . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7.3.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7.3.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7.4 Sensitivity study on operative costs . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
7.4.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
7.4.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

8 Kerosene Parallel Hybrid Results 93

8.1 The Naive Kerosene Hybrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
8.1.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
8.1.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
8.2 Thermal managed kerosene hybrid . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.2.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.2.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
CONTENTS xv

9 Hydrogen Parallel Hybrid Results 117

9.1 Thermal managed hydrogen hybrid . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
9.1.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
9.1.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

10 Hydrogen Fuel Cell Results 137

10.1 Thermal managed hydrogen hybrid . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
10.1.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
10.1.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

11 Architecture comparison 159

11.1 Aircraft mass comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
11.2 Aicraft energy comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
11.3 Aircraft emissions comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
11.4 Architecture selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

12 Conclusions 183
12.1 Overall Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
12.2 Future Work and Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184

A Executive summary 209

B Project Specifications 211

B.1 Project Object . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
B.2 Administrative Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
B.3 Software and Mathematical Specifications . . . . . . . . . . . . . . . . . . . . . . . 213
B.4 Execution Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214

C Budget 215
C.1 Personnel cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
C.2 Computational cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
C.3 Software cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
C.4 Total cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217

D Aligment with SDGs 219

E Selected aircraft drawings 221

LIST OF FIGURES

1.1 Comparison of sequential (left) and multidisciplinary (right) design optimization for
a simple two-discipline wing. Extracted from [37] . . . . . . . . . . . . . . . . . . . 2
1.2 Optimizer trajectories for the sequential and the multidisciplinary optimization.
Iso-lines represent mission range (objective function). Extracted and adapted from [37] 3

2.1 Yearly CO2 emissions derived from aviation. The first trend is based on [75], the
second trend is extracted from [76], the third trend is based on [77] . . . . . . . . . 10
2.2 Aviation demand in revenue passenger per kilometer (RPK) and aviation carbon
intensity in grams of CO2 per revenue passenger per kilometer over the years. Data
from [79] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3 Left: Heart HS-30 aircraft concept [89]. Right: Elysisan E9X aircraft concept [98]. . 12
2.4 Serial hybrid, parallel hybrid, and parallel-serial hybrid powerplant architectures. . 13
2.5 Left: Ampere EcoCaravan during takeoff [107]. Right: Airbus-Safran EcoPulse
during flight testing [106]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.6 Comparison of gravimetric density (in MJ/kg) and volumetric density (MJ/L) for
various fuels. Adapted from [112]. . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.7 Left: System description of the modified B75B. Right: modified B75B during ground
taxiing [113]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.8 Left: Tu-155 fuel storage and distribution system [114]. Right: Tu-155 during flight
testing [115]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.9 Left: Airbus zero-e aircraft concept [123]. Right: Universal Hydrogen modified
Dash-8 [124]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.10 Energy production carbon intensity for various nations. Data adapted from [131] . 18
2.11 Energy mix as 2024 for various nations. Adapted from [131]. . . . . . . . . . . . . . 18
2.12 Comparison of emitted CO2 per kg of produced H2 for various energy sources [136]. 19
2.13 Aircraft types operating in the EUROCONTROL area. Avg. daily flights 2024. Data
from [139]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.14 Comparison of carbon intensity between various models of turboprop and turbojet
regional aircraft. Data from [142]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.15 Trends for the regional turboprop and regional turbofan aircraft. The database
contains aircraft from longer regionals to pure commuters. Designs range from 1960
to 2021. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.16 Dash-8 400 [143] and ATR72-600 [144]. . . . . . . . . . . . . . . . . . . . . . . . . . 22

xvii
xviii LIST OF FIGURES

3.1 Aerodynamic forces model for steady flight. Forces are assumed to be applied over
the center of mass. The horizontal line represents the horizon, and the curved line
represents the flight trajectory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2 Aerodynamic forces model for ground acceleration . . . . . . . . . . . . . . . . . . 25
3.3 Reference nomenclature for the pitching moments and the center of masses. . . . . 27
3.4 a) Propeller geometry and propeller J-CP map for a MIL twist and chord optimized
propeller; b) Propeller geometry and propeller J-CP map for the modified MIL twist
optimized propeller. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.5 Modification of the blade chord distribution c/R along the radii and the blade twist
distribution β. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.6 Thermodynamic properties for the operation of the Kerosene turboshaft gas turbine. 31
3.7 Thermodynamic properties for the operation of the LH2 turboshaft gas turbine. . . 32
3.8 Geometric description of the batteries module with TMS cooling. . . . . . . . . . . 33
3.9 Conceptual description of the cooling jacket for the electric motor . . . . . . . . . . 35
3.10 Description of the cold plate cooling system for the electric inverter. . . . . . . . . . 35
3.11 Fuel cell efficiency as a function of throttle δF C . . . . . . . . . . . . . . . . . . . . 36
3.12 Cooling system configuration for fuel cell stack thermal management, and unit cell
for the unit coolant cell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.13 Heat exchanger model and fin unit definition: a single finned cold plate exchanger,
with options for offset fin strips on both plates. . . . . . . . . . . . . . . . . . . . . . 38
3.14 Model for hydrogen boil-off. Rates in black are assumed constant during the ODE
solve; rates in red are the unknowns to solve. . . . . . . . . . . . . . . . . . . . . . . 39
3.15 Definition of mission segments by means of modifying the equivalent airspeed and
flight path angle. Markers at the left plots indicate control nodes handled by the
optimizer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.16 Example of 600nm nominal mission with a 300nm extended mission including a 30
min holding segment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

4.1 Description of the Kerosene conventional architecture. . . . . . . . . . . . . . . . . 45

4.2 Description of the twin turboprop architecture in the trim mode and LTO mode. . . 46
4.3 Description of the Kerosene Parallel Hybrid architecture. Cabin assembly remains
as in the conventional Kerosene. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.4 Description of the Kerosene Parallel Hybrid architecture. Cabin assembly remains
as in the conventional Kerosene. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.5 Schematic of the hydrogen parallel-hybrid architecture. The cabin layout remains
identical to that of the conventional kerosene reference. . . . . . . . . . . . . . . . . 51
4.6 Description of the Hydrogen Parallel Hybrid architecture. Cabin assembly remains
as in the conventional Kerosene. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.7 Description of the Hydrogen Parallel Hybrid architecture. Cabin assembly remains
as in the conventional Kerosene. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
LIST OF FIGURES xix

4.8 Architecture description of the electric system of the fuel cell aircraft (top) and its
corresponding thermal management system (bottom). . . . . . . . . . . . . . . . . . 54

5.1 Implicitly coupled system between disciplines A and B, where discipline B depends
on outputs from discipline A, and discipline A depends on outputs from discipline B. 60
5.2 Modular decomposition of OPALO into distinct computational phases. . . . . . . . . 61
5.3 Pseudo MDA structure for the mission states multidisciplinary analysis . . . . . . . 63
5.4 Pseudo MDA structure for the LTO states multidisciplinary analysis. . . . . . . . . 64
5.5 Pseudo MDA analysis of the overall aircraft sizing within OPALO. . . . . . . . . . . 65

6.1 Climb and descent speed and flight path angle profiles based on the ATR72-600
performance from BADA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
6.2 Altitude profile and Mach number profile for the validation mission of the ATR72-600. 69
6.3 Resultant external geometry of the optimized aircraft with the inner layout and
powerplant. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.4 Comparison between the ATR72-600 from drawings and an ATR72-600-like gener-
ated in TOPAZ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.5 Upper left: comparison of aerodynamic polar at cruise condition. Upper right:
comparison of thrust requirement at climb condition. Lower left: comparison of fuel
flow at climb condition. Lower right: propeller efficiency η at climb condition. . . . 73

7.1 Aspect wing ratio ARw across the resultant M T OW vs nominal block fuel pareto
front for the conventional cantilever wings. . . . . . . . . . . . . . . . . . . . . . . . 78
7.2 Evolution of wing area and turboshaft rating along the M T OW -Nominal block fuel
Pareto front for the conventional cantilever wings. . . . . . . . . . . . . . . . . . . . 78
7.3 Aerodynamic efficiency and altitude profile for the conventional cantilever wings. . 79
7.4 Aspect wing ratio ARw across the resultant Pareto front for the conventional
kerosene turboprop aircraft for both conventional and strut-braced wings. . . . . . 80
7.5 Evolution of wing area and turboshaft rating along the M T OW -Nominal block fuel
Pareto front for the SBW architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . 80
7.6 Aerodynamic efficiency and altitude profile for the SBW Kerosene architecture. . . 80
7.7 Evolution of the fuel mission intensity (in terms of kg of Kerosene per nm) as a
function of the design nominal mission range, colored with the maximum takeoff
weight M T OW . The figure also shows the aicraft geometry for three design conditions. 82
7.8 Altitude and Mach number flight profiles for the various design nominal ranges,
colored with nominal block fuel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
7.9 Fuel intensity (in terms of kg fuel per nm) as a function of range, colored by the
wing area (left) and turboshaft power rating (right). . . . . . . . . . . . . . . . . . . 83
7.10 Fuel mission intensity (kg Kerosene/nm) vs. design nominal mission, colored by
T OW . ▲ marks aircraft optimized for the nominal range; ■ marks off-design
performance of an aircraft sized at 1000,nm. . . . . . . . . . . . . . . . . . . . . . . 84
xx LIST OF FIGURES

7.11 Altitude and Mach number vertical profiles for range-specific designs (left column)
and off-design performance of the 1000 nm sized aircraft (right column), all colored
by fuel intensity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
7.12 Left: Fuel intensity (normalized with range and total passenger count). Right: Aircraft
design MTOW. Both as a function of number of passengers and nominal design range. 86
7.13 Various resultant aircraft shapes from various passenger and range design requirements. 86
7.14 Left: Wing design area. Right: Wing design aspect ratio AR. Both as a function of
the number of passengers and the nominal design range. . . . . . . . . . . . . . . . 87
7.15 Left: Turboshaft power rating. Right: Cruise altitude. Both as a function of the
number of passengers and the nominal design range. . . . . . . . . . . . . . . . . . 87
7.16 Results from the multi-objective optimization between operational costs and fuel
costs, colored by MTOW. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
7.17 Results from the multi-objective optimization between operational costs and fuel
costs, colored by cruise altitude (left) and mean velocity (right). . . . . . . . . . . . 89
7.18 Altitude (left) and Mach number (right) profiles for the optimized aircraft along the
costs Pareto front. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
7.19 Wing aspect ratio (left) and turboshaft rating (right) for the optimized aircraft along
the costs Pareto front. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7.20 Fuel intensity (normalized with range and total passenger count) as a function of
number of passengers and nominal design range under the mean speed constraint. 91
7.21 Left: Wing design area. Right: Wing design aspect ratio AR. Both as a function
of the number of passengers and the nominal design range under the mean speed
constraint. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7.22 Left: Turboshaft power rating. Right: Cruise altitude. Both as a function of the
number of passengers and the nominal design range under the mean speed constraint. 92

8.1 Evolution of battery cell weight and nominal block fuel for various cell’s specific
energies along the geometries for three technology levels. Batteries are highlighted
in green with the inverter and the electric motor in orange and yellow, respectively. 95
8.2 Evolution of component mass with cell-specific energy. Structural mass includes
wing, fuselage, and tail weight along with subsystems, furnishing, and operational
items. Powerplant includes turboshaft, gearbox, electric motors and inverters. Bat-
tery mass is expressed at pack level, and payload includes passenger and luggage
mass. For the Kerosene Parallel Hybrid with no thermal management. . . . . . . . . 96
8.3 Energy provided by the battery and by the kerosene (left). Energy intensity from the
battery, kerosene, and total energy (right). For the Kerosene Parallel Hybrid with no
thermal management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
8.4 Evolution of the state of charge and hybridization factor for the various aircraft
specific energy configurations for the Kerosene Parallel Hybrid with no thermal
management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
LIST OF FIGURES xxi

8.5 Evolution of electric motor power rating and gas turbine power rating with respect to
the cell specific-energy for the Kerosene Parallel Hybrid with no thermal management. 98
8.6 Gas turbine throttle and PSFC along the mission for the various aircraft specific
energy configurations for the Kerosene Parallel Hybrid with no thermal management. 98
8.7 Vertical altitude profile and Mach number profile for the various aircraft specific
energy configurations for the Kerosene Parallel Hybrid with no thermal management. 99
8.8 Grid of battery cell mass (left) and nominal block fuel (right) as a function of nominal
mission range and cell specific energy level for the Kerosene Parallel Hybrid with
no thermal management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
8.9 Total required mission energy (left) and fraction from it that comes from the battery
(right) as a function of nominal mission range and cell specific energy level for the
Kerosene Parallel Hybrid with no thermal management. . . . . . . . . . . . . . . . . 100
8.10 Aircraft MTOW (left) and increase over 23000 kg baseline (right) vs. mission range
and cell specific energy for the kerosene parallel hybrid (no thermal management). 100
8.11 Turboshaft (left) and electric motor (right) power ratings vs. mission range and cell
specific energy for the kerosene parallel hybrid (no thermal management). . . . . . 100
8.12 Relative decrease in block fuel (left) and increase in total energy requirement (right)
versus mission range and lower-bound cell specific energy for the Kerosene Parallel
Hybrid with no thermal management. . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.13 Evolution of battery cell weight and nominal block fuel for various cell’s specific
energies for both the thermal managed and non-thermal managed architectures. . 104
8.14 Evolution of the temperatures of the powertrain components along the mission for
the 628 Wh/kg battery cell design at 758nm design range. . . . . . . . . . . . . . . . 105
8.15 Evolution of the temperatures of the powertrain components along the mission for
each of the cell specific energies for the Parallel Hybrid. . . . . . . . . . . . . . . . . 106
8.16 Evolution of the state of charge and hybridization factor for the various aircraft spe-
cific energy configurations for the Kerosene Parallel Hybrid with thermal management. 107
8.17 Evolution of the altitude and Mach profile for the various aircraft specific energy
configurations for the Kerosene Parallel Hybrid with thermal management. . . . . . 107
8.18 Grid of battery cell mass (left) and nominal block fuel (right) as a function of nominal
mission range and cell specific energy level for the Kerosene Parallel Hybrid with
thermal management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
8.19 Grid of total energy (left) and hybridization level (right) as a function of nominal
mission range and cell specific energy level for the Kerosene Parallel Hybrid with
thermal management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
8.20 Evolution of the temperatures of the powertrain components along the mission
for the 700 Wh/kg battery cell design at 429nm. As it can be seen the variations
in hybridization, altitude and different components being constrained at different
points lead to high temperature fluctuations. . . . . . . . . . . . . . . . . . . . . . . 109
8.21 Aircraft MTOW (left) and increase over 23000 kg baseline (right) vs. mission range
and cell specific energy for the kerosene parallel hybrid with TMS. . . . . . . . . . 110
xxii LIST OF FIGURES

8.22 Turboshaft (left) and electric motor (right) power ratings vs. mission range and cell
specific energy for the kerosene parallel hybrid (thermal managed). . . . . . . . . . 110
8.23 Increase in fuel consumption and total energy between the thermal managed parallel
hybrid electric and the non thermal managed parallel hybrid. . . . . . . . . . . . . . 110
8.24 Emissions of CO2 from the burnt (top), and equivalent total emissions from the
energy production for the electric batteries considering the three energy mix carbon
intensity scenarios (mid and lower rows). . . . . . . . . . . . . . . . . . . . . . . . . 112
8.25 Total energy consumed for the energy-focused kerosene hybrid optimization (left)
and comparison with the fuel minimization optimization results. . . . . . . . . . . . 113
8.26 Installed battery mass for the energy-focused kerosene hybrid optimization (left)
and comparison with the fuel minimization optimization results. . . . . . . . . . . . 113
8.27 Consumed fuel for the energy-focused kerosene hybrid optimization (left) and com-
parison with the fuel minimization optimization results. . . . . . . . . . . . . . . . . 114
8.28 Emissions of CO2 from the burnt kerosene (top), and equivalent total emissions from
the energy production for the electric batteries considering the three energy mix
carbon intensity scenarios (mid and lower rows). Made for the energy efficient aicraft. 115
8.29 Comparison of the relative decrease in total CO2 emissions of the fuel minimization
with respect to the energy minimization. . . . . . . . . . . . . . . . . . . . . . . . . 116

9.1 Evolution of battery cell weight and nominal block fuel for various cell’s specific
energies along the geometries for three technology levels. Batteries are highlighted
in green with the inverter and the electric motor in orange and yellow, respectively.
For the hydrogen parallel hybrid with thermal management. . . . . . . . . . . . . . 120
9.2 Evolution of the installed battery cell mass (left) and nominal block fuel (right) for
the kerosene and LH2 architectures for both nominal only and nominal + extended.
And both considering TMS. Shaded areas represent parts where the extended mission
is penalizing the addition of batteries. . . . . . . . . . . . . . . . . . . . . . . . . . . 120
9.3 Evolution of component mass with cell-specific energy. Structural mass includes
wing, fuselage, and tail weight along with subsystems, furnishing, and operational
items. Powerplant includes turboshaft, gearbox, electric motors, inverters, battery
tank and TMS systems. Battery mass is expressed at pack level, and payload includes
passenger and luggage mass. For the Hydrogen Parallel Hybrid with TMS. . . . . . 121
9.4 Energy provided by the battery and by the hydrogen (left). Energy intensity from
the battery, hydrogen, and total energy (right). For the Hydrogen Parallel Hybrid
with thermal management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
9.5 Evolution of the state of charge and hybridization factor for the various aircraft
specific energy configurations for the Hydrogen direct combustion with TMS. . . . 122
9.6 Electric motor and gas turbine power ratings vs. cell specific energy for LH2 PH-TMS 123
9.7 Temperature evolution of powertrain components and for the 628 Wh/kg battery at
758 nm design range for LH2 PH-TMS. . . . . . . . . . . . . . . . . . . . . . . . . . 123
LIST OF FIGURES xxiii

9.8 Evolution of the temperatures of the powertrain components along the mission for
each of the cell specific energies for the LH2 Parallel Hybrid. . . . . . . . . . . . . . 124
9.9 Evolution of the temperatures of the powertrain components along the mission for
each of the cell specific energies for the LH2 Parallel Hybrid. . . . . . . . . . . . . . 125
9.10 Evolution of the installed cell battery mass (left) and nominal block fuel (right) along
our range-cell specific energy design space for the PH-TMS LH2 regional aicraft. . . 125
9.11 Evolution of the M T OW (left) and relative increase with respect to an aircraft with
an M T OW of 23 tons (right) along our range-cell specific energy design space for
the PH-TMS LH2 regional aicraft. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
9.12 Evolution of the total gravimetric efficiency (left) and nominal gabrimetric efficiency
(right) along our range-cell specific energy design space for the PH-TMS LH2 regional
aicraft. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
9.13 Evolution of the total consumed energy (left) and fraction from it that comes from
the battery (right) along our range-cell specific energy design space for the PH-TMS
LH2 regional aicraft. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
9.14 Evolution of the tuboshaft power rating of each engine (left) electric motor power
rating from each engine (right) along our range-cell specific energy design space for
the PH-TMS LH2 regional aicraft. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
9.15 Comparison of the equivalent CO2 emissions for the various LH2 PH architectures
under various electricity and hydrogen production scenarios. . . . . . . . . . . . . . 129
9.16 Total energy demand (left) and relative increase in energy consumption (right) for
the LH2 PH TMS architecture optimized for minimal energy compared to the one
for minimal fuel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
9.17 Installed battery cell mass (left) and installed delta (right) for the LH2 PH TMS
architecture optimized for minimal energy compared to the one for minimal fuel. . 131
9.18 Nomial block fuel (left) and nominal block fuel change (right) for the LH2 PH TMS
architecture optimized for minimal energy compared to the one for minimal fuel. . 132
9.19 Comparison of the equivalent CO2 emissions for the various LH2 PH architectures
under various electricity and hydrogen production scenarios for the energy mini-
mization optimization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
9.20 Comparison of the change in the relative equivalent CO2 between the fuel-minimization
and energy-minimization optimizations for the Hydrogen PH-TMS architecture. . . 134

10.1 Evolution of battery cell weight and nominal block fuel for various cell’s specific
energies along the geometries for three technology levels. Batteries are highlighted
in green with the inverter and the electric motor in orange and yellow, respectively.
For the hydrogen parallel fuel cell architecture with thermal management. . . . . . 140
10.2 Evolution of the state of charge and hybridization factor for the various aircraft
specific energy configurations for the Kerosene Parallel Hybrid with no thermal
management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
xxiv LIST OF FIGURES

10.3 Evolution of component mass with cell-specific energy. Structural mass includes
wing, fuselage, and tail weight along with subsystems, furnishing, and operational
items. Powerplant includes fuel cell, electric motors, inverters, battery tank, and TMS
systems. Battery mass is expressed at pack level, and payload includes passenger
and luggage mass. For the Hydrogen fuel cell with TMS. . . . . . . . . . . . . . . . 141
10.4 Evolution of the powertrain (power components) for the fuel cell and direct hydrogen
combustion architectures as a function of the battery cell specific energy. . . . . . . 142
10.5 Evolution of the state of charge and hybridization factor for the various aircraft-
specific energy configurations for the Hydrogen Fuel Cell with thermal management 142
10.6 Energy provided by the battery and by the hydrogen (left). Energy intensity from
the battery, hydrogen, and total energy (right). For the Hydrogen Parallel Hybrid
with thermal management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
10.7 Evolution of the total energy intensity (upper left), hydrogen consumed in nominal
mission (upper right), battery cell mass (bottom left), and fuel cell rating (bottom left). 144
10.8 Temperature evolution of powertrain components and for the 628 Wh/kg battery at
758 nm design range for FC LH2 PH-TMS. . . . . . . . . . . . . . . . . . . . . . . . 145
10.9 Evolution of the temperatures of the powertrain components along the mission for
each of the cell specific energies for the LH2 FuelCell Parallel Hybrid. . . . . . . . . 146
10.10 Evolution of the hydrogen tak state of fill and the tank pressure along the nominal
mission and specific energies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
10.11 Evolution of the installed cell battery mass (left) and nominal block fuel (right) along
our range-cell specific energy design space for the FuelCell aircraft. . . . . . . . . . 148
10.12 Evolution of the MT OW (left) and relative increase with respect to an aircraft with
an MT OW of 23 tons (right) along our range-cell specific energy design space for
the fuel cell architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
10.13 : Evolution of the fuel cell power rating of each engine (left) electric motor power
rating from each engine (right), along our range-cell specific energy design space
for the fuel cell regional aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
10.14 Evolution of the total consumed energy (left) and fraction from it that comes from
the battery (right) along our range-cell specific energy design space for the PH-TMS
LH2 regional aicraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
10.15 Comparison of the equivalent CO2 emissions for the various LH2 PH architectures
under various electricity and hydrogen production scenarios. . . . . . . . . . . . . 150
10.16 Relative change in total equivalent CO2 emissions (fuel cell vs direct hydrogen
combustion) for hydrogen produced by reforming natural gas and direct hydrogen
combustion (both in fuel minimization). . . . . . . . . . . . . . . . . . . . . . . . . . 152
10.17 Total energy demand (left) and relative increase in energy consumption (right) for
the fuel cell architecture optimized for minimal energy compared to the one for
minimal fuel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
10.18 Installed battery cell mass (left) and installed delta (right) for the fuel cell architecture
optimized for minimal energy compared to the one for minimal fuel. . . . . . . . . 153
LIST OF FIGURES xxv

10.19 Nominal block fuel (left) and nominal block fuel change (right) for the fuel cell
architecture optimized for minimal energy compared to the one for minimal fuel. . 154
10.20 Comparison of the equivalent CO2 emissions for the various fuel cell architectures
under various electricity and hydrogen production scenarios for the energy mini-
mization optimization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
10.21 Comparison of the change in the relative equivalent CO2 between the fuel-minimization
and energy-minimization optimizations for the fuel cell architecture. . . . . . . . . 156
10.22 Comparison of the change in the relative equivalent CO2 between the fuel-minimizing
and the energy-minimizing considering electrolysis. For both the direct hydrogen
combustion and the fuel cell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

11.1 Comparison of the design M T OW for all four aircraft architectures: conventional
kerosene, hybrid kerosene, hybrid hydrogen combustion, and hybrid fuel cell. . . . 160
11.2 Comparison of the installed battery cell mass for the three hybrid architectures:
kerosene combustion, hydrogen combustion, and hydrogen fuel cell. . . . . . . . . . 160
11.3 Comparison of the powetraint toal mass for the four powertrain architectures. For
the kerosene conventional, this includes propeller masses, gearbox, and turboshaft
core. then for both the the kerosene hybrid and the hydrogen combsution hybrid the
TMS system mass plus the electric motor and inverter are added towards this mass.
For the case of the fuel cell this mass includes the mass of the fuel cells, electric
motors, inverters, propellers, and TMS system. . . . . . . . . . . . . . . . . . . . . . 161
11.4 Comparison of the structural and secondary mass for all four architectures. This
structural mass considers the mass of the nacelles, fuselage, tail, landing gear, system
masses, furnishing, and operational items. For the case of the hydrogen architectures,
the structural mass of the hydrogen tank is also added. . . . . . . . . . . . . . . . . 162
11.5 Relative change in M T OW between the hybrid architectures and the optimized
conventional kerosene architectures. . . . . . . . . . . . . . . . . . . . . . . . . . . 163
11.6 Comparison of the consumed nominal block fuel (either kerosene or hydrogen) for
each of the four aircraft architectures. . . . . . . . . . . . . . . . . . . . . . . . . . . 164
11.7 Comparison of the payload fraction (computed as WP L /M T OW ) for each of the
four aircraft architectures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
11.8 Comparison of the total energy demand for each of the four aircraft architectures. . 166
11.9 Comparison of the energy demand from the turboshaft/fuel cell (left) and the battery
energy demand (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
11.10 Comparison of the mission demand energy intensity (in MJ/nm, computed as the
integrals of the provided powers). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
11.11 Comparison of the consumed nominal block fuel (either kerosene or hydrogen) for
each of the four aircraft architectures. . . . . . . . . . . . . . . . . . . . . . . . . . . 168
xxvi LIST OF FIGURES

11.12 Comparison of the relative change of carbon intensity with respect to the conven-
tional kerosene architecture for all three hybrid architectures. For the case of the
hydrogen architectures the emissions from natural gas reforming and for electrolysis
is considered. For the case of grid carbon intensity of 28 gCO2 /kWh. . . . . . . . . 170
11.13 Comparison of the relative change of carbon intensity with respect to the conven-
tional kerosene architecture for all three hybrid architectures. For the case of the
hydrogen architectures the emissions from natural gas reforming and for electrolysis
is considered. For the case of grid carbon intensity of 146 gCO2 /kWh. . . . . . . . . 172
11.14 Comparison of the relative change of carbon intensity with respect to the conven-
tional kerosene architecture for all three hybrid architectures. For the case of the
hydrogen architectures the emissions from natural gas reforming and for electrolysis
is considered. For the case of grid carbon intensity of 384 gCO2 /kWh. . . . . . . . . 173
11.15 Resultant aircraft geometries for a specific energy of 471Wh/kg and a nominal
mission design range of 428nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
11.16 Relative change of total equivalent emissions across the on-design mission ranges
for a battery cell technology of 471Wh/kg. All of them compare the on-design
range optimized hybrid architecture with its corresponding on-design conventional
kerosene. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
11.17 Weight breakdown of the proposed aicraft architectures at 471Wh/kg and an on-
design mission range of 428nm. Powertaint includes the mass of propeller, gearboxes,
turboshafts, fuel cells, TMS and so on, the only excluded mass is the batteries and
the structural mass of the cryogenic hydrogen tank. . . . . . . . . . . . . . . . . . 179
11.18 Comparison of the total equivalent CO2 emissions for the proposed architectures
with their optimized geometry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
11.19 Relative change of total equivalent CO2 emissions for the various architectures.
As a function of the specific energy and all relative to the optimized conventional
kerosene. Red gradient represents increasing emissions, and blue gradient represents
decreasing emissions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
LIST OF TABLES

2.1 Comparison of specific Energy and Power of fuels with batteries (at cell level). . . 12

3.1 Propeller cruise operating conditions for the ATR72-600 at its max cruise altitude. . 29
3.2 PyCycle gas turbine surrogate model training data. . . . . . . . . . . . . . . . . . . 32
3.3 Specifications of Commercial Battery Cells . . . . . . . . . . . . . . . . . . . . . . . 34
3.4 Specifications of commercial electric machines currently on the state of the art.
Power density is computed considering the maximum continuous power. . . . . . . 34
3.5 Specifications of inverters for state-of-art aviation electric motors. . . . . . . . . . . 35
3.6 Specifications and density metrics for fuel cell stack modules. . . . . . . . . . . . . 37
3.7 Specific power and specific volume of various propulsion components . . . . . . . . 40
3.8 Trajectory optimization and control variables. . . . . . . . . . . . . . . . . . . . . . 43

6.1 Flight path angle and equivalent airspeed for each mission control node ψi . . . . . 68
6.2 Design variables and bounds (left), and associated constraints with limits (right). . . 70
6.3 Comparison of optimized design variables with ATR72 reference values. Tangential
propeller Mach number at sea level for reference data computed assuming 1000 RPM
operation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.4 Comparison of mass breakdown between the TOPAZ model and ATR72-600 reference
values [193]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

7.1 Design variables and bounds (left), and associated constraints with limits (right).
Values in salmon denote variables defined per mission segment, values in pink
represent constraints enforced over a vector of mission segments, values in orange
refer to variables or constraints defined separately for each mission failure mode. . 77

8.1 Design variables and bounds (left), and associated constraints with limits (right).
Values in salmon denote variables defined per mission segment, values in pink
represent constraints enforced over a vector of mission segments, values in orange
refer to variables or constraints defined separately for each mission failure mode. . 94
8.2 Design variables and bounds (left), and associated constraints with limits (right).
Values in salmon denote variables defined per mission segment, values in pink
represent constraints enforced over a vector of mission segments, values in orange
refer to variables or constraints defined separately for each mission failure mode. . 102

xxvii
xxviii LIST OF TABLES

8.3 Total equivalent CO2 emissions for two study cases (943 nm and 300 nm) at 700 Wh/kg,
for different electricity carbon intensities. The relative change (%) is given with
respect to the corresponding 250 Wh/kg baseline aircraft. . . . . . . . . . . . . . . . 113
8.4 Total equivalent CO2 emissions for the energy-optimized aircraft at 300 nm and
700 Wh/kg, for different electricity carbon intensities. The relative change (%) is
with respect to the 250 Wh/kg baseline aircraft. . . . . . . . . . . . . . . . . . . . . 114

9.2 Design variables and bounds (left), and associated constraints with limits (right).
Values in salmon denote variables defined per mission segment, values in pink
represent constraints enforced over a vector of mission segments, values in orange
refer to variables or constraints defined separately for each mission failure mode. . 119
9.3 Comparison of total equivalent CO2 emissions for the higher-hybridization design
(700 Wh/kg, 300 nm) against the kerosene-only case, under different electricity
carbon intensities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
9.4 Comparison of total equivalent CO2 emissions for the higher-hybridization design
(700 Wh/kg, 300 nm) between hydrogen production by means of reforming and
electrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
9.5 Comparison of total equivalent CO2 emissions for the higher-hybridization design
(300 Wh/kg, 1200 nm) between hydrogen production by means of reforming and
electrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
9.6 Relative change in total equivalent CO2 emissions (fuel-minimized vs. energy-
optimized aircraft) for hydrogen produced by reforming natural gas, at two repre-
sentative design points. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134

10.2 Design variables and bounds (left), and associated constraints with limits (right).
Values in salmon denote variables defined per mission segment, values in pink
represent constraints enforced over a vector of mission segments, values in orange
refer to variables or constraints defined separately for each mission failure mode. . 139
10.3 Comparison of total equivalent CO2 emissions for the higher-hybridization design
(700 Wh/kg, 300 nm) between the direct hydrogen combustion and the fuel cell
architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

11.1 Comparison of emissions change [%] across architectures for two representative
design cases at 28gCO2 /kWh: high-specific-energy short range (700 Wh/kg, 300 nm)
and low-specific-energy long range (300 Wh/kg, 1200 nm). . . . . . . . . . . . . . . 170
11.2 Comparison of emissions change [%] across architectures for two representative
design cases at 146 gCO2 /kWh: high-specific-energy short range (700 Wh/kg, 300
nm) and low-specific-energy long range (300 Wh/kg, 1200 nm). . . . . . . . . . . . 171
11.3 Comparison of the lowest emissions decrease at 146 gCO2 /kWh across architectures
with their corresponding specific energy and range design points. . . . . . . . . . . 171
LIST OF TABLES xxix

11.4 Comparison of emissions change [%] across architectures for two representative
design cases at 384 gCO2 /kWh: high-specific-energy short range (700 Wh/kg, 300
nm) and low-specific-energy long range (300 Wh/kg, 1200 nm). . . . . . . . . . . . 174
11.5 Comparison of the lowest emissions decrease across architectures with their corre-
sponding specific energy and range design [Link] at 384 gCO2 /kWh. . . . . . 174
11.6 Relative change in total equivalent CO2 emissions with respect to the optimized
kerosene architecture, for two battery technology scenarios (414 and 471 Wh/kg).
Negative values indicate reductions in emissions. . . . . . . . . . . . . . . . . . . . 175
11.7 Comparison of main design parameters for the four reference architectures. Power-
taint includes the mass of propeller, gearboxes, turboshafts, fuel cells, TMS and so
on, the only excluded mass is the batteries and the structural mass of the cryogenic
hydrogen tank. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178

B.1 Software versions employed in project execution . . . . . . . . . . . . . . . . . . . . 213

B.2 Mathematical tolerances for the optimization runs. . . . . . . . . . . . . . . . . . . 214

C.1 Personel cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215

C.2 Computational cost cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
C.3 Software cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
C.4 Total cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217

D.1 Aligment of the project with the sustainable developement goals. . . . . . . . . . . 219
 CHAPTER

INTRODUCTION AND
OBJECTIVES 1
Contents
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Multidisciplinary optimisation: a necessary tool. . . . . . . . . . . . . . . 2
1.2.1 Previous studies and limitations . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Thesis Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

1.1 MOTIVATION

lean, cleaner, cleaniest — these superlatives aren’t just linguistic progression but an impera-
C tive for the aviation industry. In a world increasingly aware of its environmental footprint,
the quest for sustainable solutions has never been more urgent. In 2023, aviation accounted for
approximately 2.5% of global CO2 emissions [1], exhibiting growth rates surpassing those of all
other transportation sectors [2]. Without decisive intervention, the International Civil Aviation
Organization (ICAO) forecasts international aviation emissions could triple by 2050 relative to
2015 [3]. Compounding the issue, the industry also significantly contributes to non-CO2 climate
impacts (such as contrail formation and nitrogen oxide (NOx) emission) thereby exacerbating global
warming. These environmental challenges are further amplified by the aviation sector’s intrinsic
characteristics: extended vehicle lifecycles, prolonged development timelines, and reliance on histor-
ical operational data. Such factors collectively create a substantial gap between prevailing industry
practices and urgent climate policy objectives, especially when introducing new technologies such
as hybridization or hydrogen.
The conclusion is inescapable: we must accelerate the transition to cleaner aircraft immediately!.
However, this imperative faces formidable technical challenges, as aircraft represent extraordinarily
complex systems whose performance emerges from the intricate interplay of multiple disciplines:
aerodynamics, structures, propulsion, and weight distribution among others. Any novel technology
cannot be evaluated in isolation but must be assessed within this complex system of interdepen-
dencies. Fortunately, the evolution of sophisticated multidisciplinary optimization frameworks has

1
2 CHAPTER 1. INTRODUCTION AND OBJECTIVES

enabled the acceleration of the design process for innovative configurations. By simultaneously
considering multiple disciplines, these frameworks enable a comprehensive assessment of overall
system performance improvements, bridging the critical gap between environmental urgency and
technical feasibility.

1.2 MULTIDISCIPLINARY OPTIMISATION: A NECESSARY TOOL.

The development of multidisciplinary analysis and optimization (MDAO) emerged in the 1960s
and 1970s as a response to weight-critical applications in aerospace engineering. Pioneering work
by Schmit and Haftka [4] expanded traditional structural optimization frameworks to incorporate
additional disciplines, including drag minimization [5], flutter-constrained design methodologies
[6], [7], and thermal considerations [8]. Currently, multidisciplinary optimization has evolved into a
rapid growing domain encompassing diverse aeronautical applications such as gas turbine design [9],
[10], comprehensive aircraft system optimization [11]–[15], low sonic boom configurations [16], [17],
aero-stealth integration [18]–[20], aero-propulsive coupling [21]–[23], and aero-acoustic propeller
optimization [24]–[27]. Beyond aerospace and aeronautical engineering, MDAO methodologies
have been successfully applied to diverse engineering domains, including wind turbine blade design
[28]–[31], electric vehicle development [32], and energy infrastructure optimization [33]–[35].
Despite the breadth of existing literature on multidisciplinary design optimization (MDO), a
fundamental question remains: Why is MDO necessary? To answer this, consider a canonical example:
the aerodynamic optimization of a cantilever beam wing. The goal is to maximize the aircraft’s
range by optimizing both the wing’s beam wall thickness and twist distribution, while accounting
for static aeroelastic coupling and material limits [36], [37]. This problem can be approached
in two fundamentally different ways. One option is a sequential process (Fig.1.1, left) in which
aerodynamicists aim to minimize drag and structural engineers independently aim to minimize
weight, exchanging design updates iteratively. The other option is to consider the coupled system and
update both aerodynamic and structural design variables within a single unified multidisciplinary
optimization process (Fig.1.1, right).

Figure 1.1: Comparison of sequential (left) and multidisciplinary (right) design optimization for a
simple two-discipline wing. Extracted from [37]
1.2. MULTIDISCIPLINARY OPTIMISATION: A NECESSARY TOOL. 3

The distinction between the results of these two strategies is illustrated in Fig. 1.2, where
optimization over a two-discipline design space (twist and wing-box thickness) is shown. In the
sequential approach, the solution path tends to follow the aerodynamic disciplinary optimum,
gradually reducing thickness, decreasing structural weight, and improving efficiency and range.
In contrast, the MDO approach more aggressively explores the design space, even venturing into
infeasible regions (where the stress constraint is active) before converging to the global optimum
(maximum range subject to the stress constraint). It should be emphasized that the multidisciplinary
optimum does not coincide with the sequential optimum (which produces the aerodynamic optimum
subject to the structural constraint). This underscores the value of MDO: in highly coupled systems,
the global optimum is not simply the combination of each discipline’s individual best solution, but
rather a careful tradeoff between all the components to achieve the system optimum.

Figure 1.2: Optimizer trajectories for the sequential and the multidisciplinary optimization. Iso-lines
represent mission range (objective function). Extracted and adapted from [37]

This ability to reason at the system level, rather than within isolated disciplinary islands, is
fundamental to modern engineering design. For this reason, we regard Multidisciplinary Design
Analysis and Optimization (MDAO) not merely as a computational technique, but as a design
philosophy, a framework that enables engineers to explore and understand the behavior of highly
complex, tightly coupled, and high-dimensional design spaces in a precise and holistic matter. This
perspective is particularly crucial in the context of aircraft design, where the evaluation of novel
technologies (such as alternative powerplant architectures) demands a fully coupled analysis across
all relevant disciplines, including propulsion, aerodynamics, structural weight, and mission trajectory.
Only by capturing these interdependencies can designers properly account for integration trade-offs,
such as additional volume requirements, shifts in aircraft balance, or constraints on the mission
profile. In this way, MDAO is an indispensable tool for assessing the true system-level impact of
technological innovation.
4 CHAPTER 1. INTRODUCTION AND OBJECTIVES

1.2.1 Previous studies and limitations

In contemporary aircraft design practice, MDAO has become one of the essential tools for conceptual
and preliminary design (with even mainstream authors such as Raymer acknowledging its relevance
[38]). This philosophy is embodied in what can be described as an aircraft design framework: an
integrated, object-oriented computational environment in which models for weight estimation, pow-
ertrain analysis, aerodynamics, and mission performance are combined to enable optimization over
a chosen performance metric. In many cases, such frameworks are built on top of established MDAO
platforms such as OpenMDAO [39] or GEMSEO [40], making use of their optimization capabilities and
representing each implemented model as a discipline class within the back-end MDAO framework.
Emong these kinds of tools we can find the following:

• SUAVE: Released in 2015 by the Aerospace Design Lab at Stanford University, SUAVE [41] is an
aircraft design framework built for multi-fidelity aircraft performance analysis using an object-
oriented approach, with a standalone interface to optimization codes. Over the years, SUAVE
has incorporated a wide range of capabilities, including multiple aerodynamic performance
estimation methods (zero-fidelity analysis, lifting-line theory, panel methods, etc.), diverse
powertrain modules (electric motors, turbofans, scramjets) with various energy sources
(batteries, fuel cells), as well as noise estimation and mission optimization. While current
development seems to have stopped, SUAVE has been implemented for the evaluation and
optimization of: small transitioning UAVs [42], eVTOLs [43], [44], aero-propulsive optimization
[45], [46], batteries [47], [48], and noise [49]. Currently SUAVE is being substituted by RCAIDE
[50], developed by the Laboratory for Electric Aircraft Design and Sustainability from the
University of Illinois Urbana-Champaign.

• OpenConcept: Released in 2018 by the MDOLab at the University of Michigan, OpenConcept

is an OpenMDAO-based aircraft design framework initially developed for the conceptual
optimization of hybrid-electric kerosene aircraft [13]. It has since expanded its mission analysis
capabilities to include thermal modeling [51], [52], low-fidelity1 aero-structural optimization
[53], [54], podded heat exchanger modeling [55], and hydrogen-powered configurations [56].

• FastOAD: Developed by ONERA and ISAE-Supaero from initial work in 2015 (originally under
the name FAST), FastOAD is an OpenMDAO-based aircraft sizing framework that has been
applied to the analysis and optimization of general aviation aircraft [57], hydrogen-fueled
aviation [58]–[60], turbo-electric aircraft [61], and blended-wing configurations [62], among
others.

• Aviary: developed by NASA as the new in-house platform for aircraft modelling, Aviary is an
OpenMDAO-based aircraft sizing tool based on the NASA legacy tools GASP [63] (General
Aviation Synthesis Program) and FLOPS [64] (Flight Optimization System). The objective of
Aviary is to expand these codes in a more flexible Python environment that can be connected
to additional OpenMDAO libraries such as PyCyle [10], OpenAeroStruct [65] or Dymos [66].
1
Using VLM and beam structural elements.
1.2. MULTIDISCIPLINARY OPTIMISATION: A NECESSARY TOOL. 5

Aviary has also been able to perform robust optimization [67] by means of NASA’s UQPCE
tool [68].

Additional aircraft sizing MDAO tools include DLR’s OpenAD [15], MIT’s TASOPT [69]
(Transport Aircraft System OPTimization), and UC3M’s MOTIVATION [70]. Beyond this, MDAO is
also the focus of various European projects, including the AGILE [71] project, which focused on
advancing collaborative multidisciplinary design and integrating stakeholders across the aeronautical
supply chain; the COLOSSUS [72] project, which extends MDAO methods towards the analysis and
optimization of complex systems-of-systems; and ODE4HERA [73], which applies MDAO within a
digital engineering framework for the design and certification of hybrid-electric regional aircraft.
Despite the breadth of capabilities offered by these frameworks, the current state of the art
presents several limitations:

• A common issue is that the aircraft layout is often represented in an overly simplified manner:
although additional system masses are sometimes included in the analysis, they are frequently
not physically allocated within the geometry. This abstraction is often adopted to preserve
generality and maintain an object-oriented architecture, but it can lead to substantial simplifi-
cations that overlook key design constraints (such as the volumetric requirements for battery
installation).

• Another important limitation is the strong dependency on OpenMDAO. Most tools are built on
top of this framework, relying heavily on its methods and code structure, with the system
decomposed into numerous groups of explicit and implicit discipline blocks, often with
handwritten derivatives. While this facilitates efficient optimization and coupling, it also
reduces portability outside the OpenMDAO ecosystem. Moreover, relatively few frameworks
are based on GEMSEO; and given that GEMSEO originated as a European (Airbus) standard, this
constrains broader integration with industrial workflows in the European aerospace sector.

• Finally, although there are many studies comparing regional hybrid-electric architectures,
thermal management approaches, hydrogen combustion, and fuel-cell propulsion, compar-
atively few evaluate these options within the same framework. This lack of a common
modeling environment hampers direct comparison: divergences in reported results often
reflect framework-specific assumptions and biases rather than intrinsic differences in aircraft
design.

In order to solve some of these limitations, TOPAZ (the in-house GEMSEO-based aircraft sizing
tool used on this project) was presented in SciTech 2025 [74]. The objective of this tool was to
perform geometry and layout-aware aircraft sizing to consider effects as hydrogen tank trimming
and battery allocation limitations. However, the lack of thermal management, the hydrogen tank
model, and reduced optimization runs (only comparing direct hydrogen combustion and fuel cells)
left a lot of design space to explore.
6 CHAPTER 1. INTRODUCTION AND OBJECTIVES

1.3 OBJECTIVES

The central objective of this master’s thesis is to investigate, design, and optimize novel regional
aircraft architectures based on hybrid-electric and hydrogen propulsion, considering a range of
mission profiles, carbon intensities, and battery technology levels, with the aim of determining the
optimal architecture for near-future applications.
In order to achieve this general goal, the project has set several specific goals:

i. Develop and implement new physics-based models that more accurately represent the aero-
dynamics, weights, and powertrain systems (especially their thermal management and the
hydrogen storage).

ii. Define parametric aircraft architectures that capture the essential design variables of hybrid-
electric and hydrogen configurations, and integrate these within a multidisciplinary analysis
and optimization (MDAO) environment.

iii. Quantify the life-cycle carbon impact (CO2 -equivalent emissions) of each optimized archi-
tecture across different battery technologies, operational ranges, and electricity generation
mixes to assess their environmental viability under various energy transition scenarios.

A secondary objective involves contributing to the ongoing development of the in-house aircraft
synthesis tool TOPAZ by enhancing its capabilities with integrated thermal management modeling
and algorithmic differentiation. These enhancements will result in the advanced framework OPALO,
which will serve as the primary computational platform for this research.

1.4 THESIS LAYOUT

To ensure that the objectives are properly addressed and to guide the reader step by step through
the study, the following structure has been adopted:

• Reference aircraft: Chapter 2 introduces the baseline aircraft architectures and reviews the
current state of electricity and hydrogen production, providing the starting point and the
technological context for the work.

• Modeling methodology: Chapter 3 describes the new models developed to represent the
systems and technologies required for the next generation of aircraft.

• Aircraft architectures: Chapter 4 formalizes how each aircraft configuration is arranged,

outlining its layout and showing how the previously developed models are integrated.

• Multidisciplinary analysis: Chapter 5 introduces the concept of multidisciplinary analysis

and explains how it is implemented in this work, both conceptually and computationally.
1.4. THESIS LAYOUT 7

• Reference aircraft validation: Chapter 6 validates the multidisciplinary optimization frame-

work against existing aircraft with known performance and mass data.

• Optimization results: Chapters 7, 8, 9, and 10 present the optimization results for each
specific architecture, followed by a comparative synthesis in Chapter 11.

• Conclusions and future work: Chapter 12 summarizes the findings, highlights the main
contributions, and outlines potential avenues for further development.

The proposed structure follows a progression from context to detail and finally to integration:
first establishing the state of the art in aviation, then developing the component-level models,
and ultimately assembling them into a unified MDAO framework. Each architecture is examined
independently, and the thesis closes with a comparative analysis and overarching conclusions.
In addition, the thesis includes the following supplementary documents:

(I) Executive summary: Appendix A.

(II) Technical specifications: Appendix B.

(III) Project budget: Appendix C.

(IV) Alignment with the SDGs: Appendix D.

 CHAPTER

STATE OF THE ART AND

REFERENCE AIRCRAFT 2
Contents
2.1 A View on Sustainable Aviation . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.1 SAFs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.2 The electric and the hybrids . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.3 Hydrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 The environmental cost of energy . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 The regional turboprop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

The objective of this chapter is to introduce the reader to the topic of sustainable aviation, provide
an overview of the primary strategies for reducing the environmental impact of air transport, and
present the rationale behind the selection of the reference aircraft used for the optimization study.

2.1 A VIEW ON SUSTAINABLE AVIATION

Aviation has become a significant contributor to global greenhouse gas emissions. Between 1965 and
2020, annual CO2 emissions from air transport increased by approximately 340% (Fig.2.1). During
the same period, aviation’s share of global CO2 emissions rose from 1.22% to 2.47%[75]. At first
glance, this upward trend may suggest that aviation is becoming less sustainable. But is this really
the case?
A closer analysis reveals that the increase in emissions is primarily driven by rising demand.
Population growth and the broader accessibility of air travel (especially through the expansion of
low-cost carriers) have steadily increased passenger traffic, and therefore total emissions (Fig.2.2
left axis). This does not mean, however, that the aviation sector has ignored its environmental
responsibilities. On the contrary, engineers and regulators have made consistent progress toward
improving the efficiency of flight. Advances in engine technology, lighter airframes, and more
effective operational practices have all contributed to a long-term reduction in carbon intensity
(Fig.2.2 right axis), measured as the amount of CO2 emitted per revenue passenger-kilometer[78].

9
10 CHAPTER 2. STATE OF THE ART AND REFERENCE AIRCRAFT

Figure 2.1: Yearly CO2 emissions derived from aviation. The first trend is based on [75], the second
trend is extracted from [76], the third trend is based on [77]

Despite these improvements, the pace of technological progress has not been fast enough to
offset the rapid growth in demand. While modern aircraft emit less CO2 per passenger than earlier
generations, the overall emissions continue to rise. Part of the reason lies in the long operational
lifespan of commercial aircraft; airlines tend to retain older models to redeem some of the initial
investment, meaning that more efficient technologies take time to make a significant impact across
the global fleet, giving the industry some inertia. As a result, aggregate emissions continue to grow,
even if newer technologies are constantly being introduced.
This situation highlights a fundamental challenge in achieving sustainable aviation. To reach
a net reduction in emissions, technological improvements must outpace the growth in demand.
Incremental efficiency gains, while important, are no longer sufficient. Meeting future environmental

Figure 2.2: Aviation demand in revenue passenger per kilometer (RPK) and aviation carbon intensity
in grams of CO2 per revenue passenger per kilometer over the years. Data from [79]
2.1. A VIEW ON SUSTAINABLE AVIATION 11

goals will require transformative changes in how aircraft are powered and operated. In this thesis,
the term sustainable aircraft refers to designs capable of achieving net-zero CO2 emissions through
advanced propulsion technologies, while still fulfilling the performance and reliability requirements
of modern air transport.
Sustainable aviation can be therefore pursued through several pathways: the use of sustainable
aviation fuels (SAFs), which aim to produce a net-zero increase in CO2 emissions1 ; hydrogen-
powered aircraft2 ; and fully electric aircraft. These propulsion advances are complemented by
improvements in aerostructures and aerodynamics, creating synergistic effects that enhance overall
efficiency.

2.1.1 SAFs

Among the various pathways toward sustainable aviation, sustainable aviation fuels (SAFs) represent
the most immediately deployable solution [80], [81]. Unlike hydrogen or fully electric aircraft, which
require extensive technological maturation and infrastructure development, SAFs can serve as drop-
in replacements or blends with conventional jet fuel in existing aircraft and engines with minimal
modifications [82]. This compatibility with current fleets and established fueling infrastructure [83]
positions SAFs as an attractive near-term option for aviation decarbonization.
However, SAFs face significant barriers that limit right-away adoption. The primary challenge
is economic: production costs remain substantially higher than conventional jet fuel, with pricing
variations dependent on geographic location and regional energy mix characteristics [84], [85].
This cost differential creates hesitation among airlines operating in an inherently cost-sensitive
industry. Additionally, the sustainability credentials of SAFs are intrinsically linked to feedstock
sourcing practices. Biomass production for SAF can trigger land-use changes that displace food
crops, increase food prices, and threaten natural ecosystems [86]–[88].
While SAFs offer a pragmatic pathway to near-term emission reductions, realizing their full cli-
mate potential requires overcoming these economic, logistical, and sustainability-related challenges.

2.1.2 The electric and the hybrids

Another alternative to conventional kerosene is the use of electric batteries, with the potential
of zero CO2 emissions (as long as the energy is produced from green energy sources), electric
batteries have the possibility of producing a true zero-emission aircraft3 . The zero emissions here
would only consider the operation of the powertrain, with the manufacturing within the Life Cycle
Assessment still producing a significant amount of emissions. However, with lower specific energies
and specific powers (see Table 2.1), higher masses of batteries are required to match the same energy
stored in traditional aviation fuels. Therefore, fully electric aviation is heavily limited to shorter
operational routes where the total energy is reduced. An example of this approach from current
aircraft companies is the 30-seating regional ES-30 aircraft by Heart Aerospace (Figure 2.3), which
1
However, SAFs may still result in other pollutants such as NOx , CO, and unburned hydrocarbons (UHCs).
2
When based on fuel cell technologies, these aircraft emit only H2 O at the exhaust.
3
Considering that they do not emit CO2 nor do they form other forms of emissions such as contrails
12 CHAPTER 2. STATE OF THE ART AND REFERENCE AIRCRAFT

targets a maximum range of 200 km in all-electric and 400 km in a hybrid-electric configuration

[89]. Other companies such as Elysian are proposing bigger concepts such as the E9X with 90 seats
and a range up to 800 km. While these ranges might seem large, they are still underneath the 1345
km of an ATR42-600 [90] (with 42 seats, comparable to regional aircraft like the ES-30) or the 4537
km of an Embraer E190 [91] (with 100 seats, comparable to the E9X).

Technology Year Spec Energy (Wh/kg) Spec power (W/kg) Ref

Jet-A (kerosene) - 12 000 - [92]
H2 - 33 000 - [93]
B787 batt (LPV65) 2013 102 510 [94]
X57 batt (INR18650-30Q) 2018 258 507 [95]
Zenlab eVTOL batt 2022 260 2500 [96]
HER concept batt by 2030 423 1269 [97]

Table 2.1: Comparison of specific Energy and Power of fuels with batteries (at cell level).

Figure 2.3: Left: Heart HS-30 aircraft concept [89]. Right: Elysisan E9X aircraft concept [98].

Given the current limitations in fully electric aircraft range capabilities, most "high-power
electrified" aircraft employ hybrid electric architectures 4 . These hybrid configurations involve
partial electrification of the powerplant system, enabling the deployment of a downsized gas turbine
that operates more efficiently during cruise conditions [99]–[101]. This downsizing is feasible
because the electric propulsion system provides supplementary power during high-demand flight
phases such as takeoff and climb, effectively functioning as an electric "booster", reducing the
capacity requirement of the electric battery. However, the question wether to hybridize or not to
hybridize depends strongly on the range and the electric technology, as low specific energies and
long ranges mean that the battery will be quickly depleted, leading to empty mass that will only
increase drag5 and lead to a higher overall block fuel [13].
Hybrid powerplant architectures can be broadly classified into three main configurations6 [102].
In a series hybrid system (Figure 2.4, top left), where a gas turbine produces mechanical power
4
This is also the case with the Heart aircraft concept, where a traditional turboshaft can be operated in tandem with
the electrical motor.
5
By means of higher induced drag.
6
Other architectures, such as turboelectric, are excluded due to their limited relevance in larger aircraft (regional and
above).
2.1. A VIEW ON SUSTAINABLE AVIATION 13

that is converted into electricity, which, along with battery power, feeds an electric bus supplying
the propeller-driving electric motors. In a parallel hybrid system (Figure 2.4, top right), where a
gas turbine and an electric motor (powered by a battery) deliver shaft power via a combination
gearbox to an inboard propeller. And in a parallel-series hybrid configuration (Figure 2.4, bottom),
a combination of both is present: the electric motor connected to the gearbox can operate as a
generator, supplying power to additional electrically driven outboard propellers.

Figure 2.4: Serial hybrid, parallel hybrid, and parallel-serial hybrid powerplant architectures.

In terms of overall performance, parallel hybrid architectures generally outperform series

configurations [70], [103], [104]. This advantage stems from the fact that series systems must size
their electrical components to meet the total power demand, whereas parallel hybrids only need
to accommodate the peak boost power from the electric system. As a result, parallel architectures
enable a lighter powerplant, which can allocate more weight to batteries within a fixed maximum
takeoff weight. Incorporating a series subsystem into a parallel architecture becomes particularly
attractive for low-noise distributed propulsion applications, where spreading thrust across multiple,
larger, and less heavily loaded propellers reduces tip Mach numbers and, consequently, the overall
sound pressure level (OSPL) [105].
Several hybrid aircraft configurations are currently in various stages of development and certifi-
cation, demonstrating the practical viability of these architectural approaches. The Airbus-Safran
EcoPulse [106] project exemplifies series-hybrid distributed propulsion implementation, utilizing
a modified Daher TBM 900 turboprop as the base platform. This configuration employs electric
propulsors powered by a turbogenerator system that operates independently from the primary
14 CHAPTER 2. STATE OF THE ART AND REFERENCE AIRCRAFT

turboprop engine, creating a series-hybrid architecture focused on distributed thrust generation.

Figure 2.5: Left: Ampere EcoCaravan during takeoff [107]. Right: Airbus-Safran EcoPulse during
flight testing [106].

The Ampaire EcoCaravan [107] represents another significant development in hybrid aviation
technology (currently progressing through FAA certification processes). This aircraft modifies the
Cessna 208 Caravan platform by integrating the AMP-H570 hybrid powerplant system [108]. The
EcoCaravan employs a parallel-series hybrid arrangement that is designed towards operational
flexibility, allowing the primary gas turbine to function either in boost mode with electric system
assistance or in generator mode for battery charging and electric propulsion.

2.1.3 Hydrogen

When considering alternative fuel sources, hydrogen emerges as the next logical choice. With its
high specific energy (see Table 2.1) and zero CO2 emissions, hydrogen holds significant promise
as a key enabler of a decarbonized aviation sector. However, the main challenge associated with
hydrogen-powered aviation lies in its low volumetric energy density. While each kilogram of
hydrogen contains a substantial amount of energy, storing this energy requires a relatively large
volume (see Figure 2.6). Hydrogen is typically stored either as a compressed gas at pressures ranging
from 350 to 700 bar, or as a cryogenic liquid in insulated tanks to increase its density and improve
storage efficiency.
Although hydrogen is often perceived as a premature or emerging technology, interest in its
application to aviation actually dates back to the mid-20th century. To fully appreciate the current
momentum surrounding hydrogen propulsion, it is worth taking a brief historical detour.
One of the earliest significant efforts occurred in the 1950s, when the United States began
investigating hydrogen as a fuel for high-altitude flight due to its favorable combustion characteristics
at low atmospheric pressures [109]. This research culminated in the development of the Bee project,
which aimed to modify a B-57B aircraft to operate on either liquid hydrogen or JP-4 using a specially
adapted Curtiss-Wright J65 engine (see Figure 2.7) [110]. While the program demonstrated technical
feasibility, high operational strains did not push for more prototypes.
Additional concepts—particularly those targeting high-Mach flight, where hydrogen could also
serve as a coolant—were explored and, in some cases, received funding, such as the CL-400 project
2.1. A VIEW ON SUSTAINABLE AVIATION 15

[111]. However, the challenges associated with storing, handling, and integrating liquid hydrogen
proved too significant for widespread adoption at the time.

Figure 2.6: Comparison of gravimetric density (in MJ/kg) and volumetric density (MJ/L) for various
fuels. Adapted from [112].

Figure 2.7: Left: System description of the modified B75B. Right: modified B75B during ground
taxiing [113].

A few decades later (in 1988), hydrogen-powered aviation re-emerged on the global stage with
the Soviet Union’s development of the Tupolev Tu-155 [116]. This aircraft was a modified Tu-154B
airframe that replaced one of its three turbofan engines with a specially designed NK-88 engine
capable of operating on both liquid hydrogen and liquid methane stored in a 18 m3 cryogenic fuel
tank (see Figure 2.8). While the project maintained the dual-use of hydrogen in aviation, there was
already a significant environmental vision of hydrogen as a fully restorable energy carrier that could
be used to vector the energy produced from nuclear sources. However, with the dissolution of the
16 CHAPTER 2. STATE OF THE ART AND REFERENCE AIRCRAFT

Figure 2.8: Left: Tu-155 fuel storage and distribution system [114]. Right: Tu-155 during flight testing
[115].

Figure 2.9: Left: Airbus zero-e aircraft concept [123]. Right: Universal Hydrogen modified Dash-8
[124].

Soviet Union only a few years later, the program was discontinued before further development or
operational integration could occur.
Today, interest in hydrogen as the aviation fuel of the future has been revitalized, supported
by both public initiatives and private industry. On the policy and research side, the European
Union’s Alliance for Zero-Emission Aviation (AZEA) [117] plays a central role in coordinating efforts
to accelerate the deployment of hydrogen-powered aircraft. This vision is further materialized
through large-scale projects funded under Horizon Europe and the Clean Aviation Joint Undertaking,
including GOLIAT [118], CAVENDISH [119], FASTER-H2 [120], H2ELIOS [121], and HEROPS [122].
In parallel, several private-sector initiatives are pursuing hydrogen propulsion technologies, such as
Airbus’s ZEROe (Fig.2.9 left) and smaller demonstrator aircraft from companies such as ZeroAvia,
H2FLY, and Universal Hydrogen (Fig.2.9 right).
Among the various hydrogen propulsion concepts, fuel cell-based systems have gained popularity
over direct combustion approaches for several compelling reasons. Primarily, fuel cells do not produce
NOx emissions, as they circumvent the high-temperature combustion process that generates these
pollutants through the Zeldovich mechanism [125]. Additionally, fuel cell systems offer greater
flexibility for aircraft integration, particularly in dual-fuel configurations where hydrogen fuel cells
2.2. THE ENVIRONMENTAL COST OF ENERGY 17

can complement conventional kerosene-based powertrains. A notable example is the HERA [126]
concept, where a fuel cell is positioned beneath the passenger cabin to generate electricity and
augment the primary kerosene propulsion system [127].
However, despite current improvements in hybridization and the evolving optimization of
fuel cell systems, fuel cell-based aircraft still face significant limitations. The primary challenge,
while common to all hybrid architectures, disproportionately affects fuel cell systems: thermal
management. Efficient fuel cells achieve peak efficiency levels of approximately ηF C ≈ 0.6 [128],
with decreasing efficiency as operating power increases. When compared to the peak efficiency
levels of electric machines (ηEM ≈ 0.95 [129]), this disparity indicates that fuel cell systems require
extensive thermal management to dissipate the substantial residual heat load generated during
operation. This thermal management challenge, combined with the inherently low volumetric
power density of fuel cells, necessitates the distribution of propulsion systems across multiple
pods that must also incorporate heat exchangers, as exemplified in the ZEROe concept (Fig. 2.9
left). Consequently, the resulting increase in surface area—from additional nacelles and extended
fuselage length—along with the pressure losses introduced by heat exchangers, leads to overall
lower aerodynamic efficiency in fuel cell-based systems compared to their conventional kerosene
counterparts.
The second challenge that affects hydrogen aircraft more broadly is liquid hydrogen distribution
and storage. Operating at cryogenic temperatures, liquid hydrogen requires specialized containment
tanks and metal fracture-resistant piping systems [130]. Furthermore, hydrogen in civil aviation
demands substantial support infrastructure that is not currently present in the aviation sector. This
creates a classic "chicken-and-egg" problem: the absence of infrastructure makes hydrogen aircraft
commercially unattractive, while the lack of operational aircraft reduces the incentive to invest in
the necessary ground infrastructure.

2.2 THE ENVIRONMENTAL COST OF ENERGY

As will be studied in the following sections, environmental aviation can sometimes get overly hyped
when considering that energy from batteries and hydrogen is carbon-free. While there is some truth
to this statement (as electricity can be produced from clean sources, and hydrogen can be produced
by electrolysis) the reality is that energy production is a complex topic where many energy sources
and network stability constraints impose additional CO2 emissions. Within this thesis, we will
explore a first approximation to an LCA (Life Cycle Assessment) of the emissions of hybrid electric
and hydrogen aircraft, where we will consider additional overheads in the production of the energy
(electricity) or energy vector (hydrogen). However, we will not consider additional emissions due to
transport and infrastructure generation, which represent important areas for future research.
Regarding the cost of electric energy, we can observe that different countries have vastly different
carbon intensities (gCO2 per kWh, see Fig.2.10). This variation is due to the energy mix of each
country (Fig.2.11), where countries such as India and Poland that rely heavily on fossil fuels tend
to have high carbon intensities, while other countries such as Iceland, which produces most of its
18 CHAPTER 2. STATE OF THE ART AND REFERENCE AIRCRAFT

708
CO2 Intensity [g CO2 / kWh]

615
600 560

384
400

200 146
96 103
28 44
0

SA
ce

ia
d
d

ia

il

n
az

in

n
n

ai

d
ak
an

la
ela

Ch

In
Sp
Br
ov

Po
Fr
Ic

Sl

Figure 2.10: Energy production carbon intensity for various nations. Data adapted from [131]

energy via geothermal and hydroelectric sources, or France, which has a heavy nuclear contribution
to its energy mix, demonstrate lower carbon intensities.
This is pivotal when considering hybrid-electric aviation, as for the aircraft to make an envi-
ronmental impact, it must not only reduce direct emissions from fuel combustion but also do so
with an electric source that does not produce higher emissions than the reductions achieved from
fuel savings. For further comparisons, we will consider four scenarios: a very-low carbon intensity
scenario of 28 g CO2 /kWh, a moderate-low scenario of 146 g CO2 /kWh, a moderate-high scenario
of 384 g CO2 /kWh, and a high scenario of 708 g CO2 /kWh.

100 Renewables Fossil Fuels Nuclear

80
Energy share [%]

60

40

20

0
SA
ce

ia
d
ia

il

n
y

az

in
wa

n
ai

d
ak
an

la
U

Ch

In
Sp
Br
or

ov

Po
Fr
N

Sl

Figure 2.11: Energy mix as 2024 for various nations. Adapted from [131].

Regarding hydrogen production, we can see that the current scenario remains predominantly
carbon-intensive, with hydrogen classified by "color codes" reflecting production methods and
2.3. THE REGIONAL TURBOPROP 19

Carbon Intensity [kg CO2 / kg H2 ]

33.7

30
22.7
20
14.5
11.4
10 6.9

1.3
0

ia
d

SA
s

n
ga
ai
n

tio

d
ela

Sp

In
ca
al
Ic

is

is
ur

sifi
is

ys

ys
at
is

ys

ga
ol

ol
ys

N
ol

tr

tr
ol

tr

al
ec

ec
tr

ec

Co
El

El
ec

El
El

Figure 2.12: Comparison of emitted CO2 per kg of produced H2 for various energy sources [136].

associated emissions. Gray hydrogen, which accounts for approximately 60≈ % of global production,
is produced via steam methane reforming (SMR) without carbon capture. Black and brown hydrogen
derive from coal gasification (with brown specifically denoting lignite-based processes), representing
about 20% of production. Blue hydrogen employs SMR with carbon capture and storage (CCS),
while turquoise hydrogen uses methane pyrolysis to capture carbon in solid form. Green hydrogen,
produced through renewable-powered electrolysis, represents the cleanest option but currently
comprises only 0.4% of global production. Pink hydrogen refers to nuclear-powered electrolysis,
though it remains less commonly deployed [132]–[134].
Despite green hydrogen’s potential for aviation decarbonization, electrolysis remains econom-
ically challenging compared to fossil fuel-based methods. Moreover, emissions calculations for
electrolytic hydrogen can be misleading when assuming direct renewable energy sourcing, which
may not reflect actual grid energy mixes and demand patterns. For emission estimation purposes, this
analysis assumes electrolytic hydrogen production efficiency of approximately 70% [135], requiring
47.57 kWh to produce 1 kg of H2 (equivalent to hydrogen’s lower heating value). This enables com-
parison of carbon intensities between fossil fuel-based production methods and electrolysis under
various energy mix scenarios (Fig.2.12). The analysis reveals that electrolytic hydrogen production
using carbon-intensive electricity grids can result in higher overall CO2 emissions than conventional
fossil fuel-based methods.
Considering now that a kilogram of burned Kerosene emits 3.15 kg CO2 , and that hydrogen has
2.79 times more energy (by means of lower heating value), only hydrogen sources with a carbon
intensity lower than 8.79 kg CO2 /kg H2 can compete with kerosene in terms of overall carbonization.

2.3 THE REGIONAL TURBOPROP

Seeing the current aviation scenario and the cost of energy production, we must now go back to
selecting our desired aircraft architecture. Looking back to the previous projects it is evident that
many current novel aircraft architectures are focused on the regional turboprop formula, from the
20 CHAPTER 2. STATE OF THE ART AND REFERENCE AIRCRAFT

HS-30 and E9X to the ZEROe and Universal Hydrogen’s modified Dash-8. The question is why?.
Examining the aircraft that operate daily flights (Fig. 2.13), it can be observed that regional
turboprops correspond to only 6.77% of total daily flights, while narrow-body aircraft account for
65.97% of operations. Despite this lower frequency, regional turboprop flights serve a crucial role in
commuter operations, particularly in markets characterized by short-range, high-frequency routes.
A notable example is the Spanish Canary Islands, where the airline Binter [137] operates inter-island
flights using the ATR72 for routes that span only a few dozen kilometers. Similarly, the Dash-8
Q400 was operated by the Polish airline LOT [138] on short domestic routes, such as the Gdansk to
Warsaw corridor. Flights that would not fill a narrowbody but are still necessary.

20000 19,270
Number of aircrafts

15000

10000

5000
2,783
1,420 1,977
1,158 1,196 593 600
216
0
on
P

el i

er
rc ge

dy

dy

et
je

th
lj

H
st
ai lar
ft

bo

bo

rb
al

O
ra

al

Pi
Tu
on
ry

Sm
w
e
id

ro
Ve

gi
W

ar

Re
N

Figure 2.13: Aircraft types operating in the EUROCONTROL area. Avg. daily flights 2024. Data from
[139].

From an efficiency perspective, turboprops occupy a unique position in the aviation landscape.
While several studies demonstrate that turboprop aircraft can outperform equivalent turbofan
counterparts [140], [141], regional turboprops (and regional aviation in general) tend to exhibit
higher carbon intensity compared to narrow-body aircraft in operational practice. This phenomenon
occurs because regional aircraft typically operate missions with shorter cruise segments and lower
load factors, resulting in regional aviation (both turboprops and jets) having carbon intensities in
the order of 158 g CO2 /RPK, approximately 88% higher than narrow-body aircraft [142].
However, when comparing the carbon intensity of regional turbojets and turboprops specifically,
regional turboprops demonstrate lower carbon intensities (Fig. 2.14). This advantage stems partially
from the fact that turboprop aircraft are less penalized during lower-altitude operations, as variable-
pitch propellers enable more efficient operation at the reduced speeds during the LTO and airport
procedures.
Examining the typical operational envelope of regional turboprop aircraft reveals their passenger
capacity ranges from small commuter aircraft carrying just a few passengers on short routes to larger
variants accommodating 70-80 passengers (Fig. 2.15 top left). This capacity flexibility, combined with
their characteristically lower wing loadings (Fig. 2.15 top right), provides commuter and regional
2.3. THE REGIONAL TURBOPROP 21

200
2013
CO2 intensity [g CO2 / RPK] 2019
150

100

50

J9 dair

J2 dair

J7 dair
95 r

90 r

75 r

70 r

J1 aer
-6 R

40 d

E1 rae

E1 rae

E1 rae

E1 rae
8- n
72 AT

00

00

00

45
00

h illa

ER br
CR na

CR na

CR na
b

b
Em

Em

Em

Em

Em
as v

Ca

Ca

Ca
D a H
e
D

Figure 2.14: Comparison of carbon intensity between various models of turboprop and turbojet
regional aircraft. Data from [142].

Figure 2.15: Trends for the regional turboprop and regional turbofan aircraft. The database contains
aircraft from longer regionals to pure commuters. Designs range from 1960 to 2021.

turboprops with considerable operational versatility in the short-range market segment. However,
this adaptability comes at the cost of being limited to lower cruise speeds and reduced operational
22 CHAPTER 2. STATE OF THE ART AND REFERENCE AIRCRAFT

Figure 2.16: Dash-8 400 [143] and ATR72-600 [144].

ranges compared to their turbofan counterparts (Fig. 2.15 bottom left and bottom right).
However, the prevalence of turboprop platforms in clean aviation demonstrators can be attributed
to a fundamental operational characteristic: they operate at lower power levels. These reduced power
requirements enable the development of propulsion systems that can scale progressively from
kilowatt to megawatt levels. Given that power requirements scale linearly with velocity for a given
thrust level, turboprop demonstrators allow researchers to validate new technologies within lower
power envelopes as the technology matures. This approach is particularly advantageous for fuel
cell concepts, where the thermal management system (TMS) heat dissipation requirements scale
with operating power. Lower heat loads consequently reduce the technological complexity, initial
capital investment, and volumetric requirements for novel powerplants. This consideration becomes
especially critical for retrofit demonstrators, where hybrid architectures with inherently lower
volumetric power densities must be integrated within existing airframe constraints.
For the present thesis, this evolutionary approach is maintained by selecting a representative
regional turboprop as the baseline for analysis. Among the most prevalent turboprop aircraft in
current operation are the ATR72-600 and the Dash-8 Q400 (Fig. 2.16). Between these two platforms,
an aircraft configuration similar to the ATR72 has been selected, featuring 72 seats arranged in
18 rows of four seats each and a design maximum range of 758 nautical miles with maximum
payload. The ATR72 platform was chosen due to its lower cruise speeds and reduced installed power
requirements, which align better with the technological constraints of emerging propulsion systems.
While this baseline aircraft configuration has a slightly lower passenger capacity than the HERA,
ZEROe, and E9X concepts, it represents a mature, state-of-the-art turboprop design that provides an
appropriate foundation for comprehensive comparative studies and technological assessment.
Regarding the propulsion architectures to be evaluated, this study will examine four distinct
configurations: the conventional twin turboprop architecture serving as the baseline reference,
a parallel hybrid architecture utilizing kerosene fuel, a parallel hybrid architecture employing
direct combustion hydrogen, and a distributed fuel cell architecture featuring four hydrogen fuel
cell powerplants. The objective in this case will be to use optimization to see "the best" of each
architecture and then, from what we have seen of carbon intensity, evaluate when each architecture
can be optimal.
 CHAPTER

MODELLING
METHODOLOGY 3
Contents
3.1 Aerodynamic modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1.1 Coefficient estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1.2 Balance and static stability . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2 Powerplant modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.1 Propeller model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.2 Gas Turbine Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2.3 Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2.4 Electric motors and inverters . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.2.5 Fuel cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.6 Heat exchangers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.2.7 Hydrogen tanks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3 Weight estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.4 Trajectory definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.5 Aicraft control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

This chapter will describe all the methods and models implemented to predict aircraft perfor-
mance. The presentation begins with aerodynamic performance methods (Section 3.1), followed by
the power-plant model (Section 3.2), the weight estimation models (Section 3.3), and the trajectory
and control modeling (Section 3.4 and 3.5). The objective of the chapter is to explain the building
blocks that are then assembled into an architecture and an MDAO framework.

3.1 AERODYNAMIC MODELING

The objective of aerodynamic modeling is to estimate the forces acting on the aircraft that dictate
performance requirements throughout the mission. To achieve this, OPALO estimates performance
using two distinct approaches: a steady flight aerodynamics model (operating in wind axes for flight

23
24 CHAPTER 3. MODELLING METHODOLOGY

segments such as climb and cruise) and a ground acceleration model (for ground roll acceleration
and deceleration segments).
The steady flight model (Figure 3.1) considers forces with respect to the wind axis, which is
tangential to the aircraft trajectory. The primary forces in this axis are the perpendicular lift force L
and the tangential drag force D. The model assumes that the aircraft is oriented at an angle of attack
α relative to the wind axis, while the wind axis is oriented at a flight path angle γ with respect to
the horizon. Additional forces include a pitching moment M and thrust T , with the thrust either
projected at the angle of attack α, or assumed to be coaxial with the drag. The weight component is
projected between the global and wind axes using the flight path angle γ.

Figure 3.1: Aerodynamic forces model for steady flight. Forces are assumed to be applied over the
center of mass. The horizontal line represents the horizon, and the curved line represents the flight
trajectory.

All of these forces and moments (in wind and global axes) can be arranged in wind axes (Eq.3.1c).
Where in this case δe represents the horizontal tail deflection angle (also illustrated in Figure 3.1).
By solving these equations for equilibrium conditions (Fzw = Fxw = CMyw = 0), the trim values
of α, δe , and Treq can be determined, where Treq is the thrust required for steady flight.

q · Sref · CL(α, δe ) −W · cos(γ) = Fzw (3.1a)

| {z }
L

Treq − q · Sref · CD(α, δe ) −W · sin(γ) = Fxw (3.1b)

| {z }
D

CM (α, δe ) = CMyw (3.1c)

For flight conditions where we have non-zero acceleration (for example, if we are increasing the
velocity during cruise conditions), the total thrust is given by Equation 3.2. This consideration is
particularly important for trajectory optimization as it accounts for the rate of change of aircraft
velocity and penalizes rapid, nonphysical changes in velocity.
3.1. AERODYNAMIC MODELING 25

Ttotal = Treq + W · accel (3.2)

For segments involving ground contact (such as takeoffs and landings), a ground acceleration
model is employed (Figure 3.2). This model calculates the overall residual force in the horizontal
component Fx , later to be used to compute the acceleration of the aircraft. In terms of the considered
forces, for the vertical component lift L and weight W forces are present; the negative residual
sum of these forces represents the normal force, which can be distributed between the front wheel
normal force N1 and the main wheel normal forces N2 , or aggregated into a single normal force N .
From this normal force and the ground coefficient of friction µ, the friction force Fr can be
computed. The residual axial force is then calculated using the current imposed thrust T and the
friction and drag forces. It is important to note that in this ground model, no angle of attack α or
tail trim δe is considered as a state variable. Instead, the wing operates at a fixed lift coefficient of
CL = 0.3 (approximately the coefficient that minimizes takeoff run length), while the tail is assumed
to be axial and non-force-generating.

Figure 3.2: Aerodynamic forces model for ground acceleration

The total acceleration can therefore be expressed as Ec.3.3, where q is the dynamic pressure1 ,
Sref is the reference wing area, CD is the drag coefficient, and m is the aircraft mass.

T − q · Sref · CD − Fr
accel = (3.3)
W
As will be seen later during the implementation Chapter 5, the implementation of these aero-
dynamic models requires different computational approaches. While steady flight aerodynamics
necessitate a root-finding algorithm to determine the trim states, the ground acceleration model
requires convergence within the MDA. This complexity arises because the thrust produced is a
function of flight velocity, which in turn depends on the calculated acceleration, creating a coupled
system that must be solved iteratively.

3.1.1 Coefficient estimation

The force equations presented in Equations 3.1c and 3.3 assume that the lift coefficient CL , drag
coefficient CD , and pitching moment coefficient CM can be expressed as functions of angle of attack
α and control surface deflection δe . While these functions could be computed through full aircraft
1
Computed as half the square of the velocity times the density.
26 CHAPTER 3. MODELLING METHODOLOGY

CFD simulations and surrogate model training, such an approach would result in high-dimensional
models requiring extensive training data.
To address this challenge, a polynomial regression approach is implemented for lifting surfaces
(main wing and tails), combined with a drag buildup method for additional components (fuselage,
nacelles, etc.). The polynomial regression assumes that the lift coefficient varies linearly with angle of
attack, yielding terms CL0 and CLα . The drag coefficient is expanded as a second-order polynomial,
providing three regression terms: CD0 , CDα , and CDαα .
For the main wing, the coefficients are defined as:

CL(α) = CL0 + CLα · α (3.4a)

CD(α) = CD0 + CDα · α + CDαα · α2 (3.4b)

For moving lifting surfaces affected by downwash, the angle α must be substituted by the
effective incidence angle. The analysis assumes that the freestream angle of attack is reduced by a
downwash angle ϵ and that the lifting surface deflects by an angle δe . Following the same polynomial
expansion, the coefficients for these surfaces can be expressed as:

CL(α, δe , ϵ) = CL0 + CLα · (α − ϵ + δe ) (3.5a)

CD(α, δe , ϵ) = CD0 + CDα · (α − ϵ + δe ) + CDαα · (α − ϵ + δe )2 (3.5b)

The coupling between lifting surfaces (i.e., the estimation of the downwash angle ϵ) is modeled by
assuming that the front wing lift distribution is elliptical. This allows simplification of the flowfield
as the induced vertical velocity (and thus downwash angle) is assumed to be constant, with the
downwash being estimated as a linear regression of two terms, ϵ0 and ϵα :

ϵ(α) = ϵ0 + ϵα · α (3.6)

These terms are computed as functions of the front wing lift coefficients (CL0,w and CLα,w ) and
the wing aspect ratio ARw :

2 · CL0,w
ϵ0 = (3.7a)
π · ARw
2 · CLα,w
ϵα = (3.7b)
π · ARw

This approach allows the aircraft lift coefficient CL and drag coefficient CD to be expressed as:

CL = CLw (α) + CLHT P (α, δe , ϵ) (3.8)

CD = CD0 + CDw (α) + CDHT P (α, δe , ϵ) (3.9)

3.1. AERODYNAMIC MODELING 27

For the pitching coefficients, the moments produced by the wing and empennage are projected
onto the center of mass. Given that the pitching moment is referenced with respect to the aerody-
namic center, both the wing and empennage may have residual pitching moments that are aggregated
into a CM0 term. The total pitching coefficient is defined as the residual pitching moment plus the
pitch moment contributions of the wing CMw and tail CMHT P :

CM (δe , α) = CM0 + CMw (α) + CMHT P (δe , α) (3.10)

The pitching moment contributions of the wing and tail are defined as:

lxac,w→cdg
CMw (α) = (CL0,w + CLα,w · α) (3.11)
cw

lxac,HT P →cdg · SHT P

CMHT P (α) = (CL0,HT P + CLα,HT P · (α − ϵ(α) + δe )) (3.12)
cw · Sw

where lxac,w→cg and lxac,HT P →cg are the distances from the aerodynamic centers to the center
of gravity (see Figure 3.3), and cw is the main wing mean aerodynamic chord. The subscripts w and
HT P refer to the wing and horizontal tail plane, respectively.

Figure 3.3: Reference nomenclature for the pitching moments and the center of masses.

For the estimation of the additional coefficient terms OPALO currently implements vortex-derived
coefficient estimation according to DATCOM [145], with parasitic terms being computed using well-
known handbook formulas [146], [147]. These formulas include the contribution of the interference
factors, the friction factors (for wings, fuselages, tails, and nacelles). For the case of the flaps and
landing gear contributions, correction terms were generated from regression analysis based on [148].
The framework also allows more sophisticated approximations based on CFD solutions and
multiple-fidelity approaches; in this case, this is implemented for conventional cantilever and strut-
braced wings. The polar coefficients for these wings are obtained through surrogate models trained
with optimal wing designs. The training dataset was generated by performing aerodynamic shape
optimizations using VSPAERO [149]. Each optimization is performed at a fixed aspect ratio (AR)
using a parametric model composed of a main wing and supporting strut.
The main wing features a constant chord up to the mid-section (y = b/4), beyond which it
tapers at a constant taper ratio (λw ). The strut has a constant chord equal to 0.5 times the wing
28 CHAPTER 3. MODELLING METHODOLOGY

root chord. Design variables for these discipline-level optimizations include wing twists at the root
θw,y=0 , mid-section θw,y=b/4 , and tip section θw,y=b/2 , the strut twist at the root, and the outer wing
taper ratio λw . Additional details on the surrogate model are presented in [150].
Subsequently, the coefficients CL0 , CLα , CD0 , CDα , and CDαα from each optimized configuration
are fitted across AR for system-level optimization. This methodology emulates a bi-level approach,
with the inner discipline-level design variables remaining hidden from the system-level optimization,
which only manages the aspect ratio.
While not implemented in this project, considering higher fidelity models is pivotal for proper
sizing, as some interference factors (such as wing-propeller interaction or nacelle-strut interaction)
can act as enablers or disablers of a given technology.

3.1.2 Balance and static stability

For the calculation of the balance, the center of gravity is defined with respect to the aircraft’s nose
as the mass-weighted sum of component positions:

P
mi · xcdg,i + (mf uel − mf uel,burnt ) · xcdg,f uel
xcg = P (3.13)
mi + mf uel − mf uel,burnt

where mi is the mass of each component, xcg,i is the distance of its center of gravity from the
aircraft nose, and the fuel term accounts for fuel burn during the mission.
The static margin is computed using the neutral point, defined as the point where the pitching
moment is independent of the angle of attack (i.e., ∂CM /∂α = 0):

SHTP
(lxac,w→cdg + lcdg ) · CLα,w + Sw · (lxac,HT P →cdg + lcdg ) · CLα,HTP · (1 − ϵα )
xnp = (3.14)
CLα,w + SSHTP
w
· CLα,HTP · (1 − ϵα )

The static margin is then computed throughout the mission as:

xnp − lcdg
SM = (3.15)
cw

This analysis assumes a minimum static margin requirement of SM > 0.35 to ensure adequate
stability. The critical point where this static margin constrains the problem depends heavily on the
aircraft architecture. For aircraft with hydrogen tanks, the static margin will be most restrictive at
the start of the mission, as the aircraft tends to be nose-heavy as it burns fuel (the center of gravity
shifts forward). Conversely, for aircraft with kerosene in wetted wings, this static margin will be
slightly more constrained at the end of the mission, as the aircraft tends to be tail-heavy (the center
of gravity shifts rearward) as fuel is consumed.
3.2. POWERPLANT MODELLING 29

3.2 POWERPLANT MODELLING

The following section will define the different models used to then assemble the various powerplant
architectures described in Chapter 4.

3.2.1 Propeller model

For propeller performance modeling, OPALO employs a surrogate model based on propeller per-
formance maps generated using XROTOR [151]. XROTOR is a Fortran-based computational tool
capable of estimating propeller performance for a given geometry through multiple theoretical
formulations, including Graded Momentum Theory, Potential Flow Theory, and Vortex Theory
[152]–[154]. Additionally, XROTOR can generate optimized propeller geometries based on minimum
induced loss formulations.
For the initial propeller sizing, XROTOR requires specification of a design condition from which
the chord and twist distributions are optimized to generate a minimum induced losses (MIL) propeller.
In this study, an operating point representative of the ATR 72-600 at its maximum flight altitude
[155], [156] is selected as the design condition (Table 3.1).

Parameter Value Units

Propeller diameter 4.11 m
Propeller hub diameter 0.82 m
Number of blades 6 –
Cruise speed 130.2 m/s
Cruise altitude 7620 m
Rotational speed (82% NP) 984 RPM
Single propeller thrust 5.83 kN

Table 3.1: Propeller cruise operating conditions for the ATR72-600 at its max cruise altitude.

However, sizing under these conditions can often result in a MIL propeller with high twist
distribution and a very thin root chord. To enhance propeller feasibility2 , the propeller blade chord
is modified to remain constant from the previously maximum chord position to the propeller root.
This modification generates a propeller geometry similar to the ATR 72’s Hamilton Sundstrand
568F-1 [155] or the Dash 8-400 Dowty Aerospace Propellers Model R408/6-123-F/17 [157]. The
modified propeller is then re-optimized for twist distribution to obtain a MIL propeller. While this
method may lead to lower propeller efficiency η, it evaluates propeller performance using a more
structurally reasonable geometry.
The resulting propeller geometries can be observed in Figure 3.4, where the original chord and
twist optimized MIL propeller exhibits a reduced root chord, while the modified propeller features a
more reasonable straight section. This compromise is achieved at the expense of a slight reduction
in propeller propulsive efficiency η. To provide greater flexibility during the optimization process,
2
In terms of generating a propeller blade with sufficient structural integrity
30 CHAPTER 3. MODELLING METHODOLOGY

Figure 3.4: a) Propeller geometry and propeller J-CP map for a MIL twist and chord optimized
propeller; b) Propeller geometry and propeller J-CP map for the modified MIL twist optimized
propeller.

Figure 3.5: Modification of the blade chord distribution c/R along the radii and the blade twist
distribution β.
3.2. POWERPLANT MODELLING 31

this procedure is repeated with design disk loadings of 0.5 times and 1.5 times the original disk
loading. This approach allows the optimizer to select either a propeller with thin blades and high
efficiency or a propeller with wider blades and higher maximum operating power coefficients CP ,
albeit at the expense of lower efficiencies η. The resultant twist and chord distributions for the
un-modified and modified propellers can be seen in Fig.3.5, where it can be observed that after the
chord modification, the overall blade twist is decreased.
Regarding implementation details, the surrogate model is implemented as a regular grid in-
terpolator in the J, CP , and disk loading directions. Power and thrust forces are computed using
conventional non-dimensional coefficients. For power-input conditions (such as takeoff acceleration),
the input power is transformed into CP , and with the given advance ratio J, the efficiency η and
thrust coefficient CT are computed and transformed into thrust. For conditions where thrust is
the input parameter (as in steady flight aerodynamics) rather than power, a Newton root-finding
method is implemented to determine the required power that produces the specified thrust.

3.2.2 Gas Turbine Model

The gas turbine performance is characterized through a surrogate model derived from PyCycle [10]
simulations. PyCycle is a gas turbine performance estimation code based on a port of the Numerical
Propulsion System Simulation (NPSS) code [158] into OpenMDAO.

Figure 3.6: Thermodynamic properties for the operation of the Kerosene turboshaft gas turbine.

From all the available gas turbines, the following study will assume that the current turbine within
our architectures corresponds to the multi-spool turboshaft configuration (already implemented
within the Girhub repo of PyCyle). This configuration features a three-stage compression system
(low-pressure axial, high-pressure axial, and high-pressure centrifugal), combustion chamber, and
expansion through high-pressure, low-pressure, and power turbines. Bleed air extraction for turbine
cooling is included, with default bleed values maintained for this study.
32 CHAPTER 3. MODELLING METHODOLOGY

Figure 3.7: Thermodynamic properties for the operation of the LH2 turboshaft gas turbine.

Regarding implementation, the surrogate model is implemented as two complementary models.

The first model uses a Clough-Tocher 2D interpolator to provide the fraction of power that the
gas turbine can deliver at a given altitude and Mach number relative to its power output at static
sea-level conditions (gas turbine rating condition). The second surrogate model provides operating
variables (PSFC, T4t, p5, etc.) as functions of normalized power operating conditions and dimensional
flight conditions, where normalization occurs with respect to the maximum power available at each
altitude and Mach number combination.
This approach allows the operating condition surrogate models to be normalized between zero
and unity, where unity represents the maximum operating power achievable at a given condition.
This normalization is what we will refer to as the gas turbine throttle level δgt . This gas turbine
throttle can then be scaled with the first surrogate model to refer it to the maximum power that
would be generated at static sea level, which in turn allows us to scale back to dimensional power
using the gas turbine power rating.

Parameter Range/Value Unit

On-design Mach number 0.6 —
On-design altitude 7600 m
On-design power 2000 hp
Training altitude range 0–8000 m
Training Mach number range 0–0.7 —
Fuel types Kerosene, Hydrogen —

Table 3.2: PyCycle gas turbine surrogate model training data.

The surrogate model training parameters are summarized in Table 3.2. The model is trained
with an on-design condition of M = 0.6, an altitude of h = 7600 m, and a power rating of 2000 hp.
3.2. POWERPLANT MODELLING 33

Two models are implemented: one using the implemented kerosene model (Fig. 3.6) and one using
the implemented hydrogen model (Fig. 3.7). For dimensions, both gas turbines are sized with the
specific volume of a PW127 turboshaft engine [159], consisting of a core attached to a gearbox.

3.2.3 Batteries

For electric battery modeling, OPALO considers the battery first as a set of individual cells, then as a
complete pack. Cells are characterized by specific power and specific energy as top-level design
variables, with cell mass (Wbatt,cells ) treated as a design variable. These cells are then configured into
a pack by first determining the total number of cells in the pack (Eq. 3.16), followed by calculating
the number of cell rows arranged in parallel 3 .

Wbatt,cells
ncells =  2 (3.16)
π · hcell · Dcell
2 · ρcell

For this specific arrangement implemented, the cells are assumed to be distributed in a bandoleer
configuration (Fig.3.8 with staggered cells, each being in contact with the side of a cooling channel
of width tcool (treated as a design variable). Cell height (hcell ) and diameter (Dcell ) are treated as
top-level variables and are kept constant during the optimization.

Figure 3.8: Geometric description of the batteries module with TMS cooling.

Regarding reference values, Tab.3.3 shows some of the values for various high-performance
commercial cells available in the market, and their specific properties. While the specific energy
will be varied across the various studies that will be performed, the specific power is fixed in a value
of 1200 W/kg, with the density being fixed to a value of 2.55 kg/L. For the cell dimensions, a cell
diameter of 18mm with a total stack height of 0.6m is considered.
3
Here parallel refers to the physical distribution rather than electrical wiring, as the code assumes that the cell pack is
formed by a series of rows arranged serially in the spanwise direction and distributed in parallel in the fuselage direction.
34 CHAPTER 3. MODELLING METHODOLOGY

Cell Model Spec Ene (Wh/kg) Vol Ene (Wh/L) Spec Pow (W/kg) Density (kg/L)
Tesla 4680[160] 233 626 835 2.67
Samsung INR18650-30Q[161] 237 645 1650 2.72
Amprius SA03[162] 399 992 1130 2.48
A123 26700 NCA GL[163] 143 341 6090 2.38
Farasis P79B3[164] 323 829 1360 2.56
V4Drive V21700HEP-5.X(A0) [165] 261 767 2100 2.93

Table 3.3: Specifications of Commercial Battery Cells

For thermal management, the current model assumes that the heat loss of the battery is about
5% of its operating power. Regarding materials, the battery case is constructed from aluminum with
a cell-to-battery mass scaling factor of 0.78 [166], meaning 78% of the total pack mass consists of
active cell material and 22% accounts for structural components, thermal management hardware,
and electrical connections. For the heat capacity, the battery mass is assumed to have a heat capacity
of 897 J/kg/K, with the aluminum casing having a heat capacity of 900 J/kg/K. For the heat transfer
model, the model considers the eNTU [167] method with the channel walls as heat transfer surfaces
in a single4 channel arrangement.

3.2.4 Electric motors and inverters

For the electric motors, Tab.3.4 provides some reference values for current aviation state-of-the-
art machines. While technology evaluation trends point towards power densities in the order
of 16 kW/kg for 2030 [168], here a value of 12kW/kg and a density of 1.1 kg/L will be explored.
For volumetric considerations, the electric motor is assumed to scale with the gearbox assembly5 ,
providing an electric motor with a diameter of Dem and a length of lem .

Model Diam (mm) Len (mm) Cont Pow (kW) Weight (kg) Pow Dens (kW/kg) Dens (kg/L)
SP260D [169] 418 300 260 50 5.2 1.2145
magni 500 [170] 500 600 560 135 4.15 1.1459
ENGINeUS100B1 [171] 394 354 125 51.6 2.42 1.1955
H3X HPDM-350 [172] 560 226 350 50 7.0 0.8982
H3X HPDM-1500 [173] 560 380 1500 125 12.0 1.3356
H3X HPDM-2300 [174] 780 312 2300 224 10.27 1.5025
H3X HPDM-3000 [175] 914 304 2800 220 12.73 1.1030
Zero Avia 660 [176] 680 500 600 153 3.92 0.8426

Table 3.4: Specifications of commercial electric machines currently on the state of the art. Power
density is computed considering the maximum continuous power.

In terms of thermal management, the electric motor is assumed to have a 5% heat loss (as the
electric motor is assumed to have a constant6 efficiency ηem = 0.95 %) with the cooling being
performed by a cooling jacket (Fig.3.9) exchanging heat with the outer side area (π · Dem · lem ). For
the thermal mass, the model considers the thermal mass of the electric motor and assumes that from
4
The model considers that there are multiple channels along the length of the battery pack, however they all have the
same surface temperature and as such are assumed to all have similar wall temperatures.
5
For the case of the fuel cell, the electric motor is assumed to have the same diameter as the width of the fuel cell.
6
While electric motor efficiency maps can be implemented currently OPALO does not implement any tradeoff sizing
with torque or rpm, this leads the optimizations to be ill-defined by modifying the motor map beyond feasible designs.
3.2. POWERPLANT MODELLING 35

Figure 3.9: Conceptual description of the cooling jacket for the electric motor

it 30% of the mass is steel, 30% of the mass is aluminum, 20% is copper, and the remaining 20% is
magnets, providing an aggregated heat capacity of cpem = 528 J/kg/K. Regarding the specific case
cooling the current implementation implements the methods and coefficients estimated in Brelje’s
thesis [177].
For the electric motors to operate, inverters are required to convert direct current to alternating
current. These power electronic components are modeled as single electric units with an efficiency
of ηinv = 0.95, where the remaining power is dissipated as heat loss. Regarding inverter technology
levels, Tab.3.5 provides reference values for contemporary systems. Based on current aviation
applications, a volumetric density of 0.9 kg/L and a power density of 20 kW/kg are assumed for the
analysis.

Modelo Pow (kW) Mass (kg) Dims (mm) Pow. Dens. (kW/kg) Dens (kg/L)
Zero Avia 225 kW Inv [178] 200 10 275×305×44 20.00 0.8280
Zero Avia 400kW Inverter [179] 400 20 275×305×254 20.00 0.9388
Pipistrel E-811 power driver [180] 50 7 245×126×230 7.14 0.9859

Table 3.5: Specifications of inverters for state-of-art aviation electric motors.

Figure 3.10: Description of the cold plate cooling system for the electric inverter.

In terms of thermal management, the inverters utilize a multi-channel cold plate configuration
for heat dissipation. The inverter thermal mass is characterized by a constant heat capacity of
cpinv = 528 J/kg/K (same as the electric motor, due to the lack of specific information). While the
36 CHAPTER 3. MODELLING METHODOLOGY

cold plate itself is assumed to be massless for computational simplicity, the analysis accounts for the
mass contributions of the case, cooling fins, and coolant fluid. The design incorporates fin height
hhx,f in and width whx,f in as optimization variables, while maintaining fixed dimensions for the
case thickness thx,plate and fin thickness thx,f in at 1 mm and 0.1 mm, respectively. For the heat
transferred and the pressure loses, regressions based on the hydraulic diameter are utilized [181].

3.2.5 Fuel cells

For the fuel cell system, rather than directly solving the implicit electrochemical stack equations,
the current implementation employs a pre-computed efficiency curve as a function of the fuel cell
throttle δF C . The efficiency characteristics are derived from [182] (seen in Fig.3.11), where the
throttle represents the ratio of actual power (PF C ) output to maximum available power (PF C,max ):

PF C
δF C = (3.17)
PF C,max

The fuel cell efficiency is subsequently expressed as:

PF C,max · δF C
ηF C = (3.18)
ΛH2 · ṁH2

This formulation ensures that for a given throttle setting δF C , the fuel cell operates at efficiency
ηF C , which, combined with the lower heating value of hydrogen ΛH2 , determines the required
hydrogen mass flow rate ṁH2 .

Figure 3.11: Fuel cell efficiency as a function of throttle δF C

To account for altitude effects on system performance, an additional constraint is imposed

on the required air mass flow rate7 ṁair , where the demanded airflow (from the fuel cell) must
not exceed the maximum air that can be supplied to the fuel cell (from the diverted air from the
rammed air at the fuel cell nacelle). To compensate for pressure variations with altitude, an electric
turbocompressor with constant efficiency ηcomp = 0.76 and a gravimetric index of 1.03 kW/kg is
incorporated [183], with power consumption given by:
7
The analysis assumes constant oxygen fraction in air with an air-to-H2 stoichiometric ratio of 34.3 kg air/kg H2 .
3.2. POWERPLANT MODELLING 37

  κ−1 
P (h) κ
ṁair · cp,air · T (h) · PSL −1
Pcomp (h) = (3.19)
ηcomp
Regarding fuel cell technology specifications, Tab.3.6 presents reference values for current stack
systems. For this analysis, a specific power of 2 kW/kg and volumetric power density of 0.87 kW/L are
selected. While these values may appear optimistic compared to current automotive fuel cell systems
(such as Ballard Power Systems), they remain conservative relative to future target specifications,
with development programs targeting system-level values of approximately 3 kW/kg [184]. These
specifications are, however, still below the 4.27 kW/kg specific power and 1.74 kW/L volumetric
power density of the ATR72 turboprop engine8 .

Model Pow (kW) Mass (kg) Dims (mm) Pow. Dens. (kW/kg) Dens. (kW/L)
ZeroAvia super stack flex [185] 150 170.0 619×670×414 0.88 0.87
ZeroAvia super stack [186] 400 190.6 950×800×600 2.10 0.88
Ballard Power Fcmove XD [187] 120 238.0 895×736×550 0.50 0.36

Table 3.6: Specifications and density metrics for fuel cell stack modules.

Figure 3.12: Cooling system configuration for fuel cell stack thermal management, and unit cell for
the unit coolant cell.

In terms of thermal management, the fuel cell is modeled as a thermal mass with homogeneous
temperature distribution and a mean9 thermal capacity of cp,F C = 836 J/kg/K [188]. Heat generation
corresponds to the efficiency losses (1 − ηF C ) of the electrochemical process. For cooling, parallel
cold plates are positioned between fuel cell stacks (Fig. 3.12), with fixed geometric parameters
including case thickness thx,plate = 1 mm and fin thickness thx,f in = 0.1 mm. The design variables
8
For equivalent installed power, this necessitates either a larger nacelle or modification of the aircraft architecture
from a twin-engine to a quad nacelle-propeller configuration.
9
Considering the structure, the membranes and the electrolyte solution.
38 CHAPTER 3. MODELLING METHODOLOGY

comprise the number of parallel cold plates and the channel dimensions, specifically width whx,f in
and height hhx,f in . For estimation of the cooling capacity of the cooling plates, the model employs
once more the NTU method and estimates pressure losses using the hydraulic diameter approach.
The analysis assumes uniform cold plate design with equal coolant flow distribution among all
plates.

3.2.6 Heat exchangers

To dissipate the heat accumulated in the coolant fluid, fluid-to-fluid heat exchangers [189] are
implemented within the thermal management system. The current model employs an offset strip
heat exchanger configuration, where alternating "levels" of cold plates conducting ram air flow are
intercalated with hot plates through which the heated coolant is channeled (Fig.3.13). For ODE
stability, the heat exchanger casing is assumed to be mass-less, while the design variables include
the channel width whx,cold and height hhx,cold for the cold side, and channel width whx,hot and
height hhx,hot for the hot side, as well as the number of offset strips Nstrips for both cold and hot fin
arrangements. The geometric configuration maintains fixed case thickness thx,plate and fin thickness
thx,f in at 1 mm and 0.1 mm, respectively, for both cold and hot sides.

Figure 3.13: Heat exchanger model and fin unit definition: a single finned cold plate exchanger, with
options for offset fin strips on both plates.

The thermal analysis incorporates pressure losses on both the cold and hot sides to account
for ram air drag effects and cooling loop pressure losses, respectively. Heat transfer calculations
are performed using the Number of Transfer Units (NTU) method in accordance with compact
heat exchanger correlations from established references [181]. In terms of mass estimation, the
implementation tracks the mass contributions from the case and fin material (both fabricated from
aluminum) and the mass from the coolant volume.
In addition to the heat exchangers, the current implementation incorporates piping systems to
track pressure losses [190] along the coolant distribution network using standard correlations based
on the hydraulic diameter Dh and to account for both the mass of coolant within the pipes and the
pipe structural mass.
3.2. POWERPLANT MODELLING 39

3.2.7 Hydrogen tanks

For the hydrogen tank system, the following implementation considers the system of equations
introduced in [191], with the exception of bulk boiling and condensation phenomena (which in-
troduce additional instabilities that can compromise ODE solver stability). This model treats the
hydrogen tank as a multi-phase system consisting of a control volume of liquid hydrogen and a
control volume of gaseous hydrogen that exchange energy through an interphase where hydrogen
boil-off occurs (Fig.3.14).

Figure 3.14: Model for hydrogen boil-off. Rates in black are assumed constant during the ODE solve;
rates in red are the unknowns to solve.

In terms of geometry, the tank is assumed to be spherical until the maximum tank diameter
DLH2,max is reached, at which point the tank becomes cylindrical with hemispherical end caps.
Regarding material specifications, the tank employs cryogenic-capable aluminum construction with
vacuum-insulated insulation incorporating 10 MLI (Multi-Layer Insulation) layers. For operational
characteristics, it is assumed that all hydrogen consumed by the aircraft is in liquid form, while all
vented hydrogen is in gaseous form. A heater with negligible mass is incorporated to regulate tank
pressure, which is maintained at 2 bar above atmospheric pressure through optimization constraints
controlling venting and heating operations.
While real architectures can employ liquid hydrogen as a coolant10 , the present implementation
assumes that the tank thermodynamic cycle operates independently of the coolant loop. This
assumption is expected to yield a conservative estimate of the architecture’s peak performance, with
systems incorporating integrated cooling (e.g., through a helium or nitrogen cycle [192]) possibly
demonstrating reduced ram air requirements.
To improve the stability of the fuel tank implementation in the MDA, a pseudo-consistent ap-
proach is adopted. In this implementation, the hydrogen tank geometry is fixed at each optimization
iteration, with capacity managed through state-of-fill constraints. Variations in hydrogen mass are
tracked throughout the MDA to estimate center-of-gravity changes by integrating consumed liquid
hydrogen and vented hydrogen. Upon convergence of the mission MDA, an ODE solver computes
the actual temperatures, pressures, and hydrogen masses (both liquid and gaseous phases). This
10
This approach provides the benefit of conditioning the fuel to meet injection and reaction requirements.
40 CHAPTER 3. MODELLING METHODOLOGY

approach maintains total mass conservation in the control volume while avoiding computationally
expensive ODE evaluations under non-converged conditions. Consequently, venting and heating
are treated as design variables during mission optimization rather than control-loop operations,
with their effects evaluated subsequently through the ODE solver11 .

3.3 WEIGHT ESTIMATION

For aircraft structural mass estimation, OPALO employs the Raymer weight estimation method
[146] with corrections based on reference aircraft data. The wing, fuselage, tail, nacelle, and landing
gear mass predictions are calibrated using weight breakdowns from the ATR72 [193], ATR42, and
Dash 7 aircraft [194]. For the case of the wings, Raymer’s formula estimation has shown not to scale
properly12 with high AR values, therefore, the formula was corrected with the results from [195],
which also allows us to estimate the wing mass of a strut-braced wing, further details on the model
used can be seen in [150].
Secondary system masses are estimated through calibrated13 weight regressions as functions of
MTOW, while luggage and operating items are determined based on both MTOW and passenger
count, utilizing the aforementioned reference aircraft weight breakdowns. For passenger and luggage
modeling, the analysis assumes a passenger mass of 83.3 kg and a combined passenger-plus-luggage
mass of 104.86 kg per passenger.

Component Specific Power (kW/kg) Specific Volume (kW/L)

Turboshaft 3.60 2.365
Power addition gearbox 16.45 15.73
Inverter 20.00 18.00
Electric motor 12.00 13.20
Fuel cell 2.00 0.87

Table 3.7: Specific power and specific volume of various propulsion components

For powertrain mass estimation, the model employs component-specific power densities. For
the power addition gearbox, a specific power of 16.447 kW/kg [196] is considered. The turboshaft
engine assumes a specific power of 4.15 kW/kg based on the dry mass of the PW100 family [159].
To account for installation effects, this specific power is reduced by a factor of 0.87. Battery, inverter,
electric motor, and fuel cell specific power values are adopted from the specifications detailed in
the aforementioned sections. The specific powers and volumes are gathered in Tab.3.7. For the
estimation of the hydrogen tank mass, the model employs the weight estimation from [191].
To validate the weight estimation methods, a comparison against the ATR72-600 is presented in
Chapter 6. It is important to note that these estimations are calibrated specifically for aircraft within
the tube-and-wing regional turboprop design envelope, requiring further tuning for alternative
11
Venting and heating effects are incorporated in the mission performance analysis.
12
This is likely due to the formula regression only considering conventional aspect ratios in the order of AR ≈ 9.
13
With the values predicted by Raymer and the aforementioned aircraft weight breakdowns.
3.4. TRAJECTORY DEFINITION 41

aircraft concepts.
Regarding structural considerations, the rear hydrogen tank installation will increase structural
loads during pitch-up maneuvers [197]. While no specific correction factor is available for this
effect, the hydrogen tank is treated as a non-structural component. When integrated into the aircraft
structure, the additional load-bearing capacity would partially offset these increased loads, with the
mass of removed fuselage material redistributed to strengthen the fuselage-wing bulkhead fuselage
section.

3.4 TRAJECTORY DEFINITION

For trajectory generation, the current branch of OPALO supports trajectory creation through the
modification of control nodes. This approach is implemented for two types of mission segments:
altitude change segments (including climbs and descents) and horizontal distance segments (primarily
consisting of cruise segments).

Figure 3.15: Definition of mission segments by means of modifying the equivalent airspeed and
flight path angle. Markers at the left plots indicate control nodes handled by the optimizer.

For altitude change segments, the user defines an initial altitude (hstart ) and a final altitude (hend ),
along with the start, mid, and end control values for both the flight path angle (γ) and the horizontally
projected equivalent airspeed (EASproj ). The system then generates a vector of nn altitudes (where
nn represents the number of nodes for each phase) uniformly distributed between the start and end
42 CHAPTER 3. MODELLING METHODOLOGY

altitudes. The EASproj and γ values are interpolated using a second-order polynomial along these
equally spaced altitude vectors (see in Fig.3.15, left).
At this stage, the user has defined the altitudes and the horizontal and vertical velocities. The
tool then calculates, using flight kinematics14 , the time required to transition from one altitude
point to the next, which is subsequently used to compute the horizontal distance and acceleration.
With the altitude and speed values, the tool can also compute derived parameters such as the Mach
number and flight conditions (density, pressure, temperature) using the ISA atmosphere model (see
in Fig.3.15, right). This step is performed first for all the climb and descent segments where the
horizontal distance is an unknown but the speeds are prescribed.
Once the horizontal distance for the climbs and descents has been computed, the horizontal
distance segments are calculated. The approach is based on the fact that if the nominal mission has
a known horizontal distance, the cruise distance can be calculated by subtracting the computed
horizontal distance from the climb and descent phases15 . With this horizontal distance, a distance
array similar to the one previously created for altitude can be generated. Using initial, mid, and end
EAS values, a polynomial interpolated array is created once more. This allows calculation of the
time required to travel from one horizontal distance point to another, along with all derived values.
While this thesis assumes that these segments are flown at constant altitude, it is also possible to
prescribe initial and end altitudes. In this case, the vertical speed is computed by assuming that the
altitude increases linearly along the segment at each node position. A variation of this segment
called the holding segment is also implemented, where we fly at constant altitude and zero extra
distance flown, and we just fly for a given duration of time.

Figure 3.16: Example of 600nm nominal mission with a 300nm extended mission including a 30 min
holding segment.
14
For this, the tool assumes that each segment is defined by a start and an end position where we have linear constant
acceleration. For the vector of times, the tool assumes that the start position is t0 = 0 and that the time between two
points is appended to the end node. For the acceleration, the tool computes this value by fitting the trajectory with a
spline and computing its gradient.
15
At this stage, a blending function is imposed to avoid negative cruise distances, which would cause problems when
dealing with slow climbs and descents. While an additional constraint can be imposed on this initially calculated cruise
distance, applying a blending function already penalizes the mission by extending the flight duration, and is therefore
avoided by the optimizer.
3.5. AICRAFT CONTROL 43

By combining these segments, a nominal and extended mission profile can be constructed for
the aircraft to fly (Fig. 3.16). The objective is for the user to prescribe the nominal and extended
mission range and then optimize the altitude and the horizontal (EASproj ) and vertical (γ) speed
profiles to aid with the function minimization. Trajectory optimization then results not so much
in an optimization in itself, but rather in sizing the trajectory with the aircraft to avoid generating
artificial sizing constraints for the aircraft16 .

3.5 AICRAFT CONTROL

In addition to the trajectory states (EASproj and γ), the user schedules a set of control variables
along the mission (which can include the revolutions of the propeller, how much electric boost
the powertrain is having, or the pressure relief mechanisms in the hydrogen tank). Regarding the
parametrization we have a similar parametrization to the mission trajectory, where each control is
specified at the segment boundaries (start and end position), and the values at the internal nodes are
obtained by linear interpolation over the interior points of the segment phase array.

Variable Segments Interpolation Aircraft

EASproj All Quadratic All
γ Altitude change Quadratic All
Mtip multiply All Linear All
Hybridization All Linear PH, PH LH2, FC LH2
Ram door aperture All Linear PH, PH LH2, FC LH2
Battery heating All Linear PH, PH LH2, FC LH2
LH2 tank heating All Linear PH LH2, FC LH2
LH2 tank venting All Linear PH LH2, FC LH2
FC ram door aperture All Linear FC LH2
FC flow split All Linear FC LH2

Table 3.8: Trajectory optimization and control variables.

Table 3.8 represents the trajectory and operational control17 variables that define the trajectory
and operational profiles, as well as the segments and aircraft architectures to which they apply.
Control in this case does not refer to the response of the dynamical system, but to the predetermined
prescription of aircraft operation. More details on the control of aircraft systems will be given in
the following chapter. It is important to mention that all aircraft-related profiles are interpolated
linearly; however, given that the array separation across the phases is either constant with altitude
∆h (for altitude change segments) or with constant horizontal distance ∆x (for horizontal distance
segments), when represented against time, the control may appear to have parabolic profiles.

16
This is especially important when considering thermals, as the aircraft must trade heat generation from a faster climb
against being able to reach the colder air sooner.
17
 CHAPTER

AICRAFT ARCHITECTURES 4
Contents
4.1 Conventional Kerosene turboprop . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2 Kerosene Parallel Hybrid architecture . . . . . . . . . . . . . . . . . . . . . 47
4.3 Hydrogen Parallel Hybrid Architecture . . . . . . . . . . . . . . . . . . . . . 50
4.4 Hydrogen Fuel Cell architecture . . . . . . . . . . . . . . . . . . . . . . . . . 53

This chapter introduces the set of aircraft architectures considered in this work and outlines the
modeling assumptions applied to each configuration, providing the foundation for the optimization
studies presented in the subsequent chapters.

4.1 CONVENTIONAL KEROSENE TURBOPROP

Figure 4.1: Description of the Kerosene conventional architecture.

45
46 CHAPTER 4. AICRAFT ARCHITECTURES

The first architecture analyzed is the conventional kerosene turboprop (turboshaft) configuration,
illustrated in Fig. 4.1. This layout represents a state-of-practice regional aircraft comparable to
the ATR-72, which serves as the primary reference model. Two sub-architectures are considered:
one with a conventional wing and another with a strut-braced wing. In both cases, propulsion is
provided by a propeller driven by a turboshaft gas turbine through a reduction gearbox.
The cabin arrangement mirrors that of the ATR-72. At the front, the cockpit accommodates the
flight crew and connects to the passenger cabin via a corridor passing through the forward luggage
compartment. The main cabin provides seating for 72 passengers, distributed in 18 rows with a
four-abreast layout. Aft of the seating area, two lavatories (similar in configuration to those of an
A320) are followed by a rear galley, which provides space for cabin crew operations and storage of
catering and service items.
For the aerodynamic design, a high-wing configuration is adopted to ensure adequate propeller
ground clearance. The empennage follows a T-tail arrangement, with the horizontal stabilizer
mounted at 95% of the vertical tail height.

Figure 4.2: Description of the twin turboprop architecture in the trim mode and LTO mode.

The mathematical and programmatic implementation of the twin-turboprop powertrain is con-

ceptually illustrated in Fig.4.2. The system is represented through three interconnected components:
a propeller object, which contains a performance map describing operational characteristics as a
function of design parameters (see Sec.3.2.1); a gearbox object, which couples the power turbine
shaft to the propeller shaft; and a turboshaft object, which relies on surrogate models to describe
maximum power output and performance characteristics across the Mach–altitude mission envelope
(see Sec. 3.2.2).
The powertrain operates in two distinct modes: trim and LTO. In trim mode, the system adjusts
its operating point to meet the thrust demand. The required propeller thrust (Treq,prop ), together with
the current flight conditions (altitude, Mach number, airspeed, etc.), is passed to the propeller object.
Since the propeller map is expressed in terms of the power coefficient (CP ) and the advance ratio (J),
the power needed to generate the required thrust is computed using a Newton root-finding algorithm
along the map, effectively analogous to adjusting the propeller blade pitch. This power requirement
4.2. KEROSENE PARALLEL HYBRID ARCHITECTURE 47

is then passed through the gearbox, where an efficiency loss of ηgbx = 0.98 is applied. At this
stage, the required input power (Pin,gbx ) and the gearbox throttle setting (δgbx ) can be computed by
comparing the transmitted power against the gearbox power rating, PRgbx :

Pout,gbx Pin,gbx
Pin,gbx = δgbx = (4.1)
ηgbx P Rgbx
This input gearbox power must then be delivered by the gas turbine, such that Pin,gbx = Pout,gt .
The throttle setting of the turboshaft can therefore be computed using the turboshaft power rating
P Rgt and the operational map Ξ(h, M ) as:

Pout,gt · Ξ(h, M )
δgt = (4.2)
P Rgt
Here, the operational map represents the ratio between the available power at the given flight
condition (altitude and Mach number) and the power at sea-level static conditions. This throttle
level is also used to evaluate the power-specific fuel consumption (P SF C), which in turn provides
the instantaneous fuel flow rate.
In the second operating mode, LTO, the gearbox throttle is prescribed: δgbx = 0.9 in the all-
engines-operating (AEO) case, and δgbx = 1.0 in the one-engine-inoperative (OEI) case. In this mode,
the flow of information is reversed compared to trim: the imposed gearbox output power (Pout,gbx )
defines the propeller load, and the corresponding input power required from the turboshaft (Pin,gbx )
follows directly, both quantities being calculated from Eq.4.1. As illustrated in Fig.4.2, the distinction
between the two modes lies not in the physical transmission of power (represented by continuous
red lines), but in the direction of the information flow used to compute the component operating
points (represented by dashed red lines).
For kerosene-based architectures (both the conventional and the hybrid), the maximum fuel tank
volume is estimated between the front spar (0.2c) and the rear spar (0.7c), using a GAW2 airfoil with
a 0.18 root chord, constant up to the wing kink. The takeoff field length is set to 1600m, providing
sufficient margin for the heavier hybrid architectures compared to the ATR72 or Dash8 reference
aircraft.1

4.2 KEROSENE PARALLEL HYBRID ARCHITECTURE

For the parallel-hybrid configuration (Fig.4.3), the conventional kerosene powertrain is augmented
with an electric motor, one integrated into each of the original turboshaft nacelles. In this setup,
hybridization2 is assumed to occur at the gearbox, where the shaft power from the power turbine
(PT)3 is combined with the power delivered by the electric motor (see Fig.4.4).
1
An argument can be made that next-generation turboprop architectures should retain the capability to operate from
shorter regional airfields, in line with the traditional niche of turboprops. However, in the near-term transition period, it
is more realistic to assume that such aircraft will initially operate from larger airports, where charging and, more critically,
hydrogen distribution infrastructure can be made available.
2
Defined as the proportion of mechanical power supplied by the combustion engine versus electrical power from the
battery.
3
The turbine shaft connected to the power turbine.
48 CHAPTER 4. AICRAFT ARCHITECTURES

To describe this coupling, a hybridization factor (HF ) is introduced, defining the fraction of
gearbox input power that is provided by the electrical system. Accordingly, the power contributions
from the electric motor and gas turbine can be expressed as

HF · Pin,gbx (1 − HF ) · Pin,gbx
Pout,EM = , Pout,gt = , (4.3)
ηgbx ηgbx
with the total gearbox output power equal to the propeller input power, i.e., Pout,gbx = Pin,prop ,
and the gearbox is assumed to have a set of efficiency losses such that Pout,gbx = Pout,gbx /ηgbx .

Figure 4.3: Description of the Kerosene Parallel Hybrid architecture. Cabin assembly remains as in
the conventional Kerosene.

This formulation allows the system to draw power either from the electrical lines or from the
gas turbine system4 . On the electrical side, each motor is connected to an inverter that performs DC-
to-AC conversion. Both inverters are supplied by a common electrical bus, which in turn interfaces
with the batteries5 . The electrical network also powers an electric coolant pump, which compensates
for the pressure losses in the thermal loop.
Accordingly, the total power demanded from the battery is given by

Pbatt = 2 · Pin,inv + Ppump , (4.4)

where Pin,inv is the input power of each inverter and Ppump is the coolant pump requirement.
This power demand is then supplied by the battery, and the corresponding rate of discharge is
calculated based on the battery specific energy (Espec ) and the installed battery cell mass (mcell,bat ).
4
The powertrain draws its full output from the electric system when HF → 1, and entirely from the gas turbine
when HF → 0.
5
While modeled as a single battery pack in this conceptual framework, in practice this would correspond to two
independent, thermally insulated packs.
4.2. KEROSENE PARALLEL HYBRID ARCHITECTURE 49

Figure 4.4: Description of the Kerosene Parallel Hybrid architecture. Cabin assembly remains as in
the conventional Kerosene.

The state of charge (SOC) evolution is obtained through

t
Pbatt (τ )
Z
SOC(t) = SOC(t0 ) − , dτ, (4.5)
t0 Espec · mcell,bat · ηbatt
where ηbatt represents the internal efficiency of the battery. This time integration is performed
segment by segment, allowing each mission phase to be treated as an independent MDA group
while maintaining continuity in SOC tracking across the mission profile.
For the thermal management system, a closed-loop liquid cooling circuit is employed to maintain
all power electronics within their thermal limits. The majority of the cooling is provided by a
fluid-to-air heat exchanger, which rejects heat from the coolant to the ambient flow. Ram air is
ingested through a dedicated intake duct (visible in Fig. 4.3), guided across the cooling fins, and
exhausted back to the atmosphere. The aerodynamic penalty of this installation is captured by
modeling the pressure losses across the heat exchanger core, together with the additional losses
introduced by ducting.
The coolant loop begins at a reservoir, from which coolant is pumped toward the inverter—cooled
through a top-mounted cold plate—and the electric motor, which is equipped with a cooling jacket.
After collecting heat from these components, the warmed coolant flows into the fluid-to-air heat
exchanger, where it transfers thermal energy to the ram-air stream. Downstream of the heat
exchanger, the coolant is directed through dedicated battery cooling channels before returning to
the reservoir, closing the cycle.
The cooling system is parametrically defined to allow for trade studies and optimization. Key
variables include the coolant mass flow rate, the internal diameter of the cooling pipes, and the
50 CHAPTER 4. AICRAFT ARCHITECTURES

cold-plate geometry (height and number of parallel fins). Battery cooling is modeled via channel
width variations, while the heat exchanger is represented as a cross-flow configuration in which the
number of layers (“floors”), as well as the channel height and width on both the hot and cold sides,
can be scaled6 . The maximum allowable inlet ram area is fixed at 1.5m2 for the fuselage ram doors.
For the thermal dynamics, the model assumes that the inverter, electric motor, battery, reser-
voir, and coolant pipes each possess a finite thermal mass. As a result, their temperatures evolve
non-instantaneously, following time derivatives dTem /dt, dTinv /dt, dTbatt /dt, dTreservoir /dt, and
dTpipe /dt. Collectively, the system can be expressed in state-space form with a vector of temperatures
and their time derivatives:

dTem
   
Tem (t) dt
dTinv
 Tinv (t) 
   
 dt 
dTbatt
   

T(t) = 
 T batt (t) 
 Ṫ(t) = 

dt
 = f T(t), u(t), θ
 (4.6)
 dTreservoir 
Treservoir (t)
 
 dt 
dTpipe
Tpipe (t) dt

This can be directly integrated by an ODE solver, giving a formal per-component solution as:
Z t

T(t) = T(t0 ) + f T(τ ), u(τ ), θ dτ (4.7)
t0

Here, u(t) represents the time-dependent external inputs, such as component heat generation
rates (qEM (t), qinv (t), qbatt (t)), coolant mass flow rate (ṁcool(t)), and ram-air inlet conditions (mass
flow and temperature, T inram (t)). The parameter vector θ contains the fixed physical and geometric
properties of the system, including thermal masses mi , specific heats cp,i , heat exchanger area AHX ,
material properties, efficiencies, and other constants.
A key challenge in this formulation is that, depending on the ram-air fraction and operating
conditions, the thermal ODE system can become stiff. This stiffness may degrade Jacobian accuracy
or even cause solver divergence. To mitigate this, a sigmoid limiter is applied to the heat-generation
rates, which prevents unrealistically large, nonphysical temperature derivatives while preserving
smooth gradient information. Furthermore, to avoid convergence issues in the coupled mission
MDA chain, the thermal ODE solver is decoupled from the mission loop. This prevents spurious
inputs (such as nonphysical or even negative heat rates arising during early iterations of the mission
convergence) from destabilizing the thermal solution.

4.3 HYDROGEN PARALLEL HYBRID ARCHITECTURE

Expanding upon the previous models, Fig.4.5 illustrates the system layout of the hydrogen parallel-
hybrid aircraft. The configuration builds directly on the kerosene-based parallel hybrid of Fig.4.3,

6
These dimensions are prescribed as multipliers of a reference baseline channel size.
4.3. HYDROGEN PARALLEL HYBRID ARCHITECTURE 51

but is modified to accommodate a single LH2 tank positioned aft of the rear galley7 . Although the
volume required for the hydrogen distribution system is not explicitly computed, dedicated space is
reserved below the galley and lavatories to allow for its installation.

Figure 4.5: Schematic of the hydrogen parallel-hybrid architecture. The cabin layout remains
identical to that of the conventional kerosene reference.

The powertrain and thermal management system are represented in Fig.4.6. As shown, the
operation of the hybrid-electric drivetrain follows the same principles described earlier for Fig.4.4,
with the key addition of the electrical load required by the hydrogen tank heater. Accordingly, the
total power drawn from the battery becomes:

Pbatt = 2 · Pin,inv + Ppump + Pheater (4.8)

This power demand is then tracked through the state-of-charge (SOC) integration process already
defined in Eq. 4.5.
On the thermal side, the loop shown in Fig. 4.6 (left) maintains the same cooling architecture
as in the kerosene-based parallel hybrid. As discussed in previous sections, the hydrogen system
is kept thermally independent from the powertrain cooling loop. While this assumption removes
the potential benefits of using hydrogen as a heat sink in, for example, a nitrogen Brayton cycle, it
provides a more conservative and safety-oriented estimate of system performance. Importantly, this
separation ensures that the aircraft retains full electric-operability even in the event of a complete
hydrogen tank venting scenario8 .
7
For larger aircraft, multiple hydrogen tanks could be advantageous, since in the event of a tank failure one could be
vented without losing the entire onboard hydrogen supply.
8
In such a case, while not explicitly modeled here, it is assumed that the aircraft could incorporate a venting system
52 CHAPTER 4. AICRAFT ARCHITECTURES

Figure 4.6: Description of the Hydrogen Parallel Hybrid architecture. Cabin assembly remains as in
the conventional Kerosene.

Focusing on the integration of the hydrogen tank into the system, the sizing logic follows a
process analogous to that of the batteries. At each MDO iteration, the optimizer selects a tank
capacity, which, based on the prescribed geometry, is used to estimate the structural mass of the
hydrogen tank. This mass is then fixed throughout the mission. Consistency between the design
variable (capacity) and the mission simulation is enforced by integrating the sum of the consumed
(ṁcons) and vented hydrogen (ṁvent) to track the tank’s state of fill SOF (Eq. 4.9). A capacity
constraint is then imposed such that 0 < SOF < 1 is always satisfied.

t
1
Z
SOF(t) = SOF(t0 ) − (ṁcons (τ ) + ṁvent (τ )) dτ (4.9)
mtank t0

Safe tank operation also requires ensuring that the internal pressure remains within acceptable
limits. To capture this, a nested ODE solver is employed to compute the thermodynamic evolution
of the liquid and gaseous hydrogen phases (temperatures and pressures). These updated states are
then used to recalculate the heat fluxes within the tank9 , following the methodology described in
[191]. At each MDO iteration, the optimizer specifies both a venting and a heating profile along the
mission timeline. The mission MDA computes the rate of liquid hydrogen consumption, and once

similar to that of the Tu-155 prototype or the latest Airbus Zero-E concepts.
9
This routine is algorithmically differentiated in reverse mode with a fixed number of outer iteration steps for
convergence.
4.4. HYDROGEN FUEL CELL ARCHITECTURE 53

convergence is reached, the ODE solver provides the detailed state evolution: liquid and gaseous
hydrogen masses, tank temperatures, and pressures. The latter two are also enforced as constraints
to prevent unphysical solutions, such as negative fluid mass.
As with the thermal management system, the nested ODE can become stiff depending on the
system inputs, particularly for extreme venting or heating profiles. To mitigate this, smoothing
and blending functions are applied both to the temperature derivatives and to the effective heat-
exchange areas whenever SOF becomes negative. This guarantees that the model always yields
valid outputs with usable Jacobian information, even for infeasible intermediate states. Importantly,
these adjustments do not affect the physical consistency of the optimization, as the SOF tracking
constraint within the MDA already ensures that the optimizer converges toward feasible operating
conditions (SOF > 0).

4.4 HYDROGEN FUEL CELL ARCHITECTURE

Figure 4.7: Description of the Hydrogen Parallel Hybrid architecture. Cabin assembly remains as in
the conventional Kerosene.

Finally, examining the fuel cell architecture, Fig. 4.7 shows the geometry of the selected config-
uration. Owing to the larger volume required for fuel cell installation, the powertrain transitions
from the twin-nacelle arrangement of the previous architectures to a quad-fuel-cell layout. In this
design, each nacelle houses a fuel cell together with its electric motor, inverter, dedicated cooling
system, and an additional heat exchanger.
The role of this extra heat exchanger is to pre-cool the coolant before it enters the fuel cell,
thereby increasing the temperature differential ∆T between the coolant and the fuel cell stack.
54 CHAPTER 4. AICRAFT ARCHITECTURES

Figure 4.8: Architecture description of the electric system of the fuel cell aircraft (top) and its
corresponding thermal management system (bottom).
4.4. HYDROGEN FUEL CELL ARCHITECTURE 55

This larger temperature gradient improves the rate of heat transfer and thus enhances cooling
performance10 . The corresponding electrical and cooling system layouts are depicted in Fig. 4.8.
With regard to aircraft control, hybridization is managed at the nacelle level, directly upstream
of each inverter. It is assumed that an electronic switch enables a junction where the inverter’s
demanded power can be supplied either by the fuel cell or via a connection to the main electrical bus.
As in previous architectures, this distribution is governed by a linearly segment-varying hybridization
factor HF , which determines how the inverter demand is split. Accordingly, the power provided by
the fuel cell (Pout,FC ) and the power drawn from the bus (Pout,BUS ) can be expressed as:

Pout,EM · HF Pout,EM · (1 − HF )
PBU SS,out = Pout,F C = (4.10)
ηEM · ηinv ηEM · ηinv
Where the electric lines already consider the efficiencies of the electric motor (ηEM ) and inverter
(ηinv ). This way the power demanded by the battery can be expressed as:

Pbatt = 4 · PBU SS,out + Ppump + Pheater (4.11)

Considering the thermal management system, the ODE formulation from Eq. 4.6 is extended
to include the fuel cell temperature as an additional state variable. The main difference relative to
previous architectures arises from the implementation of the fuel-to-fuel heat exchanger within
the nacelle. In this case, the ram air is no longer directed entirely toward or away from the heat
exchanger, but must instead be split between the fuel cell and the heat exchanger. This introduces
a system with two separate ramps11 . Constraints are imposed both on the loss coefficients across
the heat exchanger and on the margin between the mass flow required by the fuel cell and the
portion diverted to the exchanger. In addition, the inclusion of the fuel cell expands the cooling
system model to account for its own dedicated thermal management. This is represented as a set
of n parallel cold plates, for which the optimizer can vary both the plate height and the number
of parallel channels. For the fuel cell heat exchanger, the same design variables from the fuselage
heat exchanger are implemented, for the maximum ram inclet area it is considered as a duct with an
elliptical cross area with the width and height of the heat exchanger.
For the hydrogen tank, the same controls and assumptions described earlier are retained, with
constraints imposed on tank pressure, SOF , and hydrogen mass at the end of the mission. Addi-
tionally, for all hydrogen and kerosene-based powertrain architectures (fuel-cell hybrid, hydrogen
parallel hybrid, kerosene parallel hybrid, and conventional kerosene), variations in the aircraft center
of mass are considered. This inclusion ensures that the shifting mass distributions associated with
hydrogen storage, battery installation, and fuel burn are consistently captured, thereby contributing
to the overall design trends of the studied architectures.

10
It is worth noting that thermal losses from fuel cells are approximately an order of magnitude greater than those
from other powertrain components, rising from about 5% to nearly 60%.
11
Each ramp is again controlled by linear interpolation across the mission segment nodes, with the split defined as a
fraction of the total mass flow in each direction.
 CHAPTER

AIRCRAFT
MULTIDISCIPLINARY
ANALYSIS 5
Contents
5.1 A simplified introduction to multidisciplinary analysis . . . . . . . . . . . 57
5.1.1 Gradient computation strategies . . . . . . . . . . . . . . . . . . . . . . . 58
5.1.2 Jacobians of a coupled system . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.2 Aircraft multidisciplinary analysis within OPALO . . . . . . . . . . . . . . 61
5.2.1 Pre-processor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.2.2 Steady flight multidisciplinary analysis . . . . . . . . . . . . . . . . . . . . 62
5.2.3 Takeoff multidisciplinary analysis . . . . . . . . . . . . . . . . . . . . . . . 63
5.2.4 Overall Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

The following chapter will present a quick overview of multidisciplinary analysis and opti-
mization. The objective will be to answer two main questions: why these methods are necessary
for aircraft design, and how they are realized in practice. This will be done by first showing the
theoretical motivation for multidisciplinary approaches, and then examining how these concepts
are implemented within the OPALO system. For detailed mathematical formulations, the reader is
directed to [198].

5.1 A SIMPLIFIED INTRODUCTION TO MULTIDISCIPLINARY ANALYSIS

Starting up front, the objective of Multidisciplinary Design Analysis and Optimization (MDAO) is to
systematically find the best design that simultaneously considers multiple interacting disciplines.
This is done by means of numerical optimization.
The numerical optimization is the process that seeks to determine the optimal values of a set of
design variables that minimize (or maximize) an objective function, while satisfying a set of (equality
or inequality) constraints. This can be formally stated as the following optimization problem:

57
58 CHAPTER 5. AIRCRAFT MULTIDISCIPLINARY ANALYSIS

min f (x)
x

subject to gi (x) ≤ 0, i = 1, 2, . . . , m
(5.1)
hj (x) = 0, j = 1, 2, . . . , p
xL ≤ x ≤ xU
where x ∈ Rn represents the vector of design variables (e.g., wing aspect ratio, structural thick-
ness, engine bypass ratio), f (x) is the objective function to be minimized (e.g., aircraft weight, fuel
consumption, lifecycle cost), gi (x) are the inequality constraints (e.g., stress limitations, performance
requirements), hj (x) are the equality constraints (e.g., equilibrium conditions), and xL and xU define
the lower and upper bounds on the design variables, respectively.
The solution of this optimization problem requires the systematic variation of design variables
x through the use of numerical optimization algorithms. These algorithms can be broadly classified
into two categories based on their information requirements:
Gradient-based algorithms, that exploit derivative information (how the objective and con-
straints vary with the design variables) to guide the search toward optimal solutions. These methods
require the computation of gradients ∇f = ∂f /∂x and constraint ∇g = ∂g/∂x Jacobians . No-
table examples include Sequential Quadratic Programming (SQP) [199], Interior Point Optimization
(IPOPT) [200], and the Method of Moving Asymptotes (MMA) [201]. The benefit of these algorithms
is that they typically exhibit better local minima (lower values) with overall fewer iterations for
smooth, well-behaved problems (with continuous analytical derivatives being the SOA for optimal
performance).
Gradient-free algorithms, that operate without derivative information, relying instead on
function evaluations of f (x) and g(x). This category encompasses evolutionary algorithms (such
as genetic algorithms and differential evolution), direct search methods (like COBYLA [202]), and
population-based metaheuristics (including particle swarm optimization and genetic algorithms).
While generally more robust to discontinuities, noise, and local minima, these methods typically
require significantly more function evaluations to achieve convergence, and scale worse as the
number of design variables and constraint complexity increases (typical in OAD problems).

5.1.1 Gradient computation strategies

The effectiveness of gradient-based optimization is fundamentally dependent on the quality and
computational cost of gradient information. Several approaches exist for obtaining the required
derivatives:
Analytical derivatives represent the most accurate and computationally efficient approach
when feasible. These are user-defined expressions for ∇f and ∇g derived through mathematical
differentiation of the governing equations (usually performed using symbolic mathematic packages).
However, analytical derivatives can become increasingly hard to derive once the complexity increases
or solve methods are introduced.
Finite difference approximations provide a practical alternative when analytical derivatives
are unavailable (as, for example, running external codes) or prohibitively complex. The gradient
5.1. A SIMPLIFIED INTRODUCTION TO MULTIDISCIPLINARY ANALYSIS 59

components can be approximated using forward differences:

∂f f (x + hei ) − f (x)
≈ (5.2)
∂xi h

where h is the finite difference step size and ei is the i-th unit vector. While straightforward to
implement, this approach requires n + 1 function evaluations for an n-dimensional design space,
significantly increasing computational cost. Moreover, the accuracy is sensitive to the choice of step
size h, which must balance truncation error (too large h) against round-off error (too small h). For
codes that support complex arithmetic, complex-step differentiation can eliminate round-off errors
by using an imaginary perturbation step.
If the source code is available and consists of "basic" mathematical operations, Algorithmic
Differentiation (AD) offers a numerical method to differentiate the code in order to obtain the exact
derivatives. This is done by representing the set of mathematical operations as a graph and using
either forward (from inputs to outputs, where the number of outputs is grater than the number of
inputs) or backward (from outputs to inputs, when the number of inputs is grater than the number
of outputs) to obtain the derivatives. While we have said "basic mathematical operations," current
AD packages such as JAX [203] include additional libraries that include differentiated solvers such
as Newton method in the package Diffrax [204] and ODE solvers in the package Optimistix
[205], enabling for differentiation of complex pipelines.

5.1.2 Jacobians of a coupled system

The gradient computation strategies discussed above apply generally to individual disciplines,
regardless of whether they operate in isolation or as part of a larger system. However, when
multiple disciplines are present in a multidisciplinary system, their individual Jacobians must be
assembled into system-level Jacobians that capture the complete sensitivity information of the
coupled problem. In aeronautical engineering, these disciplines are typically implicitly coupled
through interdisciplinary data exchange, where the output of one discipline serves as input to
another, creating feedback loops that must be resolved through iterative solution procedures. This
coupling mechanism fundamentally distinguishes multidisciplinary systems from collections of
independent explicit analyses.
Consider a coupled system comprising disciplines A and B, where discipline A requires state
variables from discipline B (y2 ) as inputs, while discipline B simultaneously depends on state variables
from discipline A (y1 ), as illustrated in Figure 5.1. This bidirectional dependency creates an implicit
coupling that cannot be resolved through simple sequential execution—instead, both disciplines
must be solved simultaneously until convergence is achieved.
Once convergence is achieved for the implicitly coupled system, the individual discipline Jaco-
bians must be assembled into system-level total derivatives that properly account for interdisciplinary
coupling effects. This assembly process requires solving either the Global Sensitivity Equations
(GSE) [206] or the Multidisciplinary Analysis Under Uncertainty Direct (MAUD) method [207], both
of which yield identical system Jacobians that capture the complete interdisciplinary dependence.
The resulting system Jacobians reflect not only the direct sensitivities within each discipline but also
60 CHAPTER 5. AIRCRAFT MULTIDISCIPLINARY ANALYSIS

Figure 5.1: Implicitly coupled system between disciplines A and B, where discipline B depends on
outputs from discipline A, and discipline A depends on outputs from discipline B.

the indirect effects arising from the interdisciplinary dependencies. This approach, where disciplines
are fully converged before derivative computation, corresponds to the Multidisciplinary Feasible
(MDF) architecture, which ensures that the coupled system satisfies all interdisciplinary constraints
at each optimization iteration.
An alternative to solving the MDA at each iteration is the Individual Discipline Feasible (IDF)
architecture, where the coupling variables are treated as additional design variables in the optimiza-
tion problem, eliminating the need for iterative convergence at each optimization step. The IDF
formulation modifies the standard optimization problem as follows:

min f (x, y)
x,y

subject to gi (x, y) ≤ 0, i = 1, 2, . . . , m
hj (x, y) = 0, j = 1, 2, . . . , p
(5.3)
ck (x, y) = 0, k = 1, 2, . . . , q
xL ≤ x ≤ xU
yL ≤ y ≤ yU
where x ∈ Rn represents the vector of design variables (e.g., wing aspect ratio, structural
thickness, engine bypass ratio), y ∈ Rs represents the interdisciplinary coupling variables (e.g.,
loads transferred between structural and aerodynamic analyses), f (x, y) is the objective function,
gi (x, y) and hj (x, y) are the original inequality and equality constraints, and ck (x, y) = 0 are the
interdisciplinary consistency constraints that enforce coupling satisfaction. In this formulation, each
discipline can be evaluated independently using its portion of the design and coupling variables, while
the optimizer ensures interdisciplinary consistency through the coupling constraints ck (x, y) = 0,
which require that the coupling variables used as inputs to each discipline match the outputs
computed by the coupled disciplines. While this formulation can potentially lead the optimization to
explore and traverse faster the design space, the success or not of the IDF is really problem-specific
[208], [209].
A hybrid approach combining MDF and IDF principles can be particularly effective for systems
containing couplings with varying numerical characteristics. In such implementations, strongly
coupled subsystems that converge reliably are treated using MDF, while problematic couplings
that may cause convergence difficulties or numerical stiffness are handled through IDF constraints.
5.2. AIRCRAFT MULTIDISCIPLINARY ANALYSIS WITHIN OPALO 61

This selective approach is especially valuable when dealing with unstable feedback loops, such
as aerodynamic-structural interactions where drag-dependent variables create positive feedback
mechanisms that can destabilize traditional iterative solution procedures.
Regardless of whether an MDF, IDF, or mixed architecture is employed, a key advantage of
MDAO frameworks is that each discipline is implemented as an independent class within the
system. This modularity allows the optimizer to evaluate only the disciplines needed for specific
Jacobian entries at each iteration, rather than recomputing the entire coupled system. As a result,
computational efficiency is improved and flexibility is enhanced, since discipline classes can be
reused or interchanged across different optimization problems and platforms.

5.2 AIRCRAFT MULTIDISCIPLINARY ANALYSIS WITHIN OPALO

Looking now at our specific implementation for aircraft design, the primary objective of the multi-
disciplinary analysis framework developed in OPALO is to determine the aircraft flight performance
characteristics across all mission points. These include, but are not limited to, component operating
limits, stability margins, and aerodynamic performance. Additionally, the framework computes
integrated mission-level quantities, such as total fuel consumption and battery state-of-charge
evolution, while accounting for component integration aspects, including geometric clearances
between subsystems and the impact of components with larger volumetric requirements.
To achieve this objective, OPALO decomposes the analysis process into three sequential phases,
each with distinct computational characteristics and coupling requirements. The preliminary analy-
sis phase operates without implicit couplings, establishing the mission trajectory and computing
fundamental aircraft geometry and mass properties from the design variables. The subsequent
analysis phase constitutes the core multidisciplinary computation, where implicit aircraft perfor-
mance characteristics are resolved through iterative coupling using a Gauss-Seidel MDF coupling.
Finally, the post-processing phase computes additional derived quantities that do not require iterative
solution, evaluates objective functions, and assesses constraint satisfaction for the optimization
framework.

Figure 5.2: Modular decomposition of OPALO into distinct computational phases.

This modular decomposition, illustrated in Figure 5.2, enables computational efficiency by isolat-
ing the computationally intensive implicit coupling solving to the analysis phase, while maintaining
62 CHAPTER 5. AIRCRAFT MULTIDISCIPLINARY ANALYSIS

the rest of the trajectory and geometry management classes as independent disciplines that feed
and are fed by the mission performance. From an implementation perspective, the assembly and
orchestration of the complete system is managed by the GEMSEO framework, where each module is
composed of Discipline classes that communicate through a structured namespace system. This
object-oriented architecture facilitates seamless data exchange between disciplines while enabling
code reusability, allowing individual discipline instances to be reconfigured for different mission
phases through attribute modification rather than code duplication. Additionally, surrogate models
are instantiated during the initialization phase to maintain trained models in memory, thereby
eliminating data loading overhead during execution and ensuring consistent performance across
multiple evaluations within the optimization process. The integration workflow begins with loading
all required surrogate models, followed by instantiation of Discipline objects with their respective
model attributes and namespace connections. During this initialization phase, JAX-enabled disci-
plines are compiled for enhanced computational performance. Subsequently, the Design space is
generated with all design variables, constraints are added to the optimization framework, and the
system is executed. Upon completion, all mission state variables, design variables, and performance
metrics are systematically stored in a database for further analysis and optimization iterations.

5.2.1 Pre-processor
This initial phase of the analysis first calculates the overall geometric properties of the aircraft based
in the indicated parametric description1 of the architecture, at this stage all powerplant components
are also sized given their current power rating and their specific power and volumes. With this, the
rest of the aircraft component masses are evaluated using the methods explained in Sec.3.3, and for
disciplines that require of initial states (such as the hydrogen tank) the initial states are generated2 .
At this stage, the mission is also generated and its variables (M , q, T∞ , ρ∞ ..) are also computed
along with the total mission length and time in order to calculate the mean velocity (if mean velocity
constraints are to be imposed).

5.2.2 Steady flight multidisciplinary analysis

As indicated before, many problems in engineering present an implicit behavior; this is the case of
the aircraft performance estimation, as the fuel consumed (the main metric that we are interested in)
depends on the drag force, which depends on the aircraft mass, which depends on the fuel consumed.
To resolve this implicit coupling, each mission segment is formulated as a multidisciplinary
analysis block, as illustrated in Figure 5.3. The analysis sequence begins with the computation of
polynomial coefficients for lift (CL ), drag (CD ), and pitching moment (Cm ) coefficients, utilizing an
initial estimate of the center of gravity position xcg . These aerodynamic coefficients are subsequently
employed to formulate the trim equilibrium equations, which are solved using a Newton root-finding
method implemented through the Optimistix package. The trim calculation is performed with an
initial guess of the current aircraft mass W .
1
This parametric description can be formed by a combination of dependent and independent design variables.
2
This generation consists mainly in computing the initial adiabatic equilibrium of the hydrogen tank. For other simpler
models, as the TMS, this consists of setting all the initial temperatures to ambient.
5.2. AIRCRAFT MULTIDISCIPLINARY ANALYSIS WITHIN OPALO 63

Figure 5.3: Pseudo MDA structure for the mission states multidisciplinary analysis

From the trim solution, the resultant total drag force Dtot is combined with the axial force
required for aircraft acceleration (a · Waircraf t ) to determine the total3 thrust requirement Treq . This
thrust demand is then processed by the propeller trim discipline, where a root-finding algorithm
interrogates the propeller Cp -J performance map to determine the required propeller power Preq
necessary to generate the specified thrust.
The computed power requirement serves as a top-level requirement for the powerplant system,
which, when combined with the current mission phase control parameters (hybridization level, bat-
tery venting, thermal management, etc.), allows the powerplant to solve the input and output power
of each component, finally providing the fuelflow (and battery discharge rate). These consumption
rates are subsequently integrated over the mission segment duration and fed to the center of gravity
update module, which recalculates the updated xcg position for the next iteration of aerodynamic
coefficient computation. The integrated mass is similarly updated and passed to the aerodynamic
trim system, completing the coupling loop. The iterative process is solved using a Gauss-Seidel
algorithm (in order to maintain robustness across the convergence) and continues until convergence
is achieved across all coupled variables, ensuring that the mass-drag-fuel consumption dependencies
are resolved.

5.2.3 Takeoff multidisciplinary analysis

For the case of the take-off MDA, a similar approach to that of the main mission analysis is adopted.
Owing to the comparatively short duration of the take-off sequence, it is assumed that the aerody-
namic coefficients, the centre of gravity location, and the aircraft mass remain constant. The analysis
begins by computing the take-off aerodynamic coefficients, which are then passed to a sub-MDA
block defined for each phase of the take-off (e.g., V0 –V1 , V1 –VR , etc.). This sub-MDA (see it in the
pink group at Fig.5.4) block first evaluates the powertrain output power, which may be either fixed
(for example, in the twin-turboprop configuration δgbx = 1 under OEI conditions) or governed by
3
For cases where the TMS system is implemented the aerodynamic trim also computes the TMS drag, with the value
added to the aircraft drag.
64 CHAPTER 5. AIRCRAFT MULTIDISCIPLINARY ANALYSIS

Figure 5.4: Pseudo MDA structure for the LTO states multidisciplinary analysis.

a control parameter (for example, in PH architectures the turboprop operates at δgbx = 1 under
OEI conditions, but the hybridisation level is treated as a design variable). Using this output power,
the propeller map is evaluated to obtain an initial estimate of the aerodynamic trim (see Sec.3.1 for
further details). This trim calculation yields either the acceleration (for segments in which excess
thrust is used to accelerate, e.g., V0 –V1 ) or the flight path angle (for segments where excess thrust is
used to gain altitude, e.g., VR –V2 ).
The computed accelerations are subsequently passed to a kinematics integrator (see it in the
darker raspberry colored block at the lower right at Fig.5.4), which incorporates a nested velocity
solver already implemented in [74]. At the top level, the solver adjusts V1 as a variable within a
Newton root-finding method, aiming to determine the V1 velocity such that the continued take-off
distance under OEI conditions equals the rejected take-off distance. The second level of the nested
solver takes for each segment the current 4 V1 value at each iteration and computes the constant
time intervals required to transition from one velocity to the next, given the vector of accelerations.
Upon convergence of the nested solver, the kinematics module outputs the V1 velocity together with
the velocity vector for each take-off phase. This information is then used to recompute the propeller
thrust and drag values, which in turn provide an updated estimate of the accelerations.
The main advantage of this approach (computing the accelerations outside the nested velocity
solver) is that it allows the comparatively inexpensive velocity convergence to be performed without
re-evaluating the more computationally intensive performance models. This not only improves
computational efficiency5 , but also enhances robustness. Specifically, the propeller model and
aerodynamic trim are only evaluated using converged velocities. Combined with a velocity-limiting
strategy6 , this prevents the evaluation of models at negative velocities, which could otherwise cause
4
The updated value of V1 is only important to the V0 –V1 , V1 –VR and V1 –V0 segments.
5
The performance gain is partially offset by the need to solve the outer MDA coupling between the aircraft performance
models and the kinematics integrator.
6
This is particularly important in cases of poor propeller design, where the aircraft drag exceeds the available thrust,
leading to negative accelerations during segments that should exhibit positive acceleration. To address this, a sigmoid
5.2. AIRCRAFT MULTIDISCIPLINARY ANALYSIS WITHIN OPALO 65

divergence in the MDA process.

5.2.4 Overall Integration

For the overall integration of the system, Fig. 5.4 expands upon the conceptual structural separation
introduced in Fig. 5.2. As previously discussed, the preprocessing stage generates all the required
geometrical and sizing information for each analysis group. The code then executes the LTO aircraft
performance group and connects it with each of the mission phases7 . Within each phase group, state
information is exchanged through the consistent namespace. Once the segment MDA loops have
converged, the ODE solvers are executed, and the objective function and consistency constraints are
computed. Combined with the additional constraints evaluated throughout the performance MDAs,
these results provide the final objective function f (x) together with the constraint sets g(x) and
h(x).

Figure 5.5: Pseudo MDA analysis of the overall aircraft sizing within OPALO.

It is important to note that Fig. 5.5 provides a conceptual representation of the MDA architecture.
In practice, certain components (such as the ODE solvers and the geometry analysis) are implemented
as multiple Discipline classes. This modular design enables the reuse of specific modules across

blending function is applied at low acceleration levels to ensure that the computed distance remains positive.
7
Here, only the climb, cruise, and descent phases of the nominal mission are shown; in practice, the extended mission
segments are also included.
66 CHAPTER 5. AIRCRAFT MULTIDISCIPLINARY ANALYSIS

different architectures, while also reducing computational cost when evaluating selected entries of
the total Jacobian.
Overall, the use of generalized modules, together with the aggregation of disciplines into MDA
blocks (also known as hierarchical modular system decomposition [210]), reduces the coupling
diagonal length in the N2 diagram, and facilitates the integration of additional segment groups. This
modular approach not only enables code reuse through object-oriented discipline definitions, but also
allows for efficient Jacobian approximations by combining inexpensive finite-difference disciplines
with more complex, algorithmically differentiated disciplines. As a result, the computational cost is
reduced and convergence is improved, leading to optimizations that are both faster (fewer overall
iterations) and more accurate, owing to the reduced numerical noise compared to treating the entire
framework as a black box for finite-differencing.
In the following sections, we abstract from the internal implementation details of the framework
and adopt the perspective of an end user. Accordingly, only the objective functions to be used will be
presented, together with a description of the model’s design space, including the objective function
and the set of constraints considered for each optimization sweep.
 CHAPTER

VALIDATION RESULTS 6
Contents
6.1 Flight Profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.2 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
6.3 Optimization Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

The objective of the following chapter is to evaluate the predictive capabilities of OPALO using
the ATR72-600 as a reference aircraft. Predicted fuel consumption rates, component weights, and
performance metrics are compared against BADA data and values available in the literature to quantify
accuracy. The chapter is organized by first describing the mission profile, then the sizing design
variables, and finally the optimized results are discussed and compared against the reference data.

6.1 FLIGHT PROFILE

In order to obtain a meaningful comparison of fuel consumption and thrust requirements, comparing
performance against known data is mandatory. For this reason, in this comparison, climb and
descent segments were developed based on BADA performance data, as illustrated in Fig. 6.1. In order
to maintain consistency with OPALO flight model and with the polynomial-based mission definition
(already seen in Sec.3.4), for the case of the climb profile, the initial points from the low-speed
procedures were excluded from the fitting of the velocity profile 1 . For the case of the descent
segment, the final approach speed was slightly increased to avoid unrealistic deceleration rates,
which also adds consistency with the fact that OPALO does not model spoiler deployment (used
typically to bleed energy during rapid descent segments) and assumes a clean aircraft configuration
throughout all mission segments.
For the intermediate segments, the climb and descent phases are connected via a cruise segment
at constant altitude hcruise = 6000 m. Within this segment, airspeed is increased linearly in EAS
from the Top of Climb (TOC) to the Top of Descent (TOD). For mission range, a total nominal
distance of 758 nm is imposed to match the maximum range capability of the ATR72. The resultant
1
It is important to remember here that while we are comparing against TAS data, the trajectory is defined using the
EAS data.

67
68 CHAPTER 6. VALIDATION RESULTS

complete mission profile is shown in Fig. 6.2. The selected values for constructing the mission are
presented in Table 6.1, where the point ψ1 corresponds to the start of climb, ψ2 correspond to the
mid-point of climb, ψ3 corresponds to the TOC, ψ4 corresponds to the mid of cruise, ψ5 to the TOD,
with ψ6 and ψ7 corresponding to the mid-descent and end of descent segment2 .

Control node ψ1 ψ2 ψ3 ψ4 ψ5 ψ6 ψ7
γ [deg] 6.00 2.66 0.75 – -3.50 -3.55 -2.27
EAS [m/s] 87.8 101.7 118.7 137.1 161.1 129.2 81.5

Table 6.1: Flight path angle and equivalent airspeed for each mission control node ψi

Figure 6.1: Climb and descent speed and flight path angle profiles based on the ATR72-600 perfor-
mance from BADA.

Since no information was available for an extended mission segment, only the nominal mission
was considered and simulated in this validation. To account for the additional fuel required for
alternate and reserve operations, a fixed fuel margin was added 3 . This value was derived by first
computing the nominal mission fuel via optimization, then applying a pre-determined ratio of
2
As it can be seen the speed is interpolated across the mission segments, with the gamma only reserved for the altitude
change mission segments.
3
In practice, the reserve fuel mass depends on the airline operation, season, and expected weather. This means that,
for example, winter operations might increase the reserve fuel just in case a longer diversion due to icing is expected.
6.2. OPTIMIZATION STATEMENT 69

Figure 6.2: Altitude profile and Mach number profile for the validation mission of the ATR72-600.

nominal to total fuel based on previous design studies. For this case, this value is expected to be 520
kg of additional fuel4 .

6.2 OPTIMIZATION STATEMENT

The main objective of this validation optimization is to determine how closely OPALO can reproduce
the performance of the reference aircraft. To this end, the wing aspect ratio ARw , wing area Sw,w ,
and propeller diameter were fixed to match those of the ATR72-600 (this avoids different performance
due to higher aspect ratios or design wing loadings), with the other variables (such as tail surface
areas and powerplant component ratings) allowed to be sized by the tool. This approach validates
both the aerodynamic stability constraints and the internal powerplant sizing methodology within
the optimization process, and ensures that if we fix and freeze some of the design characteristics of
the ATR72-600, we should get the other variables to have similar values.
Table 7.1 summarizes the design variables and constraints used in the optimization. Values in
black correspond to scalar design variables or constraints applied across the entire mission. Values
in salmon denote variables defined per mission segment; for example, variations in Mtip (and thus
propeller RPM) that are linearly interpolated within each phase. Values in pink represent constraints
enforced over a vector of mission segments (one constraint per phase). Lastly, values in orange refer
to variables or constraints defined separately for each mission failure mode (which in this case and
the following optimization will be just the OEI; for the multi-failure case scenario optimization, see
[150]). This color notation will be held constant throughout the following sections.
For this and subsequent optimizations, a constraint is imposed on the mass margin, defined as
the residual between the M T OW value selected by the optimizer and the sum of all Wi aircraft
components (structural, powerplant, operational, etc.), fuel Wf uel and payload WP L (this would
represent the takeoff weight T OW ). This approach decouples mass consistency from the MDA,
4
This procedure is only considered for this validation run, with further optimization runs considering the simulation
of the extended mission and its trajectory optimization.
70 CHAPTER 6. VALIDATION RESULTS

Design variable (units) Lower Upper Constraint (units) Lower Upper

MTOW (kg) 18000 100000 δ gas turbine (-) — 0.9/1
Prop Mtang (-) 0.4 0.9 δ gearbox (-) — 0.9/1
Prop factor (-) 0.76 1.74 Propeller Cp (-) — 3
Wing ndpos (-) 0.2 0.55 Propeller Mtrue (-) — 0.75
HTP Sref (m2 ) 5 50 Static margin (-) 0.35 -
VTP Sref (m2 ) 5 50 Mass margin (m) 0.01 —
Turboshaft rating (MW) 0.02 200 BFL (m) — 1400
Gearbox rating (MW) 0.02 200
Mtip_start (-) 0.8 1.2
Mtip_end (-) 0.8 1.2

Table 6.2: Design variables and bounds (left), and associated constraints with limits (right).

treating MTOW as a design variable to be converged using an IDF formulation5 . This ensures that
the optimizer not only sizes components to operate at their efficiency optima, but also achieves
minimum mass 6 .
 X 
Mass Margin = M T OW − Wi − Wfuel − WPL , (6.1)
i∈components

The optimization objective is to minimize mission block fuel (Eq. 6.2) while satisfying constraints
on throttle settings δ, propeller operation, mass margin, and balanced field length.

f (Wfuel,block ) = Wfuel,block (6.2)

6.3 OPTIMIZATION RESULTS

After defining the optimization problem, the optimization was ran, yielding the aircraft geometry
shown in Fig. 6.3. This 3D model depicts the optimized values of the design variables, which are
summarized in Table 6.3.
Reference values for the ATR72-600 were obtained from [211], and the static shaft power rating
for the PW127XT engine was taken from [212]. The optimized outer mold line (OML) values
(particularly the tail areas and wing position) closely match those of the reference aircraft. Figure 6.4
visually compares the two configurations. This close agreement validates the sizing7 of the control
surfaces relative to the center of gravity estimation
.
5
The constraint is not imposed as an equality constraint; however, since the optimization seeks to minimize MTOW to
reduce fuel consumption, this constraint remains active throughout the optimization process.
6
As demonstrated later, this is not universally true, as a heavier but more efficient aircraft may consume less fuel than
a lighter, less efficient configuration.
7
In this case, it validates that the static margin that we have selected yields a sufficient tail volume compared to the
reference turboprop; for a more in-depth design, we shall consider additional handbook cases, dynamic stability, and pilot
handling feeback.
6.3. OPTIMIZATION RESULTS 71

Figure 6.3: Resultant external geometry of the optimized aircraft with the inner layout and power-
plant.

Design Variable Final Value ATR72 Value Difference (%)

MTOW (kg) 23,332.4 23,000 +1.44
Propeller Mtang 0.62 0.60 +3.33
Propeller factor 0.77 — —
Wing ndpos 0.42 0.41 +2.44
HTP Sref (m2 ) 11.13 11.59 −3.96
VTP Sref (m2 ) 16.30 16.73 −2.57
Turboshaft rating (MW) 2.37 2.05 +15.6
Gearbox rating (MW) 2.41 — —

Table 6.3: Comparison of optimized design variables with ATR72 reference values. Tangential
propeller Mach number at sea level for reference data computed assuming 1000 RPM operation.

Regarding variations from the reference aircraft data, we can see that all the values closely
match the expected outcome, with the only big difference in this case appearing in the turboshaft
power rating, for which the optimization yields a value 15.6% higher than the static sea-level rating
reported in the reference data.
To investigate this discrepancy, we can check if the difference comes from the aerodynamic
model. Starting with the coefficient of drag CD , we can see in Fig. 6.5 (top left) that the obtained
polar values closely align with those obtained from the BADA polar data8 , with only a 2.1% increase
in CD at constant CL . If we look at the required thrust (already considering the flight path angle and
required velocity acceleration), we can see in Fig. 6.5 (top right) that we also have a good agreement
with reference data, with only minor deviations at lower altitudes where the trajectory exhibits a
higher flight path angle ( Fig. 6.1, top right).
This means that the elevated power requirement is then likely due to the higher airspeed
maintained during low-altitude segments (see Fig. 6.1, upper left), which means that for a constant
thrust, the required power is higher. This, combined with a relatively low propeller efficiency
8
That assumes a simple parabolic polar.
72 CHAPTER 6. VALIDATION RESULTS

Figure 6.4: Comparison between the ATR72-600 from drawings and an ATR72-600-like generated in
TOPAZ

Component TOPAZ value Ref value Rel difference (%)

Fuselage mass 3427.2 3340.6 2.59
Wing mass 2215.09 2080.6 6.46
HTP mass 186.789 192.6 -3.02
VTP mass 298.296 271.7 9.79
Nacelle mass 371.194 409.4 -9.33
LDG mass 628.68 675.8 -6.97
Overall structure mass 7127.25 6970.7 2.25
Overall powerplant mass 1725.38 1557.8 10.76
Systems mass 2048.58 2118 -3.28
Furnishing 1886.5 1925 -2.00
Empty mass 12787.71 12571.5 1.72
Operating items 648.82 636 2.02
Passenger mass 5997.6 — —
Luggage mass 1552.4 — —
Total fuel (nominal+reserve) 2409.93 — —
MTOW 23332.4 23000 1.45

Table 6.4: Comparison of mass breakdown between the TOPAZ model and ATR72-600 reference
values [193].

(Fig. 6.5, lower right), yields a greater shaft power demand and, consequently, a higher fuel flow
(Fig. 6.5, lower left). What matters from the fuel flow is that the value at TOC (where we do have
the same thrust and flight regime) is quite similar, which will be crucial in order to ensure that we
are considering a gas turbine of similar performance.
6.3. OPTIMIZATION RESULTS 73

Figure 6.5: Upper left: comparison of aerodynamic polar at cruise condition. Upper right: comparison
of thrust requirement at climb condition. Lower left: comparison of fuel flow at climb condition.
Lower right: propeller efficiency η at climb condition.

Weight breakdown data is gathered in Table 6.4, where it can be seen that the masses predicted
by OPALO exhibit only slight deviations from reference ATR72-600 data [193]. All components
demonstrate variations below ±10%, except for the overall power-plant mass (where the higher
power rating of the gas turbine yields a 10.76% increase). These deviations are to be expected,
given that handbook-based weight methods provide only first-order approximations. In addition,
the calibration of this handbook’s formulas was performed with a mix of aircraft with different
technology levels (e.g., aluminum and CFRP structures), so the obtained values only give a "mean"
mass. In order to better capture the aircraft mass, physics-based FEM methods should be introduced
in the loop. However, even with this caveat, the predicted maximum take-off weight (MTOW)
closely matches the reference value, lending confidence to OPALO’s internal mass breakdown and
center-of-gravity9 (CG) estimation.
In summary, this validation study demonstrates that OPALO accurately predicts key performance
metrics and mass distribution for a turboprop configuration representative of the ATR72-600. How-
ever, it should be noted that the models were calibrated for aircraft operating in similar flight regimes,
and significant deviations from this baseline may result in larger discrepancies in performance pre-
dictions.
9
Since detailed mass allocation data were unavailable in the literature, cabin-component centers of gravity were
approximated using geometric centers observed in available schematics. Onboard systems mass was allocated at the
fuselage center of gravity, while operating items mass was positioned at the cabin center of gravity.
 CHAPTER

KEROSENE TURBOPROP
RESULTS 7
Contents
7.1 Is Lighter Better? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.1.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
7.1.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
7.2 Sensitivity study on range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
7.2.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
7.2.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
7.3 Sensitivity study on passenger count . . . . . . . . . . . . . . . . . . . . . . 85
7.3.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7.3.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7.4 Sensitivity study on operative costs . . . . . . . . . . . . . . . . . . . . . . . 88
7.4.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
7.4.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

The objective of this chapter is to present the optimization results for regional kerosene turboprop
aircraft configurations, focusing on the trade-offs between maximum takeoff weight (MTOW) and
fuel consumption as functions of design range and passenger capacity. The chpater begins with a
parametric study of MTOW versus fuel burn weighting for a baseline 72-passenger aircraft at 752 nm
range, establishing the fundamental trade-off characteristics. Subsequently, the analysis examines
range sensitivity by varying mission distance from 250 to 2000 nm at fixed capacity, followed by
capacity studies spanning 40 to 120 passengers at the baseline range. The investigation concludes
with a combined exploration of the coupled range-capacity design space to identify the evolution of
the trends in the regional turboprop architecture.

7.1 IS LIGHTER BETTER?

In multidisciplinary aircraft design optimization, mass minimization is typically pursued as a primary

objective under the assumption that reduced weight directly translates to improved performance

75
76 CHAPTER 7. KEROSENE TURBOPROP RESULTS

metrics. For conventional aircraft configurations, this relationship generally holds: lighter structures
enable increased payload capacity or reduced fuel consumption for equivalent mission profiles.
However, the coupling between increasing or decreasing the aircraft mass and system performance
can lead to some non-trivialities that require careful examination.
The complexity arises from the interdisciplinary nature of aircraft design, where mass reduction
in one subsystem may impose performance penalties in others. For instance, aircraft with higher
structural mass may accommodate increased aspect ratio wings, resulting in reduced induced drag
coefficients and improved aerodynamic efficiency. Similarly, the additional mass budget may permit
more powerful propulsion systems that enable steeper climb gradients to optimal cruise altitudes,
thereby reducing time spent at less efficient flight conditions.
These aerodynamic and operational improvements can, under certain design conditions, offset
the direct penalties associated with increased aircraft mass, potentially yielding net reductions in
mission fuel consumption. However, higher maximum takeoff weight (MTOW) invariably increases
both acquisition costs and operational expenses, including landing fees, insurance, and maintenance
requirements. This creates a fundamental tension between minimizing fuel consumption and
minimizing aircraft weight, necessitating a multi-objective optimization approach.

7.1.1 Optimization statement

To explore this multi-objective trade-off space, the optimization problem is formulated using a
weighted sum approach that combines block fuel consumption and maximum takeoff weight into a
single objective function. The composite objective function is expressed as:

Wf uel,block M T OW
f (Wf uel,block , M T OW ) = +ϱ· (7.1)
Wf uel,ref M T OWref
where Wf uel,block represents the mission block fuel consumption, M T OW is the maximum
takeoff weight, and ϱ ∈ [0, 1] is the weighting parameter that adjusts the relative importance
between minimizing fuel consumption (ϱ → 0) and minimizing aircraft weight (ϱ → 1). The
reference values Wfuel,ref and M T OWref provide normalization to ensure both terms contribute
equally to the objective function magnitude. For this investigation, the reference values are selected
as:

Wfuel,ref = 2000 kg, M T OWref = 23000 kg (7.2)

These values represent anticipated values and allow for a more reasonable normalization between
0 and 1.
For the optimization design variables, both aircraft design variables and trajectory design
variables are selected in order to capture the coupled nature of configuration and operational
optimization1 . Table 7.1 summarizes the complete design variable set and associated constraints. The
design space incorporates key geometric parameters, including propeller diameter, wing reference
1
This approach is essential, as prescribing a fully fixed trajectory may introduce design tendencies that merely
compensate for excessively demanding mission segments rather than reflecting optimal aircraft characteristics.
7.1. IS LIGHTER BETTER? 77

area, and aspect ratio, which were held constant in previous studies (Chapter 6). Additionally, the
trajectory optimization introduces equivalent airspeed (EAS) and flight path angle (γ) as design
variables at discrete control points ψi (same nomenclature as in Chapter 6 is being conisdered here)
throughout the mission profile (for further details see section 3.4 and section 3.5).
To ensure operational realism, certain trajectory parameters remain fixed: the initial climb speed
is constrained to EAS1 = 3Vstall , while the final approach speed is set to EAS7 = 1.6Vstall . Similarly,
the initial climb angle is fixed at γ1 = 6◦ and the final descent angle at γ7 = −2.75◦ . For cruise
operations, altitude is maintained constant throughout the segment, as ATC typically authorizes
multi-altitude cruise profiles only for longer flights, whereas regional operations prefer single cruise
altitudes. In terms of flight distance, a nominal mission of 758 nm with an extended mission of
200 nm is selected. These constraints ensure mission profiles consistent with typical turboprop
operational procedures and air traffic control requirements.

Design variable (units) Lower Upper Constraint (units) Lower Upper

MTOW (kg) 18000 100000 δ gas turbine (-) — 0.9/1
Prop Mtang (-) 0.4 0.9 δ gearbox (-) — 0.9/1
Prop diam (m) 2.5 5 Propeller Cp (-) — 3
Prop factor (-) 0.76 1.74 Propeller Mtrue (-) — 0.75
Wing ndpos (-) 0.2 0.55 Static margin (-) 0.35 -
Wing Sref (m2 ) 30 90 Mass margin (m) 0.01 —
Wing AR (-) 8 22 BFL (m) — 1400
HTP Sref (m2 ) 5 50
VTP Sref (m2 ) 5 50
Turboshaft rating (MW) 0.02 200
Gearbox rating (MW) 0.02 200
hcruise (m) 0.02 200
EASstart (m/s) 0.8 1.2
EASmid (m/s) 0.8 1.2
EASend (m/s) 0.8 1.2
γstart (-) 0.8 1.2
γend (-) 0.8 1.2
M tipstart (-) 0.8 1.2
M tipend (-) 0.8 1.2

Table 7.1: Design variables and bounds (left), and associated constraints with limits (right). Values in
salmon denote variables defined per mission segment, values in pink represent constraints enforced
over a vector of mission segments, values in orange refer to variables or constraints defined separately
for each mission failure mode.

The problem includes 52 design variables: 11 from aircraft design, 14 for the velocity profile,
13 for the flight path angle and altitudes, and 16 for propeller control. For the constraints, with
mn = 10 nodes per segment, 8 mission phases, and 5 state variables, we obtain 10 × 8 × 5 = 400
mission-related constraints, plus mass margin and takeoff distance, giving a total of 402 constraints.
This study is conducted for both conventional and strut-braced wings to assess the effect of the
new configuration, with further analyses focusing solely on the strut-braced variant.
78 CHAPTER 7. KEROSENE TURBOPROP RESULTS

7.1.2 Optimization results

For the conventional wing configuration optimization, Fig. 7.1 shows the effect of varying the
weighting parameter ϱ. The multi-objective results reveal a clear Pareto front between MTOW and
nominal block fuel2 , where reduced maximum takeoff mass is traded for lower fuel consumption.
As indicated by the colored markers in both Fig. 7.1 and Fig. 7.2, this trade-off primarily involves
increasing the wing aspect ratio while maintaining constant wing area3 .

Figure 7.1: Aspect wing ratio ARw across the resultant M T OW vs nominal block fuel pareto front
for the conventional cantilever wings.

Figure 7.2: Evolution of wing area and turboshaft rating along the M T OW -Nominal block fuel
Pareto front for the conventional cantilever wings.

The reasoning behind this is that higher aspect ratio wings offer improved aerodynamic efficiency
(Fig. 7.3, left) at the cost of increased structural mass. This means that, depending on how structural
mass scales with AR, there exists an optimal aspect ratio where the additional aircraft mass (which
2
Due to convergence and constraint tolerances, some noise is expected within the Pareto front values.
3
The maintenance of constant wing area seems to be also partially aided by the fact that as AR increases, the optimal
CL,cruise of the wing increases. Thus, even though the aircraft becomes heavier, the wing area remains approximately
constant to maintain the higher operational CL .
7.1. IS LIGHTER BETTER? 79

translates to extra induced drag) is offset by the higher efficiency and reduced induced losses
from the high aspect ratio wing. However, this also creates a cascading effect, as higher M T OW
increases the structural mass of other aircraft components (as the additional loads require structural
reinforcements). Thus, when M T OW minimization is prioritized, the wing aspect ratio is quickly
reduced to minimize the structural mass requirements from the wing and the cascading components.

Figure 7.3: Aerodynamic efficiency and altitude profile for the conventional cantilever wings.

This multidisciplinary cascading effect also extends to the powertrain, as additional aircraft
mass requires higher installed power to maintain performance (Fig.7.2). This means that mass
minimization leads not only to a lighter, lower aspect ratio wing, but also to slight downsizing of the
gas turbine4 . This downsizing is further exploited through a slightly lower cruise altitude (Fig.7.3,
right), where the gas turbine can be reduced in size by operating at higher inlet density, leading
to a smaller gas turbine core that further assists mass minimization. However, this comes at the
expense of lower combustion chamber ∆T , thus also slightly increasing fuel consumption through
marginally higher PSFC.
The addition of strut bracing to the wings reduces cantilever bending loads, thereby enabling
structural mass reduction for a given aspect ratio. Re-running the optimization with strut-braced
wings (Fig. 7.4) reveals higher achievable aspect ratios at lower structural masses. This results in
lighter and more efficient aircraft5 , which leads to reduced block fuel consumption, thus shifting the
Pareto front toward the lower left corner.
Examining how the aircraft design evolves, it is noteworthy that the strut-braced aircraft exhibits
an overall lower design wing area (Fig. 7.5, left), which stems partially from the reduced MTOW, but
more significantly from the additional strut area6 . This additional lifting surface enables the aircraft
to operate at higher CL0 and CLα at constant main wing projected area.
For installed power, the previous trends persist, leading to gas turbine downsizing as mass
minimization is prioritized (Fig. 7.5, right), coupled with a slightly lower cruise altitude (Fig. 7.6,
right) in order to avoid reducing the maximum power to be provided at cruise. Comparing Fig. 7.2
4
The power rating handled by the optimizer is defined at near static sea level conditions, not at top-of-climb maximum
power conditions, thus we can downsize the gas tubrine but still generate the same power at a lower altitude.
5
Efficiency gains stem from reduced induced drag due to longer wings and lower lift requirements per unit wing area,
as shown in Fig. 7.6, left.
6
As in this case our reference area for the wing is the main wing area.
80 CHAPTER 7. KEROSENE TURBOPROP RESULTS

Figure 7.4: Aspect wing ratio ARw across the resultant Pareto front for the conventional kerosene
turboprop aircraft for both conventional and strut-braced wings.

Figure 7.5: Evolution of wing area and turboshaft rating along the M T OW -Nominal block fuel
Pareto front for the SBW architecture.

Figure 7.6: Aerodynamic efficiency and altitude profile for the SBW Kerosene architecture.

(right) and Fig. 7.5 (right) reveals overall downsizing when transitioning from cantilever to strut-
braced wings, which stems from the reduced aircraft mass (decreasing power requirements during
7.2. SENSITIVITY STUDY ON RANGE 81

climb) and higher aerodynamic efficiencies that reduce thrust requirements.

This improvement is also reflected in the performance metrics: the SBW achieves aerodynamic
efficiencies exceeding 21.33 (for nominal block fuel minimization). This means that for the Nominal
block fuel minimization design, SBW can increase the aerodynamic efficiency7 in the order of
2.94% while reducing the M T OW mass in 1.03%, leading to a decrease of nominal fuel of 3.43% (as
compared with the optimized cantilever wing).

7.2 SENSITIVITY STUDY ON RANGE

Having examined the effects of MTOW minimization, it is valuable to investigate trends that emerge
with varying design range. To accomplish this, the following section performs a sweep analysis over
nominal mission ranges from 200 nm to 1400 nm, maintaining a constant extended mission of 200
nm, with the objective of evaluating fuel mileage across various nominal design ranges.
An additional study will evaluate the off-design performance of shorter flights typically operated
by regional aircraft to determine the fuel mileage of an aircraft designed for a nominal mission but
operating at reduced mission distances.

7.2.1 Optimization statement

For the optimization formulation, the aforementioned two cases are considered: the first case
examines the conventional kerosene strut-braced aircraft at different nominal design ranges with
identical design variables and constraint bounds as in the previous studies (Table 7.1). For all cases,
trajectories are modified through trajectory control design variables (in EAS and γ, cnsidering
also the previous trajectory feasibility considerations). For the objective function, from Eq. 7.1,
the parameter ϱ is maintained at ϱ → 0 such that only normalized nominal fuel minimization is
performed:

Wfuel,block
f (Wfuel,block ) = (7.3)
Wfuel,ref
For the second study, one of the synthesized aircraft at one of the specific nominal design ranges
is held constant in design (wing area, propeller design, powertrain ratings, etc.) and evaluated across
shorter ranges to assess its off-design performance. In this case, trajectory design variables remain
free to vary, allowing the same aircraft to be flown with different optimized trajectories for each
range. The aircraft is also left to take off at a lower TOW than its design MTOW8 ; with all aircraft
component masses remain fixed. In terms of the objective the same formulation as in Eq. 7.3 is
employed, transforming the optimization problem into a trajectory optimization for nominal block
fuel minimization, evaluating thus the optimal performance across different missions.
7
This model based on VSPAero solution does consider the vortex interaction between the strut and the main wing,
however it does not consider the additional compressibility interactions between the strut and the wing, nor the possible
shock waves that could appear with installed nacelles. For this case this model should be kept for below-transonic
conditions in order to avoid false performance improvements.
8
Maintaining always the mass margins considerations with respect to this new T OW .
82 CHAPTER 7. KEROSENE TURBOPROP RESULTS

7.2.2 Optimization results

Examining the optimization results for the first study, Fig. 7.7 represents the kerosene consumption
per nautical mile. As shown, fuel intensity increases with nominal design range. The primary
reason is that increased range requirements demand higher maximum fuel capacity, leading to a
higher M T OW that require heavier structures to sustain the additional loads, and larger wings
to accommodate both the increased mass and fuel volume requirements. Consequently, as the
aircraft maximum range increases, its on-design fuel consumption increases (due to higher energy
requirements from flying the heavier aicraft).

Figure 7.7: Evolution of the fuel mission intensity (in terms of kg of Kerosene per nm) as a function
of the design nominal mission range, colored with the maximum takeoff weight M T OW . The
figure also shows the aicraft geometry for three design conditions.

Figure 7.8: Altitude and Mach number flight profiles for the various design nominal ranges, colored
with nominal block fuel.

This trend holds except for very short on-design nominal ranges, where the brief mission
distances result in cruise segments of negligible duration (Fig. 7.8, left); meaning that the aicraft
7.2. SENSITIVITY STUDY ON RANGE 83

spends most of its time in very inefficient climb and descent flight segments.
Examining the vertical flight profile characteristics across all ranges, the cruise Mach number
remains approximately constant at 0.5 across all design ranges. However, the cruise altitude exhibits
more complex behavior: for shorter missions, the optimizer selects progressively higher cruise
altitudes as design range increases, but this trend reverses beyond 1100 nm, where cruise altitude
begins to decrease with further range extensions.

Figure 7.9: Fuel intensity (in terms of kg fuel per nm) as a function of range, colored by the wing
area (left) and turboshaft power rating (right).

These trends are reflected in the aircraft design evolution, where wing area increases monotoni-
cally9 with design range (Fig. 7.9, left), while the turboshaft power rating (Fig. 7.9, right) initially
increasing with range but subsequently decreasing beyond the threshold nominal range of 1100 nm.
This reduction occurs because the aircraft now operates at lower cruise altitudes with higher inlet
air density, enabling the gas turbine to generate equivalent power output with reduced core sizing
requirements (thus slightly lightening the powertrain10 ).
Looking now at the effect of off-design operation, Fig. 7.10 shows the previous fuel intensity for
each specific range, along with the fuel intensity of an aircraft that was initially sized for 1000 nm. As
can be seen, for ranges close to the 1000 nm on-design point, the fuel intensity remains approximately
constant; however, as the range decreases, not only is the fuel per nautical mile higher than the
range-specific sized aircraft, but the fuel intensity actually increases from the plateau of 1.95 kg/nm.
The main reason for this is that the oversized aircraft operates with higher structural mass relative
to the shorter mission requirements. Consequently, the aircraft must expend additional energy to
climb this unnecessary mass11 to cruise altitude, thereby increasing fuel consumption throughout
the mission profile. This effect can be seen at 270 nm, where the aircraft sized for that nominal
mission achieves a fuel intensity of 1.836 kg/nm, while the off-design aircraft requires 1.990 kg/nm,
representing approximately 8.37% higher fuel intensity.
This means that the main driver for this on- and off-design performance difference is not the
flown trajectory, as shown in Fig. 7.11 where the optimal flight profiles remain unchanged, but
rather the additional structural mass penalty. This reveals that designing an aircraft for a given
9
This behavior is expected, as the aircraft must accommodate both the higher MTOW and the additional fuel volume
required for extended mission ranges.
10
Further details on this trend are studied in the following range and pax grid analysis.
11
Given that the aicraft is now flying at a T OW underneat its M T OW
84 CHAPTER 7. KEROSENE TURBOPROP RESULTS

Figure 7.10: Fuel mission intensity (kg Kerosene/nm) vs. design nominal mission, colored by T OW .
▲ marks aircraft optimized for the nominal range; ■ marks off-design performance of an aircraft
sized at 1000,nm.

Figure 7.11: Altitude and Mach number vertical profiles for range-specific designs (left column) and
off-design performance of the 1000 nm sized aircraft (right column), all colored by fuel intensity.

nominal range represents a fundamental compromise: selecting a short nominal range limits the
ability to fly longer missions (as they would exceed the design MTOW), while designing for long
range and subsequently operating significantly shorter missions results in suboptimal performance
7.3. SENSITIVITY STUDY ON PASSENGER COUNT 85

due to higher fuel intensity. This represents a crucial design consideration driven primarily by
airline operational requirements12 , as maintaining longer range capabilities within a regional aircraft
can facilitate fleet maintenance standardization and enable access to unusual routes that may be
chartered opportunistically.

7.3 SENSITIVITY STUDY ON PASSENGER COUNT

As we have seen on the previous segment, the variation on design range dictates some interesting
design criteria (as the design mass and operational flight altitude), in this section, this study will be
expanded by including the effect of passenger count into the mix.

7.3.1 Optimization statement

For the optimization statement, the design space and constraints from (Table 7.1) are considered to
remain unchanged, with the objective function being the one from Ec.7.3.

lcabin = lcabin,nom · χ (7.4)

WP L = (WP L,pass + WP L,luggage ) · χ (7.5)

The main modifications from previous studies will be that the payload and fuselage cabin and
luggage lengths (aggregated into lcabin ) will grow proportionally to the ratio χ between the new
passenger count and the nominal design passenger count (72 for this case). In this case the study
will sweep the number of rows (with 4 seats each) from 12 to 25 rows.

7.3.2 Optimization results

Regarding the optimization results, Fig. 7.12 (left) represents the fuel intensity normalized by both
passenger count and nominal design distance, leading to the aircraft geometries that can be seen in
Fig. 7.13. As can be seen, the lowest fuel intensities occur at shorter distances and higher passenger
capacities. This behavior is expected based on the previous analysis, as shorter ranges result in
lower MTOW aircraft (Fig. 7.12, right), which exhibit reduced induced drag losses. Examining the
passenger count dependency, fuel intensity decreases as passenger capacity increases. This trend
occurs because many operational systems and structural components scale with passenger count
at a lower rate than the passenger count itself, enabling the aircraft to carry additional passengers
without proportionally increasing operational mass requirements.
Selecting an aircraft within this tendency must be consistent with airline operational constraints,
as while the lowest fuel-intensive design might be a high-capacity aircraft for short routes, that
aircraft becomes constrained to those specific short routes, thereby limiting its market applicability
where demand may be insufficient. Furthermore, if this same high-capacity aircraft is designed for
12
Which is not strictly studied in this thesis.
86 CHAPTER 7. KEROSENE TURBOPROP RESULTS

longer routes and subsequently operated at shorter ranges, the fuel intensity increases significantly.
This effect is compounded by the fact that if the aircraft cannot achieve high load factors, the fuel
intensity per passenger will increase substantially. These considerations demonstrate that even
highly efficient aircraft designs must align with market demand characteristics, which explains
why the regional market encompasses aircraft capacities ranging from 8 to 80 passengers to match
diverse route and demand profiles.

Figure 7.12: Left: Fuel intensity (normalized with range and total passenger count). Right: Aircraft
design MTOW. Both as a function of number of passengers and nominal design range.

Figure 7.13: Various resultant aircraft shapes from various passenger and range design requirements.

Regarding the evolution of design parameters across the design space, while already evident
in the geometries from Fig. 7.13, wing area increases with both range and passenger count to
7.3. SENSITIVITY STUDY ON PASSENGER COUNT 87

accommodate higher MTOWs (Fig. 7.14, left). More intriguingly, the aspect ratio exhibits distinct
regions of higher and lower values (Fig. 7.14, right). These lower aspect ratio regions appear to
stem from the optimizer prioritizing mass minimization to achieve fuel reduction, particularly for
configurations with low passenger counts and high ranges13 . This phenomenon also emerges at
higher passenger counts, suggesting a sensitivity threshold where mass minimization becomes
the dominant strategy for fuel reduction, likely occurring when certain component mass fractions
relative to total aircraft mass reach critical values14 .

Figure 7.14: Left: Wing design area. Right: Wing design aspect ratio AR. Both as a function of the
number of passengers and the nominal design range.

Figure 7.15: Left: Turboshaft power rating. Right: Cruise altitude. Both as a function of the number
of passengers and the nominal design range.

Examining the turboshaft rating and cruise altitude, this trend becomes even more evident. For
the turboshaft power rating (Fig. 7.15, left), the power requirement increases with passenger count
as MTOW becomes greater; however, for low passenger counts and high ranges, this power rating
actually decreases. This behavior is also reflected in the cruise altitudes (Fig. 7.15, right), where
13
Recall that for these runs ϱ → 0. To verify this was not an initial value sensitivity leading to local minima, the grid
was re-run with different starting conditions, consistently reproducing the same results.
14
Due to time and computational budget constraints, this could not be fully explored through sensitivity propagation
or made model agnostic.
88 CHAPTER 7. KEROSENE TURBOPROP RESULTS

maximum altitudes occur at intermediate ranges, decreasing as passenger count increases, with
particularly pronounced reductions for lower capacity, shorter range configurations.
These optimized aircraft configurations must, however, still consider additional operational
constraints from airline operations, as while it might be beneficial from an aircraft design perspective
to operate at lower cruise altitudes, increased turbulence at these altitudes may result in passenger
discomfort and pilot fatigue that outweigh the fuel efficiency gains.

7.4 SENSITIVITY STUDY ON OPERATIVE COSTS

Finally, in order to close our study of the regional turboprop architecture, it is interesting to study
the effect that operational costs have on aircraft operation. This is important as while an aircraft
with lower flight speeds might have better fuel economy, the added operational time might increase
costs associated with maintenance and crew operations.

7.4.1 Optimization statement

For the optimization statement, this study will initially consider only our baseline case of 72 pas-
sengers and 758 nm of nominal range with 200 nm of extended range. For the cost estimation, the
following study will perform a multi-objective optimization between fuel cost and operational costs.
As before, the objective function will be formulated with a weighting factor ϱ, such that the objective
function is expressed as:

f (Costf uel , Costoper ) = Costf uel + ϱ · Costoper (7.6)

Regarding the design space and constraints, the same design variables as shown in Tab. 7.1
are being considered. For the estimation of operational costs, a value of 702 USD per flight hour
for the ATR72 equipped with the PW127 [213] is used15 , which when adjusted for inflation yields
1088 USD/FH. For fuel cost, a value of 675 USD/ton is considered.
After performing this study, an additional study will be performed over the number of passengers
and design range (same design space as in previous section), but now considering operational costs
in the form of a minimum mean velocity constraint16 .

7.4.2 Optimization results

Regarding the optimization results, Fig. 7.16 shows the results from the multi-objective optimization.
As can be seen, as operational costs are decreased, overall fuel costs increase. In order to understand
this trade-off, we can examine how the cruise altitude and mean mission velocity17 vary within the
15
This represents an initial estimate for operational costs, as the strut-braced wing configuration, theoretical engine
installations, and specific airline operations may result in higher operational costs than those considered here. This value
is taken as an example to study the multi-objective trade-offs.
16
The objective of this study is to examine if there are notable changes in aircraft design and efficiency after imposing
higher operational speeds.
17
Computed by taking the mission distance and dividing it by the flight mission time.
7.4. SENSITIVITY STUDY ON OPERATIVE COSTS 89

Pareto front. As shown in Fig. 7.17, operational costs are decreased by increasing the mean velocity
(which reduces mission time and thus reduces time-dependent costs) and slightly reducing the
cruise altitude (which allows the aircraft to operate at an altitude with higher sound speed, enabling
the aircraft to achieve faster true airspeeds without proportionally increasing the Mach number
and compressibility drag). This, however, forces the aircraft to operate at higher speeds, which in
turn increases compressibility drag terms and requires higher installed power for equivalent thrust
output, thereby increasing fuel requirements and associated fuel-related costs.

Figure 7.16: Results from the multi-objective optimization between operational costs and fuel costs,
colored by MTOW.

Figure 7.17: Results from the multi-objective optimization between operational costs and fuel costs,
colored by cruise altitude (left) and mean velocity (right).

Considering operational costs is, however, crucial when evaluating low-emission aircraft and
trajectory optimization. Examining the two extreme values from Fig. 7.16, when operational cost
considerations are introduced, operational costs decrease from 3253 USD to 2679 USD (approxi-
mately 17.64% lower operational costs), while fuel costs only increase from 969 USD to 977 USD
(approximately 1.03% higher fuel cost).
90 CHAPTER 7. KEROSENE TURBOPROP RESULTS

The primary reason for this significant shift is that without operational cost considerations, the
optimization can result in excessively long descent phases (see Fig. 7.18, left) that achieve marginally
lower fuel consumption but at significantly higher mission times18 .

Figure 7.18: Altitude (left) and Mach number (right) profiles for the optimized aircraft along the
costs Pareto front.

Regarding the rest of the optimization points from the Pareto front, Fig. 7.18 (right) shows once
more that in order to reduce operational costs, the flight Mach number is increased with reduced
cruise altitude and a steeper descent profile. However, it is also interesting to examine the design
characteristics of each aircraft configuration. As shown in Fig. 7.16, as fuel cost considerations are
introduced, there is initially a reduction in MTOW, partially due to a slight decrease in wing area,
but more importantly due to general aircraft downsizing (this can also be seen in the wing AR and
turboshaft rating19 in Fig. 7.19) to attain these higher speeds. This trend continues until the lower
cruise altitude design variable bound is reached, where MTOW begins to increase due to higher fuel
requirements (resulting from increased compressibility drag and reduced propeller performance),
accompanied by increased turboshaft rating to maintain the higher thrust requirements.
For the second proposed study in this section (the one that considers once more the sweep over
the number of passengers and range), the study from Sec. 7.3 is repeated with now a minimum
mission mean speed of 160 m/s (311 kn TAS). While this speed does not minimize the total mission
costs, it is sufficient to eliminate the initially long descent segments that lead to higher operational
costs. The results from the new grid can be seen in Fig. 7.20, where it can be seen that we have a
similar trend in fuel intensity with only minor differences in the lower operational ranges, where
the short missions have to be flown at higher overall speeds in order to compensate for the relatively
lower speeds of the climb and cruise segments, with minor variations for the higher design ranges.
Regarding the aircraft design, Fig. 7.21 shows the wing design (in terms of area and AR) and
Fig. 7.22 shows the turboshaft power rating and cruise altitude. Starting with the wing design
and comparing with Fig. 7.14, it can be observed that for the wing area, the values remain mostly
18
This is also crucial when considering turboprop propulsion, as the lower flight speeds can make turboprop routes
unattractive to passengers.
19
Once more, given that the turboshaft operates at lower altitude, the aircraft can downsize the gas turbine and
compensate with higher inlet mass flow.
7.4. SENSITIVITY STUDY ON OPERATIVE COSTS 91

Figure 7.19: Wing aspect ratio (left) and turboshaft rating (right) for the optimized aircraft along the
costs Pareto front.

Figure 7.20: Fuel intensity (normalized with range and total passenger count) as a function of number
of passengers and nominal design range under the mean speed constraint.

Figure 7.21: Left: Wing design area. Right: Wing design aspect ratio AR. Both as a function of the
number of passengers and the nominal design range under the mean speed constraint.

unchanged with just a minor decrease in wing area at the very lowest nominal ranges. For the case
of the wing AR, we can see that the trend of the lower right corner remains, with now an additional
decrease in AR at higher passenger counts and nominal ranges. This trend is influenced by the
92 CHAPTER 7. KEROSENE TURBOPROP RESULTS

operational cost considerations that also lead to a decrease in cruise altitude (Fig. 7.22) at the very
low design ranges; this is to be expected as the aircraft needs to compensate for the low speeds
in cruise and descent by minimizing the time spent in these segments (which also drives weight
downsizing in order to reach cruise altitude more rapidly).

Figure 7.22: Left: Turboshaft power rating. Right: Cruise altitude. Both as a function of the number
of passengers and the nominal design range under the mean speed constraint.

Within this chapter, we have observed several trends and demonstrated the case-specific nature
of aircraft design. While the results are subject to the models employed, some interesting trends
emerge that will be carried forward to the following chapters. In decreasing order of discussion: the
mean velocity constraint, while quite demanding for short ranges that need to compensate for their
limited mission distance, allows elimination of optimization artifacts such as unrealistically slow
descent profiles and helps align eco-aircraft designs with airline operational considerations. Second,
the passenger range and capacity will be maintained at 72 passengers and 758 nm, as while higher
passenger capacities are more efficient, it is uncertain whether at these higher capacities turboprop
demand will be sufficient or if regional turbofans and narrowbodies would be preferred by airlines at
these higher demand levels. This fixed range and passenger capacity also aligns the designed aircraft
with the current state-of-the-art and reference aircraft, thus allowing better insights into different
newer architectures if an evolutionary aircraft is to be considered in the future. Finally, for the
MTOW weighting, the following studies will consider ϱ = 0 in order to avoid initial penalizations,
particularly for hybrid aircraft that typically exhibit higher MTOWs.
 CHAPTER

KEROSENE PARALLEL
HYBRID RESULTS 8
Contents
8.1 The Naive Kerosene Hybrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
8.1.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
8.1.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
8.2 Thermal managed kerosene hybrid . . . . . . . . . . . . . . . . . . . . . . . 101
8.2.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.2.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

This chapter presents the results of the optimization study for the Kerosene Hybrid architecture.
The main objective is to assess how the system’s performance evolves with increasing battery-specific
energy. Additionally, the analysis explores the impact of incorporating thermal management systems,
trends as a function of range, and the evolution of total energy requirements with technology, along
with a first approximation to the LCA emissions of hybrid aircraft.

8.1 THE NAIVE KEROSENE HYBRID

In this section, the hybrid kerosene-electric architecture is evaluated using a simplified or "naive"
modeling approach, where the electronic components are assumed to simply draw power from the
battery (given a mission-prescribed hybridization) without accounting for thermal management
system mass requirements or cooling drag considerations. This simplified model serves as a baseline
for comparison against more detailed models that incorporate thermal management and energy
integration strategies.

8.1.1 Optimization statement

For the optimization of the Kerosene Hybrid architecture, the optimization statement already
presented in Chap.7 is expanded to include the components of the hybrid electric powerplant, mainly
the power ratings of the electric motor, inverter, and battery cell weight. For the control of the
aircraft, an additional hybridization design variable is introduced as already presented in Sec.4.2.

93
94 CHAPTER 8. KEROSENE PARALLEL HYBRID RESULTS

This parameter will vary linearly across each segment in the mission, with each segment having as
design variables a start (HFstart ) and end value (HFend ).

Design variable (units) Lower Upper Constraint (units) Lower Upper

MTOW (kg) 18000 100000 δ gas turbine (-) — 0.9/1
Prop Mtang (-) 0.4 0.9 δ gearbox (-) — 0.9/1
Prop diam (m) 2.5 5 δ inveter (-) — 0.9/1
Prop factor (-) 0.76 1.74 δ electric motor (-) — 0.9/1
Wing ndpos (-) 0.2 0.55 δ battery (-) — 0.9/1
Wing Sref (m2 ) 30 90 Battery SOC (-) 0.2 1
Wing AR (-) 8 22 Propeller Cp (-) — 3
HTP Sref (m2 ) 5 50 Propeller Mtrue (-) — 0.75
VTP Sref (m2 ) 5 50 Static margin (-) 0.35 -
Turboshaft rating (MW) 0.02 200 Mass margin (m) 0.01 —
Gearbox rating (MW) 0.02 200 Mean velocity (m/s) 160 —
Electric motor rating (MW) 0.02 200 BFL (m) — 1400
Inverter rating (MW) 0.02 200
Battery cell mass (kg) 0 200000
hcruise (m) 1000 7000
EASstart (m/s) 10 300
EASmid (m/s) 10 300
EASend (m/s) 10 300
γstart (deg) -10 10
γend (deg) -10 10
M tipstart (-) 0.8 1.2
M tipend (-) 0.8 1.2
HFstart (-) 0 1
HFend (-) 0 1

Table 8.1: Design variables and bounds (left), and associated constraints with limits (right). Values in
salmon denote variables defined per mission segment, values in pink represent constraints enforced
over a vector of mission segments, values in orange refer to variables or constraints defined separately
for each mission failure mode.

The resultant design space can be seen in Fig.8.1, where we can see that the problem now includes
73 design variables: 14 from aircraft design, 14 for the velocity profile, 11 for the flight path angle,
2 for the altitudes, and 32 for both propeller control and hybridization. For the constraints, with
mn = 10 nodes per segment, 8 mission phases, and 9 state variables, we obtain 10 × 8 × 9 = 720
mission-related constraints, plus the throttles for the takeoff, margins, and the mean velocity
constraint derived in Chapter.7, giving a total of 923 constraints.
For the technology level of the electric motor, inverter, and the cell to pack scaling, the values from
Chap.3.2 are utilized. As for the objective function, we will initially only consider the minimization
of nominal block fuel, borrowing the objective function from Eq.7.3, and not considering in any case
any M T OW considerations1 .

1
This will be done in order to avoid early penalization of battery mass installation.
8.1. THE NAIVE KEROSENE HYBRID 95

8.1.2 Optimization results

For the optimization results, we can begin by comparing the effect of aircraft designs as a function
of battery technology. As shown in Fig.8.1, as the battery cell specific energy increases, the amount
of installed battery cell weight also increases, while the nominal block fuel decreases. However,
this increase in stalled battery mass does not occur linearly or smoothly, but rather we have an
initial slight increase, till 500Wh/kg, where we have a sudden bump in installed battery cell mass.
This phenomenon occurs primarily because at low specific energies, batteries provide little energy
relative to their total mass; thus, once depleted, they act as "dead mass" that must be carried by the
aircraft, leading to an overall increase in aircraft fuel consumption. This changes at a tipping point
where the batteries begin holding sufficient energy such that their extra mass can out-compete the
additional fuel required by the aircraft to carry them. At this point, the optimizer decides to increase
battery utilization and the decrease in nominal block fuel increases more rapidly2

Figure 8.1: Evolution of battery cell weight and nominal block fuel for various cell’s specific energies
along the geometries for three technology levels. Batteries are highlighted in green with the inverter
and the electric motor in orange and yellow, respectively.

It is important to note that this battery mass addition shows that this aircraft design is a heavily
multidisciplinary problem, as the additional battery mass not only increases the aircraft’s M T OW
by the value of that battery mass, but also increases the structural mass (since the structure must
now withstand the additional loading). Furthermore, additional wing area is required to lift this
extra mass, leading to a mass increase beyond simply the battery installation itself, as demonstrated
in Fig.8.2. This multidisciplinary coupling is evident in Fig.8.1, where aircraft with higher specific
energies exhibit correspondingly higher wing areas. This disciplinary coupling further penalizes the
battery installation at low specific energies, whereas it can be seen that the optimized decision is to
maintain the aircraft mostly Kerosene-based.
2
As can be seen in the bigger slope (in absolute value) in the nominal block fuel decrease.
96 CHAPTER 8. KEROSENE PARALLEL HYBRID RESULTS

Figure 8.2: Evolution of component mass with cell-specific energy. Structural mass includes wing,
fuselage, and tail weight along with subsystems, furnishing, and operational items. Powerplant
includes turboshaft, gearbox, electric motors and inverters. Battery mass is expressed at pack level,
and payload includes passenger and luggage mass. For the Kerosene Parallel Hybrid with no thermal
management.

Figure 8.3: Energy provided by the battery and by the kerosene (left). Energy intensity from the
battery, kerosene, and total energy (right). For the Kerosene Parallel Hybrid with no thermal
management.

Considering now an energy aircraft demand perspective, we can compute the effective energy
obtained from the kerosene part of the powertrain as the integral of the gas turbine output shaft
power3 Pgt along the total nominal mission time tf ; and the effective energy provided by the battery
Pbatt as the integral of the output battery electric power along the total nominal mission time tf
(Eq.8.1).
Z tf Z tf
eff
Eker = Pgt (t) dt eff
Ebatt = Pbatt (t) dt (8.1)
t0 t0
3
This already considers the efficiency from the gas turbine in extracting energy from the lower heating value of the
kerosene combustion.
8.1. THE NAIVE KEROSENE HYBRID 97

This allows us to obtain the total breakdown of the aircraft total nominal energy usage (Fig.8.3
left) as a function of the specific energy. Where it can be seen that similar to the trend of the
cumulative mass, there appears to be a bump in energy demand at 500 Wh/kg cell level, where the
total energy demand rapidly increases from around 18,000 MJ at 300 Wh/kg to about 27,000 MJ at
700 Wh/kg, leading to a 50% increase in energy demand. This increase is to be expected given that
the aircraft is now substantially heavier (about 46% heavier), and thus requires substantially more
energy to carry out the mission. If we examine the energy intensity (Fig.8.3 right), we can see that
while there is a significant increase in total energy, the energy required from kerosene is able to
decrease about 18% from 250 Wh/kg to 700 Wh/kg, meaning that indeed the aircraft is able to rely
less on energy provided by fossil fuels. However, this comes with a great caveat, as if the energy
provided by the battery comes from an energy mix that relies heavily on fossil fuels, the aircraft
could end up having a substantially higher carbon intensity compared to the conventional kerosene,
going from a decrease of fuel dependence by 18% to an increase of 50%. This means that, as we will
see later, the hybridization of regional aircraft is a region-dependent problem, as in areas with low
renewable energy, these kinds of architectures lead to an overall less sustainable aircraft.
Examining where this extra electric energy is being introduced in the mission, Fig.8.4 shows
the evolution of the state of charge (SOC) of the battery and the hybridization factor along the
mission. As can be seen, the optimizer properly maintains the battery within its utilization range
SOC ∈ [0.2, 1]. It is interesting to observe that at low specific energies, the battery is being depleted
mostly during the descent segment, with low hybridization factors along the rest of the mission.
As we increase the specific energy, we see that the battery starts being depleted more quickly at
the start of the climb, and then at the cruise segment, while the energy utilized during the descent
segment slowly decreases.

Figure 8.4: Evolution of the state of charge and hybridization factor for the various aircraft specific
energy configurations for the Kerosene Parallel Hybrid with no thermal management.

The rationale behind this behavior stems partially from the modeling assumptions that we
have applied for the gas turbine. During the descent segment, the required powerplant power is
relatively low, which means that the gas turbine is forced to operate at a low power setting where
the power-specific fuel consumption is relatively high, as can be seen in Fig.8.6. This operating point
is quite undesirable; thus, when extra energy is present, the optimizer tries to utilize the battery
98 CHAPTER 8. KEROSENE PARALLEL HYBRID RESULTS

system more during this segment to reduce the gas turbine’s power production at low power levels
(however, at this condition the gas turbine is still consuming fuel at its idle condition). Then, as more
energy becomes available, this energy is introduced to provide power during the climb segment,
thus partially reducing the power rating requirements on the gas turbine (however, until the tipping
point, it can be seen that the gas turbine power rating is still increasing due to the increase in aircraft
overall mass). Once the tipping point appears, however, it can be seen that the battery starts to be
utilized more during the cruise segment, with the electric component power ratings up-sizing and
the gas turbine downsizing, as seen in Fig.8.5.

Figure 8.5: Evolution of electric motor power rating and gas turbine power rating with respect to
the cell specific-energy for the Kerosene Parallel Hybrid with no thermal management.

Figure 8.6: Gas turbine throttle and PSFC along the mission for the various aircraft specific energy
configurations for the Kerosene Parallel Hybrid with no thermal management.

This effect can also be observed in the behavior of the gas turbine: as it is downsized, its
utilization during descent increases, since it can now provide power at a higher throttle setting,
thereby reducing the degree of hybridization in the descent segment. At the same time, across all
specific energy levels, the gas turbine operates during climb and cruise at its higher-throttle, lower-
P SF C configuration. This occurs because the optimizer is able to introduce small adjustments in the
altitude velocity profile, to better match the power demand (Fig. 8.7), underscoring the importance
of coupled aircraft trajectory optimization.
8.1. THE NAIVE KEROSENE HYBRID 99

Figure 8.7: Vertical altitude profile and Mach number profile for the various aircraft specific energy
configurations for the Kerosene Parallel Hybrid with no thermal management.

Repeating this study now for a range of battery specific energies from 200 to 700 Wh/kg and
ranges from 275 to 1200nm Fig.8.8 is obtained, where it can be seen that this non-linear bump
manifests itself at different specific energy levels for different ranges. This is to be expected, as for
shorter flights the time the battery dead mass has to be carried is lower, while for longer flights the
additional mass means a heavier aircraft that will consume more, thus we see that a diagonal tipping
point line appears, where we only see fuel reductions at the right of the line.

Figure 8.8: Grid of battery cell mass (left) and nominal block fuel (right) as a function of nominal
mission range and cell specific energy level for the Kerosene Parallel Hybrid with no thermal
management.

Examining the results in terms of energy (using the same integration formulation as before),
it is evident that for a constant range, the total energy demand increases with cell-specific energy
(Fig. 8.9), due to the additional batteries raising the M T OW . The highest energy demands appear
on the upper right side of the tipping line, while the highest hybridization levels occur at the lower
right corner. This indicates that hybrid aircraft are particularly attractive for very short regional
flights, where most of the energy can be supplied by the battery, the kerosene engine operates for
relatively little time, and the higher M T OW penalty (Fig. 8.10) can be managed. At the same point,
we also observe the greatest upsizing of the electrical system at the lower right corners, following
100 CHAPTER 8. KEROSENE PARALLEL HYBRID RESULTS

the same trend with respect to the tipping line (Fig. 8.11).

Figure 8.9: Total required mission energy (left) and fraction from it that comes from the battery
(right) as a function of nominal mission range and cell specific energy level for the Kerosene Parallel
Hybrid with no thermal management.

Figure 8.10: Aircraft MTOW (left) and increase over 23000 kg baseline (right) vs. mission range and
cell specific energy for the kerosene parallel hybrid (no thermal management).

Figure 8.11: Turboshaft (left) and electric motor (right) power ratings vs. mission range and cell
specific energy for the kerosene parallel hybrid (no thermal management).
8.2. THERMAL MANAGED KEROSENE HYBRID 101

Figure 8.12: Relative decrease in block fuel (left) and increase in total energy requirement (right)
versus mission range and lower-bound cell specific energy for the Kerosene Parallel Hybrid with no
thermal management.

Examining the relative changes (by comparing performance levels against the baseline values
obtained at the lowest cell-specific energy for each range) reveals that the most significant block fuel
reductions occur at high specific energy and low mission range combinations (Fig.8.12 left), with
reductions of up to 79%. This explains why some regional aircraft concepts, such as the Heart ES-30
with its 107 nm all-electric and 431 nm hybrid capability, target the short-range market segment.
However, from an energy perspective (Fig.8.12 right), the sustainability debate becomes crucial:
regions with energy grids heavily dependent on fossil fuels will experience a net increase in overall
energy demand, potentially undermining the environmental benefits of hybridization and leading to
overall increases of up to 77% in fossil-fuel-derived power consumption4 .

8.2 THERMAL MANAGED KEROSENE HYBRID

The previous analysis has provided crucial insights into the trade-offs between battery cell mass,
aircraft mass, energy levels, and hybridization tipping points. However, all of these studies have
considered an optimistic scenario where thermal management systems were not a constraint. In
practice, electric and electronic systems require cooling, necessitating the use of cold air from
the freestream, which leads to additional TMS-related drag that increases overall drag and fuel
consumption. This section examines the impact of incorporating thermal management systems into
the hybrid aircraft design.

8.2.1 Optimization statement

For the optimization statement, we expand the design space from Tab. 8.1 to include the thermal
management system design variables and constraints (Tab. 8.2). These design variables include
overall TMS design variables (such as coolant pump power rating, coolant mass flow, and reservoir
4
This will be further explored in the following section and will depend on the energy production mix.
102 CHAPTER 8. KEROSENE PARALLEL HYBRID RESULTS

Design variable (units) Lower Upper Constraint (units) Lower Upper

MTOW (kg) 18000 100000 δ gas turbine (-) — 0.9/1
Prop Mtang (-) 0.4 0.9 δ gearbox (-) — 0.9/1
Prop diam (m) 2.5 5 δ inveter (-) — 0.9/1
Prop factor (-) 0.76 1.74 δ electric motor (-) — 0.9/1
Wing ndpos (-) 0.2 0.55 δ battery (-) — 0.9/1
Wing Sref (m2 ) 30 90 Battery SOC (-) 0.2 1
Wing AR (-) 8 22 Propeller Cp (-) — 3
HTP Sref (m2 ) 5 50 Propeller Mtrue (-) — 0.75
VTP Sref (m2 ) 5 50 Static margin (-) 0.35 -
Turboshaft rating (MW) 0.02 200 Mass margin (m) 0.01 —
Gearbox rating (MW) 0.02 200 Mean velocity (m/s) 160 —
Electric motor rating (MW) 0.02 200 BFL (m) — 1400
Inverter rating (MW) 0.02 200 Inverter Temp (ºC) - 70
Battery cell mass (kg) 0 200000 Emotor Temp (ºC) - 70
Coolant pump (kW) 0 200000 Battery Temp (ºC) 10 60
˙
mcoolant (kg/s) 0.01 100 F2F HX k design margin 0 -
Reservoir coolant (kg) 10 1000 F2F HX width clear 0 -
Coolant hose diameter (m) 0.01 10 F2F HX height clear 0 -
InvTMS HeightRatio (-) 0.1 1
InvTMS nparallel (-) 1 100
BattTMS tchannel (m) 0.001 0.1
F2F HX nlong,c (-) 1 1000
F2F HX nwide,c (-) 1 1000
F2F HX ntall (-) 1 1000
F2F HX design k (-) 1 1000
F2F HX multiplierc (-) 0.1 1000
F2F HX multiplierh (-) 0.1 1000
hcruise (m) 1000 7000
EASstart (m/s) 10 300
EASmid (m/s) 10 300
EASend (m/s) 10 300
γstart (deg) -10 10
γend (deg) -10 10
M tipstart (-) 0.8 1.2
M tipend (-) 0.8 1.2
HFstart (-) 0 1
HFend (-) 0 1
Ramopen,start (-) 0.001 1
Ramopen,end (-) 0.001 1
Battheating,start (kW) 00.01 1000
Battheating,end (kW) 0 1000

Table 8.2: Design variables and bounds (left), and associated constraints with limits (right). Values in
salmon denote variables defined per mission segment, values in pink represent constraints enforced
over a vector of mission segments, values in orange refer to variables or constraints defined separately
for each mission failure mode.
8.2. THERMAL MANAGED KEROSENE HYBRID 103

coolant mass), inverter-specific parameters (the height ratio between the inverter cold plate height
and the height of the inverter, which defines the height spacing; and the number of parallel fins
nparallel in the cold plate, which defines the horizontal spacing), the width of the battery cooling
channels tchannel , and the design variables of the fluid-to-fluid heat exchanger (where we define the
number of fins in the horizontal nwide,c and lengthwise nlong,c cold side directions, the number of
cold and hot stacks ntall , a set of multipliers over the height and width distances for the cold and
hot sides, and the design coefficient of losses k). Among the design variables, the control of the ram
air inlet ramp is also handled as a design variable to vary linearly across each segment, along with
the battery heating to maintain the battery within the operational temperature range at altitude.
Regarding the constraints, additional requirements are added for the powertrain components to
operate within their specific maximum temperatures. Moreover, in order to decouple the ram drag
from the MDA, an additional constraint is imposed over the pressure losses coefficient, such that
the design coefficient k handled as a design variable must be greater than the pressure coefficient k
obtained from solving the ODE. While this means that the MDA is not as consistent, this approach
provides a conservative solution and significantly increases the convergence of the stiff ODE. A
constraint is also enforced for the heat exchanger, such that its width and height does not exceed its
allocated space within the fuselage section.
This means that we now have 27 aicraft-level design variables, 14 design variables for the
velocity profile, 11 for the flight path angle profile, 2 for the altitudes and 64 design variables for
the propeller, hybridization, HX doors, and battery heating. Totaling 118 design variables. For
the constraints we now consider that each phase is discretized into nn = 20 nodes per segment5 ,
this means that for 13 state variables to control and the 8 phases, we have just from state variable
constraints 20 × 8 × 13 = 2080, for the takeoff segment we only keep track of the component
throttles, meaning that for the 4 segments we have 400 constraints; adding this with the clearances
and the margins, means that we total 2485 constraints.

Efuel · ηfuel + Ebatt · ϱ

f (Wfuel,block , Wcell mass ) = (8.2)
Enorm
where:

Efuel = Wfuel,block · LHV [total chemical energy stored in fuel] (8.3)

Ebatt = Wcell mass · SpecEnergy [total electrical energy stored in battery] (8.4)
Enorm = normalization energy reference [baseline energy for comparison] (8.5)

For the objective function, the fuel minimization optimization can be generalized with an energy
perspective (Ec.8.2). Where Ef uel is the chemical energy stored in the fuel, ηf uel is the efficiency
from extracting that energy6 , and Ebatt is the energy stored in the battery. This function can perform
fuel minimization for ϱ = 0 and energy minimization7 for ϱ = 1. In this case we will perform
5
This is done to increase the resolution for solving the ODE.
6
In this case a mean efficiency of the gas turbine to convert chemical into mechanical energy, in this case considered
to be ηf uel = 0.3.
7
In this case, we are not minimizing the integral of the power, but rather the weighted masses of battery cells and fuel;
104 CHAPTER 8. KEROSENE PARALLEL HYBRID RESULTS

fuel minimization for the variation of specific energy at a constant range of 758nm, and for the
range-specific energy grid, performing only energy optimization for the grid optimization, in order
to compare the total emissions.

8.2.2 Optimization results

Starting with the optimization results for the fuel minimization, Fig. 8.13 shows the new tendency
for the thermal-managed parallel hybrid architectures. As can be seen, the introduction of thermal
management increases the nominal block fuel consumption, with this increase being about 4.31% for
the low specific energy levels (250 Wh/kg) and about 8.12% for the higher energy levels (700 Wh/kg).
The reason behind this increase is that for the low battery technology, even if the ram air is diverted
and not passed through the main heat exchanger, there are additional drag and parasitic losses that
increase the aircraft’s CD0 . This, combined with the additional coolant lines and heat exchangers
(which, while small, still contribute to the aircraft’s installed mass), means that the aircraft is
operating at a marginally higher M T OW (at these low specific energies).

Figure 8.13: Evolution of battery cell weight and nominal block fuel for various cell’s specific energies
for both the thermal managed and non-thermal managed architectures.

This additional penalty also means that the threshold specific energy (where the optimizer starts
adding batteries to the aircraft) is now also shifted slightly toward higher specific energies with
lower peak battery mass values (about 10.62% lower), which also translates to a marginal MTOW
reduction (3.64% lower). The reason behind this is simple: as the addition of TMS adds additional
constraints and penalties8 to the electrification of the aircraft, the decision technology level shifts
toward higher values with more conservative (higher) battery levels.
Regarding the thermal operation of the powertrain, Fig. 8.15 shows the evolution of component
temperatures and heat exchanger drag throughout the mission flight time. All components operate
the reason for this is that it seems to be more stable.
8
In the sense of an additional TMS cooling loop and ram air requirements.
8.2. THERMAL MANAGED KEROSENE HYBRID 105

Figure 8.14: Evolution of the temperatures of the powertrain components along the mission for the
628 Wh/kg battery cell design at 758nm design range.

within their temperature limits, with most heat exchanger drag occurring during climb conditions9 .
To examine the trends with specific energy more closely, Fig. 8.15 shows the operation of the inlet
ram door and the component heat loads and temperature evolutions.
As observed from the left columns, heat loads increase with specific energy, primarily due
to increased hybridization (as shown in Fig. 8.16, right), which raises the required heat rejection.
Consequently, the inlet ram doors must open wider to increase mass flow through the heat exchanger
(see Fig. 8.15, top right). Interestingly, the highest mass flows occur at lower specific energies due to
the proportionally smaller required heat load, allowing the heat exchanger to be designed with lower
fin density that barely permits components to reach their temperature limits at descent completion
(as seen in the component temperatures for low specific energies in Fig. 8.15).
It is noteworthy that in this architecture, the electric motor operates at the highest temperature,
as it receives coolant that has already passed through the battery and inverters, creating a thermal
bottleneck10 . While not being a bottleneck11 in this architecture, this could motivate the introduction
of an additional heat exchange between the inverter and the electric motor or the coolant mass flow
split between the motor and the inverter.
Regarding hybridization and control of other state variables, it is noteworthy that the introduction
of thermal management does not significantly alter aircraft control strategies. For hybridization, the
9
For this specific design at the given specific energy.
10
This effect is particularly pronounced at higher specific energies where thermal masses become relevant in system
dynamics.
11
In the sense of the optimizer not being able to find a feasible solution.
106 CHAPTER 8. KEROSENE PARALLEL HYBRID RESULTS

Figure 8.15: Evolution of the temperatures of the powertrain components along the mission for each
of the cell specific energies for the Parallel Hybrid.

aircraft maintains higher hybridization during descent at lower specific energies, with the battery
being predominantly discharged during this segment (Fig. 8.16, left). The hybridization gradually
converges to uniform values across climb and cruise phases at higher specific energies. Similarly,
cruise altitude and Mach number exhibit comparable behavior, as the new trajectories obtained in
Fig. 8.17 closely resemble those from Fig. 8.7. This behavior may have numerical origins, as the
TMS system influences the objective function only indirectly through heat exchanger design. For
this specific optimization, hybridization control and component sizing dominate over TMS control
strategy synthesis.
Examining the evolution of battery mass and nominal block fuel, Fig. 8.18 shows a similar trend
8.2. THERMAL MANAGED KEROSENE HYBRID 107

Figure 8.16: Evolution of the state of charge and hybridization factor for the various aircraft specific
energy configurations for the Kerosene Parallel Hybrid with thermal management.

Figure 8.17: Evolution of the altitude and Mach profile for the various aircraft specific energy
configurations for the Kerosene Parallel Hybrid with thermal management.

to that observed in Fig. 8.8, where the threshold value in cell specific energy varies diagonally
along our nominal range - specific energy map. Reductions in nominal block fuel occur once more
primarily at the shorter ranges and higher specific energies, where the aircraft can achieve higher
hybridization factors (see Fig. 8.19).
These regions (characterized by high hybridization and short mission ranges) are particularly
interesting as they reveal one of the limitations of the thermal control parameterization: high-
temperature oscillations. This is especially apparent at 700 Wh/kg and 429 nm (Fig. 8.20), where
thermal constraints initially arise during the mission’s initial climb. High power demands and
elevated ambient air temperatures create a temperature peak at the electric motor that gradually
decreases as the aircraft gains altitude. However, since the electric motor is positioned at the end
of the cooling loop, the initial higher ∆T of the cooled coolant is delayed because the battery and
inverter continue exchanging their thermal mass.
This delay, combined with battery limitations imposed by the initial energy peak at descent
initiation, causes the linear control to bounce between constraints, resulting in high-temperature
loop oscillations. This effect is magnified by the cooling system’s ability to converge with relatively
low coolant thermal mass (which leads to rapid coolant temperature variations and higher changes
108 CHAPTER 8. KEROSENE PARALLEL HYBRID RESULTS

Figure 8.18: Grid of battery cell mass (left) and nominal block fuel (right) as a function of nominal
mission range and cell specific energy level for the Kerosene Parallel Hybrid with thermal manage-
ment.

Figure 8.19: Grid of total energy (left) and hybridization level (right) as a function of nominal mission
range and cell specific energy level for the Kerosene Parallel Hybrid with thermal management.

in the heat exchange rates, which could be undesirable when having off-design performance).
An alternative approach would involve modifying the trajectory through free-form deformation,
which might be able to generate a more tightly controlled logic with a more homogeneous thermal
management12 .
Returning to the grid analysis, Fig. 8.21 shows the MTOW within the design study range, where
similar trends to those observed in Fig. 8.10 are evident: for the aircraft placed at the left of the
battery threshold line the MTOW increases with range, however, for the aircraft placed at the right
of the battery threshold line a big increase in MTOW appears as the battery cell mass is added,
with the MTOW remaining approximately constant. The same pattern emerges for turboshaft and
electric motor component ratings (Fig. 8.22), which exhibit predictably similar behavior to Fig. 8.11:
higher battery installations lead to increased hybridization, which results in a downsizing of the gas
turbine system with an increase in the electric motor rating that slowly provides more percentage
12
An additional way of improving the thermal management could be the introduction of a constraint over the rate of
change of the components, however this was excluded from the optimization in order to avoid artificially stable thermal
systems.
8.2. THERMAL MANAGED KEROSENE HYBRID 109

Figure 8.20: Evolution of the temperatures of the powertrain components along the mission for the
700 Wh/kg battery cell design at 429nm. As it can be seen the variations in hybridization, altitude
and different components being constrained at different points lead to high temperature fluctuations.

of the mission operating power. It is also worth noting that the increase in MTOW drives up the
overall installed power capacity, shifting from peak gas turbine powers of around 2.6MW to peak
electric motor powers in the order of 3MW.
Observing now the overall effect of the introduction of thermal management Fig.8.23 (left) shows
that this TMS results in an overall increase in nominal block fuel (driven mainly due to the additional
TMS drag and the extra TMS systems mass) for all specific energies and ranges. However, if we
observe this in terms of aircraft demanded energy, Fig.8.23 (right) shows that the installation of the
TMS system leads to a decrease in the total demanded energy. This reduction arises from the more
conservative battery sizing strategy, as the penalties introduced by the TMS result in lower installed
battery cell mass.
While we have demonstrated that thermally consistent parallel hybrid aircraft are feasible (given
110 CHAPTER 8. KEROSENE PARALLEL HYBRID RESULTS

Figure 8.21: Aircraft MTOW (left) and increase over 23000 kg baseline (right) vs. mission range and
cell specific energy for the kerosene parallel hybrid with TMS.

Figure 8.22: Turboshaft (left) and electric motor (right) power ratings vs. mission range and cell
specific energy for the kerosene parallel hybrid (thermal managed).

Figure 8.23: Increase in fuel consumption and total energy between the thermal managed parallel
hybrid electric and the non thermal managed parallel hybrid.

appropriate hybridization levels at specific nominal ranges and cell-specific energies) and that they
can achieve reductions in nominal mission fuel requirements, this does not automatically imply
that they are more eco-friendly. The reason is that, although reduced fuel burn directly lowers the
8.2. THERMAL MANAGED KEROSENE HYBRID 111

associated CO2 emissions from kerosene combustion, the additional energy carried in the batteries
is also subject to a certain carbon intensity, depending on how the electricity is generated.
To account for both contributions, the total equivalent CO2 emissions can be expressed as:

CO2,total = Wfuel · EFkerosene + Wbattery · Esp · CIelectricity (8.6)

where:

• Wfuel is the nominal block fuel consumption [kg]

• EFkerosene is the emission factor for kerosene combustion (3.15 kg CO2 /kg kerosene)

• Wbattery is the installed battery cell mass [kg]

• Esp is the battery specific energy [kWh/kg]

• CIelectricity is the carbon intensity of electricity generation [kg CO2 /kWh]

Using this formulation, the equivalent CO2 emissions are computed for the four carbon intensity
levels introduced in Chapter 2 (28, 156, 384, and 708 gCO2 /kWh), and the resulting emissions maps
are presented in Fig. 8.24. As anticipated, the total emissions from kerosene combustion alone are
reduced in the regions where battery cell mass is installed and the nominal block fuel decreases (as
shown in Fig. 8.18). However, as the carbon intensity of electricity increases, the total equivalent
CO2 emissions gradually converge towards a profile resembling the total aircraft demanded energy
(Fig. 8.19).
Here, the main elephant in the room becomes evident: along each constant-range horizontal line,
for carbon intensities of 28 and 156 gCO2 /kWh, increasing the cell specific energy (and thus entering
the region of battery mass installation) still results in a net decrease in total emissions, demonstrating
that battery installation is effective under low-carbon electricity scenarios. In contrast, for carbon
intensities of 384 and 708 gCO2 /kWh, the total emissions actually increase. This occurs because the
hybridized aircraft, which is sized to minimize nominal block fuel, demands more overall energy,
and the additional battery energy (depending on the carbon intensity of electricity) may carry a
higher equivalent CO2 that offsets, or even exceeds, the reductions achieved by lower fuel burn.
Taking as an example the aircraft sized at 700 Wh/kg and 943 nm (the case with the highest
energy demand, as shown in Fig. 8.19), the CO2 emissions from kerosene combustion amount to
5663 kg of CO2 (about 2.31% lower than the corresponding value at 250 Wh/kg, where no battery is
installed). However, when accounting for the carbon intensity of electricity, the total equivalent
emissions rise to 5742, 6060, 6718, and 7606 kg of CO2 for carbon intensities of 28, 156, 384, and
708 gCO2 /kWh, respectively. Compared to the aircraft sized at 250 Wh/kg, this corresponds to
changes of −1.82%, +3.57%, +14.81%, and +29.98% in total equivalent carbon emissions for the
respective carbon intensities.
For more favorable13 shorter ranges, such as 300 nm, we find that at 700 Wh/kg the fuel-related
carbon emissions are 231 kg of CO2 (an 87.39% reduction in fuel-related emissions compared to the
13
In terms of the increase in battery specific energy leading to a significant decrease in nominal block fuel.
112 CHAPTER 8. KEROSENE PARALLEL HYBRID RESULTS

Figure 8.24: Emissions of CO2 from the burnt (top), and equivalent total emissions from the energy
production for the electric batteries considering the three energy mix carbon intensity scenarios
(mid and lower rows).

aircraft sized at the same range but with 250 Wh/kg). However, when accounting for the carbon
intensity of electricity, the total equivalent emissions rise to 360, 893, 1954, and 3410 kg of CO2 for
carbon intensities of 28, 156, 384, and 708 gCO2 /kWh, respectively. Compared to the aircraft sized
at 250 Wh/kg, this corresponds to changes of −80.39%, −51.26%, +6.65%, and +86.14% in total
equivalent carbon emissions for the respective carbon intensities. As expected, the total emissions
are highly sensitive to the carbon intensity of electricity, leading to very high decreases under clean
electricity and very increases under carbon-intensive electricity.
8.2. THERMAL MANAGED KEROSENE HYBRID 113

Range Carbon Intensity Emissions [kg] Change [%]

28 gCO2 /kWh 5742 −1.82
156 gCO2 /kWh 6060 +3.57
943 nm
384 gCO2 /kWh 6718 +14.81
708 gCO2 /kWh 7606 +29.98
28 gCO2 /kWh 360 −80.39
156 gCO2 /kWh 893 −51.26
300 nm
384 gCO2 /kWh 1954 +6.65
708 gCO2 /kWh 3410 +86.14

Table 8.3: Total equivalent CO2 emissions for two study cases (943 nm and 300 nm) at 700 Wh/kg,
for different electricity carbon intensities. The relative change (%) is given with respect to the
corresponding 250 Wh/kg baseline aircraft.

Figure 8.25: Total energy consumed for the energy-focused kerosene hybrid optimization (left) and
comparison with the fuel minimization optimization results.

Figure 8.26: Installed battery mass for the energy-focused kerosene hybrid optimization (left) and
comparison with the fuel minimization optimization results.

To validate that this is not simply an ill-defined optimization problem, the range vs. specific
energy study is repeated, this time not for energy minimization (ϱ = 1). The resulting energy
distribution across the design space is shown in Fig. 8.25 (left). Unlike before, the previously
114 CHAPTER 8. KEROSENE PARALLEL HYBRID RESULTS

observed peak in energy demand at the right side of the design space disappears, yielding a more
homogeneous energy demand profile that is mainly driven by the increase in total energy with range.
This reduction in energy demand originates primarily from the removal of most of the battery cell
mass (Fig. 8.26), which in the previous case had added significant additional energy requirements.
Battery installation is now limited to the short-range, high-specific-energy region, where its benefits
remain the most pronounced.

Figure 8.27: Consumed fuel for the energy-focused kerosene hybrid optimization (left) and compari-
son with the fuel minimization optimization results.

However, while we are now minimizing the energy demand, it must be emphasized that this
does not correspond to the solution that minimizes fuel consumption. As shown in Fig. 8.27, the
reduction in nominal block fuel is now almost negligible, and in fact, the nominal block fuel increases
in the regions where the batteries have been removed. This outcome validates the earlier hypothesis:
fuel-minimizing hybrid aircraft can indeed reduce the consumed fuel, but at the expense of a much
higher total energy demand. When the optimization instead targets energy minimization, this higher
demand is avoided, leading to the near-complete removal of the electric batteries.

Carbon Intensity Emissions [kg] Change [%]

Fuel only 1599 −18.98
28 gCO2 /kWh 1623 −17.73
156 gCO2 /kWh 1750 −11.30
384 gCO2 /kWh 2001 +1.42
708 gCO2 /kWh 2344 +18.81

Table 8.4: Total equivalent CO2 emissions for the energy-optimized aircraft at 300 nm and 700 Wh/kg,
for different electricity carbon intensities. The relative change (%) is with respect to the 250 Wh/kg
baseline aircraft.

To assess whether this new energy-optimized aircraft is more or less eco-friendly, the analysis of
the total equivalent carbon emissions (Eq. 8.6) is repeated under the same carbon intensity scenarios.
The resulting emissions are shown in Fig. 8.28, where it can be observed that, as with the total energy
and the nominal block fuel, the main driver for the increase in total emissions is the nominal mission
range. To illustrate the effect of the carbon intensity, consider once more the case at 300 nm and
8.2. THERMAL MANAGED KEROSENE HYBRID 115

700 Wh/kg. Here, the fuel-related emissions amount to 1599 kg of CO2 (an 18.98% decrease compared
to the corresponding aircraft sized at 250 Wh/kg). When accounting for electricity production, the
total equivalent emissions increase to 1623, 1750, 2001, and 2344 kg of CO2 for carbon intensities of
28, 156, 384, and 708 gCO2 /kWh, respectively. These correspond to variations of −17.73%, −11.30%,
+1.42%, and +18.81% relative to the 250 Wh/kg baseline. The data from this study is summarized
in Table 8.4.

Figure 8.28: Emissions of CO2 from the burnt kerosene (top), and equivalent total emissions from
the energy production for the electric batteries considering the three energy mix carbon intensity
scenarios (mid and lower rows). Made for the energy efficient aicraft.

Comparing the total emissions between the energy- and fuel-optimized parallel hybrids yields
116 CHAPTER 8. KEROSENE PARALLEL HYBRID RESULTS

Fig. 8.29, which shows the relative change in total CO2 emissions of the fuel-minimizing design with
respect to the energy-minimizing one. As expected, for low carbon intensities (28 and 156 gCO2 /kWh)
the fuel-minimizing aircraft achieves lower total carbon emissions. However, as the carbon intensity
increases, aircraft designed near the battery cell technology threshold begin to exhibit higher total
emissions, with the difference reaching over 55% at high carbon intensities14 .

Figure 8.29: Comparison of the relative decrease in total CO2 emissions of the fuel minimization
with respect to the energy minimization.

This dependency of total emissions on carbon intensity highlights the complex nature of hybrid
aircraft architectures, reinforcing the point already anticipated: hybrid aircraft design is a regional
problem. In practice, while it is possible to design a feasible aircraft at a given range and specific
energy, its actual effectiveness depends on its position in the battery cell technology map and,
critically, on the carbon intensity of the electricity used. Exploiting reductions in total emissions
therefore depends on how close the energy source is to being zero-emissions. As will continue to be
discussed in the following chapters, it can already be stated that the decarbonization of the skies
begins with the decarbonization of the ground. In other words, reductions in equivalent carbon
emissions, and the introduction of hybrid aircraft architectures, are only meaningful in countries
with sufficiently clean energy production infrastructures.

14
At 528 Wh/kg and 300 nm.
 CHAPTER

HYDROGEN PARALLEL
HYBRID RESULTS 9
Contents
9.1 Thermal managed hydrogen hybrid . . . . . . . . . . . . . . . . . . . . . . . 117
9.1.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
9.1.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

This chapter presents the results of the optimization study for the hydrogen parallel hybrid
with direct gas turbine combustion. The main objective is to reassess how system performance
evolves with an increase in battery specific energy, the impact of optimizing for fuel or energy, and
its efficiency with respect to the kerosene hybrids.

9.1 THERMAL MANAGED HYDROGEN HYBRID

9.1.1 Optimization statement

For the optimization statement, we retain the previous energy multi-objective function and perform
two studies: one with ϱ = 0 to minimize fuel consumption (using the LHV of hydrogen) and one
with ϱ = 1 to minimize total energy. Although direct hydrogen combustion does not produce CO2
emissions, the motivation for fuel minimization lies in its NOx emissions from nitrogen thermal
reactions, which remain an environmental concern.1

Efuel · ηfuel + Ebatt · ϱ

f (Wfuel,block , Wcell mass ) = (9.1)
Enorm
where:

Efuel = Wfuel,block · LHV [total chemical energy stored in fuel] (9.2)

Ebatt = Wcell mass · SpecEnergy [total electrical energy stored in battery] (9.3)
Enorm = normalization energy reference [baseline energy for comparison] (9.4)
1
Moreover, as will be shown later, hydrogen production itself has an associated carbon footprint that also influences
the trends in total equivalent CO2 emissions.

117
118 CHAPTER 9. HYDROGEN PARALLEL HYBRID RESULTS

For the design variables and constraints, the design space from Tab. 8.2 is expanded to include
both the hydrogen tank design variables (tank capacity and venting and heating profiles) and their
operational constraints (operational pressure above and below ambient). The expanded design space
is shown in Tab.9.2.

Design variable (units) Lower Upper Constraint (units) Lower Upper

MTOW (kg) 18000 100000 δ gas turbine (-) — 0.9/1

Prop Mtang (-) 0.4 0.9 δ gearbox (-) — 0.9/1
Prop diam (m) 2.5 5 δ inveter (-) — 0.9/1
Prop factor (-) 0.76 1.74 δ electric motor (-) — 0.9/1
Wing ndpos (-) 0.2 0.55 δ battery (-) — 0.9/1
Wing Sref (m2 ) 30 90 Battery SOC (-) 0.2 1
Wing AR (-) 8 22 Propeller Cp (-) — 3
HTP Sref (m2 ) 5 50 Propeller Mtrue (-) — 0.75
VTP Sref (m2 ) 5 50 Static margin (-) 0.35 -
Turboshaft rating (MW) 0.02 200 Mass margin (m) 0.01 —
Gearbox rating (MW) 0.02 200 Mean velocity (m/s) 160 —
Electric motor rating (MW) 0.02 200 BFL (m) — 1400
Inverter rating (MW) 0.02 200 Inverter Temp (ºC) - 70
Battery cell mass (kg) 0 200000 Emotor Temp (ºC) - 70
Coolant pump (kW) 0 200000 Battery Temp (ºC) 10 60
˙
mcoolant (kg/s) 0.01 100 F2F HX k design margin 0 -
Reservoir coolant (kg) 10 1000 F2F HX width clear 0 -
Coolant hose diameter (m) 0.01 10 F2F HX height clear 0 -
InvTMS HeightRatio (-) 0.1 1 ∆ Tank Pressure (bar) 0 2
InvTMS nparallel (-) 1 100 Rank SOF (-) 0.05 1
BattTMS tchannel (m) 0.001 0.1
F2F HX nlong,c (-) 1 1000
F2F HX nwide,c (-) 1 1000
F2F HX ntall (-) 1 1000
F2F HX design k (-) 1 1000
F2F HX multiplierc (-) 0.1 1000
F2F HX multiplierh (-) 0.1 1000
LH2 mass (kg) 1 10000
LH2 press multiplier (-) 1 10000
hcruise (m) 1000 7000
EASstart (m/s) 1 300
EASmid (m/s) 1 300
EASend (m/s) 1 300
γstart (deg) -10 10

Continued on next page

9.1. THERMAL MANAGED HYDROGEN HYBRID 119

Design variable (units) Lower Upper Constraint (units) Lower Upper

γend (deg) -10 10

M tipstart (-) -10 10
M tipend (-) 0.8 1.2
HFstart (-) 0 1
HFend (-) 0 1
Ramopen,start (-) 0.01 1
Ramopen,end (-) 0.01 1
Battheating,start (kW) 0 1000
Battheating,end (kW) 0 1000
T ankventing,start (kg/s) 0 1000
T ankventing,end (kg/s) 0 1000
T ankheating,start (kW) 0 1000
T ankheating,end (kW) 0 1000

Table 9.2: Design variables and bounds (left), and associated constraints with limits (right). Values in
salmon denote variables defined per mission segment, values in pink represent constraints enforced
over a vector of mission segments, values in orange refer to variables or constraints defined separately
for each mission failure mode.

9.1.2 Optimization results

Starting with the optimization results, Fig. 9.1 presents the installed battery cell weight and the
nominal block fuel2 as functions of the cell specific energy. The observed trends closely resemble
those previously discussed in Fig. 8.1 and Fig. 8.13. For low values of cell specific energy, no batteries
are installed; however, once the cell specific energy exceeds a certain threshold (slightly beyond
500Wh/kg), the installed battery mass increases sharply. For the case of the fuel mass consumed we
can see that is approximately one third3 of the amount that we had for the kerosene architecture.
Comparing the Kerosene PH-TMS configuration (see Sec. 8.2) with the LH2 PH-TMS, it can be
observed that the installed battery mass at a specific energy of 700 Wh/kg is slightly higher in the
latter case, increasing from 5231 kg to 5779 kg (a 10.47% increase). This is accompanied by a greater
reduction in fuel consumption: 13.5% for the Kerosene PH-TMS compared to 16.4% for the hydrogen
PH-TMS. At first glance, this result may appear counterintuitive, as one might expect the kerosene
configuration to benefit more from battery technology, given that each kilogram of battery displaces
more kilograms of kerosene than of hydrogen. However, this reasoning neglects the impact of the
extended mission. Assuming the battery is fully depleted during the nominal mission4 , the hydrogen
architecture requires less additional fuel mass than the kerosene configuration to supply the same
extra energy for the extended mission.
2
In this context, hydrogen is still referred to as a fuel.
3
Owing to the energy density of hydrogen being 120 MJ/kg, compared to the 43 MJ/kg of kerosene.
4
This has already been proven for the kerosene architecture, but will also be checked for the hydrogen one.
120 CHAPTER 9. HYDROGEN PARALLEL HYBRID RESULTS

Figure 9.1: Evolution of battery cell weight and nominal block fuel for various cell’s specific energies
along the geometries for three technology levels. Batteries are highlighted in green with the inverter
and the electric motor in orange and yellow, respectively. For the hydrogen parallel hybrid with
thermal management.

Figure 9.2: Evolution of the installed battery cell mass (left) and nominal block fuel (right) for the
kerosene and LH2 architectures for both nominal only and nominal + extended. And both considering
TMS. Shaded areas represent parts where the extended mission is penalizing the addition of batteries.

To validate this reasoning, the extended mission segment can be disabled and the sweep analysis
recomputed, yielding the results shown in Fig. 9.2. The optimizations considering only the nominal
mission (shown as dashed lines) exhibit overall higher installed battery cell masses, converging to
approximately the same maximum value at 700 Wh/kg. The threshold specific energy is also lower
in this case, with a smoother gradient (i.e., a less abrupt change). This behaviour arises because,
9.1. THERMAL MANAGED HYDROGEN HYBRID 121

when the extended mission is not considered, batteries can be added and fully depleted during the
nominal mission without incurring the penalty of carrying them into the extended mission, where
they would require additional fuel to transport. Conversely, when the extended mission is included,
the technology must not only offset the penalty of additional batteries increasing the nominal
mission fuel requirement, but also mitigate the further penalty of increased fuel requirements for
the extended mission.
Since the additional kerosene mass required is greater than the additional hydrogen mass, the
overall reduction in installed battery cell mass from the nominal optimum is larger for the kerosene
architecture than for the hydrogen configuration. However, it can also be observed that, even though
the high-specific-energy optima are lower for the kerosene architecture, the installed battery masses
for specific energies between 400 Wh/kg and 500 Wh/kg are higher for kerosene than for hydrogen5 .
This is consistent with our earlier assumption that kerosene benefits more from the removal of
fuel mass, which also explains why its installation curve is less steep compared to hydrogen. In
the hydrogen case, where the benefit per kilogram of battery is lower, a higher-performing battery
technology is required to compensate for the additional battery mass; thus, as the fuel energy density
increases, the threshold jump becomes steeper and shifts towards higher specific energy values.

Figure 9.3: Evolution of component mass with cell-specific energy. Structural mass includes wing,
fuselage, and tail weight along with subsystems, furnishing, and operational items. Powerplant
includes turboshaft, gearbox, electric motors, inverters, battery tank and TMS systems. Battery mass
is expressed at pack level, and payload includes passenger and luggage mass. For the Hydrogen
Parallel Hybrid with TMS.

Looking again at the evolution of the structural mass (Fig. 9.3), we observe that as the battery cell
specific energy increases, the overall cumulative mass also increases, reaching values very similar to
those obtained for the kerosene parallel hybrid configuration (Fig. 8.21). Although one might expect
the M T OW to be lower due to the higher specific energy density of hydrogen, it must be noted
5
This is true for both the nominal-only and the complete mission cases.
122 CHAPTER 9. HYDROGEN PARALLEL HYBRID RESULTS

that the hydrogen architecture requires the additional mass of the hydrogen tanks and a longer
fuselage, which, in this case, offsets the benefit of the lower fuel mass.
As a result, the trends in energy consumption (Fig. 9.4) are also comparable to those shown in Fig. 8.3.
The additional installed batteries increase the aircraft mass, thereby raising the total energy demand.
With high battery specific energies (700 Wh/kg), it is again possible to supply approximately half of
the total mission energy from the batteries, albeit at a higher total energy requirement.

Figure 9.4: Energy provided by the battery and by the hydrogen (left). Energy intensity from
the battery, hydrogen, and total energy (right). For the Hydrogen Parallel Hybrid with thermal
management.

Figure 9.5 shows that the hybridization and battery discharge profile is similar to that in Fig. 8.16.
The battery is initially discharged during the descent segment, followed by a gradual additional
discharge during climb, and finally discharged evenly across the climb and cruise segments, with
minimal discharge during descent6 . This leads to the same tendency observed previously: as the
powertrain relies more heavily on the battery, the optimizer reduces the gas turbine rating and
increases the rating of the electric motor and inverter (Fig. 9.6).

Figure 9.5: Evolution of the state of charge and hybridization factor for the various aircraft specific
energy configurations for the Hydrogen direct combustion with TMS.

6
This is because the turboshaft engine design is the same, and therefore the P SF C trends already studied for the
Kerosene PH NoTMS configuration remain valid.
9.1. THERMAL MANAGED HYDROGEN HYBRID 123

Figure 9.6: Electric motor and gas turbine power ratings vs. cell specific energy for LH2 PH-TMS

Figure 9.7: Temperature evolution of powertrain components and for the 628 Wh/kg battery at 758
nm design range for LH2 PH-TMS.

In the context of thermal management and hydrogen pressure control, Fig. 9.7 shows the evolution
of the component temperatures throughout the mission. In this case, the main thermal constraint is
124 CHAPTER 9. HYDROGEN PARALLEL HYBRID RESULTS

imposed by the electric motor7 , which reaches its allowable limit during the initial climb segment.
This behaviour dictates the control of the ram door aperture: it opens fully at the start of the
climb, then progressively closes as the aircraft gains altitude, the power demand decreases, and the
freestream temperature increases. During the remainder of the climb segment, the ram door remains
open just enough to maintain the battery within its thermal limit.

Figure 9.8: Evolution of the temperatures of the powertrain components along the mission for each
of the cell specific energies for the LH2 Parallel Hybrid.

The temperature profiles for the full sweep of specific energies are presented in Fig. 9.8. The trends
are very similar to those in Fig. 8.15: as the cell specific energy increases, the heat load from the
electric components also increases, requiring greater ram air duct opening. The higher specific
energies also result in increased cruise hybridization levels and additional battery discharge during
7
Once more due to it taking the additional heat removed from the battery and inverter.
9.1. THERMAL MANAGED HYDROGEN HYBRID 125

the cruise segment (see Fig. 9.5), leading to thermal constraints not only in the climb segment but
also in the cruise phase. For the control of the hydrogen tank Fig.9.9 shows the evolution of the
state of fill of the hydrogen tank and the pressure evolution of the pressure along the mission for
each specific energy, what can be observed is that for all cases we can observe an initial decrease in
tank pressure due to the heavier hydrogen consumption from the climb, with then a rise in tank
pressure; as it will be seen in the fuel cell architecture this control trend is not universal and what
mostly matters for our case is that the pressure is within the prescribed limits.

Figure 9.9: Evolution of the temperatures of the powertrain components along the mission for each
of the cell specific energies for the LH2 Parallel Hybrid.

Looking now onto the evolution of the design along our battery technology - range design space,
Fig.9.10 shows the evolution of the installed battery cell mass and nominal block fuel. As we can see
here, there is also a technology threshold that presents itself as a diagonal along our design space,
with the design space left of the diagonal showing little battery installation, and the design space
right of the diagonal bumping the installed battery power, leading to a decrease in nominal block
fuel as we move to the right (higher specific energies) and to the bottom (lower design ranges) of
the design space.

Figure 9.10: Evolution of the installed cell battery mass (left) and nominal block fuel (right) along
our range-cell specific energy design space for the PH-TMS LH2 regional aicraft.

This, however, is achieved at a significantly higher M T OW (Fig. 9.11) due to the additional battery
126 CHAPTER 9. HYDROGEN PARALLEL HYBRID RESULTS

mass, which in turn increases the required aircraft structural mass. As noted earlier, and as can
be confirmed by comparing the M T OW values in Fig. 9.11 with those in Fig. 8.21, the aircraft is
designed to a very similar M T OW . While it would have been reasonable to assume that the higher
energy density of hydrogen would yield an overall lighter aircraft, the slightly greater installed cell
mass, combined with the need for an enlightened fuselage, and the additional penalty from the tank
structural mass, means that the initial lightening is offset by the additional structural8 mass.

Figure 9.11: Evolution of the M T OW (left) and relative increase with respect to an aircraft with an
M T OW of 23 tons (right) along our range-cell specific energy design space for the PH-TMS LH2
regional aicraft.

To assess how this tank structural mass penalises the design across the design space, we define
a nominal gravimetric efficiency, ζnom , and a total gravimetric efficiency, ζtot , as given in Eq. 9.5,
where WLH2 ,nom is the hydrogen consumed in the nominal mission, WLH2 ,tot is the total hydrogen
contained in the tank, and WLH2 ,struct is the total structural mass of the tank.

WLH2 ,tot WLH2 ,nom

ζtot = ζnom = (9.5)
WLH2 ,tot + WLH2 ,struct WLH2 ,tot + WLH2 ,struct
Evaluating these efficiencies over the design space yields the results shown in Fig. 9.12. For the total
gravimetric efficiency—which measures the efficiency of the tank design for storing the total amount
of hydrogen—we obtain ζtot ≈ 0.56 for short ranges (300 nm). This value decreases with increasing
nominal range, reaching ζtot ≈ 0.44 for a 1200 nm range. The drop in efficiency occurs because the
aircraft uses a single tank; as the total hydrogen requirement increases, the tank geometry transitions
from more spherical to more cylindrical, reducing the load-bearing contribution of the spherical
section relative to the cylindrical portion, and thus increasing the mean tank wall thickness.
Overall, due to the relatively similar design M T OW and the marginally higher battery cell mass,
the total mission energy9 (Fig. 9.13, left) remains approximately equal for both configurations. The
highest total energy occurs at 700 Wh/kg and a nominal range of 943 nm, where the Kerosene
PH-TMS consumes 32 123 MJ and the LH2 PH-TMS consumes 32 861 MJ.
8
Fuselage and tank.
9
Computed by summing the integrated energy from the turboshaft effective output power and the battery electrical
output power.
9.1. THERMAL MANAGED HYDROGEN HYBRID 127

Figure 9.12: Evolution of the total gravimetric efficiency (left) and nominal gabrimetric efficiency
(right) along our range-cell specific energy design space for the PH-TMS LH2 regional aicraft.

The total fraction of energy supplied by the electrical system10 is shown in Fig. 9.13 (right). Hybridiza-
tion is concentrated within the technology-threshold battery installation region, where the installed
battery mass is similar for both configurations. Moving right in the figure (higher specific energy)
increases the installed battery energy, while moving downward (lower nominal range) increases the
fraction of energy supplied electrically, as the total nominal energy requirement decreases. This
combination enables fully electric operation during the nominal mission for high specific energy
values and low nominal ranges, showing once more the motivation of these hybrid architectures for
the shorter design ranges.

Figure 9.13: Evolution of the total consumed energy (left) and fraction from it that comes from
the battery (right) along our range-cell specific energy design space for the PH-TMS LH2 regional
aicraft.

This behaviour also explains the observed powertrain design trends: as hybridization increases,
installed turboshaft power is progressively traded for electric motor power (Fig. 9.14). It should
be noted, however, that higher hybridization strategies also lead higher total installed power due
to the increase in M T OW . For example, at 300 Wh/kg and a nominal range of 942 nm (almost
10
Analogous to the hybridization factor.
128 CHAPTER 9. HYDROGEN PARALLEL HYBRID RESULTS

Figure 9.14: Evolution of the tuboshaft power rating of each engine (left) electric motor power rating
from each engine (right) along our range-cell specific energy design space for the PH-TMS LH2
regional aicraft.

no installed battery power), the turboshaft power rating is approximately 2.4 MW with no electric
motor installed. In contrast, for high hybridization cases (700 Wh/kg and a nominal range of 300 nm),
the turboshaft power rating decreases to approximately 1.5 MW, while the electric motor power
rating increases to approximately 3 MW. This is because, even in a nominally electrified mission,
the aircraft must retain sufficient turboshaft capability to supply power by means of burning fuel at
the extended mission.
Regarding the climatic impact of the PH LH2 architectures, Fig. 9.15 presents the equivalent CO2
emissions for the LH2 PH-TMS configurations across the specific energy–range design space. The
results are shown for two hydrogen production pathways: natural gas reforming (left column) and
electrolysis (right column); and under various electricity carbon intensity scenarios: 28, 146, and 384
gCO2 /kWh (in each of the rows).
For the natural gas reforming case, the carbon emissions map is qualitatively similar to that obtained
for the Kerosene PH-TMS. In the region without battery installation, emissions increase primarily
with range and remain nearly constant across the different electricity carbon intensity scenarios.
Variations appear predominantly in the battery-installed region, where, at low electricity carbon
intensities (28 gCO2 /kWh), the total emissions are reduced in the lower-right portion of the design
space. As the electricity carbon intensity increases, this trend disappears, with the area with installed
batteries having slightly greater total carbon emissions. This means that for Natural Gas Reforming,
where the emissions associated with the production of hydrogen are fixed11 , installing batteries
would make sense if we have a good energy grid, as if the electricity has a high carbon intensity, the
overall higher energy demand of a hybrid aircraft will add to the total emissions.
While there will be a chapter reserved for comparisons across architectures, we can do some initial
assessments between hydrogen and kerosene. Starting with a point with little to no hybridization
(300Wh/kg at 1200nm) we can see that the carbon emitted due to the burn of kerosene accounts for
7701 kg of CO2 , while for the direct hydrogen combustion, the carbon associated with the reforming

11
Same as we had with the kerosene usage, which had a fixed rate per kilogram burnt.
9.1. THERMAL MANAGED HYDROGEN HYBRID 129

Figure 9.15: Comparison of the equivalent CO2 emissions for the various LH2 PH architectures
under various electricity and hydrogen production scenarios.

of hydrogen accounts for a total of 9923 kg of CO2 (about a +28.85% increase over the kerosene
architecture). The reason for this is that while kerosene is considered to emit 3.15 kg of CO2 per
kg of kerosene burnt, hydrogen reforming is considered to emit 11.4 kg of CO2 per kg of hydrogen
produced. This means that even if we adjust for their specific energy ratios, reforming hydrogen is
+29.71% carbon-intensive compared to burning kerosene.
Looking now into a higher hybridization design (the one designed at 700Wh/kg and 300nm) we
can see that the total emissions at our three carbon intensity scenarios (28, 146, and 384 gCO2 per
kWh) are 201, 757, and 1883 kg of equivalent CO2 . This corresponds to a relative change of -44.16%,
-10.21%, and -3.63% in the total emissions (further details are arranged in Tab.10.3). The main reason
for this architecture having better performance (even though the higher carbon intensity of carbon
130 CHAPTER 9. HYDROGEN PARALLEL HYBRID RESULTS

reforming) corresponds to the higher installed battery masses that we were able to attain in the
hydrogen architecture12 . However, as we can see, this benefit decreases rapidly with the carbon
intensity of the produced energy, as the higher installed battery hydrogen architecture is more
sensitive to the grid intensity.

Carbon Intensity Emissions PH-Kero [kg] Emissions PH-LH2 [kg] Change [%]
28 gCO2 /kWh 360 201 −44.16
156 gCO2 /kWh 893 757 −10.21
384 gCO2 /kWh 1954 1883 −3.63

Table 9.3: Comparison of total equivalent CO2 emissions for the higher-hybridization design
(700 Wh/kg, 300 nm) against the kerosene-only case, under different electricity carbon intensi-
ties.

Looking now onto the production of hydrogen by means of electrolysis, Fig. 9.15 (right) shows the
total equivalent CO2 emissions for the three carbon intensity scenarios. As can be seen, the total
carbon emissions are heavily driven by the carbon intensity of the electricity used to generate both
the hydrogen and to recharge the batteries. Taking some reference values, we can check the aircraft
sized at 700 Wh/kg and 300nm, where we have that the total emissions are 140, 730, and 1921 kg of
equivalent CO2 emissions, or about a change in -30.35%, -2.35%, and +2.02% in the total equivanet
CO2 emissions.

Carbon Intensity Emissions Ref [kg] Emissions Elect [kg] Change [%]
28 gCO2 /kWh 201 140 −30.35
156 gCO2 /kWh 757 730 −2.35
384 gCO2 /kWh 1883 1921 +2.02

Table 9.4: Comparison of total equivalent CO2 emissions for the higher-hybridization design
(700 Wh/kg, 300 nm) between hydrogen production by means of reforming and electrolysis

Carbon Intensity Emissions Ref [kg] Emissions Elect [kg] Change [%]
28 gCO2 /kWh 9923 1132 −88.59
156 gCO2 /kWh 9923 5910 −40.44
384 gCO2 /kWh 9923 15541 +56.62

Table 9.5: Comparison of total equivalent CO2 emissions for the higher-hybridization design
(300 Wh/kg, 1200 nm) between hydrogen production by means of reforming and electrolysis

Looking now at the behavior at the higher ranges and the lower specific energies (300Wh/kg and
1200nm), we get that the value of total emissions from the natural gas reforming stays the same (as
here the emissions come solely from the production of hydrogen), while the emissions for electrolysis
at the three carbon intensities are of 1132, 5910 and 15541 kg of CO2 , which account for reductions
of -88.59%, -40.44% and +56.62% in the total equivalent emissions. Here we can see what was already
12
Let us recall, due to the lesser comparative extra mass needed for the extended mission frown on fuel.
9.1. THERMAL MANAGED HYDROGEN HYBRID 131

to be shown in the shorter ranges (and higher hybridized) aircraft, electrolysis can drastically reduce
the emissions both with and without electrification, but this reduction is subject to a threshold value,
as if the carbon intensity resulting from generating the hydrogen by electrolysis is higher than the
one from the reforming process; doing the maths with the referenced 47.57 kWh/kg LH2 reveals
that in order to have a lower carbon intensity than gas reforming, the carbon intensity must be
beneath 239.64 gCO2 /kWh. The reason why this impact affects less to the higher hybrid aircraft
(the ones that we studied at Tab.9.4) is that the gas turbine is not completely efficient, and thus we
only get a fraction of the energy employed to produce the hydrogen, compared to this, batteries are
almost completely efficient, and we can extract almost all of the grid energy. This means that for the
electrolysis of hydrogen, we end up needing to spend more kWh (over three times more, assuming a
standard efficiency of a gas turbine) to get the same MJ of energy as we would get in a standard
battery. This is why even if the total energy required by our hydrogen aircraft is bigger (as seen in
Fig.9.13), this does not translate as badly into the total emissions (as per each unit of effective shaft
energy from the hydrogen, we had to spend more than three times that energy into producing the
hydrogen, and as such fuel minimization is still relevant).

Figure 9.16: Total energy demand (left) and relative increase in energy consumption (right) for the
LH2 PH TMS architecture optimized for minimal energy compared to the one for minimal fuel.

Figure 9.17: Installed battery cell mass (left) and installed delta (right) for the LH2 PH TMS architec-
ture optimized for minimal energy compared to the one for minimal fuel.
132 CHAPTER 9. HYDROGEN PARALLEL HYBRID RESULTS

Focusing now on the optimization for energy minimization (ϱ = 1), the total energy demand is
shown in Fig. 9.16. The results indicate substantial reductions in energy demand13 , with the energy
map increasing only as a function of nominal range. This behaviour arises because, owing to
the higher energy density of hydrogen, the optimizer avoids installing batteries altogether14 , as
illustrated in Fig. 9.17.
Again, this does not mean that what we were doing before was wrong, as if we see now the fuel
consumed and the fuel delta with respect to the fuel minimization, we can see that we were able
to minimize the fuel consumed during the nominal mission segment Fig.9.18. However, a more
energy-intensive aircraft.

Figure 9.18: Nomial block fuel (left) and nominal block fuel change (right) for the LH2 PH TMS
architecture optimized for minimal energy compared to the one for minimal fuel.

To evaluate if that aircraft was or was not more polluting, we can repeat the equivalent CO2 emissions
map from before, getting Fig.9.19. As we could have expected, we find that the emissions are now
practically constant for the natural gas reforming architectures, with range as the main driver for
the increase in total equivalent CO2 emissions. For the case of electrolysis, we can observe once
more that the carbon emissions increase with the carbon intensity of the produced energy.
To compare if doing fuel minimization was or was not a good idea, we can make a comparison as
we did for kerosene in Fig.8.29, obtaining Fig.9.20. Here we can observe that for hydrogen produced
by reforming natural gas, we get a very similar result to the one obtained for the kerosene, where
the highest impact (in terms of decreasing the total CO2 emissions) of adding batteries occurs at the
low-ranges high-specific energies, where at low carbon intensities (28 gCO2 /kWh) we get that the
fuel minimized aircraft has emissions up to 91.9% lower as compared with the energy optimized
aircraft (both at 700Wh/kg and 300nm). This reduction, however, decreases as we increase the
carbon intensity, with 146 gCO2 /kWh having a reduction of only -68.8% in equivalent total CO2
emissions, and at 384 gCO2 /kWh having a reduction of only -22.9%. This reduction in the reduction
of the total emissions is mainly due to the emissions from electricity production getting slowly
13
At shaft power, this calculation keeps the integral energy calculation that, as we will have seen and we will see later,
might not mean a reduction in the total energy needed to produce the hydrogen and recharge the batteries.
14
While the kerosene architecture exhibited marginal battery installation, the higher energy density of hydrogen
reinforces the fact that each kilogram of battery displaces less hydrogen mass than kerosene mass.
9.1. THERMAL MANAGED HYDROGEN HYBRID 133

Figure 9.19: Comparison of the equivalent CO2 emissions for the various LH2 PH architectures under
various electricity and hydrogen production scenarios for the energy minimization optimization.

closer to the emissions from the production of hydrogen, slowly offsetting the benefit of the addition
of batteries.
A more interesting point to study is the transition between the areas of low battery installation and
high battery installation (what we call the battery threshold line). Taking the design point at 528
Wh/kg and 685 nm we get that the decreases in total equivalent CO2 are now -6.9%, -1.2%, and +10.1%,
for each of the three carbon intensity scenarios, respectively. The reason or this change is that at
this point, the installation of batteries is not as efficient, and while the required hydrogen is reduced,
the higher energy required to carry the additional batteries makes it so that the battery recharge
emissions quickly reach the hydrogen production. This shows that the decrease of total emissions
is a matter of when (in terms of the design range that we want, the specific energy technology
134 CHAPTER 9. HYDROGEN PARALLEL HYBRID RESULTS

Figure 9.20: Comparison of the change in the relative equivalent CO2 between the fuel-minimization
and energy-minimization optimizations for the Hydrogen PH-TMS architecture.

level that we have reached, and the carbon intensity our grid is operating at). As a result, if we are
producing hydrogen by reforming, and our energy grid has a high carbon intensity, we might need
to wait till higher specific energies, even if we have already reached the level where we can start
fuel minimizing.

Design Point 28 gCO2 /kWh 156 gCO2 /kWh 384 gCO2 /kWh
700 Wh/kg, 300 nm −91.9% −68.8% −22.9%
528 Wh/kg, 685 nm −6.9% −1.2% +10.1%

Table 9.6: Relative change in total equivalent CO2 emissions (fuel-minimized vs. energy-optimized
aircraft) for hydrogen produced by reforming natural gas, at two representative design points.
9.1. THERMAL MANAGED HYDROGEN HYBRID 135

Looking now at the results from the production of hydrogen by electrolysis (Fig.9.20 right), it can be
observed that the ratio between the fuel-minimizing and energy-minimizing stays constant with
the carbon intensity. This is to be expected as in both cases the total emissions scale proportionally
only to the carbon intensity of the energy grid, as when natural gas reforming, we were fixing the
carbon intensity from the reforming process. As a result, we get that the benefit of fuel-minimizing
(against energy-minimizing) is now independent of the carbon intensity level (however, the total
emissions still scale with it), and what we can see is that for the low-ranges high-specific energies
(700Wh/kg and 300nm), fuel minimizing actually yields an aircraft that has up to -49.5% lower total
carbon intensities, while for aircrafts in the the battery threshold line (528 Wh/kg and 685 nm), fuel
minimizing leads to an actual increase in total emissions of about +3.5%.
This means that for direct combustion of hydrogen produced via electrolysis, the region of emission
improvements begins further into the battery-installation design space, with aircraft at the transition
line remaining too energy-intensive relative to the fuel savings they achieve. The reason these
aircraft (the ones deeper into the battery-installation design space) can still reduce emissions, even
though their total shaft energy demand is higher (see Fig. 9.13), is that the power delivered by the
turboshaft is less efficient: the gas turbine cannot convert the full LHV of hydrogen into shaft power,
and therefore requires more kWh of input energy to deliver the same MJ at the shaft compared to the
batteries. The decision to install batteries thus depends on the trade-off between the additional mass
of this more efficient 15 energy vector, and the higher energy density of hydrogen, which requires
more kWh to provide the same MJ of shaft power.

15
In the sense of directly using the original electric kWh from the ground-based electricity grid.
 CHAPTER

HYDROGEN FUEL CELL

RESULTS 10
Contents
10.1 Thermal managed hydrogen hybrid . . . . . . . . . . . . . . . . . . . . . . . 137

10.1.1 Optimization statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

10.1.2 Optimization results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

The following chapter will present the results of the optimization study for the hydrogen fuel cell
parallel hybrid architecture. The main objective of this chapter will be to understand and compare
how the system performance evolves with an increase in cell-specific energy and range. This chapter
will also focus on the importance of energy vs fuel optimization.

10.1 THERMAL MANAGED HYDROGEN HYBRID

10.1.1 Optimization statement

For the optimization statement, we will maintain the multi-objective (fuel vs. energy) approach,
using the formalized energy objective function already introduced in the previous chapter (Eq. 9.1).
The only modification to this function is the efficiency in the extraction of energy from the fuel,
ηfuel , which in this case will be set to a mean reference value of ηfuel = 0.55,1 representative of the
average efficiency of the propulsion system.
For these optimization cases, we will perform two main studies by applying the objective function of
Eq. 10.1 under the conditions of ϱ = 0 and ϱ = 1. In the first study, the specific energy will be varied
from 250 Wh/kg to 700 Wh/kg at a constant range of 758 nm. In the second study, both the specific
energy (300–700 Wh/kg) and the mission range (300–1200 nm) will be varied simultaneously. In
addition, an intermediate optimization sweep will be conducted with ϱ = 0.6 at constant range, to
provide justification for some of the arguments later made regarding fuel-cell architecture trends.
1
This value aligns more closely with the efficiency curve discussed in the previous chapter.

137
138 CHAPTER 10. HYDROGEN FUEL CELL RESULTS

Efuel · ηfuel + Ebatt · ϱ

f (Wfuel,block , Wcell mass ) = (10.1)
Enorm
where:

Efuel = Wfuel,block · LHV [total chemical energy stored in fuel] (10.2)

Ebatt = Wcell mass · SpecEnergy [total electrical energy stored in battery] (10.3)
Enorm = normalization energy reference [baseline energy for comparison] (10.4)

For the design variable and constraints, the design space from Tab.9.2 is expanded in order to include
the design variables of the fuel cell, and the additional design variables of the heat exchanger under
each fuel cell nacelle, the new design space is presented under Tab.10.2.

Design variable (units) Lower Upper Constraint (units) Lower Upper

MTOW (kg) 18000 100000 δ Fuel Cell (-) — 0.9/1

Prop Mtang (-) 0.4 0.9 δ inveter (-) — 0.9/1
Prop diam (m) 2.5 5 δ electric motor (-) — 0.9/1
Prop factor (-) 0.76 1.74 δ battery (-) — 0.9/1
Wing ndpos (-) 0.2 0.55 Battery SOC (-) 0.2 1
Wing Sref (m2 ) 30 90 Tank SOF (-) 0.05 1
Wing AR (-) 8 22 Propeller Cp (-) — 3
HTP Sref (m2 ) 5 50 Propeller Mtrue (-) — 0.75
VTP Sref (m2 ) 5 50 Static margin (-) 0.35 -
Fuel Cell rating (MW) 0.02 200 Mass margin (kg) 0.01 —
Electric motor rating (MW) 0.02 200 Mean velocity (m/s) 160 —
Inverter rating (MW) 0.02 200 BFL (m) — 1400
Battery cell mass (kg) 0 200000 Fuel Cell Temp (ºC) - 70
Coolant pump (kW) 0 200000 Inverter Temp (ºC) - 70
˙
mcoolant (kg/s) 0.01 100 Emotor Temp (ºC) - 70
Reservoir coolant (kg) 10 1000 Battery Temp (ºC) 10 60
Coolant hose diameter (m) 0.01 10 F2F HX k design margin 0 -
Inv TMS HeightRatio (-) 0.1 1 FC HX k design margin 0 -
Inv TMS nparallel (-) 1 100 F2F HX width clear 0 -
FC TMS WidthtRatio (-) 0.1 1 F2F HX height clear 0 -
FC TMS nparallel (-) 1 100 FC HX width clear 0 -
FC TMS nplates (-) 1 100 FC HX height clear 0 -
BattTMS tchannel (m) 0.001 0.1 FC HX len clear 0 -
F2F HX nlong,c (-) 1 1000 ∆ Tank Pressure (bar) 0 2
F2F HX nwide,c (-) 1 1000

Continued on next page

10.1. THERMAL MANAGED HYDROGEN HYBRID 139

Design variable (units) Lower Upper Constraint (units) Lower Upper

F2F HX ntall (-) 1 1000

F2F HX design k (-) 1 1000
F2F HX multiplierc (-) 0.1 1000
F2F HX multiplierh (-) 0.1 1000
FC nacelle HX nlong,c (-) 1 1000
FC nacelle HX nwide,c (-) 1 1000
FC nacelle HX ntall (-) 1 1000
FC nacelle HX design k (-) 1 1000
FC nacelle HX multiplierc (-) 0.1 1000
FC nacelle HX multiplierh (-) 0.1 1000
LH2 mass (kg) 1 10000
LH2 press multiplier (-) 1 10
hcruise (m) 1000 7000
EASstart (m/s) 10 300
EASmid (m/s) 10 300
EASend (m/s) 10 300
γstart (deg) -10 10
γend (deg) -10 10
M tipstart (-) -10 10
M tipend (-) 0.8 1.2
HFstart (-) 0 1
HFend (-) 0 1
Ramopen,start (-) 0.01 1
Ramopen,end (-) 0.01 1
RamF Copen,start (-) 0.01 1
RamF Copen,end (-) 0.01 1
DivertF Copen,start (-) 0.01 1
DivertF Copen,end (-) 0.01 1
Battheating,start (kW) 0 1000
Battheating,end (kW) 0 1000
T ankventing,start (kg/s) 0 1000
T ankventing,end (kg/s) 0 1000
T ankheating,start (kW) 0 1000
T ankheating,end (kW) 0 1000

Table 10.2: Design variables and bounds (left), and associated constraints with limits (right). Values in
salmon denote variables defined per mission segment, values in pink represent constraints enforced
over a vector of mission segments, values in orange refer to variables or constraints defined separately
for each mission failure mode.
140 CHAPTER 10. HYDROGEN FUEL CELL RESULTS

This corresponds to 47 aircraft level design variables, 25 trajectory level design variables, and 128
control design variables (8 controls, linearly interpolated from start to end of 8 phases); leading to a
total of 200 design variables. For the trajectory definition, we discretize each phase into nn = 20
nodes, considering that we impose constraints over 22 state variables for the mission, and 6 during
the takeoff, we have a total of 3528 constraints for the mission, 480 constraints for the takeoff, with
the addition of 9 consistency a clearance constraints, leading to a total of 4017 constraints.

10.1.2 Optimization results

Starting with the discussion of the optimization results, Fig. 10.1 shows the evolution of the battery
cell mass and the nominal block fuel2 as a function of the cell-specific energy. As the battery-specific
energy increases, the nominal block fuel decreases, a trend that is then accelerated by a sudden
bump in the installed battery mass. This behavior is almost identical to the one already observed in
previous sections (Fig. 9.1 and Fig. 8.13), with the main difference being that the threshold specific
energy has shifted down from about 550 Wh/kg (in Fig. 9.1 and Fig. 8.13) to roughly 430 Wh/kg. In
addition, the battery mass installed below this threshold is non-zero, reaching values on the order of
1400 kg.

Figure 10.1: Evolution of battery cell weight and nominal block fuel for various cell’s specific
energies along the geometries for three technology levels. Batteries are highlighted in green with
the inverter and the electric motor in orange and yellow, respectively. For the hydrogen parallel fuel
cell architecture with thermal management.

To understand why this is happening, we can start looking at what happens with the electric motor
and the fuel cell. As we can see from Fig.10.2 (left), as we increase the cell-specific energy, the
electric motor power rating slightly increases, mimicking the trend that we have with the installed
battery mass at Fig.10.1. This is to be expected as, when fuel minimization, the addition of battery
2
In this case, the fuel is once again hydrogen.
10.1. THERMAL MANAGED HYDROGEN HYBRID 141

Figure 10.2: Evolution of the state of charge and hybridization factor for the various aircraft specific
energy configurations for the Kerosene Parallel Hybrid with no thermal management.

cell mass generally has an associated increase in the aircraft’s M T OW (see Fig.10.3). This means
that the component that has to provide power to the propeller (in this case, the electric motor), has
to increase its power rating in order to climb the heavier aircraft into altitude3 . Observing now the
trend of the fuel cell power rating, Fig.10.2 (right), it can be observed that now there is an initial
rapid decrease in fuel cell power rating, followed by a plateau, that once more decreases as the
battery cell threshold is surpassed. This trend is quite different from the trends that we got for the
direct hydrogen combustion (Fig.9.6), where the turboshaft rating remained relatively constant until
the cell battery threshold was reached.

Figure 10.3: Evolution of component mass with cell-specific energy. Structural mass includes wing,
fuselage, and tail weight along with subsystems, furnishing, and operational items. Powerplant
includes fuel cell, electric motors, inverters, battery tank, and TMS systems. Battery mass is expressed
at pack level, and payload includes passenger and luggage mass. For the Hydrogen fuel cell with
TMS.

3
And also to maintain the balanced field length constraint
142 CHAPTER 10. HYDROGEN FUEL CELL RESULTS

Figure 10.4: Evolution of the powertrain (power components) for the fuel cell and direct hydrogen
combustion architectures as a function of the battery cell specific energy.

The reason for this sudden change in the trend of the main power generator (the fuel cell) can be
understood by examining the propulsion system weight breakdown (Fig. 10.4). Compared to the
results obtained for fuel minimization of direct hydrogen combustion at constant range in Chap. 9,
the mass of the optimized fuel cell powertrain is significantly higher (about 2.03 times larger at
250 Wh/kg and 1.43 times larger at 700 Wh/kg). This increase in mass arises not only from the
considerably lower specific energy of the fuel cell, but also from the fact that the electric motor
and inverter must be sized to deliver the peak total power, rather than just the electric boost. As a
result, in this specific fuel-cell-based architecture, the addition of batteries and the improvement
of their specific energy act as strong enablers for downsizing the powertrain. This leads to two
distinct reductions in total powertrain mass: one from 250 Wh/kg to 415 Wh/kg, and another from
440 Wh/kg to 700 Wh/kg. These reductions in turn translate into slight decreases in block fuel
consumption, as seen in the two slopes of Fig. 10.1.

Figure 10.5: Evolution of the state of charge and hybridization factor for the various aircraft-specific
energy configurations for the Hydrogen Fuel Cell with thermal management

The way this battery energy is spent can be seen in Fig.10.5, where it can be observed that the
cell energy is initially spent solely on the climb segment, allowing the fuel cell to be downsized
10.1. THERMAL MANAGED HYDROGEN HYBRID 143

and reduce its mass (and as we will later see, reducing also the thermal load). Once the battery
technology threshold is reached, the extra battery energy is also spent during the cruise segment,
with hybridizations up to 80% during the start of climb, till cruise hybridizations around 40% during
the cruise. This is again a different behaviour to what we saw from the turboshaft, with the reason
also stemming from the fact that the fuel cell can be turned off, as opposed to the gas turbine, which
was kept at idle.

Figure 10.6: Energy provided by the battery and by the hydrogen (left). Energy intensity from
the battery, hydrogen, and total energy (right). For the Hydrogen Parallel Hybrid with thermal
management.

This extra benefit from power downsizing does not mean that the aircraft is going to be operating
at a lower energy. In fact, if we check the energy consumed by the aircraft (Fig.10.6, left), we can
see that even if the aircraft is consuming less fuel compared to the direct hydrogen combustion (at
250 Wh/kg we go from 514 kg to 452 kg, a 13.71% decrease in nominal block hydrogen needed),
the aircraft is operating at a higher hydrogen energy demand (going from 17832 MJ of energy in
Fig.9.4 to 1912 MJ of energy now, or a 7.23% increase in energy, just from the energy provided by
the hydrogen). This difference stems from the fact that we are considering the energy provided
by the hydrogen (the one that is then transformed into shaft power) rather than the energy stored
in the hydrogen (the one stored as LHV). This means that due to the higher efficiency of the fuel
cell as compared to the turboshaft, we can obtain more energy from the same hydrogen, and as
such, we can operate at a higher total aircraft energy demand while reducing fuel required. This
also means that if the carbon intensity of hydrogen production is per kg of hydrogen, we could be
reducing our equivalent carbon emissions while operating at a slightly higher energy due to the
heavier powertrain. Regarding the energy split (Fig.10.6, right), we can see that once more there is a
trade between energy from the fuel and energy from the battery, with a more parabolic curve as we
start to downsize the powertrain.
Repeating this same sweep, but now running it also for ϱ = 1 and ϱ = 0.6 we get the trends from
Fig.10.7 where we can observe that for the energy minimization the optimizer removes the battery
mass, meaning that the nominal block fuel remains constant, and the fuel cell now has a higher
power rating (about 16.23% higher at 250Wh/kg). If we now consider a mixed approach where we
144 CHAPTER 10. HYDROGEN FUEL CELL RESULTS

Figure 10.7: Evolution of the total energy intensity (upper left), hydrogen consumed in nominal
mission (upper right), battery cell mass (bottom left), and fuel cell rating (bottom left).

optimize for energy but consider only partial of the battery energy, we get that the optimizer initially
does not consider any battery mass, until we get to the threshold battery cell specific energy, where
the optimizer adds battery cell mass.
It is interesting to compare the full fuel minimization and the mixed minimization, comparing with
respect to the "constant" energy-oriented optimization, the fuel minimization adds at 700Wh/kg
5871kg of battery cell mass and reducing the nominal hydrogen in 122.77 kg (about a 27.1% decrease
wrt to the energy optimization) but increases the energy intensity from 0.355 MJ/pax/nm to 0.529
MJ/pax/nm (about a 49% increase). Meanwhile, for the mixed optimization at 700Wh/kg, the optimizer
adds 1794.15 kg of batteries, that reduces the nominal hydrogen by 56.10 kg (a 12.41% decrease wrt
to energy optimization) while only raising the energy intensity to 0.396 MJ/pax/nm (or about a
11.54% in energy demand). This means that if we get the ratio between the fuel decrease divided by
the energy increase, the fuel optimization yields a value of 0.55, while the mixed optimization yields
a value of 1.075. This means that there will most likely be, for a given specific energy, an aircraft
that maximizes the decrease in energy while minimizing the energy increase, and that aircraft will
most likely exist in between the fuel minimization, and the energy minimization4 .
Turning now to the thermal management, Fig. 10.8 shows the evolution of the component tempera-
tures. As can be seen, all components reach their thermal limits at the start of the climb segment,
after which they operate within their allowable temperature ranges. Interestingly, no thermal

4
Due to the time constraints and computational budget this study could not be fully performed.
10.1. THERMAL MANAGED HYDROGEN HYBRID 145

Figure 10.8: Temperature evolution of powertrain components and for the 628 Wh/kg battery at 758
nm design range for FC LH2 PH-TMS.

limitations appear during the cruise segment, with both the fuselage and the fuel cell ram doors
remaining open throughout cruise.
While this implies an additional drag penalty due to the continuous mass flow through the heat
exchangers, the behavior remains consistent across the optimized specific energies. In these cases,
the ram door aperture levels (for both the fuselage and the fuel cell nacelles) gradually decrease as
the total heat load (dominated by the fuel cell) is reduced (see the top two plots in Fig. 10.9). Due
to time limitations, it was not possible to rerun the optimization with alternative initial guesses to
146 CHAPTER 10. HYDROGEN FUEL CELL RESULTS

Figure 10.9: Evolution of the temperatures of the powertrain components along the mission for each
of the cell specific energies for the LH2 FuelCell Parallel Hybrid.

ensure a fully consistent feasible solution, which might have helped avoid convergence towards this
local minimum. Examining the temperature profiles in more detail (rows 2 to 5 of Fig. 10.9), we can
see that the thermal behavior closely reflects the evolution of the design variables. The fuel cell heat
10.1. THERMAL MANAGED HYDROGEN HYBRID 147

load decreases progressively with increasing specific energy, while the battery, inverter, and electric
motor all exhibit a discontinuity (or "step") in their trends as the battery specific energy threshold
is crossed, coinciding with the sudden increase in installed battery mass and in the ratings of the
electric motors and inverters.

Figure 10.10: Evolution of the hydrogen tak state of fill and the tank pressure along the nominal
mission and specific energies.

Regarding hydrogen tank control, Fig. 10.10 (left) shows that all designed aircraft exhibit an almost
constant hydrogen consumption during the climb and cruise segments, with a reduction during
descent. As in the direct-hydrogen-combustion case, the aircraft consumes only about 60% of the
available hydrogen during the nominal mission, leaving the remainder for the extended mission and
the 5% reserve. The pressure evolution, shown in Fig. 10.10 (right), follows a similar trend: the tank
pressure decreases steadily as hydrogen is consumed and then rises slightly during descent, when
extraction is minimal. This profile, however, differs from the one obtained in Fig. 9.9, suggesting the
presence of a local trend in pressure management across the mission.5
Looking now at the fuel-minimization (ϱ = 0) optimization, Fig. 10.11 shows the evolution of the
installed battery cell mass and the nominal block fuel across the battery-specific energy-range design
space. An interesting difference compared to the previous architectures is that three distinct regions
of battery installation appear. The first is located in the upper-left corner, where battery installation
is minimal. The second extends to the right, up to the line connecting 300 Wh/kg and 500 nm with
700 Wh/kg and 1200 nm, where the optimizer installs battery masses on the order of 1500 kg. Finally,
a third region emerges further to the right, where the optimizer begins to significantly increase the
installed battery mass. The reason for this phenomenon can be seen in the tendency of fuel cell
downsizing, which holds true except for the very long ranges, where the battery cell mass capacity
to downsize is surpassed by its dead mass along the long mission.
The same trends observed in the constant-range study can also be identified in the nominal block
fuel distribution (Fig. 10.11, right). A homogeneous gradient of block fuel reduction extends from
the upper-left to the lower-right corner of the design space, with the slope becoming steeper as the
5
In practice, pressure control is only partially defined, since the tank pressure remains very close to the design limits.
A change in mission or trajectory could therefore drive the tank outside its allowable pressure range, requiring additional
venting or active heating. For future optimizations, it would be advisable to narrow the allowable pressure-band width to
include a buffer margin and thus increase the robustness of the optimized architecture.
148 CHAPTER 10. HYDROGEN FUEL CELL RESULTS

Figure 10.11: Evolution of the installed cell battery mass (left) and nominal block fuel (right) along
our range-cell specific energy design space for the FuelCell aircraft.

design enters the cranking-up battery mass region just to the left of the battery cell threshold. This
is again due to the aircraft initially benefiting from the batteries by downsizing the fuel cell, and
then benefiting once more with additional battery mass as the installation of battery cells becomes
more profitable6 .

Figure 10.12: Evolution of the MT OW (left) and relative increase with respect to an aircraft with
an MT OW of 23 tons (right) along our range-cell specific energy design space for the fuel cell
architecture.

Regarding the aircraft mass, Fig. 10.12 once again shows that the installation of battery cell mass
produces a patch of higher M T OW in the battery-installation region. As before, this patch appears
at approximately constant M T OW . What is particularly interesting is the comparison of M T OW
growth relative to the reference. In the direct-hydrogen-combustion case (Fig. 9.11, right), the
architecture maintained a similar M T OW to the reference aircraft outside the high-battery-mass
region, even showing a slight reduction in some areas. In contrast, for the new fuel cell architecture
(Fig. 10.12, right), the increase in M T OW is present across the entire design space, highlighting the

6
In the sense of not acting as pure dead mass once they are depleted.
10.1. THERMAL MANAGED HYDROGEN HYBRID 149

substantially higher total mass of the fuel cell architecture.7

For the evolution of the power ratings of the fuel cell and electric motors, Fig. 10.13 shows three
distinct regions for the fuel cell, which evolve inversely with the installed battery mass. In contrast,
the electric motor ratings exhibit a single region of upsizing that coincides with the area of increased
M T OW . This behavior is consistent with the trends previously observed at constant range: the
optimizer exploits the installed battery mass (both in the partial- and full-installation regions) to
downsize the fuel cell. For the electric motors, however, the upsizing is mainly driven by the sudden
increase in aircraft M T OW and the corresponding requirement for additional power to climb the
heavier aircraft to cruise altitude.

Figure 10.13: : Evolution of the fuel cell power rating of each engine (left) electric motor power
rating from each engine (right), along our range-cell specific energy design space for the fuel cell
regional aircraft

Figure 10.14: Evolution of the total consumed energy (left) and fraction from it that comes from the
battery (right) along our range-cell specific energy design space for the PH-TMS LH2 regional aicraft

Regarding the energy consumption of the aircraft, Fig. 10.14 shows that, once again, the addition of
battery mass increases the total energy demand, with energy rising towards the right-hand side of
the design space at constant range. A key difference here is that the energy peak now occurs at the
7
Further comparisons will be provided in the architectures comparison chapter.
150 CHAPTER 10. HYDROGEN FUEL CELL RESULTS

highest specific energy and range, whereas in the kerosene and hydrogen hybrid cases, the peak
appeared at the maximum specific energy but at a range slightly below the maximum. Moreover,
hybridization is no longer confined to the battery-installation regions, as was the case for the
kerosene and hydrogen hybrids, but instead appears as a gradient extending from the upper-left to
the lower-right corner of the design space (Fig. 10.14, right). As with the nominal block fuel, this
gradient becomes steeper upon entering the high-battery-mass region (see Fig. 10.11). In terms of
absolute energy demand, the peak for the hydrogen-combustion hybrid was 32,861 MJ (at 700 Wh/kg
and 943 nm), whereas for the fuel cell hybrid it is now 38,042 MJ (at 700 Wh/kg and 1200 nm),
corresponding to a 15.76% increase. Comparing at the same range, at 943 nm and 700 Wh/kg, the
energy demand rises to 33,592 MJ, which represents a smaller increase of 2.17% relative to the
hydrogen-combustion hybrid.

Figure 10.15: Comparison of the equivalent CO2 emissions for the various LH2 PH architectures
under various electricity and hydrogen production scenarios.
10.1. THERMAL MANAGED HYDROGEN HYBRID 151

Looking now at the emissions from the fuel-minimizing fuel cell architecture, Fig. 10.15 presents the
results for hydrogen produced either by natural gas reforming or by electrolysis. Focusing first on the
natural-gas-reforming case (left column in Fig. 10.15), we observe a gradient of decreasing emissions
extending from the upper-left to the lower-right corner of the design space. The slope of this
gradient decreases as the carbon intensity of electricity increases, since the benefit of reducing fuel
consumption is progressively offset by the higher emissions associated with electricity generation.
To illustrate this, consider the design point at 700 Wh/kg and 300 nm. At carbon intensities of 28, 146,
and 384 gCO2 /kWh, the total equivalent emissions are 181, 740, and 1880 kg of CO2 , respectively.
Compared to the direct-hydrogen-combustion case (Fig. 9.15), these correspond to relative changes
of −9.95%, −2.25%, and −0.16%.

Carbon Intensity Emissions PH-LH2 [kg] Emissions FC LH2 [kg] Change [%]
28 gCO2 /kWh 201 181 −9.95
156 gCO2 /kWh 757 740 −2.25
384 gCO2 /kWh 1883 1880 −0.16

Table 10.3: Comparison of total equivalent CO2 emissions for the higher-hybridization design
(700 Wh/kg, 300 nm) between the direct hydrogen combustion and the fuel cell architecture.

This decrease in the decrease of equivalent total emissions is due to this point being heavily hy-
bridized, as such, while for low carbon intensity, we are able to see the initial benefit from the
additional efficiency of the fuel cell, as we increase the carbon intensity, the emissions associated
with the higher battery cell mass (6832 kg for the fuel cell, and 6768 kg for the direct hydrogen
combustion) start to fade the improvements due to the production of hydrogen. To isolate the direct
benefit of fuel cell efficiency (without the influence of fuel-cell downsizing), we can examine the
point at 300 Wh/kg and 1200 nm, where no battery mass is installed in either case. Here, the fuel cell
architecture results in 8638 kg of equivalent CO2 emissions across all three scenarios, corresponding
to a −12.94% reduction relative to direct hydrogen combustion.
To assess the full effect across the design space, this analysis was repeated for each range–specific-
energy combination, yielding Fig. 10.16. Here the largest relative decrease in emissions occurs
at 28 gCO2 /kWh and 300 nm, but not at 700 Wh/kg; instead, it is found at 528 Wh/kg. At this
point, the fuel cell reduces the nominal hydrogen requirement from 73.49 kg to 52.05 kg, while
the battery mass remains essentially constant (6865 kg versus 6817 kg, a mere −0.699% change).
The reason why this maximum improvement appears at 528 Wh/kg rather than at 700 Wh/kg is
that, at the highest specific energies, the architectures are transitioning toward an almost fully
electric configuration. As the system becomes more electric, the absolute reduction in hydrogen
consumption (which is where the fuel cell’s efficiency advantage lies, since it can deliver more
MJ per kilogram of hydrogen) becomes smaller. Consequently, the relative change is increasingly
dominated by the carbon impact of the electricity required to charge the batteries. Similarly, the
reason why this significant improvement does not appear in the case of a pure fuel-cell installation
without batteries (300 Wh/kg at 1200 nm) is that, in that regime, the batteries can no longer synergize
with the downsizing. Instead, when only replacing the gas turbines with fuel cells while consuming
152 CHAPTER 10. HYDROGEN FUEL CELL RESULTS

hydrogen alone, the penalty of the heavier powertrain8 offsets part of the efficiency benefits.

Figure 10.16: Relative change in total equivalent CO2 emissions (fuel cell vs direct hydrogen com-
bustion) for hydrogen produced by reforming natural gas and direct hydrogen combustion (both in
fuel minimization).

Studying now the trend with the electrolysis (Fig.10.15, right), it can be seen that, similar to what
we had with the direct hydrogen combustion, the emission trends are just scaled linearly with the
carbon intensity of the electricity used. To check if this is now done better or worse than for the
direct hydrogen combustion, we can look at Fig.10.16 where it can be seen that once more for 300nm
fuel cells can improve over direct hydrogen combustion (as the higher efficiency allows them to
obtain more energy per kilogram of hydrogen, and thus recover more energy from the original
energy used for electrolysis) but just until a local minima from which increasing the specific energy
increases the hybridization and thus starts to compare a higher fraction of battery related emissions.
8
And the heavier demand in terms of thermal cooling.
10.1. THERMAL MANAGED HYDROGEN HYBRID 153

An interesting observation is that, precisely at the boundary of battery installation (the battery
threshold line), the reduction in total emissions is actually smaller than in the case without batteries.
The reason for this is that fuel-cell architectures shift the threshold line toward lower ranges and
specific energies, leading to earlier and larger battery installations. This early addition of battery
mass diminishes the net benefit from the higher fuel-cell efficiency. A similar trend was observed
in the case of hydrogen produced by natural gas reforming (Fig. 10.15, left), where increasing the
grid carbon intensity amplified the battery-related emissions in the regions where the fuel-cell
architecture installed additional batteries, while the direct-hydrogen-combustion case did not. This
contrast will be analyzed in more detail in the following chapter.

Figure 10.17: Total energy demand (left) and relative increase in energy consumption (right) for the
fuel cell architecture optimized for minimal energy compared to the one for minimal fuel.

Figure 10.18: Installed battery cell mass (left) and installed delta (right) for the fuel cell architecture
optimized for minimal energy compared to the one for minimal fuel.

To further explore the interaction between batteries and fuel cells, we analyze the results of the
energy-minimization case (ϱ = 1). Considering first the evolution of total energy (Fig. 10.17), we
observe that, as in previous studies, energy minimization produces a profile driven almost exclusively
by the mission range. This outcome is achieved by removing most of the battery mass (Fig. 10.18),
which pushes the solutions toward the lower bound of the design space. The result is a smooth
154 CHAPTER 10. HYDROGEN FUEL CELL RESULTS

Figure 10.19: Nominal block fuel (left) and nominal block fuel change (right) for the fuel cell
architecture optimized for minimal energy compared to the one for minimal fuel.

energy gradient extending from the upper left to the lower right, with two distinct slopes: one
associated with light battery installation and a steeper one as the design enters the regime of heavier
battery installation.
As highlighted in earlier chapters, this reduction in total energy demand9 comes at the cost of higher
fuel consumption. The trade-off is particularly evident here (Fig. 10.19): without battery installation,
the fuel cell is not downsized, and the fuel savings previously observed in the partially and heavily
battery-installed regions of the fuel-minimization cases no longer occur.
Determining whether this reduction in energy at the expense of higher fuel consumption is beneficial
is not straightforward. The complexity stems from the fact that, so far, our analysis has only shown
that fuel cells can achieve a lower carbon impact than direct hydrogen combustion thanks to their
higher efficiency: they consume less hydrogen and recover more of the initial electrical energy
invested in hydrogen production via electrolysis. As fuel-cell efficiency increases, the gap between
the input electrical energy required to produce hydrogen and the energy ultimately delivered at
the shaft becomes smaller. Consequently, the aircraft that minimizes total energy demand at the
propeller shafts begins to align with the aircraft that minimizes the total initial electrical energy
demand, which now includes both the energy for charging the batteries and the energy for producing
hydrogen.
In the theoretical limit of perfect fuel-cell efficiency, hydrogen and batteries would recover identical
shaft energy per unit of input electricity. Under such conditions, both energy sources would exhibit
the same carbon intensity per unit of delivered energy, and therefore removing the batteries (with
their lower specific energy) would10 reduce the total equivalent emissions.
To evaluate how far the current system is from this idealized efficiency limit, the total emissions of
the energy-optimized architecture are recomputed, as shown in Fig. 10.20. Since energy minimization
removes the batteries, the case of hydrogen produced via natural gas reforming (Fig. 10.20, left)
results in an emissions map that is independent of the grid carbon intensity, varying only with the
9
Computed as the integral of the electric power delivered by the fuel cells and the batteries.
10
Provided, as shown earlier, that the addition of batteries for downsizing the fuel cell does not reduce the total energy
demand but only reduces fuel consumption.
10.1. THERMAL MANAGED HYDROGEN HYBRID 155

Figure 10.20: Comparison of the equivalent CO2 emissions for the various fuel cell architectures under
various electricity and hydrogen production scenarios for the energy minimization optimization.

nominal mission range. In contrast, for hydrogen produced through electrolysis (Fig. 10.20, right),
the emissions profile also increases smoothly with mission range (reflecting the greater hydrogen
demand) but is additionally scaled by the carbon intensity of the electricity used for hydrogen
production.
A more insightful comparison is obtained by evaluating the total emissions of the fuel-minimizing
architecture relative to the energy-minimizing one (Fig. 10.21). Focusing first on hydrogen produced
via natural gas reforming (Fig. 10.21, left), the largest reductions in equivalent total emissions occur
for highly hybridized architectures at short ranges and high specific energies (e.g., 700 Wh/kg and
300 nm). Quantitatively, the relative changes in emissions at this design point are −91.5%, −64.5%,
and −10.5% for carbon intensities of 28, 146, and 384 gCO2 /kWh, respectively. As expected, the
156 CHAPTER 10. HYDROGEN FUEL CELL RESULTS

Figure 10.21: Comparison of the change in the relative equivalent CO2 between the fuel-minimization
and energy-minimization optimizations for the fuel cell architecture.

benefit from fuel minimization diminishes as the grid carbon intensity increases, since the additional
emissions from charging the batteries gradually offset the reductions achieved by lowering the
emissions from hydrogen reforming.
For the case of electrolysis, the trends resemble those observed for direct hydrogen combustion.
As shown in Fig. 10.21 (right), there exists a region in the lower-right corner of the design space
where the addition of batteries (under the fuel-minimizing strategy) reduces total carbon emissions,
even though the aircraft becomes more energy demanding overall (Fig. 10.14). At the same time, a
distinct line of increased relative emissions emerges along the threshold of battery mass installation,
mirroring the behavior noted for direct hydrogen combustion. In this region, the initial reduction in
fuel consumption is no longer sufficient to offset the additional emissions associated with battery
charging.
10.1. THERMAL MANAGED HYDROGEN HYBRID 157

Figure 10.22: Comparison of the change in the relative equivalent CO2 between the fuel-minimizing
and the energy-minimizing considering electrolysis. For both the direct hydrogen combustion and
the fuel cell.

A direct comparison of this threshold line between the fuel-cell and direct-combustion cases is
shown in Fig. 10.22. Two main differences stand out: first, the line shifts toward the upper right of
the design space, consistent with the previously observed displacement of the battery installation
threshold; and second, the maximum relative reduction in emissions decreases when moving from
direct combustion to fuel-cell architectures, particularly in the highly hybridized configurations
(e.g., 700 Wh/kg and 300 nm). The reason for this lies in the efficiency of hydrogen utilization.
In the direct-combustion case, the conversion efficiency from the initial electrical input (used for
hydrogen production) to shaft power is relatively low. By contrast, the fuel cell extracts more useful
energy from the same hydrogen. As a result, when the architectures become almost fully electric
and hydrogen is replaced by batteries, the relative change in the required initial electrical energy is
smaller for the fuel-cell case than for direct combustion, leading to a reduced relative benefit.
Within this chapter, we have shown that fuel-cell architectures are not only feasible but also benefit
significantly from hybridization. In particular, the addition of batteries enables downsizing of the
otherwise heavy fuel-cell powertrain, improving overall performance once the battery installation
threshold is surpassed. When compared to direct hydrogen combustion, fuel-cell architectures under
fuel minimization generally yield lower relative emissions, regardless of whether the hydrogen is
produced via natural gas reforming or electrolysis (Fig. 10.16). However, as in the direct-combustion
case, the placement of batteries must be carefully considered. The most advantageous region lies in
the lower-right corner of the design space, where short ranges and high specific energies maximize
the benefits of hybridization, while the initial battery installation region should be avoided, since here
the additional battery mass does not yet translate into meaningful emission reductions (Fig. 10.22).
 CHAPTER

ARCHITECTURE
COMPARISON 11
Contents
11.1 Aircraft mass comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
11.2 Aicraft energy comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
11.3 Aircraft emissions comparison . . . . . . . . . . . . . . . . . . . . . . . . . . 169
11.4 Architecture selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

Having analyzed each optimized aircraft architecture individually, this chapter now presents a
comprehensive comparison across all configurations. The objective is to establish clear conclusions
on which architecture performs best under which conditions. In addition, based on the resulting
emission profiles, a representative design point will be selected for these novel architectures, for
which further details and geometry will be provided.
The comparison begins with an assessment of how the masses evolve across the design space,
followed by an analysis of the resulting energy demand and its implications for total emissions.
Finally, fixing a representative point in the range–specific-energy plane, detailed solution data will
be presented.

11.1 AIRCRAFT MASS COMPARISON

Starting with the comparison of maximum takeoff mass (M T OW ), Fig. 11.1 shows its evolution
as a function of range and battery specific energy. For the conventional kerosene architecture, the
increase in mission ranges increases the mass of fuel required at takeoff, this in turns makes the
M T OW to rise in order to carry this additional fuel mass, which in turn drives up wing area (in
order to lift the additional mass) and structural mass (due to the higher wing surface and the higher
reinforcement for the greater loads). For the hybrid architectures, however, a different behavior is
observed. As discussed in the individual chapters, looking at the design space of optimal solutions
for range and specific energies, it can be observed that the space is essentially split in half by a
diagonal, just as we denominated the battery threshold line. At the upper-left region of this line,
we experience a gradual increase in M T OW , where, due to the predominance of lower battery

159
160 CHAPTER 11. ARCHITECTURE COMPARISON

Figure 11.1: Comparison of the design M T OW for all four aircraft architectures: conventional
kerosene, hybrid kerosene, hybrid hydrogen combustion, and hybrid fuel cell.

Figure 11.2: Comparison of the installed battery cell mass for the three hybrid architectures: kerosene
combustion, hydrogen combustion, and hydrogen fuel cell.
11.1. AIRCRAFT MASS COMPARISON 161

allocations, mass mostly increases just with range (as again, longer ranges require higher masses
as a result of the additional fuel load, which generates the snowball effect of mass addition). In
contrast, for the bottom-right region of the design space (right of the battery threshold line) it can be
seen that the optimizer generates a sharp rise in M T OW (that comparatively remains constant)
due to the additional allocation of batteries (with maximums of 6452.74 at 700Wh/kg at 428nm for
the kerosene hybrid, 6951.74kg kg at 700Wh/kg at 428nm for the kerosene hybrid, and 7011.41kg at
642Wh/kg at 300nm for the fuel cell architecture) that lead to a sudden increase in takeoff weights.

Figure 11.3: Comparison of the powetraint toal mass for the four powertrain architectures. For the
kerosene conventional, this includes propeller masses, gearbox, and turboshaft core. then for both
the the kerosene hybrid and the hydrogen combsution hybrid the TMS system mass plus the electric
motor and inverter are added towards this mass. For the case of the fuel cell this mass includes the
mass of the fuel cells, electric motors, inverters, propellers, and TMS system.

This sudden increase in battery cell mass is illustrated in Fig. 11.2, where it can be seen that for the
kerosene-combustion hybrid (top rows of Fig. 11.2), two distinct regions can be identified: one of
low battery installation and another of high installation, separated by a well-defined threshold line
in the range–specific-energy space. In contrast, the fuel-cell architecture exhibits a slightly different
behavior. Here, the region of high battery installation is shifted toward the upper-right corner of the
design space, and an additional intermediate zone of medium battery installation appears. This new
area of battery allocation appears, as we have previously discussed, in order to downsize the fuel
cell powertrain. As a result, if we see the evolution of the total powertrain mass from the fuel cell
architecture (lower right corner from Fig.11.3) we can see that the highest powertrain mass appears
for the case of the hydrogen fuel cell at the upper left corner (300Wh/kg at 1200nm, where battery
mass installation was minimal), with it decreasing in the area of intermediate battery allocation, and
162 CHAPTER 11. ARCHITECTURE COMPARISON

in the direction of higher specific energies and lower ranges. This decrease is however, interrupted
just at the boundary of the higher battery mass installation, where the increase in M T OW (Fig.11.1
lower right) forces the powertrain to increase its power rating to climb the heavier aircraft, and then
the downsizing and lightening continues as the aircraft mass stabilizes and the optimizer starts to
heavily trade fuel cell rating for battery mass (as seen in Fig.10.2 and Fig.10.4).

Figure 11.4: Comparison of the structural and secondary mass for all four architectures. This
structural mass considers the mass of the nacelles, fuselage, tail, landing gear, system masses,
furnishing, and operational items. For the case of the hydrogen architectures, the structural mass of
the hydrogen tank is also added.

Turning now to the powertrain masses across the different architectures, Fig. 11.3 (upper left) shows
the case of the conventional kerosene aircraft. As expected, the powertrain mass varies only with
mission range: longer ranges require heavier aircraft, which in turn demand larger engines to
provide sufficient climb performance. Across the design space, this results in a relative increase
in powertrain mass of +10.94% between 300 nm and 1200 nm. For the kerosene and hydrogen
direct-combustion hybrids (Fig. 11.3, upper right and lower left), a distinct line of powertrain mass
increase appears along the boundary of the battery-installation region. This effect is similar to what
was observed for the fuel-cell case: as the aircraft becomes heavier with additional batteries, a larger
powertrain is required to meet climb performance. Interestingly, at the bottom-right corner of the
design space (where hybridization is highest), the powertrain mass increases rather than decreases.
In theory, a highly hybridized aircraft should exhibit a lighter powertrain, since it would rely more
on electric components with higher specific energies (as compared to the gas turbine). However,
in practice, the electric system must still be sized to provide the full mission power, while the gas
turbine must retain sufficient capacity to cover the extended mission once the batteries are fully
11.1. AIRCRAFT MASS COMPARISON 163

Figure 11.5: Relative change in M T OW between the hybrid architectures and the optimized
conventional kerosene architectures.

discharged. This dual sizing requirement drives the unexpected increase in powertrain mass, and it
is consistent with some proposed aircraft concepts, such as the Heart ES-30, which has gas turbines
for extended ranges and diversion missions.
As a result, the increase in powertrain mass combined with the addition of batteries leads to a growth
in the structural component mass within the regions of significant battery installation (Fig. 11.4), with
the largest increase observed for the fuel-cell architecture. An interesting comparison arises between
the kerosene and hydrogen direct-combustion hybrids: although the hydrogen-hybrid aircraft
exhibits a lower M T OW (Fig. 11.5, upper right), its structural mass is higher. This counterintuitive
result stems from the need to enlarge the fuselage to accommodate the fuel-cell tanks, which
increases the aircraft’s empty mass. However, once fully loaded, the reduced fuel requirement of
the hydrogen hybrid (see Fig.11.6) results in a lighter takeoff weight overall. At the same time, this
configuration implies that the aircraft typically lands at higher weights, since the load-alleviation
effect of fuel burn in wetted wings is no longer available. Consequently, the wings are subjected to
greater structural strain during landing maneuvers, which may impose stricter requirements on
wing design and fatigue management.1
It is important to note that the increase in M T OW leads to a drastic reduction in the payload
fraction of the hybrid architectures (Fig. 11.7), with only minor improvements observed for the direct
hydrogen combustion case in the low-battery-allocation region (as this architecture is able to exploit

1
As previously discussed, integrating the fuel tank structure within a reinforced fuselage could help reduce the
structural penalty, partly offsetting the mass increase due to the higher landing loads—an effect not captured by traditional
mass-estimation methods.
164 CHAPTER 11. ARCHITECTURE COMPARISON

Figure 11.6: Comparison of the consumed nominal block fuel (either kerosene or hydrogen) for each
of the four aircraft architectures.

Figure 11.7: Comparison of the payload fraction (computed as WP L /M T OW ) for each of the four
aircraft architectures.
11.2. AICRAFT ENERGY COMPARISON 165

the lower fuel mass, even with the longer fuselage and cryogenic tank penalty). This suggests that
future architectures exploiting battery technology will likely do so at a much smaller fraction of the
aircraft’s M T OW . The reduction in payload fraction also carries significant economic implications:
because the sensitivity of fuel requirements to payload (or the passenger fullness factor) is reduced,
these new architectures become more penalized when not operating at full load.2 . This is especially
detrimental for the fuel cell architecture, where large decreases in payload fraction load can be seen
even outside of the high battery mass allocation area.

11.2 AICRAFT ENERGY COMPARISON

Regarding the energy demand of the various architectures, Fig. 11.8 shows the total energy consumed
by each configuration. This is computed as the integral of the power delivered out by the turboshaft
(or fuel cell) plus the integral of the power delivered by the batteries (for those architectures that
include them). The objective here is to asses not how much energy has to be provided at the ground
(by means of the production of the fuel and the electricity), but rather, evaluate how much energy
the aircraft is asking to stay afloat.
Starting with the conventional kerosene architecture (Fig. 11.8, left), it can be observed that the main
change in energy demand is the mission range, where we have that at 300nm the energy demand is
6423 MJ, and for 1200nm the energy demand is 29440 MJ. Relative to the nominal mission range, the
energy intensity grows from 21.41 MJ/nm to 24.53 MJ/nm (a +14.57% increase). This growth has
already been explored in its relative chapter and relates mostly to the additional penalty from the
higher structure and fuel mass.
For the hybrid kerosene architecture, the maximum energy demand occurs at 700 Wh/kg and 942 nm,
where the aircraft requires 32123 MJ, or 34.1 MJ/nm. For the same nominal range, the conventional
kerosene configuration demands only 22239 MJ (23.6 MJ/nm), corresponding to a 44.45% increase
in energy demand per nautical mile. An even more dramatic effect is observed at 700 Wh/kg and
300 nm, where (as shown in the right-hand side of the second row in Fig. 11.8) almost all of the energy
comes from the batteries. Here, the energy demand rises from 6423 MJ to 16,844 MJ, representing a
162.2% increase. This sharp growth is a direct consequence of the low energy density of batteries:
substituting kerosene with batteries3 drives up M T OW significantly, which in turn raises the
total energy requirement. This implies that unless batteries can reach an effective energy density
comparable to kerosene4 , i.e. around 3583 Wh/kg, hybrid aircraft will inherently be more energy
demanding.
Something similar happens with the direct hydrogen combustion hybrid, where the peak energy
demand happens once more at 700Wh and 942nm, where the aircraft now requires 32861 MJ, or
34.88 MJ/nm (a 47.79% increase relative to the conventional kerosene). While as explained before, we
2
This is due to the fact that a reduction in payload weight produces a comparatively smaller reduction in fuel burn, as
the fuel requirement is dominated by the heavier baseline M T OW .
3
In this case, the hybridization makes for an aircraft that is almost fully electric.
4
Accounting for the fact that only about 30% of the chemical energy in kerosene can be converted to shaft power by
the gas turbine.
166 CHAPTER 11. ARCHITECTURE COMPARISON

Figure 11.8: Comparison of the total energy demand for each of the four aircraft architectures.

should have expected to have a decrease in energy demand when going from Kerosene to Hydrogen,
here the additional installed battery mass5 , plus the additional fuselage mass and fuel tank mass
make it so the aircraft is a bit more energy hungry.
If we now switch the hydrogen turbo-machinery for fuel cells, it can be seen that the new energy
peak demand moves to 1200nm and 700Wh, where the energy demand is now 38042MJ (or 31.7
5
Allowed due to the lower penalty of hydrogen aircraft in the additional fuel mass for the extended mission.
11.2. AICRAFT ENERGY COMPARISON 167

Figure 11.9: Comparison of the energy demand from the turboshaft/fuel cell (left) and the battery
energy demand (right).

MJ/nm). This means that at the same range and specific energy, the fuel cell architecture has a
29.23% higher energy demand than the standard kerosene combustion, a 19.84% higher energy
demand than the hybrid kerosene, and a 25.79% higher energy demand than the direct hydrogen
combustion. At the previous peak of the hybrid architectures (700Wh/kg and 942nm), we get that
168 CHAPTER 11. ARCHITECTURE COMPARISON

Figure 11.10: Comparison of the mission demand energy intensity (in MJ/nm, computed as the
integrals of the provided powers).

Figure 11.11: Comparison of the consumed nominal block fuel (either kerosene or hydrogen) for
each of the four aircraft architectures.
11.3. AIRCRAFT EMISSIONS COMPARISON 169

the mission intensity for the fuel cell is 35.7MJ/nm, leading to an increase of +51.3% with respect
to the conventional kerosene, a +4.69% increase with respect to the kerosene hybrid, and a +2.23%
increase with respect to the hydrogen hybrid. The reason for this strange evolution of the required
total energy is that, while the energy required from the turboshaft/fuel cell has decreased first with
the required range, and then decreases as we enter into the battery allocation area; for the energy
demanded by the batteries, we have the addition of energy demand increases as we move diagonally
deeper into the erea of battery allocation. As a result, we end up adding more energy demand from
the battery as the one that we are removing from the turboshaft/fuel cell, adding energy intensity as
we move towards the bottom right (see Fig.11.10).
This means that, as expected, if we compare the energy intensity of the hybrid architectures to the
conventional kerosene architectures, we get that for all cases there is an increase in energy demand
at the bottom right corner (see Fig.11.11). It is interesting to note that due to the extended mission
penalty, the highest increase in energy demand appears for the kerosene hybrid, the reason being
that due to the need to carry the extra fuel for the extended mission; this means that if we have
different architectures that use different main fuels, even if they all converge to the same almost
electric aircraft, there will still be a difference in performance depending on who has to carry the
leas amount of extra fuel mass for the extended mission. And as such, we get that for the fuel cell
hybrid, we have a lower increase in total energy, given that the hydrogen not only reduces the mass
needed to store that energy, but the higher efficiency of the fuel cell allows the effective energy
density to increase even more (even at the higher powertrain mass of the fuel cells).

11.3 AIRCRAFT EMISSIONS COMPARISON

As previously discussed, a higher energy demand does not necessarily translate into higher emissions;
this is because different energy sources carry distinct carbon intensities and production inefficiencies.
In this chapter, the total emissions are computed following the same methodology as in the preceding
chapters. For kerosene, a carbon impact of 3.15 kgCO2 per kilogram of fuel consumed is assumed.
For hydrogen, two production pathways are considered: natural gas reforming, with an equivalent
carbon impact of 11.4 kgCO2 per kilogram of hydrogen produced (see Chapter 2); and electrolysis,
where producing 1 kg of hydrogen requires 47.57 kWh of electricity, and the carbon intensity is
computed with the grid energy needed. For the grid energy (used to charge the batteries and the
electrolysis), three carbon intensities are considered: 28, 146, and 384 gCO2 /kWh.
Starting with the lowest carbon intensity (28 gCO2 /kWh), Fig. 11.12 presents the relative change in
total equivalent emissions with respect to the optimized conventional kerosene combustion aircraft.
It can be observed that, across all hybrid architectures, the largest reductions in emissions occur
in the bottom-right corner of the design space, precisely where the mission energy demand per
nautical mile is at its highest (see Fig. 11.10 and Fig. 11.11). This outcome illustrates one of the
central arguments developed throughout this thesis: total energy demand does not directly translate
into emissions. By replacing a more carbon-intensive energy vector with a cleaner one, it is possible
to achieve meaningful reductions in emissions, even if the overall energy consumption increases.
170 CHAPTER 11. ARCHITECTURE COMPARISON

Figure 11.12: Comparison of the relative change of carbon intensity with respect to the conventional
kerosene architecture for all three hybrid architectures. For the case of the hydrogen architectures
the emissions from natural gas reforming and for electrolysis is considered. For the case of grid
carbon intensity of 28 gCO2 /kWh.

Architecture 700 Wh/kg, 300 nm 300 Wh/kg, 1200 nm

Comb Kerosene −79.4 1.0
Comb Hydrogen (NG) −88.7 30.1
Fuel Cell (NG) −89.8 13.1
Comb Hydrogen (elect.) −91.9 −85.1
Fuel Cell (elect.) −92.0 −87.1

Table 11.1: Comparison of emissions change [%] across architectures for two representative design
cases at 28gCO2 /kWh: high-specific-energy short range (700 Wh/kg, 300 nm) and low-specific-
energy long range (300 Wh/kg, 1200 nm).

Taking some representative values (Tab. 11.1), it can be observed that the largest decreases in
equivalent CO2 emissions occur at 700 Wh/kg and 300 nm. Even under the assumption of hydro-
11.3. AIRCRAFT EMISSIONS COMPARISON 171

gen production via natural gas reforming, both hydrogen-based architectures still outperform the
kerosene hybrid: direct hydrogen combustion achieves a reduction of 11.17%, while the fuel-cell
architecture achieves a slightly larger reduction of 11.31% relative to the kerosene-combustion
hybrid. The advantage is even greater for electrolysis, with reductions of 11.57% and 11.59% for
direct combustion and fuel cell, respectively. The reason for this behavior is that, at high levels
of hybridization, hydrogen architectures are less penalized by the additional mass carried for the
extended mission. As a result, they outperform the kerosene-based hybrid, with the switch from
natural gas reforming to electrolysis further reducing the impact of the small fuel-related emissions
that remain. This is not the case for regions of the design space with almost no battery (300Wh/kg
at 1200nm), where the emissions from hydrogen architectures depend heavily on the source of
hydrogen, with natural gas reforming leading to higher overall emissions (as compared with an
optimized kerosene conventional), while electrolysis-based ones still lead to important decreases.
It is important to mention that here (for hydrogen from electrolysis), the decreases in emissions
are not as high as with battery allocation. The reason is that, while also from electricity, the higher
losses from the production of hydrogen and the lower energy recovery lead to higher overall losses,
which result in lower decreases.

Architecture 700 Wh/kg, 300 nm 300 Wh/kg, 1200 nm

Comb Kerosene −48.9 1.1
Comb Hydrogen (NG) −56.5 30.2
Fuel Cell (NG) −57.3 13.1
Comb Hydrogen (elect.) −57.9 −22.5
Fuel Cell (elect.) −58.3 −32.7

Table 11.2: Comparison of emissions change [%] across architectures for two representative design
cases at 146 gCO2 /kWh: high-specific-energy short range (700 Wh/kg, 300 nm) and low-specific-
energy long range (300 Wh/kg, 1200 nm).

Architecture Lower decrease [%] Specific Energy [Wh/kg] Range [nm]

Comb Kerosene 16.1 300 300
Comb Hydrogen (NG) 35.2 300 300
Fuel Cell (NG) 15.9 414 428
Comb Hydrogen (elect.) −15.8 357 428
Fuel Cell (elect.) −27.5 300 428

Table 11.3: Comparison of the lowest emissions decrease at 146 gCO2 /kWh across architectures
with their corresponding specific energy and range design points.

Increasing the carbon intensity now leads us to Fig. 11.13, where it can be observed that for the
kerosene hybrid (top figure) there is still a reduction in overall emissions in the bottom-right
corner of the design space (corresponding to high battery mass installation at low range and high
specific energy). The same behavior is found for the hydrogen architectures, both in the natural-
gas-reforming and electrolysis cases. However, as the carbon intensity of the grid increases, these
relative reductions become smaller (see Tab. 11.2, left column). In particular, the benefit of hydrogen
172 CHAPTER 11. ARCHITECTURE COMPARISON

electrolysis is reduced, since the electricity used to produce LH2 is now more carbon-intensive (the
penalty is particularly visible in the regions of low battery mass installation, where inefficiencies in
energy recovery combined with electrolysis reduce the decreases in equivalent emissions relative
to the lower-intensity case). Nonetheless, electrolysis remains the more efficient option overall,
showing decreases over the whole design space (with and without electrification).

Figure 11.13: Comparison of the relative change of carbon intensity with respect to the conventional
kerosene architecture for all three hybrid architectures. For the case of the hydrogen architectures
the emissions from natural gas reforming and for electrolysis is considered. For the case of grid
carbon intensity of 146 gCO2 /kWh.

Something interesting that happens is that the highest increases (or lowest decreases) do not happen
at the area of low battery allocation (as we had in Fig.11.12), but at the battery allocation threshold
line (see Tab.11.3, where there is a crest in the threshold of battery installation). This means that
we should halt the initial battery allocation if the carbon intensity of the grid is high, as we will be
initially reducing the fuel emissions less than we introduce with the recharge of batteries. This also
means that a hybrid architect will be better or worse depending on the country and grid it operates
11.3. AIRCRAFT EMISSIONS COMPARISON 173

in, with flight being less carbon-intensive flying from country A to B, and then less efficient from B
to A.

Figure 11.14: Comparison of the relative change of carbon intensity with respect to the conventional
kerosene architecture for all three hybrid architectures. For the case of the hydrogen architectures
the emissions from natural gas reforming and for electrolysis is considered. For the case of grid
carbon intensity of 384 gCO2 /kWh.

This crest in higher emissions becomes particularly evident at elevated carbon intensities (see
Fig.11.14). In this regime, emissions relative to the conventional kerosene baseline are not only uni-
versally higher (Tab.11.4), but their maximum values are concentrated precisely along the threshold
of battery mass installation (see Tab. 11.5). This highlights the need for caution: both with respect to
the placement of the battery-installation threshold, and more broadly regarding the carbon intensity
of the grid itself. The results show that under unfavorable conditions, hybridization can lead to
highly inefficient architectures. This effect is especially pronounced in the case of electrolysis, where
the combination of energy conversion losses and carbon-intensive electricity may result in emissions
that exceed those of natural gas reforming, and in fact yield substantially higher emissions across
174 CHAPTER 11. ARCHITECTURE COMPARISON

Architecture 700 Wh/kg, 300 nm 300 Wh/kg, 1200 nm

Comb Kerosene 12.6 1.2
Comb Hydrogen (NG) 8.6 30.4
Fuel Cell (NG) 8.3 13.2
Comb Hydrogen (elect.) 10.6 103.7
Fuel Cell (elect.) 9.7 77.0

Table 11.4: Comparison of emissions change [%] across architectures for two representative design
cases at 384 gCO2 /kWh: high-specific-energy short range (700 Wh/kg, 300 nm) and low-specific-
energy long range (300 Wh/kg, 1200 nm).

Architecture Lower decrease [%] Specific Energy [Wh/kg] Range [nm]

Comb Kerosene 34.8 414 300
Comb Hydrogen (NG) 53.4 357 300
Fuel Cell (NG) 30.6 300 428
Comb Hydrogen (elect.) 121.5 357 428
Fuel Cell (elect.) 90.6 300 428

Table 11.5: Comparison of the lowest emissions decrease across architectures with their correspond-
ing specific energy and range design [Link] at 384 gCO2 /kWh.

the entire design space (with and without battery mass installation).
This strong sensitivity to grid carbon intensity underscores a key point already discussed in earlier
chapters: the decarbonization of the skies begins with the decarbonization of the ground. The same air-
craft architecture can yield outcomes ranging from extremely large reductions in emissions (−92%)
to equally extreme increases (+90.6%). This means that advances in clean aviation technologies are
of limited value if they are not accompanied by investments ensuring that both hydrogen production
and battery recharging are supplied by genuinely low-carbon 6 energy sources. This idea isn’t new,
as it was already studied during the 90s, where we go back to the Tu-155 original paper [116]:

Hydrogen is a fully restorable energy carrier. During combustion of hydrogen, water

is formed which returns to the land. The hydrogen resources are unlimited if it is
recovered by electrolyses or thermochemical methods from water. The primary energy
can be only nuclear, therefore the hydrogen power of the future is considered as
nuclear–hydrogen, one where hydrogen manifests itself as a “convenient” energy
carrier. This is particularly important to aviation since the nuclear reactors remain
on land.[...] Finally, the use of the hydrogen can result in a significant reduction of
detrinental emissions
—Extracted from [V. Sosounov, 1990]

6
This is particularly important for the production of hydrogen by electrolysis, as while it could be obtained from
purely green energy, the high energy demand would usually require energy extraction from a central grid.
11.4. ARCHITECTURE SELECTION 175

11.4 ARCHITECTURE SELECTION

Having examined each architecture individually, compared them against one another, and assessed
their emissions, the next step is to select a representative design point. While the project began with
an initial set of top-level requirements (mostly 758nm as maximum range), the use of MDAO has
now provided sufficient knowledge of the design space for clean aircraft architectures to challenge
and refine these initial assumptions.
The first stage will be freezing the specific energy value selected for our batteries and the carbon
intensity of our reference grid. Starting with the specific energy, we can see that while we could
speculate over the future increase in specific energy, this specific energy is fixed by the electro-
chemistry of the battery and the density of its materials. As such, if we consider that the current
technology level is in the order of 400Wh/kg, we can propose two scenarios: a conservative one,
where we extract the aircraft’s sized for 414Wh/kg; and an optimistic one, where we consider the
aircraft’s sized for 471Wh/kg.

Architecture 414 Wh/kg (conservative) 471 Wh/kg (optimistic)

300 nm
Kerosene Hybrid +0.6% −12.0%
Direct H2 Combustion (NG Reforming) +7.4% −6.2%
Fuel Cell (NG Reforming) −25.4% −39.0%
Direct H2 Combustion (Electrolysis) −26.0% −33.4%
Fuel Cell (Electrolysis) −46.3% −52.9%
428 nm
Kerosene Hybrid +7.6% +0.2%
Direct H2 Combustion (NG Reforming) +26.4% +16.3%
Fuel Cell (NG Reforming) −1.7% −11.4%
Direct H2 Combustion (Electrolysis) −23.5% −24.7%
Fuel Cell (Electrolysis) −32.7% −40.2%

Table 11.6: Relative change in total equivalent CO2 emissions with respect to the optimized kerosene
architecture, for two battery technology scenarios (414 and 471 Wh/kg). Negative values indicate
reductions in emissions.

For the conservative one, assuming a grid energy intensity of 146 gCO2 /kWh. We have that the
kerosene aircraft would emit about +0.6% more than the conventional kerosene at 300nm, with
higher emissions overall for longer ranges7 . For the direct hydrogen combustion under natural gas
reforming, we have a similar scenario, where now we would be emitting a minimum of 7.4% more8 .
For the fuel cell architecture under natural gas reforming, things are a bit better as now we can, at
the fixed technology of 414Wh/kg, emit either -25.4% less at 300nm, or -1.7% less at 428nm (with
upper range values now emitting more). For electrolysis in direct gas combustion at 414Wh/kg
we can emit less by either by flying with a battery oriented design up to 300nm (where there is a
7
This means that for this case we should avoid the kerosene hybrid as it does not provide any benefits in emissions
8
The reason for this is the higher carbon impact of reforming the natural gas as compared to burning the kerosene.
176 CHAPTER 11. ARCHITECTURE COMPARISON

-26.% of emissions), or fly without batteries (the reason is to avoid the initial inefficiencies of battery
battery cell mass installation where the total emissions are actually higher), where we could fly up
to any range with a decrease in average of -23.5%. For fuel cell under electrolysis, we have even
better results, where we can fly with a decrease of -46.3% in emissions at a range of 300nm, or fly
without batteries up to a range of 1200nm where we have decreases in the order of -32.7%.
This means that with the current battery technology level, only the hydrogen architectures under
electrolysis are able to fly at lower emissions as compared to an optimized hybrid kerosene.
If we now optimistically increase the battery technology level to 471 Wh/kg, we have slightly better
news. Starting with the kerosene hybrid, we get that at 300nm we are to decrease by -12%, with
the emissions at 428nm then increasing just by +0.2% (all of them with respect to the optimized
kerosene architectures). For direct hydrogen combustion with natural gas reforming, we also get
reductions, with at 300nm emitting -6.2% less, and at 428nm +16.3% (with then values at longer
ranges going beyond this). For electrolysis, we have that we should fly with batteries at 300nm and
428nm, where the decreases are of -33.4% and -24.7% (with longer ranges already benefiting from the
removal of batteries within the threshold line). For a fuel cell with natural gas reforming, we have
for 300nm -39%, and for 428nm -11.4, just having overall higher emissions beyond 557nm, where
the increase is of +3.4%. With electrolysis, these decreases are now -52.9% and -40.2%, for 300 and
428nm, respectively.

Figure 11.15: Resultant aircraft geometries for a specific energy of 471Wh/kg and a nominal mission
design range of 428nm.

From these results, it becomes clear that a modest increase in battery-specific energy enables
11.4. ARCHITECTURE SELECTION 177

Figure 11.16: Relative change of total equivalent emissions across the on-design mission ranges for
a battery cell technology of 471Wh/kg. All of them compare the on-design range optimized hybrid
architecture with its corresponding on-design conventional kerosene.
178 CHAPTER 11. ARCHITECTURE COMPARISON

meaningful emission reductions at longer ranges. This effect is illustrated in Fig.11.16, which shows
the emissions across different on-design mission ranges. As it can be seen, all architectures exhibit
a local minimum around 428nm, where the synergy with the battery cells is most effective. This
behavior is particularly evident in the fuel-cell architecture, which achieves noticeable emission
reductions even when hydrogen is produced via natural gas reforming (Fig. 11.16, bottom left).
While this 428nm of nominal range could feel like a reduced cruise distance, as it could only cover
ranges from

Parameter Conv. Kerosene Kerosene Hybrid Direct LH2 Burn Fuel Cell
MTOW (kg) 22562.4 34436.3 34448.9 34670
Propeller diameter (m) 4.00 4.00 4.00 3.07
Propeller Mtip 0.63 0.610 0.656 0.604
Propeller factor 0.761 0.760 0.760 0.700
Wing area (m2 ) 49.92 75.48 75.51 75.99
Aspect ratio 21.72 18.58 18.54 17.89
Wing position 0.430 0.415 0.455 0.442
HTP area (m2 ) 6.47 9.23 12.51 10.91
VTP area (m2 ) 12.19 17.15 20.98 18.85
Turboshaft rating (MW) 2.34 1.80 1.65 –
Gearbox rating (MW) 2.27 3.11 3.17 –
Battery cell mass (kg) – 5864.7 6417.9 6355.8
Electric motor rating (MW) – 1.45 1.70 1.71
Inverter rating (MW) – 1.53 1.79 1.79
Fuel Cell rating (MW) – – – 0.497
Nominal block fuel (kg) 789.8 669.7 215.0 155.0
Nominal + ext. fuel (kg) 1050.4 1186.0 426.9 317.8
Wing mass (kg) 2995.7 4254.9 4245.7 4110.7
Empennage mass (kg) 367.7 604.7 625.4 637.9
Fuselage mass (kg) 3372.0 4165.7 4295.9 4297.7
Nacelles mass (kg) 365.7 307.1 297.0 280.6
Powertrain mass (kg) 1698.8 1915.6 1747.0 2681.2
Systems mass (kg) 1981.0 3023.5 3024.6 3044.0
Landing gear (kg) 645.8 928.9 929.2 934.4
Payload mass (kg) 7550 7550 7550 7550
Finishing & oper. (kg) 2535.2 2535.2 2535.2 2535.2
Battery total mass (kg) – 7676.6 8400.7 8319.3
LH2 tank struct. mass (kg) – – 371.3 279.0

Table 11.7: Comparison of main design parameters for the four reference architectures. Powertaint
includes the mass of propeller, gearboxes, turboshafts, fuel cells, TMS and so on, the only excluded
mass is the batteries and the structural mass of the cryogenic hydrogen tank.

Based on this, we select a battery specific energy of 471Wh/kg and a nominal design range of 428nm
as the representative design point. The resulting geometries for this configuration are shown in
Fig. 11.15, with the main sizing data and the weight breakdown in Tab.11.7. As it is evident from the
data, the non-conventional hybrid architectures present an overall greater mass, this can be further
explored in Fig.11.18, where it can be seen that the addition of the batteries heavily penalized the
11.4. ARCHITECTURE SELECTION 179

Figure 11.17: Weight breakdown of the proposed aicraft architectures at 471Wh/kg and an on-design
mission range of 428nm. Powertaint includes the mass of propeller, gearboxes, turboshafts, fuel cells,
TMS and so on, the only excluded mass is the batteries and the structural mass of the cryogenic
hydrogen tank.

aircraft structural mass, and while going from kerosene to hydrogen leads to a reduction in fuel
load, the addition of the hydrogen tank and the higher battery allocation provides an overall higher
M T OW .

Figure 11.18: Comparison of the total equivalent CO2 emissions for the proposed architectures with
their optimized geometry.

Regarding the total emissions (see Fig.11.18), we can see that for our specific energy, range, and grid
180 CHAPTER 11. ARCHITECTURE COMPARISON

carbon intensity, the kerosene hybrid architecture generates more overall emissions. This is due to
the fact that at this range, the aircraft is unable to fully exploit the addition of the batteries, with the
removed fuel mass emissions being compensated by the battery emissions9 .
For this reason, the main pathway towards reducing emissions (at this range, specific energy and
energy grid) is the substitution of the kerosene fleet with a hydrogen aircraft fleet that should
first start with electrolysis direct hydrogen combustion, and as the technology matures, should
be substituted by fuel cells that can produce reductions in total emissions either with natural gas
reforming (that currently concerns the major percentage of the produced hydrogen) or by electrolysis
(more desirable).

Figure 11.19: Relative change of total equivalent CO2 emissions for the various architectures. As a
function of the specific energy and all relative to the optimized conventional kerosene. Red gradient
represents increasing emissions, and blue gradient represents decreasing emissions.

It is important to insist that this is very case sensitive, as if we were to consider an energy mix
from the USA (see Fig.11.14), we would be extremely penalized by the production of hydrogen by
electrolysis, and if we were to select an energy mix from Norway (see Fig.11.12) we would see that
kerosene hybrid could make sense. This trend can be seen in Fig.11.19, where we can see that the lines
of relative decreases (for a fixed range and specific energy) are a linear function of the grid carbon
intensity, with the kerosene hybrid being less pollutant than kerosene hybrid just at 136 gCO2 /kWh,
direct hydrogen combustion being less pollutant at values lower than 7 gCO2 /kWh (using hydrogen
from natural gas reforming) or 190 gCO2 /kWh (using hydrogen from electrolysis), and for fuel cells
being less pollutant at 244 gCO2 /kWh (for both electrolysis and natural gas reforming). What is
really interesting to see from this figure is the slopes, as we have that for kerosene hybrid and natural
gas reforming, we have a lower sensitivity10 to the specific energy, this is to be expected as the less
of the total energy here is storing grid energy, and we also have that from the electrolysis ones (the
one with the higher slopes) the hydrogen direct combustion is more affected as the turboshaft has a
9
It has to also be said that, as we saw in previous chapters, the addition of the TMS increases the fuel requirement due
to the TMS drag that, while diverting most of the fuel-flow still impacts the aircraft flow resistance.
10
First order derivative.
11.4. ARCHITECTURE SELECTION 181

lower recovery efficiency of the stored electrical energy at the hydrogen.

Precisely for this reason, if we were in the position of a regulatory body such as the European
Commission, the logical pathway would be to prioritize the development and deployment of fuel-
cell–based hybrid architectures. Unlike incremental improvements, this option directly incentivizes
the parallel decarbonization of the energy grid, ensuring that progress in aviation translates into
genuine reductions in emissions. In this sense, fuel-cell hybrids not only represent a technological
step forward for aircraft design but also act as a driver for systemic change in energy production,
making them a cornerstone for achieving meaningful and lasting decarbonization of the skies.
 CHAPTER

CONCLUSIONS 12
Contents
12.1 Overall Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
12.2 Future Work and Development . . . . . . . . . . . . . . . . . . . . . . . . . . 184

Throughout the previous chapters of this thesis, the development of new methods, the formulation of
different architectures, and the exploration of design spaces have provided valuable insights into the
trends and feasibility of novel regional aircraft concepts. The results show that such architectures are
technologically viable when considering near-future technologies. Among them, hydrogen-based
aircraft powered by hydrogen produced through electrolysis emerge as the most promising pathway.

12.1 OVERALL CONCLUSIONS

Regarding the specific objectives of this thesis, the development of new physics-based models—most
notably the thermal management system (TMS) and the multi-phase hydrogen tank model—proved
to be critical for ensuring realistic feasibility assessments. In particular, the TMS was essential
for capturing the drag and cooling penalties associated with batteries and fuel cells, while the
hydrogen tank model accounted for the structural, integration, and thermal aspects of cryogenic
storage. These models played a key role in shaping the battery sizing process and in shifting the
conditions under which battery installation became beneficial. By embedding these elements into
the MDAO framework, the simulations produced a more realistic representation of mass distribution,
aerodynamic penalties, and powertrain sizing across the candidate aircraft architectures.
For the second specific objective (the development of new aircraft architectures), four parametric
configurations were defined and systematically explored: the conventional kerosene baseline, a
kerosene parallel hybrid, a hydrogen direct-combustion hybrid, and a fuel-cell hybrid. Each was im-
plemented within the multidisciplinary analysis and optimization environment, ensuring consistent
treatment and enabling fair comparisons across the design space. This approach made it possible
to examine how each architecture responded to changes in range and battery technology, and to
analyze the trade-offs between mass, fuel burn, and total energy demand. Crucially, optimizing
a conventional kerosene reference alongside the novel concepts also allowed for direct emission

183
184 CHAPTER 12. CONCLUSIONS

reductions to be assessed between comparable optimized architectures, minimizing biases from the
modeling framework and avoiding misleading comparisons with existing aircraft that are sized for
different technology levels.
The third specific objective (assessing the environmental viability of the proposed architectures)
was addressed by extending the results to a life-cycle perspective, where mission fuel and energy
requirements were translated into equivalent CO2 emissions. This enabled a direct evaluation
of each configuration under different electricity mixes and hydrogen production pathways. The
findings revealed that kerosene hybrids, while technically feasible, generally fail to deliver emission
reductions once the carbon intensity of the supporting grid is accounted for. Hydrogen direct
combustion offered emission benefits in certain scenarios, particularly when hydrogen was produced
via electrolysis using clean electricity, though its efficiency limitations constrained the magnitude of
those gains. In contrast, fuel-cell hybrids consistently emerged as the most promising pathway: by
combining higher conversion efficiency with the potential for substantial emission reductions, they
directly tie aviation’s decarbonization to the broader effort of decarbonizing the electricity grid.
The overarching lesson from this work can be condensed into a single guiding motto: the de-
carbonization of the skies begins with the de-carbonization of the ground. The analysis has shown that
hybrid and hydrogen-based architectures can indeed deliver substantial reductions in emissions,
but only if the energy vectors they rely upon—electricity for batteries and electrolysis, or hydrogen
itself—are produced with low carbon intensity. In other words, the environmental benefit of replacing
kerosene with novel propulsion systems is not guaranteed by the aircraft alone, but by the cleanliness
of the energy system that supports it. Novel hybrid architectures therefore make sense only if
the substituting energy source is genuinely green, aligning the transformation of aviation with
the broader transition of global energy production. Without such parallel progress on the ground,
advances in aircraft technology risk being undermined, but with it, aviation can become a meaningful
contributor to climate neutrality.

12.2 FUTURE WORK AND DEVELOPMENT

Although the main objectives of this thesis have been achieved, several limitations remain. A
complete and accurate modeling of all subsystems in novel aircraft architectures is a vast task, and
in many cases could each justify an individual thesis. The most relevant limitations identified in this
work are the following:

• Mass estimation methods. Current estimates rely heavily on regression models derived from
conventional aircraft. Consequently, certain effects unique to novel architectures—such as the
additional structural loads imposed by new components, most notably cryogenic hydrogen
tanks—cannot be fully captured. Similarly, estimating secondary masses (auxiliary functional
and non-structural parts) remains challenging due to the limited availability of detailed weight
breakdowns and component-level data. This required approximations regarding the placement
and distribution of such masses, calibrated to the reference aircraft, but potentially introducing
offsets that are difficult to quantify. The absence of data from non-conventional aircraft, beyond
12.2. FUTURE WORK AND DEVELOPMENT 185

simple retrofits, amplifies this uncertainty.

• Electrical model. The electrical model did not include voltage, current, or resistance dis-
tribution across the circuits, preventing accurate wiring sizing and precluding an explicit
estimation of wiring mass (currently included only through systems mass regressions). In
addition, electric motor modeling was limited by the lack of available data for sizing state-of-
the-art aviation motors as a function of torque, RPM, and geometry. This prevented a detailed
trade-off between motor length and diameter, and constrained the use of motor efficiency
maps (available in prior studies, but excluded here due to the absence of sizing trade-offs).

• Scope of aircraft concepts. The study focused on regional turboprop-class concepts, and
although the trends identified provide meaningful insights, the lack of exploration of nar-
rowbody concepts (with higher passenger capacities but also operating on regional routes)
restricts the generality of the conclusions. For now, the findings are therefore most directly
applicable to regional aircraft, and further research is needed to extend the framework to
larger categories.

While the first limitation is inherently difficult to overcome (given the secrecy surrounding detailed
aircraft mass breakdowns and the lack of publicly available high-fidelity industrial data) the other
two points can be addressed through targeted model improvements. In particular, expanding the
electrical and thermal models with higher-fidelity estimations, and incorporating expertise from
specialists in different domains, would fully leverage the multidisciplinary and multi-departmental
nature of MDAO.
Finally, another important direction is the refinement of the automatic differentiation framework.
Although the most computationally demanding routines of the codebase have already been differen-
tiated, achieving a fully differentiable and pre-compiled pipeline remains an open challenge. Such
an advancement would significantly reduce optimization costs, improve scalability, and make the
MDAO framework far more suitable for large-scale design space explorations and industrial-level
applications.
BIBLIOGRAPHY
[1] International Energy Agency, Aviation – Energy System, Transport, Last updated on 16 January
2025, 2025. [Online]. Available: [Link]
aviation.

[2] European Commission, Reducing emissions from aviation, Accessed on 2025-02-21, 2025.
[Online]. Available: https : / / climate . ec . europa . eu / eu - action / transport /
reducing-emissions-aviation_en.

[3] International Civil Aviation Organization, Trends in Emissions that affect Climate Change,
Accessed on 2025-02-21, 2025. [Online]. Available: https : / / www . icao . int /
environmental-protection/Pages/ClimateChange_Trends.aspx.

[4] J. Martins and A. Lambe, “Multidisciplinary design optimization: A survey of architectures,”

AIAA Journal, vol. 51, pp. 2049–2075, Sep. 2013. doi: 10.2514/1.J051895.
[5] R. T. Haftka, “Optimization of flexible wing structures subject to strength and induced drag
constraints,” AIAA Journal, vol. 14, no. 8, pp. 1106–1977, 1977. doi: 10.2514/3.7400.
[6] R. T. Haftka, “Automated procedure for design of wing structures to satisfy strength and
flutter requirements,” NASA Langley Research Center, Hampton, VA, Tech. Rep. TN D-7264,
1973.
[7] R. T. Haftka, J. H. Starnes Jr., F. W. Barton, and S. C. Dixon, “Comparison of two types of
optimization procedures for flutter requirements,” AIAA Journal, vol. 13, no. 10, pp. 1333–
1339, 1975. doi: 10.2514/3.60545.
[8] R. T. Haftka and C. P. Shore, “Approximate methods for combined thermal/structural design,”
NASA, Tech. Rep. TP-1428, Jun. 1979.
[9] F. Bertini, L. D. Mas, L. Vassio, and E. Ampellio, Multidiscipinary optimization for gas turbines
design, 2014. arXiv: 1402.0420 [[Link]]. [Online]. Available: [Link]
abs/1402.0420.

[10] E. S. Hendricks and J. S. Gray, “Pycycle: A tool for efficient optimization of gas turbine engine
cycles,” Aerospace, vol. 6, no. 8, 2019, issn: 2226-4310. doi: 10.3390/aerospace6080087.
[Online]. Available: [Link]
[11] B. J. Brelje, “Multidisciplinary design optimization of electric aircraft considering systems
modeling and packaging,” PhD dissertation, University of Michigan, 2021.

187
188 BIBLIOGRAPHY

[12] C. David, S. Delbecq, S. Defoort, P. Schmollgruber, E. Benard, and V. Pommier-Budinger,

“From FAST to FAST-OAD: An open source framework for rapid overall aircraft design,” IOP
Conference Series: Materials Science and Engineering, vol. 1024, no. 1, p. 012 062, Jan. 2021.
doi: 10.1088/1757-899x/1024/1/012062. [Online]. Available: [Link]
10.1088/1757-899x/1024/1/012062.

[13] B. J. Brelje and J. R. R. A. Martins, “Development of a conceptual design model for aircraft
electric propulsion with efficient gradients,” in Proceedings of the AIAA/IEEE Electric Aircraft
Technologies Symposium, Cincinnati, OH, Jul. 2018. doi: 10.2514/6.2018-4979.
[14] C. Recine, J. Kirk, J. Jasa, and J. Gratz, NASA’s Aviary Takes Flight: A Public Software for Aircraft
Design, Presentation at AirVenture 2024, Oshkosh, WI. NASA Document ID: 20240009217,
Jul. 2024.
[15] T. Zill, D. Böhnke, and B. Nagel, “Preliminary aircraft design in a collaborative multidisci-
plinary design environment,” in 11th AIAA Aviation Technology, Integration, and Operations
(ATIO) Conference, AIAA Paper 2011-6959, Sep. 2011. doi: 10.2514/6.2011-6959.
[16] W. Li and K. Geiselhart, “Multiobjective multidisciplinary optimization of low-boom su-
personic transports using multifidelity models,” doi: 10 . 2514 / 6 . 2022 - 2097. eprint:
https : / / arc . aiaa . org / doi / pdf / 10 . 2514 / 6 . 2022 - 2097. [Online]. Available:
[Link]

[17] Y. Sun and H. Smith, “Low-boom low-drag optimization in a multidisciplinary design analysis
optimization environment,” Aerospace Science and Technology, vol. 94, p. 105 387, 2019, issn:
1270-9638. doi: [Link] [Online]. Available:
[Link]

[18] C. Thoulon, G. Roge, and O. Pironneau, “Gradient-based aero-stealth optimization of a sim-

plified aircraft,” Fluids, vol. 9, no. 8, 2024, issn: 2311-5521. doi: 10.3390/fluids9080174.
[Online]. Available: [Link]
[19] M. Li, J. Chen, X. Feng, F. Qu, and J. Bai, “An efficient adjoint method for the aero-stealth
shape optimization design,” Aerospace Science and Technology, vol. 118, p. 107 017, 2021, issn:
1270-9638. doi: [Link] [Online]. Available:
[Link]

[20] C. Yuan, D. Ma, Y. Jia, and L. Zhang, “Stealth unmanned aerial vehicle penetration efficiency
optimization based on radar detection probability model,” Aerospace, vol. 11, no. 7, 2024,
issn: 2226-4310. doi: 10.3390/aerospace11070561. [Online]. Available: [Link]
[Link]/2226-4310/11/7/561.

[21] J. S. Gray, G. K. Kenway, C. A. Mader, and J. R. R. A. Martins, “Aero-propulsive design

optimization of a turboelectric boundary layer ingestion propulsion system,” in 2018 Aviation
Technology, Integration, and Operations Conference. doi: 10.2514/6.2018-3976. eprint:
https : / / arc . aiaa . org / doi / pdf / 10 . 2514 / 6 . 2018 - 3976. [Online]. Available:
[Link]
BIBLIOGRAPHY 189

[22] A. H. Lamkin, A. Yildirim, J. R. Martins, and N. A. Wukie, “Advancements in coupled

aeropropulsive design optimization for high-bypass turbofan engines,” in AIAA AVIATION
2023 Forum. doi: 10.2514/6.2023-3591. eprint: [Link]
10.2514/6.2023-3591. [Online]. Available: [Link]
2514/6.2023-3591.

[23] Y.-Y. A. Ko, “The multidisciplinary design optimization of a distributed propulsion blended-
wing-body aircraft,” PhD dissertation, Virginia Polytechnic Institute and State University,
Blacksburg, Virginia, 2003.
[24] P. YU, J. PENG, J. BAI, X. HAN, and X. SONG, “Aeroacoustic and aerodynamic optimization
of propeller blades,” Chinese Journal of Aeronautics, vol. 33, no. 3, pp. 826–839, 2020, issn:
1000-9361. doi: [Link] [Online]. Available:
[Link]

[25] M. J. Groom, B. Y. Zhou, and L. V. Lopes, “Aeroacoustic optimization of installed propeller

configurations with scattering and shielding effects,” in 30th AIAA/CEAS Aeroacoustics
Conference (2024). doi: 10.2514/6.2024-3039. eprint: [Link]
pdf/10.2514/6.2024-3039. [Online]. Available: [Link]
10.2514/6.2024-3039.

[26] S. F. Pullin, A. Ghiglino, B. Y. Zhou, and M. Azarpeyvand, “Rapid optimisation of propellers

with tonal and broadband aeroacoustic constraints,” in AIAA AVIATION 2023 Forum, Paper
No. AIAA 2023-3834, American Institute of Aeronautics and Astronautics, 2023. doi: 10.
2514/6.2023-3834.

[27] D. Ingraham, J. Gray, and L. V. Lopes, “Gradient-based propeller optimization with acoustic
constraints,” in AIAA SciTech 2019 Forum, NASA Report No. GRC-E-DAA-TN63365, Document
ID: 20190004936, American Institute of Aeronautics and Astronautics, San Diego, CA, 2019.
[28] S. M. Hermansen and E. Lund, “Multi-material and thickness optimization of a wind turbine
blade root section,” Structural and Multidisciplinary Optimization, vol. 67, p. 107, 2024. doi:
10.1007/s00158-024-03811-0.

[29] D.-G. Caprace, A. Cardoza, A. Ning, M. Mangano, S. He, and J. R. Martins, “Incorporating
high-fidelity aerostructural analyses in wind turbine rotor optimization,” in AIAA SCITECH
2022 Forum. doi: 10.2514/6.2022-1290. eprint: [Link]
10.2514/6.2022-1290. [Online]. Available: [Link]
2514/6.2022-1290.

[30] A. Ojo, M. Collu, and A. Coraddu, “Multidisciplinary design analysis and optimization of
floating offshore wind turbine substructures: A review,” Ocean Engineering, vol. 266, p. 112 727,
2022, issn: 0029-8018. doi: [Link]
[Online]. Available: https : / / www . sciencedirect . com / science / article / pii /
S0029801822020108.
190 BIBLIOGRAPHY

[31] K. Patryniak, M. Collu, and A. Coraddu, “Multidisciplinary design analysis and optimisation
frameworks for floating offshore wind turbines: State of the art,” Ocean Engineering, vol. 251,
p. 111 002, 2022, issn: 0029-8018. doi: [Link]
111002. [Online]. Available: [Link]
pii/S0029801822004267.

[32] P. G. Anselma, C. B. Niutta, L. Mainini, M. Fioriti, T. De Filippi, and L. Gagliardi, “Multi-

disciplinary design optimization for hybrid electric vehicles: Component sizing and multi-
fidelity frontal crashworthiness,” Structural and Multidisciplinary Optimization, vol. 62, no. 6,
pp. 2149–2166, Oct. 2020. doi: 10.1007/s00158-020-02603-6.
[33] F. Farhadi, S. Wang, R. Palacin, and P. Blythe, “Data-driven multi-objective optimization for
electric vehicle charging infrastructure,” iScience, vol. 26, no. 10, p. 107 737, 2023, issn: 2589-
0042. doi: [Link] [Online]. Available:
[Link]

[34] M. Sandberg, J. Mukkavaara, F. Shadram, and T. Olofsson, “Multidisciplinary optimization

of life-cycle energy and cost using a bim-based master model,” Sustainability, vol. 11, no. 1,
2019, issn: 2071-1050. doi: 10.3390/su11010286. [Online]. Available: [Link]
[Link]/2071-1050/11/1/286.

[35] M. Park, Z. Wang, L. Li, and X. Wang, “Multi-objective building energy system optimization
considering ev infrastructure,” Applied Energy, vol. 332, p. 120 504, 2023, issn: 0306-2619.
doi: [Link] [Online]. Available:
[Link]

[36] I. R. Chittick and J. R. R. A. Martins, “An asymmetric suboptimization approach to aerostruc-

tural optimization,” Optimization and Engineering, vol. 10, no. 2, pp. 133–152, Mar. 2009. doi:
10.1007/s11081-008-9046-2.

[37] J. R. R. A. Martins and J. T. Hwang, A very short course on multidisciplinary design optimization,
Short course material, 2016.
[38] D. P. Raymer, “Enhancing aircraft conceptual design using multidisciplinary optimization,”
Ph.D. dissertation, Kungliga Tekniska Högskolan (Royal Institute of Technology), Stockholm,
Sweden, May 2002, isbn: 91-7283-259-2.
[39] J. S. Gray, J. T. Hwang, J. R. R. A. Martins, K. T. Moore, and B. A. Naylor, “OpenMDAO: An
open-source framework for multidisciplinary design, analysis, and optimization,” Structural
and Multidisciplinary Optimization, vol. 59, no. 4, pp. 1075–1104, Apr. 2019. doi: 10.1007/
s00158-019-02211-z.

[40] F. Gallard, C. Vanaret, D. Guénot, et al., “Gems: A python library for automation of multidis-
ciplinary design optimization process generation,” in 2018 AIAA/ASCE/AHS/ASC Structures,
Structural Dynamics, and Materials Conference, 2018.
[41] A. Wendorff, A. Variyar, C. Ilario, et al., Suave: An aerospace vehicle environment for designing
future aircraft, version 2.1, 2020. [Online]. Available: [Link]
SUAVE.
BIBLIOGRAPHY 191

[42] E. M. Botero and J. J. Alonso, “Conceptual design and optimization of small transitioning uavs
using suave,” in 18th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference.
doi: 10.2514/6.2017-4149. eprint: [Link]
6.2017- 4149. [Online]. Available: [Link]
2017-4149.

[43] M. Clarke, J. Smart, E. M. Botero, W. Maier, and J. J. Alonso, “Strategies for posing a well-
defined problem for urban air mobility vehicles,” in AIAA Scitech 2019 Forum. doi: 10.2514/
6.2019- 0818. eprint: [Link] 0818.
[Online]. Available: [Link]
[44] J. M. Vegh, E. Botero, M. Clarke, J. Smart, and J. Alonso, “Current capabilities and challenges
of ndarc and suave for evtol aircraft design and analysis,” in AIAA Propulsion and Energy
2019 Forum. doi: 10.2514/6.2019-4505. eprint: [Link]
10.2514/6.2019-4505. [Online]. Available: [Link]
2514/6.2019-4505.

[45] R. M. Erhard, M. A. Clarke, and J. J. Alonso, “A low-cost aero-propulsive analysis of distributed

electric propulsion aircraft,” in AIAA Scitech 2021 Forum. doi: 10.2514/6.2021- 1200.
eprint: https : / / arc . aiaa . org / doi / pdf / 10 . 2514 / 6 . 2021 - 1200. [Online].
Available: [Link]
[46] M. A. Clarke, R. M. Erhard, J. T. Smart, and J. Alonso, “Aerodynamic optimization of wing-
mounted propeller configurations for distributed electric propulsion architectures,” in AIAA
AVIATION 2021 FORUM. doi: 10.2514/6.2021-2471. eprint: [Link]
doi/pdf/10.2514/6.2021-2471. [Online]. Available: [Link]
abs/10.2514/6.2021-2471.

[47] M. Clarke and J. Alonso, “Lithium–ion battery modeling for aerospace applications,” Journal
of Aircraft, vol. 58, pp. 1–13, Jul. 2021. doi: 10.2514/1.C036209.
[48] M. Clarke and J. Alonso, “Forecasting the operational lifetime of battery-powered electric
aircraft,” Journal of Aircraft, vol. 60, pp. 1–9, Jun. 2022. doi: 10.2514/1.c036851.
[49] M. Clarke and J. Alonso, “Evaluating the performance and acoustic footprint of aircraft for
regional and urban air mobility,” Aug. 2021. doi: 10.2514/6.2021-3205.
[50] M. Clarke, E. Botero, J. Smart, and R. Erhard, RCAIDE-LEADS: Research Community Aircraft
Interdisciplinary Design Environment, https : / / github . com / leadsgroup / RCAIDE _
LEADS, Version: latest. Developed by the Lab for Electric Aircraft Design and Sustainability
and collaborators. Licensed under AGPL-3.0., 2025.
[51] B. J. Brelje, J. P. Jasa, J. R. R. A. Martins, and J. S. Gray, “Development of a conceptual-
level thermal management system design capability in OpenConcept,” in NATO Research
Symposium on Hybrid/Electric Aero-Propulsion Systems for Military Applications (AVT-RSY-
323), Trondheim, NO, Oct. 2019. doi: 10.14339/STO-MP-AVT-323.
192 BIBLIOGRAPHY

[52] E. J. Adler, B. J. Brelje, and J. R. R. A. Martins, “Thermal management system optimization

for a parallel hybrid aircraft considering mission fuel burn,” Aerospace, vol. 9, no. 5, Apr. 2022,
issn: 2226-4310. doi: 10.3390/aerospace9050243.
[53] E. J. Adler and J. R. R. A. Martins, “Aerostructural wing design optimization considering full
mission analysis,” in AIAA SciTech Forum, Jan. 2022. doi: 10.2514/6.2022-0382.
[54] E. J. Adler and J. R. R. A. Martins, “Efficient aerostructural wing optimization considering
mission analysis,” Journal of Aircraft, Dec. 2022, issn: 1533-3868. doi: 10.2514/1.c037096.
[55] E. J. Adler, A. H. R. Lamkin, and J. R. R. A. Martins, “Aircraft ducted heat exchanger aerody-
namic shape and thermal optimization,” ASME Journal of Heat and Mass Transfer, vol. 147,
no. 1, p. 011 902, Jan. 2025, issn: 2832-8450. doi: 10.1115/1.4066438.
[56] E. J. Adler and J. R. R. A. Martins, “Blended wing body configuration for hydrogen-powered
aviation,” Journal of Aircraft, vol. 61, no. 3, May 2024. doi: 10.2514/1.C037582.
[57] F. Lutz, J. Jézégou, A. Reysset, A. B. Ramos, and V. Pommier-Budinger, “Fast-oad-ga: An open-
source extension for overall aircraft design of general aviation aircraft,” in 33rd Congress of the
International Council of the Aeronautical Sciences (ICAS), Manuscript number: ffhal-03828230,
Stockholm, Sweden, Sep. 2022.
[58] M. Oak, A. Fabre, M. Delavenne, E. N. Van, E. Benard, and S. Defoort, “Aapplication of
FAST-OAD code to the conceptual design of a hydrogen fuelled commercial aircraft,” in 11th
EASN International Conference on Innovation in Aviation and Space for Opening New Horizons,
ser. IOP Conference Series: Materials Science and Engineering, vol. 1226, Virtual and Corfu,
Greece: IOP Publishing, Sep. 2022, p. 012 027. doi: 10.1088/1757-899X/1226/1/012027.
[59] R. Parello, Y. Gourinat, E. Benard, and S. Defoort, “Integrating cryogenic tanks model in
hydrogen aircraft design for parametric performance analysis,” Sep. 2024.
[60] R. Parello, S. Defoort, E. Benard, and Y. Gourinat, “Design and integration of a liquid hydrogen
tank on an aircraft,” Jan. 2024. doi: 10.2514/6.2024-2638.
[61] E. N. Van, S. Defoort, C. Viguier, S. Mezani, N. Takorabet, et al., “Design and performances of
a turbo-electric and distributed propulsion aircraft in the small and medium range segment:
Intermediate results from the european project imothep,” in AIAA SciTech 2024 Forum, HAL
Id: hal-04462774, Orlando, United States, Jan. 2024. doi: 10.2514/6.2024-2313.
[62] J. Gauvrit-Ledogar, A. Tremolet, F. Moens, et al., “Multidisciplinary design analysis and
optimization process dedicated to blended wing body configurations,” Sep. 2022.
[63] D. Hague, “GASP – General Aviation Synthesis Program. Volume 1: Main program. Part 1:
Theoretical development,” NASA, Washington, D.C., Contractor Report NASA-CR-152303-
VOL-1-PT-1, 1978, Contract/Grant: NAS2-9352.
[64] D. P. Wells, B. L. Horvath, and L. A. McCullers, “The Flight Optimization System Weights Es-
timation Method,” National Aeronautics and Space Administration, Hampton, VA, Technical
Memorandum NASA/TM-2017-219627/VOL1, Jun. 2017.
BIBLIOGRAPHY 193

[65] J. P. Jasa, J. T. Hwang, and J. R. R. A. Martins, “Open-source coupled aerostructural optimiza-

tion using Python,” Structural and Multidisciplinary Optimization, vol. 57, no. 4, pp. 1815–1827,
Apr. 2018. doi: 10.1007/s00158-018-1912-8.
[66] R. Falck, J. S. Gray, K. Ponnapalli, and T. Wright, “Dymos: A python package for optimal
control of multidisciplinary systems,” Journal of Open Source Software, vol. 6, no. 59, p. 2809,
2021. doi: 10.21105/joss.02809. [Online]. Available: [Link]
joss.02809.

[67] B. D. Phillips, J. N. Schmidt, R. D. Falck, and E. Aretskin-Hariton, “Design under uncertainty

for conceptual aircraft design leveraging analytical gradients,” in AIAA SCITECH 2025 Forum.
doi: 10.2514/6.2025-1956. eprint: [Link]
6.2025- 1956. [Online]. Available: [Link]
2025-1956.

[68] National Aeronautics and Space Administration, UQPCE: Uncertainty Quantification with
Polynomial Chaos Expansion, [Link] Version 1.0.0, LAR-
20514-1, 2024.
[69] M. Drela, TASOPT: Transport Aircraft System OPTimization, https : / / web . mit . edu /
drela / Public / web / tasopt/, Version 2.16. Released under the GNU General Public
License, 2020.
[70] R. Quiben Figueroa, R. Cavallaro, and A. Cini, “Feasibility studies on regional aircraft
retrofitted with hybrid-electric powertrains,” Aerospace Science and Technology, vol. 151,
p. 109 246, 2024, issn: 1270-9638. doi: https : / / doi . org / 10 . 1016 / j . ast . 2024 .
109246. [Online]. Available: [Link]
pii/S1270963824003778.

[71] AGILE 4.0 Consortium, AGILE 4.0: Towards an Integrated Cyber-Physical Aeronautical Supply
Chain, [Link] European Union Horizon 2020 Research
and Innovation Programme, Grant Agreement No. 815122, 2025.
[72] COLOSSUS Consortium, COLOSSUS: Collaborative System-of-Systems Exploration of Aviation
Products, Services and Business Models, [Link] sos- [Link]/, Euro-
pean Union Horizon Europe Research and Innovation Programme, Grant Agreement No.
101097120. Swiss participation supported by SERI under contract No. 22.00609, 2023.
[73] ODE4HERA Consortium, ODE4HERA: Open Digital Environment for Hybrid-Electric Regional
Architectures, [Link] Co-funded by the European Union under
Grant Agreement No. 101140510, 2024.
[74] P. N. Simon, R. Q. Figueroa, A. Cini, and R. Cavallaro, “Multidisciplinary design and optimiza-
tion of h2-powered regional aircraft architectures,” in AIAA SCITECH 2025 Forum. doi: 10.
2514/6.2025-0360. eprint: [Link]
0360. [Online]. Available: [Link]

[75] H. Ritchie, “What share of global co emissions come from aviation?” Our World in Data, 2024,
[Link]
194 BIBLIOGRAPHY

[76] M. Klöwer, M. Allen, D. Lee, S. Proud, L. Gallagher, and A. Skowron, “Quantifying aviation’s
contribution to global warming,” Environmental Research Letters, vol. 16, p. 104 027, Oct. 2021.
doi: 10.1088/1748-9326/ac286e.
[77] E. Terrenoire, D. Hauglustaine, T. Gasser, and O. Penanhoat, “The contribution of carbon
dioxide emissions from the aviation sector to future climate change,” Environmental Research
Letters, vol. 14, Aug. 2019. doi: 10.1088/1748-9326/ab3086.
[78] M. Weinold, P. Rohrer, and R. McKenna, “Aircraft fuel efficiency improvements: The pathway
to evidence-based forecasts,” in Aerospace Europe Conference 2023 – 10 EUCASS – 9 CEAS,
Paul Scherrer Institute and ETH Zurich, EUCASS, Lausanne, Switzerland, 2023. doi: 10.
13009/EUCASS2023-466.

[79] C. Bergero, G. Gosnell, D. Gielen, S. Kang, M. Bazilian, and S. J. Davis, “Pathways to net-
zero emissions from aviation,” Nature Sustainability, vol. 6, no. 4, pp. 404–414, 2023. doi:
10.1038/s41893-023-01080-1.

[80] International Air Transport Association (IATA), Net zero 2050: Sustainable aviation fuels – fact
sheet, [Link] Accessed: 2025-05-12, Dec. 2024. [Online].
Available: [Link]
[81] International Civil Aviation Organization (ICAO), Annex 16 — Environmental Protection,
Volume IV, 2nd ed. Montréal, Canada: International Civil Aviation Organization, 2023, p. 124,
Standards and Recommended Practices on CO emissions monitoring and offsetting for
international aviation, isbn: 978-92-9258-611-9.
[82] M. Aigner, P. L. Clercq, and C. Arndt, Sustainable aviation fuels (saf): Options and chal-
lenges, Presented at the ISCC Global Sustainability Conference, ISCC Global Sustainability
Conference, February 24, 2021, Feb. 2021.
[83] K. Moriarty and R. McCormick, “Sustainable aviation fuel blending and logistics,” National
Renewable Energy Laboratory (NREL), Golden, CO, Tech. Rep. NREL/TP-5400-90979, Sep.
2024, Prepared for the U.S. Department of Energy, Office of Energy Efficiency and Renewable
Energy, Bioenergy Technologies Office. [Online]. Available: [Link]
docs/fy24osti/[Link].

[84] K. Seymour, M. Held, B. Stolz, G. Georges, and K. Boulouchos, “Future costs of power-to-
liquid sustainable aviation fuels produced from hybrid solar pv-wind plants in europe,”
Sustainable Energy Fuels, vol. 8, pp. 811–825, 4 2024. doi: 10.1039/D3SE00978E. [Online].
Available: [Link]
[85] F. Yang and Y. Yao, “Sustainable aviation fuel pathways: Emissions, costs and uncertainty,”
Resources, Conservation and Recycling, vol. 215, p. 108 124, 2025, issn: 0921-3449. doi: https:
/ / doi . org / 10 . 1016 / j . resconrec . 2025 . 108124. [Online]. Available: https :
//[Link]/science/article/pii/S0921344925000035.
BIBLIOGRAPHY 195

[86] International Civil Aviation Organization (ICAO), Sustainable aviation fuels guide, Part of
the Transforming Global Aviation Collection (4 of 4), Produced within the ICAO-UNDP-
GEF project: Transforming the Global Aviation Sector: Emissions Reductions from Inter-
national Aviation, 2017. [Online]. Available: [Link]
protection/pages/[Link].

[87] O. R. Calderon, L. Tao, Z. Abdullah, et al., “Sustainable aviation fuel (saf) state-of-industry
report: State of saf production process,” National Renewable Energy Laboratory (NREL),
Golden, CO, Tech. Rep. NREL/TP-5100-87802, Jul. 2024, Prepared for the U.S. Department
of Energy, Office of Energy Efficiency and Renewable Energy. [Online]. Available: https:
//[Link]/docs/fy24osti/[Link].

[88] International Civil Aviation Organization (ICAO), Corsia supporting document: Corsia eligible
fuels – life cycle assessment methodology, version Version 5, Referenced in Annex 16 —
Environmental Protection, Volume IV, Part II, Paragraph 3.3.2, Jun. 2022. [Online]. Available:
https : / / www . icao . int / environmental - protection / CORSIA / Documents /
CORSIA_LCA_Methodology_V5.pdf.

[89] Heart Aerospace, ES-30™: Introducing the ES-30 – A clean-sheet hybrid-electric design engi-
neered to reignite regional aviation, [Link] Accessed:
2025-05-14, Fredsflottiljens Väg 19, 417 46 Göteborg, Sweden, 2025.
[90] ATR, ATR 42-600: The Ideal Local Commuter, https : / / www . atr - aircraft . com /
aircraft-services/aircraft-family/atr-42-600/, Accessed: 2025-05-14, 1 Allée
Pierre Nadot, 31712 Blagnac Cedex, France, 2025.
[91] Embraer Commercial Aviation, E190: Commercial Jet, https : / / www .
[Link]/commercial-jets/e190/, Accessed: 2025-05-14,
São José dos Campos, Brazil, 2025.
[92] J. Holladay, Z. Abdullah, and J. Heyne, “Sustainable aviation fuel: Review of technical
pathways,” U.S. Department of Energy, Bioenergy Technologies Office, Tech. Rep., 2020,
Prepared by Pacific Northwest National Laboratory, National Renewable Energy Laboratory,
and University of Dayton. [Online]. Available: https : / / www . energy . gov / eere /
bioenergy/sustainable-aviation-fuel-review-technical-pathways.

[93] X. Vigor, M. Petitjean, D. Hertogs, D. Roma, and P. Crespi, “Renewable hydrogen for avi-
ation,” in ICAO Environmental Report 2022, Chapter Seven: Climate Change Mitigation –
Sustainable Aviation Fuels, Published by ICAO, available online at [Link]
int/environmental-protection/Pages/[Link], International Civil
Aviation Organization (ICAO), 2022, pp. 220–223.
[94] “Incident report - auxiliary power unit battery fire japan airlines boeing 787-8, ja829j boston,
massachusetts, january 7, 2013,” National Transportation Safety Board, Tech. Rep., 2014,
NTSB Aviation Incident Report.
196 BIBLIOGRAPHY

[95] D. Hernandez Lugo, S. Clarke, T. Miller, M. Redifer, and T. Foster, “X-57 maxwell battery:
From cell level to system level design and testing,” in Proceedings of the AIAA/IEEE Electric
Aircraft Technologies Symposium (EATS), Presented by Dr. Dionne Hernandez Lugo, American
Institute of Aeronautics and Astronautics, Cincinnati, OH, 2018.
[96] A. Scerri. “A look at zenlabs’ evtol battery technology that caught lilium’s attention.” In-
terview with Michael Sinkula, Zenlabs Energy Inc. (Jun. 2022), [Online]. Available: https:
//[Link]/q-and-a/look-at-zenlabs-evtol-battery-technology-
that-caught-lilium-attention/.

[97] H. Kühnelt, F. Mastropierro, N. Zhang, S. Toghyani, and U. Krewer, “Are batteries fit for
hybrid-electric regional aircraft?” Journal of Physics: Conference Series, vol. 2526, no. 1,
p. 012 026, Jun. 2023. doi: 10.1088/1742- 6596/2526/1/012026. [Online]. Available:
[Link]

[98] Elysian Aircraft, E9X: 90-Passenger Battery-Electric Aircraft Concept, https : / / www .
[Link]/, Accessed: 2025-05-14, Hoeksteen 40, 2132MS Hoofddorp, The
Netherlands, 2024.
[99] K. Milios, C. Hall, A. Burrell, et al., “Modeling and simulation of a parallel hybrid-electric
propulsion system - electrified powertrain flight demonstration (epfd) program,” in 2022 IEEE
Transportation Electrification Conference and Expo (ITEC), 2022, pp. 682–687. doi: 10.1109/
ITEC53557.2022.9813981.

[100] J. Gladin, E. Aydin, J. K. Jr., et al., “Reassessment of parametric optimization and performance
analysis methodology for mild hybrid aircraft propulsion systems,” in Proceedings of the
34th Congress of the International Council of the Aeronautical Sciences (ICAS), Florence, Italy,
2024. [Online]. Available: [Link]
papers/icas2024_1172_paper.pdf.

[101] D. F. Finger, C. Bil, and C. Braun, “Initial sizing methodology for hybrid-electric general
aviation aircraft,” Journal of Aircraft, vol. 57, no. 2, pp. 245–255, 2020.
[102] K. Abu Salem, G. Palaia, and A. A. Quarta, “Review of hybrid-electric aircraft technolo-
gies and designs: Critical analysis and novel solutions,” Progress in Aerospace Sciences,
vol. 141, p. 100 924, 2023, Special Issue on Green Aviation, issn: 0376-0421. doi: https :
/ / doi . org / 10 . 1016 / j . paerosci . 2023 . 100924. [Online]. Available: https :
//[Link]/science/article/pii/S0376042123000404.

[103] A. Batra, R. Raute, and R. Camilleri, “Series or parallel hybrid-electric aircraft propulsion
systems? case studies of the atr42 and atr72,” in Proceedings of the 33rd Congress of the
International Council of the Aeronautical Sciences (ICAS), Stockholm, Sweden, Sep. 2022.
[Online]. Available: https : / / www . icas . org / icas _ archive / ICAS2022 / data /
papers/ICAS2022_0163_paper.pdf.
BIBLIOGRAPHY 197

[104] M. Fouda, E. J. Adler, J. H. Bussemaker, et al., “Automated hybrid propulsion model con-
struction for conceptual aircraft design and optimization,” in Proceedings of the 33rd Congress
of the International Council of the Aeronautical Sciences (ICAS), Stockholm, Sweden, Sep.
2022, pp. 1249–1269. [Online]. Available: https : / / www. icas . org / icas _ archive /
ICAS2022/data/papers/ICAS2022_0593_paper.pdf.

[105] P. N. Simon and R. Cavallaro, “Local air quality and noise improvements via optimization
of strut-braced wings with distributed electric propulsion,” in AIAA SCITECH 2025 Forum.
doi: 10.2514/6.2025-0456. eprint: [Link]
6.2025- 0456. [Online]. Available: [Link]
2025-0456.

[106] Airbus, Daher, and Safran, EcoPulse: A new approach to distributed propulsion for aircraft,
[Link] transition/hybrid- and-
electric - flight / ecopulse, A hybrid-electric aircraft demonstrator developed by
Airbus, Daher, and Safran to evaluate distributed propulsion technologies. Project ended in
December 2024 after 50 test flights., 2025.
[107] Ampaire Inc., Eco Caravan: Plug-in hybrid-electric aircraft with amp-h570 powertrain, https:
//[Link]/vehicles/eco-caravan, A 9-seat hybrid-electric aircraft based
on the Cessna 208B Grand Caravan, equipped with the AMP-H570 powertrain. Offers up to
70% fuel savings and 40% OPEX reduction, with no need for charging infrastructure., 2024.
[108] Ampaire Inc., AMP-H570: Hybrid-electric engine for regional aviation, https : / / www .
[Link]/amp-h570, A hybrid-electric engine delivering 764 SHP (570 kW) at takeoff
and reducing fuel burn by 50–70%. First hybrid-electric engine to reach FAA G1 certification
stage. Designed for retrofit at cost parity with traditional PT6 engines., 2024.
[109] E. W. Hall and A. Silverstein, “Liquid hydrogen as a jet fuel for high-altitude aircraft,”
National Advisory Committee for Aeronautics, Cleveland, OH, Research Memorandum
NACA-RM-E55C28A, Apr. 1955.
[110] J. L. Sloop, “Liquid hydrogen as a propulsion fuel, 1945–1959,” Scientific, Technical Informa-
tion Office, National Aeronautics, and Space Administration, Washington, D.C., Tech. Rep.
NASA SP-4404, Jan. 1978.
[111] D. B. Snow, B. D. Avery, L. A. Bodin, P. Baldasare, and G. F. Washburn, “A study of subsonic
transport aircraft configurations using hydrogen (h2 ) and methane (ch4 ) as fuel,” NASA
Langley Research Center, Hampton, Virginia, Tech. Rep. NASA TM X-71994, Aug. 1974,
NASA Technical Memorandum.
[112] U.S. Department of Energy, Hydrogen and Fuel Cell Technologies Office, Hydrogen storage,
[Link] Accessed: 2025-
06-02, 2025.
[113] G. D. Brewer, Hydrogen Aircraft Technology. Boca Raton, FL: CRC Press, 1991. doi: 10.1201/
9780203751480.
198 BIBLIOGRAPHY

[114] flateric et al., Tupolev tu-155 internal layout and okb drawings, Secret Projects Forum, Thread
started by Antonio, contributions by flateric et al., 2008. [Online]. Available: [Link]
[Link]/threads/tupolev-tu-155.3889/ (visited on 08/16/2025).

[115] Energie Inside. “At tupolev, the hydrogen future happened 30 years ago.” Swiss maga-
zine for energy topics. (Nov. 2023), [Online]. Available: [Link]
ch / e - mobilitaet / energiespeicher / bei - tupolew - fand - die - wasserstoff -
zukunft-vor-30-jahren-statt (visited on 08/16/2025).

[116] V. SOSOUNOV and V. ORLOV, “Experimental turbofan using liquid hydrogen and liquid
natural gas as fuel,” in 26th Joint Propulsion Conference. doi: 10 . 2514 / 6 . 1990 - 2421.
eprint: https : / / arc . aiaa . org / doi / pdf / 10 . 2514 / 6 . 1990 - 2421. [Online].
Available: [Link]
[117] Directorate-General for Defence Industry and Space, European Commission, The alliance
for zero-emission aviation launches its vision towards electric and hydrogen flight in europe,
[Link]
aviation- launches- its- vision- towards- electric- and- hydrogen- flight-
europe-2024-06-06_en, Accessed: 2025-06-02, Jun. 2024.

[118] European Commission, Ground operations of liquid hydrogen aircraft (goliat), https://
[Link]/project/id/101138379, Horizon Europe Project ID: 101138379,
coordinated by Airbus, funded under the Climate, Energy and Mobility programme, Apr.
2024. doi: 10.3030/101138379.
[119] Rolls-Royce Deutschland Ltd and Co KG, Consortium for the advent of aero-engine demonstra-
tion and aircraft integration strategy with hydrogen (cavendish), [Link]
eu/project/id/101102000, Grant Agreement ID: 101102000, Funded under Horizon
Europe – Climate, Energy and Mobility, Jan. 2023. doi: 10.3030/101102000.
[120] European Commission, Fuselage, rear fuselage and empennage with cabin and cargo archi-
tecture solution validation and technologies for h2 integration (faster-h2), [Link]
[Link]/project/id/101101978, Horizon Europe Project ID: 101101978, coordinated
by Airbus Operations GmbH, funded under the Climate, Energy and Mobility programme,
Jan. 2023. doi: 10.3030/101101978.
[121] European Commission, Hydrogen lightweight and innovative tank for zero-emission aircraft
(h2elios), https : / / cordis . europa . eu / project / id / 101102003, Horizon Europe
Project ID: 101102003, coordinated by Aciturri Engineering SL, funded under the Climate,
Energy and Mobility programme, Jan. 2023. doi: 10.3030/101102003.
[122] European Commission, Hydrogen-electric zero emission propulsion system (herops), https:
//[Link]/project/id/101140499, Horizon Europe Project ID: 101140499,
coordinated by MTU Aero Engines AG, funded under the Climate, Energy and Mobility
programme, Jan. 2024. doi: 10.3030/101140499.
BIBLIOGRAPHY 199

[123] Airbus, Zeroe: Our hydrogen-powered aircraft, https : / / www . airbus . com / en /
innovation / energy - transition / hydrogen / zeroe - our - hydrogen - powered -
aircraft, Accessed: 2025-08-16, 2025.

[124] Universal Hydrogen, “Lightning McClean,” Universal Hydrogen’s zero-emission Dash 8-300,
cruising the skies of Mojave, Calif. Photograph, Universal Hydrogen Photo, 2023.
[125] M. C. Cameretti, R. De Robbio, V. Ferrara, and R. Tuccillo, “Performance and emissions
evaluation of a turbofan burner with hydrogen fuel,” Aerospace, vol. 12, no. 3, 2025, issn:
2226-4310. doi: 10 . 3390 / aerospace12030231. [Online]. Available: https : / / www .
[Link]/2226-4310/12/3/231.

[126] European Commission, Hybrid-electric regional architecture (hera), https : / / cordis .

europa . eu / project / id / 101102007, Grant Agreement ID: 101102007, DOI:
10.3030/101102007, Jan. 2023.
[127] G. Grazioso, M. Di Stasio, F. Nicolosi, and S. Trepiccione, A mathematical model for hybrid-
electric propulsion system for regional propeller-driven aircraft, Mar. 2025. doi: 10.1016/j.
ecmx.2025.100957. [Online]. Available: [Link]
100957.

[128] M. Schröder, F. Becker, and C. Gentner, “Optimal design of proton exchange membrane fuel
cell systems for regional aircraft,” Energy Conversion and Management, vol. 308, p. 118 338,
2024, issn: 0196-8904. doi: [Link]
[Online]. Available: https : / / www . sciencedirect . com / science / article / pii /
S0196890424002796.

[129] C. K. Sain, S. Kazula, and L. Enghardt, “Electric propulsion for regional aircraft – critical
components and challenges,” in Deutscher Luft- und Raumfahrtkongress 2022, DocumentID:
570151, German Aerospace Center (DLR), Germany, 2022.
[130] H. Degirmenci, A. Uludag, S. Ekici, and T. H. Karakoc, “Challenges, prospects and potential
future orientation of hydrogen aviation and the airport hydrogen supply network: A state-
of-art review,” Progress in Aerospace Sciences, vol. 141, p. 100 923, 2023, Special Issue on
Green Aviation, issn: 0376-0421. doi: [Link]
100923. [Online]. Available: [Link]
pii/S0376042123000398.

[131] H. Ritchie and P. Rosado, “Electricity mix,” Our World in Data, 2020,
[Link]
[132] M. Espitalier-Noël, J. Fonseca, D. Fraile, et al., “Clean hydrogen monitor 2024,” Hydrogen
Europe, 2024, Design: Think Things Studio Barcelona. [Online]. Available: [Link]
[Link]/clean-hydrogen-monitor-2024/.

[133] “The european hydrogen market landscape 2023,” Clean Hydrogen Joint Undertaking, 2024,
Prepared as part of the European Hydrogen Observatory. [Online]. Available: https://
[Link]/.
200 BIBLIOGRAPHY

[134] “Global hydrogen review 2024,” International Energy Agency, 2024. [Online]. Available:
[Link]

[135] K. Mazloomi, N. B. Sulaiman, and H. Moayedi, “Electrical efficiency of electrolytic hydrogen

production,” International Journal of Electrochemical Science, vol. 7, no. 4, pp. 3314–3326,
2012, issn: 1452-3981. doi: [Link] 3981(23)13957- 5.
[Online]. Available: https : / / www . sciencedirect . com / science / article / pii /
S1452398123139575.

[136] International Energy Agency, Comparison of the emissions intensity of different hydro-
gen production routes, 2021, https : / / www . iea . org / data - and - statistics /
charts/comparison-of-the-emissions-intensity-of-different-hydrogen-
production-routes-2021, Last updated 29 Jun 2023, Licence: CC BY 4.0, 2023.

[137] Binter Canarias. “Flota de binter canarias.” Spanish. Accedido el 4 de julio de 2025, Bin-
ter Canarias. (2025), [Online]. Available: https : / / www . bintercanarias . com / es /
corporativo/flota.

[138] LOT Polish Airlines. “History of the bombardier q400.” Accessed: 2025-07-04, LOT Polish
Airlines. (2023), [Online]. Available: [Link]
lot/fleet/bombardier-q400.

[139] Performance Review Commission, Performance Review Report 2024: An Assessment of Air
Traffic Management in Europe, Report, Published by EUROCONTROL, Brussels. Available
at [Link] navigation- services- performance-
review, EUROCONTROL, Mar. 2025.

[140] G. Atanasov, Aircraft modeling results, shortrange turboprop baselines (d250-321tp-2040, d220-
320tp-2040), EXACT Mid Term Review, Presented on March 31st, 2022, 2022.
[141] G. Atanasov, D. Silberhorn, P. Wassink, et al., “Short medium range turboprop-powered
aircraft as an enabler for low climate impact,” in AIAA AVIATION 2023 Forum, Session: Ultra
Efficient Aircraft Design and Dynamics, AIAA, 2023.
[142] B. Graver, D. Rutherford, and S. Zheng, “CO2 Emissions from Commercial Aviation: 2013,
2018, and 2019,” International Council on Clean Transportation, Washington, DC, Tech. Rep.,
Oct. 2020, Supplemental data available on the ICCT website. [Online]. Available: https:
//[Link].

[143] Actualidad Aeroespacial. “Image of dash 8-400 aircraft.” Image published in the news article
Longview Aviation suspends production of the Dash 8-400 and Twin Otter. (Mar. 24, 2020),
[Online]. Available: [Link]
suspende-la-produccion-de-los-dash-8-400-y-twin-otter/.

[144] ATR. “Image of atr 72-600 aircraft.” Image published on the official ATR website, ATR 72-600
Aircraft Family. (Jun. 20, 2025), [Online]. Available: [Link]
aircraft-services/aircraft-family/atr-72-600/.
BIBLIOGRAPHY 201

[145] R. D. Finck, “USAF Stability and Control DATCOM,” McDonnell Douglas Corporation, Dou-
glas Aircraft Division, Wright-Patterson Air Force Base, Ohio, Technical Report AF33(616)-
6460, F33615-76-C-3061, Apr. 1978, Originally published October 1960. Revised April 1978.
Project No. 8219, Task No. 821901.
[146] D. P. Raymer, Aircraft design: A conceptual approach (AIAA education series). Reston, Va.:
American Institute of Aeronautics and Astronautics, 2006, isbn: 1563478293. [Online]. Avail-
able: [Link]
[147] E. Torenbeek, Synthesis of Subsonic Airplane Design. Delft: Delft University Press, 1988.
[148] EUROCONTROL, Base of aircraft data (bada), [Link]
bada, Accessed: 2025-06-17.

[149] R. A. McDonald and J. R. Gloudemans, “Open vehicle sketch pad: An open source parametric
geometry and analysis tool for conceptual aircraft design,” in AIAA SCITECH 2022 Forum.
doi: 10.2514/6.2022-0004. eprint: [Link]
6.2022- 0004. [Online]. Available: [Link]
2022-0004.

[150] N. S. Pablo, R. Cavallaro, D. Zheng Zhang, and J. Brižs, “Multidisciplinary design and opti-
mization of strut-braced hybrid-electric aircraft under multi-trajectory scenarios,” in AeroBest
2025 III ECCOMAS Thematic Conference on Multidisciplinary Design Optimization of Aerospace
Systems, Conference proceeding, Version v1, Creative Commons Attribution 4.0 International
License, Lisbon, Portugal: Zenodo, Apr. 2025. doi: 10.5281/zenodo.15182834. [Online].
Available: [Link]
[151] M. Drela and H. Youngren, XROTOR: An interactive program for the design and analysis of
ducted and free-tip propellers and windmills, [Link]
web/xrotor/, Version 7.69, Accessed: 2025-03-31, 2022. [Online]. Available: [Link]
[Link]/drela/Public/web/xrotor/.

[152] E. E. Larrabee, “A pictorial account of the fundamentals of propeller theory and its application
to the design of a pedal driven airplane propeller,” Massachusetts Institute of Technology,
Cambridge, Massachusetts, Tech. Rep., Jan. 1980, Lecture notes or internal report.
[153] H. Glauert, Elements of Aerofoil and Airscrew Theory. Cambridge: Cambridge University
Press, 1926.
[154] H. Glauert, “Airplane propellers,” in Aerodynamic Theory, ser. Division L, W. F. Durand, Ed.,
vol. IV, Also reprinted by Dover Publications, Springer-Verlag, 1935.
[155] EASA, “Type-certificate data sheet no. easa.a.084 for atr 42 and atr 72,” European Union
Aviation Safety Agency (EASA), Cologne, Germany, Type-Certificate Data Sheet EASA.A.084,
Sep. 2023, Issue 12, 7 September 2023. Type Certificate Holder: ATR-GIE Avions de Transport
Régional.
202 BIBLIOGRAPHY

[156] ATR-GIE Avions de Transport Régional, Atr 72 flight crew operating manual (fcom), Manual,
Flight Crew Operating Manual, July 1999 edition, ATR-GIE Avions de Transport Régional,
Jul. 1999.
[157] EASA, “Type-certificate data sheet no. [Link].a.191 for dhc-8,” European Union Aviation
Safety Agency (EASA), Cologne, Germany, Type-Certificate Data Sheet [Link].A.191, Feb.
2023, Issue 16, 03 Feb 2023. Type Certificate Holder: De Havilland Aircraft of Canada Limited.
[158] L. J. Stauber and C. G. Naiman, Numerical propulsion system simulation (npss): An award
winning propulsion system simulation tool, NASA Glenn Research Center, Cleveland, OH,
Published in *Research and Technology 2001*, Mar. 2002. [Online]. Available: https://
[Link]/citations/20050214739.

[159] European Union Aviation Safety Agency (EASA), “Type Certificate Data Sheet No. IM.E.041
for Pratt and Whitney Canada PW100 Series Engines,” European Union Aviation Safety
Agency, Cologne, Germany, Type Certificate Data Sheet IM.E.041, version Issue 07, Dec. 2023,
[Link].00052-001 © European Union Aviation Safety Agency, 2019. All rights reserved.
ISO9001 Certified. [Online]. Available: [Link]
library/ty....

[160] M. Ank, A. Sommer, K. Abo Gamra, et al., “Lithium-ion cells in automotive applications:
Tesla 4680 cylindrical cell teardown and characterization,” Journal of The Electrochemical
Society, vol. 170, no. 12, p. 120 536, Dec. 2023. doi: 10.1149/1945-7111/ad14d0. [Online].
Available: [Link]
[161] Batemo GmbH, Samsung INR18650-30Q: High-Precision Battery Model, Data, and Report,
https : / / www . batemo . com / products / batemo - cell - explorer / samsung -
inr18650-30q/, Accessed: 2025-05-25, 2021. [Online]. Available: [Link]
com/products/batemo-cell-explorer/samsung-inr18650-30q/.

[162] Batemo GmbH, Amprius sa03 lithium-ion battery cell, [Link]

amprius-sa03, Datasheet and Batemo Cell Model v1.513 (Released 30 June 2025), 2025.

[163] Batemo GmbH, A123 systems 26700 nca gf lithium-ion battery cell, [Link]
com / es / products / batemo - cell - explorer / a123 - systems - 26700 - nca - gf/,
Datasheet, Batemo Cell Model v1.313 (Released 30 September 2024), and Teardown Report,
2024.
[164] Batemo GmbH, Farasis p79b3 lithium-ion battery cell, [Link]
products/batemo-cell-explorer/farasis-p79b3/, Datasheet, Batemo Cell Model
v1.312 (Released 31 March 2024), and Teardown Report, 2024.
[165] Batemo GmbH, V4drive v21700hep-5.x(a0) lithium-ion battery cell, [Link]
com / es / products / batemo - cell - explorer / v4drive - v21700hep - 5 - xa0/,
Datasheet and Batemo Cell Model (validation pending), 2025.
[166] J. Chin, K. Look, E. McNichols, D. Hall, J. Gray, and S. Schnulo, “Battery cell-to-pack scaling
trends for electric aircraft,” Aug. 2021. doi: 10.2514/6.2021-3316.
BIBLIOGRAPHY 203

[167] F. P. Incropera, D. P. DeWitt, T. L. Bergman, and A. S. Lavine, Fundamentals of Heat and Mass
Transfer, 6th ed. Hoboken, NJ: John Wiley and Sons, 2007, isbn: 978-0-471-45728-2.
[168] C. L. Pastra, C. Hall, G. Cinar, J. Gladin, and D. N. Mavris, “Specific power and efficiency
projections of electric machines and circuit protection exploration for aircraft applications,”
in 2022 IEEE Transportation Electrification Conference and Expo (ITEC), 2022, pp. 766–771.
doi: 10.1109/ITEC53557.2022.9813927.
[169] O. Otto, “Electrifying propulsion,” Siemens eAircraft, Tech. Rep., 2018, Restricted © Siemens
AG 2018.
[170] R. Jackson, “Magnix magni350, 650 and magnidrive 100,” E-Mobility Engineering, 2021, Feature
article on magniX electric propulsion units for aviation.
[171] “Type-certificate data sheet no. e.251 for engineus100 engine series (model engineus100b1),”
European Union Aviation Safety Agency (EASA), Tech. Rep. E.251, Issue 02, 2025, TCDS
issued 17 February 2025. Type Certificate Holder: Safran Electrical and Power S.A., Blagnac
Cedex, France.
[172] H3X Technologies Inc., Hpdm-350: 350 kw high power density machine, https : / / h3x .
tech/products/hpdm-350, Datasheet. 350 kW electric motor with integrated inverters,
fault-tolerant multisector architecture, and 7 kW/kg specific power., 2025.
[173] H3X Technologies Inc., Hpdm-1500: 1.5 mw high power density machine, [Link]
tech/products/hpdm-1500, Datasheet. 1.5 MW electric motor with integrated inverters,
12 kW/kg specific power, and eight independent drive sectors., 2025.
[174] H3X Technologies Inc., Hpdm-2300: 2.3 mw high power density machine, [Link]
tech/products/hpdm-2300, Datasheet. 2.3 MW electric motor with integrated inverters,
ultra-high specific power, and fault-tolerant multisector architecture., 2025.
[175] H3X Technologies Inc., Hpdm-3000: 3.0 mw high power density machine, [Link]
tech/products/hpdm-3000, Datasheet. 3.0 MW electric motor with integrated inverters,
>10 kW/kg specific power, and eight or more independent drive sectors., 2025.
[176] “Za600 motor datasheet,” ZeroAvia, Inc., Tech. Rep. DATASHEET-ZA-0003-REV-00, 2024, 660
kW 4x3 Phase Permanent Magnet Synchronous Machine, compliant with FAA SC part 33
and EASA E19.
[177] B. Brelje, “Multidisciplinary design optimization of electric aircraft considering systems
modeling and packaging,” Available via Deep Blue at the University of Michigan Library,
Ph.D. dissertation, University of Michigan, 2021. doi: 10.7302/2703. [Online]. Available:
[Link]

[178] “Za600 inverter datasheet,” ZeroAvia, Inc., Tech. Rep. DATASHEET-ZA-0001-REV-02, 2024,
225 kW Silicon Carbide (SiC) aerospace inverter, 20 kW/kg power density, compliant with
DO160G, DO178C, and DO254.
204 BIBLIOGRAPHY

[179] “400 kw bidirectional inverter datasheet,” ZeroAvia, Inc., Tech. Rep. DATASHEET-ZA-0002-
REV-00, 2024, SiC-based dual-side-by-side aerospace inverter, 400 kW continuous (450 kW
peak), 20 kW/kg power density, designed for high-altitude and unpressurized operation.
[180] “Type-certificate data sheet no. easa.e.234 for e-811 engine (model 268mvlc),” European
Union Aviation Safety Agency (EASA), Tech. Rep. EASA.E.234, 2020, Type Certificate Holder:
Pipistrel Vertical Solutions d.o.o., Vipavska cesta 2, 5270 Ajdovščina, Slovenia.
[181] J. E. Hesselgreaves, R. Law, and D. A. Reay, Compact Heat Exchangers: Selection, Design and
Operation, 2nd. Oxford, UK: Butterworth-Heinemann, an imprint of Elsevier, 2017, isbn:
978-0-08-100305-3.
[182] M. Chiara Massaro, S. Pramotton, P. Marocco, A. H. A. Monteverde, and M. Santarelli,
“Optimal design of a hydrogen-powered fuel cell system for aircraft applications,” Energy
Conversion and Management, vol. 306, p. 118 266, 2024, issn: 0196-8904. doi: https : / /
[Link]/10.1016/[Link].2024.118266. [Online]. Available: [Link]
[Link]/science/article/pii/S0196890424002073.

[183] M. C. Massaro, S. Pramotton, P. Marocco, A. H. A. Monteverde, and M. Santarelli, “Optimal

design of a hydrogen-powered fuel cell system for aircraft applications,” Energy Conversion
and Management, 2024, Available online 18 March 2024, issn: 0196-8904. doi: 10.1016/
j . enconman . 2024 . 118157. [Online]. Available: https : / / doi . org / 10 . 1016 / j .
enconman.2024.118157.

[184] Development and optimisation of a dedicated fuel cells for aviation: Disruptive next-gen high
temperature fuel cells technology for future aviation, https : / / ec . europa . eu / info /
funding - tenders / opportunities / portal / screen / opportunities / topic -
details/horizon- jti- cleanh2- 2022- 03- 08, HORIZON-JTI-CLEANH2-2022-03-
08, Horizon Europe Programme, Call: HORIZON-JTI-CLEANH2-2022-1, Type of Action:
HORIZON-JU-RIA, 2022.
[185] “Superstack flex fuel cell system,” ZeroAvia, Inc., Tech. Rep., 2024, Modular hydrogen fuel
cell system for electric aviation applications. Continuous power output: 150 kW at sea level,
115 kW at 14,000 ft, 85 kW at 25,000 ft. Total system weight: 170 kg. Specific power: 1 kW/kg.
Service life: >2,000 h.
[186] “Superstack fuel cell system,” ZeroAvia, Inc., Tech. Rep. DATASHEET-ZA-0009-REV-00,
2024, Max continuous power: 400 kWe at 1100–600 VDC, operating current up to 600 A.
Gravimetric power density: 2.1 kW/kg, volumetric power density: 0.87 kW/L. Weight: 190.6
kg, dimensions: 950 × 800 × 600 mm. DO160-G compliant for vibration, temperature, and
environmental conditions.
[187] Ballard Power Systems, Ballard introduces new scalable FCmove®-XD fuel cell engine with
class-leading power density at iaa transportation 2024, [Link] Press
release announcing the FCmove®-XD scalable PEM fuel cell engine (120–360 kW) with
class-leading gravimetric (0.48 kW/kg) and volumetric (0.36 kW/L) power density, 30,000+
hour lifetime, and modular integration for heavy-duty trucks., Aug. 2024.
BIBLIOGRAPHY 205

[188] S. Boulmrharj, M. Khaidar, M. Siniti, M. Bakhouya, and K. Zine-dine, “Towards perfor-

mance assessment of fuel cell integration into buildings,” Energy Reports, vol. 6, pp. 288–
293, 2020, The 6th International Conference on Energy and Environment Research - En-
ergy and environment: challenges towards circular economy, issn: 2352-4847. doi: https:
/ / doi . org / 10 . 1016 / j . egyr . 2019 . 08 . 058. [Online]. Available: https : / / www .
[Link]/science/article/pii/S2352484719305803.

[189] A. Link, M. Staggat, and S. de Graaf, “Characterisation of thermal management system

specific heat rejection for electric propulsion architectures,” in Aerospace Europe Conference
2023 – 10th EUCASS – 9th CEAS, Lausanne, Switzerland: EUCASS Association, 2023. doi:
10.13009/EUCASS2023-262.

[190] G. K. Srinivasan and O. Bertram, “Preliminary design and system considerations for an
active hybrid laminar flow control system,” Aerospace, vol. 6, no. 10, p. 109, 2019. doi:
10 . 3390 / aerospace6100109. [Online]. Available: https : / / doi . org / 10 . 3390 /
aerospace6100109.

[191] E. J. Adler and J. R. R. A. Martins, “Liquid hydrogen tank boil-off model for design and
optimization,” Journal of Thermophysics and Heat Transfer, Nov. 2024, issn: 1533-6808.
[192] Airbus, Airbus takes superconductivity research for hydrogen-powered aircraft a step further,
[Link] releases/2024- 05- airbus-
takes - superconductivity - research - for - hydrogen - powered - aircraft - a -
step-further, Press release, Airbus UpNext, May 2024.

[193] F. Nicolosi, A. De Marco, L. Attanasio, and P. D. Vecchia, “Development of a java-based

framework for aircraft preliminary design and optimization,” Journal of Aerospace Information
Systems, vol. 13, no. 6, pp. 234–242, 2016. doi: 10.2514/1.I010404.
[194] E. F. Obert and R. Slingerland, Aerodynamic design of transport aircraft, eng. Amsterdam: IOS
Press, 2009, isbn: 9781586039707.
[195] G. P. Chiozzotto, “Conceptual design method for the wing weight estimation of strut-braced
wing aircraft,” in 5th CEAS Air and Space Conference, Conference presentation; full text not
available from repository, Delft, The Netherlands, Sep. 2015. [Online]. Available: https:
//[Link]/98248/.

[196] Next Generation Turboprop Gearboxes, vol. Volume 2: Aircraft Engine; Marine; Microturbines
and Small Turbomachinery, Turbo Expo, Apr. 1982, V002T02A013. doi: 10 . 1115 / 82 -
GT- 236. eprint: [Link]
pdf / GT1982 / 79573 / V002T02A013 / 2393964 / v002t02a013 - 82 - gt - 236 . pdf.
[Online]. Available: [Link]
[197] T. Oom Ortiz de Montellano, A. Heidebrecht, and M. Hoogreef, “Structural analysis of a novel
integral tank concept for hydrogen storage onboard commercial aircraft,” in Proceedings of the
AIAA SciTech 2025 Forum, AIAA Science, Technology Forum, and Exposition, AIAA SciTech
Forum 2025: American Institute of Aeronautics and Astronautics Inc. (AIAA), 2025. doi:
206 BIBLIOGRAPHY

10.2514/6.2025- 1244. [Online]. Available: [Link]

1244.

[198] J. R. R. A. Martins and A. Ning, Engineering Design Optimization. Cambridge, UK: Cambridge
University Press, Jan. 2022, isbn: 9781108833417. doi: 10.1017/9781108980647. [Online].
Available: [Link]
[199] P. T. Boggs and J. W. Tolle, “Sequential quadratic programming,” Acta Numerica, vol. 4,
pp. 1–51, 1995. doi: 10.1017/S0962492900002518.
[200] A. Wächter and L. T. Biegler, “On the implementation of an interior-point filter line-search
algorithm for large-scale nonlinear programming,” Mathematical Programming, vol. 106,
no. 1, pp. 25–57, Mar. 2006, issn: 1436-4646. doi: 10.1007/s10107-004-0559-y. [Online].
Available: [Link]
[201] K. Svanberg, “The method of moving asymptotes—a new method for structural optimiza-
tion,” International Journal for Numerical Methods in Engineering, vol. 24, no. 2, pp. 359–
373, 1987. doi: https : / / doi . org / 10 . 1002 / nme . 1620240207. eprint: https : / /
[Link]/doi/pdf/10.1002/nme.1620240207. [Online]. Avail-
able: [Link]
[202] M. J. D. Powell, “A direct search optimization method that models the objective and constraint
functions by linear interpolation,” in Advances in Optimization and Numerical Analysis, S.
Gomez and J.-P. Hennart, Eds., Dordrecht: Kluwer Academic, 1994, pp. 51–67.
[203] J. Bradbury, R. Frostig, P. Hawkins, et al., JAX: Composable transformations of Python+NumPy
programs, version 0.3.13, 2018. [Online]. Available: [Link]
[204] P. Kidger, “On Neural Differential Equations,” Ph.D. dissertation, University of Oxford, 2021.
[205] J. Rader, T. Lyons, and P. Kidger, “Optimistix: Modular optimisation in jax and equinox,”
arXiv:2402.09983, 2024.
[206] J. Sobieszczanski-Sobieski, “On the sensitivity of complex, internally coupled systems,” NASA
Langley Research Center, Hampton, VA, United States, Technical Memorandum NASA-TM-
100537, 1988, Also published as AIAA Paper 88-2378 and NAS 1.15:100537. [Online]. Available:
[Link]

[207] J. T. Hwang and J. R. R. A. Martins, “A computational architecture for coupling heterogeneous

numerical models and computing coupled derivatives,” ACM Transactions on Mathematical
Software, vol. 44, no. 4, Article 37, Jun. 2018. doi: 10.1145/3182393.
[208] J. S. Gray, K. T. Moore, T. A. Hearn, and B. A. Naylor, “A standard platform for testing
and comparison of mdao architectures,” NASA Glenn Research Center, Cleveland, OH,
United States, Technical Memorandum NASA/TM-2012-217652, Aug. 2012, Also presented
as AIAA Paper 2012-1586 at the 53rd AIAA Structural Dynamics Conference, Honolulu,
HI, April 23–26, 2012. [Online]. Available: [Link]
20120013824/downloads/[Link].
BIBLIOGRAPHY 207

[209] F. Torrigiani and P. D. Ciampa, “Mdo architectures comparison on analytical test case and
aerostructural aircraft system design problem,” in Proceedings of the 6th CEAS Aerospace
Europe Conference, CEAS 2017 paper no. 851, Council of European Aerospace Societies
(CEAS), Bucharest, Romania: CEAS, 2017.
[210] J. Sobieszczanski-Sobieski, “Multidisciplinary optimization for engineering systems: Achieve-
ments and potential,” NASA Langley Research Center, Hampton, VA, USA, NASA Technical
Memorandum NASA-TM-101566, Mar. 1989, Presented at the DFVLR Seminar on Optimiza-
tion, Bonn, Germany, June 7–8, 1989. [Online]. Available: [Link]
citations/19890015137.

[211] ATR, Atr 72-600: The most fuel efficient regional aircraft – first choice for operators, Brochure,
ATR, Blagnac Cedex, France, 2025.
[212] European Union Aviation Safety Agency, “Type-certificate data sheet no. im.e.041 for pw100
series engines,” European Union Aviation Safety Agency, Technical Report IM.E.041, Dec.
2023, Issue 07; [Link].00052-001.
[213] Aircraft Commerce, “Atr 42 and 72 maintenance analysis and budget,” Aircraft Operator’s
and Owner’s Guide, no. 49, Dec. 2006, December 2006 / January 2007. [Online]. Available:
[Link]