2019 06 25 - DissSchmDruckv3

Analysis of a Hybrid Propulsion
Lunar Sample Return Mission
A thesis accepted by the Faculty of Aerospace Engineering and Geodesy of

the University of Stuttgart in partial fulfillment of the requirements for the
degree of Doctor of Engineering Sciences (Dr.-Ing.)
by
Dipl.-Ing. Christian Schmierer
born in Heilbronn
Committee Chair: Prof. Dr.-Ing. Stefan Schlechtriem

Committee Members: Prof. Dr.-Ing. Oskar Haidn
Prof. Dr. ir. Johan Steelant
Date of Defence: 11.03.2019
Insitute of Space Systems

University of Stuttgart
2019
Acknowledgment
The foundation for the research and work for this dissertation has been established at the DLR
Institute of Space Propulsion in Lampoldshausen, the ESA ESTEC facilities and the Institute
of Space Systems at the University of Stuttgart. I thank Prof. Dr. Stefan Schlechtriem for
his professional, helpful and kind supervision and his great interest to advance the hybrid
propulsion technology. The evaluation of this work by Prof. Dr. Oskar Haidn is highly
appreciated. The work was vastly supported by colleagues of the department of rocket
propulsion at the DLR and therefore I thank Prof. Dr. Michael Oschwald for his guidance
and the system analysis team for their constant support. A great amount of work was also
supported by the department of propellants of Dr. Helmut Ciezki and especially by the M11
test bench team and the hybrid propulsion team under the guidance of Dr. Mario Kobald.
The help of many students is also highly appreciated, who did studies with the software or
prepared and conducted the tests at M11. A special thanks to Prof. ir. Johan Steelant and his
team who gave me the opportunity to work at the ESA ESTEC facilities and supported me
with their knowledge and experience in a great manner. Astos Solutions GmbH provided the
software ASTOS for this work and the support by the team was highly valuable.
II
Contents
Table of Contents III
Abstract VI
Kurzzusammenfassung VIII
List of Figures X
List of Tables XII
Nomenclature XIV
1. Introduction 1
2. Sample Return Missions to the Moon 4

2.1. Orbit Mechanics of Space Flight to the Moon . . . . . . . . . . . . . . . . . . . . 4
2.1.1. The Moon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.2. Transfer Orbits to the Moon . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.3. Moon Orbit Injection and Landing . . . . . . . . . . . . . . . . . . . . . . 6
2.1.4. Return Trajectory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2. Historic Missions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.1. Scientific Gains of Sample Return . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.2. Propulsion System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.3. Mission Phases and Trajectory . . . . . . . . . . . . . . . . . . . . . . . . . 12
3. Theory of Hybrid Rocket Engines 15

3.1. Hybrid Rocket Engine Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2. Hybrid Rocket Liquid Oxidizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3. Hybrid Rocket Fuels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.4. Hybrid Rocket Engine Components . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.5. Advantages and Disadvantages of Hybrid Rocket Engines . . . . . . . . . . . . . 22
3.5.1. Safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.5.2. Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.5.3. Cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.6. Hybrid Rocket Propulsion Research . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.7. Internal Ballistics of Hybrid Rocket Engines . . . . . . . . . . . . . . . . . . . . . 24
3.7.1. Modeling Laws for Classical Hybrid Fuels . . . . . . . . . . . . . . . . . . 25
III
3.7.2. Regression Law for Liquefying Fuels . . . . . . . . . . . . . . . . . . . . . 27
3.7.3. Correction of the Classical Regression Rate Law . . . . . . . . . . . . . . 31
4. Software 33
4.1. Analysis, Simulation and Trajectory Optimization Software for Space Applica-
tions (ASTOS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.1.1. Moon Sample Return Analysis with ASTOS . . . . . . . . . . . . . . . . . 33
4.1.2. User Propulsion Component for Hybrid Rocket Engines . . . . . . . . . . 34
4.1.3. Analytical Hybrid Propulsion Component . . . . . . . . . . . . . . . . . . 34
4.2. EcosimPro & ESPSS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.2.1. Implementation of Hybrid Rocket Engines . . . . . . . . . . . . . . . . . . 36
4.2.2. Improved Modeling of Classical Fuels in ESPSS . . . . . . . . . . . . . . . 37
4.2.3. Enthalpy Balance at the Solid Fuel’s Surface . . . . . . . . . . . . . . . . . 39
4.2.4. Modeling of Liquefying Fuels in ESPSS . . . . . . . . . . . . . . . . . . . . 39
4.2.5. Regression Rate Model Comparison . . . . . . . . . . . . . . . . . . . . . 40
4.2.6. Problems of the Regression Rate Models . . . . . . . . . . . . . . . . . . . 46
4.3. ASTOS & ESPSS Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.4. Software Verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.4.1. Simulation of Moon Transfer and Landings . . . . . . . . . . . . . . . . . 50
4.4.2. Combined Optimization of Trajectory and Propulsion . . . . . . . . . . . 57
4.4.3. Simulation of a Hybrid Rocket Test . . . . . . . . . . . . . . . . . . . . . . 62
5. Experiments with a Hybrid Rocket Engine Demonstrator 72

5.1. Test Bench Set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5.1.1. Paraffin Fuel Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.1.2. Test Sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.1.3. Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.2. Test Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.2.1. Test Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.2.2. Regression Rate Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.2.3. Throttling Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6. Hybrid Propulsion Lunar Mission Analysis 89

6.1. Reference Mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.1.1. Launch with Ariane 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.1.2. Moon Orbit Injection And Landing . . . . . . . . . . . . . . . . . . . . . . 91
6.2. Lunar Lander Spacecraft Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.2.1. Mass Estimations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.2.2. Hybrid Propulsion Spacecraft Variants . . . . . . . . . . . . . . . . . . . . 98
6.2.3. Hybrid Propulsion System Performance . . . . . . . . . . . . . . . . . . . 102
6.2.4. Liquid Propulsion Spacecraft Variants . . . . . . . . . . . . . . . . . . . . 105
6.2.5. Spacecraft Variants Summary . . . . . . . . . . . . . . . . . . . . . . . . . 105
IV
6.2.6. Return Rocket . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6.2.7. Impact of Engine Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . 111
7. Conclusion 113
Appendices 122
A. Moon Missions 122
B. Hybrid Rocket Research Groups 132
C. ESPSS: Enthalpy Balance at the Fuel Grain 136
D. Lunar Lander Optimization 139
E. Test Results 142
F. Hybrid Propulsion Lunar Lander 145
V
Abstract
A sample return mission to the Moon brings many scientific gains to different disciplines. The
Moon and its history have come back into the view of science lately. For 20 years now the
interest in space missions to the Moon has constantly grown bringing new orbiters and even
landing missions. The advantage of returning soil samples to Earth is widely acknowledged
as the samples can be analyzed with the wide range of instruments on Earth. Advancements
in analysis instruments in the future bring benefits to the investigation of the soil samples.
In the past manned (Apollo, USA) and unmanned (Luna, USSR) sample return missions were
conducted. These missions demonstrated great technological advancements for the nations
in the 1960’s and 1970’s. Today space missions are also driven by economic and ecological
perspectives. Low cost is a major driver in the new space era. Although economic aspects
have not the highest priority for exploration missions, cost reductions are still essential.
Especially the choice of propellant is important. Environmentally friendly propellants on the
one hand support the low-cost processes and on the other hand make sure that no personnel or
environments are exposed to hazardous materials like hydrazine. Therefore new propellants
and even new propulsion systems are developed in order to reduce cost and hazards. Hybrid
propulsion technology is very promising in that matter. Therefore, a hybrid propulsion lunar
sample return mission is proposed and analyzed. First a survey of Moon missions and their
history is conducted focusing on established propulsion systems using traditional, storable
fuels and oxidizers. Then the theory of hybrid propulsion is presented. Hybrid propulsion
is dominated by the effects of having a liquid and a solid propellant. This brings many
advantages: an explosion hazard is impossible and the propulsion system layout becomes
simpler. However, at the same time the hybrid propulsion system poses challenges: The
volumetric efficiency and the thrust to weight ratio are lower as the regression rates of classical
fuels are too low and the surface of the fuel grain needs to be increased. Liquefying paraffin-
based fuels form a melt layer and increase the regression rate with droplet entrainment by
a factor of up to six. New advances with those fuels show that hybrid propulsion can gain
significant advantages from this discovery. The chamber design is simpler if liquefying fuels
are used. This strengthens the low-cost approach even further. Different regression rate
prediction models are presented after a literature research. New models for classical and
liquefying fuels are implemented in an existing simulation model in ESPSS in order to enhance
future simulation capabilities for both pre-design and test result comparison. A tool for
simulation and optimization of hybrid propulsion systems in combination with the spacecraft’s
trajectory is developed using the ASTOS software. An interface is created for coupling both
software tools, in order to simulate and optimize hybrid propulsion systems. The software
tools are verified by simulating and optimizing different trajectories. The simulation of hybrid
VI
propulsion systems in ESPSS is compared and evaluated with hybrid rocket engine tests at
the DLR test bench M11. An experimental campaign was conducted in order to get data
about paraffin-based fuels in combustion with gaseous oxygen. A special focus was put on
longer burn durations of up to 40 s. The longer burn duration creates new problems with
thermal and mechanical loads on materials, but is needed in order to make hybrid propulsion
fit for a wider range of applications. Another investigated feature of hybrid rocket engines
is throttleability resp. deep-throttling. Throttling is quite easy to achieve in hybrids, as only
one liquid propellant needs to be actively controlled. However, due to the fact that the fuel
mass flow is directly linked to the flow conditions, a shifting in mixture ratio is inevitable.
This results in performance losses. With the established tools and software six variants of
hybrid propulsion landers were designed and compared to each other, as well as to two liquid
propulsion concepts using storable and cryogenic propellants. Cryogenic propellants are a
possible choice as long as the mission duration is limited to a few days and the cryogenic
temperature is held with insulation and boil-off. The six hybrid propulsion variants are using
three different staging concepts and are analyzed with hydrogen peroxide and liquid oxygen
as the oxidizer. The results are promising. While hybrid rocket engines have a very good
ideal specific impulse, the combustion efficiency is usually lower than in liquid propulsion.
Future advancements on that matter can bring hybrid propulsion nearly on an equal level
with liquid propulsion. The dry mass of a hybrid propulsion system is still a bit higher than
of a liquid propulsion system as the storage of fuel in a cylindrical shape is less optimal
than in a spherical tank. In addition insulation material is necessary, as the chamber wall
cannot be regeneratively cooled. This further increases the structural mass. But even with
these restrictions the performance is comparable to liquid systems and the low-cost approach,
non-toxicity and lack of explosion hazard improve the hybrid propulsion’s position among
the possible concepts. A hybrid propulsion lunar sample return mission is a viable option
for low-cost spacecraft development and can become a future project of the European space
community.
VII
Kurzzusammenfassung
Eine Probenrückführungsmission zum Mond bietet viele Erkenntnisse für verschiedene wis-
senschaftliche Disziplinen. Der Mond und seine Geschichte haben sich wieder in das Blickfeld
der Wissenschaft bewegt. In den letzten 20 Jahren ist das Interesse an Raumfahrtmissionen
zum Mond stetig gewachsen, weswegen viele Satelliten und sogar Landefahrzeuge zum
Mond geschickt wurden. Die Rückführung von Bodenproben bringt einige Vorteile, da
in den Laboren auf der Erde das ganze Spektrum an Instrumenten zur Untersuchung der
Proben zur Verfügung steht. Die Weiterentwicklung von Instrumenten bringt außerdem
in der Zukunft die Möglichkeit, bereits untersuchte Proben noch genauer zu analysieren.
In der Vergangenheit wurden bemannte (Apollo, USA) und unbemannte (Luna, UdSSR)
Mondmissionen mit Probenrückführung durchgeführt. Die Missionen zeugen von den großar-
tigen technologischen Fortschritten der Nationen in den 1960er und 1970er Jahren. Heute
werden wissenschaftliche Raumfahrtmissionen auch von wirtschaftlichen und ökologischen
Faktoren beeinflusst. Obwohl bei Explorationsmissionen die Ökonomie nicht im Vordergrund
steht, sind Kostenreduktionen in den unterschiedlichen Bereichen einer Mission unerlässlich.
Besonders die Treibstoffwahl ist in mehrerer Hinsicht entscheidend. Umweltfreundliche
Treibstoffe reduzieren die Kosten bei der Handhabung und stellen andererseits sicher, dass
Personal und Umwelt keinen gefahrvollen Stoffen wie Hydrazin ausgesetzt werden. Deshalb
werden neue Treibstoffe und sogar neue Antriebssysteme entwickelt, um Kosten und Gefahren
zu reduzieren. Die Hybridantriebstechnologie ist diesbezüglich vielversprechend. Daher wird
eine Probenrückführungsmission mit Hybridraketenantrieben zum Mond vorgeschlagen und
untersucht. Hierfür wird zunächst eine Übersicht über Mondmissionen und ihre Geschichte
erstellt. Dabei wird der Fokus auf die etablierten Antriebssysteme mit traditionellen, lager-
fähigen Brennstoffen und Oxidatoren gelegt. Anschließend werden die Grundlagen der Hybri-
dantriebstechnologie präsentiert. Hybridraketenantriebe werden von ihrem Aufbau mit einer
flüssigen und festen Treibstoffkomponente geprägt. Dies bringt einerseits viele Vorteile: es gibt
keine Explosionsgefahr und der Aufbau des Antriebssystems ist einfacher. Gleichzeitig stellt
das Hybridraketenantriebsystem einige neue Herausforderungen: die volumenspezifische
Effizienz und das Verhältnis von Schub zu Gewicht ist niedrig, da die Abbrandrate klassischer
Brennstoffe gering ist und die Oberfläche des Brennstoffkerns erhöht werden muss. Wie
neue Fortschritte mit Paraffin-basierten Brennstoffen zeigen, gewinnen Hybridantriebe mit
dieser Brennstoffart erhebliche Vorteile, da diese Brennstoffe eine Flüssigkeitsschicht bilden.
Dies erhöht die Abbrandrate durch Tropfenbildung um einen Faktor von bis zu sechs. Wenn
verflüssigende Brennstoffe genutzt werden, ist die Geometrie des Brennstoffs daher einfacher.
Dies verringert die Kosten von Hybridraketenantrieben weiter. Verschiedene Modelle, um
die Abbrandrate vorherzusagen, werden nach einer Literaturrecherche vorgestellt. Neue
VIII
Modelle für klassische und verflüssigende Brennstoffe werden in ein existierendes ESPSS
Modell implementiert, um die zukünftige Simulationsmöglichkeiten sowohl für Vorentwurf
als auch für den Vergleich mit Testergebnissen zu verbessern. Ein Programm zur Simulation
und Optimierung von Hybridraketenantriebssystemen in Kombination mit der Trajektorie des
Raumfahrzeugs wird mit dem Programm ASTOS entwickelt. Eine Schnittstelle wird erstellt,
um beide Programme gekoppelt zu benutzen und hybride Antriebssysteme zu simulieren
und zu optimieren. Die Programme werden mit Simulationen und Optimierungen ver-
schiedener Trajektorien verifiziert. Die Simulation der hybriden Antriebssysteme in ESPSS
wird mit Antriebstests am DLR Prüfstand M11 verglichen und bewertet. Eine experimentelle
Testkampagne wurde durchgeführt um mehr Daten über Paraffin-basierte Brennstoffe bei
der Verbrennung mit gasförmigem Sauerstoff zu bekommen. Um Hybridraketenantriebe in
einem breiter gefächerten Bereich zur Anwendung zu bringen, ist eine längere Brenndauer der
Triebwerke notwendig. Diese längere Brenndauer erzeugt neue Probleme durch thermische
und mechanische Lasten auf Bauteile. Als weitere Eigenschaft von Hybridraketentrieb-
werken wird die Drosselbarkeit untersucht. Drosseln ist mit Hybridraketentriebwerken
einfach möglich, da nur eine Flüssigkeit aktiv gesteuert werden muss. Dennoch ergibt
sich eine Schwierigkeit: Da der Brennstoffmassenstrom direkt mit den Strömungsbedingun-
gen verbunden ist, ist eine Veränderung des Mischungsverhältnisses unvermeidbar. Das
resultiert in Leistungsverlusten. Mit den erstellten Programmen wurden sechs Varianten
von Landefahrzeugen mit Hybridraketenantrieben entworfen und miteinander verglichen,
so wie mit zwei Konzepten mit Flüssigantriebssystemen mit lagerfähigen und kryogenen
Treibstoffen. Kryogene Treibstoffe sind nutzbar, solange die Missionsdauer auf wenige Tage
begrenzt ist und die Treibstofftemperatur mit Isolation und Verdampfung gehalten wird.
Die sechs Varianten mit Hybridantrieben nutzen drei verschiedene Stufungskonzepte und
werden jeweils mit Wasserstoffperoxid und flüssigem Sauerstoff als Oxidator untersucht.
Die Ergebnisse sind vielversprechend. Während Hybridraketenantriebe einen guten idealen
spezifischen Impuls haben, ist die Verbrennungseffizienz gewöhnlich niedriger als bei Flüs-
sigtreibstoffen. Weitere Verbesserungen in diesem Bereich können Hybridraketenantriebe
auf dasselbe Leistungsniveau wie Flüssigraketentriebwerke bringen. Die Trockenmasse von
Hybridantriebssystemen ist dennoch etwas höher als bei Flüssigantriebssystemen, da die
Lagerung des Brennstoffs in der zylindrischen Brennkammer weniger optimal als in einem
Kugeltank ist. Außerdem wird für die Brennkammer auch Isolationsmaterial benötigt, da eine
regenerative Kühlung nicht möglich ist. Dies erhöht die Strukturmasse weiter. Aber selbst
mit diesen Einschränkungen ist die Leistungsfähigkeit vergleichbar zu Flüssigantrieben. Die
niedrigeren Kosten, die Ungiftigkeit der Treibstoffe und das Vermeiden von Explosionsgefahr
verbessern die Stellung von Hybridraketenantrieben unter den verschiedenen möglichen
Antriebskonzepten. Eine Probenrückführungsmission zum Mond mit Hybridraketenantrieb-
stechnologie ist eine umsetzbare Option für die Entwicklung kostengünstiger Raumfahrzeuge
und kann in der Zukunft ein Projekt der Europäischen Raumfahrtgemeinschaft werden.
IX
List of Figures
2.1. Exemplary mission profile for a Lunar sample return, [1] . . . . . . . . . . . . . 7
3.1. Hybrid rocket engine layout,[21] . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3.2. Specific vacuum impulse with PB-5% over mixture ratio . . . . . . . . . . . . . . 20
3.3. Schematic of classical fuel’s reaction process [45] . . . . . . . . . . . . . . . . . . 26
3.4. Schematic of liquefying fuel’s reaction process [49] . . . . . . . . . . . . . . . . . 28
4.1. Schematic of ESPSS model for regression rate comparison . . . . . . . . . . . . . 41

4.2. Regression rate and pressure of different models for classical fuels, part 1 . . . . 43
4.3. Regression rate and pressure of different models for classical fuels, part 2 . . . . 44
4.4. Regression rate and pressure of different models for liquefying fuels . . . . . . . 47
4.5. Apollo 11 reconstruction results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.6. Descent trajectory of Apollo 11 Lunar Module close to the Moon’s surface in
ASTOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.7. Velocities of Apollo 11 Lunar Module during the final descent in ASTOS . . . . 53
4.8. Trajectory of Luna 24 on the Proton-K launch vehicle in ASTOS . . . . . . . . . . 54
4.9. Luna 24 spacecraft data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.10. Hybrid micro launcher data, part 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.11. Hybrid micro launcher data, part 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.12. Schematic of ESPSS model of the M11.3 test bench . . . . . . . . . . . . . . . . . 65
4.13. Updated schematic of ESPSS model of the M11.3 test chamber . . . . . . . . . . 65
4.14. Test No. 3 simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.15. Comparison of Experiment and Simulation: Test No. 3 simulation results . . . . 70
4.16. Comparison of Experiment and Simulation: Test No. 4 simulation results . . . . 71
5.1. Flow plan of the M11.3 test bench . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

5.2. Technical drawing of the experimental combustion chamber . . . . . . . . . . . 74
5.3. Paraffin production [73] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.4. Process flow of the script to analyze the test data . . . . . . . . . . . . . . . . . . 78
5.5. Ideal temperature & characteristic velocity with NASA CEA . . . . . . . . . . . . 79
5.6. Test 1, 3 and 4 measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.7. Test 3 and 4 measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.8. Regression rate results and exponential fit . . . . . . . . . . . . . . . . . . . . . . 86
5.9. Regression rate data and test No. 7 measurements . . . . . . . . . . . . . . . . . 88
6.1. Moon transfer trajectory of a Ariane 5 launched spacecraft . . . . . . . . . . . . . 90
X
6.2. Hyperbolic Moon trajectory after launch with Ariane 5 . . . . . . . . . . . . . . . 91
6.3. Altitude over time of the Moon lander (total and detail) . . . . . . . . . . . . . . 92
6.4. Speed over time of the Moon lander (total and detail) . . . . . . . . . . . . . . . . 93
6.5. Attitude over independent phase variable . . . . . . . . . . . . . . . . . . . . . . 93
6.6. Oxidizer mass flow of a single lander engine . . . . . . . . . . . . . . . . . . . . . 94
6.7. ESPSS schematic for a Moon lander kick stage . . . . . . . . . . . . . . . . . . . . 97
6.8. Lunar lander spacecraft design sketch, [23] . . . . . . . . . . . . . . . . . . . . . 98
6.9. Altitude comparison between Variant 1 and 2 . . . . . . . . . . . . . . . . . . . . 102
6.10. Mass flow computation comparison between ESPSS and ASTOS model . . . . . . 104
6.11. Thrust computation comparison between ESPSS and ASTOS model . . . . . . . 104
6.12. Specific impulse computation comparison between ESPSS and ASTOS model . . 104
6.13. Mass comparison of different variants . . . . . . . . . . . . . . . . . . . . . . . . . 106
6.14. Early design of the kick stage’s propulsion system . . . . . . . . . . . . . . . . . . 107
6.15. Cross section of a H2 O2 /PB-5% engine with 13 kN thrust . . . . . . . . . . . . . 108
6.16. Return rocket trajectory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
A.1. Historical landing sites on the Moon’s surface, [78] . . . . . . . . . . . . . . . . . 122

A.2. Moon’s gravity map by GRAIL, [79] . . . . . . . . . . . . . . . . . . . . . . . . . . 122
A.3. Overview of lunar missions in the past and future . . . . . . . . . . . . . . . . . 127
A.4. Landers of Apollo and Luna Moon programs, [83, 84] . . . . . . . . . . . . . . . 130
A.5. Luna 24 lander module replica, [85] . . . . . . . . . . . . . . . . . . . . . . . . . . 131
B.1. Research on hybrid propulsion in Europe . . . . . . . . . . . . . . . . . . . . . . 132
D.1. Hybrid lunar lander data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
E.1. Test 2 and 5 measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

E.2. Test 2 and 5 measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
F.1. Thrust and pressure simulation for hybrid propulsion kick stage . . . . . . . . . 145
F.2. Mass over time for hybrid propulsion kick stage . . . . . . . . . . . . . . . . . . . 145
F.3. Mass flow, IV ac and c∗ for hybrid propulsion kick stage . . . . . . . . . . . . . . . 146
F.4. Regression rate and fuel port diameter for hybrid propulsion kick stage . . . . . 146
XI
List of Tables
2.1. ∆V [ ms ] requirements to the Moon from [1] . . . . . . . . . . . . . . . . . . . . . . 6

2.2. Summary of landed spacecraft on the Moon . . . . . . . . . . . . . . . . . . . . . 9
2.3. Propulsion system parameters of Apollo 11 . . . . . . . . . . . . . . . . . . . . . 11
2.4. Propulsion system parameters of Luna 16 . . . . . . . . . . . . . . . . . . . . . . 12
3.1. Propellant options for hybrid rocket engines, [23] . . . . . . . . . . . . . . . . . . 17

3.2. Propellant combinations [23] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.3. Hydrogen peroxide combinations with PB-5% . . . . . . . . . . . . . . . . . . . . 19
3.4. Propellant options for hybrid rocket engines, [23] . . . . . . . . . . . . . . . . . . 30
4.1. Equations of analytical hybrid rocket engine model . . . . . . . . . . . . . . . . . 35

4.2. Additional input parameters for the hybrid rocket component . . . . . . . . . . 39
4.3. Input parameters for the classical regression rate comparison . . . . . . . . . . . 41
4.4. Input data for the ESPSS deck . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.5. Launch details for Apollo 11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.6. Launch details for Luna 24 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.7. Trajectory phases[69] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.8. Optimization constraints for a micro launch vehicle[69] . . . . . . . . . . . . . . 59
4.9. Initial and optimized micro launcher masses[69] . . . . . . . . . . . . . . . . . . 59
4.10. Initial and optimized micro launcher propulsion system[69] . . . . . . . . . . . . 60
4.11. Geometrical engine parameters[69] . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.12. Optimization correction factors[69] . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.13. Valve sequence for the simulation of test No. 3 . . . . . . . . . . . . . . . . . . . 66
4.14. Important test bench simulation parameters . . . . . . . . . . . . . . . . . . . . . 67
5.1. Sensor positions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

5.2. Combustion chamber parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.3. Combustion chamber design parameters . . . . . . . . . . . . . . . . . . . . . . . 75
5.4. Test sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.5. Test parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.1. Orbital elements after launch with the Ariane 5 ECA . . . . . . . . . . . . . . . . 91

6.2. Moon landing optimization constraints . . . . . . . . . . . . . . . . . . . . . . . . 95
6.3. Mass estimation for spacecraft design . . . . . . . . . . . . . . . . . . . . . . . . . 96
6.4. Propellant comparison in the chosen kick stage design . . . . . . . . . . . . . . . 97
6.5. Lunar lander Variant 1 mass overview . . . . . . . . . . . . . . . . . . . . . . . . 100
XII
6.8. Fuel grain data kick stage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
6.9. ∆V [ ms ] per trajectory phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
6.10. Return rocket mass overview: part 1 . . . . . . . . . . . . . . . . . . . . . . . . . 110
6.11. Return rocket mass overview: part 2 . . . . . . . . . . . . . . . . . . . . . . . . . 111
6.12. Comparison of 90 % and 95 % combustion efficiency 3-staged variant using LOX 112
A.1. Soviet lunar landing missions from 1959 to 1976 [6, 80] . . . . . . . . . . . . . . 123
A.2. US lunar landing missions from 1962 to 1972 [6, 80] . . . . . . . . . . . . . . . . 124
A.3. Missions to Moon from 1990 to 2010, data from [6] . . . . . . . . . . . . . . . . . 125
A.4. Missions to Moon from 2010, data from [6] . . . . . . . . . . . . . . . . . . . . . . 126
A.5. Planned lander missions to Moon [80] . . . . . . . . . . . . . . . . . . . . . . . . 126
A.6. Launch / orbit data of Apollo 11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
B.1. Hybrid rocket propulsion research groups . . . . . . . . . . . . . . . . . . . . . . 133

D.1. Initial and optimized lunar lander masses[69] . . . . . . . . . . . . . . . . . . . . 140

D.2. Lunar lander engine parameters[69] . . . . . . . . . . . . . . . . . . . . . . . . . 140
XIII
Nomenclature
Formulas D Flow diameter m

H2 O2 Hydrogen peroxide e0 Entrainment coefficient
J
N2 O Nitrous oxide Ea Activation energy mol
N2 O4 Dinitrogen tetroxide F Thrust N

O2 Oxygen Fr Heat transfer correction
Variables factor for surface
roughness
a Regression law parameter
kg
G Mass flux density m2 s
a Semi-major axis m
ht Melt layer thickness mm
A Flow area m2
W
h Heat transfer coefficient m2 K
Af Fuel surface area m2
J
hent Total heat of entrainment kg
B Blowing parameter
J
hmelt Total heat of melting kg
Bg Evaporation blowing
J
parameter hvap Total effective heat of kg
gasification
C1 Regression law coefficient
W
k Heat conductivity mK
CB1 , CB2 Blowing correction
coefficients Kox Oxidizer concentration
CH , CH0 Stanton number with and l Length m

J
without blowing Lm , Lv Latent heat of melting kg
cf Skin-friction coefficient and vaporization

kg
cp Specific heat J M Molar weight mol
kg K
kg
c? Characteristic velocity m ṁ Mass flow s
s
d Fuel grain diameter m N ul , N ut Nusselt number (laminar,

turbulent)
XIV
O/F Mixture ratio Ta Average gas temperature K
p Pressure Pa Tmelt Melting temperature K
Pd Dynamic pressure Pa Tvap Vaporization temperature K
Pr Prandtl number Ts Fuel grain surface K
Q̇r , Q̇c Radiative and convective J temperature
m2 s
m
heat transfer u Flow velocity s
J m
Q̇w Wall heat transfer m2 s
ue Velocity at the edge of the s
Q̇vs Heat transfer from the W boundary layer

m
vapor of solid fuel to the uc Velocity at the flame zone s
gas flow v1 Velocity before maneuver m
s
Q̇sp Heat transfer from the W v2 Velocity after maneuver m
s
vapor of solid fuel to the
x Position on x-axis m
solid fuel
x Position on fuel grain m
r Distance to Earth center m
mm Xe Entrainment parameter N0.5
ṙ Regression rate s
mm y Position on y-axis m
ṙcl Regression rate, classical s
z Position on z-axis m
Rhe , Rhv Ratio of effect. Heats of
gasification for δ Thermal thickness mm
entrainment and ∆v Velocity change m
s
vaporization J
∆h Difference of enthalpy kg
Re, Rex Reynolds number (at between the surface and
position x) flame zone
J
Ru Universal gas constant mol K ∆Hfθ Heat of formation J
mol
tb Burning duration s ∆T1 , ∆T2 Temperature differences K
T Temperature K
XV
α Inclination change rad e Edge of the boundary
β Angle in orbit plane rad layer
κl Thermal diffusivity m2 ent Entrainment

s
λl Thermal conductivity W f Final (burn end)

mK
µg Gas flow viscosity Pa s fu Fuel
µl Liquid layer viscosity Pa s g Gas
µ Gravitational parameter m3 inj Injector (cavity)

s2
φc Relative flame position in l Liquid
the boundary layer E Earth
ρ Distance to Moon center m m Mean
kg
ρl , ρ s Density of liquid and m3
M Moon
solid fuel o, ox Oxidizer
m
σ Surface tension or N out Outside
hard-sphere diameter
RV Control valve
ζ Pressure loss coefficient
s, sp Solid
Indices
t Throat
0 Initial
vap Vapor, vaporization
amb Ambience
Exponents
b Bulk
k, l, m, n Regression law exponents
c Combustion zone
Abbreviations
cc Combustion chamber
ASTOS® Analysis, Simulation and
Cor Coriolis sensor Trajectory Optimization
e Nozzle exit Software for Space
Applications
XVI
CEA Chemical Equilibrium UTC Coordinated Universal
with Applications Time
DLR German Aerospace Center
ESPSS European Space
propulsion System
Simulation
FFT Fast Fourier
Transformation
HyEnD Hybrid Engine
Development
HTPB Hydroxyl-terminated
polybutadiene
JPL Jet Propulsion Laboratory
LEO Low Earth Orbit
LOX Liquid Oxygen
M11 Test bench M11 at DLR
Lampoldshausen
MON Mixture oxides of
nitrogen
NASA National Aeronautics and
Space Administration
PB-5% Paraffin based fuel with
5% addiditves
RP Rocket propellant
STERN Student Experimental
Rockets Program of DLR
XVII
1. Introduction
Planetary exploration includes a wide range of scientific disciplines, which search together
for answers to basic questions of origin, genesis and evolution of planets and their moons.
Next to remote exploration from Earth, for more than 50 years spacecraft have been used to
explore space and our solar system’s celestial bodies from a closer distance. The scientific and
technological gains of these missions are enormous and in the future much more is to come.
One of the most important targets of space exploration is the Moon. Since the 1960’s, many
missions have been conducted in Moon orbit but also on the surface, in the form of impactors,
landers and even sample return missions. The first successful sample return mission from
the Moon was the manned US mission Apollo 11, which brought back 22 kg of sample material
from the Moon’s surface. Together with the other five manned Apollo landings a total of 326 kg
lunar soil material was brought back to Earth, but since Apollo 17 in 1972 no manned mission
has been made and only the soviet Luna missions returned samples from the Moon to the
Earth. From 1970 to 1974, the Luna program conducted three sample return missions (Luna
16, 20 & 24) and brought back a total of 325 g of regolith to Earth [1]. Yet after Luna 24,
no spacecraft touched down on the lunar surface for a long time. Only in the past 25 years,
the Moon received again increased attention in the scientific and space faring world. Several
nations and space agencies sent spacecraft to the Moon for remote exploration: Hiten (1990,
ISAS (now JAXA)), Clementine (1994, DoD/NASA), Luna Prospector (1998, NASA), SMART-1
(2004, ESA), Kaguya (2007, JAXA), Chang‘e-1 (2007, CNSA), Chandrayaan-1 (2008, ISRO),
Lunar Reconnaissance Orbiter, LCROSS (2009, NASA), Chang‘e-2 (2010, CNSA), GRAIL (2012,
NASA), LADEE (2014, NASA), Chang‘e 5-T1 (2014, CNSA) and 4M (2014, OHB). Some of these
missions were also impactors but the only soft landing so far in the new century was Chang‘e 3
of China in 2013 with the rover Yutu. Analyses of the Moon regolith samples back at Earth are
more exact and detailed than in situ analyses. Therefore, sample return missions are of high
interest for science. The mission to the Moon is described in Chapter 2.
Hybrid rocket engines are using solid fuel and liquid oxidizer in their combustion chamber.
This unique combination of solid and liquid state of matter has some benefits: The solid
propellant is incapable of causing explosions and the range of suitable solid propellants is
quite large. Hybrid rocket engines don’t need chamber wall cooling and have only one liquid
propellant which needs a feed system with pressurization. Therefore hybrid rocket engines
can be developed at very low cost. Recent advancement in liquefying fuels using paraffin
wax increase the regression rate. With this the thrust per engine volume and weight is
improved. This enables hybrid rocket propulsion to be applied in spacecraft and launcher
applications more easily. Therefore hybrid propulsion shall be evaluated in the application
1
of a lunar sample return spacecraft. In order to analyze a complete sample return mission to
the Moon including hybrid rocket propulsion, both the trajectory and the propulsion system
must be simulated and optimized. The trajectory optimization must include different phases
for impulsive maneuvers and coasting arcs. The hybrid propulsion must be analyzed and
optimized regarding diverse set-ups, starting from thrust and pressure level until the choice
of propellant combinations. For trajectory simulation and optimization, "Analysis, Simulation
and Trajectory Optimization Software for Space Applications" (ASTOS® ) by Astos Solutions
GmbH is an adequate choice. This software is highly modular, which allows modeling any
kind of spacecraft trajectory from launchers over orbit maneuvers to landings on planets. In its
current state, ASTOS allows different ways of implementing rocket engines. The performance
of rocket engines is either modeled as constant, time depending or following a burning law
(e.g. for solid rocket boosters). Yet, it cannot simulate the transient behavior of a rocket engine
during start up or throttling in a detailed way. EcosimPro with its European Space Propulsion
System Simulation library (ESPSS), which is developed by Empresarios Agrupados at behalf of
the European Space Agency, is a system simulation tool with high flexibility. The ESPSS library
allows designing and simulating many different rocket propulsion systems, like pressurized
systems, turbo-pump fed cycles with a gas generator or an expander up to air-breathing hybrid
rocket engines. In order to have the toolset for analyzing and optimizing a complete space
mission, it is of high interest, to allow trajectory optimization in a loop with the simulation and
optimization of a propulsion system. For this, an interface between the trajectory optimization
software ASTOS and the propulsion cycle simulation tool ESPSS is created. A special focus
hereby is put on hybrid propulsion. Firstly, hybrid propulsion offers new opportunities in
the near future, while secondly understanding of processes in hybrid rockets advances at
high speed. The vast amount of possible propellant combinations is very promising for the
future design of high performance rocket engines, which are environmental friendly too.
Hybrid rocket engines inherit a very dynamic performance from the basic architecture of
their working principle: During the burning time, solid fuel burning inside the combustion
chamber regresses and changes thereby the inner geometry of the engine. Hence, it is essential
to have the toolset to simulate and optimize these dynamic engine processes with regard to the
trajectory, in order to find not only the best design for the engine itself, but also for the entirety
of vehicle and mission design. The ESPSS model of the regression rate of hybrid rocket engine
fuel is improved and expanded. Existing regression rate laws are evaluated. New advanced
regression rate laws for both classical and liquefying hybrid rocket fuels, which are considered
state of the art in literature, are implemented in ESPSS. This further improves the possibilities
to use the tool for predesign and design of hybrid rocket engines and propulsion systems.
Chapter 3 elaborates on hybrid rocket engines while Chapter 4 explains the software tools
created for this work and presents the verification of the software tools.
A experimental test campaign at the DLR Lampoldshausen has been conducted to support this
thesis with experimental data. A new hybrid rocket engine demonstrator has been designed,
constructed, built and tested at the test bench M11. The engine was specifically designed to
offer a longer burn duration than previously tested engines, where the maximum burn duration
2
was about 20 seconds. The newly designed engine burns up to one minute or longer, depending
on the throttle setting. First tests were done to analyze the regression rate behavior of the new
chamber design, as well as testing the heat loads on the combustion chamber parts. It was
also tested if throttling of the engine changes the combustion roughness of this experimental
set-up. A comparison of simulation results with experimental results was done as well, to
see if the simulation models are adequately capable of modeling test benches and propulsion
systems. To compare the simulation with test results, first tests of the test campaign for this
work have been used. Other publications include comparisons of other engine test results, also
focusing on the tank simulation of ESPSS [2]. The experiment set-up and results are depicted
in Chapter 5.
A mission to the Moon’s surface is defined and evaluated. The baseline is a launch with the
Ariane 5 launch vehicle. However, the created models are scalable to any launch vehicle in the
same range of payload mass. The launch vehicle trajectory is optimized to bring the largest
possible spacecraft in a hyperbolic Moon flyby trajectory. The mission objective is a soft lunar
landing with enough payload capacity to include a small rover, instrumentation and a sample
return rocket. A typical descent trajectory is defined, including a hovering phase to find a
suitable landing spot. The return rocket is then supposed to be loaded with lunar soil samples
that will be brought back to the Earth’s surface after ascending from the Moon and returning in
a reentry capsule to the Earth. Six variants of the spacecraft set-up are compared using hybrid
propulsion, as well as one spacecraft using only storable liquid propulsion and one spacecraft
using cryogenic propellants for the first part of the mission. The short flight duration to the
Moon allows the use of cryogenic propellants, as it is possible to cool them with boil-off and
the losses can be compensated by the higher specific impulse. The hybrid propulsion variants
are using either hydrogen peroxide or cryogenic liquid oxygen as the oxidizer. Propellant for
the use in reaction control thrusters is considered too. The returning spacecraft (referred to
as "return rocket" in the following) is always modeled using storable propellants. For the
spacecraft variants using hybrid propulsion, three staging concepts are evaluated: A three
staged lander concept, a two staged lander and a two staged lander with jettisoning of emptied
hybrid combustion chambers. The total sample return mass to Earth is taken as a measurement
to compare the concepts with each other. However, the analysis is focused mainly on the
performance of the spacecraft. While it is expected that liquid propulsion systems with similar
or better specific impulse than hybrid propulsion systems have a higher performance, it is also
necessary to take into account, that hybrid rocket propulsion systems are safer, cheaper to
develop and use propellants that are not hazardous. The sample return mission analysis is
presented in Chapter 6.
3
2. Sample Return Missions to the Moon
2.1. Orbit Mechanics of Space Flight to the Moon
With the use of chemical propulsion in impulsive maneuvers (i.e. short burning time, high
thrust) the orbits of a spacecraft with destination Moon are altered quickly in a short time. In
contrast, transfers to the Moon with electric propulsion have very long thrusting phases with
very low accelerations. This generally increases the ∆V requirements, which is compensated by
the extremely high specific impulse of electric propulsion. Transfers with chemical propulsion
usually take 2.5 to 5 days whereas a transfer with electric propulsion takes up weeks or months.
Up to date the landing on the Moon’s surface is always requiring the high thrust of chemical
propulsion. Change of orbit parameters is done with a change of velocity and/or velocity
direction at a certain orbit point. The velocity change ∆v is calculated by adding the velocities
vectorially:
p
∆v = v1 + v2 + 2v1 v2 cos α cos β, (2.1)
where v1 and v2 are the velocities before and after the maneuver. Here α is the change of
inclination and β is the change of the velocity direction in the orbit plane. During a flight to
the Moon both the force introduced by the Earth’s and the Moon’s gravity have a big impact on
the spacecraft’s dynamics. Therefore the transfers from the Earth to the Moon and backwards
have to be calculated as a three-body problem. The dynamics are described by the following
differential equations where x, y, z are the spacecraft’s coordinates in a coordinate system with
the Moon’s orbit spanning the x-y plane [1]:
µE x µM µM xM
ẍ = − 3
+ 3 (xM − x) − 3 (2.2)
r ρ aM
µE y µM µM yM
ÿ = − 3
+ 3 (yM − y) − 3 (2.3)
r ρ aM
µE z µM z
z̈ = − − 3 (2.4)
r3 ρ
In these equations, µE and µM are the gravitational parameters of the Earth and the Moon. r
and ρ describe the distance of the spacecraft to the centers of the Earth and the Moon. xM
and yM give the coordinates of the Moon’s center. The semi-major axis of the Moon is given as
aM . As found by many missions to the Moon’s orbit and surface, the gravitational field of the
Moon has high local deviations, which have a crucial impact especially on low altitude orbits.
Also perturbations by the Sun’s gravity and solar radiation pressure can be considered for more
detailed analyses. Figure A.2 shows the gravity of the Moon mapped by the GRAIL mission.
4
From the GRAIL data a detailed gravitational field model with spherical harmonics could be
created by Konopliv et al [3]. This model can be used for more detailed analysis of Moon
missions’ trajectories, but for this work a simplified homogeneous gravity model is used.
2.1.1. The Moon
The Earth and the Moon form a double planet system and both orbit around the barycenter,
which lies some 4700 km from the Earth’s center. The rotation of the system has a length of
27.3 days. The Moon’s semi-major axis is 384 748 km, it has an eccentricity of 0.055 and an
inclination of 5.1° in reference to the ecliptic. The Moon’s orbital plane has a precession with a
periodic duration of 18.6 years, which results in an inclination in the Earth’s equatorial system
of 18.3° to 28.3°. A co-planar transfer without additional plane change maneuvers is always
possible from a launch place below 18.3° latitude. Of course, if a transfer is done during a
time when the inclination is near 18.3°, the ∆V from the Earth to the Moon is a bit smaller.
The Moon has a diameter of 1737.1 km (mean) and a gravitational parameter only 1/80 of the
Earth’s (µE = 3.986e14 m3/s2 ; µM = 4.904e12 m3/s2 ). The gravity acceleration on the surface of the
Moon is 1.625 m/s2 [4].
2.1.2. Transfer Orbits to the Moon
Often a Low Earth Orbit (LEO) is chosen as a starting point for a Moon transfer, since this
allows for easy corrections of mistakes of the spacecraft’s orbit insertion with a launcher. Under
certain circumstances a direct transfer can be favorable, e.g. if the launcher’s configuration is
more suited for this or if re-ignition of the upper stage is not possible. The orbit insertion
into the transfer orbit can result in a hyperbolic, parabolic or elliptic transfer orbit (with
reference to the Earth). Hyperbolic orbits result in a shorter transfer time (2.5 days), but in
higher ∆V requirements at both orbit insertion and Moon orbit injection. A special hyperbolic
transfer orbit is the free return trajectory, which guarantees a return of the spacecraft to the
Earth after orbiting around the Moon without additional impulsive maneuvers (except of
possible necessary course corrections). Free return trajectories are a safe choice for manned
mission. Elliptical transfer orbits (Hohmann-Transfer) have the lowest ∆V budget but the
longest transfer times (with chemical propulsion maneuvers) of about 5 days. The best timing
for a transfer to the Moon is when the ascending node of the spacecraft and the Moon have
an angle of 0° or 180°. If a spacecraft is launched directly without parking orbit in LEO, the
best launch window is while the Moon is passing the equatorial plane (i.e. the ascending or
descending node in the equatorial reference system). This happens every 13.7 days [5].
5
2.1.3. Moon Orbit Injection and Landing
In order to bring a spacecraft into a stable orbit around the Moon, an injection burn has
to be done after the orbit transfer is finished. Optimally this injection burn is done at the
wanted periselen at the Moon. Both elliptical and circular orbits can be chosen depending on
the planned mission. By entering a polar orbit around the Moon and adjusting the periodic
duration of the orbit (by optimizing eccentricity for a given periselen) almost every landing
spot on the Moon can be reached within a few days. Theoretically, the periselen can be chosen
between 10 and 20 km, because there is no atmospheric drag and this allows for low ∆V
landings. Yet, the Moon’s heterogeneous gravity field can cause high disturbances of the orbit
at such low altitudes, therefore usually an orbit at 100 km to 200 km altitude will be chosen,
with only small losses in ∆V for the landing. When initiating a landing the periselen is chosen
to be above the landing site at low altitude. The lower the periselen, the lower is the total ∆V .
However, at the same time the sensitiveness to course deviations is higher. The periselen can
even be below the surface, resulting in a direct descent from the elliptic/circular orbit. Table
2.1 shows an overview of approximated ∆V requirements from low Earth orbit to the Moon
and back.
2.1.4. Return Trajectory
For the return trajectory, the spacecraft can be launched from the Moon’s surface either in an
intermediate orbit or directly into a hyperbolic escape orbit, which has a low perigee (in or
below the atmosphere of the Earth). Usually a flat reentry angle at the Earth’s atmosphere is
better, since it has lower maximum heat and mechanical loads. To reduce gravitational loss
the launch trajectory on the Moon should be as flat as possible, turning the vehicle from a
vertical into a horizontal orientation very quickly. The escape trajectory from the Moon is
orientated against the Moon’s velocity vector. With an optimal transfer orbit, the spacecraft
needs again about five days to the Earth and reenters the atmosphere with about 11 km/s. Figure
2.1 illustrates the sequence of a possible sample return mission.
Table 2.1.: ∆V [ ms ] requirements to the Moon from [1]

Maneuver Intermediate orbit Direct transfers
Injection into a transfer orbit to the Moon 3200 3200

Course correction 100 160
Injection into a Moon orbit 1000 0
Moon landing 2100 3000
Launch into a Moon orbit 2000 0
Injection into a transfer orbit to the Earth 1000 2900
Total 9400 9260
6
Figure 2.1.: Exemplary mission profile for a Lunar sample return, [1]
2.2. Historic Missions
The first lunar crash landing with an unmanned probe was conducted in 1959 by the Soviet
Union with Luna 2, after Luna 1 had failed earlier in the same year. In the following years,
several robotic landing missions to the Moon were initiated by the United States and the Soviet
Union, resulting finally in the landing of the first men on the Moon on 20th July 1969. An
overview of all Soviet lunar landings from Luna-2 in 1959 to Luna 24 in 1976 is given in Table
A.1. The Soviet Union launched many more probes to the Moon, but only named them Luna
with a number after a successful launch into Earth’s orbit, in order to hide the failed launch
attempts with the Semyorka launcher until 1966 and the Proton-K for later launches. Therefore
the table shows only successfully launched missions. There are ten missions on Semyorka and
four missions on Proton unaccounted for. A total of eight successful robotic soft landings were
conducted with the Luna missions. According to the ASA online archive [6], all spacecraft since
Luna 15 used resembling propulsion systems based on a very similar spacecraft design. Earlier
spacecraft based on the Ye-6 series used amine-based fuel and nitric acid as oxidizer with the
engine KTDU-5A. It had a turbopump-fed engine and surface tension propellant management
7
devices in the tank [7]. Since Luna 15 the lander series Ye-8-5 (also E-8-5) was used for the
sample return missions. The engine setup of Ye-8-5 is described in section 2.2.2. The lander
Ye-8 for the Lunokhod rovers was similar, using the same pump-fed engine KTDU-417.
Table A.2 lists all lunar landing missions of the US until Apollo 17. As an example,
Apollo 11 is explained in more detail regarding the trajectory and propulsion system in
section 2.2.2. The first Moon missions of the Ranger program had only a hydrazine (N2 H4 )
monopropellant system and a solid retro rocket. The Surveyor landers had a bipropellant
engine with monomethyl hydrazine monohydrate (MMH with water) and MON-10 (90 %
N2 O4 and 10 % nitric oxide), which ignites hypergolically. The engine was pressure-fed with
helium pressurization and bladders in the tank for zero-g conditions. For the greater part
of the ∆V requirements they also used solid retro rockets [8]. Some of the most important
missions to Moon in the recent years are listed in Tables A.3 and A.4. In 1990 Hiten was the
first Japanese lunar probe. It was also the first lunar probe since Luna 24, as well as the first
lunar probe launched by a different nation than USA and USSR. The probe conducted several
gravity assist maneuvers and aero brake maneuvers in Earth and Moon orbit. Its scientific
goal was detecting and measuring micro-meteorite particles. After an orbital maneuver failed,
NASA JPL helped IASA to conduct the first low energy transfer to Moon Orbit, by using very
low ∆V to enter the orbit. This was done by a transfer through Moon-Earth-System Lagrange
points. A controlled impact on the Moon’s surface was done on 10th April 1993. Many other
orbiters and impactors followed, as listed in the table. Before 2000, three more orbiter missions
to the Moon were done. All of them used hydrazine as a monopropellant for their attitude and
orbit control. Clementine also had a N2 O4 & hydrazine bipropellant liquid rocket engine for
orbit maneuvers. However, it failed its primary objective of meeting with asteroid Geographos.
A new approach of lunar missions was undertaken with SMART-1 in 2004 by the European
Space Agency. Its main objective was the technology demonstration of an ion drive to bring
the spacecraft into a Moon orbit. It used a Hall-effect thruster with a thrust of 70 mN using
xenon. It had a specific impulse of 1600 s. After the launch on Ariane 5 as a secondary payload
into Geostationary Transfer Orbit (GTO), the orbit was raised over 14 months to use also
three lunar resonance maneuvers to minimize propellant use. During a resonance maneuver
the spacecraft gets so close to the Moon that the Moon’s gravity has a significant impact on
the spacecraft’s orbit around the Earth. SMART-1 did not leave the sphere of influence of
Earth during these encounters however the orbital parameters like inclination, apogee and
perigee could be drastically changed without spending propellant. When SMART-1 arrived at
Moon orbit, the ion thruster was again fired for 4.5 days to lower the orbit. The ion engine
went through 5000 hours of operation and 843 starts and stops. After its extended mission
duration SMART-1 was used as an impactor in September 2006. In the following years many
orbiter and impactors were sent to the Moon. The scientific goals were always to improve
the understanding of Moon’s environment by analyzing the surface, the magnet field, the
gravitational field and radiation. Table 2.2 counts all landings on the Moon so far while Table
A.5 sums up the planned lander missions in the coming years.
8
Table 2.2.: Summary of landed spacecraft on the Moon
Nation Crash Impact Lander Rover Sample Return
USSR 6 6 2 3
USA 9 6 3 (manned) 6 (manned)
Japan 2 0 0 0
ESA 1 0 0 0
China 1 1 1 0
India 1 0 0 0
Total 20 13 6 9
Beside to ESA and SMART-1, the Chinese Space Agency (CNSA) conducted several technology
demonstrator missions. Chang’e 1 and Chang’e 2 were orbiters to demonstrate China’s ability
to send spacecraft to the Moon. These missions tested the technology for China’s first lunar
landing mission Chang’e 3 in 2013, which brought also a rover to the Moon’s surface and was
the first lunar landing since Luna 24 in 1976. The future Chang’e 5 mission is planned to be
the next lunar sample return mission. It will be an unmanned landing spacecraft which is
expected to land on the Moon in 2019. The Chinese launcher Long March 5 will be used as
a launching vehicle which has a total lift off mass up to 810 tons (CZ-504 [9]). However, it is
not clear yet which launcher setup will be used. The spacecraft itself will be composed of an
orbiter module, a lander and a return rocket module. The return rocket module is planned
to rendezvous and dock with the orbiter after ascent from the Moon’s surface. This would
be the first docking of an automated spacecraft which landed before on a celestial body. One
reason for this approach is that the technology for future manned Chinese missions shall be
demonstrated. Until now only two space programs included successful sample return from
Moon’s surface, namely the manned Apollo 11-12 & 14-17 (US) and the robotic Luna 16,
20 & 24 (USSR). Figure A.3 shows an overview over all historic and planned lunar missions
(including orbiters and flyby missions). The x-axis is the launch date and the y-axis represents
a logarithmic view of the launch mass ranging from 100 kg to the 30 t of Apollo. Figure A.1
shows the historical landing sites on the Moon’s surface.
2.2.1. Scientific Gains of Sample Return
The interest in the Moon is reasoned with a lot of possibilities to research the past of our
solar system and especially the genesis of planets. A lot of clues about Earth’s development
can be taken from analyzing the structure of the Moon’s surface. Erosion is absent on the
Moon’s surface. Hence, some surface areas of the Moon are older than three billion years. By
investigating these old soils, the history of the Moon’s and Earth’s genesis and evolution can be
reconstructed. Additionally, the Moon is considered a model for all terrestrial, rocky planets.
By researching the surface of the Moon it will be possible to use that knowledge to analyze
other planet’s surface better remotely. Beside the Moon regolith, meteorite material and cosmic
9
dust on the surface give insights in the development of the solar system. The Moon also
offers possibilities for technology demonstration of robotic missions, sample return missions,
manned missions and in-situ resource utilization (ISRU). All of those are useful later when
landing on other planets like Mars. The majority of the Moon’s surface regions like craters
and plateaus are unexplored. More data about the surface composition needs to be gathered in
order to choose the right landing spots for future manned science or ISRU missions. The age
determination of the Moon’s surface is done remotely by investigating the density of craters.
For this, more surface samples are necessary to calibrate the methods, which will increase the
precision of remote observations. Similarly this also improves the precision for observations of
other celestial bodies, when the Moon’s surface is taken as reference. An interesting landing
area for a sample return mission is the South Pole–Aitken basin, which is an enormous impact
crater on the far side of the Moon with a diameter of 2500 km. It is expected from results
of impact simulations and cartography results of Clementine and Lunar Prospector that in
the basin mantle rock is on the surface. Additionally, due to the high basin wall mountain
range, there are areas where the sun never shines. Water ice could be located in these areas.
The main advantage of sample return is that not only many more instruments can be used
in laboratories on the Earth than on a lander which is limited in mass and energy, but also
that the experience of researchers helps a lot to improve the gained results. The instruments
that are used on a spacecraft are also less diverse and advanced than stationary instruments
on the Earth. Additionally with samples on the Earth, they can be analyzed in the future,
when improved instruments are available. Sample return missions are also very interesting for
different celestial bodies like Mars or Jupiter’s moon Europa being two of the most interesting
targets in the search for evidence of extraterrestrial life. A sample return mission to the Moon
is also a technology demonstration for a future mission to Mars or Europa in search for life.
2.2.2. Propulsion System
Apollo Program
The Apollo program was launched with the strongest launching system ever developed, even
up to date. From 1969 to 1972, six Apollo Lander Modules landed on the Moon and returned
successfully to the Earth. Figure A.4 shows an illustration of the lunar lander. The dedicated
launcher Saturn V was able to launch 47 t into a Moon transfer orbit. The spacecraft consisted
of a command/service module (CSM) and the Lunar Module. The CSM offered space for three
astronauts. The launch vehicle Saturn V itself was built of the first stage S-IC, the second
stage S-II and the third stage S-IVB. Very detailed information about all Apollo missions is
found in the book “Apollo by the numbers” by Richard W. Orloff [10]. A short overview of
the data for Apollo 11 spacecraft is given in Table 2.3. The sample return was realized with
a docking of the LM with the CSM after launch from the Moon, which was a necessity due to
the manned nature of the mission. Due to the high requirements for all systems, since Apollo
was a manned program, it is not possible to compare the Apollo missions directly with small
10
robotic missions for sample return. Yet, the Apollo missions give a great example for what is
possible within the Earth-Moon system. Some parameters like the engine data are very helpful
to design future landing missions on the Moon’s surface. The AJ10-137 rocket engine of the
CSM used the hypergolic propellant combination of N2 O4 and Aerozine A-50. The engine
AJ10-137 was operated at relatively low pressure of 7 to 9 bar with a pressure-fed cycle [11].
Aerozine A-50 is a mixture of 50 % UDMH and hydrazine, which is chemically more stable than
pure hydrazine and has a higher density and boiling point than pure UDMH [12]. It is highly
toxic and carcinogenic. In addition, the oxidizer N2 O4 is toxic and environmentally harmful,
too. This is why currently a lot of effort is put in finding substitute substances for N2 O4 and
hydrazine based fuels in space propulsion. In Europe this is accelerated by REACH [13]. The
Lunar Module Descent Engine (LMDE) or Descent Propulsion System (DPS) was a variable
throttle hypergolic rocket engine, based on the same propellants as the AJ10-137. It could be
throttled down to 10 % of the full thrust [14], which is of high advantage for a soft landing
and high maneuverability near ground. The high throttle ability was realized with a pintle
injector, whose flow area could be adjusted. The pressure-fed engine had a pressure of about
7.1 bar and a thrust of 4.5 to 43.8 kN. The Lunar Module Ascent Engine (LMAE) or Ascent
Propulsion System (APS) was a smaller engine, similar to LMDE using the same propellants
and being pressure fed.
Luna Program
The Soviet robotic Moon probes of the Luna program were launched with the three-stage
launch vehicle Proton-K with a total payload into Moon transfer orbit of 5.7 t. This launcher
set-up used the Blok-D as upper stage. Luna 16 was launched in 1970, Luna 20 in 1972 and
Luna 24 in 1976, being the last sample return vehicle from the Moon ever since. Table 2.4
Table 2.3.: Propulsion system parameters of Apollo 11

Parameter S-IVB CSM LM-Landing LM-Ascent
Mass (total) [kg] 119119 30329 10281 4495

Dry mass [kg] 30329 11900 2033 2179
Engine J-2 AJ10-137 LMDE1 LMAE2
Engine cycle Gas generator pressure-fed pressure-fed pressure-fed
Propellants LOX/LH2 N2 O4 /A-50 N2 O4 /A-50 N2 O4 /A-50
Engine mass [kg] 1579 295 179 82
Vacuum thrust [kN] 1023 91.2 4.5-43.8 15.6
Ivac [s] 421 312 311 311
Burn duration [s] 475 585 n/a n/a
Chamber pressure [bar] 49.9 6.7 7.1 8.3
Expansion ratio 27.5 62.5 46 45.6
1 Lunar Module Descent Engine
2 Lunar Module Ascent Engine
11
summarizes the propulsion system data of Luna 16. The KTDU main engine on the Ye-8-5
series’ spacecraft was designed with two autonomous blocks. The main block was used with
high-pressure combustion for orbit changes and bigger course corrections. The engine cycle
was a turbo-pump fed engine setup. Due to its hypergolic propellant’s combination, it could be
reignited up to 11 times in nominal mission design. The low-pressure combustion chambers
of KTDU/417B consisted of two combustion chambers with lower thrust for a soft landing.
The engine was only throttleable in small range, which sufficed because it was of low thrust
and only used for the final descent and not for bigger ∆V maneuvers. Helium was used
as pressurizing gas. The sample return vehicles of the Luna program used the toxic and
carcinogenic propellants N2 O4 and UDMH, too. Similarly, the KRD-61 engine of the ascent
and return vehicle used the same propellants and had a turbo-pump fed cycle, too. A photo of
a mockup of Luna 24 is shown in Figure A.5.
2.2.3. Mission Phases and Trajectory
Apollo Program
The orbit maneuvers and some related information of Apollo 11 are listed in Table A.6 in
the appendix. The manned Apollo mission had some particularities that are not necessarily
practical for an unmanned mission. For example the spacecraft was very heavy, which won’t
be necessary for an unmanned mission. Apollo’s third stage was ignited at least twice. The
first burn inserted the spacecraft in an orbit, while the second burn brought the spacecraft into
the trans-lunar trajectory. Arriving at the Moon the Control/Service module (CSM) conducted
the orbit insertion with two burns (orbit insertion and circularization). Afterwards the Lunar
Module (LM) undocked and did a descent burn with the Descent Propulsion System. The
Table 2.4.: Propulsion system parameters of Luna 16

Parameter Blok-D Ye-8-5 Landing Ascent
Mass (total) [kg] 13360 5725 s. Ye-8-5 512

Dry mass [kg] 11273 1880 s. Ye-8-5 245
Engine RD-58 KTDU-417 KTDU-417B KRD-61
Engine cycle staged comb. open cycle [15] pressure.-fed open cycle[16]
Propellants LOX/RP2 N2 O4 /UDMH N2 O4 /UDMH N2 O4 /UDMH
Engine mass [kg] 300 81 s. Ye-8-5 42
Vacuum thrust [kN] 83.4 7.35-18.92 2.06-3.43 19.1
Ivac [s] 349 308-314 252-254 313
Burn duration [s] 600 650 30 30
Chamber press. [bar] 78 82 8.9 91
Expansion ratio 189 n/a n/a n/a
12
LM descended the last kilometers with a powered descent using the deep-throttling ability
of the Descent Propulsion System. The Ascent Propulsion System lifted the LM again in
Moon orbit and preparing Rendezvous and Docking by several burn maneuvers. Leaving the
Ascent module in the Moon orbit, the CSM started the trans-earth injection. One mid-course
correction was necessary before the CM and SM could separate close to the Earth and the
Astronauts returned to the ground.
Luna Program
The Luna 16 spacecraft landed on the Moon’s surface by ignition of its engine KTDU-417 in
the periselen. After a burning time of 263 s the engine was shut down at an altitude of 2.45 km.
This was followed by a 43 s unpropelled coast. At 600 m altitude, the engine was reignited and
run with throttled thrust. After the descent to 20 m, the main engine was shut off and the low-
pressure combustion chambers of KTDU-417B were ignited. The 1880 kg heavy lander touched
ground with 4.8 m/s. The lander was also the launch pad of the return rocket, which was a 2 m
tall open structure including the capsule with a control system, three spheric propellant tanks
and a non-throttleable rocket engine KRD-61. Four RCS thrusters were used to control the
attitude. After a burn of 53 s the rocket reached the necessary velocity of 2700 m/s to reach
the Moon escape and Earth reentry trajectory. Reentry into Earth’s atmosphere was done with
the 35 kg heavy reentry capsule after 84 h from lift-off at the Moon. Reentry happened with a
velocity of 10 950 m/s [1].
13
Deep-throttling
Deep-throttling is an important feature of descent engines for a soft lunar landing. The Apollo
descent engine LMDE was throttleable down to nearly 10 % of its maximum thrust. The
descent engine of Luna KTDU-417 was only throttleable to about 40 %. However, it had a
second engine unit KTDU-417B which was used for the final descent and had much lower
thrust level than the main engine. Throttling the engine has several purposes:
• Following optimized thrust profiles during boost phases of trajectories
• Decreasing maximum loads when propellant mass is reducing
• Rendezvous in orbit
• Versatility: High ∆V and fine trajectory control burns; this saves also engine mass, since
one engine fills several roles
• Descent on planets; multi-purpose use of single engine
– Hovering for landing site selection
– Obstacle evasion
– Soft landing
– Landing abort
Betts and Frederick [17] state, that for liquid rocket engines throttling to 25 % usually is easily
achievable without putting a lot of effort in valve and injector design. However, at even lower
pressures, so called deep-throttling, the pressure drop over the injector becomes very small.
This affects propellant vaporization and mixing. These problems are mitigated by using high
pressure drop injectors, actuated injectors (e.g. pintle-injector) or gaseous injection (where the
oxidizer is vaporized before passing the injection). Also chamber cooling is an issue with low
mass flow. Control valves need to have high precision with low mass flow rates. Turbo-pumps,
gas generators, pre-burners or heat-exchangers need a wide range of performance in order to
support throttling.
14
3. Theory of Hybrid Rocket Engines
3.1. Hybrid Rocket Engine Layout
The principle of hybrid rocket engines is to use solid fuels and liquid oxidizers. In some rare
set-ups, this is inversed and the oxidizer is solid and burnt with liquid fuel, but these cases shall
be neglected within this work. The liquid oxidizer is stored in a tank, while the solid fuel is
stored inside the combustion chamber as illustrated in Figure 3.1. The oxidizer will be injected
into the combustion chamber and the combustion will be ignited. The oxidizer flows along the
solid fuel surfaces, which consequently regresses and adds fuel mass flow to the combustion
process. At the end of the solid fuel, there is an optional post combustion chamber and then
the combustion gases are expanded through a nozzle. The liquid oxidizer is either pressure-fed
or pump-fed. The pressure feeding is either realized by a pressurization system or by using a
self-pressurizing liquid (e.g. nitrous oxide). This is suitable for sounding rocket and spacecraft
applications. For some high power applications like launch vehicle boosters, turbo-pumps are
conceivable to feed the oxidizer into the combustion chamber. Recent studies also indicate
that electric pumps might reduce the gross lift off mass of hybrid rocket propulsion systems,
as no heavy pressurization tank is needed [18]. Even if the lift off mass is not reduced by an
electric pump, it could also reduce the rocket length, if this is a critical design parameter. A
detailed description of the history and development of hybrid rocket engines is found in [19].
Hybrid rocket engines sometimes are even seen as the fail-safe space transportation option,
which could raise the safety and reliability of launchers and spacecraft up to the levels of civil
aviation aircrafts [20].
Figure 3.1.: Hybrid rocket engine layout,[21]
15
3.2. Hybrid Rocket Liquid Oxidizers
In principle, every liquid oxidizer is usable with hybrid rocket engines. The aim of hybrid
rocket engines is to surpass the low energetic solid rocket motors. Therefore, in most cases
oxidizers are chosen which offer high energetic combustion:
• Liquid, cryogenic oxygen LOX
– gaseous oxygen GOX is also used for laboratory scale engines
• Hydrogen peroxide H2 O2 with high concentration ("High test peroxide", HTP) from 85 %
to 98 % with the rest of the liquid being water and stabilizers
• Nitrous oxide N2 O
Liquid oxygen of course offers the highest specific impulse, yet its cryogenic storability makes
it less attractive for some applications (e.g. long-time storable rocket engines). Hydrogen
peroxide has a lower maximum specific impulse than liquid oxygen, but it is longer storable,
although it also has a slow, natural decay into water and oxygen. With stabilizers the storability
can be years with only little degradation. An advantage of hydrogen peroxide is that it is
ignited when passing a catalyst bed. This means that hybrid rocket engines with hydrogen
peroxide do not need a pyrotechnical or electrical igniter. Hydrogen oxide is caustic and needs
higher safety measurements during operations and storage as it can start a highly exothermic
decomposition process. A little less problematic than H2 O2 is nitrous oxide. Stored under
high pressure (~50 bar), nitrous oxide is liquid at room temperature. Yet, it is not toxic
or caustic, which makes it favorable for low cost operations. Table 3.1 lists the most used
oxidizers and their characteristics. In general, also other storable oxidizers like dinitrogen
tetroxide N2 O4 are possible as oxidizers and can improve the performance of storable hybrid
rocket engines. However, since hybrid rocket engines are investigated as an alternative to
storable liquid propellants like hydrazine and dinitrogen tetroxide, the advantage of having
“green” oxidizers is predominant. It reduces operation costs during the development of rocket
engines and during ground operations on the launch site by a high amount. The impact on
the environment is also reduced. However, for certain in-space applications, none of the above
mentioned "green" oxidizers can be used. A recent design study by NASA JPL for a Mars ascent
vehicle for a sample return vehicle suggests a hybrid rocket engine using paraffin-based fuel
"SP7" and Mixed Oxides of Nitrogen (MON30) as oxidizer [18]. According to the study, the
advantage of MON30 is that it has a very low melting temperature and is usable as a liquid
oxidizer down to −71 ◦C. This simplifies the thermal management while landed on the planet
Mars significantly. Also the paraffin-based fuel SP7 was specifically designed to withstand
cold temperatures on Mars’ surface. MON30 is a mixture of 70 % N2 O4 and 30 % NO. Of
course this oxidizer increases operations costs and test campaign costs. However, it extends
the feasibility of hybrid rocket engines for a very broad area of environmental conditions of in-
space applications. Some even more advanced ideas include also air-breathing hybrid (rocket)
engines for the use in Earth’s atmosphere [22].
16
Table 3.1.: Propellant options for hybrid rocket engines, [23]
Oxidizers Advantages & disadvantages
+ highest possible specific impulse

Liquid Oxygen - cryogenic
- medium density (1140 kg/m3 )
- optimal mixture ratio 2.5-3 (long fuel grains)
+ storable at standard conditions
+ highest density (>1400 kg/m3 )
+ usable as monopropellant
Hydrogen Peroxide + (re-)ignition with catalyst
+ optimal mixture ratio 7-9 (short fuel grains)
- caustic
- (slow) natural decomposition
- lower specific impulse
+ self-pressurizing
+ usable as monopropellant
+ (re-)ignition with catalyst
Nitrous Oxide + optimal mixture ratio 7-9 (short fuel grains)
- lowest density (<900 kg/m3 )
- high storing pressure (~50 bar)
+ operation temperature down to −71 ◦C
+ high density (1300-1500 kg/m3 [24])
MON30 (70 % N2 O4 , 30 % NO) + optimal mixture ratio 4-5
- toxic, carcinogenic
Fuels Advantages & disadvantages
+ high mechanical resistance

+ well processible with machines
HTPB + density (~930 kg/m3 )
+ well mixable with additives
- very slow regression rate
+ very high regression rate
pure paraffin wax + density (~900 kg/m3 )
- low mechanical resistance
- less mixable with additives (sedimentation)
+ high regression rate
Paraffin-based fuels + density (~900-930 kg/m3 )
(+) higher mechanical resistance due to polymeric additives
(+) quite mixable with additives (little sedimentation)
17
3.3. Hybrid Rocket Fuels
The amount of propellant combinations of hybrid rockets exceeds the one of solid rocket
engines. This is mainly because of the huge number of possible fuels. There are many
possibilities to compose fuel grains for hybrid rocket engines. Starting with wood, which
was tested in some of the first rocket engines [19], over pure polymeric fuels, to fuel blends
with metallic and other energetic additives, there is a broad spectrum of fuel combinations.
The most common hybrid rocket fuels are polymeric fuels, either pure or with the addition
of energetic materials to improve burning rates and exhaust velocity. The most common
polymeric fuels in hybrid rocket engines technology are the ones based on butadiene. The most
popular one with high availability and low cost is hydroxyl-terminated polybutadiene (HTPB).
A new class of solid fuels for hybrid rocket engines is liquefying fuel. This was first identified
as a possible game changer by Karabeyoglu et al. [25, 26]. At the DLR Lampoldshausen
liquefying fuels are studied for many years [27, 28, 29]. While the first liquefying fuels where
frozen gases or liquids, in order to provide maximum exhaust velocities to hybrid rockets, the
cost for cryogenic combustion chambers was much too high. By now the paraffin-based fuels
found their way into hybrid rocket engines, offering a similar exhaust velocity like polymers.
Due to the liquid melt layer, which they form on the burning surface, their mass flow to the
combustion zone is much higher than that of polymeric fuels, which enables a more compact
and versatile engine design. Especially where there is the demand for high thrust and shorter
burn durations, like in boosters and sounding rockets, liquefying fuels have a high advantage
over classical hybrid rocket fuels. While pure paraffin offers the highest regression rate (up
to factor 6 compared to HTPB), it has a very low mechanical strength. Therefore, advanced
paraffin-based fuels have been developed. Paraffin can easily be mixed with polymers up to a
certain concentration. Adding more polymers increases the mechanical strength, but reduces
the regression rate [30, 31]. The regression rate reduces with increasing viscosity of the melt
layer. A compromise is found depending on the application. In paraffin-based fuels the mixing
of additives like energetic metals is also easier than in pure paraffin, as the high viscosity
prohibits the sedimentation during the casting process. Table 3.1 compares the three fuel
options. The performance of several propellant combinations is given in Table 3.2, where O/F
is the mixture ratio, c∗ is the characteristic velocity in the combustion chamber and Ivac is the
specific impulse in vacuum conditions. The evaluated performance is under the conditions
of a chamber pressure pcc = 15 bar and an expansion ratio of = 60. As can be seen, the
performance is very similar to liquid carbohydrate fuels like RP-1. It must be noted that the
performance calculation for H2 O2 is done with a pureness of 95 % and the oxidizer is injected
as a liquid.
Table 3.3 lists the different hydrogen peroxide combinations with a paraffin-based fuel
including 5 % of polymeric additives (PB-5%), taking into account liquid and gaseous injection,
as well as two concentrations. The state of matter of the H2 O2 at injection has a quite high
impact. By this, injection of H2 O2 with a catalyst for the whole oxidizer mass flow should
increase the rocket engine’s performance, as no energy is needed in the combustion chamber
18
Table 3.2.: Propellant combinations [23]
Oxidizer Fuel O/F opt c∗ [ ms ] Ivac [s] ∆Ivac
LOX HTPB 2.4 1725.4 356.3 98.4 %

LOX HTPB/Al(30 %) 1.5 1727.2 356.4 98.4 %
LOX PB-5% 2.75 1752.2 361.7 99.9 %
LOX PB-5%/Al(20 %) 2.05 1757.1 362.1 100.0 %
LOX RP-1 2.75 1763.6 363.1 100.3 %
H2 O2 HTPB 6.8 1610.6 326.3 90.1 %

H2 O2 HTPB/Al(30 %) 5 1618.6 333 92.0 %
H2 O2 PB-5% 7.6 1621.6 327.5 90.4 %
H2 O2 PB-5%/Al(20 %) 6.4 1623.6 331.7 91.6 %
N2 O HTPB 8.2 1565.7 317.1 87.6 %

N2 O HTPB/Al(30 %) 5.6 1587.4 323.1 89.2 %
N2 O PB-5% 9.2 1574.7 318.2 87.9 %
N2 O PB-5%/Al(20 %) 7.4 1584.5 321.8 88.9 %
MON30 HTPB 4.1 1641.50 337.8 93.3 %

MON30 HTPB/Al(30 %) 2.6 1661.60 341.5 94.3 %
MON30 PB-5% 4.7 1652.2 340.9 94.1 %
MON30 PB-5%/Al(20 %) 3.6 1667.60 343.2 94.8 %
Calculation Settings:
pcc = 15 bar, = 60, frozen equilibrium, 95 % H2 O2 in liquid state
to vaporize the oxidizer. It has to be experimentally validated that this theoretical ideal values,
calculated with NASA CEA can be actually reached with an catalyst injection system. It must
be also taken into account that behind the catalyst the oxidizer is not pure H2 O2 anymore. The
oxidizer flow is a mixture of gaseous H2 O2 steam, water steam and gaseous oxygen. This also
alters the performance calculation.
Table 3.3.: Hydrogen peroxide combinations with PB-5%

Oxidizer Conc. State O/F opt c∗ [ ms ] Ivac [s] ∆Ivac
H2 O2 95 % liquid 7.6 1621.6 327.5 90.4 %

H2 O2 87.5 % liquid 8.3 1579.5 315.4 87.1 %
H2 O2 95 % gas 7.3 1723.4 352.9 97.5 %
H2 O2 87.5 % gas 8.15 1692.8 344.7 95.2 %
Calculation Settings:
pcc = 15 bar, = 60, frozen equilibrium, ∆Ivac compared to LOX PB-5%/Al(20 %)
19
Figure 3.2 shows the specific impulse over the mixture ratio. It is noteable, that for oxygen
the gradients are much higher. This means, that O/F shift is of higher impact with oxygen
as the oxidizer than with H2 O2 or N2 O. Several hybrid sounding rockets are currently under
development, some of which already use the advantages of liquefying fuels. The student group
Hybrid Engine Development (HyEnD) at the University of Stuttgart uses paraffin-based fuels
(with N2 O) [2, 32, 33, 34, 35], as well as the Peregrine project at University of Stanford and
NASA Ames (with LOX) [36]. The student project Stratos at TU Delft is using N2 O with
sorbitol-based liquefying fuel [37]. Commercial projects until now focus on the better-known,
better-understood classical hybrid fuels. One European example is Nammo and their sounding
rocket project North Star with the combination HTPB and H2 O2 [38]. Nammo plans using
a modular design developed in steps to finally have a hybrid rocket engine based launcher
to polar orbits for small payloads. The European Union project HYPROGEO is focused on
developing a hybrid rocket engine optimized for space application [39, 40]. In order to reach
long burning time, they analyze a front burning hybrid rocket engine. Another project focuses
on using hybrid rocket engines for space debris removal or prevention [41]. The Spartan project
was aimed to develop an advanced throttling engine for soft planetary landing [42]. NASA JPL
is conducting studies for a Mars Ascent Vehicle (MAV) using hybrid rocket propulsion with
paraffin-based fuel and MON30. Both the oxidizer and the fuel have been chosen to withstand
the extreme temperature changes on the Mars surface [18].
Figure 3.2.: Specific vacuum impulse with PB-5% over mixture ratio
20
3.4. Hybrid Rocket Engine Components
The oxidizer tank for hybrid rocket engines usually is pressurized to feed the oxidizer into the
combustion chamber without pumps. This allows for simple fluid control systems. However,
it adds extra mass, as the tanks need to withstand high loads. Especially when used in
atmosphere and engine pressure needs to be high, oxidizer tanks become very heavy if they
need to withstand high pressure loads. When nitrous oxide is used as an oxidizer, even higher
working pressures of up to 60 bar need to be considered. However, with the saving of a
pressurization system, nitrous oxide still stays a viable option for small hybrid rocket engines.
For larger hybrid rocket engines like in boosters for heavy launch vehicles or in first stages for
a hybrid small launcher, it is possible to use turbo pumps for feeding the oxidizer. The same
engine cycles like in liquid rocket engines are possible, except that always only one liquid needs
to be pumped, reducing the necessary power [43]. This allows compact turbo pumps. Staged
combustion cycles with a pre-burner, gas generator cycles or expander cycles are possible. Pre-
burners and gas generators are either hybrid rocket engines as well or catalytic monopropellant
combustors of H2 O2 . A very important component for hybrid rocket engines is the injector.
Since the liquid oxidizer is mixed along the chamber with fuel in the boundary layer, the
injection conditions are very different compared to liquid rocket engines. The injector needs
to be optimized allowing the best possible mixing of fuel and oxidizer. A fine atomization
and fast vaporization of the oxidizer helps the mixing and combustion, which is increasing
engine efficiency and reducing engine length. Shower head injectors with small injection holes
offer a good atomization and are often used with laboratory scale engines. Bigger engines
are often used with swirl injectors and impingement injectors. A very interesting topic is the
injection behavior when the fuel geometry changes a lot during burning time. This is especially
important with quickly regressing fuels like liquefying fuels and if the engine is designed for
a long burning time. Then the difference of flow area in the beginning and end of the burning
time is very high. For a similar combustion efficiency during all the burning time, the injection
conditions need to be very good in the beginning and in the end of the engine’s burn duration.
This has to be considered when designing a long burning hybrid rocket engine. Swirl injection
helps here to keep the oxidizer close to the solid fuel’s surface. For throttleable engines pintle
injectors like in deep-throttling liquid rocket engines are an option, given the mixing with the
fuel is good. For deep-throttling it is also a possibility to heat the oxidizer before passing the
injector, so that a already (partially) vaporized fluid is injected. This increases the pressure
drop over the injector also for low mass flows. With N2 O this is especially easy, since it is
usually at saturation pressure.
The combustion chamber of hybrid rocket engines often is thermally insulated and not using
regenerative cooling. The fuel fills out most part of the combustion chamber and therefore
isolates and cools the chamber ablatively. For other parts of the engine, ablative isolation
materials are used like composite materials with phenolic resin. Also the nozzles are usually
constructed of ablative materials. However, since a liquid oxidizer is used, it is theoretically
possible for high power hybrid rocket engines to use regenerative or film cooling. Yet, this
21
increases system complexity, which is not desired in hybrid rocket engine applications. The
standard shape for hybrid rocket fuels is a hollow cylinder. The edges can be sloped to decrease
flow resistance and vortex development. More sophisticated designs include segmented fuels,
where each segment has a different diameter and sometimes mixing devices are installed
between segments. However, usually mixing devices are installed behind the fuel slab, at the
front of a post combustion chamber, where the mixed fuel and oxidizer have residing time for
the combustion reactions to complete. Longer post combustion chambers usually maximize
combustion efficiency, but increase also engine weight, length and thermal loads. For slow
burning fuels like HTPB it is necessary to have more sophisticated fuel designs in order to
reach high thrust levels. Usually for HTPB fuels star shapes (or other shapes that increase the
surface of the fuel port) are used. Also multi-port designs, where the oxidizer flows through
several channels in the fuel, increase the fuel mass flow and allow for highest thrust levels with
classical fuels. These multi-ports design result in a lot of residual fuel mass in the chamber
after the burning time. The burn has to be stopped before the structure of the fuel collapses
and large fragments of fuel damage the nozzle or block the nozzle throat.
3.5. Advantages and Disadvantages of Hybrid Rocket Engines
The characteristics of hybrid rocket engines lead to the following advantages and disadvan-
tages in the three categories safety, performance and cost [23].
3.5.1. Safety
+ Inert fuel: The fuel grain is produced, transported and handled at high safety & low cost
+ Different state of matter: Liquid oxidizer and solid fuel cannot form an explosive mixture
during any time of production, assembly, ground operation or flight of the spacecraft
+ Grain robustness: The regression rate is not depending on the chamber pressure and the
boundary layer combustion prohibits combustion inside of cracks. Thus, unlike solid
rocket motors, hybrid rocket engines are much less sensitive to cracks in the solid fuel.
3.5.2. Performance
+ Throttling: Hybrid rocket engines are throttleable by controlling the oxidizer mass flow
into the chamber, which is a great advantage over solid rocket motors. In comparison to
liquid rocket engines, only one liquid must be controlled.
+ Shut-down and re-ignition: By closing the oxidizer supply the combustion is stoppable.
Depending on the oxidizer, re-ignition is very easy to achieve (catalytic decomposition of
H2 O2 or N2 O), without hypergolic reactions.
22
+ Propellant versatility: The range of possible propellant combinations is very high for
hybrid rocket engines. This opens up a lot of possible improvements to the propulsion
systems performance.
+ Storability: Derived from the previous point, storable propellants are available. The
propulsion system is simple and storable for a long time without maintenance. This
is mandatory for interplanetary missions.
− Low regression rate: The classical polymeric fuels of hybrid rocket engines like HTPB
offer low regression rates. This results in either low thrust density or complex fuel
geometries, like star-shape or multi-ports. These fuel shapes often result in high residuals
at the end of the burn time, which reduces greatly the performance of the engine.
Also the size of the engine is growing with these shapes. This is improved by current
advancements in the field of liquefying hybrid rocket fuels, which increase the regression
rate by a factor up to six.
− Regression rate prediction: Further problems with the regression rate emerge during
the design of a rocket engine. The regression rate is dependent on the propellant
combination, the injector design, the chamber design and the fuel geometry. It is very
difficult to predict the exact regression rate without doing both lab-scale and full-scale
tests.
− Limited burn duration: With long burn durations, the change of the combustion chamber
internal geometry is very high. It is not proven, that the combustion conditions in a
hybrid rocket engine with long burn duration is at high efficiency at all time, since the
injection behavior of the liquid oxidizer, the boundary flow conditions and the mixing of
the propellants will change over time. Additionally, degradation of combustion chamber
inserts (like mixing devices) or ablation-cooled nozzle parts over a long burn time have a
negative impact on the efficiency. Experimental investigations need to verify the behavior
of hybrid rocket engines with long burn duration.
− Dynamic combustion chamber conditions: Similar to the previous point, over the burn
duration, usually the mixture ratio of a hybrid rocket engine is not constant. This results
in losses of specific impulse compared to the optimal mixture ratio. This is reduced by
controlling the mass flow, but it cannot be avoided completely since the fuel mass flow is
always coupled directly to the oxidizer mass flow.
− Combustion efficiency: The mixing of fuel and oxidizer happens in the boundary layer
flow alongside the fuel grain inside the hybrid rocket engine. The mixing needs more
residing time in the combustion chamber than in a liquid propellants engine. Hence,
for high combustion efficiency a long post combustion chamber is needed. Additional
mixing devices improve the efficiency but increase also the engine weight.
23
3.5.3. Cost
+ High safety: the high safety of the propellants leads to a reduced cost in all phases of
development, testing, manufacturing, and operation.
+ Green propellants: the use of propellants with no or low impact on health and the
environment, make it also possible to reduce the cost at all times.
+ Lower system complexity compared to liquid propulsion systems: Lower system com-
plexity simplifies development and production.
3.6. Hybrid Rocket Propulsion Research
Figure B.1 shows a map of all European activities on hybrid propulsion to the knowledge of
the author. All activities are listed in Table B.1. A short overview over the different working
fields is given. This list is without any claim to completeness as the situation is changing every
few months with new groups and companies emerging and others stopping their work. In
summary a lot of universities work on hybrid rocket engines for education of students, for
testing small to midscale rocket engines and to bring forward fundamental research in the
field. Similarly institutional research centers around the world work on fundamentals, larger
tests and applications for hybrid rocket engines. Some companies already try to use hybrid
rocket engines for commercial applications, but the number is still small and the projects are
mostly in an early stage.
3.7. Internal Ballistics of Hybrid Rocket Engines
This section is giving a short overview of the combustion processes of hybrid rocket engines
and its modeling. Detailed information about classical hybrid theory (i.e. polymeric fuels) is
found in "Fundamentals of Hybrid Rocket Combustion and Propulsion" [19]. Karabeyoglu et
al. give a detailed description of the processes and models of liquefying fuels inside hybrid
rocket engines in [25, 26]. The combustion of hybrids is defined by the boundary layer of
the oxidizer flow along the surface of the fuel Af , where the fuel mass flow ṁf is increased
continuously. The fuel mass flow is calculated with the surface, the fuel density ρs and the
regression rate ṙf . Fuel and oxidizer react in the flame zone inside the boundary layer. Below
the flame zone, the flow is very fuel rich, while above in the core of the port the flow is oxidizer
rich. The most important parameter is the regression rate of the fuel, which is the velocity
with which the fuel burns normal to the fuel surface. Therefore, it is the parameter, which
determines together with the fuel surface area the mass flow of the fuel. The total mass flow
ṁ of the hybrid rocket engine is calculated with the sum of the oxidizer mass flow ṁox and the
fuel mass flow ṁf :
ṁ = ṁox + ṁf = ṁox + ṙf ρs Af (3.1)
24
This regression rate is dependent on the propellant combination and most of the time has to
be determined experimentally for each propellant combination. In addition, the combustion
chamber layout with injectors, and other flow manipulating geometries, have an impact on the
regression rate. Usually experiments with small-scale rocket engines yield higher regression
rates than larger scale engines. With the current state of the art, it is mandatory to do
experiments with a new rocket engine to measure the regression rate and iterate the design
until the planned operating conditions are met. Modeling laws without relying on empirical
data can only be used for preliminary designs.
3.7.1. Modeling Laws for Classical Hybrid Fuels
Marxman et al. investigated hybrid rocket’s combustion processes in the 1960’s [44]. By
evaluating the energy conservation on the fuel surface and neglecting heat fluxes by radiation
and conduction, they showed that the regression rate ṙf is approximated with the following
model (the equation was derived with imperial units, for SI units the constant 0.036 needs to
be adjusted). It uses the instantaneous mass flux density G, which is the total mass flow at a
position x along the fuel, divided by the flow section diameter. The viscosity of the gas flow µ
has a small effect on the regression rate as well as the ratio of the velocity of the gas in the core
stream and the flame ue/uc and the ratio of total enthalpy difference between flame and fuel to
the effective heat of vaporization ∆h/hv [45].
!−0.2 !0.23
G0.8 x ue ∆h
ṙf = 0.036 (3.2)
ρs µ m uc hvap
Experimental data over the last decades since Marxman’s publication deviate from the results
gained by this equation. Experiments show that the exponent of G should be smaller and
the exponent of x is not exactly -0.2. In reality the dependency of the regression rate on the
position is smaller. An explanation for this is given by Zilliac and Karabeyoglu [45]. They note
that Marxman derived his equations for a flat fuel slab. In a typical hybrid rocket engine, the
fuel grain will be cylindrical, which changes the values of the exponents. Therefore, Zilliac and
Karabeyoglu elaborated a more detailed model for regression rates of classical, non-liquefying
fuels in cylindrical combustion chambers. Figure 3.3 shows all the included processes in the
model. They state that the regression rate of the fuel is approximated by:
 !−0.22 
A  Gox D x 
ṙf = 1 + C1 e−0.4 D  Gk+1 D k (3.3)

ρs µm
In this equation they respected the circular port shape as well as heat flows due to convection
and radiation from the flame zone to the fuel grain. Additionally, the position of the flame zone
in the boundary flow has been included in the derivation of the equation. In tests, the flame
zone was observed to be located approximately in the middle of the boundary zone. The tube
diameter D as well as a new coefficient k have an increased influence on the regression rate
approximation. The parameter A, which is assumed to be a constant for a certain test setup is
25
Figure 3.3.: Schematic of classical fuel’s reaction process [45]
defined as: !0.23

0.022 ∆h
A= Pr −0.6
φc0.77 (3.4)
µkm (Ts/Tmelt )l hvap
The temperature of the solid propellant surface Ts is either a measured value or it is close to the
vaporization temperature of the fuel. The mean temperature of the gas flow Tmelt is calculated
as the equilibrium temperature of the flow. The Prandtl number P r is either approximated by
1 or calculated by cp µ/kg with the mean heat capacity cp and the mean heat conductivity of the
gas flow kg . The position of the flame sheet relative to the fuel surface φc is defined as:
1.22 (O/F ) ∆h
h uc
v
φc = = (3.5)
Kox + (O/F + Kox ) ∆h
hv
ue
The mixture ratio O/F is the local mixture ratio in the flame zone along the fuel’s x-axis.
Currently there exists no model for the calculation of this variable. However, the stoichiometric
value for the fuel and oxidizer combination is used until a better model is available. The
concentration of the oxidizer Kox is usually close to 1, except when the oxidizer inflow is diluted
by an inert medium. For some laboratory set-ups nitrogen is used to dilute the oxidizer in order
to decrease combustion temperature. However, although this model is much more detailed and
doesn’t overestimate the influence of the fuel length on the regression rate anymore, the model
is highly dependent on the viscosity µ in the flow, which is very hard to calculate precisely, and
the coefficient k, which is approximated with:
!
Ea
k = −0.005 − 0.08 (3.6)
Ru Ts
The activation energy of the chemical reaction Ea is a result of the propellant combination and
depends on the scale of the heat flow [45, 46]. So it is not a constant value for a given propellant
combination. Ru is the universal gas constant and Ts the solid fuel surface temperature,
which is also hard to determine. A first assumption, without conducting measurements in
experiments, is to assume that the surface temperature is close to the vaporization temperature
of the fuel. Zilliac and Karabeyoglu conclude that their model is very well suited for classical
hybrid fuels. However, the model is very sensitive to the fuel surface temperature Ts . It is
additionally dependent on the local O/F in the flame zone, for which there is currently no
26
model available. The activation energy Ea and the viscosity µm are also hard to obtain with high
precision. Hence, comparisons with tests need to be evaluated to get a suitable model for the
regression rate for the corresponding propellant combination. Zilliac and Karabeyoglu state,
that their model is still not sufficient to supplant determination of regression rates in rocket
engine burning tests. Therefore a simplified equation with empirically determined coefficients
a, n, and m is widely used for engine design and test evaluation:
ṙf = aGn xm (3.7)
Going along the fuel grain, more fuel will be added to the mass flow. Thus, the mass flux
term Gn increases and xm decreases. It is found that for most propellant combinations this two
effects seem to even themselves out, resulting in a mostly constant regression rate over the fuel
length. In addition, it requires very sophisticated instrumentation to measure the regression
rate exact in time and space. This is why the regression rate data is mostly determined time-
and space-averaged over a whole test and fuel (section), as well as the average oxidizer mass
flux Gox . Since these measured values are used to calculate and design new engines, often the
regression rate law is further simplified to the following:
n
ṙf = aGox (3.8)
Zilliac and Karabeyoglu note that the regression rate also has a high dependency on scale of
the engine. Therefore the scale factor α is added.
n
ṙf = αaGox (3.9)
These simplified regression rate laws is used either to simulate existing engines, where the
coefficients are known from experiments, or to design new engines. Since the scale effects are
not exactly known before, all regression rate models need to be validated in experiments. Then
the fuel geometry and oxidizer mass flow needs to be adjusted accordingly, in order to reach
the designated thrust and mixture ratio.
3.7.2. Regression Law for Liquefying Fuels
The equations mentioned before are only valid for classical hybrid rocket fuels. They do not
predict the regression rate of liquefying fuels. A great part of the fuel mass transfer with
liquefying fuels is not due to vaporization at the flame zone inside the boundary layer, but
due to droplet entrainment. The entrained droplets burn inside the oxidizer rich zone with
the necessary heat being transferred through the boundary layer. The increased mass transfer
from the boundary layer to the core flow consequently increases the regression rate. Higher
regression rates from a factor 2 to 5 have been found in experiments with liquefying fuels both
by experiments of Karabeyoglu et al. as well as DLR Lampoldshausen by Kobald et al. [25,
27, 28]. The droplet entrainment is currently investigated with optical measurement methods
27
at DLR Lampoldshausen by Kobald and Petrarolo [47] and previously the instability of the
liquid film has been analyzed in order to understand the combustion in more detail [48]. The
aim of many research groups around the world is to improve existing modeling theories about
liquefying hybrid fuels.
An extended theory about regression rate has been formulated by Karabeyoglu, Altman and
Cantwell in 2002 [25]. This theory includes also fuels that form a melt layer. First, they give an
empirical equation for the critical conditions, which define the starting point for entrainment
processes, from Nigmatulin et al. [50]:
1.3 0.6
1 ρg µ l σ
G1.6 h0.6
t ≥ 2.5 · 10
−3
, (3.10)
cf0.8 ρl0.3 µg
where ht is the thickness of the liquid melt layer, cf is the skin-friction coefficient, ρg is the
gas flow density, ρl is the liquid fuel density, µl is the liquid fuel viscosity and σ is the surface
tension of the liquid melt layer. The skin-friction coefficient is calculated as follows [51].
cf = 0.074Rex−0.2 (3.11)
The total regression rate with droplet entrainment ṙf is the sum of vaporization and entrain-
ment regression rates ṙv and ṙent . The vaporization regression rate hereby stays the same like
in Marxman’s classical theory.
ṙf = ṙv + ṙent (3.12)
Further, Karabeyoglu et al. conducted a thermal analysis of the thickness of the melt layer. By
evaluating the convective and radiative heat flux, two limits of the melt layer thickness could
be stated. In the first case, the radiation heat flux is absorbed completely in the liquid layer,
Figure 3.4.: Schematic of liquefying fuel’s reaction process [49]
28
which equals an opaque liquid layer. In applications with liquefying hybrid rocket fuel, the
paraffin is most of the time blackened with e.g. Carbon Black. This is done to reduce heating
of the complete fuel block in an instant, by increasing the absorption at the fuel’s surface.
Therefore for most practical applications of liquefying fuels in hybrid rocket engines, this first
case for the thickness of the melt layer ht in mm is the fitting one:
cp,l ∆T1
!
ht = δl ln 1 + , (3.13)
hmelt
∆T1 = Tvap − Tmelt , (3.14)

κ l ρl
δl = , (3.15)
ṙf ρs
where cp,l is the specific heat of liquid fuel and hmelt is the total melting heat. ∆T1 is the
temperature difference between the melting and vaporization temperatures Tmelt , Tvap . δ1 is
the thermal thickness in mm. κl is the thermal diffusivity of the fuel.
λl
κl = (3.16)
cp,l ρl
In the second case, the radiation absorption is very small, so that nearly all radiation is
absorbed by the whole solid fuel. This applies for solid fuels, that are translucent. In this
case the thickness of the liquid layer is
cp,l ∆T1
!
ht = δ1 ln 1 + , (3.17)
hmelt − hvap Q̇r/Q̇w
where hv is the total heat of vaporization and Q̇r/Q̇w is the ratio of radiative to the total heat
transfer between fuel and core flow. In a next step, the analysis of Karabeyoglu and his
colleagues was focused on the stability of the liquid layer and the connected entrainment of
fuel droplets into the oxidizer stream. They state, that the mass flow by droplet entrainment
ṁent is depending on the mass flow inside the liquid melt layer ṁl and the flow conditions
in the gaseous flow. For this they introduce the parameter Xe which is calculated with the
dynamic pressure Pd , the surface tension σ and the ratio Tg/Tvap of the gas and vaporization
temperature. An empirical relation e0 (Xe ) is defined. The melt layer mass flow per unit area
ṁl can be calculated by the shear force balance at the liquid gas interface.
Pd cf h2t ρl
ṁl = (3.18)
2µl
ṁent = 13.3e0 (Xe ) ṁl (3.19)

!0.25
(2Pd )0.5 Tg
Xe = (3.20)
σ Tvap
e0 (Xe ) = 1 − exp [−1.06e−4 (Xe − 2109)] (3.21)
29
The formulation of Xe in [25] is inconsistent compared with the original publication of Gater
and L’Ecuyer [52]. Karabeyoglu used the dynamic pressure Pd while Gater used a momentum
factor M0 = ρu 2 , which is M0 = 2 ∗ Pd . For this work, the factor of 2 was considered. Also the
exponent of the fraction Tg/Tvap is unclear in Karabeyoglu’s publication. Therefore it was also
chosen as Gater published it. A correction of this relation has to be done for high mass transfer
rates. The reason is that the entrainment mass equation does not regard the effects of reduction
in the liquid mass flow rate along the axial direction. The correction was done with empirical
data and resulted in the following formulation:
ṁent = −13.3 ln (1 − e0 (Xe )) ṁl (3.22)
With substitution it follows
ṁent = ρl ṙent = 1.41e−3 (Xe − 2109) ṁl (3.23)
ṁl
ṙent = 1.41e−3 (Xe − 2109) . (3.24)
ρl
If Xe < 2109 there is no entrainment mass flow. The entrainment mass flow is proportional to
the dynamic pressure Pd , the liquid film thickness ht and the fuel properties viscosity µl and
surface tension σ .
β
Pdα ht
ṁent γ (3.25)
µl σ π
Table 3.4 shows the parameters for this equation. The values for the four coefficients given by
Gater and L’Ecuyer are the ones derived with the theory. Karabeyoglu et al adjusted the values
to fit their empirical data. This means, while α can be chosen as 1.5, for the application of the
theory, the coefficient β is an additional input parameter. Following the Equation 3.18 β should
be 2. However Karabeyoglu et al. [25] did a fit to experimental results and found that a β of 1.0-
1.5 fits better. This behaviour shows, that the equations do only approximate the real processes
and are missing some parts and details. This of course makes it more difficult to apply the
theory to propellant combinations, where no experimental data is available. However, the
model anyway gives a good prediction for liquefying fuels, much better than the regression
models for classical fuels.
Table 3.4.: Propellant options for hybrid rocket engines, [23]

Source α β γ π
Karabeyoglu [25] 1.5 1.5 >π <γ

Gater and L’Ecuyer [52] 1.5 2 1 1
Nigmatulin [25] 1 1 - -
30
3.7.3. Correction of the Classical Regression Rate Law
Karabeyoglu et al. suggest three modifications to the classical theory. Firstly, the ratio of
enthalpy difference to effective heat of vaporization is altered since the effective heat of
vaporization is reduced, because the evaporation energy for the same fuel mass transfer is
highly reduced by the entrainment process. Secondly, the blowing factor is altered due to
the two-phase flow. They assume that the droplets do not vaporize below the flame sheet.
Therefore, the blowing parameter only needs to respect the vaporization mass transfer from
the surface to the flame zone. A third modification is the increase in heat transfer to the surface,
since waves on the liquid surface increase the surface area. Further, they extend the blowing
factor curve to match the blowing parameter areas of liquefying fuel. With these modifications,
they state the classical regression rate ṙcl with the ratio of radiative to convective heat transfer
Q̇r/Q̇c , the blowing parameter B and the blowing correction coefficient CB1 .
0.03µ0.2
!
g Q̇r
ṙcl = 1+ BCB1 G0.8 x−0.2 (3.26)
ρf Q̇c
The blowing correction coefficients CB1 and the blowing parameter B are defined as:
2
CB1 = (3.27)
2 + 1.25B0.75
ṙf ρf
B= cf (3.28)
2 G
Setting up the energy balance at the liquid gas interface a relation for the vaporization and
entrainment regression rate can be found with the ratio of effective heats of gasification for
entrainment and vaporization Rhe and Rhv :
0.03µ0.2
!
ṙv Q̇r C

g
ṙv + Rhe + Rhv ṙent = Fr 1+ B H G0.8 x−0.2 (3.29)
ṙ ρf Q̇c CH0
ṙ = ṙv + ṙent (3.30)
The blowing factor CH/CH0 is calculated as
CH 2 CB1
= = , (3.31)
CH0 2 + 1.25B0.75
g CB1 + CB2 (ṙv/ṙcl )0.75
where the blowing correction coefficients CB2 nd the vaporization blowing parameter Bg are
defined as:
1.25B0.75
CB2 = (3.32)
2 + 1.25B0.75
ṙv ρf
Bg = cf (3.33)
2G
31
The heat-transfer correction factor for surface roughness Fr is defined as:
14.1ρg0.4
Fr = 1 + 0.2 (3.34)
Tg
G0.8 Tvap
With this set of differential equations, the regression rate of vapor for the liquefying fuel ṙv
can be calculated. The chemical properties are taken for the mean value between the melting
and vaporization temperature for the liquid phase. The liquid surface tension is evaluated at
the boiling point according to [25]. For many liquefying fuels like paraffin-based fuels, during
regular combustion chamber conditions in a rocket engine, the pressure is higher than the
critical pressure. This means that the regression happens under supercritical conditions where
there is no defined border between liquid and gas phase. Experiments show that the regression
rate of liquefying fuels is still higher, which means that even under supercritical conditions
something similar to droplet entrainment is happening. “Droplets” with higher density fluid
from the solid fuel surface are brought to the oxidizer mass flow by turbulent processes. Adachi
and Shimada investigated supercritical behavior of fuels also by CFD simulations and could
confirm this behavior [53]. Optical investigations of supercritical combustion by Chandler
et al. revealed that there is not the same mechanism of droplet forming like at atmospheric
pressure [54]. However, there were elongated filaments of the flame sheet along the fuels and
upward burst flames, where there was a localized reaction because of a combustible oxidizer
ratio. Therefore, it is assumed that also under supercritical conditions the entrainment model
for liquefying fuels is applicable with the same input data, as increased regression rate could
be confirmed in many experiments.
32
4. Software
4.1. Analysis, Simulation and Trajectory Optimization Software for

Space Applications (ASTOS)
In order to design, optimize and analyze a sample return mission to the Moon, the trajectory
optimization software ASTOS (version 8) is used. ASTOS includes an extensive library of
aerospace vehicle differential equations with 3-DOF or 6-DOF equation of motions. With
boundary constraints, path constraints and cost functions the user can design and optimize
a trajectory without additional coding. The range of possible mission and trajectories that
can be designed and optimized is very broad, including aerodynamically stabilized sounding
rockets, orbital launchers, reentry vehicles, interplanetary missions and orbit transfers. The
mission is designed by setting up a vehicle with the model library by putting together the
respective components like engines, tanks, payload etc. For flight in atmosphere, the full
modelling of aerodynamic configurations is included. There is the possibility to model rocket
engines with predefined thrust and mass flow profiles, or engines with a given specific impulse
and other possibilities to variously model rocket and air breathing engines. An interface for
a completely user defined model is also integrated. The flight dynamics can be modeled with
different equations of motions, best suited to the current flight phase. Controls like thrust and
attitude are optimizable as well as stage sizing and trajectory phases durations. Optimization
is done with the included optimizer CAMTOS [55]. It allows multiple shooting optimization
for phases with high derivatives and the collocation optimization method for phases with low
derivatives but long phase duration. The user can define cost functions for many parameters,
e.g. maximum payload, minimum fuel consumption, heat load reduction etc. Phase constraints
like orbit parameters or maximum acceleration are defined as well. Summarizing, ASTOS
offers a wide palette of possibilities to model any vehicle in atmosphere and / or space and
optimize its trajectory. All the equations of motions and dynamics and disturbances are
already included in ASTOS. Nevertheless, in order to model and analyze a mission trajectory
with ASTOS deep understanding of orbital physics and mechanics are necessary. A very good
introduction to this topic is given by Curtis [56].
4.1.1. Moon Sample Return Analysis with ASTOS
With its vast possibilities, ASTOS allows to model a complete mission from launch on the
Earth until landing on the Moon and return until reentry in Earth’s atmosphere. Yet it
is advisable to model, simulate and optimize parts of the mission that are not necessarily
33
hardly connected to each other separately, since this will reduce optimization time greatly. A
verification that ASTOS optimizes such missions without problems has been done [57]. Some
problems regarding coordinate transformations were identified and fixed by Astos Solutions
GmbH preceding the start of these studies. A sample return mission with hybrid propulsion,
not taking into account the dynamic behavior of hybrid rocket engines, but instead using
average values for mass flow and specific impulse in ASTOS, has been analyzed by Tomilin
[1].
4.1.2. User Propulsion Component for Hybrid Rocket Engines
ASTOS offers an interface in C programming language to include propulsion models, which are
more sophisticated than simple fixed thrust and mass flow curves. This is used for including
a hybrid rocket engine model. Different levels of detail are available for this. A first approach
is to use analytical equations to simulate the time dependent behavior. Models with higher
detail are possible to be implemented by creating an interface between ASTOS and ESPSS
models. Hereby the connected ESPSS models can be as simple as a combustion chamber and
an oxidizer inflow. They can be increased in detail with tanks, pressurization systems, and
even a complete spacecraft’s propulsion system with attitude control thrusters is realizable.
4.1.3. Analytical Hybrid Propulsion Component
The first step in creating a user propulsion component for hybrid rocket engines was to create
a hybrid rocket engine model with analytic equations, avoiding differential equations. The
oxidizer mass flow is a control value, which is set and changed with ASTOS. At the same
time, also the fuel mass in the rocket engine is a state that is transferred from ASTOS to the
model as well as the ambient pressure. The rocket engine geometry, the efficiency and the
propellant combination are specified in an input text file. The regression rate is defined with
two regression parameters a and n (sometimes a and b). The equations are used as described
in Table 4.1. The combustion data and combustion pressure are iterated. The combustion data
is read from a table created with NASA CEA [58]. The model was also extended with a tank
model for self-pressurizing nitrous oxide using Peng-Robinson equations of state according to
[59, 60]. In that case, the mass flow is not a control value of ASTOS. Instead, it is calculated
with the pressure difference between the tank and the combustion chamber with
p
ṁox = CD Ainj 2ρox ∆p. (4.1)
The injector area can be an optimizable parameter in the ASTOS model.
34
Table 4.1.: Equations of analytical hybrid rocket engine model
π 2
Nozzle throat area At with the nozzle throat At = d (4.2)
4 t
diameter dt :
s
4mf
Fuel inner diameter di , outer diameter da : di = da2 − (4.3)
πρf lf
π 2
Flow area Af low : Af low = d (4.4)
4 i
ṁ
Oxidizer mass flux GOx : GOx = Ox (4.5)
Af low
n
Regression rate ṙ: ṙ = aGox (4.6)
Fuel mass flow ṁf u with fuel length lf : ṁf u = ṙρf πdi lf (4.7)
Total mass flow ṁtot : ṁtot = ṁf u + ṁox (4.8)

ṁox
Mixture ratio O/F :
O/F = (4.9)
ṁf u
h i
Combustion data1 : cF , c∗ , Isp , Ivac , γ = f (pcc , , O/F ) (4.10)
ṁtot ηc∗ c∗
Combustion pressure pcc : pcc = (4.11)
At
Thrust for adapted conditions Fad : Fad = ṁtot ηIsp Isp (4.12)
pe
Nozzle exit pressure pe : pe = pcc (4.13)
pcc
r !− γ1  ! γ−1 −0.5
pe γ − 1 pe  p e
γ 
Pressure ratio equation pcc :
=Γ 1 −  (4.14)
γ + 1 pcc  pcc 
1
! γ−1
2
Function Γ : Γ= (4.15)
γ +1
Thrust for ambient conditions Famb : Famb = Fad + (pe − pamb ) At (4.16)
Condition for separated nozzle flow: pe < 0.4pamb (4.17)

 ! 1 
 2 γ−1 pamb 
Thrust for separated nozzle flow: Fsep = pcc At 2
 −  (4.18)
γ +1 pcc 
1 Thrust coefficient cF , characteristic velocity c∗ , specific impulse Isp , vacuum impulse Ivac and heat
capacity ratio γ from NASA CEA [58]
35
4.2. EcosimPro & ESPSS
EcosimPro by Empresarios Agrupados is a simulation platform for modeling 0D or 1D systems

of differential-algebraic equations [61]. With different libraries, that include components for
modeling complex systems, different fields of applications can be served:
• Space applications with propulsion or life-support
• Aeronautics applications (gas turbines, electric systems etc.)
• Power plants, electronics, thermos-hydraulics etc.
For modeling spacecraft and launcher propulsion systems, the European Space Propulsion
System Simulation library (ESPSS) is used. It is developed at behalf of the European Space
Agency (ESA) [62]. ESPSS offers a large palette of components to model rocket engine cycles
with combustion chambers, nozzles, fluid management parts, turbo pumps, gas generators et
cetera in different levels of detail.
4.2.1. Implementation of Hybrid Rocket Engines
In its current version, ESPSS 3.1 includes a 1D non-adiabatic component for hybrid combus-
tors, either as a pre-burner without nozzle or with a nozzle for supersonic expansion (see
ESPSS User Manual section 8.3.6) [63]. It includes the combustion chamber and an injector
with a cavity volume. The chamber can be set with user-defined amount of nodes in the
subsonic chamber and supersonic nozzle sections. For the combustion process, two options
are available:
• All vapors of fuel and oxidizer will react instantly in the node
• Reaction time delay between the equilibrium and the actual burned gases composition
The latter is a good approach to simulate real hybrid rocket engine’s combustion processes
where the efficiency of the engine depends on the reaction speed and mixing of fuel and
oxidizer along the length of the engine. For every node in the combustion chamber, the fuel
thickness can be set. The fuel composition can be either one of a few pre-defined fuels or a
user-defined fuel. The regression law can be either chosen as dependent on the coefficients a
and b as in (User Manual section 8.2.5):
ṙf = aGb (4.19)
Here, G is the local instantaneous mass flux. This results in higher regression rate towards
the end of the fuel. In contrast to the widely used regression laws derived from the initial
model of Marxman this model has no dependency on the position along the fuel grain. As an
option also the oxidizer mass flux as the injector can be chosen as G, which is preferable, when
the coefficients a and b were determined by time and space averaged experiments with the
measurement of only Gox (see section 3.7). However this approach causes the node’s fuel mass
36
flow to react instantly to a change in the injector’s oxidizer flow, while in reality some finite time
will pass until the changed mass flow density reaches the node’s position. The model optionally
includes also lateral regression and changes the solid fuel’s geometry accordingly over time.
An advanced model for regression is implemented, which assumes a thin vapor layer between
the gases and the grain, and establishes the conservation equations for convective heat and
enthalpy mass flows at both sides of the layer. The vaporization temperature was chosen to be
the saturation temperature of the fuel and the vaporization energy is set to equal the enthalpy
of formation, which is not accurate. This model is a first step in modelling the regression rate
from fuel properties rather than experimental determined coefficients. However, it is not as
accurate as Marxman’s regression laws for classical hybrid polymeric fuels and it also has a
different approach which yields regression rates with offsets of an order of magnitude in some
cases. For liquefying fuels there is currently no advanced model.
4.2.2. Improved Modeling of Classical Fuels in ESPSS
The standard regression rate model, called "std_Hybrid" in ESPSS, has been enhanced by the
possibility to set a regression rate coefficient separately for every node. This enables simulation
of a combustion chamber setup with different, segmented fuels. The same principle has been
applied to the density. For example, the first nodes could be filled with a polymeric solid fuel
like HTPB while the last nodes contain a post combustion chamber with a very slow regressing
insulator like a phenolic composite. It also allows the modeling of mixing devices made
from e.g. graphite or ceramics, which have nearly no regression. In order to analyze hybrid
rocket engines in a wide field of possible applications, the versatility of the ESPSS hybrid
rocket module was improved. Mainly, it should be possible to model hybrid rocket engines
without knowing the regression rate coefficients of a propellant combination. Therefore, the
regression models for non-liquefying fuels introduced in section 3.7 have been included into
ESPSS. This enables modeling of classical hybrid rocket fuels using chemical properties instead
of empirical regression coefficients as an input for the model. Namely next to Marxman’s law
also Zilliac’s and Karabeyoglu’s model has been implemented (see section 3.7). While some
variables in these models are averaged over space and time, in ESPSS these values actually are
calculated with the states in the combustion chamber at the different nodes. Therefore, the
following adaptions have been made to the models. The mean viscosity µm in both models is
the mean viscosity of the flow, including the flame zone, the fuel vapor and the core flow over
the complete burning time and the x-axis of the fuel. Zilliac and Karabeyoglu [45] calculated
the viscosity according to the Kinetic Theory of Gases [64] as:
r
MTb
µe = 26.69e−7 (4.20)
σ2
The molecular weight M was defined as the average molecular weight over the x-axis. The
hard-sphere diameter was chosen as σ = 5. The bulk temperature Tb was approximated as the
mean temperature of the flame temperature and the surface temperature of the solid fuel. With
37
the ESPSS model both the viscosity and the molecular weight are determined at every node.
However, the nodes have an uniform state and do not represent the reality with the highly
inhomogeneous state in radial direction. In order to calculate the viscosity µm for a node, there
are several approaches implemented.
• The same equation as in the original model can be used. The flame temperature and the
surface temperature are inputs of the user, since the combustion in the boundary layer is
not modeled in ESPSS. ESPSS only provides mean values for the temperature per node.
This way of calculating the viscosity shows no dependency on the position on the x-axis.
• The viscosity can be represented as the node’s content’s viscosity. In this way the flame’s
and solid fuel vapor’s viscosity are not included. This model showed large discrepancies
to empirical regression rates. Especially at the injector side’s part of the fuel, where the
flow is dominated by cold oxidizer the local viscosity is highly different to the flame’s and
boundary layer’s viscosities.
• The viscosity can be a mean value of the solid fuel’s vapor viscosity with the vapor
temperature Tvap , the local node’s viscosity and the viscosity with the mean "bulk"
temperature Tb as above. The molecular weight M can still be calculated for every node
within the ESPSS code, instead of using a mean value.
The last option, using a mean value of the three viscosities that play a role in the model,
produces the most realistic prediction of the regression rate for the compared data. However,
further verifications with test results are necessary to evaluate the quality of the model’s
prediction. Both Marxman’s and Zilliac/Karabeyoglu’s model still depend heavily on assump-
tions and simplifications, so both are only suitable for preliminary design. However, since no
regression coefficients a and n are needed, these models are usable with every possible classical,
non-liquefying solid rocket fuel.
The oxidizer concentration Kox is calculated from the injector inflow, by dividing the oxidizer
mass flow by the total injector inflow (which is including also non-condensable gas). For both
Marxman’s and Zilliac/Karabeyoglu’s model future improvements are possible:
• Implementation of a model for the flame mixture ratio
• Modeling the flame temperature
• Implementation of a model for actual surface temperature instead of assuming that it is

a constant value roughly equal to the vaporization temperature
Even though these improvements still are possible, it is questionable how much effect and
improvement it would have on the results. It is expected that also in the future the only way
to get reliable regression rate models, especially for each unique combustion chamber set-up,
is only via experimental tests for every combustion chamber. Table 4.2 shows the additional
inputs that are necessary for the new models, as well as the chosen standard input for HTPB,
according to Zilliac et al [45].
38
Table 4.2.: Additional input parameters for the hybrid rocket component
Parameter Symbol Stand. input HTPB Unit
J [45]
Activation energy Ea 203000 mol
J [65]
Heat of formation Hfθ,sp -380000 kg
J [45]
Heat of vaporization hvap 1812000 kg
Mixture ratio of flame zone O/F 2.7 with O2
J [45]
Specific heat of flame gas cp,c 7880 kg K
J [45]
Specific heat of fuel vapor cp,sp 2386 kg K
Temperature of combustion Tc 3701 K[45]
Temperature of vaporization Tvap 930 K[45]
4.2.3. Enthalpy Balance at the Solid Fuel’s Surface
During development of the models it was also recognized that the current implementation of
the conservation of energy equations at the solid fuel slab was not consistent (ESPSS version
3.2.0). As Figure 3.3 shows, there are several enthalpy flows between the core gas, the flame
zone, the fuel vapor and the solid fuel. As a reference the layer close to the fuel’s surface
is evaluated. The net enthalpy which flows into this layer sublimates and decomposes the
solid fuel. In case of a liquefying fuel the fuel is first melted and then vaporized. For typical
hybrid rocket solid fuels like HTPB or paraffin-based fuels these processes are endothermic.
Additionally a very small fraction of the reaction enthalpy in the combustion chamber is
conducted into the solid fuel, similarly to the combustion chamber wall. To sum up, heat
is coming by convection, radiation and conduction from the flame zone to the reference zone
where the heat is splitting up in four processes:
• Heating up the solid fuel block
• Melting fuel (for liquefying fuels)
• Sublimation or vaporization of the fuel
• Pyrolising fuel (e.g C and H atoms for paraffin)
The ESPSS model was updated with a new heat and enthalpy balance at the solid fuel grain’s
surface. In addition the regression rate model based on vaporization has been corrected and
expanded. The description of the new equations can be found in Appendix C.
4.2.4. Modeling of Liquefying Fuels in ESPSS
The implementation of the regression rate model for liquefying fuels by Karabeyoglu, Altman
and Cantwell [25] is only possible by inserting equations in the "CONTINUOUS"-code of the
hybrid rocket component in ESPSS. This is due to the fact that the model forms an additional
nonlinear set of equations. This set of equations is based on equations 3.12, 3.19 and 3.29
39
including all the equations for the corresponding factors. These factors are calculated in
a newly implemented function. Additionally the relation for the start of entrainment 3.10
according to [50] is implemented in the model. When the entrainment condition is not
active, the regression rate is calculated by the "HybridZilliac" model. This model also replaces
equation 3.26 in the liquefying model, as it has proven to be more accurate while testing
the model, especially since it is considering the effects of a cylindrical fuel geometry. The
liquefying regression rate model is much more sensitive to the combustion port properties.
For example, if used with cryogenic oxidizer, an igniter mass flow must be set, to start the
combustion. Otherwise the cold temperature in the fuel port will reduce the fuel regression
rate. Once the regression rate is high enough, the produced fuel reacts with the oxidizer and
produce heat, which keeps the regression right up, even after the starter mass flow is dropping.
4.2.5. Regression Rate Model Comparison
The regression rate models for classical fuels have been compared to each other with HTPB and
LOX as a reference. No experimental data with HTPB/LOX were available, however literature
values for the regression coefficients are. Table 4.3 lists the input parameters for the models.
As HTPB is a polymer, there are many possibilities of composition regarding the average length
of the polymer molecules. In literature different values are found for heat of formation. The
same applies to heat of vaporization and melting. The stored value in ESPSS was calculated
for a HTPB with an average chain length of 10 carbon atoms. This is rather short, hence
the stored heat of formation is rather low with −380 kJ/kg. Some sources even state positive
values like 363 kJ/kg in [65]. This means the HTPB in this configuration is chemically unstable
and reacts exothermic when decomposed. An often found and cited value is 20.92 kJ/kg which
corresponds to 5000 calories per kg [66]. This value is also positive. This corresponds to the
HTPB binder before curing. The HTPB heat of formation therefore needs careful evaluation
before simulation. Khan et al. calculated the heat of formation for several HTPB configurations
with different lengths of polymers [67]. They also cite −51.9 kJ/mol for [C10 H15.4 O0.08 ]n . This is
the same value as used in ESPSS of −380 kJ/kg.
The models for classical hybrid fuels in ESPSS were named the following:
• vapModel: The default vaporization model, which with realistic values for all input
parameters produces unrealistic values.
• vapModelimproved: Improved model which respects the correct heat flow needed for
vaporizing the fuel mass flow.
• stdHybrid: Model using coefficients a and b for regression rate calculation
• HybridMarxman: Marxman’s law for classical hybrid fuels [44].
• HybridZilliac: Zilliac’s and Karabeyoglu’s improved regression rate law for classical
hybrid fuels according to [45], see Equation 3.3.
40
Table 4.3.: Input parameters for the classical regression rate comparison
Parameter Symbol Input HTPB Unit Used by model
J [45]
Activation energy Ea 203000 mol Zilliac
kJ [45]
Heat of vaporization hvap 1812 kg all
Mixture ratio of flame zone O/F 2.7 with O2 Marxman, Zilliac
J [45]
Specific heat of flame gas cp,c 7880 kg K Marxman & Zilliac
J [45]
Specific heat of fuel vapor cp,vap 2386 kg K Marxman & Zilliac
Marxman, Zilliac,
Temperature of combustion Tc 3701 K [45]
vapModelimproved
Temperature of vaporization Tvap 930 K [45] all
kJ [67, 68]
Heat of formation Hfθ,sp -380 kg all
J [65]
Specific heat of solid fuel cp,s 2900 kg K all
W [65]
Thermal conductivity of solid fuel ks 0.14 mK all
Regression rate coefficient a asp 0.0903 [45] stdHybrid
Regression rate coefficient b bsp 0.527 [45] stdHybrid
To compare the regression rate a simplistic ESPSS schematic has been created. It consists only
of an oxidizer inflow, a mass flow control junction, the hybrid rocket engine model and two
insulations, as shown in Figure 4.1. The hybrid rocket engine has 10 subsonic nodes in the
chamber, of which nodes 2 to 7 are with solid fuel (HTPB). Nodes 1, 8, 9 and 10 are used
as pre- and post-combustion chamber. The solid fuel has an inner diameter of 80 mm and
the chamber has an inner diameter of 120 mm. The nozzle throat diameter is 60 mm and the
LOX mass flow is set to 3.5 kg/s. As the fuel is a single port design, the fuel needs to be 1.8 m
long for this oxidizer mass flow. This results in a thrust of about 10 kN. Figure 4.2a and
Figure 4.2b show the regression rate of the hybrid rocket engine with the "stdHybrid" model
which uses the regression rate law ṙ = asp Gbsp . In Figure 4.2a Gox was used for the mass flux,
which is defined with the oxidizer mass flow per flow area. This is an often used correlation
in experimental investigations. For Figure 4.2b the total mass flux Gtot was used. As the total
mass flux is increased by the transferred fuel mass flow in every node, it is visible that the
regression rate is getting higher along the fuel grain. As mentioned before, the first option is
more precise if the regression coefficients are determined by experimental investigations. The
Figure 4.1.: Schematic of ESPSS model for regression rate comparison
41
second option is more realistic, as the regression rate in a node along the fuel grain will not
react instantaneously to a change in the injected oxidizer mass flow but only after a time delay,
when the mass flow actually reaches a node. For a very long fuel grain however, this model
will not predict the regression rate very well, as the mass flux Gtot will always increase and
therefore the regression rate will always increase, which is not the case in reality, where other
effects will reduce the regression rate (like the blocking effect). Figure 4.2c and Figure 4.2d
show the results with the regression rate models by Marxman and Zilliac. Marxman’s law was
developed for a flat fuel slab and has higher deviations to experimental results in cylindrical
fuel shapes. In the left graph it is visible that the regression rate in the first node is quite high
and then the blocking effect reduces the regression rate in the following nodes. These effects
were avoided by Zilliac in his improved model. The right graph shows that the regression rate
is lower in the first node and is again increasing towards the rear end. While it looks here
like Zilliac’s model would neglect the blocking effect, simulations with a longer fuel show that
the regression rate at the rear end of the fuel grain will reach a maximum. Additionally the
models using the mass flux density G as an input parameter, also are dependent on the amount
of nodes in the model. More nodes make these laws more convergent and precise.
In Figures 4.3a and 4.3b the regression rate for the vaporization model based on the thermal
balance near the fuel is depicted. It is easy to detect that the basic ESPSS model "vapModel" in
Figure 4.3a is overestimating the regression rate by one magnitude with realistic input
values. This is because it neglects the heat of formation of the decomposed gases. The
improved vaporization model in Figure 4.3b predicts the regression rate much better, but still
overestimates it. This is explained by the missing effects which reduce the heat transfer from
the gas flow to the surface, like the blowing effect and the growing thickness of the boundary
layer.
Figure 4.3c shows the regression rate and Figure 4.3d the pressure at the 4th node of the
combustion chamber for the different regression rate models. These graphs give a feeling
for the ability of the models to predict a combustion process in a hybrid rocket engine with
classical fuels. The models by Ziliac and Marxman give the closest approximation and the
deviation to the "stdHybrid" model, which is using the parameters gained by experimental
data, is quite small. The advantage of these models is that they can be applied for new fuels
where no experimental data were available. In these cases they give a first idea of the fuel’s
performance. The vaporization model is much simpler and therefore cannot give a very precise
approximation. In total, the model of Zilliac and Karabeyoglu is promising to give a good
prediction for a cylindrical fuel geometry. The drawback is that this model relies on the largest
amount of fuel properties as an input. This is difficult if the fuel properties are not so well
known yet. These models can only be applied to classical fuels that do not form a melt layer
on the surface. It will also not be possible to design a hybrid rocket engine only with one of
these models without basic research on the selected propellant combination’s behavior, both
theoretically and experimentally.
42
(a) ṙ with "stdHybrid" and Gox option (b) ṙ with "stdHybrid" and Gtot option
43
(c) ṙ with "HybridMarxman" (d) ṙ with "HybridZilliac"
Figure 4.2.: Regression rate and pressure of different models for classical fuels, part 1
(a) ṙ with "vapModel" (b) ṙ with "vapModelimproved"
44
(c) ṙ at the 4th node with different models (d) pressure at the 4th node with different models
Figure 4.3.: Regression rate and pressure of different models for classical fuels, part 2
In the following the simulation of liquefying, paraffin-based fuel PB-5% is displayed. While all
other hybrid rocket engine parameters stay the same compared to the chamber using classical
fuel, the chamber length is reduced to only 0.5 m. The fuel thickness must be increased to
35 mm to sustain a burn time of at least five seconds. The advantage of a liquefying fuel
immediately becomes clear: the fuel is much shorter for the generation of the same thrust
level. However, if the diameter stays the same, the burning duration is of course much lower,
because the fuel mass is also much lower. To have the same thrust for the same burn duration,
a hybrid rocket engine with liquefying fuel needs a much higher diameter than with a classical
fuel. If the engine dry mass is the critical factor for the application of a rocket engine, it is
possible that the optimal chamber length to diameter ratio is reached for regression rates in
between liquefying and classical fuels.
Figure 4.4a and Figure 4.4b illustrate the results of a simulation of a hybrid rocket engine
with liquefying fuel using the model by Karabeyoglu for liquefying fuel’s regression rate. The
regression rate is between 1.5 and 3 times higher than with the classical fuel regression rate
models. Going along the fuel with the increasing nodes from 2 to 7, the regression rate is
increasing. This is due to the high dependency of the modeled regression rate on the local
mass flux G. Since the optimal mixture ratio of oxygen and the paraffin-based fuel is around
2.9, the increase in the mass flux caused by the fuel mass flow has a high impact. The last fuel
nodes regress faster and therefore are depleted after a shorter time. This results in this step-
like depletion, which is also clearly visibly in the pressure curves. This high dependency of the
regression rate on the mass flux and therefore on the position of the fuel length is not congruent
with experimental results, however the trend is correct. With tuning the model’s coefficients α
and β, experimental results could be matched better. For this simulation β = 1.85 was chosen.
One particularity of the shown regression rate curves in Figure 4.4a is, that the regression rate
of a node increases in steps every time a node at the end of the fuel grain is depleting. The
reason for this is found in the equations. The entrainment regression is dependent largely on
the parameter Xe in the Equation 3.24. This parameter again is following the inverse of the
density in the gas flow core via the dynamic pressure Pd . Obviously when a node is depleting
in the simulation, the fuel mass flow drops. This causes a pressure drop in the combustion
chamber. Hence, the density drops which increases the dynamic pressure, and therefore the
regression rate in the remaining fuel nodes. This effect could not be observed in experiments
right now. Mainly, because the depletion of fuel grain segments does not happen abruptly like
in the simulation. For the future a more refined model of the fuel surface could improve this.
One suggestion is to simulate the fuel not as cylindrical segments but as trapezoids. However
this requires larger modifications to the ESPSS component model. For an evaluation of the
influence of the node number a simulation with 50 instead of 10 subsonic chamber nodes has
been done. Figure 4.4c shows that the regression rate is a bit smaller than with more nodes,
however the overall trend is similar. Only 5 out of 43 regression rates for the different nodes
have been plotted for clarity. Since the discretization is much finer, the pressure curves in
Figure 4.4d have much more but smaller steps.
45
To compare the simulation’s result with experimental data, a simulation with the "stdHybrid"-
model using the coefficients a = 0.117 · 10−3 and b = 0.62 has been done, representing the
empirical regression rate law. The results are displayed in Figures 4.4e and 4.4f. It must be
noted that these regression coefficients were compiled from small scale hybrid rocket engine
tests. The regression rate law was used with the total mass flux G, which is why there is also a
dependency of the regression rate on the position along the fuel length. The magnitude of the
regression rate is well comparable to the results of the simulation with Karabeyoglu’s model.
4.2.6. Problems of the Regression Rate Models
The regression rate models implemented for classical and liquefying fuels all rely on a high
number of fuel input parameters like viscosity, surface tension etc. Additionally the model for
classical fuels by Zilliac and the liquefying fuel model by Karabeyoglu need data that could
only be modeled in a 2D or 3D discretized chamber like flame sheet temperature, heat transfers
between core flow, flame sheet, fuel vapor, fuel liquid layer and solid fuel. Simulation of such
values that are dependent on the boundary flow and flame sheet inside the boundary flow
needs a spatial discretization. With a 2D or 3D discretization however, it will be possible to
simulate the flames and calculate some values, that by now are user input: Flame temperature
Tc , local flame mixture ratio O/F c,x , boundary layer viscosity µ etc. Different oxidizers are
not considered in the models. The models are therefore only valid with the oxidizer which
they were compared to. In the presented cases this is oxygen. To some extent the choice of
oxidizer has also an impact on the models (e.g. gas flow density or viscosity, temperature)
but it needs to be carefully evaluated if the models actually are predicting realistic values.
The liquefying model shows a very high dependency on the mass flux density parameter G.
This results in an increasing regression rate downstream of the fuel. This effect is observable
in experimental investigations, but not as drastic as in the simulation results. All newly
implemented regression models also show a high dependency on the number of nodes. In
some cases, simulations with more nodes do not necessarily match experimental results better
than with fewer nodes. Concluding, the hybrid regression rate models in literature are not at a
point, where it is possible to claim that the processes in a hybrid rocket engine are completely
understood and modeled. There are still a lot of simplifications and assumptions. This is
why the regression rate models, which do not rely on empirical data for the regression rate
coefficients a and n, are only useful as a first approach. They cannot replace experimental
work for determining the regression rate of a hybrid rocket engine at this current state.
46
(a) Results for ṙ with Karabeyoglu’s model (b) Results for pressure with Karabeyoglu’s model
(c) Results for ṙ with a finer mesh (d) Results for pressure with a finer mesh
(e) Results for ṙ with with "stdHybrid" (f) Results for pressure with "stdHybrid"
Figure 4.4.: Regression rate and pressure of different models for liquefying fuels
47
4.3. ASTOS & ESPSS Interface
In order to link an EcosimPro/ESPSS model to the ASTOS "User Propulsion Interface", the
EcosimPro "deck" function is used. The concept of the deck is, that the supplier of an
EcosimPro model can create a deck and send it to his customer. The customer is only running
the deck by setting input variables and evaluating the output variables. He is not changing the
simulation model itself. The advantage of a deck is, that it needs no installation and license of
EcosimPro. It is also possible to run a deck directly from a C++ code. This interface option is
used to link the deck with ASTOS. With this it is possible to link any deck to ASTOS. Starting
from basic models, which consist only out of a combustion chamber and a few lines, up to
complex models, with tank pressurization or pumps, all kind of models can be used to link a
deck to ASTOS.
For optimization of a trajectory with a hybrid propulsion system the direct loop with ASTOS
and a ESPSS deck is not a viable. During optimization ASTOS’ algorithm changes many
parameters, controls and states at various points of the trajectory. This results in a very large
number of propulsion system calculations for a very short time step. The disadvantage is that
ESPSS is especially slow when called only for short integration intervals. In total the expected
optimization time with a direct loop of trajectory optimization with ASTOS and propulsion
system simulation with an ESPSS deck is extremely high. Therefore another approach is
chosen: ASTOS is using the analytical model of section 4.1.3 to simulate the initial guess
and afterwards optimize the trajectory. Then in a second step the thrust and mass flow of
the hybrid propulsion system is simulated with the respective ESPSS deck and the propulsion
system parameters and control gained by the optimization. Table 4.4 lists the inputs that are
passed from ASTOS to the deck. The simulation of the ESPSS deck will result in a curve for
the mass flow and the thrust that has higher precision due to the following points:
• Gradients on the thrust instead of instantaneous changes with control adaption
• Much more precise performance calculation
– Combustion properties like c∗ , cF or Tcc
– Thermal losses to the chamber wall
• 1D discretization
– Impact on gradients: Very long chamber will act differently than a short one
– More precise information about regression rates per node
• Lateral fuel regression possible
• Liquid oxidizer vaporization rate
• Different (initial) fuel diameters possible
• Multiple evaluation points (oxidizer dome, injector,pre- & post-combustion chamber) for
temperatures, pressures, massflows, heat flows etc..
48
With more complex ESPSS models the decks could also simulate:
• Tanks & pressurization systems (helium or other gases)
• Pumps (electrical or with gas generator/preburner)
• Clustered combustion chambers
• Effects of reaction control thrusters on the main engines
Since the thrust and mass flow curve of the ESPSS deck simulation is in most cases not
congruent another optimization process can be done afterwards. Optimization will again be
conducted with the analytical model, since the ESPSS model itself is not working with the
optimization of ASTOS. Therefore, correction factors for the vacuum specific impulse, oxidizer
and fuel mass flow mass flow for the average values are evaluated and the analytical model’s
results are corrected with these factors.
4.4. Software Verification
Both ASTOS (Analysis, Simulation and Trajectory Optimization Software for Space Applica-
tions) and ESPSS (European Space Propulsion Systems Simulation) are commercially available
tools. It is started from the premise that both software tools are verified for their field of
application. In case of ASTOS this means that the modeling of equations of motions, coordinate
system transformations and all other aspects coming with the software are not doubted in their
physical and mathematical correctness. However, to test the integrity of modeling trajectories
from Earth to Moon, a simulation and optimization of Apollo 11 and Luna 24 trajectories
is done in order to detect possible problems. For example, in earlier versions of ASTOS
problems with the coordinate system change from an Earth to a Moon centered orbit were
Table 4.4.: Input data for the ESPSS deck

Name Symb. Type Name Symb. Type
Burning time tb Param. Injector zeta coefficient ζ Const.
Chamber length Lc Param. Nozzle efficiency ηcF Const.
Chamber to throat diam. ratio Dc/Dt Param. Nozzle expansion ratio Param.
Combustion efficiency ηc ∗ Const. Nozzle length Ld Const.
Fuel density ρf Const. Nozzle ambient pressure pamb Param.
Fuel initial thickness tsp,ini Param. Nozzle throat diameter dt Param.
Fuel length lf u Param. Oxidizer inflow pressure Pin Const.
Fuel regression coefficient a Const. Oxidizer inflow temperature Tin Const.
Fuel regression coefficient n Const. Oxidizer mass flow ṁox Control
Igniter massflow mign Const. Working fluid Oxidizer
Injector area Ainj Const. Working fluid non-condens. Pressurization gas
Injector volume Vinj Const.
49
found. These problems were already fixed in a recent ASTOS update. For ESPSS it is assumed
that all aspects of modeling and solving the systems of equations are correctly implemented.
The verification for the user propulsion model that was implemented for ASTOS including
an interface to an ESPSS deck was done with simulation and optimization of a hybrid micro
launcher and a lunar lander module. No data was available for comparison but the main goal
was to verify the functionality of combined optimization of the trajectory and the propulsion
system parameters. Parts of this work were also published in [69].
4.4.1. Simulation of Moon Transfer and Landings
Apollo 11
In Chapter 2 the Apollo 11 mission was already described in some detail. Many information
about the ascent, orbit, mass balance and impulsive maneuvers of the Saturn V launch vehicle
as well as the Apollo 11 spacecraft are listed in "Apollo by the numbers"[10]. However not
all information is available in such a high detail that the trajectories could be reconstructed
without any free parameters. But even with some missing information the launch with
Saturn V, the translunar orbit as well as the landing of the Lunar Module of Apollo 11 could
be reconstructed with ASTOS. Table 4.5 shows some important parameters which were needed
to simulate the launch of Apollo 11 on the Saturn V. The first stage S-IC was burning for about
160 s. However, after 135 s the center engine was cut off and only four engines continued
burning. This is why in the simulation of the launch there are the two phases 3. and 4.,
which represent these two trajectory parts of the launch. In the same way the second stage
S-II had a burning phase with five engines and four engines (6. and 7.) too. The reason for
this is, that the acceleration needed to be reduced for the astronauts. If all five engines would
have continued to burn the acceleration would have been too high at the end of the burning
phase. The Saturn V rocket first launched the spacecraft into a low Earth orbit which is shown
in Figure 4.5a. The simulation and optimization could reach the target orbit with a similar
amount of used propellants like stated in [10]. The Saturn V launch was highly complex with
ullage motors for stage separation, ejection of the escape rocket, ejection of stage adapters and
others. Therefore it was not possible to reconstruct the launch in absolute detail in ASTOS
since precise information about every process was not available, but the result is adequate. In
the ASTOS simulations the used propellants for S-IC is ca. 2091.3 t, for S-II it is ca. 436.7 t and
for S-IVB it is ca. 32.6 t. For the actual launch of Saturn V the propellant usage for the first
stage was calculated as 2113.8 t, for the second stage as 436.7 t and for third stage as 31.4 t.
Therefore the simulation shows a very high agreement with the real launch.
After the ascent of Saturn V into Earth orbit, the third stage S-IVB stayed attached to the
spacecraft and after about 2 h30 min the S-IVB was ignited again to inject the spacecraft onto
its translunar orbit. In this phase of the simulation 71 450 kg of propellants were used, while
in reality about 71 050 kg were necessary. Afterwards the spacecraft was separated from the
rocket. The orbit transfer was a free-return trajectory. This means the spacecraft was aimed
50
Table 4.5.: Launch details for Apollo 11
Apollo 11 45 693 kg Launch date 16th July 1969 Launch time 13:32 UTC
Launch site longitude −80.6041° Launch site latitude 28.447°
Ascent phases 3. Gravity turn (5 eng.) 6. 2nd stage (5 eng.)

1. Vertical takeoff 4. Gravity turn (4 eng.) 7. 2nd stage (4 eng.)
2. Pitch over 5. Separation 8. Separation & 3rd stage burn
Target orbit
Inclination 32.55° Apogee 185.9 km Perigee 183.2 km
Launch vehicle information

Stage Empty mass Gross mass Propellant mass Vacuum thrust Isp,vac
S-IC (1st ) 138 713 kg 2 261 991 kg 2 123 278 kg 39.9 MN 304 s
S-II (2nd ) 39 856 kg 483 091 kg 443 235 kg 5.2 MN 421 s
S-IVB (3rd ) 11 943 kg 119 038 kg 107 095 kg 916 kN 421 s
to fly around the Moon in such a way, that after passing the Moon the spacecraft was again
on an Earth-centered orbit with a low perigee. This would allow a return of the crew in case
of an engine failure. The free return trajectory is also faster than a more propellant efficient
trajectory like a Hohmann Transfer, which takes about five days. The Apollo 11 spacecraft was
underway for three days until the lunar orbit insertion. In the simulation with ASTOS the free
trajectory return was reproduced. However some orbital element information was missing,
therefore it cannot be analyzed how close the simulation is in reference to the real mission.
The periselen at the lunar flyby is 100 km. After leaving the Moon’s sphere of influence the
apogee of the free return trajectory is about 550 Mm while the perigee is below 80 km, which
provides an atmospheric reentry without further orbit maneuvers. Figure 4.5b shows the free
return trajectory of Apollo 11 as simulated by ASTOS. The blue circle represents the Earth,
while the black curve is the orbit of the Apollo 11 spacecraft. The green dotted line represents
the Moon’s orbit around the Earth. The view is from above the Earth’s North pole. The Moon
is therefore moving from the upper side to the lower side. ASTOS’ ability to simulate and
optimize multibody orbits proves very useful to analyze this kind of mission. A bit more than
three days after lift-off the spacecraft of Apollo 11 was injected into a Moon orbit by a burn
of the AJ10-137 engine of the Command & Service Module. This burn took about 358 s in the
flight, while it takes 418 s in the optimized simulation. The reason for this is, that the throttle
control for this engine was given to the optimizer and it reduced the throttle in the beginning.
The exact throttle of the engine and amount of propellant in the real flight is not available
in literature. The total propellant consumption in the simulation is 11.6 t. The orbit after
insertion is defined with a periselen of 111.1 km and an aposelen of 314.3 km. However, the
inclination and argument of periapsis and right ascension node is not listed in the reference.
The inclination at the Moon in the simulation results in 174.3°, which is a retrograde orbit
near the equatorial plane of the Moon. After 4 h of orbiting, which represents roughly two
revolutions, the orbit was changed another time to circularize it further. In the simulation this
51
(a) Ascent trajectory of Apollo 11 on Saturn V (b) Free return trajectory
Figure 4.5.: Apollo 11 reconstruction results
burn takes 24 s, while it took 17 s in the actual mission. Again, this difference is caused by the
free throttle control in the simulation. The final orbit then is 100.9 km × 122.4 km. 688 kg of
propellant are used in the simulation of this circularization burn. The propellant usage of the
real spacecraft is not listed in [10]. After this burn the Lunar Module is decoupled from the
CSM and commences its descent to the surface.
In the description of the lunar descent it is reported, that the lander descended to about 750 m
above the surface, when a horizontal flight was initiated to find a better landing spot. The
exact details are not given, but the ASTOS simulation was set up in a way to represent this
horizontal flight. This final descent and the horizontal flight phase is shown in Figure 4.6.
Figure 4.7 illustrates the velocities during this final descent. Not enough data was available
to find out in which direction the final horizontal flight happened. Therefore it was simply
modeled that the spacecraft was flying to its final landing spot, which is on 0.674° North and
23.473° East. When reaching an altitude of about 750 m the radial velocity is reduced close
to zero and it is kept for about a minute. Then the lander descends to the surface and lands
with about 1 m/s. The propellant reserves after landing of Apollo 11 were about 350 kg, while
in the ASTOS simulation 500 kg remain. An explanation is, that the ASTOS simulation does
not account for reaction control thrusters. The optimization progress worked flawlessly to find
this trajectory. In sum ASTOS is a very powerful tool to reconstruct, simulate and optimize
translunar trajectories like the Apollo 11 mission. Even if not all details are reported for every
orbital maneuver, with the use of the optimization process, the trajectory and its missing data
can be found by fulfilling the boundary and path constraints.
52
Figure 4.6.: Descent trajectory of Apollo 11 Lunar Module close to the Moon’s surface in ASTOS
Figure 4.7.: Velocities of Apollo 11 Lunar Module during the final descent in ASTOS
Luna 24
The Luna landing and sample return missions are as well described in in Chapter 2. The
trajectory of Luna 24 was reconstructed with the help of the ASTOS software. In the first
ASTOS scenario the launch of Luna 24 with a Proton-K launch vehicle was simulated and
optimized to reach the correct orbits. Table 4.6 lists the necessary data for the ascent trajectory
optimization. All stages of the rocket used N2 O4 and UDMH as propellants. The information
has been gathered from different online sources [70, 71, 72]. The model of the launcher was set
up with this information. Details about aerodynamics of the launcher were missing. Therefore
a rough estimation of the drag coefficients was done. The aerodynamic losses are relatively
low. This is why the impact of the aerodynamic model is small. The simulation of the launch
was necessary, since detailed information about the transfer from Earth orbit to the Moon
53
were not available. By simulating the launch, the starting orbit for the transfer, which is the
Earth centered nearly circular orbit of 243 km × 188 km, could be determined more closely. The
missing orbital elements could be determined in this way.
Figure 4.8 shows the ascent trajectory on a 3D globe. The launch site Baikonur allows only
launches in certain corridors, which explains the high inclination. The trajectory of Luna
24 after the launch was reconstructed. For the simulation of the transfer to the Moon, the
missing orbital elements were determined and are listed in Table 4.6 too. These values do
not necessarily match with the real orbital elements of Luna 24, except for the known elements
inclination (51.5°), apogee (243 km), perigee (187.5 km) and the time of orbit insertion. But this
set of elements allows to further simulate the trajectory from a very realistic starting point. The
transfer trajectory with the burn of Blok-D upper stage as well as the course correction, orbit
injection and descent to the Moon’s surface was simulated in one single ASTOS scenario, which
is a great proof for the usefulness of this tool for such multi-body trajectory problems. The
Table 4.6.: Launch details for Luna 24

Luna 24 5795 kg Launch date 9th Aug. 1976 Launch time 15:04 UTC
Launch site longitude 63.34° Launch site latitude 45.92°
Ascent phase 2. Pitch over 4. 2nd stage burn

1. Vertical Takeoff 3. Gravity turn 5. 3rd stage burn
Missing orbital elements

Argument of perigee 85.0° Ascending node −163.2°
Mean anomaly −1.7° 3rd stage separation 15:13:30 UTC
Target orbit
Inclination 51.5° Apogee 243 km Perigee 188 km
Launch vehicle information

Stage Empty mass Gross mass Propellant mass Vacuum thrust Isp,vac
1st 31 100 kg 450 510 kg 419 410 kg 10 470 kN 316 s
2st 11 715 kg 167 828 kg 156 113 kg 2399 kN 327 s
3st 4185 kg 50 747 kg 46 562 kg 613.8 kN 325 s
Blok-D 2140 kg 17 360 kg 15 220 kg 83.6 kN 363.5 s
Figure 4.8.: Trajectory of Luna 24 on the Proton-K launch vehicle in ASTOS
54
transfer was divided in the following phases: 1. Coast on the original Earth-centered orbit to
reach the best point for the impulsive maneuver of Blok-D; 2. Burn of Blok-D and Separation;
3. Coasting Phase; 4. Course Correction; 5. Coast to the Moon’s vicinity (Sphere of Influence
is about 60 000 km); 6. Coast (from now on all phases are simulated Moon-centered); 7. Moon
orbit injection burn; 8. Orbit lowering burn; 9. Descent coast; 10. Deceleration Burn; 11. Final
vertical descent coast; 12. Soft landing phase. Because no exact information about phasing
orbits around the Moon was given, the orbit injection to a 115 km circular orbit is directly
followed by the orbit change to a 12 km × 120 km orbit. In the real mission there were phasing
orbits for several days to reach the final landing destination. The orbit injection at the Moon
was on 14th August 1976 while the landing was on 18th August. This was neglected in the
simulation, but it does not affect the necessary ∆V . The landing site however is not the correct
one in the simulation. The orbit and spacecraft data that was found in the before mentioned
online sources are:
• Initial lunar orbit: 115 km circular with 120° inclination
• Final lunar orbit: 12 km × 120 km with 120° inclination
• Deorbit and soft touch down: 6 min duration and about 3 m/s touch down
speed
• Luna 24 spacecraft mass: 5795 kg including 3606 kg propellant
The Luna 24 spacecraft also contains the return rocket with 515 kg.
Figure 4.9a shows the altitude of the Luna 24 spacecraft in the ASTOS simulation after
optimization. The optimization was done with regard to all known orbit elements as
constraints and in addition the used fuel mass was optimized, to have more reserve propellant.
The vertical lines show the phase borders. After one day the course correction is simulated.
The course correction is not necessary in the optimization within ASTOS, as the simulation and
optimizer always reach the optimum attitude and impulse. However, to meet the propellant
consumption better, a course correction phase was included, and the results showed, that with
this course correction phase, the optimizer was actually able to find the optimal result faster.
The change from a Earth-centered coordinate system to a Moon-centered one is visible after
about 3.5 days. Figure 4.9b enlarges the altitude curve in the last 60 min after surpassing
200 km. The orbit injection burn begins already at about 180 km, due to the high radial
velocity in the hyperbolical orbit. The margins for the correct burns are quite small. Start,
end and total impulse of the burn need to be very precise in such a mission. After the 280 s
long orbit injection burn, the impulsive maneuver to lower the perigee to 12 km requires only
a short burn duration of 10 s. Then the coast phase follows over half a revolution around the
moon. Afterwards the final deceleration burn lowers the velocity close to standing still. This
is illustrated in Figure 4.9c. The flight path speed of the Luna 24 spacecraft when reaching the
Moon’s vicinity is about 2.4 km/s, which means an excess velocity of 0.9 km/s in the hyperbolic
trajectory. The speed after the orbit injection is 1.6 km/s. The plotted east velocity is relative to
the Moon’s surface. The final soft landing is visible nearly 60 min after getting below 200 km
55
altitude. First, the orbital velocity of the 12 km × 120 km is reduced to a standstill relative
to the Moon’s surface. The real Luna 24 mission included a similar approach, although the
precise data was not available. The standstill in the optimization is at 4.7 km altitude. It could
be lowered, if the KTDU-417 engine would be throttled down. Then a thrust-less descent
is following, until at about 1 km above the surface the engine KTDU-417 is ignited again to
slow down the spacecraft, since after the standstill the Moon’s gravity has again accelerated
the spacecraft to 110 m/s vertical descent velocity. The secondary engine KTDU-417B has just
enough thrust to keep the vertical velocity at a constant value. It was only ignited few meters
above the surface after KTDU-417 was shut down. In this optimization this was neglected
and the KTDU-417 shuts down only few meters above the ground when the velocity of the
spacecraft is below 4 m/s. Figure 4.9d shows the last 10 minutes, when the spacecraft descends
below 20 km. While Luna 24 back in 1976 did a full automatic landing without analysis of
the landing site, which was probably the reason for the failed landing of Luna 18, today a
hovering at some kilometer altitude would also allow to pick a suitable landing site in the
close area. The Luna 24 mission could be reconstructed with ASTOS from lift-off in Baikonur
to the landing on the Moon surface. Even if some orbital parameters and information was
missing, the reconstructed trajectory seems very plausible.
(a) Altitude (b) Altitude close to the Moon
(c) Speed close to the Moon (d) Speed and altitude of the final descent
Figure 4.9.: Luna 24 spacecraft data
56
4.4.2. Combined Optimization of Trajectory and Propulsion
Optimization Process
The optimization of the following two test cases, a hybrid micro launcher and a lunar lander,
was done in these steps[69]:
1. Creation of an initial guess: For the initial guess with ASTOS all mission phases, vehicle
parameters like propellant masses or engine geometries need to be set. The initial guess
for the trajectory should usually be already very close to the optimum result, in order to
increase convergence of the gradient based CAMTOS optimization. Therefore it is quite
time consuming to find a good initial control law for the attitude control and adapt all
phase lengths to create a good initial guess. Boundaries need to be set for all optimizable
parameters and controls. For the optimization in the beginning it is helpful to have rather
close bounds for some parameters and controls.
2. Optimization of the trajectory: The first optimization has the goal to calculate a trajectory
which meets the final constraints and requirements. The process is faster, if cost
functions or constraints for optimizing the total vehicle or payload mass are inactive
at first. It also is advisable to have wider path constraints in the beginning and slowly
narrow them down with each optimization run.
3. Optimization of initial mass or payload mass: When the trajectory meets all constraints,
the cost term for minimum initial mass or maximum payload mass is activated. Some
optimization runs will be necessary to find the optimum. It is possible that some
parameters or controls run against the boundaries that were initially set closer to
accelerate the optimization process. Now they can be loosened.
4. Simulation with ESPSS: When the trajectory and masses are optimized, a simulation with
the ESPSS interface is done in order to get the values of the correction factors.
5. Repetition: This process is iterative and is repeated until it converges and a new ESPSS
simulation does not change the values of the correction factors anymore.
Hybrid Micro Launcher
Model Set-Up In order to test the implemented ASTOS analytical tool and ESPSS interface a
hybrid propulsion micro launcher was designed in a first draft. The market for small launch
vehicles is potentially growing a lot in the next years. The number of launched satellites per
year is growing continuously and large satellite constellations are announced like "OneWeb".
These constellations pose specific challenges: the satellites are rather small (<500 kg) and have
many different orbits. This means launches with multi-payload adapters could be suboptimal
for satellite lifetime, as the satellite needs more propellant to reach its final orbit. Also a
constellation of hundreds of satellites raises the question of replacing satellites which fail or
reach end-of-life. Here is where the small launch vehicles possibly have an advantage over
57
heavy lift launchers. Therefore, the interest in this kind of flexible launch vehicle has grown
worldwide. According to the definition for micro satellites, which have a mass of 10 to 100 kg,
a micro launcher based on hybrid rocket engines, with a payload of 25 kg to a 250 km circular
polar orbit was defined, simulated and optimized. In contrast to liquid propulsion rockets,
hybrid rockets are always limited in their burn duration. For longer burning times, hybrid
rocket engines simply get too large. Hence, a three-stage concept was initiated. Table 4.7
shows the trajectory phases of the optimized mission. The first stage burns during the first four
phases, after which the stage separation is conducted. While setting up the initial guess, first
simulations made clear, that a relatively large first stage causes the rocket to raise its apogee
in the beginning very quickly. Therefore, the vehicle’s apogee is already close to the final orbit
altitude right after the burn of the first stage. Hence, after depletion of the first stage, an ascent
coast is inserted. The second stage is only ignited in high altitude, to raise the perigee. Another
coast phase is included between the second and third stages’ burn. The third stage circularizes
the orbit. The arrow symbol ”→” indicates the change from the initial guess to the optimized
parameter value. For a successful optimization process, several boundary constraints are set in
the ASTOS model. Table 4.8 presents a list of the active constraints in this model. None of the
mentioned boundary constraints are unique to a hybrid rocket. These constraints are rather
the minimum set necessary to optimize a three stage launch vehicle to a circular, polar orbit.
Table 4.9 shows the initial and optimized mass parameters for the micro launcher concept.
Only rough estimations were done for the structural masses. The structural mass includes
also masses of GNC electronics & instruments, batteries etc. Even if the total launcher mass
changed, these structural masses for all three stages were not optimized, as it is not of high
importance in order to demonstrate that the hybrid propulsion tools are working as expected.
Additionally, the aerodynamics of the launcher were also very rough estimations. In Table
4.10 the initial and optimized engine parameters are listed. The propulsion system is designed
to use paraffin-based fuel and H2 O2 as an oxidizer. The first stage uses seven parallel rocket
engines fed from one oxidizer tank, while the second stage uses four parallel engines. The
upper stage uses a single engine. The optimizer can also change the throttle of all three stages.
Table 4.7.: Trajectory phases[69]

Name Description Burning Duration
1. Lift-Off Vertical take-off 1st stage 0.2 s → 0.1 s

2. Pitch Over Constant pitch rate 1st stage 11.8 s → 4.1 s
3. Pitch Constant Transition to gravity turn 1st stage 6s → 3.3 s
4. Gravity Turn Angle of attack is zero 1st stage 47 s → 42.5 s
5. Ascent Ascent with stage & fairing separation 130 s → 100.2 s
6. Stage 2 burn Orbit raising 2nd stage 35 s → 43.8 s
7. Coast Coast to apogee & stage seperation 20 s → 34.6 s
8. Stage 3 burn Final orbit raising 3rf stage 65 s → 98.8 s
58
Table 4.8.: Optimization constraints for a micro launch vehicle[69]
Type Constraint Bounds Active in Phase
Initial Initial position 0 m, 68 ◦ N, 21 ◦ E 1

Initial Initial radial, north and relative east velocity 0 m/s(landed) 1
Path Pitch and yaw rate ±2.5 °/s 2,3,6,8
Path Gravity turn cond. ±0.3 ◦ 4
Final Maximum orbital inclination 95 ◦ 4,6,8
Final Residual propellant in all stages 0-10 kg 4,6,8
Path Maximum acceleration 130 m/ s2 8
Final Apogee 250 ± 5 km 8
Final Altitude 250 ± 5 km 8
Final Perigee 250 ± 5 km 8
Final Orbit inclination 94.5 ± 0.5 ◦ 8
Table 4.9.: Initial and optimized micro launcher masses[69]

Stage Component Optimizable Initial mass Optimized
1st Propellant Yes 13400 kg → 8268.3 kg
Tank structure Fixed ratio 7.53 % of oxidizer mass
Engine structure Yes 620 kg → 400 kg
Additional Structures No 310 kg
Total stage mass 15232 kg → 9548.9 kg
2nd Propellant Yes 1525 kg → 1538.2 kg

Total stage mass 1883.2 kg → 1881.05 kg
3rd Propellant Yes 132 kg → 179.1 kg

Engine structure Yes 13.5 kg → 15.5 kg
Total stage mass 216.9 kg → 270.1 kg
Payload No 25 kg
Fairing No 10 kg
Total lift off mass 17332 kg → 11700 kg
Total prop. mass 15057 kg → 9985.6 kg
59
The nominal thrust, nominal mean mixture ratio and the burn time per engine are listed in the
table. Table 4.11 shows the optimizable parameters that define the hybrid rocket engines of
the three stages.
Optimization Results The initial guess of the trajectory did not reach a circular orbit. The
initial trajectory was a suborbital trajectory with an apogee at 420 km and a perigee at −490 km.
The optimization goal was therefore to reach the defined final orbit and at the same time reduce
the launcher mass. Some of the optimization results are already shown in the tables 4.7, 4.9,
4.10 and 4.11. To reach the final orbit, the burn phases of the second and third stages were
prolonged, while the first stage’s burn duration was reduced by 15 seconds. The maximal
thrust level of the first and second stage was reduced, due to the lowered total mass. The
total launcher mass was reduced by 5632 kg (32.5 %). Most of this was saved in propellant
mass, where 5071.4 kg (33.7 %) was saved. This was reached by optimizing the mixture ratio
O/F and optimizing the expansion ratio as well as the chamber pressure. The expansion ratios
however were limited by the stage diameters. There is one limitation in the current model: The
optimizer increased the chamber pressure by reducing the throat diameter in all engines, see
Table 4.11. This increases the vacuum impulse and decreases engine mass, because the nozzle
is getting smaller. However, this requires a higher tank pressure or a pump, which is currently
Table 4.10.: Initial and optimized micro launcher propulsion system[69]

Stage Engine thrust Mixture ratio Burning time
1st 73 kN → 61 kN 7.6 → 9.0 65 s → 50 s

2nd 30 kN → 28 kN 7.6 → 7.4 35 s → 43.8 s
3rd 5 kN → 6 kN 7.6 → 7.6 65 s → 98.8 s
Table 4.11.: Geometrical engine parameters[69]

Stage Component Initial value Optimized value
1st Fuel grain length 1.6 m → 1.25 mm

Fuel grain diameter 480 mm → 415 mm
Nozzle throat diameter 139 mm → 135 mm
Nozzle expansion ratio 6 → 4
2nd Fuel grain length 965 mm → 915 mm

Fuel grain outer diameter 300 mm → 285 mm
3rd Fuel grain length 412 mm → 412 mm

Fuel grain outer diameter 270 mm → 274 mm
60
not taken into account for the optimization’s mass balance. This is a necessary step to improve
the model. However, the first optimization results still are very promising. In total, the first
stage was reduced most in size, while the propellant of stage two and three was increased.
This is expected, as the first stage has the lowest specific impulse due to the limited expansion
ratio. Additionally a larger first stage increases aerodynamic and gravitational losses. The
optimization of the model was successful and it is a proof that the combined optimization
of the trajectory and the engine with the analytical model works as expected. The thrust of
the first stage is throttled down 20 seconds after lift-off. Figure 4.10a displays exemplarily
the optimized acceleration, thrust and oxidizer mass flow of the second stage during its burn.
The mass flow is for a single engine. The oxidizer mass flow, which is the controlled input of
the engine model, is throttled down after optimization during the beginning and in the end
of the phase. While for the throttling down at the end the cause is quite obvious, the reason
for the small down throttling at the beginning is not clear. There are many possible reasons.
The optimization is a multidimensional problem, and it is not always easy to understand
the mechanisms that prefer one solution over another. This is also one reason for using
numerical optimization, because this kind of problem cannot be solved analytically. A possible
explanation for the reduced thrust in the first stage and in the beginning of the second stage’s
burn is that at this part of the trajectory gravitational losses are higher. At the end of the second
burn the vehicle’s acceleration increases as the mass is reduced every second. Therefore the
acceleration reaches the limit of 130 m/s2 and the four engines of the second stage are throttled
down. The acceleration at the phase’s end is approximately 130 m/s2 . The limit of 130 m/s2
was chosen to demonstrate that the path constraints work. The limit is still a bit too high
for real applications. For future optimizations an applicable limit according to the payload’s
requirements will be used. The regression rates and mixture ratios of the three stages’ engines
are shown in Figure 4.10b. The regression rate of stage 1 and 2 are in a similar range, only the
regression rate of the third stage is lower. This is due to the lower mass flux G in this engine,
which is also necessary to sustain longer burn times. The mixture ratios of stage one and two
show, that the throttling has a major effect on the mixture ratio. The mixture ratio drops by
about 10 %. For the propellant combination H2 O2 and paraffin-based fuel the gradients of the
function of specific impulse over mixture ratio are quite small. Therefore this oxidizer could
be especially useful for throttling engines. The loss in specific impulse in this case is only
1 % from 2900 m/s to 2870 m/s. Figure 4.10c displays some trajectory parameters over time to
visualize the optimized ascent to the low Earth orbit. Figure 4.10d illustrates the layout of the
micro launcher. It also shows the proportions of stages, engines and tanks. The base diameter
is 1.5 m without fins and the length is 12.5 m. Table 4.12 lists the applied correction factors
determined with the ESPSS interface [69]. The pitch and yaw of the launcher is displayed in
Figure 4.11a. It is visible that in the third stage’s burn there is a great change from lower pitch
and yaw values to higher values and then it levels out at a medium value. From experience this
behavior is an indication that the density of control points in this phase is not large enough.
More control points can be added to improve this. The specific impulse curves in Figure 4.11b
show the increase of exit velocity with increasing expansion ratio in every stage. For the second
61
Table 4.12.: Optimization correction factors[69]
Stage Oxidizer mass flow Fuel mass flow Specific impulse
1st 99.30 % 98.97 % 103.16 %

2nd 99.91 % 99.36 % 100.83 %
3rd 100.31 % 99.67 % 108.8 %
and third stage the specific impulse of ambient and vacuum conditions is identical, therefore
the vacuum impulse for these stage is not plotted. Figure 4.11c shows the mass balance of the
stages and propellants. A three-dimensional depiction of the launcher’s trajectory is shown
in Figure 4.11d. The line’s color shows the acceleration. The blue zones are the coast phases,
where only Earth’s gravity is exerting influence. Due to the three stage set-up, the ascent is
quite steep and only the second and third stages increase the horizontal velocity.
Lunar Lander
Similarly to the micro launcher optimization a hybrid propulsion lunar lander has been
optimized. The results have always been used for the optimizations in Chapter 6. The
optimization of this lunar lander is described in Appendix D.
4.4.3. Simulation of a Hybrid Rocket Test
Simulation Model
To evaluate the implemented models in the EcosimPro library ESPSS the test results from test
bench M11.3 at the DLR Lampoldshausen were used to compare test and simulation data. Test
No. 3 and test No. 4 which are later described in Section 5.2.1 have been used to compare
to the simulations. Figure 4.12 shows the ESPSS schematic of the test bench simulation. All
parts of the test bench have been modelled with measurements and data sheets of the tubing,
valves, coriolis sensor and the orifices. The test bench is divided into three lines: the oxygen
supply line, which includes an inflow valve (311), a release valve (313), a coriolis flow meter,
which was modelled as a bended tube, a control valve (RV01), which has the smallest flow
diameter in the system, and the main valve (312), which is opened at test begin. Additionally
there is a nitrogen line for purging the lines and the hybrid rocket engine. Finally there is the
line from the main valve to the engine, which includes several pipes, a measuring probe and
the engine. The subsonic part of the combustion chamber was discretized with nine nodes,
of which nodes three to five contain the cylindrical fuel. The measurement of the coriolis
mass flow meter is simulated with an integration of the measured signal. The coriolis mass
flow meter is averaging the mass flow measurement and therefore does not show peaks and
instabilities. To compare the simulation and the measurement, a similar integration procedure
is implemented in the ESPSS model and schematic. The exact formulation to reproduce the
62
(a) Acceleration, thrust and oxidizer mass flow during second stage burn (b) Regression rates and mixture ratios of the three stages[69]
63
(c) Altitude, apogee, perigee and flightpath speed[69] (d) A preliminary design of a three stage micro launcher[69]
Figure 4.10.: Hybrid micro launcher data, part 1

(a) Yaw and pitch angle over time (b) Specific impulse of the micro launcher stages
64
(c) Total and propellant mass of the micro launcher over time (d) 3D-trajectory colorized with acceleration data
Figure 4.11.: Hybrid micro launcher data, part 2

Figure 4.12.: Schematic of ESPSS model of the M11.3 test bench
coriolis measurement’s signal is not known. Figure 4.13 shows a technical drawing of the
combustion chamber assembly with the simulation nodes’ positions. The simulation model
outline is also marked with violet color. This is valid only for tests No. 1 to 4 without fuel
grain protection on the lateral regression surfaces. The part numbers 1, 2, 5 and 12 mark
the aluminum housing and numbers 4, 6, 7, 8 and 10 are showing insulation parts. Number
9 marks the graphite nozzle insert, while number 11 highlights the solid fuel block. For a
simulation with lateral combustion turned on, the node before the fuel block and after the
fuel block actually needs to be set to the fuel’s outer diameter. Therefore two additional nodes
without fuel have been inserted each at the front and the backside of the fuel block. If this is
not done, the lateral combustion surface is not correctly calculated, as the node’s diameter is
always evaluated in the middle of the node. When first simulations were done with the model,
the following points were identified:
• The pressure drop over the combustion chamber’s injector naturally has a high impact
on the simulations result. The pressure drop in ESPSS is modelled with a pressure loss
coefficient ζ. This coefficient is dependent on the injector’s geometry but also on pressure
ratio, Reynolds number and Mach number. Therefore, for a stationary flow it is a constant
Figure 4.13.: Updated schematic of ESPSS model of the M11.3 test chamber
65
but needs to be adjusted for every test. In case of throttling it even needs to be adjusted
for every flow condition. For future analyses a study of the pressure loss coefficient for
several possible injector conditions could be done.
• The pressure in the chamber was higher in test measurements than in the first simulation
runs, even though the mass flows and mixture ratio was matched. The reason was
identified in the nozzle shape. The used nozzle is conic and has no ideally shaped nozzle
throat but a sharp edge. Therefore the flow in the throat separates from the surface
and forms a constriction of the flow. The effective nozzle diameter therefore is smaller
than the geometrical measure. A nozzle discharge coefficient of roughly 95 % has been
introduced to match the pressures in the combustion chamber.
Test No. 3 has been simulated with the following options: first the "stdHybrid" regression rate
model was used. The regression coefficient was already determined in previous experiments
as asp,0 = 0.157 mm/s and b = 0.58. The model has been set to use lateral combustion. The front
and backside of the fuel grain in this chamber configuration with a relatively large diameter
to length ratio add a crucial part to the total surface area of the fuel. Since the front and
backside have a slower regression rate, the lateral combustion factor was set to 70 % for test
No. 3 and 40 % for test No. 4. This factor is dependent on many factors like ratio of inside
and outside diameter of the fuel, mass flux density and flow conditions in the chamber. For
example a recirculation zone will cause a different regression rate than if the oxidizer is directly
injected onto the lateral fuel surface. Therefore this factor will be empirically determined
with more tests. The oxidizer mass flux Gox as a reference for the regression rate produces a
constant regression rate along the x-axis. The test results quickly showed, that this is not a good
approximation for the set-up, as the regression rate along the x-axis is highly increasing with
distance from the injector plate. The reason for this is assumed in the injection behaviour and
the large diameter to length ratio. Therefore the total mass flux G was used as a reference. After
this change, the coefficient asp,0 = 0.157 mm/s was too large, as it was gained from tests with a
different combustion chamber size as well as using the Gox and therefore needed scaling. It
was multiplied with a factor of 0.86, which provided very good results for test No. 3. A second
simulation has been done with the regression rate model for liquefying fuels, see Section 3.7.2.
Table 4.13 shows the valve opening percentages during the test sequence. The openings are
Table 4.13.: Valve sequence for the simulation of test No. 3

Time Inflow 311 Release 313 Control RV01 Main 312 Nitrogen
0.0 s 100 % 0% 17.6 % 0% 0%
0.1 s 100 % 0% 17.6 % 100 % 0%
15 s 100 % 0% 17.6 % 100 % 0%
15.1 s 0% 0% 17.6 % 0% 0%
17.0 s 0% 0% 17.6 % 0% 0%
17.1 s 0% 100 % 17.6 % 0% 100 %
66
linearly interpolated between the table values. The most important parameters of the test
bench simulation are listed in Table 4.14. The pipes between the inflow and main valve have
a flow diameter of 18 mm and a summed up length of about 3.2 m. Due to the large diameter
they do not add a lot of pressure losses. Behind the main valve 26 cm of 8 mm diameter pipe
and 80 cm of 6 mm diameter pipe are following. Especially the 6 mm pipe adds some pressure
losses. Then a 15 cm long 12 mm diameter pipe, which has the connections for the injector
pressure sensor and the injector thermocouple is connecting the test bench to the rocket engine
demonstrator. The pressure loss coefficient ζ for the valves and orifices that are not explicitly
mentioned in Table 4.14 were automatically calculated by ESPSS. Experience shows, that this
can only be used if the orifice or valve is having a relatively low pressure loss or flow diameter
ratio. The simulation model of the control valve RV01 needs to have a choked flow, meaning
that the orifice flow velocity is limited by the sonic velocity. This is only possible with the
ESPSS option for the valve called "CV_option" being true.
Simulation Results
Figure 4.15a shows the pressure measurements of test No. 3 in comparison to the simulation
results. Three positions have been compared: downstream of the control valve, upstream of
the injector and inside the combustion chamber. The pressure is in very good agreement at
all positions over the whole test sequence. Figure 4.15b illustrates the mass flow curves of
the test measurement and simulation. The mass flow in the test set-up is only measured at
the coriolis sensor. The coriolis sensor does not measure the instantaneous mass flow but an
averaged resp. integrated mass flow. This can be seen in the slow rise of the measured mass
flow in the beginning. The measurement needs 2.5 s for the mass flow to rise to the steady
state. The simulated mass flow is plotted both for the coriolis sensor’s position and the injector.
The simulated mass flow at the coriolis sensor immediately rises to its maximum limited only
by the sonic condition in the control valve’s orifice. In the schematic of this simulation an
integrator is implemented. The integrated mass flow in the simulation is also plotted and
it has a much better congruence with the measured signal than the other simulated curves.
However, there is still a difference in time but the curves are nearly parallel in the beginning
Table 4.14.: Important test bench simulation parameters

Parameter Value Parameter Value
Inflow valve A0 20 mm2 Length of combustion chamber 353 mm
Control valve A0 16 mm2 Length of solid fuel 150 mm
Main valve A0 20 mm2 Thickness of solid fuel 51.6 mm
Main valve ζ 2.4 Nozzle throat diameter 5.44 mm
Hand lever A0 18 mm2 Injector area 3.14 mm2
Pipe rugosity 15 µm Injector ζ 3.3
67
of the test. It is undocumented why the coriolis sensor is producing such a curve. For more
precise information the coriolis needs to be dismounted and tested in a separate set-up. This
is currently impossible, as the test bench is also used for other test campaigns. The good
agreement of the pressure curves at all three compared positions however suggests that the
mass flow of the simulation is also close to the actual mass flow in the test bench set-up. The
length and inner diameter of the fuel grain is depicted in Figure 4.15c. The measurement of
the experiment, which were done before and after the test are plotted as points. There is a
very good agreement in the length, which is due to the factor of 70 % which was imposed on
the regression rate for the lateral combustion for test No. 3. The agreement of the inner fuel
diameter is in good agreement as well. The diameter at the front and back side is evaluated
at the node’s center in the simulation, while on the real hardware it is manually measured at
the outside. Therefore the measurement and simulation value have a higher discrepancy. The
fuel consumption of experiment and simulation is in quite good agreement: In the experiment
455.3 g of paraffin-based fuel were used, while in the simulation 466.4 g were burned, which
represents a small deviation of 2.4 %. Figure 4.15d shows the regression rates. The regression
rate of the experiment is averaged in time. Therefore the points were plotted in the mid
of the burn time. The average point has been calculated with the total burned mass and
the assumption of a cylindrical shape of the fuel port. The regression rate at the injector’s
and nozzle’s side was measured directly, as well as the lateral regression, which however is
averaged as the lateral regression is not homogeneous on the whole fuel grain’s front and back
side either. The simulated regression rate matches the measured mean values quite well. The
before explained difference between the node’s diameter and the front and back side diameter
also explains the larger differences for the injector’s and nozzle’s side’s regression rate. Overall
the regression rate of the cylindrical part is a bit too low in the simulation, which explains also
that the used fuel mass is a bit lower. The regression coefficients asp,0 = 0.157 mm/s and b = 0.58
can be improved with more test results. Figure 4.14a shows the simulated characteristic
velocity c∗ . The mixture ratio which is gained in the simulation is shown in Figure 4.14b.
The mixture ratio is shown for all three fuel nodes. Naturally the mixture ratio is decreasing
along the fuel grain. The shift in the mixture ratio is visible, but not very large. However, the
mixture ratio shift, including the gradient in the startup of the engine, makes it difficult to
evaluate the characteristic velocity and the combustion efficiency from the test results alone.
The simulation and comparison with test results improves understanding of the effects that
influence the mixture ratio, characteristic velocity and efficiency. Test No. 3 was also simulated
using the regression rate model for liquefying fuels. However, the simulation results have no
good agreement with the experiments. The simulated fuel mass flow is only about half of the
mass flow in the experiment. The reason for this is, that the simulation with the liquefying
model does not meet the conditions for entrainment according to Equation 3.10, which reveals
that the equation is not valid in all operational ranges of hybrid rocket engines using paraffin-
based fuel. In the experiment entrainment is increasing the fuel mass flow. Tests with this
experimental combustion chamber are in a very low range of mass flux density G. Therefore
the simulation with the liquefying model is out of its intended range of mass flux conditions
68
(a) Simulation characteristic velocity c∗ (b) Simulation mixture ratio
Figure 4.14.: Test No. 3 simulation results
and the results are not compared in more detail. Test No. 4 is also compared to the simulation
results with the regression rate model stdHybrid. The main difference between test No. 3 and
4 is the mass flow of the oxidizer. The pressure before the control valve was set to 93 bar and
this increased the oxidizer mass flow to 50 g/s. Figure 4.16a shows the pressure curves of test
No. 4. The agreement between simulation and experiment was reached very quickly. Test
No. 4 had a higher regression rate than expected. The reason for this can be a manufacturing
problem of the paraffin fuel. Therefore the regression rate coefficient asp,0 = 0.157 mm/s was
increased by 6 % in order to match the fuel consumption of the experiment. By this, the
pressure curves were automatically matching to each other. The mixture ratio is again shown in
Figure 4.16b. It is very similar to test No. 3. However, this is only due to the higher regression
rate. A higher mixture ratio was expected with a higher oxidizer mass flow. Figure 4.16c
illustrates the fuel geometry over time. The lateral combustion factor was set to 0.4 for test
No. 4. It is assumed that the lateral combustion factor is a coefficient which is very depending
on fuel geometry, mass flux density and possibly other factors. Therefore for every test the
lateral consumption factor needs to be adjusted. The diameter was in quite good agreement.
The regression rates in 4.16d are in a less good agreement. Especially the evaluated mean
regression rate of the experiment is much higher than the simulated regression rates. It is
possible that this is also caused by problems with the manufactured fuel grain. The density
could be inhomogeneous, which makes the evaluation of the data affected with errors. In
total the simulation of hybrid rocket engines with EcosimPro & ESPSS works very well. Some
coefficients like the valve’s pressure loss coefficients and the parameters for regression rates of
solid fuels in hybrid combustors need to be evaluated with test results. Simulations in order
to predict the performance of a test set-up or new hybrid combustor are always relying on this
data and only are precise if all parameters and coefficients are well known. As a conclusion
the tests could be very well reconstructed in the simulation with ESPSS. In order to find the
right regression rate coefficients a large amount of tests needs to be done and evaluated in the
future.
69
(a) Pressure (b) Mass flow
70
(c) Fuel geometry (d) Regression rate
Figure 4.15.: Comparison of Experiment and Simulation: Test No. 3 simulation results
(a) Pressure (b) Mixture ratio
71
(c) Fuel geometry (d) Regression rate
Figure 4.16.: Comparison of Experiment and Simulation: Test No. 4 simulation results
5. Experiments with a Hybrid Rocket Engine
Demonstrator
5.1. Test Bench Set-up
Existing test bench infrastructure at the test complex M11 of the DLR Institute of Space
Propulsion in Lampoldshausen was used for a small experimental campaign. The experiments
were conducted at the test position M11.3. Previously this test bench was used for works
with optical analysis of hybrid rocket engine combustion as well as for analyzing different fuel
compositions with a small radial combustor [27, 30, 47]. The test bench was expanded with
a new hybrid rocket engine, which sustains long burning durations at relatively low oxidizer
mass flow and thrust. The flow plan of the test bench M11.3 can be seen in Figure 5.1. The
oxidizer is gaseous oxygen which is supplied from a large high pressure reservoir, where the
pressure is considered steady during the test duration. The oxidizer for the combustion is
fed through a line with a pressure regulator PV11, the reservoir valve 0.311, a control valve
RV01 and the main valve 0.312. A coriolis mass flow sensor is positioned between valve 0.311
and the control valve RV01. A second gas line is supplying nitrogen for cleaning and purging
the lines with an additional valve 0.314. A further set of lines is provided for a hydrogen
and oxygen torch, which is used for the ignition of the engine. The test bench is equipped
with temperature and pressure sensors for each line segment between valves. A coriolis mass
flowmeter is installed between valve 0.311 and the control valve RV01. Table 5.1 lists the
sensor positions, which were installed to measure the pressure, mass flow and temperature
at several positions along the experimental set-up. The fuel usage inside the combustor is
not monitored. The fuel grain is measured before and after the test to get average regression
rates and the average fuel mass flow. For the purpose of testing long burning times with a
hybrid combustor a new combustion chamber was designed, constructed and built [73]. The
chamber is shown in a technical drawing in Figure 5.2. The numbers represent the parts listed
in Table 5.2. The combustion chamber is an engineering model design with a very thick wall.
Mechanical loads therefore are no issue for the chamber wall. Since the engine was designed
for long burn durations the whole chamber is protected with phenolic composite parts on the
inside. The surface of the phenolic composite pyrolyzes very slowly and the heat conductivity
of the material is very low, so that the temperature of the metal parts are low for several
minutes of burn time. The nozzle is fixed between two tubes of phenolic composite. These
two tube sections can be modified in their length and by this the length of the combustion
chamber can be adjusted. This is especially useful for adapting the combustion chamber to
different hybrid solid fuels. Each fuel has a unique regression rate behavior. One way to
72
Figure 5.1.: Flow plan of the M11.3 test bench
Table 5.1.: Sensor positions

Measured value Position Symbol Sample Rate
Pressure Control valve exit pRV 20 kHz

Pressure Injector cavity pinj 20 kHz
Pressure Combustion chamber pcc 20 kHz
Mass flow Before control valve inlet ṁox 10 Hz
Temperature Ambience Tamb 10 Hz
Temperature Control valve exit TRV 01 10 Hz
Temperature Coriolis TCor 10 Hz
Temperature Oxygen inflow TOx,in 10 Hz
Temperature Chamber wall Twall 10 Hz
73
modify the mixture ratio inside the chamber is adjusting the length of the fuel grain, increasing
or decreasing the fuel’s burning surface. Hence the combustion chamber length needs to be
modified as well. The nozzle throat is produced from graphite to sustain the high heat loads
at the throat. The used graphite has nearly no degradation over a test, which was verified in
earlier test campaigns with other combustion chambers. The injector plate is produced from
brass in order to avoid melting or burning metal in case of back flows or recirculations during
the transient phases in the test. Brass has a high heat capacity and high heat conductivity
which reduces the risk of local hot spots. The injector has a showerhead design with only four
holes. The injector plate does not need a very good spraying characteristic as the oxidizer is
already gaseous. The only purpose of the injector plate is to generate a high enough pressure
drop to decouple oxidizer feed line pressure from the combustion chamber pressure in case
pressure oscillations occur. In hindsight the injector could have been modified to improve the
distribution of oxygen on the fuel port. The engine is ignited with a hydrogen/oxygen torch
which is not shown in the drawing. The hot gas flow from the torch enters the pre-combustion
chamber at an angle of 45° to the oxidizer inflow. The driving design parameters of the engine
are listed in Table 5.3. Thrust was not a design factor therefore the nozzle and its expansion
ratio are neglectable.
5.1.1. Paraffin Fuel Production
The investigated fuel is produced at the test complex M11. The contents are industrial paraffin
of the manufacturer Sasol Wax GmbH. For the first tests the paraffin wax 6805 was used. The
fuel used for the tests had a long-chained polymer and color additive of 5 % by mass, hence
the name in this document "PB-5%". Adding this polymer improves the mechanical properties
of the fuel grain, as well as the regression behavior. One important factor for the regression
behavior is the fuel’s viscosity in the liquid state. The melt layer on the grain’s surface increases
regression rate depending on the viscosity. The lower the viscosity, the higher the droplet
entrainment. Therefore, by adding high viscosity polymer to the paraffin, the regression rate
Figure 5.2.: Technical drawing of the experimental combustion chamber
74
Table 5.2.: Combustion chamber parts
Number Part description Material
1 Inflow cone Aluminum

2 Injector cavity Aluminum
3 Combustion chamber Brass
4 Pre combustion chamber insulation Phenolic composite
5 Combustion chamber wall Aluminum
6 Paraffin fuel insulation Phenolic composite
7 Post combustion chamber insulation Phenolic composite
8 Nozzle insulation Phenolic composite
9 Nozzle insert Graphite
10 Outflow insulation Phenolic composite
11 Fuel grain Paraffin based
12 Flange Aluminum
Table 5.3.: Combustion chamber design parameters

Design Parameter Symbol Value
Maximum burning duration tb 180 s

Average mixture ratio O/F 2.0-2.5
Average chamber pressure pcc 30 bar
Average thrust F 100 N
Regression coefficient, estimation a 0.117
Regression exponent, estimation n 0.62
Fuel length lf u 150 mm
Fuel outer diameter dout 175 mm
Nozzle throat diameter dt 5.4 mm
Nozzle exit diameter de 12.2 mm
75
is slowed down. Another additive is 1 % blackening additive. This is to create an opaque fuel
grain in order to limit the melting process to the surface. Otherwise the radiative heat transfer
into the fuel grain is too large and larger fuel parts could melt and detach. The formulation of
this fuel was gained in previous works at the DLR Lampoldshausen [30]. The paraffin is melted
and mixed with the polymer before the blackening additive is added. In this way it is made
sure that despite the higher melting point of the polymer the fuel is homogeneously mixed.
Figure 5.3 shows four images of the fuel production. The pure paraffin is melting in the first
two images until the polymeric additive is added in the third image, when the paraffin is hot
enough. Magnetic stirrers make sure the mixing is fast and successful. In the last picture color
is added. After the melting and mixing the fuel is cast into a cylindrical shape.
5.1.2. Test Sequence
In order to control the test an automatic sequence was used. Table 5.4 shows an example
sequence for the first tests. In advance of the automated test sequence, the lines are manually
flushed with nitrogen for cleaning purposes. Additionally, the pressure in front of the valve
0.311 is set to the desired value. This pressure in combination with the opening of the control
valve is the main influence on the oxygen mass flow. The oxidizer temperature also influences
the mass flow by a small percentage. The control valve is used to precisely set the mass flow,
Figure 5.3.: Paraffin production [73]
76
however it is moving very slowly. Therefore the control valve cannot be used to realize quick
throttling of the engine. In order to throttle the engine, the pressure at the valve 0.311 is
changed, which is responding much faster.
5.1.3. Methodology
In order to analyze a test conducted at the test bench, the following information and data is
analyzed and processed:
• Measurements of the fuel grain before and after the test
– Fuel mass: The fuel mass is determined with a scale before and after the test. It is
weighed together with the fuel insulation and the fuel insulation mass is deducted.
– Length: The length of the fuel grain after the test varies depending on the measuring
position. An averaged value is used for the analysis. The length is measured at
positions rotated by 90°.
– Inner diameter (at front and aft end): The diameter is also averaged by measuring
in different angles.
– Outer diameter: The outer diameter is not changing during a burning test.
• Regression of insulation parts: The regression of insulation parts is neglectable for the
analysis of the combustion process. However, the inspection of the parts is necessary to
replace them for future tests, if they are burned too much.
• Nozzle throat diameter: The graphite insert inside the nozzle has a very small regression.
This is considered for the evaluation of the chamber performance.
• Pressure sensor data: Pressure sensor data is stored in several text files by the measure-
ment system of the test bench M11.3.
• Temperature sensor data: Temperature sensor data is stored in several text files by the
measurement system of the test bench M11.3.
Table 5.4.: Test sequence

Time in [s] Event Comment
-20 Start of automated sequence Control valve opening changed to 17 %

-14 Opening 0.311, 0.319 & 0.339 Reservoir connections opened
-3 Activating spark plug Inside the hydrogen/oxygen torch
-0.2 Opening 0.320 Oxygen flow to torch opened
0 Opening 0.312 and 0.340 Hydrogen flow to torch, oxygen flow to CC opened
0.2 Closing 0.339 and 0.340 Hydrogen flow and torch flame stopped
0.3 Closing 0.319 and 0.320 Oxygen flow to torch stopped
5 Closing 0.311 and 0.312 Oxygen flow to combustion chamber stopped
77
• Mass flow data from Coriolis sensor upstream of the control valve ṁox .
The sensors’ positions for pressure, temperature and mass flow are listed in Table 5.1. The
manually measured data is entered in a table file. Another table was created using NASA CEA
containing ideal combustion data including gas density, sonic velocity, specific impulse and
the characteristic velocity for several pressure and mixture ratios. These data is specific for
the combination of gaseous oxygen and the used paraffin-based fuel PB-5%. Both these table
files and the data files of the sensors are read and processed by a script written in Python
programming language. Figure 5.4 shows the process chain of the script and Figure 5.5 depicts
the ideal temperature and characteristic velocity of the combustion with gaseous oxygen and
PB-5% for different pressures in the range from 10 bar to 50 bar. In order to evaluate the
connection of the regression rate ṙ to the oxidizer mass flux density Gox the regression rate
needs to be reconstructed, as no device is available to measure the regression rate in real time.
Therefore both regression rate and the mass flux density need to be averaged over the whole
test time. The regression rate is averaged as
df − d0
ṙ = . (5.1)
tb
The oxidizer mass flux density is calculated with
16ṁox
Gox = 2 . (5.2)
π df + d0
Therefore, a long test duration reduces the impact of measurement errors (e.g. for the
dimensions of the fuel grain), but of course increases the error of the averaging technique, since
during the test beginning and end the deviation from the averaged value is quite high. Shorter
test duration however increases the impact of measurement errors and transient phenomena
Figure 5.4.: Process flow of the script to analyze the test data
78
Figure 5.5.: Ideal temperature & characteristic velocity with NASA CEA
during ignition and shut-off of the engine. The frequency analysis of the combustion chamber
pressure is realized with Python’s own included frequency analysis functions. The Fourier
transformation is done with the function "np.fft.fft" of the numpy package [74].
5.2. Test Results
5.2.1. Test Description
After the first test of the experimental combustion chamber an injector redesign was necessary.
The injector was produced from two different brass parts screwed together with steel screws.
The steel screws were not withstanding the high heat load and melted, which changed the
injector area. The gas flow at the injector is pure oxygen. This increased the reactivity of
the steel screws. Figure 5.6a shows the pressure records of the first test. After analyzing all
available data, the events that lead to this strange pressure curves could be possibly explained
like follows.
• Section I: When opening valve 0.312, the control valve pressure drops quickly. The
pressure behind the control valve rises slowly due to the mass flow limitation that the
choked flow creates. This limited mass flow needs some time to fill the large volume of
the combustion chamber and all the pipes to reach a steady pressure. In all following
tests it is usually taking 4 s to 5 s to reach a steady pressure level.
• Section II: Injector pressure rises in a short time, while the chamber pressure is reduced.
The reason cannot be clearly identified but one explanation is that one of the five injector
holes where obstructed by metal fragments from the screws. This explains why there is
less mass flow with higher a pressure drop.
• Section III: The hole stays closed and the pressure rises slowly. As it is still in the first
five seconds, this is expected.
• Section IV: The closed hole is most likely open again and the pressure drop is reduced to
its regular value.
79
• Section V: The pressure is again steady for a short period of time. The chamber pressure
is higher than the design point of 30 to 35 bar. Later analysis and comparison showed,
that the nozzle’s effective throat area is smaller than the geometrical throat area, due to
a vena contracta. This was neglected during design, but has a high impact due to the
chosen conical nozzle geometry.
• Section VI: The injector pressure drop is getting smaller and approaches zero. This is
when the steel screw in the injector finally melts and burns away and a large hole is
inside the injector plate.
• Section VII: The main valve 0.312 is closed and the pressure drops. The residual oxidizer
in the line and the injector dome keep the combustion going for nearly a second until the
nitrogen flushing starts.
One interesting observation of the first test was that even when the steel screws melted and
burned, the brass plate did not melt, although the brass’ melting point is lower than that
of steel. The reason for this is that the heat conductivity of brass is much higher and the
injector plate never reached the melting point temperature. This led to further improvements
of the combustion chamber design, where steel is exposed to the hot gas. The injector and
the preburner inlet were redesigned to include only brass parts where contact to the hot gas
is imminent. Also the torch flame inlet was reconstructed with brass and phenolic composite
protective covering. For the evaluation of the following tests a nozzle throat area coefficient of
0.95 was added in order to include the effects of the vena contracta. It must be kept in mind,
that this is simply an estimation and cannot be determined in more precision. However, if the
value is kept constant, a comparison between the different tests is possible. Figure 5.6b shows
a Fourier analysis of the combustion chamber pressure signal. It shows that the magnitudes
of the frequencies are very small. There are some peaks up to about −50 dB. Opposing the
problems in the combustion chamber during the first test, the combustion itself was still very
smooth and no high combustion chamber instabilities were occurring. For comparison in
Figure 5.6c and 5.6d the Fast Fourier Transform (FFT) analysis for test No. 3 and test No. 4
are shown. It is visible that the oscillations in these two tests are even lower in magnitude. All
other tests have a very similar FFT graph like these two tests. During test No. 2 some leakage in
the oxidizer line reduced the mass flow inside the chamber and the test procedure was changed
accordingly to double check all connections in advance of the test. Starting with test No. 3 the
results were in good agreement with the expectations. However, in contrast to most hybrid
rocket engines tests conducted at the DLR Lampoldshausen before this test campaign, the fuel
block has a very high diameter to length ratio. This is the reason, why the lateral combustion at
the front and end side of the fuel grain has a major impact on the combustion behavior. While
in earlier tests with very long but thin fuel grains the lateral combustion could be neglected
during the design of the engine, with the new combustion chamber layout the fuel regression
on the lateral sides added a larger fuel mass flow to the combustion. Additionally the fuel
regression rate is a bit higher than expected, since the regression rate behavior was not known
for the region of low Gox . These two factors changed the mixture ratio of oxidizer to fuel mass
80
flow in the combustion chamber compared to the design point to a much lower value of about
1 to 1.1 in the first tests. The design point was at about 2 for the mixture ratio. As can be seen
in Figure 5.5 the optimum for the characteristic velocity c∗ is between 2 and 2.5. In order to
reach a higher mixture ratio and a high mass flux density Gox the combustion chamber was
modified after test No. 3. The only change was the nozzle throat diameter to 9 mm and the
pressure before the control valve was adjusted to 90 bar from 60 bar in order to enlarge the
oxidizer mass flow without increasing the chamber pressure to much. However, even a higher
oxidizer mass flow could not increase the mixture ratio, as the regression rate climbed nearly
at the same rate. Figure 5.7a, 5.7b, 5.7c and 5.7d show the measured pressure and mass flow
signals of test No. 3 and 4. It is again visible that the pressure graph has a slow rise to a
constant value during the first five seconds of the test. This is due to the fact, that the mass
flow is limited by the choked conditions in the control valve. As the mass flow is limited, the
initial "filling" of volumes of the combustion chamber and pipes with high pressure gas takes
some seconds. The mass flow in test No. 4 after the changes in the operating point pressure
g g
and nozzle throat diameter rose to roughly 50 s in comparison to 30 s during test No. 1 to
No. 3. In both mass flow graphs a transient phase at the beginning of the test after the main
valve opening is visible. The duration of the transient phase is roughly 3 s. This transient
behavior related to the integration of the coriolis sensor’s internal signal. The coriolis sensor
was pre-installed. This coriolis sensor type is rather meant to measure constant mass flows over
a long time. It is not very suitable for measuring quickly changing mass flows. The chamber
pressure in test No. 4 was only 20 bar compared to 40 bar as before. The reason for this is, that
the mass flow was increased, but the oxidizer ratio was not. With a higher oxidizer ratio the
pressure would have been higher. In Test 5 a protection was mounted on the front side of the
fuel, to reduce the lateral regression and increase the mixture ratio. Test No. 1 to 5 used the
same nozzle and insulation parts, which corresponds to a total testing time of 45 s until the
beginning of test No. 5, which was initally set for a burn duration of 25 s. However, a material
defect in the nozzle insulation part (part No. 8 in Figure 5.2) caused the creation of a hole in
the material at the contact side between graphite and insulation material. This hole finally
burned through during test No. 5 quite at the beginning of the test sequence. The hole slowly
increased until the test was aborted after 24 s. The pressure inside the combustion chamber
accordingly was not stable but decreased continuously during the test duration. An evaluation
of combustion efficiency and other characteristic numbers is hence not possible. The values for
regression rate however are useful. For test No. 6 on 21st January 2018 a test duration of 30 s at
54 g/s was realized which was the longest test duration with full mass flow for a hybrid rocket
engine at the DLR Lampoldshausen up to this date. Inspection of the fuel grain showed, that
the front side protection, which was introduced in test No. 5 was very useful. On the rear side
of the grain however, the regression was still too large. During long test duration the heat loads
on the post combustion chamber insulation are very high and also the nozzle becomes white
glowing. This increases the heat flux into the back side of the grain. Therefore it was decided to
introduce the fuel grain protection also on the back side. Test No. 7 was the first test which had
a changing supply pressure in order to simulate throttling of the engine. At the same time the
81
(a) Test 1 Pressure (b) Test 1 FFT Analysis
82
(c) Test 3 FFT Analysis (d) Test 4 FFT Analysis: Regression rate
Figure 5.6.: Test 1, 3 and 4 measurements

(a) Test 3 Pressure (b) Test 3 Oxidizer Mass Flow
83
(c) Test 4 Pressure (d) Test 4 Oxidizer Mass Flow
Figure 5.7.: Test 3 and 4 measurements

burn duration was prolonged to a planned 40 sec. Unfortunately the nozzle which was used in
several earlier tests did not withstand the accumulated heat and mechanical loads and broke
and the test sequence was stopped after 14 sec. The test campaign for this work was concluded
after test No. 7. All test parameters are summarized in Table 5.5. Additional plots of test No. 2,
5 and 6 have been added to the appendix E.
5.2.2. Regression Rate Tests
One important part of the test campaign for this work was to further identify the regression
rate law for the used propellant combination at the DLR Lampoldshausen. The fuel production
is described in section 5.1.1. The propellant combination has been previously used in a
small chamber test campaign. The data points won by this campaign are also included as a
reference. The tests were conducted with a very small combustion chamber (100 mm long and
30 mm in diameter), which is abbreviated "MiniCC" in the following. The most important
numbers to predict the behavior of a propellant combination in hybrid rocket engines from
an experimental point of view are the regression coefficients a and n. The averaged regression
rates of every test is plotted against the average oxidizer mass flux in order to do a curve fit
according to the equation
n
ṙ = aGox . (5.3)
Figure 5.9a shows the regression rate of the tests conducted for this work. The Figure 5.9b
diagram compares the actually measured regression rates with the ones being corrected for the
mixture ratio. The mixture ratio correction is suggested by Karabeyoglu [75] as
1−n
fc (O/F ) = 1−n . (5.4)
1 + O1/F − 1 O/F
It must be noted that this correction always yields a factor larger than one. Therefore the
equation given in the paper by Karabeyoglu is not complete. It must be normalized by a factor
evaluated at a reference mixture ratio. From the paper it is extracted that O/F = 2.05 has been
chosen as the reference point. The completed equation is then
n fc (O/F )
ṙ = aGox . (5.5)
fc (2.05)
As all test conducted within this work had a lower mixture ratio than 2.05 the regression rate is
always higher than it would be predicted by the regression rate law. Therefore in the diagram
all data points of the corrected regression rate are lower than the actually measured ones. The
scatter of the data points is quite high, which is only natural since most tests had caused a
change to the chamber set-up for the following tests like changing the nozzle diameter, adding
the fuel protection on the front and back side and changing the combustion duration every
test. Figure 5.8 depicts the regression rate data of both the miniCC and the current tests
in one diagram. Both data sets are printed with and without correction. Additionally the
84
Table 5.5.: Test parameters
Test No. 1 2 3 4 5 6 7
Test Date 18.01.2017 21.02.2017 08.03.2017 16.05.2017 25.10.2017 23.01.2018 15.02.2018
Fuel parameters at t0
Inner diameter di 65 mm 69.3 mm 69.8 mm 62.5 mm 100 mm 100 mm 62 mm
Outer diameter do 172 mm 172 mm 173 mm 174 mm 173.6 mm 173 mm 173.5 mm
Fuel length lf 149.8 mm 148 mm 149.8 mm 150 mm 141 mm 141 mm 132 mm
Initial mass mf 2.797 kg 2.657 kg 2.732 kg 2.851 kg 2.047 kg 2.037 kg 2.529 kg
kg kg kg kg kg kg kg
Fuel density ρf 937 m3
922 m3
927 m3
924 m3
918 m3
923 m3
929 m3
Fuel parameters at tb
Fuel length lf 148 mm 146.8 mm 140.5 mm 139.3 mm 118.3 mm 97 mm 132 mm
di injector side 68.1 mm 74.2 mm 81.5 mm 80 mm 109 mm 124 mm 82 mm
di nozzle side 70.5 mm 74.8 mm 89.5 mm 99 mm 136 mm 168 mm 108 mm
di averaged 69.3 mm 74.5 mm 85.5 mm 89.5 mm 122.5 mm 146 mm 95 mm
di calculated 69.9 mm 76.1 mm 87.6 mm 96.3 mm 124.6 mm 140.4 mm 98.6 mm
Final mass mf 2.691 kg 2.53 kg 2.277 kg 2.105 kg 1.247 kg 0.719 kg 1.964 kg
Nozzle throat dt 5.4 mm 5.4 mm 5.4 mm 9 mm 9.1 mm 8.6 mm 8.7 mm

Burn time tb 5s 5s 15 s 15 s 25 s 30 s 14 s
Used oxygen mox 0.129 kg 0.145 kg 0.488 kg 0.753 kg 1.314 kg 1.677 kg 0.666 kg
g g g g g g g
Ox. mass flow ṁox 22.7 s 25.9 s 30.2 s 47.3 s 53.7 s 54.3 s 44.9 s
Ox. temp. Tox,inj 8.1 ◦C 16.1 ◦C 11.6 ◦C 14.92 ◦C 11.9 ◦C 12.8 ◦C 15.0 ◦C
Time averaged pressures

Aft. contr. valve prv 37.6 bar 38.0 bar 54.7 bar 71.5 bar 81.8 bar 84.5 bar 71.1 bar
At injector pinj 36.2 bar 35.9 bar 50.7 bar 65.1 bar 73.1 bar 75.6 bar 62.3 bar
In combustor pcc 30.5 bar 26.9 bar 38.6 bar 22.8 bar 16.5 bar 27.6 bar 21.1 bar
Performance data
Mean mixt. ratio o/f 1.23 1.15 1.07 1.01 1.64 1.27 1.18
m m m m m m
Charact. vel. c∗ 1511 s 1134 s 1355 s 1376 s 1109 s 1476 s 1350 ms
Efficiency ηc∗ 101.3 % 78.5 % 95.6 % 99.3 % 64.2 % 96.7 % 92.6 %
kg kg kg kg kg kg kg
Mass flux Gox 6.4 m2 s 6.2 m2 s 6.2 m2 s 9.5 m2 s 5.4 m2 s 4.8 m2 s 8.9 m2 s
Regression rate ṙf 0.43 mms 0.61 mms 0.55 mms 1.06 mms 0.50 mms 0.65 mms 1.23 mm
s
85
literature regression rate law of the paraffin-based fuel SP-1 [75] is shown. The regression
rates of the used fuel with 5 % polymer are only slightly higher than the SP1a regression rate
would predict. This suggests that a similar mixture has been found. The current data set does
not allow yet to create a reliable curve fit to get the a and n coefficients for the used paraffin-
based mixture. More data points with much higher Gox need to be gathered for the curve fit to
work properly. However, in Figure 5.8 the curves for an exponential fit for the gathered data
points are shown. The best curve fit with the highest coefficient of determination of 0.89 is
ṙ = 0.3G0.42 which is probably predicting too slow regression rates for high mass flux density
Gox > 100 kg/m2 s. If a regression is conducted with a fixed exponent equal to the literature value
0.62 . The coefficient of determination is only 0.8. Tests with
of SP-1a, the equation is ṙ = 0.13Gox
higher oxidizer mass flux density are necessary to improve this prediction. The effect of using
LOX instead of gaseous oxygen needs to be investigated as well.
5.2.3. Throttling Tests
Test No. 7 on the 15th February 2018 was the first test with a sequence including throttling
of the engine. The throttling was realized with changing the supply pressure of the oxidizer
inflow at the valve PV11. The reasoning for this approach is that the control valve’s motor is
too slow and therefore not an option in order to regulate the mass flow in short time frames. As
mentioned before the graphite nozzle broke in this test and therefore the planned sequence of
40 s was manually stopped at 14 s. The planned sequence was including four different supply
pressures for 10 s each: 80 bar to 100 bar to 60 bar and finally again 80 bar. As the test was
stopped prematurely only the first two levels were executed. Figure 5.9c shows the pressure
diagram of test No. 7. The diagram shows the already familiar slow ramp up of the pressure
which takes about 4 s to get to a stationary operation of the rocket engine. The pressure at
Figure 5.8.: Regression rate results and exponential fit
86
the control valve sets up at 68 bar and the chamber pressure at around 23 bar. At 10 s the
feed pressure is changed to 100 bar and the control valve pressure starts increasing to 86 bar
while the chamber pressure rises to 29 bar. The pressure rise again takes nearly 4 s to 5 s. At
13.6 s the nozzle breaks open which is clearly identified in the chamber pressure drop and the
rise in the injector pressure. It can also be seen that the engine pressure drop has nearly no
influence on the injector pressure and no effect at all on the pressure at the control valve. This
confirms that the valve is working at choked flow conditions as it is supposed to. The throttling
process itself did not impose any problems or instabilities on the chamber, however, the change
of pressure is quite slow as the whole piping and chamber volumes are quite large compared
to the oxidizer volume flow. Figure 5.9d illustrates the mass flow of the gaseous oxygen in
test No. 7. The oxygen mass flow increases as expected with the higher feed pressure. The
stationary level in the first 10 s was 45 g/s and after the rise to the higher supply pressure it
was 58 g/s. In order to realize more realistic throttling tests, a new engine design with a larger
oxidizer gas volume flow compared to the piping and engine volume is necessary, in order to
decrease the duration of the transient phases. Alternatively the engine could be tested with
LOX as oxidizer on a different test bench. The nozzle throat graphite part was redesigned after
test No. 7. The part had a quite sharp edge close to the point of the highest heat load. This was
replaced with a conical shape. More tests in the future to include throttling at the test bench
M11.3 are planned.
87
(a) Regression rates (b) Corrected regression rates
88
(c) Test 7 Pressure (d) Test 7 Mass flow
Figure 5.9.: Regression rate data and test No. 7 measurements

6. Hybrid Propulsion Lunar Mission Analysis
6.1. Reference Mission
6.1.1. Launch with Ariane 5
The concept being investigated is a Moon spacecraft launched by the Ariane 5 launcher. There
are several options possible to consider when launching a spacecraft to the Moon with a heavy
launcher like Ariane 5.
1. Launch to LEO and from there the spacecraft conducts its own maneuvers to reach the
Moon
• highest spacecraft mass
• high payload
• high spacecraft complexity, larger ∆V budget
• short transfer time
• launch window flexibility
2. Direct launch into a Moon transfer trajectory
• lower spacecraft and payload mass
• less spacecraft complexity and less ∆V budget
3. Low-thrust transfer using electrical engines from LEO to Moon orbit
• very long transfer time
• high payload
• electric propulsion system and high power demand necessary
For the mission analysis the direct launch into a Moon transfer trajectory was chosen as a
reference. The spacecraft is launched at Guiana Space Center. At the apogee of the Hohmann-
transfer ellipse, the spacecraft is then caught up by the Moon. This is comparable to the
Luna 24 mission. The aim is to enable a lower cost Moon mission. This is reached by
reducing the spacecraft’s complexity and cost. A direct transfer orbit injection lower’s the
∆V requirements and allows for a simpler spacecraft design. The Ariane 5 launch capability
to the Moon has been analyzed. An exemplary launch date has been chosen to be the
31st December 2024 at about 4:30 am. This launch window is optimal to reach a maximum
payload as the Moon crosses the orbital plane when the launched spacecraft crosses the Moon’s
89
path. However, this is only an exemplary date. A similar launch window is opening every
two weeks, when the Moon is close to its ascending or descending node. The possible payload
was determined to be 8800 kg with the current Ariane 5 launcher data. The Moon orbit of the
spacecraft would have a periselen of 100 km altitude and an excess velocity of 860 m/s. A polar
inclination of 90° was chosen for this reference mission as every landing site could be reached
from this orbit if phasing orbits are done (but only if all propellants are storable for up to one
month). Different inclinations around the Moon could be easily reached as only a slight change
in the transfer orbit will increase or decrease the inclination around the Moon by a lot. This
reference trajectory for the Moon transfer is shown in Figure 6.1. The 3D visualization shows
the highly elliptical transfer from the Earth’s surface to the vicinity of the Moon. At a distance
of roughly 60 000 km to the Moon the reference coordinate system is changed as this represents
roughly the borderline of the sphere of influence of the Moon. This is marked with the red
line. At this point the reference center of the trajectory is the Moon. The Moon is highlighted
with a white circle. Figure 6.2 displays the hyperbolic trajectory that the spacecraft will have
when arriving at the Moon’s sphere of influence after being injected into a transfer orbit with
Ariane 5. The trajectory has been optimized, so that the excess velocity is at a minimum at
860 ms and the periselen of the spacecraft is at about 100 km. The orbit inclination to the
Moon’s equatorial plane is close to 90°. With this inclination phasing orbits could be conducted
to reach any landing spot on the surface. Therefore for the future landing analysis no specific
landing spot had been chosen, as this is totally open from a spacecraft propulsion system point
of view. The inclination of the spacecraft at the Moon is easily adapted with very small changes
in the transfer orbit. The trajectory optimization resulted in a maximum payload of 8800 kg.
This payload includes the whole spacecraft mass that is sent to the hyperbolic flyby at the
Moon. The spacecraft itself must conduct course corrections while on its transfer orbit, as well
as the orbit injection at the Moon. This is described in the next section. The transfer orbit
Figure 6.1.: Moon transfer trajectory of a Ariane 5 launched spacecraft
90
Figure 6.2.: Hyperbolic Moon trajectory after launch with Ariane 5
parameters after separation from the Ariane 5 Upper Stage are listed in Table 6.1, as well as
all the orbital elements when arriving at the Moon’s orbit periapsis. The complete duration of
the orbit transfer is about 5 days. As a note, the Ariane 5 Upper Stage is not reignitable. If it
was, the payload could possibly be increased by choosing different orbit maneuvers. A future
update of the mission profile could be done with the performance values of Ariane 6, when
they are released.
6.1.2. Moon Orbit Injection And Landing
When the spacecraft reaches the periselen at about 100 km, several maneuvers are performed
to land on the Moon’s surface. The first impulsive maneuver is conducted to stabilize the
spacecraft in an orbit around the Moon. For this the rocket simply fires backwards and slows
down. The resulting orbit is either highly elliptical or circular. Another possibility is a direct
landing trajectory after the first burn. In all other cases the spacecraft can conduct phasing
Table 6.1.: Orbital elements after launch with the Ariane 5 ECA
Parameter Value Unit Parameter Value Unit
Reference center Earth center - Reference center Moon center

Apoapsis altitude 366539 km Apoapsis altitude -16738 km
Periapsis altitude 460 km Periapsis altitude 100 km
Periapsis argument 72.9 ◦ Periapsis argument 175.4 ◦
Ascending node 96.5 ◦ Ascending node 76.2 ◦
Inclination 4.9 ◦ Inclination 90 ◦
91
orbits to reach nearly every spot on the Moon’s surface. Then another burn is performed
to lower the periselen to a very low altitude above the surface. A coast arc follows and the
spacecraft travels half way around the Moon to reach the periselen. Close to the lowest altitude
the engines are fired again and the spacecraft is slowed down. This burn can be separated into
two phases with a coast arc in between, if the mission makes it necessary (e.g. if the spacecraft
needs to descend closer to the surface). In order to be able to choose the right landing spot with
suitable surface conditions, a hovering phase at 1 km to 0.85 km altitude is added. During this
phase the lander roughly stays at the same position in the reference trajectory. In the real
mission, this phase can be used to adapt the position to the surface conditions. After the
hovering phase a coasting phase is added, during which the lander is pulled down by Moon’s
gravity and in a final engine burn the spacecraft will land softly on the surface. For the analysis
of a lander spacecraft using hybrid propulsion, a three stage concept is the baseline. The reason
for this lies in the limited burning duration of hybrid rocket engines. The hybrid rocket engine
imposes low-cost combustion chamber technology but introduces higher staging complexity.
The three stages are used as follows: The first stage is used to conduct the orbit insertion and
lower the periselen to below 30 km. The exact value is left to the optimizer to find. The second
stage is used to slow down and initiate the final landing. The third stage is the actual lander. It
needs very little propulsion power. However, it needs to be maneuverable and to touch down in
a soft and controlled way. Eight spacecraft concepts have been established and are compared
in the next section. The altitude over time of this landing maneuver is shown in Figure 6.3.
On the left the whole landing trajectory is shown, on the right side a detail of the final 4 km
of the descent is shown. The different phases, which are defined in ASTOS, are highlighted
with color. The plots refer to the optimized trajectory of a three staged spacecraft. For other
variants the trajectory looks similar, even if the phases’ duration might differ. In the detailed
view of the altitude in the last 3 min during the landing, it is visible, that at 1 km altitude
the spacecraft’s sinking speed is reduced and for 1 min the spacecraft is nearly staying at the
same altitude. Figure 6.4 illustrates the corresponding velocities and speed. The flight path
speed is shown together with the three velocity components in northern, eastern and radial
direction. The east velocity is relative to the surface, taking into account the rotation of the
Moon. It can be seen that the larger amount of ∆V is done at the second stage burns. The
Figure 6.3.: Altitude over time of the Moon lander (total and detail)
92
first stage is performing a smaller ∆V . The first stage burn is capturing the spacecraft into
a circular Moon orbit and afterwards lowers the periselen even more. The first burn of the
second stage is reducing the velocity, to lower the periselen farther and leave the elliptical
orbit around Moon. The second burn of the second stage is then reducing the vertical velocity
during the descent. The east velocity rises to a maximum during the first coast arc while the
north velocity passes through zero. This is when the spacecraft flies very closely to the south
pole of the Moon and the negative north velocity (i.e. heading south) is flipping into a positive
north velocity. The detailed view shows the final landing, where from 3550 s roughly a vertical
descent is conducted. This is identified as the relative east and north velocities are zero. This
was a constraint given to the optimizer. The optimization in ASTOS does not only optimize the
spacecraft’s components, but also the controls like attitude and thrust level. Figure 6.5 shows
exemplarily the pitch and yaw controls of this trajectory. A smooth control is usually a sign for
an optimal solution. If there are spikes and jumps it is an indicator that the trajectory is not
fully optimized yet. The curves in Figure 6.5 are considered very well optimized. The x-axis
value is the independent variable. This variable is the "optimization time". It starts at zero and
every phase is exactly one unit long. Figure 6.6 depicts the oxidizer mass flow of a single lander
stage engine during the last phases of the descent.
Figure 6.4.: Speed over time of the Moon lander (total and detail)
Figure 6.5.: Attitude over independent phase variable
93
Figure 6.6.: Oxidizer mass flow of a single lander engine
ASTOS Model Set-Up
The ASTOS model for this reference landing mission has the constraints for the different phases
listed in Table 6.2, where the abbreviations IBC is an initial boundary constraint. FBC is a final
boundary constraint respectively and PC marks path constraints. The phases are named as
follows:
A Orbit Capture Burn 1st Stage D Coast Arc #2 G Lander Hovering

B Coast Arc E 2nd Stage Burn #2 H Descent Coast Arc
C 2nd Stage Burn #1 F Lander Burn I Lander Burn & Vertical Descent
Phase B is initialized with 29 major grid nodes in addition to the first and last one. Accordingly,
phase B is optimized with the "Collocation Hermite" method (see [55]). All other phases are
initialized with only one major grid node and hence optimized with the "Runge Kutta 4/5"
multiple shooting method. This proved to be the fastest way of optimizing this type of
trajectory. For very long coast phases with low gradients in the acceleration the collocation
method works best as it replaces the integration of the real trajectory with a polynomial curve
with much less integration work. In this case the Moon’s gravitational acceleration changes
only by about 9 % during the nearly one hour long coast. The trajectory was not optimized
with all constraints active initially. The first optimization was done with wider ranges of the
constraints, as well as attitude controls as parameters instead of control parameters. This made
the first optimization to reach the rough trajectory very quick. Afterwards it was reinitialized
with the parameters from this first result to introduce more precise constraints and the attitude
control parameters. The control is refined with 9 nodes in every phase.
94
Table 6.2.: Moon landing optimization constraints
Constr. A De-orbit burn F G I
Mass IBC1 8500 kg

Altitude FBC 99.5 km to 100.5 km 0.85 km to 1.0 km 0.85 km to 1.0 km 0.0 m to 1.0 m
PC 0.85 km to 1.0 km
Aposelen FBC 99.5 km to 100.5 km
Periselen FBC 10 km to 30 km
Radial vel. FBC −0.5 ms to 0.5 ms −0.5 ms to 0.5 ms
PC −5 ms to 0 ms
North vel. PC −0.1 ms to 0.1 ms ; −0.1 ms to 0.1 ms
m m
East vel. PC −0.1 s to 0.1 s −0.1 ms to 0.1 ms
Flightp. sp. FBC 0 ms to 0.5 ms
1 For the cryogenic variants the vaporized propellant needs to be substracted, 300 kg are substracted
for the orbit correction maneuvers
Note: IBC: Initial Boundary Constraint; FBC: Final Boundary Constraint; PC: Path Constraint
6.2. Lunar Lander Spacecraft Design
6.2.1. Mass Estimations
In order to validate the spacecraft’s design and trajectory, some spacecraft subsystem masses
were approximated with simple estimations. Table 6.3 shows the assumptions which were used
in the calculation of the spacecraft masses. For the subsystems not related to the propulsion
system, the mass estimation is very preliminary. The estimations are based on comparisons of
existing missions and experience values.
Propulsion System Model
In order to correctly predict the hybrid propulsion system performance for the trajectories
the hybrid analysis tool for ASTOS has been used and the results and mass estimations
have been validated with ESPSS simulations. Especially the pressurization gas mass, hybrid
engine specific impulse and thrust have been confirmed with the EcosimPro/ESPSS model.
The simulation model schematic is shown in Figure 6.7. The simulation model includes a
pressurization tank, valve and pressure regulator, an oxidizer tank and the piping connecting
the tank to the engine. As the stage has multiple engines, only one engine was modeled while
the respective mass flows to the remaining engines were simply mirrored with a fixed mass flow
component. The pressurization gas is stored in a 300 bar tank. The oxidizer tank is pressurized
to 30 bar after the start of the simulation sequence. The initial filling level with oxidizer is 90 %
of the total tank volume. The valves and orifices cross sections have been modeled in order to
reach a 10 bar pressure drop over the injector and an engine pressure of about 15 bar. The
95
Table 6.3.: Mass estimation for spacecraft design
Hybrid rocket engine
Chamber wall Cylinder shaped carbon reinforced plastic wall, scaled with pressure
loads
Injector dome Mass estimation based on CAD models with aluminum casing and
brass injector plates
Chamber insulation Mass estimation based on CAD models for insulation parts (phenolic
composite, graphite)
Nozzle Simple mass estimation of a simple laval nozzle
Oxidizer supply
Oxidizer tank Spheric composite tank, safety factor 2, metallic liner, 25 bar pressure
Oxidizer feed system 15 kg for the kick stage, 5 kg for the lander and return rocket
Oxidizer pressurization Helium for pressurization and corresponding high pressure tank
Structural parts
Staging adapter 1.5 % of the stage mass

General structure 7.5 % of the stage mass
Landing legs 130 kg
Other subsystems
Power supply 35 kg Others 25 kg

Flight computer unit 8 kg Driller 70 kg
Thermal control 100 kg Manipulator arm 70 kg
Telemetry & communication 35 kg Scientific instruments 150 kg
GNC 11 kg Rover 100 kg
oxidizer mass flow is actively controlled by a proportional-integral-derivative (PID) controller.

The PID controller uses proportional, integral and derivative terms to calculate the control
value. In this way the controller is reducing the risk of an overshoot when the set point of the
controller’s input changes. The controller’s output is then connected to the valve’s opening
position input. For the simulation of the regression rate in the hybrid rocket engine the
standard regression law with experimental coefficients a and n has been used. The exemplary
simulation results are shown in the Appendix F. Due to the long burning time the ratio of final
to initial inner diameter of the fuel grain is about 1.73. This means the port cross area at the
end of the burn is about three times higher than in the beginning. If the initial diameter is
reduced further the size of the engine can be shrinked but it will increase the diameter ratio
even more, which might influence the combustion behaviour negatively as the flow changes
drastically during the burn duration. This can cause areas of combustion instability during
the burn and needs to be verified with large scale engine tests. A similar EcosimPro/ESPSS
model was created for a liquid bipropellant kick stage using nitrogen tetroxide and hydrazine
as well as cryogenic oxygen and methane. The results won with the simulations of this model
96
Figure 6.7.: ESPSS schematic for a Moon lander kick stage
were used to do an optimized trajectory with ASTOS in order to compare the performance
of the used hybrid propellants with classical storable liquid propellants and also cryogenic
propellants with higher performance but the need to cool the propellants. Table 6.4 compares
some data between hybrid and liquid combustion. For the spacecraft performance another
major difference lies in the propulsion systems’ dry mass which is usually smaller for a pressure
fed liquid bipropellant engine, even if it’s complexity is higher. The engine efficiency has been
chosen as 95 % for the N2 H4 /N2 O4 engine and the LOX/Methane engine, as this is an usually
achieved efficiency in liquid rocket engines. For the hybrid rocket engines it is more difficult
to reach a higher efficiency. This is why only 90 % engine efficiency has been considered.
However, this is conservative, and further advances in hybrid rocket engines might push
the efficiency to higher values as well. The optimal mixture ratio for N2 H4 and N2 O4 with
an expansion ratio of 100 is at about 1.4 and for LOX and Methane it is at 3.4 . For the
hybrid rocket engines the mixture ratio is not fixed, as it is a function of the combustion time.
However, the stage’s propellant ratio is fixed and is between 7.5 to 7.6 for H2 O2 and 2.8 to 2.9
for LOX. The concentration of the hydrogen peroxide is chosen as 95 %.
Table 6.4.: Propellant comparison in the chosen kick stage design

Parameter NTO & N2 H4 LOX & LCH4 H2 O2 & PB-5% LOX & PB-5%
storable bi-liquid cryogenic bi-liquid storable hybrid cryogenic hybrid
Ivac 3300 ms 3553 ms 2950 ms 3280 ms

ηeng 95 % 95 % 90 % 90 %
O/F 1.4 3.4 7.5 to 7.6 2.8 to 2.9
100 100 100 100
97
6.2.2. Hybrid Propulsion Spacecraft Variants
Variant Hybrid 1: Three staged Lander
A first concept design for a lunar lander vehicle with hybrid propulsion had been introduced
in [23]. This design still serves as the first reference in this work even if it has been updated.
Figure 6.8 shows the design sketch of the vehicle. The left upper half of the figure shows
the return rocket module with four hybrid rocket engines, a RCS thruster system, an oxidizer
tank with storable H2 O2 (95 % concentration) and the return capsule on top. Below the return
rocket the lander with extended legs is shown. At the bottom left the kick stage design is
shown. There is a LOX and H2 O2 oxidizer version for the kick stages and the lander module.
The return rocket always is in all cases using H2 O2 as the oxidizer, as the return rocket is
stored for several days or weeks before being launched back towards the Earth. For the other
stages LOX could be used, when it is cooled by boil-off. On the right side of the figure, the
assembled spacecraft is illustrated. It is already visible, that a LOX version of the spacecraft
would be longer, as the hybrid rocket engines are longer due to the lower mixture ratio and the
resulting long fuel grains. In comparison to the already published preliminary data, the mass
estimations have been updated. Table 6.5 shows the overview of the mass distribution of the
spacecraft. The initial mass of 8800 kg for the H2 O2 version accounts also 300 kg of H2 O2 for
Figure 6.8.: Lunar lander spacecraft design sketch, [23]
98
course correction maneuvers with monopropellant thrusters prior to reaching the Moon flyby
position. The additional mass for the lander is a sum of all structural and subsystem mass,
according to the Table 6.3. The H2 O2 for course corrections and steering of the spacecraft in
all phases is used from the current active stage, if the oxidizer of the hybrid rocket engines
is H2 O2 as well. This is an advantage in mass, as the spacecraft doesn’t need an additional
H2 O2 tank on the lander stage. In the LOX version the H2 O2 is always drained from a tank on
the lander module, which weighs additional 22 kg. The LOX version also includes an estimated
LOX boil-off mass which is about 7.5 % of the total tank contents (for 5 days of flight to the
Moon, roughly 1.5 % per day). Studies show, that with multi layer insulations the boil-off is
reduced to 1 % to 3 % per month for liquid oxygen [76]. Usually the vaporization rate for LOX
in a tank varies from 1 % to 10 % in 5 days depending on the tank design, the environment and
the total tank volume. The masses in the table are valid for the lift-off. The H2 O2 for course
correction and RCS, as well as the vaporizing LOX are drained over time while transferring
from LEO to the Moon. Therefore the stages have a smaller mass, when they are entering the
Moon’s orbit. In summary the spacecraft with LOX as the main engine’s oxidizer has a higher
performance and hence a greater return rocket mass. However, the mission conducted with
LOX has more restrictions on the time windows: Phasing orbits in Low Earth or Low Moon
Orbit would not be easily realizable. Also the filling of the tanks on the launch pad would be
more difficult as it has to be done very late during launch preparations. Another issue with
LOX is the ignition. A reignitable engine with H2 O2 is realized quite easily and reliably with
a catalyst bed. For a LOX based hybrid rocket engine an additional ignition system must be
installed (i.e. laser ignition). The thrust of the four engines was first estimated to 13 kN each
and the optimizer kept close to this value.
Variant Hybrid 2: Two staged Lander
Another variant is using a two staged spacecraft. This version is expected to have a smaller
return rocket mass, as it is less effective, because more dry mass needs to be slowed down.
However, it could be more optimal from a financial point of view, as a two staged spacecraft
might be essentially cheaper than a three staged spacecraft. The optimization was conducted
to investigate if the kick stage is still optimal when it provides the ∆V for both the orbital
change and the landing velocity reduction, or if the optimizer scales up the lander vehicle,
which in return could then do the latest ∆V maneuvers on its own. Having a single stage has
some advantages: Due to the spherical shape the oxidizer tank has a higher volume to mass
ratio than two single tanks. Also there is only one feed system, one pressurization system, and
only one staging adapter to the lander, which in total is saving some dry mass. However, as
mentioned, the mass is never jettisoned and therefore the total ∆V of the kick stage is less
than two kick stages in the three staged Variant 1. This also implies that the lander itself has to
provide more ∆V . This is confirmed with the ASTOS simulations and can be seen in Figure 6.9.
This Figure compares the descent of Variant 1 and Variant 2 by showing the altitude over time.
The x-axis zero point is the touch down event. Table 6.6 lists the defining masses for the two
99
Table 6.5.: Lunar lander Variant 1 mass overview
Vehicle system Component Mass H2 O2 version Mass LOX version
Total spacecraft mass 8800 kg 8800 kg
First kick stage 2986 kg 2512 kg

H2 O2 for course correction 300 kg
H2 O2 for RCS 50 kg
Boil-off LOX for cooling 106 kg
Main engine oxidizer 1883 kg 1410 kg
Main engine fuel 247 kg 481 kg
Propellant reserve 20 kg 20 kg
Structure engines 90 kg 131 kg
Structure oxidizer 64 kg 55 kg
Additional structural mass 332 kg 309 kg
Second kick stage 3154 kg 3019 kg

H2 O2 for RCS 100 kg
Lander stage 1337 1829 kg

H2 O2 for RCS 150 kg 300 kg
Structure H2 O2 tank 20 kg
Additional mass 966 kg 963 kg
Return rocket 1323 kg 1440 kg
100
staged Variant 2. The altitude is drawn from the beginning of the lander burning phase. The
first phase in Variant 1 is less than 2 s, while for Variant 2, where the lander has to deliver
higher ∆V , the first lander phase before the hovering phase is about 30 s long. The two-staged
Variant 2 has eight engines with 13 kN each on its kick stage in order to deliver the same ∆V
and having the same burning time per engine as before. The engines can be fired in groups of
four to reduce the acceleration.
Variant Hybrid 3: Engine Jettisoning
The hybrid propulsion system with the fuel being stored in the rocket engine allows for a
simple modification of the two staged model: the engines are ignited and burnt sequential and
jettisoned when emptied. This could be realized by having a similar layout of the spacecraft
like in Figure 6.8, except that the first kick stage has no tank of its own. This variant offers

Vehicle system Component Mass H2 O2 version Mass LOX version
Kick stage 6292 kg 5641 kg

Lander stage 1341 kg 1804 kg

101
Figure 6.9.: Altitude comparison between Variant 1 and 2
only one oxidizer tank for the kick stage but still allows jettisoning empty dry mass of the first
four engines. To realize this in the ASTOS model, first the total dry mass of the kick stage
was increased, as additional staging and structural parts are necessary. Then a certain fraction
of the first stage’s dry mass representing half the engine and part of the structural mass was
ejected after the first burn of the kick stage. Even when the total dry mass of the kick stage is
higher in this variant, the total return rocket mass is still higher than in the Variant 2. This and
all the other respective masses are listed in Table 6.7.
6.2.3. Hybrid Propulsion System Performance
A comparison between an EcosimPro/ESPSS simulation and the ASTOS hybrid tool calcula-
tions has been conducted with the simulation model detailed in Section 6.2.1. The second
stage of Variant 1 with three stages using H2 O2 as an oxidizer has been used. The values in
the following Figures have been changed in the time stamp: The first burn starts 10 s after
the simulation starts. It takes 57 s like in the trajectory simulation. Then the second ignition
takes place at 100 s and the second burn takes 95.5 s. In Figure 6.10 the mass flow of both
ASTOS and ESPSS model is compared. It is visible that at this zoom level, no discrepancy
is seen. The oxygen mass flow at the injector is a control value in the ASTOS model and an
actively controlled value as well in the ESPSS model. However, it depends on the pressure in
the tank as well and it can be seen that in the last few seconds of the second burn, that the
pressurization pressure is too low to keep up the mass flow. However, the difference is only
very small. Another difference is the gradient at start-up and shut-down of the engine. For
long burn duration the effect is very small, but for a different engine burn duration the effect
can have an impact on the trajectory. The fuel mass flow model is the same in both simulations.
This means as long as the oxygen mass flow and the fuel diameter and length is the same, the
simulation result is also congruent. This could be changed, if EcosimPro/ESPSS uses a different
regression model. However, the current state of the art is, that regression is best modelled with
102
Vehicle dystem Component Mass H2 O2 version Mass LOX version
Kick stage 6237 kg 5569 kg

Lander stage 1356 kg 1815 kg

n . Figure 6.11 depicts the thrust curve

empirical regression rate coefficients according to ṙ = aGox
of ASTOS and ESPSS in comparison. Here a difference can be seen in the according zoom level
of the diagram. This is caused by three things: the before mentioned gradients at start-up and
shut-down, the mass flow drop at the end due to missing high pressure in the pressurization
system and lastly a slightly different calculation of the thrust in ESPSS. ASTOS uses an ideal
thrust calculation with an efficiency coefficient. ESPSS also uses the same procedure. However,
small differences in the exit velocity of the exhaust gases have been recognized in comparisons
between ESPSS and NASA CEA results. The corresponding specific impulse is shown in the
next Figure 6.12. While ASTOS’ specific impulse is merely a function of expansion ratio,
chamber pressure and mixture ratio, the specific impulse in ESPSS considers also gradients
and losses. This is why there is a small difference in the specific impulse similar to the thrust.
103
Figure 6.10.: Mass flow computation comparison between ESPSS and ASTOS model
Figure 6.11.: Thrust computation comparison between ESPSS and ASTOS model
Figure 6.12.: Specific impulse computation comparison between ESPSS and ASTOS model
104
6.2.4. Liquid Propulsion Spacecraft Variants
Variant Storable Propellants
To compare the hybrid propulsion system with conventional storable liquid propellant systems
using nitrogen tetroxide (N2 O4 ) and hydrazine (N2 H4 ) a comparable two-staged spacecraft
with storable propellants was optimized and simulated. The mass assumptions were similar
except that a fixed engine mass was estimated. For the kick stage an engine mass of 130 kg was
estimated. This is comparable to the Aestus engine. For the lander spacecraft the mass was
estimated to be 20 kg, which is roughly similar to the Luna 24 engine. The storable propellants
propulsion system has a lower structural mass and a higher specific impulse (mainly due to
efficiency) and therefore the spacecraft has an increased return rocket mass of up to 1545 kg.
This is completely explained with the more lightweight engine structure and the higher specific
impulse. From a narrow performance point of view, storable propellants are preferable to
the LOX/paraffin combination due to storability and performance, and also preferable to the
H2 O2 /paraffin combination due to much higher performance. However, with environmental
effects, health threats, ground operation difficulties, cost of transport, storing and cost of
engine development, the hybrid rocket engines seem to be a viable alternative, if the loss in
payload is considered acceptable.
Variant Cryogenic Propellants
In addition to classical storable liquid propellants, which have a long history of application on
spacecraft flying to the Moon, a cryogenic liquid propellant combination was analyzed as well.
The cryogenic propellants liquid oxygen and liquid methane were used for an optimization
of the landing trajectory. The mass estimations were similar to the storable variant, however,
the tanks mass was increased to consider insulation and a 7.5 % boil-off was considered for
all propellants. As the specific impulse is higher and the boil-off mass is comparably low,
the total return rocket mass was 1566 kg. Possibly the mass estimation for the boil-off and
tank insulation mass is underestimated. However, it seems that cryogenic liquid bi-propellant
systems might be a viable alternative to storable systems, if the mission duration is limited to
a few days. The return rocket itself in this variant is again using storable liquid propellant
N2 H4 and N2 O4 as the return rocket might reside for many days or even weeks on the lunar
surface before flying back to Earth.
6.2.5. Spacecraft Variants Summary
Three spacecraft staging variants using hybrid propulsion have been compared on a landing
trajectory starting from a hyperbolic flyby trajectory around the Moon. The most complex
spacecraft with three stages, namely two kick stages and a lander, provide the highest payload
mass with a fixed initial launch mass from Earth of 8800 kg. Figure 6.13 shows the different
masses of the variants in direct comparison. The differences are comparably small, the only
105
large one is the oxidizer to fuel ratio between the H2 O2 and LOX variants. It is also visible, that
the structural mass of the LOX variants is slightly higher even though the propellants mass
is smaller. The reason for this is that the larger fuel proportion means that the combustion
chambers are larger. The combustion chamber have more structural mass per propellant mass,
as the design is including thermal insulation materials and the long, slender cylinders of the
combustion chamber is less weight optimal than spherical tanks. The ASTOS simulations and
optimizations were done with the hybrid optimization tool. The fuel grain sizes as well as
the resulting mixture ratios were optimizable. Table 6.8 lists the data for the fuel grains
in the different variants. The kick stage fuel grains are taken as a reference. For the three
staged variant, the second kick stage has been chosen. One fundamental advantage of the
H2 O2 engines is the high mixture ratio, which leads to a short combustion chamber. This is
often an advantage in order to keep structural mass low, shorten the spacecraft if the length
is limited e.g. by a payload fairing and for cheaper and easier fuel production. Figure 6.14
Figure 6.13.: Mass comparison of different variants
Table 6.8.: Fuel grain data kick stage

Phase Var. 1 Var. 2 Var. 3
H 2 O2 LOX H2 O2 LOX H2 O2 LOX
Mass per grain 74 kg 150 kg 70.1 kg 140.3 kg 69.9 kg 142.3 kg

Length 0.64 m 1.51 m 0.64 m 1.52 m 0.65 m 1.58 m
Outer diameter 451 mm 450 mm 450 mm 450 mm 450 mm 450 mm
Inner diameter 210 mm 257 mm 227 mm 272 mm 232 mm 279 mm
106
illustrates an early design sketch of the propulsion system of the second stage in the three
staged variant using H2 O2 . The engines with the nozzle and injector inflow connections have a
total length of about 2 m long. The length of the whole propulsion system is about 3.5 m, if the
pressurization tank is located between the engines. The largest diameter is at the nozzle exit
with 2.1 m. With the LOX version, the tanks’ sizes stay nearly the same. Much less oxidizer
mass is needed. However, the density of LOX is lower. The major difference is the fuel length,
which is 0.9 m longer. This is also the difference in the total length of the propulsion system
and results in 4.4 m for the LOX propulsion system. A cross section of a 13 kN hybrid rocket
engine using H2 O2 and PB-5% as propellants is shown in Figure 6.15. The only difference in a
LOX version is the length and inner diameter of the paraffin. The injector inflow is connected
to the supply pipe for the oxidizer. An injector cavity is following downstream. For the
H2 O2 engines the catalytic material is stored in this cavity as a mesh in order to decompose
the oxidizer. If the oxidizer is LOX this cavity is not necessary. Behind the injector cavity the
injector head with an injector plate is following. There are different options for the injector
plate. With hybrid rocket engines, all kinds of injectors have been tested in the past, from
simple showerhead via impinging and swirling injectors to pintle injectors. It depends on
the chamber design and the oxidizer inflow conditions to choose the right injector type. The
rocket chamber itself is made from the fuel grain and insulation material, which is not burning
and carbonizing and therefore withstanding the high heat loads. It also needs very low heat
conductivity. In front of the fuel grain, a pre-chamber is typically found, where the oxidizer
vaporizes and spread evenly over the fuel port. At the end of the fuel grain another section
is found, which is used to give the combustion gases longer residing time inside the chamber
Figure 6.14.: Early design of the kick stage’s propulsion system
107
in order to mix and burn completely. Then a Laval nozzle is ending the chamber. The nozzle
throat is protected by a graphite insert, which withstands the high temperature and heat load
and still has a very slow regression rate. The outside of the chamber is wrapped in carbon
fiber reinforced plastic to complete the light weight design. Table 6.9 lists the ∆V per phase
and spacecraft variant. It shows that the total required ∆V is about 2800 ms for this kind of
landing on the Moon. The differences between the variants are caused by the spacecraft set-up
but also by thrust to weight ratio in different trajectory sections. One remaining issue with a
three staged spacecraft is the remaining of the stages. Currently there is no regulation about
space debris in Moon orbit. Space debris in orbit poses a threat to future missions to the Moon
as the Moon’s gravitational field is very inhomogeneous [77]. Therefore orbits are not stable for
a long time. It is possible for all debris parts like empty stages or inactive satellites to hit the
Moon’s surface after some months or years. This is threatening future surface based missions,
both manned and unmanned. Therefore it is preferred by mission planners to have all empty
Figure 6.15.: Cross section of a H2 O2 /PB-5% engine with 13 kN thrust
Table 6.9.: ∆V [ ms ] per trajectory phase

Phase Var. 1 Var. 2 Var. 3
H2 O2 LOX H2 O2 LOX H 2 O2 LOX
A 856 856 853 854 854 853

C 462 572 216 248 222 245
E 1265 1158 1468 1449 1464 1451
F 2 1 54 53 52 55
G 97 97 98 97 98 97
I 129 151 110 88 116 87
Total 2811 2835 2799 2789 2806 2788

0% 0.85 % −0.43 % −0.78 % −0.18 % −0.82 %
108
stages either leave the Moon orbit (and possibly Earth orbit) or to put them down on the Moon’s
surface in a controlled crash. In the trajectory of Variant 2 no debris remains in orbit. The first
stage is separated in one piece, when the periselen is already below the surface. Hence, the
first stage will splash down closely to the landing site of the lander. By further optimizing
the horizontal velocity before separation the distance between splash down and landing site
is manipulated. Variant 1 and 3 have a separation after the first burn, where the periselen is
still above 10 km. Only a very small impulsive maneuver is enough to bring the first stage or
drop off mass to a impact on the Moon’s surface. This could be realized by a small cold gas
thruster system, a very small solid rocket motor or a further short ignition of the hybrid rocket
engines.
6.2.6. Return Rocket
The return rocket vehicle has the same design in all six before aforementioned hybrid rocket
variants. As seen in Figure 6.8 it has four hybrid rocket engines with about 4 kN thrust each,
a single oxidizer tank with H2 O2 and a reentry capsule for the Moon soil samples. In the
liquid bipropellant rocket engine set-up the return rocket has a rocket engine using N2 H4 and
N2 O4 . Its lift-off mass from Moon is determined by the lander variant, because it is equivalent
to the payload of the lander. Tables 6.10 and 6.11 list the masses of the return rocket in the
different variants. The names of the mission variants are not to be confused with the actual
propellant choice for the return rocket. The return rocket always uses storable H2 O2 as the
oxidizer in the hybrid configuration and storable N2 H4 and N2 O4 for the liquid propulsion
versions. For the hybrids, the total mass is at its maximum with the three staged LOX spacecraft
variant, as the specific impulse with LOX is higher. The average specific impulse with the LOX
spacecraft is about 11 % higher with LOX than with H2 O2 . Considering the whole spacecraft
design, the payload is also 10 % to 20 % higher for the different variants with LOX. The liquid
bipropellant spacecraft have a much higher return rocket mass. This higher return rocket mass
supported by the lower structural mass for the propulsion system nearly doubles the capsule
mass of the return rocket compared to the hybrid rocket engine variants. The structural mass
of the hybrid rocket engines is very high, as the long cylinder is not an optimal shape for a
pressure vessel. Additionally the whole chamber needs to be insulated. An additional fact is
the lower efficiency, which is a possible improvement point for the calculated trajectories. If
it is proven that hybrid rocket engines reach a very high efficiency like liquids without adding
a lot of structural mass for better injection, preheating liquid oxidizers or mixing devices in
the chamber, the performance of the hybrid rocket engine might be improved a lot and could
compare more easily with a storable liquid propellant spacecraft. For the return rocket, the
same holds true like for the descend vehicle: If a mission is targeted at maximum performance
in terms of payload mass, sample return mass and therefore lowest structural mass and highest
specific impulse, the best variant is to use cryogenic liquid propellants with boil-off on the
landing vehicle and storable liquid propellants on the return vehicle. However, if low-cost and
low hazards during development, ground operations and rocket launch have a high priority
109
as well, the hybrid propellants offer a good alternative to the storable liquid propellants.
Table 6.10.: Return rocket mass overview: part 1

Variant Var. 1 Var. 2
Name H 2 O2 LOX H2 O2 LOX
Oxidizer H 2 O2 H 2 O2 H2 O2 H2 O2
Fuel PB-5% PB-5% PB-5% PB-5%
Total mass 1323 kg 1440 kg 1167 kg 1355 kg

Relative 100 % 108.8 % 88.2 % 102.4 %
Oxidizer mass 654 kg 715 kg 572 kg 670 kg

Fuel mass 87 kg 95 kg 76 kg 89 kg
RCS propellant mass 100 kg 100 kg 100 kg 100 kg
Propellant reserve 40 kg 40 kg 40 kg 40 kg
Tank mass 19 kg 20 kg 17 kg 19 kg
Engine mass 34 kg 37 kg 29 kg 34 kg
Feed system mass 30 kg 30 kg 29 kg 30 kg
Power supply 24 kg 24 kg 24 kg 24 kg
Thermal control 8 kg 8 kg 8 kg 8 kg
Electronics 8 kg 8 kg 8 kg 8 kg
Staging 15 kg 15 kg 15 kg 15 kg
Structure 99 kg 108 kg 88 kg 102 kg
Reentry capsule 205 kg 240 kg 161 kg 216 kg
Parachute 41 kg 48 kg 32 kg 43 kg
Heat protectection 62 kg 72 kg 49 kg 65 kg
Soil sample 45 kg 54 kg 34 kg 48 kg
The trajectory of the returning rocket vehicle is shown in Figure 6.16. The depicted snap-shot
shows the position of the spacecraft just at the change of the sphere of influence. Hence the
blue line representing the trajectory is switching the coordinate frame as well at that position.
The rocket is ascending from the lunar surface with a sharp gravity turn into a Moon escape
orbit with a periselen of about 10 km. The ∆V of this ascent from this launch position is
2600 m/s. The escape trajectory points in the contrary direction of the Moon’s velocity vector
in the Earth fixed coordinate system. The vehicles orbit when leaving the Moon’s sphere of
influence is a highly elliptical orbit with −2000 km perigee and 415 000 km apogee. The orbit
is still in prograde direction. The perturbation forces after leaving the sphere of influence of the
Moon are still high and hence when the spacecraft reaches an altitude of 10 000 km above the
Earth’s surface, the perigee has increased to 100 km (which was the optimization’s boundary
constraint) while the apogee has decreased slightly to 409 000 km. With a perigee at such low
altitudes a reentry is certain (even if more than revolution might be necessary, depending on
110
Table 6.11.: Return rocket mass overview: part 2
Variant Var. 3 Bi-liquid 1 Bi-liquid 2
Name H 2 O2 LOX Storable Cryo
Oxidizer H 2 O2 H 2 O2 N2 O4 N 2 O4
Fuel PB-5% PB-5% N2 H4 N2 H4
Total mass 1207 kg 1416 kg 1545 kg 1566 kg

Relative 91.3 % 107.0 % 116.8 % 118.4 %
Oxidizer mass 593 kg 703 kg 464 kg 469 kg

Fuel mass 79 kg 93 kg 332 kg 335 kg
RCS propellant mass 100 kg 100 kg 100 kg 100 kg
Propellant reserve 40 kg 40 kg 40 kg 40 kg
Tank mass 17 kg 20 kg 26 kg 26 kg
Engine mass 30 kg 36 kg 20 kg 20 kg
Feed system mass 30 kg 30 kg 31 kg 31 kg
Power supply 24 kg 24 kg 24 kg 24 kg
Thermal control 8 kg 8 kg 8 kg 8 kg
Staging 15 kg 15 kg 15 kg 15 kg
Reentry capsule 172 kg 233 kg 361 kg 373 kg
Parachute 34 kg 47 kg 72 kg 75 kg
Heat protection 52 kg 70 kg 109 kg 112 kg
Soil sample 37 kg 52 kg 84 kg 87 kg
the ballistic coefficient of the capsule). The capsule has heat protection to endure the high heat
loads and will descend on a parachute safely to ground. The rest of the return rocket structure
will burn up in the atmosphere.
6.2.7. Impact of Engine Efficiency
The engine efficiency has a direct impact on the specific impulse and therefore on the ∆V -
budget of each rocket stage. As shown in Table 6.4 a lower efficiency has been chosen for the
hybrid rocket engine variants, as it is harder to increase the efficiency of a hybrid rocket engine
compared to a liquid propellant engine, as the boundary layer combustion along the fuel grain
makes effective mixing of the propellants more difficult. However, in order to estimate the
effect of this lower efficiency on the hybrid rocket variants performance, the three staged
variant with LOX as the oxidizer has also been simulated and optimized with an combustion
efficiency of 95 %. Some parameters are compared for the three staged LOX using variant in
111
Figure 6.16.: Return rocket trajectory
Table 6.12. It is easily visible that the efficiency has a high impact. In comparison the return
rocket’s mass gains 10.4 % and the soil sample mass even 38.9 %. If a comparison to Tables 6.10
and 6.11 is done it is visible that the higher efficiency hybrid rocket variant even reach similar
values like the liquid biproppelant variants for the return rocket and the sample return mass.
However, the liquid variants are only two-staged. A three staged liquid propulsion system will
have an even higher payload capability at higher spacecraft costs.
Table 6.12.: Comparison of 90 % and 95 % combustion efficiency 3-staged variant using LOX
Efficiency Lander Return rocket Soil sample Specific impulse
propellant (LOX)
90 % 4734 kg 1440 kg 54 kg 3280 ms

95 % 4597 kg 1590 kg 75 kg 3460 ms
112
7. Conclusion
In the 60’s and 70’s, the space race led to several missions to the Moon, including impactors,
autonomous and manned landings. The Apollo program brought back many samples of lunar
soil. The Luna program included three successful sample return missions. Analytical results on
lunar soil back on the Earth are much better than in-situ analyzes. The variety and exactness
of instruments is much higher in laboratories than on a spacecraft. Acquisition of scientific
insights about the Moon’s surface improves the understanding of genesis, development and
evolution of the Moon, the Earth, other terrestrial planets and the inner solar system. The
Moon is seen as a technology testing ground for interplanetary space applications like manned
missions or in-situ ressource utilization. Soft landings on a celestial body require safe,
reliable and high-performance propulsion systems. For example, deep-throttling is needed
to land softly with the options of hovering over landing sites or evading obstacles. In the
past, only hazardous propellants were used on the spacecraft, taking advantage of storability,
monopropellant use and hypergolic ignition. In the last twenty-five years many missions to
the Moon showed, that the interest of science and space-faring organizations is very high to
conduct further missions. Several robotic, soft landing missions on the Moon are planned in
the near future by many different organizations or nations.
Hybrid rocket engines are a good choice for replacing hazardous, toxic liquid propellant
propulsion systems. Several fuel and oxidizer combinations provide similar specific impulse
like storable liquid propellants. Hybrid rocket engines share also the ability to stop, reignite
and throttle, which is very important for landing applications. The simple design of hybrid
rocket engines as well as the safe operations with its fuels and oxidizers make them an
attractive choice for in-space propulsion as they increase safety at ground testing and in space.
They also reduce costs of development and operation. The modeling of regression rates in
design tools and simulation programs is one of the key aspects in developing propulsion
systems with hybrid rocket engines. Regression models for classical and liquefying fuels
have been established in the past. The implementation of these models in ESPSS creates
a good tool to simulate hybrid rocket propulsion systems in different levels of details with
more detailed regression rates. The simplified model that relies solely on regression rate
coefficients always needs experiments to determine these coefficients beforehand. Models to
predict the performance of hybrid rocket engines with high precision are necessary to develop
hybrid propulsion systems for certain applications before experiments with that engine and
propellants are conducted. However, this does not come without a drawback: These regression
models based directly on the theory of a boundary layer flame need many input parameters.
In case of liquefying fuels, the models have to consider droplet entrainment as well and
113
need additional input parameters. These input parameters are mainly fuel properties like
viscosity or heat of gasification, but there are also obscure inputs like flame temperature or
flame mixture ratio. In general the regression rate models are very flexible and applicable for
predesign. To simulate a real rocket engine or propulsion system and compare measurements
with the simulation, the regression rate coefficients need to be determined by experiments.
Then the coefficients are used as an input for the simulation. The coupling of trajectory
simulation and optimization with rocket cycle analysis in ESPSS allows optimizing hybrid
rocket engines and spacecraft trajectories in a direct loop. This enables engineers to maximize
the performance not only of the hybrid rocket engine itself, but of all the propulsion system
and rocket stages depending on propellant combinations and trajectory requirements.
In order to verify if the used software is suitable to simulate, optimize and analyze a hybrid
propulsion sample return mission to the Moon, different software tools were tested. First,
ASTOS was used to reproduce the trajectories of Apollo 11 and Luna 24, two of the most
important missions to the Moon in the last century. Apollo 11 was the first manned mission to
land on another celestial body and Luna 24 was one of the first conducted robotic sample return
missions ever. For both missions detailed trajectory data is available. However, still not all the
necessary information to reconstruct the trajectories with full precision is attainable. With the
given information it is well possible to simulate, optimize and analyze the trajectories within
ASTOS. The optimization process helps out, where data is missing, because by fulfilling the
known boundary and path constraints, missing parameters are found by the optimizer. ASTOS
has proven to be a very powerful trajectory analysis tool to simulate translunar missions. The
second step was to test the newly developed interface for ASTOS to simulate and optimize
hybrid rocket engines. For verifying this interface’s functionality no reference mission was
available. Instead, an example was produced to evaluate the optimization result based on
experience. The first example is a hybrid propulsion micro launcher. Small launch vehicles
are a promising application for hybrid propulsion, as they offer low cost and high safety. A
hybrid micro launcher concept was set up with ASTOS and the tool for hybrid propulsion
simulation and optimization. The trajectory and the vehicle could be optimized reliably. While
the trajectory boundary constraints were reached, the hybrid rocket engines were optimized as
well as the staging of the launcher. By this the total vehicle mass for a payload of 25 kg could
be optimized from 17.3 t to 11.7 t. Although some parameters like the structure mass ratio
are just educated guesses, the trend was correct and the optimization tool for hybrid rocket
engines worked very well. Ultimately, a lunar lander was simulated and optimized with the
hybrid propulsion tool as well. In this case the final descent from a 100 km × 100 km orbit
around the Moon to the surface touchdown was considered. The tool worked as intended, the
hybrid propulsion system was optimized, as well as the control and the sizing of the lander. In
summary, the tools for simulating, optimizing and analyzing a hybrid propulsion lunar sample
return mission, namely ASTOS and the hybrid propulsion tool, were tested and verified to work
as intended.
114
An experimental campaign using a test chamber for long burn durations has been conducted
at the test bench M11.3 in Lampoldshausen. The test chamber uses gaseous oxygen in
combination with the paraffin-based fuel PB-5%. The mass flow is set to a reproducible
operation point with a control valve and the supply pressure. Several pressure sensors were
recording test data and the mass flow of the oxidizer was measured with a coriolis sensor.
The fuel usage was determined by weighing the paraffin fuel grain before and after each test.
Frequency analysis has been done for the different tests but no instabilities with high amplitude
were recorded. The regression rate depending on the propellant combination and the chamber
design were evaluated. A mixture ratio correction was necessary, since the mixture ratio was
too fuel rich for most of the tests. The chamber geometry was adjusted in order to decrease
the fuel surface and hence fuel mass flow. The reason for this was that the regression rate
was a bit higher than predicted and that the fuel’s front and back side added a fuel mass flow
portion that in earlier tests was always negligible. In this campaign however, the chamber
design included a fuel grain with a larger diameter compared to the fuel length and therefore
the fuel front and back side added a larger area relatively to the total solid fuel burning surface
area. A throttling test was conducted as well. However, the planned test duration of 40 s was
not reached due to a failure of the nozzle graphite insert. Finally two out of four pressure
levels were demonstrated in the test. Throttling itself did not inflict any problems on the test
chamber. Changing the pressure did not increase oscillations. However, the throttling with
a gaseous oxidizer is expected to be less prone to oscillations. Also the gas volume flow is
low compared to the pipe and engine volumes, which causes the pressure in the chamber to
always change in slow gradients. In order to test throttling under more realistic conditions a
higher oxygen flow or a change to liquid oxygen is necessary , which is currently not possible at
the test bench M11.3. The experimental results have also been compared to simulations done
with a corresponding EcosimPro/ESPSS model in order to validate the ESPSS models. The
experiments at the DLR test complex M11 showed that longer burn durations are possible but
also bring new challenges like material life cycles and insulation problems. In order to obtain
more data in the future, different engines should be tested with both GOX and LOX in order to
characterize regression rates in a wider range and also investigate deep-throttling further.
Eight different Moon sample return mission variants have been analyzed with ASTOS, the
analytical tool for hybrid rocket engines, and ESPSS models. The baseline for the launch of
the spacecraft is an Ariane 5 ECA launch into a direct Moon transfer with about 8.8 t payload
in a hyperbolical Moon flyby trajectory. The transfer time is about five days, which allows the
spacecraft to carry either storable or cryogenic propellants, where the cryogenic propellants
need to be cooled with boil-off. The landing trajectories of all compared spacecraft variants
are quite similar: At 100 km altitude the rocket engines are ignited to decelerate the spacecraft
into an elliptical orbit with a low periselen of about 10 km to 20 km. Close to the periselen,
the spacecraft is further decelerated and begins its descent onto the surface. At 1 km altitude,
the spacecraft initiates a hovering phase of one minute, which allows to choose the correct
landing spot. After the soft landing, ground operations with a rover, driller and manipulator
arm are conducted. When soil samples are filled into the return capsule, a return rocket is
115
launched towards the Earth and enters the Earth’s atmosphere. Six variants of the hybrid
propulsion spacecraft have been compared of which three use hydrogen peroxide and three
use liquid oxygen as the oxidizer. Additionally the staging was divided in three options: three-
staged, two-staged and two-staged with jettisoning of depleted hybrid combustion chambers.
The return rocket is always using hydrogen peroxide in the hybrid propulsion concepts and
is always a single stage rocket. In addition to the hybrid propulsion concepts, two liquid
bipropellant propulsion concepts have been analyzed. The first one is using storable hydrazine
and dinitrogen tetroxide in all stages whereas the second one is using cryogenic oxygen and
methane on the landing stages. Both concepts use hydrazine and dinitrogen tetroxide in the
return rocket. The soil sample mass is used as an indicator to compare the performance of
the concepts. The sample mass of the liquid bipropellant concepts are the highest. This is
due to the higher combustion efficiency and specific impulse as well as the smaller structural
mass of the propulsion system. With the cryogenic liquid bipropellant concept a soil sample
mass of 87 kg is reached while the storable liquid bipropellant concept reaches a sample mass
of 84 kg. The difference is quite small. The cryogenic concept has a more complex thermal
set-up and will probably require insulated tanks, pipes and valves but offers less hazardous
propellants compared with the toxic and carcinogenic hydrazine in the storable concept. The
largest soil sample mass with hybrid propulsion is reached with the three-staged concept using
liquid oxygen. The sample mass is 54 kg and therefore 62 % of the two-staged cryogenic
liquid bipropellant concept. The two-staged hybrid concept using LOX and jettisoning of
empty combustion chambers reaches a similar value of 52 kg. The lowest sample mass is
reached with the two-staged concept using hydrogen peroxide as an oxidizer with 34 kg soil.
From a narrowed perspective regarding performance, the hybrid propulsion concepts are
useful, but not competitive with an all liquid propellant concept. They could be improved
by raising the efficiency of combustion which was still conservatively low in these calculations.
A simulation with a three staged lander spacecraft using LOX and an increased efficiency of
95 % showed that the sample return mass is 75 kg, which is competitive to the liquid concepts.
This shows that if development of hybrid rocket engines in the near futures brings them to a
level, where high combustion efficiency is the standard, hybrid rocket engines become a real
alternative to liquid rocket engines. If the perspective is widened and other factors next to
performance are considered, hybrid propulsion might be a good choice even at the cost of lower
performance. Storable liquid propellants like dinitrogen tetroxide and hydrazine are very toxic
and carcinogenic. The advantage of the hybrid propellant combination with hydrogen peroxide
as an oxidizer is the much lower toxicity. This reduces cost drastically during storage, transport,
engine development and launch operations. The hybrid rocket engine still has the advantage
of needing only one fluid management system, which also reduces the cost and complexity of
the system. There will always be advantages and disadvantages to every options of hybrid and
liquid rocket engines and for every application all suitable solutions need to be considered and
compared.
116
Bibliography
[1] K. Tomilin. “Vorentwurf eines [9] Space Launch Report CZ-5-5-7 Data
Mondlandefahrzeugs mit hybrid sheet. Website. 2016. url: www .
Raketenantriebssystem für eine spacelaunchreport.com/cz5.html.
Probenrückführungsmission”. MA [10] R. Orloff. Apollo By The Numbers. 2016.
thesis. University of Stuttgart and
[11] AJ10 Wikipedia. website. Mar. 2016.
DLR Lampoldshausen, 2016.
url: en.wikipedia.org/wiki/AJ10.
[2] Mario Kobald, Christian Schmier, Ul-
[12] Aerozine 50 Wikipedia. website. Mar.
rich Fischer, Konstantin Tomilin, and
2016. url: en . wikipedia . org / wiki /
Anna Petrarolo. “A Record Flight of
Aerozine_50.
the Hybrid Sounding Rocket HEROS
3”. In: 31st ISTS Japan. 2017. [13] European Commission. REACH.
Website. Mar. 2016. url: ec . europa .
[3] A. Konopliv, R. Park, D. Yuan, S.
eu / environment / chemicals / reach /
Asmar, M. Watkins, and et al. “The
reach_en.htm.
JPL lunar gravity field to spherical
harmonic degree 660 from the GRAIL [14] Descent Propulsion System Wikipedia.
Primary Mission”. In: Journal of website. Mar. 2016. url: en.wikipedia.
Geophysical Research: Planets 118 org/wiki/Descent_Propulsion_System.
(2013), pp. 1415–1434. [15] KTDU-417. Website. 2017. url:
[4] NASA. Moon Fact Sheet. Website. Mar. www. space - ru . com / russian - rocket-
2016. url: nssdc . gsfc . nasa . gov / engines/rocket-engine-ktdu-417.
planetary/factsheet/moonfact.html. [16] KRD-61. Website. 2017. url: www .
[5] ESA Concurrent Design Facility. Lunar space - ru . com / russian - rocket -
Exploration Study - Cargo Transporta- engines/rocket-engine-krd-61.
tion System. Tech. rep. 2005. [17] E. Betts and R. Frederick. “A Historical
[6] NASA Space Science Data Coordinated Systems Study of Liquid Rocket En-
Archive. Website. 2016. url: nssdc . gine Throttling Capabilities”. In: AIAA
gsfc.nasa.gov/nmc/. Joint Propulsion Conference. 2010.
[7] KTDU-5A, encyclopedia astronautica. [18] A. Chandler Karp, B. Nakazono, J.

Website. 2016. url: www. astronautix . Benito, R. Shotwell, D. Vaughan, and
com/engines/ktdu5a.htm. G. Story. “A Hybrid Mars Ascent
Vehicle Concept for Low Temperature
[8] TD-339 (SURVEYOR VERNIER)
Storage and Operation”. In: 52nd
ROCKET ENGINE. Website. 2016.
AIAA/SAE/ASEE Joint Propulsion
url: heroicrelics . org / info / surveyor /
Conference. 2016.
surveyor-vernier.html.
117
[19] M. J. Chiaverini. “Fundamentals [28] M. Kobald, C. Schmierer, H. Ciezki,
of Hybrid Rocket Combustion S. Schlechtriem, E. Toson, and L.
and Propulsion”. In: ed. by M. J. De Luca. “Evaluation of Paraffin-based
Chiaverini and K. K. Kuo. Progress Fuels for Hybrid Rocket Engines”.
in Astronautics and Aeronautics. In: 50th AIAA/ASME/SAE/ASEE Joint
American Institute of Aeronautics and Propulsion Conference. 2014.
Astronautics, 2007. [29] A. Petrarolo, M. Kobald, and C.
[20] A. Takahashi and T. Shimada. Schmierer. “Characterization of
“Essentially Non-explosive Propulsion Advanced Hybrid Rocket Engines”. In:
Paving a Way for Fail-Safe Space 6th European Conference for Aeronautics
Transportation”. In: 31st ISTS Japan. and Space Sciences. Krakow, Poland,
2017. 2015.
[21] Hybrid Rocket. Website. Apr. 2015. url: [30] Mario Kobald, Christian Schmierer,
en . wikipedia . org / wiki / Hybrid % 5C _ Helmut Ciezki, Stefan Schlechtriem,
rocket. Elena Toson, and Luigi De Luca.
[22] Yen-Sen Chen, J. Lin, and S. Wei. “De- “Viscosity and Regression Rate of
velopment of an Airbreathing Hybrid Liquefying Hybrid Rocket Fuels”. In:
Rocket Booster”. In: 31st ISTS Japan. Journal of Propulsion and Power 33.5
2017. (2017), pp. 1245–1251.
[23] C. Schmierer, K. Tomilin, J. Steelant, [31] Y. Tang, S. Chen, and L. T. DeLuca.
and S. Schlechtriem. “Analysis and “Mechanical Modifications of Paraffin-
Preliminary Design of a Hybrid based Fuels and the Effects on Com-
Propulsion Lunar Lander”. In: Space bustion Performance”. In: Propellants,
Propulsion Conference. 2016. Explosives, Pyrotechnics (2017).
[24] Alfred Wright. Nitric Acid/Nitrogen [32] C. Schmierer, M. Kobald, K. Tomilin,
Tetroxide Oxidizers. Martin Marietta U. Fischer, M. Rehberger, and S.
Corporation, 1977. Schlechtriem. “HEROS - Sounding
[25] M. A. Karabeyoglu, D. Altman, and Rocket Development by the HyEnD
B. J. Cantwell. “Combustion of lique- Project”. In: 6th European Conference
fying hybrid propellants: Part 1, gen- for Aeronautics and Space Sciences.
eral theory”. In: Journal of Propulsion 2015.
and Power 18.3 (2002), pp. 610–620. [33] M. Kobald, C. Schmierer, and A. Pe-
[26] M. A. Karabeyoglu and B. J. Cantwell. trarolo. “Test campaign of a 10000 N
“Combustion of liquefying hybrid Hybrid Rocket Engine”. In: 6th Eu-
propellants: Part 2, Stability of liquid ropean Conference for Aeronautics and
films”. In: Journal of Propulsion and Space Sciences. Krakow, Poland, 2015.
Power 18.3 (2002), pp. 621–630. [34] M. Kobald, U. Fischer, K. Tomilin, A.
[27] M. Kobald. “Combustion phenomena Petrarolo, and C. Schmierer. “Hybrid
of advanced hybrid rocket fuels”. PhD Experimental Rocket Stuttgart: A
thesis. University of Stuttgart, 2015.
118
Low-Cost Technology Demonstrator”. [42] G. Parissenti, M. Pessana, E. Gaia, and
In: Journal of Spacecraft and Rockets et al. “Throttleable hybrid engine for
55.2 (2018), pp. 484–500. planetary soft landing”. In: 4th Euro-
[35] M. Kobald et al. “Sounding Rocket pean Conference for aerospace sciences.
"HEROS" - A Low-Cost Hybrid 2011.
Rocket Technology Demonstrator”. [43] R. Hahn, G. Waxenegger-Wilfing,
In: 53rd AIAA/ASME/SAE/ASEE Joint C. Schmierer, and J. Deeken. “Cycle
Propulsion Conference. 2017. Analysis and Feasibility Evaluation of
[36] J. Dyer et al. “Status Update Report for Pump Fed Hybrid Propulsion System”.
the Peregrine 100km Sounding Rocket In: SpacePropulsion Conference. 2018.
Project”. In: 44th Joint Propulsion Con- [44] G. Marxman and et al. “Fundamentals
ference and Exhibit. AIAA 2008-4892. of hybrid boundary combustion”. In:
Hartford, CT, 2008. Progress in Astronautics and Aeronau-
[37] T. Knop, J. Wink, and et al. “Failure tics, 1964.
Mode Investigation of a Sorbitol-based [45] G. Zilliac and M. A. Karabeyoglu.
Hybrid Rocket Flight Motor for the “Hybrid Rocket Fuel Regression
Stratos II Sounding Rocket”. In: 51st Rate Data and Modeling”. In:
AIAA Joint Propulsion Conference. 42nd AIAA/ASME/SAE/ASEE Joint
2015. Propulsion Conference and Exhibit.
AIAA-2006-4504. Sacramento, CA,
[38] O. Verberne, A. Boiron, M. Faenza, and
2006.
B. Haemmerli. “Development of the
North Star Sounding Rocket: Getting [46] M. J. Chiaverini, N. Serin, D. Johnson,
ready for the first demonstration Y. Lu, K. Kuo, and G. Risha. “Regres-
Launch”. In: 51st AIAA Joint sion Rate Behavior of Hybrid Rocket
Propulsion Conference. 2015. Solid Fuels”. In: Journal of Propulsion
and Power 16.1 (Jan. 2000).
[39] Hyprogeo. Website. Dec. 2015. url:
www.hyprogeo.eu. [47] A. Petrarolo. “Optical Analysis of Hy-
brid Rocket Combustino with Decom-
[40] Karine Odic et al. “HYPROGEO
position Methdos”. In: Space Propul-
Hybrid Propulsion: Project Objectives
sion. 2016.
& Coordination”. In: Space Propulsion
Rome. 2016. [48] M. Kobald, I. Verri, and S.
Schlechtriem. “Theoretical and
[41] M. Faenza, O. Verberne, S. Tonetti,
Experimental Analysis of Liquid
S. Cornara, and T. Langener. “Hybrid
Layer Instabilities in Hybrid Rocket
Propulsion Solutions for Space Debris
Engines”. In: CEAS Space Journal 7.1
Remediation Applications”. In:
(Mar. 2015), pp. 11–22.
Industrial Days ESA ESTEC. 2016.
[49] M. A. Karabeyoglu, B. J. Cantwell,
and J. Stevens. “Evaluation of
Homologous Series of Normal-
Alkanes as Hybrid Rocket Fuels”.
119
In: 41st AIAA/ASME/SAE/ASEE Joint [58] Chemical Equilibrium with
Propulsion Conference and Exhibit. Applications. Website. 2015. url:
AIAA-2005-3908. Tucson, AZ, 2005. www.grc.nasa.gov/WWW/CEAWeb/.
[50] R. Nigmatulin, B. Nigmatulin, [59] M. M. Fernandez. “Propellant Tank
Y. A. Khodzaev, and V. Kroshilin. Pressurization Modeling for a Hybrid
“Entrainment and Deposition Rates Rocket”. MA thesis. Rochester, NY:
in a Dispersed-Film Flow”. In: Rochester Institute of Technology,
International Journal of Multiphase 2009.
Flow 22.1 (1996), pp. 19–30. [60] J. E. Zimmerman, B. S. Waxman,
[51] J. Anderson. Introduction to Flight. B. J. Cantwell, and G. Zilliac.
1979. Chap. 4.17, p. 210. “Review and Evaluation of
[52] R. A. Gater and M. R. L. L’Ecuyer. Models for Self-Pressurizing
“A Fundamental Investigation of the Propellant Tank Dynamics”. In:
Phenomena that Characterize Liquid 49th AIAA/ASME/SAE/ASEE Joint
Film Cooling”. In: International Journal Propulsion Conference and Exhibit.
of Heat and Mass Transfer 13.3 (1970), AIAA 2013-4045. San Jose, CA, 2013.
pp. 1925–1939. [61] EcosimPro. Website. 2016. url: www .
[53] M. Adachi and T. Shimada. “Liquid ecosimpro.com/products/ecosimpro.
FFilm Instability Analysis of Liquiefy- [62] European Space Propulsion System
ing Hybrid Rocket Fuels Under Super- Simulation, ESPSS. Website. 2016. url:
critical Conditions”. In: AIAA Journal. www.ecosimpro.com/products/espss.
2015. [63] ESPSS User Manual, Volume 1. Empre-
[54] A. A. Chandler, E. T. Jens, B. J. sarios Agrupados. 2016.
Cantwell, and G. S. Hubbard. [64] R.C. Reid, J.M. Prausnitz, and T.K.
“Visualization of the Liquid Layer Sherwood. The Properties of Gases and
Combustion of Paraffin Fuel at Liquids. McGraw-Hill, 1977.
Elevated Pressures”. In: 63rd
[65] P.A. Ramakrishna, P.J. Paul,
International Astronautical Congress.
and H.S. Mukunda. Sandwich
Naples, Italy, 2012.
Propellant Combustion: Modeling
[55] P. Gath. “CAMTOS - A Software Suite and Experimental Comparison. Vol. 29.
Combining Direct and Indirect Tra- Proceedings of the Combustion
jectory Optimization Methods”. PhD Institute. 2002.
thesis. University of Stuttgart, 2002.
[66] Nicholas P. Cheremisinoff. Handbook of
[56] H. Curtis. Orbital Mechanics for Engi- Engineering Polymeric Materials. Mar-
neering Students. Elsevier, 2010. cel Dekker, Inc., 1997.
[57] C. Schmierer. Modellierung einer
Mondmission mit Hybridantrieben.
Tech. rep. DLR Lampoldshausen and
University of Stuttgart, 2015.
120
[67] Abdul Khan, Abhijit Dey, Javaid [76] Patrick Chai and Alan Wilhite. “Cryo-
Athar, and Arun Kanti Sikder. genic thermal system analysis for or-
“Calculation of enthalpies of bital propellant depot”. In: Acta Astro-
formation and band gaps of polymeric nautica 102 (2014).
binders”. In: RSC Advances (2014). [77] NASA Aims to Keep Moon’s Skies Junk-
[68] H.G. Ang and S. Pisharath. Energetic Free. Website article. Oct. 2008. url:
Polymers: Binders and Plasticizers for https://www.space.com/6031- nasa-
Enhancing Performance. WILEY-VCH, aims-moon-skies-junk-free.html.
2012. [78] NASA. Human Artifacts. Website.
[69] Christian Schmierer, Mario Kobald, Jo- Mar. 2016. url: history . nasa . gov /
han Steelant, and Stefan Schlechtriem. humanartifacts.html.
“Combined Trajectory Simulation and [79] NASA. NASA’s GRAIL Creates Most
Optimization for Hybrid Rockets us- Accurate Moon Gravity Map. Website.
ing ASTOS and ESPSS”. In: 31st ISTS 2016. url: www . nasa . gov / mission _
Japan. 2017. pages / grail / multimedia / pia16587 .
[70] Proton-K Wikipedia entry. Website. html.
May 2017. url: https://en.wikipedia. [80] Moon Landing, Wikipedia. Website.
org/wiki/Proton-K. 2016. url: en . wikipedia . org / wiki /
[71] Blok-D Wikipedia entry. Website. May Moon_landing.
2017. url: https://en.wikipedia.org/ [81] LADEE Spacecraft Overview. Website.
wiki/Blok_D. 2016. url: spaceflight101.com/ladee/
[72] Luna 24 information page. Website. spacecraft-overview/.
May 2017. url: http : / / www . [82] Chang’e 3. Website. 2016. url:
russianspaceweb.com/luna24.html. spaceflight101.com/change/change-
[73] Jonas Breitinger. “Development and 3/.
Test of an Experimental Hybrid Rocket [83] NASA. Apollo Images. Website. Mar.
Engine for Orbital Application”. 2017. 2016. url: history.nasa.gov/ap11ann/
[74] Discrete Fourier Transform in Python kippsphotos/apollo.html.
with Fast Fourier Transform algorithm. [84] Capcomespace. Lunar probe images.
SciPy. url: https : / / docs . scipy. org / Website. Mar. 2016. url: www .
doc / numpy - 1 . 13 . 0 / reference / capcomespace.net/dossiers/espace_
generated/numpy.fft.fft.html. US/apollo/annexes/15_les_sondes_
[75] M. A. Karabeyoglu, G. Zilliac, B. J. lunaire_2.htm.
Cantwell, S. DeZilwa, and P. Castel- [85] Skyatnight Magazine. Gallery of
lucci. “Scale-up tests of high regres- Luna 24. Website. Mar. 2016. url:
sion rate paraffin-based hybrid rocket www . skyatnightmagazine . com /
fuels”. In: Journal of Propulsion and gallery / cosmonauts - exhibition -
Power 20.6 (2004), pp. 1037–1045. opens-science-museum.
121
A. Moon Missions
Figure A.1.: Historical landing sites on the Moon’s surface, [78]
Figure A.2.: Moon’s gravity map by GRAIL, [79]
122
Table A.1.: Soviet lunar landing missions from 1959 to 1976 [6, 80]
Mission Launch date Mass Spacecraft Objective Result
Luna 1 02.01.1959 361 kg Ye-1 Impact Flyby
Luna 2 12.09.1959 390 kg Ye-1A Impact Success
Luna 3 04.10.1959 270 kg Ye-2A Flyby Success
Sputnik 25 04.01.1963 1500 kg n/a Landing LEO
Luna 4 02.04.1963 1422 kg Ye-6 Landing Flyby
Cosmos 60 12.03.1965 6530 kg n/a Landing LEO
Luna 5 09.05.1965 1475 kg Ye-6 Landing Impact
Luna 6 08.06.1965 1440 kg Ye-6 Landing Flyby
Luna 9 31.01.1966 1580 kg Ye-6 Landing Success
Luna 13 21.12.1966 1580 kg Ye-6M Landing Success
Luna 15 13.07.1969 5700 kg Ye-8-5 Sample Return Impact
Cosmos 300 23.09.1969 5600 kg Ye-8-5 Sample Return LEO
Cosmos 305 22.10.1969 5600 kg Ye-8-5 Sample Return LEO
Luna 16 12.09.1970 5600 kg Ye-8-5 Sample Return Success
Luna 17 10.11.1970 5700 kg Ye-8 Lunar Rover Success
Luna 18 02.09.1971 5750 kg Ye-8-5 Sample Return Impact
Luna 20 14.02.1972 5727 kg Ye-8-5 Sample Return Success
Luna 21 08.01.1973 5950 kg Ye-8 Lunar Rover Success
Luna 23 28.10.1974 5800 kg Ye-8-5 M Sample Return Landing
Luna 24 09.08.1776 5800 kg Ye-8-5 M Sample Return Success
123
Table A.2.: US lunar landing missions from 1962 to 1972 [6, 80]
Mission Launch date Mass Propulsion Objective Result
Unmanned
Ranger 4 23.04.1962 331 kg solid rocket Landing Impact

Ranger 7 28.07.1964 367 kg 
 Impact Success
N2 H4 mono-


Ranger 8 17.02.1965 367 kg 
 prop Impact Success


Ranger 9 21.03.1965 367 kg  Impact Success

Surveyor 1 30.05.1966 995 kg 


 Landing Success

Surveyor 2 20.04.1967 ~1000 kg Landing Impact





Surveyor 3 20.04.1967 ~1000 kg



 solid rocket Landing Success
and MMH


Surveyor 4 14.07.1967 ~1000 kg Landing Signal lost
& MON-10




Surveyor 5 08.09.1967 ~1000 kg 


 vernier Landing Success

Surveyor 6 07.11.1967 ~1000 kg Landing Success






Surveyor 7 07.01.1968 ~1000 kg Landing Success


Manned

Apollo 11 16.07.1969 45 702 kg 

 Sample Ret. Success


Apollo 12 24.11.1969 45 870 kg 



Apollo 14 31.01.1971 46 305 kg Sample Ret. Success



see section 2.2.2


Apollo 15 26.07.1971 48 599 kg 











124
Table A.3.: Missions to Moon from 1990 to 2010, data from [6]
Mission, Function Propulsion system Launcher
Organisation Payload
Hiten, 1990, Orbiter, impactor: Gravity N2 H4 monoprop.: M-3SII

ISAS (now JAXA) assist and aero braking 8x23 N, 4x3 N 197 kg
Clementine, 1994, Orbiter: Imaging, 489 N N2 O4 & N2 H4 , Titan II-G

BMDO, NASA gravimetry, mineralogy RCS: 12 mono-N2 H4 227 kg
Luna Prospector, Orbiter, impactor: Surface N2 H4 monoprop.: Athena 2

1998 NASA mapping, gravimetry, etc. 6x22 N 296 kg
SMART-11 , 2003, Orbiter, impactor: Ion Hall-effect 70 mN Xenon Ariane 5
ESA drive demonstrator RCS: 8 mono-N2 H4 366.5 kg
Kaguya, 2007, Orbiter, impactor: Global 500 N N2 O4 & N2 H4 H2-A

JAXA survey of the Moon RCS: 12x20 N biprop. 2885 kg
Chang’e 1, 2007, Orbiter, impactor: n/a LM 3A

CNSA Technology demonstrator, 2350 kg
imaging
Chandrayaan-1, Orbiter, impactor: Surface Bipropellant propulsion PSLV-XL

2008, ISRO topograhpy etc. system 1380 kg
LRO2 , LCROSS3 Orbiter, impactor4 : N2 H4 monoprop. Atlas V

2009, NASA Surface mapping, resource 1846 kg /
detection 621 kg
Chang’e 2, 2010, Orbiter: High resolution n/a LM 3C

CNSA images 2480 kg
1 Small Missions for Advanced Research in Technology-1
2 Lunar Reconnaissance Orbiter
3 Lunar Crater Observation and Sensing Satellite
4 The Centaur upper stage crashed on Moon too
125
Table A.4.: Missions to Moon from 2010, data from [6]
Mission, Function Propulsion system Launcher
Organisation Payload
ARTEMIS1 1&2, Orbiter: Analysis of Monoprop. Delta II, 5 satellites,

2011, NASA magnetic phenomena 126 kg each
GRAIL A & GRAIL Orbiter, Impactor: High 22 N thruster, Delta II,

B, 2011, quality gravitational field N2 H4 monoprop., low 202.4 kg
NASA mapping energy transfer each
LADEE2 , 2013, Orbiter, Impactor: Bipropellant MON-3 & Minotaur-V,

NASA Analysis of atmoshpere MMH [81] 383 kg
and dust around Moon
Chang’e 3, 2013, Lander, Rover: Technology throttleable engine, Long March 3B,
CNSA demonstrator, optical bipropellant UDMH & ~3800 kg
investigations N2 O4 [82]
Chang’e 5-T1 3 , Flyby, Earth Reentry n/a Long March 3C

2013, Capsule, Orbiter:
CNSA Technology demonstrator
1 Acceleration, Reconnection, Turbulence and Electrodynamics of the Moon’s Interaction with the Sun
2 Lunar Atmosphere and Dust Environment Explorer
3 Piggyback mission 4M by OHB with radio experiment
Table A.5.: Planned lander missions to Moon [80]

Mission Organisation Function Intended launch1
Chang’e 5 CNSA Sample Return 2017

Google Lunar X Prize Several teams Landers & Rovers Planned for 20172
Chandrayaan-2 ISRO Lander, Rover 2018
SELENE-2 JAXA Lander, Rover 2018
Chang’e 4 CNSA Lander, Rover Before 2020
Chang’e 6 CNSA Sample Return 2020
Exploration Mission-2 NASA Manned Landing 2023
Lunar Mission One UK Lander 2024
Luna-Glob 1 Roscosmos Lander Late 2020’s
Internat. Lunar Network NASA & others Landed stations n/a
Moon village ESA & others Manned base n/a
1 No launch date is officially confirmed and fixed
2 Two teams have launch contracts for 2017: Team Moon Express plans to launch on Electron and
SpaceIL plans a Falcon 9 launch
126
Figure A.3.: Overview of lunar missions in the past and future
127
Table A.6.: Launch / orbit data of Apollo 11
1st stage S-IC Time 0:00:00

Burn time 168 s Max. acceleration 3.94 g
Thrust 33600 kN Max. velocity 2.4 km/s

2nd stage S-II Time 0:02:42

3rd stage S-IVB – 1st burn Orbit insertion Time 0:09:09
Earth Orbit Data Time 0:11:49
Perigee 183.2 km Apogee 185.9 km
Descending Node 123.1◦ Inclination 32.5◦
3rd stage S-IVB – 2nd burn Translunar Injection Time 2:44:16

Transposition, docking and extraction Time 3:15:23

The Command/Service Module separated from the S-IVB, turned by 180◦ and docked with the
Lunar Module again. Afterwards the CSM and LM would together separate from S-IVB.
CSM Evasive maneuver Time 4:40:02

Burn time 2.93 s ∆V -6 m/s
Thrust 91 kN
CSM Lunar orbit insertion Time 75:49:51

Burn time 357.5 s ∆V -889.254 m/s
Altitude cutoff 60.1 n mi
CSM Orbit circularization Time 80:11:37

Burn time 16.9 s ∆V -48.4 m/s
Aposelen 66.1 n mi Periselen 54.5 n mi
CSM/LM undocking Time 100:12:00

Burn time 9s ∆V 0.8 m/s
LM Descent orbit insertion Time 101:36:14
Burn time 30 s ∆V -23.3 m/s
128
Thrust 43.9 kN max (throttle 10%-60%)

LM Powered descent Time 102:33:05

Burn time 756.4 s
LM Ascent Time 124:22:01

Burn time 434.9 s ∆V 1850.2 m/s
Thrust 14.3 kN
Aposelen 48 n mi Periselen 9.4 n mi
LM Coelliptic sequence init. Time 125:19:35

Burn time 47 s ∆V 51.5 m/s
LM Const. differential height Time 126:17:50

Burn time 17.8 s ∆V 6.1 m/s
LM Terminal phase init. Time 127:03:52

LM Midcourse corrections
∆V #1 0.3 m/s Time 127:18:31
∆V #2 0.46 m/s Time 127:33:31
LM Terminal phase finalize Time 127:46:10

CSM/LM Docking Time 128:03:00

CSM/LM Ascent Stage Jettisoned Time 130:09:31
CSM/LM Separation Time 130:30:01
CSM Transearth injection Time 135:23:42

129
CSM Midcourse corrections Time 150:29:57.4

CM/SM Separation Time 194:49:12.7
Figure A.4.: Landers of Apollo and Luna Moon programs, [83, 84]
130
Figure A.5.: Luna 24 lander module replica, [85]
131
B. Hybrid Rocket Research Groups
Figure B.1.: Research on hybrid propulsion in Europe
132
Table B.1.: Hybrid rocket propulsion research groups
No. Marker Organisation Type, Nation Field of Work
Europe
1 1 HyEnD University of Student group, Hybrid sounding rockets

Stuttgart Germany (32 km altitude record),
paraffin-based fuels
2 2 WARR University of Munich Student group, Hybrid sounding rockets
Germany
3 3 ERIG University of Student group, Hybrid sounding rockets
Braunschweig Germany
4 4 ZARM University of Bremen Student group, Hybrid sounding rockets
Germany
5 5 Aquasonic City Uni. of Student group, Hybrid sounding rockets
Bremen Germany
6 6 HyCOMET City Uni. of Student group, Hybrid sounding rockets
Augsburg Germany
7 7 DLR Lampoldshausen Research Institute, Advanced paraffin-based
Germany fuels, fundamental research,
hybrid rocket engine tests
8 8 DLR Braunschweig Research Institute, Simulation, advanced fuel
Germany geometries
9 9 DLR Trauen Research Institute, Hybrid rocket engine tests
Germany
10 10 CISAS, University of Padova University, Italy Fundamental research,
numerics, tests, HYPROGEO
11 11 Politecnico di Milano University, Italy Fundamental research,
numerics, tests
12 12 Sapienza - University of University, Italy Fundamental research,
Rome numerics, tests
13 13 Starthclyde University University, UK Fundamental research,
HYPROGEO
14 14 Surrey University University, UK Fundamental research,
alternative geometry
15 15 PERSEUS (in coop. with Student project, Hybrid sounding rocket,
CNES, ONERA) France student education
16 16 DARE, University of Delft Student group, Hybrid sounding rockets
Netherlands
17 17 Nammo Raufoss Industry, Norway Northstar Microlauncher
Project, HYPROGEO
133
18 18 Space Forest Gdynia Industry, Poland Small scale hybrid rocket

engines
19 19 Institute of Aviation Warsaw Research Institute, Fundamental research, small
Poland scale tests
20 20 SPARTAN: (Coordinator: Industry, EU Hybrid propulsion lander
Thales Alenia) demonstrator
21 21 HYPROGEO: (Coordinator: Industry, EU Hybrid propulsion system
ASL) for orbit applications
22 22 HYPROGEO: MOOG Industry, EU
23 23 HYPROGEO: Von Karman Research Institute,
Institute EU
24 24 Bloodhound SSC Project Industry, UK Speed Record Car
25 25 University of Naples University, Italy
World
26 30 Pennsylvania State University, USA

University
27 31 Stanford University University, USA
28 32 Purdue University Hybrid Student group, USA
Propulsion
29 33 HYSOR University of Student group, USA
Colorade, Boulder
30 34 NASA FSGC Hybrid Motor Student Project,
Rocket Competition USA
31 35 Sierra Nevada Corporation Industry, USA
32 36 ORBITEC Industry, USA
33 37 Space Propulsion Group Industry, USA
34 38 Virgin Galactic / Scaled Industry, USA
Composites
35 39 NASA Ames (Peregrine Research Institute,
Rocket Project) USA
36 40 NASA JPL (Mars Ascent Research Institute,
Vehicle) USA
37 41 NASA Glenn (Mars Ascent Research Institute,
Vehicle) USA
38 42 NASA Marshall (Mars Research Institute,
Ascent Vehicle) USA
134
39 50 Keldysh Research Center Research Institute,

Russia
40 60 JAXA ISAS Sagamihara Research Institute,
Japan
41 61 JAXA Tsukuba Research Institute,
Japan
42 62 Tokyo Metropolitan University, Japan
University
43 63 Tokai University University, Japan
44 64 Saitama Institute of University, Japan
Technology
45 65 Hokkaido University of University, Japan
Science
46 66 Gas Turbine Div. Kawasaki University, Japan
Heavy Industries
47 67 Advanced Rocket Research Research Institute,
Center Taiwan
48 70 ISRO VSSC Research Institute,
India
49 71 ISRO LPSC Research Institute,
India
50 80 Gilmour Space Tech Industry, Singapore
Singapore
51 81 Gilmour Space Tech Industry, Australia
Australia
52 82 HyPER, University of Student group,
Adelaide Australia
53 90 INPE LCP Cachoeira Research Institute,
Paulista Brasil
54 91 University of Brasilia University, Brasil
55 92 Hybrid Propulsion Team Student group,
University of Brasil Brasil
56 93 University of Kwazulu-Natal University, South
Africa
135
C. ESPSS: Enthalpy Balance at the Fuel Grain
In the ESPSS model (function "GasSol_exch") the heat of formation of the decomposed fuel is
notated with dH. For clarity in this document the notation is ∆Hfθ,dec . The heat of formation
can be given in literature for solid, liquid or gaseous state. Usually for fuels that are solid at
ambient conditions, the heat of formation is given for the solid state as ∆Hfθ,sp . The heat of
formation of atomic, gaseous carbon and hydrogen is unequal to zero. Only solid carbon and
molecular hydrogen have a heat of formation of zero. The heat of formation of the solid fuel
at room temperature is named LHsp in ESPSS, which is confusing, since LH stands for latent
heat. Latent heats however are the heats of phase changes like vaporization and melting, not
decomposition. Therefore in this document the heat of formation is again renamed to ∆Hfθ,sp .
In the model for the hybrid combustor the the decomposed fuel mass flow is calculated with
one of the regression rate laws. The heat flow from the flame zone to the surface zone therefore
is directly bound to the fuel mass flow. In the ESPSS 3.2 component the heat flow Q̇vs is
calculated as follows and its positive direction is from the surface zone to the flame zone:

Q̇vs = ṁf ∆Hfθ,dec − Q̇trans = ṁf ∆Hfθ,dec − Af hc Tg − Tvap (C.1)
The first term incorporates the heat flow bound to the mass flow of decomposed fuel. The
second term Qtrans is the heat flow by conduction and radiation from the flame to the surface
zone. One condition that is mandatory to be fullfilled is, that this heat flow must be equal to
the heat flows necessary to melt, vaporize and pyrolyse the fuel, as well as heating up the solid
fuel grain. The heat flow to the solid fuel grain is:

Q̇sp = Af hsp Tvap − Tg (C.2)
The following equation can be obtained:
Q̇trans + ṁf ∆Hfθ,sp = ṁf ∆Hfθ,dec + Q̇sp (C.3)
On the left side the transfered heat due to conduction and convection is added to the heat of
formation, which is set free for decomposition of the solid fuel. For most of the solid fuels,
the heat of formation is negative, since the decomposition is endothermic. The right side
represents the enthalpy that flows out of the zone: The heat bound to the mass flow in form
of heat of formation and the heat flow to the solid fuel. If the heat of formation of the solid
fuel is unknown, and only the gaseous or liquid heat of formation is known, it can be simply
136
rearranged:
∆Hfθ,sp = ∆Hfθ,liq + hmelt = ∆Hfθ,gas + hmelt + hvap (C.4)
In that case, additionally the latent heat for melting and vaporization is accounted for. This
equations are not always fullfilled in the ESPSS model.
The following problems were identified with the ESPSS code:
• As Qtrans is calculated simply with heat conduction and ṁf is calculated with the
regression model, there is no link between both and therefore the enthalpy balance is
not kept.
• The heat transfer coefficient hc from the gas flow in the port to the vapor layer is
calculated via a Nusselt correlation with the gas flow properties of the node:
k 1
hc = N ul16 + N ut16 16 (C.5)
D
with the laminar Nusselt number N ul = 4, the turbulent Nusselt number N ut , heat
conductivity k and the flow diameter D. The turbulent Nusselt number is calculated
with the Reynolds number Re and the Prandtl number P r:
N ut = 0.023Re0.8 P r 0.4 (C.6)
This simplified heat transfer coefficient which is calculated for a homogeneous tube flow
cannot represent the inhomogeneous heat transfer situation in the hybrid combustion
chamber with many different processes.
• The temperature Tvap was identified with a saturation temperature Tsat , which does not
exist for all solid fuels. The name vaporization temperature Tvap is much clearer, and
different to the saturation temperature, if it exists for the fuel. It can be assumed that the
temperature close to the solid fuel is the vaporization temperature. For liquefying fuels
also a melting temperature Tmelt can be determined. Respectively the temperature close
to the surface of a liquefying fuel is close to the melting temperature.
The correct way to calculate the enthalpy balance and heat flows depends on the regression rate
law. For the "stdHybrid" regression law, the regression rate is based on experiments and can be
assumed as correct approximation. In this case the heat flows are justified by experimental
data. For the models of Marxman and Zilliac/Karabeyoglu the heat flow and enthalpy
conservation was the starting point of deriving the equations. Therefore also here the heat
flow is correct, when it is matched to the necessary heat of vaporization and decomposition.
The heat flow from the vapor zone to the solid stays the same, and therefore the heat flow from
the gas to the vapor zone can be modified:

Q̇trans = ṁf ∆Hfθ,dec − ∆Hfθ,sp + Q̇sp (C.7)
137
The enthalpy flow Q̇vs is then:
Q̇vs = −ṁf ∆Hfθ,sp + Q̇sp (C.8)
If this equation is compared to Equation C.1, it is visible, that the wrong heat of formation is
used in the original code. The heat of formation of the decomposed fuel ∆Hfθ,dec is not occuring
in the heat flow, because this heat is first flowing from the hot gas to the fuel grain in order to
decompose the fuel but then is going back to the gas flow bound to the fuel mass flow. In
this context also the problems with the advanced regression rate model "vapModel" can be
highlighted. This model actually accounts for energy balance and calculates the vaporization
of fuel with the enthalpy flow into the boundary layer. The mistake is, that this model does only
respect heat of formation of the solid fuel but not the heat of formation of the decomposition.
However, that enthalpy must be provided too, if it is unequal to zero (like in H-atoms).
Therefore it produces regression rates much higher than observed in experiments. Solved for
the mass flow in the vaporization model "vapModel" this results in:
Q̇trans − Q̇sp
ṁf = (C.9)
∆Hfθ,dec − ∆Hfθ,sp
The regression model "vapModel" was improved:

hc Tg − Tvap − hsp Tvap − Ts
ṁf = Af (C.10)
∆Hfθ,dec − ∆Hfθ,sp
Instead of the mean gas temperature Tg , the flame sheet temperature Tc is used to calculate
the heat flow. The fuel mass flow is highly dependent on the heat transfer coefficient hc , which
increases with viscosity. So the regression rate is increasing with the chamber temperature.
As mentioned before, the heat transfer coefficient of a tube flow with isothermic gas is not
precise for a hybrid combustor. For the other regression laws, where the regression rate is
not computed with the heat balance, the heat transfer from the gas to the vapor zone can be
calculated directly with the needed heat for vaporization and decomposition of the fuel mass
flow.
138
D. Lunar Lander Optimization
A hybrid propulsion lunar lander with a sample return rocket was introduced in a previous
work [23]. The mission is designed to be launched with an Ariane 5 or comparable launcher
to a direct Moon orbit transfer. The initial Moon orbit of the vehicle is a hyperbolic flyby
with a low periselen of about 100 km. At the periselen the first of two hybrid rocket kick
stages is ignited to lower the periselen to about 10 km altitude. Then the emtpy stage is
ejected and after a coast the second kick stage is ignited at the new periselen of 10 km. The
second stage is reducing the horizontal and vertical velocity close to zero, then it is ejected
and the lunar lander starts its final vertical descent. The mass of the lander and the kick
stages in total is 8300 kg with paraffin-based fuel and 95% hydrogen peroxide as oxidizer.
Previously the simulation and optimization of the lander and its trajectory was done using
only rough estimations of the propulsion systems’ performance. A constant performance was
assumed. Now the simulation and optimization has been renewed with the new ASTOS and
ESPSS interface for hybrid propulsion systems. One change that has been done to the original
model was the engine set-up. The four parallel engines on each kick stage were replaced by
one larger hybrid rocket engine. This change was possible, since new regression rate results
in experiments showed, that it can be feasible to have one large long burning hybrid rocket
engine. The optimum in terms of structural mass to fuel ratio has to be found in the future.
Table D.1 displays the lunar lander mass before and after optimization of the trajectory. In total
200 kg of H2 O2 are included for RCS thrusters. The payload includes the return rocket as well
as power supply and instrumenation of the lander. It can be seen that the payload and masses
were not improved a lot by the optimizer. The reason for this is, that the inputs were already
taken from the optimized trajectory of the previous work. However the trajectory itself was
optimized, as well as the rocket engine’s geometries which are listed in Table D.2. Figure D.1a
shows the altitude and periapsis altitude of the lander. Both initial guess and optimized curves
are printed. The initial guess is not very close to a realistic descent on the Moon’s surface.
The descent rate at the end is increasing, which can be also seen in Figure D.1b. Even though
the initial guess was not very good, the optimization with ASTOS found an optimal trajectory
which would meet the conditions of a soft touch down. Figure D.1c shows a detailed view of
the final landing. It features the end of the second kick stage’s burn until 3300 s, a short coast
of 25 s and a nearly vertical descent until touch down with a flight path speed of about 1.5 m/s.
A first impression of the possible landing module with a return rocket is pictured in Figure
D.1d. The combustion chambers and nozzles as well as the H2 O2 tanks are in true scale both
for the landing craft and the return rocket. The diameter of the truss is about 2.5 m and the
height is about 3.1 m.[69]
139
Table D.1.: Initial and optimized lunar lander masses[69]
Stage Component Optimizable Initial mass Optimized
1st Propellant Yes 2300 kg → 2302 kg

Tank structure Fixed ratio 4 % of oxidizer mass
Total Stage Mass 2852 kg → 2851 kg
2nd Propellant Yes 2550 kg → 2497 kg

Tank structure Fixed ratio 4 % of oxidizer mass
3rd Propellant Yes 290 kg → 171 kg

Payload Yes 1217 kg → 1200 kg
Table D.2.: Lunar lander engine parameters[69]

Stage Component Initial value Optimized value
1st Fuel length 850 mm → 830 mm

Fuel diameter 720 mm → 720 mm
Throat diameter 104 mm → 104 mm
Expansion ratio 180 → 182.5
2nd Fuel length 850 mm → 849 mm

Fuel diameter 720 mm → 740 mm
Throat diameter 104 mm → 100.6 mm
Expansion ratio 180 → 190
3rd Fuel length 155 mm → 160 mm

Fuel diameter 140 mm → 144.5 mm
Throat diameter 18 mm → 15.1 mm
Expansion ratio 190 → 200
140
(a) Altitude and periapsis of the lander before and after optimization[69] (b) Speed and descent rate of the lander before and after optimization[69]
141
(c) Velocities and altitude of the lander in the final landing sequence[69] (d) Preliminary illustration of the lunar lander module with a return rocket[69]
Figure D.1.: Hybrid lunar lander data

E. Test Results
142
143
Figure E.1.: Test 2 and 5 measurements

144
Figure E.2.: Test 2 and 5 measurements

F. Hybrid Propulsion Lunar Lander
H2 O2 was used as the oxidizer. Figure F.1 shows the thrust on the left with an exemplary
thrust level of 13 kN. The engine is ignited in the simulation after 10 s. Before that the oxidizer
tank is pressurized to the feed pressure which is illustrated on the right. The pressurization
tank’s pressure is 300 bar. Figure F.2 illustrates the propellant masses. On the left the liquid
oxidizer in the tank is compared to the solid propellant in all four engines. On the right the
pressurization gas is shown, which is reduced in the pressurization tank and flowing into the
main oxidizer tank. The mass flows of the stage is shown in Figure F.3 on the left. The injector
and nozzle mass flow are for a single engine with a mixture ratio of about 7.9. Figure F.4 depicts
the regression rate and the fuel port diameter over time.
Figure F.1.: Thrust and pressure simulation for hybrid propulsion kick stage
Figure F.2.: Mass over time for hybrid propulsion kick stage
145
Figure F.3.: Mass flow, IV ac and c∗ for hybrid propulsion kick stage
Figure F.4.: Regression rate and fuel port diameter for hybrid propulsion kick stage
146

2019 06 25 - DissSchmDruckv3

Uploaded by

Copyright:

Available Formats

2019 06 25 - DissSchmDruckv3

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2019 06 25 - DissSchmDruckv3

Uploaded by

Copyright:

Available Formats

Analysis of a Hybrid Propulsion

Lunar Sample Return Mission

A thesis accepted by the Faculty of Aerospace Engineering and Geodesy of

Dipl.-Ing. Christian Schmierer

Committee Chair: Prof. Dr.-Ing. Stefan Schlechtriem

Insitute of Space Systems

Table of Contents III

List of Tables XII

2. Sample Return Missions to the Moon 4

3. Theory of Hybrid Rocket Engines 15

5. Experiments with a Hybrid Rocket Engine Demonstrator 72

6. Hybrid Propulsion Lunar Mission Analysis 89

A. Moon Missions 122

B. Hybrid Rocket Research Groups 132

C. ESPSS: Enthalpy Balance at the Fuel Grain 136

D. Lunar Lander Optimization 139

E. Test Results 142

F. Hybrid Propulsion Lunar Lander 145

2.1. Exemplary mission profile for a Lunar sample return, [1] . . . . . . . . . . . . . 7

3.1. Hybrid rocket engine layout,[21] . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

4.1. Schematic of ESPSS model for regression rate comparison . . . . . . . . . . . . . 41

5.1. Flow plan of the M11.3 test bench . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

6.1. Moon transfer trajectory of a Ariane 5 launched spacecraft . . . . . . . . . . . . . 90

A.1. Historical landing sites on the Moon’s surface, [78] . . . . . . . . . . . . . . . . . 122

B.1. Research on hybrid propulsion in Europe . . . . . . . . . . . . . . . . . . . . . . 132

D.1. Hybrid lunar lander data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

E.1. Test 2 and 5 measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

2.1. ∆V [ ms ] requirements to the Moon from [1] . . . . . . . . . . . . . . . . . . . . . . 6

3.1. Propellant options for hybrid rocket engines, [23] . . . . . . . . . . . . . . . . . . 17

4.1. Equations of analytical hybrid rocket engine model . . . . . . . . . . . . . . . . . 35

5.1. Sensor positions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

6.1. Orbital elements after launch with the Ariane 5 ECA . . . . . . . . . . . . . . . . 91

B.1. Hybrid rocket propulsion research groups . . . . . . . . . . . . . . . . . . . . . . 133

D.1. Initial and optimized lunar lander masses[69] . . . . . . . . . . . . . . . . . . . . 140

Formulas D Flow diameter m

N2 O4 Dinitrogen tetroxide F Thrust N

CH , CH0 Stanton number with and l Length m

cf Skin-friction coefficient and vaporization

d Fuel grain diameter m N ul , N ut Nusselt number (laminar,

Q̇vs Heat transfer from the W boundary layer

κl Thermal diffusivity m2 ent Entrainment

λl Thermal conductivity W f Final (burn end)

µg Gas flow viscosity Pa s fu Fuel

µl Liquid layer viscosity Pa s g Gas

µ Gravitational parameter m3 inj Injector (cavity)

2.1. Orbit Mechanics of Space Flight to the Moon

2.1.1. The Moon

2.1.2. Transfer Orbits to the Moon

2.1.4. Return Trajectory

Table 2.1.: ∆V [ ms ] requirements to the Moon from [1]

Injection into a transfer orbit to the Moon 3200 3200

Total 9400 9260

2.2. Historic Missions

2.2.1. Scientific Gains of Sample Return

2.2.2. Propulsion System

Table 2.3.: Propulsion system parameters of Apollo 11

Mass (total) [kg] 119119 30329 10281 4495

2.2.3. Mission Phases and Trajectory

Table 2.4.: Propulsion system parameters of Luna 16