Master of Science Thesis

Assisted-Launch Performance Analysis

Using Trajectory and Vehicle Optimization

Jan Vandamme

6-12-2012

Faculty of Aerospace Engineering Delft University of Technology

Assisted-Launch Performance Analysis
Using Trajectory and Vehicle Optimization

Master of Science Thesis

For obtaining the degree of Master of Science in Aerospace

Engineering at Delft University of Technology

Jan Vandamme

6-12-2012

Faculty of Aerospace Engineering Delft University of Technology

Cover figure:
Pegasus Engine Ignites after Drop from B-52 Mothership (modified from [34])
 Delft University Of Technology
Department Of
Space Exploration

The undersigned hereby certify that they have read and recommend to the Faculty
of Aerospace Engineering for acceptance a thesis entitled Assisted-Launch Perfor-
mance Analysis by Jan Vandamme in partial fulfillment of the requirements for the
degree of Master of Science.

Dated: 6-12-2012

Chairman of thesis comittee:

Prof. ir. B.A.C. Ambrosius

Thesis supervisor:
Dr. ir. E. Mooij

Member of thesis comittee:

Ir. B.T.C. Zandbergen

Member of thesis comittee:

Dr.-Dipl.Ing. S. Erb
 Summary

At the time of writing 55 years have passed since the launch of Sputnik 1. While space-
flight has changed significantly in many ways since then, further improvement is still
needed. This thesis study aims to provide an answer to the question if launch-assist sys-
tems could provide such an improvement and if so, how the launch vehicles should be
designed.
Launch assists systems come in all shapes and sizes. They can be provided with
an assist from the air with help of a fly-back booster, rocket plane, airplane, zeppelin
or a balloon or from the ground with a rail or a gun. Most of the studies about the
concept of assisted launch focus on the first leg, the assist itself and what is available
for use on the market today or what could be built with current technology. However,
most launch-assist platforms available today have been built for other purposes and are
in no way optimal for launch assist. Therefore, it is interesting to investigate whether a
launch-assist system purpose built for the job is needed and what performance require-
ments it should meet. There are two factors that come into play when answering this
question. These factors are the cost of building and operating both the assist platform
and the launch vehicle as well as the systems total performance. This thesis focuses on
the performance of the entire system by assuming certain performance parameters of the
assist platform. All of the concepts proposed have one thing in common. By giving the
launch vehicle an initial velocity and/or an initial altitude (often referred to in this study
as the launch-assist parameters), they bring the launcher closer to its destination, ener-
getically speaking, and they might diminish atmospheric and gravitational losses. This
thesis studys main objective is to shed some clarity onto the relationship between the
launch-assist parameters, the launch-vehicle performance and the launch-vehicle param-
eters. Therefore, the research question for this work is formulated as follows:

How does a launch assist affect the performance and the design of a launch vehicle?

To answer the research question, extensive analysis is performed. Both a set of launch-
vehicle parameters and a trajectory is optimized for a variety of given sets of launch-assist
parameters. The simulation of the trajectory takes place by propagating the state vector

v
vi Summary

by means of an integrator. This trajectory is dictated by the guidance program, which

consists out of a pitch program that follows a 4th -degree polynomial. The trajectory is
subjected to several constraints, which are implemented with current-day technology in
mind. The problem of optimizing all parameters of that trajectory as well as those of the
launch vehicle is tackled by the use of a differential-evolution algorithm.
The results provide an overview for systems that could be used today with respect
to the technology available. This means that no results are obtained from simulated
launch assists that have a higher than 2,000 m/s assist speed or an assist altitude higher
than 40,000 m. The results that were obtained showed that the launch-performance
parameters have a large effect on the both the design and the performance of the vehicle.
The payload-to-initial-mass ratio increased more than six-fold for hydrolox launchers and
more than seven-fold for kerolox launchers between the no-assist case and the case of
maximum assist. A high assist velocity affects the launch-vehicle performance more than
a high-altitude assist almost to the extent that 1 m/s in velocity is as valuable as 100 m
in altitude. Although the results widely vary from case to case, the hydrolox launchers
outperform the kerolox launchers at every launch-assist combination. In general, small
dimensions, low thrust-to-weight ratios, high chamber pressures and low exit pressures
were often found to be the characteristics of an optimal launcher. In addition, there
are also multiple cases where the optimal trajectory of the launcher is such that a non-
negligible amount of lift is generated or that only a very low heat flux occurs so that no
fairing protection is needed for the satellite.
The study is intended to be used as a tool in future concept studies and gives
significant insight on important design considerations regarding both the launch-assist
system as well as the launch vehicle itself. Through this work, it is my hope that the
reader is convinced that current day launchers, which are often an update of an older
launcher or designed around an engine that just so happens to be available, have room for
improvement left by tackling the design process from a different angle. The main concepts
that the study aims to support are the railed launch-assist concepts, the airplane/rocket
plane launch-assist concepts and to some extent low V flyback boosters.
 Acknowledgements

This document represents the findings from a master thesis study on assisted launch
performance conducted as a part of my Master of Science at the Delft University of Tech-
nology. The work focused on establishing a means to accurately model a wide array of
launchers and analyze their performance given certain initial conditions. The work docu-
ments how this is modeled in a simulation program that optimizes the trajectory and the
launch vehicle given certain required launch assist and vehicle parameters, in addition the
work documents the results extensively and provides a detailed analysis.
I would like to take this moment to thank my supervisor Dr. Ir. Erwin Mooij, who has
helped and assisted me during this year and has showed me many places where I should
shift my attention to. I would also like to thank my friends of the 9th floor both those
who have stayed and those who have left for the help they offered me from time to time
and more importantly, for the countless enjoyable massive music quiz sessions. When,
not if, I will succeed in obtaining the mythical perfect score, I am sure that I will hear an
echo of the perfect score chants ringing in my memories. Last but not least, I would like
to offer my sincere gratitude to my family and my close friends for helping me enjoying
my life outside of work, they are the kind of people that one should never take for granted.

Delft, The Netherlands Jan Vandamme

6-12-2012

vii
viii Acknowledgements
 Contents

Summary v

Acknowledgements vii

List of Figures xv

List of Tables xviii

Nomenclature xix

1 Introduction 1
1.1 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Layout of the thesis report . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Flight Mechanics 5
2.1 Reference frames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 State variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1 Orbital elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Equations of motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3.1 Euler Angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3.2 Frame transformations . . . . . . . . . . . . . . . . . . . . . . . . . 13

3 Models 17
3.1 Environment models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1.1 Atmospheric model . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1.2 Gravitational model . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2 Aerodynamic model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2.1 Aerodynamic forces . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.3 Launch-vehicle model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

ix
x Contents

3.3.1 Performance Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.3.2 Mass Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3.3 Launch Vehicle Model Verification . . . . . . . . . . . . . . . . . . 37
3.4 Thrust model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.5 Analysis of the velocity budget . . . . . . . . . . . . . . . . . . . . . . . . 46
3.6 Guidance model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.6.1 Initial conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.6.2 Coasting phase(s) and circularization . . . . . . . . . . . . . . . . . 49
3.6.3 Verification of the Pitch Polynomial . . . . . . . . . . . . . . . . . 50
3.7 Trajectory constraint model . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.7.1 Bending load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.7.2 Axial acceleration . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.7.3 Angle of attack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.7.4 Heat Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.7.5 Altitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.7.6 Sidenote . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

4 Numerical Integration 57
4.1 Runge-Kutta-4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.2 Selection of the time step . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

5 Optimization Techniques 61
5.1 General optimization problem . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.1.1 Global optimization methods . . . . . . . . . . . . . . . . . . . . . 62
5.1.2 Comparison of algorithms in literature. . . . . . . . . . . . . . . . 65
5.1.3 Problem formulation . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.1.4 Comparison between Differential Evolution and Particle Swarm Op-
timization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.1.5 Settings of the DE algorithm . . . . . . . . . . . . . . . . . . . . . 66

6 Tool Implementation and External Software 69

6.1 Missile DATCOM 97 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
6.2 Tudat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
6.3 PaGMO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
6.4 CEA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.5 Code architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Contents xi

7 Results 75
7.1 Introduction to the results . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.2 Discussion of the launch-vehicle parameters . . . . . . . . . . . . . . . . . 79
7.2.1 Fuel type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
7.2.2 Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
7.2.3 Slenderness and diameter . . . . . . . . . . . . . . . . . . . . . . . 81
7.2.4 Nozzle exit-to-tank ratio / Nozzle exit diameter . . . . . . . . . . . 83
7.2.5 Oxidizer-to-fuel ratio . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7.2.6 Chamber Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7.2.7 Exit Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
7.2.8 Initial flight-path angle . . . . . . . . . . . . . . . . . . . . . . . . 88
7.2.9 Thrust-to-weight ratio . . . . . . . . . . . . . . . . . . . . . . . . . 89
7.2.10 Tank mass and engine mass . . . . . . . . . . . . . . . . . . . . . . 90
7.2.11 Specific impulse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
7.2.12 Drag Loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7.2.13 Gravity Loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
7.3 Notable flights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
7.3.1 No fairing trajectories . . . . . . . . . . . . . . . . . . . . . . . . . 95
7.3.2 Low-assist flights . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
7.3.3 Altitude anomalies . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
7.3.4 Effect of change in pitch angle . . . . . . . . . . . . . . . . . . . . 98
7.4 Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

8 Conclusions and Recommendations 103

8.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
8.2 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
8.2.1 For further study on the subject: . . . . . . . . . . . . . . . . . . . 106
8.2.2 For launch-vehicle design: . . . . . . . . . . . . . . . . . . . . . . . 108

A CEA Tables 109

B Main Results 123

C Figures 137

Bibliography 151
xii Contents
 List of Figures

2.1 Representation of the Earth centered inertial reference frame [33] . . . . . 6

2.2 Representation of the vehicle-carried normal Earth reference frame for a
spherical Earth. Adapted from [33] . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Body fixed reference frame for an aircraft [33] . . . . . . . . . . . . . . . . 8
2.4 Orbital elements [38] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.5 Cartesian and spherical coordinate systems [38] . . . . . . . . . . . . . . . 10

3.1 Pressure (a) and density (b) to geometric height profiles generated by the
model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2 A comparison of the temperature (in Kelvin) - geometric height profile of
the model (a) and from literature (b) [8], which also shows the changes
between US 1976 and US 1962. . . . . . . . . . . . . . . . . . . . . . . . . 21
3.3 Location of P and Q [23] . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.4 Propellant tanks depiction . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.5 Hydrolox engine weight according to the original method (Eq. (3.44)), the
optimized method (Eq. (3.68)) and the actual data points of the 3 hydrolox
engines according to [15]. . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.6 Kerolox engine weight according to the original method (Eq. (3.43)), the
optimized method (Eq. (3.69)) and the actual data points of the 3 hydrolox
engines according to [15]. . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.7 Comparison of the two polynomial types that obtained the best objective
function value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.8 Example of how the axial-acceleration constraint is met for hydrolox vehicles. 53
3.9 Example of how the angle-of-attack constraint is met for hydrolox vehicles.
The legend contains case identifiers using the h/V notation. . . . . . . . 53
3.10 Example of how the heat-flux constraint is met. The legend contains case
identifiers using the h/V (type of fuel) notation. . . . . . . . . . . . . . 54

5.1 Global and local optima of a two-dimensional function [60]. . . . . . . . . 63

xiii
xiv List of Figures

5.2 Evolution of the objective value of the hydrolox maximum assist case (a),
showing the early lead of the PSO algorithm being overtaken by the DE
algorithm and the evolution of the most difficult case, the no assist kerolox
case (b), where the PSO algorithm does not find a solution during 10,000
generations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.3 3D visualization of the results obtained from different weight and crossover
probability DE algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

6.1 Hydrolox engine weight according to the original method (Eq. (3.44)), the
optimized method obtained through PaGMO (Eq. (3.68)), the actual data
points of the 3 hydrolox engines according to [15] and the Macht function
from Microsoft Excel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.2 Flowchart of the simulation tool. . . . . . . . . . . . . . . . . . . . . . . . 74

7.1 Altitude vs time plots of all optimized hydrolox simulations . . . . . . . . 77

7.2 Cubic spline fit of the payload-mass to initial-mass ratio of optimized hy-
drolox launchers and launch trajectories for different combination of assist
velocities and altitudes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
7.3 Cubic spline fit of the payload-mass to initial-mass ratio of optimized
kerolox launchers and launch trajectories for different combination of assist
velocities and altitudes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
7.4 Length (m) of the optimized hydrolox vehicles (a) and kerolox vehicles (b). 80
7.5 Slenderness of the optimized hydrolox vehicles (a) and kerolox vehicles (b). 81
7.6 Diameter (m) of the optimized hydrolox vehicles (a) and kerolox vehicles
(b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
7.7 Nozzle exhaust-to-tank diameter ratio of the optimized hydrolox vehicles
(a) and kerolox vehicles (b). . . . . . . . . . . . . . . . . . . . . . . . . . . 84
7.8 Nozzle exhaust diameter (m) of the optimized hydrolox vehicles (a) and
kerolox vehicles (b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
7.9 Oxidizer-to-fuel ratio of the optimized hydrolox vehicles (a) and kerolox
vehicles (b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7.10 Chamber pressure (bar) of the optimized hydrolox vehicles (a) and kerolox
vehicles (b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
7.11 Dimensionless exhaust velocity as a function of pressure ratio . . . . . . . 87
7.12 Exit pressure (Pa) of the optimized hydrolox vehicles (a) and kerolox ve-
hicles (b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
7.13 Initial flight-path angle (deg) of the optimized hydrolox vehicles (a) and
kerolox vehicles (b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
7.14 Exhaust velocity of the optimized hydrolox vehicles (a) and kerolox vehicles
(b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
7.15 Drag loss (m/s) of the optimized hydrolox vehicles (a) and kerolox vehicles
(b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
7.16 Altitude vs Velocity plots of various launch cases for optimized hydrolox
vehicles (a) and kerolox vehicles (b). . . . . . . . . . . . . . . . . . . . . . 93
7.17 Gravity loss (m/s) of the optimized hydrolox vehicles (a) and kerolox ve-
hicles (b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
List of Figures xv

7.18 Summation of the gravity and drag losses (m/s) of the optimized hydrolox
vehicles (a) and kerolox vehicles (b). . . . . . . . . . . . . . . . . . . . . . 94
7.19 Lift-to-weight ratio of liquid hydrogen vehicles utilizing different launch
assist altitudes and velocities. The legend identifies the specific simulations
with their respective assist parameters documented in the following format:
h [m] / V [m/s]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
7.20 Altitude vs Time plots of hydrolox vehicles (a) and kerolox vehicles (b),
indicating some anomalies. . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
7.21 Comparison plots between optimal trajectories with an initial flight-path
angle of zero and with optimized flight-path angle. Top left: Altitude-
velocity plot, top right: Pitch angle - time plot, bottom left: Bending load
- time plot and bottom right: lift-to-weight ratio - time plot. . . . . . . . 99
7.22 Mean of the total standardized effect of parameters on the variance of the
model with indications of the maximum and minimum values . . . . . . . 101

C.1 Mean of the first-order standardized effect of the launch-vehicle parame-

ters on the variance of the model with indications of the maximum and
minimum values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
C.2 Total standardized effect of parameters on the variance of the model with
various hypercube sample sizes. . . . . . . . . . . . . . . . . . . . . . . . . 138
C.3 Total effect sensitivity indices of exit pressure parameter at various values
of optimized exit pressure. . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
C.4 Inertial velocity plots of all optimized hydrolox simulations . . . . . . . . 139
C.5 Inertial velocity plots of all optimized kerolox simulations . . . . . . . . . 139
C.6 Eccentricity plots of all optimized hydrolox simulations . . . . . . . . . . . 140
C.7 Eccentricity plots of all optimized kerolox simulations . . . . . . . . . . . 140
C.8 Altitude plots of all optimized hydrolox simulations . . . . . . . . . . . . . 141
C.9 Altitude plots of all optimized kerolox simulations . . . . . . . . . . . . . 141
C.10 Total acceleration plots of all optimized hydrolox simulations . . . . . . . 142
C.11 Total acceleration plots of all optimized kerolox simulations . . . . . . . . 142
C.12 Pitch angle plots of all optimized hydrolox simulations . . . . . . . . . . . 143
C.13 Pitch angle plots of all optimized kerolox simulations . . . . . . . . . . . . 143
C.14 Angle of attack plots of all optimized hydrolox simulations . . . . . . . . . 144
C.15 Angle of attack plots of all optimized kerolox simulations . . . . . . . . . 144
C.16 Logarithmic dynamic pressure plots of all optimized hydrolox simulations 145
C.17 Logarithmic dynamic pressure plots of all optimized kerolox simulations . 145
C.18 Logarithmic bending load plots of all optimized hydrolox simulations . . . 146
C.19 Logarithmic bending load plots of all optimized kerolox simulations . . . . 146
C.20 Logarithmic heat flux plots of all optimized hydrolox simulations . . . . . 147
C.21 Logarithmic heat flux plots of all optimized kerolox simulations . . . . . . 147
C.22 Position plots in Cartesian coordinates of all optimized hydrolox simula-
tions in the equatorial plane . . . . . . . . . . . . . . . . . . . . . . . . . . 148
C.23 Position plots in Cartesian coordinates of all optimized kerolox simulations
in the equatorial plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
C.24 Lift to weight plots of all optimized hydrolox simulations. . . . . . . . . . 149
C.25 Lift to weight plots of all optimized kerolox simulations. . . . . . . . . . . 149
C.26 Altitude vs velocity plots of all optimized hydrolox simulations. . . . . . . 150
C.27 Altitude vs velocity plots of all optimized kerolox simulations. . . . . . . . 150
xvi List of Figures
 List of Tables

3.1 Defined reference levels and gradients [11] . . . . . . . . . . . . . . . . . . 18

3.2 Numerical values of low degree gravitational coefficients of the Earths grav-
ity field [37] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.3 Missile DATCOM database configuration matrix . . . . . . . . . . . . . . 26
3.4 Comparison launch vehicle catalogue values and vehicle model for first stages 42
3.5 Comparison launch vehicle catalogue values and vehicle model for first stages 43
3.6 Comparison launch vehicle catalogue values and vehicle model for first stages 44
3.7 Comparison launch vehicle catalogue values and vehicle model for first
stages with updated engine . . . . . . . . . . . . . . . . . . . . . . . . . . 45

4.1 Comparison of the results from different time step RK4 runs . . . . . . . 60

5.1 Comparison of the results obtained from different weight and crossover
probability DE algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

7.1 Length, diameter and slenderness of various first stages [15] . . . . . . . . 82

7.2 Drag loss, gravity loss and performance comparison of various launcher
cases with either an optimized initial flight-path angle or an initial flight-
path angle of zero. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

A.1 Molar mass of hydrogen-oxygen mixtures . . . . . . . . . . . . . . . . . . 109

A.2 Molar mass of hydrogen-oxygen mixtures . . . . . . . . . . . . . . . . . . 110
A.3 Molar mass of hydrogen-oxygen mixtures . . . . . . . . . . . . . . . . . . 110
A.4 Molar mass of hydrogen-oxygen mixtures . . . . . . . . . . . . . . . . . . 111
A.5 Ratio of specific heats of hydrogen-oxygen mixtures . . . . . . . . . . . . . 111
A.6 Ratio of specific heats of hydrogen-oxygen mixtures . . . . . . . . . . . . . 112
A.7 Ratio of specific heats of hydrogen-oxygen mixtures . . . . . . . . . . . . . 112
A.8 Ratio of specific heats of hydrogen-oxygen mixtures . . . . . . . . . . . . . 113

xvii
xviii List of Tables

A.9 Chamber temperature of hydrogen-oxygen mixtures . . . . . . . . . . . . 113

A.10 Chamber temperature of hydrogen-oxygen mixtures . . . . . . . . . . . . 114
A.11 Chamber temperature of hydrogen-oxygen mixtures . . . . . . . . . . . . 114
A.12 Chamber temperature of hydrogen-oxygen mixtures . . . . . . . . . . . . 115
A.13 Molar mass of kerosene-oxygen mixtures . . . . . . . . . . . . . . . . . . . 115
A.14 Molar mass of kerosene-oxygen mixtures . . . . . . . . . . . . . . . . . . . 116
A.15 Molar mass of kerosene-oxygen mixtures . . . . . . . . . . . . . . . . . . . 116
A.16 Molar mass of kerosene-oxygen mixtures . . . . . . . . . . . . . . . . . . . 117
A.17 Ratio of specific heats of kerosene-oxygen mixtures . . . . . . . . . . . . . 117
A.18 Ratio of specific heats of kerosene-oxygen mixtures . . . . . . . . . . . . . 118
A.19 Ratio of specific heats of kerosene-oxygen mixtures . . . . . . . . . . . . . 119
A.20 Ratio of specific heats of kerosene-oxygen mixtures . . . . . . . . . . . . . 119
A.21 Chamber temperature of kerosene-oxygen mixtures . . . . . . . . . . . . . 120
A.22 Chamber temperatures of kerosene-oxygen mixtures . . . . . . . . . . . . 120
A.23 Chamber temperatures of kerosene-oxygen mixtures . . . . . . . . . . . . 121
A.24 Chamber temperatures of kerosene-oxygen mixtures . . . . . . . . . . . . 121
A.25 Chamber temperatures of kerosene-oxygen mixtures . . . . . . . . . . . . 122

B.1 Optimized launch vehicle parameters and characteristics using kerolox pro-
pellants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
B.2 Optimized launch vehicle parameters and characteristics using kerolox pro-
pellants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
B.3 Optimized launch vehicle parameters and characteristics using kerolox pro-
pellants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
B.4 Optimized launch vehicle parameters and characteristics using kerolox pro-
pellants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
B.5 Optimized launch vehicle parameters and characteristics using kerolox pro-
pellants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
B.6 Optimized launch vehicle parameters and characteristics using kerolox pro-
pellants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
B.7 Optimized launch vehicle parameters and characteristics using hydrolox
propellants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
B.8 Optimized launch vehicle parameters and characteristics using hydrolox
propellants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
B.9 Optimized launch vehicle parameters and characteristics using hydrolox
propellants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
B.10 Optimized launch vehicle parameters and characteristics using hydrolox
propellants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
B.11 Optimized launch vehicle parameters and characteristics using hydrolox
propellants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
B.12 Optimized launch vehicle parameters and characteristics using hydrolox
propellants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
 Nomenclature

Latin symbols

a Semi-major axis [m]

C Thrust-to-weight ratio [-]
CD Drag coefficient [-]
CF Thrust coefficient [-]
CL Lift coefficient [-]
CS Side force coefficient [-]
c Characteristic velocity [m/s]
D Drag [N]
D Diameter [m]
E Youngs modulus [Pa]
e Eccentricity [-]
F Force vector [N]
G Universal gravitational constant 6.668 1011 [Nm2 /kg2 ]
g Gravitational acceleration vector [m/s2 ]
h Geopotential height [m]
h Height [m]
Habl Heat of ablation [J/kg]
Isp Specific Impulse [s]
i Inclination [rad]
i, j, k Cartesian unit vectors [-]
j Number of equality constraints [-]
Jn,m Gravitational scaling parameter [-]
k Boltzmann constant 1.380622 1023 [Nm/K]
L Lift [N]
L Length [m]
LM Molecular scale temperature gradient [K]
M Molecular weight [kg/kmol]
M Mass [kg]
M Moment [Nm]
m Mass flow [kg/s]
n Number of constraints [-]

xix
xx Nomenclature

NA Avogadro constant 6.022169 1026 [kmol1 ]

p Pressure [Pa]
Pn,m (x) Legendre function [-]
Q Heat flux [W/m2 ]
q Dynamic pressure [Pa]
R Specific gas constant [J/mol K]
r Distance [m]
r Radius [m]
R Gas constant 8.31432 103 [Nm/kmol K]
RN Nose radius [m]
Sref Aerodynamic reference surface [m2 ]
T Thrust [N]
t Thickness [m]
t Time [s]
TM Molecular scale temperature [K]
TT OT Vacuum thrust [kg]
u Control variable several
V Velocity vector [m/s]
V(P) Gravitational potential vector [J/kg]
v Mean Anomaly [-]
x n-vector of independent variables several
x State variable [-]
x, y, z Cartesian coordinates [m]
Z Geometric height [m]

Greek symbols

Angle of attack [rad]

Side-slip angle [rad]
Vandekerckhove function [-]
Ratio of specific heats [-]
Flight-path angle [rad]
V Change in velocity [m/s]
Latitude [rad]
 Measure of the curvature of tan() profile [-]
 Correction factor [-]
Longitude [rad]
Pitch angle [rad]
n,m Gravitational orientation parameter [-]
Slenderness ratio [-]
Standard gravitational parameter [m3 /s2 ]
Density [kg/m3 ]
Stress [Pa]
Bank angle [rad]
c Time interval needed for vaporization [s]
Longitude [rad]
 xxi

Poisson ratio [-]

Vernal equinox or First point of Aries [-]
Output [-]
Increment function [-]
Centrifugal potential vector [J/kg]
Latitude [rad]
Roll angle [rad]
Heading [rad]
Yaw angle [rad]
Longitude of the ascending node [rad]
Rotational velocity vector [rad/s]
Argument of pericenter [rad]

Acronyms and Abbreviations

AIAA American Institute of Aeronautics and Astronautics

ASTOS AeroSpace Trajectory Optimization Software
CEA Chemical Equilibrium with Applications
DARPA Defense Advanced Research Projects Agency
DE Differential Evolution
DoF Degrees of Freedom
GA Genetic Algorithm
ESA European Space Agency
ICAO International civil aviation organization
ITAR International Trade in Arms Regulation
LH2 Liquid Hydrogen
LOX Liquid Oxygen
NASA National Aeronautics and Space Administration
PaGMO Parallel Global Multiobjective Optimizer
PSO Partical Swarm Optimization
RK Runge-Kutta
RP1 Refined Petroleum
Tudat TUDelft Astrodynamics Toolbox

Indices

A Aerodynamic force
A Aerodynamic (air-path) reference frame
B Body-fixed reference frame
b Combustion efficiency
C Earth-centered, Earth-fixed reference frame
E Vehicle carried normal Earth reference frame
F Thrus coefficient
G Gravitational force
I Inertial reference frame
xxii Nomenclature

i Initial
p Payload
T Thrust force
T Trajectory reference frame
 Chapter 1

Introduction

This thesis is meant to lay the groundwork for future concept-design studies for launch-
vehicle development with a focus on assisted-launch concepts. Major fields of interest
include the study of optimal-ascent trajectories and launch-vehicle design. The thesis
study is supposed to seek an answer to the following research question:

How does a launch assist affect the performance and the design of a launch vehicle?

How this question will be answered is explained in the scope of this introduction, which
is discussed in section 1.1. In section 1.2 the layout of the report is presented.

1.1 Scope

During the 1950s and 1960s, the act of traveling into outer space by rocket was widely
regarded as the epitome of human achievement. Just as it was for that other great inven-
tion of the aerospace world, the airplane, half a century before. At the dawn of the rocket
age, flight by airplane had through constant innovation evolved into an every day activity
for the average man. Using this as a precedent, the expectations of the general public
with regard to spaceflight by the end of the century were quite high. In their minds,
little of these expectations have actually come to fruition and space flight has become a
disappointment.
Access to space is to this day not self-evident in current society. Not only are the
risks far higher than those for an average airplane ride, a cost comparison of the two
aerospace activities unveils perhaps an even greater discrepancy. One of the main reasons
for the financial difference between the two is that space flight, 55 years after its first
occurrence, still depends on expendable vehicles, whereas heavier than air flight relied
on the usage of reusable vehicles from the very beginning. As if this fact on itself is not
enough to constrain the industry from blossoming, the velocity needed to reach orbit is
so high that a conventional launch vehicle exists of two or even more separate vehicles

1
2 Introduction

mounted on top of or next to each other while the actual payload contains only a small
fraction of the total vehicle mass.
Arguably, spaceflight has seen very little innovation since the day it started. Launch
vehicles still have the same shape, still use comparable propellants, are still launched the
same way and are still mostly or fully expendable. If the past will be extrapolated into
the future, it could take a long time before a threshold in efficiency, cost and safety will be
reached that can significantly change the industry. For this very reason, other concepts
of launch vehicles using different means of reaching orbit should be further investigated.
One of these alternatives is the adaptation of the strategy of giving an initial assist
to the launch vehicle. Assisted launch is a technique that still uses a rocket vehicle as a
means of transporting a payload into space, just as a conventional launch does, but its
main difference is that it uses another method of transportation to give the launch vehicle
a better orientation, position and velocity (launch-assist parameters) at the time of igni-
tion of the launch vehicle. Not only brings this the rocket closer to its destination point
both in a spatial sense as well as in an energetic sense, but it also creates an opportunity
to further increase its efficiency by minimizing drag and gravity losses.
Most launch-assist platforms available today are designed for other purposes and
are in no way optimal for launch assist. They consist out of either civil aviation planes
built for long range flights, thus having significant extra structural mass due to the large
fuel tanks, military fighter jets that traded in some speed and altitude capability for ma-
noeuvreability and stealth or suborbital rocket planes that have added structural mass
to house their passengers. Launch vehicles are then also not only designed to fit these
sub-optimal assist platforms but are often designed around an already existing engine, in
the hope to cut costs, or with the requirement of being manufactured in certain existing
facilities, in the hope to provide work. Therefore, it is interesting to investigate whether
a launch-assist system purpose built for the job is needed and what performance require-
ments it should meet. There are two factors that come into play when answering this
question. These factors are the cost of building and operating both the assist platform
and the launch vehicle as well as the systems total performance. This thesis focuses on
the performance of the entire system by assuming certain performance parameters of the
assist platform and it should be noted that the cost aspect is a significant restriction that
should be incorporated in further studies.
Assisted launch unfortunately remains a topic that has not undergone a large
amount of research and interest in the space community. With the exception of the
Pegasus launcher no concept has found its way off the drawing board into space. In
addition, the vast majority of the studies conducted so far focused on the feasibility of
the concept at hand taking into account the technology and infrastructure of the day.
Only a small number of studies has published a performance analysis of some kind and
they generally followed the approach of only optimizing the trajectory for a given set of
launch-assist parameters. In these studies, the launch vehicle remained mostly unchanged
for all evaluated launch assists [6] or was just optimized for one set of launch-assist pa-
rameters. This particular set of parameters was determined by external factors, such as
the performance of the airplane that the launch vehicle is supposed to be launched from.
To answer the research question posed at the start of this chapter, a number of
launch simulations need to be performed. To this purpose, a program was coded in C++
that can simulate the launch of an array of vehicles with different values for the launch-
vehicle parameters along different trajectories starting with different initial conditions.
1.2 Layout of the thesis report 3

The simulation of the trajectory takes place by propagating the state vector by means of
an integrator. This trajectory is dictated by the guidance program. The guidance pro-
gram is the model used to guide the launcher into the desired orbit. Because it is required
that an optimal trajectory, starting out from a whole variety of launch-assist parameters,
can be flown, the model needs to be very versatile. To provide this versatility, a 4t h-degree
polynomial is used to describe the pitch angle through time. The guidance program is
also subjected to a number of constraints, namely an altitude constraint, a bending load
constraint, an axial acceleration constraint, an angle-of-attack constraint and a heat flux
constraint.
There are a lot of combinations of the 4 degrees of the polynomial and the multiple
launch-vehicle parameters that can be made that describes a launch vehicle and a tra-
jectory that leads to insertion of a certain payload in the desired final orbit. To identify
the combination that provides the highest initial payload-to-initial-mass ratio into the de-
sired orbit, an optimization process is needed. The optimization algorithm used to fulfill
this task is a differential evolution algorithm, which is a population based evolutionary
algorithm that improves the solution by modifying the population throughout various
generations.
Due to the exponential nature of the complexity of a launch vehicle versus the
amount of V required, a relatively small assist with small compounding effects of mini-
mized losses can still open up the possibility of a significantly more simple launch vehicle.
Perhaps to the extent that the vehicle only requires one stage to be competitive with
respect to conventionally launched vehicles. Therefore the study focused on the opti-
mization of single stage launch vehicles in order to give a clearer view at which assist
parameters these types of vehicles could become competitive and by doing so the results
also show continuous trends through the range of launch-assist parameters.
The results obtained from the optimizer are sets of optimal launch-vehicle design
and trajectory parameters that represent a single stage vehicle and its trajectory that is
supposed to be launched at a given set of launch-assist parameters. These results are
subjected to an Extended Fourier Amplitude Sensitivity Test or EFAST sensitivity anal-
ysis in order to provide more insights into the trends that become apparent from the field
of optimal vehicles generated. This is an analysis, described in [46] that quantifies the
effect various parameters have on the total performance. These insights on why a certain
combination of launch-assist parameters generates an interesting compounding effect of
added assist energy, minimized losses and maximized vehicle efficiency can be important
information for the design of a future launch system. Through this manner, the study
purports to help future studies regarding both the launch-assist system as well as the
launch vehicle itself. It does not give a detailed description of a specific launch system
concept, although the conclusions and recommendations contain some information in that
regard. For an overview of specific concepts and launch-assist methods already designed
or proposed, the reader is referred to [56].

1.2 Layout of the thesis report

To find an answer to the research question, this body of work will begin in Chapter 2 with
the basic theory used to propagate a vehicle, ranging from the reference frames and their
4 Introduction

relevant transformation matrices to the equations of motion. In Chapter 3 the atmospheric

model, the gravitational model, the aerodynamic model, the launch-vehicle model, the
thrust model, the guidance model and the trajectory constraint model will be introduced
and discussed. A presentation of the numerical integrator is given in Chapter 4 followed by
an introduction to the chosen optimization algorithm and its settings in Chapter 5. The
assembly of the final simulation tool using external software components, such as PagMO
(Parallel Global Multiobjective Optimizer), Tudat (TU Delft Astrodynamics Toolbox),
CEA (Chemical Equilibrium with Applications) and Missile DATCOM, is documented in
Chapter 6. Finally, the results of the study are presented in Chapter 7 followed by the
conclusions and recommendations in Chapter 8.
 Chapter 2

Flight Mechanics

This chapter will provide a formulation of the ascent trajectory problem beginning with
a description of several reference frames that can be used to describe the motion and
the dynamics of the system in section 2.1. This is followed by the definition of the state
variables and an outline of the equations of motion in sections 2.2 and 2.3, respectively.

2.1 Reference frames

When a trajectory needs to be simulated, one needs to be able to define a motion or

position and at least one frame of reference is therefore needed. The option of using
multiple reference frames gives the added benefit that certain equations of motion are
easier to be described in one type of reference frame and can then just be translated
into another through transformation of vector coordinates from one frame to another.
Another reason for using multiple reference frames is that certain definitions of vectors
are more sensible in a certain reference frame, while they are not in others. For example,
the thrust force vector is easily defined in the body fixed reference frame and a rockets
velocity vector with respect to the Earths surface has a direct physical meaning. The
following chapter will describe the reference frames used in the simulation program. All
frames are right-handed orthogonal Cartesian frames:

Inertial reference frame / Earth-centered inertial reference frame, FI

The first frame used is the Earth-centered inertial reference frame. The definition
of an inertial reference frame can be derived from Newtons first law. An inertial
reference frame is thus a reference frame w.r.t. which a particle remains at rest or in
uniform rectilinear motion if no resultant force acts upon that particle. All reference
frames that perform a uniform rectilinear translational (non-rotating) motion w.r.t.
an inertial reference frame and in which the time differs only by a constant from
the time in the inertial reference frame, are inertial.
The Earth-centered inertial reference frame (OXI YI ZI ) is a pseudo inertial reference

5
6 Flight Mechanics

frame that can only be used as an inertial one as long as the influence of the
movement of the Earth around the Sun is very small on the flight of a rocket. Since
this is considered to be the case in this study, the frame is approximated as an
inertial one. It logically has its origin at the center of mass of the Earth and has
its ZI -axis directed North along the spin-axis of the Earth, which is assumed to be
fixed. The XI -axis passes through the equator at the line where the ecliptic and
the equator intersect, it is directed toward the vernal equinox, which is also referred
to as the First Point of Aries and denoted by . The YI -axis is perpendicular to
both the XI and ZI axes and completes the right-handed frame. This is the frame
in which the state vector is propagated.

Figure 2.1: Representation of the Earth centered inertial reference frame [33]

Earth-centered, Earth-fixed reference frame, FC

The second reference frame used is the Earth-centered, Earth-fixed reference frame
(OXC YC ZC ). This frame is used to express the motion of the launch vehicles
center of mass w.r.t the Earths surface. The frame is very similar to the Earth-
centered inertial reference frame with equal definitions for both the origin as the
Z-axis (ZC = ZI ). The difference is the X-axis: the XC -axis is defined to intersect
the Greenwich meridian in the equatorial plane and not the vernal equinox, which
makes it a frame rotating along with the Earth. The YC -axis is perpendicular to
the XC and ZC axes.

Vehicle-carried normal Earth reference frame, FE

The vehicle-carried normal Earth reference frame (OXE YE ZE ) is also known as the
vertical reference frame and has its origin in the vehicles center of mass. The frame
is used as an intermediate reference frame to store the state vector in during state
vector transformations. The XE YE plane is tangent to the Earth surface. For this
the geoid is taken as a reference. The ZE -axis is directed along the local gravity
2.1 Reference frames 7

Figure 2.2: Representation of the vehicle-carried normal Earth reference frame for a spherical Earth.
Adapted from [33]

vector as seen by the vehicles center of gravity. The XE -axis points to the direction
of North and the YE -axis is perpendicular to the XE -axis to the East. The frame
is depicted in figure 2.2 where is the longitude, latitude and O is the center of
the Earth, which is not the origin of the currently discussed reference frame.

Body-fixed reference frame, FB

The Body-fixed reference frame (OXB YB ZB ) has a vehicle reference point as its
origin and is used to define the orientation of the thrust vector, which is assumed
to align with the XB -axis at all times. The vehicle reference point is mostly the
center of mass at t=0. The frame is fixed to the vehicle. The XB -axis lies in the
symmetry plane and points forward, the ZB -axis also lies in the symmetry plane
and points downwards. The YB -axis completes the right-handed system.

Aerodynamic (air-path) reference frame, FA

Due to the absence of wind assumed in this study, the groundspeed-based and the
airspeed-based aerodynamic reference frames (OXA YA ZA ) are one and the same.
This frame is coupled to the aerodynamic velocity, which is defined as the velocity
of the center of mass relative to the undisturbed air and is thus used in the study
to define the direction of the drag vector. The origin of the frame is the same as
the one of the Body-fixed reference frame and the XA -axis is in the direction of
the aerodynamic velocity. The ZA -axis is collinear with the aerodynamic lift but
opposite in direction and the YA -axis completes the right-handed system.

Trajectory reference frame, FT

The origin of the trajectory reference frame (OXT YT ZT ) lies in the center of mass
of the vehicle. The XT -axis is directed along the velocity vector relative to the
rotating Earth-centered, Earth-fixed reference frame. The ZT -axis is in the vertical
plane directed downwards and the YT -axis completes the right-handed system. Due
to the assumption of no wind during this study, there is no distinction between a
groundspeed-based and an airspeed-based trajectory reference frame.
8 Flight Mechanics

Figure 2.3: Body fixed reference frame for an aircraft [33]

2.2 State variables

In this section, the possible ways of expression of the position and velocity of an object
will be discussed. The means presented here are orbital elements, Cartesian components
and spherical components.

2.2.1 Orbital elements

Since the goal of a launch is reaching a certain target orbit, a description of the motion
of a body around a planet in an elliptical orbit would be needed. Orbital elements are
very useful for this purpose and consist out of the following state variables [32]:

e: the eccentricity, defines the shape of the elliptical orbit. (0 e < 1)

a: the semi-major axis, defines the size of the orbit. (a > Re )

i: the inclination, defines the orientation of the orbit w.r.t. the Earths equator.
(0 i 180 )

: argument of pericenter, defines where the perigee of the orbit is w.r.t. the Earths
surface. (0 < 360 )

: the longitude of the ascending node, defines the location of the ascending and
descending orbit locations w.r.t. the Earths equatorial plane. (0 < 360 )

: true anomaly, defines where the satellite is within the orbit (0 < 360 )

The state variables that denote the position and velocity of the object depend on the
coordinate system used.
2.2 State variables 9

Figure 2.4: Orbital elements [38]

There are many types of coordinate systems, the two most common three dimensional
coordinates are Cartesian coordinates and spherical coordinates. In a Cartesian coordi-
nate system a point is specified by the coordinates x, y and z. These values are measured
from the origin, which is dependent on the type of reference frame that is used. A vector
at a certain point can be specified in terms of three unit vectors i, j and k that are all
perpendicular to each other.
For the Cartesian system the following variables define position and velocity:

Position: x, y, z

Velocity: x, y, z

In a spherical coordinate system a point is specified by the coordinates r, and . Where

r is measured from the origin, is measured from the z-axis and is measured from the
x-axis. A vector at a certain point can also be specified in terms of three unit vectors r,
and that are all perpendicular to each other. These unit vectors form a right handed
set.
For the spherical coordinate system the following variables are used and are expressed
w.r.t. the rotating Earth-centered reference frame:

Position: distance r, longitude and latitude

Velocity: speed V , flight-path angle and heading

The latitude is measured along the meridians and is zero at the equator. It is positive in
the northern direction (90 < 90 ) The longitude is measured along the circles of
latitude positive in the eastward direction (180 180 ). The distance, r, is the
distance between the center of mass of the launcher and the center of mass of the Earth.
V is the relative velocity (i.e., the modulus of the velocity vector V), is the angle between
V and the local horizontal plane (90 G 90 ) and the heading angle, , is the
projection of V in the local horizontal plane w.r.t. the local North (180 G < 180 ).
10 Flight Mechanics

Figure 2.5: Cartesian and spherical coordinate systems [38]

At = 90 , the vehicle is moving parallel to the equator to the East.

For simulation of a launch trajectory around a near spherical body, the use of a spherical
coordinate system would give a better understanding to the problem. Unfortunately this
coordinate system evokes singularities when defining the equations of motion, for instance
when the flight-path angle is 90 and -90 [33]. Therefore, Cartesian coordinates will be
used to express the equations of motion. Note that and depend on the velocity vector.
Therefore, due to the difference in the flight-path angle between the one in the FI frame
and the one in the FC frame, two spherical state vectors exist.

2.3 Equations of motion

The equations of motion are based on the three laws of Newton and do not incorporate
relativistic effects. These laws are listed below [58]:

Newtons 1st law: Every particle continues in its state of rest or uniform motion in a
straight line relative to an inertial reference frame, unless it is compelled to change
that state by forces acting upon it.
Newtons 2nd law: The time rate of change of linear momentum of a particle relative
to an inertial reference frame is proportional to the resultant of all forces acting upon
that particle and is collinear with and in the direction of the resultant force.
Newtons 3rd law: If two particles exert forces on each other, these forces are equal
in magnitude and opposite in direction (action = reaction).

It must be mentioned that the first two laws are only valid in an inertial reference frame.
Normally inertial reference frames are based on stellar references, however for this study it
2.3 Equations of motion 11

will also be assumed that the frame in which the motion of the vehicle is propagated, the
Earth centered inertial reference frame FI , is, as the name implies, also an inertial frame
and thus the absolute rotational velocity of FI is assumed negligible. This assumption
was made due to two reasons, first of all the study only concerns the launch, which has a
small time frame in the order of minutes or hours compared to the time frame, a year, it
takes for the Earth to orbit the Sun. Secondly, the fictitious forces in FI are very small.
Considering a non-inertial reference frame B that rotates with angular velocity and a
particle, P , that has a relative position rrel with a relative velocity vrel and a relative
acceleration arel , the acceleration of a particle is described as follows:

ap = aB + ( rrel ) + ( rrel ) + 2( vrel ) + arel (2.1)

The second term involving , which is called the Euler acceleration, can in the case of
FI be disregarded since the angular velocity of the Earth around the sun is constant
( 1.991 107 rad/s). Using a relative position of 6.5 103 km, the third term, the
centrifugal acceleration becomes 2.57 107 m/s2 . Using a relative velocity of 7.8 km/s,
the fourth term, the acceleration due to the coriolis force becomes 3.1 103 m/s2 . These
low values are from hereon out deemed negligible during launch by assuming the Earth
centered inertial reference frame to be truly inertial.
In the case of motion of a rigid body, the accompanying equations can be split up in
translational and rotational equations. Whilst simulating a launch trajectory, the main
interest is in the motion of the center of mass of the vehicle instead of its attitude motions.
Neglecting the motion of the vehicle around this center of mass reduces the degrees of
freedom to 3 (3 degrees of freedom or 3 DoF). In this study the vehicle is thus assumed
to be a point mass because rotational dynamics would require precise knowledge of the
launch vehicle configuration for which a launch trajectory is being developed. It would
require detailed information about the geometry as well as the mass distribution of the
vehicle throughout time. It is far beyond the scope of this study to accurately model
this information since it is the objective of this study to not only analyze one single
vehicle but a whole array of vehicles with different parameters. This leads to a choice
of the 3 DoF simulation, which introduces the assumption that the control system can
generate the required moments to change the attitude of the vehicle toward the on-coming
flow. This means that all attitude changes are performed instantaneously, which is an
oversimplification of reality. To diminish the effect of this oversimplification, the guidance
program, which is dictated by the pitch attitude of the vehicle, is modeled as a smooth
polynomial, see section 3.6.3.
The governing equation of motion for a rigid body with fixed-mass distribution is [32]:

d2 rI dVI
FI = m 2
= m (2.2)
dt dt
where,

FI : Total of the external forces expressed in components of the

inertial frame [N]

rI : Position w.r.t. the inertial reference frame [m]

12 Flight Mechanics

VI : Velocity w.r.t. the inertial reference frame [m/s]

This is a dynamic equation that describes the motion of a body under the influence of
external forces. The kinematic equation can be derived from it, which describes the rate
of change of position of this body:

drI
= VI (2.3)
dt
Using the definition of Cartesian position and velocity given in 2.2.1 the dynamic and
kinematic equations of motion can be described as follows:

x
dVI I 1
= yI = [FA,I + FG,I + FT,I ] (2.4)
dt m
zI

x
drI I
= yI (2.5)
dt
zI

where forces F with the indices A, G and T stand for the aerodynamic, gravitational and
thrust force respectively, expressed in the Earth-centered inertial reference frame (index
I). These forces are usually expressed in different reference frames. The aerodynamic
force is often expressed in the body-fixed reference frame (index B) or the aerodynamic
reference frame (index A), the gravitational force is often expressed in the vehicle carried
normal Earth reference frame (index E) and the thrust force is also easily expressed in
the body-fixed reference frame.
To this purpose the force vectors must be transformed by means of transformation matri-
ces into vectors that apply to the correct reference frame. The required transformations
are:

FA,I = CI,B FA,B (2.6)

FT,I = CI,B FT,B (2.7)

FG,I = CI,E FG,E (2.8)

2.3.1 Euler Angles

The attitude of a vehicle relative to an inertial reference frame can be described by three
successive rotations through three Euler angles1 . Only the orientation of the reference
frame is described by these Euler angles, no translation takes place through Euler angles.
1
Named after Leonhard Euler (April 15, 1707 - September 18, 1783), a Swiss mathematician and
physicist.
2.3 Equations of motion 13

Each rotation should be around a different axis than the one before, resulting in 12
possible sets of Euler angles [33]. Six sets of symmetric Euler angles which use two
(non-consecutive) rotations about the same axis:

x y x , y x y , z x z
(2.9)
x z y , y z x , z y x

And six sets of asymmetric Euler angles, which use rotations around all axes:

x y z , y x z , z x y
(2.10)
x z x , y z y , z y z

These sets are also often referred to as a combination of numbers, where 1 represents an
Euler angle rotation around the X-axis, 2 an Euler angle rotation around the Y -axis and
3 an Euler angle rotation around the Z-axis

2.3.2 Frame transformations

The information of a single rotation through an Euler angle depends on the particular
axis around which the rotation is performed. This information is transferred through
the form of transformation matrix, which transforms the original definition of a vector
in the original reference frame. In these transformation matrices the angle , is the
angle by which the reference frame is rotated around the X, Y or Z axes. The following
matrices are transformation matrices for rotation around the X-axis, Y -axis and Z-axis
respectively [33] [32].

1 0 0
CX (x ) = 0 cos x sin x (2.11)
0 sin x cos x

cos y 0 sin y
CY (y ) = 0 1 0 (2.12)
sin y 0 cos y

cos z sin z 0
CZ (z ) = sin z cos z 0 (2.13)
0 0 1

The transformation matrix Ci,j denotes a transformation from the j-frame to the i-frame.
This matrix is equal to both the inverse as well as the transpose of the transformation
matrix from the i-frame to the j-frame, see Eq. (2.14).

Cj,i = C1 T
i,j = Ci,j (2.14)
14 Flight Mechanics

With this information the transformation matrix for an entire Euler angle set can be cal-
culated. Consider the rotation sequence x y x also called a 3-2-1 transformation.
The transformation for the entire rotation is described by [33]:

v2 = CX (x ) CY (y ) CZ (z )

1 0 0 cos y 0 sin y cos z sin z 0
= 0 cos x sin x
0 1 0 sin z cos z
0
0 sin x cos x sin y 0 cos y 0 0 1

cos y cos z cos y sin z sin y
= sin x sin y cos z cos x sin z sin x sin y sin z + cos x cos z sin x cos y
cos x sin y cos z + sin x sin z cos x sin y sin z sin x cos z cos x cos y
(2.15)

Earth centered Inertial reference frame (FI ) to Earth-Centered, Earth-fixed

reference frame (FC ).

The two frames are nearly identical to each other with the sole difference that the FC -
frame is rotating along with the central body. They are defined to coincide at t = 0.
The transformation matrix can be derived using the following variables,

t = rotational rate of the Earth (rad/s)

t = time from epoch (s)

cos t t sin t t 0
CC,I = CZ (t t) = sin t t cos t t 0 (2.16)
0 0 1

Earth centered Earth-fixed reference frame (FC ) to vehicle carried normal

Earth reference frame (FE )

For this transformation, two consecutive rotations are required.

about the ZC -axis

2 + about the YE -axis

The transformation matrix then becomes [32]:

sin sin sin cos cos

CE,C = CY ( + ) CZ ( ) = sin sin cos sin cos (2.17)
2
cos 0 sin
2.3 Equations of motion 15

Vehicle carried normal Earth reference frame (FE ) to body-fixed reference

frame (FB )

For this transformation, three consecutive rotations are required.

Yaw angle about the ZE -axis

pitch about the YE0 -axis2
roll angle about the XB -axis

The transformation matrix then becomes the same as in (2.15) [33]:

CB,E = CX () CY () CZ ()

cos cos cos sin sin
= sin sin cos cos sin sin sin sin + cos cos sin cos
cos sin cos + sin sin cos sin sin sin cos cos cos
(2.18)

Earth centered Inertial reference frame (FI ) to vehicle carried normal Earth
reference frame (FE )

This transformation matrix can be obtained by simply multiplying two of the previously
found transformation matrices.

CE,I = CE,C CC,I (2.19)

Earth centered Inertial reference frame (FI ) to body-fixed reference frame

(FB )

This transformation matrix can be obtained by simply multiplying two of the previously
found transformation matrices.

CB,I = CB,E CE,I (2.20)

Vehicle carried normal Earth reference frame (FE ) to trajectory reference

frame (FT )

To make a transformation between these two reference frames the following matrix needs
to be used [38]:

cos cos cos sin sin
CT,E = CY () CZ () = sin cos 0 (2.21)
sin cos sin sin cos
where is the flight-path angle and is the heading angle.
2
The YE0 -axis is the Y-axis after the first rotation.
16 Flight Mechanics

Aerodynamic reference frame (FA ) to vehicle carried normal Earth reference

frame (FE )

To make a transformation between these two reference frames, three consecutive rotations
are required.

A aerodynamic bank angle about the XA -axis

A aerodynamic flight-path angle about the YA0 -axis3

A aerodynamic heading angle about the ZE -axis

These rotations result in the following matrix [32]:

CA,E = CZ (A ) CY (A ) CX (A )

cos A cos A sin A cos A cos A sin A sin A sin A sin A + cos A sin A cos A
= sin A cos A cos A cos A sin A sin A sin A cos A sin A + sin A sin A cos A
sin A cos A sin A cos A cos A
(2.22)

3
The YA0 -axis is the Y-axis after the first rotation.
 Chapter 3

Models

This chapter provides an overview of all the different models that are used to simulate
the conditions that occur during space launch. The chapter begins with an overview
of the environment in section 3.1, which consists out of an atmospheric model (section
3.1.1) and the gravitational model (section 3.1.2). This is followed by the aerodynamic
model in section 3.2 and an elaborate discussion of the launch vehicle model in section
3.3 where two sub models are presented, namely the launch vehicle performance model
(section 3.3.1) and the launch vehicle mass model (section 3.3.2). The chapter is then
completed with a description of the guidance model in section 3.6 and the trajectory
constraint model in section 3.7.

3.1 Environment models

The environment in which the flight of the rocket is undertaken, has a significant impact
on the performance of the launch system. For this reason it must be modeled accurately.
In this section, both the atmospheric model and the gravitational model will be discussed.

3.1.1 Atmospheric model

Due to the influence of the atmosphere on the flight path of a vehicle, a thorough under-
standing of atmospheric properties is required and the atmosphere needs to be adequately
modeled to provide realistic results and comparisons regarding different launch methods.
The real atmosphere does not remain constant. At a different time, a change in tem-
perature, density and pressure at any height and place on Earth will take place. In the
real atmosphere, the actual performance of a vehicle does not provide a reliable basis in
comparison with other vehicles. For these reasons, the need for standardization arises and
a number of atmospheric models has been developed. These models can be put in either
one of two groups. Either the model is a standard atmospheric model or it is a reference

17
18 Models

atmospheric model. The standard models only give atmospheric properties with chang-
ing position while the reference model can also take temporal, seasonal, geomagnetic
and solar effects into account. Some simpler reference atmosphere models just involve
a temperature or humidity offset from a standard atmosphere model and its standard
day temperature gradient. This takes into account that the geometric height, Z, of the
tropopause is increased on a hot day since increasing temperature causes the atmosphere
to expand upwards. However, since this simulations aims to provide a comparison between
multiple launches, they all are required to undergo the same circumstances. Therefore
the simplicity of the models become a more important factor and the simplicity of the US
1976 standard atmosphere model was the reason why it was decided in [56] that it would
be used in the simulations.

Standard atmospheric model, US 1976

One of the standard atmospheric models in common use is the US standard atmosphere
of 1976, which will be shortly presented here [11]. This model is defined in terms of
the International System (SI) of units and is identical to the standard atmosphere of
the International Civil Aviation Organization (ICAO) below 32 km. The model presents
values for atmospheric parameters from the Earths surface up to an altitude of 1000
km at a latitude of 45 North. The temperature of this model is given by the following
equation below 86 km geometric altitude [11]:

TM = TM,b + LM,b (H Hb ) (3.1)

where TM is the molecular scale temperature with the value of subscript b ranging from 0
to 6 in accordance with each of the seven successive layers that are defined below 86 km,
TM0 is equal to T0 and has the value of 288.15 K. LM is the molecular scale temperature
gradient and is given in Table 3.1. Hb stands for the geopotential height, which is an
adjustment of geometric height using the variation of gravity with elevation.

Table 3.1: Defined reference levels and gradients [11]

Subscript b Geopotential Molecular Form of

Height Hb scale function
[km] temperature relating T
gradient to H
LM,b
[K/km]
0 0 -6.5 Linear
1 11 0 Linear
2 20 +1 Linear
3 32 +2.8 Linear
4 47 0 Linear
5 51 -2.8 Linear
6 71 -2.0 Linear
7 84.852 -6.5 Linear
3.1 Environment models 19

The unit of measurement of geopotential altitude is the standard geopotential meter [m],
which represents the altitude over which the same work is done for lifting a unit mass as
would be done when lifting that unit mass 1 geometric meter, through a region in which
the acceleration of gravity is uniformly 9.80665 m/s2 .
A conversion from geopotential altitude and back can be made using the following formula
[11]:

r0 H
Z= (3.2)
r0 H

where Z is the geometric altitude and r0 is the radius of the Earth and has the value of
6,356,766 m for the US 1976 standard atmosphere.
For higher altitudes Eq. (3.1) does not longer hold. From 86 to 91 km there is an isother-
mal layer modeled where the temperature stays at 186.8673 K. For the layer between 91
an 110 km Eq. (3.3) holds [11].

 2 !(1/2)
Z Z8
T = Tc + A 1 (3.3)
a

where A = -76.3232 K and a = -19.9429 km.

For the next layer between 110 and 120 km the temperature follows the following rela-
tionship [11]:

T = T9 + LK,9 (Z Z9 ) (3.4)

And finally for the layer between 120 and 1000 km [11]:

T = T (T T10 ) exp () (3.5)

where = LK,9 /(T T10 ) and = (Z Z10 ) (r0 + Z10 )/(r0 + Z).
In the above expressions, T is the exospheric temperature defined at 1000 K.
The model has two different equations for computing pressure at various regions below
an altitude of 86 km. The first of these equations is the following [11]:

 
g0 M0

 
TM,b R LM,b
P = Pb (3.6)
TM,b + LM,b (H Hb )

where R is the gas constant with the value 8.31432 103 Nm/kmol K. This equation is
used when the molecular scale temperature lapse rate is not equal to zero [11].
 
g0 M0 (H Hb )
P = Pb exp (3.7)
R TM,b

In these equations g0 , M0 and R are single valued constants, while LM,b and Hb are
multi valued constants, meaning that they have a different value for each layer. M0 is the
20 Models

molecular weight of the atmosphere at sea level which stays constant throughout layers
0-6 up to a geometric altitude of 86 km and holds the value of 28.964 kg/kmol. The
reference value for Pb for b = 0 is the defined sea level value, P0 = 101325 N/m2 . Values
for Pb with b = 1 through 6 can be obtained by applying these equations to find the value
of the upper boundary of the previous layer.
For an altitude between 86 and 1000 km the pressure is computed as a function of geo-
metric altitude and involves the altitude profile of kinetic temperature T rather than that
of TM [11]:

ni R T
X X P
P = Pi = ni k T = (3.8)
NA

where the Boltzmann constant, k = 1.3806221023 Nm/K, NA is the Avogadro constant,

6.022169 1026 kmol1 and
P
ni = the sum of the number densities of the individual gas
species comprising the atmosphere at an altitude above 86 km.
For all layers that make up the US 1976 model, the temperature and pressure are known
by the aforementioned equations. The only relevant parameter concerning aerodynamic
forces is the density. This can be solved by assuming the atmosphere to be a perfect gas
where the ideal gas law holds:

P M
= (3.9)
R T
The model that was used in the simulation and is based on these equations summarized
above follows all equations up to a geometric height of 120 km. Above this altitude the
atmosphere is assumed to be non-existent.

4 4
x 10 x 10
12 12

10 10
Geometric Height (m)

Geometric Height (m)

8 8

6 6

4 4

2 2

0 0
4 2 0 2 4 6 8 6 4 2 0 2
10 10 10 10 10 10 10 10 10 10 10 10
Pressure (Pa) Density (kg/m3)

(a) (b)

Figure 3.1: Pressure (a) and density (b) to geometric height profiles generated by the model.

3.1.2 Gravitational model

Gravity is a very important force that acts on the vehicle. It is in most situations the
main external force that determines the movement of a vehicle through space. This force
3.1 Environment models 21

4
x 10
12

10

8
Geometric Height (m)

0
150 200 250 300 350 400
Temperature (K)

(a) (b)

Figure 3.2: A comparison of the temperature (in Kelvin) - geometric height profile of the model (a)
and from literature (b) [8], which also shows the changes between US 1976 and US 1962.

can be written as follows:

FG = m g (3.10)

where m is the mass of the vehicle and g is the gravitational acceleration vector.
One can model this force by just assuming that the value of g is a constant (e.g., g = g0 =
9.80665 m/s2 ). This is a major oversimplification since in fact the value of g varies with
position. Another way to model the gravitational acceleration vector is by assuming the
Earth to be a point mass or homogeneous sphere with a mass of 5.9742 1024 kg (ME )
[55]. In this case, the value of g would be governed by the gravitational force equation
that Newton established in 1666 and describes the force of gravity between two point
masses, A and B as follows:

MA MB
FA = G 2 eAB (3.11)
rAB

where G stands for the Universal Gravitational Constant and has a value of 6.668 1011
Nm2 /kg2 , MA and MB are the masses of the two objects and rAB is the distance between
22 Models

the two objects that attract each other. [58]

In the case of attraction between the Earth and a man-made vehicle, the mass of the
vehicle is negligible compared to the mass of the Earth and Eq. (3.11) becomes:

ME
FG = G r (3.12)
r3

This model is called the central gravity field model. It is incorrect to a certain extent
because it does not take the heterogeneous distribution of mass of the Earth into account.
Only a more complex model can take effects as the flattening of the Earth into account.
Gravity can sometimes be defined as the sum of the gravitational acceleration with the
centrifugal acceleration. In this definition gravity will be the force that is perceived by
the observer The centrifugal acceleration is given by Eq. (3.13)

= ( r) (3.13)

This gives for the centrifugal potential written in rectangular coordinates:

2 2
= (x + y 2 ) (3.14)
2

The gravity potential is then defined as the sum of the gravitational potential and the
centrifugal potential.
This gravitational potential can be written as following:

Z Z Z
(Q) X
V(P) = G d (3.15)
lP Q
Q

Figure 3.3: Location of P and Q [23]

This is an integration performed over the whole body (i.e. Earth) presented in figure 3.3.
The vector lP Q between the body with negligible mass in P and a random element at Q,
part of the main body, can be rewritten in a series expansion with legendre polynomials
and eventually a so-called harmonic expansion model arises where the satellite position
3.1 Environment models 23

is expressed in spherical coordinates w.r.t. the center of mass of the Earth: r (distance),
(latitude) and (longitude): [37]

V(P) =
X
n
 n  n !
X Re X Re
1 Jn Pn (sin ) + Jn,m Pn,m (sin ) cos (m( n,m ))
r r r
n=2 n=2 m=1
(3.16)

where

dm Pn (x)
Pn,m (x) = (1 x2 )m/2 (3.17)
dxm

and

1 dn
Pn (x) = (1 x2 )n (3.18)
(2)n n! dxn

Pn,m (x) is a Legendre function. When m is zero one speaks of legendre polynomials. The
parameters n and m denote the degree and order of the expansion terms, respectively.
The coefficients Jn,m and n,m are scaling and orientation parameters.
It should be noted that the terms (n = 1, m = 0) and (n = 1, m = 1) are absent since
the origin of the gravity field is assumed to coincide with the center of mass of the Earth.
Based on measurements from satellite data, aerial data and terrain data, values for co-
efficients can be assigned. Since the function uses an infinite sum, the gravity field can
be copied perfectly. Due to the accuracy of the measurements only values for a degree of
around 3600 can be established, which provides a very high accuracy. In Table 3.2 the
values of degree 2 and 3 are given.

Table 3.2: Numerical values of low degree gravitational coefficients of the Earths gravity field [37]

n (degree) m (order) Jn,m [106 ] n,m [deg]

2 0 1082.6267 0.0
2 1 0.0016 98.9
2 2 1.8154 -14.9
3 0 -2.5327 0.0
3 1 2.2090 7.0
3 2 0.3744 -17.2
3 3 0.2214 21.0

As one can see, J2,0 is far larger than any of the other scaling coefficients. This is due to
the fact that J2,0 represents the effect of the flattening of the Earth due to the Earths
rotation. The J3,0 coefficient, which is the second largest, gives the gravity field a pear
shape which is predominantly due to the location of the continents on Earth.
24 Models

Since J3,0 already has a very small value, some models only use J2,0 in their model. This
leads to the following equation:
  n
Re
V J2 = 1 J2 P2 sin (3.19)
r r
Using the definition of the Legendre polynomial P2 (t) [26],

3t2 1
P2 (t) = (3.20)
2
one obtains

J2 R2e
 
2
VJ2 = 1+ [1 3 sin ] (3.21)
r 2r2
Replacing sin with z/r and adding centrifugal potential due to Earth rotation to VJ2 ,
one obtains the gravity potential.

J2 R2e 3z 2 2 2
 

UJ2 = 1+ [1 ] + (x + y 2 ) (3.22)
r 2r2 r2 2
The three components of the gravity vector are then the three partial derivatives of UJ2 :

3J2 R2e 15J2 R2e z 2

 
x
gx = 3 1+ + x 2 (3.23)
r 2r2 2r4

3J2 R2e 15J2 R2e z 2

 
y
gy = 3 1 + + y 2 (3.24)
r 2r2 2r4

9J2 R2e 15J2 R2e z 2

 
z
gz = 3 1 + (3.25)
r 2r2 2r4
It has been numerically determined that the difference due to the J2 effect is less than 5
g in all three components [26]. This means that in the event of a launch trajectory where
the gravity loss amounts to 1000 m/s, the J2 effect amounts to just 0.005 m/s in V. The
other coefficients are, as already pointed out, significantly smaller and not incorporating
them in the model would only evoke a small discrepancy between the model and the actual
Earth gravity field. The J3 effect for instance, which is about one five hundredth of the
J2 effect, amounts to 0.00001 m/s and is thus totally negligible. For the same reason also
the J2 effect is discarded in favor for implementation of the central gravity field model.

3.2 Aerodynamic model

During launch, the vehicle penetrates the atmosphere to reach its goal: space orbit. Every
vehicle that moves through the atmosphere is being influenced by aerodynamic forces that
in turn have an effect on the trajectory of the vehicle. The magnitude of these forces are
dependent on the velocity w.r.t. the atmosphere, the attitude and shape of the vehicle
and the properties (temperature, density and pressure) of the atmosphere.
3.2 Aerodynamic model 25

3.2.1 Aerodynamic forces

Aerodynamic forces arise due to the pressure and the skin friction of air when an object
moves through it. A rocket mostly does this while having an incident angle. This so called
angle of attack should be kept small in order to avoid high drag losses. The resultant
force that the airflow implements upon the vehicle is in this model divided into three
components: these are called the lift, the drag and the side force components. The side
force component is during the model always equal to zero since there is no wind considered
and the rocket is standard directed to fly at constant azimuth with a non-existent side
slip angle. However, the model is capable to cope with the introduction of a side slip
angle. The drag is defined parallel to the velocity vector, while the lift and the side force
are defined perpendicular to it in the aerodynamic reference frame, FA , as follows:

D = FX,AA (3.26)

S = FY,AA (3.27)

L = FZ,AA (3.28)

They satisfy the following relations:

1
D = CD V 2 Sref (3.29)
2

1
S = CS V 2 Sref (3.30)
2

1
L = CL V 2 Sref (3.31)
2
where CL is the lift coefficient ,CD is the drag coefficient, CS is the side force coefficient,
is the density of the atmosphere, V is the velocity relative to the atmosphere, and Sref
is the aerodynamic reference surface., which corresponds to the cross-sectional area of the
body.
The three coefficients are dependent on the Mach number, the Reynolds number, the
altitude, vehicle configuration, control surfaces, the side slip angle, , and the angle of
attack, [44].
In order to obtain these coefficients, the off-the-shelf program code, Missile DATCOM is
used for the simulations. The Missile DATCOM program is obtained as a supplement
CD of the book Design Methodologies for Space Transportation Systems, by Walter E.
Hammond [24]. The program can be directly implemented in the larger simulation tool
although it was quickly discovered that this severely increases the computational load.
Therefore, it was chosen to construct a database of several launcher configurations. A
configuration contained in that database (all are listed in table 3.3) is chosen to serve as
the aerodynamic model when the estimated length and diameter of the stage are closest
to values of the specific configuration. All these configurations were standard equipped
26 Models

Table 3.3: Missile DATCOM database configuration matrix

L (m) 10 12.5 15 20 25 30 40 50 60 80 100

0.5 0.625 0.75 1 1.25 1.5 2 2.5 3 4 5
0.625 0.78125 0.9375 1.25 1.5625 1.875 2.5 3.125 3.75 5 6.25
0.75 0.9375 1.125 1.5 1.875 2.25 3 3.75 4.5 6 7.5
0.875 1.09375 1.3125 1.75 2.1875 2.625 3.5 4.375 5.25 7 8.75
1 1.25 1.5 2 2.5 3 4 5 6 8 10
1.25 1.5625 1.875 2.5 3.125 3.75 5 6.25 7.5 10 12.5
1.5 1.875 2.25 3 3.75 4.5 6 7.5 9 12 15
D (m) 1.75 2.1875 2.625 3.5 4.375 5.25 7 8.75 10.5 14 17.5
2 2.5 3 4 5 6 8 10 12 16 20
2.5 3.125 3.75 5 6.25 7.5 10 12.5 15 20 25
3 3.75 4.5 6 7.5 9 12 15 18 24 30
3.5 4.375 5.25 7 8.75 10.5 14 17.5 21 28 35
4 5 6 8 10 12 16 20 24 32 40
4.7 5.9 7 9.5 12 14 19 23 28 37 47
5.5 6.875 8.25 11 13.75 16.5 22 27.5 33 44 55

with a tangent ogive nosecone without any lifting surfaces and the sole difference between
one and another are their respective lengths and diameters. This method of working
with a database instead of working with a computationally intensive integrated Missile
DATCOM program was validated by using both methods to run a simulation. The
simulation performed had as assist conditions a velocity of 2,000 m/s and an altitude of 40
km. The simulation utilizing the Missile DATCOM database found after 1000 generations
of 20 individuals a maximum payload to initial mass ratio injected into a circular 200 km
high orbit of 0.18965 and a population average of 0.17535. The simulation with the
integrated Missile DATCOM program found a maximum payload ratio of 0.18855 and
a population average of 0.17534. These differences between these results fall within the
noise of two different optimization runs and can therefore be assumed to be negligible.
In section 6 the Missile DATCOM and its implementation in the simulation program is
further discussed.

3.3 Launch-vehicle model

The computer simulation program is designed to perform multi disciplinary optimization.

This means, in this case, that it optimizes both the launch trajectory as well as the launch
vehicle model. This section gives an explanation of that launch-vehicle model, which
subdivisions it has, and how these interact with each other and how the whole model
interacts with the trajectory further on explained in the guidance model, see section 3.6.
The launch vehicle model is subdivided into two smaller sections or models. These are
identified as follows:

The performance model

3.3 Launch-vehicle model 27

The mass model

The main focus of the performance model is providing a decent assumption for the thrust
of the launch vehicle. The mass model and the aerodynamic model evidently focus on
an estimate of the mass and the aerodynamic characteristics of the vehicle, respectively.
The vehicle model requires the following inputs:

Choice of propellant between a refined petroleum oxygen mixture (RP1/LOX) and

a hydrogen oxygen mixture (LH2/LOX), often referred to as kerolox and hydrolox.

The oxidizer-to-fuel ratio

The chamber pressure

The combined length of the propellant tanks

The diameter of the tanks

The nozzle exit pressure

The nozzle exit diameter

3.3.1 Performance Model

The choice of propellant in combination with a given oxidizer to fuel ratio and chamber
pressure is enough to establish , the ratio of specific heats, the molar mass, M , and the
chamber temperature, Tc . This is done effectively by linear interpolation of values stored
in a data sheet. The data include values for all of these three variables for a total of 6271
combinations of oxidizer-to-fuel ratios and chamber pressures for both propellant choices.
These values were gathered from the NASA Glen Equilibrium Program also known as
CEA [7]. Further explanation of this program is given in section 6. The chamber pres-
sures used range with incremental steps of 8 bar from 8 up to 264 bar. Likewise, the
oxidizer to fuel ratios range with incremental steps of 0.5 from 1 up to 10. The results
from this program are published in Appendix A. Having this information in place, the rest
of the performance model only exists of relations between change in pressure, tempera-
ture and thermal energy over the nozzle to the exhaust velocity, i.e., change in velocity of
the gas flow using basic laws of mechanics and thermodynamics. These relations are also
known as ideal rocket motor theory. The assumptions needed for this theory are quoted
below [65]:

The exhaust gases are homogeneous and have a constant composition. As in many solid
propellant rockets metal powder is added, which is expelled in solid or liquid state, the
combustion gases are not always homogeneous. As the temperature decreases when the
gases are expanded through the nozzle, the chemical equilibrium changes and thus the
composition of the gases is not constant. - The gas or gas mixture expelled obeys the
ideal gas law. The ideal gas law, or universal gas equation, is an equation of state of an
1
19 different oxidizer-to-fuel ratios for each of all 33 chamber pressures
28 Models

ideal gas. It relates the pressure p of a gas with the volume V it occupies, the number
to moles of the gas n, and the gas temperature1. The ideal gas law is valid for ideal
gases only.Real gases obey this equation only approximately, but its validity increases as
the density of the gas tends to zero. - The heat capacity of the gas or mixture of gases
expelled is constant. In reality the heat capacity depends on the temperature and composi-
tion of the expelled matter and neither of them is constant. - The flow through the nozzle
is one-dimensional, steady and isentropic. Only with a specially shaped nozzle, the flow
can be made one-dimensional. During motor start-up and stop the gas flow is not steady.
Though relatively small, there is some heat exchange with the surroundings causing the
flow not to be isentropic.

What follows is a guide that follows ideal rocket motor theory through which thrust can
be calculated with only the chamber pressure c , the nozzle exit pressure e , the ratio
of specific heats , the molar mass M and the chamber temperature Tc . The relation
between this ideal rocket motor theory and the actual behavior of a rocket motor will be
accounted for by using correction factors in the design process. Two will be used, namely
the combustion efficiency, b , and the nozzle efficiency or nozzle quality, F .
From the molar mass, the specific gas constant of the gas can be easily calculated with
Eq. (3.32).

RA
R= (3.32)
M
where R is the specific gas constant of a gas, RA the universal gas constant which is equal
to 8,314.5 m3 Pa/kgmolK. Similarly, a value for leads to a value for the Vandenkerckhove
function given by:
 
+1

 
2 2 (1)
= (3.33)
+1
The ideal characteristic velocity can then be calculated with the following formula:

1 p
cid = R Tc (3.34)

However in the model, this ideal characteristic velocity is multiplied by a correction factor
which accounts for some imperfections of the rocket motor. This factor is the combustion
efficiency, b , and has been allotted a value of 0.95. This correction factor will be subjected
to further scrutiny in section 3.3.3. The chamber density can be calculated as follows:

Pc
c = (3.35)
Tc R
where Pc is the chamber pressure. The expansion will be assumed to be isentropic thus
using the Poisson relation shown in Eq. (3.36) the exit density can be derived as shown
in Eq. (3.37).
 
  1  (1)
Te Pe e
= = (3.36)
Tc Pc c
3.3 Launch-vehicle model 29

 
  1
Pe
e = c (3.37)
Pc

The area ratio can be given as follows:

Ae
=s (3.38)
At  
2

1

2
     
pe pe
1 pc 1 pc

While the equation for ideal thrust coefficient is given by Eq. (3.39):

v  
u   1
u 2 Pe + P e Ae
u
CFid = t 1 (3.39)
1 Pc Pc At

It must again be noted that in the model used, this ideal thrust coefficient is multiplied
by a correction factor which accounts for some imperfections of the rocket motor. This
thrust coefficient correction factor is the nozzle quality or nozzle efficiency, F , and has
been allotted a value of 0.97. Just as b , F will be subjected to further scrutiny in section
3.3.3. Another remark required to be made here is that the model does not address the
effect of the ambient pressure on the thrust coefficient. This is done so because ambient
pressure can only be determined in the launch trajectory model and information transfer
between models is being minimized. For total clarity, the thrust force this model calculates
is the maximum vacuum thrust force and the ambient pressure effect is accounted for
outside this model, in the trajectory simulation without any loss of accuracy. From all
the information acquired thus far, the exit velocity can be calculated as the product of
the characteristic velocity and the thrust coefficient:

Ve = c CF (3.40)

The mass flow can also be calculated using the conditions found at the exit of the nozzle.

m = e Ve Ae (3.41)

where Ae is the exit area, which is one of the inputs required. Finally, all information is
present to find the maximum vacuum thrust force the vehicle can produce according to
this model, which is the product of the mass flow and the exit velocity.

Fmax = m Ve (3.42)
30 Models

3.3.2 Mass Model

The mass model consists out of several parametric formulas that can be used for mass
estimation of several launch vehicle components. These components are the following:

The engine
The thrust structure
The propellants
The propellant tanks
The nozzle
The vehicle equipment bay
The fairing
The heat shield

As can be seen in various of the following subsections, many of the parametric formulas
drastically depend on some of the values found in the performance model discussed earlier.

Engine mass

Several methods for estimating the engine mass were discussed in [56], finally the most
simple method was selected, a small recap is given below. Two parametric formulas were
found for each of the possible propellant choices in [68]. For a kerolox engine the following
connection was found between engine mass and thrust delivered:

m = 0.0011 F 0.9992 (3.43)

This formula has a standard error of estimate (SEE) of 23.9%
For a hydrolox engine the following equation needs to be used:

m = 0.0114 F 0.8573 (3.44)

which has an SEE of 18.6%.

Thrust structure

For the thrust structure, a similar parametric formula was found in [31]:

m = C TT OT (3.45)
where C is the rocket engine thrust scaling coefficient, which depends on the material
used, and TT OT is the total vacuum thrust. C is 0.003 for aluminum and 0.0024 for
composites. This formula was implemented in the model using the latter value, since the
source of the formula is outdated and it is deemed that current technology would certainly
be able to comply with the low requirement given.
3.3 Launch-vehicle model 31

Figure 3.4: Propellant tanks depiction

Propellant mass

The propellant mass can be estimated using the tank dimensions, which are inputs for
the model. The density of the propellants that the model needs to deal with are given by
[66]:

Liquid Oxygen: 1.141 kg/l

Liquid Hydrogen 0.071 kg/l

Dodecane (Kerosene) 0.749 kg/l

Using the oxidizer-to-fuel ratio, an average propellant density can be established according
to the following formula:

(O/F )ratio + 1 1
prop = (O/F )ratio
+ (3.46)
f uel
oxidizer

Assuming that the propellant tanks have the shape of cylinders of which the ends are
spherical caps and assuming that there is no space between the top of the spherical cap of
the fuel tank and the bottom of the spherical cap of the oxidizer tank, it can be calculated
how much volume the propellants can occupy.
Regarding the calculations of the tank mass, which will be shown in section 3.3.2, it
would be more exact if the exact dimensions of both tanks were calculated with regard to
thermal stresses. However, to keep it simple, it was opted to take a more general approach
to the estimation methods.
The propellant volume available within the input for dimensions (length, L, and diameter,
D) is thus not the entire volume of a spherically capped cylinder as can be seen in figure
3.4. Namely, the volume depicted in blue, needs to be accounted for. This volume can
be calculated by subtracting a sphere of diameter D from a cylinder with length and
diameter D. A calculation that is performed in Eq. (3.47).

D2 4 D3 1
Vloss = Vcylinder Vsphere = D = D3 (3.47)
4 3 8 12
32 Models

where Vcylinder stands for the volume of a cylinder with both length and diameter equal to
the tanks diameter, Vsphere stands for the volume taken in by a sphere with the diameter
of the tank and Vloss stands for the resulting difference in volume between those two
bodies. The total volume of the both propellant tanks with diameter, D, and sum of
lengths, L, is then calculated as the volume of a cylinder with length L D and diameter
D added with the volume of a sphere with diameter D, minus the difference in volume
calculated in Eq. 3.47.

Vtotal = Vcylinder + Vsphere Vloss

D2 4 D3 1
= (L D) + D3 (3.48)
4 3 8 12
1 1
= D2 L D3
4 6

Having acquired values for both the average propellant density and the volume available
in the tanks, the propellant mass can be calculated via a simple multiplication.

mprop = Vtotal prop 0.9 (3.49)

where the factor of 0.9 is a typical value for the fill ratio, the ratio between the propellant
volume and the volume of the storage system, of liquid pump systems. This fill ratio
accounts for the ullage2 volume needed to keep some pressurant in the tank and also to
allow for expansion of the liquids [69].

Tank Mass

One of the most often tank materials used is 6Al-4V titanium alloy [65]. Therefore, the
tank mass will be estimated based on the properties this alloy. These properties are a
tensile yield strength of 900 MPa, a density of 4428 kg/m3 , a modulus of elasticity of 110
6
GPa, a poisson ratio of 0.31 [65] and a coefficient of thermal expansion of 8.8 10C .
Two assumptions must be established. The first one is that the tank shell is part of
the structure of the launcher and must be able to withstand the large compressive loads
during launch. The second one is that the system is assumed to be pump fed due to the
fact that this is normally used with large stages because the advantage of low tank mass
increases significantly in that case [67]. This leads to the assumption that the tank is put
under a maximum ullage pressure of 0.3 MPa.
There are basically two stresses that determine the thickness of the tank. First, there
is the circumferential stress also known as the hoop stress, the other stress is the axial
stress, which will be will be loaded under tension due to the assumption of a load carrying
tank. The axial stress will thus be subjected to the critical buckling stress criterion.
There can also be a natural-frequency requirement, but this requires assumptions concern-
ing the natural frequency and [62] notes that the driving requirements for the monocoque
cylinder are bending rigidity and compressive stability.
2
In liquid rockets, ullage is the space within a fuel tank above the liquid propellant.
3.3 Launch-vehicle model 33

Circumferential stress limit The hoop or circumferential stress must stay below the
given tensile yield strength of 900 MPa and is given by the following formula [52]:

Pc D
y = (3.50)
2 tcircum
In this equation, Pc is equal to the design burst pressure inside the tank times a safety
factor, D is the diameter of the tank and tcircum is the wall thickness needed to keep the
stress below the level required for the 6Al-4V titanium alloy. The design burst pressure
is the pressure at which the shell is likely to fail catastrophically. It is likely to be 1.25 - 2
times (the burst safety factor) higher than the design pressure. The exact value depends
on how conservative the design is chosen to be. For this study, the model was chosen to
not be overly conservative, leading to the selection of a value of 1.5. The design pressure
is set equal to the sum of the hydrostatic pressure given in Eq. (3.51) and the vapor
pressure, which was assumed to be 0.3 MPa.
The hydrostatic pressure is calculated in Eq. (3.51)

Phydro = f luid H a (3.51)

where H is height of the fluid and a is the acceleration the fluid in the tank undergoes
(this would be 9.81 m/s2 on the surface of the Earth). Due to the fact that the propellant
mass is not all the mass of the rocket, H will decrease faster than a will increase under
constant thrust. Therefore, maximum hydrostatic pressure will occur at lift off. This
means that the acceleration at this moment needs to be known. Unfortunately, at this
moment, only the maximum thrust is known, which is given in Eq. (3.42). The total
mass of the launcher is unknown, since this is the purpose of the calculations performed
in the mass model. Therefore, the total vehicle mass used to obtain the acceleration at
lift off is estimated in the model by using the assumption that the propellant mass, which
is known, is 90 % of the total vehicle mass. This value is based on existing launch vehicles
[15].

Critical buckling stress The equation for theoretical cylinder buckling stress, cr , is:

E t
cr = (3.52)
3 (1 ) R

where E is youngs modulus, R is the cylinder radius, is the poisson ratio, t is the wall
thickness and is a reduction factor used to correlate theory to test results and depends
on a parametric parameter, , for cylinders 3 :
r
1 R
= (for R/t < 1500 and L/R < 5) (3.53)
16 t
 
= 1 0.910 1 e (3.54)
3
No other formula was found for cylinders with an L/R ratio larger than 5 or a R/t ratio larger than
1500, so this formula holds through for all inputs in the model
34 Models

with L as the length of the shell.

When this stress given by (3.52) becomes larger than the one due to thrust, bending
moment and thermal stress for a given thickness, it means that this thickness is enough
for the tank to withstand buckling. The stress given by the thrust is straightforwardly
given by (3.55)

Fmax
= (3.55)
Dt
Thermal stress is given by [65]:

E T
= (3.56)
2 (1 )
where is the coefficient of thermal expansion (8.8 106 1K for 6AL-4V titanium) and E
is the modulus of elasticity, 110 GPa and T is the difference between the temperature of
the propellant and the outside temperature. Since the model does not create a distinction
between the fuel tank and the oxidizer tank two assumptions are made. Firstly, T is
set equal to 298.15 20.27 = 277.88 and secondly, a factor is used to lessen the effect of
thermal stress on the system using the non-cryogenic kerosene fuel. This factor was taken
to be the proportion of the total propellant volume needed to contain oxygen. This is
done in the following formula:

kerosene (O/F )ratio 1

f actor = kerosene (O/F )ratio
(3.57)
liquidoxygen 1+ liquidoxygen

The hull tank also has to deal with an additional stress given by the aerodynamic bending
moment due to dynamic pressure.

mbending = Cm q S c = (Cm0 + Cm ) q S c (3.58)

where Cm0 is approximately zero, is the angle of attack, q is the dynamic pressure, S
is the surface area and c is the reference length. This additional load was not modeled
into the mass model since knowledge about the maximum bending load is not known
before optimization of the launch trajectory. Therefore it was assumed that the launchers
would undergo the same order of stresses that conventional launch vehicles undergo and in
section 3.3.3 the entire mass model was then scaled to correspond with current launchers.
The model compares the two stresses for a thickness ranging from 0.0001 m incrementally
increasing with 0.0001 m until the critical buckling stress becomes larger than the stress
induced by thrust and thermal properties. Once this thickness is found, it is compared
with the thickness found from the circumferential stress requirement and the largest value
is selected. This thickness is assumed to be uniform along the outer hull of the tank. The
two spherical caps that do not need to transfer any thrust loads or bending loads are
therefore assumed to be significantly thinner (a thickness of 1 mm). The surface of the
outer hull is modeled as a cylinder with length, L, and diameter, D, with two spherical
caps at the end. This gives for the formula of the surface:

Shull = (L D) D + D2 = L D (3.59)
3.3 Launch-vehicle model 35

This then gives for total tank mass:

mtank = titanium (t (D L ) + D2 0.001) (3.60)

where titanium = 4428 kg/m3 .

Nozzle mass

The only dimensions obtained through the inputs and the performance model are values
for both the exit area and the area ratio. The model uses these in combination with the
chamber pressure through the following formula to estimate the mass of the nozzle [65]
p415:

 
0.1475
mnozzle =  Fmax 0.00225 + (3.61)
Pc

where  is the area ratio, Fmax is the maximum thrust and Pc is the chamber pressure.
This formula is valid for regeneratively cooled nozzles.

Vehicle equipment bay

The vehicle equipment bay mass is modeled using the following method given by [67]:

0.6814
m = 0.404 Mdry (3.62)

This formula has a SEE of 34%.

Fairing

In [56] the fairing estimation method was presented that will be used here. The formula
for fairing mass is as follows

MF airing = C SF airing (3.63)

where S is the wetted area in m2 , M is the mass in kg and C is the fairing unit weight. This
method found very good values estimating various Ariane 5 ECA fairing configurations
by assuming a value of 13.3 kg/m2 for the fairing unit weight and by achieving wetted
area estimates of the fairing by modeling the fairing as having a conical shape in the top
halve and a cylindrical shape in the bottom half [56]. For this reason, the same value of
13.3 kg/m2 was used in the model.
In addition, [56] also mentioned that the payload density within a fairing can be assumed
36 Models

to be around 128.15 kg/m3 . This leads to the following equation for payload mass to
fairing mass ratio:

MP ayload 128.15 VF airing

=
MF airing 13.3 SF airing
(3.64)
128.15 32 r2 h 6.42 r h
= =
2 2
13.3 ( r r + 0.25 h + h r ) r + 0.25 h2 + h
2

Using the total launcher mass and an initial estimation of its payload mass ratio, a fairing
mass estimate can be obtained.

Heat shield mass

The standard vehicle that will be modeled in this study is a non-reusable vehicle. This
being the case, the logical choice for type of heat shield results in an ablative heat shield.
In an ablative heat shield, the total heat transferred must be balanced with the total
amount of material released:

Q = r t S Habl = Mabl Habl (3.65)

where Habl is the heat of ablation of the ablative material. As a standard, a value of 2500
kJ/kg will be used [65].
The peak heat flux occurs at the stagnation point. Heat flux usually has two components,
a radiative component and a convective component. Due to the fact that below orbital
velocity the convective heat flux is far larger than the radiative component, only the
convective component is taken into account. This convective heat flux can be calculated
at the stagnation point (where the flow is laminar) with the following equation [51]:
 n  m
V 1
Q = C (3.66)
0 1000 RN

where for the free stream:

C = 21,969 W/m3/2

n = 0.5

m = 3.0

0 = 1.225 kg/m3

RN is the nose radius of the launcher, which was set to 10% of the radius of the launcher.
In [36] a maximum value in the order of 1 Btu/(ft2 s) or 11,360 W/m2 was found to
occur during launch. This compares with the maximum heat flux allowed after fairing
separation for the Vega launcher, which is 1,135 W/m2 [3]. While the model used in
3.3 Launch-vehicle model 37

Missile DATCOM assumes an ogive nosecone, a spherical cap will be assumed to exist on
top of the ogive, having a radius equal to 10% of the launchers diameter.
In [2], C was modified with the factor xT0.2 in order to find the heat flux downstream the
turbulent boundary layer. Part of this heat flux can be emitted radiatively, the rest will
need to be absorbed through ablation of the heat shield. In [51] the following maximal
radiation heat loss was established for an aluminum tank:

Q =  T 4 (3.67)

where  is the emissivity (0.8 for anodized aluminum), is the Stefan Boltzmann con-
stant (5.6704 108 W/m2 K4 ) and T is the melting temperature (aluminum (933.5 K)).
Inserting the values for aluminum gives a maximal radiation heat loss of 6,075 W/m2 .
The heat shield mass is modeled in a way that all surplus convective heat flux need to be
absorbed by the ablative heat shield following Eq. (3.65).
Most positions along the launchers nosecone have a far lower heat flux than what occurs
at the stagnation point. This is shown via an analytical method and through experiments
[13]. In this reference two spherically blunted cones with different semi-apex angles were
investigated. As a rule of thumb, a doubling of the semi-apex angle resulted in a doubling
of the heating rate ratio at identical chord positions. However, the stagnation point does
not always occur at the tip of the nosecone due to different angles of attack that can
occur during flight. Therefore, the mass of the heat shield was modeled as if the entire
surface of the nosecone must be able to deal with the surplus convective heat flux in the
stagnation points. In all but one simulation, the simulated heat shield mass was found to
be lower than 0.1% of the total gross mass of the launcher.

3.3.3 Launch Vehicle Model Verification

In order to verify that this model accurately approaches reality its results will be matched
with data from ESA Launch Vehicle Catalogue [15]. For this verification, six launcher
stages, 3 kerolox and 3 hydrolox, were selected (based on the availability of the data) to
compare the simulation with. These launcher stages are all first stages from the following
launchers: the Atlas 5, the Zenit, The Delta IV, The Ariane V ECB, the Soyuz and the
H2. The data of these stages and their simulated counterparts is displayed in Tables 3.4
and 3.5 and shows that there is already a close match between the real and simulated data
points. However, this can be further improved by changing some values that were assumed
throughout section 3.3. For example, the specific impulse of the simulations is fairly close
to the actual values with a standard deviation of 7.5 seconds. Increasing the simulated
values with 1.35% would however minimize the standard deviation to just 5.3 seconds.
This can be done by increasing the nozzle efficiency from 0.97 to 0.983 while leaving the
combustion efficiency at 0.95. More importantly, the dry mass of stages is consistently
underestimated by the simulation. The same holds for the engine mass, where on average
the engine mass is estimated to be 16.23% lower than in reality. Besides the fact that
there is a very close match for liquid hydrogen propelled stages that use boosters for their
assist there is no other remarkable connection to be distinguished. It was therefore chosen
to optimize the parameters in Eq. (3.44) and (3.43) to minimize the standard deviation
38 Models

Figure 3.5: Hydrolox engine weight according to the original method (Eq. (3.44)), the optimized
method (Eq. (3.68)) and the actual data points of the 3 hydrolox engines according to
[15].

with the data of the six launchers.

This optimization showed that the following formula is optimal for hydrolox engines:

m = 0.00005755 F 1.235275 (3.68)

which has a standard deviation of 4.26%. Note that this formula now indicates an ex-
ponential growth in mass for a linear increase in thrust, which was not the case in the
original Eq. (3.44). The following is then likewise found for kerolox engines:

m = 0.000345 F 1.093032 (3.69)

which has a standard deviation of 7.51%. The improvement with the original method
for both types of engines is visualised in figures 3.5 and 3.6 where the data points, the
actual engine weights of the six launchers according to [15], are also plotted for reference.
Implementation of both the improvements in the mass estimation method as well as the
thrust coefficient correction factor leads to tables 3.6 and 3.7. Here it can still be seen that
the dry mass of the vehicles is still underestimated. It can not accurately be determined
which factors weigh how much in this discrepancy. Therefore it was chosen to apply a
certain factor to the dry mass in order to minimize the standard deviation between the
simulated results and the actual vehicles. This minimization of the standard deviation
requires a 30% increase of estimated dry mass resulting in a decrease of the standard
deviation from 4.44% to 1.95%. At this point it is interesting to perform a preliminary
opimization of a launcher stage with the objective of maximizing the V and comparing
the results with the actual launcher stages that have the maximum V from the group of
six. These are the Ariane V ECB for hydrolox stages and the Atlas V for kerolox stages.
When optimizing the hydrolox launcher stage, the following values for the six parame-
ters are found using a differential evolution algorithm from an external software package
named PaGMO (Parallel Global Multiobjective Optimizer), which is further discussed in
chapter 6. The results presented are those obtained after the algorithm has optimized
3.3 Launch-vehicle model 39

Figure 3.6: Kerolox engine weight according to the original method (Eq. (3.43)), the optimized
method (Eq. (3.69)) and the actual data points of the 3 hydrolox engines according to
[15].

a population of 20 individuals over 1000 generations. The reasoning behind the search
spaces is presented in Chapter 7:

Length = 100 m (search space: 10 - 100 m)

Slenderness: 12.49 or a diameter of 8.01 m (search space: 1.82 - 20)

Nozzle-to-tank-diameter ratio: 0 (search space: 0 - 1)

Oxidizer to fuel ratio: 7 (search space: 1 - 10)

Chamber pressure: 254.06 (search space: 8 - 264 bar)

Exit pressure: 0 Pa (search space: 0 - 100,000 Pa)

These values give a maximum V of 17,674 m/s, which is well in excess of the Ariane Vs
simulated V of 10,761 m/s. Also interesting to note is that the optimal values approach
the boundary of their respective search spaces on three occasions. Namely, the upper
boundary for the length, the lower boundary for the nozzle-to-tank-diameter ratio and
the lower boundary for the exit pressure.
The reason why the the nozzle-to-tank diameter is minimized is because this generates a
positive effect for the engine mass (see Eq. (3.68)). Secondly, this also means that there
is almost no thrust generated which leads to almost no hydrostatic pressure and thus a
very small tank thickness.
The reason why the length is being maximized is due to two reasons. Due to the square-
cube law, which dictates that when an object increases proportionally in size, the new
volume is proportional with the cube of the multiplier while the surface area is propor-
tional with its square. And this causes the construction mass ratio to drop in this case.
Normally, this effect is counteracted by the fact that length influences the hydrostatic
pressure by a 1 to 1 ratio due to the formula given in (3.51), which causes a thicker tank
thickness for longer tanks, however in this case the hydrostatic pressure is non-existent
40 Models

and there is no incentive to keep tank length low.

The reason why the exit pressure is minimized is due to the desired maximization of the
specific impulse and it also helps to minimize the thrust.
The same was done with the kerolox launcher stage. The optimized values found can be
seen below:

Length = 100 m (search space: 10 - 100 m)

Slenderness: 9.36 or a diameter of 10.69 m (search space: 1.82 - 20)

Nozzle-to-tank-diameter ratio: 0 (search space: 0 - 1)

Oxidizer-to-fuel ratio: 2.5 (search space: 1 - 10)

Chamber pressure: 8 bar (search space: 8 - 264 bar)

Exit pressure: 0 Pa (search space: 0 - 100,000 Pa)

The kerolox stage is found to have a maximum V of 21,209 m/s. Again, this is way
higher than the simulated Atlas V rocket stage, which has an estimated V of just 9,092
m/s. A striking difference between the optimized kerolox launcher and the hydrolox
launcher is that in addition to length, nozzle-to-tank-diameter ratio and exit pressure
also the chamber pressure for the kerolox stage finds the lower boundary to be the most
optimal solution. The chamber pressure is identified to have two effects on the model.
These are an effect on the nozzle mass through Eq. (3.61) and an effect on the specific
impulse through the pressure ratio as will be further explained in section 7.2.6. However
both effects do not contribute in this optimization since the maximum thrust is near
zero leading to a near zero nozzle mass. Since the exit pressure is also near zero, the
pressure ratio is very large for all possible values for the chamber pressure and at very
large pressure ratios, the change in specific impulse with further increase of the pressure
ratio is almost non-existent. Therefore, it can be concluded that the chamber pressure
does not influence the optimization process when the exit pressure and thrust are close
to zero.
As mentioned, both optimizations leave a high V but ineffective rocket stage with no
thrust. For this reason a constraint was put in place that demanded the vacuum initial
thrust-to-weight ratio to be higher than 1.15. This resulted in much closer matches of V
when comparing to the existing launchers. Putting the constraint in place generated a
maximum V for the hydrolox launcher of 10,618 m/s with the following characteristics:

Length = 12.23 m (search space: 10 - 100 m)

Slenderness: 6.17 or a diameter of 1.98 m (search space: 1.82 - 20)

Nozzle-to-tank-diameter ratio: 0.46 (search space: 0 - 1)

Oxidizer-to-fuel ratio: 6.95 (search space: 1 - 10)

Chamber pressure: 243.36 bar (search space: 8 - 264 bar)

Exit pressure: 9,416 Pa (search space: 0 - 100,000 Pa)

3.3 Launch-vehicle model 41

It should be pointed out that although the real Ariane V first stage has a larger V ,
it does not have an adequate initial thrust to weight ratio and it needs to be assisted
during the first phase of its launch by solid rocket boosters. The results obtained now are
vastly different than when generated without the constraint. There are no parameters
that approach a boundary of the search space because the requirement of high thrust
leads to significant values of exit pressure and nozzle-to-tank-diameter ratio because if
one of those two would be equal to zero so would the maximum thrust be. The thrust
requirement also leads to significant hydrostatic pressure for launchers with a large length.
This is the reason why the length of the launcher retreats from its maximum value of 100
m to only 12.23 m.
Also for the kerolox launcher a close match with the Atlas V V of 9,092 m/s was found.
The optimized kerolox launcher has a V of 10,327 m/s with the following values:

Length = 12.75 m (search space: 10 - 100 m)

Slenderness: 7.58 or a diameter of 1.68 m (search space: 1.82 - 20)

Nozzle exit-to-tank diameter ratio: 0.58754 (search space: 0 - 1)

Oxidizer-to-fuel ratio: 2.5 (search space: 1 - 10)

Chamber pressure: 255.57 bar (search space: 8 - 264 bar)

Exit pressure: 15,557 Pa (search space: 0 - 100,000 Pa)

This result is also vastly different compared with the original when the thrust-to-mass
constraint is put in place. It shows that just like in the case of the hydrolox launcher
and for the same reason, none of the parameters approach any of their search spaces
boundaries.
 42
Table 3.4: Comparison launch vehicle catalogue values and vehicle model for first stages

Atlas 5 (Real) Atlas 5 Zenit Zenit Delta IV Delta IV

[15] (Simulated) (Real) (Simulated) (Real) [15] (Simulated)
[15]
Type of Fuel RP1 RP1 RP1 RP1 LH2 LH2
O/F Ratio 2.72 2.72 (+ 0.0%) 2.63 2.63 (+ 0.0%) 6 6 (+ 0.0%)
Vacuum Thrust (kN) 4,152 4,244 (+ 2.2%) 7,911 7,930 (+ 0.2%) 3,314 3,320 (+ 0.2%)
Expansion Ratio 36.87 36.87 (+ 0.0%) 37 36.97 (- 0.1%) 21.5 21.49 (+ 0.0%)
Chamber Pressure (bar) 257 257 (+ 0.0%) 250 250 (+ 0.0%) 97.2 97.2 (+ 0.0%)
Dry Mass (kg) 21,173 13,980 (- 34.0%) 28,080 21,402.5 (- 23.8%) 26,760 18,031 (- 32.6%)
Propellant Mass (kg) 284,089 284,405 (+ 0.1%) 318,800 318,427 (- 0.1%) 199,600 200,477 (+ 0.4%)
Engine Mass (kg) 5,390 4,611 (- 14.5%) 13,000 8,613 (- 33.8%) 6,600 4,441 (- 32.7%)
Length (m) 32.46 30.25 (- 6.8%) 33 32.3 (- 2.1%) 40.8 33.2 (- 18.6%)
Diameter (m) 3.81 3.81 (+ 0.0%) 3.9 3.9 (+ 0.0%) 5.13 5.13 (+ 0.0%)
Vacuum Isp (sec) 337.8 330 (- 2.3%) 337 330.9 (- 1.8%) 410 410 (+ 0.0%)
Nozzle to Tank Ratio - 0.65 - 0.88 - 0.547
Exit Pressure (Pa) - 51,000 - 49,300 - 35,800

Models
 3.3 Launch-vehicle model
Table 3.5: Comparison launch vehicle catalogue values and vehicle model for first stages

Ariane 5 ECB Ariane 5 ECB H2 (Real) H2 Soyuz Soyuz

(Real) [15] (Simulated) [15] (Simulated) (Real) [15] (Simulated)
Type of Fuel LH2 LH2 LH2 LH2 RP1 RP1
O/F Ratio 6.1 6.1 (+ 0.0%) 6 6 (+ 0.0%) 2.39 2.39 (+ 0.0%)
Vacuum Thrust (kN) 1,350 1,351 (+ 0.0%) 1,079 1,083 (+ 0.3%) 971 961 (- 1.0%)
Expansion Ratio 60 60.14 (+ 0.2%) 52 52.22 (+ 0.4%) 18.94 18.88 (- 0.1%)
Chamber Pressure (bar) 116.2 116.2 (+ 0.0%) 127 127 (+ 0.0%) 51 51 (+ 0.0%)
Dry Mass (kg) 12,300 11,744 (- 4.5%) 11,900 7,138 (- 40.0%) 6,500 3,381 (- 48.0%)
Propellant Mass (kg) 173,000 173,228 (+ 0.1%) 86,200 86,086 (- 0.1%) 94,500 94,496 (+ 0.0%)
Engine Mass (kg) 2,040 2,055 (+ 0.7%) 1,714 1,700 (- 0.1%) 1,250 1,045 (- 16.4%)
Length (m) 30.5 26.6 (- 12.8%) 28 23.7 (- 15.4%) 27.8 22.5 (- 19.1%)
Diameter (m) 5.4 5.4 (+ 0.0%) 4 4 (+ 0.0%) 2.555 2.55 (+ 0.0%)
Vacuum Isp (sec) 434 432.3 (- 0.4%) 446.1 430.7 (- 3.5%) 315 315 (+ 0.0%)
Nozzle to Tank Ratio - 0.48 - 0.52 - 0.77
Exit Pressure (Pa) - 13,200 - 16,900 - 21,700

4
This was not taken from the ESA launch vehicle catalogue but rather from [42]

43
5
Varying from 2.95 to 2.15
 44
Table 3.6: Comparison launch vehicle catalogue values and vehicle model for first stages

Atlas 5 (Real) Atlas 5 Zenit Zenit Delta IV Delta IV

[15] (Simulated) (Real) (Simulated) (Real) [15] (Simulated)
[15]
Type of Fuel RP1 RP1 RP1 RP1 LH2 LH2
O/F Ratio 2.72 2.72 (+ 0.0%) 2.63 2.63 (+ 0.0%) 6 6 (+ 0.0%)
Vacuum Thrust (kN) 4,152 4,225 (+ 1.8%) 7,911 7,887 (- 0.3%) 3,314 3,323 (+ 0.3%)
Expansion Ratio 36.87 36.87 (+ 0.0%) 37 36.97 (- 0.1%) 21.5 21.49 (+ 0.0%)
Chamber Pressure (bar) 257 257 (+ 0.0%) 250 250 (+ 0.0%) 97.2 97.2 (+ 0.0%)
Dry Mass (kg) 21,173 14,622 (- 30.9%) 28,080 24,790 (- 11.7%) 26,760 20,234 (- 24.4%)
Propellant Mass (kg) 284,089 284,405 (+ 0.1%) 318,800 318,427 (- 0.1%) 199,600 201,150 (+ 0.8%)
Engine Mass (kg) 5,390 6,026 (+ 11.2%) 13,000 11,923 (- 8.3%) 6,600 6,544 (- 0.8%)
Length (m) 32.46 30.25 (- 6.8%) 33 32.3 (- 2.1%) 40.8 33.2 (- 18.6%)
Diameter (m) 3.81 3.81 (+ 0.0%) 3.9 3.9 (+ 0.0%) 5.13 5.13 (+ 0.0%)
Vacuum Isp (sec) 337.8 334.5 (- 1.0%) 337 335.3 (- 0.5%) 410 415 (+ 1.2%)
Nozzle to Tank Ratio - 0.64 - 0.866 - 0.54
Exit Pressure (Pa) - 51,000 - 49,300 - 35,800

Models
 3.3 Launch-vehicle model
Table 3.7: Comparison launch vehicle catalogue values and vehicle model for first stages with updated engine

Ariane 5 ECB Ariane 5 ECB H2 (Real) H2 Soyuz Soyuz

(Real) [15] (Simulated) [15] (Simulated) (Real) [15] (Simulated)
Type of Fuel LH2 LH2 LH2 LH2 RP1 RP1
O/F Ratio 6.1 6.1 (+ 0.0%) 6 6 (+ 0.0%) 2.39 2.39 (+ 0.0%)
Vacuum Thrust (kN) 1,350 1,359 (+ 0.7%) 1,079 1,070 (- 0.8%) 971 961 (- 1.0%)
Expansion Ratio 60 60.14 (+ 0.2%) 52 52.22 (+ 0.1%) 18.96 18.88 (- 0.1%)
Chamber Pressure (bar) 116.2 116.2 (+ 0.0%) 127 127 (+ 0.0%) 51 51 (+ 0.0%)
Dry Mass (kg) 12,300 11,866 (- 3.5%) 11,900 7,043 (- 40.8%) 6,500 3,538 (- 45.6%)
Propellant Mass (kg) 173,000 173,228 (+ 0.1%) 86,200 86,086 (- 0.1%) 94,500 94,496 (+ 0.0%)
Engine Mass (kg) 2,040 2,170 (+ 6.4%) 1,714 1,614 (- 5.8%) 1,250 1,194 (+ 4.5%)
Length (m) 30.5 26.6 (- 12.8%) 28 23.7 (- 15.4%) 27.8 22.5 (+ 19.1%)
Diameter (m) 5.4 5.4 (+ 0.0%) 4 4 (+ 0.0%) 2.557 2.55 (+ 0.0%)
Vacuum Isp (sec) 434 438.1 (+ 0.9%) 446.1 436.4 (- 2.2%) 315 319 (+ 1.2%)
Nozzle to Tank Ratio - 0.475 - 0.51 - 0.76
Exit Pressure (Pa) - 13,200 - 16,900 - 21,700

6
This was not taken from the ESA launch vehicle catalogue but rather from [42]

45
7
Varying from 2.95 to 2.15
46 Models

3.4 Thrust model

The actual launch vehicle that uses an assist will still need a propulsive force to reach Earth
orbit. This propulsive force is considered to originate singularly from rocket propulsion
and is provided according to Eq. (3.70), the momentum thrust, and Eq. (3.71), the
pressure thrust.

Tmomentum = m Ue (3.70)

Tpressure = Ae (pe pa ) (3.71)

where m is the mass flow, Ue is the exhaust velocity, Ae is the nozzle exit area, pe is the
nozzle exit pressure and pa is the ambient pressure. Rocket thrust for all stages of flight
will be modeled as follows w.r.t the body fixed reference frame:

Tmomentum + Tpressure
T = 0 (3.72)
0

The simulation model only provides the option to choose between two types of fuels in
order to generate thrust. Both fuels, kerosene and liquid hydrogen, are liquid fuels. This
makes the thrust model quite straightforward since liquid engines generally burn through
their fuel at a steady rate due to the constant pressure in the combustion chamber. More
advanced and expensive types of engines can be throttled as desired. Although in reality
this cannot be done at full efficiency through the whole range of thrust from 0 N to max-
imum thrust, full efficiency and throttle ability is assumed in this study.
All forms of rocket propulsion operate according to the third law of Newton and acceler-
ate mass and expel it, resulting in an equally great force in the opposite direction that
propels the rocket. The speed to which the mass is accelerated out of the rocket is vitally
important for the effectiveness of the rocket as was shown by Konstantin Tsiolkowski:

mi
V = Isp g0 ln( ) (3.73)
mf

where Isp is the specific impulse in seconds, g0 is the gravity at sea level, namely 9.80655
m/s2 , mi is the initial mass in kg and mf is the final mass. The product of the specific
impulse and the gravity at sea level is the exit velocity, which was given in Eq. (3.40) in
section 3.3.1.

3.5 Analysis of the velocity budget

The main focus of this study is on the velocity or V budget and how initial conditions
that could be created by use of assisted launch techniques effect that budget. Therefore, a
3.5 Analysis of the velocity budget 47

thorough study of this budget is definitely in order. The mission budget for space launch
to a certain Earth orbit is often represented as follows:

V = Vinj + Vd + Vg + Vs (3.74)

where Vinj is the injection velocity required for the desired orbit, Vd is the drag loss,
Vg is the gravity loss and Vs is the steering loss up until the point of injection.
In section 3.4, Eq. (3.73) gave a definition of the launcher V budget. This budget
must always be larger or equal to the mission V budget for the obvious reason that the
desired payload would not be able to reach the desired orbit, and the mission would thus
not be successful, if otherwise. This means that the mission V budget needs to be as
small as possible, and drag and gravity losses need to be minimized.
Drag loss can be defined by the following equation:
Z t=te
D
Vd = dt (3.75)
t=t0 m

According to Eq. (3.29) the drag force, D, is positively correlated to the velocity, density,
reference surface area and the aerodynamic coefficient. However, only the former two are
related to the launch trajectory. The mass of the vehicle, m, depends on the dry mass
and the thrust profile of the launcher but will continuously decrease. This means that to
optimize Eq. (3.75) the velocity must be kept low and the mass high until the density
severely decreases.
Another interesting aspect about drag loss can be derived from the drag forces linear
dependence on the reference surface area and its independence on the length of the vehicle
(assuming skin friction drag to be negligible8 ):

Sref Sref 1
Vd = = (3.76)
m0 Sref l rocket l rocket
where m0 is the take-off mass and obviously has a direct relationship with the instanta-
neous mass, which is a factor in Eq. (3.75). In words, imagine three launchers, launcher
A, B and C that all have the same thrust-to-weight ratio. Launcher B is four times as tall
as launcher A, while launcher C is twice as wide (i.e., the mass of each is four times the
mass of launcher A). This means that launcher B and C both need four times as much
thrust but only launcher C experiences a drag force four times as large. This implies that
the drag loss does not increase with increasing diameter and decreases with increasing
length. In addition, also average density of the launcher turns out to be inversely related
to drag loss.
Unfortunately, rather than only have to optimize for the drag loss, the trajectory problem
also needs to optimize for both the thrust loss and the gravity loss. This is quite complex
so it might be beneficial to have a quick look at the equation of gravity loss separately
[58]:
Z t=te
Vg = g sin dt (3.77)
t=t0
8
Varying length would result in varying skin friction drag and thus a different drag coefficient.
48 Models

where g is the local acceleration due to gravity and is the local flight-path angle, which
is defined from the triangle of velocities: tan = VZ /VX (where VX and VZ are the X-
and Z-components of the total velocity V respectively). The local acceleration, g, only
slightly changes with height during the propulsive phase of the launch, with the exception
of an injection burn at geostationary altitude, which is done by some launch vehicles [59].
However, since both of the equations parameters are expressed in a non-inertial reference
frame, the fictitious centrifugal force must be implemented in the formula, if not, eccentric
orbits would have a periodical change in gravity loss. This equation is presented as follows:

Z t=te
Vg = (g r 2 ) sin dt (3.78)
t=t0

where r is the distance to the center of the Earth and is the angular velocity.
The determining factor with regard to gravity loss is the flightpath angle. To keep gravity
losses low, one must keep the flight-path angle low. In the ideal scenario the flight-path
angle would be kept zero at all times (horizontal flight) and gravity losses would amount
to zero. This scenario, which is not viable since it needs an impulsive shot in order to
keep the vehicle from hitting the ground, quite clearly contradicts the scenario of low drag
loss discussed before.
Another aspect of gravity loss is that it not only is a measure for efficiency of the launcher
w.r.t. gravity, but it also incorporates the aspect of kinetic energy traded in for potential
energy. Integrating for gravity loss along a non-disturbed elliptical orbit would result in
a gravity loss between perigee and apogee and an equal gravity gain between apogee and
perigee. In other words, a comparison between the gravity loss of a system that takes a
payload to geostationary orbit and one that takes a payload to low Earth orbit is com-
paring apples with oranges.

3.6 Guidance model

In this section the guidance program as coded into the simulation will be discussed. As
an introduction to guidance programs a trajectory breakdown of the most essential flight
phases will be performed briefly. These flight phases occur at different times for each single
vehicle and the occurrence of some even depend on the configuration and architecture of
a launch vehicle (number of stages, number of boosters, etc.). Nonetheless almost all
conventionally launched vehicles go through the most essential phases summarized below:

Vertical lift-off

Kick phase (Flight phase where the vehicle pitch angle is actively changed so the
right conditions are met to insert into the gravity turn)

Gravity turn (Flight phase with no angle of attack, flight-path angle is only dis-
turbed by the force of gravity)

Control law
3.6 Guidance model 49

Coasting phase(s) and circularization

One problem with this approach is that the first phase (the vertical lift-off) already
requires from the initial condition that the flight-path angle is 90 . This might be the
case for most rocket launches but it is not for those that need to be simulated in this study.
Optimization using the gravity turn and a certain control law was for this study dismissed
in favor of pure optimized pitch control using polynomials. For further information on
these flight phases the reader is referred to [56].

3.6.1 Initial conditions

After the vehicle is separated from its assist platform it will have a certain position,
velocity, mass and attitude. These initial conditions are very important for the shape of
the launch trajectory and need to be correctly inserted into the model that describes the
entire trajectory ascent problem.

3.6.2 Coasting phase(s) and circularization

The coasting phase and circularization burn are the final phases of the guidance program.
To circularize an orbit, a burn must be made at the apogee of the launch trajectory.
There are no longer any gravitational or aerodynamic disturbing forces and the objective
is simply to reach the necessary horizontal velocity and minimal eccentricity. For certain
combinations of launch vehicles and desired orbits, the time it takes to reach the altitude
of the required orbit is larger than the time the vehicle can burn. In these instances,
the vehicle starts to coast to a higher altitude where it then provides a circularization
burn. To do this, the guidance program must keep track of the apocenter given by the
Keplerian elements of the vehicle state vector. When this apocenter equals the apocenter
of the target orbit, the thrust should be deactivated so that the coasting phase begins.
After some time, the launcher will reach its apogee and the circularization burn can begin.
When the circularization burn is modeled as an impulsive shot, the right burn at apogee
will lead to a circular orbit. However, due to the fact that this is not the case in the
simulation it was chosen to initiate the burn when the launch vehicle altitude is higher
than 98% of the apogee altitude. During this burn, the pitch angle of the vehicle is set
equal to the flight-path angle. The burn ends when the orbit criteria are met. These
are the final eccentricity and the apogee altitude. These can be defined as wished and
correspond in this study with the values 0.01 or lower for final eccentricity and 250 km or
higher for the apogee altitude. For fully optimized simulations (also with initial flight path
angle optimization) the circularization burn can be assumed to be optimally very short.
This due to the fact that it occurs at the highest altitude during the burn sequence which
is the least favorite position to burn fuel in (gravity loss, gravity well, gravity assist).
From [58] we know that:

ra rp
e= (3.79)
ra + rp
Due to the eccentricity requirement of 0.01 a minimum perigee of 120 km can only be
insured if the apogee is higher than 250 km.
50 Models

The optimization of the circularization process requires the simulation to be run for over
one half of the orbital period. Therefore total launch simulation time is set to 3000
seconds.

3.6.3 Verification of the Pitch Polynomial

In order to avoid spiky pitch profiles that can be generated due to node optimization
[38] and would imply instantaneous attitude changes, the pitch was modeled as a smooth
continuous polynomial. Two pitch polynomials have been implemented during the opti-
mization process. The first one presented is a third-degree polynomial of the following
form:

= initial (1 + (a 0.5) w 0.00314 t (b 0.5) x 0.00000628 t2

(3.80)
+ (c 0.5) y 0.00000001256 t3 )

where a, b and c are all optimized parameters with a search space between 0 and 1. Their
weights are chosen so that each of the three parameters will exhibit an equal effect on the
pitch angle when t is 500 s and can have at least a magnitude of 180 after 1000 seconds.
The parameters w, x and y are values that are assigned to a specific optimization by
means of trial and error so that none of the a, b and c parameters find an optimum value
close to their boundaries.
The second polynomial has an added term, making it a fourth degree polynomial:

= initial (1 + (a 0.5) w 0.00314 t + (b 0.5) x 0.00000628 t2

(3.81)
+ (c 0.5) y 0.00000001256 t3 + (d 0.5) z 0.00000000002512 t4 )

In the three degree configuration, the search space was expanded by using the following
values as the w, x and y parameters given in 3.80. w = 3, x = 10 and y = 30, the
added fourth configuration has parameter z set to 1809 . The average optimal value of
2 runs of 5000 generations with a population of 20 individuals was 0.200085 while the
quadruple pitch parameter simulation found an average optimal value of 0.20111. Based
on this result, fourth-order polynomials will be used as the method of choice to establish
the pitch program. The resulting polynomials are shown in figure 3.7.

9
These parameters were for the final results reset to w = 8, x = 20, y = 60 and z = 180 based on the
experience that with using these search spaces very rarely a global optimum was found in the vicinity of
one of the boundaries of the search spaces. It is however recommended for future research to use a more
robust method to establish the search space of these variables.
3.7 Trajectory constraint model 51

0.5
4thdegree polynomial
0.45 3rddegree polynomial

0.4

0.35
Pitch Angle (rad)

0.3

0.25

0.2

0.15

0.1

0.05

0
0 50 100 150 200 250 300
Time

Figure 3.7: Comparison of the two polynomial types that obtained the best objective function value

3.7 Trajectory constraint model

3.7.1 Bending load

One of the requirements that should be enforced throughout all phases of flight is a
structural constraint. Here we run into the problem that a structural study of each of
the launch vehicles cannot be done within the scope of this study. To circumvent this
problem a rule of thumb can be used known as the q limit, this rule originates from the
bending load due to dynamic pressure given in Eq. (3.82).

Mbending = Cm0 + Cm q S c (3.82)

where Cm0 is almost zero, is the angle of attack, q is the dynamic pressure, S is the
surface area and c is the chord length. The stability derivative Cm is the slope of the
moment curve Cm = Cm () at the respective angle of attack at which Cm0 = 0 [33]. Since,
S, c and Cm are nearly constant, the q limit is linearly connected with the bending
load. The multiplication of the dynamic pressure with the angle of attack is therefore
an important structural constraint. Assigning a certain value for this limit will have a
definite impact on the results of the optimization. As a guideline, from a study by both
NASA and DARPA on horizontal launch systems that all used lifting surfaces it could
be revealed that an assumption was used of a bending load limit of 5010 lbs deg/ft2
or 36.135 Pa rad [35]. It should also be noted that the relevant angle of attack in Eq.
(3.82) is the total angle of attack which can be calculated as follows [38]:

t = arccos(cos cos ) (3.83)

52 Models

However, since the side slip angle was during all simulations set to zero, the total angle
of attack is nothing more than the normal angle of attack. The bending load constraint
was not approached by any of the optimal simulations.

3.7.2 Axial acceleration

Another frequently used constraint in trajectory optimization is an axial acceleration

constraint given in the following equation:

T (t) D(t)
ax = (3.84)
m g0

During launch, fuel is drained from the launch vehicle reducing its weight significantly.
All the while the engine can still perform at the same thrust causing the axial acceleration
to increase. At some point, the thrust need to be throttled to prevent critical failure of
either the launcher or the payload. During most simulations this constraint has been set
at 50 m/s for all simulations. This value was based on 6 different launchers, which had an
average maximum axial acceleration of 4.86 g [38]. Implementation of this constraint was
due to the nature of the thrust-force model different than implementation of the other
constraints. The thrust force model assumes constant thrust with adjustments due to
air pressure unless the acceleration constraint would be violated. In case of violation,
the thrust and mass flow is modified to its highest possible value without violating the
acceleration constraint. In figure 3.8 all hydrolox axial-acceleration profiles can be viewed.
The small peaks that violate the constraint are due to the fairing jettisoning event, where
the thrust was still calculated based on the old mass (with fairing) while the mass was
already decreased in that time step to the launcher mass without fairing.

3.7.3 Angle of attack

Normally the most meaningful way to implement an angle-of-attack constraint is how it

is implicitly done in the bending moment constraint, see Eq. (3.82). However, due to the
limits of the Missile DATCOM simulator the angle of attack has been hard constrained
to 30 degrees or 0.5236 rad. Figure 3.9 displays several angle-of-attack profiles that are
subjected to this constraint. It should be noted that the only cases that came close to
the limit are the ones that do not utilize a velocity assist. All angle-of-attack profiles
can be viewed in Appendix C. This constraint is implemented by discarding all solutions
that violate it, which is different from the previously discussed implementation of the
axial-acceleration constraint.

3.7.4 Heat Flux

The heat flux already has an effect on the launch vehicle mass model through the heat
shield mass, which was discussed in section 3.3.2. Thus as long as the model is able to
3.7 Trajectory constraint model 53

50

40
Total Acceleration (m/s2)

30

20

10

0 200 400 600 800 1000 1200

Time (sec)

Figure 3.8: Example of how the axial-acceleration constraint is met for hydrolox vehicles.

0.6
0 m / 0 m/s
5,000 m / 0 m/s
0.5
10,000 m / 0 m/s
20,000 m / 0 m/s
0.4 40,000 m / 0 m/s
0 m / 300 m/s
Angle of Attack (rad)

0.3

0.2

0.1

0.1

0.2
0 50 100 150 200 250 300
Time (s)

Figure 3.9: Example of how the angle-of-attack constraint is met for hydrolox vehicles. The legend
contains case identifiers using the h/V notation.
54 Models

1200
40,000 / 0 (RP1)
40,000 / 0 (LH2)
20,000 / 300 (RP1)
1000 20,000 / 300 (LH2)
40,000 / 300 (RP1)
40,000 / 300 (LH2)
40,000 / 600 (RP1)
800 40,000 / 600 (LH2)
Heat Flux (W/m3)

600

400

200

0
0 50 100 150 200 250 300 350 400
Time (sec)

Figure 3.10: Example of how the heat-flux constraint is met. The legend contains case identifiers
using the h/V (type of fuel) notation.

cope with the extra weight of the heat shield, the heat flux is essentially not constrained.
The heat flux was only modeled as a hard constraint after fairing jettisoning. The value
of this constraint was the same as the jettisoning criteria for the fairing, namely 1135
W/m2 . This resulted in a few interesting launch trajectory profiles further discussed in
section 7.3. These trajectory profiles can also be seen in figure 3.10, where for some cases,
the heat flux is clearly modeled to stay below the jettison criteria.

3.7.5 Altitude

The trajectory of the launcher is logically hard constrained to not go under the radius of
the Earth. This constraint was not approached by any of the optimal simulations.

3.7.6 Sidenote

Other constraints that trajectory simulation software sometimes takes into account is
staging re-entry locations and visibility for ground stations. In addition, a major require-
ment will be to not hit nor endanger the assist platform. This can be a quite significant
constraint in the case of plane assist. These constraints require specific knowledge about
the launch-vehicle geometry, type of assist and/or geography of the launch site. To reduce
complexity of this study, it is decided to not incorporate this kind of info as inputs of the
simulation nor will some consequences be taken into account. It is recommended to take
these constraints into account in an advanced stage of a specific system design study. An-
other often used constraint is the dynamic pressure. While this was contemplated during
the design process it was quickly found out that the maximum dynamic pressure of all
3.7 Trajectory constraint model 55

optimized cases is the initial condition of the most extreme launch assist10 . Therefore,
it did not seem prudent to use dynamic pressure as a direct constraint but was it rather
implemented indirectly through the bending moment, see Eq. (3.82).

10
The most extreme launch assist in this case is meant to be the combination of an assist velocity of
2,000 m/s at an altitude of 0 m.
56 Models
 Chapter 4

Numerical Integration

The computation of the trajectory has to be done with a certain accuracy. To this
purpose, a small overview was presented in [56] about the usage of numerical methods
for the solution of the equations of motion. It was then decided that the main purpose of
the integrator in this model was to maximize the speed and the single step fourth order
Runge-Kutta (RK) method was then chosen since it outperforms multi-step methods on
this criteria partially due to the fact of its stability at larger step sizes. This chapter will
introduce the reader to the RK4 method in section 4.1 and will provide an explanation
with regard to which step size was used for the simulations in section 4.2.

4.1 Runge-Kutta-4

During single step method numerical integration, no use is made of function values calcu-
lated in earlier steps to predict the value of the function at the next step. This means that
all integration steps are completely independent of one another. To discuss the various
single-step methods the following differential equation will be introduced.

y = f (t, y) with y, y, f Rn (4.1)

Starting from initial values y0 = y(t0 ) at time t0 a numerical integration method seeks to
approximate y at a later time ,t0 +h1 , as closely as possible by using an increment function,
, that closely approximates the slope of the secant through (t0 , y0 ) and (t0 +h, y(t0 +h)).
In a mathematical notation this means:

y(t0 + h) y0 + h = (t0 + h) (4.2)

where (t0 + h) is thus the approximate solution. One of the most straightforward meth-
ods just equates the increment function to the slope of y at its initial point. This
1
h is the time step

57
58 Numerical Integration

may deviate considerably from the real slope and needs a very small step size in order
to guarantee great accuracy. This method is called the Euler forward method and is
mathematically presented in Eq. (4.3)

y(t0 + h) y0 + h y0 = y0 + h f (t0 , y0 ) (4.3)

There are other variations from this method such as the Eulers backward method and the
higher order modified Euler method. For more information the reader is referred to [57].
To further improve on the accuracy of numerical integration, a series of methods have
been developed by mathematicians Carl Runge and Wilhelm Kutta. The method series,
aptly named the Runge-Kutta methods, are based on the slopes at various points within
the integration step. As opposed to just the slope at the initial point, the approach used
in the Euler method, the RK4 method is based on slopes in four points and the increment
function is calculated as a weighted mean of the four slopes, k1 , k2 , k3 and k4 :
1
RK4 = (k1 + 2k2 + 2k3 + k4 ) (4.4)
6

The four slopes are defined as follows:

k1 = f (t0 , y0 ) (4.5)

k2 = f (t0 + h/2, y0 + hk1 /2) (4.6)

k3 = f (t0 + h/2, y0 + hk2 /2) (4.7)

k4 = f (t0 + h, y0 + hk3 ) (4.8)

The Runga-Kutta series all use the same build-up but can be extended with even more
stages. The general formulation of an s-stage RK formula is as follows:

k1 = f (t0 + c1 , y0 ) (4.9)

i1
X
ki = f (t0 + ci h, y0 + h aij kj ) with (i = 2, ..., s) (4.10)
j=1

And the increment function is defined as follow:

s
X
= bi ki (4.11)
i=1

The coefficients are usually determined such that they obey the following relations:

s
X i1
X
bi = 1; c1 = 0; ci = (i > 1) (4.12)
i=1 j=1
4.2 Selection of the time step 59

As can be seen in Eq. (4.5) to Eq. (4.8), the order of h in ki usually increases with every
increase in i, there are however exceptions and the real order of each method needs to be
determined through an elaborate mathematical examination. The local truncation error
of RK4 is given as follows

eRK4 = |y(t0 + h) (t0 + h)| const h5 (4.13)

The error is thus bound by a term of order h5 .

Now say that a certain s-stage RK formula will have an approximation of the exact
solution up to terms of order hs . This requires that bs 6= 0. However, another Runge-
Kutta method can have the same number of stages as the previous one but with bs = 0.
This RK method now has an approximation of the exact solution up to terms of order
hs1 . Two of these methods with the same number of stages but with a different order,
p and p + 1, give the following truncation errors:

e1 = |y(t0 + h) 1 (t0 + h)| const hp+1 (4.14)

e2 = |y(t0 + h) 2 (t0 + h)| const hp+2 (4.15)

Now it can be said that e2 is smaller than e1 by the order of h (a small value) it results
in the following error estimation:

e1 = |y | | 2 1 | (4.16)

This means that the estimate of the local truncation error is based on the difference of the
two solutions. This estimate of error can in turn be used for step size control. Suppose
that this error value is larger than a certain tolerance, . In this case the step needs to
be repeated with a smaller step size, h , for the tolerance to be met. The new step size
is found to be:
r r
 
h = p+1 h p+1 h (4.17)
e(h) | 2 2 |

where e(h) is the error due to the original step size. In practice about 0.9 times this value
is commonly used. This still does require the user to define the initial step size but this
choice might be directed by experience with earlier simulations of the same problem with
the same integrator. Some of the more popular methods that use this technique are called
the Runge-Kutta-Fehlberg methods.

4.2 Selection of the time step

To select the time step with which the model would be integrated, 100 randomly varying
inputs were generated for the four variables of the pitch profile, see section 3.6.3. These
inputs were varied around the found optimal values with a maximum deviation of -0.005%
and 0.005% of their search spaces. All inputs were used to simulate the hydrolox case
60 Numerical Integration

utilizing an assist velocity of 300 m/s and an assist altitude of 20 km. Each input was
simulated with eight different time steps. The different time steps were 0.01, 0.02, 0.05,
0.1, 0.2, 0.5, 1.0 and 2.0 seconds. The results of the simulations with a time step of 0.01
seconds were used as a measure of approximate reality. These results had a mean altitude
of 242,710 m with a standard deviation of 1,537 m and a payload ratio of 0.0534 with a
standard deviation of 1.426 105 . The results of the other runs were tabulated in table
4.1 with the mean difference w.r.t the 0.01 second run and with the standard deviation
around its own mean.
Table 4.1: Comparison of the results from different time step RK4 runs

Time Mean altitude Standard Mean payload Standard

Step difference deviation altitude ratio difference deviation payload
ratio
0.02 1.1584 103 1.3618 103 -4.3841 105 2.7498 105
0.05 6.3627 103 3.2122 103 -2.0301 104 5.0271 105
0.1 1.3318 104 1.0369 104 -4.5621 104 1.5352 104
0.2 3.9164 104 9.2364 103 -0.001 1.6368 104
0.5 1.8296 106 1.8468 106 -0.0271 0.0258
1.0 1.5456 106 1.3029 106 -0.0315 0.0247
2.0 3.6826 106 1.1369 105 -0.0534 1.4188 105

For time steps larger than or equal to 0.5 seconds the standard deviations are about the
same magnitude as the mean values. This occurs because the orbit criteria, see section
3.6.2, are sometimes not met resulting in large discrepancies at the end of the simulation.
The inability to meet the orbit criteria is due to the circularization burn, which is defined
to start when the altitude is higher than 98% of the desired apogee altitude. This way,
the burn is not modelled as an impulsive shot but as a realistic continuous burn. During
this burn the pitch angle is set equal to the flight-path angle. In the case of circularization
with an impulsive shot, it is optimal to burn exactly at the apogee. However since the
modelled burn is not impulsive it does not exactly take place at the apogee and some
variations of the optimal solution burn too long or too short, taking the vehicle over the
desired apogee altitude causing a failure to insert into the desired orbit.
With a time step of 0.2 seconds or less, the orbit criteria are always met, as should be
the case. Furthermore, the mean difference in the payload ratio decreases to less than
one percent of its relative value of 0.0534 as soon as a time step equal to or shorter than
0.1 seconds is used. This difference is actually mainly caused by the more precise thrust
cutoff when the launcher reaches the required orbit instead of being a consequence of the
different trajectory generated by using a different trajectory. Decreasing the time step
naturally causes computational time to increase. A decrease in the time step with a factor
of ten translates to an increase in evaluation time of 5 to 6 for the entire optimization
problem. This is due to some calculations that only need to be done once and not
during every time step. The entire optimization was also run with different time steps
and created an increase in optimized payload ratio of about the same magnitude of the
observed increase in the single simulation. This indicates that the time step does not
influence the optimization process. Therefore it was concluded that a time step of 0.1
seconds is sufficient for the purpose of this study.
 Chapter 5

Optimization Techniques

The goal of this study is to compare the performance of various launchers utilizing various
assist cases. To achieve this goal with a high degree of accuracy, the launch vehicle and
the trajectory that this vehicle will follow in every case needs to be optimized. This
chapter gives the reader an introduction to optimization and the optimization algorithm
used.

5.1 General optimization problem

The general optimization problem is to find a set of independent variables that either
minimize or maximize the outcome of a system characteristic that is dependent on these
variables. The mathematical formulation of an optimization problem, in this case the
minimization problem, can thus be written as follows:

min f (x) (5.1)

subject to x (5.2)

The vector x is an n-dimensional vector of independent variables, that is, x = (x1 , x2 , ...,
xn )T Rn . The function f : Rn R that we wish to minimize is called the objective
function. The set is a subset of Rn , called the constraint set [10]. The constraints
of this set can be in the form of equality constraints, denoted by h(x) and inequality
constraints, denoted by g(x):

= x : h(x), g(x) (5.3)

where

h(x) = 0, for i = 1, 2, ..., j (5.4)

61
62 Optimization Techniques

g(x) 0, for i = j + 1, ..., n (5.5)

where j is the number of equality constraints and n j is the number of inequality con-
straints. There is no limit to the number of inequality constraints whereas the number
of equality constraints can not surpass the number of variables, since each equality con-
straint defines a variable and more equality constraints than variables would render the
system redundant. An optimization problem can also require maximization of the ob-
jective function. These problems have the same representation as displayed above, since
maximizing f is equivalent to minimizing f . The above problem is the general form
of a constrained optimization problem. If = , then the problem is referred to as an
unconstrained optimization problem.
Optimization problems are often rather complex and therefore analytical techniques sel-
dom suffice to obtain a solution to the problem. This is the reason that one generally
has to rely on numerical methods to obtain solutions for the more difficult problems.
Trajectory optimization definitely falls into that category.

5.1.1 Global optimization methods

There are two general types of optimization techniques that can be used to solve a complex
problem. These are called global and local optimization techniques. When the problem
is convex1 i.e. when all the functions in the problem are convex, any local minimizer
is global and local optimization techniques can be used as well as global optimization
techniques. However, for problems that are not convex, such as the maximization of
payload-to-weight-ratio that this study is dealing with, the local optimization technique
can never be relied upon to find the global minimum. The difference between a global
peak and a local peak can be explained using figure 5.1.
A local optimizer also needs an initial guess that acts as a starting point from which
it can start the search to the local optimum in its vicinity. This requires a minimum
knowledge about the problem at hand. Global optimizers do not need initial guesses
since these are created randomly in the search space. Therefore the following discussion
of numerical optimization techniques will be limited to global optimizers. For information
about local optimizers the reader is referred to [56].

Evolutionary algorithms One type of evolutionary algorithms are the genetic algo-
rithms, which were the first global optimization technique to be developed [20]. The
algorithm has, just as the name reveals, its roots in genetics. The algorithm starts out
with an initial set of points, called the initial population P (0), within the constraint set .
The size of the population is usually taken to be twice to four times the number of design
variables. The method then continues to iteratively perform operations of crossover and
mutation on each population in order to produce a new population until a termination
criterion is met.
Genetic algorithms do not work directly with points in the set but rather with an encod-
ing. The set is mapped onto another set consisting of strings of symbols of equal length
1
A real-valued function f (x) is convex if the graph of the function lies below the line segment joining
any two points of the graph.
5.1 General optimization problem 63

Figure 5.1: Global and local optima of a two-dimensional function [60].

(L). The strings are called chromosomes and the symbols form the so called alphabet.
This can be done by transforming them into binary chromosomes, decimal chromosomes
or into floating point representations. The latter is most promising for more complex
problems like the one dealt with in this study. Each chromosome (x) has a value of the
objective function referred to as the fitness (f (x)) of the chromosome. After the choice of
chromosome length, alphabet, encoding and initial population has been made, an oper-
ation called selection is applied. Different methods exist to perform this operation. One
is called the tournament selection, another is the roulette wheel selection. The first one
uses random selection of two individuals after which the best of the two goes to the next
round. The second method uses a probability selection based on the fitness on all individ-
uals The second stage is called evolution, where the crossover and mutation operations
are performed. Crossover is when a pair of chromosomes are taken and replaced by a
pair of offspring chromosomes. The operation involves exchanging substrings of in initial
chromosomes. The initial chromosomes are chosen randomly.
The newer genetic algorithms also implement an idea called elitism. Elitism consists of
the use of an external repository where the best solutions from every run are stored so
they cannot disappear due to successive iterations. Because normally a good individual
has a large probability to reproduce to do this the good individual is destroyed but re-
production does not always lead to the production of an even better individual. With
elitism this cannot happen.

Differential Evolution The idea behind this algorithm is to simulate a population

of solutions, that can produce offspring which replaces the original population in case
it is better than its parent. The algorithm was first introduced by Storn and Price in
1996 [39]. Such a differential evolution (DE) algorithm uses a pair of vector populations
that both contain N p D-dimensional vectors of real valued parameters. The current, Px ,
64 Optimization Techniques

population is composed out of those vectors, xi,g that are acceptable as initial points.

Px = xi,g with i = 0, 1, ..., N p 1 and g = 0, 1, ..., gmax (5.6)

xi,g = xj,i,g with j = 0, 1, ..., D 1 (5.7)

where g indicates the generation or number of iteration and each vector is assigned a
population index, i, and where parameters within vectors are indexed j. After initializa-
tion, DE mutates randomly chosen vectors to produce an intermediary population. Each
vector in the current population is then recombined with a mutant of this intermediary
population to produce a trial population. The trial vectors of the trial population over-
write the mutant vectors to reduce complexity. The initial value of the j th parameter of
the ith vector can be written as follows:

xj,i,g = r (bj,U bj,L ) + bj,L (5.8)

Here, r is a random number larger than zero and smaller or equal to one and bj,U and
bj,L indicate the upper and lower boundary respectively for the parameter. The mutant
vector can be created in many ways, one of them is given below:

vi,g = xbest + F (xr1 xr2 ) (5.9)

where F is the scale factor or weight, a real number that controls the rate of evolution
that has no upper limit but seldom has a value greater than 1. Finally crossover into the
trial vector only happens when the random number r is higher than the user defined value
of the crossover probability. These two factors, the weight and the crossover probability,
have a significant effect on the performance of the algorithm but their optimal values are
problem specific. The trial vector is evaluated by the objective function and if a more
optimized value is generated than that for the current population, the trial vector replaces
the initial vector in the next generation. If not the initial vector remains in place [39].

Particle Swarm Optimization Particle swarm optimization (PSO) is a global opti-

mization method that is based on the behavior of a swarm of insects, a flock of birds or a
school of fish. The particle denotes the individual in the swarm, flock or school and the
idea behind the optimization method is that each such individual behaves in a distributed
way based on both its own and the swarms intelligence. In the algorithm, each of these
particles is assumed to have a certain position and a certain velocity. During its travel
through the design space it remembers its best position it has discovered regarding the
objective function. These positions are communicated between particles, which adjust the
positions and velocities of the whole swarm during iterations. Initially, a certain number
of particles, N , are defined. Normally, this number is taken equal to about 20 or 30 in
order to have a high possibility of quick conversion without an astronomical amount of
function evaluations that need to be performed. These particles are assigned a random
position between the constraints. All of their initial velocities (the speed with which they
are moving toward the supposed maximum of the objective function) is 0. The objective
function is each iteration evaluated for all the particles. After the first iteration, the value
5.1 General optimization problem 65

of the objective function is memorized as that particles personal best, Pbest , while the
best value of all is denoted as the groups best, Gbest . The velocity of particle j in the
ith iteration can be written down as follows [41]:

Vj (i) =Vj (i 1) + c1 r1 [Pbest,j Xj (i 1)] + c2 r2 [Gbest,j Xj (i 1)]

(5.10)
with j = 1, 2, ..., N

where Xj is the position of the particle and r1 and r2 are random variables between 0 and
1 that illustrate the importance of the personal best and the group best to the particle in
order to change its velocity toward the target. c1 and c2 are so-called cognitive and social
group learning rates, which denote the importance of the memory of the personal and
group best. These values are often chosen to be equal to 2, so that the particle overflies
the target about half of the time. The equation shows that a particle gets deflected from
its previous direction toward the updated best values. The iterations are terminated once
the whole population is converging toward the same values.

5.1.2 Comparison of algorithms in literature.

Genetic Algorithms, Particle Swarm Optimization and Differential Evolution are three of
the most known global optimization techniques, but are by no means the only ones. Other
global optimization methods include simulated annealing and ant colony optimization but
these were rarely considered for trajectory optimization in the reviewed studies (only in
[40] for simulated annealing). GA has been used for launch trajectory optimization [25]
(sometimes in conjunction with the gradient method [64]) and for multidisciplinary op-
timization of both trajectory as well as vehicle design [5]. PSO is also frequently used
for ascent trajectory optimization [9] [38] as is the case for DE [21]. GA was most often
discussed in the literature but this might be the case because its the best known method
of the three and the most available. In addition, the work of Tusar and Filipic [54] clearly
shows that GA underperformed DE by a significant margin when trying to perform a
multi objective optimization. It is also stated that it also underperformed DE for single
objective optimization. This work compared the genetic algorithms NSGA-II, SPEA2
and IBEA with their differential evolution based variants DEMON SII , DEMOSP 2 and
DEMEIB [54]. A comparison between DE and PSO was performed in [49]. Here it was
demonstrated by means of optimization of a solar sail mission that DE also outperforms
PSO significantly with 3.24%. The same was found in a similar orbit trajectory optimiza-
tion study [27]. Since comparison of methods is problem specific, this does not imply that
DE is better than PSO for multidisciplinary optimization of both trajectory and vehicle
design and therefore these methods are compared by optimization of the problem at hand.

5.1.3 Problem formulation

The objective function that the algorithms need to minimize in the model is the following:

Mp
f [0] = 100, 000, 000, 000 (5.11)
Mi
66 Optimization Techniques

where, Mp is the payload mass and the Mi is the initial launcher mass. However, this
objection function is only evaluated when the eccentricity of the orbit gets below 0.01
while the apogee of the orbit is higher than an input value (in all simulations this value is
250 km unless indicated otherwise). If this is not the case the individual is evaluated with
the following objective function in order to guide its evolution toward the first objective
function:

f [0] = a (5.12)

where, a is the semi-major axis. If on the other hand any kind of constraint is violated
during the simulation the objective function gets a penalty value assigned to it in order
to make sure the algorithm discards solutions that violates constraints.

5.1.4 Comparison between Differential Evolution and Particle Swarm

Optimization

This section compares data gathered by implementation of the DE algorithm with data
from the PSO algorithm. At first, the PSO algorithm tends to outperform the DE algo-
rithm but after a number of generations the DE starts to outperform the PSO while the
PSO often mis converges in a local minimum. In addition, when the optimizers were given
the task to optimize the no assist cases, which are the most difficult to find a solution for
that does not violate any constraints, the PSO algorithm had more difficulty to find any
set of parameters that leads to orbit insertion of a payload and in the case of the no assist
kerolox case never did during its first 10,000 generations. The evolution of the fitness
values using both techniques can be seen in figure 5.2, where for every tenth generation
the fitness values of all the members of the population are plotted. Since DE did manage
to optimize all problems and outperformed the PSO algorithm in general, the decision
was made to use the DE algorithm for all cases.

5.1.5 Settings of the DE algorithm

In this section the selection of the weight coefficient, the crossover probability and the
size of the population will be discussed.
The means to estimate good values for these parameters was deemed to be via comparison
of how different values faired for when optimizing for certain launch conditions. The
conditions evaluated were an altitude of 40,000 m and a speed of 2,000 m/s. The optimizer
ran a population of 20 individuals over 5,000 generations in order to find the optimum for
a hydrolox launcher. The decision was based on the sum of both the best and the worst
fitness values in the final population. To start, a total of 25 simulations were run, which
are depicted in table 5.1.
As can be seen in figure 5.3, the solutions form a convex curve. And there is a triangular
area where a global optimum of the weight and crossover settings is likely to take place
and where any setting is likely to have a stable and good performance. In this triangular
area (governed by the points (0.5, 0.75), (0.75, 0.75) and (0.75, 0.99)) four additional
runs were made. These were in the middle of the triangle (at (0.67 0.67)) and in the
5.1 General optimization problem 67

9 9
x 10 x 10
0 0
PSO PSO
2 DE DE

4 0.5

8
Fitness value

Fitness value
1

10

12 1.5

14

16 2

18

20 2.5
0 500 1000 1500 2000 2500 3000 3500 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Number of Generations Number of Generations

(a) (b)

Figure 5.2: Evolution of the objective value of the hydrolox maximum assist case (a), showing the
early lead of the PSO algorithm being overtaken by the DE algorithm and the evolution
of the most difficult case, the no assist kerolox case (b), where the PSO algorithm does
not find a solution during 10,000 generations.

Table 5.1: Comparison of the results obtained from different weight and crossover probability DE
algorithms

hhhh
Crossover
hhhhh Probability
hhhh 0.01 0.25 0.5 0.75 0.99
Weight hhhh
0.01 0.1741 0.1709 0.1773 0.1429 0.0626
0.25 0.1913 0.1912 0.1977 0.1925 0.151
0.5 0.1852 0.1889 0.1968 0.2005 0.192
0.75 0.1787 0.1847 0.1972 0.1997 0.2004
0.99 0.1818 0.1844 0.1959 0.1969 0.1851

respective midway points between the middle of the triangle and the three corner points
((0.71, 0.71), (0.71, 0.83) and (0.58, 0.71)). The best result obtained by these four runs
was 0.20132 achieved by using a weight of 0.71 and a crossover of 0.83. A second run
confirmed this result by finding an optimum value of 0.2009.
Another setting that is important in evolutionary optimization is the size of the popula-
tion. To investigate its effect 3 different population sizes were selected and each of these
were run until the objective function was evaluated 100,000 times (e.g. 5,000 generations
with a population of 20, 10,000 generations with a population of 10). Rarely improve-
ments are found when larger populations than 40 individuals are generated. Therefore,
the three population sizes of choice were 10, 20 and 40. The average optimized value
of both runs for the different population sizes were 0.1976, 0.20111 and 0.19984. This
confirms that the original setting, using 20 individuals as a population size, is an efficient
setting.
68 Optimization Techniques

Figure 5.3: 3D visualization of the results obtained from different weight and crossover probability
DE algorithms
 Chapter 6

Tool Implementation and External

Software

This chapter presents the reader a quick overview of the external software that was used
to interact with the program code written in C++ by the author and how the overall tool
works. The overview of the external software includes the aerodynamic prediction tool
Missile DATCOM in 6.1, the astrodynamics toolbox Tudat in 6.2, the global optimization
toolbox PaGMO 6.3 and the Chemical Equilibrium with Applications (CEA) tool 6.4. The
chapter finishes with an overview of the code architecture of the simulation tool in section
6.5.

6.1 Missile DATCOM 97

Missile DATCOM is an aerodynamic prediction tool that predicts aerodynamic forces,

moments and stability derivatives as a function of angle of attack and Mach number for
missile configurations. It has been under continual development for more than 20 years.
The code, which is written in FORTRAN, is also subject to ITAR (International Trade
in Arms Regulations), but the 1997 version is available free of charge in a supplement
CD of the book Design Methodologies for Space Transportation Systems, by Walter
E. Hammond [24]. Missile DATCOM can generate aerodynamic force coefficients for up
to 20 different angles of attack and up to 20 different Mach numbers for a total of 400
coefficient sets per configuration. Missile DATCOM is software that was found to have
some difficulty in predicting axial force error1 , but overall, it followed experimental data
quite closely concerning normal force and center-of-pressure locations. Another downside
of the code is that the method used to calculate the axial coefficient is only valid up
to an angle of attack of 30 . Beyond this value the program switches to an empirically
derived method with far greater uncertainty. For this study, the force components are
1
In [14] the code was found to have a double Mach peak at about 1.7 Mach for the axial force
coefficient, CA , which is not expected to actually occur.

69
70 Tool Implementation and External Software

most important. The axial force is one of the most difficult aerodynamics forces to predict
[48] and an angle of attack range of 30 is deemed large enough for the purpose of this
study. Therefore, Missile DATCOM is chosen to be a sufficient basis for the aerodynamic
model in the simulations presented in this thesis.
Implementation of the Missile DATCOM program was done through code provided by
Frank Engelen, which was verified in [14]. This code block has as function to process
the Missile DATCOM files. After reading in the input file, all coefficients are stored in a
lookup table as a function of Mach number and angle of attack. The code does not allow
to also use the aerodynamic coefficients dependent on the Reynolds number. Therefore
the Reynolds number was assumed constant through the entire flight and a coefficient
only depends on two variables. In order to estimate the dependency on these variables
when their values fall between generated data points, bilinear interpolation is used. If
a value of the independent variables is beyond the boundaries of the database the limit
values are chosen and no extrapolation is performed.

6.2 Tudat

The TUDelft Astrodynamics Toolbox (Tudat) is an C++ library developed at the faculty
of Aerospace Engineering of the TU Delft. The program is defined in the Tudat User
Tutorial as follows [53]:
Tudat (TU Delft Astrodynamics Toolbox) is a C++ library that provides functionality to
perform astrodynamics simulations. Tudat is developed within the Astrodynamcis & Space
missions chair at the Faculty of Aerospace Engineering, Delft Universtity of Technology,
the Netherlands. Tudat is set up with particular focus on modularity and robustness of
code.
The toolbox has a very wide range and can therefore be implemented in various appli-
cations such as interplanetary trajectories, launcher ascent trajectories, vehicle re-entry,
etc. Information between the users and developers of Tudat is being exchanged through
multiple venues. First there are the bi-weekly Working Group Meetings where all active
developers and interested users discuss progress of the Tudat code and further develop-
ment. Then there are also monthly management meetings meant to discuss the overall
progress and longevity of Tudat. The internet is also used as a means of communication.
The Tudat website can be accessed at tudat.tudelft.nl. The website hosts a forum with
development discussions and also a wiki is available to explain the concept of Tudat to
newcomers. The website is also used to create issues concerning the Tudat development
and these can be created by all users. Minutes of meetings are also documented online.
In addition, various external libraries such as Eigen, Boost and CMake are needed to use
the functionality of Tudat.

6.3 PaGMO

PaGMO stands for Parallel Global Multiobjective Optimizer and is a C++ platform2
with the intention to easily perform parallel computations of optimizations tasks. For
2
PaGMO also has a python alter ego named PyGMO.
6.3 PaGMO 71

Figure 6.1: Hydrolox engine weight according to the original method (Eq. (3.44)), the optimized
method obtained through PaGMO (Eq. (3.68)), the actual data points of the 3 hydrolox
engines according to [15] and the Macht function from Microsoft Excel.

this purpose PaGMO contains many state of the art optimization algorithms, both global
and local algorithms, and also includes an extended set of optimization problems. PaGMO
can also interface with other optimization frameworks/algorithms such as NLOPT, SciPy,
SNOPT, IPOPT, GSL. Developed within the European Space Agency, the code was
initially coded for the automated design of interplanetary trajectories and spacecraft
transfers in general. The user can implement his own problem and algorithm both in
C++ and in Python. The reason that the decision was made to use PaGMO was driven
by a couple of factors. First of all, PaGMO has been tested and validated for a number
of space mission scenarios, secondly making use of such a platform leaves freedom to
compare how different algorithms fare whilst trying to optimize the problem at hand. In
addition, together with several members of the Tudat team it was decided that PaGMO
was the most interesting optimization toolbox to interface with Tudat.
Verification of PaGMO was done by optimizing the standard deviation of the hydrolox
engine data points as was done in section 3.3.3 and comparing this with the Microsoft
Excel data-fit, which was shown in figure 3.5. Here it can be seen that the optimized
method to minimize the standard deviation with PaGMO, depicted in blue, coincides
with Microsoft Excels internal function depicted in black. Another verification was by
maximizing the area-to-circumference ratio of a rectangle, a problem where the solution
has a maximal objective value if the rectangle is a square. This problem was made
more difficult by extrapolating it to hyper rectangles. PaGMOs optimization algorithms
found that the ratio between the hyper volume and the hyper circumference was always
maximized when the hyper rectangle has the same size in all dimensions.
72 Tool Implementation and External Software

6.4 CEA

The program CEA, which stands for Chemical Equilibrium with Applications, is also
known by the name of the NASA Glenn Equilibrium Program. The program is written
in ANSI standard FORTRAN by Bonnie J. McBride and Sanford Gordon [7]. CEA
calculates complex chemical equilibrium product concentrations from any set of reactants
and determines thermodynamic and transport properties for the product mixture. It is
used in this study to tabulate values for Tc , M and for both RP 1 as well as LH2
fuel and a whole array of chamber pressures and area ratios. Assumptions made in the
tool are one dimensional forms of the continuity, energy and momentum equations, zero
velocity at the combustion chamber inlet, complete and adiabatic combustion, isentropic
expansion in the nozzle, homogeneous mixing, ideal-gas law and zero temperature and
velocity lags between condensed and gaseous species [22]. It is also used by the ASTOS
(AeroSpace Trajectory Optimization Software) library as a 3rd party model [63] and it
was also subjected to verification and comparison with another software industry standard
and gave very similar results [61]. Therefore it was deemed to be useful as a means of
obtaining the thermodynamic data needed for the vehicle model.
The CEA database used in the simulation program consists out of a total of 3648 data
points. These were created at each increment of 0.5 in the oxidizer to fuel ratio, ranging
from a value of 1 to a value of 10, and at each increment of 8 bar of chamber pressure,
ranging from a value of 8 bar up until a value of 264 bar. These data points contain values
of the ratio of specific heats, , the chamber temperature, Tc and the molar mass, R. The
method that was used to interpolate linearly between these data points was extensively
tested by Dominic Dirkx.

6.5 Code architecture

The code was developed in C++ in order to be compatible with Tudat. The flowchart
is depicted in 6.2. The Input of the program, the launch simulator, consists out of two
parts. First there is the individual generated by the differential evolution algorithm, which
contains 11 to be optimized parameters and secondly there is a set of inputs defined by the
user, which consists out of the fuel type and the initial state of the launcher. The inputs
are given to the vehicle model in order to construct a configuration of the launcher and
to the control block, which defines the settings of the thrust throttle and the orientation
of the vehicle. The information from the vehicle model is transmitted into the launch
simulator loop, depicted in blue, by passing it on to the body information, which also
contains the state, and communicates the state vector with the environment model. After
an update of the state vector, the body information code block transforms the state vector
to different reference frames and coordinate systems. The other sub models can then also
be updated. The Missile DATCOM Database provides new aerodynamic coefficients in the
aerodynamic reference frame to the aerodynamic force code block, which has as output the
aerodynamic force vector in the inertial frame. The same holds true for the rocket motor,
which provides the magnitude of the thrust force to the thrust force code block, which
converts it to a force vector in the inertial frame. Together with the gravitational force,
these force models pass the force vectors on to the translational equations of motion from
which the acceleration of the vehicle is derived. The propagator collects this acceleration
6.5 Code architecture 73

and the integrator then computes a new state in the inertial frame which is then evaluated
so that it does not violate any hard constraints, if it only violates for example the axial
acceleration constraint this information is transmitted to the control block, which then
alters the throttle setting. Other parts of the state can also influence the pitch control or
the throttle. If no abort is made the state is resubmitted in the Body Information code
block and the loop starts again. During optimization the objective function gets evaluated
and the differential evolution algorithm minimizes its value by altering the optimizable
parameters. If a constraint is violated the objective value gets a penalty value assigned to
it during the function evaluation process. The output of the optimization process needs
to be inputted in a copy of the launch simulator, which generates additional data on the
launch in data files. Postprocessing of these data files is then done with Matlab.
 74
Tool Implementation and External Software
Figure 6.2: Flowchart of the simulation tool.
 Chapter 7

Results

In this chapter the main results and a short introduction to how they were obtained are
presented in section 7.1. This is followed by a thorough discussion of their implications
in section 7.2 and finally, the model is subjected to a sensitivity analysis in section 7.4.
The data on which most of the plots presented in this chapter are based, can be found in
Appendix B.

7.1 Introduction to the results

The objective of the study is to optimize the performance of several launch vehicles
that utilize the assisted-launch technique and draw conclusions on the effect that the
launch-assist parameters have on the vehicle design and total system performance. For
this reason, a launch simulation of these vehicles must be performed. This simulation
is subjected to a number of constraints, which were discussed in section 3.7. These
constraints are the bending-load limit, the axial-acceleration limit, the angle-of-attack
limit, the heat-flux limit and the altitude limit. Each launch vehicle will be simulated with
different input parameters. These inputs1 are the propellant type (hydrolox or kerolox),
the initial velocity and the initial altitude of the launcher, also referred to as the launch-
assist parameters. During the optimization process a set of 11 parameters, consisting
out of 5 trajectory parameters and 6 launch-vehicle parameters, is altered, generating a
different flight path of the launch vehicle and a different launch vehicle performance and
mass. The different flight path and launch-vehicle characteristics change the performance
of the assist technique, which is defined as the ratio between the payload mass and
the total launch-vehicle mass, from now on referred to as the payload mass ratio. For
every unique set of inputs, a unique set of these 11 parameters will be found by the DE
optimization algorithm that describe the optimal vehicle and trajectory required to get
the maximum payload mass ratio into orbit considering the given inputs and applicable
1
Another input needed for the model is an estimation of the optimized payload mass ratio, which is
required for the estimation of the fairing mass.

75
76 Results

constraints. The settings of this DE algorithm are a population of 20 with a crossover of

0.83 and a weight of 0.71 as discussed in section 5.1.5.
A total of 60 input sets were chosen, leading to a total of 60 sets of the aforementioned 11
parameters. Through this chapter, the author hopes to provide the reader with a thorough
understanding of how the input parameters influence the launch-vehicle parameters, how
the launch-vehicle parameters influence each other and how the input and launch-vehicle
parameters influence the objective function, the payload mass ratio. The 60 sets of inputs
consist out of the 30 combinations that can be made with the following launch-assist
parameters for both the kerolox type of vehicle as well as the hydrolox type:

Altitude: 0, 5000, 10000, 20000, 40000 [m]

Velocity: 0, 300, 600, 1000, 1500, 2000 [m/s]

These parameters were chosen to have this range, because the outermost values roughly
correspond with what is possible with current day technology for conceivable assist sys-
tems. The altitude limit of 40 km roughly corresponds with altitude records of high air
balloons and air-breathing engines2 while the speed limit was based on the record of the
X-15 rocket plane3 . The parameters also have increasing leaps between each other since it
was expected beforehand that the further one gets from the no-assist case, the smaller the
increments in the objective function will be, as can be seen in figures 7.2 and 7.3, where
the surface is visibly steeper at the origin than at the outermost point4 . This has several
reasons, firstly it is in part due to the logarithmic nature of Tsiolkowskis equation, see
Eq. (3.73), but it is also due to the nature of the sum of the drag and gravity losses,
seen in figure 7.18, which are maximized at low altitudes and to a lesser extent at low
velocities. To give a good overview of the objection functions impact by the parameters,
more data points were deemed needed in the expected high gradient environments. The
11 trajectory and launch-vehicle parameters that need to be optimized are summarized
below accompanied by their respective search spaces:

The 4 Pitch variables (see section 3.6.3) (search space: 0 - 1)

The length of the propellant tanks (search space: 10 - 100 m)

The slenderness of the propellant tanks (search space: 1.82 - 20)

The ratio between the exit diameter of the nozzle and the diameter of the propellant
tanks (search space: 0 - 1)

The ratio between the oxidizer and the fuel mass (search space: 1 - 10)

The chamber pressure (search space: 8 - 264 bar)

The nozzle exit pressure (search space: 0 - 100,000 Pa)

2
The altitude record for balloons is pending verification at approximately 39,000 m and for aircraft it
stands at 37,650 m [16]
3
The fastest speed achieved by the X-15 was 2,020.62 m/s [43]
4
This would be even more obvious when plotted logarithmically, which would indicate the relative
increase in the payload mass ratio
7.1 Introduction to the results 77

The initial flight-path angle5 (search space: 0 - 1.5 rad)

For every set of input parameters the resulting launch-vehicle and trajectory parameters
are tabulated in Appendix B together with other relevant information. Taking a look at
the outermost left column of table B.1 reveals that for a launch of a kerolox launcher,
ignited at an assist altitude of 0 m with an assist velocity of 0 m/s, the optimal length is
18.56 m. The optimal slenderness is 10.38, the optimal diameter is 1.79 m, the optimal
nozzle diameter is 0.88 m, the optimal oxidizer-to-fuel ratio is 2.50, the optimal chamber
pressure is 263.98, the optimal exit pressure is 45861 Pa and the optimal initial-flight-path
angle is 84.81 deg. This optimal configuration leads to an injection of 859.9 kg into an orbit
with an apogee larger than 250 km and an eccentricity smaller than 0.01. This results in
an optimized payload mass ratio of 0.021. The table also reveals other information such
as the maximum dynamic pressure, the maximum heat flux, the maximum bending load
and the times of their occurrences, the gravity and drag losses, several masses, the initial
and maximum acceleration, the moment when the fairing is jettisoned, the nozzle area
ratio, the maximum thrust, and the maximum angle of attack.

5
x 10
3

2.5

2
Pitch angle (rad)

1.5

0.5

0
0 500 1000 1500 2000 2500 3000
Time (sec)

Figure 7.1: Altitude vs time plots of all optimized hydrolox simulations

In Appendix C a lot of additional information about all optimal launch trajectories is

plotted. This includes the inertial velocity, the eccentricity, the altitude, the total accel-
eration, the pitch angle, the angle of attack, the dynamic pressure, the bending load, the
heat flux, the position, the lift-to-weight ratio and altitude vs velocity. An example of
this is given by the altitude plots of the 30 hydrolox vehicles, which is shown in figure 7.1.
5
unless stated otherwise this is also the initial pitch angle since these were mostly set at the same
values at t=0.
78 Results

In this figure it can be seen that 3 particular cases deviate from the norm by performing
a continual rise past the 120 km mark, while most follow a trajectory that stays in this
region for an extended amount of time. This information is helpful to create a better
understanding about the subject of assisted launch and some imperfections of the models
used. This specific deviation is further explained in section 7.3.3. The color codes of the
lines in the figure are only for enhanced visibility.
The evolution of the objective function, the payload mass ratio, is visualized in figure 7.2
for hydrolox launchers and 7.3 for kerolox launchers. In these figures the values of all 60
optimized payload mass ratios are interpolated with a cubic spline function and plotted
in accordance to their respective launch-assist parameters. The importance of the assist
altitude and velocity on the launchers efficiency is quantified and it can be seen that the
payload ratio increases both with increasing assist velocity as with increasing assist alti-
tude. The optimized launcher for the no-assist case has a maximum payload mass ratio
of only 0.021 for kerolox vehicles and 0.031 for hydrolox vehicles. This rises to values of
0.15 and 0.19 for kerolox and hydrolox vehicles for the highest assist case. To get a sense
of the importance of both assist parameters with respect to each other, a comparison of
the results utilizing the lowest assist altitude and those with the highest assist altitude
shows the following. To get the same performance increase by adding velocity instead of
altitude one should add a velocity of about 400 m/s for kerolox launchers and 540 m/s
for hydrolox launchers for the entire performance increased caused by a change in assist
altitude of 0 km to 40 km. The difference between the two types of launchers has to do
with the drag losses, which will be discussed in section 7.2.12.

0.18
0.2

0.16

0.15
0.14
Diameter(m)

0.1 0.12

0.05 0.1

0.08
0
4
3 0.06
2 2000
4 1500
x 10 1 1000
500 0.04
0 0
Altitude (m)
Velocity (m/s)

Figure 7.2: Cubic spline fit of the payload-mass to initial-mass ratio of optimized hydrolox launchers
and launch trajectories for different combination of assist velocities and altitudes.
7.2 Discussion of the launch-vehicle parameters 79

0.15

0.16

0.14

0.12
Diameter(m)

0.1 0.1

0.08

0.06

0.04

0.02
4 0.05
3
2 2000
4 1500
x 10 1 1000
500
0 0
Altitude (m)
Velocity (m/s)

Figure 7.3: Cubic spline fit of the payload-mass to initial-mass ratio of optimized kerolox launchers
and launch trajectories for different combination of assist velocities and altitudes.

7.2 Discussion of the launch-vehicle parameters

In this section all optimized launch-vehicle parameters and some extra parameters of the
model will be discussed together with the trends that they follow, which will be analyzed
and explained. All plots shown are cubic spline interpolations of 30 data points, meaning
that the displayed 3-D surface intersects with all data points.

7.2.1 Fuel type

The model used two types of fuel, kerolox and hydrolox. This parameter is the sole
parameter that was not optimized by the algorithm but was fixed. Therefore, the altitude-
velocity plots of all parameters are presented in this chapter twice, once for the hydrolox
launcher case (always depicted in the left subfigure) and once for the kerolox case (always
presented in the right subfigure). As could already be seen in figures 7.2 and 7.3, all
altitude-velocity combinations uniformly show that hydrolox vehicles provided a greater
payload ratio than kerolox vehicles. Relatively speaking, the type of fuel gives a far
smaller effect for high-assist systems. With no assist, the difference in the payload ratio is
0.01006 or in relative terms this means that a switch from a kerolox launcher to a hydrolox
launcher gives a relative performance gain of 48.4 %. At an assist altitude of 40 km with a
velocity of 2,000 m/s the difference in payload ratio is less than 0.039 corresponding to just
a relative improvement of 25.7%. This effect is even greater when adjusting for residual
propellant mass not accounted for in this study. The results of the vehicles published in
Appendix B are also split up according to their fuel type. The tables B.1 to B.6 give the
results for the kerolox vehicles while the results of the hydrolox vehicles are published in
tables B.7 to B.12.
80 Results

7.2.2 Length

4 4
x 10 x 10
4 4
65 45

3.5 60 3.5
40
55
3 3

50 35
2.5 2.5
45
Altitude (m)

Altitude (m)
30
2 2
40

1.5 35 1.5 25

30
1 1
20
25
0.5 0.5
20 15

0 0
0 500 1000 1500 2000 15 0 500 1000 1500 2000
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.4: Length (m) of the optimized hydrolox vehicles (a) and kerolox vehicles (b).

The search space of the length of the tank was taken to vary from 10 - 100 m. This is
based on the dimensions of the six launchers, given in section 3.3.3, on which the model
was based. These launchers are roughly 3 times longer than the lower boundary and 3
times shorter than the upper boundary of the search space used. The variation of length
along the different vehicles optimized for different assist parameters shows a couple of
clear trends. Looking at figure 7.4, the most attention is being grabbed by the high
length peak in the high-velocity/low-altitude region. It appears at first glance that this
happens to minimize drag losses. This square cube law effect of minimizing drag losses
by increasing launcher dimensions was explained in section 3.5 and basically states that
drag loss decreases with increasing length. However, when looking at the drag loss in
figure 7.15 it can be seen that not one but two local peaks occur in drag loss while only
one (the smaller one) translates in larger dimensions. Further discussion on this anomaly
is provided in section 7.2.3.
Next to this global peak, the rest of the figure looks like irresponsive space. A look at the
data presented throughout Appendix B reveals that there are the trends in this space of
decreasing length with increasing altitude and decreasing length with decreasing velocity
over the entire spectrum. This can be explained by a combination of the drag loss effect
and by the results displayed in section 3.3.3. In this section, two vehicles were optimized
for maximum V . The first set with a thrust-to-weight ratio of zero had a maximum
length of 100 m due to the tank surface-to-volume ratio that gets maximized with large
tanks. The second set, had a thrust-to-weight ratio of 1.15 and found that the drive for
a larger length to minimize the surface-to-volume ratio was counteracted by the drive to
minimize the hydrostatic pressure, which causes a thicker tank thickness for longer tanks,
and the drive for smaller mass to minimize the engine mass ratio (see Eqs. (3.69) and
(3.68)) to the degree that the optimal length was found to be between 12 and 13 meter.
This means that in the absence of severe drag loss, the thrust-to-weight ratio plays an
important role in the optimization process of the length. With the launches that occur at
7.2 Discussion of the launch-vehicle parameters 81

40 km, drag loss is almost negligible and the length is indeed observed to increase with
decreasing thrust-to-weight ratios.

7.2.3 Slenderness and diameter

4 4
x 10 x 10
4 20 4

16
3.5 3.5
18

14
3 3
16

2.5 2.5
12
Altitude (m)

Altitude (m)
14
2 2
10
12
1.5 1.5

10 8
1 1

0.5 8 0.5
6

0 0
0 500 1000 1500 2000 6 0 500 1000 1500 2000
Velocity (m/s) Velocity (m/s) 4

(a) (b)

Figure 7.5: Slenderness of the optimized hydrolox vehicles (a) and kerolox vehicles (b).

4 4
x 10 x 10
4 3.4 4
6.5
3.2
3.5 3.5
6

3
3 3 5.5

2.8
5
2.5 2.5
Altitude (m)

Altitude (m)

2.6 4.5
2 2
2.4 4

1.5 1.5
2.2 3.5

1 1 3
2

2.5
0.5 1.8 0.5

2
0 1.6 0
0 500 1000 1500 2000 0 500 1000 1500 2000 1.5
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.6: Diameter (m) of the optimized hydrolox vehicles (a) and kerolox vehicles (b).

The search space of the slenderness, , ranges from 1.82 to 20. This was chosen based
on data from [15] of several launch vehicles shown in table 7.1. In this table it can be
seen that the maximum slenderness is 12.44 while the minimum slenderness is 2.73. It
was deemed necessary to have at least a margin of 50% on both the minimum and the
maximum value in the search space and due to programatic convenience the values of
1.82 and 20 were chosen. In figures 7.5 and 7.6, the optimal values of the slenderness
82 Results

ATLAS 5 Zenit K1 Delta IV Angara

Length 32,46 33 18,3 40,8 25,14
Diameter 3,81 3,9 6,7 5,13 2,9
Slenderness 8,52 8,46 2,73 7,95 8,67
Long March Long March H2A H2 Soyuz
CZ-5-200 CZ-500 booster
Length 28 31 37,2 28 19,6
Diameter 2,25 5 4 4 2,68
Slenderness 12,44 6,2 9,3 7 7,31
Soyuz Energya Energya Delta 2 Ariane
strap on core 5 ECB
Length 27,8 38 59 26,08 30,5
Diameter 2,55 4 8 2,44 5,4
Slenderness 10,9 9,5 7,38 10,67 5,65

Table 7.1: Length, diameter and slenderness of various first stages [15]

and diameter can be seen over the launch parameter spectrum. The most noticeable
trend of the slenderness parameter is that it maximizes in the high drag loss areas, which
are located in the bottom corners of the plots as shown in figure 7.15, and that this is
more the case for hydrolox launchers than for kerolox launchers. The explanation for the
fact that kerolox launchers have less slender dimensions could be twofold. First of all, for
kerolox vehicles, the nozzle-to-tank-diameter ratio is maximized for various simulations in
the high velocity / low altitude spectrum implying that the only possible way to increase
the nozzle exit diameter is to also increase the vehicles tank diameter and thus have a
slightly less slender body than desired. Secondly, the drag losses are significantly lower
for the high-density kerolox vehicles than for the lower-density hydrolox vehicles. And as
a final reason, hydrolox launchers have a greater performance than kerolox launchers and
the higher the performance of the launcher, the more slenderness will tend to follow the
drag-loss trend, which will be explained further on.
Another noticeable fact is that while the slenderness follows the drag loss it does this
more so in the high velocity drag loss peak than in the low velocity drag-loss peak. To
answer why this is the case one must take a look at Tsiolkowskis equation, see Eq.
(3.73). Consider two launchers, both have a construction mass ratio of 7%. Launcher A
is launched in the no-assist case and brings a payload to orbit of just 3%. Launcher B is
launched with a high-assist velocity and brings a payload to orbit with a mass equal to
18% of the total launcher mass. Both launchers would be able to decrease their drag loss
by 100 m/s by making the vehicle slender, which would result in a higher construction
mass ratio of 7.5%. Assuming a specific impulse of 450 seconds this means that the V
of launcher A evolves as follows:

   
1 1
V = 450 9.81 ln = 10, 165 V = 450 9.81 ln = 9, 949 (7.1)
0.1 0.105

This formula shows that the performance increase due to the lower drag loss is more than
canceled out by the performance decrease of 216 m/s due to the increased mass ratio.
7.2 Discussion of the launch-vehicle parameters 83

For launcher B this performance decrease is only 88 m/s, see Eq. (7.2) causing its slender
version to have a greater overall performance.

   
1 1
V = 450 9.81 ln = 6, 120 V = 450 9.81 ln = 6, 032 (7.2)
0.25 0.255

Furthermore, in the kerolox plots of slenderness and diameter there are two large anoma-
lies identified at 5,000 m altitude with assist velocities of 1,000 m/s and 1,500 m/s.
Combined with the large optimal length that was found at a velocity of 1,500 m/s and
an altitude of 5,000 m, this leads to a very high global peak in the diameter plot. Since
no physical explanation can be found for these anomalies, which are only observed with
the kerolox launcher and not with its hydrolox counterpart, it might imply that the DE
algorithm failed to converge at the true global optimum for this configuration resulting
in an odd value for the slenderness. Furthermore, the simulation with an assist velocity
of 1,000 m/s also has an anomaly regarding orbit insertion as can be seen in figure C.9,
where it is displayed by the red graph having the highest altitude of all simulations from
about 2,300 seconds into the simulation onwards. The fact that the slenderness parameter
was identified by the sensitivity analysis as having a fairly small impact on the overall
solution might be the reason why the algorithm failed to converge for the slenderness
parameter on these two instances.
Other noticeable anomalies can be found in the hydrolox diameter plot. A region of high
optimal diameters can be found for the high-altitude/low-velocity cases, this is mainly
due to a region of local slenderness minima. Also at 2,000 m/s a large-diameter ridge
is being displayed, which is a result of the large length tendency due to low optimal
thrust-to-weight ratios found in this zone. This ridge would be even more pronounced if
the slenderness at the data point utilizing an assist velocity of 2,000 m/s and an assist
altitude of 20 km would not be so high. At this case, th hydrolox vehicle has a slender-
ness of 15.79 which compares with the kerolox vehicle that has a slenderness of just 5.33.
The reason for this is because as was previously discussed, the slenderness of hydrolox
launchers follows the drag loss trend more closely than kerolox launchers. And secondly,
hydrolox launchers also have more drag loss than their kerolox counterparts.

7.2.4 Nozzle exit-to-tank ratio / Nozzle exit diameter

This ratio is optimized in a search space between 0 and 1, these values were chosen because
this means that the nozzle never disrupts the airflow in a significant way by sticking out.
This ratio scales the nozzle of the launcher and impacts the maximum thrust of the
launcher (nozzle throat), the nozzle exit pressure (nozzle exit) and the nozzle mass. The
ratio is displayed in figure 7.7 and shows that it gets influenced both by the length, see
figure 7.4, and slightly by the exit pressure, see figure 7.12. The length effect can be
recognized by the characteristic bottom-right global peak with an upward offshoot, while
the exit pressure effect can be recognized by the distinct global minimum at the no-assist
case.
The reason why it is influenced by the length is because an increase in length causes a
cubic increase in mass and in order to keep the thrust-to-weight ratio to a decent level,
the nozzle exit area, which has a great effect on the thrust, needs to be maximized. On
84 Results

4 4
x 10 x 10
4 4
1

3.5 0.9 3.5

0.95

3 3 0.9
0.8
0.85
2.5 2.5
Altitude (m)

Altitude (m)
0.7 0.8
2 2
0.75

1.5 0.6 1.5

0.7

1 1 0.65
0.5

0.6
0.5 0.5
0.4 0.55
0 0
0 500 1000 1500 2000 0 500 1000 1500 2000
0.5
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.7: Nozzle exhaust-to-tank diameter ratio of the optimized hydrolox vehicles (a) and kerolox
vehicles (b).

4 4
x 10 x 10
4 4

3.5 3 3.5 6

3 3

2.5 5
2.5 2.5
Altitude (m)

Altitude (m)

2 2 4
2

1.5 1.5
3
1.5
1 1

2
0.5 0.5
1

0 0
0 500 1000 1500 2000 0 500 1000 1500 2000 1
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.8: Nozzle exhaust diameter (m) of the optimized hydrolox vehicles (a) and kerolox vehicles
(b).

the other hand when the length is rather short the effect of the exit pressure can be seen.
This is due to the fact that exit pressure peaks near the no-assist case implying the need
for a small area ratio and together with a small thrust needed to propel the small vehicle,
this translates into very small nozzle-to-tank ratios of 0.32 and 0.49 for the non-assisted
hydrolox and kerolox cases respectively. Furthermore it was already noted that for the
kerolox launcher the ratio approaches its upper boundary in a few assist cases (while that
only happens in the most extreme hydrolox case) in these cases the ratio is deemed to
interact with the slenderness of the launcher since they start to depend on each other but
have opposite trends. The global maximum shown in the kerolox plot of the nozzle-exit
diameter is due to the large optimal diameter found there, which was discussed in section
7.2 Discussion of the launch-vehicle parameters 85

7.2.3.

7.2.5 Oxidizer-to-fuel ratio

4 4
x 10 x 10
4 4 2.509

6.45
3.5 3.5 2.508

6.4
2.507
3 3
6.35
2.506
2.5 2.5
6.3
2.505
Altitude (m)

Altitude (m)
2 6.25 2
2.504

6.2
1.5 1.5 2.503

6.15
2.502
1 1
6.1
2.501
0.5 0.5
6.05
2.5

0 6 0
0 500 1000 1500 2000 0 500 1000 1500 2000 2.499
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.9: Oxidizer-to-fuel ratio of the optimized hydrolox vehicles (a) and kerolox vehicles (b).

The search space of this parameter was set from 1 to 10 to have a large certainty that the
optimum would be found for both types of fuels where the value tends to range from a
low of 2.39 to 6.1, see tables 3.4 and 3.5. This parameter has two main influences on the
performance of the rocket. Firstly, it will influence the specific impulse significantly and
secondly it will influence the average propellant density and thus propellant tank volume
and tank mass needed. The very minute variations shown in the plots, see figure 7.9 and
note the legend, seem to indicate that not much of a trade off is performed by the model
and that the value of the parameter is established independent of outside factors that
change by choosing different assist parameters. The only real mentionable observation
that can be made is about the hydrolox plot, where four of the six data points that have
a value of 6.5 instead of 6 are grouped at the bottom right. The other two are regarded
as noise from an imperfect model. The four data points coincide with the region where
the drag loss and the dimension of the launcher is exceptionally high. This makes sense
because higher oxidizer-to-fuel ratio leads to a higher density and lower drag loss, as shown
by Eq. 3.76. No variation in the kerolox plot may indicate that either a higher resolution
than 0,5 is needed in the CEA database or a more accurate interpolation method than
the currently used linear interpolation method should be implemented.

7.2.6 Chamber Pressure

The search space for the chamber pressure was defined to lie between 8 and 264 bar.
The effect of this parameter is twofold. First and foremost the chamber pressure has
a significant impact on the specific impulse of the launcher. In [65] the effect of the
86 Results

4 4
x 10 x 10
4 4
264

3.5 260 3.5

263.5

3 3
255
263

2.5 2.5
250 262.5
Altitude (m)

Altitude (m)
2 2
262
245
1.5 1.5
261.5

1 240 1
261

0.5 0.5
235
260.5
0 0
0 500 1000 1500 2000 0 500 1000 1500 2000
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.10: Chamber pressure (bar) of the optimized hydrolox vehicles (a) and kerolox vehicles (b).

pressure ratio on the exhaust velocity is investigated by introducing the dimensionless

exhaust velocity:
v  
u   1
pe
u
(Ue )dim = t2 1 (7.3)
u
1 pc

The effect of Eq. (7.3) is plotted in 7.11 for various ratios of specific heats. As can
be seen in this figure, the effect of the chamber pressure on the exhaust velocity almost
flatlines at higher chamber pressures. For example, between a pressure ratio of 5,000 and
one of 6,000 the advantage in specific impulse is only in the order of 0.5%. The other
effect that the chamber pressure has on the model is on the nozzle mass, shown by Eq.
(3.61). At very high pressure ratios the effect on the nozzle mass starts to contribute
more to the optimization process than that of the specific impulse. This is why the low
regions in the hydrolox chamber pressure plot, the left of figure 7.10, coincides with the
low regions of the nozzle-exit pressure. Although this is not evident from figure 7.12 due
to the high global peak at the no-assist case, careful examination of the data displayed in
Appendix B led to the aforementioned conclusion that the local minima of the chamber
pressure plots coincide with local minima found in the exit pressure data. In 2 of the 3
cases where exit pressure is modeled to be lower than 6,000 Pa, the chamber pressure is
modeled to be lower than 240 bar (pressure ratios in excess of 4,000). But overall, the
optimal chamber pressure is found to be around the maximum of the search space with
few exceptions. This is an odd result since current engines, notably upper stage engines
often operate at far lower chamber pressures. Examples are the Vulcain 2 (Ariane 5 first
stage engine), which has a chamber pressure of 117.3 bar [4] and tests have been run
for the Ariane 5 upper stage engine, Vinci, at around 55 bar [29]. The reason for this
discrepancy with reality is up to speculation. It might be due to an underrepresentation
of the chamber pressure in the mass model, since it would lead to a closer match with
reality if the engine mass would be dependent on the chamber pressure. Another reason
7.2 Discussion of the launch-vehicle parameters 87

could be that for real engines, the additional small improvement might not be deemed
important enough compared with regard to higher manufacturing costs. Whether or
not the underrepresentation of the chamber pressure in the mass model has a genuine
influence, the model should be expanded with a cost model to provide a definitive answer
on the question whether or not current launchers utilize suboptimal chamber pressures.

3.5

3
Dimensionless exhaust velocity ()

2.5

1.5

1
= 1.15
0.5 = 1.2
= 1.25
= 1.3
0
0 1000 2000 3000 4000 5000 6000
Chamber to nozzle exit pressure ()

Figure 7.11: Dimensionless exhaust velocity as a function of pressure ratio

7.2.7 Exit Pressure

4 4
x 10
4 x 10 x 10
4 x 10
4 4 4.5
4.5

3.5 3.5
4
4

3 3
3.5
3.5

2.5 2.5
3
3
Altitude (m)

Altitude (m)

2 2
2.5 2.5

1.5 1.5
2 2

1 1
1.5 1.5

0.5 0.5
1
1
0 0
0 500 1000 1500 2000 0.5 0 500 1000 1500 2000
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.12: Exit pressure (Pa) of the optimized hydrolox vehicles (a) and kerolox vehicles (b).
88 Results

The search space of the exit pressure was established to be between 0 and 100,000 Pa. The
upper limit was established because an engine can not perform optimally from the Earths
surface all the way to space when the exit pressure is always higher than the ambient
pressure, which decreases rapidly from its maximum value of 101,325 Pa. Therefore, it
could be assumed beyond reasonable doubt that all optimal values lie within the region
of the search space. An assumption that was rapidly confirmed as the first simulations
were performed. This parameter also effects the exhaust velocity through the chamber-
pressure-to-exit-pressure ratio depicted in figure 7.11. In addition, it also has an extra
effect through the pressure thrust, see Eq. (3.71). In figure 7.12 it can be seen that this
pressure thrust effect causes the optimal exit pressure to increase significantly in cases
where the launcher spends a long period of time at low altitude, which happens for the
low-assist cases. The reason why the optimal exit pressure decreases rapidly from its
maximum even with the a relatively small increase in the assist parameters is due to the
burn time in the densest parts of the atmosphere, which is rapidly diminished by even the
smallest assist. The no-assist hydrolox launcher reaches 5 km after 58.6 seconds and 10
km after 79.5 seconds. For comparison, the optimal launcher launched at sea level with an
initial velocity of 300 m/s reaches 5 km after 19.4 seconds and 10 km after 38.8 seconds.
The launcher with assist altitude of 5 km and no assist velocity reaches the 10 km mark
after 53.5 seconds without having to traverse the densest 5 km of the atmosphere. For
a certain fixed assist velocity it is sometimes observed6 that the optimized exit pressure
achieves a minimum at 20,000 m altitude instead of 40,000 m altitude with cases at 10,000
m often optimizing about the same value as for cases at 40,000 m. Therefore it could
be concluded that most if not all of the advantage of the pressure thrust component is
already achieved at launch altitudes of 10,000 m.

7.2.8 Initial flight-path angle

4 4
x 10 x 10
4 4
80
80
3.5 75 3.5
75

3 70 3
70

65
2.5 2.5 65
Altitude (m)

Altitude (m)

60 60
2 2
55 55

1.5 1.5
50 50

1 45 1 45

40
40
0.5 0.5
35
35
0 0
0 500 1000 1500 2000 0 500 1000 1500 2000 30
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.13: Initial flight-path angle (deg) of the optimized hydrolox vehicles (a) and kerolox vehicles
(b).
6
All for hydrolox but only at 300 and 1500 m/s for kerolox
7.2 Discussion of the launch-vehicle parameters 89

The initial flight-path angle was chosen from a search space of 0 - 85.94 (0-1.5 radians)
for convenience after it was established that even for the no-assist cases this angle was
optimized to be lower than 85.94 . In the end, at burnout in a circular orbit, this param-
eter is desired to be 0. Also from the equation of gravity loss, Eq. (3.77), it can be seen
that a large flight-path angle is detrimental to the launchers performance. However, it
must naturally be large enough to escape the atmosphere and with respect to drag losses
and pressure thrust, the sooner the better. In figure 7.13 a clear trend can be seen where
the flight-path angle decreases rapidly with increased assist velocity and also slightly de-
creases with increasing assist altitude. This smaller effect due to the altitude is higher
at high-assist velocities than at low-assist velocities. For instance, the difference between
the initial flight-path angle at 40,000 m and the one at 0 m is only 4.7 at no-assist
velocity but more than 11.5 radians at an assist velocity of 2000 m/s7 . There are some
anomalies to be found in the plots and the data displayed in Appendix B. The initial
flight-path angle turns out to be higher than expected with an assist velocity of 300 m/s
at altitudes of 20 and 40 km for both fuel type vehicles. This is due to the fact that the
optimizer found that it was better to gain altitude rapidly and keep the heat flux below
the limit where a fairing would be needed. This is further explained in section 7.3.1. A
couple of other flights also deviated slightly from the overall trend e.g. the case of 300
m/s at 0 m altitude for a kerolox vehicle where the initial flight-path is lower than its 600
m/s counterpart. This might be explained by the higher lift that gets produced at lower
altitudes, as is discussed in section 7.3.2, but this explanation does not stay consistent
with all high-lift cases. What can be noted about all other anomalies is that they take
place in the low velocity part of the grid. This is also logical because the same deviation
at a higher velocity has a greater impact due to the magnitude of velocity vector.

7.2.9 Thrust-to-weight ratio

The thrust-to-weight ratio is rather low for the whole range of vehicles ranging from a high
of 1.372 to a low of 0.728. Considering the V losses are mostly dominated by gravity loss,
one would expect that a high thrust-to-weight ratio would be beneficial. In conventional
vehicles gravity loss is larger than drag loss by a ratio of about ten to one [28]. This
means that a higher thrust-to-weight ratio would shorten the time of flight and would be
beneficial for gravity losses to a much greater degree that the increased velocity would be
detrimental to drag losses. Even though the results show that the ratio of gravity losses
to drag losses is considerably smaller than in conventional vehicles this is not believed
to be the primary reason for the low results for thrust-to-weight ratios. There is namely
another argument to be made in this discussion. While a higher thrust-to-weight ratio is
mostly beneficial to the overall losses that occur during flight, an increase in this ratio can
be achieved in two ways. The first is an increase in engine thrust, which increases engine
mass according to Eq. (3.44) and (3.43), increases the thrust structure mass according
to Eq. (3.45) and the tank thickness, Eq. (3.55). The second is to make the propellant
tanks smaller for the same engine, which results in higher area to volume ratios for the
propellant tanks and thus also higher construction mass ratios. In addition, in section
3.3.3 a general drive to minimize the thrust-to-weight ratio was discovered when only
trying to maximize the V of a launcher stage. This was caused due to a minimization
7
This is for the kerolox types of vehicles, but the hydrolox vehicles share the same observations.
90 Results

of the thrust structure mass and a low nozzle mass for a high specific impulse.
Even though costs are not taken into account, where the engines are usually the more
costly component, the optimization model already hints at the use of small engines for
the launch vehicle from a performance standpoint.
Another important influence that the thrust-to-weight ratio has is the impact on the
optimal length of the vehicle. This was discussed in section 7.2.2 with the argument
that when the thrust-to-weight ratio increases, the drive for a larger length to minimize
the surface-to-volume ratio is increasingly counteracted by the drive to minimize the
hydrostatic pressure and the drive to minimize the engine mass ratio.

7.2.10 Tank mass and engine mass

The tank-to-propellant mass ratio of the kerolox launchers are vastly superior to the
mass ratios of the hydrolox launchers. For the optimized kerolox launchers the value
ranges from 33.48 to 43.95 while for the hydrolox launchers the value stays a lot more
constant in the range of 12.04 to 14.39. No significant trends could be established. The
engine-to-propellant mass has peaks of 62.93 and 70.01 and lows of 44.64 and 48.02 for
the hydrolox and kerolox vehicles respectively and also showed no significant trends. The
only conclusion that can be drawn from this is that the mass ratios are clearly favorable
for kerolox vehicles but eventually do not bridge the performance gap of the higher specific
impulse of hydrolox vehicles.

7.2.11 Specific impulse

4 4
x 10 x 10
4 4

4550 3550
3.5 3.5

3 4500 3
3500
2.5 2.5
4450
Altitude (m)

Altitude (m)

2 2
3450
4400

1.5 1.5

4350
3400
1 1

4300
0.5 0.5

3350
0 4250 0
0 500 1000 1500 2000 0 500 1000 1500 2000
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.14: Exhaust velocity of the optimized hydrolox vehicles (a) and kerolox vehicles (b).

The specific impulse depends on the exit velocity of the gas in the nozzle. This is given
by Eq. (3.40) and from Eqs. (3.34), (3.38) and (3.39), from which it can be seen that
the only parameters it influences are the exit pressure, the chamber pressure and the
oxidizer-to-fuel ratio (through the ratio of specific heats , chamber temperature, Tc , and
7.2 Discussion of the launch-vehicle parameters 91

molar mass, R). Since the variations in the chamber pressure and oxidizer-to-fuel ratio
are rather low, the reflection of the exit pressure plots is most prevalent when looking
at the chamber pressure plots. In the results given in Appendix B and plotted in figure
7.14, it can be seen that the exhaust velocity is often rather high. When no-assist was
simulated, the exhaust velocities of the optimal solutions are the lowest for all assist
combinations investigated. This is due to the fact that the launcher remains a longer
time period in denser regions of the atmosphere. The lowest exhaust velocities found
were 3,321.7 m/s for the kerolox launch vehicle and 4,228.2 m/s for the hydrolox launch
vehicle, corresponding to a specific impulse of 338.6 and 431 s respectively8 .
Since these simulations were run without any assist the launch conditions are the same
to those of other launchers. Therefore, these values could logically be compared with
the values given for the 6 first stages presented in tables 3.4 and 3.5. The average first
stage liquid hydrogen specific impulse is 430 seconds with a maximum of 446.1 swhile
the first stage kerosene fueled engines achieve an average of 329.9 and a maximum of
337.8 s. The exhaust velocities are found to increase with the magnitude of the assist and
reach a maximum when the assist consist out of an altitude of 40,000 m and a velocity of
2,000 m/s. Here the optimal vehicles had exhaust velocities of 4,587.4 and 3,578 m/s for
hydrolox and kerolox stages respectively. This translates to specific impulses of 467.6 and
364.7 s. For the hydrolox launcher this compares well with the expected efficiency of the
Vinci engine, which is 465 s. For the kerolox vehicle, the RD-0124 kerolox engine will be
used as a benchmark, which is one of the most efficient kerolox engines in the world and
reaches a specific impulse of 357 s [50]. All in all, the found optimal engine efficiencies are
in correlation with the industry standard. However, a case could be made that the model
slightly overestimates the efficiency of the kerolox launcher. On the other hand, other
upper stage launch designs must take into account the larger interstage mass needed for
a longer nozzle. This does not effect the optimization process of a single stage, which is
performed in this study, resulting in large rocket nozzles with high area ratios. To have a
sense of how important the specific impulse is, the thrust coefficient correction factor was
lowered from 0.983 to 0.96 for the kerolox case of 1,000 m/s assist velocity and 10,000 m
altitude. This leads to a reduction of the payload ratio of 0.384% or a relative decrease
of 4.4%. In the case of just 300 m/s at 10,000 m it has a larger effect, namely a reduction
of 0.505% or a relative decrease of 10.6%.

7.2.12 Drag Loss

The nature of drag loss was already discussed previously in section 3.5. From the results
obtained and shown in figure 7.15, two main peaks can be established for both vehicle
types. These peaks are located at the lowest altitude and at the outermost extremes of
the velocity spectrum investigated. In addition, a valley of local minima can be seen at
velocities between 500 and 1,000 m/s. This can be explained by examining Eq. (3.75),
where drag loss is a function of three parameters: Drag(V 2 ), mass(t) and time itself.
Therefore it is logical that to minimize drag loss a certain compromise must be reached
between the velocity of traveling through the atmosphere and the time needed to escape
it. In Eq. (3.76) it was also shown that drag loss is inversely related to the average
density of the launcher and this can be seen when comparing the drag loss results of
8
All specific impulses used in this section are acquired in vacuum conditions
92 Results

4 4
x 10 x 10
4 4
400

3.5 3.5 300

350

3 3
250
300

2.5 2.5
250 200
Altitude (m)

Altitude (m)
2 2
200
150
1.5 1.5
150

1 1 100
100

0.5 0.5
50 50

0 0
0 500 1000 1500 2000 0 0 500 1000 1500 2000
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.15: Drag loss (m/s) of the optimized hydrolox vehicles (a) and kerolox vehicles (b).

the set of hydrolox launchers with the set of kerolox launchers. The hydrolox launchers
have an average drag loss of 143.6 m/s while the kerolox launchers have an average drag
loss of just 99.8 m/s. Another anomaly in the drag loss results is that the drag loss is
higher than expected at low-assist velocities. At the no-assist cases the drag loss reaches
values as high as 340.8 m/s and 428.9 m/s for kerolox and hydrolox launchers respectively,
which compares poorly with the values given for conventional launchers in [28] of 100 -
150 m/s. The reason for this anomaly is the high-lift trajectory that some of the low-
assist simulations tend to follow. These lifting trajectories, which are further discussed
section 7.3.2, would benefit smaller launchers more than larger launchers. This is again
due to the square-cube law that relates the surface area that can provide lift with the
volume/mass of the launcher. These trajectories also result in a flight profile where the
launcher accumulates more velocity at lower velocities and thus accumulate more drag
loss. In the altitude vs velocity plots displayed in figure 7.16, the 3 simulations with
no assist velocity and the lowest assist altitudes can be seen to have significantly lower
profiles. This is more so in the kerolox plot, displayed on the right than in the hydrolox
plot. In the trade off between these high-lift trajectories vs low-drag loss trajectories,
the pendulum swings more in the direction of high-lift trajectories for kerolox launchers
than it does for hydrolox launchers. This is due to the fact that kerolox launchers are
less sensitive to increased drag losses. This is also displayed by the comparison between
figures C.24 and C.25 where it is displayed that the lift-to-weight is about the same
magnitude for kerolox and hydrolox despite the severe density difference between both
launcher configurations.

L S ld 1
2
= = (7.4)
M V ld d l
Eq. (7.4) shows that the lift-to-mass ratio is dependent on the slenderness, , to length
ratio. And while this is not the only driver for the slenderness and the length it might
give an indication why the length anomaly of high drag and short length appears in the
low assist corner of the assist spectrum provided.
7.2 Discussion of the launch-vehicle parameters 93

5 5
x 10 x 10
3 3
0 m / 0 m/s 0 m / 0 m/s
5,000 m / 0 m/s 5,000 m / 0 m/s
10,000 m / 0 m/s 10,000 m / 0 m/s
2.5 20,000 m / 0 m/s 2.5 20,000 m / 0 m/s
40,000 m / 0 m/s 40,000 m / 0 m/s
0 m / 300 m/s 0 m / 300 m/s
0 m / 2,000 m/s 0 m / 2,000 m/s
2 2
Altitude (m)

Altitude (m)
1.5 1.5

1 1

0.5 0.5

0 0
0 1000 2000 3000 4000 5000 6000 7000 8000 0 1000 2000 3000 4000 5000 6000 7000 8000
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.16: Altitude vs Velocity plots of various launch cases for optimized hydrolox vehicles (a)
and kerolox vehicles (b).

7.2.13 Gravity Loss

4 4
x 10 x 10
4 4 750

700
700
3.5 3.5
650
650
3 3
600 600
2.5 2.5
550 550
Altitude (m)

Altitude (m)

2 2
500
500

1.5 1.5 450

450

1 1 400
400
350
0.5 0.5
350
300
0 0
0 500 1000 1500 2000 300 0 500 1000 1500 2000
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.17: Gravity loss (m/s) of the optimized hydrolox vehicles (a) and kerolox vehicles (b).

The gravity-loss difference between the different type of vehicles is almost negligible with
an average of 482.2 m/s for hydrolox vehicles and 467 m/s for kerolox vehicles. However,
when combining both types of losses, it can clearly be seen that kerolox launchers (566.8
m/s of average total losses) have an advantage to hydrolox launchers (625.8 m/s) of
about 10%. Maxima of both launcher types are nearly the same with 721,5 m/s for the
hydrolox launcher (at 0 altitude and 300 m/s) and 764.5 m/s for the kerolox launcher (at
0 altitude and 600 m/s). This compares with values of conventionally launched vehicles,
which have gravity losses in the order of 1,000 - 1,500 m/s [28]. The trend in increasing
gravity losses with decreasing assist velocity, which occurs from 2000 m/s all the way
down to 600 m/s and in lesser extent 300 m/s is easy to explain. Furthermore if this
94 Results

trend would have materialized in the simulations with no-assist velocity, the gravity loss
expected for the no-assist case would fall within or come close to the range of values
of conventionally launched vehicles9 . This trend can shortly be explained to be due to
various causes. Firstly, there is the fact that at higher assist velocities, the duration of the
flight can be shorter, since the difference between initial velocity and orbital velocity is
smaller. Secondly, the initial flight-path angle and the flight-path profile in general can be
significantly lower, since the initial vertical velocity component is already large at smaller
flight-path angles and because the force against which the rocket must climb is smaller
due to the centrifugal component. Looking at figure 7.18 a noticeable ridge is created
at a velocity of 300 m/s indicating that at all altitudes with the exception of 40 km,
the combined losses due to drag and gravity are highest at this assist velocity, an assist
velocity often used in launch assist proposals. The reason for the anomaly encountered
at low-assist velocities is more difficult to explain. This is most likely due to the major
difference in the altitude profile these cases have when compared to other simulations.
This altitude anomaly is further explained in section 7.3.3 and can be observed in figures
C.8 and C.8. The gravity-loss anomaly is according to this hypothesis due to the higher
thrust-to-weight ratio these launchers have, which results in a shorter burn time and
causes them to burn more of their fuel at lower altitudes than the other launchers, which
can be seen in the altitude-velocity plots 7.16.

4 4
x 10 x 10
4 1000 4

900
3.5 3.5
900

3 3 800
800

2.5 2.5 700

Altitude (m)

Altitude (m)

700
2 2
600
600
1.5 1.5

500
1 500 1

400
0.5 0.5
400

0 0 300
0 500 1000 1500 2000 0 500 1000 1500 2000
Velocity (m/s) Velocity (m/s)

(a) (b)

Figure 7.18: Summation of the gravity and drag losses (m/s) of the optimized hydrolox vehicles (a)
and kerolox vehicles (b).

9
This with a launch to the fairly low target altitude of 250 km, which as explained in section 3.5 is
also of influence on the gravity loss.
7.3 Notable flights 95

7.3 Notable flights

7.3.1 No fairing trajectories

There are eight flights10 that have in common that the heat flux never surpasses the value
of 1,135 W/m2 . As can be seen in figure 7.19a, this is partially due to the atmospheric
model, which only assigns a density up to a geometric height of 120,000 m. In this figure all
heat-flux profiles of these eight flights are plotted and it clearly shows that the optimizer
optimized a second local maximum heat flux upon leaving the atmospheric model. This
not only demonstrates the limits of the model and the inaccuracies that it brings forth,
a more accurate atmospheric model would require some of the optimal trajectories to be
slightly higher in order to keep the heat flux low. But the results plotted in 7.19a also
demonstrates the capability and flexibility of the pitch program and the resultant flight
profile. While figure 7.19a shows the lowest 8 heat flux profiles, figure 7.19b shows the
runners up with peak heat fluxes in excess of 3,500 W/m2 indicating that the limit of
1,135 W/m2 is not trivial. The value for 1,135 W/m2 was used as the constraint for fairing
jettisoning. Modeling the trajectory in a way that does not surpass this low heat-flux limit
has as a consequence that the fairing can be jettisoned before ignition and does not need
to be accelerated along the way to orbit. To keep the heat flux below the aforementioned
limit, the flight path needs to be different compared with other launch simulations that
do not jettison the fairing before ignition. This can be seen in the form of a higher than
usual optimal initial flight-path angle and a steeper trajectory in general, causing the
launcher to travel sooner in less dense atmospheric conditions. This requirement of a
steeper ascent profile is greater for the low altitude launch, since it is more difficult for
this configuration to keep the heat flux below the limit. The fact that the lower the initial
altitude, the quicker the trajectory guides the rocket to denser atmospheric conditions
explains the counterintuitive decrease of optimal exhaust exit pressure with decreasing
altitude, which is a trend opposite to the trend seen in all other configurations. The

1200 5000
40,000 / 0 (RP1) 20,000 / 0 (RP1)
40,000 / 0 (LH2) 4500 20,000 / 0 (LH2)
20,000 / 300 (RP1)
1000 20,000 / 300 (LH2)
40,000 / 300 (RP1) 4000
40,000 / 300 (LH2)
40,000 / 600 (RP1) 3500
800 40,000 / 600 (LH2)
Heat Flux (W/m3)

Heat Flux (W/m3)

3000

600 2500

2000

400
1500

1000
200

500

0 0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300
Time (sec) Time (sec)

(a) Heat flux of eight optimized simulations with (b) Heat flux of 2 (LH2 and RP1) optimized simu-
varying altitude and initial velocity demonstrat- lations utilizing a launch assist of 20,000 m alti-
ing the models ability to cope with the hard tude with no velocity.
heat flux constraint after jettisoning.

10
Flights cases h-V of 40,000-300, 20,000-300, 40,000-600 and 40,000-0 for both fuel types.
96 Results

steep trajectory also translates to a higher than usual gravity loss. This interesting trade-
off between high gravity loss and no fairing mass has further less noticeable effects on
additional parameters. The increased importance of gravity loss might cause the optimal
T /W ratio to be higher, the decreased importance of drag loss might cause the slenderness
to be lower, etc.
It can basically be said that the optimizer makes a trade-off between the optimal solution
with jettisoning and the optimal solution without jettisoning. Looking at it in this manner
explains why of the flights at 20,000 m only the flights with a velocity of 300 m/s found
a low heat flux trajectory to be optimal. Logically one would assume that the optimum
without jettisoning after an assist at 20,000 with 0 m/s initial velocity would deviate less
from the optimum with jettisoning than would be the case of the 300 m/s counterpart.
However the advantage gained, no fairing mass during acceleration, also depends naturally
on the mass of the fairing, which is a lot lower in the case of no velocity assist.
The difference in heat flux by choosing to use a fairing is demonstrated in figure 7.19b.
Also note that the final heat flux is around the heat flux constraint implying that at
the edge of the atmospheric model, the heat flux of launchers traveling at near orbital
velocities already tends to fall below the constraint. This means that with an atmospheric
model extending to higher altitudes, similar optimal solutions regarding whether or not
a fairing needs to be used, are likely to be found.

7.3.2 Low-assist flights

While the tendency for gravity loss is to get higher with lower assist flights this does not
hold for the lowest of them all with low-assist altitudes and velocities. At low altitudes
and velocities a reversal of the trend in gravity loss is seen and drag loss increases more
than expected as was discussed in sections 7.2.13 and 7.2.12. In these cases, gravity losses
are kept low by utilizing the lift of the vehicle as a means for keeping the flight-path
angle low. This method does not occur anymore at higher velocities and altitudes as can
be seen in figure 7.19, where a comparison of different lift-to-weight ratios is displayed
and some of the higher assist cases are clearly identified of generating little to no lift.
These high lift trajectories are not seen in current conventional launcher profiles but was
already examined in other studies [18] [19], where a high lift trajectory with vertically
launched axis-symmetric launch vehicles is deemed to be capable to enhance performance.
The reason why modern launchers do not seem to fully utilize the lift generated by the
launchers body is up to speculation, perhaps excessive vibrations or bending loads lie
at the cause. Other reasons could be a flawed assumption that a gravity turn is always
optimal, aerodynamic instabilities or the fact that the vertical launch, which is always
performed by modern launchers with exception of the MU-V, in combination with kick-
rate constraints leads to a radically different ascent profile.

7.3.3 Altitude anomalies

In Appendix C a number of plots for all simulations is shown, one of these is the altitude
vs time plot of a hydrolox vehicle and of a kerolox vehicle, which are shown in figure 7.20.
A couple of anomalies are spotted in these plots. One of these, the kerolox case with
an assist velocity of 1,000 m/s at an altitude of 5 km recognizable by having the highest
7.3 Notable flights 97

0/0
0.4 5,000 / 0
10,000 / 0
0.35 0 / 300
5,000 / 300
10,000 / 300
0.3
0 / 1000
20,000 / 300
0.25
Lift to weight ratio

0.2

0.15

0.1

0.05

0.05

0.1
0 20 40 60 80 100 120 140 160 180 200
Time (sec)

Figure 7.19: Lift-to-weight ratio of liquid hydrogen vehicles utilizing different launch assist altitudes
and velocities. The legend identifies the specific simulations with their respective assist
parameters documented in the following format: h [m] / V [m/s].

altitude after 3,000 seconds, was already discussed in section 7.2.3. The other anomalies
tend to rise to apogee altitude, while most decreased their rate of climb significantly once
they were higher than 120 km. These anomalies were identified to be cases with no-
assist velocity at the four lowest altitudes11 for kerolox launchers and the three lowest for
hydrolox vehicles. These vehicles do not gain altitude rapidly due to their initial stationary
position. In addition, these also have the highest thrust-to-weight ratios, resulting in the
shortest times to orbit. This burn is in fact so short that these rockets cannot undertake
the normal trajectory where launchers still provide thrust at perigee, which is slightly
higher than the modeled atmosphere. This hypothesis is further strengthened by the
observed difference in burnout times of the hydrolox and kerolox vehicles with an assist
velocity of 0 m/s and an altitude of 20 km. The hydrolox launcher, which shows normal
behaviour, has a burn time of 296.2 seconds while the kerolox launcher, which shows an
anomaly, has a burn time of only 284.9 seconds.
Based on these cases the conclusion can be made that implementing a more intelligent
thrust profile with perhaps even coasting arcs would further improve the objective function
for these launchers.

11
The four lowest altitudes are 0, 5, 10 and 20 km
98 Results

5 5
x 10 x 10
3 3

2.5 2.5

2 2
Pitch angle (rad)

Altitude (m)
1.5 1.5

1 1

0.5 0.5

0 0
0 500 1000 1500 2000 2500 3000 0 500 1000 1500 2000 2500 3000
Time (sec) Time (sec)

(a) (b)

Figure 7.20: Altitude vs Time plots of hydrolox vehicles (a) and kerolox vehicles (b), indicating some
anomalies.

7.3.4 Effect of change in pitch angle

In this section the effects are discussed of having a pitch angle non-equal to the flight-path
angle. To have a clear view on this effect 4 simulations are run with an initial flight-path
angle of 0 for relatively low speed assists, since low speeds assist are most likely to be
assisted by aircraft where the optimized initial flight-path angle would be difficult to
achieve for most of the flights. These simulations were compared with their optimized
flight-path angle counterparts. The 8 resulting pitch angle profiles are visualized in the
top right of figure 7.21, showing an overcompensation in the second part of powered flight
for the sub-optimal initial pitch angle setting. All results presented in this section were
obtained by simulating hydrolox launchers. The results of these simulations show that
the drag loss increases substantially with only a slight improvement in gravity loss, which
can be seen in table 7.2. Although other factors play a role, this is deemed to be the main
reason that causes the objective value, the payload ratio, to fall. This drop in payload
ratio is lower than expected and can again be partially explained due to the fact that the
launchers utilize a high lift-to-weight trajectory significantly higher than their optimized
initial flight-path angle counterparts, which can be seen in the bottom right of figure 7.21.
These high-lift trajectories are partially explained by the altitude vs velocity plot in the
top left of figure 7.21, where it is shown that the cases with an initial flight-path angle
of 0 have a much higher velocity at the same altitude, enabling them to generate more
lift at comparable angles of attack. This causes the launcher to undergo higher bending
loads due to high angles of attack and dynamic pressures, see bottom left of figure 7.21,
but neither the bending load nor the angle of attack comes near any constraints.
To draw watertight conclusions from the high lift-to-weight benefit that sometimes seems
to occur it is recommended to create a better structural mass model with regard to the
high bending loads.
7.4 Sensitivity Analysis 99

Case (Identified by assist Drag Gravity Payload

parameters. Notation: V - h) loss (m/s) loss (m/s) ratio (-)
300 m/s - 10,000 m, = 0 1089.2 522.3 0.0661
300 m/s - 10,000 m 295.5 585.2 0.0681
300 m/s - 20,000 m, = 0 1328.6 364.4 0.0738
300 m/s - 20,000 m 17.2 643.8 0.0789
600 m/s - 10,000 m, = 0 1112.9 509.1 0.077
600 m/s - 10,000 m 61.8 670.1 0.0883
600 m/s - 20,000 m, = 0 1035.6 387.9 0.0851
600 m/s - 20,000 m 19.2 575.6 0.0962

Table 7.2: Drag loss, gravity loss and performance comparison of various launcher cases with either
an optimized initial flight-path angle or an initial flight-path angle of zero.

Figure 7.21: Comparison plots between optimal trajectories with an initial flight-path angle of zero
and with optimized flight-path angle. Top left: Altitude-velocity plot, top right: Pitch
angle - time plot, bottom left: Bending load - time plot and bottom right: lift-to-weight
ratio - time plot.

7.4 Sensitivity Analysis

The results that were discussed so far in this chapter, were also subjected to a sensitivity
analysis in order to know how sensitive the solution is to a change of one of the launch-
vehicle parameters. The sensitivity analysis was performed with Matlab code [45] utilizing
the extended Fourier amplitude sensitivity test (EFAST) method published in the paper
100 Results

of Saltelli, Tarantola and Chan [46]. This method is a variation on the FAST method,
which was developed in the 1970s. FAST computes the effect of each input factor to the
variance of the output. The variance of the output is generated by varying the parameters
around their optimal values in the form of a latin hypercube. The calculation of the effect
of each input parameter is then performed through Sobol sensitivity indices. There are
two types of indices, the first one is sometimes called first-order sensitivity indices and
is given by Eq. (7.5) and the second one is the total-effect sensitivity index given in Eq.
(7.6).
Vi Vi
Si = (7.5) ST i = 1 (7.6)
V () V ()
where is the output of the system and V () is the variance of the output. Vi is
the fraction of V () due to factor xi only while Vi indicates the contribution to the
variance due to all factors but xi and all the higher-order effects in which it is involved
[45]. This means that the sum of the total-effect sensitivity indices can be higher than
1 because the effect that the combination of parameter a and parameter b contributes
to the variance is incorporated both in the total-effect sensitivity index of parameter a
as well as in the total-effect sensitivity index of parameter b. The focus in this paper
will lie on the total-effect sensitivity indices shown in figure 7.22. The first-order effect
sensitivity indices are displayed in figure C.1. The sensitivity analysis is as discussed,
performed with variations around the global optimum found by the differential evolution
algorithm. When a constraint is met, the objective function is assigned a penalty value
so that the differential evolution algorithm discards that solution. For the sensitivity
analysis this penalty was slightly altered to a maximum of 10% per constraint compared
with the expected result. The sensitivity analysis was performed in two parts. The
first part analysed the sensitivity of the launch vehicle parameters and the second part
analysed the launch trajectory components, namely the initial flight-path angle and the
four pitch parameters. All optimal solutions were again simulated together with a total
of 24,966 variations of these optimal solutions. This number was determined obtained
by starting out with the minimum value for the amount of parameters as recommended
in [46]. Several runs were made with several multitudes of this minimum value used as
the number of data points within the hypercube. These results are shown in figure C.2.
After this analysis, 24,966 variations was the chosen amount to conduct the sensitivity
analysis. These 24,966 variations were spaced in a hypercube with a size of 1% of the
search space in accordance to the EFAST algorithm. For each optimal solution, a different
hypercube was generated, resulting in a total of 30 hypercubes12 . The results of the overal
importance (including combined effects) of the parameters on the model are given in figure
7.22. As can be seen from this figure, the two most important parameters are the length
and the exit pressure. Two notable differences can be detected between the kerolox and
hydrolox results. These are a smaller effect for kerolox vehicles due to the length of
the vehicle as well as due to the oxidizer-to-fuel ratio. The second effect can be solely
attributed to the fact that the variations performed during the sensitivity analysis were
for all parameters and for all optimal points, scaled to be 1% of the initial search space
of the parameter space. This results in smaller relative variations around the optimal
oxidizer-to-fuel ratio of hydrolox vehicles, which have oxidizer-to-fuel ratios of 6 to 6.5,
12
The same set of hypercubes were used to analyze both the hydrolox simulations and the kerolox
simulations
7.4 Sensitivity Analysis 101

Figure 7.22: Mean of the total standardized effect of parameters on the variance of the model with
indications of the maximum and minimum values

compared to the relative variations for the kerolox vehicles, which have oxidizer-to-fuel
ratios of around 2.5 (about a relative difference of a factor of 2.5). In fact, when scaling
the effect of the parameter to their relative sizes, one founds that both types of vehicles
are about equally sensitive to the parameter. Total sensitivity to the oxidizer-to-fuel ratio
was found to be 16.1% for the kerolox launcher, a comparable factor 2.3 larger than the
7.1% found for the hydrolox launcher. The difference in the effect that length has on
launcher performance can basically be contributed to the same effect. Optimized values
for the length of hydrolox vehicles are slightly larger than optimized values of kerolox
vehicles. Therefore for a given variation the greater effect on the model will be given by
the shorter launcher.
The overall results show that length has a bigger total effect than slenderness. While
they are two comparable parameters, there are several reasons why this could be the
case. These could have to do with accumulated drag loss and changes in the launchers
mass ratio, but the reason deemed most important for this effect is the thrust-to-weight
ratio. With increased slenderness, the diameter increases, leading to an increase in mass,
but so does the nozzle diameter and thrust thus the thrust-to-weight ratio stays the same.
With an increase in length, both the length and the diameter13 increase leading to a more
rapid mass increase than thrust increase. The only other parameter whose sensitivity

13
When slenderness remains the same value, the diameter of the vehicle will increase in correspondence
with the length.
102 Results

hinges greatly on the thrust-to-weight ratio effect is the nozzle-to-diameter ratio. This
parameters sensitivity, which is also effected by changes in the construction mass ratio
due to a change in engine power, is the third greatest and thus further demonstrates the
likely importance of the thrust-to-weight ratio on the model. The largest impact on the
model is made by variations in the exit pressure. This parameter showed a clear trend of
becoming increasingly important at high-assist simulations. This is because at this high
assists the optimum exit pressure is rather low and the variation of 1% of the search space,
a variation of 1000 Pa, has far greater effects when this occurs around an optimum value
of 5,000 Pa w.r.t an optimal value of 50,000 Pa. This effect can be seen in figure C.3. This
implies that the magnitude of the importance of the exit pressure mainly comes from the
fact that the variation is relatively speaking quite large for this parameter. Nonetheless,
even in the four cases with the largest optimal exit pressure (> 15000 Pa) the average
sensitivity of the model to the exit pressure was still close to 20%.
 Chapter 8

Conclusions and Recommendations

This work began with the formulation of the research question:

How does a launch assist affect the performance and the design of a launch vehicle?

In section 8.1 the conclusions that were drawn during this thesis work are given. This
is followed by section 8.2 where the recommendations are summarized for future work
related to this research in 8.2.1 and for future launch-vehicle designs in section 8.2.2.

8.1 Conclusions

1. The differential evolution algorithm was shown to be better suited to tackle this
particular problem than the particle swarm optimization technique.
DE was found to be suited to find solutions for even the most difficult cases, while
the PSO algorithm did not found any solution that obeyed the constraints. For the
more easier cases the PSO method found better solutions faster but mis converged
at local optima.

2. Variations in the length and the exit pressure were found to have the largest impact
on the sensitivity of the model.
Some of the conclusions concerning the sensitivity analysis were less transparent
due to the fact that some parameters had more optimal solutions near the lower
boundary of the search space than others, making the variations performed during
the analysis larger in a relative sense. Nonetheless, the sensitivity analysis made
clear that the model is more sensitive to the length and the exit pressure than the
chamber pressure, the oxidizer-to-fuel ratio and the slenderness, which were found
to have the least effect. The nozzle exit-to-tank diameter ratio was also found to
have a significant effect due to its influence on the thrust-to-weight ratio.

103
104 Conclusions and Recommendations

3. The overall effect of the chamber pressure on the performance is rather marginal
but the effect of increased specific impulse with increased chamber pressure almost
uniformly outweighs the construction-mass ratio increase.
The model was found to be the least sensitive to the chamber pressure and this was
due to the fact that at large pressures, where the optimum was uniformly found, the
chamber pressure exerts only very slight differences on the specific impulse. Only
in a couple of cases this drive toward optimal high chamber pressures was to some
extent counteracted by the negative effect the chamber pressure has on the mass
ratio.

4. The performance of the launcher measured by the payload mass-to-initial mass ratio
increases both with increased altitude as well with increased velocity.
The payload ratios always increase with an increase in assist altitude as it does
with an increase in assist velocity. The payload ratios rise from 3.1% and 2.1% for
the no assist case to a maximum of 19% and 15.1% for the hydrolox and kerolox
launchers respectively. The maxima both occur at the highest assist velocity and
highest assist altitude simulated, which were 2,000 m/s and 40,000 m.

5. Launchers with a low density hydrogen-oxygen propellant mixture were found to

provide greater payload mass ratios to a low Earth orbit than launchers containing
a high density kerosene-oxygen propellant mixture for all assist cases investigated.
The difference in performance ranged from a relative improvement of 25.7% at the
maximum assist case to 48.4% at the no assist case.

6. Combinations of a high-assist altitude and a low-assist velocity led to optimal launcher

configurations without a fairing. Eight out of the sixty trajectory simulations were
found to perform better when a trajectory is flown that keeps the heat flux below
the fairing jettisoning constraint for the entire flight profile. These flights, where
the fairing is allowed to be jettisoned prior to launch, had assist altitudes of 40 km
combined with assist velocities of 0, 300 and 600 m/s, the 7th and the 8th flight had
assist altitudes of 20 km and assist velocities of 300 m/s.

7. The assist velocity has a larger impact on the performance of the launcher than
the assist altitude. This holds for both launcher types investigated but more so for
kerolox launchers than hydrolox launchers.
The performance surplus generated by an assist altitude of 40 km compared with an
assist altitude of 0 km roughly corresponds with the performance increase generated
by having an assist velocity of 540 m/s instead of 0 m/s for hydrolox launchers. For
a kerolox launcher this performance increase is already met at assist velocities of
about 400 m/s. This implies that when given a fast assist platform, the trade-off
between a kerolox launcher and a hydrolox launcher is more in favor of the kerolox
launcher than it would be when a high assist platform would be used.

8. The gravity loss and the optimal thrust-to-weight ratio decreases with larger launch-
assist altitude and velocity.
Gravity loss and thus its negative effect on launcher performance gets smaller with
increasing assist velocity and altitude. Smaller thrust-to-weight ratios cause a de-
crease in the construction mass ratio while it causes an increase in gravity loss. At
8.1 Conclusions 105

large launch-assist altitudes and velocities, the lower construction mass effect be-
comes more important compared with the gravity loss effect, leading to a reduction
in the thrust-to-weight ratio.

9. The overall tendency regarding the dimensions of the launchers is to minimize them.
The average length of all optimized launcher tanks is 22.9 m, while their diameter
is on average 2.35 m and these values are mainly that high due to a few aberran-
cies, 36 of the 60 launchers have shorter tank lengths than 20 m. Optimal kerolox
launchers are generally smaller than optimal Hydrolox launchers. A major driver
to keep dimensions low is a high thrust-to-weight ratio, but this was found to be
counteracted in high-drag-loss environments to minimize the drag loss.

10. The importance of construction mass ratio is greater for low-performance systems
than in their high-performance counterparts causing the slenderness to follow the
drag loss trends more closely in high performance areas.
A certain increase in the construction mass leads to a higher increase in V for
low-performance systems than for high-performance systems. A more slender body
improves on the systems drag loss regardless of performance. Therefore, optimiza-
tion of the slenderness with the purpose of minimizing drag loss plays a greater role
for high performance systems. This leads to a close match between the drag loss
and the slenderness of the vehicle in the high-performance and high-drag-loss region
of the launch-assist spectrum.

11. At constant altitude, the drag loss reaches a minimum for assist velocities between
500 and 1,000 m/s.
Low assist velocities cause the launcher to spend large amount of times in the denser
parts of the atmosphere. However, since drag force is proportional to the second
power of the velocity, the velocity effect on the drag loss gets larger than the effect
of the longer time spent in denser parts of the atmosphere and causes the drag loss
to rise again with increased assist velocity.

12. At constant altitude, the sum of gravity loss and drag loss reaches a maximum for
assist velocities at about 300 m/s.
The sum of gravity and drag loss tends to get greater with decreasing velocity. How-
ever it reaches a maximum just before the no-assist velocity cases. This maximum is
due to a combination of the high-lift trajectories that are being flow from no-assist
velocity cases, which increase drag losses but decrease gravity losses, and from the
higher thrust-to-weight the vehicles have when utilizing an assist without any initial
velocity, which decreases gravity losses even further.

13. Low assist flights and flights with a 0 initial flight-path angle prefer high-lift trajec-
tory profiles
Several solutions for flights using assists at low altitudes and low velocities utilized
the ability of the launcher to generate lift to its advantage. This entailed lower than
normal flight profiles and higher than normal drag losses, bending loads and angles
of attack. This phenomenon was even magnified when launch assist were simulated
with 0 flight-path angles. While this might indeed be a better solution than the
standard gravity turn, its validity is questioned due to the mass models inability
to cope with the influence of bending loads.
106 Conclusions and Recommendations

14. The optimal initial flight-path angle decreases both with increasing assist altitude
and even more so with increasing assist velocity.
At the no assist case the initial flight path angle is close to vertical with values of
85 and 82 for kerolox and hydrolox launchers respectively. This decreases both
with increased assist altitude as with increased assist velocity. The decreasing effect
with increasing assist altitude is quite small in general but gets more prominent
with increased assist velocity. The optimal flight path angles for the highest assist
cases are 28 and 30 for kerolox and hydrolox launchers respectively.

15. The slenderness of the vehicle and the nozzle exit diameter have opposite trends,
which conflict with launch assists that use high-assist velocities and low-assist alti-
tudes.
The slenderness of the launcher increases with high-velocity and low-altitude assists,
due to the higher drag losses that occur and the high payload ratio that causes a
smaller effect of the construction mass ratio. In the same region, the optimal exit
pressure is low and the mass is high for small tank diameters. To have adequate
thrust-to-weight ratios and high specific impulses, the nozzle exit-to-tank diameter
ratios reach the upper boundary of the search space, parity, in certain cases.

16. The effect on the specific impulse is dominated by the optimal exit pressure.
Due to the fact that the oxidizer-to-fuel ratio and the chamber pressure show little to
no variation across the assist spectrum, the exit pressure has the greatest influence
on the specific impulse. The specific impulse is large when the exit pressure is low
and vice versa. The exit pressure shows a singular peak around the no assist case
since these are the cases where the propelled flight spends a more significant portion
in a high density environment.

8.2 Recommendations

8.2.1 For further study on the subject:

Include the mass and aerodynamic model of an attachable wing to investigate its
effect. This is deemed to be the most important recommendation, since the results
that were obtained, implied that aerodynamic lift can have an important effect on
the trajectory and consequently on the overall performance.

Include a more realistic mass model effect for the chamber pressure. Currently, it
only has an almost negligible effect on the nozzle mass.

Include a more realistic engine mass model. In this model, engine mass is dependent
solely on the thrust. This implies that when a given engine increases its efficiency,
it also increases its mass.

Include a more realistic mass model effect for the bending load. Perhaps by modeling
an exoskeleton that can carry the loads and can be jettisoned in a similar fashion as
the fairing. This would save mass and was also proposed in the form of a strongback
structure of a winged vehicle in [35].
8.2 Recommendations 107

Estimate the residual propellant mass so that this can be subtracted from the final
inert mass, where the payload mass was now based upon. It should be noted that an
inclusion of the residual propellant mass model would only affect the trends display
in the results if it would be influenced by launch-vehicle parameters. If it would be
modeled as for instance a standard 1% of the initial mass, all optimized solutions
would stay the same1 .

Create either a higher resolution in the CEA database for the oxidizer-to-fuel ratio
or a more accurate interpolation method. The results displayed a rather unrealistic
picture that showed little to no variation in the parameter.

Include a constraint concerning maximum pitch rate.

Invent a more rigorous approach to define the search spaces of the pitch variables.
It should be noted that further increases in these search space have been shown to
lead to slower optimization times and higher chances for mis convergence.

Adjust the thrust model so that the thrust does not have an immediate response.

Implement a lower boundary constraint for the thrust throttle setting.

Model the thrust as a series of optimizable nodes connected by a non-linear inter-

polation method instead of modeling it at constant thrust or constant axial accel-
eration.

Include a staging event and either optimize the second stage in the same manner as
was done in this study or simulate a standard upper stage.

Include dependency on the Reynolds number on the model. This can be done
by adding an extra dimension to the aerodynamic coefficient database, an extra
dimension to the interpolation, and by using multiple Missile DATCOM input files.
This will improve the accuracy of the aerodynamic coefficients.

Interpolate, based on length and diameter, between the files generated by Missile
DATCOM.

Implement a different aerodynamic coefficient generator. It has been indicated that

Missile DATCOM does not always estimate reality with great success. One of the
most noticeable errors is the miscalculation of the axial coefficient, where double
Mach peaks have been observed between 1 and 2 Mach [14]. Implementation of a
newer version of the Missile DATCOM code would also be an improvement.

Perform a sensitivity analysis on the vehicle model to identify the sensitivity of

the model to certain assumptions such as fill ratio, ullage pressure, safety factors,
efficiency factors, etc.
1
It might have a slight influence if it was modeled as a more realistic 1% of the initial propellant mass.
108 Conclusions and Recommendations

8.2.2 For launch-vehicle design:

Raising the altitude from sea level to 5,000 m altitude for the cases with 0 m/s
assist velocity results in a relative increase of the payload ratio of 27% and 21% for
hydrolox and kerolox respectively. Combined with the fact that the optimal vehicles
for these cases are rather small, 12.1 and 39.8 tonnes, there might be a business case
for an elevated launch site for small launchers.

It should be further investigated if the requirement concerning heat flux for payloads
of current and proposed assisted launch vehicles could not be raised to the extent
that these vehicles would not need to provide a fairing during their ascent trajectory.
A significant rise in the heat flux constraint will get rid of the need for a fairing
for even lower and faster assisted vehicles and perhaps also horizontally launched
vehicles.

The results make the case for the use of relatively small engines from a performance
point of view. Although it is advised to make a final decision on this topic after
further investigation on the implementation of an optimized thrust profile.

The results indicate several possibilities for launch-assist parameters that could lead
to a launch-vehicle payload ratio greater than 10%. If this could be demonstrated
the next logical step would be to trade some payload for reusability systems such as
a heat shield, parachutes, de-orbit propellant mass, etc. Transforming the system
into the first fully reusable space system, assuming reusability of the assist platform.
 Appendix A

CEA Tables

In this appendix the tables generated with the online CEA tool are shown.

Table A.1: Molar mass of hydrogen-oxygen mixtures

HH Pc
H
8 16 24 32 40 48 56 64
O/F HH
H
1 4,032 4,032 4,032 4,032 4,032 4,032 4,032 4,032
1,5 5,04 5,04 5,04 5,04 5,04 5,04 5,04 5,04
2 6,047 6,047 6,047 6,047 6,047 6,048 6,048 6,048
2,5 7,051 7,052 7,053 7,053 7,054 7,054 7,054 7,054
3 8,04 8,046 8,049 8,051 8,052 8,053 8,054 8,054
3,5 8,998 9,015 9,024 9,029 9,033 9,036 9,038 9,04
4 9,915 9,948 9,965 9,976 9,984 9,99 9,996 10
4,5 10,786 10,838 10,866 10,884 10,898 10,909 10,918 10,926
5 11,613 11,683 11,723 11,75 11,771 11,787 11,8 11,812
5,5 12,393 12,483 12,534 12,57 12,598 12,62 12,638 12,654
6 13,129 13,236 13,298 13,342 13,376 13,404 13,427 13,447
6,5 13,821 13,942 14,014 14,066 14,105 14,138 14,166 14,189
7 14,47 14,603 14,683 14,74 14,785 14,821 14,852 14,879
7,5 15,079 15,221 15,307 15,368 15,416 15,456 15,489 15,518
8 15,416 15,456 15,489 15,518 16,003 16,045 16,08 16,11
8,5 16,191 16,343 16,434 16,499 16,551 16,593 16,628 16,66
9 16,7 16,853 16,945 17,011 17,062 17,105 17,14 17,171
9,5 17,18 17,334 17,426 17,491 17,542 17,584 17,62 17,65
10 17,635 17,788 17,879 17,943 17,994 18,035 18,07 18,1

109
110 CEA Tables

Table A.2: Molar mass of hydrogen-oxygen mixtures

HH P
H c 72 80 88 96 104 112 120 128
O/F HH H
1 4,032 4,032 4,032 4,032 4,032 4,032 4,032 4,032
1,5 5,04 5,04 5,04 5,04 5,04 5,04 5,04 5,04
2 6,048 6,048 6,048 6,048 6,048 6,048 6,048 6,048
2,5 7,054 7,054 7,054 7,054 7,054 7,054 7,054 7,054
3 8,055 8,055 8,056 8,056 8,056 8,057 8,057 8,057
3,5 9,041 9,043 9,044 9,045 9,046 9,047 9,048 9,048
4 10,003 10,007 10,009 10,012 10,014 10,016 10,018 10,019
4,5 10,932 10,938 10,943 10,947 10,951 10,955 10,958 10,961
5 11,822 11,83 11,838 11,845 11,851 11,857 11,862 11,867
5,5 12,667 12,679 12,69 12,7 12,709 12,717 12,724 12,731
6 13,465 13,48 13,494 13,507 13,518 13,529 13,539 13,548
6,5 14,21 14,229 14,246 14,261 14,275 14,288 14,3 14,312
7 14,903 14,924 14,943 14,961 14,977 14,992 15,006 15,019
7,5 15,544 15,567 15,588 15,607 15,625 15,641 15,656 15,671
8 16,137 16,161 16,183 16,203 16,222 16,239 16,255 16,27
8,5 16,687 16,712 16,734 16,754 16,773 16,79 16,807 16,822
9 17,199 17,223 17,246 17,266 17,285 17,302 17,318 17,333
9,5 17,677 17,702 17,724 17,744 17,762 17,779 17,795 17,81
10 18,126 18,15 18,172 18,191 18,209 18,226 18,241 18,256

Table A.3: Molar mass of hydrogen-oxygen mixtures

HH P
H c 136 144 152 160 168 176 184 192
O/F HH H
1 4,032 4,032 4,032 4,032 4,032 4,032 4,032 4,032
1,5 5,04 5,04 5,04 5,04 5,04 5,04 5,04 5,04
2 6,048 6,048 6,048 6,048 6,048 6,048 6,048 6,048
2,5 7,054 7,055 7,055 7,055 7,055 7,055 7,055 7,055
3 8,057 8,057 8,058 8,058 8,058 8,058 8,058 8,058
3,5 9,049 9,049 9,05 9,05 9,051 9,051 9,052 9,052
4 10,021 10,022 10,023 10,025 10,026 10,027 10,028 10,029
4,5 10,964 10,967 10,969 10,971 10,973 10,975 10,977 10,979
5 11,872 11,876 11,88 11,884 11,887 11,89 11,893 11,896
5,5 12,738 12,744 12,75 12,755 12,76 12,765 12,769 12,774
6 13,556 13,565 13,572 13,579 13,586 13,592 13,599 13,604
6,5 14,322 14,332 14,341 14,35 14,359 14,367 14,374 14,382
7 15,031 15,043 15,054 15,064 15,074 15,083 15,092 15,101
7,5 15,684 15,697 15,709 15,72 15,731 15,741 15,751 15,76
8 16,284 16,297 16,309 16,321 16,333 16,344 16,354 16,364
8,5 16,836 16,85 16,862 16,874 16,886 16,897 16,907 16,917
9 17,348 17,361 17,374 17,386 17,397 17,408 17,418 17,428
9,5 17,824 17,837 17,849 17,861 17,872 17,883 17,893 17,902
10 18,269 18,282 18,294 18,305 18,316 18,326 18,336 18,346
 111

Table A.4: Molar mass of hydrogen-oxygen mixtures

HH P
H c 200 208 216 224 232 240 248 256 264
O/F HH H
1 4,032 4,032 4,032 4,032 4,032 4,032 4,032 4,032 4,032
1,5 5,04 5,04 5,04 5,04 5,04 5,04 5,04 5,04 5,04
2 6,048 6,048 6,048 6,048 6,048 6,048 6,048 6,048 6,048
2,5 7,055 7,055 7,055 7,055 7,055 7,055 7,055 7,055 7,055
3 8,058 8,058 8,058 8,059 8,059 8,059 8,059 8,059 8,059
3,5 9,053 9,053 9,053 9,053 9,054 9,054 9,054 9,055 9,055
4 10,029 10,03 10,031 10,032 10,033 10,033 10,034 10,035 10,035
4,5 10,981 10,982 10,984 10,985 10,987 10,988 10,989 10,991 10,992
5 11,899 11,902 11,904 11,907 11,909 11,911 11,914 11,916 11,918
5,5 12,778 12,782 12,786 12,789 12,793 12,796 12,799 12,802 12,805
6 13,61 13,615 13,62 13,625 13,63 13,634 13,639 13,643 13,647
6,5 14,389 14,395 14,402 14,408 14414 14419 14425 14430 14435
7 15,109 15,117 15,124 15,131 15,138 15,145 15,152 15,158 15,164
7,5 15,769 15,778 15,786 15,794 15,802 15,81 15,817 15,824 15,831
8 16,373 16,382 16,391 16,4 16,408 16,416 16,423 16,431 16,438
8,5 16,927 16,936 16,945 16,954 16,962 16,97 16,977 16,985 16,992
9 17,438 17,447 17,455 17,464 17,472 17,48 17,488 17,495 17,502
9,5 17,912 17,921 17,929 17,937 17,945 17,953 17,96 17,968 17,975
10 18,354 18,363 18,371 18,379 18,387 18,394 18,402 18,409 18,415

Table A.5: Ratio of specific heats of hydrogen-oxygen mixtures

HH P
H c 8 16 24 32 40 48 56 64
O/F HH H
1 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587
1,5 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157
2 1,2813 1,2817 1,2819 1,282 1,2821 1,2822 1,2822 1,2822
2,5 1,2504 1,2528 1,2538 1,2545 1,2549 1,2552 1,2555 1,2557
3 1,2183 1,2237 1,2263 1,228 1,2292 1,2301 1,2308 1,2313
3,5 1,1891 1,1963 1,2002 1,2027 1,2046 1,206 1,2072 1,2081
4 1,1668 1,1741 1,1782 1,1811 1,1832 1,1849 1,1863 1,1875
4,5 1,1508 1,1574 1,1613 1,164 1,1661 1,1678 1,1693 1,1705
5 1,1395 1,1452 1,1487 1,1511 1,1531 1,1547 1,156 1,1571
5,5 1,1317 1,1366 1,1395 1,1417 1,1434 1,1448 1,1459 1,147
6 1,1265 1,1307 1,1332 1,135 1,1365 1,1377 1,1387 1,1396
6,5 1,1232 1,1269 1,1291 1,1307 1,1319 1,133 1,1338 1,1346
7 1,1213 1,1247 1,1267 1,1281 1,1293 1,1302 1,1309 1,1316
7,5 1,1204 1,1236 1,1255 1,1269 1,1279 1,1287 1,1295 1,1301
8 1,12 1,1232 1,1251 1,1264 1,1274 1,1282 1,1289 1,1295
8,5 1,12 1,1232 1,1251 1,1264 1,1274 1,1283 1,129 1,1296
9 1,1202 1,1234 1,1253 1,1267 1,1277 1,1286 1,1293 1,1299
9,5 1,1205 1,1238 1,1258 1,1272 1,1282 1,1291 1,1299 1,1305
10 1,1209 1,1243 1,1263 1,1277 1,1288 1,1297 1,1305 1,1311
112 CEA Tables

Table A.6: Ratio of specific heats of hydrogen-oxygen mixtures

HH P
H c 72 80 88 96 104 112 120 128
O/F HH H
1 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587
1,5 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157
2 1,2823 1,2823 1,2823 1,2823 1,2823 1,2824 1,2824 1,2824
2,5 1,2559 1,256 1,2562 1,2563 1,2564 1,2565 1,2565 1,2566
3 1,2318 1,2322 1,2326 1,2329 1,2332 1,2334 1,2337 1,2339
3,5 1,209 1,2097 1,2103 1,2108 1,2113 1,2118 1,2122 1,2126
4 1,1885 1,1894 1,1902 1,1909 1,1916 1,1922 1,1927 1,1932
4,5 1,1716 1,1725 1,1734 1,1741 1,1748 1,1755 1,1761 1,1766
5 1,1582 1,1591 1,1599 1,1606 1,1613 1,1619 1,1625 1,1631
5,5 1,1479 1,1487 1,1494 1,1501 1,1507 1,1513 1,1518 1,1523
6 1,1403 1,1411 1,1417 1,1423 1,1428 1,1433 1,1438 1,1442
6,5 1,1352 1,1358 1,1364 1,1369 1,1373 1,1378 1,1382 1,1385
7 1,1322 1,1327 1,1332 1,1336 1,134 1,1343 1,1347 1,135
7,5 1,1306 1,1311 1,1315 1,1319 1,1323 1,1326 1,1329 1,1332
8 1,1301 1,1305 1,131 1,1313 1,1317 1,132 1,1323 1,1326
8,5 1,1301 1,1306 1,131 1,1314 1,1318 1,1321 1,1324 1,1327
9 1,1305 1,131 1,1314 1,1318 1,1322 1,1325 1,1328 1,1331
9,5 1,1311 1,1316 1,132 1,1324 1,1328 1,1331 1,1335 1,1338
10 1,1317 1,1322 1,1327 1,1331 1,1335 1,1339 1,1342 1,1345

Table A.7: Ratio of specific heats of hydrogen-oxygen mixtures

HH P
H c 136 144 152 160 168 176 184 192
O/F HH H
1 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587
1,5 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157
2 1,2824 1,2824 1,2824 1,2824 1,2824 1,2824 1,2824 1,2825
2,5 1,2567 1,2567 1,2568 1,2568 1,2569 1,2569 1,257 1,257
3 1,2341 1,2342 1,2344 1,2346 1,2347 1,2348 1,2349 1,2351
3,5 1,2129 1,2132 1,2135 1,2138 1,214 1,2143 1,2145 1,2147
4 1,1937 1,1941 1,1945 1,1949 1,1952 1,1955 1,1959 1,1961
4,5 1,1771 1,1776 1,1781 1,1785 1,1789 1,1793 1,1796 1,18
5 1,1636 1,164 1,1645 1,1649 1,1653 1,1657 1,1661 1,1664
5,5 1,1528 1,1533 1,1537 1,1541 1,1544 1,1548 1,1551 1,1555
6 1,1446 1,145 1,1454 1,1457 1,1461 1,1464 1,1467 1,147
6,5 1,1389 1,1392 1,1395 1,1398 1,1401 1,1404 1,1406 1,1409
7 1,1353 1,1356 1,1359 1,1361 1,1363 1,1366 1,1368 1,137
7,5 1,1335 1,1337 1,134 1,1342 1,1344 1,1346 1,1348 1,135
8 1,1329 1,1331 1,1333 1,1336 1,1338 1,134 1,1342 1,1343
8,5 1,1329 1,1332 1,1334 1,1337 1,1339 1,1341 1,1343 1,1344
9 1,1334 1,1337 1,1339 1,1341 1,1343 1,1346 1,1348 1,1349
9,5 1,134 1,1343 1,1346 1,1348 1,135 1,1352 1,1354 1,1356
10 1,1348 1,1351 1,1353 1,1356 1,1358 1,136 1,1362 1,1364
 113

Table A.8: Ratio of specific heats of hydrogen-oxygen mixtures

HH P
H c 200 208 216 224 232 240 248 256 264
O/F HH H
1 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587 1,3587
1,5 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157 1,3157
2 1,2825 1,2825 1,2825 1,2825 1,2825 1,2825 1,2825 1,2825 1,2825
2,5 1,257 1,2571 1,2571 1,2571 1,2572 1,2572 1,2572 1,2573 1,2573
3 1,2352 1,2353 1,2354 1,2354 1,2355 1,2356 1,2357 1,2358 1,2358
3,5 1,2149 1,2151 1,2153 1,2155 1,2156 1,2158 1,2159 1,2161 1,2162
4 1,1964 1,1967 1,1969 1,1972 1,1974 1,1976 1,1979 1,1981 1,1982
4,5 1,1803 1,1806 1,1809 1,1812 1,1814 1,1817 1,1819 1,1822 1,1824
5 1,1667 1,1671 1,1674 1,1677 1,1679 1,1682 1,1685 1,1687 1,169
5,5 1,1558 1,1561 1,1564 1,1566 1,1569 1,1572 1,1574 1,1577 1,1579
6 1,1473 1,1475 1,1478 1,148 1,1483 1,1485 1,1487 1,1489 1,1491
6,5 1,1411 1,1413 1,1415 1,1418 1,142 1,1422 1,1423 1,1425 1,1427
7 1,1372 1,1374 1,1376 1,1377 1,1379 1,1381 1,1382 1,1384 1,1385
7,5 1,1352 1,1354 1,1355 1,1357 1,1358 1,136 1,1361 1,1363 1,1364
8 1,1345 1,1347 1,1348 1,135 1,1351 1,1353 1,1354 1,1355 1,1357
8,5 1,1346 1,1348 1,1349 1,1351 1,1352 1,1354 1,1355 1,1357 1,1358
9 1,1351 1,1353 1,1355 1,1356 1,1358 1,1359 1,1361 1,1362 1,1363
9,5 1,1358 1,136 1,1362 1,1363 1,1365 1,1366 1,1368 1,1369 1,1371
10 1,1366 1,1368 1,137 1,1371 1,1373 1,1375 1,1376 1,1378 1,1379

Table A.9: Chamber temperature of hydrogen-oxygen mixtures

HH P
H c 8 16 24 32 40 48 56 64
O/F HH H
1 977,49 977,49 977,49 977,49 977,49 977,49 977,49 977,49
1,5 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9
2 1797,3 1797,4 1797,5 1797,6 1797,6 1797,6 1797,7 1797,7
2,5 2138,5 2140,5 2141,4 2142 2142,3 2142,6 2142,8 2143
3 2430,9 2439,3 2443,2 2445,7 2447,4 2448,6 2449,6 2450,4
3,5 2668,3 2688,6 2698,7 2705 2709,5 2712,9 2715,6 2717,9
4 2853,8 2888,7 2906,9 2918,7 2927,3 2933,9 2939,3 2943,7
4,5 2995,7 3045,2 3072 3089,9 3103,1 3113,5 3122 3129,1
5 3102,2 3164,9 3199,7 3223,5 3241,4 3255,6 3267,3 3277,2
5,5 3179,6 3253,3 3295,2 3324,3 3346,4 3364,1 3378,9 3391,4
6 3233,1 3315,5 3363,1 3396,5 3422,2 3442,9 3460,2 3475,1
6,5 3267,7 3356 3407,7 3444,3 3472,6 3495,6 3514,9 3531,6
7 3287,3 3379,3 3433,5 3472,1 3502 3526,5 3547,1 3565
7,5 3295,7 3389,3 3444,7 3484,2 3514,9 3540 3561,2 3579,6
8 3295,7 3389,4 3444,9 3484,4 3515,2 3540,4 3561,7 3580,2
8,5 3289,7 3382,4 3437,2 3476,2 3506,6 3531,4 3552,4 3570,5
9 3279,2 3370,2 3423,8 3462 3491,6 3515,8 3536,2 3553,9
9,5 3265,6 3354,3 3406,5 3443,5 3472,2 3495,6 3515,4 3532,4
10 3249,6 3335,9 3386,4 3422,2 3449,9 3472,5 3491,4 3507,8
114 CEA Tables

Table A.10: Chamber temperature of hydrogen-oxygen mixtures

HH P
H c 72 80 88 96 104 112 120 128
O/F HH H
1 977,49 977,49 977,49 977,49 977,49 977,49 977,49 977,49
1,5 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9
2 1797,7 1797,7 1797,7 1797,7 1797,7 1797,7 1797,7 1797,7
2,5 2143,1 2143,3 2143,4 2143,5 2143,5 2143,6 2143,7 2143,7
3 2451,1 2451,6 2452,1 2452,6 2452,9 2453,3 2453,6 2453,9
3,5 2719,8 2721,4 2722,8 2724 2725,1 2726,1 2726,9 2727,8
4 2947,5 2950,7 2953,6 2956,1 2958,4 2960,4 2962,3 2964
4,5 3135,2 3140,5 3145,2 3149,3 3153,1 3156,5 3159,6 3162,5
5 3285,7 3293,3 3299,9 3305,9 3311,3 3316,3 3320,8 3325
5,5 3402,4 3412 3420,6 3428,4 3435,5 3441,9 3447,9 3453,4
6 3488,1 3499,7 3510 3519,4 3528 3535,9 3543,1 3549,9
6,5 3546,3 3559,3 3571,1 3581,7 3591,5 3600,5 3608,9 3616,7
7 3580,7 3594,7 3607,4 3618,9 3629,5 3639,3 3648,4 3656,9
7,5 3595,9 3610,4 3623,5 3635,5 3646,5 3656,6 3666,1 3674,9
8 3596,5 3611 3624,2 3636,2 3647,3 3657,5 3667 3675,9
8,5 3586,6 3600,9 3613,8 3625,6 3636,5 3646,5 3655,8 3664,6
9 3569,4 3583,3 3595,9 3607,3 3617,8 3627,5 3636,6 3645
9,5 3547,4 3560,8 3572,9 3583,9 3594 3603,3 3612 3620
10 3522,2 3535,1 3546,6 3557,2 3566,8 3575,7 3584 3591,7

Table A.11: Chamber temperature of hydrogen-oxygen mixtures

HH P
H c 136 144 152 160 168 176 184 192
O/F HH H
1 977,49 977,49 977,49 977,49 977,49 977,49 977,49 977,49
1,5 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9
2 1797,8 1797,8 1797,8 1797,8 1797,8 1797,8 1797,8 1797,8
2,5 2143,8 2143,8 2143,9 2143,9 2144 2144 2144 2144,1
3 2454,1 2454,4 2454,6 2454,8 2455 2455,1 2455,3 2455,4
3,5 2728,5 2729,2 2729,8 2730,4 2730,9 2731,4 2731,9 2732,4
4 2965,5 2967 2968,3 2969,5 2970,7 2971,8 2972,8 2973,8
4,5 3165,1 3167,6 3169,9 3172 3174 3175,9 3177,7 3179,4
5 3328,8 3332,5 3335,8 3339 3342 3344,8 3347,4 3349,9
5,5 3458,6 3463,4 3467,9 3472,1 3476,1 3479,9 3483,5 3486,9
6 3556,2 3562,2 3567,7 3573 3577,9 3582,6 3587,1 3591,4
6,5 3624 3630,8 3637,3 3643,4 3649,2 3654,7 3659,9 3664,9
7 3664,9 3672,4 3679,4 3686,1 3692,5 3698,5 3704,3 3709,8
7,5 3683,2 3691 3698,4 3705,4 3712,1 3718,4 3724,4 3730,2
8 3684,2 3692 3699,5 3706,5 3713,2 3719,6 3725,6 3731,4
8,5 3672,7 3680,4 3687,7 3694,6 3701,2 3707,4 3713,3 3719
9 3652,9 3660,3 3667,4 3674 3680,3 3686,3 3692,1 3697,5
9,5 3627,6 3634,7 3641,5 3647,8 3653,9 3659,6 3665,1 3670,3
10 3598,9 3605,7 3612,1 3618,2 3623,9 3629,4 3634,6 3639,6
 115

Table A.12: Chamber temperature of hydrogen-oxygen mixtures

HH P
H c 200 208 216 224 232 240 248 256 264
O/F HH H
1 977,49 977,49 977,49 977,49 977,49 977,49 977,49 977,49 977,49
1,5 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9 1411,9
2 1797,8 1797,8 1797,8 1797,8 1797,8 1797,8 1797,8 1797,8 1797,8
2,5 2144,1 2144,1 2144,1 2144,2 2144,2 2144,2 2144,2 2144,3 2144,3
3 2455,6 2455,7 2455,9 2456 2456,1 2456,2 2456,3 2456,4 2456,5
3,5 2732,8 2733,2 2733,6 2733,9 2734,3 2734,6 2734,9 2735,2 2735,5
4 2974,7 2975,6 2976,4 2977,2 2977,9 2978,6 2979,3 2979,9 2980,5
4,5 3180,9 3182,5 3183,9 3185,3 3186,6 3187,8 3189 3190,1 3191,2
5 3352,3 3354,6 3356,8 3358,8 3360,8 3362,7 3364,5 3366,3 3368
5,5 3490,1 3493,2 3496,1 3499 3501,7 3504,3 3506,8 3509,2 3511,5
6 3595,4 3599,3 3603 3606,6 3610 3613,3 3616,5 3619,6 3622,5
6,5 3669,7 3674,2 3678,6 3682,8 3686,9 3690,8 3694,5 3698,2 3701,7
7 3715,1 3720,1 3725 3729,7 3734,2 3738,5 3742,7 3746,8 3750,7
7,5 3735,7 3741,1 3746,2 3751,1 3755,8 3760,4 3764,8 3769,1 3773,2
8 3737 3742,4 3747,5 3752,5 3757,2 3761,8 3766,3 3770,6 3774,8
8,5 3724,5 3729,7 3734,7 3739,6 3744,3 3748,8 3753,1 3757,3 3761,4
9 3702,8 3707,8 3712,7 3717,3 3721,8 3726,1 3730,3 3734,3 3738,3
9,5 3675,3 3680,1 3684,7 3689,2 3693,4 3697,6 3701,5 3705,4 3709,1
10 3644,3 3648,9 3653,2 3657,5 3661,5 3665,4 3669,2 3672,8 3676,3

Table A.13: Molar mass of kerosene-oxygen mixtures

HH P
H c 8 16 24 32 40 48 56 64
O/F HH H
1 15,329 15,461 15,563 15,646 15,718 15,782 15,839 15,891
1,5 17,645 17,657 17,663 17,667 17,669 17,671 17,672 17,674
2 20,506 20,62 20,682 20,725 20,757 20,783 20,804 20,821
2,5 22,506 22,701 22,817 22,9 22,965 23,019 23,064 23,103
3 23,984 24,209 24,345 24,444 24,522 24,586 24,641 24,689
3,5 25,166 25,4 25,542 25,645 25,727 25,794 25,852 25,902
4 26,145 26,379 26,522 26,624 26,706 26,773 26,83 26,88
4,5 26,971 27,201 27,339 27,44 27,519 27,584 27,639 27,687
5 27,674 27,897 28,03 28,126 28,202 28,264 28,316 28,362
5,5 28,278 28,491 28,617 28,708 28,779 28,838 28,887 28,93
6 28,799 29 29,119 29,204 29,27 29,324 29,369 29,409
6,5 29,249 29,437 29,547 29,626 29,686 29,736 29,778 29,814
7 29,638 29,812 29,914 29,985 30,04 30,085 30,123 30,155
7,5 29,975 30,134 30,227 30,291 30,34 30,38 30,414 30,443
8 30,266 30,41 30,493 30,55 30,594 30,629 30,659 30,684
8,5 30,516 30,646 30,719 30,769 30,807 30,838 30,863 30,885
9 30,731 30,846 30,909 30,953 30,986 31,012 31,034 31,052
9,5 30,914 31,014 31,069 31,107 31,135 31,157 31,175 31,19
10 31,07 31,156 31,203 31,234 31,258 31,276 31,291 31,304
116 CEA Tables

Table A.14: Molar mass of kerosene-oxygen mixtures

HH P
H c 72 80 88 96 104 112 120 128
O/F HH H
1 15,939 15,983 16,024 16,063 16,1 16,135 16,168 16,2
1,5 17,675 17,676 17,676 17,677 17,678 17,678 17,679 17,679
2 20,837 20,85 20,862 20,873 20,882 20,891 20,899 20,906
2,5 23,138 23,169 23,197 23,222 23,246 23,267 23,288 23,307
3 24,732 24,77 24,805 24,837 24,867 24,894 24,92 24,944
3,5 25,947 25,987 26,024 26,057 26,088 26,117 26,144 26,17
4 26,924 26,964 27,001 27,034 27,065 27,094 27,12 27,146
4,5 27,73 27,769 27,804 27,836 27,866 27,893 27,919 27,943
5 28,403 28,439 28,473 28,503 28,531 28,557 28,581 28,604
5,5 28,968 29,002 29,033 29,061 29,087 29,111 29,133 29,154
6 29,444 29,475 29,504 29,529 29,553 29,575 29,595 29,614
6,5 29,845 29,874 29,899 29,923 29,944 29,964 29,982 29,999
7 30,184 30,209 30,232 30,252 30,271 30,289 30,305 30,32
7,5 30,468 30,49 30,51 30,528 30,545 30,56 30,574 30,588
8 30,706 30,725 30,742 30,758 30,773 30,786 30,798 30,809
8,5 30,904 30,921 30,935 30,949 30,961 30,972 30,982 30,992
9 31,068 31,082 31,095 31,106 31,116 31,126 31,134 31,142
9,5 31,204 31,215 31,226 31,235 31,243 31,251 31,258 31,265
10 31,315 31,324 31,333 31,34 31,347 31,354 31,359 31,365

Table A.15: Molar mass of kerosene-oxygen mixtures

HH P
H c 136 144 152 160 168 176 184 192
O/F HH H
1 16,23 16,259 16,287 16,314 16,34 16,365 16,389 16,412
1,5 17,68 17,68 17,68 17,681 17,681 17,681 17,682 17,682
2 20,913 20,92 20,926 20,931 20,936 20,941 20,946 20,95
2,5 23,324 23,341 23,357 23,372 23,386 23,4 23,413 23,425
3 24,967 24,988 25,008 25,028 25,046 25,064 25,081 25,097
3,5 26,194 26,216 26,238 26,258 26,278 26,297 26,315 26,332
4 27,169 27,192 27,213 27,233 27,253 27,271 27,289 27,306
4,5 27,966 27,987 28,008 28,027 28,046 28,063 28,08 28,096
5 28,625 28,645 28,664 28,682 28,7 28,716 28,732 28,747
5,5 29,174 29,193 29,21 29,227 29,243 29,258 29,272 29,286
6 29,632 29,649 29,665 29,68 29,695 29,708 29,721 29,734
6,5 30,015 30,03 30,044 30,058 30,071 30,083 30,094 30,106
7 30,334 30,348 30,36 30,372 30,383 30,394 30,404 30,414
7,5 30,6 30,611 30,622 30,633 30,642 30,651 30,66 30,669
8 30,82 30,83 30,839 30,848 30,856 30,864 30,871 30,878
8,5 31,001 31,009 31,017 31,024 31,031 31,038 31,044 31,05
9 31,15 31,156 31,163 31,169 31,175 31,18 31,185 31,19
9,5 31,271 31,277 31,282 31,287 31,292 31,296 31,3 31,304
10 31,37 31,374 31,378 31,382 31,386 31,39 31,393 31,396
 117

Table A.16: Molar mass of kerosene-oxygen mixtures

HH P
H c 200 208 216 224 232 240 248 256 264
O/F HH H
1 16,435 16,457 16,479 16,5 16,52 16,54 16,559 16,578 16,596
1,5 17,682 17,683 17,683 17,683 17,684 17,684 17,684 17,684 17,685
2 20,955 20,959 20,962 20,966 20,97 20,973 20,976 20,979 20,982
2,5 23,437 23,448 23,459 23,47 23,48 23,49 23,499 23,508 23,517
3 25,113 25,128 25,142 25,156 25,17 25,183 25,195 25,208 25,22
3,5 26,348 26,364 26,379 26,394 26,408 26,422 26,436 26,449 26,461
4 27,322 27,338 27,353 27,367 27,381 27,395 27,408 27,421 27,433
4,5 28,112 28,127 28,141 28,155 28,169 28,182 28,194 28,206 28,218
5 28,762 28,775 28,789 28,802 28,814 28,826 28,838 28,849 28,86
5,5 29,299 29,312 29,324 29,336 29,348 29,359 29,369 29,38 29,39
6 29,746 29,757 29,768 29,779 29,789 29,799 29,809 29,818 29,827
6,5 30,116 30,126 30,136 30,146 30,155 30,163 30,172 30,18 30,188
7 30,423 30,432 30,44 30,449 30,457 30,464 30,471 30,479 30,485
7,5 30,677 30,684 30,691 30,699 30,705 30,712 30,718 30,724 30,73
8 30,885 30,891 30,898 30,904 30,909 30,915 30,92 30,925 30,93
8,5 31,056 31,061 31,066 31,071 31,076 31,081 31,085 31,089 31,093
9 31,195 31,199 31,204 31,208 31,212 31,215 31,219 31,222 31,226
9,5 31,308 31,312 31,315 31,318 31,321 31,324 31,327 31,33 31,333
10 31,399 31,402 31,405 31,408 31,41 31,413 31,415 31,417 31,419

Table A.17: Ratio of specific heats of kerosene-oxygen mixtures

HH P
H c 8 16 24 32 40 48 56 64
O/F HH H
1 1,2724 1,2597 1,2528 1,2484 1,2451 1,2426 1,2406 1,239
1,5 1,2405 1,2464 1,2493 1,2511 1,2524 1,2533 1,2541 1,2547
2 1,1526 1,1598 1,1642 1,1673 1,1698 1,1718 1,1734 1,1749
2,5 1,1271 1,1311 1,1335 1,1352 1,1365 1,1376 1,1386 1,1394
3 1,1214 1,1247 1,1266 1,128 1,129 1,1299 1,1306 1,1312
3,5 1,1195 1,1226 1,1245 1,1257 1,1267 1,1275 1,1282 1,1288
4 1,1188 1,1219 1,1237 1,125 1,126 1,1268 1,1275 1,128
4,5 1,1187 1,1219 1,1237 1,1251 1,1261 1,1269 1,1276 1,1282
5 1,1192 1,1224 1,1244 1,1257 1,1268 1,1276 1,1284 1,129
5,5 1,1201 1,1235 1,1255 1,1269 1,1281 1,129 1,1298 1,1304
6 1,1214 1,125 1,1272 1,1287 1,1299 1,1309 1,1317 1,1324
6,5 1,1233 1,1271 1,1294 1,131 1,1323 1,1334 1,1343 1,135
7 1,1256 1,1297 1,1321 1,1339 1,1353 1,1364 1,1374 1,1382
7,5 1,1284 1,1328 1,1355 1,1374 1,1389 1,1401 1,1412 1,1421
8 1,1318 1,1365 1,1394 1,1415 1,1431 1,1444 1,1455 1,1465
8,5 1,1357 1,1409 1,1439 1,1461 1,1479 1,1493 1,1505 1,1515
9 1,1402 1,1458 1,149 1,1514 1,1532 1,1547 1,1559 1,157
9,5 1,1454 1,1512 1,1547 1,1571 1,159 1,1606 1,1618 1,1629
10 1,1511 1,1572 1,1608 1,1633 1,1653 1,1668 1,1681 1,1692
118 CEA Tables

Table A.18: Ratio of specific heats of kerosene-oxygen mixtures

HH Pc
H
72 80 88 96 104 112 120 128
O/F HH
H
1 1,2376 1,2363 1,2352 1,2342 1,2333 1,2325 1,2318 1,2311
1,5 1,2552 1,2557 1,2561 1,2564 1,2567 1,257 1,2572 1,2574
2 1,1762 1,1773 1,1783 1,1793 1,1801 1,1809 1,1816 1,1823
2,5 1,1401 1,1408 1,1414 1,1419 1,1424 1,1429 1,1433 1,1437
3 1,1317 1,1322 1,1326 1,133 1,1334 1,1337 1,134 1,1343
3,5 1,1293 1,1297 1,1301 1,1305 1,1308 1,1312 1,1314 1,1317
4 1,1286 1,129 1,1294 1,1298 1,1301 1,1304 1,1307 1,131
4,5 1,1287 1,1292 1,1296 1,13 1,1303 1,1307 1,131 1,1313
5 1,1296 1,1301 1,1305 1,1309 1,1313 1,1316 1,1319 1,1322
5,5 1,131 1,1315 1,132 1,1325 1,1329 1,1332 1,1336 1,1339
6 1,1331 1,1336 1,1341 1,1346 1,135 1,1354 1,1358 1,1362
6,5 1,1357 1,1363 1,1369 1,1374 1,1379 1,1383 1,1387 1,1391
7 1,139 1,1396 1,1402 1,1408 1,1413 1,1418 1,1422 1,1426
7,5 1,1429 1,1436 1,1442 1,1448 1,1454 1,1459 1,1463 1,1468
8 1,1473 1,1481 1,1488 1,1494 1,15 1,1505 1,151 1,1515
8,5 1,1524 1,1532 1,1539 1,1546 1,1552 1,1558 1,1563 1,1568
9 1,1579 1,1588 1,1595 1,1602 1,1608 1,1614 1,162 1,1625
9,5 1,1639 1,1648 1,1655 1,1662 1,1669 1,1674 1,168 1,1685
10 1,1702 1,171 1,1718 1,1725 1,1731 1,1737 1,1742 1,1747
 119

Table A.19: Ratio of specific heats of kerosene-oxygen mixtures

HH P
H c 136 144 152 160 168 176 184 192
O/F HH H
1 1,2304 1,2298 1,2292 1,2287 1,2282 1,2277 1,2272 1,2268
1,5 1,2576 1,2578 1,258 1,2581 1,2583 1,2584 1,2585 1,2586
2 1,1829 1,1835 1,1841 1,1846 1,1851 1,1856 1,186 1,1865
2,5 1,1441 1,1445 1,1448 1,1452 1,1455 1,1458 1,1461 1,1463
3 1,1346 1,1348 1,1351 1,1353 1,1355 1,1357 1,1359 1,1361
3,5 1,132 1,1322 1,1324 1,1326 1,1328 1,133 1,1332 1,1334
4 1,1313 1,1315 1,1317 1,1319 1,1321 1,1323 1,1325 1,1327
4,5 1,1315 1,1318 1,132 1,1322 1,1324 1,1326 1,1328 1,133
5 1,1325 1,1328 1,133 1,1333 1,1335 1,1337 1,1339 1,1341
5,5 1,1342 1,1345 1,1347 1,135 1,1352 1,1355 1,1357 1,1359
6 1,1365 1,1368 1,1371 1,1374 1,1376 1,1379 1,1381 1,1383
6,5 1,1394 1,1398 1,1401 1,1404 1,1407 1,1409 1,1412 1,1414
7 1,143 1,1434 1,1437 1,144 1,1443 1,1446 1,1449 1,1452
7,5 1,1472 1,1476 1,1479 1,1483 1,1486 1,1489 1,1492 1,1495
8 1,1519 1,1524 1,1527 1,1531 1,1535 1,1538 1,1541 1,1544
8,5 1,1572 1,1576 1,1581 1,1584 1,1588 1,1591 1,1595 1,1598
9 1,1629 1,1634 1,1638 1,1642 1,1645 1,1649 1,1652 1,1655
9,5 1,1689 1,1694 1,1698 1,1702 1,1706 1,1709 1,1712 1,1716
10 1,1752 1,1756 1,176 1,1764 1,1767 1,1771 1,1774 1,1777

Table A.20: Ratio of specific heats of kerosene-oxygen mixtures

HH P
H c 200 208 216 224 232 240 248 256 264
O/F HH H
1 1,2263 1,2259 1,2255 1,2251 1,2248 1,2244 1,2241 1,2237 1,2234
1,5 1,2588 1,2589 1,259 1,259 1,2591 1,2592 1,2593 1,2594 1,2594
2 1,1869 1,1872 1,1876 1,188 1,1883 1,1886 1,189 1,1893 1,1896
2,5 1,1466 1,1468 1,1471 1,1473 1,1475 1,1478 1,148 1,1482 1,1484
3 1,1363 1,1364 1,1366 1,1367 1,1369 1,137 1,1372 1,1373 1,1374
3,5 1,1335 1,1337 1,1338 1,134 1,1341 1,1342 1,1344 1,1345 1,1346
4 1,1328 1,133 1,1331 1,1333 1,1334 1,1336 1,1337 1,1338 1,1339
4,5 1,1332 1,1333 1,1335 1,1337 1,1338 1,1339 1,1341 1,1342 1,1343
5 1,1343 1,1345 1,1346 1,1348 1,135 1,1351 1,1353 1,1354 1,1355
5,5 1,1361 1,1363 1,1365 1,1367 1,1368 1,137 1,1372 1,1373 1,1375
6 1,1386 1,1388 1,139 1,1392 1,1394 1,1395 1,1397 1,1399 1,14
6,5 1,1417 1,1419 1,1421 1,1423 1,1425 1,1427 1,1429 1,1431 1,1433
7 1,1454 1,1457 1,1459 1,1461 1,1464 1,1466 1,1468 1,147 1,1472
7,5 1,1498 1,15 1,1503 1,1505 1,1508 1,151 1,1512 1,1514 1,1516
8 1,1547 1,155 1,1552 1,1555 1,1557 1,156 1,1562 1,1564 1,1566
8,5 1,1601 1,1604 1,1606 1,1609 1,1612 1,1614 1,1616 1,1619 1,1621
9 1,1658 1,1661 1,1664 1,1667 1,1669 1,1672 1,1674 1,1676 1,1679
9,5 1,1719 1,1721 1,1724 1,1727 1,1729 1,1732 1,1734 1,1736 1,1739
10 1,178 1,1783 1,1785 1,1788 1,179 1,1793 1,1795 1,1797 1,1799
120 CEA Tables

Table A.21: Chamber temperature of kerosene-oxygen mixtures

HH P
H c 8 16 24 32 40 48 56 64
O/F HH H
1 1451,14 1478,37 1498,59 1514,89 1528,64 1540,58 1551,16 1560,68
1,5 2479,39 2487,63 2491,48 2493,84 2495,47 2496,7 2497,66 2498,44
2 3160,42 3222,46 3256,67 3279,86 3297,17 3310,86 3322,1 3331,58
2,5 3352,22 3446,76 3502,57 3542,28 3573,1 3598,27 3619,52 3637,9
3 3377,63 3478,45 3538,77 3582,12 3616,06 3643,97 3667,69 3688,33
3,5 3357,55 3457,06 3516,67 3559,55 3593,14 3620,8 3644,31 3664,78
4 3321,78 3417,66 3474,97 3516,15 3548,36 3574,85 3597,37 3616,95
4,5 3279,4 3370,74 3425,15 3464,14 3494,58 3519,58 3540,79 3559,22
5 3233,78 3320,16 3371,38 3407,97 3436,46 3459,81 3479,59 3496,74
5,5 3186,35 3267,53 3315,4 3349,48 3375,93 3397,56 3415,83 3431,65
6 3137,72 3213,55 3257,99 3289,48 3313,86 3333,72 3350,46 3364,93
6,5 3088,15 3158,53 3199,49 3228,37 3250,63 3268,72 3283,92 3297,02
7 3037,74 3102,61 3140,05 3166,31 3186,47 3202,78 3216,46 3228,21
7,5 2986,51 3045,82 3079,76 3103,41 3121,48 3136,05 3148,22 3158,64
8 2934,43 2988,19 3018,65 3039,74 3055,76 3068,63 3079,34 3088,48
8,5 2881,5 2929,75 2956,79 2975,38 2989,43 3000,66 3009,97 3017,89
9 2827,7 2870,54 2894,28 2910,47 2922,63 2932,31 2940,3 2947,08
9,5 2773,07 2810,67 2831,24 2845,16 2855,56 2863,79 2870,56 2876,29
10 2717,68 2750,26 2767,86 2779,68 2788,45 2795,36 2801,03 2805,8

Table A.22: Chamber temperatures of kerosene-oxygen mixtures

HH P
H c 72 80 88 96 104 112 120 128
O/F HH H
1 1569,35 1577,32 1584,69 1591,57 1598,01 1604,07 1609,8 1615,23
1,5 2499,09 2499,65 2500,13 2500,55 2500,93 2501,26 2501,57 2501,84
2 3339,74 3346,88 3353,21 3358,88 3364 3368,65 3372,92 3376,85
2,5 3654,09 3668,54 3681,58 3693,47 3704,38 3714,45 3723,81 3732,55
3 3706,6 3722,99 3737,86 3751,46 3764 3775,63 3786,48 3796,64
3,5 3682,92 3699,19 3713,97 3727,49 3739,96 3751,53 3762,32 3772,44
4 3634,29 3649,84 3663,94 3676,85 3688,75 3699,78 3710,07 3719,7
4,5 3575,51 3590,11 3603,34 3615,44 3626,58 3636,9 3646,51 3655,52
5 3511,88 3525,44 3537,7 3548,9 3559,2 3568,74 3577,62 3585,92
5,5 3445,6 3458,06 3469,32 3479,58 3489,01 3497,74 3505,85 3513,42
6 3377,65 3388,99 3399,23 3408,55 3417,1 3425 3432,33 3439,17
6,5 3308,52 3318,76 3327,98 3336,35 3344,03 3351,1 3357,66 3363,78
7 3238,49 3247,63 3255,84 3263,29 3270,11 3276,38 3282,18 3287,59
7,5 3167,75 3175,81 3183,05 3189,6 3195,58 3201,07 3206,15 3210,87
8 3096,44 3103,48 3109,78 3115,47 3120,65 3125,41 3129,79 3133,86
8,5 3024,77 3030,84 3036,25 3041,13 3045,57 3049,63 3053,37 3056,84
9 2952,95 2958,11 2962,7 2966,84 2970,58 2974,01 2977,16 2980,08
9,5 2881,23 2885,56 2889,41 2892,87 2896 2898,85 2901,47 2903,89
10 2809,91 2813,5 2816,68 2819,54 2822,12 2824,47 2826,62 2828,61
 121

Table A.23: Chamber temperatures of kerosene-oxygen mixtures

HH P
H c 136 144 152 160 168 176 184 192
O/F HH H
1 1620,4 1625,32 1630,03 1634,54 1638,87 1643,04 1647,05 1650,92
1,5 2502,1 2502,33 2502,55 2502,75 2502,93 2503,11 2503,28 2503,43
2 3380,48 3383,86 3387,02 3389,97 3392,74 3395,35 3397,82 3400,16
2,5 3740,74 3748,44 3755,71 3762,6 3769,13 3775,34 3781,27 3786,93
3 3806,2 3815,22 3823,76 3831,87 3839,59 3846,96 3854,01 3860,76
3,5 3781,95 3790,94 3799,45 3807,54 3815,24 3822,59 3829,63 3836,37
4 3728,77 3737,32 3745,42 3753,12 3760,44 3767,43 3774,11 3780,52
4,5 3663,98 3671,96 3679,51 3686,68 3693,5 3700,01 3706,22 3712,18
5 3593,72 3601,07 3608,02 3614,61 3620,87 3626,85 3632,55 3638,01
5,5 3520,53 3527,22 3533,54 3539,54 3545,23 3550,64 3555,82 3560,76
6 3445,59 3451,62 3457,31 3462,7 3467,81 3472,67 3477,31 3481,74
6,5 3369,5 3374,87 3379,94 3384,73 3389,28 3393,59 3397,71 3401,63
7 3292,64 3297,37 3301,83 3306,05 3310,03 3313,82 3317,43 3320,86
7,5 3215,27 3219,39 3223,27 3226,93 3230,39 3233,68 3236,8 3239,77
8 3137,66 3141,2 3144,54 3147,67 3150,64 3153,45 3156,12 3158,65
8,5 3060,06 3063,08 3065,9 3068,56 3071,07 3073,44 3075,69 3077,83
9 2982,79 2985,31 2987,68 2989,9 2992 2993,98 2995,85 2997,63
9,5 2906,14 2908,23 2910,19 2912,02 2913,75 2915,38 2916,93 2918,39
10 2830,44 2832,16 2833,76 2835,26 2836,67 2838 2839,25 2840,45

Table A.24: Chamber temperatures of kerosene-oxygen mixtures

HH P
H c 200 208 216 224 232 240 248 256
O/F HH H
1 1654,66 1658,27 1661,78 1665,17 1668,47 1671,67 1674,78 1677,81
1,5 2503,58 2503,72 2503,85 2503,98 2504,1 2504,22 2504,34 2504,44
2 3402,37 3404,48 3406,48 3408,39 3410,22 3411,97 3413,65 3415,26
2,5 3792,35 3797,55 3802,54 3807,33 3811,95 3816,41 3820,71 3824,86
3 3867,24 3873,47 3879,47 3885,25 3890,83 3896,23 3901,45 3906,51
3,5 3842,84 3849,06 3855,05 3860,83 3866,41 3871,8 3877,03 3882,08
4 3786,66 3792,57 3798,26 3803,75 3809,04 3814,16 3819,11 3823,91
4,5 3717,89 3723,38 3728,66 3733,75 3738,66 3743,4 3747,99 3752,43
5 3643,24 3648,26 3653,09 3657,75 3662,24 3666,57 3670,76 3674,81
5,5 3565,5 3570,04 3574,41 3578,62 3582,67 3586,58 3590,36 3594,01
6 3485,98 3490,05 3493,95 3497,71 3501,33 3504,81 3508,18 3511,43
6,5 3405,38 3408,98 3412,43 3415,75 3418,94 3422,01 3424,98 3427,84
7 3324,15 3327,29 3330,3 3333,19 3335,97 3338,65 3341,23 3343,72
7,5 3242,6 3245,32 3247,91 3250,41 3252,8 3255,1 3257,32 3259,46
8 3161,07 3163,38 3165,6 3167,72 3169,75 3171,7 3173,59 3175,4
8,5 3079,87 3081,81 3083,67 3085,45 3087,16 3088,8 3090,38 3091,89
9 2999,33 3000,94 3002,49 3003,96 3005,38 3006,74 3008,04 3009,3
9,5 2919,79 2921,11 2922,38 2923,59 2924,75 2925,86 2926,93 2927,95
10 2841,58 2842,66 2843,68 2844,67 2845,6 2846,5 2847,37 2848,2
122 CEA Tables

Table A.25: Chamber temperatures of kerosene-oxygen mixtures

HH P
H c 264
O/F H
H
H
1 1680,76
1,5 2504,55
2 3416,8
2,5 3828,88
3 3911,41
3,5 3886,99
4 3828,56
4,5 3756,74
5 3678,74
5,5 3597,55
6 3514,58
6,5 3430,61
7 3346,13
7,5 3261,52
8 3177,15
8,5 3093,36
9 3010,5
9,5 2928,94
10 2849
 Appendix B

Main Results

In this appendix the main results generated with the program are tabulated. As a side
note, the maximum thrust is always presented without the exit pressure term, thus denotes
maximum thrust at design altitude. And the thrust-to-mass ratio is calculated with
included fairing mass even for the solutions that jettison the fairing before ignition

123
124 Main Results

Table B.1: Optimized launch vehicle parameters and characteristics using kerolox propellants

Velocity (m/s) 0 0 0 0 0
Altitude (m) 0 5000 10000 20000 40000
Pitch 1 0,1013 0,0792 0,084 0,2187 0,3183
Pitch 2 0,7251 0,7144 0,6867 0,6723 0,3612
Pitch 3 0,347 0,4241 0,4838 0,3022 0,4957
Pitch 4 0,5604 0,5239 0,497 0,573 0,5482
Length (m) 18,558 14,823 13,667 10,816 11,094
Slenderness 10,38 7,5495 6,686 4,7284 4,8734
Diameter (m) 1,7879 1,9635 2,0441 2,2875 2,2764
Nozzle diameter (m) 0,8804 1,0507 1,1559 1,3034 1,5026
Oxidizer to fuel ratio 2,4988 2,5023 2,5018 2,5019 2,5
Chamber Pressure (bar) 263,98 263,13 263,95 263,77 263,99
Exit Pressure (Pa) 45861 27546 20978 14089 9975,6
Initial Flight path angle (Deg) 84,814 83,553 81,791 80,401 80,083
Optimized payload Ratio 0,0208 0,0251 0,0292 0,3427 0,0429
Estimated payload Ratio 0,021 0,028 0,032 0,035 0,04
Target Altitude (m) 250000 250000 250000 250000 250000
Gravity Loss (m/s) 445,1 410,1 380,4 490,9 428,4
Drag Loss (m/s) 340,8 297,9 242,4 61,2 3,4
Maximum Dynamic Pressure (Pa) 48940 26584 14211 2513 134,4
Time max dynamic Pressure (s) 85,7 78 78 87,6 88,3
Initial Acceleration (m/s2 ) 11,7 12,05 12,33 12,5 12,5
Maximum Acceleration (m/s2 ) 50 50 50 50 50
Initial Mass (kg) 41351 38902 38431 36380 37231
Final Mass (kg) 3039,5 3040,6 3161,5 3184 3625,1
Jettisoning (s) 365 352,2 350,9 222,8 0
Fairing mass (kg) 224,6 281,2 317,1 328,1 0
Maximum Bending load (Pa rad) 8120 5958 3746,5 950 48,1
Time max Bending (s) 63,8 61 60,8 89,8 75,7
Heat shield Mass (kg) 9 8,7 8 0 0
mass flow (kg/s) 155,5 145 139,9 127,6 127
Payload Mass (kg) 859,9 978,3 1123 1246,9 1596,2
Max thrust (N) 516546 491948 479830 443686 446702
Exhaust Velocity (m/s) 3321,7 3392,8 3428,9 3477,9 3518,1
Maximum Angle of Attack (rad) 0,515 0,502 0,517 0,52 0,523
Delta V (m/s) 8652,8 8623,6 8536,4 8440,2 8158,1
Area Ratio 41,1 63,9 81,2 114,9 155,3
Mass Engine (kg) 787,67 746,85 726,7 667,16 672,1
Mass Nozzle (kg) 77,48 114,92 142,35 186,16 253,24
Mass Tank (kg) 957,84 859,3 833,43 765,83 778,83
Max Heat flux (W/m2 ) 37426 35523 32303 4260 1135
Time max heat flux (s) 212,5 215 78 152,8 99,1
T/M (N/kg) 12,492 12,646 12,486 12,196 11,998
 125

Table B.2: Optimized launch vehicle parameters and characteristics using kerolox propellants

Velocity (m/s) 300 300 300 300 300

Altitude (m) 0 5000 10000 20000 40000
Pitch 1 0,40141 0,37675 0,40895 0,42489 0,42309
Pitch 2 0,607 0,58421 0,50483 0,32181 0,42187
Pitch 3 0,2543 0,25299 0,3067 0,51772 0,4512
Pitch 4 0,59075 0,60361 0,59393 0,52871 0,53997
Length (m) 16,4992 13,527 11,7915 15,4315 10,8795
Slenderness 8,12269 6,18082 5,76523 5,99709 5,21521
Diameter (m) 2,03125 2,18855 2,04527 2,57316 2,08611
Nozzle diameter (m) 1,15298 1,25848 1,10898 1,86945 1,21478
Oxidizer to fuel ratio 2,5 2,50012 2,50011 2,50057 2,50003
Chamber Pressure (bar) 263,998 263,32 264 262,631 263,941
Exit Pressure (Pa) 21249,6 16043,4 15680,6 10209,3 12260,1
Initial Flight path angle (Deg) 64,635 68,4453 65,4663 69,19 56,3973
Optimized payload Ratio 0,03983 0,04412 0,04751 0,05431 0,0617
Estimated payload Ratio 0,04 0,045 0,05 0,055 0,06
Target Altitude (m) 250000 250000 250000 250000 250000
Gravity Loss (m/s) 703,4 678,4 630,8 613,7 415
Drag Loss (m/s) 262,1 138,4 106,3 16,5 3,9
Maximum Dynamic Pressure (Pa) 55081,5 33118 18583 3965 173,7
Time max dynamic Pressure (s) 0 0 0 0 0
Initial Acceleration (m/s2 ) 7,7 8,79 9,775 10,4 11,6
Maximum Acceleration (m/s2 ) 50 50 50 50 50
Initial Mass (kg) 46524,4 43140,1 32601,7 68070,8 31013,1
Final Mass (kg) 4056,5 3989,3 3126,7 7089,1 3477,7
Jettisoning (s) 240,4 221,9 211,4 0 0
Fairing mass (kg) 478,9 498,9 418,4 0 0
Maximum Bending load (Pa rad) 6244 3046 1934 362 50,7
Time max Bending (s) 28,3 25,8 27 27,6 51,7
Heat shield Mass (kg) 0 0 0 0 0
mass flow (kg/s) 140,7 132,6 101 200,2 98,7
Payload Mass (kg) 1853,2 1903,3 1549 3696,7 1913,4
Max thrust (N) 482411 458940 350092 703796 344754
Exhaust Velocity (m/s) 3427,5 3462,1 3465,3 3514,6 3494,5
Maximum Angle of Attack (rad) 0,274 0,289 0,3157 0,341 0,4148
Delta V (m/s) 8326,4 8202,5 8079,3 7900,2 7592
Area Ratio 80,4 102,4 104,7 151,5 129,7
Mass Engine (kg) 730,99 692,25 514,93 1104,61 506,35
Mass Nozzle (kg) 141,57 171,73 133,9 389,61 163,28
Mass Tank (kg) 983,71 889,72 664,43 1412,32 632,84
Max Heat flux (W/m2 ) 1859 2918 2632 1132,6 1134
Time max heat flux (s) 0 150 143 142 140,8
T/M (N/kg) 10,369 10,6384 10,7385 10,3392 11,1164
126 Main Results

Table B.3: Optimized launch vehicle parameters and characteristics using kerolox propellants

Velocity (m/s) 600 600 600 600 600

Altitude (m) 0 5000 10000 20000 40000
Pitch 1 0,38633 0,43518 0,28762 0,33299 0,47947
Pitch 2 0,32955 0,44634 0,59118 0,54415 0,35633
Pitch 3 0,59578 0,41728 0,44257 0,4784 0,5138
Pitch 4 0,49037 0,54802 0,51713 0,50516 0,52504
Length (m) 19,0521 18,0421 13,609 13,4834 11,2983
Slenderness 8,74287 8,12332 5,31686 4,81443 6,26334
Diameter (m) 2,17916 2,22102 2,5596 2,80063 1,80387
Nozzle diameter (m) 1,51719 1,61023 1,51041 1,72514 1,22024
Oxidizer to fuel ratio 2,50588 2,50382 2,50011 2,50623 2,49999
Chamber Pressure (bar) 262,78 263,262 263,58 263,25 264
Exit Pressure (Pa) 14294,7 12074,8 12654,1 11461,6 8808,88
Initial Flight path angle (Deg) 68,8358 58,9612 66,0217 58,1154 45,923
Optimized payload Ratio 0,05639 0,06093 0,06543 0,07044 0,07806
Estimated payload Ratio 0,055 0,065 0,07 0,075 0,08
Target Altitude (m) 250000 250000 250000 250000 250000
Gravity Loss (m/s) 746,5 646,3 654,5 539,6 387,3
Drag Loss (m/s) 97,3 98,1 42,4 14,6 3,5
Maximum Dynamic Pressure (Pa) 219979 132358 74218 15851,1 694
Time max dynamic Pressure (s) 0 0 0 0 0
Initial Acceleration (m/s2 ) 2,1 5,2 6,2 9,22 11
Maximum Acceleration (m/s2 ) 50 50 50 50 50
Initial Mass (kg) 62477,3 61267,4 58396,9 68437 24822
Final Mass (kg) 6470,1 6654,7 6462,2 8007,7 3205
Jettisoning (s) 175,4 180,6 77,9 52,8 0
Fairing mass (kg) 880,8 1018,2 1043,8 1309 0
Maximum Bending load (Pa rad) 3091 4319 585 343,5 53
Time max Bending (s) 15,5 12,5 17,9 14,8 23,3
Heat shield Mass (kg) 0,4 0,3 0 0 0
mass flow (kg/s) 174,8 171 156,6 187,9 75,4
Payload Mass (kg) 3522,9 3732,8 3821,1 4820,4 1937,5
Max thrust (N) 607596 597900 546555 657957 266435
Exhaust Velocity (m/s) 3475,2 3495,6 3490,6 3501,4 3531,9
Maximum Angle of Attack (rad) 0,211 0,302 0,2596 0,293 0,3515
Delta V (m/s) 7831,2 7701,3 7620,8 7444,6 7157
Area Ratio 113,1 131,2 126 137,3 173
Mass Engine (kg) 940,68 924,3 837,85 1026,22 382,07
Mass Nozzle (kg) 251,03 286,52 251,55 329,94 168,35
Mass Tank (kg) 1324,7 1284,92 1157,91 1371,11 501,15
Max Heat flux (W/m2 ) 14325 11020,5 7688 3401 963
Time max heat flux (s) 0 0 0 0 86,5
T/M (N/kg) 9,72507 9,75886 9,35932 9,61405 10,7338
 127

Table B.4: Optimized launch vehicle parameters and characteristics using kerolox propellants

Velocity (m/s) 1000 1000 1000 1000 1000

Altitude (m) 0 5000 10000 20000 40000
Pitch 1 0,35945 0,34764 0,38198 0,35717 0,45402
Pitch 2 0,50485 0,51959 0,42348 0,59188 0,46149
Pitch 3 0,49209 0,49963 0,57817 0,41108 0,46954
Pitch 4 0,50727 0,50282 0,48242 0,52992 0,52141
Length (m) 19,0725 18,1594 15,7677 14,3292 10,3185
Slenderness 8,47571 11,2901 6,57739 6,3481 5,25643
Diameter (m) 2,25025 1,60844 2,39726 2,25725 1,96302
Nozzle diameter (m) 1,64924 1,21861 1,64386 1,57863 1,22087
Oxidizer to fuel ratio 2,5092 2,5 2,50713 2,50594 2,49958
Chamber Pressure (bar) 263,944 263,598 260,514 263,214 260,426
Exit Pressure (Pa) 11334,3 9758,82 10770,4 8856,84 8057,48
Initial Flight path angle (Deg) 55,6807 54,3499 49,9696 47,507 39,6175
Optimized payload Ratio 0,07837 0,08262 0,08655 0,09138 0,09645
Estimated payload Ratio 0,075 0,085 0,09 0,095 0,1
Target Altitude (m) 250000 250000 250000 250000 250000
Gravity Loss (m/s) 629,3 593,5 529,9 482,7 361,3
Drag Loss (m/s) 133,2 77,6 52,2 13,7 2,2
Maximum Dynamic Pressure (Pa) 610791 367453 206199 44024,7 1926,9
Time max dynamic Pressure (s) 0 0 0 0 0
Initial Acceleration (m/s2 ) -5,84 -0,25 3,8 8,05 9,69
Maximum Acceleration (m/s2 ) 50 50 50 50 50
Initial Mass (kg) 66754,8 33237,8 61363 49307,2 26261,1
Final Mass (kg) 8258,8 4227 8123 6769,2 3769,8
Jettisoning (s) 130,1 119,1 112,6 89,7 58,1
Fairing mass (kg) 1276,8 718,7 1403 1188,5 665,485
Maximum Bending load (Pa rad) 3915 2166 1874 415 52,2
Time max Bending (s) 11,4 8,5 8,4 9,7 14,1
Heat shield Mass (kg) 7,5 5,4 3,7 0,9 0
mass flow (kg/s) 170,2 82 161,7 126,7 69,9
Payload Mass (kg) 5231,4 2723,1 5310,7 4505,9 2532,8
Max thrust (N) 596123 288571 567022 447431 247528
Exhaust Velocity (m/s) 3502,8 3520,3 3506,8 3530,4 3539,4
Maximum Angle of Attack (rad) 0,225 0,229 0,1975 0,2411 0,3
Delta V (m/s) 7252,3 7182,5 7010 6924 6779,4
Area Ratio 139 158 143,6 171,8 184,8
Mass Engine (kg) 921,31 416,91 872,17 673,27 352,56
Mass Nozzle (kg) 302,51 166,53 298,09 280,93 167,44
Mass Tank (kg) 1372,15 680,68 1231,88 969,54 509,21
Max Heat flux (W/m2 ) 65221,4 59904,1 36790,4 17534,9 3935,9
Time max heat flux (s) 0 0 0 0 0
T/M (N/kg) 8,93004 8,68201 9,24045 9,07435 9,42565
128 Main Results

Table B.5: Optimized launch vehicle parameters and characteristics using kerolox propellants

Velocity (m/s) 1500 1500 1500 1500 1500

Altitude (m) 0 5000 10000 20000 40000
Pitch 1 0,30391 0,30399 0,39171 0,37522 0,41225
Pitch 2 0,66193 0,69279 0,45214 0,57429 0,49043
Pitch 3 0,43878 0,4096 0,5638 0,46067 0,5573
Pitch 4 0,50729 0,51477 0,48501 0,50979 0,47704
Length (m) 44,1592 30,6699 16,5618 12,9515 11,203
Slenderness 14,6486 4,4899 5,61482 5,29576 5,65653
Diameter (m) 3,01456 6,83086 2,94965 2,44564 1,98054
Nozzle diameter (m) 3,00448 6,82098 2,13671 1,7007 1,29488
Oxidizer to fuel ratio 2,5036 2,5046 2,50368 2,50308 2,50035
Chamber Pressure (bar) 263,515 262,317 263,994 263,122 260,322
Exit Pressure (Pa) 13105,3 7556,54 8579,93 6781,5 6899,27
Initial Flight path angle (Deg) 47,0941 44,0082 42,2663 39,0715 34,0136
Optimized payload Ratio 0,09871 0,10537 0,11115 0,11828 0,12363
Estimated payload Ratio 0,1 0,105 0,11 0,115 0,12
Target Altitude (m) 250000 250000 250000 250000 250000
Gravity Loss (m/s) 537,7 485,2 469,6 413,2 317,4
Drag Loss (m/s) 156,1 112,7 107,2 34,4 2,4
Maximum Dynamic Pressure (Pa) 1374140 826421 463360 99003 4334,1
Time max dynamic Pressure (s) 0 0 0 0 0
Initial Acceleration (m/s2 ) -14,5 -8,67 -8,5 3,39 8,45
Maximum Acceleration (m/s2 ) 50 50 50 50 50
Initial Mass (kg) 289399 923909 96002,9 51255,9 29391,3
Final Mass (kg) 41854,1 141590 14897,5 8316,7 4937,5
Jettisoning (s) 99,6 87,5 92 83,6 61,9
Fairing mass (kg) 7333,7 24552,4 2669,3 8316,7 889,3
Maximum Bending load (Pa rad) 1703 586 1916 419 41
Time max Bending (s) 4,2 3,5 6,6 7,5 10,2
Heat shield Mass (kg) 91,9 61,8 17,9 5,9 0,4
mass flow (kg/s) 637,9 2069,7 226,2 117,5 69
Payload Mass (kg) 28567,2 97356,5 10671,1 6062,7 3633,7
Max thrust (N) 2223760 7341460 799493 418323 245475
Exhaust Velocity (m/s) 3486,2 3547,2 3534,5 3559,3 3556,11
Maximum Angle of Attack (rad) 0,263 0,2535 0,174 0,22 0,267
Delta V (m/s) 6651,4 6557,9 6485,7 6367,9 6234,25
Area Ratio 122,2 196,7 177,1 216,7 211,5
Mass Engine (kg) 3884,53 14331,2 1269,84 625,56 349,31
Mass Nozzle (kg) 992,94 5279,69 516,88 331,24 190,06
Mass Tank (kg) 7025,33 20734,5 1880,06 967,33 552,5
Max Heat flux (W/m2 ) 190152 98044 111726 56810 13219
Time max heat flux (s) 0 0 0 0 0
T/M (N/kg) 7,68406 7,94609 8,3278 8,16146 8,35196
 129

Table B.6: Optimized launch vehicle parameters and characteristics using kerolox propellants

Velocity (m/s) 2000 2000 2000 2000 2000

Altitude (m) 0 5000 10000 20000 40000
Pitch 1 0,3255 0,33417 0,36363 0,4223 0,34945
Pitch 2 0,63251 0,68485 0,64775 0,47284 0,7104
Pitch 3 0,46093 0,39942 0,41382 0,54445 0,37438
Pitch 4 0,5033 0,52014 0,51836 0,48891 0,52691
Length (m) 45,8342 39,9775 37,6945 21,8635 13,3736
Slenderness 16,9713 13,8723 12,9079 5,32786 5,49624
Diameter (m) 2,70068 2,88182 2,92026 4,10361 2,43322
Nozzle diameter (m) 2,66964 2,88159 2,91504 3,72837 1,84071
Oxidizer to fuel ratio 2,50802 2,5 2,50074 2,50371 2,50078
Chamber Pressure (bar) 263,725 264 263,801 263,167 262,892
Exit Pressure (Pa) 13104 10508,9 10108,6 6380,98 5663,47
Initial Flight path angle (Deg) 39,8954 37,4962 34,6504 31,692 28,0966
Optimized payload Ratio 0,12255 0,13022 0,13626 0,14485 0,15115
Estimated payload Ratio 0,125 0,13 0,135 0,14 0,19
Target Altitude (m) 250000 250000 250000 250000 250000
Gravity Loss (m/s) 464,5 432,2 391,6 337,1 251
Drag Loss (m/s) 237,9 167,9 101,1 58 7,3
Maximum Dynamic Pressure (Pa) 2440970 1468370 824127 175941 7704
Time max dynamic Pressure (s) 0 0 0 0 0
Initial Acceleration (m/s2 ) -30,6 -20,25 -10 -1,27 7,4
Maximum Acceleration (m/s2 ) 44,9 42,7 42,2 42,5 41,7
Initial Mass (kg) 243845 240024 231939 245677 53696,3
Final Mass (kg) 40651,7 41685 41783,5 46611,6 10468,8
Jettisoning (s) 95,3 92,2 92,7 82,6 66,7
Fairing mass (kg) 7675,5 7847,6 7865 8628,6 2527,7
Maximum Bending load (Pa rad) 4174 1475 768 718,4 9,24
Time max Bending (s) 4,7 4 9 6,3 17,6
Heat shield Mass (kg) 246,7 178,2 134,6 40 11,8
mass flow (kg/s) 503,5 487,9 483,2 536,6 118,3
Payload Mass (kg) 29884,3 31255,1 31604 35585,7 8116,2
Max thrust (N) 1755290 1713450 1699020 1913420 423276
Exhaust Velocity (m/s) 3485,9 3512,2 3516,4 3565,6 3578
Maximum Angle of Attack (rad) 0,239 0,2365 0,2255 0,1908 0,229
Delta V (m/s) 6133,5 6031,7 5905,7 5799,2 5677,3
Area Ratio 122,4 148,4 153,4 228,6 253,3
Mass Engine (kg) 2999,36 2921,36 2894,45 3296,02 633,62
Mass Nozzle (kg) 784,42 928,2 951,73 1597,83 391,82
Mass Tank (kg) 5840,12 5462,99 5230,42 4928,69 989,82
Max Heat flux (W/m2 ) 475633 357501 266343 103914 28263
Time max heat flux (s) 0 0 0 0 0
T/M (N/kg) 7,19838 7,13866 7,32529 7,78836 7,88278
130 Main Results

Table B.7: Optimized launch vehicle parameters and characteristics using hydrolox propellants

Velocity (m/s) 0 0 0 0 0
Altitude (m) 0 5000 10000 20000 40000
Pitch 1 0,07812 0,32335 0,06294 0,26318 0,18037
Pitch 2 0,97471 0,5423 0,99274 0,6158 0,59087
Pitch 3 0,162 0,42165 0,16668 0,29534 0,40533
Pitch 4 0,58484 0,5356 0,58168 0,59142 0,55579
Length (m) 20,431 17,0035 16,8899 15,3777 12,894
Slenderness 11,5255 10,4661 10,4019 7,35343 5,95251
Diameter (m) 1,77268 1,62462 1,62374 2,09122 2,16615
Nozzle Diameter (m) 0,56763 0,60583 0,72687 1,05332 0,93048
Oxidizer to fuel 6,00637 6,07103 6,03563 6 6
Chamber Pressure (bar) 263,871 263,945 263,758 264 263,961
Exit Pressure (Pa) 47986,6 27294,7 17255,4 10705,4 12573,6
Initial flight path angle 0,95786 0,92411 0,90566 0,88745 0,94210
Optimized payload Ratio 0,03086 0,07408 0,04632 0,05341 0,06527
Estimated payload Ratio 0,035 0,077 0,041 0,055 0,055
Target Altitude (m) 250000 250000 250000 250000 250000
Gravity Loss (m/s) 576,1 714 500,8 504,6 429,1
Drag Loss (m/s) 428,9 176,7 275,9 98,5 5,1
Maximum Dynamic Pressure (Pa) 36112 219780 11985 2461 138
Time of dynamic pressure (s) 81,8 0 66,5 64,7 76,7
Initial Acceleration (m/s2 ) 12,15 -0,8 13,3 13,5 13,9
Maximum Acceleration (m/s2 ) 50 50 50 50 50
Initial Mass (kg) 17237,5 17300 11978,1 17610,4 15465,1
Final Mass (kg) 2131,2 2842,5 1708,6 2659,6 2513,4
Jettisoning (s) 270,5 196,5 253,7 213,7 0
Fairing mass (kg) 155,5 339,5 126,3 248,3 0
Maximum Bending load (Pa rad) 5042 2500 1644 850 41
Time max Bending (s) 62,5 89,6 49 56 66,8
Heat shield Mass (kg) 0,6 0,3 0,2 0 0
mass flow (kg/s) 52,7 43 37 52,1 46,5
Payload Mass (kg) 531,9 1281,6 554,9 940,6 1009,4
Max thrust (N) 222945 190011 162636 233168 207099
Exhaust Velocity (m/s) 4228,2 4419 4401,1 4476,4 4452,8
Maximum Angle of Attack (rad) 0,523 0,2026 0,513 0,524 0,357
Delta V (m/s) 8800,4 7893,3 8524,2 8398,2 8027,3
Area Ratio 39,5 125,1 96,1 145,6 126,6
Mass Engine (kg) 302,38 204,62 204,75 319,54 275,99
Mass Nozzle (kg) 32,11 38,35 57,07 124,02 95,68
Mass Tank (kg) 1039,22 722,15 717,21 1038,7 918,32
Max Heat flux (W/m2 ) 11502 17104,8 9349 3432 1135
Time max Heat flux (s) 174,6 0 156,4 142,5 135,9
T/W (N/kg) 12,9337 10,9833 13,5778 13,2404 13,3914
 131

Table B.8: Optimized launch vehicle parameters and characteristics using hydrolox propellants

Velocity (m/s) 300 300 300 300 300

Altitude (m) 0 5000 10000 20000 40000
Pitch 1 0,39105 0,39673 0,44301 0,16242 0,32132
Pitch 2 0,47397 0,57324 0,58853 0,81901 0,66486
Pitch 3 0,42892 0,32393 0,26956 0,24882 0,3023
Pitch 4 0,53736 0,56674 0,58656 0,57511 0,56941
Length (m) 23,9634 22,4141 21,9255 15,8012 13,3072
Slenderness 15,4058 14,1052 13,7668 7,39214 5,52631
Diameter (m) 1,55548 1,58907 1,59263 2,13757 2,40797
Nozzle Diameter (m) 0,74068 0,84203 0,89999 1,08414 1,05158
Oxidizer to fuel 6,00284 6,0029 5,99981 6,03337 6,10186
Chamber Pressure (bar) 263,715 263,809 263,795 261,293 262,941
Exit Pressure (Pa) 18914,1 12853,3 10812,7 9072,29 11141,4
Initial flight path angle 0,77484 0,71766 0,62157 0,88618 0,65645
Optimized payload Ratio 0,05572 0,06284 0,06805 0,07886 0,08787
Estimated payload Ratio 0,055 0,06 0,065 0,07 0,075
Target Altitude (m) 250000 250000 250000 250000 250000
Gravity Loss (m/s) 721,5 666,5 585,15 643,8 405,5
Drag Loss (m/s) 278,9 279,9 295,5 17,2 6,9
Maximum Dynamic Pressure (Pa) 55067,3 33122 18596 3966 174
Time of dynamic pressure (s) 0 0 0 0 0
Initial Acceleration (m/s2 ) 7,5 8,45 9,82 11,3 12,5
Maximum Acceleration (m/s2 ) 50 50 50 50 50
Initial Mass (kg) 15867,2 15445,2 15192,5 18941 19753
Final Mass (kg) 2334,2 2380,9 2437 3238,4 3582,6
Jettisoning (s) 240,6 236,2 233,4 0 0
Fairing mass (kg) 223,7 237,2 252,5 0 0
Maximum Bending load (Pa rad) 3973 3332 2882,5 115 47,2
Time max Bending (s) 23 24,2 24,3 42,7 51,1
Heat shield Mass (kg) 0 0 0 0 0
mass flow (kg/s) 41,4 38,8 38,3 47,9 53,5
Payload Mass (kg) 884,1 970,6 1033,9 1493,7 1735,7
Max thrust (N) 181934 172583 171576 215405 238836
Exhaust Velocity (m/s) 4389,7 4449,2 4474,8 4495 4460,6
Maximum Angle of Attack (rad) 0,201 0,2583 0,303 0,232 0,3644
Delta V (m/s) 8350,9 8250,3 8113,9 7858,3 7529
Area Ratio 88,7 124,1 144,3 166,9 140,6
Mass Engine (kg) 235,17 220,35 218,79 289,77 329,16
Mass Nozzle (kg) 58,89 78,26 90,35 131,56 122,72
Mass Tank (kg) 953,81 915,2 898,43 1090,44 1147,9
Max Heat flux (W/m2 ) 4124 3520,5 3802 1113 1134
Time max Heat flux (s) 153,2 149,8 141,5 129,6 133,4
T/W (N/kg) 11,466 11,1739 11,2935 11,3724 12,0911
132 Main Results

Table B.9: Optimized launch vehicle parameters and characteristics using hydrolox propellants

Velocity (m/s) 600 600 600 600 600

Altitude (m) 0 5000 10000 20000 40000
Pitch 1 0,32335 0,33016 0,37681 0,26812 0,41944
Pitch 2 0,5423 0,44484 0,3587 0,62903 0,36403
Pitch 3 0,42165 0,55244 0,62475 0,45641 0,61293
Pitch 4 0,5356 0,49294 0,47275 0,50601 0,47329
Length (m) 26,105 22,5873 22,6798 17,1344 15,9576
Slenderness 17,1263 14,1067 10,0785 8,86207 8,38882
Diameter (m) 1,52426 1,60117 2,25032 1,93345 1,90224
Nozzle Diameter (m) 0,88808 0,90924 1,22481 1,04277 0,91027
Oxidizer to fuel 6,34116 6,00449 6,0005 6,09309 6,00419
Chamber Pressure (bar) 263,937 263,607 262,097 261,069 260,032
Exit Pressure (Pa) 12899,4 10357,7 10073,2 7611,77 11315,1
Initial flight path angle (deg) 0,77171 0,76686 0,75405 0,71724 0,54241
Optimized payload Ratio 0,07408 0,0823 0,08834 0,09617 0,10833
Estimated payload Ratio 0,077 0,085 0,1 0,1 0,11
Target Altitude (m) 250000 250000 250000 250000 250000
Gravity Loss (m/s) 714 686,8 670,1 575,6 386,9
Drag Loss (m/s) 176,7 110,5 61,8 19,2 4,5
Maximum Dynamic Pressure (Pa) 219780 132226 74204 15849 695
Time of dynamic pressure (s) 0 0 0 0 0
Initial Acceleration (m/s2 ) -0,8 2,73 5,51495 9,1 12,2
Maximum Acceleration (m/s2 ) 50 50 50 50 50
Initial Mass (kg) 17300 15896,6 30892,6 17238,9 15446
Final Mass (kg) 2842,5 2741,9 5402 3170,7 3087,3
Jettisoning (s) 196,5 171,5 82,6 57 0
Fairing mass (kg) 339,5 343,7 782,9 436,9 0
Maximum Bending load (Pa rad) 2500 1448 1080 171 44
Time max Bending (s) 89,6 12,3 12 17 21
Heat shield Mass (kg) 0,3 0,2 0 0 0
Mass flow (kg/s) 43 37,7 66,9 38,2 40,7
Payload Mass (kg) 1281,6 1308,3 2729 1657,9 1673,3
Max thrust (N) 190011 169102 299750 172432 181515
Exhaust Velocity (m/s) 4419 4480,4 4483,7 4513,8 4465,1
Maximum Angle of Attack (rad) 0,2026 0,202 0,2593 0,268 0,278
Delta V (m/s) 7893,3 7776,1 7703,4 7527 7063,2
Area Ratio 125,1 149,7 152,6 194,7 137
Mass Engine (kg) 248,17 214,89 435,76 220,09 234,52
Mass Nozzle (kg) 86,71 92,43 167,18 122,85 91
Mass Tank (kg) 1013,48 929,37 1752,92 966,29 888,81
Max Heat flux (W/m2 ) 17104,8 12960,2 8197,5 4093 984
Time max Heat flux (s) 0 0 0 0 72
T/W (N/kg) 10,9833 10,6376 9,70297 10,0025 11,7516
 133

Table B.10: Optimized launch vehicle parameters and characteristics using hydrolox propellants

Velocity (m/s) 1000 1000 1000 1000 1000

Altitude (m) 0 5000 10000 20000 40000
Pitch 1 0,23955 0,23968 0,28056 0,33391 0,31838
Pitch 2 0,64456 0,68503 0,71886 0,56343 0,60287
Pitch 3 0,47257 0,44593 0,36922 0,5122 0,50763
Pitch 4 0,49862 0,50291 0,52882 0,48932 0,48299
Length (m) 24,834 22,314 21,7038 19,4735 16,2907
Slenderness 15,5514 9,93228 10,4793 8,40724 8,47795
Diameter (m) 1,59689 2,24661 2,07111 2,31627 1,92154
Nozzle Diameter (m) 0,84521 1,28746 1,20377 1,55823 0,99838
Oxidizer to fuel 6,02932 5,97869 6,04586 6,03333 6,33393
Chamber Pressure (bar) 263,069 263,609 262,369 230,059 258,877
Exit Pressure (Pa) 12465,1 8944,39 8005,56 5183,71 8834,2
Initial flight path angle (deg) 0,71886 0,65456 0,60851 0,54487 0,47909
Optimized payload Ratio 0,09854 0,10687 0,11327 0,12032 0,12638
Estimated payload Ratio 0,1 0,115 0,125 0,13 0,135
Target Altitude (m) 250000 250000 250000 250000 250000
Gravity Loss (m/s) 659,7 583,4 551,9 462,1 351,5
Drag Loss (m/s) 213,1 145,6 81,9 24,3 2
Maximum Dynamic Pressure (Pa) 609491 366899 206023 44022 1927
Time of dynamic pressure (s) 0 0 0 0 0
Initial Acceleration (m/s2 ) -15,9 -7,6 -0,26 7,6 10,9
Maximum Acceleration (m/s2 ) 50 50 49,2 48,3 50
Initial Mass (kg) 17503,8 30368,4 25381,1 28114 16649,6
Final Mass (kg) 3235,7 5912,3 5057,8 5891,2 3557,5
Jettisoning (s) 136,1 124,6 111,3 96,7 58,3
Fairing mass (kg) 443,6 881,7 798,9 919,2 564,6
Maximum Bending load (Pa rad) 1843 710 494 339 1927
Time max Bending (s) 4,3 4 14,3 9,8 23
Heat shield Mass (kg) 3,6 3,2 2,4 0,7 14,6
mass flow (kg/s) 38,1 67 53,2 60,4 39,4
Payload Mass (kg) 1724,9 3245,5 2875 3382,7 2104,1
Max thrust (N) 169385 301628 240073 274940 176271
Exhaust Velocity (m/s) 4450,8 4502,3 4512 4554,6 4471,1
Maximum Angle of Attack (rad) 0,215 0,214 0,255 0,243 0,224
Delta V (m/s) 7399,4 7234,9 7133,8 6966,6 6746,2
Area Ratio 127,3 169,9 186,8 243,9 171,1
Mass Engine (kg) 215,28 439,14 331,24 391,69 226,2
Mass Nozzle (kg) 78,78 187,07 163,93 252,07 110,5
Mass Tank (kg) 1014,26 1723,15 1418,82 1565,33 915,98
Max Heat flux (W/m2 ) 77175,5 50572 39530,5 17308,3 3979,3
Time max Heat flux (s) 0 0 0 0 0
T/W (N/kg) 9,67704 9,9323 9,45873 9,77947 10,5871
134 Main Results

Table B.11: Optimized launch vehicle parameters and characteristics using hydrolox propellants

Velocity (m/s) 1500 1500 1500 1500 1500

Altitude (m) 0 5000 10000 20000 40000
Pitch 1 0,26315 0,4355 0,31794 0,45157 0,38607
Pitch 2 0,7496 0,37634 0,59079 0,31079 0,60365
Pitch 3 0,37443 0,61196 0,50277 0,68332 0,43963
Pitch 4 0,52318 0,47439 0,49226 0,4535 0,51314
Length (m) 61,2933 59,409 22,1837 20,5464 18,2075
Slenderness 19,7046 18,9321 7,54095 9,76108 8,42877
Diameter (m) 3,11061 3,13801 2,94177 2,10493 2,16016
Nozzle Diameter (m) 2,86134 2,69587 1,79433 1,22462 1,24882
Oxidizer to fuel 6,38482 6,05353 5,99883 6,05638 6,01475
Chamber Pressure (bar) 262,489 262,701 259,439 262,669 263,257
Exit Pressure (Pa) 8524,06 9587,98 6494,75 7191,78 6137,07
Initial flight path angle (deg) 0,59427 0,50780 0,51512 0,43410 0,39219
Optimized payload Ratio 0,11903 0,12749 0,13879 0,15009 0,1581
Estimated payload Ratio 0,12 0,14 0,15 0,155 0,16
Target Altitude (m) 250000 250000 250000 250000 250000
Gravity Loss (m/s) 593,7 504 478,1 393,9 319,3
Drag Loss (m/s) 250,1 170 175,8 57 3,5
Maximum Dynamic Pressure (Pa) 1371020 825636 462483 98984 4334,5
Time of dynamic pressure (s) 0 0 0 0 0
Initial Acceleration (m/s2 ) -31 -15,8 -22,5 1,7 9,12
Maximum Acceleration (m/s2 ) 40,4 39,8 40,8 40,4 38,7
Initial Mass (kg) 172799 166164 51029,7 24892,4 22919,8
Final Mass (kg) 36154,9 36384,3 11434,7 5848,5 5584,9
Jettisoning (s) 90,9 98,8 90,4 92,7 62,5
Fairing mass (kg) 5228,2 5835,9 1915,5 964,3 915,4
Maximum Bending load (Pa rad) 5150 6294 264,2 867 36,7
Time max Bending (s) 5 6,4 13,8 6,8 11
Heat shield Mass (kg) 51,2 50,2 11,2 5,2 0,3
Mass flow (kg/s) 314,5 310,5 99,1 50,3 45,9
Payload Mass (kg) 20568 21184,9 7082,3 3736,1 3623,6
Max thrust (N) 1407250 1393060 450163 227735 208672
Exhaust Velocity (m/s) 4474,6 4486,3 4542,5 4525,9 4551,1
Maximum Angle of Attack (rad) 0,275 0,2384 0,2341 0,204 0,246
Delta V (m/s) 6862,3 6653,6 6620,6 6376,4 6240,4
Area Ratio 179 159,9 221,5 205,4 235,9
Mass Engine (kg) 2943,85 2907,19 720,2 310,44 278,59
Mass Nozzle (kg) 920,66 813,93 365,43 170,95 179,79
Mass Tank (kg) 10518,8 10292,2 2811,9 1370,59 1258,01
Max Heat flux (W/m2 ) 186557 144448 111558 61217,7 12659
Time max Heat flux (s) 0 0 0 0 0
T/W (N/kg) 8,14385 8,38365 8,82159 9,14878 9,10444
 135

Table B.12: Optimized launch vehicle parameters and characteristics using hydrolox propellants

Velocity (m/s) 2000 2000 2000 2000 2000

Altitude (m) 0 5000 10000 20000 40000
Pitch 1 0,2675 0,36671 0,32329 0,30479 0,29821
Pitch 2 0,72366 0,52829 0,69758 0,73082 0,80278
Pitch 3 0,41177 0,5273 0,4058 0,3908 0,32293
Pitch 4 0,5131 0,49081 0,51628 0,51882 0,53671
Length (m) 67,5273 57,5372 52,6081 40,9385 19,5537
Slenderness 19,7418 19,9974 18,8752 15,7936 8,20976
Diameter (m) 3,42053 2,87723 2,78716 2,59209 2,38177
Nozzle Diameter (m) 3,40298 2,81383 2,55416 2,47068 1,52727
Oxidizer to fuel 6,49743 6,37116 6,37712 6,01009 5,99978
Chamber Pressure (bar) 248,023 263,836 247,8 232,32 263,95
Exit Pressure (Pa) 7381,21 6363,49 6681,1 4502,45 4688,85
Initial flight path angle (deg) 0,52247 0,46303 0,43010 0,39687 0,35130
Optimized payload Ratio 0,14426 0,15582 0,16523 0,17767 0,19006
Estimated payload Ratio 0,15 0,16 0,17 0,18 0,2
Target Altitude (m) 250000 250000 250000 250000 250000
Gravity Loss (m/s) 536,1 466,7 420 364,6 290,2
Drag Loss (m/s) 341 257,6 161,6 67,3 10,3
Maximum Dynamic Pressure (Pa) 2434590 1465530 823028 175911 7703,9
Time of dynamic pressure (s) 0 0 0 0 0
Initial Acceleration (m/s2 ) -56,1 -39,3 -23 -3,5 7,4
Maximum Acceleration (m/s2 ) 33,2 32,7 31,4 31,1 31,5
Initial Mass (kg) 233855 139794 119739 77701,4 30019,1
Final Mass (kg) 54268,5 34017,5 29960,4 20574,4 8241,1
Jettisoning (s) 81,2 86,4 85,7 77,6 58,7
Fairing mass (kg) 8778 5583,2 5068,4 3473,9 1413,1
Maximum Bending load (Pa rad) 11228 1005,3 1084 389 22
Time max Bending (s) 4,8 8,9 3,9 4,8 6
Heat shield Mass (kg) 158,5 111 82,6 31 2,3
Mass flow (kg/s) 387,8 238 201,6 135,1 54,7
Payload Mass (kg) 33735,2 21782,7 19785 13805,1 5705,3
Max thrust (N) 1737760 1074770 907988 618107 250971
Exhaust Velocity (m/s) 4481,6 4516,8 4503 4576,3 4587,4
Maximum Angle of Attack (rad) 0,2725 0,2575 0,261 0,2365 0,222
Delta V (m/s) 6375,1 6199,4 6044 5871,7 5709
Area Ratio 194,4 232 211 277,8 298,6
Mass Engine (kg) 3820,31 2110,16 1713,4 1065,48 349,96
Mass Nozzle (kg) 1249,56 910,52 708,5 644,02 273,65
Mass Tank (kg) 13995,9 8233,03 6902,09 4439,37 1618,63
Max Heat flux (W/m2 ) 420977 356748 272084 130696 28565,7
Time max Heat flux (s) 0 0 0 0 0
T/W (N/kg) 7,43093 7,68824 7,58306 9,10444 8,36038
136 Main Results
 Appendix C

Figures

This appendix contains all figures that were deemed to add information to the study
discussed in the main text. The first three figures contain sensitivity indices. The other
figures displayed contain information about all launch trajectories. Every page contains
two plots, the upper contains the hydrolox cases while the bottom one contains the kerolox
cases. Various colours are used to display the cases, which only has as purpose to help
the reader follow a case of interest.

Figure C.1: Mean of the first-order standardized effect of the launch-vehicle parameters on the vari-
ance of the model with indications of the maximum and minimum values

137
138 Figures

Figure C.2: Total standardized effect of parameters on the variance of the model with various hy-
percube sample sizes.

Figure C.3: Total effect sensitivity indices of exit pressure parameter at various values of optimized
exit pressure.
 139

8000

7000

6000
Inertial Velocity (m/s)

5000

4000

3000

2000

1000

0
0 500 1000 1500 2000 2500 3000
Time (sec)

Figure C.4: Inertial velocity plots of all optimized hydrolox simulations

8000

7000

6000
Inertial Velocity (m/s)

5000

4000

3000

2000

1000

0
0 500 1000 1500 2000 2500 3000
Time (sec)

Figure C.5: Inertial velocity plots of all optimized kerolox simulations

140 Figures

0.8
Eccentricity

0.6

0.4

0.2

0
0 500 1000 1500 2000 2500 3000
Time (sec)

Figure C.6: Eccentricity plots of all optimized hydrolox simulations

0.8
Eccentricity

0.6

0.4

0.2

0
0 500 1000 1500 2000 2500 3000
Time (sec)

Figure C.7: Eccentricity plots of all optimized kerolox simulations

 141

5
x 10
3

2.5

2
Pitch angle (rad)

1.5

0.5

0
0 500 1000 1500 2000 2500 3000
Time (sec)

Figure C.8: Altitude plots of all optimized hydrolox simulations

5
x 10
3

2.5

2
Altitude (m)

1.5

0.5

0
0 500 1000 1500 2000 2500 3000
Time (sec)

Figure C.9: Altitude plots of all optimized kerolox simulations

142 Figures

50

40
Total Acceleration (m/s2)

30

20

10

0 200 400 600 800 1000 1200

Time (sec)

Figure C.10: Total acceleration plots of all optimized hydrolox simulations

50

40
Total Acceleration (m/s2)

30

20

10

0 500 1000 1500 2000 2500 3000

Time (sec)

Figure C.11: Total acceleration plots of all optimized kerolox simulations

 143

1.4

1.2

1
Pitch angle (rad)

0.8

0.6

0.4

0.2

0.2
0 50 100 150 200 250 300 350 400
Time (sec)

Figure C.12: Pitch angle plots of all optimized hydrolox simulations

1.5

1
Pitch Angle (rad)

0.5

0 50 100 150 200 250 300 350

Time (sec)

Figure C.13: Pitch angle plots of all optimized kerolox simulations

144 Figures

0.6

0.5

0.4
Angle of Attack (rad)

0.3

0.2

0.1

0.1

0.2
0 50 100 150 200 250 300 350 400
Time (sec)

Figure C.14: Angle of attack plots of all optimized hydrolox simulations

0.6

0.5

0.4
Angle of attack (rad)

0.3

0.2

0.1

0.1

0.2
0 50 100 150 200 250 300 350
Time (sec)

Figure C.15: Angle of attack plots of all optimized kerolox simulations

 145

6
10

5
10
Dynamic Pressure (Pa)

4
10

3
10

2
10

1
10

0
10
0 50 100 150 200 250
Time (sec)

Figure C.16: Logarithmic dynamic pressure plots of all optimized hydrolox simulations

6
10

5
10
Dynamic Pressure (Pa)

4
10

3
10

2
10

1
10

0
10
0 200 400 600 800 1000 1200
Time (sec)

Figure C.17: Logarithmic dynamic pressure plots of all optimized kerolox simulations
146 Figures

4
10

3
10
Bending load (Pa*rad)

2
10

1
10

0
10
0 20 40 60 80 100 120 140 160 180 200
Time (sec)

Figure C.18: Logarithmic bending load plots of all optimized hydrolox simulations

4
10

3
10
Bending load (Pa*rad)

2
10

1
10

0
10
0 50 100 150 200 250 300 350 400
Time (sec)

Figure C.19: Logarithmic bending load plots of all optimized kerolox simulations
 147

6
10

5
10

4
10
Heat Flux (W/m2)

3
10

2
10

1
10

0
10
0 50 100 150 200 250 300
Time (sec)

Figure C.20: Logarithmic heat flux plots of all optimized hydrolox simulations

6
10

5
10

4
10
Heat Flux (W/m2)

3
10

2
10

1
10

0
10
0 200 400 600 800 1000 1200
Time (sec)

Figure C.21: Logarithmic heat flux plots of all optimized kerolox simulations
148 Figures

6
x 10
8

2
Y

4
8 6 4 2 0 2 4 6 8
X 6
x 10

Figure C.22: Position plots in Cartesian coordinates of all optimized hydrolox simulations in the
equatorial plane

6
x 10
8

2
Y

4
8 6 4 2 0 2 4 6 8
X 6
x 10

Figure C.23: Position plots in Cartesian coordinates of all optimized kerolox simulations in the equa-
torial plane
 149

0.4

0.35

0.3

0.25
Lift to weight ratio

0.2

0.15

0.1

0.05

0.05

0.1
0 20 40 60 80 100 120 140 160 180 200
Time (sec)

Figure C.24: Lift to weight plots of all optimized hydrolox simulations.

0.4

0.35

0.3

0.25
Lift to weight ratio

0.2

0.15

0.1

0.05

0.05
0 20 40 60 80 100 120 140 160 180 200
Time (sec)

Figure C.25: Lift to weight plots of all optimized kerolox simulations.

150 Figures

5
x 10
3

2.5

2
Altitude (m)

1.5

0.5

0
0 1000 2000 3000 4000 5000 6000 7000 8000
Velocity (m/s)

Figure C.26: Altitude vs velocity plots of all optimized hydrolox simulations.

5
x 10
3

2.5

2
Altitude (m)

1.5

0.5

0
0 1000 2000 3000 4000 5000 6000 7000 8000
Velocity (m/s)

Figure C.27: Altitude vs velocity plots of all optimized kerolox simulations.

 Bibliography

[1] American National Standards Institute, Guide to Reference and Standard Atmo-
sphere Models, ANSI/AIAA Paper G-003C-2009, Feb. 3. 2003.

[2] Anderson, J. D., Hypersonic and high temperature gas dynamics, 2001. Reprint. New
York: McGraw-Hill, 1989. Print.

[3] ARIANESPACE, Vega Users Manual, Issue 3 / Revision 0, March, 2006. 188 pp.

[4] ASTRIUM Space Transportation, VULCAIN 2: Thrust Chamber, 4pp.

[5] Bayley, D.J., Design Optimization of Space Launch Vehicles Using a Genetic Algo-
rithm, Auburn, Alabama, Aug. 2007.

[6] Carter, P., Owen, B., Eremenko, P., Hudson, J., Lynch, R., Design Trade Space
Analysis for Air-Launched Spacelift Vehicles, 2nd Responsive Space Conference, Los
Angeles, CA, Apr. 2004.

[7] NASA, Chemical Equilibrium with Applications (CEA)., NASA N.p., n.d. Web. 11
Oct. 2012. http://www.grc.nasa.gov/WWW/CEAWeb/

[8] Champion, K.S.W., Cole, A.E., Kantor, A.J., Standard and Reference Atmo-
spheres, Handbook of Geophysics and the Space Environment, pp. 14-1 - 14-
43. Air Force Geophys. Lab., Bedford, Massachussets, 1985. retrieved from:
http://www.cnofs.org/Handbook of Geophysics 1985/Chptr14.pdf

[9] Chenglong, H., Xin, C., Leni, W. Optimizing RLV Ascent Trajectory Using PSO
Algorithms College of Automation Engineering, NanJing University of Aeronautics
and Astronautics, Nanjing, China, 2008.

[10] Chong, K.P., Zak, H.S., An Introduction to Optimization, New York: Wiley, 1996.
Print.

[11] COESA, 1976: U.S. Standard Atmosphere, 1976. NOAA, 227 pp.

151
152 BIBLIOGRAPHY

[12] Cornelisse, J.W., Schyer, H.F.R., Wakker, K.F., Rocket Propulsion and Spaceflight
Dynamics, Pitman, London. 1979.

[13] DeJarnette, F.R., Davis, R.M., A simplified method for calculating laminar heat
transfer over bodies at an angle of attack, Langley Research Center, NASA TN D-
4720, Washington D.C., Aug. 1968.

[14] Engelen, F.M. Quantitative risk analysis of unguided rocket trajectories, Master of
Science Thesis, Delft University of Technology. 2012.

[15] ESA, Esa Launch Vehicle Catalogue, European Space Agency, Paris.

[16] FAI RECORDS. FAI portal. N.p., n.d. Web. 10 Oct. 2012.
http://www.fai.org/records

[17] Ferguson, F., Corbett, T., Lindsay, H., The development of waveriders using an
axisymmetric flowfield, Center for Aerospace Research, North Carolina A&T State
University.

[18] Filatyev, A.S, Golikov, A.A., New Possibilities of Significant Improvement of

Aerospace Launcher Efficiency by Rigorous Optimization of Atmospheric Flight.,
Central Aerohydrodynamic Institute (TsAGI), 24th International Congress of The
Aeronautical Sciences, 2004.

[19] Filatyev, A.S, Golikov, A.A., Aerospace Trajectory Optimization: Novel view on a
role of atmospheric flight, Central Aerohydrodynamic Institute (TsAGI), 26th Inter-
national Congress of The Aeronautical Sciences, 2008.

[20] Gen, M., Cheng, R., Genetic Algorithms & Engineering Optimization, John Wiley
& Sons, Inc. 2000. Print.

[21] Giri, R., Ghose, D., Differential Evolution Based Ascent Phase Trajectory Optimiza-
tion for a Hypersonic Vehicle, Springer-Verlag Berlin Heidelberg, 2010.

[22] Gordon, S., McBride, B.J., Computer Program for Calculation of Complex Chemical
Equilibrium Compositions and Applications - I. Analysis, NASA, Reference Publica-
tion 1311, Oct. 1994.

[23] Gunter, B., Hanssen, R., Simons, D., Introduction to Earth Observation: Lecture 5,
Gravity field - Physics, TU Delft, Delft. 2008. Lecture.

[24] Hammond, W. E., Design methodologies for space transportation systems, Reston,
VA: AIAA, 2001. Print.

[25] Hartfield, R.J., Ahuja, F., Albarado, K.M., Walsh, T., A Method for Optimizing
Launch Vehicle Aero-Assist Including Path Optimization, Auburn University, AIAA
2011-6566, Aug. 2011.

[26] Hsu, D.Y., Comparison of Four Gravity Models, IEEE, Litton Guidance & Control
Systems. 1996
BIBLIOGRAPHY 153

[27] Izzo, D., Becerra, V. M., Myatt, D. R., Nasuto, S. J., and M., B. J., Search Space
Pruning and Global Optimisation of Multiple Gravity Assist Spacecraft Trajectories,
Journal of Global Optimization, 38:283296. 2007.

[28] Kirk, D.R., MAE 4262: Rockets and Mission Analysis - Rocket Equa-
tion and Losses, Mechanical and Aerospace Engineering Department,
Florida Institute of Technology, Jan. 22, Lecture. retrieved from:
http://www.docstoc.com/docs/123207503/Rocket-Equation-and-Losses

[29] Krhsel, G., Schfer, K., Kronmller, H., Zimmermann, H., P4.1 Test Facility for Alti-
tude Simulation of Vinci Engine - Bench Development, Institute of Space Propulsion,
European Conference for Aero-Space Sciences (EUCASS), German Aerospace Center
(DLR), Germany.

[30] Mendenhall, M.R., Chou, H.S.Y., Love, J.F., Computational Aerodynamic Design
and Analysis of Launch Vehicles, Nielsen Engineering & Research, Inc, AIAA Paper
2000-0385, Jan. 2000.

[31] Merkl, A., Mass Model for TSTO Vehicle Concepts, ESA Reference: FESTIP-SS-50,
31 Jan. 1995.

[32] Mooij, E., The Motion of a Vehicle in a Planetary At-

mosphere, Delft University Press, 1997. retrieved from:
http://repository.tudelft.nl/view/ir/uuid%3Ae5fce5a0-7bce-4d8e-8249-
e23293edbb55/

[33] Mulder, J.A, van Staveren, W.H.J.J., van der Vaart, J.C., de Weerdt, E., Flight
Dynamics - Lecture Notes, February 5, 2007, Delft

[34] NASA, NASA Photo: EC91-348-4, NASA Dryden Flight Research Cen-
ter Photo Collection, N.p, n.d. Web. date last accessed: 15 Nov. 2012.
http://www1.dfrc.nasa.gov/Gallery/Photo/Pegasus/Small/EC91-348-
4.jpg.

[35] NASA, DARPA, Report of the Horizontal Launch Study - Interim Report, June 2011

[36] Noffz, G.K., Curry, R.E., Summary of Aerothermal Test Results from the First Flight
of the Pegasus Air-Launched Space Booster, AIAA paper 91-5046, NASA Dryden
Flight Research Facility, AIAA Third International Aerospace Planes Conference,
Orlando, FL, 3-5 Dec. 1991

[37] Noomen, R., Chu, Q.P., Kamp, A., Space Engineering and Technology III - AE3-803
Version 2.0 Lecture Notes, TU Delft, June, 2007, Delft

[38] Pagano, A., Global Launcher Trajectory Optimization for Lunar Base Settle-
ment, Master of Science Thesis, Delft University of Technology. 2010 retrieved
from: http://repository.tudelft.nl/view/ir/uuid%3Afa96b280-af15-4520-
b528-f93b07752e8d/

[39] Price, K.V., Storn, R.M., Lampinen, J.A., Differential Evolution - A Practical Ap-
proach to Global Optimization, Springer-Verlag Berlin Heidelberg, 2005. Print.
154 BIBLIOGRAPHY

[40] Rafique, A.F., He, L.S., Zeeshan, Q., Kamran, A., Nisar, K., Multidisciplinary design
and optimization of an air launched satellite launch vehicle using a hybrid heuristic
search algorithm, Engineering Optimization, 43:3, 305-328, Taylor & Francis, Informa
Ltd. 2011. http://dx.doi.org/10.1080/0305215X.2010.489608
[41] Rao, S.S., Engineering Optimization - Theory and Practice, Fourth edition, Johm
Wiley & Sons, Inc., 2009. Print.
[42] RD-108-8D75K. Encyclopedia Astronautica. N.p., n.d. Web. 11 Oct. 2012.
http://www.astronautix.com/engines/rd18d75k.htm
[43] Resnick, B., North American X-15: Record Breaking Vehicles, Pop-
ular Mechanics, Popular Mechanics. N.p., n.d. Web. 10 Oct. 2012.
http://www.popularmechanics.com/technology/engineering/extreme-
machines/worlds-top-12-fastest-vehicles-north-american-x-15#slide-5
[44] de Ridder, S., Study on Optimal Trajectories and Energy Management Capabilities
of a Winged Re-Entry Vehicle during the Terminal Area, Master of Science Thesis,
Delft University of Technology. July 2009.
[45] Ridolfi, G., Mooij, E., Corpino, S., Complex-Systems Design Methodology
for Systems-Engineering Collaborative Environment, Systems Engineering -
Practice and Theory, Boris Cogan (Ed.), ISBN: 978-953-51-0322-6, InTech,
http://www.intechopen.com/books/systems-engineering-practice-and-
theory/complex-systems-design-methodology-for-systems-engineering-
collaborative-environment
[46] Saltelli, A., Tarantola, S., Chan, K.P.S., A Quantitative Model-Independent Method
for Global Sensitivity Analysis of Model Output, Joint Research Centre of the Euro-
pean Comission, Italy, 1999.
[47] Shaughnessy, J.D., Pinvknry, S.Z., McMinn, J.D., Cruz, C.I., Kelly. M. Hypersonic
Vehicle Simulation Model: Winged-Cone Configuration, NASA Technical Memoran-
dum 102610, November. 1990.
[48] Sooy, T.J., Schmidt, R.Z. Aerodynamic Predictions, Comparisons, and Validation
using Missile DATCOM (97) and Aeroprediction 98 (AP98) Journal of Spacecraft
and Rockets, Vol.42, No.2, March-April 2005.
[49] Spaans, J., Improving Global Optimization Methods for Low Thrust Trajectories,
Delft University of Technology, Delft, The Netherlands, Aug. 2009.
[50] Starsem, RD-0124 An Optimized Propulsion System for Soyuz/ST, Versailles, May
14, 2002, Lecture.
[51] Stern, R.G., Aerothermal Heating for Satellite Reentry Conditions, The Aerospace
Corporation, Aerospace Report No. TR-2004(8506)-2, El Segundo, California, 22nd
Sept. 2004.
[52] Thin-Walled Pressure Vessels., eFunda: The Ultimate On-
line Reference for Engineers. N.p., n.d. Web. 11 Oct. 2012.
www.efunda.com/formulae/solid mechanics/mat mechanics/pressure vessel.cfm
BIBLIOGRAPHY 155

[53] Tudat Working Group, Tudat User Tutorial, Aug. 25, 2011.

[54] Tusar, T., Filipic B., Differential Evolution versus Genetic Algorithms in Multiobjec-
tive Optimization, Springer-Verlag Berlin Heidelberg, Jozef Stefan Institute, Ljubl-
jana, Slovenia, 2007.

[55] Vallado, D.A., McClain, W.D., Fundamentals of astrodynamics and applications,

Dordrecht: Kluwer Academic Publishers, 2001. Print.

[56] Vandamme, J., Study on the Effect of Asssisted Launch on Launch Vehicle Perfor-
mance, Literature Study, TU Delft, Delft, Feb. 2nd 2012.

[57] Vuik, C., van Beek, P., Vermolen, F., van Kan, J., Numerical Methods for Ordinary
Differential Equations, Delft, VSSD, 2007. Print.

[58] Wakker, K.F., Astrodynamics - I - AE4-874, Part 1, Lecture Notes, TU Delft, Au-
gust, 2007, Delft

[59] Wakker, K.F., Astrodynamics - II - AE4-874, Part II, Lecture Notes, TU Delft,
August, 2007, Delft

[60] Weise, T., Global Optimization Algorithms: Theory and Application. 2008.

[61] Werley, B. L., A Comparison of CEA and HSC Software for Oxidant
Safety Thermo/Equilibrium Analysis, Self-published opinion, BWOpinion Website,
www.enter.net/ bwerley, 2010, 31 pages.

[62] Wertz, J. R., Larson, W.J., Space mission analysis and design. 3rd edition. Torrance,
CA, Microcosm, 1999. Print.

[63] Wiegand, A., Practical Problem Solving on Fast Trajectory Optimization 3rd Inter-
national Workshop on Astrodynamics tools and Techniques. Lecture. 2006.

[64] Yokoyama, N., Flight Trajectory Optimization Using Genetic Algorithm Combined
with Gradient Method, University of Tokyo, Japan, 2001.

[65] Zandbergen, B.T.C., Thermal Rocket Propulsion (version 2.04) - AE4S01, Lecture
Notes, TU Delft, February, 2010, Delft

[66] Zandbergen, B.T.C., Storage tanks for liquids and gases - AE4S01, TU Delft, Delft,
1 May 2004, Lecture.

[67] Zandbergen, B.T.C., Aerospace Design & Systems Engineering Elements I - AE1-
201, Part: Launcher design and sizing, Lecture Notes, TU Delft, Delft

[68] Zandbergen, B.T.C., TU Delft: Thruster mass estimation., TU Delft: LR. N.p.,
n.d. Web. 31 Aug. 2011. http://lr.tudelft.nl/en/organisation/departments-
and-chairs/space-engineering/space-systems-engineering/expertise-
areas/space-propulsion/system-design/analyze-candidates/dry-mass-
estimation/powered-systems/thruster-mass/.
156 BIBLIOGRAPHY

[69] Zandbergen, B.T.C., TU Delft: Data for rocket propulsion system

size estimation ., TU Delft: LR. N.p., n.d. Web. 10 Oct. 2012.
http://www.lr.tudelft.nl/en/organisation/departments-and-chairs/space-
engineering/space-systems-engineering/expertise-areas/space-
propulsion/design-data/size-estimation/.