MASTER'S THESIS

Health Monitoring of Re-entry Vehicles

Michael McWinnie

Master of Science Space Engineering - Space Master

Luleå University of Technology Department of Computer Science, Electrical and Space Engineering Health Monitoring of Re-entry Vehicles

Michael McWhinnie Supervisor: Dr. Peter Roberts School of Engineering Cranfield University

A thesis submitted for the degree of: MSc Astronautics and Space Engineering

2011 August

c Cranfield University, 2011. All rights reserved. No part of this publication may be reproduced without the written permission of the copyright holder. Abstract

The inevitable return of space hardware from low earth orbit via targeted re-entries or naturally decaying orbits, creates a risk to human casualty and property damage. Fragmentation models that are used to assess this risk, however, contain a large degree of uncertainty and lack experimental data for calibration. A recently proven method for a↵ordably recording this crit- ical re-entry data is the use of an automated, compact re-entry health mon- itoring system passively attached to a host . The device records a variety of data during the breakup, only to release and transmit its data volume to a commercial satellite constellation for retrieval and analysis.

Using the Automated Transfer Vehicle re-entry scenario, this study explores the limitations of such a concept by identifying and analysing system and scenario-related constraints. Although realisable, two major findings were made evident. First, the transmittable data volume, though sucient in this scenario, is likely to be quickly exceeded in more data-intensive scenarios. Second, power consumption on a passive device is a major constraint for longer duration missions.

The data recorded is expected to provide invaluable insights leading to model calibration and spacecraft ‘design-for-demise’ techniques. Given the necessity to mitigate casualty risk and comply with agency guidelines, the aim here is to relieve spacecraft operators of manoeuvring costs and sub- sequent payload reductions. There is also potential to use the device as a ‘black-box’ for reusable re-entry vehicles of the future. Such a device could reduce developmental costs and risk to manned-spaceflight. Acknowledgements

To all who have directly or indirectly assisted in the creation of this work: My supervisor Dr. Peter Roberts for his insights and guidance, everyone in Meeting Group 1 for ideas transpired through discussion, SpaceMasters past and present, Tweefontein for their understanding and hospitality, Emily for her love and support, and of course, my family. Thank you. ii Contents

List of Figures v

List of Tables vii

Notation ix

Acronyms xi

1 Introduction 1 1.1 Thespacedebrisproblem ...... 1 1.2 The re-entry health monitoring problem ...... 2 1.3 Statement of purpose ...... 3 1.4 Overviewofthestudy ...... 4

2 Background and Context 5 2.1 Debrishazard...... 5 2.2 Design-for-demise...... 7 2.3 Fragmentation models ...... 8 2.3.1 Object-orientedcodes ...... 8 2.3.1.1 DAS ...... 8 2.3.1.2 ORSAT ...... 9 2.3.1.3 SESAM ...... 10 2.3.2 Spacecraft-oriented codes ...... 10 2.3.2.1 SCARAB ...... 10 2.4 Reentry observation campaigns ...... 12 2.5 The Automated Transfer Vehicle (ATV) ...... 12 2.6 TheRe-entryBreakupRecorder(REBR) ...... 13

iii CONTENTS

2.6.1 Overview ...... 14 2.6.2 Mission profile ...... 15 2.6.3 Flight history ...... 15

3 Research Design 17 3.1 Researchsummary ...... 17 3.2 Scopeofstudy ...... 18

4 Mission Concept 21 4.1 Missionobjectives ...... 21 4.2 MissionConcept ...... 22 4.2.1 Physical properties ...... 22 4.2.2 Payload ...... 22 4.2.3 On-orbit phase ...... 23 4.2.4 Atmospheric entry phase ...... 23 4.2.5 Data recovery phase ...... 24

5SystemRequirements 27 5.1 Systempassivity ...... 27 5.2 Aerothermalrequirements ...... 29 5.3 Dynamicrequirements ...... 30 5.4 Fragmentation requirements ...... 31 5.5 Communicationrequirements ...... 32 5.6 Summary ...... 33

6 Trajectory Modelling 35 6.1 Trajectory model rationale ...... 35 6.2 Planartrajectorymodel ...... 36 6.2.1 Formulation of dynamic model and assumptions ...... 36 6.2.2 Formulation of atmospheric model and assumptions ...... 40 6.2.3 Initial conditions ...... 40 6.2.4 Trajectoryresults...... 42 6.2.5 Discussion...... 46 6.2.5.1 Maximum transmittable data volume ...... 46 6.2.5.2 Aeroshell design ...... 48

iv CONTENTS

6.2.5.3 Thermal protection system ...... 49

7 Payload 51 7.1 Attitudedetermination...... 51 7.1.1 Attitude rates of change ...... 52 7.1.2 Integration technique ...... 54 7.2 Temperaturemeasurement...... 55 7.2.1 Device housing and thermocouple type ...... 56 7.2.2 Thermocouple amplifiers ...... 57 7.3 Heat flux measurement ...... 57 7.3.1 The lumped system approach to heat transfer ...... 58 7.3.2 Model Validation ...... 60 7.3.2.1 Error sources and assumptions ...... 61 7.3.2.2 Calculated heat flux output for the RAFLEX tempera- tureprofile ...... 62 7.3.3 Discussion...... 63 7.3.4 Applicability to the ATV re-entry environment ...... 65 7.3.5 Conclusion ...... 66 7.4 Pressuresensor ...... 66 7.5 Data transceiver and Antenna ...... 67 7.6 Data generation and transmission ...... 69 7.6.1 Sampling rate lower-bound ...... 69 7.6.2 Sampling rate upper-bound ...... 70 7.6.3 Estimated data volume generated ...... 70 7.7 Micro-controller and data storage ...... 71

8 Power 73 8.1 Payload power consumptions ...... 73 8.2 Payload activation scheduling ...... 74 8.2.1 Low-power payload schedule ...... 77 8.2.2 Medium-power payload schedule ...... 77 8.2.3 High-power payload schedule ...... 77 8.3 Batteryselectionandsizing ...... 78 8.4 Discussion...... 79

v CONTENTS

9 Mass 81 9.1 Payload mass ...... 81 9.2 Thermal protection system mass ...... 81 9.3 Housing mass ...... 82

10 Conclusion 85 10.1 Thesiscontribution...... 85 10.2 Looking forward ...... 87 10.3Futurework...... 88 10.4 Final thoughts ...... 89

References 91

A Additional Trajectory Results 95

B Ariane-5 Performance Envelope 101

vi List of Figures

2.1 Impact of de-orbit timeframe mitigation strategies on manned-mission debris impact flux for objects 1cm (1)...... 6 2.2 Di↵ering heat loads for a full re-entry, and a re-entry exposure starting at a generic altitude (2)...... 9 2.3 A typical spacecraft model generated in SCARAB (2)...... 11 2.4 Overview of the ATV design and layout (3)...... 13 2.5 Exploded view of the REBR with copper housing (4)...... 14 2.6 REBR mission profile overview as initially conceived in it’s 2005 patent (5)...... 16

5.1 Panel shadowing in SCARAB from geometric surface area projections (6) 31

6.1 Inertial coordinate system for re-entry...... 37 6.2 Drag coecient as a function of Knudsen number as calculated by various fragmentation models, for an AA7075 Aluminium sphere of 1m diameter (2)...... 38 6.3 ISA density model with an exponential extension above 20km altitude. . 41 6.4 Re-entry trajectory profiles. Note unequal axes for presentation purposes. 43 6.5 Velocity profiles versus time over full entry angle range...... 43 6.6 Trajectoryaltitude-timehistory...... 44 6.7 Mach number profiles, where the intersections with the black line indi- cates the beginning of the subsonic flight region...... 44 6.8 G-loadinghistory...... 45 6.9 Flight path angle evolution during re-entry...... 45

vii LIST OF FIGURES

6.10 Altitude profile for lifting body with CL =0.05. Note the inflexion at moderate altitudes, causing an increase in total flight times...... 49

7.1 Ariane-5 cryogenic main stage EPC body angular rates during re-entry (7)...... 54 7.2 MIR body angular rates during re-entry (8)...... 55 7.3 Calorimeter design for the RAFLEX air data system (9) ...... 60 7.4 Specific heat of Copper as a function of temperature (9)...... 62 7.5 Replication of temperature data generated by a calorimetric probe KFA2 during the RAFLEX II experiment on the MIRCA re-entry capsule (10). 63 7.6 Aerothermal heat flux output from Matlab model, with relevant compo- nents...... 64 7.7 Post-processed aerothermal heat flux output from KFA2 calorimeter on- board RAFLEX (10)...... 65 7.8 Iridium coverage at sea-level using the 48 beams for all 66 satellites. . . 68

A.1 Drag force versus time over full entry angle range...... 95 A.2 Gravitational acceleration profiles over full entry angle range...... 96 A.3 Ground track profiles versus time over full entry angle range...... 96

A.4 Ground-track in lifting-body scenario for CL =0.05...... 97 A.5 Stagnation-point heat flux profiles over full entry angle range...... 97 A.6 Stagnation-point temperatures over full entry angle range...... 98 A.7 Comparison of exponential and ISA (black) density models...... 98

A.8 Lifting body re-entry flight-path history for CL =0.05...... 99

A.9 Lift force in lifting-body scenario for CL =0.05...... 99

B.1 Ariane-5 sustained launch accelerations (11)...... 101 B.2 Ariane-5 shock envelope across vibration frequency spectrum (11). . . . 102

viii List of Tables

2.1 SCARAB aerothermal models for various flow regimes (6)...... 11

4.1 MissionObjectives...... 21

5.1 Tradeo↵ table comparing fully passive and active RHMS systems. . . . . 28 5.2 System requirements summary, with corresponding sections...... 33

6.1 Initial conditions for low altitude trajectory ...... 39 6.2 Initial conditions for RHMS trajectory in ATV re-entry scenario. . . . . 41 6.3 Trajectory results for various initial entry angles. Entry and impact angles are relative to local horizontal...... 42 6.4 Estimated data transmission times and subsequent maximum transmit- table data volumes, using flight-path angle as the main constraint. Times includea30secondconnectiondelay...... 48

7.1 ADIS16375 Performance specifications (12)...... 53 7.2 Melting temperature, thermal conductivity and penetration times for some common materials...... 56 7.3 Calorimetric head parameters used in RAFLEX II and Matlab validation model(10)...... 61 7.4 Iridium 9522B Specifications (13) ...... 67 7.5 Data volume generated when IMU outputs are sampled at 15Hz. . . . . 71

8.1 Power consumption estimates for payload items in both normal and sleep modeswhereapplicable...... 75 8.2 ATV-1 Jules Verne undocking and re-entry timeline (14)...... 76

ix LIST OF TABLES

8.3 Low-power payload schedule option. No re-entry detection is used in the standbyphase...... 77 8.4 Medium-power payload schedule option. 50% of standby time is used forre-entrydetection...... 78 8.5 High-power payload schedule option. Re-entry detection mode is used throughout the entire standby phase...... 78 8.6 Energizer EA91 specifications (15)...... 79 8.7 Battery sizing for individual payload activation schedules...... 79

9.1 DS-II and PREP re-entry results for TPS selection (16)...... 82 9.2 Payload mass for di↵erent power scheduling scenarios...... 83

x "0 Permittivity of vacuum [F/m] A Sensor area [m2]

2 Ap,us Panel unshadowed area [m ] Bi Biot number

C Sensor material specific heat capac- Notation ity [J/kg.K]

Cb Ballistic coecient

CD Drag coecient

M1 Mach number before shock CL Lift coecient

P1 Pressure before shock [pa] cp Specific heat of panel material [J/kg.K] P2 Pressure after shock [pa] cp Constant pressure specific heat 1 Atmospheric scale height [m] [J/kg.K] x Conduction distance [m] cv Constant volume specific heat 2 q˙aer Total aerothermal heat flux [W/m ] [J/kg.K]

2 D Drag force [N] q˙cal Calorimetric heat flux [W/m ]

flink Communication link frequency [Hz] q˙loss Radiation and conduction heat flux 2 losses [W/m ] fp Plasma frequency [Hz]

2 Q˙ p,cond Conductive heat rate of panel ele- g Gravitational acceleration [m/s ] ment [W] gs Sea-level gravitational acceleration [m/s2] Q˙ p,conv Convective heat rate of panel element [W] h Spacecraft altitude [m] ˙ Qp,rad Radiative heat rate of panel element hc Convection heat transfer coecient [W] [W/m2.K]

Q˙ p,total Total heating rate of panel element L Lift force [N] [W] Lc Characteristic length [m] ✏ Material emissivity m Mass [kg]

Ratio of specific heats me Electron mass [kg]

E Flight path angle relative to local mp Panel mass [kg]

horizontal [deg] 3 ncrit Critical electron density [e/m ]  Thermal conductivity [W/m.K] p Free-stream pressure [pa] 1 3 ⇢ Atmospheric density [kg/m ] q Electron charge [C]

⇢s Sea-level atmospheric density qm Panel material specific heat of melt- [kg/m3] ing [J/kg]

Stefan-Boltzman constant [W/m2.K4] R Radius of the earth [m]

xi NOTATION

r Radius from earth centre to space- Tw Wall temperature [K] craft [m] T Free-stream temperature [K] 1 S Planform area [m2] Tenv Environmental temperature for con- s Ground-track distance [m] duction [T] st Stanton Number V Inertial velocity [m/s] T Temperature [K] V 0 Velocity relative to atmosphere t Time [s] [m/s2]

T0 Stagnation temperature [K] 3 Vs Sensor volume [m ]

Tp Temperature of panel element [K] V Free-stream velocity[m/s] 1 tp Penetration time [s]

xii MADS modular auxiliary data system. 2 MEMS micro-electro-mechanical system. 22 MIRCA Micro Re-entry Capsule. 61

NASA National Aeronautics and Space Admin- istration. 2, 5–9, 12, 14, 35

Acronyms ORSAT object re-entry survival analysis tool. 7–9, 17, 30, 35

PICA phenolic impregnated ceramic ablator. ATV Automated Transfer Vehicle. 2, 4, 12, 13, 82 15, 18, 22–24, 29, 30, 32, 40, 41, 52–57, 66, PREP Pico Re-entry Probe. 82 67, 70, 73, 74, 85 RAFLEX Re-entry Aerodynamic Flow Experi- COESA Committee on Extension to the Stan- ment. 60–63, 65 dard Atmosphere. 40 REBR Re-entry Breakup Recorder. 3, 14, 15, COTS commercial o↵ the shelf. 19, 51, 57, 73, 18, 24, 25, 39, 85, 90 78, 89 RF radio frequency. 2, 32, 46 DAS debris assessment software. 7–10, 17, 35 RHMS re-entry health monitoring system. 3, DOF degrees of freedom. 36, 52 4, 18, 21–24, 27–33, 35, 36, 38–42, 46, 47, DRAMA debris risk assessment and mitigation 50–52, 54–57, 66, 67, 69, 71, 73, 74, 79, analysis. 35 81–83, 85–89 DS-II Deep Space II. 82 SCARAB spacecraft atmospheric re-entry and EEPROM electrically erasable programmable aerothermal breakup. 8, 10, 12, 17, 27, memory. 72 29–31, 35, 53 EOM end of mission. 5, 15 SESAM spacecraft entry survival analysis mod- EPC ´etage principal cryotechnique. 53 ule. 8, 10, 35 ESA . 2, 8, 10, 12, 35 SIRCA silicone impregnated reusable ceramic ablator. 82 GEO geosynchronous earth orbit. 1, 25 SPI serial peripheral interface. 72 GPS Global Positioning System. 15, 33, 71, 75, 76, 83 STK Satellite Tool Kit. 35

ICAO International Civil Aviation Organiza- TDRS Tracking and Data Relay System. 2, 25 tion. 40 TPS thermal protection system. 49, 50, 81, 83, ICC integrated cargo carrier. 13, 29 89 IMU inertial measurement unit. 51, 52, 69, 70, UARS upper atmosphere research satellite. 1, 72, 83 23 ISA international standard atmosphere. 40 UART universal asynchronous receiver/transmitter. ISS International Space Station. 1, 2, 12, 13, 67, 72 15, 23 USART universal synchronous/asynchronous LEO low earth orbit. 1, 5, 22, 23, 47, 65, 67, 86 receiver/transmitter. 67

xiii Acronyms

xiv 1

Introduction

1.1 The space debris problem

The space debris population surrounding earth has been growing steadily since humans first took to space in the 1950s. Objects including launcher stages and decommissioned satellites, as well as particle debris from explosions and collisions, makeup the bulk of this population. Current figures suggest that around 5,500 tonnes of‘ debris remain at various altitudes around the earth between low earth orbit (LEO) and geosynchronous earth orbit (GEO), many of which continue to be used by active satellites and manned spacecraft. The collision risk in LEO is such that avoidance manoeuvres have become a common necessity on the International Space Station (ISS). The 2009 collision between the active Iridium 33 and inactive Kosmos-2251 satellites in this region also highlighted the worsening issue. The LEO debris environment is somewhat alleviated by significant atmospheric drag at altitudes under around 500km. This tends to limit the orbital lifetime of inactive or damaged spacecraft that lack a boost capability. By eventually disintegrating in the earth’s atmosphere, the potential for collision and additional debris is removed. LEO satellites above 500km however, can remain in slowly decaying orbits for many years. An example is the upper atmosphere research satellite (UARS) spacecraft. Launched in 1991 to a 600km orbit and decommissioned in 2005, the spacecraft remains in a decay mode that should lead to an uncontrolled re-entry between 2011-2012. In this case, the spacecraft was left with insucient fuel to accelerate it’s decay.

1 1. INTRODUCTION

Currently, the National Aeronautics and Space Administration (NASA) and Euro- pean Space Agency (ESA) guidelines state that any risk to human casualty for falling space debris with kinetic energy 15J should be less than 0.0001%. A primary exam- ple is ESA’s Automated Transfer Vehicle (ATV), which provides a resupply capability to the ISS. The ATV typically remains in orbit for less than a month before being guided to a controlled re-entry. During re-entry, spacecraft ablate due to the aerother- mal environment that is generated. It is generally thought that between 10 and 40% of spacecraft mass survives re-entry (17), hence giving rise to a potential casualty area. Traditionally, these guidelines are met by altering the mission to ensure su- cient demise. This can be costly to operators, and is not always possible for inactive satellites.

1.2 The re-entry health monitoring problem

Many re-entry vehicles use some form of system health monitoring. Often, this comes in the form of either live streaming data, or recorded data that is later recovered. Recorded data is often required to overcome radio frequency (RF) blackout during re-entry. Operators then use this information to improve both current and future operations through the iterative design of systems and physical components. The Space Shuttle for example, used a data recorder known as the modular auxiliary data system (MADS) to obtain many forms of telemetry during re-entry. A small amount of critical telemetry was also transmitted to the Tracking and Data Relay System (TDRS) satellite constellation for real-time tracking, but was often subject to severe RF interference. Recorded data was then downloaded and analysed post-flight. The disintegration of Space Shuttle Columbia during re-entry in 2003 highlighted the limitations of such a system. The MADS system was not designed to survive an accident, and was not designed to be easily located. Due to a favourable impact angle however, a large volume of data was recovered, revealing crucial information regarding the event. This was made possible only after an extended and costly search for debris (18). This challenge still stands for reusable re-entry vehicles of the future. Re-entry data must be easily recoverable, particularly in the event of unmanned prototype failures,

2 1.3 Statement of purpose which can impart invaluable data. Lacking such a capability can lead to development stagnation and cost blowouts.

1.3 Statement of purpose

Potential solutions to the space debris and re-entry health monitoring problems de- scribed above contain some common elements. A solution that will serve both needs is therefore sought after, and has led to an emerging area of study. The commonal- ity between these is made evident by first considering a solution to the space debris problem. Current spacecraft fragmentation models contain a great deal of uncertainty. An accurate prediction of the spacecraft mass fraction that is expected to survive is not possible. This is due in great part to lack of empirical data from actual re-entries. Cur- rent models are therefore laden with assumptions and approximations that limit their reliability. A better understanding of‘ the complexities of the fragmentation process can therefore lead to dramatic improvements in these models. Such a development could lead to significant advances in the so called ’design- for-demise’ strategy. This represents a shift in the approach used by operators to reduce casualty areas. Rather than change the mission profile, the spacecraft can be designed to ablate completely during re-entry. Extraordinarily low casualty areas are then complimented by maximised satellite operational lifetimes. There is a clear need for the calibration of existing models with real-world fragmentation data. This calls for a survivable re-entry health monitoring system (RHMS). It must record critical re-entry data, survive the spacecraft fragmentation, and easily return the recorded data for analysis. The existing Re-entry Breakup Recorder (REBR) device discussed in Chapter 2 is aimed at filling this need. Only a brief overview of the device is available to the public however, leaving ample room for related research activities to occur. Such a device could also serve as a prototype ’black-box’ for reusable re-entry ve- hicles. Here, critical telemetry could be easily recovered in the event of a loss. RHMS research is likely to stimulate advancements in such spino↵ technologies. The purpose of this study can therefore be summarised as follows:

3 1. INTRODUCTION

1. Identify RHMS requirements for the purpose of re-entry fragmentation model cal- ibration.

2. Analyse RHMS performance for a given re-entry scenario.

3. Evaluate RHMS limitations for the given scenario, and for future applications

1.4 Overview of the study

This study was carried out in a highly sequential manner, much like the feasibility phase in any space project. Chapter 2 contains background research on current fragmentation models, existing re-entry recording devices and the ATV. The ideas that emerge from this review are then summarised in Chapter 3, Research Design. This completes the initial literature review. A general mission concept for the RHMS is then developed in Chapter 4, and is broken into a set of system requirements in Chapter 5. The chapters that follow provide findings based purely on the system requirements identified. This includes the analysis of trajectory, payload, data generation, power and mass to deduce parameters of interest. The conclusion then draws together all the aspects mentioned above. It seeks to evaluate the feasibility of the device analysed, and will revisit the ’black-box’ concept in Section 1.3 for application in future reusable re-entry vehicles.

4 2

Background and Context

2.1 Debris hazard

The degree to which the space debris problem is managed varies significantly between agencies. Processes and requirements relating to spacecraft disposal at end of mission (EOM) are inconsistent, and tend to be cost-dependent. Management of LEO missions tends to limit the cost burden of disposal however, leading to more standardised reg- ulations between agencies. The NASA Safety Standard 8719.14 - Process for Limiting Orbital Debris (19) contains re-entry debris findings and regulations that are generally accepted across most agencies. This standard was therefore used to quantify acceptable ground-risk, and identify the current processes used to mitigate this risk. Spacecraft disposal at EOM can be achieved using one of three methods: atmo- spheric re-entry, manoeuvre to a disposal orbit or direct retrieval. In the context of debris hazard to humans, only atmospheric re-entry was considered in this background research. Two forms of atmospheric re-entry are available for a given mission. The first option is for the spacecraft to remain in, or adopt an orbit that will lead to a natural re-entry within a specified timeframe. The NASA standards state that this is the most energy-ecient method for spacecraft in orbits below 1400km, and dictate that re-entry must occur within 25 years of EOM, or 30 years from launch. The second is to actively direct the spacecraft to a re-entry at a specified time and/or attitude. Spacecraft with average orbit altitudes of under 600km will usually satisfy the first requirement auto- matically if no specific disposal action is taken. Spacecraft with average altitudes over 700km will usually require a manoeuvre to satisfy this requirement. The purpose of

5 2. BACKGROUND AND CONTEXT the 25 year timeframe is to limit the growth of the debris population over the next 100 years as shown in Figure 2.1, while also considering disposal cost.

Figure 2.1: Impact of de-orbit timeframe mitigation strategies on manned-mission debris impact flux for objects 1cm (1).

Re-entry debris hazard however, also depends on the re-entry option chosen, and is defined based on a predicted debris casualty area. If the casualty probability is less than 1:10,000 and no object with kinetic energy greater than 15J is predicted in populated areas, a natural 25 year re-entry is acceptable. If the casualty area or kinetic energy exceeds these thresholds, additional action must be taken. In this case, the mitigation strategies used by NASA include:

1. Perform a controlled re-entry. Manoeuvre spacecraft to re-entry trajectory with perigee no higher than 50km altitude. This prevents atmospheric skipping and provides a degree of control over re-entry location.

2. Implement a ’design-for-demise’ strategy, discussed in Section 2.2. This involves using design methods that increase the likelihood that components will fully ab- late during re-entry.

3. Breakup spacecraft into smaller components immediately prior to re-entry, ensur- ing that no components remain in orbit. This increases the surface area exposed to heating, and reduces the debris footprint.

6 2.2 Design-for-demise

The casualty thresholds are calculated using NASA’s debris assessment software (DAS) and object re-entry survival analysis tool (ORSAT) software packages and are discussed in Section 2.3.

2.2 Design-for-demise

The design-for-demise strategy involves applying spacecraft design practices that ensure components fully ablate during re-entry. This is necessary to reduce the probability of human casualty and property damage as discussed in Section 2.1. Material type, as well as object size and shape all impact ablation properties. For a given material volume, materials with a higher melting temperature will generally be more likely to survive re-entry. Aluminium for example will tend to demise during re-entry, while titanium, beryllium and stainless steal tend to survive (20). Other material properties including specific heat capacity, heat fusion, density and thermal conductivity also have a notable impact (19). With an understanding of these properties, materials can be selected that are conducive to demise.

A complete understanding of the ablation process however, requires an understand- ing of the heat loads experienced during re-entry. Convective heat flux results from the spacecraft’s emersion in a flow heated by rapidly decelerated air. This is the most significant heat load. Radiative heat flux from the preceding bow shock can also pro- vide a significant but lesser heat load, and contribute to total and instantaneous heat loads. The layout and design of the spacecraft dictate which surfaces and materials will be exposed to these loads. Consequently, a combination of design strategy and understanding of the re-entry environment can be used to influence ablation. Section 2.3 shows however, that current models used to simulate such an environment contain critical uncertainty.

The design-for-demise strategy is stipulated in NASA Safety Standard 8719.14 (19). No formal process for implementing the strategy exists however, and has resulted in only a loose application in typical spacecraft design phases (20).

7 2. BACKGROUND AND CONTEXT

2.3 Fragmentation models

Several re-entry fragmentation models have been developed by NASA and ESA in recent years, and remain in common use today. NASA’s standard packages include DAS and ORSAT, while ESA’s include the spacecraft entry survival analysis module (SESAM) and the spacecraft atmospheric re-entry and aerothermal breakup (SCARAB) tool. These models vary considerably in their approach to fragmentation modelling and the fidelity that they o↵er. As such, each model can be classified under one of two categories which broadly define their approach to fragmentation modelling. These include object- oriented codes, and spacecraft-oriented codes. Details of these categories is provided below, along with a description of the relevant models.

2.3.1 Object-oriented codes

Object-oriented codes focus on analysing individual spacecraft components, and assume a generic spacecraft breakup altitude between 75 and 85km (2). Thermal analysis of individual components begins only after this altitude is reached, and a uniform breakup event occurs. Figure 2.2 indicates how this impacts the total heat load for a given spacecraft, when compared with a full re-entry. In using this method, these models neglect the rapid heating of the outer skin in the upper atmosphere and the shielding of inner components. A conservative debris footprint is predicted as a result, meaning that a higher component survivability is calculated. The DAS, ORSAT and SESAM models are all examples of object-oriented codes. Following is a brief description of each.

2.3.1.1 DAS

Originally developed by Lockheed Martin, DAS assesses mission risk according NASA safety standards (19). It is available to the public, and is generally accepted as a low-fidelity first approximation to ground risk (21). Spacecraft are modelled as a col- lection of geometric elements, including plates, spheres, cylinders and boxes, each with mass, material and dimensional properties. All material properties are considered to be independent of temperature. Thermal analysis of all elements begins at the same predefined altitude, and uses a lumped-mass approach for solid objects. This means that an average density approach

8 2.3 Fragmentation models

Figure 2.2: Di↵ering heat loads for a full re-entry, and a re-entry exposure starting at a generic altitude (2). is used for both hollow objects, and objects comprising of several materials. A result of the lumped-mass approach is a uniform temperature assumption for all elements. A temperature gradient is only present between elements. The model does not consider partial melting of elements, but measures the cumulative absorbed heat for each ele- ment. An element is considered to have demised if its heat of ablation is exceeded. Its full mass is otherwise considered to have survived re-entry. This contributes greatly to its conservative output (22).

2.3.1.2 ORSAT

ORSAT was originally designed by NASA Johnson Space Centre. It is used as a high-fidelity analysis tool when DAS has previously indicated a non-compliance with NASA safety standards (19) (23). ORSAT uses similar element geometries to DAS, but includes element kinetics for some, including spinning and tumbling. Material properties are mostly temperature dependent. The thermal analysis uses either the lumped-mass approach, or one-dimensional heat conduction depending on the element’s geometry. Partial melting is also consid- ered using a demise-factor, hence removing some of the conservatism found in DAS.

9 2. BACKGROUND AND CONTEXT

Although limited to ballistic, non-lifting re-entries, the aerodynamic model considers various flow regimes throughout the re-entry. This results in an improved estimate of drag and incident heat flux. A limited capability to allow for heat exchange between structural elements is also provided. Although breakup altitudes remain user-defined, various altitudes can be allocated to specific element groups.

2.3.1.3 SESAM

SESAM is similar in fidelity and application to DAS, and is used to assess debris risk as per the European Space Debris Safety and Mitigation Standard (1). It uses a simplified but otherwise identical aerothermal model.

2.3.2 Spacecraft-oriented codes

Spacecraft-oriented codes model the entire spacecraft an a single entity. Geometries are not simplified approximations, but are complex representations based on the actual spacecraft design. Flow-fields around the spacecraft are calculated to determine accu- rate aerodynamic and thermal characteristics. Fragmentation events are based on sim- ulated melting and mechanical demise rather than a single event at a pre-determined altitude. SCARAB is currently the only known spacecraft oriented code, and is de- scribed below.

2.3.2.1 SCARAB

SCARAB is an ESA re-entry tool developed by HTG, and has been in continual devel- opment since 1995. It is a numerical simulation comprising of high fidelity aerodynamic, aerothermal, thermodynamic, dynamic and structural models (6). The package con- sists of two primary components: spacecraft design and re-entry analysis. The design component provides tools to generate complex shapes from basic primitive geometries, as shown in Figure 2.3. A triangular mesh of variable resolution volume panels is cre- ated. Temperature dependent material properties are available, and can be manually defined by the user if necessary. The re-entry analysis consists of various dynamic and thermal components. The dynamic analysis generates trajectory and attitude simulations. Output is given by integrating full six-degrees-of-freedom equations of motions. The thermal

10 2.3 Fragmentation models

Figure 2.3: A typical spacecraft model generated in SCARAB (2). analysis calculates individual panel temperatures using conductive, radiative and con- vective heat flux models. Heat exchange between neighbouring panels is considered, but only radiative cooling is used. There is no radiative heat exchange between neighbour- ing or adjacent panels. Radiative heating from shocks are neglected, and can impact thermal analysis for high velocity re-entries from interplanetary missions. A complex aerodynamic model addresses several flow regimes, and uses complex bridging functions to smooth transition between them. The aerothermal model uses various expressions for stagnation-point heat flux depending on the flow regime. These are summarised in Table 2.1 below.

Flow regime Model Continuum flow Detra-Kemp-Riddell Laminar flow Blunt-body heat transfer Turbulent flow Detra-Hidalgo Free-molecular flow Daley

Table 2.1: SCARAB aerothermal models for various flow regimes (6).

The aerodynamic and thermal contribution of individual panels is subject to a shad-

11 2. BACKGROUND AND CONTEXT owing analysis. Only panels exposed to the flow contribute loads. A simple projection model is implemented, and does not account for complexities such as flow expansion. Melted panels are removed from the analysis when their heat capacity is exceeded. This exposes internal panels to the flow. A significant strong-point is the inclusion of tank bursting, explosion and sloshing in the SCARAB analysis. Recent studies have shown that these events can have a major impact on spacecraft fragmentation (6). Significant uncertainties are known to exist and require further calibration (24).

2.4 Reentry observation campaigns

One method used to better understand re-entry fragmentation is the re-entry obser- vation campaign. In recent years, several campaigns have been established, aiming to collect optical data for model calibration. The most recent being the joint ESA/NASA airborne ATV-1 Jules Verne Re-entry Campaign, conducted on September 29, 2008. Two aircraft used 22 instruments to capture image, spectroscopic and photometric data. The primary objective was to characterise fragmentation and explosive events, and determine their likely cause. The secondary objectives were to reconstruct the re- entry trajectory, measure rotation rates and identify ablation products through spectral analysis (25). The results highlighted the clear advantage of such campaigns: the ability to re- construct the fragmentation. Very high Chromium emissions early in the re-entry suggested that the propulsion system was the first to be penetrated. This was consis- tent with other observations which stated that the bottom of the ATV was the first to disintegrate. Other spectral events were used to identify similar events throughout the re-entry. Comparisons with existing models such as the study by Koppenwallner et al. (26) are yet to be published (25).

2.5 The Automated Transfer Vehicle (ATV)

ESA’s ATV is an unmanned, expendable spacecraft designed primarily to resupply the ISS at intervals of seventeen months. It’s resupply capabilities include water, propellant, experimental payloads and oxygen among others. It also provides a boost capability

12 2.6 The Re-entry Breakup Recorder (REBR) to raise the orbit of the ISS as it slowly decays due to atmospheric drag. Figure 2.4 shows the main components of the ATV, including the integrated cargo carrier (ICC) and the Propulsion Module (PM) (14).

Figure 2.4: Overview of the ATV design and layout (3).

The ATV is typically launched on the Ariane-5 and weighs approximately twenty tonnes when it docks to the ISS. The ICC is a pressurised vessel and is accessible to ISS crew when the ATV is docked. It contains eight payload racks which are are used to deliver useful payloads upon arrival. The ATV is designed to remain docked for up to six months, at which time its payload racks are loaded with up to 6.5 tonnes of waste from the ISS. After undocking, the ATV remains in orbit for up to a month as operators prepare for a controlled re-entry in which the majority of the structure and waste is destroyed. Typically, any surviving debris impacts the Pacific Ocean (14).

2.6 The Re-entry Breakup Recorder (REBR)

The re-entry fragmentation models discussed in Section 2.3 contain a great number of simplifications, assumptions and unknowns. Current simulations that are based on

13 2. BACKGROUND AND CONTEXT

first-principles fail to explain many observations that are made during spacecraft re- entries. These include the time and altitude of fragmentation events, heating rates and spacecraft attitude. Without real re-entry data recorded from spacecraft during fragmentation, sucient calibrations cannot be made. The REBR designed by the Aerospace Corporation in partnership with NASA Ames and Boeing, was intended to fulfil this need, and has flown two missions to date. This section discusses aspects of the REBR’s design and operation where publicly available. The proprietary nature of the device means that many details of it’s design remain concealed, leaving only general performance indicators from published papers.

2.6.1 Overview

The REBR shown in Figure 2.5 is a small, autonomous device designed to record sensory data from onboard a host spacecraft during re-entry and fragmentation. It is designed to survive the fragmentation event, separate from the host and return recorded data for analysis via a data-call to the Iridium network (4).

Figure 2.5: Exploded view of the REBR with copper housing (4).

The REBR consists of an aeroshell which provides dynamic stability during descent

14 2.6 The Re-entry Breakup Recorder (REBR) and thermal protection for the internal payload. Although the specific payload used is unavailable, it is thought to collect a minimum of temperature, acceleration, rotation rate and Global Positioning System (GPS) data. An Iridium transceiver is used to transmit the recorded data during descent. The device has a mass of 4kg and a diameter of 30cm without copper housing.

2.6.2 Mission profile

Figure 2.6 depicts the main phases of the REBR mission profile. This depiction was illustrated in the original 2005 patent (5), and shows only approximate times for an unspecified re-entry scenario.The REBR is designed to be launched and stored onboard a host spacecraft for the duration of its primary mission. At EOM, the host spacecraft is either manoeuvred to a controlled re-entry or placed in a decaying orbit for eventual re-entry. During this phase, the device enters ’standby’ mode as it attempts to detect conditions representative of re-entry (temperature/acceleration). At this point the device wakes up and begins recording re-entry and fragmentation data. During the main fragmentation event, the device is released from the spacecraft through melting of its copper housing at 75-85km altitude (4). It then enters a ballistic free-fall as it decelerates through the atmosphere. The onboard transceiver then initiates a satellite data-call to upload all recorded data before the device impacts the ground. The device is not designed to survive impact, and is not intended to be recoverable.

2.6.3 Flight history

The REBR has flown twice to date: the first onboard the unmanned Japanese Kounotori- 2 in March, 2011, and the second onboard ATV-2 Johannes Kepler in June, 2011. In both cases, the device was activated by ISS crew immediately prior to undocking (27). The first mission was a success, having recorded and transmitted re-entry data during it’s descent. The second device however, failed to make contact for causes unknown. As of the completion date for this paper, it is unclear whether data from the first re-entry will be made available to the public.

15 2. BACKGROUND AND CONTEXT

Figure 2.6: REBR mission profile overview as initially conceived in it’s 2005 patent (5).

16 3

Research Design

This section draws together findings from the background research in Chapter 2. This is achieved by defining a narrow scope to shape the study that follows

3.1 Research summary

Background research indicated several areas in which a detailed knowledge of the re- entry environment is necessary. The debris hazard is perhaps the most pertinent, where ine↵ectiveness of mitigation strategies can potentially cause injury or death on earth. Techniques for reducing casualty areas can also come at a large cost to operators who seek to maximise the operational times of their satellite assets. Specific spacecraft design practices can assist in this pursuit, but are rarely applied. The lack of for- mal design requirements and only loose integration of these practices in typical design phases, appear to be largely responsible for this trend. Although this can be attributed to a number of factors, an incomplete knowledge of the relationship between spacecraft materials and the re-entry environment appears to be a major contributor. The investigation into commonly used re-entry fragmentation models reiterated this notion. The assumptions used in low-fidelity codes such as DAS provide conservative casualty area estimates. This often results in the need for analysis with high-fidelity codes such as ORSAT or SCARAB, and additional time and cost. Even these high precision codes contain a great number of unknowns, and have led to inconsistencies between model outputs and re-entry observations.

17 3. RESEARCH DESIGN

This scenario led to the development of the Aerospace Corporation’s REBR. The re- entry data that it collects from its host counterpart, is aimed directly at the verification and calibration of these existing models. The localised re-entry detail it can achieve compliments the macro-scale data provided by observation campaigns. The potential to replace these expensive campaigns with such a device is also an attractive possibility. The 0.5 success rate of REBR missions to date shows that the concept of a re-entry fragmentation recorder is feasible. The limitations on such a system and a specific set of system requirements are however yet to be made widely available. Such information could serve as a valuable input to the study of emerging systems, such as flight data recorders for re-usable re-entry vehicles.

3.2 Scope of study

Many di↵erent spacecraft re-entry scenarios exist. Spacecraft mass, volume and orbital orientation are all variable, and can potentially impact RHMS design. Equally, several di↵erent approaches to re-entry health monitoring have been used in the past. An example is the ATV-1 Jules Verne re-entry observation campaign which used electro- magnetic spectroscopy to analyse aspects of its fragmentation. As a result, this section seeks to define a narrow scope that guides the direction of this study. This study will focus on the development of a standalone recording device. This was motivated in part by the successful test flight of the REBR discussed in Chapter 2. Very little information is available regarding specifics of its design and performance. The limited data available was therefore used as a concept guide, and as a point of comparison and reference for the RHMS. The ATV is solely considered as the host spacecraft for the RHMS scenario. The ATV was selected due to the wealth of data available in the public domain. This includes performance and design data, as well as independent studies that have emerged from past ATV re-entry observation campaigns. The ATV was also the host spacecraft used in one REBR flight, making it a logical choice for comparison purposes. A systems engineering approach is adopted in this study to better understand the relationship between system parameters. A typical spacecraft design project is highly iterative and requires such a holistic approach. Where possible, a small-satellite de- sign philosophy was also adopted. The intention was to encourage a compact, low-cost

18 3.2 Scope of study design, realisable over a short timescale using commercial o↵ the shelf (COTS) tech- nologies where relevant. It is not the purpose of this study to identify specific COTS products for use. Specific items were however used when typical results were needed for calculation purposes.

19 3. RESEARCH DESIGN

20 4

Mission Concept

The purpose of this chapter is to develop a general mission concept for a RHMS device. The chapter opens with a broad set of mission objectives. The mission concept is then formulated to meet these objectives. Chapter 5 will go on to to draw this mission concept into a set of system requirements that can be used evaluate the feasibility of such a device.

4.1 Mission objectives

The setting of mission objectives shapes the development of the RHMS concept. The mission objectives were therefore broadly developed with consideration of the aim and scope of this study defined in Section 3.2, and the background research conducted in Section 2. Table 4.1 shows the resulting mission objectives.

Objective 1 Collect re-entry data relevant to spacecraft fragmen- tation. This should include the recording of dynamic, thermal, environmental and/or mechanical data on host spacecraft during re-entry/fragmentation. 2 Return collected data to earth for analysis 3 Minimise dependence on the host spacecraft and be sensitive to certification constraints.

Table 4.1: Mission Objectives.

21 4. MISSION CONCEPT

4.2 Mission Concept

The mission objectives set in Section 4.1 allow an initial mission concept to be devised. This will include any significant missions options that are available during particular phases of flight, and should provide an overview of what an RHMS mission might consist of. This section also attempts to capture both the limitations and demands that are imposed on the mission profile by the mission objectives.

4.2.1 Physical properties

The RHMS should resemble a small self-contained unit, and should be comparable in volume and character to traditional black boxes found in civilian aircraft. Current LEO launch costs are estimated at around $10,000US/kg (28). This highlights the need for a low-mass system to maximise launch opportunities. An initial mass target of 4kg is stated here, and will be subject to a feasibility study. The RHMS should be easily mounted to the host spacecraft. In the case of the ATV, it is envisioned that the device will be mounted within the integrated cargo carrier. Any sensors should be local to the RHMS, such that minimal impact is incurred on the host in the way of cables, structure or function. As such, a device that can be easily relocated by crew is preferable.

4.2.2 Payload

A payload of sensors and a recorder should allow re-entry data to be collected and stored throughout the re-entry and fragmentation phases. Recorded parameters might include acceleration and rotation rate of the host spacecraft. Wall temperatures of structural elements, internal pressure, heat flux or element strain could be examined. The sample rate of these sensors should be subject to expected rates of change of the given parameters, and any memory or time limitations that are identified. Sen- sors should be of minimal mass, and use low power micro-electro-mechanical system (MEMS) technologies where possible, to minimise the size of the power system and the overall unit.

22 4.2 Mission Concept

4.2.3 On-orbit phase

It is foreseen that the RHMS will be required to sit dormant for a long period of time before use. ATV-1 Jules Verne remained on orbit for six months before re-entry (14). Many LEO satellites will remain on orbit for several years before a controlled re-entry is instigated. Others will be left in a slowly decaying orbit after they are decommissioned. An example is the UARS spacecraft, which after its 1991 launch, will re-enter the atmosphere in an orbital decay mode by around 2011 (29). The RHMS should initially be designed around the ATV scenario however. It is typical that the ATV will remain on orbit for around a month before re-entry, following undocking from the ISS. It is envisioned that crew members could activate the RHMS before ATV undocking, reducing the need for automation over longer periods. This is particularly important in a fully passive system, where changes to the system are not possible after launch by either remote or direct means. After undocking, the RHMS should be capable of managing it’s power consumption by entering a sleep mode.

4.2.4 Atmospheric entry phase

The RHMS must wake from sleep-mode once it detects re-entry. Theoretically, either temperature or acceleration increases typical of the re-entry environment could be used to trigger this event. Acceleration is seen as a preferable indicator, since measurements can be made at any location on the host. If located inside the ATV, increases in temperature will not be immediately detectable. A wakeup sensor should be used for this purpose. An analysis of its power con- sumption is necessary to determine whether this sensor can remain powered throughout sleep-mode to detect a possible re-entry. Alternatively, timers can be used when the re-entry schedule is more defined, as is in the case with the ATV. In this scenario, the wakeup sensor may be powered a matter of hours or days prior to re-entry, such that the RHMS can enter full-power mode when recording is necessary. Other scenarios may call for additional wakeup possibilities. Longer duration mis- sions may call for a system where the RHMS is periodically in ground-contact, via some means of communication. This could allow the wakeup event to occur at specific times, but may impose significant power limitations. For the purpose of this study, only the basic wakeup functions mentioned above are considered.

23 4. MISSION CONCEPT

Once the RHMS enters full power mode, it must begin recording data from the sensor payload to onboard memory. The total recording time for the ATV scenario must be identified by determining the time between ATV entry-interface, and its main fragmentation. This will contribute to the total amount of memory that is necessary. The RHMS must be designed to survive the fragmentation, and should be released during the main fragmentation event. A passive method of release is preferable. This might include the melting of the restraining housing or structure, as is thought to have been used on the REBR.

4.2.5 Data recovery phase

Following its ejection from the fragmentation of the host vehicle, the RHMS then begins its descent. Several techniques were considered regarding retrieval of the recorded data. These included:

Device recovery • Local data transfer • Satellite data transfer •

Recovery of the device following ground-impact eliminates the need for a commu- nications system, hence simplifying the overall architecture. Solid state memory can be utilised to record large amounts of data with very little mass or volume overhead. Two key drawbacks however, are the potential cost and feasibility of recovery. This is particularly true of a sea-based recovery. An additional drawback is the challenge of designing the device to remain intact and accessible following ground or water impact. For these reasons, this option was eliminated. A local data transfer was also considered. In this option, an onboard transceiver transfers data to a sea or air-based receiver station near the re-entry flight-path. While this option overcomes the challenge of surviving impact, it too requires infrastructure of considerable cost. An added challenge is the mobilisation of these assets to specific locations in potentially short periods of time. While ample warning may be available in scenarios like the ATV, this is not always the case. For these reasons, this option was also eliminated.

24 4.2 Mission Concept

Finally, a data transfer to an orbiting satellite network was considered. The space shuttle used a tail-mounted antenna to transmit telemetry to the TDRS system during re-entry, hence overcoming RF blackout to a large degree. The REBR adapted this idea by using the Iridium satellite constellation to receive a data call from the device during descent. This approach has the clear advantage of very little infrastructure overhead. It also provides a large coverage of the earth’s surface, and no human intervention. The disadvantage is the introduction of additional technical challenges. These might include automation, or any attitude/velocity limitations that are introduced. Power consumption and mass could also be a consideration, particularly if a GEO satellite is used.This option was however viewed as the most realistic and advantageous, and was selected for further investigation.

25 4. MISSION CONCEPT

26 5

System Requirements

This chapter defines the system requirements for the RHMS. The first three require- ments categories listed below were developed using the software structure of SCARAB 3.1L as a guide. These categories define the functional requirements of the payload for data collection. The final category defines the requirements for retrieving stored data from the payload for post-mission analysis. The system requirements categories therefore include:

Aerothermal • Dynamic • Fragmentation • Communication and data retrieval • Establishing a set of requirements in these areas will then allow the feasibility of various system concepts to be investigated. This will also impact the nature of the payload that is carried onboard. The following sections open with a discussion of system passivity before moving into the various system requirements. A summary will then draw these together in a final list 5.6.

5.1 System passivity

System passivity is a measure of how independent the RHMS will be from the host spacecraft. The degree of passivity will strongly influence the overall configuration of the RHMS, and therefore serves as a logical starting point.

27 5. SYSTEM REQUIREMENTS

The RHMS is to be a fully passive system and will therefore be entirely independent from the host spacecraft. A tradeo↵ table comparing degrees of passivity is shown in Table 5.1 and was used in making this determination. Reducing complexity was seen as the primary criteria. The selection of a passive system will significantly simplify the architecture of the RHMS (17), but may also constrain its functionality. This section briefly discusses these constraints.

Advantages Disadvantages Fully passive Non-invasive, minimal certi- Power limited, no access fication issues, low cost, de- to telemetry, limited sensor creased complexity placement access Active Access to host power bus, ac- Invasive, certification issues cess to telemetry, broader ac- likely, high cost, increased cess to host for sensor place- complexity ment

Table 5.1: Tradeo↵ table comparing fully passive and active RHMS systems.

Table 5.1 indicates that a fully passive system is advantageous for its simplicity and minimal impact on the host spacecraft. The obvious drawback of such a configuration is that it may limit the ability to locate sensors at various locations around the host spacecraft. Cabling and attachment points will likely raise certification issues. To max- imise RHMS flight opportunities, it is unacceptable for the host spacecraft to require redesign or recertification. The use of single-location sensors is, however, viewed as an acceptable compromise at this stage. Future revisions could potentially make use of wireless sensors for an increased number of measurement points around the host. Access to power is also a critical point. Without access to bus power from the host spacecraft, the RHMS primary power is limited to a finite source, such as non- rechargeable batteries. This inevitably limits its usable life in orbit. This is also viewed as an acceptable compromise. The implications however, are that payload selection and power management become critical design elements. These are discussed in Chapters 7 and 8 respectively. Access to telemetry from the host spacecraft could provide the RHMS with an attitude determination capability. The clear advantage here, is that an accurate at-

28 5.2 Aerothermal requirements titude reference can be obtained without the need for reference sensors (sun sensors, star trackers, etc). In a fully passive system, attitude reference must be determined independently if it is required. Onboard reference sensors are not viewed as an option. The additional weight, complexity and mounting logistics are considered to outside the scope of the system concept. Additionally, the RHMS will not have line-of-site access to any attitude references if mounted inside the host. In the ATV, this would likely be within the ICC (See Figure 2.4). The degree of attitude determination that is feasible and necessary is discussed in Section 7.1.

5.2 Aerothermal requirements

The re-entry aerothermal environment is the most significant factor to consider when analysing spacecraft fragmentation. It is also the area in which the greatest uncer- tainties exist within current fragmentation models. This is to some extent due to a lack of empirical data, but can also be attributed to modelling complexities. Both the high range of Mach numbers (typically between Mach 1 and Mach 25) and the vari- ous transitional flow regimes around the vehicle are challenges encountered by current models. The SCARAB model uses convection, conduction and radiation heat flux terms in order to calculate the total heating loads and temperature gradients. Equation 5.1 shows this relation, and can be numerically integrated to obtain Tp over time t.The total heat flux is central in analysing the conditions for thermal fragmentation of the spacecraft. This is discussed further in Section 5.4.

dT Q˙ = m c p = Q˙ + Q˙ + Q˙ (5.1) p,total p p dt p,conv p,cond p,rad Of the heat flux terms found in Equation 5.1, the convection term is generally the most dominant for vehicles re-entering at velocities < 10km/s (30). Equation 5.2 gives the expression used in SCARAB, which implements an approach using local heat transfer coecients (Stanton Number) for each panel. The coecients are calculated based on the current flow regime.

2 Q˙ p,conv = Ap,us st p V (0.5V cp,air(Tw T )) (5.2) · · 1 1 1 1

29 5. SYSTEM REQUIREMENTS

Of particular note in Equation 5.2 is the dependence on panel wall temperature

Tw. This term is clearly influential in the calculation of convective heat flux. Direct measurement of wall temperatures onboard the host are therefore necessary. Most importantly, a direct measurement of heat flux would also be useful in cali- brating the heat convection model. This would allow the accuracy of the fragmentation criterion in Equation 5.3 to be evaluated. Incident heat flux is the primary factor lead- ing to structural demise, including the disintegration of surface panels and primary structure. A simple pressure measurement onboard the RHMS could provide an in- dication of when these major events occur. These timelines can be compared with existing simulations to provide additional calibration.

5.3 Dynamic requirements

The initial attitude of the host spacecraft prior to the entry interface (generally con- sidered as altitudes > 120km) can have a significant impact on the the magnitude of the ground debris footprint. It has the potential to introduce a considerable amount of uncertainty, particularly in the case of an uncontrolled reentry where random angular rates may exist. Lips et al calculated a 21% standard deviation between the ground footprints of a variety of controlled and uncontrolled reentries (31) using SCARAB. Studies such as this, highlight the overall impact of spacecraft attitude on re-entry fragmentation. Specific knowledge of a spacecraft’s entry attitude and velocity can, however, be dicult to obtain. The uncontrolled re-entry of satellites, for example, can consist of high rates of random rotations which cannot be accurately tracked, even with precise instruments. Controlled re-entries such as the ATV however, occur in predictable manner. A tumbling motion is typically instigated to prevent atmospheric rebound, and assist in uniform and complete fragmentation (14). Measuring motion of this character is possible, and will provide general rates of change that can be fed into, or compared with existing models. Controlled re-entries usually have the added advantage of an active telemetry and guidance system onboard. Re-entry tools such as SCARAB and ORSAT make particular use of inputs such as these. It subsequently impacts the aerodynamic, aerothermal and fragmentation analysis. This is particularly important to the SCARAB tool, whereby a detailed analysis of panel shadowing and internal heating is carried out, as is simplified in

30 5.4 Fragmentation requirements

Figure 5.1. Equation 5.2 indicates how convective heat flux is dependent on unshadowed surface area Ap,us and free-stream velocity. While Ap,us can be adequately modelled with a general knowledge of attitude, velocity requires a higher precision which may not be possible. Controlled re-entries usually have the added advantage of an active telemetry and guidance system onboard, making it possible to identify these values post-flight.

Figure 5.1: Panel shadowing in SCARAB from geometric surface area projections (6)

The RHMS therefore must be capable of acquiring relevant attitude information. This should include acceleration and rotation rates in all three axes. The frame of reference will not be obtainable independently, but can be deduced with relevant input from host spacecraft telemetry, after re-entry has occurred. Although precise attitude determination is desirable, it is not required for the calibrations discussed.

5.4 Fragmentation requirements

System requirements regarding spacecraft fragmentation are driven by both thermal and mechanical breakup modes. Thermal fragmentation occurs due to melting or ab- lation, while mechanical fragmentation occurs due to acting forces and moments. The total heat flux in Equation 5.1 is used to evaluate the thermal fragmentation criterion for the demise of the spacecraft panel elements through melting. SCARAB holds panel temperature constant once its melting temperature is reached. Total heat flux then becomes dependent on the changing panel mass as it ablates, as is shown in Equation 5.3.

dm Q˙ = q p (5.3) p,total m dt This assumption is perhaps the most significant in terms of SCARAB’s aerothermal modelling, and has been subject to question. An examination of this criterion was

31 5. SYSTEM REQUIREMENTS carried out by Lips et al (31). The study found that the ground risk of debris impact was highly sensitive to variations in the criterion, showing a 59% standard deviation across various scenarios and re-entry vehicles. This is a direct inference on the sensitivity of the fragmentation process to the criterion. Direct measurements of heat flux and temperature are therefore necessary to determine the validity of this model. Due to the requirement for system passivity, it is not possible to measure the stresses and strains in components that lead to mechanical demise. Accelerations and rotation rates can however provide some measure of the mechanical environment.

5.5 Communication requirements

Communications requirements can be derived following the data recovery concept dis- cussed in Section 4.2.5. This section defines the estimated recording period and any requirements regarding the transmission of data for recovery. Re-entry is typically defined to begin at an altitude of 120km (14), where aerody- namic forces and heating begin to increase significantly. As the spacecraft descends further into the atmosphere, melting, abrasion and fragmentation of components be- gins to occur. Observations of historical re-entries, including the ATV-1 Jules Verne campaign, suggest that a main fragmentation event typically occurs between 75km and 85km (2). These two events mark the beginning and end of the main recording period, spanning an altitude of around 45km. It is during the main fragmentation event that the RHMS is designed to separate from the host and begin its descent. The time taken to descend from 120km to 75-85km is variable depending on the initial flight-path angle at entry-interface. In the ATV controlled re-entry scenario, this descent takes approximately 5-7 minutes. The device must therefore be capable of recording the data volume output from the sensor payloads over at least a 5 minute period. The data volume is dependent on the number of sensors and their sampling rate. The recorded data must then be transmitted at the earliest convenience. A trajec- tory analysis is necessary to determine the timeframe in which communications will be possible. This analysis should include any attitude or velocity constraints, and should consider the possibility of RF blackout. Once an approximate descent time is estab- lished, a transmittable data volume can be determined based on known transfer rates.

32 5.6 Summary

This should be compared with the recorded data volume to determine any limitations or margins.

5.6 Summary

Table 5.2 shows a summary of the RHMS system requirements. Some requirements, such as record time, have been derived from the requirements discussion above, and will impact subsequent analyses. Others, such as the various calculated data items, are yet to be defined. Where relevant, the corresponding sections are also provided.

System requirements Section Accelerations (x, y, z) [m/s2] Angular rates (x, y, z) [deg/s] Recorded data Heat flux [W/m2] Temperature [K] Pressure [pa] GPS (pos/lon/lat/vel) Record time 5 minutes Accelerometer 7.1 Gyroscope 7.1 Thermocouple 7.2 Calorimeter 7.2, 7.3 Payload Pressure sensor 7.4 Data transceiver 7.5 Antenna 7.5 Microcontroller 7.7 Solid-state memory 7.7 Calorimetric heat flux 7.3 Initial conditions (velocity, flight path angle) 6.2.3 Trajectory parameters 6.2.4 Data transmit time 6.2.5.1 Calculated data Max. transmittable data volume 6.2.5.1 Payload data volume generated 7.6.3 Payload data volume transmitted 7.6.3 Power budget / battery sizing 8.1, 8.3 Mass budget 9

Table 5.2: System requirements summary, with corresponding sections.

33 5. SYSTEM REQUIREMENTS

34 6

Trajectory Modelling

This chapter contains an analysis of the estimated trajectory that a RHMS might follow under various mission profiles. This acts as a precursor to the design of subsystem elements such as the communications system. Factors such as decent time, attitude and heating a↵ects can all limit the amount of recoverable data and will ultimately dictate the method of data recovery that is selected.

6.1 Trajectory model rationale

A number of models were considered for the re-entry trajectory analysis. Their usability was assessed based on availability, complexity and the level of detail required from the output. For this study, an estimate of the general decent profile was considered to be satisfactory for initial subsystem design. The Satellite Tool Kit (STK) package from Analytical Graphics, Inc was initially considered due to its availability and ease of scenario design. It was found however, that additional commercial packages were required in order to obtain propagators relevant to the re-entry environment. Several existing trajectory and fragmentation models used by ESA and NASA were also considered, including ESA’s debris risk assessment and mitigation analy- sis (DRAMA)/SESAM and SCARAB, and NASA’s ORSAT and DAS. Of these, only DAS was publicly available. DAS is closed-source however, preventing the user from seeing the details of the models used or altering certain parameters.

35 6. TRAJECTORY MODELLING

Due to the limitations imposed by the above packages, the development of a Matlab model was also investigated. The use of a 6 degrees of freedom (DOF) model was eliminated due to an unnecessary degree of accuracy and significant complexity related overhead. Applying a 3-DOF model was deemed to be unnecessary for similar reasons. Guidance and attitude control is assumed to be unavailable on a device of this size and complexity, and hence will not take full advantage of such a model. Finally, a 2-DOF semi-ballistic model was selected for use in the trajectory analysis. Without a detailed knowledge of the mass or aerodynamic properties, this approach is expected to provide results of acceptable accuracy. Details of the development and use of this model are contained in Section 6.2. The Matlab code used is included in the electronic copy of this study.

6.2 Planar trajectory model

This section describes the 2-DOF planar trajectory model selected, assumptions related to the model, its implementation and the results obtained from it.

6.2.1 Formulation of dynamic model and assumptions

The 2-DOF planar trajectory model was selected for its ease of development and ad- equate level of accuracy during the initial design phase of the RHMS. Since its scope is limited to subsystem design, a simplified inertial model was utilised rather than one in a cartesian coordinate system. This is deemed to be an appropriate simplification since coordinate transformations will not be necessary for this study. The equations of motion used in the model are shown in Equations 6.1a to 6.1d (30). These are valid only in an inertial frame and for a non-rotating planet as shown in Figure 6.1. The position, trajectory and velocity errors introduced by this assumption are minimised by providing initial conditions (V,r,) within the inertial frame. This is convenient since the trajectory profile of the RHMS begins after the entry interface following the demise of the host spacecraft. At this point, the initial conditions are either known through measurement, or can be approximated in the inertial frame. This is explained in further detail in Section 6.2.3.

36 6.2 Planar trajectory model

Figure 6.1: Inertial coordinate system for re-entry.

dr dh = V sin (6.1a) dt dt E dV D g = (6.1b) dt m sin E d 1 L V 2 E = g cos (6.1c) dt V m r E  ✓ ◆ ds R = V cos (6.1d) dt r E Equations 6.1a to 6.1d are subject to external lift and drag forces, and gravitational acceleration as per Equations 6.2a to 6.2c. Given that initial conditions are within the inertial frame, the assumption can be made that the velocity relative to the atmosphere is approximately equal to inertial velocity, as shown.

1 2 L = ⇢V 0 SC V 0 V (6.2a) 2 L ⇡ 1 2 D = ⇢V 0 SC V 0 V (6.2b) 2 D ⇡ R 2 g = g (6.2c) s r ✓ ◆

37 6. TRAJECTORY MODELLING

The expressions for lift and drag forces require an estimate of lift and drag coe- cients. Only a rough estimate of drag coecient was possible, and was based on the drag profile of known capsule geometries. This approach was necessary since an ac- curate knowledge of the RHMS geometry cannot be gained until subsystem design is complete. Additionally, a study of capsule shapes is necessary, but is beyond the scope of this study. This model neglects the a↵ects of changing flow regimes throughout the re-entry profile. Drag coecient typically changes depending on flow regime, and is represented by the Knudsen number. Typically, drag coecient is highest during the free-molecular- flow regime and lowest in the subsonic or continuum regime. A sharp peak is also typical in the transitional regime as the mach number becomes transonic. These a↵ects are visible in Figure 6.2, whereby the drag coecient is calculated for a metallic sphere entering from an altitude of 120km.

Figure 6.2: Drag coecient as a function of Knudsen number as calculated by various fragmentation models, for an AA7075 Aluminium sphere of 1m diameter (2).

Using a static drag coecient neglects these e↵ects. The impact of this assump- tion is, however, minimised by considering a trajectory only in the transitional and continuum regimes. This is possible because the RHMS is typically released after the free-molecular-regime has passed. Using an average static drag coecient over the

38 6.2 Planar trajectory model transitional and continuum regimes is therefore deemed an acceptable simplification for this purpose. To maximise flight time and hence transmission time, the highest possible drag coecient is desirable. A generally accepted value for the drag coecient of a 3D sphere is Cd =0.5, and Cd =0.8 for an angled cube. An average conservative value of

Cd =0.65 was used since a smaller drag coecient will reduce decent time. The initial mass estimate of 4kg and capsule diameter of 30cm were based on the initial mission concept in Chapter 4 and the existing REBR design (4). These parameters were then used to calculate the ballistic coecient, which is a commonly accepted measure of a body’s aerodynamic deceleration characteristics. This makes it ideal for comparison to other known geometries. The ballistic coecient is defined as:

m Cb = (6.3) SCd

2 This yielded the result Cb = 87.1kg/m for the initial estimate of the RHMS ge- ometry. This value is comparable to estimates of the REBR ballistic coecient during 2 the early phases of it’s design, where a value of Cb = 73.4kg/m was identified (32).

A lift coecient of CL = 0 was used such that a purely ballistic trajectory was modelled with zero lift force. This is also a conservative assumption since any lift will act to increase the total decent time and subsequent transmission time. The concept of using lift to increase the decent time is discussed in Section 6.2.5.2 but is neglected in this model. A summary of the parameters applied in the trajectory model are shown in Table 6.1.

Lift coecient CL 0

Drag coecient CD 0.65 2 Ballistic coecient Cb 87.1kg/m 2 Gravitational acceleration at sea level gs 9.80665m/s 2 Atmospheric density at sea level ⇢s(@15C) 1.2250kg/m

Radius of the earth RE 6, 378.14km

Table 6.1: Initial conditions for low altitude trajectory

39 6. TRAJECTORY MODELLING

6.2.2 Formulation of atmospheric model and assumptions

A simple exponential atmospheric density model was considered for use in the trajec- tory simulation, and is a function of altitude only as shown in Equation 6.4. This model was found to provide a sucient output at high altitudes of around 75 120km, which is above the maximum altitude of most commonly available atmospheric models. Similarly, it was found to provide a sucient output at low altitudes, where the factor

⇢s is used to calibrate the output at sea level. It was found however, that the model would provide significant underestimates of density in the intermediate altitude range of 5 15km where the RHMS is in free-fall. This determination was made by comparing the output of the exponential model with the international standard atmosphere (ISA) of the International Civil Aviation Organization (ICAO). Considering that this had the potential to impact the total decent time, a more general model was sought after. The Matlab function ‘atmoscoesa()’ was found to use the 1976 Committee on Extension to the Standard Atmosphere (COESA) model. The model provides an exponential exten- sion of the ISA at altitudes above 20km, and was found to be in agreement with the model in Equation 6.4. This model was subsequently used in the trajectory simulation. Figure 6.3 shows how an exponential extension to the ISA model is applied.

1 h ⇢(h)=⇢se (6.4)

A comparison of the various atmospheric density model outputs can be found in Appendix A with additional trajectory results.

6.2.3 Initial conditions

The initial conditions described here represent the initial velocity, altitude and flight path angle of the RHMS at the point of separation from the host spacecraft. A number of initial conditions could be considered for various spacecraft and mission profiles. The case considered here, however, relates to the ATV only. The ATV represents a typical destructive re-entry profile, where a main fragmentation event is expected in the 75 85km altitude range (2). It is therefore an ideal case for this study. The initial conditions for this scenario were able to be deduced from a variety of studies relating to the ATV-1 Jules Verne re-entry on 29 September 2008. The initial state is assumed to correspond with the main fragmentation event. This infers that

40 6.2 Planar trajectory model

Figure 6.3: ISA density model with an exponential extension above 20km altitude. the RHMS separates from the host at this time, and inherits the velocity and altitude of the ATV at the time of the fragmentation. The flight path angle however, may vary considerably due to the spread of debris during the fragmentation. Models derived from the re-entry observations indicate that this range lies between 4 and 20 relative to the local horizontal and in-plane with the trajectory (33). The initial conditions applied for this scenario are summarised in Table 6.2.

Inertial velocity 7.38km/s (25) Altitude 74.6km (33) Flight path angle 4 to 20 (33)

Table 6.2: Initial conditions for RHMS trajectory in ATV re-entry scenario.

41 6. TRAJECTORY MODELLING

6.2.4 Trajectory results

Contained in this section are the results yielded from the Matlab trajectory simulation for the specified range of entry angles. Table 6.3 provides a summary of these results. A brief description of some of the result parameters is first necessary.

Time to subsonic: Elapsed time between the main fragmentation event (t = 0) • and the time at which the RHMS slows to M = 1.

Time to vertical: Elapsed time between the main fragmentation event (t = 0) • and the time at which the RHMS enters a vertical free-fall. For the purpose of this model the flight-path angle is considered to be vertical when 80 .  Subsonic alt.: Altitude at which RHMS slows to M = 1. • Vertical alt.: Altitude at which RHMS enters a vertical free-fall. • Max. ’g’ loading: Maximum deceleration experienced during the simulated tra- • jectory, expressed in g’s.

Impact angle: Flight-path angle of RHMS at ground impact • Impact velocity: Velocity of RHMS at ground impact • Decent time: Total time elapsed between main fragmentation event (t = 0) and • ground impact.

= 4 = 10 = 15 = 20 E E E E Time to subsonic 2.1 min 1.3 min 1.0 min 0.76 min Time to vertical 2.6 min 1.9 min 1.5 min 1.3 min Subsonic alt. 28.7 km 28.6 km 27.9 km 26.9 km Vertical alt. 22.2 km 21.9 km 21.5 km 21.0 km Max. ’g’ loading 12 26 38 50

Impact angle 90 90 90 90 Impact velocity 33.8 m/s 33.8 m/s 33.8 m/s 33.8 m/s Decent time 8.10 min 7.31 min 6.95 min 6.68 min

Table 6.3: Trajectory results for various initial entry angles. Entry and impact angles are relative to local horizontal.

42 6.2 Planar trajectory model

Shown below are some graphical results from the Matlab re-entry trajectory sim- ulation. Figure 6.4 shows the general trajectory shape, while Figure 6.6 depicts the altitude-time history. Both mach number and attitude (flight-path angle) can impact data communications, hence these quantities are shown in Figures 6.7 and 6.9 respec- tively. Figure 6.5 shows the velocity profiles for consideration of Doppler constraints. Additional trajectory results can be found in Appendix A.

Trajectory 80 γ = −4 E γ = −10 70 E γ = −15 E γ = −20 60 E

50

40 Height (km) 30

20

10

0 0 50 100 150 200 250 300 350 400 Range (km)

Figure 6.4: Re-entry trajectory profiles. Note unequal axes for presentation purposes.

Velocity vs Time 8 γ = −4 E γ = −10 7 E γ = −15 E γ = −20 6 E

5

4

Velocity (km/s) 3

2

1

0 0 1 2 3 4 5 6 7 8 9 10 Time (min)

Figure 6.5: Velocity profiles versus time over full entry angle range.

43 6. TRAJECTORY MODELLING

Altitude vs Time 80 γ = −4 E γ = −10 70 E γ = −15 E γ = −20 60 E

50

40 Altitude (km) 30

20

10

0 0 1 2 3 4 5 6 7 8 9 10 Time (min)

Figure 6.6: Trajectory altitude-time history.

Mach no. vs Time

25 γ = −4 E γ = −10 E γ = −15 E 20 γ = −20 E

15

Mach number 10

5

0 0 1 2 3 4 5 6 7 8 Time (min)

Figure 6.7: Mach number profiles, where the intersections with the black line indicates the beginning of the subsonic flight region.

44 6.2 Planar trajectory model

G−loading vs Time 50 γ = −4 E 45 γ = −10 E γ = −15 40 E γ = −20 E 35

30

gs 25

20

15

10

5

0 0 1 2 3 4 5 6 7 8 Time (min)

Figure 6.8: G-loading history.

Entry angle vs Time 0 γ = −4 E γ = −10 −10 E γ = −15 E −20 γ = −20 E

−30

−40

−50 Entry angle (deg) −60

−70

−80

−90 0 1 2 3 4 5 6 7 8 9 10 Time (min)

Figure 6.9: Flight path angle evolution during re-entry.

45 6. TRAJECTORY MODELLING

6.2.5 Discussion

The simulated re-entry trajectory results obtained in Section 6.2.4 have a direct impact on many aspects of subsystem design. This section discusses inferences that can be drawn from these results, and their impact on the relevant subsystems.

6.2.5.1 Maximum transmittable data volume

Data transmission time is the timespan over which the RHMS will be able transmit it’s stored data to a satellite communications network such as Iridium. The data volume transmission capability during this period depends on the network data rates. Recall that the RHMS will first record re-entry data between entry-interface at 120km and the main fragmentation altitude. It will then be required to transmit its data for analysis at the earliest possible time. Velocity, Mach number and flight-path angle histories shown in Figures 6.5, 6.7 and 6.9 respectively were identified as possible constraints to transmission time for any given initial flight-path angle. Flight-path angle was found to be the most significant of these. All are discussed in detail here, and lead to an approximation of the maximum data volume transmission possible. Mach number was first considered. Attenuation of electromagnetic radiation emis- sions during re-entry due to interaction with the surrounding plasma at high mach numbers is a well documented phenomena. This is commonly referred to as ’RF Black- out’, and occurs when the radio frequency used for communications is comparable to the electron frequency of free electrons in the plasma sheath surrounding the spacecraft. This e↵ectively reduces the transmission power and alters the transmission frequency, usually resulting in a loss of communications. The critical electron density ne, above which attenuation can be expected to occur, is calculated by rearranging the definition of electron frequency:

2 1 q ne ⇡flink 2 fp = ncrit =4"0me( ) (6.5) 2⇡ s"0me ! q For the Iridium network which uses L-band frequencies between 1616 1626.5 MHz, the critical electron density is calculated as n =3.26 1010e/cm3. To gain an crit ⇥ approximation of the blackout period, the electron density generated in the plasma must be determined. A rough estimate is possible with a number of flow field solvers when spacecraft geometry and dynamics are accurately known. Without these details

46 6.2 Planar trajectory model however, the analysis is futile. For this reason the conservative assumption was made that communications are not possible for M 1. This is considered a reasonable assumption due to the rapid deceleration of the RHMS through the Mach range, having low mass and high drag characteristics. As a result, a comparatively small fraction of the total flight time occurs at supersonic velocities. This is visible in Figures 6.7 and 6.8. A more detailed analysis would be necessary in scenarios where Mach number becomes the key constraint. An example is a steep initial flight-angle, where low Mach numbers are reached before vertical flight is established. The flight-path angle history in Figure 6.9 can strongly influence visibility between an onboard antenna and the Iridium satellite network. Antenna placement is discussed in Section 7.5, and indicates that a rear-mounted omni-directional antenna will provide roughly hemispherical visibility. A vertical attitude is therefore necessary to guarantee satellite access. Inertial velocity was also examined to determine the impact of Doppler shifting on communications. As a fast moving LEO constellation, the Iridium network is capable of dealing with considerable Doppler shifts of up to 37.5kHz (34). On close examination ± of Figures 6.7 and 6.5, it was found that the 4 RHMS trajectory had the highest Mach 1 inertial velocity at 296m/s. Using the L-band frequency range stated above, this was calculated to correspond to a maximum Doppler shift of 1.60kHz, which is well under the threshold of 37.5kHz. This also provides a large bu↵er for the additional velocity component introduced by relative motion between the RHMS and satellite. This is not anticipated to be a significant factor, and was not considered in this study. Comparatively, the maximum allowable shift of 37.5kHz represents an inertial velocity of around 6.9km/s; a point very high in the RHMS trajectory. Access is also dependent on other variables not addressed here in detail. These include:

The number of visible satellites • The position of visible satellites relative RHMS flight-path • RHMS geocentric latitude/longitude • RHMS inertial heading angle •

47 6. TRAJECTORY MODELLING

These findings suggest that flight-path angle is the most significant constraint to consider for this scenario. The results in Table 6.3 show that vertical flight occurs roughly 30 seconds after subsonic velocities are reached, and approximately 7km lower in altitude. This indicates that transmission times of approximately 4.8 minutes can be expected. Included in this approximate is a conservative 30 second delay to establish a network connection (35). It was then possible to approximate the maximum data volume that can be transmitted. Iridium advertises a data rate of 2.4kbps, however a real-world rate of 1.2kbps was used for this calculation (36). Table 6.4 summarises transmission times and transmittable data volumes for the relevant range of initial flight-path angles.

= 4 = 10 = 15 = 20 E E E E Data transmit time 4.8 min 4.8 min 4.8 min 4.7 min Data volume (@1.2kbps) 354kb 352kb 349kb 346kb

Table 6.4: Estimated data transmission times and subsequent maximum transmittable data volumes, using flight-path angle as the main constraint. Times include a 30 second connection delay.

6.2.5.2 Aeroshell design

Although aeroshell design is outside the scope of this work, the trajectory results pose some interesting demands on this process. Clearly, it is desirable to maximise the drag coecient of the aeroshell, particularly during subsonic flight. This will lengthen the transmission times, and will lead to greater volumes of transmittable data. The importance of this is reiterated by examining the sensitivity of the simulated data transmission times to drag coecient. A 10% change to the nominal drag coecient of 0.65 for example, leads to a 7-8% change in both transmission time and transmittable data volume. This is a considerable amount, particularly when high data rates are unavailable on networks such as Iridium. This places a high importance on the drag characteristics of the aeroshell design. A lifting aeroshell is also possible, and can be implemented by using a high centre of gravity to generate an angle of attack. This approach does however require a degree of spacecraft guidance to ensure the centre of gravity remains in a desirable vertical

48 6.2 Planar trajectory model orientation. This concept holds the potential to increase the total descent time and hence transmission time. Although an in-depth analysis of lifting-bodies is outside the scope of this work, an initial study was carried out. The altitude profile result is illustrated in Figure 6.10 for CL =0.05, with additional results contained in Appendix A. The findings show a notable increase in total descent times by just under a minute, due to the notable inflexion visible in the altitude profile. Transmission times remained largely unchanged, however. This indicated that the period of increased flight-time occurs in the region currently constrained by velocity, Mach number or attitude, as discussed in Section6.2.5.1.

Altitude vs Time 80 γ = −4 E γ = −10 70 E γ = −15 E γ = −20 60 E

50

40 Altitude (km) 30

20

10

0 0 1 2 3 4 5 6 7 8 9 10 Time (min)

Figure 6.10: Altitude profile for lifting body with CL =0.05. Note the inflexion at moderate altitudes, causing an increase in total flight times.

6.2.5.3 Thermal protection system

Trajectory results typically serve as an input to thermal protection system (TPS) de- sign. Given that TPS design was not deemed to impose significant constraints, this task was largely excluded. Initial results for stagnation point heating rate and temperature, however, are included in Appendix A. Stagnation point heating was based on the Kemp and Riddell method for a full re-entry from orbital velocities (30), and hence provides

49 6. TRAJECTORY MODELLING an overestimate in this scenario. Stagnation temperature was derived from the First Law of Thermodynamics as shown in Equation 6.6.

T 1 0 =1+ M 2 (6.6) T 2 Further analysis is needed to calculate heat transfer from the stagnation point to the RHMS wall, and the resulting heat flux and wall temperatures. This will impact TPS material selection, but must also ensure the internal RHMS volume is kept within operational limits. These limits are driven primarily by payload items, such as thermo- couple amplifiers, which are particularly sensitive to temperature. Selection of material and thickness also impacts mass calculations as discussed in Section 9.2.

50 7

Payload

This chapter details the selection of sensors that will make up the RHMS payload. Chapter 5 developed a set of system requirements that are necessary to better under- stand uncertainties in current fragmentation models. The purpose of this chapter is to identify payloads that will help to meet these requirements. Included is an estimate of the data volume likely to be generated by the payloads based on their individual sampling requirements. This data needs to be transmitted during descent, meaning that it must then be compared to the maximum transmittable volume calculated from the trajectory analysis in Chapter 6. The payload selection will then form the basis of the power system sizing in Chapter 8.

7.1 Attitude determination

A number of attitude measurement solutions were investigated to fulfil the dynamic requirements in Section 5.3. The instruments must be capable of measuring acceleration and rotation of the host spacecraft within certain ranges, and in a convenient reference frame. The issue of whether this can be achieved through direct or indirect methods is also discussed here. A number of accelerometers, angular rate sensors and complete inertial measure- ment unit (IMU)s were investigated to provide position and attitude measurements. For the purpose of cost reduction, all sensors examined were available as COTS prod- ucts. The following technical specifications were identified as being particularly critical to performance:

51 7. PAYLOAD

Sample rate - While high sample rates can provide improved resolution and • accuracy, this may also pose a challenge for data recovery. Large amounts of recorded data will take longer to transmit.

Acceleration limits - The sensors must be able to withstand all acceleration • loads throughout the host spacecraft mission. Most importantly, their operational limits must be appropriate for the loads expected during the re-entry fragmenta- tion phase. This will likely incur significant axial and rotational accelerations.

Power - Sensors with low power consumption should be given priority. This is • necessary to reduce the size and complexity of the power system.

Interface - Sensors with common interface protocols are preferable to improve • integration of the sensor with the other supporting systems.

Mass/volume - Compact, low-mass systems are preferable to reduce the total • size and mass of the RHMS.

Both analog and digital devices were considered. A digital sensor was found to be preferable to avoid the need for analog to digital conversion. Conversion can be relatively slow, even for micro-processors with reasonably high clock rates. A high sampling rate is required from the attitude sensors for attitude tracking, making a digital sensor appropriate for this purpose. The Analog Systems ADIS16375 6-DOF IMU was identified as a typical light- weight, low-power complete inertial sensor. It was therefore used as a baseline for all power, data and mass calculations in this study. Key specifications are shown in Table 7.1, with full details contained in the ADIS16375 Specifications Document (12). This sensor is not intended for space applications, but is used here as a guide only. Critical Ariane-5 launch performance envelopes are contained in the Ariane-5 User’s Manual (11). For an ATV launch, these envelopes must be withstood by the final payload, and are contained in Appendix B for reference.

7.1.1 Attitude rates of change

It is necessary to identify the attitude rates of change expected during the re-entry of the host spacecraft. The reasons are twofold. First, it is critical to ensure that

52 7.1 Attitude determination

Specification Max. value Sample rate: 2.46kHz 3-axis accelerometer g-limit: 18g ± 3-axis gyroscope: 300 /s ± Voltage: 3.3V Shock limit: 2000g Mass: 25 grams

Table 7.1: ADIS16375 Performance specifications (12). these rates lie within the sensor’s performance limitations. Second, a suitable sample rate for the sensor can be identified to ensure all major frequency components of the signal are captured in the output. Two main components were considered. The first being the frequency introduced in two axes due to a dominant angular velocity in the remaining axis. The second being the frequency introduced by the rate of change of angular velocity. The dominant frequency is used in Section 7.6 to gain an estimate of the data volume produced by the payload. Due to the lack of availability of ATV-specific re-entry data, other historical de- structive re-entries were sourced. The aim of this investigation was to gain a general understanding of the re-entry dynamics and to evaluate whether the attitude sensors are applicable for such an environment onboard the ATV. Figures 7.1 and 7.2 show the body angular rates of the Ariane-5 ´etage principal cryotechnique (EPC) main stage and MIR space station during their re-entries. The data occurs over a 60km to 85km range which is typical for main fragmentation events. These figures were output from SCARAB simulations and verified with optical observation data. The inevitable di↵erences in mass and moments of inertia only allow for rough comparisons to the ATV. The angular rates however, approach only 30% of the 300/s limit of the sensor. Figure 7.1 is of particular relevance to the ATV case. The total body angular rate is initially 71/s due to a tumbling motion caused by tank venting around the pitch axis. This e↵ectively constrains the angular rates as aerodynamic forces increase. This increases to a maximum of around 83/s,whichintroducesthe first frequency component of around 0.23Hz. This scenario is similar to the ATV controlled re-entry, whereby a 10/s pitch rate is introduced to reduce lifting e↵ects

53 7. PAYLOAD

Figure 7.1: Ariane-5 cryogenic main stage EPC body angular rates during re-entry (7). during re-entry (37). This is a further indicator that the ATV body angular rates are not expected to vary significantly from these related cases. The second frequency component was then extracted from the gyroscope’s angular velocity output. Angular velocity is viewed as the most significant signal to analyse, since the accelerometer outputs will largely inherit their frequency components from the spacecraft rotation. The deceleration of the spacecraft in the direction of its flight path due to re-entry forces traditionally contains significantly lower rates of change. Main frequency component extracted from these cases was 0.325Hz. Being the higher of the two, this was used as the highest frequency component for subsequent calculations.

7.1.2 Integration technique

Accelerations and angular rates must be integrated to obtain the various attitude pa- rameters. It is briefly worth mentioning that there are several options as to how and when this can be done. The first option is to download raw accelerations and angular velocities from the RHMS. Once obtained, these can then be integrated during post-flight analysis. This option can potentially involve large volumes of data if high sample rates are used, and may therefore be impractical.

54 7.2 Temperature measurement

Figure 7.2: MIR body angular rates during re-entry (8).

The second option is to perform the integration onboard the RHMS. Attitude pa- rameters can then be downloaded directly. This can lead to lower data volumes, but increases the load on the onboard processor. Although the raw data could be recon- structed to some degree, some resolution will inevitably be lost. Due to the scientific nature of this study, obtaining raw data is viewed as the preferable option.It is therefore intended that the integration will be performed during post-analysis. The feasibility of this option is examined in Section 7.6.

7.2 Temperature measurement

Temperature measurements onboard the RHMS serve two primary functions. The first is to gain an idea of the thermal environment directly surrounding the device. Recall that in the scenario considered here, the RHMS is located within the ATV cargo carrier. Measurements here will provide internal temperatures during re-entry, and a means of evaluating results from existing aerothermal models (37) (6) (2). The second function is to provide a means for measuring heat flux, and is covered individually in Section 7.3. Both functions require the use of thermocouples. These are ideal for use over a wide range of temperatures. They are also rugged, making them well suited to extreme thermal and aerodynamic environments. This section identifies the types of thermocouples that are relevant for this application, and how they can be utilised.

55 7. PAYLOAD

Choice of thermocouple type depends primarily on temperature range, but also the gaseous environment and cost. In the extreme thermal environment of re-entry, the type of material being measured provides the upper-temperature limit. This is usually the melting temperature of the material, since material temperatures tend to stagnate once their thermal capacity is reached. This means that the material of interest must first be identified.

7.2.1 Device housing and thermocouple type

For a small self-contained RHMS device, the outer protective housing might be the best material to measure. This should be designed to shield the device during the frag- mentation of the host, and to release the device at a convenient melting temperature. This material could also be used for temperature and flux measurements by housing thermocouples within the material. Aerothermal studies have indicated that internal ATV temperatures are expected to reach 1300K at the approximate time of the main fragmentation event (37). This temperature was compared with the melting temperature of some common structural materials in Table 7.2.

Material Tmelt[K]  [W/m.K] tp[s] Aluminium 850 250 Titanium 1950 21.9 Stainless steel 1700 19.0 CC - Reinforced 2144 1.70 Copper alloy 1356 401 0.0875 Fibreglass 1200 0.04 Graphite Epoxy 700 0.35 Iron 1812 60.0

Table 7.2: Melting temperature, thermal conductivity and penetration times for some common materials.

From observation it can be seen that the melting temperature of copper alloy is close to this requirement. Lower melting temperatures might release the device too early, while higher melting temperatures will release it too late. Its thermal conductivity is

56 7.3 Heat flux measurement also the highest of all materials considered. This was found to be of critical importance to the measurement of heat flux, and is discussed in Section 7.3. For these reasons, copper was selected for the housing material. Having identified the melting temperature of Copper, a suitable thermocouple type could then be selected. The common k-type thermocouple has a typical temperature range of 93-1573K. It is therefore suitable for this application.

7.2.2 Thermocouple amplifiers

Thermocouples measure the temperature di↵erence between two points, and typically have millivolt outputs. It is therefore usual to amplify the output, whilst also providing a temperature reference point, or cold junction. For s small device, it will be dicult and impractical to regulate temperature to zero degrees. A thermocouple amplifier with cold junction compensation is therefore necessary. Various COTS products are available, one being the AD8479, which is designed specifically for the k-type. Error- free temperature compensation is usually possible over ambient temperatures in the range 25 100 C. This infers that the thermal environment inside the RHMS must be regulated to ensure accurate readings. Due to the high range of temperatures being measured (up to 1500K for copper), the device will operate outside of it’s linear range (+/- 2C error for 55C to 565C). As a result, the non-linearity error must be accounted for. Since cold junction compensation is carried out on the amplifier’s integrated circuit board, error correction must be carried out onboard the RHMS, rather than during post-processing. Lookup tables can be used, and must be loaded onto the micro-controller.

7.3 Heat flux measurement

Heat flux is defined as the rate of heat energy transfer through a surface, with units W/m2. As discussed in Section 5.2, measurement of incident heat flux is necessary to accurately determine the re-entry aerothermal environment. Direct measurement of heat flux is possible using a number of COTS products. Recall from Section 5.2 that a maximum heat flux range of 330 500kW/m2 and maximum temperature range of 1100 1300K is expected in the ATV cargo bay during re-entry. These requirements clearly exceed the capability of the COTS instruments examined. A ruggedised indirect

57 7. PAYLOAD method of heat flux measurement was therefore found to be necessary for this appli- cation. Calorimetric heat flux instruments use thermocouples in their measurement and are therefore capable of operating in the temperature ranges stated above. With proper sensor design, the ’lumped system approach’ to transient heat transfer can be used to determine incident heat flux. This method and its application are discussed in the following sections.

7.3.1 The lumped system approach to heat transfer

Thermal analysis of a body such as a calorimetric sensor can vary in complexity de- pending on the temperature distribution throughout the body. In the simplest case, the temperature distribution throughout the body remains uniform as it undergoes tran- sient heat transfer. That is, temperature throughout the body is a function of time, and not position. These systems behave as ’lumped bodies’ and can be analysed using the ‘lumped system approach’ (38). To determine whether the lumped system approach is applicable to a particular body, an understanding of the relevant heat transfer mechanisms must first be under- stood. During an atmospheric re-entry for example, thermal energy is first convected to the body from the aerothermal environment. The thermal energy is then conducted within the body, with some small loss due to thermal radiation from the body to the surrounding environment. It is intuitive that high thermal conductivity within a body will contribute to a small temperature gradient in the body at a given time. It is also intuitive that small bodies will have smaller thermal gradients due to smaller conduc- tion lengths and faster heating. Therefore, the ideal body to undergo lumped system analysis is a small body with high thermal conductivity. This concept is captured in the dimensionless Biot number, which acts as an initial indicator of the applicability of the approach to a particular system. The Biot number is defined as:

h V h L Bi = c s = c c (7.1) A 

The characteristic length Lc is given by the ratio of sensor volume Vs to surface area A exposed to convection, as seen in Equation 7.1. For simple geometries like cylinders where only one circular surface is exposed to convection, this is simply the cylinder length.

58 7.3 Heat flux measurement

The Biot number represents the ratio of external heat convection to internal heat conduction. A small Biot number (typically Bi 0.1) therefore represents good heat  conduction within the body and a small temperature gradient throughout the material. If fulfilled, this criterion allows the approach to be used to approximate heat transfer. As a rough guide, fulfilling the Biot criterion will allow internal temperatures to remain within 5% of the ambient temperature (38). Material selection obviously plays an important role in internal thermal conduction, and hence the Biot number. Table 7.2 compares the thermal conductivity and relative penetration times for a number of common materials. Penetration time characterises the time taken for temperature changes to permeate a given volume of material, and is defined in Equation 7.2 (39). Shorter penetration times allow higher frequency flux changes to be resolved.

L 2c t =0.1 c v (7.2) p 

In cases where the lumped system approach is found to be an acceptable assumption, time-rate of change of the sensor temperature is the only measurable quantity necessary for heat flux measurement. With a set of periodic temperature measurements at equal time intervals, heat flux can be deduced by evaluating:

dT Cm q˙ = (7.3) cal dt A

The designation C is given to specific heat, since cv = cp for incompressible sub- stances (38). The specific heats for incompressible substances depend only on temper- ature, and not pressure. Hence for improved accuracy, C(T ) can be used instead of C = Const where C is evaluated at the average measured temperature. A comparison of these approaches is included in Section 7.3.2. Equation 7.3 provides only the heat flux measurable by the calorimeter probe. Losses must be accounted for in order to evaluate the total flux incident on the probe. Losses that should be considered include heat transfer through conduction between the probe and surrounding materials, as well as radiative heat transfer away from the probe. Neither of these losses are measurable by the probe and must be estimated to

59 7. PAYLOAD improve accuracy of heat flux measurements. These losses can be modelled as:

 q˙ = ✏T 4 + (T T ) (7.4) loss x env The first term represents radiation heat losses as per Kircho↵’s law, while the second represents conduction losses as per Fourier’s Law. The total aerothermal heat flux incident on the probe is then given by:

q˙aer =˙qcal +˙qloss (7.5)

Close consideration must be given to the design of the sensors to enhance the ability of error prediction. Details of how to accomplish this are beyond the scope of this work. Figure 7.3 however, shows the basic design of the calorimeter probe used in the Re- entry Aerodynamic Flow Experiment (RAFLEX) air data system. Of particular note is the spacing between the sensor head and surrounding structure, which is intended to reduce conduction losses.

Figure 7.3: Calorimeter design for the RAFLEX air data system (9)

7.3.2 Model Validation

The thermal analysis modelled in Equations 7.1 to 7.5 was evaluated in Matlab using a known temperature profile. The Matlab script used can be found in the electronic copy of this study. The temperature profile was recorded during the RAFLEX II

60 7.3 Heat flux measurement experiment onboard the Micro Re-entry Capsule (MIRCA) in 1997. The heat flux profile output from the Matlab model was then compared with the processed output from the RAFLEX experiment for validation. RAFLEX II consisted of an air-data system for determination of free-steam aerothermal parameters. This included the deduction of heat flux from a temperature profile generated by a calorimetric probe similar to that in Figure 7.3. Table 7.3 contains the probe parameters that were used in both the RAFLEX II experiment and the Matlab validation model.

Head material Copper Head diameter 10mm Head length 36mm Penetration time 1.13sec Thermocouple sample rate 1Hz

Table 7.3: Calorimetric head parameters used in RAFLEX II and Matlab validation model (10).

7.3.2.1 Error sources and assumptions

Before introducing the temperature profile and heat flux output from the model, it is worth noting any assumptions or possible sources of error. First, the convection heat transfer coecient h, introduces considerable uncertainty into the calculation of the Biot number. A 20% error in this coecient is often considered ’normal’ (38). A typical value of 200W/m2 was applied for a cylinder-air interface. It was shown for this geometry however, that values up to around 1100W/m2 will continue to fulfil the Biot criterion, which is outside the normal range of most interfaces. This supports the validity of the criterion in this case, but a more detailed knowledge of the coecient for this application is necessary. The most significant unknown in the heat flux calculation of the RAFLEX experi- ment is the specific heat model applied. Equation 7.3 indicates that calorimetric heat flux is influenced directly by specific heat, making this a possible source of error. To gain an idea of the sensitivity of this factor, a constant averaged value and a function C(T ) were both applied as in Figure 7.4. The emissivity factor in Equation 7.4 is also

61 7. PAYLOAD unknown for the RAFLEX experiment and was assumed. A value of ✏ =0.78 was applied and corresponds to black, oxidised copper at 38C.

COPPER: Specific heat relation to temperature 3.8

3.7

3.6 .K) 3 3.5

3.4

3.3 Speficif heat (J/cm

3.2

Model C samples 3.1 C(T) C(T ) av 3 200 400 600 800 1000 1200 1400 Temperature (K)

Figure 7.4: Specific heat of Copper as a function of temperature (9).

Finally, the physical environment that influences the conduction heat flux is un- known. Equation 7.4 indicates that the temperature of the surrounding environment (such as structural elements) influences this flux component. The contribution of this component was found to be small, and hence a rough estimate was found to be suf-

ficient. Tenv was therefore modelled as being 5% cooler than the head temperature as seen in Figure 7.5. A constant thermal conductivity  was also applied, and is considered a reasonable assumption since it varies only slightly with temperature.

7.3.2.2 Calculated heat flux output for the RAFLEX temperature profile

The temperature profile from the experiment was replicated for use in the Matlab model, and is shown in Figure 7.5. The temperature profile was recorded from calorime- ter KFA2, which is the second of three calorimeters onboard RAFLEX, and registered the highest absolute temperature during the experiment. For the Copper head defined in Table 7.3, the Biot number is calculated as Bi = 0.0180. Note that since Bi 0.1, the lumped system approach applied is considered ⌧

62 7.3 Heat flux measurement

Temperature profile for flux model 1200

1100

1000

900

800

700

600 Temperature (K)

500

400 Model temp. samples 300 Thermocouple temp. Environment temp. 200 0 50 100 150 200 250 300 350 400 Time (s)

Figure 7.5: Replication of temperature data generated by a calorimetric probe KFA2 during the RAFLEX II experiment on the MIRCA re-entry capsule (10). to be a valid assumption. The heat flux generated by the Matlab model in response to the RAFLEX tem- perature profile is shown in Figure 7.6. Indicated in the figure are each of the flux components that contribute to the total aerothermal heat flux.

7.3.3 Discussion

Figure 7.7 shows the RAFLEX total aerothermal heat flux experimental result for the KFA2 calorimeter following processing. The corresponding temperature profile for this calorimeter was shown earlier in Figure 7.5, and was used to generate Figure 7.7 as per Equation 7.5. On first observation, the Matlab output in Figure 7.6 generally appears to match the experimental data in Figure 7.7. The profile shapes are similar, with a gentle increase in flux to a maximum at approximately 200 seconds elapsed time, where temperature rate-of-change is at a maximum. This is followed by a reduction in heat flux. The zero-

63 7. PAYLOAD

Heat flux components 1500 q calorimeter q radiation q 1000 conduction q total q with C(T) total 2 500

0 Heat flux (kW/m

−500

−1000 0 50 100 150 200 250 300 350 400 Time (s)

Figure 7.6: Aerothermal heat flux output from Matlab model, with relevant components. crossings both occur at approximately 245 seconds, as the probe temperature maximises and begins to reduce. Minimum heat flux values then both occur at approximately 266 seconds. The minimum heat flux appears to be sustained in the experimental case however, compared with a sharper increase in the Matlab model. This can be attributed to the spline used to reproduce the calorimeter temperature profile in Figure 7.5. The spline smoothed the curve, accentuated the inflexion point and sharpened the point of minimum flux in the Matlab output. Upon closer inspection, there are some notable discrepancies in the flux outputs. The Matlab model shows a maximum flux of 1370kW/m2,whichis8.7% higher than that represented in the experimental results. Similarly, the Matlab model minimum is lower at 609kW/m2 compared with 400kW/m2 in the experimental results. Al- though the radiation flux contribution qradiation was considered as the primary source of the discrepancy, this is more likely the result of an overestimate of the specific heat value used in the Matlab model. Component qradiation would have significantly greater

64 7.3 Heat flux measurement

Figure 7.7: Post-processed aerothermal heat flux output from KFA2 calorimeter onboard RAFLEX (10). impact on the minimum than the maximum, as is clearly seen. It may however con- tribute to the larger error in the minimum value. To this point, there is no specific RAFLEX data available to verify the contribution of qradiation in the Matlab model for calorimeter KFA3. Radiation data is however available for calorimeter KFA1. The maximum temperature recorded at this probe was 1113K, compared with 1200K at KFA2. At 1113K the Matlab model gives a maximum qradiation value that is around 7.5% lower than that in the experimental case.

This indicates that the qradiation component in the KFA2 case is also likely to be an underestimate of comparable proportion. This is likely to contribute to a small error in the minimum heat flux value. It is also worth noting that the Matlab model indicates minimal di↵erence between using a constant specific heat taken at Tav and a function C(T ). In the critical regions of maximum and minimum flux, di↵erences of only 1-3 % are present between the models.

7.3.4 Applicability to the ATV re-entry environment

The temperature and heat flux data in the RAFLEX experiment are representative of a ballistic re-entry from LEO. The calorimeters were mounted in the heat-shield and

65 7. PAYLOAD remained oriented in the direction of flight. The temperature profile therefore varies smoothly throughout the flight time, leading to a heat flux output of similar character. With the RHMS mounted in the ATV cargo-carrier during re-entry however, a very di↵erent scenario emerges. Temperature changes may occur suddenly as the ATV structure is breached by hot plasma. The tumbling motion of the ATV may also introduce a frequency component into the temperature recordings. For this reason it is important to understand the calorimeter’s limitations in regards to these inputs. With a penetration time of 1.13 seconds, and a corresponding thermocouple sample rate of 1Hz, it is theoretically not possible to detect temperature changes at higher than 0.5Hz. Slower changes will however be discernible. While there is a great deal of uncertainty in any temperature frequency components that may exist inside the ATV, an approximation is possible from the discussion of attitude changes in Section 7.1.1. If temperature changes are related to spacecraft rotation during re-entry, a temperature frequency of around 0.325Hz may be inherited. Theoretically, these components are detectable. It may also be useful to make a direct measurement of the ambient temperature surrounding the RHMS using an additional thermocouple. This will aid in the deter- mination any higher frequency temperature components.

7.3.5 Conclusion

From this discussion it is evident that the Matlab model developed is valid to an acceptable degree at this early stage. Further development is however necessary for a more precise calibration. This should include an improved estimate of specific heat. Although a precise function C(T ) is preferable, it is evident that an appropriately selected constant value will suce. Additionally, the possible error in the radiation component can be rectified with a better knowledge of material emissivity, which is dependent on the probe material and the material finish applied to the surface.

7.4 Pressure sensor

Pressure sensor data is useful for a number of reasons. First, this data will help to identify key events throughout the re-entry. Structural fissures that develop in the ATV during re-entry can be identified by pressure changes. Separation of the RHMS

66 7.5 Data transceiver and Antenna from the ATV during fragmentation may also be visible through pressure data. At this point the RHMS will exit the flow region behind the bow shock and stabilise in a descent. This leads to the second practical use of pressure data. The normal shock relation in Equation 7.6 can be used to calculate the mach number of the device as it descends (40). This requires only a knowledge of pressure altitude, which can be deduced from the fragmentation altitude.

P2 2 2 1 = M1 (7.6) P1 +1 +1 The key requirement is that the sensor must have a dynamic range from vacuum to 1 atmosphere. The Measurement Specialists 86A fits this requirement, and was subsequently used for power and data volume calculations.

7.5 Data transceiver and Antenna

A data transceiver is required to upload the data volumes calculated previously in Section 7.6. It is responsible for establishing a data call with the Iridium satellite network, preparing data for transmission, and transmitting the data. The Iridium 9522B L-band satellite transceiver is the basic modem used for data communications. As discussed in Section 4.2.5, communications with LEO systems is preferable due to their low mass and power characteristics. The 9522B specifications re- flect these characteristics, and are shown in Table 7.4. The RS-232 protocol means the micro-controller or data logger must have a universal asynchronous receiver/transmitter (UART) or universal synchronous/asynchronous receiver/transmitter (USART) inter- face, both of which are common.

Mass: 420g Temp. range: 30 C to +70 C Frequency: 1616 - 1626MHz Protocol: RS-232 Supply voltage: 5V Power average (call): 4W

Table 7.4: Iridium 9522B Specifications (13)

67 7. PAYLOAD

An antenna is necessary to establish a data call. Most earth-bound antennas have a hemispherical coverage of the sky, giving horizon to horizon coverage if the antenna is pointed vertically. Iridium’s constellation of 66 satellites provides total earth coverage at sea-level. Each satellite uses 48 beams, giving the coverage pattern shown in Figure 7.8. Each collection of 48 spot-beams provides a footprint diameter of around 4700km at a half-beam angle of around 70. Spot-beams for all satellites overlap. A vertically oriented antenna therefore has access to between 1 and 3 satellites at any given time, anywhere on the earth’s surface (41).

Figure 7.8: Iridium coverage at sea-level using the 48 beams for all 66 satellites.

Two factors can then contribute to a reduction in coverage. The first is antenna orientation. By angling the antenna away from vertical, a portion of the hemispherical volume becomes redundant. A communications dropout will occur if the satellite in use is within this portion of the sky. Satellites near the antenna’s horizon are particularly susceptible to this type of coverage loss. The second factor is the altitude of the antenna above sea-level. As altitude increases, the footprint diameter of 4700km begins to reduce. As a result, the overlap between spot-beams reduces until a point is reached whereby ‘blind-spots’ develop. An antenna in one of these regions will also experience a communications dropout. Given the orbital velocity of the satellite constellation and

68 7.6 Data generation and transmission the rotation of the earth, the position of these points change over time. Additionally, these points will develop first in equatorial regions, where overlap is at a minimum. These factors can clearly impact the RHMS communication times calculated in Section 6.2.5.1. While a detailed examination of these factors is necessary, a brief study of the coverage reduction was carried out here.The aim was to approximate the reduction in spot-beam diameter at an altitude of 22km. Recalling the trajectory analysis summary in Table 6.3, 22km was selected as the point of vertical antenna orientation. For a satellite orbital altitude of 781km (42), spot-beam diameters were found to reduce by 2.8% to 4568km at 22km altitude. Although quantitatively small, it could not be established whether this would prevent the occurrence of small blind- spots. This calls for a more detailed simulation in more advanced iterations of this study.

7.6 Data generation and transmission

This section investigates the amount of data likely to be generated by the payload sensors. This is necessary to determine the data volume generated by various sampling scenarios, and to identify how this compares to the amount of data transmittable during the decent phase (see Table 6.4). This will influence how data is processed onboard. The IMU discussed in Section 7.1 was found to generate the highest portion of payload data. Accelerations and rotation rates can be sampled at high frequencies, leading to large data volumes. It is therefore the handling of this data that places the highest load on processing and transmission. Two theoretical sample rates were first calculated to form the upper and lower bounds of the possible range. The lower bound was first calculated using the Nyquist Sampling Criterion. The upper bound was then calculated based on the maximum amount of transmittable data. This range was then used to evaluate the feasibility of various data handling techniques.

7.6.1 Sampling rate lower-bound

The lower bound on possible sampling rates was approximated by identifying the Nyquist rate. The Nyquist rate is defined as twice the highest frequency component

69 7. PAYLOAD contained in the signal of interest. From the MIR and Ariane-5 cases studied in Sec- tion 7.1, a maximum frequency component of 0.325Hz was measured for rotation rate signals. The controlled tumbling of the spacecraft before re-entry is responsible for this frequency, and is also used during ATV re-entries. This infers that a sample rate of 0.65Hz makes up the lower boundary.

7.6.2 Sampling rate upper-bound

The maximum possible sampling rate was calculated using the maximum transmittable data volume, IMU sample packet size and recording time as inputs. The maximum amount of transmittable data was calculated in Table 6.4, and is around 354kb. While a small portion of this data can be dedicated to other payloads, it was assumed for the purposes of this calculation that the entire portion is dedicated to the IMU. The IMU sample packet size was determined from the ADIS16375 data sheet, and indicated that the gyro and accelerometer use a 32-bit output format for each axis (12). This equates to a total packet size of 192 bits for all three axes of the gyro and accelerometer combined. The total recording time is the last input required. From Section 5.5 this was defined as the time elapsed between ATV entry interface at 120km altitude, and the main fragmentation event. This was found to be around 420 seconds. Using this information, a maximum sample rate of 35Hz was calculated.

7.6.3 Estimated data volume generated

Using the upper and lower boundaries, a range of 0.65-35Hz was identified for possi- ble sample rates. There are both advantages and disadvantages to sampling in each extreme. By operating at the lower portion of the range, a smaller amount of data is generated. By increasing the time available for the transceiver to establish a data call, the likelihood of a successful data upload can also be increased. The obvious drawback is the inability to detect high frequency signal components in sensor outputs. Conversely, high sample rates lead to high data volumes. In the event of an extended transceiver connection time, ground-impact may occur before the full recorded data volume is recovered. For this reason, moderate sample rates in the range 10-20Hz were

70 7.7 Micro-controller and data storage considered to be preferable. Table 7.5 shows the generated data volume for a 15Hz sampling rate of accelerations and rotation rates.

Parameter Channels Record Sample rate bps Transmitted Accelerometer x,y,z 420s 15Hz 96 75.6kb Gyroscope x,y,z 420s 15Hz 96 75.6kb GPS pos/lon/lat 420s 1Hz 96 5.04kb Thermocouple T1,T2,T3 420s 1Hz 48 2.52kb Pressure P1 420s 1Hz 32 1.68kb Total: 160.44kb

Table 7.5: Data volume generated when IMU outputs are sampled at 15Hz.

The transmission of 160kb requires around 45% of the total possible transmission time of 4.8 minutes calculated in Table 6.4. The remaining 55% equates to around 157 seconds. This time represents a bu↵er in which the transceiver can repeatedly initiate data calls if dropouts occur. Otherwise, this additional time can be used to transmit additional descent data in the period after the main fragmentation event. The question was raised in Section 7.1 as to whether the sampling and transmission of raw accelerations and angular velocities was feasible. The large range of possible sample rates and corresponding data volumes show that this is possible. A detailed communication protocol should be developed in a later iteration of this study. This should include time stamps, headers and checksums.

7.7 Micro-controller and data storage

An onboard micro-controller is necessary to provide the degree of independence and automation that is required of the RHMS. The payloads discussed earlier in this sec- tion all produce data, which must be processed, stored and then transmitted at an appropriate time. Chapter 8 will go on to describe how the activation of these payloads must be scheduled to adequately regulate power consumption. These functions are best managed using a micro-controller. While selection of a specific device is beyond the scope of this study, this section aims to outline a few basic requirements relevant to the RHMS.

71 7. PAYLOAD

A number of interfaces will undoubtedly be necessary. Many typical data transceivers, including the Iridium 9522B, use the RS-232 protocol, which use the UART interface. Many modern sensors, including the IMU examined in this study, also use the serial peripheral interface (SPI). Electrically erasable programmable memory (EEPROM) is also required. Aside from the general coding that must be stored here, sucient space must also be dedicated to lookup tables, such as those required for the thermocouple amplifiers discussed in Section 7.2.2 External memory must also be added to store the data volume calculated in Section 7.6.3. These calculations indicate that only 160kb is likely to be transmitted. The total volume stored however, could exceed 50mb if full sensor resolution is stored throughout both the recording and descent phases of flight. Finally, programmable low-power/sleep modes with numerous interrupts is essen- tial. Chapter 8 will highlight the importance of sound power management during extended portions of the mission. E↵ective design of logic-statements to drive the activation/deactivation of these modes is critical.

72 8

Power

This chapter provides findings regarding the sizing of the RHMS power system. The instantaneous power loading is driven by various combinations of active payloads, which were discussed in Chapter 7. The duration of payload activation periods must also be closely managed however, and may vary widely for di↵erent mission profiles. This is necessary to balance power conservation with the consumption levels that are necessary during both the hibernation and active fight phases discussed in Section 4.2. Recall that a non-rechargeable battery was selected as the primary power source, on the grounds that a fully passive system is required. This emphasises the need for sound power management. The primary concern of this chapter is therefore to size the RHMS power system for the ATV re-entry scenario. This should ultimately result in a battery sizing with feasible mass and volume. Subsequently, it is desirable to extract any relevant subsystem or mission profile constraints that emerge as a result. Finally, other mission profiles beyond the ATV are considered.

8.1 Payload power consumptions

A rough knowledge of the power consumption of individual payload items is necessary before scheduling or battery sizing is considered. Sizing is clearly an iterative process, however a good first approximation can be achieved using typical consumption values for various COTS products available. Table 8.1 shows the expected power consumption from the various payload items. A single Wattage value is stated when available for

73 8. POWER the payload item under consideration, but is otherwise calculated from known current and voltage values. Payloads required to operate in other specific power modes are also provided. This includes the ’standby’ and ’detect’ power phases which are applied in Section 8.2. Standby power is used for payloads required to hibernate, while Detect power refers to power consumption during re-entry detection. Table 8.1 indicates a power consumption of around 8W is expected during normal operation, 0.13W in detect mode, and around 13mW in standby mode. The scheduling of these modes is discussed in Section 8.2. Note that a 25% margin was included in the power consumption summation. This accounts for any additional loads that are unforeseen in this initial study. These loads might include voltage converters and other circuitry, as well as unexpectedly high power consumption from payload items already included. Where possible, items with common voltage ranges have been used to minimise the number of voltage line conversions that are necessary.

8.2 Payload activation scheduling

Not all payload items are required during all mission phases. Hence, management of payload activation schedules is key to minimising unnecessary power consumption while not compromising on the likelihood of mission success. The duration of these schedules is highly dependent on the the timeline of the host spacecraft before re-entry occurs. The ATV-1 Jules Verne timeline is shown in Table 8.2 from undocking to re-entry. This timeline is used to devise payload activation schedules in subsequent sections. The timeline spans for around 24 days, and is fairly typical of other ATV re-entry timelines. The assumption can be made that the RHMS is powered-on immediately prior to undocking. This implies that a conservative payload schedule must span for around a month. A period of 30 days was therefore used in the schedule development. Four general power modes were used in the schedules. These included:

1. Standby: The longest duration mode. All payload items in standby/power-down mode for a predetermined amount of time. The exceptions are the wakeup sensor and micro-controller, which may periodically enter ‘re-entry detection’ mode to measure accelerations that could indicate a possible re-entry. In some scenarios this may be necessary due to unknowns in the host spacecraft’s re-entry schedule.

74 8.2 Payload activation scheduling - - - - - 100.0 25.03 0.1128 0.1251 Detect pwr. (mW) - - - - - 10.00 37.50 0.01251 Standby 0.004800 pwr. (mW) - 4000 1000 1597 571.0 3.000 12.50 800.0 7.983 Normal pwr. (mW) - - - - - 30 Sleep 0.004 Total (W): curr. (mA) Margin (mW): - - - 2.5 173 250 0.027 Normal curr. (mA) - - 3.3 5.0 5.0 1.8 5.0 (V) Voltage 1 1 3 1 1 1 1 # Used Model example ADIS16375 Iridium 9522B AD8497 ATmega640 - MS 86A DLR Phoenix Power consumption estimates for payload items in both normal and sleep modes where applicable. Table 8.1: Payload item IMU Data transceiver Thermocouple amp. Microcontroller Wakeup sensor Pressure sensor GPS

75 8. POWER

Event Date/Time Undocking 5 September 2008 (21:32 GMT) Departure burn 5 September 2008 (21:33 GMT) First orbit transfer burn 15 September 2008 (11:38GMT) Second orbit transfer burn 15 September 2008 (12:33GMT) First mid-course burn 19 September 2008 Second mid-course burn 24 September 2008 Orbital alignment burn 27 September 2008 Re-entry interface burn 28 September 2008 De-orbit burn 1 29 September 2008 (10:11 GMT) De-orbit burn 2 29 September 2008 (13:09 GMT) Re-entry at 120km 29 September 2008 (13:42 GMT) Impact of residual fragments 29 September 2008 (13:54 GMT)

Table 8.2: ATV-1 Jules Verne undocking and re-entry timeline (14).

Without periodic checks for re-entry, the event could go undetected. The length of time in detect mode has a significant impact on power consumption, hence several schedules will be considered.

2. Re-entry detection: Wakeup sensor permanently powered-on. Micro-controller enters a state of moderate power consumption as it monitors the wakeup sensor. Continual acceleration measurements are made to detect accelerations typical of re-entry.

3. Record: Re-entry accelerations are detected by the wakeup sensor. Micro- controller shifts to normal power mode, thermocouples, amplifiers, pressure sensor and GPS power-on. Solid-state memory begins recording data. Data transceiver remains o↵.

4. Record and transmit: Data transceiver powers-on, initiates an Iridium data call, and begins transmitting. Recording continues. This is the last mode and remains active until end of mission.

76 8.2 Payload activation scheduling

8.2.1 Low-power payload schedule

Table 8.3 shows the low-power payload activation schedule option. In this option, the micro-controller and wakeup sensor remain at standby power for the entire, pre- programmed 25 days. No re-entry detection is carried out during this period. An early unscheduled re-entry will therefore not be detected in this mode. A 6 Volt supply was used in all Amp-hour calculations, as is discussed in Section 8.3.

Power mode Period Watts Watt-hours Standby (0% detect) 25 days 0.0125 7.50 Re-entry detection 5 days 0.125 15.0 Record 7 min 3.00 0.350 Record and transmit 5 min 7.98 0.665 Total (W-hr) 23.5 Total (A-hr) 3.92 (@6V)

Table 8.3: Low-power payload schedule option. No re-entry detection is used in the standby phase.

8.2.2 Medium-power payload schedule

Table 8.4 shows the medium-power payload activation schedule option. In this option, a pre-programmed fraction of the standby mode is dedicated to re-entry detection. In this case, 50% of time in standby is spent in detect mode, resulting in higher power consumptions than the low-power option.

8.2.3 High-power payload schedule

Table 8.5 shows the high-power payload activation schedule option. In this option, re-entry detection is used throughout the entire 25 day standby phase. By comparing the Watt-hour and Amp-hour totals, it is clear that this option has the highest power consumption.

77 8. POWER

Power mode Period Watts Watt-hours Standby (50% detect) 25 days 0.0125/0.125 41.3 Re-entry detection 5days 0.125 15.0 Record 7 min 3.00 0.350 Record and transmit 5 min 7.98 0.665 Total (W-hr) 57.3 Total (A-hr) 9.55 (@6V)

Table 8.4: Medium-power payload schedule option. 50% of standby time is used for re-entry detection.

Power mode Period Watts Watt-hours Standby (100% detect) 25 days 0.0125/0.125 75.1 Re-entry detection 5days 0.125 15.0 Record 7 min 3.00 0.350 Record and transmit 5 min 7.98 0.665 Total (W-hr) 91.1 Total (A-hr) 15.2 (@6V)

Table 8.5: High-power payload schedule option. Re-entry detection mode is used through- out the entire standby phase.

8.3 Battery selection and sizing

A number of COTS batteries were considered to gain an approximation of the battery sizing required. The Energizer EA91 has a space heritage, having been used for exten- sively for impact detection in the space shuttle wing leading edge (43). The EA91 was therefore deemed suitable for this analysis, and has the specifications shown in Table 8.6. The operating voltages of the payload items listed in Table 8.1 indicate that supply voltages at or above 5 Volts are desirable. A 6 Volt supply was found to be convenient, given that four batteries connected in series will provide a 6 Volt output. This infers that groups of four batteries can be added in parallel until the overall power requirement is met. Table 8.7 shows the number of batteries required for each payload activation schedule examined in Section 8.2.

78 8.4 Discussion

Chemical system: LiF eS2 Volts: 1.5V Max. discharge: 1.5A Continuous Mass: 14.5g

Table 8.6: Energizer EA91 specifications (15).

Payload schedule # Batteries Amp-hrs Mass Low-power 12 4.50 174g Medium-power 28 10.5 406g High-power 44 16.5 638g

Table 8.7: Battery sizing for individual payload activation schedules.

8.4 Discussion

One key finding that emerges from the power analysis, is the imbalance in power con- sumption during the various flight phases. This is most apparent in the medium high- power payload schedules shown in Tables 8.4 and 8.5. In both cases, over 70% of power consumption occurs in the standby phase. The low-power case by comparison uses 30% in standby, where no re-entry detection is used within the first 25 days. This tends to indicate that a compromise between the low and medium-power schedules is suit- able for mission lengths of up to around 30 days. For longer mission lengths however, consumption becomes increasingly inecient. An independent power supply such as a battery, may become impractical at this point. This would increase the need for a shared power supply between the host spacecraft and the RHMS. A significant rise in total battery mass is also apparent as power consumption increases. This is due primarily to the use of re-entry detection for fractions of standby time in the medium and high-power cases. It is clear however, that mission lengths beyond 30 days will quickly blow-out battery mass to impractical fractions of total mass. Compounding this e↵ect, is the notion that periodic re-entry detection becomes increasingly important for longer missions, where orbital lifetime are highly variable. Again, this suggests that a shared power source would become preferable. It is also worth noting that the payload consumption rates used are approximate

79 8. POWER for this initial study. They were conservatively set to provide a worst-case scenario. A sound design could improve power eciency in low power modes, such as standby mode. Ultimately, this would also contribute to a reduction in battery mass.

80 9

Mass

This section provides an initial RHMS mass estimate. Findings are based primarily on the inclusion of the payloads discussed in Chapter 7, but also include mass margins for items not considered in this study and items that require a detailed analysis. Table 9.2 shows the complete mass table. A discussion of the contributions from the payload, TPS and housing are provided in subsequent sections.

9.1 Payload mass

Mass contributions were considered for all payloads included in the power analysis in Table 8.1. Battery sizing in Section 8.3 indicated that battery mass varies depending on the payload power schedule design and subsequently forms a significant fraction of the total RHMS mass. Table 9.2 summarises the estimated payload mass for each power schedule considered. Included in the payload mass calculation is a 25% margin to account for supporting or additional payload components neglected in this initial study. Such components might include wiring, PCBs, voltage converters, supporting structure, etc. The findings indicate that payload masses between around 1.4 and 2.0kg are expected.

9.2 Thermal protection system mass

Design of the TPS requires a detailed investigation that couples trajectory and thermal analyses, and is beyond the scope of this study. TPS mass fraction can however be significant, and is therefore estimated here. TPS mass depends primarily on spacecraft

81 9. MASS physical size, shape and the re-entry profile. Peak heat flux at steep re-entry angles is the primary driver behind material selection, while the total heat load throughout shallow re-entry angles drives material thickness. The Pico Re-entry Probe (PREP) and Deep Space II (DS-II) were both used in the estimation, having approximate masses of 0.85kg and 3.7kg respectively (44). A study of these spacecraft was conducted to determine their potential for science-return missions during earth re-entry (16). Initial conditions were entry altitude of 93km, inertial velocity 6.88km/s at entry angles of 0 and 39 .The two materials considered were silicone impregnated reusable ceramic ablator (SIRCA) and phenolic impregnated ceramic ablator (PICA) with maximum heat flux limits of 180W/m2 and 1200W/m2 respectively. The results from this study are summarised in Table 9.1.

DS-II DS-II PREP PREP

Entry angle 0 39 0 39 Stag. point max. flux [W/cm2] 78 271 48 177 Stag. point max load [J/cm2] 8641 2204 5317 1417 Material PICA PICA SIRCA SIRCA Thickness [cm] 4.85 1.05 1.39 0.29 Mass fraction 67% 15% 19% 4%

Table 9.1: DS-II and PREP re-entry results for TPS selection (16).

The high heat flux experienced by DS-II in the 39 entry angle was responsible for the selection of the PICA material. The maximum entry angle predicted for RHMS is however around half of this (see Section 6.2.3). Although the specific nose radius for the RHMS remains a variable at this early stage, it appears feasible that SIRCA will suce. The key constraint then becomes heat load, which drove the 19% mass fraction for a 0 PREP re-entry. Although only a 4 re-entry angle is predicted for RHMS, a conservative 20% mass fraction was applied in Table 9.2. This gives a total RHMS mass of between 1.7kg and 2.4kg.

9.3 Housing mass

This section calculates a rough estimate for the copper housing discussed in Section 7.2.1, with density 8.94g/m3. The assumption is made that the container is spherical

82 9.3 Housing mass with 300mm diameter and 2mm thickness. The volume of copper required for such a hollow sphere was calculated as approximately 670cm3. An additional 20% margin was added for mounting brackets etc, giving a total mass estimate of 5.99kg.

Payload item Low-power (g) Medium-power (g) High-power (g) IMU 25 ” ” Data transceiver 420 ” ” Antenna (transceiver + GPS) 312 ” ” Thermocouple amp. 10 ” ” Microcontroller 150 ” ” Wakeup sensor 5 ” ” Pressure sensor 11 ” ” GPS 20 ” ” Batteries 174 406 638 Margin (25%) 281 339 397 Total Payload [kg]: 1.41 1.70 2.00 TPS (20%) [kg] 0.282 0.340 0.398 Total RHMS [kg]: 1.69 2.04 2.39 Copper Housing [kg] 5.99 5.99 5.99 RHMS + Housing [kg]: 7.68 8.02 8.37

Table 9.2: Payload mass for di↵erent power scheduling scenarios.

83 9. MASS

84 10

Conclusion

This thesis examined the concept of using a small, autonomous re-entry health moni- toring system (RHMS) to record fragmentation data and return it for analysis. It was established early in the study that such a device could have numerous applications in a variety of space domains. Having identified a key research demand however; namely the need for calibration of existing fragmentation models; the study gained focus and scope. The work that ensued first involved identifying a mission concept and a corresponding set of system requirements. Great inspiration was taken from the flight-proven, but largely unpublished REBR concept. An analysis of the limitations imposed on such a system was then undertaken. The aim being to gain an understanding of its full capability under specific circumstances, then generalise about its suitability to other applications. In this final chapter, conclusions are first drawn regarding the specific scenario addressed in this study. Looking forward, these findings are considered in the context of future applications. Finally, comes a statement of future work.

10.1 Thesis contribution

A number of limitations and constraints were identified by using the ATV scenario as a case study. This essentially provided an RHMS performance envelope, giving a guide to the potential that the system holds in similar scenarios. Further, by looking at key constraints in a larger context, possible methods to exceed them became apparent. Although this might require a redesign or reconceptualisation in future applications, they are nonetheless a possibility. The key findings are summarised as follows:

85 10. CONCLUSION

1. Using an overhead satellite constellation for data acquisition holds good potential, but only for modest data volumes. In the specific scenario analysed, less than half of the transmittable volume was required. The threshold of 354kb over 4.8 minutes can be easily exceeded however if additional sensors or higher sample rates are required.

2. If the need arises to exceed the 354kb limit, a rethink of the free-fall descent phase is required. The addition of mechanisms from parachutes to rotors, will however add complexity and mass.

3. Accurate measurement of heat flux is a critical capability for model calibration. Calorimeters designed specifically to perform as a ’lumped-system’ show good potential as a possible method. Their performance can however be limited by assumptions regarding material properties and environmental factors.

4. Using batteries as a primary power source for this system configuration is real- isable, but incredibly inecient. Standby mode accounts for over 70% of power consumption. By contrast, only 1-4% is used for recording and transmitting data - the primary function of the device.

5. Minimising power consumption is critical in maintaining a fully passive RHMS design. The implication is that using a fully passive system places a definite limi- tation on its orbital lifetime. In its current configuration, the RHMS is incapable of operating in standby for durations exceeding a month. Clearly, this rules out LEO missions of 5 or more years duration, where battery mass fractions become unrealisable.

6. Point 5 infers that missions of long duration require a significant rethink of the RHMS architecture. The obvious solution is to consider sharing the host space- craft power bus. Potential consequences include increased cost, complexity and the possibility of certification issues for both parties; none of which are attractive to a possible vender. A more conservative approach would be to extend the pas- sive concept as far as possible, using improved power management, more ecient payloads, or high density power sources.

86 10.2 Looking forward

7. While the mass of the RHMS device itself is under the target of 4kg initially set, it was not foreseen that copper housing would also be required. This sets the combined mass above the 4kg target. This highlights the challenge of surviving the fragmentation process.

10.2 Looking forward

The RHMS concept has many possible applications. At the outset of this study it was identified that a definite need exists for a ’black-box’ for reusable re-entry vehicles. The Columbia disintegration highlighted a need that is growing with the large number of spaceplanes and commercial launchers in development. SpaceX’s Dragon, Scaled Composites’ SpaceShipOne and Reaction Engines’ Skylon are just a few examples of current designs. Although the direction of the study moved to focus on the need to calibrate existing fragmentation models, it is now necessary to return to these potential ’spino↵’ applications. A key di↵erence between RHMS variants for science-return and black-box applica- tions is data volume recorded. The few sensor outputs present in this study are likely to be replaced with hundreds of telemetry parameters from the host spacecraft. It is qualitatively easy to see that by recording these parameters at a moderate resolution, data volumes will far exceed the recoverable threshold via satellite identified in this study. A configuration change regarding the data recovery technique is therefore a likely requirement if large data volumes are a priority. This also raises the question of whether autonomous data recovery is even a prior- ity for reusable re-entry vehicles. The advantages are clear for science-return missions, where thin budgets and harsh environments constrain physical recovery. When con- sidering reusable re-entry vehicles however, there are added advantages to a recovery e↵ort, where debris can be collected and analysed. The RHMS could then be recovered as a matter of course, similar to the recovery of aircraft flight-data recorders. In such a case, key design drivers emerge from the need to survive both spacecraft fragmentation and ground impact. Also crucial, as was made clear in the aftermath of Columbia, is the ability to locate the device. These applications also suggest that the RHMS must become an integrated sys- tem within the host spacecraft. Both power requirements and the need to access host

87 10. CONCLUSION telemetry suggest that full passivity is no longer advantageous. A degree of indepen- dence may still be necessary to prevent fault propagation from the host through to the RHMS. Such a capability could draw inspiration from the system analysed in this study. Inevitably an array of other applications exist, from interplanetary probes to the testing of new thermal protection systems here on earth. The scenario and findings in this study can serve as a baseline for these endeavours, and underline areas that require further attention.

10.3 Future work

This study served as an initial design phase in RHMS development. As such, many assumptions, simplifications and oversights were inevitable. This final section aims to stress areas in further need of work as a direct result of this study. These areas include:

1. Power consumption - This was found to be the key constraint on mission duration and device mass/volume. While this is not anticipated to change significantly, an in-depth study of circuit design, power management techniques and sensor- specific consumptions is necessary to identify the full breadth of the issue. This could hold the potential for far-improved eciency in standby mode.

2. Autonomous re-entry detection - This is clearly a critical capability in the scenario examined in this study. The precise acceleration profile typical of a particular vehicle at 120km was not analysed here. Nor was the ability to distinguish gentle re-entry accelerations from impulse-like booster firings. While this shouldn’t be an intensive task, such algorithms are in need of development.

3. Satellite accesses - Due to time restrictions and limited access to tools, a detailed knowledge of RHMS accesses to the Iridium or similar satellite network could not be obtained. This work is necessary to either alleviate or underline the data transmission constraints due to RHMS heading, position and attitude.

4. Thermal protection system - While a preliminary and unverified thermal analy- sis was carried out and shown in Appendix A, a more accurate study is needed.

88 10.4 Final thoughts

This should include an estimate of one-dimensional heat transfer from the stag- nation point to aeroshell/heat-shield. Selection of specific TPS materials will be dependent on these findings.

5. Aeroshell design - This study found that a dynamically stable aeroshell is required for the scenario examined. A study of possible designs that will provide a stable descent from any initial orientation should be conducted. Future work could also examine the feasibility of using a lifting-body design to increase data transmission times.

6. Heat flux measurement - Further work is needed to reduce uncertainties and assumptions in the lumped-system approach to heat flux measurement. The degree of uncertainty compared to the accuracy needed for model calibration should be quantified.

7. COTS Sensors - In this initial study it was not the intention to identify spe- cific products for use. In later iterations of this study, specific sensors should be identified. These must be rated to operate in the space environment and to with- stand the rapid decelerations typical in this scenario. Any constraints placed on the power system by consumption characteristics could give way to an additional challenge.

8. Mass - The use of copper housing to control RHMS release makes up a signifi- cant portion of total mass. While this is seen as the most passive option, other techniques should be investigated to reduce the mass footprint.

10.4 Final thoughts

It is the view of the author that the RHMS concept holds good promise for scenarios both examined and envisioned. The endeavour is worthy, if only to prevent human casualty at the hand of falling debris in an increasingly mitigated and populated space environment; or equally from the sound design of future manned re-entry vehicles. Importance is also given however, to the notion of reducing operational costs by im- plementing aspects of a design-for-demise strategy. The Creation of this future relies on contributions from independent studies and corporate undertakings alike. Although

89 10. CONCLUSION the REBR has only recently taken flight, it is hoped and encouraged that findings from these flights make their way into the public domain. It is also believed that much is to be learned from the 50% success rate of REBR flights at the time this study was completed. It is hoped that by sharing information regarding likely failure modes a col- lective e↵ort can be made to advance the concept of reliable re-entry health monitoring systems.

90 References

[1] Sdunnus H. Klinkrad H. Wallker, R. Update of the ESA Space Debris Mitigation Handbook, 2002.

[2] Fritsche B. Lips, T. A comparison of commonly used re-entry analysis tools. Acta Astronautica, 57, 2005.

[3] ESA. ATV-2 Johannes Kepler Fact Sheet. Technical report, 2010.

[4] The Aerospace Corporation. REBR Fact-sheet, April 2011.

[5] The Aerospace Corporation. US Patent - Spacecraft Re-entry Breakup Recorder, May 2005.

[6] M. Homeister G. Koppenwallner H. Klinkrad T. Lips, B. Fritsche and M. Toussaint. Re-entry risk assessment for launchers - Development of the new SCARAB 3.1L.InEuropean Space Agency, (Special Publication) ESA SP, 2007.

[7] Fritsche B. Koppenwallner G. Leveau C. Lips, T. Ariane-5 EPC re- entry - A comparison of observations and SCARAB simulation results. 2007.

[8] Koppenwallner G. Fritsche, B. Computation of destructive satellite re-entries. 2001.

[9] Collocott S. J. White, G. K. Heat Capacity of Reference Materials: Cu and W. Journal of Physical and Chemical Reference Data, 13(4), 1984.

91 REFERENCES

[10] Koppenwallner G. Fritsche B. Eigner, R. M. Pressure and Heat Flux Measurement with RAFLEX II Experiment During MIRKA Re-entry. Number 426, 1998.

[11] Arianespace. Ariane 5 User’s Manual, 5 edition, 2008.

[12] Analog Devices. ADIS16375 - Six degrees of freedom inertial sensor, 2011.

[13] Iridium Communications. 9522B L-Band Data Transceiver, 2010.

[14] ESA. ESA Information Kit, ”Jules Verne” Automatmed Transfer Vehi- cle (ATV) Re-entry, September 2008.

[15] Inc Energizer Holdings. Product Data Sheet - Energizer EA91. Technical report, 2009.

[16] Kapoor V. B. Allen G. A. Venkatapathy E. Arnold J. O Ailor, W. H. and J. D. Rasky. Pico Reentry Probes: A↵ordable Options for Reentry Measurements and Testing. Technical report, 2007.

[17] Ailor W.H. Kapoor, V.B. The reentry breakup Recorder: A ”black- box” for space hardware. 2003.

[18] Columbia Accident Investigation Board. Columbia Accident Investi- gation Board - Volume 1. Technical report, 2003.

[19] NASA. NASA Technical Standard - Process for Limiting Space Debris, 2009.

[20] P. Waswa. Spacecraft Design-for-Demise Strategy, Analysis and Impact on LEO Space Missions. Master’s thesis, Massachusetts Institute of Technology, 2009.

[21] NASA Orbital Debris Program Office. Orbital Debris Re-entry, 2009.

[22] NASA Johnson Space Centre. DAS User Guide - Version 2.0, 2007.

[23] NASA Orbital Debris Program Office. Orbital Debris - ORSAT, 2009.

[24] Fritsche B. Koppenwallner G. Klinkrad H. Lips, T. Spacecraft destruc- tion during re-entry latest results and development of the SCARAB software system. Advances in Space Research, 34, 2004.

92 REFERENCES

[25] Lohle S. Marynowsky T. Rees D. Stenbeak-Nielsen H.C. Beks M.L. Hatton J. Lips, T. Assessment of the ATV-1 re-entry observation cam- paign for future Re-entry missions. 680 SP, 2010. cited By (since 1996) 0.

[26] Fritsche B. Lips T. Martin T. Francillout-L. De Pasquale E. Koppen- wallner, G. Analysis of ATV destructive re-entry including explosion events. Number 599, pages 487–492, 2005. cited By (since 1996) 0.

[27] NASA Space Operations. ISS On-Orbit Status 06/19/11, 2011. Last accessed on 8 August, 2011.

[28] Futron Corporation. Space Transportation Costs: Trends in Price Per Pound to Orbit 1990-2000, September 2002.

[29] Marichalar J. J. Johnson N. L. Rochelle, W.C. Analysis of reentry survivability of UARS spacecraft. Advances in Space Research, 34, 2004.

[30] Michael D. Griffin and James R. French. Space Vehicle Design. AIAA, 1991. Dynamics p231-254, thermal p255-268.

[31] T. Lips, G. Koppenwallner, L. Bianchi, and H. Klinkrad. Risk assess- ment for destructive re-entry, 2009.

[32] Dupzyk I. Shepard J. Newfield M. Ailor, W. REBR: An innovative, cost-e↵ective system for return of reentry data. 3, pages 2350–2357, 2007. cited By (since 1996) 0.

[33] Krag H. Lips T. De Pasquale E. Virgili, B.B. Simulation of the ATV re-entry observations. 680 SP, 2010. cited By (since 1996) 0.

[34] ICAO. Manual for ICAO Aeronautical Mobile Satellite (Route) Service - Part 2 -Iridium, 2007.

[35] NAL Research Corporation. Network Reference - Iridium Dial-up, 2011.

[36] Campbell Scientific Corp. 9522B - Iridium L-Band Data Modem, 2011.

93 REFERENCES

[37] Reynier P. Schmehl R. Marraffa L.-Steelant J. Boutamine, D. E. Computational Analysis of Automated Transfer Vehicle Reentry Flow and Explosion Assessment. 2007.

[38] Y. E. Cengel. Heat Transfer: A Practical Approach. McGraw-Hill Companies, 2002.

[39] G. Koppenwallner. Controlled hypersonic flight air data system and flight instrumentation. Technical report, 2007.

[40] A. H. Shapiro. The Dynamics and Thermodynamics of Compressible Fluid Flow, Vol. 1. Ronald, New York, 1953.

[41] D. Whelan. iGPS - Integrated Nav and Com Augmentation of GPS, 2010.

[42] Iridium Communications. The Global Network: Life Expectancy of Irid- iums Satellite Constellation. Technical report, 2010.

[43] Boeing Space Exploration. Wing Leading Edge Impact Detection Sys- tem - Presentation. Technical report, 2010.

[44] A. R. Howard. Miniaturisation of Atmospheric Entry Probes: Options for Future Planetary Exploration Missions. Technical report, 2009.

94 Appendix A

Additional Trajectory Results

Drag vs Time

2500 γ = −4 E γ = −10 E γ = −15 E 2000 γ = −20 E

1500 Drag (N) 1000

500

0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Time (min)

Figure A.1: Drag force versus time over full entry angle range.

95 A. ADDITIONAL TRAJECTORY RESULTS

g vs Time 9.9

9.85

9.8

9.75

g (m/s/s) 9.7

9.65

9.6

9.55 0 1 2 3 4 5 6 7 8 9 10 Time (min)

Figure A.2: Gravitational acceleration profiles over full entry angle range.

Ground track vs Time 400

350

300

250

200

150 Ground track (km)

100

50

0 0 1 2 3 4 5 6 7 8 9 10 Time (min)

Figure A.3: Ground track profiles versus time over full entry angle range.

96 Ground track vs Time 500

450

400

350

300

250

200 Ground track (km)

150

100

50

0 0 1 2 3 4 5 6 7 8 9 10 Time (min)

Figure A.4: Ground-track in lifting-body scenario for CL =0.05.

Stag. heating rate vs Time 500 γ = −4 E 450 γ = −10 E γ = −15 400 E γ = −20 E 350

300 ) 2

250 q (W/cm 200

150

100

50

0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Time (min)

Figure A.5: Stagnation-point heat flux profiles over full entry angle range.

97 A. ADDITIONAL TRAJECTORY RESULTS

4 x 10 Stagnation temperature vs Time 3 γ = −4 E γ = −10 E 2.5 γ = −15 E γ = −20 E

2

1.5

1 Stagnation temperature (K)

0.5

0 0 0.5 1 1.5 2 2.5 3 Time (min)

Figure A.6: Stagnation-point temperatures over full entry angle range.

Density vs Time 3

2.5

2 ) 3

1.5 Density (kg/m 1

0.5

0 0 1 2 3 4 5 6 7 8 9 10 Time (min)

Figure A.7: Comparison of exponential and ISA (black) density models.

98 Entry angle vs Time 0 γ = −4 E γ = −10 −10 E γ = −15 E −20 γ = −20 E

−30

−40

−50 Entry angle (deg) −60

−70

−80

−90 0 1 2 3 4 5 6 7 8 9 10 Time (min)

Figure A.8: Lifting body re-entry flight-path history for CL =0.05.

Lift vs Time 200 γ = −4 E 180 γ = −10 E γ = −15 160 E γ = −20 E 140

120

100 Lift (N) 80

60

40

20

0 0 1 2 3 4 5 6 7 8 9 10 Time (min)

Figure A.9: Lift force in lifting-body scenario for CL =0.05.

99 A. ADDITIONAL TRAJECTORY RESULTS

100 Appendix B

Ariane-5 Performance Envelope

Figure B.1: Ariane-5 sustained launch accelerations (11).

101 B. ARIANE-5 PERFORMANCE ENVELOPE

Figure B.2: Ariane-5 shock envelope across vibration frequency spectrum (11).

102