University of Southern California Dissertations and Theses

High temperature latent heat thermal energy storage to augment solar thermal propulsion for microsatellites

(USC Thesis Other)

High temperature latent heat thermal energy storage to augment solar thermal propulsion for microsatellites

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content HIGH TEMPERATURE LATENT HEAT THERMAL ENERGY STORAGE TO AUGMENT
SOLAR THERMAL PROPULSION FOR MICROSATELLITES
by
Matthew R. Gilpin
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(AEROSPACE ENGINEERING)
August 2015
Copyright 2015 Matthew R. Gilpin
Distribution A: Approved for public release; unlimited distribution. PA#15347
Dedication
ForMyParents -Engineersaren
0
tborn; they
0
remade:
ToKatharine -Sogladwemadeit! ii
Distribution A: Approved for public release; unlimited distribution. PA#15347
Acknowledgments
First and foremost, I must acknowledge the support I have received from the Air Force Research
Laboratory throughout my nine years at USC. I am incredibly grateful for the opportunities I have
received and the research career that I have been afforded. I especially need to thank Dr. Marcus
Young and Dr. David Scharfe for shepherding me through this project and giving a young intern
the chance to excel.
I also owe a great debt to the mentors that convinced me to pursue a Ph.D. A special thanks
goes to Dr. Anthony Pancotti for demonstrating a commitment to expanding the boundaries of
technology and pushing me to also look further. I also need to thank Dr. Taylor Lilly, Dr. Andrew
Ketsdever and Dr. Nate Selden. When I was first hired in CHAFF, I could not have imagined the
opportunity I faced, and as I reflect on my time at USC as an undergraduate, Masters and Ph.D
student, no other experience - either personal, professional or academic - has had a more profound
impact on my life goals.
In addition to the AFRL, this work would not have been possible without an extremely ded-
icated and capable team of undergraduate and graduate students. Matthew Orr, Martin Hilario,
Turner Topping, Rayed Kahn, Andy Su, Mallory Smith and Gavin Moler - your contributions were
essential to this research effort.
Finally, I must thank the ARCS foundation for their support in both my graduate and undergrad-
uate studies. By providing financial stability during times of great turmoil, you made it possible to
devote myself to advancing my research.
iii
Distribution A: Approved for public release; unlimited distribution. PA#15347
Table of Contents
Dedication ii
Acknowledgments iii
ListofFigures vii
ListofTables xi
Abstract xii
1 Introduction 1
1.1 Propulsion for High Performance Microsatellite Missions . . . . . . . . . . . . . . 1
1.2 State of Solar Thermal Propulsion . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Thermal Energy Storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Scope of Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 SolarThermalPropulsionHistory 7
2.1 Concept Proposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Initial Developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Bi-Modal Solar Thermal Propulsion . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Integrated Solar Upper Stage (ISUS) Program . . . . . . . . . . . . . . . . . . . . 16
2.5 Post-ISUS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3 Bi-ModalSolarThermalPropulsionforMicrosatellites 25
3.1 Solar Thermal Microsatellite Development . . . . . . . . . . . . . . . . . . . . . . 25
3.2 Bi-Modal Microsatellite Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3 Technological Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3.1 Solar Concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3.2 Fiber Optic Coupling and Pointing Accuracy . . . . . . . . . . . . . . . . 36
3.3.3 Thermal-to-Electric Conversion . . . . . . . . . . . . . . . . . . . . . . . 37
3.3.4 High Performance Thermal Insulation . . . . . . . . . . . . . . . . . . . . 39
3.3.5 Thermal Energy Storage Material . . . . . . . . . . . . . . . . . . . . . . 40
iv
Distribution A: Approved for public release; unlimited distribution. PA#15347
4 HighTemperatureLatentHeatThermalEnergyStorage 44
4.1 Phase Change Material Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2 Technological Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2.1 Latent Heat Application to Existing Designs . . . . . . . . . . . . . . . . . 47
4.2.2 Microsatellite Performance Impact . . . . . . . . . . . . . . . . . . . . . . 60
4.3 Existing Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.4 Research Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5 SolarFurnaceDevelopment 69
5.1 Initial Designs and Diagnostic Development . . . . . . . . . . . . . . . . . . . . . 69
5.1.1 Fresnel Lens Based Development Furnace . . . . . . . . . . . . . . . . . . 70
5.1.2 Commercial Diagnostics . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.1.3 CCD Flux Mapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.2 Final Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.2.1 DOTI Solar Concentrator . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.2.2 Heliostat Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.2.3 Characterization and Current Performance . . . . . . . . . . . . . . . . . . 85
6 MaterialStudiesandInitialExperiments 91
6.1 Material Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.1.1 Boron Compatibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.1.2 Silicon Compatibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.2 Preliminary Solar Furnace Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
7 PredictiveModel 101
7.1 Model Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
7.2 Solution Method and Boundary Conditions . . . . . . . . . . . . . . . . . . . . . 102
7.2.1 Radiation Shielding Integration . . . . . . . . . . . . . . . . . . . . . . . 105
7.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
8 MoltenSiliconTesting 116
8.1 Test Design and Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
8.2 100% Fill Factor Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
8.3 Expansion Damage Mitigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
8.3.1 Reduced Fill Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
8.3.2 High Density Graphite . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
8.3.3 Partial Freezing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
8.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
9 FutureWork 132
9.1 Convective Coupling Characterization . . . . . . . . . . . . . . . . . . . . . . . . 132
9.1.1 CSNR Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
9.1.2 Convective Coupling Optimization . . . . . . . . . . . . . . . . . . . . . . 137
v
Distribution A: Approved for public release; unlimited distribution. PA#15347
9.1.3 Convection Model Validation . . . . . . . . . . . . . . . . . . . . . . . . 139
9.2 Potential Solar Furnace Improvements . . . . . . . . . . . . . . . . . . . . . . . . 141
9.3 Developmental Roadmap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
10 Conclusions 144
ReferenceList 146
A MatlabModelFormulation 156
vi
Distribution A: Approved for public release; unlimited distribution. PA#15347
List of Figures
2.1 Ehricke’s original solar thermal spacecraft configuration . . . . . . . . . . . . . . 8
2.2 Solar thermal spacecraft design proposed in 1979 by the Air Force Rocket Propul-
sion Laboratory (AFRPL) and Rockwell International . . . . . . . . . . . . . . . . 11
2.3 Rocketdyne solar receiver concepts . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4 Cutaway drawing for the Rocketdyne solar thermal assembly . . . . . . . . . . . . 13
2.5 AFRPL solar thermal facility circa 1992 . . . . . . . . . . . . . . . . . . . . . . . 14
2.6 Receiver-absorber-converter (RAC) subsystem diagram from the Integrated Solar
Upper Stage (ISUS) program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.7 ISUS engine ground demonstration schematic . . . . . . . . . . . . . . . . . . . . 20
2.8 Experimental data for the ISUS RAC taken during hydrogen blowdown testing . . 21
2.9 Concept rendering of the Boeing / Air Force Research Lab (AFRL) Solar Orbit
Transfer Vehicle (SOTV) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.10 Flight scale concentrators developed during the Critical Flight Experiment (CFE)
program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.11 Diagram of experimental hardware from the NASA Shooting Star Experiment (SSE) 24
3.1 Rendering of the notional Surrey Space Center (SSC) solar thermal propulsion unit 28
3.2 Rendering of a microscale solar thermal deorbiting unit for a Disaster Monitoring
Constellation (DMC) satellite as proposed by Kennedy . . . . . . . . . . . . . . . 29
3.3 Fiber optic coupling configurations for a solar thermal spacecraft as proposed by
Nakamura et al. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.4 Single crystal molybdenum solar thermal receiver and prototype de-orbiting mod-
ule proposed by Sahara et al. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.5 Energy flow diagram for a bi-modal solar thermal propulsion and power system . . 34
3.6 Uncoated graphite components for the ISUS RAC . . . . . . . . . . . . . . . . . . 41
4.1 Temperature profiles for the “ISUS - Approximation” STAR-CCM+ Model . . . . 50
vii
Distribution A: Approved for public release; unlimited distribution. PA#15347
4.2 Mass flow averaged hydrogen exit temperatures for STAR-CCM+ heat exchanger
simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.3 Transient Hydrogen exit temperatures as a percentage of peak values . . . . . . . . 51
4.4 Temperature profiles for the “Boron - Constant Mass” STAR-CCM+ Model . . . . 53
4.5 Relative V delivery for representative heat exchanger models . . . . . . . . . . . 54
4.6 Temperature profiles for the “Boron - Length Reduction” STAR-CCM+ Model . . 57
4.7 Temperature profiles for the “Boron - Diameter Reduction” STAR-CCM+ Model . 59
4.8 Comparison of the proposed bi-modal solar thermal microsatellite vs. competing
technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.9 Silicon freezing model diagram presented in Chubb et al. . . . . . . . . . . . . . . 66
5.1 Path diagram for the Fresnel lens solar furnace . . . . . . . . . . . . . . . . . . . . 71
5.2 Photographs of Fresnel lens solar furnace components . . . . . . . . . . . . . . . . 72
5.3 Photographs of the USC heliostat tracking drive . . . . . . . . . . . . . . . . . . . 73
5.4 Black body calibration data for CCD camera . . . . . . . . . . . . . . . . . . . . . 77
5.5 Solar flux profiles for the Fresnel lens solar furnace . . . . . . . . . . . . . . . . . 79
5.6 Path diagram for a two stage solar furnace using a spherical concentrator . . . . . . 81
5.7 Photographs of the USC solar concentrator . . . . . . . . . . . . . . . . . . . . . . 82
5.8 Photograph of the USC heliostat mirror array . . . . . . . . . . . . . . . . . . . . 83
5.9 Solar furnace azimuth alignment vector . . . . . . . . . . . . . . . . . . . . . . . 84
5.10 USC solar concentrator coverage as a function of day of the year . . . . . . . . . . 85
5.11 Ray trace simulation of the USC solar furnace at the focal point . . . . . . . . . . 86
5.12 Relative power delivered to a 2.54 cm diameter spot vs. distance from the solar
concentrator. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
5.13 Flux map taken at the experimental location for the USC solar furnace. . . . . . . . 88
5.14 Power delivery vs. insolation for the USC solar furnace as a function of acceptable
spot size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
6.1 Tube furnace test articles to investigate silicon, graphite and boron nitride interaction 95
6.2 Steinfeld and Fletcher model geometry . . . . . . . . . . . . . . . . . . . . . . . . 97
6.3 Diagram of early solar furnace test assembly with radiation shielding . . . . . . . . 98
viii
Distribution A: Approved for public release; unlimited distribution. PA#15347
6.4 Photographs of radiation shielding test assembly . . . . . . . . . . . . . . . . . . . 99
7.1 MATLAB modeling geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
7.2 Ray trace plot illustrating view factor calculation method . . . . . . . . . . . . . . 110
7.3 Ray trace results - node to node view factors . . . . . . . . . . . . . . . . . . . . . 111
7.4 Ray trace results - shield escape percentage . . . . . . . . . . . . . . . . . . . . . 112
7.5 Thermal profiles calculated by the in-house MATLAB cooling model for the test
section given in Figure 7.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
7.6 Cut-away diagram of a cylindrical test article without phase change material (graphite
only) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
7.7 Temperature predictions for graphite control and silicon test sections . . . . . . . . 114
7.8 Comparision of experimental and modeling data for a graphite only test article . . . 115
8.1 Cut-away diagram of a cylindrical test article containing silicon PCM and boron
nitride liner . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
8.2 Photographs of test section construction . . . . . . . . . . . . . . . . . . . . . . . 119
8.3 Experimental cooling curves for the test section described in Fig. 8.1 . . . . . . . . 121
8.4 Photographs of 100 % fill factor tests . . . . . . . . . . . . . . . . . . . . . . . . . 122
8.5 Experimental heating curve for 100 % fill factor test sections . . . . . . . . . . . . 123
8.6 Comparison of MATLAB and experimental data for 100% fill factor tests . . . . . 124
8.7 Comparison of MATLAB and experimental data for 80% fill factor tests . . . . . . 125
8.8 Section photographs of an 80% fill factor test article with evidence of flowing silicon129
8.9 Cut-away diagram of a cylindrical test article with a graphite walled PCM cavity . 130
8.10 Section photographs of graphite walled test section post testing . . . . . . . . . . . 130
8.11 Comparison of MATLAB and experimental data for 80% fill factor, graphite walled
test sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
8.12 Experimental data for partial freezing tests . . . . . . . . . . . . . . . . . . . . . . 131
9.1 Temperature profiles for the “CSNR Design” STAR-CCM+ model . . . . . . . . . 135
9.2 Simulated heat exchanger temperature output performance for the proposed CSNR
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
ix
Distribution A: Approved for public release; unlimited distribution. PA#15347
9.3 Length optimization plots for Center for Space Nuclear Research (CSNR) heat
exchanger design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
9.4 Cutaway diagram of initial resistive heating test article . . . . . . . . . . . . . . . 140
A.1 Single Node . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
A.2 Adjacent nodes in the z direction . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
A.3 Adjacent nodes in the r direction . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
x
Distribution A: Approved for public release; unlimited distribution. PA#15347
List of Tables
2.1 LEO-GEO transfer analysis data reprinted from Ethridge for a space shuttle deliv-
ered 28,100 kg spacecraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.1 Potential high temperature insulation materials . . . . . . . . . . . . . . . . . . . 40
3.2 High temperature sensible heat storage materials . . . . . . . . . . . . . . . . . . 42
4.1 Relevant properties of typical phase change materials . . . . . . . . . . . . . . . . 45
4.2 Potential high temperature phase change materials . . . . . . . . . . . . . . . . . . 46
5.1 Component efficiencies of the USC solar furnace . . . . . . . . . . . . . . . . . . 89
xi
Distribution A: Approved for public release; unlimited distribution. PA#15347
Abstract
Solar thermal propulsion (STP) offers an unique combination of thrust and efficiency, provid-
ing greater total V capability than chemical propulsion systems without the order of magnitude
increase in total mission duration associated with electric propulsion. Despite an over 50 year
development history, no STP spacecraft has flown to-date as both perceived and actual complex-
ity have overshadowed the potential performance benefit in relation to conventional technologies.
The trend in solar thermal research over the past two decades has been towards simplification and
miniaturization to overcome this complexity barrier in an effort finally mount an in-flight test.
A review of micro-propulsion technologies recently conducted by the Air Force Research Lab-
oratory (AFRL) has identified solar thermal propulsion as a promising configuration for microsatel-
lite missions requiring a substantial V and recommended further study. A STP system provides
performance which cannot be matched by conventional propulsion technologies in the context of
the proposed microsatellite “inspector” requiring rapid delivery of greater than 1500 m/s V . With
this mission profile as the target, the development of an effective STP architecture goes beyond
incremental improvements and enables a new class of microsatellite missions.
Here, it is proposed that a bi-modal solar thermal propulsion system on a microsatellite plat-
form can provide a greater than 50% increase in V vs. chemical systems while maintaining
delivery times measured in days. The realization of a microsatellite scale bi-modal STP system
requires the integration of multiple new technologies, and with the exception of high performance
thermal energy storage, the long history of STP development has provided “ready” solutions.
xii
Distribution A: Approved for public release; unlimited distribution. PA#15347
For the target bi-modal STP microsatellite, sensible heat thermal energy storage is insufficient
and the development of high temperature latent heat thermal energy storage is an enabling technol-
ogy for the platform. The use of silicon and boron as high temperature latent heat thermal energy
storage materials has been in the background of solar thermal research for decades without a sub-
stantial investigation. This is despite a broad agreement in the literature about the performance
benefits obtainable from a latent heat mechanisms which provides a high energy storage density
and quasi-isothermal heat release at high temperature.
In this work, an experimental approach was taken to uncover the practical concerns associated
specifically with applying silicon as an energy storage material. A new solar furnace was built and
characterized enabling the creation of molten silicon in the laboratory. These tests have demon-
strated the basic feasibility of a molten silicon based thermal energy storage system and have
highlighted asymmetric heat transfer as well as silicon expansion damage to be the primary engi-
neering concerns for the technology. For cylindrical geometries, it has been shown that reduced
fill factors can prevent damage to graphite walled silicon containers at the expense of decreased
energy storage density.
Concurrent with experimental testing, a cooling model was written using the “enthalpy method”
to calculate the phase change process and predict test section performance. Despite a simplis-
tic phase change model, and experimentally demonstrated complexities of the freezing process,
results coincided with experimental data. It is thus possible to capture essential system behaviors
of a latent heat thermal energy storage system even with low fidelity freezing kinetics modeling
allowing the use of standard tools to obtain reasonable results.
Finally, a technological road map is provided listing extant technological concerns and poten-
tial solutions. Improvements in container design and an increased understanding of convective
coupling efficiency will ultimately enable both high temperature latent heat thermal energy storage
and a new class of high performance bi-modal solar thermal spacecraft.
xiii
Distribution A: Approved for public release; unlimited distribution. PA#15347
Chapter 1
Introduction
1.1 PropulsionforHighPerformanceMicrosatelliteMissions
The principle motivation for this research effort was a review of microsatellite propulsion technolo-
gies published by the Advanced Concepts Propulsion Group of the Air Force Research Laboratory
(AFRL) in 2009 [1]. This review highlighted the known utility and cost reductions possible when
using a microsatellite platform and identified a robust propulsion mechanism as an enabling tech-
nology for a new class of fast response missions. Enhancing propulsion capability increases the
range of orbit change maneuvers for what is inherently a lower cost, lower time to orbit spacecraft
and also increases the availability of launch options [2, 3].
The AFRL review proposed a mission profile for a 200 kg microsatellite “inspector” requiring a
1.5 km/s V capability and evaluated the readiness of various technologies to meet this aggressive
target. Chemical propulsion options in the form of miniaturized bi-propellant and monopropellant
thrusters were explored targeting high thrust, low power operation. While the miniaturized bi-
propellant system was seen as an ideal solution, it was noted that the technology was in extremely
early-stage development. Electric propulsion options were also explored but low power availability
on a microsatellite limits thrust levels, resulting in mission times measured in months and years
as opposed to days. When considering near-term technology, a high performance microsatellite
1
Distribution A: Approved for public release; unlimited distribution. PA#15347
must rely on relatively inefficient monopropellant chemical thrusters capable of meeting mission
duration requirements at the expense of reduced payload mass.
The review also highlighted the unique possibilities of a solar thermal propulsion system.
Solar thermal propulsion, even when using non-cryogenic propellants such as ammonia, main-
tains relatively high thrust levels while exceeding the efficiency of mono and bi-propellant chem-
ical systems. Thus, the mission requirements for a microsatellite “inspector” are met with only a
slight increase in mission time compared to non-chemical options while maintaining a reasonable
payload mass fraction. It was recommended that solar thermal microsatellite concept be further
explored as a potential near term solution to enable the proposed microsatellite mission.
1.2 StateofSolarThermalPropulsion
Solar thermal propulsion (STP) uses concentrated sunlight to add thermal energy to a gaseous
propellant which is then ejected through a conventional nozzle to produce thrust. This relatively
simple concept has been under development since the mid 1950s when it was first proposed as
a means of reducing launch mass by eliminating the need for a separate oxidizer. Despite over
fifty years of development, and a unique combination of thrust and efficiency, no STP systems
have flown to-date. Progress of STP technology has been limited by concerns over perceived
system complexity, storage of cryogenic propellants, power loss when in eclipse and the mission
impact of requiring a separate thermal collection system in addition to standard photovoltaics and
batteries.
The majority of early solar thermal development was focused on the deployment of large scale
spacecraft where solar thermal was used to compete with existing cryogenic bi-propellant tech-
nology. The proposed systems offered significant efficiency advantages but the scale required
the development of large solar concentrators and mechanisms for coupling collected solar energy
effectively into the working fluid. The thermal energy subsystem had to co-exist with conventional
2
Distribution A: Approved for public release; unlimited distribution. PA#15347
components on the spacecraft and as a result, solar thermal propulsion gained a reputation as a
promising yet complicated design.
A seminal report published in 1979 by Rockwell International was the first systems level effort
to define a 28,100kg solar thermal spacecraft and the findings became the basis for the next decade
of solar thermal development [4]. Work was focused on producing prototype solar receivers, heat
exchangers, inflatable solar concentrator designs and relevant ground test facilities with the goal
of a spaceflight demonstration. By 1989, the technology was declared feasible but budget cuts at
what was then the Air Force Rocket Propulsion Laboratory (AFRPL) terminated further study.
The next phase of solar thermal development focused on a bi-modal operational concept where
the thermal sub-system on board the spacecraft is used for electrical power generation in addition to
propulsion. This advancement greatly reduced the perceived complexity of a solar thermal space-
craft and the Integrated Solar Upper Stage program (ISUS) was begun to develop a flight ready
system by the close of the decade. While ultimately unsuccessful in mounting a flight mission,
the ISUS program did succeed in ground testing a prototype thermal energy receiver and storage
mechanism. In contrast to previous development efforts which viewed thermal energy storage as
unnecessary, the proposed bi-modal configuration necessitated the inclusion of graphite as a sen-
sible heat thermal energy storage material. Termination of the ISUS program marked the end of
large-scale solar thermal development. But, it should be noted that for the second time, STP was
put on hold for budgetary, as opposed to technological reasons.
In the aftermath of the ISUS program, solar thermal technology was explored as a potential
propulsion mechanism for microsatellites. Seeing microsatellites as a low cost, attainable launch
opportunity, researchers focused on simplicity in an effort to finally mount a space demonstration
of STP. Reduced power requirements resulted in inflatable solar concentrators being replaced by
rigid thin film designs and cryogenic propellants were neglected in favor of lower performance
yet more easily sortable propellants such as ammonia. Despite a unique and relatively low cost
proposal, STP development again stalled due to a lack of funding.
3
Distribution A: Approved for public release; unlimited distribution. PA#15347
In this work, the bi-modal concepts developed during the 1990’s and the recent microsatellite
efforts are combined to propose a bi-modal solar thermal microsatellite as the technological evolu-
tion of the STP concept. The deployment and the associated risks of an inherently novel spacecraft
architecture have prohibited the flight of any STP concept to-date. For such a system to be real-
ized, STP must provide significant performance benefits. In the context of a high-performance
microsatellite as proposed by the AFRL, a bi-modal solar thermal concept can provide signifi-
cant (35-60% discussed in this work) V increase over chemical propulsion while maintaining
the necessary responsivity.
Over half a century of STP development has produced “ready” solutions for the majority of
technologies required to develop a bi-modal solar thermal microsatellite and a technological review
has determined that a high performance energy storage mechanism is the only remaining hurdle to
the development of an effective STP spacecraft.
1.3 ThermalEnergyStorage
To date, all solar thermal propulsion efforts containing thermal energy storage have utilized the
sensible heat of materials such as graphite and boron carbide. Sensible heat thermal energy storage
can provide adequate performance for large spacecraft. However, on a microsatellite platform with
tight mass and volume constraints, the energy storage density is insufficient. Additionally, a bi-
modal architecture with radiatively coupled thermophotovoltaic converters is highly affected by
the large temperature swings inherent in sensible heat thermal energy storage.
Here it is proposed that a switch to latent heat thermal energy storage is required as the final
enabling technology for a high performance solar thermal microsatelite. With sufficient thermal
energy storage capacity, traditional photovoltaics and batteries can be removed from the spacecraft
reducing propulsion and power system mass. Additionally, the relative temperature stability of
latent heat release results in higher propulsive and electrical efficiency along with lower stresses
on spacecraft components.
4
Distribution A: Approved for public release; unlimited distribution. PA#15347
Transitioning to a latent heat system requires that a new class of thermal energy storage materi-
als be developed. Both boron and silicon are presented as ideal candidates due to a combination of
a high melting temperature and a high heat of fusion. In this work, boron is considered a long-term
high performance solution and silicon is presented as a near term development target with potential
terrestrial applications.
This effort is not the first to propose the use of silicon and boron as energy storage materials.
In the existing solar thermal propulsion literature, both silicon and boron have been mentioned as
thermal energy storage candidates in the context of broad conceptual studies. However, there have
been no thorough investigations due to the absence of existing research in addition to cost and
schedule constraints. For terrestrial applications, silicon is proposed as the “ideal” thermal energy
storage material for thermophotovoltaic systems [5], but the existing literature is limited to 1-D
studies and calculations based on basic material properties.
1.4 ScopeofResearch
The goal of this research is to determine the basic feasibility of high temperature latent heat thermal
energy storage and to facilitate the development of a high performance bi-modal solar thermal
microsatellite. An emphasis is placed on experimental demonstration as the existing literature
lacks discussion of practical engineering concerns and viable solutions.
This work begins with a thorough literature review tracing the history of solar thermal propul-
sion and the natural evolution of the proposed bi-modal solar thermal propulsion system. The
technological requirements of the proposed system are presented along with existing solutions
identifying high temperature thermal energy storage as the final technological limitation. Both sil-
icon and boron are identified as potential thermal energy storage candidates and two technological
comparisons are presented demonstrating the performance benefits which can be gained.
To conduct experiments with molten silicon, a new solar furnace facility was built. Both the
design and characterization are discussed. Concurrent with solar furnace development, container
5
Distribution A: Approved for public release; unlimited distribution. PA#15347
materials were investigated and initial solar furnace tests were performed highlighting the need
for multi-dimensional thermal modeling. This drove the development of an in-house MATLAB
cooling model using the “enthalpy method” to calculate the silicon phase change process.
Molten silicon tests were performed using the solar furnace facility and results were compared
against the in-house MATLAB model showing that macro system properties can be obtained using
relatively simple phase change treatment. Experiments identified asymmetrical freezing as well
as expansion damage to be the primary engineering limitations for high temperature latent heat
thermal energy storage and a reliable design condition was identified for cylindrical geometries by
reducing the fill factor of the phase change material container.
Finally, future research topics are presented with an emphasis on container development, long-
term contamination testing, and convective coupling characterization of a latent heat gas blowdown
system. Recent interest in nuclear thermal bi-modal space power systems is targeting silicon based
latent heat thermal energy storage which provides potential motivation to further investigate these
topics.
6
Distribution A: Approved for public release; unlimited distribution. PA#15347
Chapter 2
Solar Thermal Propulsion History
The concept of solar thermal propulsion has been under investigation since its proposal over 50
years ago seeking a combination of high thrust and high efficiency to reduce launch mass and
expand satellite operations.
Early solar thermal propulsion development was focused on large spacecraft and presumed that
solar thermal energy would be utilized purely as a propulsive mechanism. This resulted in highly
complex architectures and concerns about vehicle integration limited the scale of solar thermal
development and prevented in-space demonstration. Ground test activity was focused on proof of
concept solar thermal heat exchanger design and the feasibility of large scale solar concentrators.
In the early 1990s, the concept of a bi-modal solar thermal spacecraft was proposed where the
thermal subsystem could be used for both propulsive and electrical power. This concept, com-
bined with a general reduction in spacecraft scale, greatly simplified spacecraft designs. Individual
components began to reach technological maturity and test campaigns demonstrated hardware for
potential in-flight demonstrations. Ultimately, no missions were launched. However, bi-modal
system development identified solar thermal as an ideal candidate for small, high performance
satellites.
7
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 2.1: Ehricke’s original proposed configuration for a solar thermal spacecraft [6]. Note
that the location of the heat exchangers within the spherical concentrators requires extensive high
temperature, high pressure plumbing.
2.1 ConceptProposal
Krafft Ehricke first proposed the concept of a solar thermal rocket in 1956 as a way to reduce launch
mass and increase the energy supply for an orbiting satellite [6]. The proposed solar thermal rocket
consisted of three primary components: the solar collector, a receiver/absorber and a propellant
feed mechanism. The solar collector focuses energy into the receiver/absorber where the thermal
energy is transfered to a propellant before it is ejected through a nozzle. As originally proposed
by Ehricke, a prototype solar thermal spacecraft would have been relatively large and complex
with a gross weight in excess of 7000 kg, a liquid hydrogen propellant mass fraction approaching
70% and two 39 m diameter inflatable spherical solar concentrators capable of generating modest
propellant temperatures of approximately 1000 K. It is interesting to note that Ehricke’s original
proposal included a “solar battery” to provide electrical energy for the spacecraft. However, the
conversion method was undefined as Ehricke’s report pre-dates the use of spacecraft photovoltaics.
8
Distribution A: Approved for public release; unlimited distribution. PA#15347
Following Ehricke’s proposal, Electro-Optical Systems (EOS) conducted the first detailed
experimental investigation of the solar thermal concept in 1963 using funding provided by the
Air Force Rocket Propulsion Laboratory (AFRPL). This experimental effort successfully con-
ducted hot flow hydrogen tests with a solar heated molybdenum absorber and tungsten-rhenium
flow tubing. Using a 1.5 m diameter solar concentrator, EOS achieved hydrogen gas temperatures
approaching 2300 K, which corresponds to a theoretical I
sp
of approximately 700 s [7]. EOS
demonstrated the feasibility of the basic solar rocket concept. However, work was halted due to
both funding and technical concerns. In addition to large budget cuts which shifted development
money to a competing advanced propulsion concept, there existed a perception in the propulsion
community that the solar thermal rocket was an awkward design requiring extensive solar concen-
trators and long stretches of articulated plumbing. The EOS study scarcely approached the issue
of “vehicle integration” and the scientific community found outstanding technological concerns
sufficient to abandon the solar thermal rocket concept for over a decade [7].
2.2 InitialDevelopments
In the late 1970s, the advent of the space shuttle revived the solar thermal concept. The AFRPL saw
the highly capable shuttle as an ideal candidate for placing a complicated solar thermal spacecraft
into orbit and they believed that the space shuttle would provide a host of new mission scenarios
well suited to solar rocket technology [4]. In 1979, the AFRPL funded an extensive study at
Rockwell International and Ethridge’s final project report became the foundation for the next 15
years of solar thermal research. Ethridge proposed the use of off-axis parabolic concentrators
to focus sunlight into a centrally mounted receiver, greatly simplifying solar thermal spacecraft
design. As can be seen in Figure 2.2, the use of a central receiver eliminated the extensive plumbing
from Ehricke’s proposal and allowed decoupling of solar concentration from the orientation of the
spacecraft. Furthermore, off-axis parabolic concentrators offered improved optical performance
9
Distribution A: Approved for public release; unlimited distribution. PA#15347
Table 2.1: LEO-GEO transfer analysis data reprinted from Ethridge for a space shuttle delivered
28,100 kg spacecraft [4]. Note that engine performance numbers represent the authors use of future
estimates in 1979.
Engine Type LO
2
-H
2
Ion Solar 1 Solar 2
V , m/s 4270 5850 5850 4800
I
sp
, s 475 2,940 872 872
Trip Time, days 5 180 14 40
Payload to GEO, kg 9,250 20,000 9,300 13,200
over spherical designs and avoided plume impingement on the concentrators themselves and other
spacecraft components. Note that Ethridge did not include a mechanism for thermal storage.
In addition to examining the previously neglected issue of vehicle integration, Ethridge also
performed an in depth analysis of solar thermal propulsion for transfer between low Earth orbit
(LEO) and geosynchronous orbit (GEO). The LEO-GEO mission was analyzed using spacecraft
scaled by the space shuttle maximum separation weight (approximately 28,100 kg) and two solar
thermal designs were compared against competing technologies. The first solar thermal configu-
ration was optimized for a rapid trip time and the second increased the mission duration to reduce
the power requirements for the solar concentration system. Data from the Ethridge study, given in
Table 2.1, demonstrates the payload advantage solar thermal propulsion gains over chemical sys-
tems if slightly longer transfer times are permitted. Additionally, solar thermal systems maintain
maneuver durations well below those required by electric propulsion. Ethridge’s report concluded
that a solar rocket was possible with state-of-the-art technologies and specifically recommended
the near term production of an engine/absorber and prototype solar collector for evaluation [4].
Throughout the 1980s, AFRPL oversaw the construction of facilities to evaluate and experi-
mentally test the concepts laid out in Ethridge’s seminal report. The AFRPL contracted the Rock-
etdyne division of what was then Rockewell International to produce a prototype receiver and
thruster which was delivered to the AFRPL for testing in 1984 [8]. During the development of
10
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 2.2: Solar thermal spacecraft design proposed in 1979 by AFRPL and Rockwell Interna-
tional. The use of off-axis parabolic reflectors greatly simplified the design of a solar thermal
spacecraft by allowing a centrally mounted receiver and eliminating articulated plumbing. Draw-
ing reprinted from Ethridge [4].
the thruster at Rocketdyne, multiple thruster and heat exchanger concepts (shown schematically
in Figure 2.3) were considered for both indirect and direct heat absorption techniques. Indirect
or windowless concepts used solar radiation to heat a refractory heat exchanger which then trans-
fered heat to a flowing propellant. Direct and windowed concepts included the use of propellant
seedants to directly absorb solar radiation and transfer the heat to the propellant stream. Rotating
porous beds, vortex traps to retain seedants, and the use of “aerodynamic windows” to pass sun-
light directly to the propellant stream were all considered. While direct heating options showed a
possible performance benefit and were not limited by heat exchanger material properties, a simple
11
Distribution A: Approved for public release; unlimited distribution. PA#15347
(a) (b)
(c)
(d) (e) (f)
Figure 2.3: Rocketdyne solar receiver concepts. a) Windowless heat exchanger b) Windowed heat
exchanger c) Windowed vortex flow exchanger utilizing a vortex flow pattern to retain propellant
seedants and preserveI
sp
d) Windowed particulate concept which discharged seedants in the pro-
pellant stream e) Rotating bed concept f) Aerodynamic window proposed to implement a direct
heating concept with a simplistic design. Schematics reprinted from Shoji [9].
indirect heat exchanger design was chosen for cost, schedule and risk reasons. The thruster deliv-
ered to the AFRPL, shown schematically in Figure 2.4, used a coiled rhenium tube as a combined
absorber and heat exchanger. The design targeted an H
2
exit temperature of 2705 K, an I
sp
of
approximately 800 s and thrust of 3.7 N.
12
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 2.4: Cutaway drawing for the Rockeydyne solar thermal assembly. Originally printed in
Shoji [9].
To test the Rocketdyne thruster, the AFRPL was concurrently developing a solar furnace and
thrust stand facility. The initial furnace, whose specifications drove the development of the Rock-
etdyne thruster, proved unsatisfactory upon delivery and a replacement furnace was built based on
existing JPL designs. The 25 ft tall, 25 ft wide AFRPL solar concentrator, shown in Figure 2.5, was
constructed using over 228 facets and had a design power delivery of 24 kW. During characteriza-
tion, however, the furnace was only able to deliver 17.5 kW at the focal point - nearly 30% below
the input power specification for the delivered Rocketdyne thruster and receiver [10]. Despite the
low power level of the AFRPL facility, the Rocketdyne thruster was tested for over 65 hours “on-
sun” and achieved H
2
temperatures of 1810 K (67% of the target temperature) corresponding to an
estimatedI
sp
of 650 s [11].
During the ground test campaign, the AFRPL also investigated various concepts for deploy-
able solar concentrators to address optics, deployability and robustness against space debris and
13
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 2.5: AFRPL solar thermal facility circa 1992.
micrometeroid impact. Early in the AFRPL investigation an “electrostatically controlled concen-
trator” was proposed as a means to prevent concentrator deflation. This technique used embedded
electric grids to apply electrostatic forces which maintained concentrator shape and it was demon-
strated on a small scale [10]. Following these tests, development focused on designing concentra-
tors using an inflatable toroidal rim to provide primary support. Toroidal concentrator development
was successful in producing small scale reflectors capable of sustaining a suitable optic. However,
the supporting truss remained in the rudimentary stage at the close of the decade [10].
By 1989, project managers at AFRPL declared solar thermal propulsion a feasible technology
and plans were made for a fully integrated systems test in 1994 and an in-space demonstration
around 1997. Again, however, solar thermal propulsion faced laboratory budget cuts to the in-
house AFRPL program. Smaller projects still continued such as the development of direct absorp-
tion thrusters by Rocketdyne using porous disk targets. This effort leveraged existing AFRPL
14
Distribution A: Approved for public release; unlimited distribution. PA#15347
facilities and worked to capitalize on then recent improvements to refractory manufacturing tech-
niques [12]. It was noted during this period that the research funding environment became the
limiting factor for solar thermal propulsion and not the state of technological readiness [13].
2.3 Bi-ModalSolarThermalPropulsion
The next major development in solar thermal propulsion began in the early 1990s as an offshoot
of nuclear thermal technology. Nuclear thermal mission designers were considering a novel con-
cept that combined the propulsion and power generation subsystems of a spacecraft into a single
integrated system driven by a nuclear thermal energy source. Zubrin et al. proposed a “bi-modal”
approach which modified an existing 40 kWe nuclear-thermionic reactor so the core could be used
to directly heat hydrogen propellant in addition to providing electric power [14]. This configura-
tion was referred to as the Integrated Power and Propulsion Stage and allowed for similar weight
savings as electric propulsion integration while dramatically reducing transfer times due to the
relatively high thrust and efficiency of a hydrogen thermal rocket. The USAF Philips Labora-
tory (previously the AFRPL) outlined a set of performance requirements and investigated various
nuclear bi-modal systems suggesting that a bi-modal satellite bus could provide the operational
flexibility for the development of new military strategies and significantly reduce operating costs
with lower payload weights and increased satellite capabilities [15].
Despite promising technology, nuclear thermal research stalled due to funding cuts, challenges
with public perception, high regulatory costs and the difficulty of arranging nuclear ground tests
necessary for system validation [13, 14, 16]. As an alternative, BWX Technologies (now Bab-
cock & Wilcox) developed and patented the solar bi-modal concept which leveraged the extensive
investment of the previous decade in nuclear thermal propulsion, nuclear thermal power gener-
ation and solar thermal propulsion to conceptualize an integrated thermal stage [17, 18]. The
proposed solar bi-modal concept shared many technologies with its nuclear thermal counterparts
15
Distribution A: Approved for public release; unlimited distribution. PA#15347
and replaced the nuclear energy source with a central receiver to collect and store concentrated
solar energy for both thermionic electric conversion and propellant heating.
The AFRPL conducted a small study as a preliminary investigation into the solar bi-modal
concept showing that a bi-modal system held a payload mass advantage over conventional upper
stages while maintaining a highly flexible and responsive architecture [17]. More importantly, the
AFRPL identified the concept as a means to cheaply improve the economy of current space launch
systems during a period of low developmental and project budgets when the existing inventory
of launch vehicles were costly and offered limited performance. By drawing on the heritage of
previous research efforts, a solar bi-modal upper stage was seen as a way to quickly reduce the
cost of USAF space operations using near-term technologies [17].
2.4 IntegratedSolarUpperStage(ISUS)Program
To demonstrate the benefits of a solar bi-modal system (and bi-modal systems in general), the
aggressive Integrated Solar Upper Stage (ISUS) program was initiated in 1994. This collaborative
project was inspired by the Air Force Space Command’s identification of affordability and respon-
siveness as key operational deficiencies and was managed by the AFRPL with support from the
Naval Research Laboratory, Idaho National Laboratory, NASA Lewis Research Center, and vari-
ous defense contractors. Little existing research effort was targeting the modernization of upper
stages and ISUS was presented as an easily accessible breakthrough technology [16, 19].
The ISUS program’s primary focus was launch vehicle “stepdown” which would allow smaller,
yet similarly capable payloads to be launched into medium earth orbit (MEO), GEO and highly
elliptical orbits from smaller launch vehicles. The baseline mission considered in the early stages
of development suggested that a high performance solar bi-modal upper stage would allow a nom-
inal communications satellite, typically launched from a Delta II/7925 ($50M per launch in 1995),
to be “stepped down” to launch on a Titan IIG ($18-30M per launch in 1995) [16]. The idea of
a stepdown was made possible by a paradigm shift in upper stage design brought forth by the
16
Distribution A: Approved for public release; unlimited distribution. PA#15347
bi-modal concept. The orbit raising stage was typically seen as disposable, but the ISUS stage
would be integrated with the payload and used to provide power and propulsion to the spacecraft
throughout its lifetime.
The goals for the ISUS program called for relevant ground testing and development of ISUS
subsystems in the next two years and the flight demonstration of a militarily useful communications
payload on an approximately 2000 kg satellite in a 12 hour Molniya orbit by early 1998 [3].
These targets ultimately proved to be overly optimistic. However, the ISUS project did succeed in
providing a usable framework for future bi-modal development.
ISUS systems development targeted three primary areas of concern: the receiver-absorber-
converter (RAC), the hydrogen propellant feed system and the thermionic conversion subsystem.
Despite being able to draw heavily on both solar thermal and nuclear thermal research from the
previous decade, these technological areas still required experimental validation before launch.
Rigid, faceted solar concentrators similar to those developed for previous space based dynamic
power systems were selected early in the program as a way to reduce overall programmatic risk at
the cost of performance when compared to inflatables [3, 20].
The most compelling result of the ISUS program, with respect to the contents of this paper,
is the development and ground testing of a prototype RAC. The ISUS RAC unit, diagrammed in
Figure 2.6, was the first solar thermal concept to utilize a thermal energy storage (TES) material to
absorb and store concentrated sunlight. While the addition of TES to solar thermal propulsion had
been previously mentioned in the literature, it was not seen as essential to solar thermal technology
and was not thoroughly explored. Ethridge’s study postulated that the addition of TES would
simply reduce transfer times by allowing thrust during eclipse for non-impulsive burns [4]. Shoji
suggested that TES would allow for reduced concentrator size by allowing for “charging” of a solar
thermal engine during coast periods, however, only considering the reduction in concentrator size
resulted in a net mass increase from the addition of TES and the concept was abandoned [11].
17
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 2.6: Subsystem diagram of the ISUS RAC unit. Figure reprinted from Partch and Frye [21].
The inclusion of TES in the ISUS RAC design provided propulsive benefits and reduced con-
centrator size. However, the primary reason for inclusion was to provide constant electrical output
supporting the bi-modal concept. The ISUS RAC represented a completely thermal spacecraft bus
relying on TES as the primary means of energy storage and output.
The heat exchanger and absorber for the ISUS RAC was constructed using 20 kg of graphite
and 15 kg of protective rhenium coating. Graphite sensible heat thermal energy storage was
selected due to acceptable performance and the absence of previous development towards high
temperature latent heat materials. Previous investigations into latent heat thermal energy stor-
age existed from space based dynamic power system development. However, the thermal storage
media studied (LiF-CaF
2
) operated at temperatures too low to be considered for a solar thermal
rocket [22]. Graphite sensible heat TES was an existing technology and presented a low enough
technical risk to meet the aggressive program goals.
18
Distribution A: Approved for public release; unlimited distribution. PA#15347
A protective rhenium coating was required to suppress graphite reactivity in the hot hydrogen
environment and reduce carbon evaporation rates at STP temperatures. Investigations conducted
at BWX Technologies indicated that un-protected graphite would lose mass in sufficient quantities
to pose a danger to solar concentration surfaces and hydrogen flow integrity [23]. Rhenium was
shown to be the most promising coating due to a low vapor pressure, no reaction with hydrogen, a
lack of carbide formation, ductility at high temperatures and manufacturability.
In an effort to raise the ISUS system to a technology readiness level (TRL) of 6 and pave the
way for flight experiments, the ISUS engine ground demonstration (EGD) used a solar simulator
at what is now the NASA Glenn research center. A schematic of the “Tank 6” testing arrangement
is given in Figure 2.7. Original testing goals for the EGD were to integrate all aspects of the bi-
modal system (H
2
storage and feed, thermionic conversion, power management, space relevant
concentrator facets and RAC) into the ground test. However, technical and integration problems
resulted in testing only of the RAC hardware [24, 25, 26].
The ISUS EGD program culminated in hot flow hydrogen testing through the prototype RAC
and experimentally demonstrated the ability of the proposed heat exchanger to transfer heat from
the TES into a flowing hydrogen propellant. Data shown in Figure 2.8 shows how propellant
temperature (assumed during testing to be approximately the temperature of the exit tube) remained
relatively constant during the first seven minutes of the burn and approached the temperature of the
RAC receiver cavity.
Hydrogen blowdown testing showed successful thermal coupling, but temperatures remained
below program goals. The existing solar concentrator in NASA’s Tank 6 facility required upgrades
to provide enough power for the ISUS prototype RAC and the upgrades performed during the ISUS
project proved insufficient [25]. It was also found that the RAC and multi-layer insulation package
suffered from higher heat loss than design models predicted [27]. The EGD program achieved
only 56% of the desired hydrogen burns required to simulate the thrust profile of the reference
19
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 2.7: Schematic of the ISUS engine ground demonstration (EGD) experiment. Diagram
reprinted form Frye and Kudija [25].
flight mission and “on-sun” testing only reached propellant temperatures of 2150 K instead of
target temperatures of greater than 2300 K.
While the ISUS EDG succeeded in being the first ground demonstration of a solar thermal
system using thermal energy storage, it did not succeed in providing the necessary technological
validation to mount the ISUS space demonstration. Excluding follow-on thermionic testing at
the New Mexico Energy Conversion Research Lab, the ISUS program was brought to a close in
1998 [26].
20
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 2.8: ISUS RAC experimental data from hydrogen blowdown testing. RAC exit temperature
was measured at the outer surface of the exhaust tube which is assumed to be equivalent to the
hydrogen exit temperature. Graph reprinted from Frye and Kudija [25]
2.5 Post-ISUS
In the aftermath of the ISUS program, what is now called the Air Force Research Lab (AFRL)
began sponsoring two solar thermal development efforts under two separate divisions [28]. As a
direct continuation of the ISUS concept, the AFRL at Kirtland AFB provided Boeing (the then
owner of Rocketdyne) with funding to design the Solar Orbit Transfer Vehicle (SOTV) with the
intention of performing a flight validation of the bi-modal concept. Concurrently, the AFRL at
Edwards AFB, using funding through the Integrated High Payoff Rocket Propulsion Technology
Program (IHPRPT), sponsored Thiokol and SRS to perform a proof-of-concept test of an inflatable
21
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 2.9: Concept rendering of the Boeing/AFRL Solar Orbit Transfer Vehicle (SOTV).
Reprinted from Kennedy et al. [27].
solar concentrator known as the STP Critical Flight Experiment (CFE). In 2000, AFRL manage-
ment merged the two programs and replaced the rigid concentrator designed for the SOTV with
the inflatable concentrator under development by the CFE.
The Boeing SOTV , shown configured with the inflatable concentrator in Figure 2.9, was a
direct application of ISUS technologies in an effort to demonstrate the potential of solar thermal in
space. The smaller spacecraft was based on previous demonstration designs proposed during the
ISUS program when funding and scheduling gradually reduced the scope of the ISUS mission [29].
Intended for launch using a Taurus rocket, the SOTV was purely a demonstrator of bi-modal solar
thermal technology and would provide the necessary validation of hydrogen propellant handling,
solar concentrator performance and control, as well as thermionic conversion under microgravity
[30].
The CFE program was a continuation of in-house inflatable development which had been ongo-
ing at Edwards AFB since the mid 1980s. Researchers strove to produce an inflatable concentrator
of sufficient optical quality and to also develop a functional pointing mechanism. SRS and Thiokol
22
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 2.10: Flight scale concentrators developed during the CFE program. Reprinted from
Holmes [31].
succeeded in producing 2 m x 3 m and 4 m x 6 m inflatable concentrators, seen in Figure 2.10, and
tested the control system at the component level [28]. The program was set to culminate with a
fully integrated test of the inflatable concentrator, pointing mechanism and a direct gain tungsten
engine. However, before these tests were completed, the program exceeded its budget and devel-
opment was halted. It is important to note that the concentrators produced by the program were
considered to be “optical quality” and could have been used in a demonstration mission [31].
Concurrent with developmental efforts funded by the AFRL, NASA Marshall sponsored the
Shooting Star Experiment (SSE), which attempted to mount a flight demonstration of a solar ther-
mal engine using a porous heat exchanger and a deployable Fresnel lens-based solar concentrator.
This concept of a porous direct gain heat exchanger was investigated in the early 1990s by the
Phillips Laboratory and the SSE experiments added rhenium coated graphite foam thermal energy
storage [3, 32]. The use of thermal energy storage allowed for significant thrust levels within the
power confines of a Fresnel lens based concentrator. This system was tested using electric heaters
23
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 2.11: Cutaway diagram of the Shooting Star Experiment (SSE) hardware showing the loca-
tion of the rhenium foam heat exchanger and thermal energy storage. Reprinted from Tucker and
Salvail [32].
and was demonstrated “on-sun” at the NASA Marshall Solar Thermal Test Facility showing that
a rhenium foam provided a highly efficient heat exchanger configuration [33]. Despite significant
progress, funding for further testing and mounting a flight experiment did not materialize and the
project was closed.
The SOTV/CFE and SSE programs marked the end of large scale solar thermal demonstration
development. Much as Ethridge had defined the scope of solar thermal development through the
1980s, the ISUS program and bi-modal development was the primary driver throughout the 1990s.
In the early 2000s, the solar thermal concept became the focus of smaller investigations which
attempted to miniaturize the concept and apply it to microsatellites in the hopes of finally mounting
an STP space experiment.
24
Distribution A: Approved for public release; unlimited distribution. PA#15347
Chapter 3
Bi-Modal Solar Thermal Propulsion for
Microsatellites
After the ISUS program, solar thermal development shifted towards smaller projects targeting low-
cost flight demonstration on a microsatellite platform. The concept of a solar thermal microsatellite
was investigated at the Surrey Space Center (SCC) and their work defined both the potential per-
formance and the general spacecraft architecture. Based on the results of the SCC effort, the Air
Force Research Laboratory proposed that solar thermal propulsion was a promising candidate for
a high performance microsatellite [1]. Additionally, it was suggested that operating in a bi-modal
configuration could eliminate some of the operational concerns preventing the launch of previous
flight efforts. A review of the technologies required shows that, with the exception of high per-
formance thermal energy storage, the components necessary for mounting a successful bi-modal
solar thermal microsatellite exist within the current technology.
3.1 SolarThermalMicrosatelliteDevelopment
The ISUS program and subsequent follow-on efforts explored the utility of solar thermal propul-
sion in small satellite packages (200-400 kg) but failed to mount a flight demonstration. Kennedy,
25
Distribution A: Approved for public release; unlimited distribution. PA#15347
who had previously worked on the ISUS program, proposed in the year 2000, that microsatellites
(10-100 kg) were an overlooked option for finally providing experimental verification of the solar
thermal concept and that a low cost solar thermal pathfinder mission may be feasible [27]. The lack
of high V propulsion options for microsatellites made the addition of a solar thermal unit even
more compelling. A three year investigation at the Surrey Space Center (SSC) at the University of
Surrey sought to analyze the effects of a solar thermal engine on the microsatellite platform and
provide the experimental basis for a long awaited in-space experiment.
Microsatellites offer a low cost and relatively rapid development option for mission design-
ers. However, a lack of significant V capability limits their utility. Additionally, microsatellites
launch as secondary payloads with sub-optimal orbit insertion requiring hundreds of meters per
second of V for correction. Kennedy proposed that a V capability of 1-3 km/s would be an
enabling technology and dramatically increase the microsatellite operating envelope. Analysis
for multiple candidate missions showed that a robust V would allow for transfer from geosyn-
chronous transfer orbits (GTO) to geosynchronous orbit (GEO) (1,760 km/s), near earth escape
(700-1,770 km/s) and insertion into lunar orbits (2,103 km/s) [3]. Preliminary analysis showed
that this V target was well within the capabilities of a microsatellite scaled solar thermal propul-
sion system with< 50 % propellant mass fraction [34].
Previous solar thermal concepts hoped to optimize I
sp
by using hydrogen propellant, but
microsatellite weight and volume constraints preclude the use of cryogens. As an alternative,
Kennedy proposed the use of either hydrazine (N
2
H
4
) or ammonia (NH
3
) which both have a high
storage density, low molecular weight and a theoreticalI
sp
of approximately 400 s [34]. Assuming
that the theoretical performance can be achieved, Kennedy showed that STP gained an advantage
in both volume and efficiency over currently available microsatellite scaled chemical propulsion
systems. A comparison was also made against electric propulsion systems which theoretically
offer very large V capabilities (> 10 km/s). However, the low power levels available from a
26
Distribution A: Approved for public release; unlimited distribution. PA#15347
microsatellite platform would result in extremely low thrust levels and maneuver durations would
be measured in years rather than days.
After completing initial mission and systems analysis, work at SSC focused on experimental
design of a reusable solar thermal receiver to demonstrate feasibility. Limitations on concentra-
tor size and system complexity for a microsatellite platform necessitated the use of a pulsed solar
thermal rocket relying on sensible heat thermal energy storage (TES). The TES system consisted
of a packed bed heat exchanger using pellets of boron carbide (B
4
C). The volumetric heat trans-
fer benefits of a packed bed concentrator and resistance to thermal shock and expansion stresses
outweighed the disadvantages of a high pressure drop across the heat exchanger and the potential
for propellant “channeling” [3]. B
4
C was selected since it has an approximately 20% specific heat
advantage over graphite. However, reactivity concerns with the hot propellant stream required that
the B
4
C particles be coated with a protective layer of BN. A schematic of the target design is given
in Figure 3.1 showing a rigid fixed concentrator and self contained receiver, thermal energy storage
unit and thruster positioned at the focal point. This architecture intended to use the attitude control
of the spacecraft to position the solar concentrators for “thermal charging” maneuvers.
SSC produced two prototype receiver units with the purpose of evaluating material compatibil-
ity and sealing methods; the Mk I unit was consistent with the notional design and included thermal
storage elements. The Mk II unit was a scaled down spiral channel flow design that removed ther-
mal energy storage in order to reach higher overall system temperatures. Both test sections were
heated under vacuum using a central graphite resistive heating element in lieu of solar energy input
and multiple hot flow tests were conducted using ammonia propellant. The core material for both
the Mk I and Mk II receivers was a TiB
2
/BN intermetallic composite and it was determined that
flanged seals using graphite foil gaskets were the most promising construction option. While the
presence of graphite foil seals presented reactivity concerns, the small graphite surface area limited
reaction rates with the incoming propellant stream to manageable levels [3].
27
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 3.1: Rendering of the notional SCC solar thermal propulsion system with an integrated rigid
concentrator. Images taken from Kennedy [3].
The Mk I and Mk II receivers were both heated without propellant to temperatures approaching
1600 K and 2000 K respectively and showed little damage apart from the precipitation of boron
oxide binder on the surfaces of TiB
2
sections. However, during hot flow tests as chamber pressures
approached 140 psi, cracks formed in both test articles presumably due to stress concentrations at
ceramic thread locations.
The Mk II receiver was tested using ammonia propellant on a coarse resolution thrust stand
and demonstrated thrust levels in excess of 500 mN at an I
sp
of 237 s. Lower than predicted
efficiencies were attributed to propellant leakage at the highest testing temperatures due to loss of
seal integrity stemming from thermal expansion [34]. When experimental testing was completed,
it was suggested that the majority of testing issues arose from the improper engineering of ceramic
structures and that there were “ready” solutions for outstanding sealing problems [27].
Possible candidate missions for flight testing as secondary payload would necessitate a more
complicated system than the notional design tested during the ground campaign at SSC. Missions
28
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 3.2: Rendering of a microscale solar thermal deorbiting unit for a Disaster Monitoring
Constellation (DMC) spacecraft. The single mirror is sized to fit within the the DMC separation
ring and is fiber-optically coupled. Figures taken from Kennedy [3].
such as a proposed proof of concept de-orbiting unit on a Disaster Monitoring Constellation (DMC)
satellite must address packaging concerns, sun facing surface area and pointing accuracy available
on a non-purpose built spacecraft bus. Kennedy suggested the possibility of a 75 W solar thermal
system on board the DMC spacecraft with a 112 m/s V capability. The propulsion unit, shown
in Figure 3.2, would be capable of lowering the DMC spacecraft into a safely decaying orbit in
approximately 13 days using a series of apogee burns [3].
In addition to experimentally demonstrating the base feasibility of a microscale solar thermal
receiver, work at SSC evaluated the capabilities of diamond turned aluminum and formed PMA
solar concentrators coupled with fiber optic power transmission. Fiber optic development at SSC
continued through 2006 and focused on the solar tracking and receiver cavity designs required
to mount a feasible system [35]. It was found that utilizing multiple smaller concentrators saves
weight at the cost of increasing system complexity. Each smaller solar concentrator requires its
own tracking system and a low weight, low complexity tracking unit was proposed to provide
pointing accuracy on the order of 0.05

. The proposed mechanism was called Fiber Optic Image
Position Determination (FIPD) and was designed to assume tracking control for each mirror assem-
bly after the spacecraft bus provides an initial rough alignment. In this system, the target for each
29
Distribution A: Approved for public release; unlimited distribution. PA#15347
solar concentrator is a bundle of optical fibers. Each fiber has an externally mounted photodiode to
measure the photoluminesence of the fiber and give a relative intensity measurement. By compar-
ing these relative intensity measurements, the location of the solar image on the fiber bundle can be
determined and the tracking mount can be moved in a closed loop system. As a result, a notional
system was designed and the locating ability of the fiber optic bundle was proven. However, only
rudimentary tests were performed with the integrated tracking system [36].
As with solar thermal efforts before it, the work at SSC did not result in a flight test of solar ther-
mal hardware. The expense, risk, and rise in operational complexity associated with the addition of
the solar thermal propulsion unit to an existing satellite prevented the inclusion on any upcoming
missions. Despite this, the SSC effort did succeed in identifying the microsatellite as a promising
candidate for solar thermal development and proposed STP as a solution for the demonstrated need
for high V capability on the microsatellite platform.
In parallel with work at SSC, two smaller efforts also investigated the possibility of microsatel-
lite scale solar thermal propulsion. A group at the Physical Sciences Inc (PSI) headed by Nakamura
and funded through the AFRL in the aftermath of the SOTV program performed an experimen-
tal investigation into fiber optic power coupling. Unlike Kennedy’s investigation which looked
at power transfers of less than 5 W, Nakamura demonstrated power transfer of up to 200 W
and achieved fiber optically heated receiver temperatures in excess of 1400 K [37]. The work
at PSI made use of experimental hardware previously developed for in-situ lunar material pro-
cessing experiments and showed promise as a means of directly delivering heat to the core of a
STP receiver. Along with experimental results, Nakamura also published useful sizing metrics to
estimate the total concentrator and fiber optic weights for a 1 N thruster using either H
2
or NH
3
propellant. Using rigid concentrators, Nakamura estimated that combined concentrator and fiber
optic weights for an NH
3
system would be 9.46 kg for an on-axis configuration and 8.43 kg for an
off-axis configuration. Both of these designs are shown schematically in Figure 3.3.
30
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 3.3: On-axis (left) and off-axis (right) fiber optic solar thermal configuration diagrams re-
printed from Nakamura et al. [37].
Outside of the United States, the Advanced Space Technology Research Group at JAXA also
targeted solar thermal as the “most promising” propulsion design for an orbit transfer vehicle [38].
Their research focused on the development of lightweight solar concentrators and the construction
of a solar thermal engine out of single crystal refractory metals. The most interesting aspect of their
work was the creation of thin film polyester and polyamide solar concentrators with area densities
as low as 180 g/m
2
. These concentrators were vacuum formed at temperature onto paraboloidal
glass molds, taking into account forming error, and allowed to set for several days to relax stresses
in the material. This effort resulted in reflectors with concentration ratios exceeding 10,000:1 and
vibration resistance sufficient for space launch. Additionally, direct gain receivers were experi-
mentally tested on-sun at atmospheric pressures and propellant temperatures of up to 2000 K were
recorded. The JAXA effort combined thin film concentrators and the proposed STP engine into a
sample de-orbiting module, shown in Figure 3.4, that was intended for flight on board the newly
developed 50 kg-LabSat bus. However, the experiment never launched.
31
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 3.4: Single crystal Mo solar thermal receiver and prototype de-orbiting module with
receiver unit mounted centrally inside a thin film solar concentrator. Photographs taken from
Sahara et al. [38].
3.2 Bi-ModalMicrosatelliteConcept
In 2009, the Advanced Propulsion Concepts Group in the AFRL’s aerophysics branch published a
review of high thrust, high V propulsion options for microsatellites and identified solar thermal
propulsion as a promising candidate for high performance missions [1]. The baseline mission for
the review was a microsatellite “inspector” that could rapidly be diverted from its existing orbit
and rendezvous with another satellite for close proximity operations. Estimates for this mission
scenario expected a V requirement of around 1 km/s and an ideal thrust of at least 1 N to ensure
rapid mission response. Citing Kennedy’s solar thermal microsatellite development effort, the
review found that STP offered a compelling combination of efficiency and thrust that could achieve
greater V capability than chemical propulsion options with a minimum decrease in responsive-
ness. It was the recommendation of the authors to pursue further research in what was found to
be an enabling technology in “early stage” development [1]. These recommendations proved to be
the genesis of this research effort.
Budget issues aside, the historical failure of solar thermal research programs to produce a
flight mission stems primarily from the inherent complexity that accompanies mounting a novel
propulsion mechanism. Furthermore, the large scale of previous development projects positioned
32
Distribution A: Approved for public release; unlimited distribution. PA#15347
the spacecraft within a design space where the benefits of a solar thermal system did not outweigh
the risks to both cost and mission assurance. Large scale STP systems necessitated the use of
cryogenic propellants to gain significant advantage over more conventional chemical rockets and
the electrical power available on board a large spacecraft bus allows for electric propulsion systems
with adequate thrust for conventional mission scenarios.
In contrast, a strong case is made for solar thermal technology on a microsatellite platform.
A highly aggressive microsatellite requiring substantial V is currently unable to generate the
necessary power for a relatively high thrust electric propulsion system. Additionally, mass and
volume limitations require propulsive efficiency beyond what is offered by small scale chemical
rockets.
It is proposed that a bi-modal solar thermal microsatellite could serve as a low cost, high per-
formance platform for propulsion intensive mission scenarios. Such a satellite would take the
microscale solar thermal concept originally presented by Kennedy and combine it with a means
of thermal electric conversion. This would entirely replace the electrical power subsystem (photo-
voltaics and batteries) of conventional satellites with an entirely thermal-based architecture, effec-
tively reducing complexity. If coupled with a suitable high energy density thermal storage medium,
the thermal energy system would function as a conventional satellite bus and provide significant
mass savings. Whereas previous small scale solar thermal missions have been viewed as pure
demonstrations for the solar thermal concept, the performance benefits of a bi-modal system would
create an operational, low cost platform within the same mass and volume constraints.
In order to realize a low-cost, high performance microsatellite with rapid deployability, the
proposed bi-modal system architecture, illustrated in Figure 3.5, must be designed so that the
use of a novel technology has minimal mission impact. Despite higher potential performance,
cryogenic propellants have been ruled out due to insulation, pressurant and volume constraints on a
microsatellite platform. Instead, ammonia is proposed as the ideal propellant due to ease of storage,
acceptable performance and self pressurization to reasonable chamber pressures. While hydrazine
33
Distribution A: Approved for public release; unlimited distribution. PA#15347
Incoming Sunlight
`
Solar Concentrator
Fiber-optic
Transmission
High Temperature
Phase Change TSM
Insulation
Losses
Resonant
Mesh IR Filter
Thermophotovoltaics
Waste
Heat
Electricity
Output
Propulsion
Power Out
Heat Exchanger
Ammonia
Propellant
Figure 3.5: Energy flow diagram for a bi-modal solar thermal power system capable of providing
both propulsive and electrical power output.
may offer higher performance, the requirements for a pressurant increase system complexity and
its toxicity complicates testing and development.
Furthermore, in order to avoid mission impact imposed by the solar concentration system, fiber
optics must be used to decouple the solar concentration mechanism and the solar thermal engine
from the spacecraft attitude and sufficient thermal energy storage must be available to provide
continuous power to all subsystems despite changing solar flux while in orbit. If these system
goals can be realized, a novel, high-performance microsatellite will be enabled.
34
Distribution A: Approved for public release; unlimited distribution. PA#15347
3.3 TechnologicalRequirements
In order to realize a bi-modal solar thermal system at the microsatellite scale, many technological
requirements must be met. These include the ability to produce low mass, high performance solar
concentration systems, efficient fiber optic power coupling, high specific mass thermal-electric
energy conversion, advanced insulation materials and high performance thermal energy storage.
Due to the long development history of solar thermal propulsion, the majority of these technologies
have already been investigated and many of them have reached technology readiness level (TRL)
5-6.
3.3.1 SolarConcentration
First and foremost, a solar thermal energy system requires high performance solar concentration.
The development of solar concentrators for STP has a long history and it appears that concen-
trator options suitable for microsatellite use are currently available. The work of Sahara et al. at
JAXA in particular has experimentally demonstrated sub 200 g/m
2
concentrators with the 10,000:1
concentration ratios required for target temperatures of 2500 K [21, 38, 39]. Since a microsatel-
lite based solar platform would require less than 2 m
2
of total concentrator area, the use of self-
supporting thin structures is possible. With the exception of the ISUS program and its heritage
in rigid deployable concentrators from the Solar Dynamic Ground Test, the majority of histori-
cal concentrator development has focused on the development of inflatable concentrators for large
scale satellites [20]. Inflatable structures allow promising packaging possibilities and area densi-
ties of < 1 kg/m
2
have been experimentally demonstrated at the surface uniformity required for
radio transmission. Published data, however, does not include demonstrations of concentrators
with sufficient optical quality required for solar thermal applications as documented RMS slope
errors approach 0.5 degrees [3, 36, 40, 41]. Deployable rigid structures proposed by the ISUS
program had area densities of 2.5 kg/m
2
which was improved to 1.5 kg/m
2
through changes in
manufacturing techniques [25, 42]. These figures are for the mirrored facets themselves and do not
35
Distribution A: Approved for public release; unlimited distribution. PA#15347
include tracking systems, deployment mechanisms or other structures. Harris Corp, the original
manufacturer of the ISUS program concentrator array, estimated that these assemblies will add an
additional 1.5 kg/m
2
based on large 10 m
2
systems [43]. It is expected that the additional weight
estimates would be less dramatic for a micro-scale system. With minimal packaging and support
structure, the concentrator’s proposed by Sahara et al. can be used as a low mass, high performance
system and it is believed that little research effort is required in solar concentrator development
beyond packaging for a specific satellite design if inflatable structures are not required.
3.3.2 FiberOpticCouplingandPointingAccuracy
The next technological requirement for a bi-modal STP system is fiber optic power transmission
and concentrator pointing capability. Fiber optics allow for a de-coupling of solar thermal hard-
ware from spacecraft attitude, which is required if the bi-modal satellite is to be used for anything
beyond a technological demonstration. The use of a thermal energy storage design does afford the
possibility of using the spacecraft attitude control system to perform “charging” maneuvers before
burns. However, the operational impact of multiple re-positioning maneuvers limits the utility of
the platform. Work by Nakamura has demonstrated the possibility of fiber optic transmission of
concentrated sunlight at the laboratory scale and achieved total system efficiencies (sunlight to fiber
exit) of 38% using un-optimized hardware [37]. Nakamura estimates that a realistic space qual-
ified system could raise the total system efficiency to more than 70% by careful engineering and
better material selection such as higher reflectivity concentrators and switching to low-OH opti-
cal fibers. Henshall’s work at SCC also demonstrated a lab scale fiber optic transmission system
and showed an end-to-end efficiency of approximately 50% including Cassegrain concentrators
[36]. It is believed that the transmission efficiency predictions made by Nakamura are realistic and
that there are certainly improvements that can be made to increase end-to-end efficiencies through
careful engineering with existing technologies.
36
Distribution A: Approved for public release; unlimited distribution. PA#15347
It is generally proposed that a pointing accuracy of 0.1

will provide acceptable solar concen-
trator performance on-board a spacecraft and is achievable with current technology [4, 3, 44]. In
terms of body pointing capabilities, it is possible to achieve an order of magnitude higher than the
required performance as demonstrated by space based observatories such as the Herschel and Plank
missions operated by ESA [45]. Microsatellites, such as Surrey Satellite Technology’s 165 kg
CFESat are currently capable of pointing accuracies on the order of1

and the remaining point-
ing capability can be made up using commercially available pointing systems and position sensing
devices. An example of a commercially available unit is the Type 22 Antenna positioner assembly
from Moog which can achieve minimum step sizes of 0.02

(noting that the 5 kg system weight
could potentially be reduced by re-design to exact specifications) [46]. It should also be noted that
work at SSC on Fiber Optic Image Position Determination provides a road map for a low com-
plexity solar tracking solution [36]. There appears to be no fundamental limits to achieving the
pointing accuracy required for a solar thermal collection system and careful engineering should
yield the required performance.
3.3.3 Thermal-to-ElectricConversion
If the solar concentration system is to be the only source of energy for proposed bi-modal space-
craft, a means of thermal-to-electric energy conversion must be employed. Due to the high tem-
peratures involved in solar thermal propulsion, an effective system must operate with a hot side
temperature in excess of 1600 K while retaining a high energy conversion density. If only cur-
rently realizable technologies are considered, the conversion options are limited to thermionics
(proposed for ISUS), closed brayton systems (proposed for the Solar Dynamic Ground Test) and
thermophotovoltaics [47]. Both thermionic conversion and closed brayton systems have strong
developmental histories. However, they exhibit poor scaling down to microsatellite power lev-
els [47, 48]. Dynamic systems, in particular, require a condenser stage and the ability to remove
sufficient heat from the working fluid is problematic within the confines of a microsatellite [49].
37
Distribution A: Approved for public release; unlimited distribution. PA#15347
Thus, thermophotovoltaic (TPV) conversion is selected as the most promising candidate for
a bi-modal microsatellite. TPV power conversion uses photovoltaic cells that are optimized for
IR radiation from a hot radiating body. The key to peak power efficiencies in a TPV system is
matching the emission of the radiating body to the peak efficiency wavelengths of the photovoltaic
cell.
The use of TPV conversion has seen great improvements in the last two decades. In the early
1990s, work at McDonnell Douglass in partnership with NASA Lewis tested TPV cells coupled
with selectively emitting radiating bodies producing experimental cell efficiencies on the order
of 30% and solar-to-electric efficiencies specified between 12-20% depending on temperature of
the solar receiver [50, 51]. In the early 2000s, EDTEK, with funding from the California Energy
Commission, US DoD and the US DoE, investigated thermophotovoltaics as a commercial power
generation option when coupled with their patented selective IR filter technology. EDTEK devel-
oped what they called a “resonant mesh IR band pass filter:” essentially a gold screen precision
etched with a geometric pattern. At the wavelengths of interest, the geometric pattern, coupled
with the electrical properties of the material, created a resonant condition in which the sheet func-
tioned as an antenna - absorbing the desired wavelengths on one side and re-emitting them on the
other at efficiencies approaching 70% [52]. Wavelengths outside of the desired band gap were
reflected back to the radiating body at an efficiency of 98%. Using these resonant mesh filters,
EDTEK predicted solar-to-electric efficiencies on the order of 30% with theoretic TPV cell effi-
ciencies approaching 48%. Unfortunately, these numbers were based on extrapolated data from
low power experiments as operator error lead to a failure of the advanced testing hardware [52].
In 2000, a review of space power systems by Hyder specified the efficiency of a TPV system
to be approximately 19% with a specific power density of 15 W/kg including radiators; this was
comparable to the advanced three junction photovoltaic panels of the time [47, 53]. The end-to-end
efficiency of thermophotovoltaics are tied to the efficiency limits of advanced photovoltaic cells
and there is reason to believe that the specific power density of TPV technology has significant
38
Distribution A: Approved for public release; unlimited distribution. PA#15347
potential for improvement. A parallel is the specific power density growth seen during the same
period in conventional photovoltaics. By 2012 the specific power density of PV panels had risen
from 15 W/kg in 2001 to 100 W/kg in 2012 and it is believed that similar improvements can be
made in TPV system [54].
3.3.4 HighPerformanceThermalInsulation
The remaining technological requirements for a bi-modal STP system concern the receiver-
absorber-converter (RAC) which must receive, store and distribute collected solar energy. The
high temperatures needed for solar thermal propulsion require the use of advanced insulation as
well as ceramic and refractory construction materials. All materials must be able to operate for
sustained periods at temperatures between 1500 - 2600 K and careful consideration is required to
avoid adverse material interactions and degradation.
High temperature insulation materials currently exist that can provide acceptable performance
at STP temperatures. Table 3.1 lists candidate insulation materials and relevant properties. A
notable example with a long development history is carbon bonded carbon fiber (CBCF), which
was developed as insulation for NASA RTG projects [55, 56]. Other carbon based insulation such
as carbon foams and rigid networks of carbon fibers can achieve similar or better performance and
multiple ceramics are available with moderate performance but comparatively high density [57].
A new type of refractory and ceramic doped aerogel is also currently under investigation at the
AFRL in conjunction with this research effort.
In addition to solid insulation materials, the use of multi-foil insulation is common in high
temperature applications and low emissivity vacuum gaps are critical to TPV systems [52, 55]. An
appropriate insulation system will combine radiation shielding, vacuum gaps and solid insulation
to ensure minimal heat loss. Special attention must also be paid to material compatibility and
reactivity. However, a workable solution is certainly achievable with the existing state of the art.
39
Distribution A: Approved for public release; unlimited distribution. PA#15347
Table 3.1: Potential high temperature insulation materials. Thermal conductivity values given in
italics are interpolated from available data.
Material
k
th
@ 1000

C k
th
@ 1500

C k
th
@ 2000

C Density
Reference
[W/mK] [W/mK] [W/mK] [g/cm
3
]
Silicon Carbide 45 30 25 3.2 [58]
Boron Nitride 17-33 22.5 18 1.8 [59, 60]
Alumina 6.5 6.6 – 3.8 [59]
1-D Vacuum Gap 1.5 3.8 7.7 – [61]
Zirconia 2 2.5 3 5.5 [59]
Rescor 760 0.93 1.2 1.4 4 [62]
ONRL CBCF 0.17 0.2 0.26 0.2 [55]
Calcarb CBCF 0.2 0.35 0.65 0.18 [63]
Aerogel Filled
0.25 0.4 0:75 0.07 [64]
Graphite Foam
Mo / Zr0
2
Multifoil 0.001 0.05 0.1 1.4 [55, 65]
3.3.5 ThermalEnergyStorageMaterial
The selection of a thermal energy storage (TES) material/method is the final technology require-
ment and drives the overall bi-modal system architecture. The TES material is fundamental to
the overall RAC design since the RAC structure must be compatible at storage temperatures and
incorporate a suitable method of heat exchange with the propellant stream. The primary factors
which drive TES selection are desired operating temperature range, energy storage density, thermal
conductivity, reactivity, and whether sensible or latent heat is the primary means of energy storage.
To date, all solar thermal designs using a TES approach have relied on sensible heat with either
graphite or boron carbide as the TES material [3, 25, 49].
In the case of the ISUS program, the ground demonstration assembly utilized what was essen-
tially a graphite cylinder as the TES medium. The RAC assembly was designed so that the center
of the cylinder acted as a blackbody cavity for absorbing solar radiation and 195, 3 mm diameter
channels were bored through the cylinder wall to act as heat exchanger passages for the propellant
stream [66]. A photograph of this graphite assembly, showing the central cylinder and plenumn
caps is given in Figure 3.6. In order to survive the hot hydrogen testing environment, the entire
40
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 3.6: Photograph of the uncoated graphite components for the ISUS RAC showing the cen-
tral combined heat exchanger and TESM. Reprinted from Miles [23].
assembly was coated with rhenium, resulting in total mass values of 15 kg of rhenium and 20 kg
of graphite TES [25]. The protective rhenium coating provided satisfactory performance during
coupon testing. However, there were remaining difficulties with graphite sublimation sustained
through the rhenium coating at temperature [23].
The decision to utilize graphite for the ISUS program, and sensible heat in general, was a func-
tion of developmental time constraints and the low TRL level of other TES options. Table 3.2
lists the specific heat of multiple sensible heat storage materials and it can be seen that a signif-
icant operating temperature range is required to reach appreciable storage densities. It is for this
reason that sensible heat thermal energy storage systems are described in the literature as having
“moderate yet acceptable” performance [34].
41
Distribution A: Approved for public release; unlimited distribution. PA#15347
Table 3.2: High temperature sensible heat storage materials [67, 68].
Material
T
melt
C
p
@ 2500 K T required
[K] [kJ/kgK] for 1 MJ/kg
Graphite 3923 2.15 475
B
4
C 2700 2.68 380
Silicon Carbide 2818 1.01 740
Boron Nitride 3273 1.98 510
Published experimental data from the ISUS program allows for an evaluation of a sensible
heat storage system during hot flow testing. Figure 2.8 gives the results for the ISUS RAC at a
hydrogen feed rate of 1.7 g/s and it can be seen that the system produces an approximately steady
state propellant exit temperature for the first seven minutes of the burn. After this, both the TES
medium and the propellant temperatures drop rapidly. The ISUS RAC is able to maintain a steady
exit temperature because the propellant flowing through the RAC reaches thermal equilibrium with
the graphite TES before exiting the heat exchanger. As the entrance region of the graphite TES
transfers energy to the propellant and cools, the point at which the propellant reaches equilibrium
moves further forward within the TES medium. As long as “hot” thermal energy storage exists
beyond this equilibrium point, the propellant maintains a steady exit temperature. Once the pro-
pellant begins to draw heat from the RAC across the entire length of the heat exchanger, the exit
temperature begins to fall with the overall drop in RAC temperature.
When the ISUS RAC unit reaches an effective energy storage density of 1 MJ/kg approximately
15 minutes into the burn, the propellant exit temperature has dropped to under 75% of the peak
value corresponding to a 15% drop inI
sp
. SinceI
sp
varies with the square root of temperature, the
ISUS graphite system is able to maintain the stated “adequate” thermal rocket performance and
only particularly long burns will suffer from reduced efficiency.
While the large temperature swing of a sensible heat TES system has a relatively low impact
on propulsive efficiency, the effects on thermal electric conversion in a bi-modal configuration
are profound. The original specification for the ISUS system allows for converter temperatures
between 1900-2200 K at the thermionic converter hot shoe [23]. This 300 K temperature range
42
Distribution A: Approved for public release; unlimited distribution. PA#15347
would prove unacceptable on a microsatellite bi-modal platform relying on radiatively coupled
thermophotovoltaics. Based on published data for the EDTEK 1400 nm selective emitter system
and GaSb TPV cells, a temperature drop from 2200 K to 1900 K at the emitter (assumed to be
black body) will result in an over 50% drop in power output [52]. To maintain steady electrical
power on the spacecraft despite fluctuating emitter temperatures, additional weight in the form of
batteries or additional TPV units will be required.
Within the constraints of a TPV based bi-modal microsatellite platform, the additional mass
and operational inefficiencies of using a sensible heat thermal energy storage system renders it
impractical. Therefore, it is proposed that the final technological requirement for a high perfor-
mance bi-modal microsatellite is adoption of latent heat thermal energy storage. Latent heat storage
materials offer both higher energy storage densities and a relatively constant energy output temper-
ature corresponding to the melting point of the storage material. Unlike the technologies discussed
previously in this section, there is very limited research into the use of latent heat materials at STP
temperatures and there have been no direct experimental efforts. The development of an effective
latent heat thermal energy storage system is presented here as the final technological hurdle that
must be overcome to realize a high performance bi-modal solar thermal microsatellite platform.
43
Distribution A: Approved for public release; unlimited distribution. PA#15347
Chapter 4
High Temperature Latent Heat Thermal
Energy Storage
Many of the technological requirements for a solar thermal bi-modal satellite have been previously
investigated and feasible solutions appear to be readily available. However, the transition from
using sensible heat thermal energy storage to a high performance latent heat system requires an
extensive research effort. Existing latent heat thermal energy storage research is chiefly concerned
with terrestrial applications and targets temperatures well below those required by a solar thermal
system. To reach temperatures suitable for spacecraft use, a new class of high temperature latent
heat storage materials must be identified and developed.
4.1 PhaseChangeMaterialSelection
Latent heat thermal energy storage utilizes the heat of fusion released during a phase transition
(typically liquid to solid) as the primary energy storage mechanism. Extensive research has been
performed for terrestrial application of latent heat materials and thousands of potential materials
have been evaluated. Existing latent heat materials can be divided into three broad categories:
paraffin waxes, fatty acids and hydrated salts [69]. These materials, whose properties are listed in
44
Distribution A: Approved for public release; unlimited distribution. PA#15347
Table 4.1: Relevant properties of typical phase change materials [69, 70].
Class
T
melt
H
fus
k
th
[K] [kJ/kg] [W/mK]
Paraffin Waxes 317-379 72-214 0.19-0.75
Fatty Acids 268-344 45-210 0.14-0.17
Hydrated Salts 281-1170 115-492 0.46-5.0
Table 4.1, have melting temperatures far below what is required for STP as well as thermal conduc-
tivities and energy storage densities an order of magnitude below what is necessary. Even some of
the highest temperature phase change materials under consideration, such as potassium carbonate,
have melting temperatures only slightly above 1100 K and heats of fusion below 500 kJ/kg [70].
Furthermore, many of these materials suffer degradation after multiple thermal cycles. In order
to apply latent heat thermal energy storage to a solar thermal bi-modal system, a new class of
materials is required.
In the search for a new high temperature latent heat storage material, the primary considerations
were the melting (operating) temperature and energy storage density. The melting temperature of
the material defines the peak propellant temperature and thus the performance of the solar thermal
rocket. For the bi-modal microsatellite proposed, ammonia propellant has been selected due to
ease of handling and reasonable performance. Literature indicates that the optimal temperature for
an ammonia rocket is around 2500 K which results in anI
sp
in excess of 400 s dependent on the
degree of ammonia dissociation in the system [35, 37, 39]. At lower exhaust temperatures around
1500 K, an ammonia rocket is still capable of achieving approximately 300 sI
sp
, which is beyond
what microsatellite scale monopropellant thrusters can provide. Using this temperature range as a
guide, the potential phase change materials given in Table 4.2 were identified.
Silicon and boron have been selected as target high temperature phase change materials due
to their exceptionally high heats of fusion, and melting temperatures that are ideally matched for
use in an ammonia solar thermal rocket. Additionally, these materials are elemental, removing
45
Distribution A: Approved for public release; unlimited distribution. PA#15347
Table 4.2: Potential high temperature phase change materials. Material specific references are
given when applicable, otherwise, values are taken from [59, 67, 71]. Thermal conductivity values
given in italics are the closest available measurement toT
melt
.
Material
T
melt
H
o
fus
k
th
@T
melt
[K] [kJ/kg] [W/mK]
MgF
2
[72] 1536 940 3.8
Beryllium 1560 1312 69
Silicon [73] 1687 1785 20
Nickel 1728 292 83
Scandium 1814 313 16
Chromium 2180 394 48
Vanadium 2183 448 51
Boron 2350
1
4600 5-10
Ruthenium 2607 381 80
Niobium 2750 290 82
Molybdenum 2896 375 84
concerns of material breakdown, and both exhibit adequate thermal conductivities. Both silicon
and boron are considered in this work with silicon treated as a near term, moderate performance
solution and boron being the ideal long term target.
4.2 TechnologicalComparison
The use of latent heat thermal energy storage appears immediately beneficial when comparing
material properties. However, operational concerns and the effects of implementing this tech-
nology within spacecraft parameters must be considered. Switching to a latent heat medium is
a significant technological undertaking and the benefits must be strong enough to support the
needed development. Two technological comparisons have been made to demonstrate the benefits
of switching to latent heat thermal energy storage. The first comparison considers the advantages
1
The melting temperature given is for-rhombohedral boron. Some sources list the melting temperature as 2573K
corresponding to amorphous boron, however, the phase change process will occur at the-rhombohedral temperature
[74, 75]. Sands and Hoard write that from the liquid state, boron will re-crystallize into the-rhombohedral form and
a later work notes that the-rhombohedral form will irreversibly become-rhombohedral at approximately 1773K
[76, 77].
46
Distribution A: Approved for public release; unlimited distribution. PA#15347
of integrating latent heat thermal energy storage into an existing sensible heat based bi-modal
design. The second comparison, performed in conjunction with the AFRL, proposes the per-
formance impact of a boilerplate bi-modal solar thermal propulsion system on a microsatellite
platform vs. competing conventional configurations.
4.2.1 LatentHeatApplicationtoExistingDesigns
A logical starting point for the analysis of latent heat thermal energy storage is to apply the tech-
nology to the heat exchanger designed for the Integrated Solar Upper Stage Program. The ISUS
program is the most fully realized bi-modal solar thermal design to-date and even with sensible
heat thermal energy storage it provided acceptable performance. However, if the sensible heat
thermal energy storage (TES) were replaced with boron, the TES package could potentially see
an almost fourfold increase in total energy storage density (versus the designed graphite T of
600 K). It is also expected that a latent heat energy storage system will offer convective coupling
advantages resulting in a more stable propellant output temperature during the course of a burn.
Since these assumptions are based purely on material properties, analysis is required to address
the operational concerns of applying a latent heat thermal energy storage medium. Chief amongst
these concerns is the ability of a latent heat medium to effectively transfer energy into propellant
due to the relatively low thermal conductivity of solid high temperature PCMs near their melting
point. A series of conjugate heat transfer models have been completed using the commercial
multiphysics package STAR-CCM+ to quantify the effects of an ISUS like design using latent heat
thermal energy storage
ISUSApproximation
In order to efficiently model the ISUS system, the heat exchanger at the core of the receiver-
absorber-converter was modeled a single representative adiabatic heat exchanger passage. Pub-
lished specifications for the ISUS RAC state that the assembly was manufactured from three parts:
47
Distribution A: Approved for public release; unlimited distribution. PA#15347
two plenum end caps and a central graphite cylinder. This central graphite ring, shown in Fig-
ure 3.6, is the primary heat exchanger. The graphite ring is approximately 152 mm long with an
inner diameter of 140 mm and an outer diameter of 300 mm. The graphite heat exchanger con-
tained 195 flow passages with a diameter of 3 mm and was coated with 0.3 - 2.5 mm of CVD
rhenium depending on location. From these values, it was calculated that the heat exchanger core
contains approximately 15 kg of graphite and 10.1 kg of rhenium thermal energy storage.
For the STAR-CCM+ model, the macro properties of this central heat exchanger were divided
by the total number of heat exchanger passages and the result was assigned to a single 152 mm
long heat exchanger channel. The resulting model geometry, shown in Figure 4.1, consists of a
19.6 mm outer diameter, 3 mm inner diameter passage containing 129 g of total thermal energy
storage mass. The inner surface of the package is coated with a 0.25 mm rhenium layer consistent
with published coating values and the graphite bulk geometry was selected to evenly distribute
the full design’s 15 kg of graphite TES between 195 channels. Similarly, the outer surface of
the channel is coated with an additional 0.22 mm layer of rhenium to account for the remaining
CVD rhenium specified in the original system. Note that this geometry is an approximation due
to the uneven distribution of the propellant channels in the ISUS design. However, results for this
geometry demonstrate similar behavior as the macro system.
Solid properties were modeled as being fully temperature dependent and graphite properties
were kept consistent with the TM grade POCO graphite used in the ISUS experiments. Hydrogen
flow conditions were taken from published ISUS data which lists a total flow rate for the ISUS RAC
at 1.78 g/s at 23.1 psia with an inlet temperature of 500 K. For modeling a single heat exchanger
passage, this corresponds to a modeled mass flow rate of 0.009 g/s. As with the sensible TES
materials, hydrogen was modeled with temperature dependent material properties and flow was
considered to be both viscous and turbulent. Disassociation of hydrogen was not considered and
hydrogen was treated as an ideal gas.
48
Distribution A: Approved for public release; unlimited distribution. PA#15347
All solid exterior boundaries were modeled as adiabatic. The ISUS RAC utilized an advanced
multi layer insulation system with a calculated thermal loss of 4800 W at a cavity temperature of
2300 K [25]. Published hydrogren flow data, taken with a lower RAC cavity temperature of 2150 K
indicates a peak convective power draw exceeding 36,000 W suggesting that total insulation losses
are less than 15% and can be neglected for the purposes of this comparison.
The initial thermal condition for the solid TES region in the STAR-CCM+ model was isother-
mal at 2360 K. This temperature is slightly higher than the ISUS design specification (2200-
2300 K) but allows a direct comparison to a molten boron based system. The STAR-CCM+ model
was run as an 2-D axisymmetric simulation using the implicit unsteady solver after establishing an
initial starting condition for hydrogen flow at zero seconds.
The transient temperature profiles of this sensible heat model are given in Figure 4.1 and the
mass flow average hydrogen exit temperature vs. time is given in Figure 4.2. As expected from
published ISUS data, Figure 4.3 shows an initial segment where the hydrogen output is relatively
isothermal taking approximately 250 seconds to drop below 95% of the initial value. As discussed
in Section 3.3.5, hydrogen reaches relative thermal equilibrium before exiting the heat exchanger
and this equilibrium point moves upstream as the rear of the passage cools, maintaining a region of
steady performance. For comparison, the published data for the ISUS program takes approximately
380 seconds to drop by the same amount. The steeper cooling seen in the STAR-CCM+ model is
due to neglecting the graphite caps present in the ISUS experimental design resulting in the absence
of an additional 5 kg of graphite and 5 kg of rhenium as well as the additional hydrogen flow length
which supports higher sustained output temperatures.
49
Distribution A: Approved for public release; unlimited distribution. PA#15347
time = 0 s
time = 50 s
time = 100 s
time = 150 s
time = 200 s
time = 250 s
time = 300 s
Temperature (K)
Figure 4.1: Modeling geometry and transient temperature profiles for the “ISUS Approximation”
STAR-CCM+ model. Model geometry is axisymmetric and represents the red outlined region of
the cylindrical heat exchanger passage. Hydrogen flow through the passage is from left to right
and dimensions are given in millimeters.
50
Distribution A: Approved for public release; unlimited distribution. PA#15347
0 200 400 600 800 1000 1200 1400
800
1000
1200
1400
1600
1800
2000
2200
2400
Time, s
Hydrogen Exit Temperature, K

ISUS Approximation
Boron − Constant Mass
Boron − Length Reduction
Boron − Diameter Reduction
Figure 4.2: Mass flow averaged hydrogen exit temperatures for STAR-CCM+ heat exchanger sim-
ulations.
0 100 200 300 400 500 600
80
82
84
86
88
90
92
94
96
98
100
Time, s
Percent of Peak Hydrogen Exit Temperature

ISUS Approximation
Boron − Constant Mass
Boron − Length Reduction
Boron − Diameter Reduction
Figure 4.3: Mass flow averaged hydrogen exit temperature as a percentage of peak (initial) exit
temperature for STAR-CCM+ heat exchanger simulations.
51
Distribution A: Approved for public release; unlimited distribution. PA#15347
BoronIntegration-ConstantMass
To see the effects of switching the sensible design to a boron based system, graphite TES was
replaced with boron in the STAR-CCM+ model while keeping total system mass constant. Since
boron has a higher density than TM graphite, the total diameter of the model was reduced to ensure
that both the sensible and latent heat models both contained the same TES mass. Additionally, two
0.5 mm layers of boron nitride were included between rhenium layers and the boron TES to act as a
suitable container for boron when it is liquid. The reasons for selecting boron nitride are discussed
in Section 6.1. The overall geometry is given in Figure 4.4 and the total TES design contains 9.4 g
of boron nitride, 52.6 g of rhenium, and 66.8 g of boron for a total system mass of 129 g.
The temperature transforming method was used to accommodate the latent heat release from
the boron in the STAR-CCM+ model [78]. Boron was treated as a solid with a variable specific
heat accounting for the latent heat release across a “mushy” zone of 2.5 K from the melting point
(2350 K). Apart from this change, all model parameters and initial conditions were kept constant.
Figure 4.2 includes the hydrogen exit temperatures for this latent heat model (“Boron - Con-
stant Mass”) and demonstrates the effects of dramatically increasing the energy storage density
of the TES system. In addition to prolonging the length of potential propulsive burns, the quasi-
isothermal heat release during the phase change process produces an extremely stable output tem-
perature which takes 1750 seconds to drop below 95% of the maximum. Thermal gradients, given
in Figure 4.4, show that the freezing process takes in excess of 2000 seconds and that during phase
change, propellant passage wall temperatures remain high.
SinceI
sp
varies with the square root of temperature, this dramatic increase in propellant output
stability and longevity only effects total propulsive performance for long duration burns. For a
single burn of 1400 seconds, the latent heat system delivers approximately 20% more V than the
sensible heat system. But, for burns less than 600 seconds, the latent heat system provides a V
52
Distribution A: Approved for public release; unlimited distribution. PA#15347
time = 0 s
time = 400 s
time = 800 s
time = 1200 s
time = 1600 s
time = 2000 s
time = 2400 s
Temperature (K)
Figure 4.4: Modeling geometry and transient temperature profiles for the “Boron - Constant Mass”
STAR-CCM+ model. Model geometry is axisymmetric and represents the red outlined region of
the cylindrical heat exchanger passage. Hydrogen flow through the passage is from left to right
and dimensions are given in millimeters.
53
Distribution A: Approved for public release; unlimited distribution. PA#15347
0 200 400 600 800 1000 1200 1400
−20
−15
−10
−5
0
5
10
15
20
Time, s
Relative Total ∆V Delivery vs. ISUS Approximation (%)

ISUS Approximation
Boron − Constant Mass
Boron − Length Reduction
Boron − Diameter Reduction
Figure 4.5: Relative V delivery for representative heat exchanger models.
efficiency benefit of less than 5%. A comparison of relative V for all STAR-CCM+ models is
given in Figure 4.5.
BoronOptimization-LengthReduction
The dramatic improvement in temperature stability and burn longevity seen when switching from
graphite to boron based TES is expected as the energy storage density of the system is increased
more than fourfold due to the introduction of a latent heat component. However, if the performance
of the sensible heat system is in fact adequate, an increase in energy storage density can be better
used to optimize the system by reducing both total mass and volume of the heat exchanger. To
examine this possibility, two models were created targeting either length or diameter reduction.
The first optimized system model attempted to replicate ISUS performance by taking advantage
of convective coupling to a latent heat medium to reduce overall heat exchanger length. Both
Figures 4.1 and 4.4 show that the initial convective coupling of the hydrogen propellant occurs
early within the channel. At approximately one third of the total heat exchanger length, propellant
54
Distribution A: Approved for public release; unlimited distribution. PA#15347
temperatures are within 5% of the target output temperature. At the midpoint of the heat exchanger,
propellant temperatures are within 2% of this target value. It is proposed that this convective profile
is advantageous in the design of a latent heat based system. If the latent heat PCM maintains a
stable temperature profile during energy release, the total length of the heat exchanger can be cut
in half reducing both mass and volume.
A new model heat exchanger passage was created to investigate this optimization method and
the heat exchanger length was reduced to 76 mm from 152 mm. Total boron thermal energy storage
mass was sized based on steady performance from the sensible heat model for which the output
temperature remained within 5% of the maximum temperature for roughly 250 seconds requiring
67 kJ of energy delivery from the thermal energy storage (effective energy storage density of
0.52 MJ/kg). To maintain this output with a shortened heat exchanger, all of the energy must come
from the boron phase transition which sets the total boron mass for the system at 14.6 g of boron
vs. 77 g of graphite.
Other parameters for the model were held constant including rhenium coatings on the inner
and outer surfaces as well as the inclusion of the 0.5 mm BN liners. The end geometry resulted
in a total system mass of 38 g corresponding to a 70% reduction in mass vs. the full scale model.
Additionally, total volume of the heat exchanger passage was reduced by 79%.
Transient thermal profiles for this optimized case are given in Figure 4.6 and mass flow aver-
aged hydrogen exit temperatures are included in Figure 4.2 (“Boron - Length Reduction”). After
an initial brief drop in temperature before the beginning of the phase change process, this opti-
mized system maintains a steady output temperature for approximately 280 seconds, exceeding
the performance of the sensible heat system and the design goal for the optimized configuration.
However, it must be noted that after this initial period of stability, the exit temperature for the
length optimized system rapidly drops in temperature. The reasons for this are twofold; first, the
system is almost entirely reliant on latent heat energy storage due to the dramatic reduction in
55
Distribution A: Approved for public release; unlimited distribution. PA#15347
mass. Once this latent heat is exhausted, there is comparatively little sensible heat which can be
drawn from the system to maintain output temperatures.
Second, the extremely low thermal conductivity of solid boron (approximately 5W=mK near
the melting point) allows dramatic thermal gradients to form within the system supported by the
remaining still liquid PCM. When the phase change is complete, the slow diffusion of thermal
energy within the thermal energy storage medium results in maintained high temperature regions
near the exit of the heat exchanger. Convective coupling to this thermal environment is very inef-
ficient since the propellant stream does not see a high wall T until shortly before exit.
BoronOptimization-DiameterReduction
Another route to optimization simply reduces the mass of the system while leaving the heat
exchanger channel length the same. For this model, the amount of boron was sized as a direct
replacement for the graphite TES while maintaining the same convective coupling length.
The sensible heat ISUS model reaches 1 MJ/kg of effective energy storage density approxi-
mately 500 seconds into the burn. Targeting this energy storage capacity requires only 13.2 g of
boron to replace the 77 g of graphite used in the sensible case. The boron mass calculation includes
both the latent heat and the sensible heat contributions across the assumed T . Due to boron’s
relatively high specific heat, the sensible component accounts for 20% of the total energy storage.
All model parameters are kept the same as the “Boron - Constant Mass” case apart from reduc-
ing the total amount of boron. Rhenium coatings and BN protective layers are still included which
brings total exchanger mass to 51 g corresponding to only a 60% reduction over the sensible case.
Total passage volume is reduced by 75%.
Transient thermal profiles for this optimized case are given in Figure 4.7 and mass flow aver-
aged hydrogen exit temperatures are given in Figure 4.2 (“Boron - Diameter Reduction”). This
configuration yields a steady output temperature for 425 seconds exceeding both the sensible
56
Distribution A: Approved for public release; unlimited distribution. PA#15347
time = 0 s
time = 75 s
time = 150 s
time = 225 s
time = 300 s
time = 375 s
time = 450 s
Temperature (K)
Figure 4.6: Modeling geometry and transient temperature profiles for the “Boron - Length Reduc-
tion” STAR-CCM+ model. Model geometry is axisymmetric and represents the red outlined region
of the cylindrical heat exchanger passage. Hydrogen flow through the passage is from left to right
and dimensions are given in millimeters.
57
Distribution A: Approved for public release; unlimited distribution. PA#15347
design case as well as the length optimized case. It also must be noted that in the relative V
data, this optimized case provides slightly better performance over 500 seconds.
Because of the increased passage length, boron is spread throughout the system and as large
thermal gradients form, the propellant is still able to couple to high temperature regions later in the
channel. However, much like the length optimized case, the reliance on latent heat energy storage
results in a sharp drop off in propellant temperatures after the phase change process completes.
Summary
Modeling a representative conjugate heat transfer system has shown that integrating latent heat
materials into a heat exchanger design is theoretically feasible and that the high performance of
boron as a latent heat thermal energy storage material can be used to either dramatically increase
system performance or reduce total system mass.
It must be noted, however, that latent heat systems behave differently than a sensible heat
system and optimization must take into account the burn times required by the target mission.
Modeling results indicate that comparisons between sensible and latent heat systems are more
complex than the purely material property based comparisons found in the existing literature. V
delivery comparisons in Figure 4.5 show that similarly sized latent heat systems only provide
propulsive benefits for long duration burns, and the dramatic mass reductions created through
optimization are only able to match performance for certain burn durations. Certainly, any system
designed specifically for latent heat thermal energy storage will take these operating characteristics
into account.
Mass and volume are primary constraints on a microsatellite platform and the dramatic
improvements in both metrics demonstrated by switching to latent heat materials yields signifi-
cant advantages against sensible heat configurations. When targeting an aggressive V budget
on a microsatellite the mass and volume of sensible heat systems render them infeasible and thus
58
Distribution A: Approved for public release; unlimited distribution. PA#15347
time = 0 s
time = 75 s
time = 225 s
time = 300 s
time = 375 s
time = 450 s
time = 525 s
Temperature (K)
time = 150 s
Figure 4.7: Modeling geometry and transient temperature profiles for the “Boron - Diameter
Reduction” STAR-CCM+ model. Model geometry is axisymmetric and represents the red out-
lined region of the cylindrical heat exchanger passage. Hydrogen flow through the passage is from
left to right and dimensions are given in millimeters.
59
Distribution A: Approved for public release; unlimited distribution. PA#15347
latent heat thermal energy storage is an enabling technology. The following comparison seeks to
determine the impact of an implemented latent heat based system on a microsatellite platform.
4.2.2 MicrosatellitePerformanceImpact
An initial systems level analysis has been performed with the AFRL to compare the proposed
bi-modal STP system against existing propulsion technologies [79]. Theoretical bi-modal solar
thermal systems were sized using a variety of parameters (i.e. propellant budget, desired V ,
phase change material, etc.) and compared against current conventional capabilities. It has been
shown that, even when using conservative assumptions, the development of silicon and boron as
PCMs can have a strong impact on microsatellite performance.
A sample bi-modal STP system was sized for a 100 kg microsatellite with a 1500 m/s V
capability to target the high-performance microsatellite category. Energy storage and conversion
subsystems for the baseline spacecraft were required to simultaneously provide 100 W of full-time
electrical power and 100 W of thermal propulsive power in a low Earth orbit (LEO) using silicon
as the PCM. These requirements are similar to a proposed space demonstration study that was to
follow the ISUS program. The proposed mission was to launch from a Pegasus XL rocket and place
a 207 kg satellite into LEO with a buss power output of 100 W, a V capability of approximately
1600 m/s and a propulsion and power mass fraction of 61% [80].
The sizing of the bi-modal STP system, including all power and propulsion components, was
based upon previous research and readily-achievable technological solutions [1, 34, 61, 81, 82].
Parameters of interest during the mass budgeting process included the weight of the solar con-
centration system (fiber optics, primary concentrator and support structure), thermal energy stor-
age mass (PCM, container and insulation), power system weight (TPV , radiators, electronics) and
the weight of an ammonia STP propulsion system (tankage, flow system, engine and propellant
weight). Total thermal efficiency of the propulsion and power unit was approximately 17.5%
assuming 70% thermal collection efficiency, storage losses of 25% per orbit, and 20% thermal
60
Distribution A: Approved for public release; unlimited distribution. PA#15347
electric system efficiency. Calculations also assumed that thruster firing would only occur for 5%
of the mission duration at 20-times the average power (i.e. near apogee and perigee of maneuver
orbits for efficiency) to approximate an impulsive burn strategy.
The base system was sized to produce 1500 m/s V ; calculations showed that this V could
be delivered in under 23 days with a combined propulsion and power mass budget of just over
58 kg utilizing silicon as the PCM and essentially “proven” system components documented in the
literature.
Holding the propulsion and power mass fraction of 58% constant (i.e. assuming systems for
comparison must have an identical mass budget for the payload and other non-propulsion and
power systems on a 100 kg microsatellite), system budgets and capabilities were calculated for
competing technologies published in the literature, including a 1 N Hydrazine thruster [83], a 20 N
hydrazine thruster [84], a standard (non-bi-modal) STP system and a 100 W Hall Effect Thruster
[85]. It was assumed that all of these technologies included photovoltaic panels and batteries
suitable to meet the same 100 W electrical power draw while propulsion systems were active.
The results of these calculations, along with the results for the silicon bi-modal system, are
presented in Figure 4.8. As with the bimodal system sizing, the relative weights of power systems,
tankage and thrusters were determined primarily through previously established sizing metrics
[86, 87]. NASA’s year-2020 target for photovoltaic power density was used to size the traditional
power systems, saving mass relative to current available technologies [88]. Additionally, it should
be noted that for chemical and electrical propulsion cases, it was assumed that the thrusters fired
continuously to deliver the desired V . This would result in orbit-change inefficiency, and a
smaller orbit change relative to the V imparted. Alternatively, if the same 5% firing time was
enforced on these thrusters as the proposed STP systems, the maneuver times would have been
increased by a factor of 20, pushing the comparative system response significantly higher.
In spite of this analysis being bent to favor the competing technologies, Figure 4.8 shows
the unique position occupied by the STP systems. The standard STP option already occupies
61
Distribution A: Approved for public release; unlimited distribution. PA#15347
a performance gap between chemical and EP systems, allowing a slightly higher V than the
chemical systems without the years-long maneuver time of the hall thruster. When the silicon
thermal energy storage system is added and the thermophotovoltaics and batteries are replaced by
a bi-modal system, the response time drops slightly and V capability shows a 10% advantage
over chemical systems due to an increase in propellant mass and propulsive efficiency.
Figure 4.8: Systems comparison of the proposed bi-modal solar thermal system versus competing
technologies. Vertical lines indicate specific mission V requirements and the blue trendline indi-
cates the trade off between mission duration and V typical of conventional propulsion concepts.
Replacing the silicon PCM with boron reduces the amount of PCM required, allowing for
slightly more propellant within the power-and-propulsion mass budget. Simultaneously, the boron
based system operates at a higher temperature, yielding higher I
sp
(but slightly decreasing the
thrust and increasing the response time. This is consistent with the assumption that average propul-
sive thermal power draw remains constant). With this change, and still sizing other power and
propulsion components via published readily-achievable technology, the response time increases
62
Distribution A: Approved for public release; unlimited distribution. PA#15347
slightly to 34 days, but the total available V increases to over 1850 m/s. The boron system has a
35% V advantage over chemical rockets while still providing a highly-responsive architecture.
Making more optimistic estimates for system components (i.e. increasing the TPV plus radiator
power efficiency from currently-available 15 W/kg to 30 W/kg and cutting fiber optic mass from 5
to 2.5 kg) yields an advanced bi-modal boron-STP system with a response time of under 40 days
and a total V approaching 2200 m/s. At this level, the STP system can provide a V that cannot
be reasonably considered using chemical systems, while remaining several orders of magnitude
more responsive than an EP system at the same V (2200 m/s in> 500 days).
Since microsatellites can be delivered into orbit at relatively low cost by piggy-backing on
the launch vehicles of larger satellites, the ability to rapidly reposition into a drastically different
orbit (i.e. the desired orbit for the microsatellite mission) can significantly enhance the utility and
frequency of microsatellite launches. Additionally, note the sample maneuvers marked via the
vertical lines and shaded areas on Figure 4.8 [34]. Development of the proposed bimodal STP
technology could provide for a microsatellite platform that can not only piggy-back on the launch
of a conventional satellite and reposition itself accordingly, but could also result in microsatellites
capable of transferring into Lagrange Point orbits or even inserting into lunar or asteroid orbits.
4.3 ExistingLiterature
In the solar thermal propulsion literature, the use of high temperature phase change materials for
thermal energy storage has been mentioned briefly in the context of broad conceptual studies.
Silicon or boron are often proposed as promising thermal energy storage candidates based on
their material properties and then consequently rejected in favor of sensible heat options due to
operational and programmatic concerns. For example, the ISUS program stated that latent heat
storage materials would increase energy density and simplify both the design and testing of the
thermionic system due to a narrow thermal cycling range. However, the use of phase change
63
Distribution A: Approved for public release; unlimited distribution. PA#15347
materials was rejected before the design phase of the program due to the extremely low TRL level
of existing research and an ambitious project schedule.
Kennedy’s microsatellite work similarly selected sensible heat thermal energy storage due to
simplicity of design and “moderate but acceptable performance levels” [34]. Kennedy states that
complications in phase change energy storage systems stem from void formation and containment
structure bursting. However, the papers referenced only discuss the former in the context of LiF-
CaF
2
based energy storage systems [22, 89]. Kennedy does briefly mention boron and silicon
as potential storage materials and cites their material properties. But, like the ISUS program,
Kennedy’s work rejects the use of phase change materials before the design phase and makes no
estimates of their performance [3].
An early assessment from the AFRL of a bi-modal solar thermal system assumes the use of
silicon as a thermal energy storage material and makes basic estimates of system performance [17].
The paper states that a bi-modal configuration using silicon thermal energy storage would offer a
flexible and responsive spacecraft architecture. However, it appears that silicon properties were
simply used for top level analysis [17]. A contemporary paper from Rocketdyne is the first paper
to discuss boron as a thermal energy storage medium for solar thermal propulsion beyond a simple
listing of material properties [11]. Shoji’s analysis states that a boron based thermal energy storage
system would actually raise the propulsion subsystem mass. But, his analysis does not consider a
bi-modal configuration and thermal energy storage is only seen as a means for reduction in solar
concentrator area keeping with the findings proposed by Ethridge a decade prior [4].
In 2001 Ontario Engineering International proposed a tri-modal satellite bus with the potential
for silicon thermal energy storage. The study again only looked at broad material properties and
noted that the technology would be of interest in comparison with graphite [90]. Most recently a
study funded by ESTEC/EADS in 2003 looked at the possibility of a solar thermal upper stage to
reduce launch mass and overall costs [91]. The spacecraft targeted in this study did not consider
bi-modal operation, and much like in Shoji’s analysis, determined that the addition of TES system
64
Distribution A: Approved for public release; unlimited distribution. PA#15347
would reduce available payload mass using silicon or graphite. Interestingly, it was stated that
boron was an exception. However, later analysis utilized graphite based storage.
Recently, there has also been a renewed interest in nuclear thermal bi-modal spacecraft sys-
tems. Researchers at the Center for Space Nuclear Research at the Idaho National Laboratory
have proposed a bi-modal propulsion and power bus to allow interplanetary exploration of the
solar system using small satellites [92]. To increase the specific power that can be achieved with
a radioisotope thermal generator, a thermal capacitor system has been proposed to allow for high
power propulsive burns and burst generation of electrical power for communications through the
use of closed Brayton cycle dynamic thermoelectric conversion. High performance sensible heat
thermal energy storage along with boron and silicon based TES were considered with the program
ultimately settling on using silicon as the primary TES. This research is currently ongoing and
many of the concerns raised about molten silicon thermal energy storage parallel those raised in
the early stages of this research effort.
The body of available literature considering boron and silicon based thermal energy storage
for terrestrial applications is similarly narrow. Furthermore, there is very limited research into
molten boron itself, let alone its use as a PCM. Terrestrial research into high temperature phase
change materials is primarily associated with thermophotovoltaic development projects. Woodall,
in a patent assigned to IBM in 1982, is the first to mention silicon specifically as a potential energy
storage material [93]. Intended as a buffer for a TPV emitter, silicon was proposed along with
other materials meeting temperature, thermal conductivity, and energy density requirements which
included iron, manganese and chromium. Woodall also proposed potential container materials
such as boron nitride and aluminum oxide and went on to melt silicon in pyrolitic BN crucibles to
confirm the latent heat value. However, experiments ended here and the proposed buffer-storage
system was never built.
The most substantive research into the use of using molten silicon as a thermal energy storage
material was first presented by Chubb, Good and Lowe in 1995 as part of thermophotovoltaic
65
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 4.9: Diagram of of the silicon freezing model proposed in Chubb et al [5]. Note the large
reciever area (A
s
) and small emitter are (A
E
) which are intended to smooth thermal gradients
during the freezing process.
development at what was then McDonnell Douglass [5]. Similar to Woodall’s proposal, silicon
was to be a thermal energy storage material for a terrestrial TPV system and would allow for
continuous operation. In addition to selecting silicon as a PCM, a 1-D model for PCM solidification
was presented in a conical enclosure. As shown in Figure 4.9, this model assumed a large receiver
(A
S
) and smaller emitter (A
E
) which would result in higher energy flux at the emitter during
silicon solidification and a lower temperature drop between the receiver and emitter during “on-
sun” operation. This model was not silicon specific and could be applied to any PCM based system.
Multiple large assumptions were also made, including perfectly insulated walls and the neglect of
volumetric expansion effects in the PCM. The paper concluded that silicon would make an “ideal”
thermal storage material for TPV systems but no experiments were performed. This research was
granted a patent in 1999 and the combined TPV reciever and storage system was estimated to have
a combined thermal-electric conversion efficiancy of 27-51% [94].
66
Distribution A: Approved for public release; unlimited distribution. PA#15347
In 2013, a paper published by Datas, Chubb and Veeraragavan revisited the 1-D PCM model
originally proposed in 1995 to perform further analysis on the TPV buffer concept [95]. Datas
initially looked at the steady state performance of a system using the same conical geometry and
adiabatic wall assumptions stated in Chubb’s paper with the intent of mapping steady state on-sun
performance of the system and optimizing geometric factors for maximum thermophotovoltaic
efficiency. A follow-up paper in 2014 by Verraragavan, Montgomery, and Datas applied a transient
model to the conical PCM geometry to gauge night time performance of a silicon based system
and concluded that high energy storage density of silicon produces sufficient storage durations to
be feasible [96]. The authors intend to refine the PCM model beyond that originally proposed by
Chubb and include factors such as, 2-D phase front treatment, insulation losses, natural convection
in the PCM and reservoir orientation and make no mention of future experiments.
In summary, within the existing literature concerning both silicon and boron as phase change
materials there is broad agreement that their outsized energy storage densities could have sig-
nificant impact on energy systems operation. However, considering the low TRL level of high
temperature phase change thermal energy storage both materials have been either neglected, as
with solar thermal propulsion research, or relegated to side projects and white paper studies in the
case of terrestrial applications.
4.4 ResearchGoals
The use of high temperature latent heat thermal energy storage has been in the background of
both solar thermal and terrestrial thermophotovoltaic research projects for the past two decades.
Simple technological comparisons have proposed broad operational and capability advantages if
the technology were available. However, to-date there has been no motivation to begin a thorough
investigation.
This work proposes that the benefits of solar thermal propulsion can be amplified by imple-
menting a bi-modal system on a microsatellite platform and technological comparisons presented
67
Distribution A: Approved for public release; unlimited distribution. PA#15347
here demonstrate that significant performance gains are in-fact possible. Such a satellite would be
more than just a long awaited technology demonstration of solar thermal propulsion, it would be a
high performance microsatellite platform unrealizable with conventional systems.
The component technologies required for a viable bi-modal microsatellite have all reached
sufficient technological maturity with the exception of high performance latent heat thermal energy
storage. It is the goal of this study to determine the basic feasibility of this enabling technology
and facilitate the development of a bi-modal solar thermal microsatellite.
Specifically, this work seeks to perform the first substantive experiments targeting silicon as
an energy storage material. An experimental approach is being taken to uncover the practical
design concerns which are currently absent from the literature. Concentrated sunlight will be used
to generate liquid silicon in the laboratory to evaluate potential container materials and uncover
major technological challenges. Additionally, experimental results will be used to evaluate the
modeling fidelity required to capture macro system behaviors.
Experiments and analysis target silicon as opposed to boron due to the lower melting temper-
ature effectively reducing project costs by an order of magnitude when considering solar furnace
and custom material requirements. Additionally, silicon research and facility development have
broad applications when considering terrestrial potential of the technology.
68
Distribution A: Approved for public release; unlimited distribution. PA#15347
Chapter 5
Solar Furnace Development
A new solar furnace facility has been constructed at the University of Southern California in order
to create molten silicon in the laboratory and evaluate the practical concerns associated with using
high temperature latent heat materials. Using a solar furnace to drive experiments forces designs
to accommodate a single-point energy input and provides correlation with potential spacecraft
applications. Solar furnace construction was also driven by a desire for a facility which could
support future thermophotovolatic research due to the broad terrestrial applications of a silicon
based energy storage system.
The USC solar furnace has progressed through multiple iterations and makes ample use of
commercial off the shelf (COTS) and surplus components resulting in a low-cost design. The
furnace is capable of peak solar concentration ratios in excess of 4000:1 and an average power
delivery of approximately 800 W.
5.1 InitialDesignsandDiagnosticDevelopment
Construction of a new solar furnace facility at USC required the simultaneous development of both
the solar concentration system as well as relevant diagnostics. To support this process, a simple
solar furnace was constructed using a surplus Fresnel lens as the primary concentrator. While this
69
Distribution A: Approved for public release; unlimited distribution. PA#15347
furnace exhibited a total system efficiency of only 25%, it did succeed in supporting diagnostic
development, evaluation of experimental methods, and the first round of solar furnace experiments.
Not included in the discussions of solar furnace development are attempts to produce a large
scale solar concentrator through metalization of a commercially available fiberglass parabolic
antenna [48]. This effort ultimately proved unsuccessful after a multi-month fabrication process
due to difficulties associated with an even application of metalization chemicals across a large
surface using the SpectraChrome process.
5.1.1 FresnelLensBasedDevelopmentFurnace
The first iteration of the USC solar furnace facility consisted of a heliostat and a large acrylic
Fresnel lens diagrammed in Figure 5.1. The testing chamber, diagnostics and concentrator are
fixed with respect to the lab and the heliostat mirror is the only moving component. The heliostat
compensates for solar motion throughout the day by maintaining a reflected sunlight vector normal
to the Fresnel lens. This configuration simplifies design at the expense of additional reflection
losses compared to single stage designs requiring motion of both the solar concentrator and the
testing chamber.
FresnelLensandVacuumChamber
The primary concentrator during development of the solar furnace was an acrylic Fresnel spot lens
originally used in a rear-projection television. The 1.1 m by 0.84 m lens, shown in Figure 5.2, has
a focal length of approximately 1 m. The lens was framed and mounted on rails to allow for the
position of the focal point to be shifted with respect to the testing chamber.
The testing chamber is a 15.2 cm (6 in) CF cross coupled to a larger vacuum chamber via
an extension in order to place vacuum pumps and electrical feeds outside of experimental area.
Sunlight enters the testing chamber through a 15.2 cm diameter, 0.64 cm fused quartz window
which is held onto the chamber via an o-ring seal and the pressure differential between the chamber
70
Distribution A: Approved for public release; unlimited distribution. PA#15347
SUNLIGHT
Redirection Mirror
96% reflective
Heliostat
Test Section
Fresnel Lens
Figure 5.1: Path diagram for the Fresnel lens based solar furnace.
and ambient. This design allows for low cost quartz blanks to be used as opposed to integrated
vacuum chamber viewports and facilitates easy removal for cleaning.
The fresnel lens based solar furnace was capable of delivering approximately 200 W into a
19 mm diameter spot. This power level was insufficient to reach silicon melting temperatures.
However, the development furnace enabled evaluation of heliostat performance and diagnostic
development.
HeliostatDrive
An altitude-azimuth heliostat drive was obtained as surplus from the AFRL and was previously
used in the AFRL solar thermal research facility during the 1980s and early 1990s. The tracking
drive, shown in Figure 5.3, had been stored outdoors for over a decade and required a complete
rebuild of the drive motors, replacement of all control electronics, the creation of new tracking
software, and the installation of a new mirror assembly.
71
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 5.2: Photographs of the Fresnel lens based solar furnace including the first iteration of the
heliostat, the Fresnel lens, the testing chamber showing a glowing test article after furnace power
is removed, and an infrared image of the Fresnel lens output on the face of a 0.75 inch diameter
graphite test section.
The heliostat is driven by two AC gear motors with speed further reduced by the altitude and
azimuth tracking hardware. Control for each motor is handled via a variable AC motor controller
capable of accepting digital on-off control signals as well as an analog programing signal for motor
speed. Even with speed control and considerable gearing reduction, the heliostat’s minimum con-
stant slew rates (32 arcseconds per second for azimuth and 52 arcseconds per second for altitude)
are too fast to provide constant rate tracking of the sun. As an alternative to constant rate tracking, a
minimum reliable motion step was established considering spin-up and spin-down times for the AC
motors and gear trains to implement discrete tracking control. The minimum reliable movement
increment for the heliostat was established as 0.036

and 0.026

for the altitude and azimuth drives
respectively. Since the recommended and calculated total heliostat pointing accuracy is between
0.1

and 0.25

, this is sufficient for operation with minimal manual intervention [4, 97].
Heliostat control is implemented in National Instruments (NI) LabView software and is run
on a standard desktop PC. Analog and digital control signals are provided using an NI PCI-6281
72
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 5.3: Surplus heliostat tracking unit shown installed at the AFRPL solar facility (left) and at
the current USC solar furnace facility (right). The unit received a full rebuild before being installed
at USC.
multi-function DAQ via a tether to the AC motor controllers which are mounted on the Heliostat
body. The control software is capable of providing slewing capability as well as automated open-
loop solar tracking.
Automated tracking control uses published algorithms to calculate the local solar position in
real time as well as the rate of solar motion [98, 99]. The calculated local solar vector, a supplied
known target vector, and the heliostat position are then used to calculate the mirror orientation
required to provide a normal incoming solar vector to the concentrator after reflection. The algo-
rithm uses this information to calculate the necessary mirror movement rates to maintain a normal
solar vector and uses the established minimum control increments and known step time to repro-
duce the required motion.
An open-loop tracking configuration is advantageous since it only requires an accurate assess-
ment of the heliostat location, target vector and initial placement of the solar vector on the test
section. Not relying on the absolute position of the heliostat for closed loop control was par-
ticularly beneficial during these early stage tests since the heliostat arrived pre-configured with
rheostats for angle sensing. The minimum angular increment of these sensors was found to be
approximately 0.3

which is insufficient for closed-loop operation.
73
Distribution A: Approved for public release; unlimited distribution. PA#15347
The final component of the heliostat requiring restoration was the heliostat mirror panel. The
heliostat drive arrived at USC with a small circular mirror attached that originally served as a hub
for a large spoked mirror panel which can be seen in Figure 5.3. Two 1.2 m by 1.4 m aluminum
honeycomb panels were fixed to the top of the existing circular mirror to create a mounting surface
and 30.5 cm square mirror panels were adhered to create the new mirrored surface. To protect the
mirrored surface, the heliostat was stored inside the lab and rolled outside on rails when in use.
5.1.2 CommercialDiagnostics
The USC solar furnace relies on commercial diagnostics for measuring ambient solar conditions,
characterizing total solar power delivery by the furnace, and the measurement of experimental
temperatures.
Ambient solar conditions are characterized by measuring the direct normal insolation which is
defined as the power per unit area at the surface perpendicular with the solar vector. This measure-
ment, given in W/m
2
, excludes diffuse radiation and represents the sunlight available for reflection
by the heliostat. Measurements of the local solar insolation are performed by an Eppley Normal
Incidence Pyreheliometer. This instrument is considered a “secondary standard” and comes from
the factory with a supplied calibration and stated error of less than 1%. Since the pyreheliometer
must maintain alignment with the solar vector, the instrument has been mounted to an automated
commercial telescope tracking mount. The signal output for the pyreheliometer is recorded during
testing in real time using a 24 bit NI DAQ due to the low signal levels (typically 4-8 mV).
Total power delivery by the solar furnace is characterized using a commercial thermopile based
laser power meter. The Newport 818P-3KW-60 is capable of sustained measurements of up to
3 kW into a 60 mm diameter aperture with an uncertainty of approximately5%. By combining
measurements from the power meter and the pyreheliometer, a relationship between the measured
insolation and total delivered solar furnace power can be obtained. This is used during testing to
estimate power delivery as a function of time.
74
Distribution A: Approved for public release; unlimited distribution. PA#15347
Experimental temperatures are recorded by Type C and Type K thermocouples. Type K ther-
mocouples are logged by a 16 channel NI thermocouple specific DAQ (NI-9213). Type C thermo-
couples are displayed by OMEGA thermocouple controllers which output a linear voltage signal
proportional to the temperature measurement range. This signal is used to log Type C temperatures
through a 24-bit NI DAQ. All thermocouple channels are calibrated and verified using an OMEGA
thermocouple calibrator.
5.1.3 CCDFluxMapping
Complete characterization of the solar furnace system requires accurate mapping of the solar flux
at the focal plane in addition to total power delivery. Accurate flux maps are used to determine
total and peak concentration ratios for the furnace and can also be used to characterize heliostat
panel alignment and concentrator performance vs. simulations.
To provide the required flux maps for the USC solar furnace, a CCD camera based flux map-
ping technique was developed similar to those described in the literature. A typical flux mapping
system consists of a CCD camera, a lambertian surface, and a means of calibration. When taking
measurements, the solar furnace output strikes the lambertian target. The reflected image is cap-
tured with the CCD camera and the local pixel intensity of the image is ultimately converted to
incident solar flux at the imaged location using a calibration factor.
Existing systems produce this calibration factor using either a flux gauge installed within the
target [100] or a point radiometer [101, 102] to provide a real time flux value for a given location. In
contrast, the USC system utilized a bench-top black body source to provide a fixed CCD calibration
factor for a given camera and lens assembly. This simplified the experimental set up but required a
fixed system geometry and fixed components throughout testing.
Two CCD camera assemblies were used for the USC solar furnace to provide two different
fields of view and two different focal lengths. The cameras used were monochromatic CCDs
using the Sony ICX445 CCD sensor coupled to a 12 bit ADC and were selected for both CCD
75
Distribution A: Approved for public release; unlimited distribution. PA#15347
linearity and sensitivity at the IR wavelengths of interest. Each camera was fitted with a 35 mm
lens and a 980 mm bandpass filter to both reduce total intensity and avoid wavelengths affected
by atmospheric absorption. The lenses were locked at an aperture of f/22 and focal lengths were
fixed at 0.43 m and 1.40 m for the “short” and “long” CCD assemblies respectively. The “short”
CCD provided a field of view of 57 mm by 43 mm and the “long” CCD provided a field of view of
53 mm x 71 mm. Images were taken with constant camera settings, except for exposure time, and
the 1296 x 964 pixel output of the CCD chip was down-sampled to record 640 x 480 pixel images.
Calibration
Each camera was calibrated by imaging a blackbody cavity across multiple temperatures. Cameras
were set up so that the aperture for the blackbody cavity was at the designated focal length with
distance confirmed by imaging a reference grid and comparing the result against the desired field
of view. Once this was established, multiple images were taken of the black body cavity across
a range of temperatures from 900 to 1200

C ensuring sufficient time for the blackbody to reach
equilibrium at each temperature and scaling the exposure time of the CCD to ensure that images
were unsaturated.
Calibration images were processed by masking the location of the blackbody aperture and
determining the average pixel intensity per microsecond of exposure time (counts/s). At each
temperature, the black body radiation emitted from the cavity was calculated to get an estimated
irradience value at the target wavelength of 980 nm. This irradiance value and the pixel intensities
across multiple temperatures were then combined to produce a final calibration factor for the CCD
of counts/s vs. spectral irradiance (W/m
2
-nm).
It is important to note that since the geometry of the problem is fixed, the calibration factor
for the CCD is simply chip response vs. estimated intensity at the blackbody surface at pre-
determined wavelength as opposed to a quantification of energy actually striking each pixel of the
76
Distribution A: Approved for public release; unlimited distribution. PA#15347
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
0
2
4
6
8
10
12
14
16
18
20
CCD Pixel Response, counts/µs
980nm Irradiance at Focal Plane, W/m
2
−nm
Figure 5.4: Calibration data for the “long” CCD produced from 36 calibration images at black
body temperatures between 900 and 1200

C. The calibration results in a calibration factor of
59.7 counts/s/W/m
2
-nm
CCD. Calibration data for the “long” camera assembly is given in Figure 5.4 and demonstrates the
expected linear relationship between CCD response and irradiance at the imaged location.
FluxCalculation
Flux map images were taken by placing a quasi-Lambertian target at the plane of interest and cap-
turing an image of reflected sunlight using one of the previously calibrated CCDs. The measured
distance between the focal plane and the camera assembly were confirmed by imaging a grid pat-
tern to check the camera field of view. Camera settings were also kept identical to those used the
calibration process. Image exposures were set to ensure that no regions of the CCD were saturated.
The first step in processing flux map images is to convert the intensity of each pixel to an
approximate blackbody spectral irradiance at 980 nm for that location using the counts/s vs.
77
Distribution A: Approved for public release; unlimited distribution. PA#15347
spectral irradiance (W/m
2
-nm) factor determined during the calibration process. This results in an
intensity map for a specific wavelength with an assumed black body source.
Since the actual image is produced by reflected solar spectrum, this intensity map must be con-
verted. The blackbody irradiance values at 980 nm are compared to the irradiance at 980 nm of the
ASTM G-173 reference solar spectral irradiance provided the National Renewable Energy Labo-
ratory (NREL). This spectrum represents the average the solar radiation intensity per wavelength
at an approximate solar insolation of 900 W/m
2
.
The black body intensity map is divided by the ASTM G-173 reference value of 0.57 W/m
2
-nm
at 980 nm yielding a pixel by pixel map of reflected sunlight measured in “suns”. For example, if a
given pixel measured a black body spectral intensity of 57 W/m
2
-nm, comparison with the ASTM
G-173 spectrum would result in local intensity of 100 “suns” in the region imaged by that pixel.
Since a single “sun” represented in the ASTEM G-173 spectrum has a total power of 900 W/m
2
, the
intensity of each pixel is multiplied by this value giving the desired solar flux map at the lambertian
target in units of W/m
2
. Finally, the flux map image is divided by a reflectivity value to account
for reflection losses in the lambertian target yielding the final flux map for the system.
The reflectivity used for the lambertian target is set at 71% matching total power measurements
in the flux map data to those recorded by the commercial power meter. The lambertian target was
created by covering an aluminum sheet in multiple layers of matte white commercial spray paint
so this reflectivity is within the range of expected values. Noting the uncertainty of power meter
measurements, it is expected that flux mapping calculations have a minimum uncertainty of5%.
The flux mapping system was initially tested using the Fresnel lens-based development furnace
to both quantify the performance of the heliostat and find the optimal location for experimental
performance. The two flux maps given in Figure 5.5 show the output profile of the furnace at both
the geometric focus of the lens and at the region of peak intensity approximately 3 cm closer to the
lens. Flux maps at the geometric focus show low total concentration ratios as well as two separate
approximately rectangular lobes. These regions are a result of misalignment in the heliostat panel
78
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 5.5: Solar flux profiles taken for the Fresnel Lens solar development furnace at the (A)
geometric focus and (B) the location of peak performance. Note that at the geometric focus,
misalignment of the heliostat panels is visible effectively creating two images. Units displayed on
the contours are the number of suns which is equivalent to the concentration ratio.
assembly and were ultimately eliminated by adding more bracing to the panels and ensuring that
they were co-planar.
The flux profile in the region of peak intensity demonstrates the performance gains that can be
achieved with accurate flux mapping at various distances from the lens. Due to aberrations and
alignment inaccuracies, the geometric and “visual” focus points are often not the best performing
region. A series of images mapping the output of the Fresnel lens ultimately showed peak regions
with concentration ratios exceeding 2500:1 representing a more than three fold increase in peak
concentration ratio vs. the geometric focus.
79
Distribution A: Approved for public release; unlimited distribution. PA#15347
5.2 FinalDesign
Transitioning to molten silicon experiments required total power delivery and concentration ratio
improvements over the existing Fresnel lens based development furnace. Initially, a custom manu-
factured solar concentrator was targeted using either a single large glass parabolic mirror or a large
segmented array approximating an ideal parabolic concentrator. However, it was found that custom
manufactured optics on the scale required (concentrator area> 1m
2
) were both cost and schedule
prohibitive resulting in a search for COTS solutions. Ultimately, a modular spherical mirror sup-
plied by Display and Optical Technologies Incorporated (DOTI) of Georgetown, Texas was used
as the primary concentrator and provided sufficient performance for molten silicon testing despite
losses from spherical aberrations.
The USC solar furnace is a two stage design, diagrammed in Figure 5.6. In addition to devel-
oping a new primary concentrator, improvements in the heliostat were required to both increase
coverage for the larger concentrator and reduce flatness errors witnessed during CCD flux mapping
development.
5.2.1 DOTISolarConcentrator
The primary concentrator for the USC solar furnace is an array of four spherical mirrors that are
arranged to form a single spherical concentrator with an effective diameter of 1.78 m and radius of
curvature of 3.15 m. The concentrator assembly was manufactured by Display and Optical Tech-
nologies Incorporated (DOTI) due to their expertise in creating large reflectors and their existing
production capability. While parabolic concentrators are ideal for solar furnace applications, tool-
ing costs for new optics were cost prohibitive and spherical reflectors were required in order to
gain sufficient scale using existing inventory. The individual mirrors that make up the concentrator
were sourced from DOTI’s Wide Angle Collimated (WAC) display program and are a standard
component manufactured for use in high performance flight simulator displays. Adapting these
80
Distribution A: Approved for public release; unlimited distribution. PA#15347
SUNLIGHT
Redirection Mirror
96% reflective
Heliostat
Spherical
Concentrator
Test Section
Figure 5.6: Path diagram for a two stage solar furnace using a spherical concentrator.
readily available mirrors into the final USC solar concentrator produced significant cost savings as
the only custom design work was the fabrication of the mounting structure.
Each concentrator facet is an approximately 1 m x 1 m square aluminized first surface mirror on
13 cm thick glass substrate. The mirrors are specified with a mean slope error of 180 arcseconds
from ideal and a proprietary silicon dioxide based coating results in a reflectivity approaching
90% when weighted against the solar spectrum. Mirrors were shipped with steel mounting rings
cemented to the rear surface containing three threaded rods used for both mounting and alignment.
The steel mounting structure was custom manufactured by DOTI and included a rail system
allowing for the entire mirror assembly to translate along the optical axis. The center of the
mounting structure contained a small optical plate to mount diagnostics. After mounting rails
were installed and aligned, a diode laser was installed on this optical plate to define the optical
axis of the solar furnace assembly. This optical plate was also used to mount CCD flux mapping
diagnostics as well as a standard video camera with an attached neutral density filter to image the
experimental area during testing.
81
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 5.7: Photographs of the mirror array during construction (left) and as seen through the 70
inch aperture curtain. The mirrors are blue during construction due to a protective plastic film
applied before shipment. Note the testing chamber seen in the middle of the concentrator array in
the photograph on the right.
The completed mirror array, shown in Figure 5.7, was aligned using a point source placed at
the radius of curvature along the central optical axis. Each mirror was adjusted so that its reflected
image collapsed onto the point source resulting in a single unified optic.
Due to the influence of spherical aberrations, it was necessary to create an aperture for the solar
concentrator. At the extremes of the mirror array, sunlight is reflected with an effective concentrator
radius of over 1.4 m and the resulting aberrations are unmanageable. Currently, a curtain has been
installed with an aperture of 1.78 m (70 in) resulting in a usable concentrator area of approximately
2.3 m
2
when taking into account the central area used by the diagnostic mount and blockages by
the testing chamber.
5.2.2 HeliostatImprovements
To accommodate the new larger solar concentrator, the heliostat was expanded and a new 2.4 m
by 3.7 m mirror panel was constructed. The original central mirror mount was removed from the
82
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 5.8: Photograph of the USC heliostat mirror array.
heliostat and a new framing assembly was built using aluminum I-beams and a central steel plate
mounted to the heliostat drive components. This framing system can be seem in Figure 5.3. Three
2.4 m x 1.4 m aluminum honey comb panels were fixed to to the aluminum mirror frame and
the assembly was adjusted to be co-planar. Finally, six 2.4 m x 2.4 m second surface float glass
mirrors were mounted to the honeycomb panels using spray adhesive to produce the finished panel
in Figure 5.8. The larger individual mirrors reduced flatness errors compared to the mirror tiling
used in previous heliostat iterations.
The total size of the new mirror array was a trade between concentrator requirements and
available experimental space. The optical axis laser from the concentrator array was used to prop-
erly locate the center of the heliostat drive and the heliostat was permanently mounted outside
the building to accommodate the larger mirror panels. The location of the heliostat (34.020116

,
-118.287748

) was determined using a high accuracy GPS with a radial uncertainty of 1.3 m.
The final upgrade to the heliostat was replacement of the existing rheostat based angle sensors
with digital angle encoders, improving tilt sensing uncertainty to 0.0045

on both axes. To align
83
Distribution A: Approved for public release; unlimited distribution. PA#15347
Heliostat Location
Alignment Point A
Figure 5.9: Satellite photograph (Google Maps) of the USC campus illustrating the heliostat loca-
tion as well as the azimuth alignment vector.
the azimuth axis of the heliostat, a second location on the USC campus (“Point A”) was determined
within an uncertainty of 20 cm via GPS. An azimuth vector was established between the heliostat
and “Point A”, as shown in Figure 5.9, with an uncertainty of 0.27

. A laser was shone from “Point
A” to the heliostat and the heliostat mirror panel was used to reflect the laser back upon the source.
The heliostat was then considered aligned to the reference azimuth vector.
Next, the azimuth vector for the solar concentrator was set by rotating the heliotat and reflect-
ing the optical axis laser installed within the concentrator back upon itself. The relatively large
uncertainty in the azimuth vector is due to the uncertainty of the heliostat location. The surround-
ing buildings result in a weak GPS signal and the accuracy of this measurement can be improved
by using a GPS device capable of simultaneously referencing the GPS and GLONASS systems.
The current accuracy, however, is sufficient due to the use of open-loop control. The altitude axis
of the heliostat was calibrated using a machinist’s level.
84
Distribution A: Approved for public release; unlimited distribution. PA#15347
0 50 100 150 200 250 300 350 400
0
20
40
60
80
100
120
140
160
180
Day of the Year
Minutes
Illumination of Concentrator from 1PM to 4PM in 2013

>95% Coverage
>90% Coverage
>80% Coverage
>70% Coverage
Figure 5.10: Solar concentrator coverage as a function of day of the year. Heliostat size and
placement limitations result in reduced coverage from May-July. Coverage is still sufficient for
year-round experimental operations.
The reflected sunlight from the heliostat is effectively a projection of the tilted mirror assem-
bly onto the plane of the concentrator and the heliostat is unable to provide full coverage during
instances of extreme tilt due to total size limitations. Additionally, the placement of the solar fur-
nace facility in an urban campus only allows for approximately 4 hours of sunlight coverage per
day. Figure 5.10 shows the differences in coverage vs. time during the year and shows that with
the current design, winter testing is ideal. However, there are still windows of full coverage during
the summer.
5.2.3 CharacterizationandCurrentPerformance
The current solar furnace facility was characterized for total output power as a function of solar
insolation and individual component efficiencies. Because the DOTI solar concentrator array is
85
Distribution A: Approved for public release; unlimited distribution. PA#15347
148 150 152 154 156 158 160
−6
−4
−2
0
2
4
6
x, cm
y, cm
Figure 5.11: Ray trace simulation (with only 200 rays for visibility) of the focal area for the DOTI
mirror array. Note the inclusion of the quartz window at an axial distance of 149.9 cm and the
geometric focal length at 157.48 cm as noted by a dotted green line.
spherical, special attention must be given to the effects of spherical aberrations. A ray trace pro-
gram (2-D with weighting to approximate 3-D performance) was written in-house to provide a
predictive model for concentrator output and determine the optimal location for experiment place-
ment. This program takes into account the random slope error of the reflector, heliostat pointing
and flatness error, incoming solar angle distribution, blockage from the test chamber, and the influ-
ence of the quartz window.
Figure 5.11 shows the incoming ray profile of the solar furnace near he geometric focal point
with a 1.78 m aperture. It can be seen that due to spherical aberrations and manufacturing errors,
the effective spot size at the geometric focus is approximately 9 cm in diameter. At different
locations however, there are areas of increased flux density which can allow for reasonable con-
centration ratios provided that peripheral rays are neglected.
86
Distribution A: Approved for public release; unlimited distribution. PA#15347
55%
60%
65%
70%
75%
80%
85%
90%
95%
100%
150 151 152 153 154 155 156 157
Relative Power in a 2.54 cm Spot
Location, cm
Ray Trace Estimate
CCD Set 1
CCD Set 2
CCD Set 3
CCD Set 4
Figure 5.12: Relative power delivered to a 2.54 cm diameter spot vs. distance from the solar
concentrator as predicted by ray tracing and experimentally measured via CCD flux mapping.
Note that the ideal location, both predicted and measured, is approximately 3.8 cm back from the
geometric focus at 157.5 cm.
To determine the optimal experimental location, the power delivered into a 2.54 cm spot was
simulated and experimentally recorded using the CCD flux mapping technique for multiple focal
planes in the region shown in Figure 5.11. Figure 5.12 gives the relative power vs. distance for
the concentrator and indicates that the optimal experimental location is approximately 3.8 cm back
from the geometric focus noting that the predicted and experimentally measured locations coincide.
At this location of best performance, flux maps given in Figure 5.13 indicate peak concentration
ratios in excess of 4000:1 and approximately 90% of total power at the plane included within a
2.54 cm diameter spot.
The end-to-end system efficiency of the USC solar furnace is relatively low at approximately
40%. Table 5.1 lists the individual component efficiencies for the furnace and their relevant con-
tributing factors. The heliostat and the quartz chamber window are the least efficient components
of the system since they suffer from both transmission and reflectance losses. The heliostat, which
87
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 5.13: Flux maps taken at the experimental location for the USC solar furnace. Iso lines are
given in number of suns.
uses commercial second surface soda-lime glass mirrors, is not optimized for solar reflection and
represents the weakest component in the system. The quartz window is also a major loss contrib-
utor despite a high transmittance in the solar spectrum due to reflection losses brought on by the
relatively steep incident angle of rays from the solar concentrator. For future testing, it is possi-
ble to substantially raise the performance of the furnace facility by replacing the heliostat mirrors
with higher cost first surface mirrors and employing a secondary concentrator to achieve higher
concentration ratios.
Despite the low total system efficiency, power levels sufficient for molten silicon experiments
are possible with the current furnace design. Figure 5.14 gives the power delivery to a test sec-
tion vs. local direct-normal insolation and acceptable spot size. This data has been confirmed
utilizing commercially available high flux laser power meters and scaled against measurements
from an Eppley Pyrheliometer.
88
Distribution A: Approved for public release; unlimited distribution. PA#15347
Table 5.1: Component efficiencies of the USC solar furnace
Component Approx. Efficiency Contributing Factors
Heliostat Mirrors 68% Reflection Losses
Transmission Losses
Weathering
Solarization
Surface Imperfections
Dirt
Concentrator Mirrors 90% SiO
2
Coating Transmittance
Aluminum Reflectivity
Quartz Window 73% Reflection Losses
Transmission Losses
1” Acceptance 90% Spherical Abberations
Experimental Design
Total 40%
89
Distribution A: Approved for public release; unlimited distribution. PA#15347
350
450
550
650
750
850
950
1050
1150
500 550 600 650 700 750 800 850 900 950 1000
Power Delivered (W)
Insolation (W/m
2
)
1.90 cm (0.75") Spot
2.54 cm (1.00") Spot
3.18 cm (1.25") Spot
3.81 cm (1.75") Spot
Figure 5.14: Power delivery vs. insolation for the USC solar furnace as a function of acceptable
spot size. Values include losses from the quartz chamber window. Typical insolation at the facility
is between 750-950 W/m
2
depending on atmospheric conditions.
90
Distribution A: Approved for public release; unlimited distribution. PA#15347
Chapter 6
Material Studies and Initial Experiments
Material studies and initial solar furnace experiments were performed concurrently during solar
furnace development to identify a reliable experimental condition for melting silicon. A literature
review of material compatibilities succeeded in identifying potential container materials for both
silicon and boron. Tube furnace tests were conducted validating the short term compatibility of
molten silicon with boron nitride and graphite. Early solar furnace tests were designed using a
0-D thermal model for a test article in a radiation shielding cavity. The results of these tests did
not produce molten silicon. However, they did drive the creation of an in-house 2-D axisymmetric
model leading to the ultimate solar furnace experimental series using cylindrical crucibles.
6.1 MaterialStudies
The use of latent heat phase change materials requires storage containers that are both chemically
compatible and structurally sound at the PCM melting temperatures. Identifying suitable contain-
ers for molten silicon is simplified due to the relatively large quantity of industrial research from
the manufacture of silicon wafers. In contrast, experiments to-date studying liquid boron have
been primarily concerned with simply creating the material as opposed to thorough investigation
of long term material compatibilities. Even though boron experimentation is out of the scope of
91
Distribution A: Approved for public release; unlimited distribution. PA#15347
this research, material compatibility information is still included here for completeness and for the
reference of future experimental designers.
6.1.1 BoronCompatibility
A literature review has indicated that molten boron has been created in the presence of refractory
metals, graphite and various ceramics. In the case of refractory metals, specifically tungsten and
tantalum, direct contact with molten boron has resulted in limited success during effusion cell
testing for vacuum vapor deposition [103, 104, 105]. These studies measured contamination in the
thin films themselves as opposed to the bulk boron sample, but it can be inferred from published
results that molten boron actively attacked tungsten cells and that the purity of films produced using
tantalum cells was a function of favorable vapor pressures for pure boron versus tantalum boride
contaminants [106, 107]. Molybdenum has not been used in experimental molten boron testing
but contamination is likely due to the instability of molybdenum borides well below the melting
point of boron [108]. Boron sample contamination when using a refractory container appears
probable, however it is unclear if certain high temperature borides might produce a protective
layer preventing further contamination (as in the case of aluminum oxide on aluminum) or allow
for further degradation with time (as with rust on iron).
Graphite was used experimentally as a molten boron crucible material by Stout, et al. [109].
This testing yielded heavy boron carbide contamination of the bulk sample, indicating that solid
boron carbide did not remain at the boron-graphite interface, but rather moved through the molten
boron mass. It is important to note that the contamination process began only when the boron was
in the liquid state, providing an example of increased reactivity after phase change.
Due to contamination issues with graphite and the refractory metals, ceramics, particularly
boron nitride (BN), emerge as the predominant crucible material for containing molten boron.
Multiple studies cite negligible contamination of a boron sample melted in contact with BN, which
92
Distribution A: Approved for public release; unlimited distribution. PA#15347
is attributable to BN’s low reactivity at high temperatures and a resistance to attack in a boron-
rich environment [109, 110, 111, 112]. Additional experimental considerations, however, must
be made when using BN due to its strong tendency towards dissociation into liquid boron and
nitrogen gas at temperatures above 2300 K [113]. Dissociation can be prevented by maintaining a
system pressure above the equilibrium pressure of dissociated nitrogen and it is estimated that this
pressure will be between 0.1 and 10 Torr (13-1300 Pa) [114, 115]. Previous experimental efforts
have maintained a suitable system pressure by operating in inert gas environments, or in the case
of Stout et al., by sealing the boron nitride in a graphite vessel and allowing dissociated nitrogen
to pressurize the container and prevent further decomposition. As an alternative to BN, Holcombe
et al. reported the use of halfnium diboride containers with no mention of contamination issues.
Due to the extent of reported BN use with favorable results, this work suggests BN as the container
material for a future molten boron system.
6.1.2 SiliconCompatibility
Using molten silicon as a PCM also requires careful selection of a container material. However, a
large body of research exists due to industrial silicon processing. Molten silicon is routinely han-
dled in the semiconductor industry using silica (SiO
2
) crucibles for single crystal silicon production
and graphite crucibles for the industrial manufacture of large solar grade multicrystaline silicon
ingots. In the Czochralski process for producing single crystal silicon, molten silicon reacts with
the SiO
2
crucible to produce SiO which then evaporates from the bulk material [116]. While this
is acceptable during the relatively short crystal pulling process, the Si-Si0
2
reaction prevents the
use of SiO
2
for long-term storage applications. In the case of graphite crucibles, minimum sample
contamination is reported when operating in an inert environment provided that the graphite meets
certain porosity requirements to prevent wetting. Ciszek and Schwuttke state that carbon con-
tamination on the order of 20ppm is possible using graphite crucibles provided that the graphite
density is> 1.75 g/cc and that the graphite grain size is< 50m [117].
93
Distribution A: Approved for public release; unlimited distribution. PA#15347
BN also appears to be an acceptable molten silicon container providing synergy with future
molten boron testing and removing concerns about high temperature graphite reacting with other
system materials. Drevet et al. states that molten silicon in contact with BN will react to form a thin
layer of silicon nitride (Si
3
N
4
), but this reaction is self limiting, likely due to boron saturation of the
liquid silicon bulk [118]. It is also expected that a small amount of boron within the silicon sample
will have little effect on solidification. Wang et al. states that for boron concentrations below 2%,
defect free silicon crystals are expected with boron substitutions which should minimally affect
the latent heat release [119]. It is important to note, however, that these assumptions are based on
relatively small molecular dynamics models. In addition to BN, Si
3
N
4
is used as a release agent for
silicon processing and may be a usable crucible material itself. However, a lack of machineability
makes its use impractical in this work.
Initial tube furnace experiments were performed with molten silicon in both graphite and BN
crucibles to ensure experimental compatibility. The first set of tests used both bare graphite and
BN lined crucibles loaded with 99+% pure, 325 mesh amorphous silicon powder that were then
placed in the tube furnace and held at 1550

C for approximately 12 hours under an argon purge.
After sectioning the crucibles, it was seen, contrary to expectations, that the silicon had formed
into multiple smaller beads as opposed to a single silicon mass. It was believed that this behavior
was due to contamination of the silicon powder by SiO
2
. Since the particles were small (44m
and smaller), the thin SiO
2
layer formed on the outer surface of the particles was significant with
respect to the total silicon mass. A second round of testing alleviated this problem by switching to
99.9999% pure 1-3 mm silicon chips. These tests, shown sectioned in Figure 6.1, were similarly
held at 1550

C for 12 hours and the behavior of the molten silicon was as expected. Significant
wetting is shown in the bare graphite crucible which is in line with predictions made by Ciszek
about graphite suitability. The NAC-500 graphite used in these tests has a grain size well above that
suggested and an average density slightly below the requirements for non-wetting behavior. In the
case of the BN lined crucible, there was minimal wetting of the BN surface. For both test sections
94
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 6.1: Cut-away diagrams and photographs of (A) a bare graphite crucible and (B) a graphite
crucible with a BN liner as tested in a high temperature tube furnace. The photographs were taken
after heating silicon chips for 12 hours at 1550

C. Note that the grey contamination inside the BN
sleeve is a result of coolant flow during the cut-away process. Dimensions given in inches.
there appears to be no evident short term damage, but longer duration tests will be required in the
future to assess behavior across multiple cycles. These tests were originally scheduled and test
articles were created, however, budget cutbacks prevented the completion of any long term testing.
It also must be noted here that Figure 6.1 shows the formation of freezing point singularities
in the re-solidified silicon. These semi-conical structures are produced by the combined effects of
surface tension and volumetric expansion during freezing [120]. Silicon is unique in that is is one
of the few materials that expands during freezing and the resulting volume change of approximately
10% poses significant challenges if silicon is to be used as a PCM [121]. In the existing literature
95
Distribution A: Approved for public release; unlimited distribution. PA#15347
concerning silicon as a thermal energy storage media, there is no mention of this expansion issue.
As a result, this was not a design driver until late in this research campaign.
6.2 PreliminarySolarFurnaceTests
The first series of experiments were performed using the Fresnel lens based development furnace
with the goal of simply producing liquid silicon using concentrated sunlight. The design driver for
these tests was a paper published in 1990 by Steinfield and Fletcher which presented a closed form
solution for the temperature of a solid spherical solar furnace test article in a reflecting spherical
cavity [122]. Steinfeld’s paper calculated receiver temperature by performing a radiation bal-
ance between a centrally located test section, a truncated spherical radiation shield, and a circular
entrance aperture. A diagram of the modeling geometry is reprinted in Figure 6.2 and shows that
for modeling purposes, the centrally located receiver was divided into “front” (A
1f
) and “back”
(A
1b
) sections effectively dividing the system into four regions. Surface to surface radiation was
considered in the model and the paper contained Monte-Carlo ray trace results supplying view
factors for the receiver regions after reflection in the spherical radiation shield.
Steinfeld suggested the use of graphite for the test section and aluminum for the radiation
shields. Due to high thermal conductivity of both materials, all regions of the model were assumed
to be isothermal meaning that the front and back temperatures of the central receiver were identical.
Additional assumptions in the model state that all incident incoming solar radiation strikes the
receiver directly on the surface A
1f
and that surface A
1b
is positioned so it is hidden from the
aperture resulting in a a view factorF
31b
= 0. For the mathematical formulation of the model and
the supplied solution, the reader is directed to the original paper.
An experimental apparatus was constructed in an attempt to achieve molten silicon tempera-
tures by approximating the configuration in Steinfeld’s model. Figure 6.3 shows the overall system
geometry including a “bullet” shaped test section, spherical aluminum radiation shields, support
96
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 6.2: Basic three body modeling geometry for an isothermal spherical test section in a spher-
ical radiation shield. Note that the isothermal test section is broken into two sections to aid in the
formulation of the radiation balance. Diagrams reprinted from Steinfeld and Flecther [122].
structure, and the tantalum sheathed Type C thermocouple which served as a the primary temper-
ature diagnostic and sting mount for the centrally located test section. Since the Steinfeld model
required that all incident solar radiation strike the receiver, the minimum reliable spot produced by
the Fresnel lens set the test section diameter at 0.75 inches. “Bullet” shaped crucibles were used
as opposed to spherical due to the limits of in-house machining capability.
The outer diameter of the radiation shielding was determined by the availability of aluminum
hemispheres since model results were invariant with respect to shield size provided the entrance
angle remained constant. Three inch diameter aluminum hemispheres were sourced and polished
in-house using a lathe to create the necessary reflective surface. Rough sanding was first performed
with sand paper ranging from 400-2500 grit in 200 grit increments. The sanded aluminum surface
was then wet polished using 3M Zona polishing paper decreasing in grit size from 15 - 1 microns.
Finally, a finishing pass was performed using Mother’s Billet Aluminum Polish. The entrance
rim angle of 40

was also set by the Fresnel lens input geometry and was created by sectioning a
97
Distribution A: Approved for public release; unlimited distribution. PA#15347
A
A
3.500
0.750
0.475
0.750
R1.375
Type C
Thermocouple
Probe
Polished Aluminum Radiation Shield
Graphite Crucible
PCM Cavity
Steinfeld Assembly No Explode
WEIGHT:
A3
SHEET 1 OF 1 SCALE:1:1
DWG NO.
TITLE:
REVISION DO NOT SCALE DRAWING
MATERIAL:
DATE SIGNATURE NAME
DEBUR AND
BREAK SHARP
EDGES
FINISH: UNLESS OTHERWISE SPECIFIED:
DIMENSIONS ARE IN MILLIMETERS
SURFACE FINISH:
TOLERANCES:
LINEAR:
ANGULAR:
Q.A
MFG
APPV'D
CHK'D
DRAWN
Figure 6.3: 2-D cutaway schematic of the initial solar furnace test assembly showing the spherical
radiation shield, support structure, thermocouple sting mount and “bullet” style crucible. Note that
for pure graphite tests, the PCM cavity was removed and the thermocouple was placed in the center
of the test section. Dimensions given in inches.
completed hemisphere. Figure 6.4 shows the completed radiation shield along with the rest of the
test apparatus.
With the Fresnel lens furnace and the shielding system discussed above, test articles made from
solid graphite (no PCM cavity) were able to achieve molten silicon temperatures with an estimated
power input of 240 W. This power input includes both the estimated direct input to the crucible
surface as well as sunlight that initially misses the crucible yet enters the shielding cavity agreeing
with the Fresnel lens characterization values for the observed solar insolation within 5%. Cooling
curve analysis of the experimental results suggested a 55% reduction in radiation losses via the
radiation shielding.
The results of these tests showed that molten silicon temperatures could be reached. However,
they were not in agreement with model predictions. Steinfeld’s model suggested a 70% reduction
in radiation losses (with a shield reflectivity of 0.8) and a minimum required power of 188 W to
98
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 6.4: Photographs of the radiation shielding test assembly. Clockwise from top left: Radia-
tion shield half after polishing, graphite test article mounted on thermocouple sting mount, assem-
bled test article, crucible immediately after removing solar furnace power.
reach 1680 K. It was expected that the experiment would require more power and exhibit reduced
shielding efficiency since the experimental test sections have 22% more surface area and are non-
spherical. This geometry increase suggests that 83% would be required over that predicted by
Steinfeld’s model. However, experiments demonstrated molten silicon temperatures with only a
28% power increase.
Despite disagreement with Steinfeld’s predictions, experiments with this geometry continued
since molten silicon temperatures had been experimentally confirmed. Crucibles were hollowed to
create the PCM cavity shown in Figure 6.3 and small BN containers loaded with silicon chips were
inserted. Throughout multiple tests, silicon within the PCM cavity showed no evidence of melting
even though thermocouple measurements indicated temperatures above 1680 K. Later inspection
99
Distribution A: Approved for public release; unlimited distribution. PA#15347
revealed that during these tests, a combination of thermocouple bead placement within the probe
and focusing of scattered sunlight by the radiation shields was inflating true measured temperature
of the crucible.
All design and testing during this series relied on Steinfeld’s modeling assumption that the
crucible would be an approximately isothermal body. Once a PCM and a PCM container were
added to the crucible, modeling results, along with pyrometer and updated thermocouple data,
demonstrated crucible thermal gradients of 100’s of Kelvin which prevented PCM melting. As a
result, this series of tests was halted in order to construct an accurate multidimensional model of
the system to aid in test section design.
100
Distribution A: Approved for public release; unlimited distribution. PA#15347
Chapter 7
Predictive Model
After completing initial solar furnace testing using simplified formulations to solve for experi-
mental temperatures, disagreement with experimental values highlighted the need for a full multi-
dimensional and transient solution for test section performance. Thus, an in-house model was
written in MATLAB to predict test section heating and cooling. In addition to serving as a pre-
dictor for test performance, the model was also intended to assess the fidelity required in freezing
kinetics models to capture essential system behaviors.
7.1 ModelGeometry
The MATLAB model is a cylindrical, axisymmetric 2-D (r,z) simulation. A cylindrical geometry
was selected for both ease of calculation and ease of manufacture for eventual experimental test
sections. A concession was made in both the design of the model and subsequent experiments by
utilizing a flat plate solar absorber as opposed to a radiation cavity. While this significantly reduced
computational difficulty, it similarly reduced the thermal efficiency of test articles.
The geometry shown in Figure 7.1 gives the general scale of test sections and was initially
set by using SolidWorks Simulation Professional to determine maximum experimental size. Also
101
Distribution A: Approved for public release; unlimited distribution. PA#15347
shown in Figure 7.1 are the 14 adjustable regions in the model which are used to set the geometry
of individual test section components as well as their materials.
Each material is modeled with fully temperature dependent properties using values primarily
taken from the TPRC Data Series [123]. This is required to accurately represent the thermal inertia
of the system and is especially important when considering the rapid decrease in thermal conduc-
tivity of the PCM as it transitions from liquid to solid. Further assumptions in the MATLAB model
include a neglect of thermal contact resistance and thermal expansion.
7.2 SolutionMethodandBoundaryConditions
A fixed-grid, energy balance method is used to solve for the temperature profile as a function of
time [124]. Heat trasnfer between adjacent nodes is assumed to be conduction only and the relevant
conduction equations are described in Appendix A. The resulting solutions are 2-D axisymmetric
transient temperature profiles.
The modeling geometry shown in Figure 7.1 is represented by a 25 x 37 grid of nodes in the
r andz directions respectively at the coarsest setting. It is possible to independently set the grid
spacing for each of the 14 geometry regions provided that each region contains a minimum of three
nodes in both ther andz directions to accurately resolve material boundaries. Time steps are on the
order of 1 millisecond resulting in a total solution time of 6-8 hours for a full heating and cooling
cycle in the model. To produce the models shown in this paper, the grid was increased to 50 x 74
nodes for higher resolution in temperature maps and the time step was reduced to 0.2 milliseconds
to maintain stability. Thermal performance at monitoring locations between the two models was
identical showing that the coarse settings are sufficient to predict overall system behavior.
To account for the phase change process in the PCM, the “enthalpy method” is used in a similar
formulation to that presented in Elgafy et. al [125]. For computational nodes containing the PCM,
a latent heat value is assigned and treated as a source or sink when that node is in a temperature
102
Distribution A: Approved for public release; unlimited distribution. PA#15347
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
0
0.005
0.01
0.015
0.02
0.025
z distance, m
r distance, m
Figure 7.1: MATLAB modeling geometry showing both a 2D cutaway of a nominal cylindri-
cal experimental test section and the geometric representation for antisymmetric computations.
Dimensions, node spacing, and material properties can be independently set for each of the four-
teen regions shown in the lower figure.
103
Distribution A: Approved for public release; unlimited distribution. PA#15347
range known as the “mushy” zone. Once a node enters the “mushy zone”, currently defined as
T
m
0.1 K, all energy leaving or entering the node is assigned to the phase change process and the
node remains at a constant temperature. When a particular node exhausts the assigned latent heat
energy during cooling, the temperature is allowed to change and sensible heat cooling resumes.
Since the model formulation is a fixed grid with fixed control volumes, motion of the PCM
when in the liquid state is neglected. Additionally the model assumes a 100% fill factor at all times
and ignores effects from density change during melting a freezing. This simplified treatment of the
phase change process does not consider convective heat transfer within the liquid PCM.
When modeling the cooling process, neglecting convection is justified due to low thermal gra-
dients with the liquid PCM. Sensible heat quickly dissipates from the PCM when solar furnace
power is removed from the system and thermal gradients within the test section re-orient. Once the
PCM reaches the melting temperature, all thermal gradients are supported by the quasi-isothermal
phase change process and the remaining liquid PCM maintains a constant temperature within the
“mushy zone.” This very stable temperature environment and a low Prandlt number ( 0.02) result
in Rayleigh numbers less than 10
3
denoting conduction dominated heat transfer.
For heating models, it is important to consider convection effects within the liquid PCM as this
will increase the effective heat transfer rate leading to both shorter melt times and higher tempera-
tures in the rear of the test section. For the geometry considered here, temperature gradients within
the liquid PCM can be in excess of 100 K when the system is in equilibrium with the solar furnace
input power. These large thermal gradients correspond to Rayleigh numbers of approximately 10
5
and it is expected that buoyancy driven flows will be established.
However, in the context of this experiment and the ultimate goal of operating in a microgravity
environment, it is also reasonable to neglect convection during heating for the purposes of the
MATLAB model. For the geometry considered here, increasing or decreasing the effective thermal
conductivity of the liquid PCM by a factor of two results in only a<2% increase or decrease in
bulk PCM temperatures. Furthermore, after the heat source is removed, the phase change process
104
Distribution A: Approved for public release; unlimited distribution. PA#15347
dominates and container temperatures are established relative to the melting temperature of the
PCM and not the initial condition.
At the outer surfaces of the model, a radiative and convective boundary condition is applied.
The natural convection boundary condition accounts for operation in the 150 torr argon testing
atmosphere and is approximated by published empirical heat transfer coefficient correlations for
cylindrical bodies (h = 6.6 and 5.7 W/m
2
K for the vertical faces and cylindrical body respectively)
[126]. Operating in this argon atmosphere is required to prevent reaction of experimental materials
as described in Section 8.1 and convection effects account for< 10% of total heat losses.
The radiation boundary condition is calculated to account for the small shielding effect pro-
vided by the vacuum chamber. Integrating radiation shielding into the model required the cal-
culation of node to node view factors as well as integrating radiosity calculations into the finite
difference solutions at the boundary.
7.2.1 RadiationShieldingIntegration
Radiation shielding was included in the model with the option to model either a cylindrical or
spherical shield. Radiation shielding calculations are performed via a radiation energy balance
on each exterior node in a manner similar to that described in Steinfeld and Fletcher [122]. All
surfaces are considered opaque grey bodies.
For all exterior nodes 1 throughn, the radiosity per unit areaB
n
is calculated as
B
n
=
n

n
(T
4
n
T
4
amb
) +
n
H
n
(7.1)
where is the emissivity, is the Stefan-Boltzmann constant, T
n
is the node temperature,
n
is
the local reflectivity andH
n
is the total radiation incident on the node per unit area. The first term
on the right represents the local thermal emission and the second term represents the portion of
incident radiation that is reflected off of the node surface.
105
Distribution A: Approved for public release; unlimited distribution. PA#15347
The total radiation incident per unit areaH
n
is calculated as
A
n
H
n
=
shield
A
n
B
n
F
nrn
+
shield
m
X
i=1
A
m
B
m
F
mrn
(7.2)
whereA
n
is the area of noden,F
nrn
is the view factor from noden to noden after reflection in the
radiation shield, and
shield
is the reflectivity of the radiation shielding. The first term on the right
represents the portion of radiation emitted from node n which is reflected back onto itself. The
second term is a summation representing the total incident radiation after reflection from all other
nodes and view factorF
mrn
corresponds to the view factor from nodem reflected in the radiation
shield to noden.
CalculatingB
n
andH
n
for all nodes yields the following system of equations
B
1
=
1

1
(T
4
1
T
4
amb
) +
1
H
1
B
2
=
2

2
(T
4
2
T
4
amb
) +
2
H
2
.
.
.
B
n
=
n

n
(T
4
n
T
4
amb
) +
n
H
n
H
1
=
shield
(A
1
B
1
F
1r1
+A
2
B
2
F
2r1
+:::A
n
B
n
F
nr1
)=A
1
H
2
=
shield
(A
1
B
1
F
1r2
+A
2
B
2
F
2r2
+:::A
n
B
n
F
nr2
)=A
2
.
.
.
H
n
=
shield
(A
1
B
1
F
1rn
+A
2
B
2
F
2rn
+:::A
n
B
n
F
nrn
)=A
n
(7.3)
which can be solved for eachB
n
andH
n
. The net energy flux at each node can then be calculated
simply as the difference between power leaving the node and power incident upon the node.
q
n
=A
n
(B
n
H
n
) (7.4)
106
Distribution A: Approved for public release; unlimited distribution. PA#15347
All of the terms given in Equations 7.3 can be easily obtained with the exception of the node-
to-node view factors after reflection. Thus, a ray trace code was written by Dr. David Scharfe
at the AFRL in order to produce the requirednxn matrix of view factors. The view factor code
accepts the test section and shield geometry and then pseudo-randomly launches 50 million rays
distributed by total area. Rays are tracked, logging both their source and end points, and taking into
account rays requiring multiple reflections within the radiation shield to either strike the crucible or
exit the shield via the entrance aperture. The result of this ray trace code, illustrated in Figures 7.2,
7.3 and 7.4, yields a matrix of point to point view factors in the format
2
6
6
6
6
6
6
6
6
6
6
4
F
1r1
F
1r2
F
1r3
::: F
1rn
F
2r1
F
2r2
F
2r3
::: F
2rn
F
3r1
F
3r2
F
3r3
::: F
3rn
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
F
nr1
F
nr2
F
nr3
::: F
nrn
3
7
7
7
7
7
7
7
7
7
7
5
(7.5)
With view factors calculated, Equations 7.3 can then be solved. At each time step, energy balances
must be recalcualted so the solution to Equations 7.3 is reduced to a series of matrix operations
to speed computations. The first step in finding the solution is to substitude all the B
n
terms in
Equations 7.3 to get a system of equations solely in terms ofH
n
. Each equation forH
n
then takes
the form of
H
n
= (
shield
=A
n
)
m
X
i=1
A
m
F
mrn

m
(T
4
m
T
4
amb
) +A
m
F
mrn

m
H
m
(7.6)
Multiplying both sides of Equation 7.6 by (A
n
=
shield
), moving allH
n
dependent terms to the left
and all other terms to the right yields the following series of equations
107
Distribution A: Approved for public release; unlimited distribution. PA#15347
(F
1r1

1
1=
shield
)A
1
H
1
+A
2
F
2r1

2
H
2
+:::A
n
F
n
r1
n
H
n
=A
1
F
1r1

1
(T
4
1
T
4
amb
)
A
2
F
2r1

2
(T
4
2
T
4
amb
)A
n
F
nr1

n
(T
4
n
T
4
amb
)
A
1
F
1r2

1
H
1
+ (F
2r2

2
1=
shield
)A
2
H
2
+:::A
n
F
n
r1
n
H
n
=A
1
F
1r2

1
(T
4
1
T
4
amb
)
A
2
F
2r2

2
(T
4
2
T
4
amb
)A
n
F
nr2

n
(T
4
n
T
4
amb
)
.
.
.
A
1
F
1rn

1
H
1
+A
2
F
2rn

2
H
2
+::: (F
nrn

n
1=
shield
)A
n
H
n
=A
1
F
1rn

1
(T
4
1
T
4
amb
)
A
2
F
2rn

2
(T
4
2
T
4
amb
)A
n
F
nrn

n
(T
4
n
T
4
amb
)
(7.7)
It is now possible to solve for allH
n
values via matrix operations in MATLAB by converting the
above system of equations into the following
2
6
6
6
6
6
6
6
6
6
6
4
LHS
3
7
7
7
7
7
7
7
7
7
7
5
2
6
6
6
6
6
6
6
6
6
6
4
H
1
H
2
H
3
.
.
.
H
n
3
7
7
7
7
7
7
7
7
7
7
5
=
2
6
6
6
6
6
6
6
6
6
6
4
RHS
3
7
7
7
7
7
7
7
7
7
7
5
(7.8)
where each rown of theLHS matrix is in the format
[A
1
F
1rn

1
A
2
F
2rn

2
::: (F
nrn

n
1=
shield
)A
n
] (7.9)
108
Distribution A: Approved for public release; unlimited distribution. PA#15347
noting that terms along the diagonal are of a different form to accommodate an additional 1=
shield
term and each rown of theRHS matrix is defined as
[A
1
F
1rn

1
(T
4
1
T
4
amb
) A
2
F
2rn

2
(T
4
2
T
4
amb
) ::: A
n
F
nrn

n
(T
4
n
T
4
amb
)] (7.10)
Only theRHS matrix is temperature dependent. Thus, theLHS matrix can be pre-calculated
before running the time dependent simulation. Additionally, all non-temperature dependent terms
in the RHS matrix can also be pre-calculated and supplied in a matrix format to quickly build
RHS for each iteration.
Once allH
n
values have been found,B
n
values can be calculated via Equations 7.3 and the net
energy flux from radiation of each node can be calculated using Equation 7.4. These energy flux
values are then inserted into the finite difference calculations at each exterior node.
The current radiation shielding geometry is an open topped cylinder which approximates the
stainless steel testing chamber and neglects potential reflections from the quartz window. Calcula-
tions indicate an approximately 20% drop in radiation losses by including shielding effects in the
model.
7.3 Results
Cooling behavior was simulated for a test section with the same geometry as that given in Fig-
ure 7.1 after bringing the test section to thermal equilibrium with a simulated solar furnace input
power of 740 W. Figure 7.5 show the approximate freezing profile of the silicon PCM and it can be
seen that the freezing profile is highly asymmetric. Note that this deviates from the symmetrical,
109
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 7.2: Example ray trace plot for view factor calculation using only 50 rays for visibility. Rays
are launched from the “line of nodes” indicated in yellow and are tracked as they interact with the
open topped cylindrical radiation shield. Rays leaving the crucible are in blue, rays returning to the
crucible are in green, rays leaving the shield after one reflection are given in red, and rays leaving
the shield after no reflections are given in pink.
adiabatic wall treatment of silicon as a PCM in the existing literature. With “real-world” bound-
ary conditions, heat loss from all parts of the container results in a multi-dimensional phase front
ultimately leading to regions of molten silicon encased in solid silicon. As will be discussed in
Chapter 8, the expansion of this trapped liquid silicon results in high stress concentrations and the
potential for container damage.
110
Distribution A: Approved for public release; unlimited distribution. PA#15347
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
Region−Region Reflected VF, total rays = 1000000
Position on Line Of Nodes where Ray Strikes
Position on Line of Nodes where Ray Launches

Front/Unshielded Face Circumferential Face Back/Shielded Face
Front/Unshielded Face Circumferential Face Back/Shielded Face
0.005
0.01
0.015
0.02
0.025
0.03
0.035
Figure 7.3: Ray trace results for a 1,000,000 ray calculation illustrating the magnitude of node to
node reflected view factors within the modeled geometry. Because the radiation shield is modeled
as an open top cylinder, the largest node to node view factors after reflection are calculated for
nodes at the back/shielded face. Note that the scale has been set to highlight the relatively low
view factor values.
It was also determined using the MATLAB model that using polished radiation shielding as
opposed to the case ZrO
2
ceramic insulation would only result in minor performance gains in
both experimental scale and maximum achievable temperature. Using cast ceramic as the primary
insulation is less labor intensive than the radiation shield polishing process described in Chapter 6.
Thus, future experiments eliminated any radiation shielding component beyond that provided by
the testing chamber.
A control simulation was also performed for a graphite only test article. This test, which is
diagrammed in Figure 7.6, maintains the same geometry but eliminates both the PCM and the
111
Distribution A: Approved for public release; unlimited distribution. PA#15347
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
Region−Region Multi−Reflected VF, total rays = 1000000
Position on Line Of Nodes where Ray Strikes
Position on Line of Nodes where Ray Launches

Front/Unshielded Face Circumferential Face Back/Shielded Face
Exit Shield
Front/Unshielded Face Circumferential Face Back/Shielded Face
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Figure 7.4: Ray trace results for a 1,000,000 ray calculation illustrating the likelihood of rays
exiting the radiation shielding. Because the radiation shield is modeled as an open top cylinder,
the majority of rays from the front/unshielded face escape the shield. Note that the results given in
Figure 7.3 are obscured by the large scale.
PCM container. Control results for a simulated rear-mounted type C thermocouple are given in
Figure 7.7 and compared against those for a test article containing silicon PCM.
The graphite-only case initially shows a higher temperature at the simulated thermocouple due
to the elimination of liquid silicon and BN container from the test section core which both have a
relatively low thermal conductivity compared to the bulk graphite. As expected, the graphite-only
case exhibits a smooth cooling curve as sensible heat is removed from the system. In contrast, the
test section containing the silicon PCM begins at a lower temperature and demonstrates a region
112
Distribution A: Approved for public release; unlimited distribution. PA#15347
Temperature (K)
t = 20 s
t = 60 s
t = 100 s
t = 40 s
t = 80 s
t = 120 s
Figure 7.5: Thermal profiles as a function of time calculated by the in-house MATLAB model for
the test section given in Fig. 7.1. Note that the model is axisymmetric so the red outlined region
of the cutaway drawing is the region represented by the thermal maps. Also note that grey is used
in the thermal maps to represent liquid silicon and it is apparent that liquid silicon will become
trapped during the freezing process. Dimension given in inches.
of relative temperature stability due to energy release during the phase change process. Note in
both curves that there is a roughly 10 second region of temperature stability after simulated solar
furnace power is cut while thermal gradients within the test section re-orient.
The mean surface temperature of the exposed graphite surface on the front of the test section is
also affected by the phase change process. However, it must be noted that before the phase change
window, the thermal performance of the PCM under-performs the graphite only case. Calculated
mean temperature curves given in Figure 7.7 show that the silicon based case quickly drops below
the sensible heat case before being supported by latent heat release. Since the thermal conductivity
113
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 7.6: Cut-away diagram of a graphite only cylindrical test article. Dimensions given in
inches.
0 50 100 150 200 250
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
Time, s
Rear Thermocouple Temperature, K

0 50 100 150 200 250
1000
1200
1400
1600
1800
2000
Time, s
Front Surface Mean Temperature, K

With Silicon PCM
Graphite Only
With Silicon PCM
Graphite Only
Figure 7.7: Temperature predictions for both the test section shown in Figure 7.1 and a pure
graphite test section shown in Figure 7.6. Both test sections were bought to equilibrium with a
simulated solar furnace input power of 740W and furnace power was cut at time zero. The figure
on the left gives temperatures for a simulated rear mount Type C thermocouple and the figure on
the right gives a mean surface temperature for the exposed graphite surface on each test section.
114
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 7.8: Comparison of experimental and MATLAB data for graphite only test sections using
the geometry given in Figure 7.6. Note that experimental and simulated temperature traces match
within2%.
of the PCM is low, the relatively isolated graphite on the front surface is able to cool quickly
without being able to draw heat from the remainder of the system.
The USC solar furnace was used to produce experimental data for the graphite only test sections
with the same geometry as those shown in Figure 7.6. Comparison of model and experimental data,
given in Figure 7.8, shows that during the cooling process, temperatures for a rear mounted Type C
thermocouple match within2% indicating that the sensible heat calculations within the in-house
MATLAB model are correct. Validation of the latent heat components of the MATLAB model
required molten silicon testing which is described in detail Chapter 8.
115
Distribution A: Approved for public release; unlimited distribution. PA#15347
Chapter 8
Molten Silicon Testing
The completion of the USC Solar Furnace and the advent of accurate prediction capabilities
enabled a series of tests which produced molten silicon in the laboratory. The intent of these
tests was to uncover practical concerns with using molten silicon as a phase change energy storage
material that thus far had been neglected in the literature which consists primarily of simple, mate-
rial property based comparisons. Additionally, molten silicon experiments were used to validate
the results of the in-house MATLAB model and demonstrate that macro scale results relevant to
a thermal energy storage system could be obtained with relatively simple handling of the phase
change process. While experiments were limited to small quantities of silicon, they succeeded in
highlighting multiple design concerns and proposed potential solutions for future molten silicon
based designs.
8.1 TestDesignandProcedure
Molten silicon tests were performed with cylindrical test sections keeping geometry consistent
with that of the models discussed in Chapter 7. The primary design drivers for solar furnace test
sections were ease of in-house manufacture and consistency with modeling capability as opposed
116
Distribution A: Approved for public release; unlimited distribution. PA#15347
to maximum thermal efficiency. As a result, final test sections lack a cavity receiver in lieu of a flat
plane absorber.
Test sections contain approximately 9 g of silicon and were limited in scale by the affordabil-
ity of commercial silicon rod stock in addition to total solar furnace power. Test sections were
designed with a minimum solar furnace input of 750 W and models identified potential geome-
tries supporting over 30 g of molten silicon during the design phase. However, these geometries
required the use of silicon chips or powder for loading and early solar furnace testing demonstrated
that these packing methods yielded maximum fill factors (actual silicon volume vs available PCM
container volume) of approximately 60%. To achieve a 100% fill factor, silicon rod was required
and 1.6 cm diameter rod was sourced within budget. Test sections were designed around this mate-
rial which fixed the diameter of the PCM container. Figure 8.1 shows a cutaway diagram of an
experimental test section identifying the overall geometry and individual test section components.
Test section construction begins by loading a silicon rod into a cylindrical HBC grade BN
container. HBC boron nitride was selected for all BN components since it lacks a boric oxide
binder which can precipitate at high temperatures [3]. The BN container is sealed with a press fit
BN lid and the finished PCM container is then inserted into a graphite sleeve. The graphite sleeve
serves as both the solar energy receiver and primary emitter for the test section. Additionally, the
inclusion of graphite helps to spread heat within the test section due to the relatively low thermal
conductivity of solid silicon and BN. The graphite sleeve is capped with a friction fit lid containing
a hole allowing a bare wire Type C thermocouple, sheathed in a dual bore Al
2
O
3
tube, to be in
direct contact with the inner PCM container. Ultimately, the Al
2
O
3
thermocouple sheath is used as
a sting mount to support the test section in the testing chamber.
To complete construction of each test section, the inner BN and graphite assembly is placed
in a stainless steel mold and potted using commercially available Rescor 760 castable ceramic
compound. This ZrO
2
ceramic product is low cost, easy to cast, and has an acceptably low thermal
conductivity (approximately 0.93 W/mK dependent on mixing ratio and curing procedure [62]).
117
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 8.1: Cut-away diagram of a cylindrical test article sized for 9 g of silicon showing materials,
components, and overall experimental geometry. Dimensions given in inches.
However, the use of ZrO
2
ceramic in contact with graphite places limits on the experiment due
to reactivity at elevated temperatures. ZrO
2
and graphite react at temperatures as low as 1400
K and with peak experimental temperatures exceeding 2000 K, the equilibrium pressure of the
reaction is approximately 40 Torr [127]. Before this reaction was identified, solar furnace tests
resulted in irreversible contamination of the quartz vacuum chamber entrance window. To prevent
contamination, at the expense of convection losses, current tests are operated in an environment
of 150 Torr of argon resulting in an approximately 10% increase in heat loss due to convection.
Testing using a pure BN system would allow for low pressure operation without quartz window
damage. However, the high cost and relatively low thermal conductivity of BN compared with
graphite makes this approach impractical.
During the Rescor 760 casting process, additional Type K thermocouples are placed within the
ceramic to monitor insulation temperatures. Test sections are allowed to dry in ambient conditions
for 24 hours and then loaded into the solar furnace. Photographs of the test section construction
process are given in Figure 8.2.
118
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 8.2: Montage of test section construction photos. Clockwise from top-left: Silicon rod
sample, inner components after machining, silicon placement in BN sleeve, inner components
supported via slip-fit before compression into place, assembly before ceramic casting, wet ceramic
casting showing clamps for thermocouple support, completed test section face, completed test
section body.
After test section construction, the first step in the furnace testing procedure was to bake out
new test sections under vacuum at approximately 500 K using a 1000 W spot lamp. This process
evaporates proprietary water based binders from the Rescor 760 compound which can fog the
quartz vacuum chamber window and decrease power delivery. After an 8 hour bake out, the testing
chamber was opened, a freshly cleaned quartz window was installed and the chamber was re-
evacuated to approximately 10 mTorr. The solar furnace is then used to re-heat the test section to
600 K prior to vacuum pump shut down and an argon backfill to 150 Torr is applied in order to
suppress ZrO
2
-C reaction during testing [127]
After bake out, solar furnace power was gradually increased in approximately 25% increments
via an aperture curtain until the test article reaches molten silicon temperatures. Once thermal
equilibrium has been achieved, a shutter curtain was used to cut power to the solar furnace and
the cooling curve for the test section is recorded. If solar conditions permit, repeated cycles can
119
Distribution A: Approved for public release; unlimited distribution. PA#15347
be performed on each test section. All thermocouple and solar furnace power data was recorded
during testing at a rate of 2 Hz using the instrumentation described in Chapter 5.
8.2 100%FillFactorTesting
The first round of molten silicon testing was performed on sections with a 100% silicon fill factor.
Silicon rods were precision cut to be an exact fit for the BN PCM container leaving no voids or
gaps in the test articles and maximizing energy storage potential. Figure 8.3 shows experimental
data taken during one such test. The phase change process occurs from approximately t = 30 s to
t = 120 s and demonstrates the relative temperature stability expected from a latent heat system.
The curves “Cycle 1” and “Cycle 2” are taken from the same test section across two back-to-back
cycles.
Note that “Cycle 1” exhibits a temperature spike at t = 108 s. This spike corresponded with
cracking of the test section due to asymmetric expansion of liquid silicon trapped within the par-
tially frozen silicon bulk. Expansion results in both increased contact pressure within the container
and a shift in container geometry producing higher recorded temperatures. Repeated tests also
demonstrated similar temperature profiles and cracking behavior. Figure 8.4 shows photographs
taken during the testing process for a 100% fill factor test section illustrating container failure. In
the existing literature concerning silicon as a potential phase change material, volumetric expan-
sion is neglected and as a result it was not a primary concern in both modeling and experimental
efforts until this damage was observed.
In addition to demonstrating the difficulties posed by silicon expansion, 100% fill factor testing
illustrated other operational concerns. During test section heating, a rapid jump in temperature
is apparent when the rear of the test section approaches 1500 K. A temperature curve during the
heating process is given in Figure 8.5 which illustrates this temperature spike in the absence of
increasing solar furnace power. This temperature spike corresponds to the phase transition of the
120
Distribution A: Approved for public release; unlimited distribution. PA#15347
0 50 100 150 200 250
1000
1100
1200
1300
1400
1500
1600
1700
1800
Time, s
Temperature, K

Cycle 1
Cycle 2
Figure 8.3: Experimental data taken with a Type C thermocouple as labeled in Fig. 8.1. Both traces
are from the same test section in back to back cycles. Note that “Cycle 1” has a temperature spike
at approximately t = 108 s corresponding to the rapid freezing of trapped silicon.
silicon bulk and the rising rear temperature is a function of a step change in silicon’s thermal con-
ductivity from approximately 18.5 W/mK at the melting point to 51 W/mK once liquid [128]. In
order to account for this change, thermal designs must consider the lower thermal conductivity
value to ensure sufficient heat conduction throughout. In this testing series, relatively small reduc-
tions in total solar furnace power have produced incomplete melting when the system was unable
to overcome this thermal conductivity barrier.
A total of three 100% test sections were successfully brought to molten silicon temperatures
and a composite of their cooling curves is presented in Figure 8.6. This plot also contains the
results of the in-house MATLAB model and demonstrates its ability to predict representative test
section performance despite the relative simplicity of phase change calculations. It is important to
note that the thermal conductivity of the boron nitride liner was reduced by an order of magnitude
(3 vs. 25 W/mK) in the model to provide an adequate fit. Since the model neglects thermal contact
resistance and the machining process does not yield perfect mating of parts, this reduction can
121
Distribution A: Approved for public release; unlimited distribution. PA#15347
A) B) C) D)
Figure 8.4: Photographs taken during 100 % fill factor tests. A) Test section before heating. B) Test
section immediately after power cutoff. C) Infrared photograph taken during solar heating showing
a large crack formed in the test section during the previous cooling cycle. Note the relative size
and intensity of the solar furnace input. D) Image of the interior of the crucible after being cut in
half showing decreased silicon density post testing.
be justified. Additionally, the area-averaged receiver (exposed graphite) temperature only varies
by a maximum of 3.5% when reverting the boron nitride back to the literature value. Relative
insensitivity of global parameters to this thermal conductivity value indicates that variation of this
material property primarily accounts for ineffective coupling of the Type C thermocouple to the
experimental system.
8.3 ExpansionDamageMitigation
Despite being neglected in the majority of extant literature concerning silicon as a PCM, the issue
of volumetric expansion presents the greatest difficulty for realizing an effective system. Silicon
is one of the few materials that expands during freezing and the relatively large volume increase
of approximately 10% poses a significant challenge [121]. The majority of phase change mate-
rials currently in use expand when melting (including boron) and this is typically resolved by
122
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 8.5: Initial experimental heating curve for a test section at a 100% silicon fill factor. Note
the four distinct heating regions as solar furnace power was gradually increased and a 5th rapid
rise in temperature corresponding to complete melting of the silicon PCM.
incorporating an expansion area to accept and drain back the additional liquid volume during ther-
mal cycling. Water is the only other material currently considered as a commercial PCM which
undergoes freezing expansion and it is typically held in either open or flexible containers to pre-
vent system damage. In the case of a silicon based system for satellite applications, a flexible or
open topped container is likely not possible. When using a sealed and filled container, perfect
re-solidification could theoretically return the silicon to the original shape. However, in practice
asymmetrical freezing will lead to voids, trapped liquid volumes, and a decreased effective density.
In industrial applications, the difficulties of asymmetric silicon freezing are alleviated via pre-
cision control of thermal gradients leading to an approximately 1-D freezing front [129, 130]. In
the case of a thermal energy storage system, this approach is impractical as multiple heat paths
out of the silicon container will yield multiple freezing locations. In order to prevent container
damage when freezing, a silicon based latent heat system will have to employ a reduced fill factor,
123
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 8.6: Comparison of experimental and MATLAB data for 100% fill factor testing using the
geometry given in Fig. 8.1. The experimental data range includes data across three test sections
and 4 thermal cycles. The temperatures given are at the Type C thermocouple location shown in
Fig. 8.1.
precise geometry to control heat flow, safeguards to prevent complete solidification or, more likely,
a combination of all three.
8.3.1 ReducedFillFactor
Following 100% fill factor testing, experiments were performed in the hope of establishing a reli-
able testing configuration that would allow for repeated testing without test section failure. The
first step in this investigation was performing a series of tests utilizing the same geometry as in
Fig. 8.1 with gradually reduced fill factors. Tests were conducted with fill factors between 100%
and 80% decreasing in 5% increments.
During this testing series, no test sections with fill factors less than 100% showed the macro-
scale damage seen during 100% fill factor trials. But, audible cracking during the phase change
process provided an indication of internal test section damage. Once fill factors were reduced to
80% there was only a single instance of audible cracking during the phase change process across
124
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 8.7: Comparison of experimental and MATLAB data for 80% fill factor testing using the
geometry given in Fig. 8.1. The experimental data range includes data across three test sections
and 15 thermal cycles. The temperatures given are at the Type C thermocouple location shown in
Fig. 8.1.
15 thermal cycles and three test sections. However, when cut open and examined, it was seen that
small cracks had still formed in the internal boron nitride liners.
When sectioned, 80% fill factor test articles consistently indicated an interesting silicon freez-
ing behavior. Since silicon is non-wetting to the BN sleeve, the liquid silicon forms a “bead”
within the test section upon melting. Due to the reduced fill factor and increased liquid density,
this “bead” does not make contact with the upper ends of the container. The shape of silicon after
testing, as shown photographed in Fig. 8.8, suggests that during the freezing process the front of
the liquid “bead” freezes first as the receiver side of the test section is responsible for the majority
of heat loss. Once this freezes, the remainder of the liquid “bead” is isolated from the front of the
PCM cavity. This causes freezing silicon to completely fill the rear of the test section until pressure
caused by volumetric expansion cracks the front of the “bead” and the remaining liquid silicon is
extruded into the front void.
125
Distribution A: Approved for public release; unlimited distribution. PA#15347
The composite cooling curve for all cycles using this testing condition is given in Fig. 8.7. As
with 100% fill factor testing, this plot also gives MATLAB model results. In order to approximate
the effects of a lower fill factor, the latent heat available to the MATLAB model was reduced
by 20% while other parameters were kept constant. Despite this simplistic approach, the MATLAB
model similarly follows experimental results.
8.3.2 HighDensityGraphite
After 80% fill factor tests repeatedly demonstrated damage only limited to the internal BN lin-
ers, a solution was sought to eliminate this component from the system. In response to the suc-
cessful use of pure graphite crucibles during early tube furnace tests, solar furnace test articles
were constructed with a bare graphite PCM container. While eliminating synergy with potential
molten boron experiments, bare graphite containers for liquid silicon are common in the industrial
production of large silicon ingots. Literature on the casting of silicon indicates that non-wetting
behavior and carbon contamination in the silicon bulk on the order of< 20 ppm is possible using
graphite crucibles provided that the graphite density is> 1.75 g/cc and that the graphite grain size
is< 50m [117].
Figure 8.9 shows the design of bare graphite test articles. The PCM cavity was integrated
into the graphite absorber and heat spreader with a press fit lid sealing the container at the top.
SIC-6 grade graphite was sourced from Graphite Machining Services Inc (GMSI) to construct the
graphite components. SIC-6 graphite has a density of 1.85 g/cc and a grain size of 10m which
is within the published requirements for low contamination and non-wetting behavior. A BN disk
is included at the top of the test section to protect the tip of the type C thermocouple at high
temperature. The remainder of test section construction is identical to those shown in Fig. 8.1 and
an 80% fill factor condition was maintained for all test articles.
126
Distribution A: Approved for public release; unlimited distribution. PA#15347
Test sections constructed using SIC-6 graphite exhibited no audible cracking and showed no
damage after multiple freezing cycles. However, test sections exhibited behavior indicating wetting
by liquid silicon which is in contrast to predictions made in the literature. An additional test
section was constructed substituting B325 industrial graphite which is outside the recommended
specifications for density and grain size. Test section performance and silicon behavior for this
lower density graphite was qualitatively identical.
Figure 8.10 shows the interior of a SIC-6 test section after melting in the solar furnace. Like
the previous tests using BN liners, there were voids apparent in the silicon bulk. However, it was
evident that in these tests silicon had been wicked into the corners of the test article indicating
wetting behavior. While the contamination of the silicon bulk has yet to be investigated, this
wicking behavior could be beneficial in a spacecraft system. Unlike BN lined tests which formed a
central mass of liquid silicon, pure graphite containers could keep liquid silicon and the subsequent
freezing process along the walls of the container. This would aid in overall heat transfer and
maintain a central void to take up silicon expansion.
The composite cooling curve for all cycles using pure graphite PCM containers is given in
Fig. 8.11. MATLAB model results are also included. Note the relatively large experimental tem-
perature spread of4% which is a result of both differing grades of graphite as well as a 10%
variation in input power.
8.3.3 PartialFreezing
Another potential method for mitigating test section damage is to only allow for partial freezing
of the silicon by re-introducing power to the system before complete solidification. This was
attempted using a test section with a 100% fill factor and power was returned to the test section
after 50, 60, and 70 seconds with temperature data given in Figure 8.12. All three of these intervals
were successful and the test section indicated no audible cracking or physical container damage.
127
Distribution A: Approved for public release; unlimited distribution. PA#15347
Since it is estimated that the phase change process begins approximately 30 seconds after cutting
solar furnace power, the longest 70 second freezing interval represents roughly 45% of the total
phase change process length and the total energy storage achievable in the test section is reduced
accordingly. Further testing is required to determine the ultimate level of solidification possible.
This method has potential for flight systems, provided the duty cycle can be matched to the
eclipse period on orbit. However, it proves problematic for solar furnace ground demonstrations.
Since crucible failure is assured when the experiment is terminated, a limited number of cycles
is possible with each test section. This is further constrained by the limited experimental time
afforded by the placement of the USC solar furnace facility.
8.4 Summary
Testing performed using the USC Solar Furnace successfully produced samples of molten silicon
via concentrated sunlight. These experiments ultimately highlighted the asymmetry of the silicon
freezing process and identified the management of freezing expansion as the primary technological
hurdle in the design of a molten silicon based energy storage system. Multiple methods for miti-
gating freezing expansion damage in a cylindrical container were explored and satisfactory results
were obtained by reducing the total PCM cavity fill factor to 80% and constructing test sections
from high density graphite. Experiments were also used to validate the in-house MATLAB model
and showed that using the temperature transforming method along with other simplifications can
accurately predict macro-scale thermal behavior of a molten silicon based system.
128
Distribution A: Approved for public release; unlimited distribution. PA#15347
A)
B) D)
C)
Figure 8.8: Photographs taken after sectioning an 80% fill factor solar furnace test article. The
graphite absorber / heat spreader, boron nitride liner, and silicon are shown. Test section geometry
is given in Fig. 8.1. In all photographs the rear of the test section is on the left. A) Top half of the
test article. B) Bottom half of the test article in a top down view as related to the test section as a
whole. Note how the rear of the test section is filled and external voids are present at the front. C)
Top half of the silicon removed from the top of the test article and flipped vertically. The formation
towards the right indicates flowing liquid silicon. D) Top down view of the test article with silicon
restored. It is apparent that after initial silicon freezing, liquid silicon was forced from the top of
the silicon bead.
129
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 8.9: Cutaway diagram for test sections with a graphite walled PCM cavity. Note the pres-
ence of a small cutout in the outer wall of the graphite container. This was added as a retention
mechanism after repeated failures caused by a slight negative draft machined outer wall of test sec-
tions. This draft, coupled with differential thermal expansion, would separate the two components
of the container. Dimensions are given in inches.
Figure 8.10: Photographs taken after sectioning a test article with a graphite walled PCM cavity.
Note the silicon wicked into the upper corners in the front of the test section (center of image)
indicating wetting behavior. This is in contrast to the behavior seen in Figure 8.8.
130
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 8.11: Comparison of experimental and MATLAB data for 80% fill factor testing, graphite
walled PCM cavity test sections using the geometry given in Fig. 8.9. The experimental data range
includes data across three test sections and 6 thermal cycles. The temperatures given are at the
Type C thermocouple location shown in Figure 8.9
Figure 8.12: Experimental data demonstrating successful partial freezing trials compared with
experimental data with now power restoration. A star marks the point where power is restored in
each trial.
131
Distribution A: Approved for public release; unlimited distribution. PA#15347
Chapter 9
Future Work
The results of the experimental effort at USC prove the basic feasibility of the high temperature
latent heat thermal energy storage while highlighting engineering concerns such as asymmetric
freezing. This has identified several potential research questions which must be answered before
a prototype system can be developed. Chief amongst these concerns are the effects of asymmetric
energy release when trying to convectively couple to a propellant or working fluid. Note that
future work items are related to the development of silicon based thermal energy storage in general
and have benefits beyond solar thermal propulsion applications. In particular, the experimental
facility developed during this research effort is well suited to continue molten silicon based work
for terrestrial applications with minor improvements.
9.1 ConvectiveCouplingCharacterization
Recently, a group at the Center for Space Nuclear Research (CSNR) completed a conceptual design
of a nuclear thermal bi-modal spacecraft for low-cost interplanetary exploration [92]. The pro-
posed spacecraft design couples a radioisotope thermal generator with a high performance thermal
capacitor to create a high power pulsed energy source usable for both direct thermal propulsion or
high power electric generation. This configuration takes advantage of the high specific energy of a
132
Distribution A: Approved for public release; unlimited distribution. PA#15347
radioisotope system while still allowing high power operation for short periods. It is proposed that
such a bi-modal spacecraft could deliver a 10 kg scientific payload into orbit around the Saturnian
moon Enceladus with a total launch mass of only 1000 kg.
At the core of the CSNR design is a combined high performance heat exchanger and thermal
capacitor using silicon as a latent heat thermal energy storage material. The use of silicon TES
in a nuclear thermal system has similar design constraints as solar thermal application and can be
used to further illustrate convective coupling concerns when using latent heat. As illustrated by the
boron based systems comparison in Section 4.2.1, harnessing the power of a latent heat medium
requires analysis of both the total energy storage density and the effective energy storage density
as defined by conjugate heat transfer geometry.
9.1.1 CSNRDesign
The thermal power sub-system for the proposed CSNR spacecraft was sized by communication
power requirements on-orbit at Enceladus. The design specifies a pair of closed Brayton cycle
engines for electrical power generation which are fed via a Helium gas blowdown through the
molten silicon based heat exchanger. The design goal for the system is to provide pulsed power
of 25 kWe for 360 seconds enabling a 3 Mbps downlink for the transfer of scientific data every
21 hours. Since the electrical conversion efficiency is specified at 30%, this means that the heat
exchanger must supply 83.3 kW of thermal energy throughout each 360 second Helium blowdown
cycle. The total helium flow rate for the heat exchanger was specified at 0.02 kg/s with an input
temperature of 903 K.
The core of the CSNR heat exchanger / thermal capacitor is specified as an 18.5 cm diameter,
30 cm long cylinder containing the silicon TES mass, 6 PuO
2
fuel rods, and 195 approximately
5 mm diameter flow channels. The amount of silicon TES required was sized purely from the latent
heat energy storage density and is specified at 15.58 kg. Insulation is provided by a combination
of zirconia and high performance carbon aerogel.
133
Distribution A: Approved for public release; unlimited distribution. PA#15347
To determine the estimated performance of the proposed CSNR design while including convec-
tive coupling effects, STAR-CCM+ has been used to model a single representative heat exchanger
passage. Similar to the ISUS analysis presented in Section 4.2.1, the macro properties of the CSNR
heat exchanger were divided by the total number of heat exchanger passages to yield a single 5 mm
diameter ID, 30 cm long helium flow channel surrounded by 80 g of silicon TES. A 0.25 mm thick
rhenium layer and a 0.5 mm graphite layer were added to the passage design considering sealing
problems during the ISUS program and the need for a suitable container material. Additionally,
based on the results of this work, the total silicon density was reduced by 20% to approximate an
80% fill factor condition to accommodate freezing expansion. This resulted in a passage OD of
14.95 mm.
Helium gas flow through the heat exchanger was modeled as a viscous, turbulent and ideal
gas with temperature dependent properties and a flow rate of 0.103 g/s. The CSNR design spec-
ifies a heat exchanger operating pressure of 4000 kPa. However, in order to speed convergence,
STAR-CCM+ models were performed at 101 kPa. Initial flow conditions for both pressures were
compared and it was found that output temperatures varied between the two pressure conditions by
less than 1% and that peak channel Reynolds numbers varied by less than 5%. Thus, it is expected
that the thermal behavior of the simulated system will be comparable to the CSNR design.
Solid properties were modeled as being fully temperature dependent and the silicon phase
change process was accommodated via the temperature transforming method with a “mushy” zone
of 2.5 K. All outer boundaries were adiabatic which is consistent with the proposed CSNR
design at the timescale of a single blowdown. The total insulation losses were estimated to be only
795W which is approximately 1% of the desired Helium power draw.
After setting an initial solid temperature condition of 1700 K and establishing Helium flow at
zero seconds, the STAR-CCM+ model was completed as a 2-D axisymmetric implicit unsteady
134
Distribution A: Approved for public release; unlimited distribution. PA#15347
time = 0 s
time = 75 s
time = 225 s
time = 300 s
time = 375 s
time = 450 s
time = 525 s
Temperature (K)
time = 150 s
Figure 9.1: Modeling geometry and transient temperature profiles for the “CSNR Design” STAR-
CCM+ model. Model geometry is axisymmetric and represents the red outlined region of the
cylindrical heat exchanger passage. Helium flow through the passage is from left to right and
dimensions are given in millimeters.
135
Distribution A: Approved for public release; unlimited distribution. PA#15347
0 50 100 150 200 250 300 350 400
1550
1600
1650
1700
Time, s
Helium Exit Temperature, K
0 50 100 150 200 250 300 350 400
65
70
75
80
85
Time, s
Power Out, kWt

StarCCM+ Model
CSNR Target
90% of Target
Figure 9.2: Heat exchanger thermal output performance for the proposed CSNR system. Tem-
peratures are calculated as the mass flow averaged temperature for the simulated heat exchanger
passage in Figure 9.1. Thermal power output is calculated as the energy required to bring propel-
lant from the inlet temperature of 903 K to the specified output temperature at the given mass flow
rate.
simulation. Transient temperature profiles for the model are given in Figure 9.1 and thermal per-
formance of the represented heat exchanger is given in Figure 9.2.
In the model, the silicon freezing process takes in excess of 525 seconds. However, the heat
exchanger never achieves the design power output and is only capable of achieving 90% of the
desired power level for approximately 316 seconds. There are two factors affecting this perfor-
mance. The first is the overall length of the heat exchanger. At 30 cm, the initial thermal profile in
Figure 9.1 shows that peak temperature is achieved with little margin.
The second concern is asymmetric freezing of the silicon TES due to the convective coupling
profile. Like the ISUS system, higher heat draw from the inlet results in an almost 1D phase
front which moves further down the heat exchanger passage during blowdown. The low ther-
mal conductivity of solid silicon near the melting point prevents diffusion of the remaining latent
heat energy and a large thermal gradient forms within the TES. Once the phase front moves past
136
Distribution A: Approved for public release; unlimited distribution. PA#15347
a certain length, there is insufficient distance for effective heat transfer to the gas flow and the
heat exchanger output temperature drops dramatically at approximately 25 seconds as shown in
Figure 9.2.
9.1.2 ConvectiveCouplingOptimization
If the effects of asymmetric heat transfer are considered, it is possible to re-design the heat
exchanger passage to yield higher effective energy storage density and stronger thermal perfor-
mance. A basic length optimization of the heat exchanger passage has been performed via a series
of STAR-CCM+ models. Passage length was varied from 15 to 70 cm while keeping the helium
mass flow rate and the amount of silicon thermal energy storage constant. This effectively varies
the aspect ratio of the model, spreading the thermal energy storage across a wider area and provid-
ing an extended convective coupling length. The results of this length optimization are given in the
left of Figure 9.3.
The first results plot concerns the initial mass flow average exit temperature as a function of
passage length. As expected, increasing channel length increases the propellant exit temperature
until it reaches quasi-equilibrium with the initial TES temperature of 1700 K. If the acceptable
performance threshold for the heat exchanger is set at 90% of the desired power draw, an exit
temperature of at least 1624 K is required. It can be seen that at the shortest passage lengths, there
is insufficient heat transfer area to achieve minimum acceptable performance even in the initial
state.
The second set of results gives the duration of sustained power output at 90% and 95% of the
target power level. Note that as the length of the heat exchanger is increased, the same amount of
silicon thermal energy storage provides significantly higher performance. This increase in effective
energy storage density is a result of a favorable convective coupling environment.
As shown in the model for the 30 cm design case, performance is ultimately limited by differ-
ential cooling of the passage. As the rear of the passage cools, the beginning of the effective heat
137
Distribution A: Approved for public release; unlimited distribution. PA#15347
20 30 40 50 60 70
1580
1600
1620
1640
1660
1680
1700
1720
Heat Exchanger Length, cm
Initial Mass Flow Averaged Exit Temp., K

Simulation Data
Initial TES Temp.
20 30 40 50 60 70
100
150
200
250
300
350
400
450
Heat Exchanger Length, cm
Sustained Power Output, s

90% Desired Power
95% Desired Power
Run Time Target
Figure 9.3: STAR-CCM+ length optimization data for the proposed CSNR heat exchanger. Increa-
seing heat exchanger length increases the total area availible for convection producing higher initial
temperatures and increases utilization of the silicon PCM producing higher performance.
transfer region moves forward along with the point of quasi-thermal equilibrium. The phase front
within the PCM moves in step with the onset of convection since this is the location of maximum
T and the low thermal conductivity of the recently frozen PCM prevents rapid diffusion of the
remaining latent heat energy. This results in an effective minimum length of liquid PCM which
must be maintained within the heat exchanger to reach the desired temperature - a requirement that
is quite different from simply the minimum PCM required to store the needed thermal energy.
For the 90% power case, the minimum convective coupling length is slightly under 20 cm. As
a result, a 20 cm long passage is able to sustain the desired power output for a short period of
time. As the heat exchanger length increases, performance initially shows a rapid increase as the
exit temperature approaches quasi-equilibrium with the liquid PCM creating a higher initial exit
temperature margin. Once the passage length is sufficient for the fluid to reach quasi-equilibrium
with the liquid PCM (approximately 35 cm), performance gradually increases until maximum
utilization of the phase change material is reached.
138
Distribution A: Approved for public release; unlimited distribution. PA#15347
Note that the maximum performance value doesn’t plateau as graphite and rhenium layers are
considered in the model effectively adding more mass to the system producing a sizable sensible
heat energy storage component. Additionally, the sensible heat component of the silicon TES
must also be considered. At 90% power draw, latent heat energy accounts for approximately 300
seconds of the sustained power.
This simple optimization shows that proper design of a phase change material based system
must consider the motion of the phase front in relation to the convective coupling profile. Since heat
exchanger length is limited by other concerns such as spacecraft size and insulation requirements
a more detailed characterization of the convective coupling problem is required. An extended
effort is necessary to relate total PCM storage mass, passage length, passage inner diameter, and
mass flow rate for multiple propellant gases. This can provide simple approximations and relations
that can be used for future designers to balance effective energy storage density against other
requirements.
9.1.3 ConvectionModelValidation
To validate convective coupling characterization, experiments are required which mirror modeling
conditions. The primary concern for these experiments will be attempting to match adiabatic
boundary conditions and ensuring that environmental heat loss is a small fraction of the convective
power draw. Ideally, this can be accomplished through the use of the molybdenum / zirconium
oxide multifoil insulation described in Section 3.3.4 and resistive heaters to achieve molten silicon
temperatures.
In the case where advanced multi-foil insulation is unavailable, it is possible to conduct mean-
ingful experiments using alternatives with an increase in experimental complexity. An attempt
was made to reach molten silicon temperatures for an approximately 21 cm test section using cast
zirconuim oxide insulation as the primary insulator and molybdenum wire heaters. The test arti-
cle, diagrammed in Figure 9.4 was able to achieve molten silicon temperatures across the entire
139
Distribution A: Approved for public release; unlimited distribution. PA#15347
Figure 9.4: Cutaway diagram of an initial resistive heating experiment using low-cost materials to
reach molten silicon temperatures. The cylindrical test section was able to achieve molten silicon
temperatures across the entire graphite core with a power input of 1400 W.
graphite core but required approximately 1400 W of power. With a desired convective draw of
430 W, environmental losses would dominate the system and provide little insight into convective
coupling effects on the silicon storage medium.
It is proposed that the desired quasi-adiabatic conditions are possible with low-cost materials
by implementing precise control of the resistive heater. As a simplified case, consider a 1-D slice of
the assembly given in Figure 9.4. Thermocouples placed at the edge of the graphite core and within
the heater assembly can be used to determined the heat flux across the core boundary. Since the
materials and geometry are known, radial conduction equations can be solved simultaneously with
the outer boundary condition for the heater power required to maintain a zero heat flux condition.
As the core is cooled by the propellant stream, real time calculations can be used to drop the power
level in the heater so that the boundary follows the core temperature and maintains a zero flux
condition. In the experimental system, accurate thermal control would require real-time calculation
of 3-D heat transfer conditions and the resistive heater assembly would have to be divided into
multiple controllable segments to accommodate asymmetric heat loss as the PCM is cooled.
140
Distribution A: Approved for public release; unlimited distribution. PA#15347
9.2 PotentialSolarFurnaceImprovements
The USC solar furnace facility is currently capable of molten silicon based experiments and is well
suited for silicon thermal energy storage testing targeting terrestrial thermophotovoltaic systems. In
order to mount larger scale experiments and potentially reach molten boron temperatures, furnace
improvements are necessary. The heliostat is currently the least efficient component of the solar
furnace and replacing the current reflector with commercially available first surface mirrors will
increase the power delivery and concentration ratio of the furnace by 30%. Since the heliostat is
permanently mounted outdoors, special care will be required to preserve the surface of the new
reflector. Careful engineering and a protective cover for when the system is off line can prevent
damage from the elements.
Another route to improving the performance of the solar furnace is the addition of a secondary
concentrator. Adding a compound parabolic concentrator (CPC) to the system can approximately
double the concentration ratio and increase experimental temperatures to where small scale molten
boron testing is feasible. Based on the parametric equations for a CPC given by Weldord and Win-
ston, adding a secondary concentrator with an overall reflective efficiency of 80% could increase
peak concentration ratios for the solar furnace to approximately 9,600:1 [131]. By combining
a CPC with new heliostat mirror panels, concentration ratios above 10,000:1 can certainly be
achieved making molten boron experiments possible.
9.3 DevelopmentalRoadmap
The latent heat thermal energy storage component of a high performance bi-modal solar thermal
microsatelite has been identified as both an enabling and under-developed technology. This work
has succeed in demonstrating the basic feasibility of a silicon based thermal energy storage system
and identified key technological hurdles and practical engineering concerns. While the long term
goal of future research efforts should be focused on the design and test of a microsatellite scale
141
Distribution A: Approved for public release; unlimited distribution. PA#15347
bi-modal system, key near term research goals must be met before sensible engineering decisions
can be made. The following points present the chief design concerns from both this work and the
existing literature.
Expansion damage is the primary extant concern for mounting an effective silicon based ther-
mal energy storage system. This work has demonstrated successful results with reduced fill
factors in cylindrical geometries. However, results are limited to the specific tests performed.
A combined experimental and analytical approach is recommended to vary geometries and
explore the conical container sections proposed in the terrestrial literature by Chubb, Datas,
and Veeraragavan [5, 95, 96]. It is important to note that freezing asymmetry, highlighted in
this work, necessitates that future experiments emulate the environmental heat loss profile
that would be seen in a real system as opposed to uniform furnace testing.
This work proposes the use of high density graphite containers for molten silicon thermal
energy storage based on a survey of the available literature and demonstrated short-term
performance in solar furnace tests. Future study is required to quantify both contamination
levels and repeatability of this combination across thousands of cycles. Automated furnace
tests using varying grades of graphite must be conducted while accurately measuring latent
heat release, preferably with a power draw profile simulating a propellant / working fluid
blowdown.
The proper design of a high temperature PCM based heat exchanger must consider the con-
vective coupling profile in addition to total PCM mass as demonstrated by STAR-CCM+
models of the conjugate heat transfer system. An extended modeling effort is required to
determine the relationship between effective energy storage density and multiple variables
such as heat exchanger diameter and length, mass flow rate and working fluid. The resulting
data set will allow mission designers to easily trade effective energy storage density with
other design parameters without the use of time consuming transient heat transfer models.
142
Distribution A: Approved for public release; unlimited distribution. PA#15347
Molten boron experiments were outside the scope and budget of this research effort. How-
ever, the ultimate solar thermal bi-modal system requires boron based thermal energy storage
to see the proposed 35-60% increase in V capability vs. competing chemical systems. The
first stages of molten boron research should focus on a reliable container design capable of
surviving multiple cycles using the proposed combination of boron nitride liners and sealed
graphite containers. Following these tests, design of a molten boron heat exchanger can draw
from convective coupling relations established via molten silicon system development.
143
Distribution A: Approved for public release; unlimited distribution. PA#15347
Chapter 10
Conclusions
The promise of high thrust and high efficiency has driven decades of research into solar thermal
propulsion (STP). However, despite multiple flight development programs and multiple statements
of feasibility with existing technology, no STP spacecraft systems have flown to-date. Perceived
and actual system complexity, coupled with vehicle integration concerns overshadow the utility
of STP and the benefits of a mid-rangeI
sp
, high thrust propulsion mechanism are not enough to
outweigh technological and mission uncertainty.
As solar thermal propulsion has progressed, the trend has been toward miniaturization as well
as simplification and the latest research efforts have targeted microsatellite systems. In the frame-
work of a high performance microsatellite that requires both quick response time and large V
delivery, an STP system fills a unique role that cannot be matched with conventional propulsion
technologies. Implementation of an STP system at the microsatellite scale has the potential for a
greater than 50% V increase while maintaining response times that can be measured in days. In
this case, the risk of a novel flight mechanism is outweighed not by an incremental improvement,
but by the enabling of a new class of high performance, low cost spacecraft.
Proper implementation of solar thermal propulsion on board a microsallite requires a bi-modal
configuration to both reduce system complexity and provide acceptable propulsion and power mass
fractions. In this scenario, the thermal sub-systems on the spacecraft provide both propulsive and
144
Distribution A: Approved for public release; unlimited distribution. PA#15347
electric power and a review of current technology shows that ready solutions are available for
the majority of the necessary spacecraft systems with the exception of high performance thermal
energy storage.
Throughout the history of solar thermal propulsion, thermal energy storage has been primarily
an afterthought baring the bi-modal development programs of the 1990s. Designs have settled on
sensible heat based systems even though the literature frequently mentions the benefits of latent
heat thermal energy storage. This work represents the most thorough investigation of high temper-
ature latent heat thermal energy storage to date confirming both the potential gains achievable and
the basic feasibility of the concept.
An experimental approach was taken to uncover practical engineering concerns culminating in
the first molten silicon experiments exploring the thermal energy storage problem. Experiments
conducted with a newly developed solar facility succeeded in the short term demonstration of
potential container materials and highlighted the asymmetry of the silicon freezing process in a
real world system. Asymmetric freezing and container damage due to freezing expansion were
identified as the most pressing engineering concerns for silicon based thermal energy storage and
the reduction of container fill factor has been demonstrated as a viable solution in cylindrical
geometries at the expense of energy storage density.
Experimental results were also used to validate an in-house MATLAB model of the experiment
using the “enthalpy method” for calculating the phase change process. Despite the complexities of
of the silicon freezing process, simplified modeling methods are able to capture essential system
behaviors.
Ultimately, this work provides a technological basis for future design efforts by proving the
basic feasibility of a molten silicon based thermal energy storage system and presenting multiple
pathways for future development.
145
Distribution A: Approved for public release; unlimited distribution. PA#15347
Reference List
[1] Scharfe, D. and Ketsdever, A., “A Review of High Thrust, High Delta-V Options for
Microsatellites Missions,” 45th AIAA/ASME/SAE/ASEE Joint Propulsion Conference, No.
AIAA-2009-4824, American Institute of Aeronautics and Astronautics, Denver, CO, August
2009.
[2] Wertz, J. R., editor,SpaceMissionEngineering: TheNewSMAD, Microcosm Press, 2011.
[3] Kennedy, F., Solar Thermal Propulsion for Microsatellite Manouevering, Doctoral thesis,
University of Surrey, Guildford, Surrey, UK, September 2004.
[4] Ethridge, F. G.,SolarRocketSystemConceptAnalysis, Rockwell International Final Report,
AFRPL-TR-79-79, December 1979.
[5] Chubb, D. L., Good, B. S., and Lowe, R. A., “Solar Thermophotovoltaic (STPV) System
With Thermal Energy Storage,” The 2nd NREL Conference on Thermophotovoltaic Gener-
ationofElectricty, Colorado Springs, CO, July 1995.
[6] Ehricke, K. A., “The Solar-Powered Space Ship,” Semi-Annual Meeting of the American
RocketSociety, No. ARS Paper 310-56, Cleveland, OH, 1956.
[7] Selph, C. C., “The Place of Solar Thermal Rockets In Space,”JANNAFPropulsionMeeting,
New Orleans, LA, May 1981.
[8] Shoji, J. M.,SolarRocketComponentStudy, Rockwell International - Rocketdyne Division,
Final Report, AFRPL-TR-84-057, February 1985.
[9] Shoji, J. M., “Performance Potential of Advanced Solar Thermal Propulsion,” 19th
AIAA/SAE/ASMEJointPropusionConference, Seattle, WA, June 1983.
[10] Laug, K., “The Solar Propulsion Concept is Alive and Well at the Astronautics Laboratory,”
JANNAFPropulsionMeeting, Cleaveland, OH, May 1989.
[11] Shoji, J. M., Frye, P. E., and McClanahan, J. A., “Solar Thermal Propulsion Status and
Future,”AIAASpaceProgramsandTechnologiesConference, Huntsville, AL, March 1992.
[12] Shoji, J., Frye, P., Chwiedor, T., and Lim, D., “Advanced Absorber/Thruster Concepts For
Solar Thermal Propulsion,” ASME Joint Solar Engineering Conference, New York, NY ,
1993.
146
Distribution A: Approved for public release; unlimited distribution. PA#15347
[13] Fitzsimmons, M. J., “Technology Considerations For Future Upper Stages,” AIAA Space
ProgramsandTechnologiesConference, Huntsville, AL, 1994.
[14] Zubrin, R. M., Sulmeisters, T. K., Jacox, M. G., and Watts, K., “The Integrated Power and
Propulsion Stage: A Mission Driven Solution Utilizing Thermionic Technology,” Proceed-
ingsoftheNinthSymposiumOnSpaceNuclearPowerSystems, Alburquerqe, NM, January
1992.
[15] Culver, D., Kolganov, V ., D’Yakov, E., and Daragan, I., “Low Cost Space Power: High
Performance Nuclear Bimodal System With Flexible Growth Options From Demonstrated
Technologies,”30thAIAA/ASME/SAE/ASEEJointPropulsionConference, Indianapolis, IN,
June 1994.
[16] Kennedy, F., “Bimodal Power and Propulsion System Requirements,” 31st
AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, San Diego, CA,
July 1995.
[17] Laug, K. K., Holmes, M. R., and Westerman, K. O., Solar Bi-Modal System Concept: Mis-
sion Applications, A Preliminary Assessment, Technical Paper, Air Force Research Lab
Propulsion Directorate, Wright-Patterson AFB, OH, 1992.
[18] Malloy, J. D., Rochow, R. F., and Inman, J. B., “Hybrid Solar Rocket Utilizing Thermal
Storage For Propulsion And Electrical Power,” US Patent No. 5,459,996, 1995.
[19] Kennedy, F. and Jacox, M., “The Integrated Solar Upper Stage (ISUS) Program,” AIAA
1995SpaceProgramsandTechnologiesConference, American Institute of Aeronautics and
Astronautics, Huntsville, AL, September 1995.
[20] Harper, W. B. and Shatlens, R. K., “NASA Solar Dynamic Ground Test Demonstration
(GTD) Program And Its Application To Space Nuclear Power,” 10th Symposium on Space
NuclearPowerandPropulsion, Alberquerque, NM, January 1993.
[21] Partch, R. and Frye, P., “Solar Orbit Transfer Vehicle Conceptual Design,” 35th
AIAA/ASME/SAE/ASEE Joint Propulsion Conference, No. AIAA-1999-2476, American
Institute of Aeronautics and Astronautics, Los Angeles, CA, 1999.
[22] Kerslake, T. W. and Ibrahim, M. B., “Analysis of Thermal Energy Storage Material With
Change-of-Phase V olumetric Effects,” Journal of Solar Energy Engineering, V ol. 115:1,
February 1993, pp. 22–31.
[23] Miles, B. J., “Development and Testing of High Temperature Coatings for the Integrated
Solar Upper Stage Thermal Energy Storage System,” Space Technology and Applications
InternationalForum, Alburquerque, NM, January 1997.
[24] Kudija, C., “The Integrated Solar Upper Stage (ISUS) Engine Ground Demonstrator
(EGD),” 32nd AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, Lake
Buena Vista, FL, July 1996.
147
Distribution A: Approved for public release; unlimited distribution. PA#15347
[25] Frye, P. E. and Kudija, C., “Integrated Solar Upper Stage Engine Ground Demonstration
Test Results and Data Analysis,”34thAIAA/ASME/SAE/ASEEJointPropulsionConference,
No. AIAA-1998-3958, American Institute of Aeronautics and Astronautics, 1998.
[26] Wyant, F. J., Luke, J. R., and Luchau, D. W., “START-3: Operational Evaluations of the
ISUS Engine Ground Demonstration Thermionic Power System,” Space Technology and
ApplicationsInternationalForum, Alberquerque, NM, Jan-Feb 1999.
[27] Kennedy, F., Palmer, P., and Paul, M., “Results of Microscale Solar Thermal Engine ground
Test Campaign at the Surrey Space Center,” 40th AIAA/ASME/SAE/ASEE Joint Propulsion
Conference, No. AIAA-2004-4137, American Institute of Aeronautics and Astronautics,
Fort Lauderdale, FL, 2006.
[28] Wassom, S. R., Focus Control System for Solar Thermal Propulsion, Technical Paper,
Thiokol Propulsion, Brigham City, UT, ADA410345.
[29] Frye, P. E. and and, F. K., “Reusable Orbital Transfer Vehicles (ROTV): Applications of
an Integrated Solar Upper Stage (ISUS),” 33rd AIAA/ASME/SAE/ASEE Joint Propulsion
Conference, No. AIAA-1997-2981, American Institute of Aeronautics and Astronautics,
Seattle, WA, 1997.
[30] Carroll, J., “Solar Orbit Transfer Vehicle,” AIAA Space Conference, No. AIAA-2000-5110,
American Institute of Aeronautics and Astronautics, Long Beach, CA, 2000.
[31] Holmes, M., “Solar Rocket Propulsion: Ground and Space Technology Demonstration,”
12thAnnualAdvancedSpacePropulsionWorkshop, Hunstsville, AL, 2001.
[32] Tucker, S. and Salvail, P., “Solar-Thermal Engine Testing,”SpaceTechnologyandApplica-
tionsInternationalForum, Alburquerque, NM, February 2002.
[33] Gerrish, H., Schmidt, G., and Rodgers, S., “Advanced Propulsion Research at Marshall
Proulsion Research Center,” 37th Joint Propulsion Conference and Exhibit, Salt Lake City,
UT, July 2001.
[34] Kennedy, F. and Palmer, P. L., “Preliminary Design of Micro-Scale Solar Thermal Propul-
sion System,”38thAIAA/ASME/SAE/ASEEJointPropulsionConference, No. AIAA-2002-
3928, American Institute of Aeronautics and Astronautics, Indianapolis, IN, July 2002.
[35] Henshall, P., “Fibre Optic Solar Thermal Propulsion Technology Demonstrtation,” Surrey
Space Centre, School of Electronis and Physical Sciences, University of Surrey, March
2006.
[36] Henshall, P., A Proposal to Develop and Test a Fibre-Optic Coupled Solar Thermal Proul-
sionSystemforMicrosatellites, University of Surrey. AF Contract FA8655-04-1-3030 Final
Report., 2006.
148
Distribution A: Approved for public release; unlimited distribution. PA#15347
[37] Nakamura, T., Sullivan, D., and McCalahan, J., “Solar Thermal Propulsion for Small Space-
craft,” 40th AIAA/ASME/SAE/ASEE Joint Propulsion Conference, No. AIAA-2004-4138,
American Institute of Aeronautics and Astronautics, Fort Lauderdale, FL, July 2004.
[38] Sahara, H. and Shimizu, M., “Solar Thermal Propulsion System for Microsatellite Orbit
Transferring,”40thAIAA/ASME/SAE/ASEEJointPropulsionConference, No. AIAA-2004-
3764, American Institute of Aeronautics and Astronautics, Fort Lauderdale, FL, July 2004.
[39] Colonna, G., Capitta, G., Capitelli, M., Wysong, I., and Kennedy, F., “Model for Ammonia
Solar Thermal Thruster,” Journal of Thermophysics and Heat Transfer., V ol. 20, No. 4,
October-December 2006.
[40] Pearson, J., Gierow, P., and Lester, D., “Near Term In-Space Demonstration of an Inflat-
able Concentrator,” Proceedings of the 37th AIAA Aerospace Sciences Meeting, Reno, NV ,
January 1999, pp. 11–14.
[41] Pearson, J. C., Lester, D. M., Holmes, M. R., and Wong, W. A., Solar Thermal Vacuum
Testing of an Integrated Membrane Concentrator System at the NASA GRC Tank 6, SRS
Technologies - Technical Report, Huntsville, AL, June 2003.
[42] Shatlens, R. K. and Mason, L. S., “Early Results from Solar Dynamic Space Power Testing,”
JournalofPropulsionandPower, V ol. 12, No. 5, October 1996, pp. 852–858.
[43] Borell, G. J. and Campbell, J. S., “ISUS Solar Concentrator Array Development,” 32nd
AIAA, ASME, SAE and ASEE Joint Propulsion Conference and Exhibit, Lake Buena Vista,
FL, July 1996.
[44] Holmes, M. R. and Laug, K. K., “Dependence of Solar-Thermal Rocket Performance on
Concentrator Performance,”JournalofSolarEngineering, V ol. 2, 1995, pp. 837–848.
[45] ESA,HerschelObserver’sManual., Version 5, HERSCHEL-HSC-DOC-0876.
[46] Space, M. and Group, D.,Type22AntennaPointingAssemblyDataSheet, Chatsworth, CA,
2010.
[47] Hyder, A. K., Halpert, R. L., Halpert, G., Flood, D. J., and Sabripour, S., Spacecraft Power
Technologies, Imperial College Press, London, 2000.
[48] Gilpin, M., Scharfe, D., Young, M., and Pancotti, A., “Molten Boron Phase-Change Thermal
Energy Storage: Containment and Applicability to Microsatellites,” 42nd AIAA Thermo-
physics Conference, No. AIAA-2011-3637, American Institute of Aeronautics and Astro-
nautics, 2011.
[49] Preijde, J.,DesignofaSolarThermalPower-PropulsionSystemforaSmallSatellite, Mas-
ter’s thesis, Delft University of Technology, January 2015.
149
Distribution A: Approved for public release; unlimited distribution. PA#15347
[50] Stone, K. W., Fatemi, N. S., and Garverick, L. M., “Operation and Testing of a Solar Ther-
mophotovoltaic Power System,” 25th IEEE Photovoltaic Specialists Conference, Washing-
ton, DC, May 1996.
[51] Stone, K., Leingang, E., Drubka, R., Chubb, D., Good, B., and Wilt, D., “System Perfor-
mance of a Solar Themophotovoltaic System for Space and Terrestrial Applicaions,” Inter-
societyEnergyConversionEngineeringConference, V ol. 1, 1995, pp. 713–718.
[52] Horne, E., “Hybrid Thermophotovoltaic Power Systems,” Final Report P500-02-048F, Cal-
ifornia Energy Commission, 2002.
[53] Kessler, T., “An Overview of a Solar Thermal Propulsion and Power System Demonstration
Applicable to HEDS,” AIAA Space 2001 Conference, No. AIAA-2001-4777, Albuquerque,
NM, August 2001.
[54] Lyons, V . J., Gonzolez, G. A., Houts, M. G., Iannello, C. J., Scott, J. H., and Surampudi, S.,
SpacePowerandEnergyStorageRoadmap, NASA, April 2012.
[55] Wei, G. and Robbins, J., Development and Characterization of Carbon-Bonded Carbon
Fiber Insulation for Radioisotope Space Power Systems, No. ORNL/TM-9381, Oak Ridge
National Labratory and Martin Marietta, 1985.
[56] Baxter, R. I., Rawlings, R. D., Iwashita, N., and Sawada, Y ., “Effect of Chemical Vapor Infil-
tration on Erosion and Thermal Properties of Porous Carbon/Carbon Composite on Thermal
Insulation,”Carbon, V ol. 38, No. 3, 2000, pp. 441–449.
[57] Gallego, N. C. and Klett, J. W., “Carbon Foams for Thermal Management,”Carbon, V ol. 41,
20013, pp. 1461–1466.
[58] Gasch, M., Ellerby, D., Irby, E., Beckman, S., Gusman, M., and Johnson, S., “Processing,
Properties and Arc Jet Oxidation of Hafnium Diboride/Silicon Carbide Ulatra High Tem-
perature Ceramics,”JournalofMaterialsScience, V ol. 39, No. 19, 2004, pp. 5925–5937.
[59] Lide, D. R.,CRCHandbookofChemistryandPhysics, Boca Raton, FL, 77th ed., 1996.
[60] “Hot Pressed BN Shapes Data Sheet,” Momentive Performance Materials Inc., Strongsville,
OH, 2012.
[61] Gilpin, M., Scharfe, D., Young, M., and Pancotti, A., “Molten Boron Phase-
Change Thermal Energy Storage to Augment Solar Thermal Propulsion Systems,” 47th
AIAA/ASME/SAE/ASEE Joint Propulsion Conference, No. AIAA-2011-5986, American
Institute of Aeronautics and Astronautics, San Diego, CA, 2011.
[62] “Rescor Castable Ceramics,” Cotronics Corp., New York, NY , 2008.
[63] “Calcarb Rigid Carbon Thermal Insulation Tehcnical Guide,” MERSEN USA, Bay City,
MI, 2009.
150
Distribution A: Approved for public release; unlimited distribution. PA#15347
[64] “Ultramet Carbon Foam Data Sheet,” Ultramet, Pacoima, CA, 2014.
[65] Glazer, S. D., Multifoil Insulation Study for Weight Reduction, No. JPL D-1268, NASA Jet
Propulsion Laboratory, January 1984.
[66] Westerman, K. O. and Barry, “Testing of a Receiver-Absorber-Converter (RAC) for the
Integrated Solar Upper Stage (ISUS) Program,” Space Technology and Applications Inter-
nationalForum, Albuquerque, NM, January 1998.
[67] Chase, M. W., editor, NIST-JANAF ThermochemicalTables 2 Volume-Set(Journal of Phys-
icalandChemicalReferenceDataMonographs), American Institute of Physics, 1998.
[68] Lynch, J., Ruderer, C., and Duckworth, W., editors, Engineering Properties of Selected
CeramicMaterials, The American Ceramic Society, Columbus, OH, 1966.
[69] Kenisarin, M. and Mahkamov, K., “Solar Energy Storage Using Phase Change Materials,”
RenewableandSustainableEnergyReviews, V ol. 11, 2007, pp. 19113–1965.
[70] Zalba, B., Marin, J., Cabeza, L., and Mehling, H., “Review on Thermal Energy Storage
with Phase Chamge: Materials, Heat Transfer Analysis and Applications,”AppliedThermal
Engineering, V ol. 23, 2003, pp. 251–283.
[71] Ho, C., Powell, R., and Liley, P., “Thermal Conductivity of the Elements: A Comprehensive
Review,”JournalofPhysicalandChemicalReferenceData, V ol. 3, 1974.
[72] Koyama, K., “Thermal Conductivity of Magnesium Flouride Between 25 and 900 C,”Jour-
naloftheAmericanCeramicSocitey, V ol. 52, No. 4, June 2006.
[73] Glassbrenner, C. and Slack, G. A., “Thermal Conductivity of Silicon and Germanium from 3
K to the Melting Point,”PhysicalReview, V ol. 134, No. 4A, May 1964, pp. A1058–A1069.
[74] Cooper, H. S.,RareMetalsHandbook, Reinhold, London, 2nd ed., 1961.
[75] Jansen, L. H., editor, Kirk-Othmer Encyclopedia of Chemical Technology, V ol. 14, Wiley
InterScience, New York, 5th ed., 2004.
[76] Sands, D. and Hoard, J., “Rhombohedral Elemental Boron,”JournaloftheAmericanChem-
icalSociety, V ol. 79, October 1957, pp. 5582–5583.
[77] Hoard, J., “Remarks on Structure and Polymorphism in Boron,” Proceedings of the 1st
InternationalSymposiumonBoron, Asbury Park, NJ, 1959.
[78] Amundson, T. R., Computational Evaluation of the Effects of Voids on a Thermal Energy
Storage System in a Combined Solar Power and Propulsion System, Master’s thesis, Uni-
versity of Colorado Colorado Springs, 2013.
151
Distribution A: Approved for public release; unlimited distribution. PA#15347
[79] Gilpin, M., Scharfe, D., and Young, M., “Phase-Change Thermal Energy Storage
and Conversion: Development and Analysis for Solar Thermal Propulsion,” 48th
AIAA/ASME/SAE/ASEE Joint Propulsion Conference, No. AIAA-2012-3715, American
Institute of Aeronautics and Astronautics, Atlanta, GA, 2012.
[80] Frye, P., “Integrated Solar Upper Stage (ISUS) Space Demonstration Design,” Space Tech-
nologyandApplicationsInternationalForum(STAIF-97), Albaquerque, NM, Jan 1997.
[81] Nakamura, T., Sullivan, D., McClanahan, A., Shoji, J., Partch, R., and Quinn., S., “Solar
Thermal Propulsion for Small Spacecraft - Engineering System Development and Evalu-
ation,” 41st AIAA/ASME/SAE/ASEE Joint Propulsion Conference, No. AIAA-2005-3923,
American Institute of Aeronautics and Astronautics, Tucson, AZ, 2005.
[82] Mueller, J., “Thruster Options for Microspacecraft: A Review and Evaluation of Existing
Hardware and Emerging Technologies,” Tech. Rep. AIAA-97-3058, AIAA, 1997.
[83] Astrium, E., 1N Hydrazine Thruster, Technical Specifications: http://cs.astrium.eads.net.
[84] Astrium, E., 20 N Hydrazine Thruster, Technical Specifications: http://cs.astrium.eads.net.
[85] Biagioni, L., Cesari, U., and Saverdi, M., “Development Status of the HT-100 Miniaturized
Hall Effect Thruster System,”41stAIAA/ASME/SAE/ASEEJointPropulsionConverence&
Exhibit, Tucson, AZ, July 2005.
[86] Brown, C.,SpacecraftPropulsion, AIAA, Reston, V A, 1st ed., 1996.
[87] Larson, W. and Wertz, J., editors,SpaceMissionEngineering: TheNewSMAD, Microcosm
Press, 1st ed., 2011.
[88] “NASA Technology Roadmaps: Propulsion and Power Workshop,” Tech. rep., Aeronautics
and Space Engineering Board, National Research Council, March 2012.
[89] Strumpf, H. and Coombs, M., “Solar Reciever Experiment for the Space Station Freedom
Brayton Engine,”JournalofSolarEnergyEn, V ol. 112, 1990, pp. 12–18.
[90] Abbott, R., “MultiUse Solar Thermal Power Generators,” Space Technology and Applica-
tionsInternationalForum, Alburquerque, NM, February 2001.
[91] Calabro, M., “Solar Thermal Upper Stages,” International Conference on Recent Advances
inSpaceTechnologies, Istanbul, Turkey, November 2003.
[92] Jerred, N., Howe, T., Howe, S., and Rajguru, A., “Dual-Mode Propulsion System Enabaling
Cubesat Exploration of the Solar System,” Tech. rep., NASA Inovative Advanced Concepts:
Phase 1 Final Report, 2014.
[93] Woodall, “Energy Conversion,” US Patent No. 4,316,048, 1982.
152
Distribution A: Approved for public release; unlimited distribution. PA#15347
[94] Stone, K. W., Schneider, S., and Schmitt, R. J., “Solar Thermophotovoltaic Power Conver-
sion Method and Apparatus,” 1999‘.
[95] Datas, A., Chubb, D., and Veeraragavan, A., “Steady State Analysis of A Storage Integrated
Solar Thermophotovoltaic (SISTPV) System,”SolarEnergy, V ol. 96, 2013, pp. 33–45.
[96] Veeraragavan, A., Montgomery, L., and Datas, A., “Night Time Performance of a Stor-
age Integrated Solar Thermophotovoltaic (SISTPV) System,”SolarEnergy, V ol. 108, 2014,
pp. 377–389.
[97] Hukuo, N., “Design Problems of a Solar Furnace,” The Proceedings of the Solar Furnace
Symposium, Association for Applied Solar Energy, 1957.
[98] Michalsky, J. J., “The Astronomical Almanac’s Algorithm For Approximate Solar Position
(1950-2050),”SolarEnergy, V ol. 40, No. 3, 1988, pp. 227–235.
[99] Meeus, J.,AstonomicalAlgorithms, Willmann-Bell, Inc., 2009.
[100] Strachan, J. W. and Houser, R. M., “Testing and Evaluation of Large-Area Heliostats for
Solar Thermal Applications,” Tech. rep., Sandia National Laboratories, Albuquerque, NM,
February 1993.
[101] Haueter, P., Steitz, T., and Steinfeld, A., “A New High-Flux Solar Furnace for High-
Temperature Thermochemical Research,” Journal of Solar Engery Engineering, V ol. 121,
1999, pp. 77–80.
[102] Ries, H. and Schubnell, M., “The Optics of a Two-Stage Solar Furnace,” Solar Energy
Materials, V ol. 21, 1990, pp. 213–217.
[103] Muggleton, A. and Howe, F., “The Preparation of Thin Self Supporting Boron Films,”
NuclearInstrumentsandMethods, V ol. 13, August-October 1961, pp. 211–214.
[104] Searcy, A. W. and Myers, C. E., “The Heat of Sublimation of Boron and Gaseous Species of
the Boron-Boric Oxide System,” Journal of Physical Chemistry, V ol. 67, No. 7, July 1957,
pp. 957–960.
[105] Storms, E. and Mueller, B., “Phase Relations and Thermodynamic Properties of Transition
Metal Borides. I. The Molybdenum-boron system and Elemental Boron,”JournalofPhysi-
calChemistry, V ol. 81, 1977, pp. 318–324.
[106] Leitnaker, J. M. and Bowman, M. G., “High-Temperature Phase Studies in the Tantalum-
Boron System between Ta and TaB,”JournaloftheElectrochemicalSociety, V ol. 108, No. 6,
June 1961, pp. 568–572.
[107] Itoh, H., Matsudaira, T., and Naka, S., “Formation Process of Tungsten Borides by Soild
State Reaction Between Tungsten and Amorphous Boron,” Journal of Materials Science,
V ol. 22, No. 8, 1987, pp. 2811–2815.
153
Distribution A: Approved for public release; unlimited distribution. PA#15347
[108] Steinitz, R., Binder, I., and Moskowitz, D., “System Molybdenum-Boron and Some Proper-
ties of The Molybdenum-Borides,”JournalofMetals, September 1952, pp. 983–988.
[109] Stout, N. D., Mar, R. W., and Boo, W. O., “The High-Temperature Enthalpy of Fusion
of Boron by Drop Calorimetry,” High Temperature Science, V ol. 5, No. 4, August 1973,
pp. 241–251.
[110] Wald, F. and Stormont, R. W., “Investigations on the Constitution of Certian Binary Boron-
Metal Systems,” Journal of the Less Common Metals, V ol. 9, No. 6, December 1965,
pp. 423–433.
[111] Kimpel, R. F. and Moss, R. G., “Melting Point of 98.9 to 99.6% Pure Boron,” Journal of
ChemicalEngineeringData, V ol. 13, No. 2, April 1968, pp. 231–234.
[112] Mar, R., “High-temperature Thermal Analysis of High Boron Alloys Using Automatic Opti-
cal Pyrometry,”ThermochimicaActa, V ol. 4, No. 3-5, August 1972, pp. 367–376.
[113] Hildenbrand, D. L. and Hall, W., “The Vaporization Behavior of Boron Nitride and Alu-
minum Nitride,”JournalofPhysicalChemistry, V ol. 67, No. 4, April 1963, pp. 888–893.
[114] Fesenko, V . and Bolgar, A., “Evaporation Rate and Vapor Pressure of Carbides, Silicides,
Nitrides and Borides,”PowerMetallurgyandMetalCeramics, V ol. 2, No. 1, January 1964,
pp. 11–17.
[115] Paine, R. and Narula, C., “Synthetic Routes to Boron Nitride,” Chemical Reviews, V ol. 90,
No. 1, January 1990, pp. 73–91.
[116] Patrick, W. J., Scilla, S. J., and Westdrop, W. A., “Controlling the Oxygen Content
of Czochralski Process of Silicon Crystals By Sandblasting Silica Vessel,” US Patent
No.4,010,064, 1977.
[117] Ciszek, T., Schwuttke, G., and Yang, K., “Directionally Solidified Solar-Grade Silicon
Using Carbon Crucibles,”JournalofCrystalGrowth, V ol. 46, 1979, pp. 527–533.
[118] Drevet, B., V oytovych, R., Israel, R., and Eustathopoulous, N., “Wetting and Adhesion of Si
on Si
3
N
4
and BN substrates,” Journal of the European Ceramic Society, V ol. 29, February
2009, pp. 2363–2367.
[119] Wang, L., Clancy, P., Thompson, M., and Murthy, C., “Thermodynamic and Kinetic Studies
of Laser Thermal Processing of Heavily Boron-Doped Amorphous Silicon Using Molecular
Dynamics,”JournalofAppliedPhysics, V ol. 92, No. 5, September 2002, pp. 2412–2419.
[120] Enriquez, O., Marin, A., Winkels, K., and Snoeijer, J., “Freezing Singularities in Water
Drops,”PhysicsofFluids, V ol. 24, 2012.
154
Distribution A: Approved for public release; unlimited distribution. PA#15347
[121] Rhim, W.-K. and Ohsaka, K., “Thermophysical Properties Measurements of Molten Silicon
by High-Temperature Electrostatic Levitator: Density, V olume Expansion, Specific Heat
Capacity, Emissivity, Surface Tension and Viscosity,” Journal of Crystal Growth, V ol. 208,
1999, pp. 313–321.
[122] Steinfeld, A. and Fletcher, E. A., “A Spherical Reciever-Reactor with Specularly Reflect-
ing Inner Surfaces for High-Temperature Solar Furnace Applications,” Proceedings of the
NinethInternationalHeatTransferConference, Jerusalem, Israel, 1990, pp. 127–132.
[123] Touloukian, Y ., editor, Thermophysical Properties of Matter - The TPRC Data Seires, Pur-
due University, 1971.
[124] Incropera, F. P., DeWitt, D. P., Bergman, T. L., and Lavine, A. S., Introduction to Heat
Transfer, Wiley, 2006.
[125] Elgafy, A. and Mesalhy, O., “Numerical and Experimental Investigations of Melting and
Solidification Processes of High Melting Point PCM in a Cylindrical Enclosure,”Journalof
HeatTransfer, V ol. 126, October 2004, pp. 869–875.
[126] Burmeister, L. C.,ConvectiveHeatTransfer, Wiley-Interscience, 1993.
[127] Hollahan, J. R. and Gregory, N., “A Torsion Effusion Study of the Reaction of Graphite with
Oxides of Thorium and Zirconium,” The Journal of Physical Chemistry, V ol. 68, August
1964, pp. 2346–2351.
[128] Kobatake, H., Fukuyama, H., Minato, I., Tsukada, T., and Awaji, S., “Noncontact Mea-
surement of Thermal Conductivity of Liquid Silicon In A Static Magnetic Field,” Applied
PhysicsLe, V ol. 90, 2007.
[129] Oswald, M., Turek, M., and Bagdahn, J., “Numerical Simulations of Thermo-Mechanical
Stresses During The Casting of Multi-Crystaline Silicon Ingots,” 11th. Int. Conf. on Ther-
mal, Mechanical and Multiphysics Simulation and Experiments in Micro-Electronics and
Micro-Systems, Bordeaux, France, April 2010.
[130] Ciszek, T. F. and Schwuttke, G. H., “Method for Directional Solidification of Silicon,” US
Patent No. 4,243,471, 1981.
[131] Welford, W. and Winston, R., The Optics of Nonimaging Concentrators, Academic Press,
Inc., New York, NY , 1st ed., 1978.
155
Distribution A: Approved for public release; unlimited distribution. PA#15347
Appendix A
Matlab Model Formulation
Interior heat transfer for the MATLAB model discussed in Chapter 7 is handled via a 2-D axisym-
metric (r,z) conduction formulation. An energy balance method is used to calculated transient con-
duction and the “enthalpy method” is used to accommodate the phase change process [124, 125].
Consider a single node (m;n) as defined in Figure A.1.
During each time step,dt, heat transfer via conduction is calculated between node (m;n) and
the four surrounding nodes to determine the net heat transfer
q
net;(m;n)
=q
up;(m;n)
+q
down;(m;n)
+q
left;(m;n)
+q
right;(m;n)
(A.1)
Exterior nodes have additionalq
rad
andq
conv
terms taking into account radiation and convection
respectively and negate conduction terms when no adjacent node exists.
For nodes not containing a phase change material, the temperature change for the specifieddt
is then calculated as
T
(m;n)
=
q
net;(m;n)
dt
m
(m;n)
c
p;(m;n)
(A.2)
156
Distribution A: Approved for public release; unlimited distribution. PA#15347

(m, n+1)

(m, n-1)

(m+1, n)

(m-1, n)

(m, n)

Δr

Δz

Figure A.1: Diagram illustrating node (m;n) and the adjacent nodes used for computations
For nodes that do contain a phase change material, the “enthalpy method” is used to approx-
imate the phase change process. Each node is assigned a latent heat reservoir based on the mass
of phase change material (PCM) within the node. When the temperature of the node is within the
“mushy zone” (T
melt
0:1 K), all heat transfer is assigned to the phase change process and the
temperature of the node is kept constant
T (m;n) =
8
>
>
<
>
>
:
0; T
melt
0:1T
(m;n)
T
melt
+ 0:1
q
net;(m;n)
dt
m
(m;n)
c
p;(m;n)
; otherwise
(A.3)
L
f;(m;n)
=L
f;(m;n)
+q
net;(m;n)
(A.4)
When the latent heat reservoir value is either depleted in the case of cooling, or reaches a
maximum value in the case of heating, the node is considered either completely solid or completely
liquid and the node is allowed to change temperature out of the “mushy zone.”
157
Distribution A: Approved for public release; unlimited distribution. PA#15347

α

β

r
1
r
2
L

Figure A.2: Diagram illustrating two adjacent nodes in the z direction
The conduction equations used to calculate heat transfer between adjacent nodes assume one-
dimensional steady state condition for either a plane wall in thez direction, or a radial system in
the case of conduction in ther direction.
For conduction in thez direction, consider two adjacent nodes and as shown in Figure A.2.
Heat transfer from node to node can be defined as
q

=
(r
2
2
r
2
1
)(T

T

)
L(1=k

+ 1=k

)
(A.5)
withL defined as the distance between nodes,r
1
andr
2
as the upper and lower node radii,T as the
node temperature, andk as the thermal conductivity. In the model, computational time is saved by
pre-calculating temperature independent terms and storing them in a matrix so that for each node
(m;n), coefficientsa,b,c andd are defined containing this information for conduction left, right,
up, and down respectively.
For instance, conduction ”right” as defined by Equation A.5 is calculated in the software as
q

=
b

(T

T

)
(1=k

+ 1=k

)
(A.6)
withb

pre-calculated and defined as
158
Distribution A: Approved for public release; unlimited distribution. PA#15347

r
1
r
2
r
3
L

γ

δ

Figure A.3: Diagram illustrating two adjacent nodes in the r direction
b

=
(r
2
2
r
2
1
)
L
(A.7)
For conduction in ther direction, consider two adjacent nodes
and as shown in Figure A.3.
Heat transfer from node
to node can be defined as
q

=
(T

T

)
ln(r
1
=r
2
)
2Lk
+
ln(r
3
=r
2
)
2Lk

(A.8)
wherer
1
,r
2
andr
3
are defined as the location of the lower node, the location of the node bound-
ary, and the location of the upper node respectively and L is defined as the node width. Again,
to decrease computation time, temperature independent properties are pre-calculated and stored.
However, radial conduction equations require the storage of two constants to accommodate dif-
fering geometries when calculating the thermal resistance. This is accomplished by using a three
dimensional matrix in MATLAB.
For example, conduction “up” as defined by Equation A.8 is calculated in the software as
159
Distribution A: Approved for public release; unlimited distribution. PA#15347
q

=
(T

T

)
c

;1
k
+
c

;2
k

(A.9)
with constantsc

;1
andc

;2
defined as
c

;1
=
ln(r
2
=r
1
)
2L
(A.10)
c

;2
=
ln(r
3
=r
2
)
2L
(A.11)
Ultimately, at each time step, for the original node (m;n) as defined in Figure A.1, the four
following equations are used to calculate conduction with adjacent nodes.
q
up;(m;n)
=
(T
(m;n)
T
(m;n+1)
)
c
(m;n;1)
k
(m;n)
+
c
(m;n;2)
k
(m;n+1)
(A.12)
q
down;(m;n)
=
(T
(m;n)
T
(m;n1)
)
d
(m;n;1)
k
(m;n)
+
d
(m;n;2)
k
(m;n1)
(A.13)
q
left;(m;n)
=
a
(m;n)
(T
(m;n)
T
(m1;n)
)
1=k
(m;n)
+ 1=k
(m1;n)
(A.14)
160
Distribution A: Approved for public release; unlimited distribution. PA#15347
q
right;(m;n)
=
a
(m;n)
(T
(m;n)
T
(m+1;n)
)
1=k
(m;n)
+ 1=k
(m+1;n)
(A.15)
161
Distribution A: Approved for public release; unlimited distribution. PA#15347

Abstract (if available)

Abstract Solar thermal propulsion (STP) offers an unique combination of thrust and efficiency, providing greater total ΔV capability than chemical propulsion systems without the order of magnitude increase in total mission duration associated with electric propulsion. Despite an over 50 year development history, no STP spacecraft has flown to‐date as both perceived and actual complexity have overshadowed the potential performance benefit in relation to conventional technologies. The trend in solar thermal research over the past two decades has been towards simplification and miniaturization to overcome this complexity barrier in an effort finally mount an in‐flight test. ❧ A review of micro‐propulsion technologies recently conducted by the Air Force Research Laboratory (AFRL) has identified solar thermal propulsion as a promising configuration for microsatellite missions requiring a substantial ΔV and recommended further study. A STP system provides performance which cannot be matched by conventional propulsion technologies in the context of the proposed microsatellite “inspector" requiring rapid delivery of greater than 1500 m/s ΔV. With this mission profile as the target, the development of an effective STP architecture goes beyond incremental improvements and enables a new class of microsatellite missions. ❧ Here, it is proposed that a bi‐modal solar thermal propulsion system on a microsatellite platform can provide a greater than 50% increase in ΔV vs. chemical systems while maintaining delivery times measured in days. The realization of a microsatellite scale bi‐modal STP system requires the integration of multiple new technologies, and with the exception of high performance thermal energy storage, the long history of STP development has provided “ready" solutions. ❧ For the target bi‐modal STP microsatellite, sensible heat thermal energy storage is insufficient and the development of high temperature latent heat thermal energy storage is an enabling technology for the platform. The use of silicon and boron as high temperature latent heat thermal energy storage materials has been in the background of solar thermal research for decades without a substantial investigation. This is despite a broad agreement in the literature about the performance benefits obtainable from a latent heat mechanisms which provides a high energy storage density and quasi‐isothermal heat release at high temperature. ❧ In this work, an experimental approach was taken to uncover the practical concerns associated specifically with applying silicon as an energy storage material. A new solar furnace was built and characterized enabling the creation of molten silicon in the laboratory. These tests have demonstrated the basic feasibility of a molten silicon based thermal energy storage system and have highlighted asymmetric heat transfer as well as silicon expansion damage to be the primary engineering concerns for the technology. For cylindrical geometries, it has been shown that reduced fill factors can prevent damage to graphite walled silicon containers at the expense of decreased energy storage density. ❧ Concurrent with experimental testing, a cooling model was written using the “enthalpy method" to calculate the phase change process and predict test section performance. Despite a simplistic phase change model, and experimentally demonstrated complexities of the freezing process, results coincided with experimental data. It is thus possible to capture essential system behaviors of a latent heat thermal energy storage system even with low fidelity freezing kinetics modeling allowing the use of standard tools to obtain reasonable results. ❧ Finally, a technological road map is provided listing extant technological concerns and potential solutions. Improvements in container design and an increased understanding of convective coupling efficiency will ultimately enable both high temperature latent heat thermal energy storage and a new class of high performance bi-modal solar thermal spacecraft.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The effect of lattice structure and porosity on thermal conductivity of additively-manufactured porous materials

PDF

Hydrogen peroxide vapor for small satellite propulsion

PDF

Architectures and algorithms of charge management and thermal control for energy storage systems and mobile devices

PDF

Thermally driven water treatment with membrane distillation: membrane performance, waste heat integration, and cooling analysis

PDF

Non-radiative energy transfer for photovoltaic solar energy conversion: lead sulfide quantum dots on silicon nanopillars

PDF

Modular shading: using design to mitigate bus rider thermal heat stress

PDF

Advanced nuclear technologies for deep space exploration

PDF

A high-performance SuperWall: designed for a small residence at Joshua Tree National Park

PDF

Small organic molecules in all-organic redox flow batteries for grid-scale energy storage

PDF

Economic model predictive control for building energy systems

PDF

High power, highly efficient millimeter-wave switching power amplifiers for watt-level high-speed silicon transmitters

PDF

Energy control and material deposition methods for fast fabrication with high surface quality in additive manufacturing using photo-polymerization

PDF

Molecular-scale studies of mechanical phenomena at the interface between two solid surfaces: from high performance friction to superlubricity and flash heating

PDF

Direct C−H arylation for the synthesis of conjugated polymers

PDF

Resonant light-matter interactions in nanophotonic structures: for manipulating optical forces and thermal emission

PDF

Lateral sound flanking transmission at curtain wall mullions: an empirical investigation to identify controlling mechanisms

PDF

Contour crafting construction with sulfur concrete

PDF

Synthesis of high-quality nanoparticles using microfluidic platforms

PDF

Hybrid vat photopolymerization processes for viscous photocurable and non-photocurable materials

PDF

Nano-engineered devices for display and analog computing

Asset Metadata

Creator Gilpin, Matthew R. (author)

Core Title High temperature latent heat thermal energy storage to augment solar thermal propulsion for microsatellites

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Aerospace Engineering

Publication Date 07/20/2015

Defense Date 04/30/2015

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag advanced propulsion,convective coupling,latent heat,liquid boron,liquid silicon,microsatellites,molten boron,molten silicon,OAI-PMH Harvest,phase change material,radiation shielding model,solar concentration,solar furnace,solar thermal,spacecraft propulsion,thermal battery,thermal capacitor,thermal energy storage,thermal rocket

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Shiflett, Geoffrey R. (committee chair), Erwin, Daniel A. (committee member), Ronney, Paul D. (committee member), Sadhal, Satwindar S. (committee member)

Creator Email gilpin@usc.edu,matthew.gilpin.ctr@us.af.mil

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-600904

Unique identifier UC11302045

Identifier etd-GilpinMatt-3652.pdf (filename),usctheses-c3-600904 (legacy record id)

Legacy Identifier etd-GilpinMatt-3652.pdf

Dmrecord 600904

Document Type Dissertation

Format application/pdf (imt)

Rights Gilpin, Matthew R.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA