Journal of the British Interplanetary Society

VOLUME 72 NO.3 MARCH 2019 General issue

OPTICAL QUANTUM TECHNOLOGIES for Compact Rubidium Vapor-cell Frequency Standards in Space Using Small Satellites Aline Dinkelaker et al. MAGNETIC SHIELDING for Interplanetary Travel M.W. Sailer & H.M. Doss ASPIRE – AN INNOVATIVE HYPERSONIC PROPULSION PARADIGM Christopher MacLeod, Matthew Murray A REAL-TIME SPACE DEBRIS DETECTION SYSTEM for BIRALES Denis Cutajar et al.

www.bis-space.com

ISSN 0007-084X PUBLICATION DATE: 16 MAY 2019 Submitting papers International Advisory Board to JBIS

JBIS welcomes the submission of technical Rachel Armstrong, Newcastle University, UK papers for publication dealing with technical Peter Bainum, Howard University, USA reviews, research, technology and engineering in astronautics and related fields. Stephen Baxter, Science & Science Fiction Writer, UK James Benford, Microwave Sciences, California, USA Text should be: James Biggs, The University of Strathclyde, UK ■ As concise as the content allows – typically 5,000 to 6,000 words. Shorter papers (Technical Notes) Anu Bowman, Foundation for Enterprise Development, California, USA will also be considered; longer papers will only Gerald Cleaver, Baylor University, USA be considered in exceptional circumstances – for Charles Cockell, University of Edinburgh, UK example, in the case of a major subject review. Ian A. Crawford, Birkbeck College London, UK ■ Source references should be inserted in the text in square brackets – [1] – and then listed at the Adam Crowl, Icarus Interstellar, Australia end of the paper. Eric W. Davis, Institute for Advanced Studies at Austin, USA ■ Illustration references should be cited in Kathryn Denning, York University, Toronto, Canada numerical order in the text; those not cited in the Martyn Fogg, Probability Research Group, UK text risk omission. Raghavan Gopalaswami, Aerospace Researcher, India ■ Captions must be labelled with their Fig. number and should be as short as possible. Lamartine Guimarães, Institute for Advanced Studies, Brazil Mark Hempsell, Hempsell Astronautics Ltd, UK Illustrations should be: Takuto Ishimatsu, Massachusetts Institute of Technology, USA ■ Colour or mono, but should be as close to print Les Johnson, Marshall Space Flight Center, USA resolution (300 dpi) as possible. Poor-quality illustrations may compromise the acceptance of Terry Kammash, University of Michigan, USA paper for publication. Images embedded in Word Kelvin F. Long, Initiative for Interstellar Studies documents may be acceptable, but JBIS reserves Inoue Makoto, Institute of Astronomy & Astrophysics Academia Sinica, Taiwan the right to request separate higher-resolution Gregory L. Matloff, City University New York, USA image files from the author prior to publication. Koichi Mori, Nagoya University, Japan ■ Responsibility for copyright clearance of images rests entirely with the author. Richard Obousy, Richard Obousy Consulting LLC, USA Robert Parkinson, BIS, Aylesbury, UK Submission of papers George Schmidt, NASA John H Glenn Research Center, Ohio, USA ■ Papers for consideration should be sent by Paul Schuch, The SETI League Inc, USA email to [email protected] as both a Word document and as a Word PDF file (in order to Tabitha Smith, Bifrost, USA check for font anomalies), together with any Andreas Tziolas, Variance Dynamical Corporation, USA separate image files. Chris Welch, The International Space University, Strasbourg, France ■ If a paper is accepted for publication, the Friedwardt Winterberg, University of Nevada, Reno, USA author will be asked to sign a License to Publish form. This can be downloaded at www.bis- space.com/wp-content/uploads/2012/08/ WebsiteLicense.pdf. ■ Authors will receive a complimentary copy of the issue in which their paper appears. Editor Roger Longstaff Deputy Editor Duncan Law-Green Associate Editors Stephen Ashworth, We respectfully ask authors to adhere to these Keith Cooper, Stephen Gamble, Paul Gilster, Rob Swinney, Production MP3 Media guidelines. Failure to do so will result in the Promotion Gill Norman JBIS Office British Interplanetary Society, Arthur C. Clarke House, delay of acceptable papers for publication. 27-29 South Lambeth Road, London, SW8 1SZ, United Kingdom tel +44 (0)20 7735 3160 email [email protected] www.bis-space.com

Our full Guidelines for Authors can be downloaded DISTRIBUTION from www.bis-space.com JBIS is distributed worldwide by mail and may be received by annual subscription or purchase of single copies. It is available through membership of the British Interplanetary Society at much reduced rates. Subscription details for members, non-members and libraries are available from the above address.

JBIS is a publication that promotes the mission of the British Interplanetary Society. Opinions expressed in signed articles are those of the contributors and do not necessarily reflect the views of the Editor or the Council of the British Interplanetary Society. Security clearance, where necessary, is the responsibility of the author.

Published by the British Interplanetary Society. Registered Company No: 402498. Registered Charity No: 250556. Printed by Latimer Trend & Company Ltd, Estover Road, Plymouth, PL6 7PY, England. © 2019 British Interplanetary Society. No part of this magazine may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying or recording by any information storage or retrieval system without prior permission from the Publishers. contents VOLUME 72 NO.3 MARCH 2019

74 OPTICAL QUANTUM TECHNOLOGIES for Compact Rubidium Vapor-cell Frequency Standards in Space Using Small Satellites Aline Dinkelaker et al.

83 MAGNETIC SHIELDING for Interplanetary Travel M.W. Sailer & H.M. Doss

90 ASPIRE – AN INNOVATIVE HYPERSONIC PROPULSION PARADIGM Christopher MacLeod, Matthew Murray

102 A REAL-TIME SPACE DEBRIS DETECTION SYSTEM for BIRALES Denis Cutajar et al.

OUR MISSION STATEMENT The British Interplanetary Society promotes the exploration and use of space for the benefit of humanity, connecting people to create, educate and inspire, and advance knowledge in all aspects of astronautics.

JBIS Vol 72 No.3 March 2019 73 JBIS VOLUME 72 2019 PAGES 74-82

OPTICAL QUANTUM TECHNOLOGIES for Compact Rubidium Vapor-cell Frequency Standards in Space Using Small Satellites

ALINE N. DINKELAKER, AKASH KAPARTHY, SVEN E. REHER, MARKUS KRUTZIK, Humboldt-Universität zu Berlin, Department of Physics, Newtonstr. 15, 12489 Berlin, Germany Email [email protected]

AHMAD BAWAMIA, CHRISTIAN KÜRBIS, ROBERT SMOL, HEIKE CHRISTOPHER, ANDREAS WICHT, Ferdinand-Braun-Institut, Leibniz-Institut für Höchstfrequenztechnik, Gustav-Kirchhoff Straße 4, 12489 Berlin, Germany Email [email protected]

PHILIPP WERNER, JULIAN BARTHOLOMÄUS, SVEN ROTTER, MERLIN F. BARSCHKE, Technische Universität Berlin, Institute of Aeronautics and Astronautics, Marchstr. 12-14, 10587 Berlin, Germany Email [email protected]

ROBERT JÖRDENS, QUARTIQ GmbH, Rudower Chaussee 29, 12489 Berlin, Germany Email [email protected]

As part of the phase 0/A of the QUEEN mission, we evaluated our payload and satellite platform heritage and studied feasible mission scenarios for demonstrating optical frequency references aboard small satellites. We propose an optical vapor-cell frequency reference payload based on the 5S1/2 → 5D5/2 two-photon transition in 85Rb with low size, weight, and power budgets. In conjunction with an optical frequency comb, which can be used as an optical-to-microwave frequency divider, our payload could be advanced to a compact and simple vapor-cell based optical atomic clock. At the center of the payload is a laser system, based on micro-integrated laser technology, consisting of two independently frequency stabilized extended cavity diode lasers in a master-oscillator-power-amplifier configuration (ECDL-MOPA) at 778 nm. Ground-based development and environmental testing accompany the design and definition phase. Small satellites provide perfect vehicles for in-orbit tests of key technologies, due to the low associated costs compared to large satellite missions, and their potential for fast implementation. The requirements of our payload design are matched by the capabilities of the modular, flight-proven TUBiX20 satellite platform. This platform is designed to support demanding small science, technology and Earth observation missions of roughly 20 kg. Its modular approach realized in hard- and software allows scaling the platform’s capabilities for a wide variety of payloads while maintaining short development cycles. In this paper, we discuss the options and requirements for sending a compact, high-stability two-photon optical frequency reference into orbit and present advances in platform and payload design.

Keywords: Optical clock, Rubidium, Two-photon transition, Diode lasers, Quantum technology, Small satellite mission, Modular satellite platform, Frequency reference, Laser system

1 INTRODUCTION to which a large number of lasers with different functionality are stabilized. Optical quantum technologies in space have applications in navigation [1], geodesy [2, 3], optical communication [4], and Optical frequency references work by stabilizing a laser on fundamental physics. One focus is the interplay of gravity and an atomic transition such that the output frequency is stable quantum physics, such as novel approaches for gravitation- enough to act as reference for other devices. An optical fre- al wave detection [5, 6, 7, 8] or tests of general relativity with quency reference can be utilized in the same way as a micro- ultracold atoms [9, 10, 11, 12] and clock comparisons [13]. wave frequency by converting the optical frequency to RF Optical frequency references play a fundamental role in such frequencies with the use of an optical frequency comb [14]. instruments, just as they do on ground: They can be part of a There are a variety of different options for optical frequen- clock system or, in cold atom physics, operate as master lasers cy references (e.g. optical lattice clocks [15], beam standards [16, 17], Iodine standards [18], alkali-vapor cell clocks [19], see also overview by P. Gill [20]), which differ in complexity This paper was first presented at the 16th Reinventing Space and frequency stability. While stability might be a long-term (RiSpace) Conference, London, 2018. goal, portability is a deciding factor for space applications. This

74 Vol 72 No.3 March 2019 JBIS OPTICAL QUANTUM TECHNOLOGIES for Compact Rubidium Vapor-cell Frequency Standards in Space Using Small Satellites means that any optical frequency reference in orbit must be atomic transitions provide an absolute frequency reference. designed to be compact, robust, and energy-efficient. Optical However, the absolute transition frequency of the atoms be- systems, quantum technology, and frequency combs currently tween two energy levels can change as the energy levels shift undergo miniaturization efforts [19, 21, 22] and have been test- due to environmental effects. As a result, the laser is stabilized ed on sounding rockets [23, 24, 25, 26, 27], drop tower experi- to a shifting absolute transition frequency, which leads to fre- ments [28, 29], zero-g flights [30], as well as on the ISS [31, 32]. quency instabilities. To continue the transition from ground-based instruments to orbit, small satellites are ideal platforms for tests of key compo- In the laboratory, frequency noise can be controlled by keep- nents and subsystems [33, 34]. ing environmental parameters constant, installing appropriate shielding, and using low-noise electronics. In the challenging The last decades have been characterized by an ever-increas- environment of space, the implementation and adaptation of ing interest in facilitating space missions as means of gathering different techniques has to be tested. There are two main as- global observation data products, introducing worldwide com- pects when making the transition from lab to satellite based munication services or pursuing scientific goals and objectives. setups. Firstly, the technological realization of the setup has to Consequently, the number of satellites launched in recent years be adapted to the satellite environment. Limited size, weight, has been growing exponentially. This especially holds true for and power (SWaP) budgets can demand a customized design the field of small satellites ranging from 1 to 100 kg [35]. With and components and novel manufacturing and integration the advances of the electronics industry in miniaturization of methods might have to be applied. Standard optics, electron- components in both size and power consumption in conjunc- ics, and mounting components are usually not suitable for the tion with an increasing overall performance, nano- and micro- satellite payload. Additionally, there is no option to replace satellites are now capable of supporting demanding missions components or to re-align the setup manually. Secondly, the that were formerly conducted with larger spacecraft. Addition- setup is installed on a mobile platform. Effects that could be ally, new application areas are opened by these satellite classes, suppressed in the lab or that would result in a constant offset as for such missions, the overall cost is comparatively lower due might instead affect the frequency stability of the frequency ref- to reduced launch masses and shorter development times. This erence in orbit. As the environment changes throughout each also makes these missions especially interesting for technology orbit and throughout the mission, it is a technical challenge to demonstration applications. keep resulting frequency instabilities low. The Earth’s magnetic field or stray fields from the satellite platform can result in fre- Technische Universität Berlin has a long history of success- quency shifts, described e.g. by Martin et al. [39]. Temperature ful design, manufacturing, and operation of small spacecraft variations of critical components due to Sun exposure or power with a total of 16 satellites launched to date [36]. Today, satel- dissipation within the satellite can affect the frequency stability lites of Technische Universität Berlin in the range of 20 kg are as well [39]. As part of the Humboldt-Universität zu Berlin and based on the flight-proven TUBiX20 platform [37]. The mod- Ferdinand-Braun–Institut heritage from previous experiments ular design of this platform makes it well-adapted for different in the drop tower and on sounding rockets, some of these as- applications from fields like technology demonstration, sci- pects have been addressed: micro-integrated laser modules and ence, Earth observation and communications, as it facilitates complex laser systems have been built for the confined SWaP the scaling of relevant performance parameters of the space- budgets of a sounding rocket; experimental sequences for au- craft’s different subsystems. Hence, it is well-suited for hosting tonomous control and redundancy options have been tested the proposed payload suite of the QUEEN mission. during flight. With QUEEN, we aim to adjust the ground and sounding rocket designs for a small satellite payload. This in- With the QUEEN design study [38], we investigate the op- cludes accommodating the limited SWaP budget, as well as ar- tions and requirements to test a compact and stable optical va- chitecture for magnetic and radiation shielding, temperature por cell frequency reference, stabilized to an atomic transition insulation, heat transfer, and autonomous payload control. of Rb, in orbit. This would allow us to characterize environmen- tal effects on the frequency reference and monitor long-term As part of the QUEEN mission, the suitability of these de- effects on components. Additionally, in-orbit operation would signs shall be studied by monitoring electro-optical properties increase the maturity of different components and techniques of the laser diode such as diode voltage and current. Measur- that are included in the setup: Optical frequency references ing the optical output power along the path will help identify as proposed for the QUEEN payload contain e.g. laser mod- potential degradation risk of optical components in orbit. Fre- ules, optics, or timing systems. Advancing these would benefit quency measurements will show additional noise sources due other space-based quantum technology applications, such as to gradients or damage within the payload. These in-orbit tests quantum communication payloads demonstrating, e.g. optical are necessary to better understand the effects of the space en- memory on small satellites. Autonomy concepts that will have vironment on the frequency reference and the required level of to be applied to stabilize the frequency reference could be im- additional control techniques for future improvements. plemented not only in future space-based experiments but also for remote control in ground-based laboratories and devices. The QUEEN mission is designed for an operational lifetime of one year. The orbit is constrained by the use of the facilities at This paper describes the QUEEN mission goals and outline, Technische Universität Berlin as primary ground station. Con- the current design study results for the optical frequency refer- sequently, the inclination of the destination orbit is required to ence payload and the modular satellite platform, as well as an be higher than 52 degrees. Additionally, the satellite can rely on outlook with necessary steps for going to orbit. further ground stations accessible from the university’s mission control center. The orbital altitude for the QUEEN mission is 2 THE QUEEN MISSION assumed to be in the order of 500 to 600 km. This is to ensure a minimal mission duration of one year on the one hand and to The QUEEN mission aims to test an optical frequency refer- guarantee reentry within 25 years [40] without active support ence payload in orbit. Optical frequency references based on on the other hand.

JBIS Vol 72 No.3 March 2019 75 ALINE DINKELAKER ET AL

Fig.1 L. Overview of QUEEN Payload Design.

3 PAYLOAD DESIGN 5D5/2 transition in 85Rb. It can be addressed by two-photon ab- sorption at 778 nm. The 5S1/2 → 5D5/2 transition has a very small The QUEEN payload design consists of a compact optical vapor linewidth of around 300 kHz [41, 42], which directly benefits cell frequency reference. It is based on two ECDL master-os- the stability of the laser that is frequency locked to it. Fractional cillator-power-amplifier (MOPA) that can be independently instabilities on the order of 10-13 have already been achieved frequency stabilized to atomic transitions of the alkali atom- with lasers locked to Rb two-photon transitions in ground- ic species rubidium. To build a frequency reference, only one based laboratories [19, 39, 42, 43] and instabilities of 10-15 are laser is necessary. However, as we intend to gain information thought to be possible with a stronger level of environmental on the frequency stability, we implement a second laser. Light control [39]. This would provide a stable optical frequency ref- from both lasers is overlapped and the frequency difference of erence in a compact vapor cell setup. the two lasers can be measured as beat note. The mean value will be the nominal frequency difference, resulting from using To utilize atomic transitions as absolute frequency refer- different atomic transitions for stabilization and a value known ence, different spectroscopy techniques can be applied. For the from literature. proposed frequency reference payload, frequency modulation two-photon spectroscopy would be used to stabilize the laser Atomic transitions provide the possibility to stabilize (lock) [44, 45]. Laser light travels through the vapor cell filled with an oscillator (here: the laser) to an absolute value rather than atomic gas and interacts with the atoms. Part of the atoms ab- relative to a second oscillator. The absolute frequency value as sorb the light and change their internal state from ground state well as the stability of this frequency depends on the atomic to excited state. In case of the proposed two-photon transition, species, potentially the atomic isotope, and on the energy levels the atoms do not fall back to the ground level directly. Instead, of the atomic transitions. Various different energy levels exist they decay via the 6P3/2 immediate level. The resulting 6P3/2 for each atom, defined by different quantum numbers. Alka- → 5S1/2 fluorescence is recorded using a photomultiplier tube li atoms, such as rubidium, are widely used in atomic physics (PMT) or a similar device, such as a silicon photomultiplier experiments. They have the advantage of having a comparably (SiPM). By modulating the laser beam (e.g. with an electro-op- simple energy level structure due to their hydrogen-like elec- tical modulator) and demodulating the fluorescence signal, tron configuration. The atomic transition that we plan to use a feedback signal for the laser stabilization can be produced. for the QUEEN frequency reference payload is the 5S1/2 → Active stabilizing electronics for the laser current and tempera-

76 Vol 72 No.3 March 2019 JBIS OPTICAL QUANTUM TECHNOLOGIES for Compact Rubidium Vapor-cell Frequency Standards in Space Using Small Satellites

Fig.2 CAD Model of an ECDL-MOPA Module with Dimensions 30 x 80 x 10 mm³. VHBG: volume holographic Bragg grating; µ-isolator: micro-optical isolator; µ-mirror: micro-mirror. The optical fibers exiting the fiber coupler are not shown in the figure.

cillator consisting of either a monolithic diode laser chip, such as a DFB laser, or as a discrete laser such as an ECDL. The laser chips are hybrid micro-integrated onto a micro-optical bench consisting of lithographically patterned AlN substrates togeth- er with micro-optics and discrete electronics. A CAD model of a MOPA with an ECDL as master-oscillator is shown in Fig. 2.

A 2 mm long, single longitudinal mode ridge-waveguide di- ode laser chip (Chip 1) is used as gain medium of the ECDL part. The front facet is coated to provide a power reflectivity of 30%, while the rear, intra-cavity facet is AR coated and fea- tures a residual reflectivity of about 0.1%. Frequency selective feedback is provided by a Volume Holographic Bragg Grat- ing (VHBG), that provides a resonant diffraction efficiency of approximately 70%. An intra-cavity aspheric lens collimates the emission of the rear-facet of the diode laser chip onto the VHBG. The light emitted through the front facet is collimated by means of cylindrical lenses and are transmitted through a ture is required for the frequency reference. The vapor cell will micro-optical isolator (µ-isolator), before being focused into a be heated to produce a higher vapor pressure and thus higher power-amplifier chip (Chip 2) by means of cylindrical lenses fluorescence signal. In order to benefit from the small linew- identical to those mentioned above. idth and potentially low level of instability of this transition, the experimental setup requires stabilization of the laser power The power-amplifier chip is a 6mm long, single longitudinal at the vapor cell as well as magnetic shielding around the vapor mode ridge-waveguide diode laser chip. Its facets are antireflec- cell [39]. An overview of the optical setup and stabilizing elec- tion coated to a residual reflectivity of about 0.1%. Its straight tronics is shown in Fig. 1. waveguide is tilted by 3 to the facet normal in order to reduce the effective reflectivity back into the waveguide by another 3 3.1 Laser Sources orders of magnitude. This guarantees an effective reflectivity of 10-5–10-6 with respect to the formation of a parasitic Fabry-Pérot As laser sources, semiconductor diode lasers are ideal can- resonator between the facets of the power-amplifier chip. This didates: they are compact, robust, come in a variety of wave- suppression is sufficient to avoid that the amplifier reaches laser lengths and are not expected to suffer much from the relatively threshold when the seed laser is turned off (no self-lasing) even weak radiation in low Earth orbit (LEO). In QUEEN, micro-in- at the largest injection current setting. The output beam of the tegrated laser modules by the Ferdinand-Braun-Institut, Lei- power-amplifier is collimated by means of an aspheric lens with bniz-Institut für Höchstfrequenztechnik Berlin (FBH) are a focal length of f = 2.0 mm. The lens position along all three planned as laser light sources. Several such lasers have been directions is optimized for maximum fiber coupling efficiency. launched in different microgravity missions, on sounding Once the lens is bonded, the two mirrors located between the rockets and in the ZARM drop tower [24, 25, 27, 46, 47, 48]. lens and the fiber coupler are precisely aligned for maximum coupling efficiency and then bonded as well. The optical fiber is The conceptual approach to the foreseen laser modules rests single-mode and polarization-maintaining. on versatility and multi-functionality, embodied by a design incorporating two arbitrary semiconductor-based active or The laser emission transmitted through the VHBG is col- passive chips and two optical ports that can be used as input lected into a rear optical port consisting of a fiber coupler or output ports according to the requirements. For instance, a identical to the main port. The output power (typically in the MOPA architecture can be implemented, with the master-os- range of 1 mW) of the single-mode, polarization-maintaining

JBIS Vol 72 No.3 March 2019 77 ALINE DINKELAKER ET AL

Fig.3 ECDL-MOPA Laser Module for Operation at 780nm. The photograph shows the AlN ceramic body and the housing, as well as the optical and electrical feedthroughs. fiber at the rear port can be used, for instance, for the genera- ulator required for the two-photon spectroscopy are entirely tion of a beat note measurement or for performance monitor- new developments. Here we will merge available digital hard- ing purposes. ware with novel algorithms, data process and control schemes. The demands on robustness and autonomous operation re- The micro-optical bench is fully packaged into a Kovar hous- quire the development of new solutions. Open Source Software ing, with electrical signals fed through via Mini-SMP coaxial frameworks like ARTIQ [52] and Open Hardware ecosystems connectors and optical signals fed through via single mode, like Sinara [53] provide a flexible and unencumbered baseline polarization-maintaining optical fibers. The module housing to test and prototype new technologies like algorithms for au- has dimensions of 125 mm x 75 mm x 22.5 mm and a mass of tonomous and robust operations or data processing schemes. approximately 760 g. Fig.3 shows a picture of an ECDL-MOPA laser module. Such a laser module operating at a wavelength Temperature controller designs that achieve sub-mK tem- around 1064nm [49] has successfully been flown as part of an perature stability and sufficient SWaP budgets are available as iodine-based frequency reference in the JOKARUS [25] exper- open hardware within the Sinara ecosystem as well as from iment on the TEXUS 54 sounding rocket [50] in May 2018. commercial providers. These modules are typically interfaced It delivers more than 500mW output power ex-fiber within a with via Ethernet or serial protocols (SPI, I2C, UART) and will technical (FWHM) linewidth of 25 kHz measured on a times- require adaptation and qualification to be used for laser mod- cale of 1 ms (estimated from the frequency noise PSD accord- ule and vapor cell temperature control. ing to the method proposed by Di Domenico et al. [51]) and a Lorentzian linewidth of smaller than 1kHz. The laser current driver will require low current noise at fre- quencies near and above the control loop bandwidth. In addi- The ECDL-MOPA technology has already been transferred tion to the current steering signal for frequency stabilization a to the wavelength of 780 nm for experiments on Rubidium method to control the offset current for power stabilization is Bose-Einstein-Condensates (BEC) and atom interferometry, foreseen. Potentially suitable circuit designs both in open hard- planned on board a sounding rocket. At an emission wave- ware as well as proprietary modules have been identified and length of 780.24 nm (corresponding to the Rubidium D2 line), evaluation efforts with respect to the aforementioned require- an optical output power of the ECDL-MOPA of > 400 mW ments and the need for adaptation and qualification are ongoing. is achieved within a technical (FWHM) linewidth of 150kHz measured on a timescale of 1 ms (estimated from the frequency The frequency measurement unit that detects the beat fre- noise PSD according to the method proposed by Di Domeni- quency between the two redundant and independently oper- co et al. [51]) and a Lorentzian linewidth of < 1kHz. Identical ating spectroscopy systems needs to be able to operate in two diode laser chips and micro-optics can be used to produce the modes. In the first mode during clock operation when both ECDL-MOPAs emitting at 778 nm for the QUEEN payload, systems are stabilizing their laser source to the same optical with the exception of the VHBGs, that need to have their cen- atomic transition, the beat frequency is the sum of the dif- tral wavelength adjusted to 778 nm. As a result, the perfor- ferential frequency error and the AOM difference frequency. mance of the ECDL-MOPAs emitting at 778 nm is expected This permits selecting a beat frequency offset that both cancels to be identical to those emitting at 780 nm, thus fulfilling the common mode time base errors and has optimal properties for requirements for the laser modules in the QUEEN payload. the detection electronics. In the second mode where the beat frequency is scanned over a wider range to locate the operating 3.2 Payload Electronics point of the two lasers with respect to the optical transitions the unit needs to be able to detect the beat frequency with less The payload electronics implement laser drive and stabiliza- accuracy but much larger range. tion, data processing, management, and supervision tasks. Some technologies like stable laser current drivers and tem- 3.3 Signal Processing for Laser Stabilization perature controllers already exist for laboratory environments and may just need adaptation and qualification. Others like a Both analog and digital circuitry have previously been used low-power fully integrated and all-digital modulator-demod- with the laser systems by Humboldt-Universität zu Berlin to

78 Vol 72 No.3 March 2019 JBIS OPTICAL QUANTUM TECHNOLOGIES for Compact Rubidium Vapor-cell Frequency Standards in Space Using Small Satellites perform laser stabilization [25, 27]. Digital, integrated control nents and techniques and make suitable design choices. In a applications can offer great potential [54, 55] with regards to laboratory ground test bed at Humboldt-Universität zu Berlin, scalability and functionality. Based on this, it is planned that we are currently investigating different vapor cell types, as well the QUEEN payload executes laser stabilization tasks by digital as cell heating and insulation configurations. After this first run signal processing (DSP) in an embedded system. of hardware tests, frequency modulation techniques and spec- trum detection will be studied. Here, we aim to find a power Here, the payload can benefit from synergies with the and space efficient solution where we can use the same cell, TUBiX20 satellite platform, as this already uses commercial the same sensor and much of the same optics for the spectrum off-the-shelf (COTS) microcontrollers with DSP functionality generation by digitally separating the signals [56]. If no com- and flight heritage. pact solution is found, the spectroscopy part of the setup will have to be duplicated for the stabilization of two lasers. They offer a hardware floating point unit (FPU) with DSP instruction set, as well as integrated analog-to-digital (ADC) 4 SATELLITE PLATFORM and digital-to-analog converters (DAC). Including the multi- ple timers and direct memory access (DMA) channels, such The QUEEN spacecraft will base on the TUBiX20 platform of devices are well suited to control the laser frequency modula- Technische Universität Berlin. This platform is designed to sup- tion via an electro-optic modulator (EOM) or digitize the spec- port a wide array of different missions due to its flexible architec- troscopy signal. Furthermore, digital filters like finite impulse ture [37]. To this end, TUBiX20 features a high level of modular- response (FIR) or infinite impulse response (IIR) filters allow ity that is based on clear standards for its software and hardware to demodulate the spectroscopy signal or can act as a loop filter interfaces. This enables the removal or addition of components for laser frequency control. By integrating all the former analog without major redesigns of electronics or software. The core functionality in the digital domain of one microcontroller, this functionality of the TUBiX20 platform is constituted by its re- approach is expected to contribute to a smaller SWaP budget dundantly implemented computational nodes communicating and enhance the performance by digital signal processing tech- via a common data bus system. This synergizes with the modu- niques. Design parameters like coefficients of filters or control lar software approach since the same fundamental software suite loop transfer functions are programmable and depend no can be used on any of the system’s node. All nodes are located longer on analog circuitry resulting in higher flexibility. within the core avionics compartment of the satellite.

3.4 Payload Requirements The TUBiX20 platform was tested in orbit for the first time during the TechnoSat mission. TechnoSat was launched in July With customized, power efficient electronics, we estimate the 2017 and has been in operation since then to verify key com- total power consumption of the frequency reference payload ponents of the platform. Its mission is the in-orbit verifica- to 25 W. We expect the payload to fit into a volume of roughly tion of seven technology payloads [57]. Orbit results from the 8 liters and its total mass to be around 12 kg. Due to several TechnoSat mission can be found in several publications [58, temperature dependent components in the payload, these re- 59, 60, 61, 62]. quire that the temperature within the payload bay stays stable to within 10 K. Additionally, the temperature-stabilized com- The second mission implementing the TUBiX20 platform is ponents inside the payload as well as the electronics will gener- TUBIN that aims at detecting high-temperature events using ate heat that has to be transported away from the payload bay. microbolometer technology [63, 64]. A thermal management concept will be necessary to cope with this excess heat. Based on the modular approach of the TUBiX20 platform, the subsystems and components for the QUEEN mission are To test the frequency reference and its key components in configured to meet the requirements of the payload. To this space, the electro-optical properties of the laser sources will be end, the attitude determination and control system (ADCS) is measured. This includes optical output power of both lasers, equipped with a moderate accuracy sensor suite and a set of re- associated driving currents and voltages, as well as tempera- action wheels (RWS) and magnetorquers for controlling the at- tures of lasers and within the payload. Additionally, measure- titude. Furthermore, a GNSS receiver is included to determine ments regarding the frequency stabilization can be performed, the position of the spacecraft. The communications subsystem i.e. frequency stabilization feedback parameters such as cur- is constituted by four UHF transceivers for nominal TM/TC rent or fluorescence amplitude, and of course the frequency communication (COM) as well as a high-speed data downlink difference between the two lasers with different sampling rates option via S-band. Here, the S-band transmitter is directly con- and over different durations. If a scan of the laser frequency is nected to the payload data handling (PDH) node to integrate performed on the satellite, the fluorescence spectrum can be the option of forwarding payload data to the ground without recorded. The electronics and the data rate will have to handle the necessity to distribute it on the central data bus system. the payload performance requirements. The payload will also Besides fulfilling high-level tasks such as satellite mode man- require different experimental sequences (e.g. initialize, re- agement and telemetry storage, the on-board computer (OBC) cord fluorescence spectrum, activate stabilization) that can be reads a number of temperature sensors to confirm the thermal divided into different experimental modes. Autonomy will be environment of the payload. A block diagram featuring subsys- critical in orbit as direct communication to the satellite and its tems, components and payload as well as their interconnecting payload is limited. All experimental sequences will be prepro- power and data busses is given in Fig. 4 overleaf. grammed so that the payload autonomously records, analyses and sends data according to the current experimental mode. In line with the increased power consumption of the payload when compared to the preceding instalments of the TUBiX20 3.5 Ground Tests platform, the electrical power system (EPS) is equipped with a primary solar panel defining a preferential attitude towards the Several tests on ground are necessary in order to test compo- Sun. It is attached to one of the sides of the octagonal structure

JBIS Vol 72 No.3 March 2019 79 ALINE DINKELAKER ET AL

Fig.4 Systems Design of the QUEEN Spacecraft with the central power and data bus system shown in the middle and the dark blue boxed indicate computational nodes, while the light blue boxes indicate connected components (modified from [38]). and features a sufficient number of solar cells to allow contin- optical frequency reference, a two-photon transition of rubid- uous operation of the payload. However, the QUEEN satellite ium has been selected as a good candidate. The selected tran- will also survive periods of free tumbling as further solar cells sition has a small linewidth and can be utilized to provide a are mounted on the remaining surfaces of the spacecraft. The frequency reference with low instability, while coming with the thermal design of the platform is purely passive implementing generally low level of complexity of vapor cell frequency refer- adequate surface characteristics to stabilize the payload within ences. Instabilities in the order of 10-15 are expected to be possi- its desired range. However, within the payload itself the temper- ble in well-controlled environments. However, to benefit from ature is actively managed. A rendering of the proposed platform the two-photon transition, the payload has to be shielded from design created in phase 0/A of the mission is depicted in Fig. 5. changing environmental parameters such as magnetic fields A summary of the key parameters of the platform configuration and radiation. Additionally, active stabilization of vapor cell implemented for the QUEEN mission is given in Table 1. temperature (heated), laser module temperature, and optical laser power has to be implemented. The temperature value and While the thermal stability and the electrical power de- stability at different points in the payload is one of the main manded by the QUEEN payload exceeds the requirements of payload design aspects. A thermal management concept has the TechnoSat payloads, most other capabilities that are need- to be developed for the payload, modelled, and tested as part ed for QUEEN, e.g. pointing accuracy and data downlink were of thermal-balance tests. Different ground and environmental already successfully demonstrated on orbit within the Techno- tests are thus required to find appropriate design solutions and Sat mission. components that can on the one hand provide a stable frequen- cy reference and on the other hand comply with the limited 5 SUMMARY AND OUTLOOK SWaP budget on the satellite platform.

With the QUEEN design study, we have identified a potential We have estimated the power consumption of the payload payload and platform design for an optical frequency reference to be 25 W. Customized payload control electronics will be demonstrator mission in orbit. The TUBiX20 satellite platform necessary to provide full functionality and laser stabilization can be adjusted to accommodate the payload requirements. As while being power efficient. The satellite platform TUBiX20

TABLE 1 Key parameters of the platform configuration implemented for the QUEEN mission

Design lifetime 1 year Spacecraft mass 30 kg Spacecraft volume 465 x 465 x 305 mm TM/TC link UHF and S band Payload downlink S band Position sensors GNSS receiver

Attitude sensors Magnetometers, MEMS gyroscopes, Sun sensors, Fiber optic rate sensors Fig.5 Rendering of the QUEEN spacecraft that was produced in Attitude actuators Reaction wheels, magnetorquer phase 0/A of the project.

80 Vol 72 No.3 March 2019 JBIS OPTICAL QUANTUM TECHNOLOGIES for Compact Rubidium Vapor-cell Frequency Standards in Space Using Small Satellites was adapted to match the enhanced power requirements com- technologies based on alkaline earth beam standards, and pared to previous payloads. The TUBiX20 platform is modular, reference oscillators for fundamental physics tests, e.g. test of incorporates comprehensive redundancy concepts, and is flight general relativity with ultracold atoms or clock comparisons. proven. The available payload bay matches the size (8 liters) and weight (12 kg) requirements of the payload. Acknowledgements

The QUEEN mission will allow tests of a compact optical The QUEEN project is funded by the Federal Ministry for Eco- frequency reference in space. It will demonstrate the suitabil- nomic Affairs and Energy (BMWi) through the German Aer- ity of a small modular satellite platform for quantum experi- ospace Centre (DLR) on the basis of a decision of the German ment payloads and incorporate autonomy concepts for pay- Bundestag (Grant No. 50 WM 1753-1755, 50WM1857-1858 load and platform. and 50 RU 1801).

The optical frequency can be translated from the optical to The development of the TUBiX20 platform is funded by the the RF domain by adding a frequency comb. This would pro- Federal Ministry for Economic Affairs and Energy (BMWi) vide a stable, compact optical vapor cell clock in space. Hence, through the German Aerospace Center (DLR) on the basis of the QUEEN mission is an important step towards optical quan- a decision of the German Bundestag within the TechnoSat and tum technology on satellites, such as next generation GNSS the TUBIN mission (Grant No. 50 RM 1219 and 50 RM 1102).

REFERENCES 1. J.D. Prestage and G.L. Weaver, “Atomic clocks and oscillators for deep-space Cacciapouti, E. Rocco, K. Bongs, W. Ertmer and E.M. Rasel, “Differential navigation and radio science,” Proc. IEEE 95, pp. 2235–2247, (2007). atom interferometry with 87Rb and 85Rb for testing the UFF in STE- 2. N. Yu, J. Kohel, J. Kellogg and L. Maleki, “Development of an atom QUEST,” Available at arXiv:1312.5963, (2013). interferometer gravity gradiometer for gravity measurement from space,” 12. T. Schuldt, C. Schubert, M. Krutzik, L.G. Bote, N. Gaaloul, J. Hartwig, Appl. Phys. B, Vol. 84, pp. 647–652, (2006). H. Ahlers, W. Herr, K. Posso-Trujillo, J. Rudolph, S. Seidel, T. Wendrich, 3. O. Carraz, , C. Siemes, L. Massotti, R. Haagmans, and P. Silvestrin, W. Ertmer, S. Herrmann, A. Kubelka-Lange, A. Milke, B. Rievers, E. “A spaceborne gravity gradiometer concept based on cold atom Rocco, A. Hinton, K. Bongs, M. Oswald, M. Franz, M. Hauth, A. Peters, interferometers for measuring Earth’s gravity field,” Microgravity Sci. A. Bawamia, A. Wicht, B. Battelier, A. Bertoldi, P. Bouyer, A. Landragin, Technol., Vol. 26, pp. 139–145, (2014). D. Massonnet, T. Lévèque, A. Wenzlawski, O. Hellmig, P. Windpassinger, 4. J. Yin, Y. Cao, Y.-H. Li, S.-K. Liao, L. Zhang, J.-G. Ren, W.-Q. Cai, W.-Y. Liu, K. Sengstock, W. von Klitzing, C. Chaloner, D. Summers, P. Ireland, B. Li, H. Dai, G.-B. Li, Q.-M. Lu, Y.-H. Gong, Y. Xu, S.-L. Li, F.-Z. Li, Y.-Y. I. Mateos, C.F. Sopuerta, F. Sorrentino, G. M. Tino, M. Williams, C. Yin, Z.-Q. Jiang, M. Li, J.-J. Jia, G. Ren, D. He, Y.-L. Zhou, X.-X. Zhang, N. Trenkel, D. Gerardi, M. Chwalla, J. Burkhardt, U. Johann, A. Heske, E. Wang, X. Chang, Z.-C. Zhu, N.-L. Liu, Y.-A. Chen, C.-Y. Lu, R. Shu, C.-Z. Wille, M. Gehler, L. Cacciapuoti, N. Gürlebeck, C. Braxmaier and E. Peng, J.-Y. Wang and J.-W. Pan, “Satellite-based entanglement distribution Rasel, “Design of a dual species atom interferometer for space,” Exp. over 1200 kilometers,” Science, Vol. 356, pp. 1140-1144, (2017). Astron., Vol. 39, pp. 167–206, (2015). 5. P.W. Graham, J.M. Hogan, M.A. Kasevich and S. Rajendran, “New 13. L. Cacciapuoti, and C. Salomon, “Space clocks and fundamental tests: method for gravitational wave detection with atomic sensors,” Phys. Rev. the ACES experiment,” Eur. Phys. J., Vol. 172, pp. 57–68, (2009). Lett., Vol. 110, p. 171102, (2013). 14. D.J. Jones, S.A. Diddams, J.K. Ranka, A. Stentz, R.S. Windeler, J.L. Hall 6. G.M. Tino, , F. Sorrentino, D. Aguilera, B. Battelier, A. Bertoldi, Q. and S.T. Cundiff, “Carrier-envelope phase control of femtosecond mode- Bodart, K. Bongs, P. Bouyer, C. Braxmaier, L. Cacciapuoti, N. Gaaloul, locked lasers and direct optical frequency synthesis,” Science, Vol. 288, N. Gürlebeck, M. Hauth, S. Herrmann, M. Krutzik, A. Kubelka, A. 635–639, (2000). Landragin, A. Milke, A. Peters, E.M. Rasel, E. Rocco, C. Schubert, T. 15. M. Takamoto, F.-L. Hong, R. Higashi and H. Katori, “An optical lattice Schuldt, K. Sengstock and A. Wicht, “Precision gravity tests with atom clock,” Nature, Vol. 435, pages 321–324, (2005). interferometry in space,” Nuclear Phys. B – Proc. Suppl. 243–244, pp. 16. R.W. Fox, J.A. Sherman, W. Douglas, J.B. Olson, A.D. Ludlow and C.W. 203–217, (2013). Oates, “A high stability optical frequency reference based on thermal 7. S. Dimopoulos, P.W. Graham, J.M. Hogan and M.A. Kasevich, “Testing calcium atoms,” Proceedings of 2012 IEEE Frequency Control Symposium. general relativity with atom interferometry,” Phys. Rev. Lett., Vol. 98, p. Baltimore, MD, USA, (2012). 111102, (2007). 17. H. Shang, X. Zhang, S. Zhang, D. Pan, H. Chen, and J. Chen, 8. J.M. Hogan, D.M.S. Johnson, S. Dickerson, T. Kovachy, A. Sugarbaker, S.- “Miniaturized calcium beam optical frequency standard using fully- W. Chiow, P.W. Graham, M.A. Kasevich, B. Saif, S. Rajendran, P. Bouyer, sealed vacuum tube with 10−15 instability,” Optics Express, Vol. 25, Iss. B.D. Seery, L. Feinberg and R. Keski-Kuha, “An atomic gravitational wave 24, pp. 30459-30467, (2017). interferometric sensor in low Earth orbit (AGIS-LEO),” Gen. Relativ. 18. K. Döringshoff, T. Schuldt, E.V. Kovalchuk, J. Stühler, C. Braxmaier and Gravitation, Vol. 43, pp. 1953–2009, (2011). A. Peters, “A flight-like absolute optical frequency reference based on 9. D.N. Aguilera, , H. Ahlers, B. Battelier, A. Bawamia, A. Bertoldi, R. iodine for laser systems at 1064 nm,” Applied Physics B, Vol. 123, p. 183, Bondarescu, K. Bongs, P. Bouyer, C. Braxmaier, L. Cacciapuoti, C. (2017). Chaloner, M. Chwalla, W. Ertmer, M. Franz, N. Gaaloul, M. Gehler, 19. G. Phelps, N. Lemke, C. Erickson, J. Burke and K. Martin, “Compact D. Gerardi, L. Gesa, N. Gürlebeck, J. Hartwig, M. Hauth, O. Hellmig, Optical Clock with 5x10-13 Instability at 1 s,” Navigation, Vol. 65, Iss. 1, W. Herr, S. Herrmann, A. Heske, A. Hinton, P. Ireland, P. Jetzer, U. pp. 49–54, (2018). Johann, M. Krutzik, A. Kubelka, C. Lämmerzahl, A. Landragin, I. 20. P. Gill, “Optical Frequency Standards,” Metrologia, Vol. 42, pp. S125– Lloro, D. Massonnet, I. Mateos, A. Milke, M. Nofrarias, M. Oswald, A. S137, (2005). Peters, K. Posso-Trujillo, E. Rasel, E. Rocco, A. Roura, J. Rudolph, W. Schleich, C. Schubert, T. Schuldt, S. Seidel, K. Sengstock, C.F. Sopuerta, 21. J. Kitching. “Chip-scale atomic devices,” Applied Physics Reviews, Vol. 5, F. Sorrentino, D. Summers, G.M. Tino, C. Trenkel, N. Uzunoglu, W. von Iss. 3, p. 031302, (2018). Klitzing, R. Walser, T. Wendrich, A. Wenzlawski, P. Weßels, A. Wicht, E. 22. M.T. Hummon, S. Kang, D.G. Bopp, Q. Li, D.A. Westly, S. Kim, C. Wille, M. Williams, P. Windpassinger and N. Zahzam, “STE-QUEST— Fredrick, S.A. Diddams, K. Srinivasan, V. Aksyuk, and J.E. Kitching, test of the universality of free fall using cold atom interferometry,” “Photonic chip for laser stabilization to an atomic vapor with 10-11 Classical Quantum Gravity, Vol. 31, p. 115010, (2014). instability,” Optica, Vol. 5, Iss. 4, pp. 443–449, (2018). 10. S. Herrmann, H. Dittus, and C. Lämmerzahl, the QUANTUS and 23. DLR Press release - MAIUS 1: Erstes Bose-Einstein-Kondensat im All PRIMUS Teams, “Testing the equivalence principle with atomic erzeugt, http://www.dlr.de/dlr/desktopdefault.aspx/tabid-10212/332_ interferometry,” Classical Quantum Gravity, Vol. 29, p. 184003, (2012). read-20337 (accessed 11th October 2017). 11. C. Schubert, J. Hartwig, H. Ahlers, K. Posso-Trujillo, N. Gaaloul, U. 24. V. Schkolnik, O. Hellmig, A. Wenzlawski, J. Grosse, A. Kohfeldt, K. Velte, A. Landragin, A. Bertoldi, B. Battelier, P. Bouyer, F. Sorrentino, Döringshoff, A. Wicht, P. Windpassinger, K. Sengstock, C. Braxmaier, G. M. Tino, M. Krutzik, A. Peters, S. Herrmann, C. Lämmerzahl, L. M. Krutzik and A. Peters, “A compact and robust diode laser system for

JBIS Vol 72 No.3 March 2019 81 ALINE DINKELAKER ET AL

atom interferometry on a sounding rocket,” Appl. Phys. B, Vol. 122, pp. of the Doppler-free Two-photon Transitions in Rubidium,” Optics 1–8, (2016). Communications, Vol. 108, pp. 91–96, (1994). 25. V. Schkolnik, K. Döringshoff, F.B. Gutsch, M. Oswald, T. Schuldt, C. 44. M.J. Snadden, A.S. Bell, R.B.M. Clarke and E. Riis, "Frequency Braxmaier, M. Lezius, R. Holzwarth, C. Kürbis, A. Bawamia, M. Krutzik modulation two-photon spectroscopy with a modelocked laser in laser- and A. Peters, “JOKARUS - design of a compact optical iodine frequency cooled Rb," in Quantum Electronics and Laser Science Conference, R. reference for a sounding rocket mission,” EPJ Quantum Technology, Vol. Freeman and P. Corkum, eds., Vol. 10 of OSA Technical Digest (Optical 4, pp. 4-9, (2017). Society of America), paper QThF24, (1996). 26. M. Lezius, T. Wilken, C. Deutsch, M. Giunta, O. Mandel, A. Thaller, V. 45. G.C. Bjorklund, “Frequency modulation spectroscopy: a new method Schkolnik, M. Schiemangk, A. Dinkelaker, A. Kohfeldt, A. Wicht, M. for measuring weak absorptions and dispersions,” Opt. Lett., Vol. 5, pp. Krutzik, A. Peters, O. Hellmig, H. Duncker, K. Sengstock, P. Windpassinger, 15–17, (1980). K. Lampmann, T. Hülsing, T.W. Hänsch and R. Holzwarth, “Space-borne 46. E. Luvsandamdin, S. Spießberger, M. Schiemangk, A. Sahm, G. Mura, frequency comb metrology,” Optica, Vol. 3, pp. 1381–1387, (2016). A. Wicht, A. Peters, G. Erbert and G. Tränkle, “Development of narrow 27 A.N. Dinkelaker, M. Schiemangk, V. Schkolnik, A. Kenyon, K. linewidth, micro-integrated extended cavity diode lasers for quantum Lampmann, A. Wenzlawski, P. Windpassinger, O. Hellmig, T. Wendrich, optics experiments in space,” Appl. Phys. B, Vol. 111, pp. 225–260, (2013). E.M. Rasel, M. Giunta, C. Deutsch, C. Kürbis, R. Smol, A. Wicht, M. 47. E. Luvsandamdin, C. Kürbis, M. Schiemangk, A. Sahm, A. Wicht, A. Krutzik and A. Peters, “Autonomous frequency stabilization of two Peters, G. Erbert, and G. Tränkle, “Micro-integrated extended cavity extended-cavity diode lasers at the potassium wavelength on a sounding diode lasers for precision potassium spectroscopy in space,” Opt. Express, rocket,” Appl. Opt., Vol. 56, pp. 1388–1396, (2017). Vol. 22, pp. 7790–7798, (2014). 28. S. Kulas, C. Vogt, A. Resch, J. Hartwig, S. Ganske, J. Matthias, D. 48. M. Schiemangk, K. Lampmann, A. Dinkelaker, A. Kohfeldt, M. Krutzik, Schlippert, T. Wendrich, W. Ertmer, E. M. Rasel, M. Damjanic, P. C. Kürbis, A. Sahm, S. Spießberger, A. Wicht, G. Erbert, G. Tränkle and Weßels, A. Kohfeldt, E. Luvsandamdin, M. Schiemangk, C. Grzeschik, A. Peters, “High-power, micro-integrated diode laser modules at 767 and M. Krutzik, A. Wicht, A. Peters, S. Herrmann and C. Lämmerzahl, 780 nm for portable quantum gas experiments,” Appl. Opt., Vol. 54, pp. “Miniaturized Lab System for Future Cold Atom Experiments in 5332–5338, (2015). Microgravity,” Microgravity Sci. Technol, Vol. 29, pp. 37-48, (2016). 49. C. Kürbis, et al., "Extended cavity diode laser master-oscillator-power- 29. H. Müntinga, , H. Ahlers, M. Krutzik, A. Wenzlawski, A. Arnold, D. amplifier for precision iodine spectroscopy in space," to be published. Becker, K. Bongs, H. Dittus, H. Duncker, N. Gaaloul, C. Gherasim, E. 50. DLR. "Successful completion of the TEXUS 54/55 dual rocket mission", Giese, C, Grzeschik, T.W. Hänsch, O. Hellmig, W. Herr, S. Herrmann, E. https://www.dlr.de/dlr/en/desktopdefault.aspx/tabid-10081/151_read- Kajari, S. Kleinert, C. Lämmerzahl, W. Lewoczko-Adamczyk, J. Malcolm, 27969/year-all/#/gallery/30739 (posted 05 June 2018). N. Meyer, R. Nolte, A. Peters, M. Popp, J. Reichel, A. Roura, J. Rudolph, 51. G. Di Domenico, S. Schilt, and P. Thomann, “Simple approach to the M. Schiemangk, M. Schneider, S.T. Seidel, K. Sengstock, V. Tamma, relation between laser frequency noise and laser line shape,” Appl. Opt. T. Valenzuela, A. Vogel, R. Walser, T. Wendrich, P. Windpassinger, 49, 4801–4807, (2010). W. Zeller, T. van Zoest, W. Ertmer, W.P. Schleich and E.M. Rasel, “Interferometry with Bose-Einstein Condensates in Microgravity,” 52. S. Bourdeauducq, , R. Jördens, P. Zotov, J. Britton, D. Slichter, D. Physical Review Letters, Vol. 110, p. 093602, (2013). Leibrandt and J. Bohnet. ARTIQ 1.0 (Version 1.0). Zenodo. http://doi. org/10.5281/zenodo.51303, (posted 2016, May 10). 30. P. Bouyer, G. Varoquaux, R. Nyman, and J.-P. Brantut, “ICE: atom interferometry in microgravity for testing the universality of free fall,” in 53. https://sinara-hw.github.io (accessed 11th October 2017). 37th COSPAR Scientific Assembly, Montréal, Canada, 13–20 July (2008). 54. E. Perego, M. Pomponio, A. Detti, L. Duca, C. Sias and C.E. Calosso, “A 31. E.R. Elliott, M.C. Krutzik, J.R. Williams, R.J. Thompson and D.C. scalable hardware and software control apparatus for experiments with Aveline, “NASA’s Cold Atom Lab (CAL): system development and hybrid quantum systems,” Review of Scientific Instruments, Vol. 89, No. ground test status,” NPJ Microgravity, Vol. 4, No. 16, (2018). 11, p. 113116 (2018). 32. NASA’s Cold Atom Laboratory mission homepage, http://coldatomlab. 55. B.M. Sparkes, H.M. Chrzanowski, D.P. Parrain, B.C. Buchler, P.K. Lam jpl.nasa.gov (accessed 11th October 2018). and T. Symul, “A scalable, self-analyzing digital locking system for use on quantum optics experiments,” Review of Scientific Instruments, Vol. 82, 33. D.K.L. Oi, A. Ling, J.A. Grieve, T. Jennewein, A.N. Dinkelaker and M. No. 7, p. 075113 (2011). Krutzik, “Nanosatellites for quantum science and technology,” Contemp. Phys., Vol. 58, pp. 25–52, (2017). 56. M. Mihm, K. Lampmann, A. Wenzlawski and P. Windpassinger, “Note: Simultaneous modulation transfer spectroscopy on transitions of 34. H.Takenaka, A. Carrasco-Casado, M. Fujiwara, M. Kitamura, M. multiple atomic species for compact laser frequency reference modules,” Sasaki, and M. Toyoshima, “Satellite-to-ground quantum-limited Review of Scientific Instruments, Vol. 89, No. 9, (2018). communication using a 50-kg-class microsatellite,” Nature Photonics, Vol. 11, pp. 502-508, (2017). 57. M.F. Barschke, K. Gordon, M. Lehmann, and K. Brieß, “The TechnoSat mission for on-orbit technology demonstration,” 65th German Aerospace 35. E. Buchen, “Small satellite market observations,” Proceedings of the 29th Congress, Braunschweig, Germany, (2016). AIAA/USU Conference on Small Satellites. Logan, USA, (2015). 58. M.F. Barschke, P. Werner, K. Gordon, M. Lehmann, W. Frese, D. Noack, 36. M.F. Barschke, K. Brieß, and U. Renner, “Twenty-five years of satellite L. Grunwaldt, G. Kirchner, P. Wang, and B. Schlepp, in Proceedings of the development at Technische Universität Berlin,” Small Satellites Systems 31st AIAA/USU Conference on Small Satellites, Logan, USA, (2018). and Services Symposium, Valletta, Malta, (2016). 59. K. Gordon, M.F. Barschke, and P. Werner, “Upgrading TUBiX20 – 37. M.F. Barschke, and K. Gordon, “A generic systems architecture for bringing TechnoSat flight experience into the TUBIN mission,” Small a single failure tolerant nanosatellite platform,” 65th International Satellites Systems and Services Symposium, Sorrento, Italy, (2018). Astronautical Congress, Toronto, Canada, (2014). 60. D. Noack, M.F. Barschke, P. Werner, and K. Brieß, “FDA in space – 38. M.F. Barschke, A.N. Dinkelaker, J. Bartholomäus, P. Werner, H. First in-orbit results of a fluid dynamic attitude control system,” Small Christopher, and M.Krutzik, “Optical Quantum Technology in Space Satellites Systems and Services Symposium, Sorrento, Italy, (2018). using Small Satellites,” 68th International Astronautical Congress, 61. P. Wang, H. Almer, G., Kirchner, F. Koidl, M. Steindorfer, M.F. Barschke, Adelaide, Australia, (2017). P. Werner, and M. Starke, “kHz SLR Application on the Attitude Analysis 39. K.W. Martin, G. Phelps, N.D. Lemke, M.S. Bigelow, B. Stuhl, M. Wojcik, of TechnoSat,” 21st International Workshop on Laser Ranging, Canberra, M. Holt, I. Coddington, M.W. Bishop and J.H. Burke, “Compact Optical Australia, (2018). Atomic Clock Based on a Two-Photon Transition in Rubidium,” Physical 62. M.F. Barschke, J. Bartholomäus, , J.M. Crespo, K. Gordon, C. Jonglez, P. Review Applied, Vol. 9, Iss. 1, 014019, p. 014019, (2018). von Keiser, D. Költzsch, J. Leglise, M. Lehmann, C. Meumann, S. Reinert, 40. European Code of Conduct for Space Debris Mitigation, Iss. 1.0, (2004). S. Rotter, M. Starke, P. Werner, and L. Zander, “TechnoSat - Results from 41. F. Nez, F. Biraben, R. Felder and Y. Millerioux, “Optical frequency the first 18 months of operation,” 12th IAA Symposium on Small Satellites determination of the hyperfine components of the 5S1/2-5D3/2 two- for Earth Observation, Berlin, Germany, (2019). photon transitions in rubidium,” Optics Communications, Vol. 102, (1993). 63. M.F. Barschke, J. Bartholomäus, K. Gordon, M. Lehmann, and K. Brieß, 42. O. Terra, and H. Hussein, “An ultra-stable optical frequency standard for “The TUBIN mission for wildfire detection using nanosatellites,” CEAS telecommunication purposes based upon the 5S1/2 → 5D5/2 two-photon Space Journal, Vol. 9, Iss. 2, pp. 183-194, (2017). transition in rubidium,” Applied Physics B, Vol. 122, (2016). 64. J. Bartholomäus, M.F. Barschke, and M. Lehmann, “The TUBIN mission 43. Y. Millerioux, D. Touahri, L. Hilico, A. Clairon, R. Felder, F. Biraben within the context of present and future satellite-based fire detection and B. de Beauvoir, “Towards an Accurate Frequency Standard at λ systems,” 12th IAA Symposium on Small Satellites for Earth Observation, = 778 nm using a Laser Diode Stabilized on a Hyperfine Component Berlin, Germany, (2019).

Received 27 February 2019 Approved 6 March 2019

82 Vol 72 No.3 March 2019 JBIS JBIS VOLUME 72 2019 PAGES 83-89

MAGNETIC SHIELDING for Interplanetary Travel

M.W. SAILER1 & H.M. DOSS2 1Matthew Sailer 1602 Governors Dr, APT 1627, Pensacola, Fl 32514 USA; 2Point Loma Nazarene University, Physics & Engineering Department, 3900 Lomaland Dr., San Diego, CA 92106 USA. email [email protected] / [email protected]

A proposed design for radiation shielding in interplanetary travel is presented with primary shielding created by a super conducting split toroid magnetic field and unconfined magnetic fields created by two deployable superconducting loops. The split toroid’s shielding effectiveness is analyzed with calculations of mass, field, particle path, and synchrotron radiation provided. The creation of plasma regions is also taken into consideration. The design has a wire mass of 8.08×104 kg, which creates a 0.47 T shielding field, and a field less than 2.17×10-4 T in the crew area. Calculations show the shielding field deflects associated with particles of solar wind, solar flares, and high energy, high atomic number particles found in galactic cosmic radiation. This design might also be used to generate power and thrust from the radiation to create a potentially self-sufficient mobile station.

Keywords: Radiation Shielding, Interplanetary Travel, Magnetic Shielding

1 INTRODUCTION

Interplanetary travel requires radiation shielding to protect astronauts and electronics. The radiation of most concern is that from the solar wind and solar flares, which produce so- lar energetic particles (SEP) and the radiation produced in galactic events such as supernovae known as galactic cosmic radiation (GCR). SEP radiation consists mainly of protons and electrons. GCR consists of ionized H, He, as well as high atomic number energetic particles (HZE) such as Fe+26, O+8, and C+6[1]. To shield these particles both their maximum en- ergy and the energy of maximum flux need to be taken into consideration [2], as well as secondary radiation that might be produced during shielding.

Passive methods that have been considered include mass shielding [3], and active methods considered include creat- ing plasmas [4], electric fields[5], and magnetic fields [6-11] surrounding the system. No one of these systems seems to be Fig.1 Crew Area Dimensions. enough to sufficiently reduce radiation exposure. Ongoing research in new passive materials such as BNNT [3] is very promising as is the application of superconducting materials is a cylindrical shape with a 10 m radius, and a 10 m height as and new designs for fields [12-13]. seen in Fig. 1. Its surface is constructed from BNNT to mini- mize any radiation that might make it through the fields and This work proposes a combination of both passive mass any produced secondary radiation from reaching the crew shielding and active magnetic shielding with consideration of area. The thickness of this BNNT wall can be adjusted based trapped plasma as a design. The focus of this paper is on the on shielding effectiveness of the material and weight, but it is magnetic field created by the design and calculations of particle shown as 1m in Fig. 1. trajectories through the field with consideration of the second- ary radiation the particles produce. Superconducting wire is necessary to sustain a B-field dur- ing interplanetary travel. Ti-MgB2 superconducting tape has a 2 PROPOSED DESIGN high current to mass ratio as Musenich et al. and Buzea and Yamashita demonstrate [12-13]. Tables 1-2 show the limits of A shielding method consisting of a split toroid wire configu- the Ti-MgB2 superconducting tape according to both articles. ration, two deployable wire loops, and passive shielding using Boron Nitride Nanotubes (BNNT) is proposed. The crew area Tables 1-2 have data differing by about one order of mag-

JBIS Vol 72 No.3 March 2019 83 M.W. SAILER, H.M. DOSS

TABLE 1 Ti-MgB2 Musenich et al data 2015* B-field 0.25 T 0.5 T 1.0 T Max Current 885 A 788 A 594 A Temperature 16 K 16 K 16 K * calculated from Ti-MgB2 tape – improved process

TABLE 2 MgB2 Buzea and Yamashita data 2001 B-field 0.25 T 0.5 T 1.0 T Max Current 8600 A 7200 A 5800 A Temperature 25 K 25 K 25 K

Fig.2 Ti-MgB2 Split Toroid B-field surrounding the crew area. nitude. This difference comes from the titanium sheath used by Musenichet al., which allows the avoidance of helium cry- ogenics and maintains stability of the tape [12]. Table 2 shows the potential MgB2 superconductors have as wire for B-fields as radiation shields in space. Until more work is done, this paper will assume the values from the data of Musenich et al. as the limits for MgB2 superconductors. TheTi-MgB2 superconduct- ing tape will be referred to as a wire for the rest of this paper for simplicity.

A split toroid made from Ti-MgB2 superconducting wire creates a roughly uniform, 5 m thick, B-field of 0.47 T that sur- rounds the crew area. The inner loops have a radius of 10 m while the outer loops have a 15 m radius as shown in Fig. 2.

The superconducting wires of the split toroid carry a current Fig.3 Split Toroid with Adjustable Caps. of 800 A at a temperature of 16 K, with a wire density of 600 wires per meter extending vertically 10 meters, surrounding the crew area. A wire density of this magnitude will have sig- at the top and bottom of the split toroid with the same cur- nificant wire repulsive forces associated with the toroidal loops rent and wire density making the split toroid a nearly confined and will need to be optimized in a final design by adjusting B-field, however, their shape is adjustable as shown in Fig. 3. the distance between each wire. Toroidal caps deflect particles A deployable wire loop of radius 100 m, current 800 A, and no = 90 superconducting loops of wires carrying a total com- bined current of noI = 72,000 A surrounds the split toroid and produces a B-field with similar magnitude to that of Earth. As demonstrated by the SR2S project [10], a plasma of trapped particles forms surrounding the deployable loops similar to that of the Van Allen belts, Fig. 4.

A smaller wire loop of 15 m radius, current 800 A, and ni = 20 superconducting wires loops carrying a total combined current of niI = 10,800 A will cancel out the B-field in the crew area as seen in Fig. 5.

The adjustable toroidal caps have potential of producing thrust from redirected radiation spiraling down at the top and bottom of the crew area as shown in Fig. 6. The direction of Fig.4 Induced Plasma from B-field. thrust can be adjusted by manipulating the shape of the caps to change the direction of deflection of the particles.

Produced synchrotron radiation may be harnessed for ener- gy with the incorporation of photovoltaic cells [14].

While there are many parts to this design, this paper focuses on the shielding effectiveness of theB -field created by the split toroid and the two deployable wire loops.

3 CALCULATIONS

Fig.5 Two Deployable Wire Loops of Opposite Current (in red). Before any simulations were run, highly energetic “test parti-

84 Vol 72 No.3 March 2019 JBIS MAGNETIC SHIELDING for Interplanetary Travel

(2)

(3)

(4)

(5)

The total energy lost per particle due to synchrotron radia- tion was calculated by summing over the total time the power radiated multiplied by the time step. Fig.6 Particle Redirection at Caps. To calculate the B-field, a 200 × 200 × 200 matrix was cre- ated with each vector representing one square meter. A mini- cles” were chosen as a goal for the shielding method to deflect. mum toroidal radius of 10 m and a maximum radius of 15 m Based on Miller and Zeitlin’s data, particles with twice the en- was input as the dimensions of the split toroid to create the ergy of the most probable GCR particles were chosen. These toroid seen in Fig. 2. A current of 800 A was sent through the particles were Fe+26, C+6, He+2, and H+ with kinetic energies of toroidal loops using 6000 layers of wire over a vertical distance 84 GeV, 18 GeV, 12 GeV, and 1.5 GeV respectively [2]. of 10 m to give an estimate of the field produced. A deployable loop of 90 wires was added with a radius of 100 m. A second A B-field simulation was created using MATLAB coding deployable loop of opposite current and a radius of 15 m was with the Biot-Savart law in Eq. 1. added to cancel out the B-field in the center from 20 wires cre- ating the configuration seen in Fig. 5.

(1) Fe+26, C+6, He+2, H+, and e- particles were sent into the B-field in groups of ten of each particle type from the bottom left cor- ner at the matrix points [1 1 101] to [10 1 101]. The particle’s The following information was input into the code: Matrix velocity is directed towards the center of the crew area to max- dimensions, split toroid dimensions and specifications, and imize the amount of B-field traversed. Due to the unconfined deployable loop currents and dimensions. The code outputs a toroidal B-field in the simulation, the particles are initially 3-dimensional vector field using matrices. TheB -field vector redirected to a tangential path along the toroid and some are was calculated, from each wire in the split toroid and the de- pulled into the toroid as grazing particles. In a final design, this ployable loops, at each point in the matrix. effect will not occur to this extent due to the toroid being a nearly confined field. Different kinetic energies were input- un A second MATLAB code was used to send virtual particles til a maximum kinetic energy for a grazing particle was found into the B-field from various directions and observe their path. that would not enter the crew area. The values of synchrotron This code has inputs of particle mass, charge, initial kinetic radiation produced were also calculated. energy, and the B-field matrix created by the toroid and de- ployable loops. The relativistic particle’s path is plotted using The B-field was then multiplied by 1/4, 1/2, 2, and 4 to cal- time steps which calculate acceleration, new velocity, new po- culate its shielding effectiveness for a total of 5 different config- sition, produced synchrotron radiation, and change in kinetic urations. These values will be called B-field Multipliers in this energy before moving onto the next time step and continues in paper. The corresponding mass was also calculated for each this fashion until the particle leaves the vector field. We used a B-field Multiplier. The shielding effectiveness of eachB -field timestep of 0.1 m/|v| where |v| is the magnitude of the initial Multiplier is calculated in the following section. velocity of the time step which gives a reasonable estimate of the particle’s path. The particle’s path is calculated based on the To calculate the total mass of the wires, the dimensions of value of the nearest B-field vector to provide an accurate trajec- the Ti-MgB2 superconductors were used and a mass per length tory as the particle travels through the field. The code outputs of 0.037 kg/m [12]. The total length of wire of the deployable the total produced synchrotron radiation in electron volts, and loops and split toroid was added up to calculate a wire mass visual plots of the particle’s trajectory, the B-field, and the toroi- of 4.15 × 104 kg. The total mass, including an estimate of the dal and deployable wires. toroidal caps, is calculated to be 8.08 × 104 kg by doubling the mass of the toroid wires. Calculations performed to update variables are given by Eq. 2 for acceleration , Eq. 3 for velocity , Eq. 4 for position , 4 RESULTS and Eq. 5 for the synchrotron radiation power P. In what fol- lows, m represents the rest mass of the particle, t the time that The resulting B-field from the deployable loops is similar in has passed, c the speed of light in vacuum, q the charge of the magnitude to Earth’s geomagnetic field, but spread out over particle, and a smaller volume. This was calculated based on representing Earth’s outer core with the larger deployable wire loop and choosing a current to produce the same B-field magnitude at a distance of 220 m from the center of the loops representing

JBIS Vol 72 No.3 March 2019 85 M.W. SAILER, H.M. DOSS

Fig.7 Deployable wire loops creating B-field Fig.8 Center of B-field from deployable Fig.9 Center of B-field from deployable of similar magnitude to Earth. loops (side view). loops (top view).

Fig.10 Split Toroid and Deployable Loop Fig.11 Split Toroid B-field (side view). Fig.12 Split Toroid B-field (top view). B-field (side view).

Earth’s surface. The resulting field is shown in Figs. 7-9. tion of caps and added to the deployable loop’s B-field. Because the caps were not added to this calculation, the resulting B-field The maximum B-field produced by the deployable loops in- for the toroid is not fully confined. The additional field in the side the crew area’s 10 m radius is 2.17×10-4 T with the center crew area, created from the split toroid, is neglected in the total not exceeding 8.63×10-5 T. These are acceptable values accord- B-field of the crew area since the final design will be nearly fully ing to the International Standards of static B-field prolonged confined. This combined B-field can be seen in Figs 10-12. exposure from the World Health Organization[15]. Particles in groups of ten were sent into this field from the The split toroid B-field was then calculated without the addi- bottom left corner from 1-10 m on the x-axis as previously

Fig.13 Group of ten 113.8 GeV Fe+26 grazing particles traveling Fig.14 Group of ten 113.8 GeV Fe+26 grazing particles traveling through 1× created B-field (top view). through 1× created B-field at (side view)

86 Vol 72 No.3 March 2019 JBIS MAGNETIC SHIELDING for Interplanetary Travel

Fig.15 A single 113.8 GeV Fe+26 particle traveling through 1× Fig.16 Group of ten 113.8 GeVFe+26 grazing particles traveling created B-field (top view). Particle enters at x = 10 m and leaves through 1× created B-field (top view). around x = 39 m. described, and their paths were plotted in red as seen in Figs. of 1 is no change in the magnitude of the vectors from the initial 13-14. Note that MATLAB has a tendency to connect the final parameters. A B-field Multiplier of 2 is twice the magnitude of positions of each particle with a straight line to each other. This each vector in the B-field created from the necessary wires. MATLAB artifact does not obscure the path of the particles. When only one particle is sent through, no extra lines appear In Table 4, “Max KE Fe+26” is the maximum grazing particle to create a clearer picture of the particle’s path as seen in Fig. 15. kinetic energy in GeV that the B-field for the given multiplier can deflect. The Synchrotron Radiation is the total produced Particles grazing the outer deployable loop were simulated synchrotron radiation during the particles' path. to demonstrate that they are also shielded as shown in Fig. 16. TABLE 5 Data for C+6 The paths of Fe+26, C+6, He+2, H+, and e- particles for different B-field multipliers strength as described earlier were calculated. B-field Multiplier Max KE C+6 (GeV) Synchrotron Radiation (eV) The results are shown in Tables 3-8. 0.25 12.0 4.99×10-8 0.5 34.5 1.45×10-7 In Table 3, the B-field Multiplier is the number multiplied to 1 116 7.58×10-5 each vector in the created field. For instance, aB -field Multiplier 2 229 2.80×10-4 4 504 2.00×10-2 TABLE 3 The integer multiples of the B-field with corresponding wire mass and wire length at 16 K TABLE 6 Data for H+2 B-field Mass of Wires Length of wire Critical Current Multiplier (kg) (km) (A) B-field Multiplier Max KE H+2 (GeV) Synchrotron Radiation (eV) 0.25 1.73×104 5.45×102 936 0.25 0.981 3.99×10-10 0.5 3.63×104 1.09×103 890 0.5 2.92 9.87×10-9 1 8.08×104 2.18×103 800 1 7.90 4.32×10-7 2 2.10×105 4.36×103 617 2 17.9 3.10×10-6 4 1.02×106 8.72×103 253 4 39.0 1.79×10-4

TABLE 7 Data for H+ TABLE 4 Data for Fe+26, including the particle’s kinetic + energy, energy of synchrotron radiation produced B-field Multiplier Max KE H (GeV) Synchrotron Radiation (eV) 0.25 0.690 3.04×10-9 B-field Max KE Fe+26 Synchrotron Plasma Effect Multiplier (GeV) Radiation (eV) Percentage 0.5 1.96 9.75×10-8 Needed* 1 4.47 7.91×10-7 0.25 12.0 4.99×10-8 85.7% 2 9.76 4.67×10-5 0.5 34.5 1.45×10-7 58.9% 4 22.1 2.35×10-2 1 116 7.58×10-5 none

-4 2 229 2.80×10 none TABLE 8 Data for e- 4 504 2.00×10-2 none B-field Multiplier Max KE e- (GeV) Synchrotron Radiation (eV) *percentage of the test particle’s kinetic energy that would require some 0.25 1000+ 1000×109 other method, such as plasma considerations, to deter the particle

JBIS Vol 72 No.3 March 2019 87 M.W. SAILER, H.M. DOSS

An accurate idea of how much plasma is induced from the deployable loops is not achievable without real world experi- mentation. When a particle enters a plasma, its kinetic energy will change. The Plasma Effect Percentage Needed in Table 4 is the percentage decrease of the initial kinetic energy of the particle after it traverses the plasma. So, an 84 GeV Fe+26 parti- cle must decrease its kinetic energy to 34.5 GeV in the plasma, which is a 58.9% decrease from 84 GeV, to be fully shielded with a B-field Multiplier of 0.5. If no plasma is induced from the deployable loops, a Multiplier of one is required to fully shield the particle. This table gives an idea of the induced plas- ma’s shielding power required to stop the test particle for each B-field Multiplier.

C+6, He+2, H+, and e- particles were sent into the same B-field multipliers to find their maximum kinetic energies and pro- duced synchrotron radiation. The results are in Tables 5-8. Fig.17 Fe+26 particles with increasing energy interacting with a Electrons lose considerably more energy to synchrotron B-field Multiplier of 4". radiation when traveling at higher kinetic energy through a B-field. No electron kinetic energies were found that could may be deflected by a combination of plasma,B -field, and make it through the B-field. At 1000 GeV, the particle was still BNNT shielding. The produced synchrotron radiation may be shielded from the crew area since energy is lost quicker due to partially absorbed by photovoltaic cells and used as power for the γ4 term. Because of this, even with a B-field Multiplier of onboard equipment [14]. The rest will have to be deflected or 0.25, the crew area is a forbidden region for all electrons trav- absorbed by the BNNT structure housing the crew area [3]. eling through the toroidal field. This suggests a second plasma made of electrons may form inside the split toroid which would It is important to understand how high energy particles will cause unwanted heating of the wires and a greater power re- interact with this design when unshielded. Fig. 17 shows 5 Fe+26 quirement to keep the wires cool. This plasma, however, has a with energies increasing to and including 2800 GeV. potential to be used as a power source, but this requires further consideration. Some particles may make it through the split in Fig. 17 shows that some high energy particles are not shield- the toroid, but only if they are incident tangential to the toroid ed. Furthermore, substantial synchrotron radiation will be pro- at the slit. The orientation of the ship to the Sun will drastically duced and needs to be considered. These values are in Table 10. reduce the solar wind particles that make it through the slit. With the addition of a plasma surrounding the ship and The effectiveness of this shielding method can be seen when BNNT around the crew area, these very high energy particles compared to Miller and Zeitlin’s data [2]. Miller and Zeitlin may become shielded from the crew area. Knowing the char- show a graph of differential flux vs. kinetic energy per nucleon acteristics of the induced plasma and the shielding effective- of Fe+26, C+6, He+2, and H+ particles. Estimates from the graphs ness of BNNTs is essential to finding the limit of this method’s in Miller and Zeitlin’s paper for most probable particle and shielding ability. maximum kinetic energy are provided in Table 9. Additionally, a lightweight method to keep the wires un- Using the data from Miller and Zeitlin [2], the most proba- der their critical temperature will be required. The addition of ble kinetic energy GCR particles can be fully deflected using a well-designed sunshields may reduce the energy required to B-field Multiplier of 1. Additionally, Mewaldt[16] notes SEP- keep the wires below their critical temperature. proton kinetic energy has not exceeded much higher than 1 GeV during the past few major solar flares from 1956-1989. 5 CONCLUSION According to Cline and McDonald, the kinetic energy of elec- trons during solar flares are 3-12 MeV [17]. Even with a B-field The simulation shows significant shielding effectiveness of this multiplier of 0.5, the most probable kinetic energies of GCR design. A wire mass of 8.08 × 104 kg is a reasonable mass con- particle that was simulated can be fully deflected, along with sidering this is less than the mass of the ISS. Although the mass every observed proton and electron from recent solar flares. of the cooling system required to keep the Ti-MgB2 wires under There will be some particles that cannot be deflected, but they their critical temperature is not included in this calculation, it is have significantly less flux than the average GCR particle and important to note the potential of producing thrust from redi-

TABLE 9 Estimates of the probable kinetic energies and TABLE 10 Synchrotron radiation from Fe+26 particles maximum kinetic energies traveling through a B-field Multiplier of 4 Particle Most probable KE Maximum KE Fe+26 KE (GeV) Synchrotron Radiation (eV)

-4 Fe+26 42 GeV 2800 GeV 560 1.82×10 1120 1.26×10-2 C+6 9.0 GeV 12,000 GeV 1680 8.59×10-1 He+2 6.0 GeV 4000 GeV 2240 6.12 + H 0.75 GeV 5000 GeV 2800 6.65

88 Vol 72 No.3 March 2019 JBIS MAGNETIC SHIELDING for Interplanetary Travel rected radiation may make this mass acceptable. A space station These results are promising, but more considerations need similar to the ISS could be designed to move between orbits of to be made. The induced plasma and its added shielding must Earth, the Moon, Mars, or many other bodies using this shield- be studied further. Since calculations of induced plasma are ing method without the need for chemical propellants by using roughly estimated based on knowledge of the plasma trapped the momentum from redirected radiation. The induced electron in Earth’s magnetic field, a real-world experiment may be need- plasma fields and photovoltaic cells have the potential of pow- ed to get an idea of the plasma’s characteristics. This could be ering the systems onboard the ship making it a self-sufficient accomplished using a CubeSat with a deployable wire loop. mobile station. Its cylindrical crew area also has potential to be Considerations must be made of the time required for the plas- spun simulating a gravitational field and so reduces problems ma to form in the solar wind, the heating of the wires due to arising from prolonged exposure to the absence of gravity. This the plasma and synchrotron radiation, the produced neutrons Magnetically Shielded Self-Sufficient Space Station (M5S) holds from the collision of radiation with any mass on the ship, and potential to be a more practical method for interplanetary travel. the effectiveness of BNNT as a passive shielding method.

REFERENCES 1. R.A. Mewaldt, “Cosmic Rays”, in Macmillan Encyclopedia of Physics, Transactions on Applied Superconductivity, 14, 2, 2004. Simon & Schuster Macmillan, New York, 1996. 10. R.A. Bamford, B. Kellett, J. Bradford, T.N. Todd, M.G. Benton, R. 2. J. Miller & C. Zeitlin, “Twenty years of space radiation physics at the Stanord-Allen and R. Bingham, “An Exploration of the Effectiveness of BNL AGS and NASA Space Radiation Laboratory”, in Life Sciences in Artificial Mini-magnetospheres as a Potential Solar System Shelter for Space Research, 9, 12-18, 2016. Long Term Human Space Missions”, Acta Astronautica, 105, 2, 2014. 3. S.A. Thibeault, C.C. Fay, S.E. Lowther, K.D. Earle, G. Sauti, H.J. Kang, 11. R. Bruce and B. Baudouy, “Cryogenic design of a large superconducting and C. Park, “Radiation Shielding Materials Containing Hydrogen, magnet for astro-particle shielding on deep space travel missions”, Boron, and Nitrogen: Systematic Computational and Experimental Physics Procedia, 67, 264-269, 2015. Study - Phase I”, NASA Langley Research Center, NIAC Final Report, 12. R. Musenich, D. Nardelli, S. Brisigotti, D. Pietranera, M. Tropeano, A. 2012. Tumino and G. Grasso,“Ti-MgB2 Conductor for Superconducting Space 4. G.S. Janes, and R.H. Levy, “Plasma Radiation Shielding”, AIAA Journal, Magnets,” IEEE Transactions on Applied Superconductivity, 26, 4, 2015. 2. 10. pp. 1835-1838, 1964. 13. C. Buzea and T. Yamashita, “Review of superconducting properties of 5. L.W. Townsend, “Critical Analysis of Active Shielding Methods for MgB2”, Superconductors, Science & Technology, 14, 11, 2001. Space Radiation Protection”, IEEAC, Paper 1094, 2005. 14. J. Hirota, K. Tarusawa, K. Kudo and M. Uchida,“Proposal for Electric 6. R.H. Levy,“Radiation Shielding of Space Vehicles by Means of Power Generation by Using X-Rays and Gamma Rays”,Journal of Superconducting Coils”, Research Report 106, Contract No. AF 04(647)- Nuclear Science and Technology, 48,1, 103-107, 2011. 278,1961. 15. World Health Organization. (2006, March). “Electromagnetic Fields 7. F.H. Cocks. “A Deployable High Temperature Superconducting Coil and Public Health”. Retrieved from http://www.who.int/peh-emf/ (DHTSC): A Novel Concept for Producing Magnetic Shields Against publications/facts/fs299/en/ (Last Accessed 15th October 2018). Both Solar Flare and Galactic Radiation During Manned Interplanetary 16. R.A. Mewaldt, C.M.S. Cohen, A.W. Labrador, R.A. Leske, G.M. Mason, Missions”, JBIS, 44, pp. 99-102, 1991. M.I. Desai, M.D. Looper, J.E. Mazur, R.S. Selesnick and D.K. Haggerty, 8. S.G. Shepherd, &B.T. Kress, “Stormer theory applied to “Proton, helium, and electron spectra during the large solar particle magnetic spacecraft shielding”, Space Weather, 5, S04001, 2007. events of October-November 2003”, Journal of Geophysical Research 110: doi:10.1029/2006SW000273 A09S18, 2005. 9. L. Rossi, M. Sorbi, & P. Spillantini, “A Superconducting Magnetic Lens 17. T.L. Cline and F.B. McDonald, “Relativistic Electrons from Solar Flares,” for Solar Rays Protection in Manned Interplanetary Missions,” IEEE Solar Physics, 5, 4, pp.507-530, 1968.

Received 8 November 2018 Approved 11 April 2019

JBIS Vol 72 No.3 March 2019 89 JBIS VOLUME 71 2018 PAGES 90-101

ASPIRE – An Innovative Hypersonic Propulsion Paradigm

CHRISTOPHER MACLEOD, MATTHEW MURRAY, University of the Highlands and Islands, Lews Castle College Campus, Isle-of- Lewis, HS2 OXR, United Kingdom

Email [email protected]

This paper reports the results of a theoretical study into an innovative scramjet-like hypersonic propulsion system. The method, named ASPIRE (Air-breathing Supersonic Pellet Injection Rotary Engine), aims to overcome one of the main limitations of the traditional scramjet cycle – poor mixing and therefore inadequate combustion of air and fuel. The proposed system achieves this by redirecting the main airflow, injecting encapsulated or localised fuel throughout the engine-duct mixing volume and then switching the main airflow back on – this then engulfs and mixes with the fuel. The results presented here include design calculations, simulations and mathematical modelling. These show that the system performs well in simulation and gives excellent theoretical results for air-fuel mixing and airflow dynamics. A road-map to a working engine is also outlined.

Keywords: : Scramjet, Pulse detonation engine, ASPIRE, Hypersonic, Propulsion, Pellets, Air fuel mixing

1 INTRODUCTION ing summary section.

There are many formidable engineering challenges to over- 2 THE BASIC ENGINE CONCEPT, CYCLE AND come in order to obtain a working scramjet engine. These have PARAMETERS been catalogued extensively in the literature [1] - but perhaps the most pressing and difficult is that of achieving good air-fuel The original idea and its development is explained in previous mixing (and as a result, combustion) [2]. Achieving this is the papers – so only an overview will be covered here. An anima- main aim of the method discussed here. tion is also available on the internet which demonstrates the concept [6]. The basic idea is shown in figure 1. The system works by redirecting the airflow away from the main engine-duct and injecting the fuel in the form of encap- At the start of the cycle, the airflow is diverted away from the sulated gas, liquid droplets/capsules or a pelletized solid (or a main engine duct as shown in figure 1b). One way this can be mixture of these forms). The air is then redirected back into the accomplished is using a rotary nose-cone valve, as discussed in duct with the appropriate timing and engulfs the fuel, which the following sections. Next, fuel is injected or dropped into the vaporizes and so releases its load evenly throughout the mixing duct as shown in 1c), the key principle of the system is that the volume. The final stage is ignition and combustion. The engine fuel is in discrete packages – this may be in the form of solid-fu- concept has been named ASPIRE (Air-breathing Supersonic el pellets, liquid-fuel droplets, gas capsules or a mixture of these Pellet Injection Rotary Engine). The cycle is explained in the (these will all be referred to in the next sections as “pellets” for sections below. simplicity – unless it is necessary to discuss their actual nature). The injection sequence, dynamics and timing are arranged so The basis of this system has already been the subject of a that the fuel is evenly spaced in the duct volume prior to the peer-reviewed paper [3] and was also included in a review on next stage. The front valve is then opened as shown in 1d) and air-fuel mixing [4]. A layman’s explanation was carried in the air enters, surrounding and accelerating the injected pellets as magazine Analog [5] and it was featured in the magazine of the shown 1e). Because the pellets have a higher density than the Institution of Mechanical Engineers (Professional Engineer- airflow and are initially stationary, relative to that airflow, along ing), the New Statesman Magazine and by the BBC. The pur- the axis of the duct, their inertia means that the flow overtakes pose of this paper is to present further proof-of-concept results and engulfs them - but they are distributed correctly through- which show that the idea is definitively worth pursuing to the out the airflow volume. Finally, in 1f), the heat and friction of experimental verification stage. the impinging airflow heats the pellets and they vaporise, mix- ing their contents with the flow which then ignites. Both vapor- The paper outlines the basic cycle and engine structure in isation and ignition can be assisted by other technologies. the section below. It then discusses the dynamics of the re- leased fuel. Some alternative ideas for intake design are then A key advantage of this system is its flexibility. The pellets reviewed. Finally, some other applications are discussed and a can be a mixture of different types and sizes, with different -fu roadmap for further development outlined before a conclud- els (including hypergolic and pyrophoric types) – giving fine

90 Vol 72 No.3 March 2019 JBIS ASPIRE – An Innovative Hypersonic Propulsion Paradigm

Fig.1 The basic sequence of events during the engine cycle. control of fuel distribution and composition. They can be of a composite nature, with a gaseous, liquid and solid component (2) in a single unit or have, for example, a disruptive core which propels fuel outwards on activation. They can have a variety of coatings to control the timing of their load distribution or be disrupted by inductive or radio frequency heating. They can be The volume of fuel per capsule for an air volume of Va as- combined with more traditional scramjet continuous fuel in- suming n capsules is therefore: jection from duct edges or struts. Finally, they can carry loads into the flow other than combustible fuels – like electromag- (3) netically active materials for more advanced laser, microwave or magnetohydrodynamic propulsion concepts [4] and from this point of view the system outlined is an enabling technol- Or number of pellets required: ogy for these.

For a conventional combustion system, fuel calculations are (4) straightforward. The mass of any fuel required to completely burn in 1 m3 of air is: Expressions for volumes of common pellet shapes (for ex- (1) ample, spheres, cylinders, ellipsoids) can be found in any ap- propriate reference.

Where ρa is the air density under the selected flow conditions Because the system is very flexible there is a large latitude in and S is the stoichiometric ratio by mass (important examples how it can be set up. Pellet numbers per second required for 1 include: CH4 = 17.2, Petrol (Gasoline) = 14.7, H2 = 34). This MW of released combustion power, for two types of gaseous can be generalised for any volume by multiplying the result by capsule (H2 and CH4) at 1 bar internal pressure and one liquid that volume. The volume of fuel required for this mass is: type (paraffin or kerosene), are shown in Table 1.

TABLE 1 Number of fuel pellets per second required for one Megawatt of released combustion power

Pellet Radius 1 mm 2 mm 5 mm 1 cm

H2 gas at STP 18.86M 2.36M 150k 19k

CH4 gas at STP 5.97M 747k 48k 6k

Paraffin / Kerosene 6.45k 805 52 7 Notes on table: Spherical pellets, M stands for million (×106), k stands for thousand (×103)

JBIS Vol 72 No.3 March 2019 91 CHRISTOPHER MACLEOD, MATTHEW MURRAY

As can be seen from the table, liquid or solid forms of fuel To establish the definitive answer to this question a four- lead to far fewer required pellets (practical solid fuels have sim- pronged approach was adopted: 1) Simulating the pellet dy- ilar number requirements to liquids). This need not limit the namics using Computational Fluid Dynamics (CFD); 2) Using effectiveness of the combustion system when it includes cry- published measured values of supersonic drag to predict their ogenic and frozen fuel types. In a study of a similar system. embedding; 3) Modelling them using theoretically determined Bates [8] for example, suggests a cryogenic hydrogen slurry as drag values; 4) Finally, to calculate the “worst case” scenario being a useful fuel form. The information also suggests that a based on the assumption that all the available fluid energy practical system might include different fuels and pellet sizes as (kinetic, potential and thermal) is available as an accelerating already mentioned. pressure placed directly across the pellet. These approaches are considered one by one in the sections below. Obviously introducing the fuel into the mixing area involves some mechanical challenges. Liquid propellants would be the 3.1 Simulation using CFD easiest to handle, as they can be sprayed into the duct from an injection plate. Solid pellets and gas capsules can be injected in Of all the approaches discussed here, simulation using CFD is using pressurised gas (which can also form part of the fuel sys- likely to be the most accurate. This is because it uses the full tem), using the differential pressures generated by combustion Navier-Stokes model to capture a close approximation of fluid within the system or by other mechanical or electromagnetic behaviour in all the various flow regimes that the pellet pass- means. It has also been suggested that the pellets could be pro- es through. In the results presented below, the ANSYS Fluent jected ballistically into a continuous (unswitched) flow stream, commercial software package is used with a compressible flow but this has not been modelled fully yet. model (a strongly coupled energy equation). Chemical reac- tions (hypersonic modelling) are not included as the flow being Many parts of the system described above use established simulated in the mixing area is only mildly reacting at its upper and well-tested technology (for example the expansion duct velocity limit (Mach numbers of approximately 1.5 to 6.5) and and exhaust). Others (like the fuel delivery system) are chal- combustion is not being modelled. lenging, but nothing in them is outside standard mechanical engineering practice. However, there are two new key elements The pressure distribution is measured by splitting the sur- required to ensure that this system works. These are: a) That the face of the pellet into areas equal to the pressure divisions gen- pellet dynamics are such that the fuel will be well embedded erated by ANSYS and finding the pressure component in thex into the flow and that b) the concept of the switchable front (axial) direction by projecting this onto the y (cross-sectional) valve is feasible and achievable. The next sections investigate plane. The corresponding force may then be found by multi- these key enabling technologies. plying the average pressure in the area by that projection. The total force on the pellet is then the difference between front 3 PELLET DYNAMICS and back faces. Side (y-direction) forces were assumed to cancel due to the symmetrical nature of the pellets. The accel- The dynamics of the pellets in the engine duct air-stream are eration of the pellet as it speeds-up is assessed by simulating critical to the operation of the system. The problem can be the conditions at 100 m/s velocity intervals. As an example of posed as a question: Can it be shown that the pellets become the data obtained using the CFD calculations, Figure 2 below sufficiently embedded in the air-stream, so that their load is shows the total pressure on the surface of a pellet as it speeds well-distributed throughout the mixing volume? up in a Mach 5 free stream engine mixing-section (approx- imately Mach 2 in the section in question). To confirm the Some preliminary work was done on this question, and this results, full three-dimensional simulations of the pellets were was reported in our first paper [3] (which produced good re- also performed by an independent researcher, Figure 3 shows sults), but for a complete proof of concept (at least theoretical- examples of these simulations for pressure and pathlines ly) a more rigorous approach was needed. around a pellet.

3.2 Calculations using published values of supersonic drag

Knowing the measured simple (combined) coefficient of drag, one can use the drag equation to find the corresponding force produced:

(5)

Where FD is the drag force, pd is the dynamic pressure, CD is the drag coefficient and A is the frontal area.

Values for CD have been published in several older papers [9–11] and more recent re-evaluations [12] as well as govern- ment, NACA and NASA reports [13, 14]. The results reported below use the worst case (lowest drag) figures reported. Al- most all published values relate to stationary objects in a test stand (for example a wind or shock tunnel), rather than mov- ing with the flow – this might conceivably lead to some dis- crepancies as in one case the drag is retarding movement and Fig.2 Simulations showing total pressure at four airflow velocities, in the other causing it, however the results match up well with of an accelerating pellet in a Mach 5 (free-steam) ASPIRE engine. the CFD and other simulations. As with the other results, data

92 Vol 72 No.3 March 2019 JBIS ASPIRE – An Innovative Hypersonic Propulsion Paradigm

Fig.3 Examples of three-dimensional simulations used to confirm the basic results.

points were taken at velocity intervals of 100 m/s throughout Where T is the normal temperature of the flow and Cp is the the acceleration curve. constant pressure heat capacity. The maximum velocity achiev- able by such a flow in an isentropic system is: 3.3 Theoretically determined drag values (8) A number of expressions for supersonic wave drag have been produced over the years – for example by Smith [15], Jones And we can convert this into an equivalent pressure: [16], Haack [17] and Sears [18]. These adopt similar strategies and give similar results. The wave drag thus calculated can (9) be combined with known or estimated values of skin drag to produce an overall picture for simple symmetrical shapes like spheres and ellipses. Smith for example gives for an ellipse: As outlined in the previous sections, force can be calculated from this. Such an analysis could also be taken to its extreme (6) by adding the static pressure to the result – which would be equivalent to there being a vacuum behind the pellet. Such an extreme scenario is impossible in practice, but serves as an ex- Here vol is the volume of the object, A is the frontal area, l the ercise in reductio ad absurdum which is useful in proving the characteristic length, and K is a constant which can be calculat- point. ed from the geometry of the ellipse (see references for further details). This is a semi-empirical approach and gives similar val- As the pellet accelerates in the stream, its velocity relative to ues to those obtained using measured values of drag and CFD. the flow decreases. Figure 4 shows what might be expected as it speeds up. 3.4 Worst case scenarios To accommodate the constantly changing acceleration of In this case, we calculate the maximum possible acceleration, if the pellet, an iterative algorithm for calculating its velocity and all the available energy in the flow (kinetic and thermal) were displacement was used as shown in Figure 5. This was supplied converted into pressure and this were placed directly on the with the calculated or simulated accelerations from the four forward face of the pellet. methods outlined above.

The stagnation temperature of a flow can be calculated from:

(7)

Fig.4 Expected velocity profile for pellet. Fig.5 Iterative parameter calculation algorithm.

JBIS Vol 72 No.3 March 2019 93 CHRISTOPHER MACLEOD, MATTHEW MURRAY

TABLE 2 Parameters in a common engine model by Billig [7] Free stream Altitude (km) Mixing section Pressure ratio Mixing section Mixing section Mixing section Mixing section Mach number Mach number pressure (×105 Pa) temp (°C) temp (K) speed (m/s) 3 14.6 1.53 8 1.0 140 413 620 4 17.5 1.95 16 1.28 244 517 879 5 20.0 2.36 25 1.37 339 612 1158 6 22.3 2.77 35 1.35 437 710 1454 7 24.4 3.14 47 1.31 533 806 1755 10 29.1 4.14 90 1.23 815 1088 2665 15 34.8 5.50 186 1.10 1327 1600 4239 20 42.0 6.65 314 0.69 1990 2263 5989 27 54.3 7.69 473 0.22 2609 2882 8292

The parameter Δt was reduced until the results always con- radius, but volume and therefore pellet-mass increases as the verged on the same values – in practice a default value of 1 cube of radius. Therefore doubling the pellet radius increases μs was used in most cases as this was an order of magnitude frontal area by a factor of four, but mass by a factor of eight smaller than required. – and airflow penetration similarly scales (approximately dou- bles), ignoring the scaling-differences in skin and other drag A few simple extra lines can be added to this algorithm to components. calculate the progress of the main flow along the duct and its displacement relative to the pellet. The graph in Figure 6 shows the acceleration profile of a 2 mm radius pellet with a density of 850 kg/m3 (this is typical of To calculate the pellet, intake and engine parameters in this a heavier liquid fuel drop or a wax encapsulated light liquid research, a very typical standard engine-design used for refer- or gaseous fuel) in a Mach 5 (free-stream) engine. A similar ence calculations in many papers was used. This is the work acceleration graph can be produced for all the other situations published by Billig [7] and discussed by Anderson [19]. A table discussed. Only one graph is shown here out of the several doz- of this data is shown in Table 2 (converted from imperial units en from the experiments, since they all show the same trends. and reformatted). The most likely values of acceleration are the three lines in the middle with the solid lines representing the limiting values The graphs opposite are for spherical pellets. It was found (and the top – line A, being the absolute maximum theoretical- that, although irregularly shaped pellets exhibited different sta- ly possible). Line C terminates where reliable published figures bility and tumbling characteristics, they did not show much dif- for drag run out. ference in airflow penetration – this is almost certainly because the large Reynold’s number of the flow dominates the pellets Figure 7 shows the overall result for the whole flight-enve- behaviour. However, pellet size does make a difference - with lope for embedding of the same pellet in the air-stream. Here penetration increasing with size. This is because flow-presenta- the y - axis coordinate shows the distance into the air-stream tion area (and therefore force) increases as the square of pellet that the pellet has become embedded when the pellet itself

Fig.6 The acceleration profile of a typical 2 mm pellet in a Mach 5 free-stream engine.

94 Vol 72 No.3 March 2019 JBIS ASPIRE – An Innovative Hypersonic Propulsion Paradigm

Fig.7 Embedding of 2mm radius liquid or encapsulated pellets in flow at flight Mach numbers. has moved 10 cm from its original axial (x) position. It can be He has done extensive research, including experimental test- seen from this that even in the worst-case scenario the pellet ing, on the topic and particularly pointed out the advantages is embedded 35 cm into the airstream (in other words when of a solid/liquid slurry as a fuel (particularly studying He/H2 the pellet has moved 10 cm the leading edge of the airstream cryogenic mixtures). His calculations from the latent heat of has moved 35 cm past it). Two lines are shown dotted because the H2 slurries show that for small 1mm diameter pellets of this some parameters which they rely on are extrapolated from fuel only 1.2 J of heat was required for complete vaporisation. published data at lower speeds. In fact, even if we assumed that He also found that the following ratio is particularly important the entire initial (worst-case) pressure were across the pellet in the breakup process: throughout its whole acceleration profile, the pellet would still embed itself in the stream sufficiently well – at Mach 3 this cal- culation yields an embedding of around 30 cm. (10)

This exercise shows that the pellets will become completely engulfed in the airflow even in the most extreme scenarios and Here the subscript f refers to the fuel pellet and a to the that the first of the two major issues discussed in the last par- airflow. The ratio must be greater than six for rapid effective agraph of Section two is (at least theoretically) not going to be breakup and vaporisation and these characteristics can be con- a problem. trolled using this ratio and pellet size. Bates found, in his ex- periments and calculations – which used pellets of similar di- Before finishing the discussion on pellets, a few other impor- mensions to those already discussed – a complete vaporisation tant issues would benefit from some attention. Some of these distance of around 30cm at flow conditions similar to the ones were not elucidated clearly in the first papers on the method, already mentioned. Aside from this simple scenario however, others result from audience questions arising when the system fuel can be encased in a wide variety of refractory coatings or has been presented at seminars in other institutions, compa- shells – which allows breakup time to be tightly controlled. nies and universities. The current authors [20, 21] have also done extensive re- One important aspect of the system revolves around the search into the use of radio-frequency and microwave radia- mechanisms of pellet break-up and vaporisation. Fortunately tion in scramjet thermal systems including its use in aiding fuel these important parameters can be controlled closely in a num- release and mixing. Electromagnetically excitable material can ber of ways – adding greatly to the usefulness and flexibility be added to the pellets and this allows their disintegration to of the system. Controlling the disintegration time of the pellet be controlled by a duct mounted radiation source (inductive allows fine tuning of pellet penetration and therefore the whole heating could potentially be used in a similar way). combustion process. The use of pellets also allows innovative solutions to the igni- The mechanisms of breakup and fuel dispersion in simple tion and combustion stability issues which traditionally plague uncoated pellets has been studied and reported by Bates [8] in scramjet design as a second-order issue. Pellets may be loaded his work on a similar system (Bates suggested, in several pa- with hypergolic or pyrophoric fuels, combustion catalysts or pers, a system where pellets or strings of solid or semi-solid electromagnetically activated ignition agents which operate in fuel are injected into a continuous stream from the duct walls). a similar way to those mentioned in the previous paragraph.

JBIS Vol 72 No.3 March 2019 95 CHRISTOPHER MACLEOD, MATTHEW MURRAY

Fig.8 Basic concept of rotating nose-cone or rotor with duct valves.

This potentially allows an even flame-front to propagate across the combustion space – something difficult with conventional mechanisms.

Let us now consider the second point highlighted in the last paragraph of Section two – the design of the airflow switching mechanism.

4 INTAKE DESIGNS

Several options for switching the airflow into and away from the inner duct were discussed in the first paper [3]. These included using a system of shutters as a valve and a rotating wedge diverter and many other possibilities can be readily im- agined. However, simulation and calculation has shown that one particular design seems to work well and is also a particu- larly elegant solution – a rotating nose-cone or rotor. The prin- ciple of this is shown in Figure 8.

The nose-cone consists of a cone shaped external structure with standard design scramjet intake-ducts cut into it around its periphery. This structure continues internally within the body of the engine to form the isolator, mixing, combustion and expansion sections of the system as part of a rotor. Fig- ure 9 shows a simple configuration for the front (external) portion of the rotor in 3 dimensions which was used in CFD Fig.9 External part of rotor with stepped scramjet intakes. simulations.

Within this basic idea there are many different configura- - of which there are many published designs). Figure 10d shows tions possible. These include the number and shape of the in- a staggered design which is a half-way house between b and c, take ducts. Some alternatives for these are shown in Figure 10. but is easier to design than a gently curving structure. Figure

Figure 10a shows a curved, gently changing duct-profile, where the top surface of the intake is part of the engine cowl. The idea behind designs like these is that, as they turn they pro- vide a gradually changing profile to the incoming airflow and therefore avoid an abrupt disruption. However, their complex shape makes them difficult to design with traditional methods and it is planned to explore this using genetic algorithms at a later date. The intake compression surface may be a stepped topology as shown in Figure 9 or an isentropic one. Figure 10b shows a similar structure to 10a, but with the entire intake build into the rotor (and not using the engine cowl). This al- lows the top of the duct to be flat, making design simpler. Fig- Fig.10 Some of the many possible intake configurations, viewed ure 10c shows a very simple square duct (like that in Figure 9 from the front.

96 Vol 72 No.3 March 2019 JBIS ASPIRE – An Innovative Hypersonic Propulsion Paradigm

Fig.11 Cross-sections of some designs explored for external rotors.

11 shows the cross-section of a design for the forward part of ducts) volume. The power used by the system depends on the an intake similar to Figure 9 and 10c on the left and a modified speed of rotation of the nose-cone: staggered design on the right. (12) An alternative to having the whole intake rotate, is to keep the outer (hypersonic) compression surface stationary and have only the inner part (the lower-speed, supersonic section) of the surface rotate. Alternatively, the isolator section of the Where R is the rotational speed in r.p.m and ω is the angular engine could rotate. These options are shown in Figure 12. velocity in rad/s. With careful design these values are in the tens to hundreds of kilowatts for practical engine powers in the Similarly, the portion of the rotor internal to the engine has region of megawatts to tens of megawatts. many possible configurations including partitioned or contin- uous annular ducts. It should be noted, that although these de- In fact this analysis can also be extended to calculate the signs are presented here as axisymmetric, it is equally possible losses associated with a small rotation of the duct as air passes to create linear engine topologies using the same or somewhat through it, as shown in Figure 13. Here the velocity difference modified ideas. used in the equation represents the edge effect of the duct rotat- ing. It may be seen intuitively from this, that ratio of the speed The power required to turn the rotor can be calculated from of the duct movement to the speed of an air particle is impor- the aerodynamics of the situation. Consider an undiverted air- tant – if the speed of a particle through the duct is much greater flow represented by an average velocity vector va and also an- than the speed of the edge of the duct then the effect of duct other vector vb representing the diverted flow’s average velocity. rotation is minimised. This leads to the conclusion that, in this The magnitude of the difference between these two velocities respect anyway, many ducts around a relatively slow-moving represents the effect on the airflow of the switching mecha- rotor would perform well. nism. The energy required to produce this change is therefore (assuming a loss-less system): Having discussed some of the many structures possible, the aim of this section is simply to show that such a system is (11) theoretically plausible. In other words: Can a rotating ducted valve of this nature redirect the airflow in a manner consistent with the ASPIRE engine’s operation? This would show that the Where ρ is the average density of the flow and VD is the duct (or system is possible at the theoretical level – and serve as a basis

Fig.12 Intake cross sections showing low-speed rotating duct topologies.

JBIS Vol 72 No.3 March 2019 97 CHRISTOPHER MACLEOD, MATTHEW MURRAY

Fig.13 Effect of a small rotation of the duct. Fig.14 Stylised topology of duct. for practical verification and design. A simplified topology of the system is visualised in figure 14. In this diagram a is the duct width, its height is b and its Fortunately there is a good way to describe the effect of a depth c. The letter d designates the distance moved by the edge rotating intake duct on the airflow. As already stated above, if of the duct at velocity Vd during the time T that an air particle an incoming air-particle is moving much faster than the duct takes to transit the intake at velocity Va. The duct is shown as is rotating, then the duct appears nearly stationary to the par- rectangular because it is the conditions at its mouth that limit ticle (another way of looking at this is that the problem of flow its performance. through a moving duct is similar to a stationary duct on a yaw- ing vehicle). Existing scramjet theory provides a way of assessing We can see from this that T = c/Va and that the distance d this, as figures can be calculated which give the area constriction swept by the edge of the duct during the time taken for a parti- required to cause the duct to unstart or stall. This happens when cle to pass through is TVd. The (shaded) area swept by the duct the constriction in duct size is sufficient for the intake to dis- is therefore bd = bVdT = bVd(c/Va). From this and designating gorge a normal shock-wave. The theory of this is covered in Se- the ratio A0/AR as R, for the duct to be operational: gal [1] on pages 98 – 100 and in Heiser [22] on pages 250 – 251. In the case discussed here the constriction is caused by the inlet (15) area decreasing in effective size due to its rotation.

In this analysis several assumptions are made. These in- Or rearranging: clude that the duct is smooth, uses isentropic compression and that edge effects do not cause a shock sufficient to disrupt (16) the main airflow. Viscous effects (like the boundary layer) are also ignored. Segal provides several references which discuss these effects and reports a measured case where these oth- We can explore the effect of this by putting in some typical er phenomena caused an error of approximately 27% in the figures. An intake width to length ratio is around 0.06 for a long calculated figures. The maximum area constriction is calcu- system, so, for example at Mach 5, Vd must be less than 24.36 lated by first calculating the Mach number of the post nor- m/s. Now, as already stated, we probably do not wish to exceed mal-shock flow: 10% of this value, which is 2.4 m/s. In an engine with a 1m radi- us rotor this corresponds to an RPM of around 23. This figure, making the same assumptions, ranges from 10 RPM at Mach (13) 3 to 134 at Mach 20. Table 4 gives the duct values, making the same assumptions and rounded down to the nearest m/s for the Mach numbers given in Table 3. For practically sized en- Where M is the Mach number of the duct we are considering gines with several ducts this is perfectly compatible with the and γ is the ratio of specific heats. The restriction of area before fuel injection figures given in Table 1 – for example at Mach 5, unstart in an isentropic duct is then given by: for 1 MW of power, approximately 800 2mm pellets need to be deposited within the swept duct area, if there were four ducts on the nose-cone, this would be around 200 pellets per 2.4 m of swept area. (14) In many ways these figures are quite conservative. For exam- ple the output velocity of the intake has been chosen for the cal- Where A0 is the normal duct area and AR is the restricted area. culation above – in fact this only exists at its back-end and the Table 3 (opposite) gives the allowable constriction ratio over a modal average would be more accurate (and Heisler uses the range of operational Mach numbers. As can be seen there is an free-stream velocity - since the duct does not technically stall asymptotic progression towards a value of approximately 1.66. until the normal-shock appears at the front). The derivation also ignores the progression of the duct to the right (in Figure We can see from these figures that, as long as the allowa- 14) which sweeps out an area which is equal to that occluded ble apparent duct-area decrease caused by rotation is less than on the left – this would effectively half the area and double the around 28% at Mach 3 or 40% at Mach 25, during the time allowed duct speed. However, the aim of this analysis is to show taken for an air particle to transverse the intake, unstart should that the system is practical and so only the worst-case scenario not occur. Of course this is the limit, in a real system we would is described – but these points should be enough to compen- probably not wish to exceed 10% of this value. sate for ignoring the viscous effects already discussed.

98 Vol 72 No.3 March 2019 JBIS ASPIRE – An Innovative Hypersonic Propulsion Paradigm

TABLE 3 Area constrictions necessary for unstart in an isentropic duct Mach number 3 5 10 15 25 A0 /AR 1.39 1.54 1.63 1.65 1.66

TABLE 4 Maximum duct velocity values for an a/c ratio of 0.06 Mach number 3 5 10 15 25 Duct velocity Vd (m/s) 10 24 61 100 197

It is also worth pointing out at this juncture, that the above designed. Obviously this does not make for a practical engine, analysis gives some interesting pointers to the performance of but serves to prove the main points. the engine. For example: Both the intake and pellet calcula- tions show that the engine works better at higher speeds – at Several simulations were used to observe operation of the lower Mach numbers there is less pellet penetration and the duct – these included: a 3D simulation of the nose-cone sys- nose-cone must rotate more slowly (making injection timing tem using setup similar to that already shown in figure 9; 2D more critical). This could be overcome by using larger pellets sectional simulations where each section of the nose-cone was and more side-wall injection – and this is compatible with the simulated slice-by-slice as it turned and then animated to show lower airflow speed and better mixing at these speeds. flow in cross-section; finally moving-stream simulations were also attempted. Figure 15 shows one of the 2D cross-sections At the rotor speeds calculated above, transitive disruptive produced (half the cone shown) in the course of one of these edge effects are minimal. During the passage of an air particle, investigations. the edge sweeps out a wedge (it hasn’t moved when the parti- cle enters and has swept out a distance d when it leaves). This 5 DEVELOPMENT ROADMAP could cause a vertically orientated oblique shock emanating from the edge into the intake. However the wedge angle is very At this point there is little more theoretical proof-of-concept shallow and any shock therefore weak, close to the duct wall work that can be done on pellet dynamics – all our simulations and insufficient to disrupt the main flow. and modelling indicate strongly that these will operate as ex- pected. In the next stage of the project the system principles So that the dynamics of the ducts could be observed in more outlined above (pellet penetration) need to be confirmed ex- detail and some of the results discussed above could be con- perimentally. firmed, CFD simulations of the intake system were also under- taken. These used the same setup described in Section 3. Given Further theoretical work is desirable, though, to choose that the aim was a proof of principle, to simplify the problem between intake topologies and investigate the action of these a simple single speed duct was chosen. Mach 5 was selected more thoroughly. In particular the aerodynamics of the differ- as the free-stream speed because other published scramjet de- ent rotary duct types may be further illustrated with CFD sim- signs where available for comparison and this speed is achiev- ulations. It is intended to investigate this and publish the results able for testing on small sounding rockets. Only simple square in a follow-up paper. and staggered intakes were designed and simulated – as shown in figure 10c and 10d. Finally, and again to keep the design Practical confirmation of both operational aspects may be simple, stepped (two or three stage) compression ramps were done with two experiments. The first is to show that pellet -em

Fig.15 Simulation of Mach 5 open intake.

JBIS Vol 72 No.3 March 2019 99 CHRISTOPHER MACLEOD, MATTHEW MURRAY bedding is as expected. This can be achieved simply in a stand- ard shock-tunnel with a pellet dropping device and high-speed camera. The second experiment is to confirm the practical operation of the rotating nose-cone. It may be also possible to do this in a shock-tunnel – however only if the gas burst is of sufficient duration to observe the effect of rotation. One alter- native would be testing in a high-temperature, high-speed flow (the outlet of a rocket system), several organisations offer this, although obviously the flow parameters are different in such a stream. Another alternative is using a small sounding rocket for in-flight testing – both these latter alternatives would be more expensive than tunnel testing.

Once these two aspects are confirmed, a single-speed functional demonstrator launched on a small sounding rock- Fig.16 3D printed rendering of an example test design for the et would be a logical next step. Mach 4 or 5 are suitable and external nose-cone. achievable speeds for such a test (and even some amateur rock- et groups in the USA have achieved these velocities). described in the previous papers are outlined in this section.

Several possible designs for these options have already been It has already been mentioned that several other options devised. Figure 16 shows a 3D printed rendering of a test nose- for flow diversion are worth investigating (and many others cone. These designs can be machined from the same software are theoretically possible). One of these discussed above is files using CNC for actual testing. switching the flow after the hypersonic intake section, at the lower-speed compression or at the isolator stage. This may have Finally, a full demonstrator for use with a larger vehicle advantages, including less chance of flow instability and greater could be designed and tested. A road-map for developments is ease of implementing a variable topology inlet. The reverse of shown in Figure 17. this philosophy is to switch the air before the (stationary) in- take using a rotating plain or Busemann topology. A projected full-price cost for Phase A (the experimental validation of the key technologies discussed in this paper) Another area of interest is the adaption of the design to use would be in the region £100k - £200k at current costs. Phase B a variable cycle. For example, the nose-cone could be locked (testing a simple small design on a sounding rocket) would be in place using a clutch system and the internal ducts used in a around £250k. This is an important milestone, because at this pure rocket mode for final transition to the space environment. point all the key operating principles of the engine will have A more complex option for multispeed operation would be an been verified and tested. The cost of Phase C is difficult to es- internal concentric rotor within that already discussed, which timate as it requires much more serious development and test- would rotate slowly as the engine passes through its flight ing of hardware and the involvement of government or large envelop to expose a gradually changing compression surface industrial bodies. In addition to this, a sustained on-going through an open aperture in the outer rotor; the rotor assem- simulation and theoretical effort to refine current designs and bly could also change its position relative to the cowl through produce new ones would be in the region of £40k per annum. the flight envelop as implemented in the Pratt and Whitney J58 engine in the Lockheed SR71. 6 OTHER ADDITIONS AND APPLICATIONS The system described here may be used for innovative pro- Numerous other variations and additions to the basic design pulsion purposes other than fuel mixing - for example to facil- have been investigated. A few of the more important ones, not itate the introduction into the airflow [4] of:

Fig.17 Development roadmap.

100 Vol 72 No.3 March 2019 JBIS ASPIRE – An Innovative Hypersonic Propulsion Paradigm

• MHD active substances - perhaps to aid ionisation or in- This paper shows that the key untested parts of the system crease flow conductivity. are at least theoretically feasible. In fact the flexibility of the idea • E lectrically active substances - to aid inductive or similar is such that that there are a variety of possible topologies for electrical heating or acceleration. the whole system which afford many options for overcoming • Chemical catalysts to aid (for example) combustion. technical issues. It is intended to follow this paper with anoth- • Compounds which lase. er, analysing the intake section with more thoroughly through • EMA active substances. CFD and/or other appropriate modelling. There is also a clear road-map available between the current research and results 7 CONCLUSIONS and a full experimental confirmation of the engine’s operation. The route outlined would be very cost-effective, with compre- The aim of the ASPIRE engine is to overcome the main prob- hensive verification only costing in the region of £300k - £500k. lem with scramjet operation – poor air-fuel mixing. However, the technique also offers an unparalleled degree of flexibility in Acknowledgements the distribution and release of the fuel through the airflow. It is also an enabling technology for other advanced propulsion Thanks to Callum MacDonald, who undertook the confirma- systems because it can distribute a wide variety of substances tory three-dimensional pellet CFD simulations and Claire Ger- other than fuel into the flow. rard and Murdo Smith for helpful suggestions and proofreading.

REFERENCES 1. C. Segal, The Scramjet Engine, Cambridge University Press, Cambridge, 12. M. Parmar, A. Haselbacher, and S. Balachandar. “Improved Drag 2009. Correlation for Spheres and Application to Shock-Tube Experiments”, 2. J. P. Drummond, G. S. Diskin and A. D. Cutler, “Fuel-Air Mixing and AIAA Journal, Vol. 48, No. 6 (2010), pp. 1273-1276. Combustion in SCRAMJETS”, 38th AIAA/ASME/SAE/ASEE Joint 13. A. B. Bailey and J. Hiatt, “Free-Flight Measurements of Sphere Drag Propulsion Conference, Indianapolis, 7-10 July 2002. Paper AIAA - at Subsonic, Transonic, Supersonic, and Hypersonic Speeds for 2002 - 3878. Continuum, Transition, and Near-Free- Molecular Flow Conditions”, 3. C. MacLeod, N.F. Capanni and K.S Gow “Overcoming fuel-air mixing Defence technical information centre, 1971. issues with pulsed scramjets and pelletized fuel”, Journal of the British 14. H. Ashkenas and P. P. Wegener, Method of sphere-drag measurement in Interplanetary Society, 68(11), pp.354 - 362, 2015. supersonic gas flows using low-density wind tunnel, NASA, 1961. 4. C. MacLeod, C.E. Gerrard “A review of air-fuel mixing and alternative 15. J. H. B. Smith, Lift/Drag Ratios of Optimised Slewed Elliptic Wings at methods in scramjets and scramjet-like engines”, Journal of the British Supersonic Speeds, The Aeronautical Quarterly, Volume 12, Issue 3, Interplanetary Society, 69(4), pp.116 - 126, 2016. August 1961 , pp. 201-218. 5. C. MacLeod “The new space race: a case for innovation”, Analog science 16. Jones, R. T. Aerodynamic Design for Supersonic Speeds. Delivered to fiction and fact, 136(12), pp. 21 – 29. 2016. the First International Congress in the Aeronautical Sciences at Madrid 6. https://www.youtube.com/watch?v=NbQTqpmMoWM in September 1958. Published in Advances in Aeronautical Sciences, Vol. 1, Pergamon Press, London, 1959. 7. F.S. Billig, “Design and development of single stage to orbit vehicles, John-Hopkins Applied Physics Laboratory Technical Digest, 11 (3/4), 17. W. Haack, “GeschossformenkleinstenWellenwiderstandes, Bericht 139 pp.336-352, 1990. der Lilienthal-Gesellschaft fur Luftfahrt”. 8. S. C. Bates, “Assessment of Solid Hydrogen Slurry Fueling for an 18. Sears, William R.: On Projectiles of Minimum Wave Drag. Quart. Appl. Airbreathing Supersonic Combustor,” Journal of Propulsion and Power, Math., vol. IV, no. 4, Jan. 1947, pp. 361-366. 20(5), pp. 793 - 800, 2004. 19. J.D. Anderson, Fundamentals of aerodynamics (3rd ed), McGraw-Hill, 9. C. B. Henderson, “Drag Coefficients of Spheres in Continuum and New York, 2001. Rarefied Flows”, AIAA Journal, Vol. 14, No. 6 (1976), pp. 707-708. 20. C. MacLeod, “Electromagnetically Activated Hypersonic Ducts,” Journal 10. A. C. Charters. “The Aerodynamic Performance of Small Spheres from of the British Interplanetary Society, Vol. 62, No 3, 2009, pp. 99 - 109. Subsonic to High Supersonic Velocities”, Journal of the Aeronautical 21. C. MacLeod, K. S. Gow and N. F., “Some Practical Aspects of Sciences, Vol. 12, No. 4 (1945), pp. 468-476. Electromagnetic Activation,” Journal of the British Interplanetary Society, 11. A.B. Bailey and J. Hiatt. "Sphere Drag Coefficients for a Broad Range of Vol 62, No 9, 2009, pp. 320 - 331. Mach and Reynolds Numbers", AIAA Journal, Vol. 10, No. 11 (1972), 22. W. H Heiser et al, “Hypersonic Airbreathing Propulsion”, AIAA pp. 1436-1440. Education, Washington, 1994.

Received 29 August 2018 Approved 12 April 2019

JBIS Vol 72 No.3 March 2019 101 JBIS VOLUME 71 2018 PAGES 102-108

A REAL-TIME SPACE DEBRIS DETECTION SYSTEM for BIRALES

D. CUTAJAR1, A. MAGRO1, J. BORG1, K. ZARB ADAMI1,2, G. BIANCHI3, C. BORTOLOTTI3, A. CATTANI3, F. FIOCCHI3, L. LAMA3, A. MACCAFERRI3, A. MATTANA3, M. MORSIANI3, G. NALDI3, F. PERINI3, G. PUPILLO3, M. ROMA3, S. RUSTICELLI3, M. SCHIAFFINO3, P. DI LIZIA4, M. LOSACCO4, M. MASSARI4, M. REALI5, & W. VILLADEI5 1Institute of Space Sciences and Astronomy (ISSA), University of Malta; 2Dept. of Physics, University of Oxford, Oxford, OX1 3RH, United Kingdom; 3Istituto Nazionale di Astrofisica – Istituto di Radioastronomia, Via P. Gobetti 101 - 40129 Bologna, Italy; 4Department of Aerospace Science and Technology, Politecnico di Milano, Via G. La Masa 34 - 20156 Milano, Italy; 5Italian Air Staff – ITAF, Viale dell’Università 4, 00185 Rome, Italy. email [email protected] (corresponding author)

The ever increasing satellite population in near-Earth orbit has made the monitoring and tracking of cooperative and non-cooperative objects ever more important. Non-cooperative objects, or space debris, pose a threat to existing and future satellites as they cannot avoid potential collisions. Furthermore, the orbit of the smaller debris is often not actively monitored. As the population grows, the risk of a collision increases. Thus, various institutions around the world have been upgrading their space detection capabilities in order to better monitor the objects orbiting Earth down to a few centimetres in diameter. One of the latest such systems is the BIstatic RAdar for LEo Survey (BIRALES) space debris detection system based in Italy. The BIRALES system is a bistatic radar composed of a radio transmitter in Sardinia and the Medicina Northern Cross radio telescope near Bologna as the receiver. The backend of this system includes a digital beamformer able to synthetize 32 beams covering the instrument’s Field of View (FoV). As a high-velocity object transits, its Doppler shift signature (or track) can be measured. Whilst a number of streak detection algorithms have been proposed for optical telescopes, the number of detection algorithms for high-speed objects for bistatic radars is limited. This work describes the detection algorithm used in the BIRALES space debris detection pipeline. The detection algorithm takes the beamformed, channelized data as input. Firstly, the data undergoes a number of pre-processing stages before the potential space debris candidates are identified. Secondly, the candidates are validated against a number of criteria in order to improve the detection quality. The algorithm was designed to process the incoming data across 32 beams in real-time. Initial validation results on known objects are positive and the system has been shown to reliably determine orbiting objects with minimal false positives.

Keywords: Orbital debris, Radar, Detection, Clustering

1 INTRODUCTION The increasing risk of orbit collisions has led to a heavy investment in new preventive measures to mitigate the pro- Satellites have become indispensable to many areas and disci- liferation of space debris and minimize the collision risks for plines including telecommunications, climate research, nav- active satellites. Bonnal and McKnight [2] list a number of di- igation and human space exploration. Of the thousands of rect removal techniques that have been proposed throughout satellites that were put in orbit, only a fraction remain opera- the years. Furthermore, post-mission disposal and mitigation tional either due to a planned decommission or a fatal break- strategies are now taken into account during the mission de- up. In addition, Klinkrad [1] mentions several other sources of sign of new satellites. inoperative hardware within the space environment including spacecraft used to place the satellites in orbit, such as rocket It is thus important to characterize the orbital environment bodies, dust particles and leaked cooling agents. Collectively, of Resident Space Objects (RSO) through direct observations. these objects are often referred to as orbital or space debris. Measurements provide the space agencies and satellite opera- Both natural and man-made space debris pose a considera- tors with a deeper understanding of the current environment, ble threat to the active satellites and space missions. As the including growth trends [3] and accumulation regions. This number of space debris objects increase, the likelihood of a data is required not only at a mission design phase, but also to collision is also increased. assess the risk of potential collisions by predicting the debris trajectory [4].

This paper was originally presented at the 69th International Space agencies amalgamate the output of different sensors in Astronautical Congress 2018 in Bremen, Germany – paper number order to detect, track and identify both known and unknown IAC-18.A6.1.9x47208. orbiting objects. At present, the space environment is deter-

102 Vol 72 No.3 March 2019 JBIS A REAL-TIME SPACE DEBRIS DETECTION SYSTEM for BIRALES mined through both space-borne [5] and ground based optical and radar systems. Observational facilities and space surveil- lance networks monitor only a fraction of the space debris pop- ulation. Thus, the introduction of new high sensitivity instru- mentation is paramount to the better characterization of the space environment. Apart from building new high sensitivity instruments, one of the research missions of the ESA’s Space Situational Awareness (SSA) goals is to use existing systems and upgrade them for the monitoring of orbital debris.

Since 2006, the Istituto Nazionale di Astrofisica (INAF), funded by the Italian Space Agency (ASI), has been investigat- ing a number of possible radar setups at the Medicina radio astronomical station. A review of the Italian space debris ac- tivity can be found in [6] and [7]. In 2007, the Medicina 32 m parabolic antenna was used in three space debris detection tests. The observation campaigns were done in bi-static mode with the Medicina parabolic dish acting as receiver and the RT- 70 parabola in Evpatoria, Ukraine, acting as transmitter. The Fig.1 The BEST2 array within the Northern Cross in Medicina, near system has been shown to be capable of detecting small sized Bologna, Italy. debris in Low Earth Orbit (LEO) and Medium Earth Orbit (MEO) [8]. mediate 30 MHz frequency and then fed to a ROACH-based digital backend developed by [14] [15]. The signals are digi- In 2015, INAF, in collaboration with the University of Malta tized and channelized into 1024 single polarisation, 78.125 kHz and the Politecnico di Milano, embarked on an ambitious pro- wide, coarse frequency channels. ject to upgrade one of INAF radio telescopes for use in SSA. The aim of the project is the design and implementation of an The channelized data are transferred as a UDP stream to a operational bistatic radar for orbital debris in LEO as part of processing server over a 10 GBit link [16]. At the processing the European Space Surveillance & Tracking (SST) Support server, a purpose-built space debris software backend analyses Framework [9] [10]. the incoming data for radar echoes reflected off in-orbit -ob jects. This contribution describes the processes used in the detec- tion of orbital objects with the BIRALES system. First, an over- 3 A DATA PROCESSING BACKEND FOR SPACE DEBRIS: view of the radar system is given. This is followed by a descrip- PyBIRALES tion of the data processing software implemented to process the incoming data. Finally, the detection processing pipeline is 3.1 Description presented together with some preliminary results of this novel system. The data processing system, called PyBirales, processes the in- coming data from the BEST2 digital backend in real-time. The 2 BIRALES: A BI-STATIC RADAR FOR SPACE DEBRIS data rate of the instrument is in the order of tens of megabytes DETECTION per second. For this reason, PyBirales was developed in Python whilst computationally intensive tasks were developed in C++ The BIRALES is a bistatic radar which makes use of the Radio and imported into Python. Frequency Transmitter (TRF) located in the Italian Joint Test Range of Salto di Quirra (PISQ), in Sardinia as a transmitter The incoming antenna signals are processed using the Py- operating in continuous wave mode with a maximum output Birales data processing pipelines. A data processing pipeline is power of 10kW. The Basic Element for SKA Training (BEST) made up of a chain of separate processing stages as shown in [11], located at the Medicina radio astronomical station, near Fig. 2. At each processing stage, or module, the incoming data Bologna, Italy, is used as receiving component of the bistatic is mutated and passed over to the next processing stage. radar. The processing modules can be chained in any order, so The BEST project was a series of planned upgrades to a sub- long as the data container, or data blob, transferred to the sub- set of the Northern Cross antenna. The array is currently at the sequent module is compatible. A check on blob compatibility second stage of the project called the BEST2 (Fig 1.) [12]. It in-between the chained modules is performed upon the initial- consists of eight East-West oriented cylindrical concentrators isation of the pipeline. as shown in Fig. 1. Each cylinder has a reflecting surface made of 430 parallel steel wires of 0.5 mm placed 2 cm apart [13].

Four low noise receivers are installed on the focal line of each cylinder. Each receiver combines the dipole signals in groups of 16, resulting in four analogue channels per cylinder. This gives a total of 32 elements which are arranged in a 4 by 8 grid having a total collecting area of 1420 m2. The amplified RF signals travel to a receiver room, locat- ed within the central building, through analogue optical fibre Fig.2 The pipeline design pattern used in the PyBirales pipelining links 500 m long. The signals are down converted to the inter- framework.

JBIS Vol 72 No.3 March 2019 103 DENIS CUTAJAR ET AL.

The processing pipelines in PyBirales are designed to pro- normally produced once per observation. cess the data in real-time. This means that the following re- al-time condition in each module is met: The complexity of the system is shielded away from the op- erator by a web-application. This is a consolidated approach in (1) monitoring, control and administration of PyBirales. Ultimate- ly, the goal of the system was to build the necessary compo- nents that will eventually be used for the detection of orbital For instance, at a sampling rate of 78,125 samples per sec- debris. The next section gives an in-depth description of the ond, 262144 samples need to be processed in less than 3.35 s for detection pipeline within PyBirales, and how this is used for the real-time condition to be met. This condition has to be met the detection of orbital debris in LEO. at each processing stage within the pipeline. 4 DETECTION STRATEGY The creation of a data processing pipeline is facilitated by a framework that is able to generate different processing pipe- The detection of RSO objects is done through a dedicated de- lines with ease. This framework provides a standard and con- tection pipeline built using the PyBirales framework described cise method of chaining a set of modules into a pipeline by earlier. The detection pipeline makes use of a number of pro- making use of a pipeline manager that initialises the processing cessing modules and its composition is illustrated graphically modules and chains one processing module to the next. Using in Fig. 3. this system, a number of pipelines can be created, such as the detection and the correlation pipelines. The latter is an inte- The incoming data from the digital backend is consumed by gral part of the calibration routine that is used to calibrate the the receiver module. The signals from the 32 antennas in the BEST2 array. BEST2 array are beam formed into a number of beams. The number of beams generated, together with their pointing, is 3.2 Calibration configurable and can be specified by the operator of the PyBi- rales software backend. At present, the default configuration A radio interferometer array will always be liable to geometri- generates 32 coherent beams within the primary beam of the cal and instrumental delays, with the latter changing on short BEST2 array. time scales and thus requiring frequent corrections. Geometri- cal delays arise from the fact that the array itself is inherently The channel bandwidth is too wide for the detection of small composed of antennas with a quantified ground separation, signature objects such as space debris. Hence, finer channeli- with radio waves arriving at each antenna at specific time de- zation is applied at the channelizer module. A polyphase filter lays. Such introduced delays can thus be easily defined and bank channelizer is used since leakage is significantly less than corrected for with accurate knowledge of the antenna and the a standard Discrete Fourier Transform. The channelizer splits observed point source. On the other hand, instrumental delays a single coarse channel into 8192 separate channels of around are less predictable and are inherent to the instrument itself. 9.5 Hz each. This gives a temporal resolution of around 100 m. Such errors are the reason for which the calibration observa- tion of a point source passing through the beam is necessary. The output data blob of the channelizer module serves as the input to the pre-processor module. It is an n-dimensional array For calibration of the BEST2 array, a correlation matrix of consisting of 8192 channels by 32 time samples for each of the complex visibilities is first produced by the correlation pipe- 32 beams. The operations described in the following sections line. This is generated through an observation of an unresolved are applied across all the 32 beams generated by the PyBirales calibration source, such as Cassiopeia A or Taurus A, as it tran- beamformer. sits through the field of view of the array’s beam. Consequent- ly, after the point source transits at the meridian through the 4.1 Pre-processing and Filtering beam, the entire correlation matrix dataset is inspected by the calibration routine. The selection of those visibilities obtained The pre-processing module takes in the beam formed, chan- at the point source’s exact moment of passage at the meridian nelized data blobs and calculates the power P from the received ensues. These visibilities are selected on the premise that, for a antenna voltages V as: perfectly calibrated array, the phase component of the complex visibilities for all baselines during the passage of a point source (2) at zenith should be equivalent to zero. Corrections to obtain such visibilities’ phases are calculated for every antenna, taking An estimate for the background noise at a channel was taken a specific antenna as phase reference. to be the root mean square value of the power of the N samples at that channel c. This noise needs to be filtered out before a Thus, after obtaining phase coefficients through this routine, detection could be made. Filtering of this data is done at the the visibilities are phase aligned accordingly. A gain calibra- next module in the pipeline namely filtering. tion routine follows, using the phase calibrated visibilities as a starting point. The gain calibration assumes that the source The filtering module consists of a number of filters that are power observed is equal throughout the array, with its abso- applied on the data sequentially. The aim of the filters is to -re lute value dependent only on the source brightness itself. A move the background noise as much as possible. This process logarithmic implementation of a least squares estimation tech- reduces the number of data points that the detection algorithm nique is used for obtaining an estimate for gain solutions per has to process thereby simplifying the detection problem at a antenna. These are combined with the calibration coefficients later stage. obtained from the initial phase calibration routine (which themselves also solve for gain errors to some extent) in order Fig. 4 shows a subset of the raw, pre-processed data that is to obtain the final calibration coefficients. Such coefficients are depictive of the input to the filtering module. One may note

104 Vol 72 No.3 March 2019 JBIS A REAL-TIME SPACE DEBRIS DETECTION SYSTEM for BIRALES

The application of this filter clips the bulk of the data given that most of the data in this blob is noise. However, this fil- ter does not adequately filter the RFI noise introduced by the transmitter line. In observations where the transmitter is locat- ed near the receiving part of the radar, the RFI signature of the transmitter can be measured. This is usually distinguished by two unique characteristics. First, due to the relative proximity of the transmitter, the measured power is very high, usually or- ders of magnitude higher than other detected signals. Second, the transmitter frequency does not change in time and is pres- ent within the same frequency channels across time. These two features are exploited in the transmitter line filter.

The channels at which a high power is measured are identi- fied by performing a peak search on the data. This is achieved by summing the bandpass across N samples. Channels whose summed power is greater than an arbitrary threshold τ are con- sidered to be peaks. At these channels, the power Pc across all time samples is clipped (Eq. 4).

(4)

After the application of these filters, the data is still charac- terized by random and isolated pixels with a high SNR. These data points, or pixels, can be removed through binary hit-or- miss transform. This transform finds the pixels which match Fig.3 A block diagram of the detection pipeline used in the a specified pattern or mask. The mask is a representation of a BIRALES system. The modules pertaining to the detection of RSO single pixel with no immediate neighbouring pixels. are highlighted in grey. As shown in Fig. 5, the application of the aforementioned the radar signature of an RSO (NORAD 30774) as it passed three filters proved to be very effective in clipping most of the through the FoV of the instrument. The Radio Frequency In- noise thereby reducing the complexity of the detection algo- terference (RFI) contribution of the transmitter is also visible. rithm. However, some noise can still be present after this filter- ing process. Thus, the detection algorithm, needs to be robust The first component in the filtering module is the back- enough to cope for any noise artefact that are left after the fil- ground noise filter. A data point is considered to be noise if its tering stage. power value P is within four standard deviations, σ of the mean noise power Pε (Eq 3). 4.2 Detection

The detection module accepts channelized data which has (3) been pre-processed and filtered at the previous modules in the pipeline. The aim of the detection algorithm is to identify the

Fig.4 A spectrogram of a subset of the channelized, pre-processed Fig.5 The output of the filtering module after filtering the data from data showing the power received across a set of channels and time the background noise, transmitter RFI and speck noise. It can be samples. The transmitter RFI frequency and the radar signature of noted that most of the noise is eliminated and the radar signature an RSO object can be noted within this image. of the target RSO is clearly noticeable.

JBIS Vol 72 No.3 March 2019 105 DENIS CUTAJAR ET AL. radar echoes reflected by the RSO. These echoes are charac- terised by a sharp increase in intensity when compared to the background level. Given the linearity of these radar signatures, classical edge detection algorithms such as Hough transforms were investigated initially. The Hough transform solves the edge detection problem by converting the problem to a local peak detection search in parameter space [17]. In earlier work [18], this method proved to be effective in the identification of large RSOs. However, the algorithm proved to be less effective in the presence of noise and targets with a low SNR.

The manuscript presents the use of a clustering technique for the automated identification of these radar echoes. Clus- tering is the unsupervised organisation of unlabelled data into groups called clusters. Patterns within one cluster are consid- ered to be more similar to each other than to any other pattern associated with a different cluster [19].

Given that the number of radar echoes present is not known a priori, a hierarchical clustering technique was used. Conse- Fig.6 The output of the DBSCAN algorithm that is applied on the quently, an established, density-based clustering technique filtered data. The target object is identified as being Cluster 1. known as Density-Based Spatial Clustering of Applications False positives are also identified including data points identified with Noise (DBSCAN) [20] was used. An added advantage of as noise. this clustering algorithm is that it is able to classify isolated pix- els as noise thereby distinguishing them from valid clusters. time epoch. Fig. 6, shows the clusters labelled by the DBSCAN for a typical dataset. One may observe that the algorithm is The algorithm groups together data points based on their very effective in grouping most of the pixels related to the tar- proximity to each other using a distance metric a minimum get detection. Isolated, small, clusters of pixels are also correctly number of points per cluster. The Eps-neighbourhood of a classified as noise. However, one may also observe the presence point q is defined by NEps(q) ={p D|dist(q, p) ≤ Eps} where of clusters which are clearly false positives. This is especially dist(q, p) is the distance function and Eps is a constant that de- true near the channels at which the transmitter RFI frequency fines the maximum distance between two data points p and q was present. belonging to the same cluster. The Eps distance was determined empirically and set to 5 pixels. False positives are handled by a validation process on the detected clusters. The validation process consists of a number In DBSCAN, data points are classified as either being inside of criteria. For instance, clusters made up of a few data points (core points) or at the edge of a cluster (border points). A point are ignored. Furthermore, the shape of the cluster is also taken p is said to be directly density-reachable from a point q if p into account. Tracklets are expected to be linear where the fre- NEps(q) and |NEps(q)| ≥ MinPts where MinPts is the minimum quency and time of a detection are strongly correlated. number of points in a cluster. On the other hand, a point p is said to be density-reachable from a point q if there is a chain Another optimisation that was introduced was to ignore of points p1,. . ., pn, p1 = q, pn = p such that pi+1 is directly den- clusters with an unrealistic Doppler shift value. Analysis of the sity-reachable from pi. A border point can also be considered catalogued objects in orbit put the expected Doppler shift Δf to be density-connected to a point q if there is a point o such to lie between -12143 and +13245 Hz. Clusters with a Dop- that both p and q are density-reachable from o. Thus, a cluster pler shift value outside of this range were dropped. Similarly, a C is considered to be a non-empty subset of D if the following range for the rate of change in the Doppler shift of the detec- conditions are satisfied. tion was obtained. The rate of change in the measured Doppler is expected to lie between -291.47 and -69.62 Hz s-1. This pro- Maximality: p, q: if p C and q is density-reachable from cess ensures that only valid RSO tracklets are passed on to the p w.r.t. Eps and MinPts, then q C. next stage. Connectivity: p, q: if p C and q is density-connected to q w.r.t. Eps and MinPts. A single RSO transient can produce multiple tracklets across multiple beams. These tracklets can appertain to the same ra- Points which do not satisfy these conditions are regarded as dar echo, or track, of an RSO. A track can span multiple data noise such that if C1. . . Ck are clusters of D, then noise ={p blobs. As a result, different tracklets in subsequent data blobs D| i: p Ci}. Once the point p is classified as either a core, bor- can belong to the same RSO track. Thus, a system of merging, der or noise, the algorithm moves to the next point in D. Thus, or linking, these tracklets, across multiple beams and blobs, be- this approach groups points that are closely packed together, longing to the same RSO was put in place. This process is called expands clusters in any direction where there are nearby points tracklet linking. using a density-based metric. This way, it is able to deal with different shapes of clusters making it ideal for the detection of In tracklet linking, the detection tracklets are associated with radar echoes. a single RSO track or just track. This is achieved by comparing the parameters of the detection tracklets with the RSO tracks. The clusters identified by this algorithm are referred to be A RSO is associated with a parent RSO track if the cosine sim- beam tracklets or just tracklets. A tracklet is a grouping of pix- ilarity between the two is below a threshold. In so doing, the els in a beam with an associated Doppler shift, channel and track grows as new tracklets are detected and associated with

106 Vol 72 No.3 March 2019 JBIS A REAL-TIME SPACE DEBRIS DETECTION SYSTEM for BIRALES it. In the case when no track exists in the queue, an empty RSO track is created and the new cluster is associated with it.

The track is kept in memory in order to be able to compare the track with any future detections of new tracklets. The can- didate is popped out of the queue when the candidate is not updated for a specified number of iterations. This ensures that no two tracks belonging to two distinct RSO get merged by chance. Whilst this approach is adequate, a more accurate ap- proach is to calculate the theoretical time at which the space candidate exists the instrument FoV.

RSO tracks which were created or modified during an itera- tion are persisted to the MongoDB database. The detected RSO tracks and other observation details are made available to an Fig.7 A plot of power against time for the NORAD 20439 detected operator in real-time through the monitoring dashboard that on the 30 June 2017. is shipped with PyBirales. At the end of an observation, the detection made are saved in TDM format at a post-processing stage once the pipeline is stopped.

This TDM file is also the input to the orbital determination block developed by a team of researchers at the Politecnico di Milano, Italy [4]. The tailored algorithm makes use of the beam illumination sequence, SNR and Doppler shift together with the beam distribution and antenna pointing. These parameters are used to refine the orbital parameters of known RSO or per- form a preliminary orbit determination in the case of unknown objects.

5 RESULTS AND DISCUSSION

One of the earliest tests of the BIRALES system was performed Fig.8 A Doppler profile for the NORAD 20439 detected on the 30 on the 30th of June 2017 observation, in which the system June 2017. successful in detecting the OSCAR 16 (PACSAT) (NORAD 20439). Launched in January 1990, OSCAR 16 is an operation- The measured SNR profile of the detected object is shown al satellite for the amateur radio community. Having a RCS of in Fig. 7. Multiple beams are illuminated as the object passes 0.1139 m2, it is an ideal target to test the system capability thus through the instrument’s FoV. The Doppler profile of this de- far. The parameters of the observation are listed in Table 1. tection is represented in Fig. 8. In this case, the highest SNR per time epoch is selected. One may observe that a strong linear relationship with a correlation coefficient of -0.99985 obtained, TABLE 1 The parameters for the NORAD 20439 observation indicative of a hyper-velocity transient. campaign on 30 June, 2017 Transit time (UTC) 2017-06-30 12:13:46.043 A global SNR maximum of 19.795 dB was measured in the central beam at 12:13:45.869 at a Doppler shift of 8208.770 Hz. TRF Elevation (°) 72.150251 Comparing these values with the expected values in Table 1 TRF Azimuth (°) 120.76353 one finds that the transit time is off by less than half a second. The measured Doppler shift shows a similar level of agreement. BEST2 Declination (°) 29.988066 6 CONCLUSION Altitude (km) 786.75898 TRF Range (km) 821.32114 With the exponential increase in satellite launches, the popula- tion of defunct or broken material in space has grown substan- BEST2 Range (km) 1054.9398 tially. This is especially true in sensitive orbits such as the LEO Slant Range (km) 1876.2609 and GEO regimes. This material, collectively known as space debris, can pose a considerable threat to the operational satel- RCS (m2) 0.1143 lites. This makes the monitoring of these objects of the utmost importance. Doppler shift (Hz) +8192.5523 In this work, a new process for the detection of high velocity The measured SNR profile of the detected object is shown RSO is presented. The detection process makes use of new data in Fig. 7. Multiple beams are illuminated as the object passes processing software that was especially built to process the in- through the instrument’s FoV. The Doppler profile of this de- coming data from the BEST2 digital backend at the Medicina tection is represented in Fig. 8. In this case, the highest SNR per radio astronomical station. time epoch is selected. One may observe that a strong linear relationship with a correlation coefficient of -0.99985 obtained, The detection module uses a clustering algorithm, DB- indicative of a hyper-velocity transient. SCAN, to group neighbouring pixels into detection clusters or

JBIS Vol 72 No.3 March 2019 107 DENIS CUTAJAR ET AL. tracklets. These detection tracklets are validated against a num- lishing the sensitivity of this novel system, these observation ber of criteria and grouped into RSO tracks. Space debris tracks test the reliability of this new radar system within the European are saved to a database and made available to the real-time SST network. monitoring front-end that is shipped in PyBirales. Detection information can also be dumped to disk and made accessible Acknowledgements to the orbital determination block within the BIRALES project. The research activities and operations described in this work Preliminary results indicate that the BIRALES system is able was developed thanks to the funding from the European Com- to detected objects at an RCS of 0.11 m2 at a slant range of 1876 mission Framework Programme H2020 and Copernicus, “SST km. These initial results are encouraging and will be extended – Space Surveillance and Tracking” contract No. 785257-2- in future works. For instance, the data processing system will 3SST2016 and No. 237/G/GRO/COPE/16/8935-1SST2016. be optimised to accommodate a higher data rate in an eventual upgrade of the BEST2 system. Furthermore, tests on smaller The Northern Cross Radio Telescope is a facility of the Uni- objects in beam park experiments are planned. In these obser- versity of Bologna operated under agreement by the IRA-IN- vations, the performance of other detection strategies may be AF (Radio Astronomy Institute – National Institute of Astro- compared to the one presented in this work. Apart from estab- Physics).

REFERENCES 1. H. Klinkrad. Space debris. Springer, 2006, pp. 5–57. antennas”. In: Astronomische Nachrichten: Astronomical Notes 327.5-6 2. C. Bonnal and D. S. McKnight. IAA Situation Report on Space Debris. (2006), pp. 624–625. Tech. rep. International Academy of Astronautics (IAA), 2016, p. 169. 12. S. Montebugnoli et al. “BEST: basic element for SKA training”. In: Wide 3. S. Montebugnoli et al. “The bistatic radar capabilities of the Medicina Field Astronomy & Technology for the Square Kilometre Array. 2009. radiotelescopes in space debris detection and tracking”. In: Advances in 13. P. Bolli et al. “Basic element for square kilometer array training (BEST): Space Research 45.5 (2010), pp. 676–682. Evaluation of the antenna noise temperature”. In: IEEE Antennas and 4. P. Di Lizia et al. “Performance assessment of the multibeam radar Propagation Magazine 50.2 (2008), pp. 58–65. sensor birales for space surveillance and tracking”. In: Proceedings of 7th 14. G. Foster et al. “Implementation of a direct-imaging and FX correlator European Conference on Space Debris. 2017. for the BEST-2 array”. In: Monthly Notices of the Royal Astronomical 5. M. Grassi, E. Cetin, and A. G. Dempster. “Enabling orbit determination Society 439.3 (2014), pp. 3180–3188. of space debris using narrowband radar”. In: IEEE Transactions on 15. F. Perini et al. “Skads”. In: First MCCT-SKADS Training School. Vol. 59. Aerospace and Electronic Systems 51.2 (2015), pp. 1231–1240. SISSA Medialab. 2008, p. 008. 6. F. Piergentili et al. “Italian activity in space debris measurements”. In: 16. A. Magro, J. Hickish, and K. Z. Adami. “Multibeam Gpu Transient Fifth European Conference on Space Debris. Vol. 672. 2009. Pipeline for the Medicina Best-2 Array”. In: Journal of Astronomical 7. G. Pupillo et al. “The INAF contribution to the ASI Space Debris Instrumentation 02.01 (2013), p. 1350008. program: observational activities.” In: Memorie della Societa 17. J. Illingworth and J. Kittler. “A survey of the Hough transform”. In: Astronomica Italiana Supplementi 20 (2012), p. 43. Computer vision, graphics, and image processing 44.1 (1988), pp. 87–116. 8. G. Pupillo et al. “Space debris radar experiments at the medicina VLBI 18. A. Morselli et al. “Orbit Determination of Space Debris Using a Bi-Static dish”. In: European Space Agency, (Special Publication) ESA SP 672 Radar Configuration with a Multiple-Beam Receiver”. In: Proceedings SP.April (2009), pp. 2–5. issn: 03796566. of the International Astronautical Congress, IAC (2014), pp. 1–11. issn: 9. Decision. Establishing a Framework for Space Surveillance and 00741795. Tracking Support no 541/2014/EU. 19. A. K Jain, M Narasimha Murty, and Patrick J Flynn. “Data clustering: a 10. A. Morselli et al. “A new high sensitivity radar sensor for space debris review”. In: ACM computing surveys (CSUR) 31.3 (1999), pp. 264–323. detection and accurate orbit determination”. In: 2015 IEEE Metrology for 20. M. Ester et al. “A density-based algorithm for discovering clusters in Aerospace (MetroAeroSpace). June 2015, pp. 562–567. large spatial databases with noise.” In: AAAI. 1996. 11. S. Montebugnoli et al. “Italian SKA test bed based on cylindrical

Received 8 March 2019 Approved 15 March 2019

108 Vol 72 No.3 March 2019 JBIS CALL FOR PAPERS

Putting Astronauts in Impossible Locations A one day technical symposium

9:00 a.m - Wednesday 27 November2019

BIS HQ, 27/29 South Lambeth Road

While the human exploration of the Moon and Mars has been extensively examined, serious technical consideration of the rest of the solar system has been largely ignored. This symposium is designed to explore the limits of where human exploration can go in the solar system and how to overcome the challenges involved. The symposium is open to papers on the transportation requirements, the practicalities of habitation in extreme environments and any other aspects of a solar system wide civilisation. Submissions should be on the basis that there will be a completed paper delivered before the symposium as well as giving a presentation on the day. All papers will be considered for publication in the Journal of the British Interplanetary Society.

Proposed papers should be described in an abstract of no more than 400 words, and submitted to the Society at [email protected]

Submission deadline 31 July 2019 Journal of the British Interplanetary Society

VOLUME 72 NO.3 MARCH 2019

www.bis-space.com

ISSN 0007-084X PUBLICATION DATE: 16 MAY 2019