Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

In-Orbit Aerodynamic Coefficient Measurements using SOAR (Satellite for Orbital Aerodynamics Research)

In-Orbit Aerodynamic Coefficient Measurements using SOAR (Satellite for Orbital Aerodynamics... The Satellite for Orbital Aerodynamics Research (SOAR) is a CubeSat mission, due to be launched in 2021, to investigate the interaction between di erent materials and the atmospheric ow regime in very low Earth orbits (VLEO). Improving knowledge of the gas-surface interactions at these altitudes and identi cation of novel materials that can minimise drag or improve aerodynamic control are important for the design of future spacecraft that can operate in lower altitude orbits. Such satellites may be smaller and cheaper to develop or can provide improved Earth observation data or communications link-budgets and latency. In order to achieve these objectives, SOAR features two payloads: i) a set of steerable ns which provide the ability to expose di erent materials or surface nishes to the oncoming ow with varying angle of incidence whilst also providing variable geometry to investigate aerostability and aerodynamic control; and ii) an ion and neutral mass spectrometer with time-of- ight capability which enables accurate measurement of the in-situ ow composition, density, velocity. Using precise orbit and attitude determination information and the measured atmospheric ow characteristics the forces and torques experienced by the satellite in orbit can be studied and estimates of the aerodynamic coecients calculated. This paper presents the scienti c concept and design of the SOAR mission. The methodology for recovery of the aerodynamic coecients from the measured orbit, attitude, and in-situ atmospheric data using a least-squares orbit determination and free-parameter tting process is described and the experimental uncertainty of the resolved aerodynamic coecients is estimated. The presented results indicate that the combination of the satellite design and experimental methodology are capable of clearly illustrating the variation of drag and lift coecient for di ering surface incidence angle. The lowest uncertainties for the drag coecient measurement are found at approximately 300 km, whilst the measurement of lift coecient improves for reducing orbital altitude to 200 km. Keywords: Orbital Aerodynamics; Drag and Lift Coecient; Gas-Surface Interactions; Thermospheric Wind; CubeSat. 1. Introduction novel attitude and orbit control manoeuvres using the aero- dynamic forces and torques that can be generated at these The Satellite for Orbital Aerodynamics Research (SOAR) altitudes. is a scienti c CubeSat mission due to be launched in 2021 The SOAR mission is a key component of the Horizon and designed to investigate the interactions between the 2020 funded DISCOVERER project [1, 2] that aims to atmospheric ow regime in very low Earth orbits (VLEO) radically redesign Earth observation satellites for sustained and di erent materials. Secondary objectives of the SOAR operation at signi cantly lower altitudes. The experiments mission are to provide new in-situ measurements of the performed by SOAR aim to improve knowledge and under- atmospheric density and composition and variation of the standing of the gas-surface interactions (GSIs) at VLEO thermospheric wind velocity over the range of altitudes altitudes and provide valuable validation data for ground- below approximately 400 km. SOAR will also demonstrate based experiments on materials and GSIs which will be performed in the ROAR (Rare ed Orbital Aerodynamics Research) facility at The University of Manchester. The Email address: nicholas.crisp@manchester.ac.uk ROAR Facility is a unique experimental set-up that is (N.H. Crisp) designed to identify novel materials for satellite applica- Preprint submitted to Acta Astronautica arXiv:2012.07407v2 [physics.space-ph] 17 Dec 2020 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Nomenclature A Total Surface Area T Free-stream temperature T 1 A Reference area T Surface (wall) temperature ref C Force coecient F v Relative atmospheric ow velocity rel C Torque coecient T x  Linear acceleration F Force Thermal (energy) accommodation coecient I Moment of inertia Normal energy accommodation coecient l Reference length  ref Rotational acceleration m Mass Atmospheric density s Molecular speed ratio Tangential momentum accommodation coe- T Torque cient tions with a focus on improved aerodynamic properties and of the expansion and contraction of the atmosphere with atomic oxygen (AO) resistance. The facility is principally the di erent diurnal, seasonal, and solar cycles. comprised of a ultra-high vacuum (UHV) environment, In VLEO the atmosphere is signi cantly less dense an AO source capable of providing representative orbital than at the ground or conventional ight altitudes and velocities and surface interactions, and a sensor suite in- is considered to be rare ed such that the mechanics of cluding ion and neutral mass spectrometers (INMS) which continuum ow regimes can no longer be applied. The enable measurement and characterisation of the incident non-dimensional Knudsen number can be used to classify and re-emitted gas- ow on sample materials [3, 4]. di erent ow-regimes and is de ned as the ratio between the Improvements in the knowledge and understanding of mean free path (the average distance between successive gas the GSIs and identi cation of novel materials that can particle to gas particle or gas particle to surface collisions) reduce atmospheric drag, improve aerodynamic control in a ow and a characteristic physical length (e.g. the length capability, or increase aerodynamic intake eciencies are of a body in that ow). When the Knudsen number is high important steps in enabling the sustained operation of (i.e. Kn  10) the gas-surface interactions along the length spacecraft at lower orbital altitudes. This reduction in of a body are of much greater signi cance than any gas orbital altitude has been linked to numerous bene ts, for particle to gas particle interactions, including those with example reduced debris collision risk, a more favourable ra- re ected particles [7]. This regime is termed free-molecular diation environment, and aerodynamics-assisted end-of-life ow (FMF). The variation of the Knudsen number with disposal. The opportunity to incorporate novel technolo- altitude is given in Fig. 1. The lower bound of the VLEO gies such as atmosphere-electric propulsion (ABEP) and range can be de ned as the ow enters the more complex aerodynamic attitude and orbit control is also presented. transitional regime (Kn < 10), and the conditions of free- For Earth observation applications, lower altitude orbits molecular ow cannot be assumed. This is shown to occur o er the possibility of smaller and less expensive platforms, for altitudes below approximately 130 km altitude. leading to cheaper data products, or alternatively higher In the FMF regime, the forces which act on a body resolution imagery, both with a wide range of potential can be determined by simply considering the interaction commercial, environmental, and societal impact [5]. Com- between the incident molecules and satellite surfaces, and munications satellites may correspondingly bene t in their the subsequent angular distribution and velocity of the design from improved link-budgets, lower latency, and in- re-emitted or re ected particles. It has been observed that creased frequency re-use [6]. these GSIs, and the associated momentum and energy transfer, are dependent on surface roughness and cleanli- 1.1. Gas-Surface Interactions in Very Low Earth Orbit ness (particularly related to altitude-dependent AO adsorp- tion), surface composition and lattice structure, surface The upper bound of the VLEO range can be broadly temperature, gas composition, and the incident particle de ned as the altitude below which the atmosphere begins temperature, velocity, and incidence angle [9{11]. The to have a signi cant e ect on the orbital and attitude presence of ionised thruster plumes may also a ect the dynamics of a spacecraft and is typically de ned at 450 km local ow conditions and therefore the aerodynamic forces altitude. However, this de nition is somewhat fuzzy as in produced [12]. reality the the atmospheric density can vary considerably Models for these GSIs have been developed to enable at this altitude (as shown in Fig. 1) principally as a result 2 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Figure 2: Variation of drag and lift force coecients of a single-sided at-plate of area 1 m under di erent GSI models and inputs assumptions for FMF conditions (s = 10, T = 300 K, T = 600 K). Aerodynamic coecients are referred to the projected (cross-sectional) area with respect to the oncoming ow. [7] assumes di use re-emission of particles, whilst modi ed analytical equations (from Schaaf and Chambre [14]) based on the Cercignani-Lampis-Lord (CLL) model [20{22] as Figure 1: Variation of density (top) and free-stream Knudsen proposed by Walker et al. [23] can be used to represent spec- number (bottom) with altitude for di erent levels of solar activity, ular re ections. For di use re-emission but reducing energy assuming a characteristic length of 1 m and atmospheric parameters accommodation the drag force increases as the incidence calculated using the NRLMSISE-00 model [8]. Average kinetic approaches normal to the ow. Meanwhile, the lift force diameter is weighted by the number density of the atmospheric species at each altitude (values for AO and N assumed to be increases modestly for inclined surfaces to a maximum at conservatively equivalent to N ). approximately 45°. However, if quasi-specular re-emission or specular re ection properties are exhibited the drag can be signi cantly reduced for shallow incidence angles estimation and determination of the aerodynamic forces (<45°) and will increase as the incidence approaches normal which act on surfaces in these conditions. These models to the ow. Lift force generation can also be increased are used for the purpose of orbit and attitude simulation, signi cantly. spacecraft design and modelling, and in the development of Studies of in-orbit GSI performance have shown that atmospheric density and thermospheric wind models from materials commonly utilised on spacecraft have exhibited on-orbit observations [13]. Popular GSI models used in primarily di use re-emission properties with high energy the eld of orbit aerodynamics include those of Sentman accommodation (  0:8 to 1:0), particularly in low alti- [7], Schaaf and Chambre [14], Schamberg [15], Gaposchkin tude orbits where surface contamination (principally by [16], Storch [17]. Comparison and review of these models adsorbed atomic oxygen) is high [9, 24]. The prevalence of is provided by Mostaza-Prieto et al. [18], Livadiotti et al. energetic and highly-reactive atomic oxygen in low altitude [19]. orbits also introduces the issue of material erosion [25{27] In general, the re-emitted or re ected particle distribu- that can further increase accommodation and therefore re- tion is described as di use or specular with some models sult in di use re-emission. However, evidence of increasing using combinations of these de nitions. Coecients that quasi-specular re-emission behaviour has been observed for de ne the range of energy or momentum accommodation materials on spacecraft in higher altitude orbits (800 km at the surface are typically used to characterise the GSI to 1000 km) [28] where surface contamination is lower and performance given an assumed re-emission distribution and in elliptical orbits where the incident kinetic energy near therefore the forces experienced by the surface. perigee is greater [24]. Ground-based molecular beam Fig. 2 demonstrates the e ect of di erent GSI model experiments have also demonstrated such quasi-specular assumptions and given parameters on the drag and lift qualities for clean materials under UHV conditions and at force coecients for a at-panel surface. Sentman's model 3 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ energies approaching that of orbit velocity [29]. by the abundance of GSI models, but lack of consensus For the purpose of improving aerodynamic performance on their suitability and application for di erent materials, in the VLEO regime, quasi-specularly re ecting materi- surface treatment, altitude range, and period of the solar als in combination with appropriate satellite geometric cycle [44]. design, would provide the ability to reduce aerodynamic The unique combination of a test satellite (SOAR) and drag and therefore increase orbital lifetime or reduce the an experimental ground facility (ROAR) aims to improve requirements for drag compensating propulsion systems. the knowledge of GSIs and the underlying physical mecha- Alternatively, using the increased drag generated at high- nisms, leading to improved modelling of aerodynamic forces incidence angles, enhanced aerodynamics-based deorbit in VLEO. A systematic investigation of di erent materials devices could be conceived. The capability to produce lift will also seek to identify those that can provide improved forces of greater magnitude also provides the possibility aerodynamic performance through specular re ection prop- to utilise new methods of aerodynamics-based orbit and erties and have atomic oxygen erosion resistance, enabling attitude control. a new class of spacecraft that can operate sustainably at lower orbital altitudes. 1.2. On-Orbit Investigations of Gas-Surface Interactions A number of investigations of material GSI performance 2. Satellite Design and surface accommodation in the FMF regime have been performed using direct on-orbit measurements and ground- The principal scienti c objective of SOAR is to in- based observations of spacecraft. Review and comparison vestigate the variation of the aerodynamic coecients of of studies in this area have been provided by Moe et al. di erent materials and surface nish at di erent incidence [9], Moe and Moe [30]. angle to the oncoming ow and at di erent orbital altitudes. Direct measurement of the remission angle of scattered In-situ measurement of the incident ow environment will AO from a vitreous carbon surface was studied on the be used in addition to measured attitude and orbital param- STS-8 Space Shuttle ight. The di use remission spectrum eters to determine the forces and torques experienced by observed, approaching a cosine distribution, indicated that the body. By providing in-situ density measurements of the almost full accommodation was occurring at the surface oncoming ow which can be used directly in the recovery of [31]. Investigation of scattering angle from an oxidised the tted aerodynamic coecients and associated accommo- aluminium surface has also been noted as part of a larger dation coecients, this experimental methodology presents study on erosion characteristics of scattered AO which was a signi cant advantage over previous observation-based conducted on MISSE-FF (Materials International Space studies. Station Experiment Flight Facility) by Banks et al. [32, 33]. SOAR takes the form of a 3U CubeSat developed from Aerodynamic coecients resulting from the summed e ect the Dsat design of Virgili Llop and Roberts [45], previ- of GSI over a spacecraft body have also been studied. ously proposed for the QB50 programme for lower thermo- For example, the aerodynamic coecients of the Space spheric exploration and research. The basic geometry of Shuttle were measured using accelerometer data during the SOAR is shown in Fig. 3. transitional re-entry phase [34, 35]. A set of four panels that unfold after launch and de- Other studies have used observational methods to de- ployment into orbit to extend away from the satellite body termine the aerodynamic coecients of di erent spacecraft and can be rotated with respect to the satellite body (and or materials from the attitude motion or orbital trajectory. the oncoming ow) have been designed to achieve proposed GSI and surface accommodation can subsequently be inves- investigation of material aerodynamic coecients and to tigated by considering the spacecraft attitude and geometry act as aerodynamic control surfaces. These appendages are and through comparison to di erent models. These stud- termed steerable ns herein. ies have notably included Paddlewheel [36] and spherical The surfaces of these steerable ns have been coated [10, 37{39] satellites, but have also included more complex with four di erent material coatings with the con guration geometries [40, 41] and predictions for time-varying atti- of similar materials placed on opposing surfaces as indicated tude where observed or measured data was not available in Fig. 4. Through coordinated rotations of the steerable [42]. However, in the absence of measured data, the results n, each material can therefore be individually exposed obtained using these methods are typically dependent on into the ow at varying angles of incidence (neglecting the modelled atmospheric densities and are therefore subject to body of the spacecraft and parallel surfaces). their inherent biases and uncertainties [43]. Furthermore, As described by Virgili Llop and Roberts [45], the steer- as some of the analysed spacecraft may also have been able ns can be operated in pairs in two principal ways; used during the development and calibration of the density co-rotation and counter-rotation. From the minimum drag models, some circular logic may be present [13]. con guration and under stable ow-pointing conditions, There remains both a lack of knowledge of the physical co-rotation of a single opposing-pair of the steerable ns mechanisms that control GSI behaviour in VLEO and how (see Fig. 4b) exposes a single material to the ow and will these apply to di erent materials and their interactions generate a net lift or side force and therefore a torque (in in the true orbit environment. This is further exhibited 4 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ (a) Minimum drag con guration (b) Maximum drag con guration Figure 3: Design of the Satellite for Orbital Aerodynamics Research (SOAR) with forward-facing ion and neutral mass spectrometer (INMS) and the steerable ns oriented in the two nominal aerostable con gurations. Table 1: Principal geometric and system parameters of SOAR. yaw for the vertical ns or pitch for the lateral ns). The spacecraft will therefore rotate to y at an angle to the ow. Property Value Contrastingly, counter-rotation of a pair of opposing ns (see Fig. 4a) can similarly expose a single material to the Mass [kg] 2:88 ow, but creates opposing lift forces from each steerable Length (in z-axis) (L ) [m] 0:366 n, resulting in no net side-force and a rolling moment that Total Surface Area (A ) [m ] 0:225 causes the spacecraft to spin up about the ow-pointing CoM (in z-axis from rear) [m] 0:161 8 9 8 9 direction. x 0:0392 < = < = In order to provide in-situ information about the ow Principal MoIs y [kg m ] 0:0392 : ; : ; conditions, including thermospheric winds, the spacecraft z 0:0288 features a forward-facing ion and neutral mass spectrometer Residual Magnetic Dipole [A m] 18 10 (INMS), labelled in Fig. 3. This sensor, improved since the RW Max Torque [N m] 23 10 development of the QB50 satellites, includes new time-of- RW Max Ang Momentum [N m s] 1:2 10 2 9 ight (ToF) capability, enabling assessment of the incoming RW Spin Axis MoI [kg m ] 694:5 10 ow velocity in addition to the total atmospheric density and ow composition. To maintain accuracy of the INMS instrument, the spacecraft must be pointed in the direction despite not having a star tracker, principally as a result of of the oncoming ow within a given angular range (see the improved gyroscope (IMU) performance. Table 2). Simply, this requires that the spacecraft nominally A NovAtel OEM719 GPS receiver provides the precise ies in an attitude that is closely aligned with the direction position (<1:5 m) and velocity (<0:03 m s ) of the space- of the ow. craft and removes dependency of the experiment on ground- Attitude control of the spacecraft is principally enabled based observational tracking information. The accuracy by a three-axis reaction wheel assembly (tetrahedral con- and performance of such miniature commercial-o -the-shelf guration of four wheels). A three-axis magnetorquer is (COTS) GPS receivers in LEO has been discussed [48, 49] also included to perform initial detumbling operations fol- and demonstrated in orbit, for example on the PROBA-2 lowing launch and to enable desaturation and momentum [50] and CASSIOPE satellites [51]. management of the reaction wheels. Further parameters of interest relating to the spacecraft Attitude determination for SOAR is provided by ne design are summarised in Table 1. sun sensors, a magnetometer, and a high-performance IMU (Epson M-G370). Using a unscented Kalman lter (UKF), 3. Experimental Methodology the combined sensor set is expected to provide an attitude knowledge with an expected uncertainty of less than 1° The primary scienti c objective of SOAR is to pro- (3-sigma) even during eclipse. This exceeds the attitude vide in-space measurements of the GSI characteristics of knowledge performance of the antecedent GOMX-3 satellite di erent materials and surface-coatings in the VLEO en- [46] and approaches that of the GOMX-4B [47] satellite, vironment. The steerable ns of SOAR can be used to 5 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ (a) Counter-rotated con guration of the lateral ns. (b) Co-rotated con guration of the vertical ns. Figure 4: Principal experimental con gurations of the steerable ns on SOAR showing the corresponding arrangement of the four di erent test materials. expose di erent materials to the oncoming ow at varying The force F can be associated with the dimensionless incidence angle and at di erent altitudes as the orbit of force coecients C using Eq. (1), which also expresses SOAR decays. the accelerations x as a function of the dynamic pressure The orbit trajectory and attitude of the spacecraft will of the surrounding ow and the spacecraft geometry. vary depending on the con guration of the steerable ns with respect to the oncoming ow. With knowledge of F = v A C = mx (1) ref F rel the ow conditions and spacecraft position/orientation where  is the local atmospheric density, v the space- over time, the aerodynamic forces and torques experienced rel craft velocity relative to the oncoming ow, and A the by the satellite can be estimated and linked to the GSI ref reference area. characteristics of the di erent surfaces exposed to the ow. Investigation of the drag coecient of di erent materi- Reconciliation of the force and moment coecients with the true nature of the GSI mechanics still requires a model als exposed to the ow by the steerable ns was proposed by Virgili Llop and Roberts [45] for the Dsat mission. In this for the exchange of energy and momentum of the gas species method, opposing steerable ns are counter-rotated, expos- with the surface and the associated particle re ection/re- ing the same material/coating to the oncoming ow, and emission pattern. However, experimental determination nominally producing no net lift/side-forces or pitch/yaw of the aerodynamic coecients provides valuable in-situ torques but only a net torque in roll. Thus, only an in- validation data for the ground-based material experiments, creased nominal drag force is generated by the panel area in particular those that are planned for the ROAR Facility. exposed to the ow and the associated drag coecient can be determined from the variation in the spacecraft 3.1. Drag Force Coecient trajectory over a period of time using the orbit determina- A body exposed to an oncoming ow will experience tion and free-parameter tting process described later in forces of an aerodynamic nature, the magnitude and direc- Section 3.3. tion of which will be dependent on the orientation of the On SOAR, both co-rotated and counter-rotated con gu- body with respect to the direction of the oncoming ow. rations of opposing steerable ns will be considered. Given This force is often decomposed into three mutually perpen- the con guration of the material coatings shown previously dicular forces in the body axes (axial, normal, and side) (Fig. 3), the steerable ns can be rotated independently to with associated coecients. Alternatively, the components expose a single material (on two opposing ns) into the on- of the force and coecients with respect to the oncoming coming ow to investigate the variation in drag coecient ow are considered; drag, lift, and a third mutually per- with incidence angle and at di erent altitudes. pendicular component (often referred to as side-force or The drag coecient for a given orbital altitude and sometimes cross-wind). The term lift, will be used herein to con guration of the steerable ns can subsequently be describe both force components perpendicular to the drag, recovered by considering the produced aerodynamic accel- allowing commonality in terminology due to the fourth or- eration of the spacecraft, expressed by Eq. (1). However, der rotational symmetry of the spacecraft about the z-axis it should be noted that the drag coecient determined by (see Fig. 3). 6 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ this method is representative of the whole spacecraft in the body of the spacecraft does not contribute any additional given con guration and not only the materials exposed to meaningful roll torques, the rolling moment coecient can the ow. be recovered by considering the evolution of attitude in During these experiments, the attitude control actuators the roll-axis of the spacecraft. Free-parameter tting of (principally the reaction wheels) will be used to maintain the rolling moment coecient from the attitude evolution the nominal pointing direction of the satellite into the on- of the spacecraft requires an attitude dynamics model coming ow direction. However, as the steerable ns will be including models for the torques which act on the spacecraft. displaced from the minimum or maximum drag condition The orbit trajectory and perturbation models used in the to expose the di erent materials to the ow, the spacecraft drag coecient analysis are also required to provide the may have reduced aerostability and experience disturbing correct spatial and temporal reference for the selected aerodynamic torques. Furthermore, the external environ- torque models. mental perturbations may not be periodic in nature and For a co-rotated con guration of opposing steerable will vary in magnitude depending on a number of factors ns a pitching or yawing torque will be produced. In the including the orbital position, spacecraft attitude, solar absence of a correcting control torque, this pitch or yaw environment, lighting conditions (sunlight/eclipse), and torque would cause the spacecraft to rotate (and oscillate altitude. For di erent experiments at di erent altitudes about) an equilibrium angle to the ow. By considering the stability of the spacecraft may only be maintained by the measured evolution of attitude in the pitch/yaw-axis of the ADCS for a certain period of time before actuator sat- the spacecraft, the pitch or yaw coecient for a co-rotated uration occurs. The magnitude of the aerodynamic forces, con guration without attitude correction can be recovered. spacecraft stability, and the length of the possible exper- However, if the attitude of the spacecraft is perturbed from imental period are critical in determining the expected the ow-pointing condition the accuracy of the INMS will performance of the investigation. These factors are ex- be compromised and uncertainty in the incidence of the plored and their impact on the experimental performance steerable panels to the ow will be increased. estimated and discussed in Section 5. Alternatively, the reaction wheels can be used to coun- teract the torque produced by the co-rotated steerable 3.2. Lift Force Coecient ns and thus attempt to maintain a close to ow-pointing attitude of the spacecraft. A true ow-pointing attitude Aerodynamic torques experienced by the spacecraft can cannot be realised as knowledge of the oncoming ow di- be described by Eq. (2) in which C is the aerodynamic rection would be required. Under these circumstances, the moment coecient set (typically roll C , pitch C , and yaw l m measured angular momentum in the reaction wheels rather C ) and l is an additional reference length (L is used n ref z than the motion of the spacecraft body may be used in herein, see Table 1). The torque can also be de ned by the free-parameter tting process to determine the pitch the force F and associated moment arm ` or the rotational  or yaw moment coecients. acceleration  and the corresponding moment of inertia In a controlled co-rotated con guration the lift force can matrix I [52]. also be considered directly through coordinated analysis of 2 the orbital trajectory of the spacecraft and simultaneous T = ` F = v A l C = I (2) ref ref M rel parameter tting of the drag coecient and the lift force The aerodynamic moment coecients of the satellite coecient. However, as the lift force of typical materials is can be investigated by analysis of the spacecraft attitude a fraction of the drag force (indicated by the di erence in response with the steerable ns con gured at di erent magnitude between the lift and drag coecients of di use incidence angles with respect to the ow. The lift force surfaces in Fig. 2), the ability to distinguish the lift force coecient of the di erent materials exposed to the ow can from the measured orbital data in the presence of other subsequently be recovered from the aerodynamic moment sources of perturbation and uncertainty is likely to be coecients by considering the spacecraft geometry and limited. angle of incidence of the ns. Experimental determination The selection of the most suitable method to investigate of the moment coecients of SOAR can be performed the lift force coecient will be dependent on the attitude using either counter-rotated or co-rotated steerable n and stability characteristics of the spacecraft in each con- con gurations. guration and the expected uncertainty which is associated A counter-rotated con guration of opposing steerable with the di erent experimental modes and subsequent data ns can be used to analyse the rolling moment coecient. processing. The equal but opposing lift forces produced by the opposing 3.3. Orbit Determination and Free-Parameter Fitting counter-rotated ns act as a couple to generate a net rolling torque on the spacecraft. The rolling moment coecient In order to recover the aerodynamic forces and torques can therefore be recovered by considering the evolution of experienced by the satellite, the orbital position and atti- the spacecraft attitude in roll. The lift force coecient of tude of the spacecraft during an experimental period can be the exposed surfaces can subsequently be determined by analysed. However, in addition to the aerodynamic forces decomposing the spacecraft geometry. Assuming that the and torques of interest, the satellite will experience other 7 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ external perturbations of varying magnitude, for example (f ) Update state vector and free-parameter. Repeat due to the non-spherical gravitational eld of the Earth, if RMS has not converged. solar radiation pressure, and residual magnetic dipole inter- 6. If converged, output state vector and free parameter actions. The aerodynamic forces and torques experienced are best- t for the observed data and the provided by the satellite cannot therefore be simply isolated from mathematical models (propagation method). the measured position and attitude data. The accuracy to which the aerodynamic coecients can An orbit determination algorithm can be used to de- be determined by such a method is primarily dependent on termine the drag coecient as a free-parameter (also seen the quality of the experimental data that can be obtained as the solve-for parameter) from the measured orbit po- during each test-run. To compare two di erent spacecraft sition data from a given experimental run and associated con gurations over a given period of time, it is necessary con guration of the steerable ns [45]. The same method that the measured trajectories (in orbit or attitude) can can be applied to perform combined orbit and attitude rst be distinguished from each other in the presence of determination to t and recover a moment coecient of sensor noise and other uncertainties. For a di erence in the spacecraft from the measured orbit position, attitude, generated force or torque by the spacecraft in two di erent and environmental data. con gurations this therefore imposes a minimum require- These methods compare the output of model-based sim- ment on the position measurement accuracy (using GPS) ulations (orbit/attitude propagation) to the data measured and ADCS (attitude determination and control system) on-orbit. Using iterative di erential correction, best- t measurement accuracy. aerodynamic coecient values can be found by a least- The delity of the mathematical model used in the orbit squares method that provides convergence between the determination process is also critical to the orbit determi- measured orbit or attitude trajectory of the spacecraft nation process and recovery of the free-parameter (force and the mathematical model of the corresponding motion. or moment coecient). In order to provide convergence Uncertainty in the observations can be accounted for by towards the measured trajectory, it is necessary that the updates to the initial state vector used for the modelled model incorporates the relevant perturbations with their trajectory at each iteration and di erences in the sensor spatial and temporal variations over the duration of the performance for di erent state variables (e.g. position and test-run. The selection of necessary perturbations and velocity) using weighting methods. modelling delity are related to the noise in the measured The implementation of this non-linear weighted least position, velocity, and attitude. Perturbations that would squares process [53] can be brie y summarised: cause variation in the trajectory of the spacecraft of simi- 1. Import state vectors of experimentally measured or- lar or smaller magnitude than the noise in the measured bital elements, attitude quaternions, and atmospheric values can be safely neglected, simplifying the form of the density for each time step. mathematical model. 2. Initialise numerical orbit propagation method. (a) Select propagation force and torque models. 4. Attitude Stability and Control (b) Initialise environmental models. (c) Initialise spacecraft geometric models. The presence and use of the steerable ns on SOAR 3. Set initial guess of the free-parameter (e.g. drag produces a number of di erent forces and torques which coecient). need to be carefully considered to ensure stability and 4. Set weighting matrix based on expected uncertainty pointing accuracy of the spacecraft throughout its lifetime. of measured state vector parameters (from sensor Interaction of the spacecraft with the residual atmosphere, performance). solar radiation, and the magnetic and non-spherical gravity 5. Begin iterative scheme: elds of the Earth must be considered. The ability to (a) Apply small modi cations to each initial state control the attitude and stability of the spacecraft using vector component ( nite- or central-di erencing) on-board actuators also requires investigation as the expe- based on a small percentage of value or as a rienced torques vary in relative magnitude with decreasing function of the weighting matrix. orbital altitude. (b) Calculate the orbit trajectory for each varia- The concept of aerostability is employed by SOAR to tion of the initial state vector using the orbit provide passive pointing towards the oncoming ow direc- propagation method. tion in orbit. This aerostability is provided by the steerable (c) Form the partial derivative matrix from di er- ns which are located towards the aft of the spacecraft ences between each propagated state vectors at and thus generate a restoring aerodynamic torque in pitch each time step. and/or yaw in response to any misalignment of ow direc- (d) Calculate update to the initial state vector and tion with the longitudinal axis of the spacecraft. When free-parameter. each steerable n is oriented parallel to the longitudinal (e) Calculate weighted root mean square (RMS) of body axis of the spacecraft (Fig. 3a) a minimum drag con- residuals (between current iteration and mea- guration is generated for the nominal spacecraft attitude. sured trajectory) 8 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Similarly, when the steerable ns are all oriented normal to the spacecraft longitudinal axis (Fig. 3b) the maximum drag con guration is achieved. The surface coatings applied to the steerable ns repre- sent a range of expected GSI performance from complete en- ergy accommodation and di use re-emission to incomplete accommodation and more specular re ection properties. However, as the GSI properties of these surfaces have not been wholly characterised, the true attitude performance and control capability of the satellite are also uncertain. The di erent materials may therefore result in the produc- tion of di erent forces and torques when exposed to the ow. Figure 5: Pitch/Yaw moment coecient of SOAR with varying angle of incidence with respect to the ow in the minimum Modelling of the aerodynamic coecients for SOAR has (steerable ns parallel to body) and maximum (steerable ns been performed using ADBSat [54], an analytical panel- perpendicular to body) drag con gurations. method tool that can implement di erent gas-surface inter- action models and features basic shadowing analysis. Using this tool, a database of aerodynamic coecients can be calculated from a CAD model of the spacecraft for di erent orientation angles with respect to the ow (angle of attack and sideslip) and di erent con gurations of the steerable ns. Sentman's model [7] for GSIs has been used in all anal- yses unless otherwise stated. This model assumes a fully di use re-emission pattern of particles with a Maxwellian velocity distribution dependent on the thermal energy ac- commodation coecient and surface (wall) temperature T . A default accommodation coecient of = 1 and wall temperature T = 300 K have been used unless otherwise Figure 6: Aerodynamic sti ness (static stability derivative) of stated. SOAR for varying steerable n angle with respect to the ow. Direct simulation Monte Carlo (DSMC), developed by Bird [55] and applied in a number of di erent software tools the torque generated by the interaction of the spacecraft (e.g. DS3V, DAC [56, 57], dsmcFoam+ [58], and PICLas geometry with the oncoming ow. [59, 60]), could have alternatively been used to calculate The static pitching/yawing moment coecient of SOAR the aerodynamic characteristics of the satellite geometry in in the minimum and maximum drag con gurations is pre- the VLEO environment. These methods are able to provide sented in Fig. 5. It can be noted that the symmetrical increased delity of the modelled ow, including features nature of the spacecraft about the roll axis allows for equiv- such as intermolecular collisions, chemical reactions, and alence in pitch and yaw for static analysis. The negative electromagnetic/electrostatic interactions. Furthermore, in slope of the pitch/yaw moment coecient with angle of DSMC more fundamental forms of the GSIs are generally incidence (angle of attack or sideslip respectively) indicates implemented rather than analytical expressions that use the aerostable nature of these con gurations. mean ow conditions and averaged interactions over at- The concept of aerodynamic stability derivatives, or plate elements. However, whilst these methods can over- aerodynamic sti ness, for spacecraft at orbital altitudes, come some of the shortfalls of panel methods (for example can be used to further investigate the expected attitude for complex and concave geometries), they are signi cantly behaviour for varying geometry and ight conditions [61]. more computationally intensive, particularly so when the The static pitch/yaw stability derivative C , can be calcu- variation of the aerodynamic coecients with geometric T lated from the gradient of C over a small range about the con guration, attitude, and altitude are needed. Thus, as T nominal attitude ( = 0). The variation in static pitch/yaw the geometry of SOAR is relatively simple and the ow stability derivative for steerable n angles over the range can be considered rare ed and free-molecular (for altitudes of minimum to maximum drag con gurations is shown in above 150 km, see Fig. 1), the panel method could be Fig. 6. The increase in stability derivative with increasing safely applied for the purpose of the analyses herein. incidence angle demonstrates that a greater static stability is achieved when a larger panel area is presented to the 4.1. Static Stability ow. The static stability provided by di erent con gurations The e ect of thermal accommodation coecient on of the steerable ns can be investigated by considering static stability is also shown in Fig. 6, indicating that 9 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ aerodynamic sti ness is shown to increase with decreasing thermal accommodation under the assumptions of Sent- man's GSI model. This e ect is more marked at shallow steerable n incidence angle with respect to the ow. In order to characterise the performance of di erent materials and surface-coatings in orbit, during the exper- imental periods only one pair of steerable ns will be ro- tated with respect to the ow at any given time. With this con guration a total of four materials or surfaces can be characterised during the mission, two per pair of opposing steerable ns. When a single pair of steerable ns is counter-rotated with the spacecraft pointing into the direction of the oncom- ing ow a net rolling torque is generated but no net pitch or yaw torques are created. However, if the relative direc- tion of the ow changes (for example due to atmospheric co-rotation or thermospheric winds) or the attitude of the satellite is perturbed, induced torques are generated due to a variation in the projected area of the counter-rotated ns to the ow. Under these conditions a net pitching torque arises due to the di erence in area presented to the Figure 7: Aerodynamic torque coecient response for SOAR with di erent counter-rotated or co-rotated steerable n con guration. ow by the rotated ns. The plot of torques for counter- rotated vertical ns is shown in Fig. 7 (top), demonstrating that induced torques in pitch due to angle of sideslip have provided. In these simulations the forces and torques asso- a positive gradient about the equilibrium, and are there- ciated with the Earth gravitational potential (EGM96/2008 fore disturbing rather than restoring. Furthermore, these [63, 64]), solar radiation pressure, residual magnetic dipole torques grow at a faster rate than the restoring torques interactions (IGRF-11 [65]), varying atmospheric density generated by the lateral ns due to the change in pitch (NRLMSISE-00 [8]), and thermospheric winds (HWM93/07 angle. If the ow is therefore o set with respect to the [66, 67]) are implemented. It is important to recall that spacecraft body in yaw the spacecraft responds with cou- the results presented herein remain subject to the assump- pled motion in the pitch axis as a result of the set angle of tions and limitations of the implemented GSI model and the steerable ns. Equivalent behaviour is demonstrated the input parameters used and therefore may still di er for counter-rotated lateral ns and an o set in angle of substantially from the true behaviour in orbit. attack. This e ect is termed pitch-yaw coupling henceforth. For rotationally symmetric con gurations the previous Co-rotation of a pair of opposing steerable ns gener- analyses showed that aerostability ensures that restoring ates a net torque in pitch or yaw, but no net torque in torques will be produced in response to changes in the roll. Fig. 7 (bottom) shows the torques in pitch and yaw oncoming ow direction. However, due to the FMF na- for a con guration in which the lateral ns are co-rotated, ture of the surrounding atmospheric environment, natural demonstrating a small bias in pitch torque when the space- damping of any generated angular velocity is not generated. craft is aligned with the direction of the oncoming ow. Therefore, given an external perturbation and without any However, as the pitch (angle of attack) is increased by a additional damping input, the spacecraft will begin to os- small amount (3°) the pitch torque crosses zero with a cillate. The frequency of this oscillation is dependent on negative gradient. The spacecraft therefore demonstrates the initial disturbance, stability derivative, and the envi- stability in pitch at this equilibrium angle with respect to ronmental conditions [61]. the ow. The spacecraft is also shown to be stable in yaw The nominal response of SOAR in the minimum and about the oncoming ow direction. maximum drag con guration for varying orbital altitude and in the absence of further perturbing torques is pre- 4.2. Dynamic Stability sented in Fig. 8. The responses demonstrate the basic In order to understand the evolution of attitude over aerostable nature of the spacecraft in the maximum and time and in the presence of perturbing torques the dynamic minimum drag con gurations and that the oscillatory am- response of the spacecraft must be considered. The orbital plitude decreases and frequency increases with increasing and attitude response of SOAR in the VLEO environment aerodynamic sti ness and dynamic pressure. can be investigated using a 6-DOF simultaneous orbit and The oscillatory modes can be internally damped using attitude propagation method. This method is based on the attitude control actuators as discussed by Virgili Llop et al. numerical solution of the complete equations of motion for [68], signi cantly reducing the range over which the atti- an orbiting satellite (kinematic and dynamic motion [62]) tude of the spacecraft varies. However, due to the presence for which varying force and torque model inputs can be 10 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Figure 8: Nominal uncontrolled pitch response of SOAR for varying altitude and minimum/maximum drag n con guration in an equatorial orbit. Only spherical gravity and aerodynamic torques are considered in an equatorial orbit. An initial pitching rate of 0:1 ° s is applied. of further perturbing torques in the real orbital environ- feathered con gurations [68] similar to SOAR. ment (e.g. atmospheric co-rotation, thermospheric winds, A signi cant challenge in the control of an aerodynamic solar-radiation pressure, gravity gradient), and errors or spacecraft is that the true oncoming ow vector is generally incompatibilities associated with real attitude actuator sys- not known a priori. A true reference vector for three-axis tems (e.g. magnetorquer availability and cross-coupling), ow-pointing control is therefore missing. A reference the true dynamic response is more complex. vector including the e ect of atmospheric co-rotation can Methods of control for aerostable spacecraft have been be provided, however the prediction of thermospheric winds presented in the literature. Psiaki [69] presents a compass- using available models is associated with much greater like PID control method which utilises magnetorquers to uncertainty. Alternatively, a sensor which can provide provide three-axis stabilised nadir-pointing capability to a the oncoming ow vector could be used to provide in- 1U CubeSat with a shuttlecock con guration of deployed situ measurements for active control methods. However, \feathers". Auret and Steyn [70] subsequently applied this proposals for such instruments for use in VLEO are only method to a 3U CubeSat geometry which included a pair of just emerging [77, 78]. actuating \paddles" that provide control capability about SOAR will be launched to the International Space Sta- the roll axis. Aerostability with assisted damping has tion and deployed into an 51:6° inclination orbit with an also been successfully demonstrated in orbit. The DS-MO initial altitude of approximately 400 km. Whilst forecasts spacecraft featured extended aerodynamic skirts and a gy- for solar cycle 25 vary [79{81] the deployment will occur dur- rodamper mechanism [71]. The Passive Aerodynamically- ing a period of minimum solar activity and the atmospheric stabilized Magnetically-damped Satellite (PAMS) had a density at this altitude will therefore be characteristically cylindrical geometry with a biased centre-of-mass that low. provided aerostability and magnetic hysteresis rods to pro- The attitude response of SOAR in the maximum drag vide damping [72]. The Gravity eld and steady-state con guration with reaction wheel damping and propor- Ocean Circulation Explorer (GOCE) spacecraft featured tional control is shown in Fig. 9. The restoring aerody- rear-mounted aerodynamic ns and was equipped with mag- namic torques experienced by the spacecraft are shown to netorquers for damping [73]. MagSat also demonstrated be of a similar magnitude to the solar radiation pressure trim in pitch using an aerodynamic boom of variable length torques and over an order of magnitude smaller than the [74, 75]. Three-axis aerodynamic pointing control has also residual magnetic dipole torques. The use of the reaction been considered for a \shuttlecock" geometry [76] and wheels is therefore important at this altitude to assist the 11 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ stability of the spacecraft and maintain pointing of the spacecraft close to the oncoming ow direction. The accumulation of the perturbing torques may also necessitate periodic management of the angular momen- tum within the reaction wheels to avoid saturation. The typical disadvantages of magnetorquer cross-coupling and limited availability (due to instantaneous orientation of the magnetic eld) must be accommodated whilst the power consumption of the actuators must be considered for ex- tended use and through eclipse periods. As SOAR naturally decays to lower altitudes, the magnitude of the aerodynamic torques will increase, and the aerostability of the spacecraft when in the minimum or maximum drag con gurations will improve, reducing the requirements on the attitude actuators. Given the relative magnitude of the di erent external perturbations, particularly the residual magnetic dipole, it may be more dicult to perform the experiments at higher altitudes where the \signal" (aerodynamic forces and torques) is low in comparison to the sources of \noise" (other perturbing forces and torques). It may therefore be necessary to allow the satellite to initially decay in altitude before commencing the experimental operations. 4.3. Experimental Con gurations During the experimental operations co-rotated or counter- rotated con gurations of the steerable ns will be utilised to expose the di erent materials to the oncoming ow, mod- ifying the natural stability and therefore attitude dynamics of the spacecraft. In the counter-rotated con guration, when the space- craft is nominally pointed towards the oncoming ow di- rection, no net torques are generated in pitch or yaw. A rolling moment is however produced due to the opposing lift forces generated on the two exposed surfaces, which if uncontrolled will cause the spacecraft to spin up. If the spacecraft is disturbed from its equilibrium ow pointing con guration pitch and/or yaw torques will be produced and the e ect of pitch-yaw coupling will act to further dis- turb the attitude of the spacecraft from the ow-pointing direction. In a three-axis controlled mode, a maximum duration on operations with ns rotated in a counter-rotated con g- uration is imposed by the build-up of angular momentum and saturation of reaction wheels. This is a function of the atmospheric density, incidence angle of the steerable ns, and material performance. The thermospheric wind, solar activity, and other external disturbance torques also contribute to this attitude performance, but vary with greater uncertainty. At lower altitudes the time-period over which spacecraft can be operated successfully may therefore be signi cantly limited for some counter-rotated con gurations, the impact of which will be discussed later Figure 9: Environmental torques and controlled attitude of SOAR at 400 km altitude and 51:6° inclination in maximum drag in Section 5 with regards to the experimental uncertainty. con guration. In alternative co-rotated n con gurations, net pitch or yaw torques are generated by the common incidence 12 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Table 2: Summary of expected satellite sensor performance. angle of the two opposing steerable ns. This causes the spacecraft to rotate and y at an angle to the oncoming ow Instrument Uncertainty with an associated oscillatory motion about the new o set equilibrium attitude that results from the aerostability GPS Position [m] 2:5 (1) 1 3 of the spacecraft. The control actuators may be used GPS Velocity [m s ] 45 10 (1) to correct the nominal pointing direction of the satellite ADCS Angle [rad] 0:2 10 (1) 1 3 such that the INMS will be realigned close to the ow, ADCS Angular Velocity [rad s ] 25 10 (1) ensuring the accuracy of the measured density and ow 3 INMS Number Density [cm ] N + 0:7 (1) velocity information. However, as a bias in the pitch or yaw INMS Horizontal Acceptance [rad] 0.279 (FWHM) torques exists, angular momentum will again accumulate INMS Vertical Acceptance [rad] 0.035 (FWHM) in the reaction wheels, eventually causing saturation of the Steerable Fin Rotation Angle [rad] 0:015 (1) attitude control system. Contrastingly, for the investigation of lift force coe- cient, the roll axis can be left uncontrolled allowing the uncertainty and angular dependency, expressed as full- net torque to accumulate and the acceleration and angle in width half-maximum (FWHM) measures in Table 2. roll to be measured by the ADCS. Control of the pitch and This modi ed data is subsequently used to perform the yaw axes are maintained to keep the spacecraft pointing orbit and attitude determination processes. For recovery close to the oncoming ow direction. of the drag coecient only orbit propagation is performed, whilst for the moment coecient combined attitude and orbit propagation is performed. In both cases, the initial 5. Experimental Performance \measured" state vector is used to set-up the propagation The expected performance of the mission and the ex- method. The force and torque models used in the propaga- perimental determination of the aerodynamic coecients tion are modi ed to incorporate further uncertainty and can be investigated by testing the free-parameter tting to represent the reduction in or lack of knowledge of the and least-squares processes using simulated orbit and atti- true in-orbit environment and spacecraft interactions. For tude data. This data is modi ed with noise to represent the drag coecient tting, the the thermospheric winds the expected in-orbit sensor performance. A Monte Carlo are neglected, whilst for the moment coecient cases the approach is also implemented to encompass variation in residual magnetic dipole interactions are also excluded from the initial conditions (epoch, orbit, and attitude) and to the attitude propagation scheme. vary the addition of noise to the data for each simulated The expected performance of the experiments at di er- run. ent orbital altitudes and steerable n con gurations can The free-parameter tting process utilises a least-squares be obtained by considering the standard deviation of the orbit determination algorithm, described in Section 3.3. returned aerodynamic coecient after a number of Monte This process seeks to minimise the error between the ref- Carlo iterations. However, given that such a Monte Carlo erence (simulated or on-orbit) data and mathematically simulation only provides a random sample of results, the modelled data by varying the free values of the aerody- con dence of the standard deviation should also be con- namic coecients in the presence of known measurement sidered, within which the population standard deviation uncertainties. The process is iterative and is terminated would be expected to lie with the given con dence. by convergence criteria based on the weighted RMS of the Reducing the width of the standard deviation increases residuals between the reference and modelled data. Central- the resolution of the experiment and can be most simply di erencing methods are used to account for the errors in achieved by increasing the signal-to-noise ratio of the exper- the initial condition of the state vector due to measurement iment. For the on-orbit experiments this may be achieved uncertainties. by increasing the test run time or the magnitude of the Simulated on-orbit data is rst produced using the high- force or torque to be measured. However, in many cases delity attitude and orbit propagation method described these parameters are restricted by the platform design (e.g. in Section 4.2. Orbit and attitude noise is produced by steerable n size, reaction wheel capability) or on-orbit considering the performance parameters of the GPS and operations (e.g. power and downlink budget). Reducing ADCS sensors, reported in Table 2. Uncertainty on the the uncertainties associated with the measured on-orbit angular velocity of the reaction wheels and angular position data may also be helpful, for example improved in-situ of the steerable ns has also been similarly introduced. density measurements and position/attitude knowledge. The expected in-orbit performance of the INMS is also Finally, improved knowledge and modelling of the unmea- simulated. The measured density is rst produced using sured disturbances or perturbations would further reduce the NRLMSISE-00 [8] atmosphere model, informed by the uncertainties within the free-parameter tting and orbit orbit and attitude data, and modi ed for GPS and ADCS determination process, primarily requiring characterisation sensor and acquisition errors and noise. This density is of the residual magnetic dipole and solar radiation pressure then subsequently transformed using the INMS instrument interactions of the satellite. 13 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ 5.1. Drag Coecient data for di erent steerable n incidence angles (in 15° in- crements) are likely to be largely identi able and distin- The returned drag coecients from this data reduction guishable from one another. In the co-rotated con gura- process are shown in Fig. 10 for counter-rotated and co- tion smaller increments in incidence angle may also be rotated steerable n con gurations over a range of di erent discernible, particularly at shallower de ections. incidence angles and altitudes. A duration of 120 min is Di erence in the experimentally determined results from targeted, a limit imposed by the expected power balance the reference drag coecients arises primarily from the achievable by spacecraft in the experimental mode. Three- variation in the area of the satellite surfaces projected axis reaction wheel control is implemented to maintain an into the ow. This occurs as the satellite attitude varies approximately ow-pointing attitude and stability. The with respect to the ow as a result of the aerodynamic test-run is aborted if the angular range in pitch or yaw and other environmental torques. Further, the attitude exceeds the INMS instrument acceptance limit or the re- control system does not have knowledge of the oncoming action wheels approach saturation. Dashed lines for the ow direction and therefore uses the o set LVLH frame as underlying GSI model-based drag coecient (calculated us- a reference. Additional sources of error can be attributed ing Sentman's model with  = 1 and the ADBSAT tool) at to the sensor accuracy and noise parameters (INMS, GPS, each altitude and con guration are provided as a reference. and attitude) and the modelled forces used in the least- It should be noted that under the assumption of di use squares orbit determination tting process. Of the modelled re-emission properties, the counted-rotated and co-rotated forces, the solar radiation pressure exhibits the greatest con gurations an incidence of 90° are equivalent. The very uncertainties. Improved knowledge of the solar radiation minor di erences in the presented results are therefore due pressure interaction with the di erent external surfaces of to variations that arise from the implementation of the the satellite could improve the estimation of these e ects. Monte Carlo simulation. The dependence of the drag coecient with altitude The results for the counter-rotated and co-rotated con- appears to be less clearly identi able from the expected gurations are very similar for altitudes between 400 km experimental performance. This is due to the relatively to 300 km. However, at altitudes at and below 250 km small variation in drag coecient which is expected over the the results from the co-rotated con guration demonstrate available altitude range (200 km to 400 km) in comparison much smaller standard deviations and mean values closer to the experimental uncertainty. However, Sentman's GSI to the reference values. This is principally due to the longer model with a single accommodation coecient ( = 1) has experimental periods that can be maintained by the atti- been used in this analysis, representing typical di usely tude control system in the co-rotated con guration before re-emitting materials, for example contaminated metallic the reaction wheels saturate, in general more than double surfaces [30]. When complete accommodation and di use than the counter-rotated con guration. More orbital po- re-emission is assumed signi cant variation in the drag sition information is therefore provided against which the coecient with altitude is not expected and will only be least-squares orbit determination process can best t the driven by the variation in speed ratio and thermospheric experimentally determined drag coecient, reducing the temperature. experimental uncertainty. Increasing the experimental duration at higher orbital 5.2. Rolling Moment Coecient altitudes would similarly assist in improving the experimen- The results for the tted rolling moment coecient ex- tal uncertainty for both the counter-rotated and co-rotated periments are shown in Fig. 11 for varying counter-rotated steerable n con gurations. However, this is challenging steerable n incidence angles and altitude under two-axis due to the power budget of the satellite. The increase in (pitch and yaw) control. standard deviation at 400 km is due to low atmospheric The experimental uncertainty is generally seen to in- density at higher altitudes and therefore the limited e ect crease with orbital altitude. This is attributed to the e ect that the drag will have on the orbit over a period of only of unmeasured and poorly modelled attitude perturbations 120 min in the presence of the GPS position measurements, on the satellite during the experiment. These are most particularly for shallow steerable n incidence angles. The signi cant when the atmospheric density is lowest and greater relative magnitude of the additional disturbing per- the aerodynamic torques are therefore relatively low in turbations also contribute to the increased experimental magnitude. Residual magnetic dipole interactions are the uncertainty at the higher altitudes. most signi cant of these e ects and are not included in For both the co-rotated and counter-rotated con gura- the orbit determination algorithm. However, if knowledge tions the minimum experimental uncertainty is expected or modelling of the time-varying residual magnetic dipole to be achieved at approximately 300 km altitude where a of the satellite can be obtained this may be incorporated balance between the experimental duration, measurable into the analysis. Similarly to the drag coecient exper- e ect on the orbital trajectory, and magnitude of external iments, improvement of the knowledge and modelling of perturbations is found. solar radiation pressure e ects could also be bene cial. These results indicate that drag coecients determined For most steerable n con gurations the rolling moment from the on-orbit experiments and associated measured coecient is clearly identi able against the other results. 14 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Figure 10: Drag coecient determination performance for counter-rotated and co-rotated steerable n con gurations. Sample mean tted drag coecient (top), referred to reference area A =2, is given top with error-bars representing the associated standard deviation. Reference lines indicate the modelled GSI value. The standard deviation is given bottom with error bars representing the 95% con dence interval. Data points have been shifted slightly in the x-axis to allow for visibility of overlapping error bars. However, at lower incidence angles (15° and 30°) the er- of the SOAR geometry demonstrate the aerostable nature ror bars overlap indicating that these con gurations may of the design in the nominal maximum and minimum drag not be distinguishable from each other from the on-orbit modes and the use of the steerable ns in both counter- measurements. At greater incidence angles the variation rotated and co-rotated modes to perform the proposed between smaller increments in steerable n angle may be aerodynamics characterisation experiments. possible, particularly at lower altitudes. Using the combination of the INMS and the steerable n In contrast to the drag coecient results presented payloads, on-orbit experimental assessment of the aerody- previously, the dependence of torque coecient on orbital namic coecients of di erent materials at varying incidence altitude appears to be more marked, even for assumed to the oncoming ow will be performed. These experiments fully accommodated gas-surface interactions. However, will be repeated as the orbit of SOAR decays to investigate this variation may remain obscured by the experimental the variation with orbital altitude. The modelled uncer- uncertainties, particularly for larger steerable n incidence tainty of these experiments indicates that the drag and angles (60° and 75°) that vary more slowly with altitude. lift coecients at di erent incidence be determined from the measured parameters in the presence of the disturbing and perturbing forces and torques present in VLEO. The 6. Concluding Remarks uncertainty of drag coecient measurements was shown to be minimised around an altitude of 300 km, whilst the lift This paper has described the proposed method for de- coecient experiment generally demonstrates improvement termination of the aerodynamic coecients of di erent ma- as the altitude is reduced further. These insights will be terials on SOAR, a scienti c CubeSat due to be launched used to plan the operations of the SOAR mission. in 2021. The presented analysis and simulated dynamics 15 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ is not responsible for any use that may be made of the information it contains. References [1] P. C. Roberts, N. H. Crisp, S. Edmondson, S. J. Haigh, R. E. Lyons, V. T. Oiko, A. Macario-Rojas, K. L. Smith, J. Be- cedas, G. Gonz alez, I. Vazquez,  A. Brana, ~ K. Antonini, K. Bay, L. Ghizoni, V. Jungnell, J. Morsbl, T. Binder, A. Boxberger, G. H. Herdrich, F. Romano, S. Fasoulas, D. Garcia-Alminana, ~ S. Rodriguez-Donaire, D. Kataria, M. Davidson, R. Outlaw, B. Belkouchi, A. Conte, J. S. Perez, R. Villain, B. Heierer, A. Schwalber, DISCOVERER { Radical Redesign of Earth Observation Satellites for Sustained Operation at Signi cantly Lower Altitudes, in: 68th International Astronautical Congress, September, International Astronautical Federation (IAF), Ade- laide, Australia, 2017. [2] P. C. Roberts, N. H. Crisp, F. Romano, G. H. Herdrich, V. T. Oiko, S. Edmondson, S. J. Haigh, C. Huyton, S. Livadiotti, R. E. Lyons, K. L. Smith, L. A. Sinpetru, A. Straker, S. D. Worrall, J. Becedas, R. M. Dom nguez, D. Gonz alez, V. Canas, ~ V. Hanes- sian, A. Mlgaard, J. Nielsen, M. Bisgaard, A. Boxberger, Y.-A. Chan, S. Fasoulas, C. Traub, D. Garcia-Alminana, ~ S. Rodriguez- Donaire, M. Sureda, D. Kataria, R. Outlaw, B. Belkouchi, A. Conte, J. S. Perez, R. Villain, B. Heierer, A. Schwalber, DISCOVERER { Making Commercial Satellite Operations in Very Low Earth Orbit a Reality, in: 70th International Astro- nautical Congress, International Astronautical Federation (IAF), Washington, DC, 2019. [3] V. T. A. Oiko, P. C. Roberts, S. Edmondson, S. D. Worrall, D. Kataria, R. Outlaw, S. Haigh, N. H. Crisp, R. E. Lyons, S. Li- vadiotti, C. Huyton, L. A. Sinpetru, J. Becedas, G. Gonz alez, R. M. Dominguez, D. Gonz alez, L. Ghizoni, V. Jungnell, K. Bay, J. Morsbl, G. H. Herdrich, F. Romano, T. Binder, A. Boxberger, S. Fasoulas, C. Traub, D. Garcia-Alminana, ~ S. Rodriguez- Donaire, M. Sureda, R. Villain, J. S. Perez, A. Conte, B. Belk- ouchi, A. Schwalber, B. Heierer, Design and Development of a Figure 11: Rolling moment coecient determination performance Hyper-thermal Atomic Oxygen Wind Tunnel Facility, in: 14th for counter-rotated steerable n con gurations. Sample mean tted ISMSE and 12th ICPMSE, Biarritz, France, 2018. torque coecient (top), referred to reference area A =2 and [4] V. T. Oiko, P. C. Roberts, S. D. Worrall, S. Edmondson, S. J. reference length L , is given with error-bars representing the Haigh, N. H. Crisp, S. Livadiotti, C. Huyton, R. E. Lyons, associated standard deviation. Reference lines indicate the modelled K. L. Smith, L. A. Sinpetru, B. E. Holmes, J. Becedas, R. M. GSI value. The standard deviation (bottom) is given with error bars Dom nguez, D. Gonz alez, V. Canas, ~ V. Hanessian, A. Mlgaard, representing the 95% con dence interval. Data points have been J. Nielsen, M. Bisgaard, A. Boxberger, Y.-A. Chan, G. H. Her- shifted slightly in the x-axis to allow for visibility of overlapping drich, F. Romano, S. Fasoulas, C. Traub, D. Garcia-Alminana, ~ error bars. S. Rodriguez-Donaire, M. Sureda, D. Kataria, R. Outlaw, B. Belkouchi, A. Conte, J. S. Perez, R. Villain, B. Heierer, A. Schwalber, A Ground-Based Experimental Facility for Or- The purpose of this on-orbit experimentation is to pro- bital Aerodynamics Research, in: 70th International Astronau- vide valuable in-situ validation data for a more extensive tical Congress, International Astronautical Federation (IAF), Washington, DC, 2019. investigation of rare ed- ow GSIs to be performed on the [5] N. Crisp, P. Roberts, S. Livadiotti, V. Oiko, S. Edmond- ground with the aim to improve knowledge of GSI mech- son, S. Haigh, C. Huyton, L. Sinpetru, K. Smith, S. Wor- anisms and the associated models that describe this be- rall, J. Becedas, R. Dom nguez, D. Gonz alez, V. Hanes- haviour. A systematic study to identify materials that sian, A. Mlgaard, J. Nielsen, M. Bisgaard, Y.-A. Chan, S. Fasoulas, G. Herdrich, F. Romano, C. Traub, D. Garc a- can increase aerodynamic performance at lower orbital al- Alminana, ~ S. Rodr guez-Donaire, M. Sureda, D. Kataria, titudes will also be performed. SOAR will test two such R. Outlaw, B. Belkouchi, A. Conte, J. Perez, R. Villain, novel materials with promising drag-reducing characteris- B. Heierer, A. Schwalber, The bene ts of very low earth tics in-orbit. orbit for earth observation missions, Progress in Aerospace Sciences 117 (2020) 100619. URL: https://linkinghub. elsevier.com/retrieve/pii/S0376042120300312. doi:10.1016/ Acknowledgements j.paerosci.2020.100619. [6] B. Gavish, J. Kalvenes, The impact of satellite altitude on the performance of LEOS based communication systems, Wireless This project has received funding from the European Networks 4 (1998) 119{213. doi:10.1023/A:1019151905814. Union's Horizon 2020 research and innovation programme [7] L. H. Sentman, Free molecule ow theory and its application under grant agreement No 737183. This publication re ects to the determination of aerodynamic forces, Technical Report, Lockheed Missiles & Space Company, Sunnyvale, CA, 1961. only the view of the authors. The European Commission 16 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ [8] J. Picone, A. Hedin, D. P. Drob, A. Aikin, NRLMSISE-00 on spacecraft materials, Journal of Materials Science 30 (1995) Empirical Model of the Atmosphere: Statistical Comparisons 281{307. doi:10.1007/BF00354389. and Scienti c Issues, Journal of Geophysical Research 107 (2002). [26] B. A. Banks, S. K. Miller, K. K. de Groh, Low Earth Orbital doi:10.1029/2002JA009430. Atomic Oxygen Interactions with Materials, in: 2nd Interna- [9] K. Moe, M. M. Moe, S. D. Wallace, Improved Satellite Drag tional Energy Conversion Engineering Conference, American Coecient Calculations from Orbital Measurements of Energy Institute of Aeronautics and Astronautics (AIAA), Providence, Accommodation, Journal of Spacecraft and Rockets 35 (1998) RI, 2004. doi:10.2514/6.2004-5638. 266{272. doi:10.2514/2.3350. [27] S. Samwel, Low Earth Orbital Atomic Oxygen Erosion E ect [10] B. R. Bowman, K. Moe, Drag Coecient Variability at on Spacecraft Materials, Space Research Journal 7 (2014) 1{13. 175-500km from the Orbit Decay Analyses of Spheres, in: doi:10.3923/srj.2014.1.13. AAS/AIAA Astrodynamics Specialists Conference, American [28] I. Harrison, G. Swinerd, A free molecule aerodynamic investi- Astronautical Society (AAS), Lake Tahoe, CA, 2005. gation using multiple satellite analysis, Planetary and Space [11] E. K. Sutton, Normalized Force Coecients for Satellites with Science 44 (1996) 171{180. doi:10.1016/0032-0633(95)00077-1. Elongated Shapes, Journal of Spacecraft and Rockets 46 (2009) [29] V. J. Murray, M. D. Pilinski, E. J. Smoll, M. Qian, T. K. Minton, 112{116. doi:10.2514/1.40940. S. M. Madzunkov, M. R. Darrach, Gas{Surface Scattering Dy- [12] S. Andrews, L. Berthoud, Characterising satellite aerody- namics Applied to Concentration of Gases for Mass Spectrometry namics in Very Low Earth Orbit inclusive of ion thruster in Tenuous Atmospheres, The Journal of Physical Chemistry C plume-thermosphere/ionosphere interactions, Acta Astro- 121 (2017) 7903{7922. doi:10.1021/acs.jpcc.7b00456. nautica 170 (2020) 386{396. URL: https://linkinghub. [30] K. Moe, M. M. Moe, Gas-Surface Interactions in Low-Earth elsevier.com/retrieve/pii/S0094576519314675. doi:10.1016/ Orbit, in: 27th International Symposium on Rare ed Gas j.actaastro.2019.12.034. Dynamics, AIP Conference Proceedings, volume 1333, American [13] E. Doornbos, Thermospheric Density and Wind Determi- Institute of Physics, Paci c Grove, CA, 2011, pp. 1313{1318. nation from Satellite Dynamics, Springer Theses, Springer doi:10.1063/1.3562825. Berlin Heidelberg, Berlin, Heidelberg, 2012. doi:10.1007/ [31] J. Gregory, P. Peters, A Measurement of the Angular Distri- 978-3-642-25129-0. bution of 5eV Atomic Oxygen Scattered o a Solid Surface in [14] S. A. Schaaf, P. L. Chambre, Flow of rare ed gases, in: Fun- Earth Orbit, Rare ed Gas Dynamics 15 (1987). damental of Gas Dynamics, volume III, Princeton University [32] B. A. Banks, K. K. de Groh, S. K. Miller, MISSE Scattered Press, Princeton, NJ, 1958, pp. 687{739. Atomic Oxygen Characterization Experiment, TM-2006-214355, [15] R. Schamberg, A New Analytic Representation of Surface Inter- NASA Glenn Research Centre, Cleveland, OH, 2006. action for Hyperthermal Free Molecule Flow with Applications [33] B. A. Banks, K. K. de Groh, S. K. Miller, D. L. Waters, J. I. to Neutral-particle Drag Estimates of Satellites, Research mem- Kleiman, Lessons Learned from Atomic Oxygen Interaction orandum, Rand Corporation, 1959. with Spacecraft Materials in Low Earth Orbit, AIP Conference [16] E. M. Gaposchkin, Calculation of Satellite Drag Coecients, Proceedings 1087 (2009) 312{325. doi:10.1063/1.3076845. Technical Report 998, Lincoln Laboratory, MIT, Lexington, MA, [34] R. C. Blanchard, Rare ed Flow Lift-to-Drag Measurements of 1994. the Shuttle Orbiter, in: 15th ICAS Congress, American Institute [17] J. Storch, Aerodynamic Disturbances on Spacecraft in Free- of Aeronautics and Astronautics (AIAA), London, UK, 1986. Molecular Flow, TR-2003(3397)-1, The Aerospace Corporation, [35] R. C. Blanchard, J. Y. Nicholson, Orbiter Rare ed-Flow Reentry El Segundo, CA, 2002. Measurements from the OARE on STS-62, TM-110182, NASA, [18] D. Mostaza-Prieto, B. P. Graziano, P. C. Roberts, Spacecraft Hampton, VA, 1995. drag modelling, Progress in Aerospace Sciences 64 (2014) 56{65. [36] K. Moe, Absolute Atmospheric Densities Determined from doi:10.1016/j.paerosci.2013.09.001. the Spin and Orbital Decays of Explorer VI, Planetary and [19] S. Livadiotti, N. H. Crisp, P. C. Roberts, S. D. Worrall, V. T. Space Science 14 (1966) 1065{1075. doi:10.1016/0032-0633(66) Oiko, S. Edmondson, S. J. Haigh, C. Huyton, K. L. Smith, 90022-5. L. A. Sinpetru, B. E. Holmes, J. Becedas, R. M. Dom nguez, [37] K. Moe, B. R. Bowman, The E ects of Surface Composition V. Canas, ~ S. Christensen, A. Mlgaard, J. Nielsen, M. Bis- and Treatment on Drag Coecient of Spherical Satellites, in: gaard, Y.-A. Chan, G. H. Herdrich, F. Romano, S. Fasoulas, AAS/AIAA Astrodynamics Specialists Conference, American C. Traub, D. Garcia-Alminana, ~ S. Rodriguez-Donaire, M. Sureda, Astronautical Society (AAS), Lake Tahoe, CA, 2005. D. Kataria, B. Belkouchi, A. Conte, J. S. Perez, R. Vil- [38] C. Pardini, W. K. Tobiska, L. Anselmo, Analysis of the orbital lain, R. Outlaw, A review of gas-surface interaction models decay of spherical satellites using di erent solar ux proxies and for orbital aerodynamics applications, Progress in Aerospace atmospheric density models, Advances in Space Research 37 Sciences 119 (2020) 100675. URL: https://linkinghub. (2006) 392{400. doi:10.1016/j.asr.2004.10.009. elsevier.com/retrieve/pii/S0376042120300877. doi:10.1016/ [39] M. D. Pilinski, B. M. Argrow, S. E. Palo, Drag Coecients j.paerosci.2020.100675. of Satellites with Concave Geometries: Comparing Models and [20] C. Cercignani, M. Lampis, Kinetic models for gas-surface in- Observations, Journal of Spacecraft and Rockets 48 (2011) teractions, Transport Theory and Statistical Physics 1 (1971) 312{325. doi:10.2514/1.50915. 101{114. doi:10.1080/00411457108231440. [40] B. Ching, D. Hickman, J. Straus, E ects of atmospheric winds [21] R. Lord, Some extensions to the Cercignani{Lampis gas{surface and aerodynamic lift on the inclination of the orbit of the S3-1 scattering kernel, Physics of Fluids A: Fluid Dynamics 3 (1991) satellite, Journal of Geophysical Research 82 (1977) 1474{1480. 706{710. doi:10.1063/1.858076. doi:10.1029/JA082i010p01474. [22] R. Lord, Some further extensions of the Cercignani{Lampis [41] C. Pardini, L. Anselmo, K. Moe, M. M. Moe, Drag and energy gas{surface interaction model, Physics of Fluids 7 (1995) 1159{ accommodation coecients during sunspot maximum, Advances 1161. doi:10.1063/1.868557. in Space Research 45 (2010) 638{650. doi:10.1016/j.asr.2009. [23] A. Walker, P. M. Mehta, J. Koller, Drag Coecient Model Using 08.034. the Cercignani{Lampis{Lord Gas{Surface Interaction Model, [42] A. Macario-Rojas, K. L. Smith, N. H. Crisp, P. C. Roberts, Journal of Spacecraft and Rockets 51 (2014) 1544{1563. doi:10. Atmospheric interaction with nanosatellites from observed or- 2514/1.A32677. bital decay, Advances in Space Research 61 (2018) 2972{2982. [24] M. M. Moe, S. D. Wallace, K. Moe, Re nements in determining doi:10.1016/j.asr.2018.02.022. satellite drag coecients - Method for resolving density discrep- [43] E. Doornbos, H. Klinkrad, Modelling of space weather e ects on ancies, Journal of Guidance, Control, and Dynamics 16 (1993) satellite drag, Advances in Space Research 37 (2006) 1229{1239. 441{445. doi:10.2514/3.21029. doi:10.1016/j.asr.2005.04.097. [25] M. R. Reddy, Review e ect of low earth orbit atomic oxygen [44] P. M. Mehta, A. C. Walker, E. K. Sutton, H. C. Godinez, New 17 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ density estimates derived using accelerometers on board the [61] D. Mostaza-Prieto, P. C. Roberts, Methodology to Analyze At- CHAMP and GRACE satellites, Space Weather 15 (2017) 558{ titude Stability of Satellites Subjected to Aerodynamic Torques, 576. doi:10.1002/2016SW001562. Journal of Guidance, Control, and Dynamics 39 (2016) 437{449. [45] J. Virgili Llop, P. C. Roberts, Dsat, a QB50 CubeSat mission doi:10.2514/1.G001481. to study rare ed-gas drag modelling, Acta Astronautica 89 [62] F. Landis Markley, J. L. Crassidis, Fundamentals of Space- (2013) 130{138. doi:10.1016/j.actaastro.2013.04.006. craft Attitude Determination and Control, volume 33 of [46] D. Gerhardt, M. Bisgaard, L. Alminde, R. Walker, M. A. Fer- Space Technology Library, 1 ed., Springer-Verlag, New nandez, A. Latiri, J.-L. Issler, GOMX-3 : Mission Results from York, NY, 2014. URL: http://link.springer.com/10.1007/ the Inaugural ESA In-Orbit Demonstration CubeSat, in: 30th 978-1-4939-0802-8. doi:10.1007/978-1-4939-0802-8. Annual AIAA/USU Conference on Small Satellites, American [63] F. Lemoine, S. Kenyon, J. Factor, R. Trimmer, N. Pavlis, Institute of Aeronautics and Astronautics (AIAA), Logan, UT, D. Chinn, C. Cox, S. Klosko, S. Luthcke, M. Torrence, 2016. Y. Wang, R. Williamson, E. Pavlis, R. Rapp, T. Olson, [47] R. Holst, J. Nielsen, D. Gerhardt, J. A. G. Ahumada, Attitude The Development of the Joint NASA GSFC and the Na- and Orbit Control Results of the GOMX-4 Tandem CubeSat tional Imagery and Mapping Agency (NIMA) Geopotential Mission, in: 69th International Astronautical Congress, Inter- Model EGM96, NASA/TP|1998{206861 July, NASA God- national Astronautical Federation (IAF), Bremen, Germany, dard Space Flight Center, Greenbelt, MD, 1998. URL: http:// 2018. link.springer.com/10.1007/978-3-662-03482-8{_}62. doi:10. [48] O. Montenbruck, M. Garcia-Fernandez, J. Williams, Per- 1007/978-3-662-03482-8_62. formance comparison of semicodeless GPS receivers for [64] N. K. Pavlis, S. A. Holmes, S. C. Kenyon, J. K. Factor, The LEO satellites, GPS Solutions 10 (2006) 249{261. URL: development and evaluation of the Earth Gravitational Model http://link.springer.com/10.1007/s10291-006-0025-9. 2008 (EGM2008), Journal of Geophysical Research: Solid Earth doi:10.1007/s10291-006-0025-9. 117 (2012). doi:10.1029/2011JB008916. [49] O. Montenbruck, M. Markgraf, M. Garcia-Fernandez, [65] C. C. Finlay, S. Maus, C. Beggan, T. Bondar, A. Chambo- A. Helm, GPS for Microsatellites { Status and Per- dut, T. Chernova, A. Chulliat, V. Golovkov, B. Hamilton, spectives, in: R. Sandau, H.-P. R oser, A. Valenzuela M. Hamoudi, R. Holme, G. Hulot, W. Kuang, B. Langlais, (Eds.), Small Satellites for Earth Observation, Springer V. Lesur, F. Lowes, H. Luhr,  S. Macmillan, M. Mandea, Netherlands, Dordrecht, 2008, pp. 165{174. URL: http:// S. McLean, C. Manoj, M. Menvielle, I. Michaelis, N. Olsen, link.springer.com/10.1007/978-1-4020-6943-7{_}15. doi:10. J. Rauberg, M. Rother, T. Sabaka, A. Tangborn, L. T ner- 1007/978-1-4020-6943-7_15. Clausen, E. Th ebault, A. Thomson, I. Wardinski, Z. Wei, [50] O. Montenbruck, P. Swatschina, M. Markgraf, S. Santan- T. Zvereva, International Geomagnetic Reference Field: The drea, J. Naudet, E. Tilmans, Precision spacecraft navi- eleventh generation, Geophysical Journal International 183 gation using a low-cost GPS receiver, GPS Solutions 16 (2010) 1216{1230. doi:10.1111/j.1365-246X.2010.04804.x. (2012) 519{529. URL: http://link.springer.com/10.1007/ [66] A. Hedin, E. Fleming, A. Manson, F. Schmidlin, S. Avery, s10291-011-0252-6. doi:10.1007/s10291-011-0252-6. R. Clark, S. Franke, G. Fraser, T. Tsuda, F. Vial, R. Vincent, [51] O. Montenbruck, A. Hauschild, R. B. Langley, C. Siemes, Empirical wind model for the upper, middle and lower atmo- CASSIOPE orbit and attitude determination using commercial sphere, Journal of Atmospheric and Terrestrial Physics 58 (1996) o -the-shelf GPS receivers, GPS Solutions 23 (2019) 114. 1421{1447. doi:10.1016/0021-9169(95)00122-0. URL: https://doi.org/10.1007/s10291-019-0907-2http: [67] D. P. Drob, J. Emmert, G. Crowley, J. Picone, G. Shepherd, //link.springer.com/10.1007/s10291-019-0907-2. W. Skinner, P. Hays, R. Niciejewski, M. Larsen, C. She, J. W. doi:10.1007/s10291-019-0907-2. Meriwether, G. Hernandez, M. Jarvis, D. Sipler, C. Tepley, [52] NASA, Spacecraft Aerodynamic Torques, SP-8050 January, M. O'Brien, J. Bowman, Q. Wu, Y. Murayama, S. Kawamura, NASA, 1971. I. Reid, R. Vincent, An empirical model of the Earth's horizontal [53] D. A. Vallado, Fundamentals of Astrodynamics and Applications, wind elds: HWM07, Journal of Geophysical Research: Space 4 ed., Microcosm Press/Springer, Hawthorne, CA, 2013. Physics 113 (2008) 1{18. doi:10.1029/2008JA013668. [54] D. Mostaza-Prieto, Characterisation and Applications of Aero- [68] J. Virgili Llop, P. C. Roberts, Z. Hao, Aerodynamic Attitude dynamic Torques on Satellites, Phd thesis, The University of and Orbit Control Capabilities of The Dsat CubeSat, in: 37th Manchester, 2017. Annual AAS Guidance and Control Conference, AAS 14-063, [55] G. A. Bird, Molecular Gas Dynamics and the Direct Simulation American Astronautical Society (AAS), Breckenridge, CO, 2014. of Gas Flows, Clarendon Press, Oxford, UK, 1994. [69] M. L. Psiaki, Nanosatellite Attitude Stabilization Using Pas- [56] G. LeBeau, A parallel implementation of the direct simula- sive Aerodynamics and Active Magnetic Torquing, Jour- tion Monte Carlo method, Computer Methods in Applied nal of Guidance, Control, and Dynamics 27 (2004) 347{355. Mechanics and Engineering 174 (1999) 319{337. doi:10.1016/ doi:10.2514/1.1993. S0045-7825(98)00302-8. [70] J. Auret, W. H. Steyn, Design of an Aerodynamic Attitude [57] G. LeBeau, F. Lumpkin III, Application highlights of the DSMC Control System for a Cubesat, 62nd International Astronautical Analysis Code (DAC) software for simulating rare ed ows, Congress (2011). Computer Methods in Applied Mechanics and Engineering 191 [71] V. Sarychev, S. Mirer, A. Degtyarev, E. Duarte, Investigation (2001) 595{609. doi:10.1016/S0045-7825(01)00304-8. of equilibria of a satellite subjected to gravitational and aerody- [58] C. White, M. Borg, T. Scanlon, S. Longshaw, B. John, D. Emer- namic torques, Celestial Mechanics and Dynamical Astronomy son, J. Reese, dsmcFoam+: An OpenFOAM based direct simu- 97 (2007) 267{287. doi:10.1007/s10569-006-9064-3. lation Monte Carlo solver, Computer Physics Communications [72] R. R. Kumar, D. D. Mazanek, M. L. Heck, Simulation and 224 (2018) 22{43. doi:10.1016/j.cpc.2017.09.030. Shuttle Hitchhiker validation of passive satellite aerostabilization, [59] C.-D. Munz, M. Auweter-Kurtz, S. Fasoulas, A. Mirza, P. Or- Journal of Spacecraft and Rockets 32 (1995) 806{811. doi:10. twein, M. Pfei er, T. Stindl, Coupled Particle-In-Cell and 2514/3.26688. Direct Simulation Monte Carlo method for simulating reactive [73] M. R. Drinkwater, R. Haagmans, D. Muzi, A. Popescu, plasma ows, Comptes Rendus M ecanique 342 (2014) 662{670. R. Floberghagen, M. Kern, M. Fehringer, The GOCE Gravity doi:10.1016/j.crme.2014.07.005. Mission: ESA'S First Core Earth Explorer, in: 3rd International [60] S. Fasoulas, C.-D. Munz, M. Pfei er, J. Beyer, T. Binder, S. Cop- GOCE User Workshop, SP-627, European Space Agency (ESA), plestone, A. Mirza, P. Nizenkov, P. Ortwein, W. Reschke, Com- Frascati, Italy, 2007, pp. 1{7. doi:ISBN92-9092-938-3. bining particle-in-cell and direct simulation Monte Carlo for the [74] T. H. Stengle, MagSat Attitude Dynamics and Control: Some simulation of reactive plasma ows, Physics of Fluids 31 (2019) Observations and Explanations, in: J. Teles (Ed.), Fiifth Annual 072006. doi:10.1063/1.5097638. Flight Mechanics/Estimation Theory Symposium, Greenbelt, 18 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ MD, 1980. [75] B. Tossman, F. Mobley, G. Fountain, K. He ernan, J. Ray, C. Williams, MAGSAT attitude control system design and performance, in: Guidance and Control Conference, American Institute of Aeronautics and Astronautics (AIAA), Danvers, MA, 1980. doi:10.2514/6.1980-1730. [76] M. L. Gargasz, Optimal Spacecraft Attitude Control Using Aero- dynamic Torques, Msc thesis, Air Force Institute of Technology, [77] M. Eberhart, S. L ohle, A. Steinbeck, T. Binder, S. Fasoulas, Measurement of atomic oxygen in the middle atmosphere using solid electrolyte sensors and catalytic probes, Atmo- spheric Measurement Techniques 8 (2015) 3701{3714. URL: https://amt.copernicus.org/articles/8/3701/2015/. doi:10. 5194/amt-8-3701-2015. [78] R. Verker, A. Bolker, Y. Carmiel, I. Gouzman, E. Grossman, T. K. Minton, S. Remaury, Ground testing of an on-orbit atomic oxygen ux and ionizing radiation dose sensor based on material degradation by the space environment, Acta Astronautica 173 (2020) 333{343. URL: https://doi.org/10.1016/j.actaastro. 2020.04.065. doi:10.1016/j.actaastro.2020.04.065. [79] D. H. Hathaway, L. A. Upton, Predicting the amplitude and hemispheric asymmetry of solar cycle 25 with surface ux trans- port, Journal of Geophysical Research: Space Physics 121 (2016) 10,744{10,753. doi:10.1002/2016JA023190. [80] R. Cameron, J. Jiang, M. Schussler,  Solar Cycle 25: Another Moderate Cycle?, The Astrophysical Journal 823 (2016) L22. doi:10.3847/2041-8205/823/2/L22. [81] P. Bhowmik, D. Nandy, Prediction of the strength and timing of sunspot cycle 25 reveal decadal-scale space environmental conditions, Nature Communications 9 (2018) 5209. doi:10.1038/ s41467-018-07690-0. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Physics arXiv (Cornell University)

In-Orbit Aerodynamic Coefficient Measurements using SOAR (Satellite for Orbital Aerodynamics Research)

Download PDF
19 pages

Loading next page...
 
/lp/arxiv-cornell-university/in-orbit-aerodynamic-coefficient-measurements-using-soar-satellite-for-xSTomhSn4a
ISSN
0094-5765
eISSN
ARCH-3341
DOI
10.1016/j.actaastro.2020.12.024
Publisher site
See Article on Publisher Site

Abstract

The Satellite for Orbital Aerodynamics Research (SOAR) is a CubeSat mission, due to be launched in 2021, to investigate the interaction between di erent materials and the atmospheric ow regime in very low Earth orbits (VLEO). Improving knowledge of the gas-surface interactions at these altitudes and identi cation of novel materials that can minimise drag or improve aerodynamic control are important for the design of future spacecraft that can operate in lower altitude orbits. Such satellites may be smaller and cheaper to develop or can provide improved Earth observation data or communications link-budgets and latency. In order to achieve these objectives, SOAR features two payloads: i) a set of steerable ns which provide the ability to expose di erent materials or surface nishes to the oncoming ow with varying angle of incidence whilst also providing variable geometry to investigate aerostability and aerodynamic control; and ii) an ion and neutral mass spectrometer with time-of- ight capability which enables accurate measurement of the in-situ ow composition, density, velocity. Using precise orbit and attitude determination information and the measured atmospheric ow characteristics the forces and torques experienced by the satellite in orbit can be studied and estimates of the aerodynamic coecients calculated. This paper presents the scienti c concept and design of the SOAR mission. The methodology for recovery of the aerodynamic coecients from the measured orbit, attitude, and in-situ atmospheric data using a least-squares orbit determination and free-parameter tting process is described and the experimental uncertainty of the resolved aerodynamic coecients is estimated. The presented results indicate that the combination of the satellite design and experimental methodology are capable of clearly illustrating the variation of drag and lift coecient for di ering surface incidence angle. The lowest uncertainties for the drag coecient measurement are found at approximately 300 km, whilst the measurement of lift coecient improves for reducing orbital altitude to 200 km. Keywords: Orbital Aerodynamics; Drag and Lift Coecient; Gas-Surface Interactions; Thermospheric Wind; CubeSat. 1. Introduction novel attitude and orbit control manoeuvres using the aero- dynamic forces and torques that can be generated at these The Satellite for Orbital Aerodynamics Research (SOAR) altitudes. is a scienti c CubeSat mission due to be launched in 2021 The SOAR mission is a key component of the Horizon and designed to investigate the interactions between the 2020 funded DISCOVERER project [1, 2] that aims to atmospheric ow regime in very low Earth orbits (VLEO) radically redesign Earth observation satellites for sustained and di erent materials. Secondary objectives of the SOAR operation at signi cantly lower altitudes. The experiments mission are to provide new in-situ measurements of the performed by SOAR aim to improve knowledge and under- atmospheric density and composition and variation of the standing of the gas-surface interactions (GSIs) at VLEO thermospheric wind velocity over the range of altitudes altitudes and provide valuable validation data for ground- below approximately 400 km. SOAR will also demonstrate based experiments on materials and GSIs which will be performed in the ROAR (Rare ed Orbital Aerodynamics Research) facility at The University of Manchester. The Email address: nicholas.crisp@manchester.ac.uk ROAR Facility is a unique experimental set-up that is (N.H. Crisp) designed to identify novel materials for satellite applica- Preprint submitted to Acta Astronautica arXiv:2012.07407v2 [physics.space-ph] 17 Dec 2020 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Nomenclature A Total Surface Area T Free-stream temperature T 1 A Reference area T Surface (wall) temperature ref C Force coecient F v Relative atmospheric ow velocity rel C Torque coecient T x  Linear acceleration F Force Thermal (energy) accommodation coecient I Moment of inertia Normal energy accommodation coecient l Reference length  ref Rotational acceleration m Mass Atmospheric density s Molecular speed ratio Tangential momentum accommodation coe- T Torque cient tions with a focus on improved aerodynamic properties and of the expansion and contraction of the atmosphere with atomic oxygen (AO) resistance. The facility is principally the di erent diurnal, seasonal, and solar cycles. comprised of a ultra-high vacuum (UHV) environment, In VLEO the atmosphere is signi cantly less dense an AO source capable of providing representative orbital than at the ground or conventional ight altitudes and velocities and surface interactions, and a sensor suite in- is considered to be rare ed such that the mechanics of cluding ion and neutral mass spectrometers (INMS) which continuum ow regimes can no longer be applied. The enable measurement and characterisation of the incident non-dimensional Knudsen number can be used to classify and re-emitted gas- ow on sample materials [3, 4]. di erent ow-regimes and is de ned as the ratio between the Improvements in the knowledge and understanding of mean free path (the average distance between successive gas the GSIs and identi cation of novel materials that can particle to gas particle or gas particle to surface collisions) reduce atmospheric drag, improve aerodynamic control in a ow and a characteristic physical length (e.g. the length capability, or increase aerodynamic intake eciencies are of a body in that ow). When the Knudsen number is high important steps in enabling the sustained operation of (i.e. Kn  10) the gas-surface interactions along the length spacecraft at lower orbital altitudes. This reduction in of a body are of much greater signi cance than any gas orbital altitude has been linked to numerous bene ts, for particle to gas particle interactions, including those with example reduced debris collision risk, a more favourable ra- re ected particles [7]. This regime is termed free-molecular diation environment, and aerodynamics-assisted end-of-life ow (FMF). The variation of the Knudsen number with disposal. The opportunity to incorporate novel technolo- altitude is given in Fig. 1. The lower bound of the VLEO gies such as atmosphere-electric propulsion (ABEP) and range can be de ned as the ow enters the more complex aerodynamic attitude and orbit control is also presented. transitional regime (Kn < 10), and the conditions of free- For Earth observation applications, lower altitude orbits molecular ow cannot be assumed. This is shown to occur o er the possibility of smaller and less expensive platforms, for altitudes below approximately 130 km altitude. leading to cheaper data products, or alternatively higher In the FMF regime, the forces which act on a body resolution imagery, both with a wide range of potential can be determined by simply considering the interaction commercial, environmental, and societal impact [5]. Com- between the incident molecules and satellite surfaces, and munications satellites may correspondingly bene t in their the subsequent angular distribution and velocity of the design from improved link-budgets, lower latency, and in- re-emitted or re ected particles. It has been observed that creased frequency re-use [6]. these GSIs, and the associated momentum and energy transfer, are dependent on surface roughness and cleanli- 1.1. Gas-Surface Interactions in Very Low Earth Orbit ness (particularly related to altitude-dependent AO adsorp- tion), surface composition and lattice structure, surface The upper bound of the VLEO range can be broadly temperature, gas composition, and the incident particle de ned as the altitude below which the atmosphere begins temperature, velocity, and incidence angle [9{11]. The to have a signi cant e ect on the orbital and attitude presence of ionised thruster plumes may also a ect the dynamics of a spacecraft and is typically de ned at 450 km local ow conditions and therefore the aerodynamic forces altitude. However, this de nition is somewhat fuzzy as in produced [12]. reality the the atmospheric density can vary considerably Models for these GSIs have been developed to enable at this altitude (as shown in Fig. 1) principally as a result 2 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Figure 2: Variation of drag and lift force coecients of a single-sided at-plate of area 1 m under di erent GSI models and inputs assumptions for FMF conditions (s = 10, T = 300 K, T = 600 K). Aerodynamic coecients are referred to the projected (cross-sectional) area with respect to the oncoming ow. [7] assumes di use re-emission of particles, whilst modi ed analytical equations (from Schaaf and Chambre [14]) based on the Cercignani-Lampis-Lord (CLL) model [20{22] as Figure 1: Variation of density (top) and free-stream Knudsen proposed by Walker et al. [23] can be used to represent spec- number (bottom) with altitude for di erent levels of solar activity, ular re ections. For di use re-emission but reducing energy assuming a characteristic length of 1 m and atmospheric parameters accommodation the drag force increases as the incidence calculated using the NRLMSISE-00 model [8]. Average kinetic approaches normal to the ow. Meanwhile, the lift force diameter is weighted by the number density of the atmospheric species at each altitude (values for AO and N assumed to be increases modestly for inclined surfaces to a maximum at conservatively equivalent to N ). approximately 45°. However, if quasi-specular re-emission or specular re ection properties are exhibited the drag can be signi cantly reduced for shallow incidence angles estimation and determination of the aerodynamic forces (<45°) and will increase as the incidence approaches normal which act on surfaces in these conditions. These models to the ow. Lift force generation can also be increased are used for the purpose of orbit and attitude simulation, signi cantly. spacecraft design and modelling, and in the development of Studies of in-orbit GSI performance have shown that atmospheric density and thermospheric wind models from materials commonly utilised on spacecraft have exhibited on-orbit observations [13]. Popular GSI models used in primarily di use re-emission properties with high energy the eld of orbit aerodynamics include those of Sentman accommodation (  0:8 to 1:0), particularly in low alti- [7], Schaaf and Chambre [14], Schamberg [15], Gaposchkin tude orbits where surface contamination (principally by [16], Storch [17]. Comparison and review of these models adsorbed atomic oxygen) is high [9, 24]. The prevalence of is provided by Mostaza-Prieto et al. [18], Livadiotti et al. energetic and highly-reactive atomic oxygen in low altitude [19]. orbits also introduces the issue of material erosion [25{27] In general, the re-emitted or re ected particle distribu- that can further increase accommodation and therefore re- tion is described as di use or specular with some models sult in di use re-emission. However, evidence of increasing using combinations of these de nitions. Coecients that quasi-specular re-emission behaviour has been observed for de ne the range of energy or momentum accommodation materials on spacecraft in higher altitude orbits (800 km at the surface are typically used to characterise the GSI to 1000 km) [28] where surface contamination is lower and performance given an assumed re-emission distribution and in elliptical orbits where the incident kinetic energy near therefore the forces experienced by the surface. perigee is greater [24]. Ground-based molecular beam Fig. 2 demonstrates the e ect of di erent GSI model experiments have also demonstrated such quasi-specular assumptions and given parameters on the drag and lift qualities for clean materials under UHV conditions and at force coecients for a at-panel surface. Sentman's model 3 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ energies approaching that of orbit velocity [29]. by the abundance of GSI models, but lack of consensus For the purpose of improving aerodynamic performance on their suitability and application for di erent materials, in the VLEO regime, quasi-specularly re ecting materi- surface treatment, altitude range, and period of the solar als in combination with appropriate satellite geometric cycle [44]. design, would provide the ability to reduce aerodynamic The unique combination of a test satellite (SOAR) and drag and therefore increase orbital lifetime or reduce the an experimental ground facility (ROAR) aims to improve requirements for drag compensating propulsion systems. the knowledge of GSIs and the underlying physical mecha- Alternatively, using the increased drag generated at high- nisms, leading to improved modelling of aerodynamic forces incidence angles, enhanced aerodynamics-based deorbit in VLEO. A systematic investigation of di erent materials devices could be conceived. The capability to produce lift will also seek to identify those that can provide improved forces of greater magnitude also provides the possibility aerodynamic performance through specular re ection prop- to utilise new methods of aerodynamics-based orbit and erties and have atomic oxygen erosion resistance, enabling attitude control. a new class of spacecraft that can operate sustainably at lower orbital altitudes. 1.2. On-Orbit Investigations of Gas-Surface Interactions A number of investigations of material GSI performance 2. Satellite Design and surface accommodation in the FMF regime have been performed using direct on-orbit measurements and ground- The principal scienti c objective of SOAR is to in- based observations of spacecraft. Review and comparison vestigate the variation of the aerodynamic coecients of of studies in this area have been provided by Moe et al. di erent materials and surface nish at di erent incidence [9], Moe and Moe [30]. angle to the oncoming ow and at di erent orbital altitudes. Direct measurement of the remission angle of scattered In-situ measurement of the incident ow environment will AO from a vitreous carbon surface was studied on the be used in addition to measured attitude and orbital param- STS-8 Space Shuttle ight. The di use remission spectrum eters to determine the forces and torques experienced by observed, approaching a cosine distribution, indicated that the body. By providing in-situ density measurements of the almost full accommodation was occurring at the surface oncoming ow which can be used directly in the recovery of [31]. Investigation of scattering angle from an oxidised the tted aerodynamic coecients and associated accommo- aluminium surface has also been noted as part of a larger dation coecients, this experimental methodology presents study on erosion characteristics of scattered AO which was a signi cant advantage over previous observation-based conducted on MISSE-FF (Materials International Space studies. Station Experiment Flight Facility) by Banks et al. [32, 33]. SOAR takes the form of a 3U CubeSat developed from Aerodynamic coecients resulting from the summed e ect the Dsat design of Virgili Llop and Roberts [45], previ- of GSI over a spacecraft body have also been studied. ously proposed for the QB50 programme for lower thermo- For example, the aerodynamic coecients of the Space spheric exploration and research. The basic geometry of Shuttle were measured using accelerometer data during the SOAR is shown in Fig. 3. transitional re-entry phase [34, 35]. A set of four panels that unfold after launch and de- Other studies have used observational methods to de- ployment into orbit to extend away from the satellite body termine the aerodynamic coecients of di erent spacecraft and can be rotated with respect to the satellite body (and or materials from the attitude motion or orbital trajectory. the oncoming ow) have been designed to achieve proposed GSI and surface accommodation can subsequently be inves- investigation of material aerodynamic coecients and to tigated by considering the spacecraft attitude and geometry act as aerodynamic control surfaces. These appendages are and through comparison to di erent models. These stud- termed steerable ns herein. ies have notably included Paddlewheel [36] and spherical The surfaces of these steerable ns have been coated [10, 37{39] satellites, but have also included more complex with four di erent material coatings with the con guration geometries [40, 41] and predictions for time-varying atti- of similar materials placed on opposing surfaces as indicated tude where observed or measured data was not available in Fig. 4. Through coordinated rotations of the steerable [42]. However, in the absence of measured data, the results n, each material can therefore be individually exposed obtained using these methods are typically dependent on into the ow at varying angles of incidence (neglecting the modelled atmospheric densities and are therefore subject to body of the spacecraft and parallel surfaces). their inherent biases and uncertainties [43]. Furthermore, As described by Virgili Llop and Roberts [45], the steer- as some of the analysed spacecraft may also have been able ns can be operated in pairs in two principal ways; used during the development and calibration of the density co-rotation and counter-rotation. From the minimum drag models, some circular logic may be present [13]. con guration and under stable ow-pointing conditions, There remains both a lack of knowledge of the physical co-rotation of a single opposing-pair of the steerable ns mechanisms that control GSI behaviour in VLEO and how (see Fig. 4b) exposes a single material to the ow and will these apply to di erent materials and their interactions generate a net lift or side force and therefore a torque (in in the true orbit environment. This is further exhibited 4 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ (a) Minimum drag con guration (b) Maximum drag con guration Figure 3: Design of the Satellite for Orbital Aerodynamics Research (SOAR) with forward-facing ion and neutral mass spectrometer (INMS) and the steerable ns oriented in the two nominal aerostable con gurations. Table 1: Principal geometric and system parameters of SOAR. yaw for the vertical ns or pitch for the lateral ns). The spacecraft will therefore rotate to y at an angle to the ow. Property Value Contrastingly, counter-rotation of a pair of opposing ns (see Fig. 4a) can similarly expose a single material to the Mass [kg] 2:88 ow, but creates opposing lift forces from each steerable Length (in z-axis) (L ) [m] 0:366 n, resulting in no net side-force and a rolling moment that Total Surface Area (A ) [m ] 0:225 causes the spacecraft to spin up about the ow-pointing CoM (in z-axis from rear) [m] 0:161 8 9 8 9 direction. x 0:0392 < = < = In order to provide in-situ information about the ow Principal MoIs y [kg m ] 0:0392 : ; : ; conditions, including thermospheric winds, the spacecraft z 0:0288 features a forward-facing ion and neutral mass spectrometer Residual Magnetic Dipole [A m] 18 10 (INMS), labelled in Fig. 3. This sensor, improved since the RW Max Torque [N m] 23 10 development of the QB50 satellites, includes new time-of- RW Max Ang Momentum [N m s] 1:2 10 2 9 ight (ToF) capability, enabling assessment of the incoming RW Spin Axis MoI [kg m ] 694:5 10 ow velocity in addition to the total atmospheric density and ow composition. To maintain accuracy of the INMS instrument, the spacecraft must be pointed in the direction despite not having a star tracker, principally as a result of of the oncoming ow within a given angular range (see the improved gyroscope (IMU) performance. Table 2). Simply, this requires that the spacecraft nominally A NovAtel OEM719 GPS receiver provides the precise ies in an attitude that is closely aligned with the direction position (<1:5 m) and velocity (<0:03 m s ) of the space- of the ow. craft and removes dependency of the experiment on ground- Attitude control of the spacecraft is principally enabled based observational tracking information. The accuracy by a three-axis reaction wheel assembly (tetrahedral con- and performance of such miniature commercial-o -the-shelf guration of four wheels). A three-axis magnetorquer is (COTS) GPS receivers in LEO has been discussed [48, 49] also included to perform initial detumbling operations fol- and demonstrated in orbit, for example on the PROBA-2 lowing launch and to enable desaturation and momentum [50] and CASSIOPE satellites [51]. management of the reaction wheels. Further parameters of interest relating to the spacecraft Attitude determination for SOAR is provided by ne design are summarised in Table 1. sun sensors, a magnetometer, and a high-performance IMU (Epson M-G370). Using a unscented Kalman lter (UKF), 3. Experimental Methodology the combined sensor set is expected to provide an attitude knowledge with an expected uncertainty of less than 1° The primary scienti c objective of SOAR is to pro- (3-sigma) even during eclipse. This exceeds the attitude vide in-space measurements of the GSI characteristics of knowledge performance of the antecedent GOMX-3 satellite di erent materials and surface-coatings in the VLEO en- [46] and approaches that of the GOMX-4B [47] satellite, vironment. The steerable ns of SOAR can be used to 5 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ (a) Counter-rotated con guration of the lateral ns. (b) Co-rotated con guration of the vertical ns. Figure 4: Principal experimental con gurations of the steerable ns on SOAR showing the corresponding arrangement of the four di erent test materials. expose di erent materials to the oncoming ow at varying The force F can be associated with the dimensionless incidence angle and at di erent altitudes as the orbit of force coecients C using Eq. (1), which also expresses SOAR decays. the accelerations x as a function of the dynamic pressure The orbit trajectory and attitude of the spacecraft will of the surrounding ow and the spacecraft geometry. vary depending on the con guration of the steerable ns with respect to the oncoming ow. With knowledge of F = v A C = mx (1) ref F rel the ow conditions and spacecraft position/orientation where  is the local atmospheric density, v the space- over time, the aerodynamic forces and torques experienced rel craft velocity relative to the oncoming ow, and A the by the satellite can be estimated and linked to the GSI ref reference area. characteristics of the di erent surfaces exposed to the ow. Investigation of the drag coecient of di erent materi- Reconciliation of the force and moment coecients with the true nature of the GSI mechanics still requires a model als exposed to the ow by the steerable ns was proposed by Virgili Llop and Roberts [45] for the Dsat mission. In this for the exchange of energy and momentum of the gas species method, opposing steerable ns are counter-rotated, expos- with the surface and the associated particle re ection/re- ing the same material/coating to the oncoming ow, and emission pattern. However, experimental determination nominally producing no net lift/side-forces or pitch/yaw of the aerodynamic coecients provides valuable in-situ torques but only a net torque in roll. Thus, only an in- validation data for the ground-based material experiments, creased nominal drag force is generated by the panel area in particular those that are planned for the ROAR Facility. exposed to the ow and the associated drag coecient can be determined from the variation in the spacecraft 3.1. Drag Force Coecient trajectory over a period of time using the orbit determina- A body exposed to an oncoming ow will experience tion and free-parameter tting process described later in forces of an aerodynamic nature, the magnitude and direc- Section 3.3. tion of which will be dependent on the orientation of the On SOAR, both co-rotated and counter-rotated con gu- body with respect to the direction of the oncoming ow. rations of opposing steerable ns will be considered. Given This force is often decomposed into three mutually perpen- the con guration of the material coatings shown previously dicular forces in the body axes (axial, normal, and side) (Fig. 3), the steerable ns can be rotated independently to with associated coecients. Alternatively, the components expose a single material (on two opposing ns) into the on- of the force and coecients with respect to the oncoming coming ow to investigate the variation in drag coecient ow are considered; drag, lift, and a third mutually per- with incidence angle and at di erent altitudes. pendicular component (often referred to as side-force or The drag coecient for a given orbital altitude and sometimes cross-wind). The term lift, will be used herein to con guration of the steerable ns can subsequently be describe both force components perpendicular to the drag, recovered by considering the produced aerodynamic accel- allowing commonality in terminology due to the fourth or- eration of the spacecraft, expressed by Eq. (1). However, der rotational symmetry of the spacecraft about the z-axis it should be noted that the drag coecient determined by (see Fig. 3). 6 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ this method is representative of the whole spacecraft in the body of the spacecraft does not contribute any additional given con guration and not only the materials exposed to meaningful roll torques, the rolling moment coecient can the ow. be recovered by considering the evolution of attitude in During these experiments, the attitude control actuators the roll-axis of the spacecraft. Free-parameter tting of (principally the reaction wheels) will be used to maintain the rolling moment coecient from the attitude evolution the nominal pointing direction of the satellite into the on- of the spacecraft requires an attitude dynamics model coming ow direction. However, as the steerable ns will be including models for the torques which act on the spacecraft. displaced from the minimum or maximum drag condition The orbit trajectory and perturbation models used in the to expose the di erent materials to the ow, the spacecraft drag coecient analysis are also required to provide the may have reduced aerostability and experience disturbing correct spatial and temporal reference for the selected aerodynamic torques. Furthermore, the external environ- torque models. mental perturbations may not be periodic in nature and For a co-rotated con guration of opposing steerable will vary in magnitude depending on a number of factors ns a pitching or yawing torque will be produced. In the including the orbital position, spacecraft attitude, solar absence of a correcting control torque, this pitch or yaw environment, lighting conditions (sunlight/eclipse), and torque would cause the spacecraft to rotate (and oscillate altitude. For di erent experiments at di erent altitudes about) an equilibrium angle to the ow. By considering the stability of the spacecraft may only be maintained by the measured evolution of attitude in the pitch/yaw-axis of the ADCS for a certain period of time before actuator sat- the spacecraft, the pitch or yaw coecient for a co-rotated uration occurs. The magnitude of the aerodynamic forces, con guration without attitude correction can be recovered. spacecraft stability, and the length of the possible exper- However, if the attitude of the spacecraft is perturbed from imental period are critical in determining the expected the ow-pointing condition the accuracy of the INMS will performance of the investigation. These factors are ex- be compromised and uncertainty in the incidence of the plored and their impact on the experimental performance steerable panels to the ow will be increased. estimated and discussed in Section 5. Alternatively, the reaction wheels can be used to coun- teract the torque produced by the co-rotated steerable 3.2. Lift Force Coecient ns and thus attempt to maintain a close to ow-pointing attitude of the spacecraft. A true ow-pointing attitude Aerodynamic torques experienced by the spacecraft can cannot be realised as knowledge of the oncoming ow di- be described by Eq. (2) in which C is the aerodynamic rection would be required. Under these circumstances, the moment coecient set (typically roll C , pitch C , and yaw l m measured angular momentum in the reaction wheels rather C ) and l is an additional reference length (L is used n ref z than the motion of the spacecraft body may be used in herein, see Table 1). The torque can also be de ned by the free-parameter tting process to determine the pitch the force F and associated moment arm ` or the rotational  or yaw moment coecients. acceleration  and the corresponding moment of inertia In a controlled co-rotated con guration the lift force can matrix I [52]. also be considered directly through coordinated analysis of 2 the orbital trajectory of the spacecraft and simultaneous T = ` F = v A l C = I (2) ref ref M rel parameter tting of the drag coecient and the lift force The aerodynamic moment coecients of the satellite coecient. However, as the lift force of typical materials is can be investigated by analysis of the spacecraft attitude a fraction of the drag force (indicated by the di erence in response with the steerable ns con gured at di erent magnitude between the lift and drag coecients of di use incidence angles with respect to the ow. The lift force surfaces in Fig. 2), the ability to distinguish the lift force coecient of the di erent materials exposed to the ow can from the measured orbital data in the presence of other subsequently be recovered from the aerodynamic moment sources of perturbation and uncertainty is likely to be coecients by considering the spacecraft geometry and limited. angle of incidence of the ns. Experimental determination The selection of the most suitable method to investigate of the moment coecients of SOAR can be performed the lift force coecient will be dependent on the attitude using either counter-rotated or co-rotated steerable n and stability characteristics of the spacecraft in each con- con gurations. guration and the expected uncertainty which is associated A counter-rotated con guration of opposing steerable with the di erent experimental modes and subsequent data ns can be used to analyse the rolling moment coecient. processing. The equal but opposing lift forces produced by the opposing 3.3. Orbit Determination and Free-Parameter Fitting counter-rotated ns act as a couple to generate a net rolling torque on the spacecraft. The rolling moment coecient In order to recover the aerodynamic forces and torques can therefore be recovered by considering the evolution of experienced by the satellite, the orbital position and atti- the spacecraft attitude in roll. The lift force coecient of tude of the spacecraft during an experimental period can be the exposed surfaces can subsequently be determined by analysed. However, in addition to the aerodynamic forces decomposing the spacecraft geometry. Assuming that the and torques of interest, the satellite will experience other 7 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ external perturbations of varying magnitude, for example (f ) Update state vector and free-parameter. Repeat due to the non-spherical gravitational eld of the Earth, if RMS has not converged. solar radiation pressure, and residual magnetic dipole inter- 6. If converged, output state vector and free parameter actions. The aerodynamic forces and torques experienced are best- t for the observed data and the provided by the satellite cannot therefore be simply isolated from mathematical models (propagation method). the measured position and attitude data. The accuracy to which the aerodynamic coecients can An orbit determination algorithm can be used to de- be determined by such a method is primarily dependent on termine the drag coecient as a free-parameter (also seen the quality of the experimental data that can be obtained as the solve-for parameter) from the measured orbit po- during each test-run. To compare two di erent spacecraft sition data from a given experimental run and associated con gurations over a given period of time, it is necessary con guration of the steerable ns [45]. The same method that the measured trajectories (in orbit or attitude) can can be applied to perform combined orbit and attitude rst be distinguished from each other in the presence of determination to t and recover a moment coecient of sensor noise and other uncertainties. For a di erence in the spacecraft from the measured orbit position, attitude, generated force or torque by the spacecraft in two di erent and environmental data. con gurations this therefore imposes a minimum require- These methods compare the output of model-based sim- ment on the position measurement accuracy (using GPS) ulations (orbit/attitude propagation) to the data measured and ADCS (attitude determination and control system) on-orbit. Using iterative di erential correction, best- t measurement accuracy. aerodynamic coecient values can be found by a least- The delity of the mathematical model used in the orbit squares method that provides convergence between the determination process is also critical to the orbit determi- measured orbit or attitude trajectory of the spacecraft nation process and recovery of the free-parameter (force and the mathematical model of the corresponding motion. or moment coecient). In order to provide convergence Uncertainty in the observations can be accounted for by towards the measured trajectory, it is necessary that the updates to the initial state vector used for the modelled model incorporates the relevant perturbations with their trajectory at each iteration and di erences in the sensor spatial and temporal variations over the duration of the performance for di erent state variables (e.g. position and test-run. The selection of necessary perturbations and velocity) using weighting methods. modelling delity are related to the noise in the measured The implementation of this non-linear weighted least position, velocity, and attitude. Perturbations that would squares process [53] can be brie y summarised: cause variation in the trajectory of the spacecraft of simi- 1. Import state vectors of experimentally measured or- lar or smaller magnitude than the noise in the measured bital elements, attitude quaternions, and atmospheric values can be safely neglected, simplifying the form of the density for each time step. mathematical model. 2. Initialise numerical orbit propagation method. (a) Select propagation force and torque models. 4. Attitude Stability and Control (b) Initialise environmental models. (c) Initialise spacecraft geometric models. The presence and use of the steerable ns on SOAR 3. Set initial guess of the free-parameter (e.g. drag produces a number of di erent forces and torques which coecient). need to be carefully considered to ensure stability and 4. Set weighting matrix based on expected uncertainty pointing accuracy of the spacecraft throughout its lifetime. of measured state vector parameters (from sensor Interaction of the spacecraft with the residual atmosphere, performance). solar radiation, and the magnetic and non-spherical gravity 5. Begin iterative scheme: elds of the Earth must be considered. The ability to (a) Apply small modi cations to each initial state control the attitude and stability of the spacecraft using vector component ( nite- or central-di erencing) on-board actuators also requires investigation as the expe- based on a small percentage of value or as a rienced torques vary in relative magnitude with decreasing function of the weighting matrix. orbital altitude. (b) Calculate the orbit trajectory for each varia- The concept of aerostability is employed by SOAR to tion of the initial state vector using the orbit provide passive pointing towards the oncoming ow direc- propagation method. tion in orbit. This aerostability is provided by the steerable (c) Form the partial derivative matrix from di er- ns which are located towards the aft of the spacecraft ences between each propagated state vectors at and thus generate a restoring aerodynamic torque in pitch each time step. and/or yaw in response to any misalignment of ow direc- (d) Calculate update to the initial state vector and tion with the longitudinal axis of the spacecraft. When free-parameter. each steerable n is oriented parallel to the longitudinal (e) Calculate weighted root mean square (RMS) of body axis of the spacecraft (Fig. 3a) a minimum drag con- residuals (between current iteration and mea- guration is generated for the nominal spacecraft attitude. sured trajectory) 8 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Similarly, when the steerable ns are all oriented normal to the spacecraft longitudinal axis (Fig. 3b) the maximum drag con guration is achieved. The surface coatings applied to the steerable ns repre- sent a range of expected GSI performance from complete en- ergy accommodation and di use re-emission to incomplete accommodation and more specular re ection properties. However, as the GSI properties of these surfaces have not been wholly characterised, the true attitude performance and control capability of the satellite are also uncertain. The di erent materials may therefore result in the produc- tion of di erent forces and torques when exposed to the ow. Figure 5: Pitch/Yaw moment coecient of SOAR with varying angle of incidence with respect to the ow in the minimum Modelling of the aerodynamic coecients for SOAR has (steerable ns parallel to body) and maximum (steerable ns been performed using ADBSat [54], an analytical panel- perpendicular to body) drag con gurations. method tool that can implement di erent gas-surface inter- action models and features basic shadowing analysis. Using this tool, a database of aerodynamic coecients can be calculated from a CAD model of the spacecraft for di erent orientation angles with respect to the ow (angle of attack and sideslip) and di erent con gurations of the steerable ns. Sentman's model [7] for GSIs has been used in all anal- yses unless otherwise stated. This model assumes a fully di use re-emission pattern of particles with a Maxwellian velocity distribution dependent on the thermal energy ac- commodation coecient and surface (wall) temperature T . A default accommodation coecient of = 1 and wall temperature T = 300 K have been used unless otherwise Figure 6: Aerodynamic sti ness (static stability derivative) of stated. SOAR for varying steerable n angle with respect to the ow. Direct simulation Monte Carlo (DSMC), developed by Bird [55] and applied in a number of di erent software tools the torque generated by the interaction of the spacecraft (e.g. DS3V, DAC [56, 57], dsmcFoam+ [58], and PICLas geometry with the oncoming ow. [59, 60]), could have alternatively been used to calculate The static pitching/yawing moment coecient of SOAR the aerodynamic characteristics of the satellite geometry in in the minimum and maximum drag con gurations is pre- the VLEO environment. These methods are able to provide sented in Fig. 5. It can be noted that the symmetrical increased delity of the modelled ow, including features nature of the spacecraft about the roll axis allows for equiv- such as intermolecular collisions, chemical reactions, and alence in pitch and yaw for static analysis. The negative electromagnetic/electrostatic interactions. Furthermore, in slope of the pitch/yaw moment coecient with angle of DSMC more fundamental forms of the GSIs are generally incidence (angle of attack or sideslip respectively) indicates implemented rather than analytical expressions that use the aerostable nature of these con gurations. mean ow conditions and averaged interactions over at- The concept of aerodynamic stability derivatives, or plate elements. However, whilst these methods can over- aerodynamic sti ness, for spacecraft at orbital altitudes, come some of the shortfalls of panel methods (for example can be used to further investigate the expected attitude for complex and concave geometries), they are signi cantly behaviour for varying geometry and ight conditions [61]. more computationally intensive, particularly so when the The static pitch/yaw stability derivative C , can be calcu- variation of the aerodynamic coecients with geometric T lated from the gradient of C over a small range about the con guration, attitude, and altitude are needed. Thus, as T nominal attitude ( = 0). The variation in static pitch/yaw the geometry of SOAR is relatively simple and the ow stability derivative for steerable n angles over the range can be considered rare ed and free-molecular (for altitudes of minimum to maximum drag con gurations is shown in above 150 km, see Fig. 1), the panel method could be Fig. 6. The increase in stability derivative with increasing safely applied for the purpose of the analyses herein. incidence angle demonstrates that a greater static stability is achieved when a larger panel area is presented to the 4.1. Static Stability ow. The static stability provided by di erent con gurations The e ect of thermal accommodation coecient on of the steerable ns can be investigated by considering static stability is also shown in Fig. 6, indicating that 9 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ aerodynamic sti ness is shown to increase with decreasing thermal accommodation under the assumptions of Sent- man's GSI model. This e ect is more marked at shallow steerable n incidence angle with respect to the ow. In order to characterise the performance of di erent materials and surface-coatings in orbit, during the exper- imental periods only one pair of steerable ns will be ro- tated with respect to the ow at any given time. With this con guration a total of four materials or surfaces can be characterised during the mission, two per pair of opposing steerable ns. When a single pair of steerable ns is counter-rotated with the spacecraft pointing into the direction of the oncom- ing ow a net rolling torque is generated but no net pitch or yaw torques are created. However, if the relative direc- tion of the ow changes (for example due to atmospheric co-rotation or thermospheric winds) or the attitude of the satellite is perturbed, induced torques are generated due to a variation in the projected area of the counter-rotated ns to the ow. Under these conditions a net pitching torque arises due to the di erence in area presented to the Figure 7: Aerodynamic torque coecient response for SOAR with di erent counter-rotated or co-rotated steerable n con guration. ow by the rotated ns. The plot of torques for counter- rotated vertical ns is shown in Fig. 7 (top), demonstrating that induced torques in pitch due to angle of sideslip have provided. In these simulations the forces and torques asso- a positive gradient about the equilibrium, and are there- ciated with the Earth gravitational potential (EGM96/2008 fore disturbing rather than restoring. Furthermore, these [63, 64]), solar radiation pressure, residual magnetic dipole torques grow at a faster rate than the restoring torques interactions (IGRF-11 [65]), varying atmospheric density generated by the lateral ns due to the change in pitch (NRLMSISE-00 [8]), and thermospheric winds (HWM93/07 angle. If the ow is therefore o set with respect to the [66, 67]) are implemented. It is important to recall that spacecraft body in yaw the spacecraft responds with cou- the results presented herein remain subject to the assump- pled motion in the pitch axis as a result of the set angle of tions and limitations of the implemented GSI model and the steerable ns. Equivalent behaviour is demonstrated the input parameters used and therefore may still di er for counter-rotated lateral ns and an o set in angle of substantially from the true behaviour in orbit. attack. This e ect is termed pitch-yaw coupling henceforth. For rotationally symmetric con gurations the previous Co-rotation of a pair of opposing steerable ns gener- analyses showed that aerostability ensures that restoring ates a net torque in pitch or yaw, but no net torque in torques will be produced in response to changes in the roll. Fig. 7 (bottom) shows the torques in pitch and yaw oncoming ow direction. However, due to the FMF na- for a con guration in which the lateral ns are co-rotated, ture of the surrounding atmospheric environment, natural demonstrating a small bias in pitch torque when the space- damping of any generated angular velocity is not generated. craft is aligned with the direction of the oncoming ow. Therefore, given an external perturbation and without any However, as the pitch (angle of attack) is increased by a additional damping input, the spacecraft will begin to os- small amount (3°) the pitch torque crosses zero with a cillate. The frequency of this oscillation is dependent on negative gradient. The spacecraft therefore demonstrates the initial disturbance, stability derivative, and the envi- stability in pitch at this equilibrium angle with respect to ronmental conditions [61]. the ow. The spacecraft is also shown to be stable in yaw The nominal response of SOAR in the minimum and about the oncoming ow direction. maximum drag con guration for varying orbital altitude and in the absence of further perturbing torques is pre- 4.2. Dynamic Stability sented in Fig. 8. The responses demonstrate the basic In order to understand the evolution of attitude over aerostable nature of the spacecraft in the maximum and time and in the presence of perturbing torques the dynamic minimum drag con gurations and that the oscillatory am- response of the spacecraft must be considered. The orbital plitude decreases and frequency increases with increasing and attitude response of SOAR in the VLEO environment aerodynamic sti ness and dynamic pressure. can be investigated using a 6-DOF simultaneous orbit and The oscillatory modes can be internally damped using attitude propagation method. This method is based on the attitude control actuators as discussed by Virgili Llop et al. numerical solution of the complete equations of motion for [68], signi cantly reducing the range over which the atti- an orbiting satellite (kinematic and dynamic motion [62]) tude of the spacecraft varies. However, due to the presence for which varying force and torque model inputs can be 10 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Figure 8: Nominal uncontrolled pitch response of SOAR for varying altitude and minimum/maximum drag n con guration in an equatorial orbit. Only spherical gravity and aerodynamic torques are considered in an equatorial orbit. An initial pitching rate of 0:1 ° s is applied. of further perturbing torques in the real orbital environ- feathered con gurations [68] similar to SOAR. ment (e.g. atmospheric co-rotation, thermospheric winds, A signi cant challenge in the control of an aerodynamic solar-radiation pressure, gravity gradient), and errors or spacecraft is that the true oncoming ow vector is generally incompatibilities associated with real attitude actuator sys- not known a priori. A true reference vector for three-axis tems (e.g. magnetorquer availability and cross-coupling), ow-pointing control is therefore missing. A reference the true dynamic response is more complex. vector including the e ect of atmospheric co-rotation can Methods of control for aerostable spacecraft have been be provided, however the prediction of thermospheric winds presented in the literature. Psiaki [69] presents a compass- using available models is associated with much greater like PID control method which utilises magnetorquers to uncertainty. Alternatively, a sensor which can provide provide three-axis stabilised nadir-pointing capability to a the oncoming ow vector could be used to provide in- 1U CubeSat with a shuttlecock con guration of deployed situ measurements for active control methods. However, \feathers". Auret and Steyn [70] subsequently applied this proposals for such instruments for use in VLEO are only method to a 3U CubeSat geometry which included a pair of just emerging [77, 78]. actuating \paddles" that provide control capability about SOAR will be launched to the International Space Sta- the roll axis. Aerostability with assisted damping has tion and deployed into an 51:6° inclination orbit with an also been successfully demonstrated in orbit. The DS-MO initial altitude of approximately 400 km. Whilst forecasts spacecraft featured extended aerodynamic skirts and a gy- for solar cycle 25 vary [79{81] the deployment will occur dur- rodamper mechanism [71]. The Passive Aerodynamically- ing a period of minimum solar activity and the atmospheric stabilized Magnetically-damped Satellite (PAMS) had a density at this altitude will therefore be characteristically cylindrical geometry with a biased centre-of-mass that low. provided aerostability and magnetic hysteresis rods to pro- The attitude response of SOAR in the maximum drag vide damping [72]. The Gravity eld and steady-state con guration with reaction wheel damping and propor- Ocean Circulation Explorer (GOCE) spacecraft featured tional control is shown in Fig. 9. The restoring aerody- rear-mounted aerodynamic ns and was equipped with mag- namic torques experienced by the spacecraft are shown to netorquers for damping [73]. MagSat also demonstrated be of a similar magnitude to the solar radiation pressure trim in pitch using an aerodynamic boom of variable length torques and over an order of magnitude smaller than the [74, 75]. Three-axis aerodynamic pointing control has also residual magnetic dipole torques. The use of the reaction been considered for a \shuttlecock" geometry [76] and wheels is therefore important at this altitude to assist the 11 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ stability of the spacecraft and maintain pointing of the spacecraft close to the oncoming ow direction. The accumulation of the perturbing torques may also necessitate periodic management of the angular momen- tum within the reaction wheels to avoid saturation. The typical disadvantages of magnetorquer cross-coupling and limited availability (due to instantaneous orientation of the magnetic eld) must be accommodated whilst the power consumption of the actuators must be considered for ex- tended use and through eclipse periods. As SOAR naturally decays to lower altitudes, the magnitude of the aerodynamic torques will increase, and the aerostability of the spacecraft when in the minimum or maximum drag con gurations will improve, reducing the requirements on the attitude actuators. Given the relative magnitude of the di erent external perturbations, particularly the residual magnetic dipole, it may be more dicult to perform the experiments at higher altitudes where the \signal" (aerodynamic forces and torques) is low in comparison to the sources of \noise" (other perturbing forces and torques). It may therefore be necessary to allow the satellite to initially decay in altitude before commencing the experimental operations. 4.3. Experimental Con gurations During the experimental operations co-rotated or counter- rotated con gurations of the steerable ns will be utilised to expose the di erent materials to the oncoming ow, mod- ifying the natural stability and therefore attitude dynamics of the spacecraft. In the counter-rotated con guration, when the space- craft is nominally pointed towards the oncoming ow di- rection, no net torques are generated in pitch or yaw. A rolling moment is however produced due to the opposing lift forces generated on the two exposed surfaces, which if uncontrolled will cause the spacecraft to spin up. If the spacecraft is disturbed from its equilibrium ow pointing con guration pitch and/or yaw torques will be produced and the e ect of pitch-yaw coupling will act to further dis- turb the attitude of the spacecraft from the ow-pointing direction. In a three-axis controlled mode, a maximum duration on operations with ns rotated in a counter-rotated con g- uration is imposed by the build-up of angular momentum and saturation of reaction wheels. This is a function of the atmospheric density, incidence angle of the steerable ns, and material performance. The thermospheric wind, solar activity, and other external disturbance torques also contribute to this attitude performance, but vary with greater uncertainty. At lower altitudes the time-period over which spacecraft can be operated successfully may therefore be signi cantly limited for some counter-rotated con gurations, the impact of which will be discussed later Figure 9: Environmental torques and controlled attitude of SOAR at 400 km altitude and 51:6° inclination in maximum drag in Section 5 with regards to the experimental uncertainty. con guration. In alternative co-rotated n con gurations, net pitch or yaw torques are generated by the common incidence 12 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Table 2: Summary of expected satellite sensor performance. angle of the two opposing steerable ns. This causes the spacecraft to rotate and y at an angle to the oncoming ow Instrument Uncertainty with an associated oscillatory motion about the new o set equilibrium attitude that results from the aerostability GPS Position [m] 2:5 (1) 1 3 of the spacecraft. The control actuators may be used GPS Velocity [m s ] 45 10 (1) to correct the nominal pointing direction of the satellite ADCS Angle [rad] 0:2 10 (1) 1 3 such that the INMS will be realigned close to the ow, ADCS Angular Velocity [rad s ] 25 10 (1) ensuring the accuracy of the measured density and ow 3 INMS Number Density [cm ] N + 0:7 (1) velocity information. However, as a bias in the pitch or yaw INMS Horizontal Acceptance [rad] 0.279 (FWHM) torques exists, angular momentum will again accumulate INMS Vertical Acceptance [rad] 0.035 (FWHM) in the reaction wheels, eventually causing saturation of the Steerable Fin Rotation Angle [rad] 0:015 (1) attitude control system. Contrastingly, for the investigation of lift force coe- cient, the roll axis can be left uncontrolled allowing the uncertainty and angular dependency, expressed as full- net torque to accumulate and the acceleration and angle in width half-maximum (FWHM) measures in Table 2. roll to be measured by the ADCS. Control of the pitch and This modi ed data is subsequently used to perform the yaw axes are maintained to keep the spacecraft pointing orbit and attitude determination processes. For recovery close to the oncoming ow direction. of the drag coecient only orbit propagation is performed, whilst for the moment coecient combined attitude and orbit propagation is performed. In both cases, the initial 5. Experimental Performance \measured" state vector is used to set-up the propagation The expected performance of the mission and the ex- method. The force and torque models used in the propaga- perimental determination of the aerodynamic coecients tion are modi ed to incorporate further uncertainty and can be investigated by testing the free-parameter tting to represent the reduction in or lack of knowledge of the and least-squares processes using simulated orbit and atti- true in-orbit environment and spacecraft interactions. For tude data. This data is modi ed with noise to represent the drag coecient tting, the the thermospheric winds the expected in-orbit sensor performance. A Monte Carlo are neglected, whilst for the moment coecient cases the approach is also implemented to encompass variation in residual magnetic dipole interactions are also excluded from the initial conditions (epoch, orbit, and attitude) and to the attitude propagation scheme. vary the addition of noise to the data for each simulated The expected performance of the experiments at di er- run. ent orbital altitudes and steerable n con gurations can The free-parameter tting process utilises a least-squares be obtained by considering the standard deviation of the orbit determination algorithm, described in Section 3.3. returned aerodynamic coecient after a number of Monte This process seeks to minimise the error between the ref- Carlo iterations. However, given that such a Monte Carlo erence (simulated or on-orbit) data and mathematically simulation only provides a random sample of results, the modelled data by varying the free values of the aerody- con dence of the standard deviation should also be con- namic coecients in the presence of known measurement sidered, within which the population standard deviation uncertainties. The process is iterative and is terminated would be expected to lie with the given con dence. by convergence criteria based on the weighted RMS of the Reducing the width of the standard deviation increases residuals between the reference and modelled data. Central- the resolution of the experiment and can be most simply di erencing methods are used to account for the errors in achieved by increasing the signal-to-noise ratio of the exper- the initial condition of the state vector due to measurement iment. For the on-orbit experiments this may be achieved uncertainties. by increasing the test run time or the magnitude of the Simulated on-orbit data is rst produced using the high- force or torque to be measured. However, in many cases delity attitude and orbit propagation method described these parameters are restricted by the platform design (e.g. in Section 4.2. Orbit and attitude noise is produced by steerable n size, reaction wheel capability) or on-orbit considering the performance parameters of the GPS and operations (e.g. power and downlink budget). Reducing ADCS sensors, reported in Table 2. Uncertainty on the the uncertainties associated with the measured on-orbit angular velocity of the reaction wheels and angular position data may also be helpful, for example improved in-situ of the steerable ns has also been similarly introduced. density measurements and position/attitude knowledge. The expected in-orbit performance of the INMS is also Finally, improved knowledge and modelling of the unmea- simulated. The measured density is rst produced using sured disturbances or perturbations would further reduce the NRLMSISE-00 [8] atmosphere model, informed by the uncertainties within the free-parameter tting and orbit orbit and attitude data, and modi ed for GPS and ADCS determination process, primarily requiring characterisation sensor and acquisition errors and noise. This density is of the residual magnetic dipole and solar radiation pressure then subsequently transformed using the INMS instrument interactions of the satellite. 13 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ 5.1. Drag Coecient data for di erent steerable n incidence angles (in 15° in- crements) are likely to be largely identi able and distin- The returned drag coecients from this data reduction guishable from one another. In the co-rotated con gura- process are shown in Fig. 10 for counter-rotated and co- tion smaller increments in incidence angle may also be rotated steerable n con gurations over a range of di erent discernible, particularly at shallower de ections. incidence angles and altitudes. A duration of 120 min is Di erence in the experimentally determined results from targeted, a limit imposed by the expected power balance the reference drag coecients arises primarily from the achievable by spacecraft in the experimental mode. Three- variation in the area of the satellite surfaces projected axis reaction wheel control is implemented to maintain an into the ow. This occurs as the satellite attitude varies approximately ow-pointing attitude and stability. The with respect to the ow as a result of the aerodynamic test-run is aborted if the angular range in pitch or yaw and other environmental torques. Further, the attitude exceeds the INMS instrument acceptance limit or the re- control system does not have knowledge of the oncoming action wheels approach saturation. Dashed lines for the ow direction and therefore uses the o set LVLH frame as underlying GSI model-based drag coecient (calculated us- a reference. Additional sources of error can be attributed ing Sentman's model with  = 1 and the ADBSAT tool) at to the sensor accuracy and noise parameters (INMS, GPS, each altitude and con guration are provided as a reference. and attitude) and the modelled forces used in the least- It should be noted that under the assumption of di use squares orbit determination tting process. Of the modelled re-emission properties, the counted-rotated and co-rotated forces, the solar radiation pressure exhibits the greatest con gurations an incidence of 90° are equivalent. The very uncertainties. Improved knowledge of the solar radiation minor di erences in the presented results are therefore due pressure interaction with the di erent external surfaces of to variations that arise from the implementation of the the satellite could improve the estimation of these e ects. Monte Carlo simulation. The dependence of the drag coecient with altitude The results for the counter-rotated and co-rotated con- appears to be less clearly identi able from the expected gurations are very similar for altitudes between 400 km experimental performance. This is due to the relatively to 300 km. However, at altitudes at and below 250 km small variation in drag coecient which is expected over the the results from the co-rotated con guration demonstrate available altitude range (200 km to 400 km) in comparison much smaller standard deviations and mean values closer to the experimental uncertainty. However, Sentman's GSI to the reference values. This is principally due to the longer model with a single accommodation coecient ( = 1) has experimental periods that can be maintained by the atti- been used in this analysis, representing typical di usely tude control system in the co-rotated con guration before re-emitting materials, for example contaminated metallic the reaction wheels saturate, in general more than double surfaces [30]. When complete accommodation and di use than the counter-rotated con guration. More orbital po- re-emission is assumed signi cant variation in the drag sition information is therefore provided against which the coecient with altitude is not expected and will only be least-squares orbit determination process can best t the driven by the variation in speed ratio and thermospheric experimentally determined drag coecient, reducing the temperature. experimental uncertainty. Increasing the experimental duration at higher orbital 5.2. Rolling Moment Coecient altitudes would similarly assist in improving the experimen- The results for the tted rolling moment coecient ex- tal uncertainty for both the counter-rotated and co-rotated periments are shown in Fig. 11 for varying counter-rotated steerable n con gurations. However, this is challenging steerable n incidence angles and altitude under two-axis due to the power budget of the satellite. The increase in (pitch and yaw) control. standard deviation at 400 km is due to low atmospheric The experimental uncertainty is generally seen to in- density at higher altitudes and therefore the limited e ect crease with orbital altitude. This is attributed to the e ect that the drag will have on the orbit over a period of only of unmeasured and poorly modelled attitude perturbations 120 min in the presence of the GPS position measurements, on the satellite during the experiment. These are most particularly for shallow steerable n incidence angles. The signi cant when the atmospheric density is lowest and greater relative magnitude of the additional disturbing per- the aerodynamic torques are therefore relatively low in turbations also contribute to the increased experimental magnitude. Residual magnetic dipole interactions are the uncertainty at the higher altitudes. most signi cant of these e ects and are not included in For both the co-rotated and counter-rotated con gura- the orbit determination algorithm. However, if knowledge tions the minimum experimental uncertainty is expected or modelling of the time-varying residual magnetic dipole to be achieved at approximately 300 km altitude where a of the satellite can be obtained this may be incorporated balance between the experimental duration, measurable into the analysis. Similarly to the drag coecient exper- e ect on the orbital trajectory, and magnitude of external iments, improvement of the knowledge and modelling of perturbations is found. solar radiation pressure e ects could also be bene cial. These results indicate that drag coecients determined For most steerable n con gurations the rolling moment from the on-orbit experiments and associated measured coecient is clearly identi able against the other results. 14 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ Figure 10: Drag coecient determination performance for counter-rotated and co-rotated steerable n con gurations. Sample mean tted drag coecient (top), referred to reference area A =2, is given top with error-bars representing the associated standard deviation. Reference lines indicate the modelled GSI value. The standard deviation is given bottom with error bars representing the 95% con dence interval. Data points have been shifted slightly in the x-axis to allow for visibility of overlapping error bars. However, at lower incidence angles (15° and 30°) the er- of the SOAR geometry demonstrate the aerostable nature ror bars overlap indicating that these con gurations may of the design in the nominal maximum and minimum drag not be distinguishable from each other from the on-orbit modes and the use of the steerable ns in both counter- measurements. At greater incidence angles the variation rotated and co-rotated modes to perform the proposed between smaller increments in steerable n angle may be aerodynamics characterisation experiments. possible, particularly at lower altitudes. Using the combination of the INMS and the steerable n In contrast to the drag coecient results presented payloads, on-orbit experimental assessment of the aerody- previously, the dependence of torque coecient on orbital namic coecients of di erent materials at varying incidence altitude appears to be more marked, even for assumed to the oncoming ow will be performed. These experiments fully accommodated gas-surface interactions. However, will be repeated as the orbit of SOAR decays to investigate this variation may remain obscured by the experimental the variation with orbital altitude. The modelled uncer- uncertainties, particularly for larger steerable n incidence tainty of these experiments indicates that the drag and angles (60° and 75°) that vary more slowly with altitude. lift coecients at di erent incidence be determined from the measured parameters in the presence of the disturbing and perturbing forces and torques present in VLEO. The 6. Concluding Remarks uncertainty of drag coecient measurements was shown to be minimised around an altitude of 300 km, whilst the lift This paper has described the proposed method for de- coecient experiment generally demonstrates improvement termination of the aerodynamic coecients of di erent ma- as the altitude is reduced further. These insights will be terials on SOAR, a scienti c CubeSat due to be launched used to plan the operations of the SOAR mission. in 2021. The presented analysis and simulated dynamics 15 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ is not responsible for any use that may be made of the information it contains. References [1] P. C. Roberts, N. H. Crisp, S. Edmondson, S. J. Haigh, R. E. Lyons, V. T. Oiko, A. Macario-Rojas, K. L. Smith, J. Be- cedas, G. Gonz alez, I. Vazquez,  A. Brana, ~ K. Antonini, K. Bay, L. Ghizoni, V. Jungnell, J. Morsbl, T. Binder, A. Boxberger, G. H. Herdrich, F. Romano, S. Fasoulas, D. Garcia-Alminana, ~ S. Rodriguez-Donaire, D. Kataria, M. Davidson, R. Outlaw, B. Belkouchi, A. Conte, J. S. Perez, R. Villain, B. Heierer, A. Schwalber, DISCOVERER { Radical Redesign of Earth Observation Satellites for Sustained Operation at Signi cantly Lower Altitudes, in: 68th International Astronautical Congress, September, International Astronautical Federation (IAF), Ade- laide, Australia, 2017. [2] P. C. Roberts, N. H. Crisp, F. Romano, G. H. Herdrich, V. T. Oiko, S. Edmondson, S. J. Haigh, C. Huyton, S. Livadiotti, R. E. Lyons, K. L. Smith, L. A. Sinpetru, A. Straker, S. D. Worrall, J. Becedas, R. M. Dom nguez, D. Gonz alez, V. Canas, ~ V. Hanes- sian, A. Mlgaard, J. Nielsen, M. Bisgaard, A. Boxberger, Y.-A. Chan, S. Fasoulas, C. Traub, D. Garcia-Alminana, ~ S. Rodriguez- Donaire, M. Sureda, D. Kataria, R. Outlaw, B. Belkouchi, A. Conte, J. S. Perez, R. Villain, B. Heierer, A. Schwalber, DISCOVERER { Making Commercial Satellite Operations in Very Low Earth Orbit a Reality, in: 70th International Astro- nautical Congress, International Astronautical Federation (IAF), Washington, DC, 2019. [3] V. T. A. Oiko, P. C. Roberts, S. Edmondson, S. D. Worrall, D. Kataria, R. Outlaw, S. Haigh, N. H. Crisp, R. E. Lyons, S. Li- vadiotti, C. Huyton, L. A. Sinpetru, J. Becedas, G. Gonz alez, R. M. Dominguez, D. Gonz alez, L. Ghizoni, V. Jungnell, K. Bay, J. Morsbl, G. H. Herdrich, F. Romano, T. Binder, A. Boxberger, S. Fasoulas, C. Traub, D. Garcia-Alminana, ~ S. Rodriguez- Donaire, M. Sureda, R. Villain, J. S. Perez, A. Conte, B. Belk- ouchi, A. Schwalber, B. Heierer, Design and Development of a Figure 11: Rolling moment coecient determination performance Hyper-thermal Atomic Oxygen Wind Tunnel Facility, in: 14th for counter-rotated steerable n con gurations. Sample mean tted ISMSE and 12th ICPMSE, Biarritz, France, 2018. torque coecient (top), referred to reference area A =2 and [4] V. T. Oiko, P. C. Roberts, S. D. Worrall, S. Edmondson, S. J. reference length L , is given with error-bars representing the Haigh, N. H. Crisp, S. Livadiotti, C. Huyton, R. E. Lyons, associated standard deviation. Reference lines indicate the modelled K. L. Smith, L. A. Sinpetru, B. E. Holmes, J. Becedas, R. M. GSI value. The standard deviation (bottom) is given with error bars Dom nguez, D. Gonz alez, V. Canas, ~ V. Hanessian, A. Mlgaard, representing the 95% con dence interval. Data points have been J. Nielsen, M. Bisgaard, A. Boxberger, Y.-A. Chan, G. H. Her- shifted slightly in the x-axis to allow for visibility of overlapping drich, F. Romano, S. Fasoulas, C. Traub, D. Garcia-Alminana, ~ error bars. S. Rodriguez-Donaire, M. Sureda, D. Kataria, R. Outlaw, B. Belkouchi, A. Conte, J. S. Perez, R. Villain, B. Heierer, A. Schwalber, A Ground-Based Experimental Facility for Or- The purpose of this on-orbit experimentation is to pro- bital Aerodynamics Research, in: 70th International Astronau- vide valuable in-situ validation data for a more extensive tical Congress, International Astronautical Federation (IAF), Washington, DC, 2019. investigation of rare ed- ow GSIs to be performed on the [5] N. Crisp, P. Roberts, S. Livadiotti, V. Oiko, S. Edmond- ground with the aim to improve knowledge of GSI mech- son, S. Haigh, C. Huyton, L. Sinpetru, K. Smith, S. Wor- anisms and the associated models that describe this be- rall, J. Becedas, R. Dom nguez, D. Gonz alez, V. Hanes- haviour. A systematic study to identify materials that sian, A. Mlgaard, J. Nielsen, M. Bisgaard, Y.-A. Chan, S. Fasoulas, G. Herdrich, F. Romano, C. Traub, D. Garc a- can increase aerodynamic performance at lower orbital al- Alminana, ~ S. Rodr guez-Donaire, M. Sureda, D. Kataria, titudes will also be performed. SOAR will test two such R. Outlaw, B. Belkouchi, A. Conte, J. Perez, R. Villain, novel materials with promising drag-reducing characteris- B. Heierer, A. Schwalber, The bene ts of very low earth tics in-orbit. orbit for earth observation missions, Progress in Aerospace Sciences 117 (2020) 100619. URL: https://linkinghub. elsevier.com/retrieve/pii/S0376042120300312. doi:10.1016/ Acknowledgements j.paerosci.2020.100619. [6] B. Gavish, J. Kalvenes, The impact of satellite altitude on the performance of LEOS based communication systems, Wireless This project has received funding from the European Networks 4 (1998) 119{213. doi:10.1023/A:1019151905814. Union's Horizon 2020 research and innovation programme [7] L. H. Sentman, Free molecule ow theory and its application under grant agreement No 737183. This publication re ects to the determination of aerodynamic forces, Technical Report, Lockheed Missiles & Space Company, Sunnyvale, CA, 1961. only the view of the authors. The European Commission 16 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ [8] J. Picone, A. Hedin, D. P. Drob, A. Aikin, NRLMSISE-00 on spacecraft materials, Journal of Materials Science 30 (1995) Empirical Model of the Atmosphere: Statistical Comparisons 281{307. doi:10.1007/BF00354389. and Scienti c Issues, Journal of Geophysical Research 107 (2002). [26] B. A. Banks, S. K. Miller, K. K. de Groh, Low Earth Orbital doi:10.1029/2002JA009430. Atomic Oxygen Interactions with Materials, in: 2nd Interna- [9] K. Moe, M. M. Moe, S. D. Wallace, Improved Satellite Drag tional Energy Conversion Engineering Conference, American Coecient Calculations from Orbital Measurements of Energy Institute of Aeronautics and Astronautics (AIAA), Providence, Accommodation, Journal of Spacecraft and Rockets 35 (1998) RI, 2004. doi:10.2514/6.2004-5638. 266{272. doi:10.2514/2.3350. [27] S. Samwel, Low Earth Orbital Atomic Oxygen Erosion E ect [10] B. R. Bowman, K. Moe, Drag Coecient Variability at on Spacecraft Materials, Space Research Journal 7 (2014) 1{13. 175-500km from the Orbit Decay Analyses of Spheres, in: doi:10.3923/srj.2014.1.13. AAS/AIAA Astrodynamics Specialists Conference, American [28] I. Harrison, G. Swinerd, A free molecule aerodynamic investi- Astronautical Society (AAS), Lake Tahoe, CA, 2005. gation using multiple satellite analysis, Planetary and Space [11] E. K. Sutton, Normalized Force Coecients for Satellites with Science 44 (1996) 171{180. doi:10.1016/0032-0633(95)00077-1. Elongated Shapes, Journal of Spacecraft and Rockets 46 (2009) [29] V. J. Murray, M. D. Pilinski, E. J. Smoll, M. Qian, T. K. Minton, 112{116. doi:10.2514/1.40940. S. M. Madzunkov, M. R. Darrach, Gas{Surface Scattering Dy- [12] S. Andrews, L. Berthoud, Characterising satellite aerody- namics Applied to Concentration of Gases for Mass Spectrometry namics in Very Low Earth Orbit inclusive of ion thruster in Tenuous Atmospheres, The Journal of Physical Chemistry C plume-thermosphere/ionosphere interactions, Acta Astro- 121 (2017) 7903{7922. doi:10.1021/acs.jpcc.7b00456. nautica 170 (2020) 386{396. URL: https://linkinghub. [30] K. Moe, M. M. Moe, Gas-Surface Interactions in Low-Earth elsevier.com/retrieve/pii/S0094576519314675. doi:10.1016/ Orbit, in: 27th International Symposium on Rare ed Gas j.actaastro.2019.12.034. Dynamics, AIP Conference Proceedings, volume 1333, American [13] E. Doornbos, Thermospheric Density and Wind Determi- Institute of Physics, Paci c Grove, CA, 2011, pp. 1313{1318. nation from Satellite Dynamics, Springer Theses, Springer doi:10.1063/1.3562825. Berlin Heidelberg, Berlin, Heidelberg, 2012. doi:10.1007/ [31] J. Gregory, P. Peters, A Measurement of the Angular Distri- 978-3-642-25129-0. bution of 5eV Atomic Oxygen Scattered o a Solid Surface in [14] S. A. Schaaf, P. L. Chambre, Flow of rare ed gases, in: Fun- Earth Orbit, Rare ed Gas Dynamics 15 (1987). damental of Gas Dynamics, volume III, Princeton University [32] B. A. Banks, K. K. de Groh, S. K. Miller, MISSE Scattered Press, Princeton, NJ, 1958, pp. 687{739. Atomic Oxygen Characterization Experiment, TM-2006-214355, [15] R. Schamberg, A New Analytic Representation of Surface Inter- NASA Glenn Research Centre, Cleveland, OH, 2006. action for Hyperthermal Free Molecule Flow with Applications [33] B. A. Banks, K. K. de Groh, S. K. Miller, D. L. Waters, J. I. to Neutral-particle Drag Estimates of Satellites, Research mem- Kleiman, Lessons Learned from Atomic Oxygen Interaction orandum, Rand Corporation, 1959. with Spacecraft Materials in Low Earth Orbit, AIP Conference [16] E. M. Gaposchkin, Calculation of Satellite Drag Coecients, Proceedings 1087 (2009) 312{325. doi:10.1063/1.3076845. Technical Report 998, Lincoln Laboratory, MIT, Lexington, MA, [34] R. C. Blanchard, Rare ed Flow Lift-to-Drag Measurements of 1994. the Shuttle Orbiter, in: 15th ICAS Congress, American Institute [17] J. Storch, Aerodynamic Disturbances on Spacecraft in Free- of Aeronautics and Astronautics (AIAA), London, UK, 1986. Molecular Flow, TR-2003(3397)-1, The Aerospace Corporation, [35] R. C. Blanchard, J. Y. Nicholson, Orbiter Rare ed-Flow Reentry El Segundo, CA, 2002. Measurements from the OARE on STS-62, TM-110182, NASA, [18] D. Mostaza-Prieto, B. P. Graziano, P. C. Roberts, Spacecraft Hampton, VA, 1995. drag modelling, Progress in Aerospace Sciences 64 (2014) 56{65. [36] K. Moe, Absolute Atmospheric Densities Determined from doi:10.1016/j.paerosci.2013.09.001. the Spin and Orbital Decays of Explorer VI, Planetary and [19] S. Livadiotti, N. H. Crisp, P. C. Roberts, S. D. Worrall, V. T. Space Science 14 (1966) 1065{1075. doi:10.1016/0032-0633(66) Oiko, S. Edmondson, S. J. Haigh, C. Huyton, K. L. Smith, 90022-5. L. A. Sinpetru, B. E. Holmes, J. Becedas, R. M. Dom nguez, [37] K. Moe, B. R. Bowman, The E ects of Surface Composition V. Canas, ~ S. Christensen, A. Mlgaard, J. Nielsen, M. Bis- and Treatment on Drag Coecient of Spherical Satellites, in: gaard, Y.-A. Chan, G. H. Herdrich, F. Romano, S. Fasoulas, AAS/AIAA Astrodynamics Specialists Conference, American C. Traub, D. Garcia-Alminana, ~ S. Rodriguez-Donaire, M. Sureda, Astronautical Society (AAS), Lake Tahoe, CA, 2005. D. Kataria, B. Belkouchi, A. Conte, J. S. Perez, R. Vil- [38] C. Pardini, W. K. Tobiska, L. Anselmo, Analysis of the orbital lain, R. Outlaw, A review of gas-surface interaction models decay of spherical satellites using di erent solar ux proxies and for orbital aerodynamics applications, Progress in Aerospace atmospheric density models, Advances in Space Research 37 Sciences 119 (2020) 100675. URL: https://linkinghub. (2006) 392{400. doi:10.1016/j.asr.2004.10.009. elsevier.com/retrieve/pii/S0376042120300877. doi:10.1016/ [39] M. D. Pilinski, B. M. Argrow, S. E. Palo, Drag Coecients j.paerosci.2020.100675. of Satellites with Concave Geometries: Comparing Models and [20] C. Cercignani, M. Lampis, Kinetic models for gas-surface in- Observations, Journal of Spacecraft and Rockets 48 (2011) teractions, Transport Theory and Statistical Physics 1 (1971) 312{325. doi:10.2514/1.50915. 101{114. doi:10.1080/00411457108231440. [40] B. Ching, D. Hickman, J. Straus, E ects of atmospheric winds [21] R. Lord, Some extensions to the Cercignani{Lampis gas{surface and aerodynamic lift on the inclination of the orbit of the S3-1 scattering kernel, Physics of Fluids A: Fluid Dynamics 3 (1991) satellite, Journal of Geophysical Research 82 (1977) 1474{1480. 706{710. doi:10.1063/1.858076. doi:10.1029/JA082i010p01474. [22] R. Lord, Some further extensions of the Cercignani{Lampis [41] C. Pardini, L. Anselmo, K. Moe, M. M. Moe, Drag and energy gas{surface interaction model, Physics of Fluids 7 (1995) 1159{ accommodation coecients during sunspot maximum, Advances 1161. doi:10.1063/1.868557. in Space Research 45 (2010) 638{650. doi:10.1016/j.asr.2009. [23] A. Walker, P. M. Mehta, J. Koller, Drag Coecient Model Using 08.034. the Cercignani{Lampis{Lord Gas{Surface Interaction Model, [42] A. Macario-Rojas, K. L. Smith, N. H. Crisp, P. C. Roberts, Journal of Spacecraft and Rockets 51 (2014) 1544{1563. doi:10. Atmospheric interaction with nanosatellites from observed or- 2514/1.A32677. bital decay, Advances in Space Research 61 (2018) 2972{2982. [24] M. M. Moe, S. D. Wallace, K. Moe, Re nements in determining doi:10.1016/j.asr.2018.02.022. satellite drag coecients - Method for resolving density discrep- [43] E. Doornbos, H. Klinkrad, Modelling of space weather e ects on ancies, Journal of Guidance, Control, and Dynamics 16 (1993) satellite drag, Advances in Space Research 37 (2006) 1229{1239. 441{445. doi:10.2514/3.21029. doi:10.1016/j.asr.2005.04.097. [25] M. R. Reddy, Review e ect of low earth orbit atomic oxygen [44] P. M. Mehta, A. C. Walker, E. K. Sutton, H. C. Godinez, New 17 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ density estimates derived using accelerometers on board the [61] D. Mostaza-Prieto, P. C. Roberts, Methodology to Analyze At- CHAMP and GRACE satellites, Space Weather 15 (2017) 558{ titude Stability of Satellites Subjected to Aerodynamic Torques, 576. doi:10.1002/2016SW001562. Journal of Guidance, Control, and Dynamics 39 (2016) 437{449. [45] J. Virgili Llop, P. C. Roberts, Dsat, a QB50 CubeSat mission doi:10.2514/1.G001481. to study rare ed-gas drag modelling, Acta Astronautica 89 [62] F. Landis Markley, J. L. Crassidis, Fundamentals of Space- (2013) 130{138. doi:10.1016/j.actaastro.2013.04.006. craft Attitude Determination and Control, volume 33 of [46] D. Gerhardt, M. Bisgaard, L. Alminde, R. Walker, M. A. Fer- Space Technology Library, 1 ed., Springer-Verlag, New nandez, A. Latiri, J.-L. Issler, GOMX-3 : Mission Results from York, NY, 2014. URL: http://link.springer.com/10.1007/ the Inaugural ESA In-Orbit Demonstration CubeSat, in: 30th 978-1-4939-0802-8. doi:10.1007/978-1-4939-0802-8. Annual AIAA/USU Conference on Small Satellites, American [63] F. Lemoine, S. Kenyon, J. Factor, R. Trimmer, N. Pavlis, Institute of Aeronautics and Astronautics (AIAA), Logan, UT, D. Chinn, C. Cox, S. Klosko, S. Luthcke, M. Torrence, 2016. Y. Wang, R. Williamson, E. Pavlis, R. Rapp, T. Olson, [47] R. Holst, J. Nielsen, D. Gerhardt, J. A. G. Ahumada, Attitude The Development of the Joint NASA GSFC and the Na- and Orbit Control Results of the GOMX-4 Tandem CubeSat tional Imagery and Mapping Agency (NIMA) Geopotential Mission, in: 69th International Astronautical Congress, Inter- Model EGM96, NASA/TP|1998{206861 July, NASA God- national Astronautical Federation (IAF), Bremen, Germany, dard Space Flight Center, Greenbelt, MD, 1998. URL: http:// 2018. link.springer.com/10.1007/978-3-662-03482-8{_}62. doi:10. [48] O. Montenbruck, M. Garcia-Fernandez, J. Williams, Per- 1007/978-3-662-03482-8_62. formance comparison of semicodeless GPS receivers for [64] N. K. Pavlis, S. A. Holmes, S. C. Kenyon, J. K. Factor, The LEO satellites, GPS Solutions 10 (2006) 249{261. URL: development and evaluation of the Earth Gravitational Model http://link.springer.com/10.1007/s10291-006-0025-9. 2008 (EGM2008), Journal of Geophysical Research: Solid Earth doi:10.1007/s10291-006-0025-9. 117 (2012). doi:10.1029/2011JB008916. [49] O. Montenbruck, M. Markgraf, M. Garcia-Fernandez, [65] C. C. Finlay, S. Maus, C. Beggan, T. Bondar, A. Chambo- A. Helm, GPS for Microsatellites { Status and Per- dut, T. Chernova, A. Chulliat, V. Golovkov, B. Hamilton, spectives, in: R. Sandau, H.-P. R oser, A. Valenzuela M. Hamoudi, R. Holme, G. Hulot, W. Kuang, B. Langlais, (Eds.), Small Satellites for Earth Observation, Springer V. Lesur, F. Lowes, H. Luhr,  S. Macmillan, M. Mandea, Netherlands, Dordrecht, 2008, pp. 165{174. URL: http:// S. McLean, C. Manoj, M. Menvielle, I. Michaelis, N. Olsen, link.springer.com/10.1007/978-1-4020-6943-7{_}15. doi:10. J. Rauberg, M. Rother, T. Sabaka, A. Tangborn, L. T ner- 1007/978-1-4020-6943-7_15. Clausen, E. Th ebault, A. Thomson, I. Wardinski, Z. Wei, [50] O. Montenbruck, P. Swatschina, M. Markgraf, S. Santan- T. Zvereva, International Geomagnetic Reference Field: The drea, J. Naudet, E. Tilmans, Precision spacecraft navi- eleventh generation, Geophysical Journal International 183 gation using a low-cost GPS receiver, GPS Solutions 16 (2010) 1216{1230. doi:10.1111/j.1365-246X.2010.04804.x. (2012) 519{529. URL: http://link.springer.com/10.1007/ [66] A. Hedin, E. Fleming, A. Manson, F. Schmidlin, S. Avery, s10291-011-0252-6. doi:10.1007/s10291-011-0252-6. R. Clark, S. Franke, G. Fraser, T. Tsuda, F. Vial, R. Vincent, [51] O. Montenbruck, A. Hauschild, R. B. Langley, C. Siemes, Empirical wind model for the upper, middle and lower atmo- CASSIOPE orbit and attitude determination using commercial sphere, Journal of Atmospheric and Terrestrial Physics 58 (1996) o -the-shelf GPS receivers, GPS Solutions 23 (2019) 114. 1421{1447. doi:10.1016/0021-9169(95)00122-0. URL: https://doi.org/10.1007/s10291-019-0907-2http: [67] D. P. Drob, J. Emmert, G. Crowley, J. Picone, G. Shepherd, //link.springer.com/10.1007/s10291-019-0907-2. W. Skinner, P. Hays, R. Niciejewski, M. Larsen, C. She, J. W. doi:10.1007/s10291-019-0907-2. Meriwether, G. Hernandez, M. Jarvis, D. Sipler, C. Tepley, [52] NASA, Spacecraft Aerodynamic Torques, SP-8050 January, M. O'Brien, J. Bowman, Q. Wu, Y. Murayama, S. Kawamura, NASA, 1971. I. Reid, R. Vincent, An empirical model of the Earth's horizontal [53] D. A. Vallado, Fundamentals of Astrodynamics and Applications, wind elds: HWM07, Journal of Geophysical Research: Space 4 ed., Microcosm Press/Springer, Hawthorne, CA, 2013. Physics 113 (2008) 1{18. doi:10.1029/2008JA013668. [54] D. Mostaza-Prieto, Characterisation and Applications of Aero- [68] J. Virgili Llop, P. C. Roberts, Z. Hao, Aerodynamic Attitude dynamic Torques on Satellites, Phd thesis, The University of and Orbit Control Capabilities of The Dsat CubeSat, in: 37th Manchester, 2017. Annual AAS Guidance and Control Conference, AAS 14-063, [55] G. A. Bird, Molecular Gas Dynamics and the Direct Simulation American Astronautical Society (AAS), Breckenridge, CO, 2014. of Gas Flows, Clarendon Press, Oxford, UK, 1994. [69] M. L. Psiaki, Nanosatellite Attitude Stabilization Using Pas- [56] G. LeBeau, A parallel implementation of the direct simula- sive Aerodynamics and Active Magnetic Torquing, Jour- tion Monte Carlo method, Computer Methods in Applied nal of Guidance, Control, and Dynamics 27 (2004) 347{355. Mechanics and Engineering 174 (1999) 319{337. doi:10.1016/ doi:10.2514/1.1993. S0045-7825(98)00302-8. [70] J. Auret, W. H. Steyn, Design of an Aerodynamic Attitude [57] G. LeBeau, F. Lumpkin III, Application highlights of the DSMC Control System for a Cubesat, 62nd International Astronautical Analysis Code (DAC) software for simulating rare ed ows, Congress (2011). Computer Methods in Applied Mechanics and Engineering 191 [71] V. Sarychev, S. Mirer, A. Degtyarev, E. Duarte, Investigation (2001) 595{609. doi:10.1016/S0045-7825(01)00304-8. of equilibria of a satellite subjected to gravitational and aerody- [58] C. White, M. Borg, T. Scanlon, S. Longshaw, B. John, D. Emer- namic torques, Celestial Mechanics and Dynamical Astronomy son, J. Reese, dsmcFoam+: An OpenFOAM based direct simu- 97 (2007) 267{287. doi:10.1007/s10569-006-9064-3. lation Monte Carlo solver, Computer Physics Communications [72] R. R. Kumar, D. D. Mazanek, M. L. Heck, Simulation and 224 (2018) 22{43. doi:10.1016/j.cpc.2017.09.030. Shuttle Hitchhiker validation of passive satellite aerostabilization, [59] C.-D. Munz, M. Auweter-Kurtz, S. Fasoulas, A. Mirza, P. Or- Journal of Spacecraft and Rockets 32 (1995) 806{811. doi:10. twein, M. Pfei er, T. Stindl, Coupled Particle-In-Cell and 2514/3.26688. Direct Simulation Monte Carlo method for simulating reactive [73] M. R. Drinkwater, R. Haagmans, D. Muzi, A. Popescu, plasma ows, Comptes Rendus M ecanique 342 (2014) 662{670. R. Floberghagen, M. Kern, M. Fehringer, The GOCE Gravity doi:10.1016/j.crme.2014.07.005. Mission: ESA'S First Core Earth Explorer, in: 3rd International [60] S. Fasoulas, C.-D. Munz, M. Pfei er, J. Beyer, T. Binder, S. Cop- GOCE User Workshop, SP-627, European Space Agency (ESA), plestone, A. Mirza, P. Nizenkov, P. Ortwein, W. Reschke, Com- Frascati, Italy, 2007, pp. 1{7. doi:ISBN92-9092-938-3. bining particle-in-cell and direct simulation Monte Carlo for the [74] T. H. Stengle, MagSat Attitude Dynamics and Control: Some simulation of reactive plasma ows, Physics of Fluids 31 (2019) Observations and Explanations, in: J. Teles (Ed.), Fiifth Annual 072006. doi:10.1063/1.5097638. Flight Mechanics/Estimation Theory Symposium, Greenbelt, 18 ©2020. This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/ MD, 1980. [75] B. Tossman, F. Mobley, G. Fountain, K. He ernan, J. Ray, C. Williams, MAGSAT attitude control system design and performance, in: Guidance and Control Conference, American Institute of Aeronautics and Astronautics (AIAA), Danvers, MA, 1980. doi:10.2514/6.1980-1730. [76] M. L. Gargasz, Optimal Spacecraft Attitude Control Using Aero- dynamic Torques, Msc thesis, Air Force Institute of Technology, [77] M. Eberhart, S. L ohle, A. Steinbeck, T. Binder, S. Fasoulas, Measurement of atomic oxygen in the middle atmosphere using solid electrolyte sensors and catalytic probes, Atmo- spheric Measurement Techniques 8 (2015) 3701{3714. URL: https://amt.copernicus.org/articles/8/3701/2015/. doi:10. 5194/amt-8-3701-2015. [78] R. Verker, A. Bolker, Y. Carmiel, I. Gouzman, E. Grossman, T. K. Minton, S. Remaury, Ground testing of an on-orbit atomic oxygen ux and ionizing radiation dose sensor based on material degradation by the space environment, Acta Astronautica 173 (2020) 333{343. URL: https://doi.org/10.1016/j.actaastro. 2020.04.065. doi:10.1016/j.actaastro.2020.04.065. [79] D. H. Hathaway, L. A. Upton, Predicting the amplitude and hemispheric asymmetry of solar cycle 25 with surface ux trans- port, Journal of Geophysical Research: Space Physics 121 (2016) 10,744{10,753. doi:10.1002/2016JA023190. [80] R. Cameron, J. Jiang, M. Schussler,  Solar Cycle 25: Another Moderate Cycle?, The Astrophysical Journal 823 (2016) L22. doi:10.3847/2041-8205/823/2/L22. [81] P. Bhowmik, D. Nandy, Prediction of the strength and timing of sunspot cycle 25 reveal decadal-scale space environmental conditions, Nature Communications 9 (2018) 5209. doi:10.1038/ s41467-018-07690-0.

Journal

Physics – arXiv (Cornell University)

Published: Dec 14, 2020

References

  • DISCOVERER – Radical redesign of earth observation satellites for sustained operation at significantly lower altitudes
    Roberts
  • DISCOVERER – Making commercial satellite operations in very low earth orbit a reality
    Roberts