Final Report Summary - ELSA (European Levitated Spherical Actuator)
Executive Summary:
The ELSA project had the goal of improving European capacity to independently manufacture commercial and scientific satellites by bringing a new actuator for attitude and orbit control systems to a higher level of maturity. This goal is in line with ESA’s technology strategy and long term plan.
Attitude and orbit control systems (AOCS) are responsible for the orbital behavior and pointing precision of stabilized satellites. As an alternative to a traditional reaction wheels (RW) assembly, the use of a single reaction sphere (RS) was proposed. The proposed RS consists of a spherical rotor with permanent magnets (PM) that can be accelerated about any axis thanks to a multi-coil stator that also fulfils the function of magnetic bearing. The design of the EBB is based on Proba-3 requirements.
- Angular momentum ±1.3 Nms
- Torque 34 mNm
- Torque static error 0.5%/0.02°
- Specific torque per power 0.482 mNm/W
- Density 0.295 Nms/kg 0.295 Nms/kg
- Specific angular momentum per mass
Support for initial development activities for this innovative actuator technology was provided within the framework of ESA’s GSTP program. The concept consists of a levitated sphere that can be accelerated about any axis producing a resultant torque in any direction. This system aims at replacing the three (or four) traditional reaction wheels or control moment gyroscopes commonly used in satellites and spacecrafts. It reduces the mass and power supply allocated to the attitude and navigation unit. The mass and power saved can be allocated to the payload. The concept offers performance gain; it provides more momentum for a given system mass in comparison with the current inertial subsystems. It improves the system reliability as no mechanical bearings are present and allows more flexible mission planning. Within the project, a novel Elegant Breadboard was developed and manufactured. System characterization and closed loop tests were performed, allowing rotations up to 400 rpm. Functional closed-loop experimental results were performed showing the ability of simultaneously levitating the rotor while rotating it about any desired axis up to 300 rpm.
Project Context and Objectives:
Attitude and orbit control systems (AOCS) have long since been recognized as one of the main spacecraft subsystems having a major impact on the efficiency and quality of commercial and scientific space missions. The AOCS is responsible for the orbital behaviour and pointing precision of stabilized satellites. The performance of the AOCS has a direct impact on the efficient utilisation of an essential resource of the satellite, the propellant, which in turn directly affects the lifetime of the spacecraft. It is also one of the most delicate and critical subsystem as failure of one of its constituents implies the loss of the mission. The goal of the attitude control subsystem (ACS) is to orientate the main structure of the spacecraft correctly and with the adequate accuracy while being robust, reliable and lightweight. Reducing mass, power and improving reliability while keeping the required precision are some of the major challenges in today’s design of spacecrafts. As a positive side effect, reduced ACS size is likely to lead to reduced system recurring costs and increased payload size. Basically, an active ACS system consists of sensors and actuators in conjunction with the necessary sensing and drive electronics as well as control software and related algorithms. While sensors and electronics are relatively independent with respect to the size of the spacecraft, the actuators are dependent principally as a function of inertia.
Attitude control systems traditionally need a minimum of three reaction wheels (RW) or control moment gyroscopes (CMG), but in practice, four or five wheels are common for optimization and redundancy purposes. The attitude of the satellite can be changed by the reaction to the acceleration of the appropriate wheel. Another traditional approach is to use a control moment gyro consisting of a rapidly rotating wheel held by gimbals. Applying torques on the gimbal joints changes the satellite’s attitude.
The use of a single Reaction Sphere held in position by magnetic levitation is proposed in Figure 1 as a replacement to the current approach. The sphere can be accelerated in any direction by a three dimensional (3D) motor. With respect to reaction wheel/CMG devices, spheres are a very attractive alternative. Because of its unparalleled symmetry, a hollow sphere delivers constantly a maximum inertia independently of its current rotation axis. Furthermore, a hollow sphere has the natural optimal multi axis useful inertia-to-mass and -volume ratios. It turns out that, with equivalent inertia capacities within the same volume, an arrangement with a single Reaction Sphere will offer the advantage of a reduction in mass of more than 50% with respect to a three reaction wheel configuration. Moreover, a sphere can theoretically be used in reaction, momentum, and control moment gyro models all at the same time.
Reaction Spheres were proposed more than twenty years ago but none have reached the technology readiness level needed for a commercial product, especially because the proposed designs showed poor efficiency and complex coupling between bearing and motor functions leading to unusable designs. Based on the same conceptual idea, several spherical motors, mainly for robotic purposes, have also been proposed. However, such proposed motors do not behave like a Reaction Sphere, mainly because one axis has to come out of the motor to use the produced torque, which is in addition mainly static.
CSEM proposed, in the frame of an ESA project, a novel approach and proved the validity of the Reaction Sphere concept with magnetic bearings. CSEM’s first study consisted of a trade-off between several potential concepts: inductive motor, permanent magnet motor, and reluctant motor. The synchronous permanent magnet concept was finally selected with 20-pole stator and an 8-pole (permanent magnet) hollow iron-core rotor.
As the theoretical opposite to the inductive and reluctant concepts, the synchronous concept has the advantage of linearity allowing decoupling between bearing and motor functions. Moreover, it presents a higher efficiency than the inductive concept. A schematic representation of the 8-pole rotor and 20-pole stator is reported in Figure 2. The Reaction Sphere is actively levitated and its position regulated by appropriately energizing the 20 coils around the stator. Simultaneously, the same coils are employed to accelerate the rotor about any desired axis in order to produce the appropriate torque.
The Reaction Sphere is designed so that it comprises its own redundancy. For instance, stator coils can be doubled. Only the rotor, which is a pure mechanical piece, does not include redundancy; but its design is assumed to be highly reliable with a very low probability of failure.
A proof-of-concept laboratory prototype based on this principle has been designed, built, and tested. The selected synchronous concept has been sized to approximately meet rough specifications provided by various prime contractors. The first Reaction Sphere prototype was designed to obtain a torque of 0.2 Nm and an angular momentum of 23 Nms, as shown in Figure 3.
The prototype proved the validity of the concept, confirmed some theoretical assumptions, and demonstrated its feasibility. The prototype also showed that major work needs to be done at both theoretical and technological level in order to reach the level of maturity needed for such demanding flight equipment.
ELSA main objective was to bring the Reaction Sphere to Technological Readiness Level of 5 in order for such a concept to be exploitable by the space industry, relying on Proba 3 mission AOCS requirements
ELSA aimed at improving the stator and rotor design, using extensively a multi-physics simulation environment as well as an analytical model of the system (developed within ELSA). In parallel, to reach the Reaction Sphere’s challenging specification, a new design of the system control algorithm and drive electronics had to be performed.
Finally, a spatialization roadmap had to be established.
Project Results:
Reaction Sphere Elegant Breadboard specifications
In order to establish the Reaction Sphere elegant breadboard specifications, at first, a comprehensive market survey was performed, identifying the type of space missions, namely:
• Telecommunication missions
• Earth Observation missions
• Scientific missions
• Navigation missions
• Manned missions
Based on the mission review, mission advantages and disadvantages, a reference mission was selected as baseline for the Reaction Sphere.
A demonstration mission is considered an essential step for the Reaction Sphere technology before it could be considered for other applications (science or commercial). In the short term, the only demonstration mission at ESA level is Proba-3, although some others can be identified. The target at long term would be to manufacture a product that can compete in the GEO commercial market and any other commercial market that could arise (e.g. EO). However, the entry barrier is too high and only well-proven products are accepted by commercial operators.
Therefore, other applications are needed as an instrumental way to achieve this objective. In
2011, ESA started to consider the deployment of SAT-AIS and this system, with no top level requirements for the AOCS control, could be suitable for the operational demonstration of this technology, once it is demonstrated in ground and in orbit.
Next step could be the second generation of Galileo if the requirements are compatible with this application. Once these steps are fulfilled, the Reaction Sphere would be in a position to compete in the commercial market and to have access to the challenging Science missions.
This led to the following requirements:
- Angular momentum ±1.3 Nms
- Torque 34 mNm
- Torque static error 0.5%/0.02°
- Specific torque per power 0.482 mNm/W
- Density 0.295 Nms/kg 0.295 Nms/kg
- Specific angular momentum per mass
These requirements were translated into detailed technical specifications.
Main system specifications covered:
- The electromechanical system specifications (rotor and stator),
- The power electronics,
- The control electronics.
The design specifications were as follows:
- Angular momentum ±1.86 Nms
- Stator diameter 160 mm
- Mass 6 kg (spatialized version)
- Bearing constant 4-72 – 5.28 N/sqrt(W)
- Specific torque per power 44.7 mNm/sqrt(W)
- Specific angular momentum per mass 0.310 NMs/kg
- Density 0.84 kg/dm3
Reaction Sphere Elegant Breadboard design
The RS spherical rotor and a stator hemisphere are depicted in Fig. 5 and Fig. 6, respectively. The rotor consists of eight bulk permanent magnets (PMs) with a shape optimized for manufacturability and to approach the ideal octupole magnetic distribution. The PMs are mounted on an octahedral support frame so that the rotor is mostly hollowed in order to maximize its inertia-to-mass ratio. A finishing shell covers the PMs so that the rotor is perfectly spherical, which is needed for optical position sensing preformed along three independent axes. A spherical shape for the rotor is also desired in case of accidental loss of bearing for safe touchdown of a rapidly rotating rotor. The stator consists of 20 coils in dodecahedral symmetry supported by two hemispheric shells which also integrate the three position sensors as well as 15 flux sensors used to continuously reconstruct in real-time the rotor orientation and the rotor angular velocity, which are both necessary for closed-loop control. Flux density sensors are arranged according to an optimal strategy. In the current breadboard, additional sensors are also embedded to measure the RS exported torque.
The measured radius of the rotor is 56.517 mm and its mass, which includes a balancing mechanism, is 2.85 kg. The stator is non-ferromagnetic and has been machined with PEEK material. The mass of the stator is 2.42 kg. The required angular momentum can be obtained at an angular velocity of 3190 rpm
Design optimization relied extensively on simulations. The goal of the simulation was to find the rotor design which minimize the distortion of the magnetic flux density with respect to the desired fundamental harmonic. The model agreement criterion d is introduced as a quantitative evaluation of the global distortion of the magnetic flux density with respect to the spherical harmonic of degree 3 and order 2. This spherical harmonics is the fundamental for the 8-pole rotor in Figure 7 and has interesting properties that can simplify the control of the reaction sphere
To compute d, the radial component of the simulated magnetic flux density Br(rc ,θ,φ) evaluated at rc, is decomposed on the basis of spherical harmonics up to a degree N=20.
Therefore, the model agreement criterion d is defined as the ratio between the coefficient relative to the desired harmonic and the sum of the other harmonic coefficients up to the order N.
Figure 8 shows the evolution of the d parameter along the optimization process, reaching a value close to 0.999.
Similar optimization was performed for the stator optimization.
Control algorithm
Thanks to the work performed within the ELSA activity, a novel concept of spherical actuator for AOCS concept was proposed, designed, manufactured and validated through a series of open and closed loop tests.
The goal of the closed-loop control is to levitate the rotor inside the stator and to rotate it about some desired axis. A simplified control scheme is presented in Figure 8. In this control scheme there are two loops. The first loop controls the rotor position inside the stator (magnetic bearing) whereas the second loop controls the angular velocity of the rotor.
A complete analytical model of the system was developed to establish the associated control algorithm.
This implied the development of the analytical model of eddy currents in the rotor.
The inputs of the magnetic bearing controller are the reference position (in nominal operation this should correspond to the center of the stator), the estimated rotor position, which is measured using three laser optical sensors, and the estimated force characteristic matrix KF, which relates a vector of 20 currents i to the generated force F. The matrix KF is computed as a linear combination given the estimated magnetic state, which is determined linearly and in closed-form given a minimum of seven measurements of the radial component of the rotor magnetic flux density
The force output F is subsequently transformed into a vector of currents iF through the general inverse matrix MF according to an inverse model. The magnetic bearing controller is based on a state-space scheme with displacement velocity estimator. Poles are placed to obtain a desired dynamic characteristic, which includes a closed-loop bandwidth of 10 Hz. The sampling frequency is 2.5 kHz.
The inputs of the angular velocity controller are the reference ωref, the estimated magnetic state, and the torque characteristics matrix KT. Similarly to KF, the matrix KT is computed linearly and in closed-form given a minimum of seven measurements of the radial component of the rotor magnetic flux density. To validate rotation, we initially employed a proportional controller with an angular velocity estimator. The algorithm to estimate the rotor angular velocity is based on Faraday’s law, in which the magnetic flux density is decomposed on a spherical harmonic basis, whose expansion parameters are derived from measurements of the radial component of the field collected from at least seven locations.
Rotor orientation algorithm relies on the use of spherical harmonics and was the object of a Ph. D. work.
Then, given the back-EMF voltages, the rotor angular velocity is derived employing the energy conservation principle. The resulting expressions are linear and are expressed in closed-form. Finally, the controller torque output T is multiplied with the torque general inverse matrix MT to produce the relative vector of currents iT, which summed with iF gives the total control current iF,T.
Test
In the frame of the system validation, it was noticed that the developed PWM electronics presented major limitations, as all the power amplifiers could not be used at the same time, and could not deliver the maximum voltage and current.
For that reason, it was decided to refurbish the previous linear electronics, despite its power limitation so as to perform a closed loop test campaign.
The developed RS actuator integrated in its dedicated test bench setup is reported in Figure 10. The RS EBB is equipped with 15 single-axis Hall sensors arranged on the top stator hemisphere to determine the rotor orientation and the rotor angular velocity. Moreover, three optical sensors are used to determine the position of the rotor inside the stator. Finally, a system of three micrometer screws is employed to center the rotor inside the stator for initial zeroing of the position sensors.
In Figure 11 we report an example of the angular velocity profile where the rotor is accelerated in a step-wise fashion up to ±175 rpm about a fixed rotation axis. As can be observed, the estimated angular velocity during closed-loop operation is in good agreement with the reference profile. However, the estimated angular velocity is subject to undesired fluctuations (relative error approximately 5-6% throughout the observed velocity range) and cross-coupling between axes. These fluctuations results from the derivative-based angular velocity estimation approach [20], which is perturbed by unmodeled high-order harmonics of the rotor magnetic flux density and is not accurate due to the finite sampling time. As a result, with the current angular velocity estimator, the demanding pointing requirements cannot be achieved.
The displacement of the rotor relative to the same test in reported in Figure 12. As it can be noticed the displacement is well bounded within 0.2 mm. However, we observe that the performance of the magnetic bearing controller deteriorates while increasing the angular velocity.
Possible explanations include rotor-orientation dependent errors in the force and torque models (force and torque models are estimated at each sampling time taking into account the principal harmonic of the field only) and rotor non-sphericity (the ELSA rotor has a non-sphericity of approximately 0.17 mm). As a result of these force and torque errors and rotor nonsphericity, undesired forces are exported to the test bench equipment because position sensors cannot distinguish non-sphericity of the rotor from real center of gravity displacements.
In Figure 13 we report another example of closed-loop experiments where the rotor is accelerated about a time varying rotation axis that follows a sinusoidal profile.
The estimated angular velocity is again in good agreement with the reference trajectory even though unwanted perturbation can still be observed throughout.
In Figure 14 we depicted the rotor position relative to these experiments where we observe that the displacement is bounded within 2 mm.
Although not reported in this article, the maximum angular velocity achieved with the current EBB is 300 rpm.
Main advances were achieved in the on-line determination of the rotor using flux sensors positioned around the stator.
In parallel, a complete analytical model of the system was developed to establish the associated control algorithm.
This implied the development of the analytical model of eddy currents in the rotor.
A main achievement was the development of an optimization process for the stator and rotor design parameters.
Potential Impact:
Thanks to the development of the ELSA Elegant breadboard, performance tests could be achieved. The ELSA elegant breadboard will allow exploiting the potential of the Reaction Sphere in terms of exported force and torques to show the full potential of the design with respect to Reaction Wheels or Control Momentum Gyroscopes for space applications.
The ELSA application shall be a base for answering the needs of future scientific and earth observation missions in terms of exported forces and micro-vibrations.
This need was recently expressed in the ESA AOCS harmonization process as well as in the ESA contract N0 4000110761/14/NL/MH “Market Assessment for Reaction Wheels” final presentation.
Thanks to the absence of mechanical bearing, ELSA shall be a solution for providing such a system.
Moreover, by design, the Reaction Sphere presents a ration torque / power 4 time higher than a set of Reaction Wheels.
ELSA allowed progress in terms of:
- Electromagnetic design (Reaction sphere design process and optimization)
- Control algorithms for magnetic bearing and rotor position estimation (PhD subject)
- Eddy current analysis
- Manufacturing and assembly process
Targeted market
The space sector depends a lot on the public budget for space investments but it is strongly interconnected with the demand and investments from the commercial market and industry.
In order to investigate the commercial market for ELSA, first ground and in-orbit demonstrations are needed.
There is a number of possible missions that could be accessible to ELSA in the medium term: SAT-AIS and Galileo Second Generation (GSG). Both systems will require some spacecrafts (8-16 for SAT-AIS and 30 for GSG). Succeeding in these missions will pave the way for future access to GEO market.
ELSA potential space missions based on technology development
Demonstration missions
Without ground demonstration of ELSA and in-orbit demonstration, ELSA could not be considered as an option for being used in space missions, scientific or commercial.
In a first step, also not only due to technical readiness but also the high barriers of entry into the commercial market, the target is the institutional market (ESA).
The conclusion from the Deliverable D2.2 prepared by SENER summarizes that in the short term, the only demonstration mission at ESA level is Proba-3, although some others can be identified.
Therefore, other applications are needed as an instrumental way to achieve this objective. In 2011, ESA started to consider the deployment of SAT-AIS and this system, with no top level requirements for the AOCS control, could be suitable for the operational demonstration of this technology, once it is demonstrated in ground and in orbit.
Operational missions
Next step could be the second generation of Galileo if the requirements are compatible with this application. Once these steps are fulfilled, the Reaction Sphere would be in a position to compete in the commercial market and to have access to the science missions.
All the types of space missions could be taken in consideration: telecommunication, Earth Observation, scientific, navigation and others. As it comes to ELSA, the GEO commercial market and any other commercial market that could arise (e.g. EO) is the target market on long term basis. However, looking at the trends and predictions in service sector development, as well as the size of the market, commercial GEO satellites are considered as the largest market and most profitable and that should be aimed at by ELSA in long-term dimension. However, only well-proven products are accepted by commercial operators.
GEO commercial market could be addressed and the product may be offered to all major platform manufacturers, mainly EADS and Thales in Europe.
Alternative applications
ELSA developed technology could be applied for:
- High accuracy quasi-steady accelerometer
- Energy storage device for satellites
- Spherical stepper motor (actuator)
- Torque actuator in specific applications:
- Motion actuator in terrestrial / planetary exploration robots,
Exploitation
The partners of the ELSA consortium made different contributions to the exploitation of the project results, and they will have different benefits and sources of revenue:
The technology providers will use their contacts to decision makers in the space industry to promote and market the ELSA results. The research partners will make ELSA results known in the scientific community. They will benefit from ELSA by acquiring knowledge about user requirements and application fields of their methodology. We identify the role of each partner in this process
List of Websites:
http://elsa-project.eu/
The ELSA project had the goal of improving European capacity to independently manufacture commercial and scientific satellites by bringing a new actuator for attitude and orbit control systems to a higher level of maturity. This goal is in line with ESA’s technology strategy and long term plan.
Attitude and orbit control systems (AOCS) are responsible for the orbital behavior and pointing precision of stabilized satellites. As an alternative to a traditional reaction wheels (RW) assembly, the use of a single reaction sphere (RS) was proposed. The proposed RS consists of a spherical rotor with permanent magnets (PM) that can be accelerated about any axis thanks to a multi-coil stator that also fulfils the function of magnetic bearing. The design of the EBB is based on Proba-3 requirements.
- Angular momentum ±1.3 Nms
- Torque 34 mNm
- Torque static error 0.5%/0.02°
- Specific torque per power 0.482 mNm/W
- Density 0.295 Nms/kg 0.295 Nms/kg
- Specific angular momentum per mass
Support for initial development activities for this innovative actuator technology was provided within the framework of ESA’s GSTP program. The concept consists of a levitated sphere that can be accelerated about any axis producing a resultant torque in any direction. This system aims at replacing the three (or four) traditional reaction wheels or control moment gyroscopes commonly used in satellites and spacecrafts. It reduces the mass and power supply allocated to the attitude and navigation unit. The mass and power saved can be allocated to the payload. The concept offers performance gain; it provides more momentum for a given system mass in comparison with the current inertial subsystems. It improves the system reliability as no mechanical bearings are present and allows more flexible mission planning. Within the project, a novel Elegant Breadboard was developed and manufactured. System characterization and closed loop tests were performed, allowing rotations up to 400 rpm. Functional closed-loop experimental results were performed showing the ability of simultaneously levitating the rotor while rotating it about any desired axis up to 300 rpm.
Project Context and Objectives:
Attitude and orbit control systems (AOCS) have long since been recognized as one of the main spacecraft subsystems having a major impact on the efficiency and quality of commercial and scientific space missions. The AOCS is responsible for the orbital behaviour and pointing precision of stabilized satellites. The performance of the AOCS has a direct impact on the efficient utilisation of an essential resource of the satellite, the propellant, which in turn directly affects the lifetime of the spacecraft. It is also one of the most delicate and critical subsystem as failure of one of its constituents implies the loss of the mission. The goal of the attitude control subsystem (ACS) is to orientate the main structure of the spacecraft correctly and with the adequate accuracy while being robust, reliable and lightweight. Reducing mass, power and improving reliability while keeping the required precision are some of the major challenges in today’s design of spacecrafts. As a positive side effect, reduced ACS size is likely to lead to reduced system recurring costs and increased payload size. Basically, an active ACS system consists of sensors and actuators in conjunction with the necessary sensing and drive electronics as well as control software and related algorithms. While sensors and electronics are relatively independent with respect to the size of the spacecraft, the actuators are dependent principally as a function of inertia.
Attitude control systems traditionally need a minimum of three reaction wheels (RW) or control moment gyroscopes (CMG), but in practice, four or five wheels are common for optimization and redundancy purposes. The attitude of the satellite can be changed by the reaction to the acceleration of the appropriate wheel. Another traditional approach is to use a control moment gyro consisting of a rapidly rotating wheel held by gimbals. Applying torques on the gimbal joints changes the satellite’s attitude.
The use of a single Reaction Sphere held in position by magnetic levitation is proposed in Figure 1 as a replacement to the current approach. The sphere can be accelerated in any direction by a three dimensional (3D) motor. With respect to reaction wheel/CMG devices, spheres are a very attractive alternative. Because of its unparalleled symmetry, a hollow sphere delivers constantly a maximum inertia independently of its current rotation axis. Furthermore, a hollow sphere has the natural optimal multi axis useful inertia-to-mass and -volume ratios. It turns out that, with equivalent inertia capacities within the same volume, an arrangement with a single Reaction Sphere will offer the advantage of a reduction in mass of more than 50% with respect to a three reaction wheel configuration. Moreover, a sphere can theoretically be used in reaction, momentum, and control moment gyro models all at the same time.
Reaction Spheres were proposed more than twenty years ago but none have reached the technology readiness level needed for a commercial product, especially because the proposed designs showed poor efficiency and complex coupling between bearing and motor functions leading to unusable designs. Based on the same conceptual idea, several spherical motors, mainly for robotic purposes, have also been proposed. However, such proposed motors do not behave like a Reaction Sphere, mainly because one axis has to come out of the motor to use the produced torque, which is in addition mainly static.
CSEM proposed, in the frame of an ESA project, a novel approach and proved the validity of the Reaction Sphere concept with magnetic bearings. CSEM’s first study consisted of a trade-off between several potential concepts: inductive motor, permanent magnet motor, and reluctant motor. The synchronous permanent magnet concept was finally selected with 20-pole stator and an 8-pole (permanent magnet) hollow iron-core rotor.
As the theoretical opposite to the inductive and reluctant concepts, the synchronous concept has the advantage of linearity allowing decoupling between bearing and motor functions. Moreover, it presents a higher efficiency than the inductive concept. A schematic representation of the 8-pole rotor and 20-pole stator is reported in Figure 2. The Reaction Sphere is actively levitated and its position regulated by appropriately energizing the 20 coils around the stator. Simultaneously, the same coils are employed to accelerate the rotor about any desired axis in order to produce the appropriate torque.
The Reaction Sphere is designed so that it comprises its own redundancy. For instance, stator coils can be doubled. Only the rotor, which is a pure mechanical piece, does not include redundancy; but its design is assumed to be highly reliable with a very low probability of failure.
A proof-of-concept laboratory prototype based on this principle has been designed, built, and tested. The selected synchronous concept has been sized to approximately meet rough specifications provided by various prime contractors. The first Reaction Sphere prototype was designed to obtain a torque of 0.2 Nm and an angular momentum of 23 Nms, as shown in Figure 3.
The prototype proved the validity of the concept, confirmed some theoretical assumptions, and demonstrated its feasibility. The prototype also showed that major work needs to be done at both theoretical and technological level in order to reach the level of maturity needed for such demanding flight equipment.
ELSA main objective was to bring the Reaction Sphere to Technological Readiness Level of 5 in order for such a concept to be exploitable by the space industry, relying on Proba 3 mission AOCS requirements
ELSA aimed at improving the stator and rotor design, using extensively a multi-physics simulation environment as well as an analytical model of the system (developed within ELSA). In parallel, to reach the Reaction Sphere’s challenging specification, a new design of the system control algorithm and drive electronics had to be performed.
Finally, a spatialization roadmap had to be established.
Project Results:
Reaction Sphere Elegant Breadboard specifications
In order to establish the Reaction Sphere elegant breadboard specifications, at first, a comprehensive market survey was performed, identifying the type of space missions, namely:
• Telecommunication missions
• Earth Observation missions
• Scientific missions
• Navigation missions
• Manned missions
Based on the mission review, mission advantages and disadvantages, a reference mission was selected as baseline for the Reaction Sphere.
A demonstration mission is considered an essential step for the Reaction Sphere technology before it could be considered for other applications (science or commercial). In the short term, the only demonstration mission at ESA level is Proba-3, although some others can be identified. The target at long term would be to manufacture a product that can compete in the GEO commercial market and any other commercial market that could arise (e.g. EO). However, the entry barrier is too high and only well-proven products are accepted by commercial operators.
Therefore, other applications are needed as an instrumental way to achieve this objective. In
2011, ESA started to consider the deployment of SAT-AIS and this system, with no top level requirements for the AOCS control, could be suitable for the operational demonstration of this technology, once it is demonstrated in ground and in orbit.
Next step could be the second generation of Galileo if the requirements are compatible with this application. Once these steps are fulfilled, the Reaction Sphere would be in a position to compete in the commercial market and to have access to the challenging Science missions.
This led to the following requirements:
- Angular momentum ±1.3 Nms
- Torque 34 mNm
- Torque static error 0.5%/0.02°
- Specific torque per power 0.482 mNm/W
- Density 0.295 Nms/kg 0.295 Nms/kg
- Specific angular momentum per mass
These requirements were translated into detailed technical specifications.
Main system specifications covered:
- The electromechanical system specifications (rotor and stator),
- The power electronics,
- The control electronics.
The design specifications were as follows:
- Angular momentum ±1.86 Nms
- Stator diameter 160 mm
- Mass 6 kg (spatialized version)
- Bearing constant 4-72 – 5.28 N/sqrt(W)
- Specific torque per power 44.7 mNm/sqrt(W)
- Specific angular momentum per mass 0.310 NMs/kg
- Density 0.84 kg/dm3
Reaction Sphere Elegant Breadboard design
The RS spherical rotor and a stator hemisphere are depicted in Fig. 5 and Fig. 6, respectively. The rotor consists of eight bulk permanent magnets (PMs) with a shape optimized for manufacturability and to approach the ideal octupole magnetic distribution. The PMs are mounted on an octahedral support frame so that the rotor is mostly hollowed in order to maximize its inertia-to-mass ratio. A finishing shell covers the PMs so that the rotor is perfectly spherical, which is needed for optical position sensing preformed along three independent axes. A spherical shape for the rotor is also desired in case of accidental loss of bearing for safe touchdown of a rapidly rotating rotor. The stator consists of 20 coils in dodecahedral symmetry supported by two hemispheric shells which also integrate the three position sensors as well as 15 flux sensors used to continuously reconstruct in real-time the rotor orientation and the rotor angular velocity, which are both necessary for closed-loop control. Flux density sensors are arranged according to an optimal strategy. In the current breadboard, additional sensors are also embedded to measure the RS exported torque.
The measured radius of the rotor is 56.517 mm and its mass, which includes a balancing mechanism, is 2.85 kg. The stator is non-ferromagnetic and has been machined with PEEK material. The mass of the stator is 2.42 kg. The required angular momentum can be obtained at an angular velocity of 3190 rpm
Design optimization relied extensively on simulations. The goal of the simulation was to find the rotor design which minimize the distortion of the magnetic flux density with respect to the desired fundamental harmonic. The model agreement criterion d is introduced as a quantitative evaluation of the global distortion of the magnetic flux density with respect to the spherical harmonic of degree 3 and order 2. This spherical harmonics is the fundamental for the 8-pole rotor in Figure 7 and has interesting properties that can simplify the control of the reaction sphere
To compute d, the radial component of the simulated magnetic flux density Br(rc ,θ,φ) evaluated at rc, is decomposed on the basis of spherical harmonics up to a degree N=20.
Therefore, the model agreement criterion d is defined as the ratio between the coefficient relative to the desired harmonic and the sum of the other harmonic coefficients up to the order N.
Figure 8 shows the evolution of the d parameter along the optimization process, reaching a value close to 0.999.
Similar optimization was performed for the stator optimization.
Control algorithm
Thanks to the work performed within the ELSA activity, a novel concept of spherical actuator for AOCS concept was proposed, designed, manufactured and validated through a series of open and closed loop tests.
The goal of the closed-loop control is to levitate the rotor inside the stator and to rotate it about some desired axis. A simplified control scheme is presented in Figure 8. In this control scheme there are two loops. The first loop controls the rotor position inside the stator (magnetic bearing) whereas the second loop controls the angular velocity of the rotor.
A complete analytical model of the system was developed to establish the associated control algorithm.
This implied the development of the analytical model of eddy currents in the rotor.
The inputs of the magnetic bearing controller are the reference position (in nominal operation this should correspond to the center of the stator), the estimated rotor position, which is measured using three laser optical sensors, and the estimated force characteristic matrix KF, which relates a vector of 20 currents i to the generated force F. The matrix KF is computed as a linear combination given the estimated magnetic state, which is determined linearly and in closed-form given a minimum of seven measurements of the radial component of the rotor magnetic flux density
The force output F is subsequently transformed into a vector of currents iF through the general inverse matrix MF according to an inverse model. The magnetic bearing controller is based on a state-space scheme with displacement velocity estimator. Poles are placed to obtain a desired dynamic characteristic, which includes a closed-loop bandwidth of 10 Hz. The sampling frequency is 2.5 kHz.
The inputs of the angular velocity controller are the reference ωref, the estimated magnetic state, and the torque characteristics matrix KT. Similarly to KF, the matrix KT is computed linearly and in closed-form given a minimum of seven measurements of the radial component of the rotor magnetic flux density. To validate rotation, we initially employed a proportional controller with an angular velocity estimator. The algorithm to estimate the rotor angular velocity is based on Faraday’s law, in which the magnetic flux density is decomposed on a spherical harmonic basis, whose expansion parameters are derived from measurements of the radial component of the field collected from at least seven locations.
Rotor orientation algorithm relies on the use of spherical harmonics and was the object of a Ph. D. work.
Then, given the back-EMF voltages, the rotor angular velocity is derived employing the energy conservation principle. The resulting expressions are linear and are expressed in closed-form. Finally, the controller torque output T is multiplied with the torque general inverse matrix MT to produce the relative vector of currents iT, which summed with iF gives the total control current iF,T.
Test
In the frame of the system validation, it was noticed that the developed PWM electronics presented major limitations, as all the power amplifiers could not be used at the same time, and could not deliver the maximum voltage and current.
For that reason, it was decided to refurbish the previous linear electronics, despite its power limitation so as to perform a closed loop test campaign.
The developed RS actuator integrated in its dedicated test bench setup is reported in Figure 10. The RS EBB is equipped with 15 single-axis Hall sensors arranged on the top stator hemisphere to determine the rotor orientation and the rotor angular velocity. Moreover, three optical sensors are used to determine the position of the rotor inside the stator. Finally, a system of three micrometer screws is employed to center the rotor inside the stator for initial zeroing of the position sensors.
In Figure 11 we report an example of the angular velocity profile where the rotor is accelerated in a step-wise fashion up to ±175 rpm about a fixed rotation axis. As can be observed, the estimated angular velocity during closed-loop operation is in good agreement with the reference profile. However, the estimated angular velocity is subject to undesired fluctuations (relative error approximately 5-6% throughout the observed velocity range) and cross-coupling between axes. These fluctuations results from the derivative-based angular velocity estimation approach [20], which is perturbed by unmodeled high-order harmonics of the rotor magnetic flux density and is not accurate due to the finite sampling time. As a result, with the current angular velocity estimator, the demanding pointing requirements cannot be achieved.
The displacement of the rotor relative to the same test in reported in Figure 12. As it can be noticed the displacement is well bounded within 0.2 mm. However, we observe that the performance of the magnetic bearing controller deteriorates while increasing the angular velocity.
Possible explanations include rotor-orientation dependent errors in the force and torque models (force and torque models are estimated at each sampling time taking into account the principal harmonic of the field only) and rotor non-sphericity (the ELSA rotor has a non-sphericity of approximately 0.17 mm). As a result of these force and torque errors and rotor nonsphericity, undesired forces are exported to the test bench equipment because position sensors cannot distinguish non-sphericity of the rotor from real center of gravity displacements.
In Figure 13 we report another example of closed-loop experiments where the rotor is accelerated about a time varying rotation axis that follows a sinusoidal profile.
The estimated angular velocity is again in good agreement with the reference trajectory even though unwanted perturbation can still be observed throughout.
In Figure 14 we depicted the rotor position relative to these experiments where we observe that the displacement is bounded within 2 mm.
Although not reported in this article, the maximum angular velocity achieved with the current EBB is 300 rpm.
Main advances were achieved in the on-line determination of the rotor using flux sensors positioned around the stator.
In parallel, a complete analytical model of the system was developed to establish the associated control algorithm.
This implied the development of the analytical model of eddy currents in the rotor.
A main achievement was the development of an optimization process for the stator and rotor design parameters.
Potential Impact:
Thanks to the development of the ELSA Elegant breadboard, performance tests could be achieved. The ELSA elegant breadboard will allow exploiting the potential of the Reaction Sphere in terms of exported force and torques to show the full potential of the design with respect to Reaction Wheels or Control Momentum Gyroscopes for space applications.
The ELSA application shall be a base for answering the needs of future scientific and earth observation missions in terms of exported forces and micro-vibrations.
This need was recently expressed in the ESA AOCS harmonization process as well as in the ESA contract N0 4000110761/14/NL/MH “Market Assessment for Reaction Wheels” final presentation.
Thanks to the absence of mechanical bearing, ELSA shall be a solution for providing such a system.
Moreover, by design, the Reaction Sphere presents a ration torque / power 4 time higher than a set of Reaction Wheels.
ELSA allowed progress in terms of:
- Electromagnetic design (Reaction sphere design process and optimization)
- Control algorithms for magnetic bearing and rotor position estimation (PhD subject)
- Eddy current analysis
- Manufacturing and assembly process
Targeted market
The space sector depends a lot on the public budget for space investments but it is strongly interconnected with the demand and investments from the commercial market and industry.
In order to investigate the commercial market for ELSA, first ground and in-orbit demonstrations are needed.
There is a number of possible missions that could be accessible to ELSA in the medium term: SAT-AIS and Galileo Second Generation (GSG). Both systems will require some spacecrafts (8-16 for SAT-AIS and 30 for GSG). Succeeding in these missions will pave the way for future access to GEO market.
ELSA potential space missions based on technology development
Demonstration missions
Without ground demonstration of ELSA and in-orbit demonstration, ELSA could not be considered as an option for being used in space missions, scientific or commercial.
In a first step, also not only due to technical readiness but also the high barriers of entry into the commercial market, the target is the institutional market (ESA).
The conclusion from the Deliverable D2.2 prepared by SENER summarizes that in the short term, the only demonstration mission at ESA level is Proba-3, although some others can be identified.
Therefore, other applications are needed as an instrumental way to achieve this objective. In 2011, ESA started to consider the deployment of SAT-AIS and this system, with no top level requirements for the AOCS control, could be suitable for the operational demonstration of this technology, once it is demonstrated in ground and in orbit.
Operational missions
Next step could be the second generation of Galileo if the requirements are compatible with this application. Once these steps are fulfilled, the Reaction Sphere would be in a position to compete in the commercial market and to have access to the science missions.
All the types of space missions could be taken in consideration: telecommunication, Earth Observation, scientific, navigation and others. As it comes to ELSA, the GEO commercial market and any other commercial market that could arise (e.g. EO) is the target market on long term basis. However, looking at the trends and predictions in service sector development, as well as the size of the market, commercial GEO satellites are considered as the largest market and most profitable and that should be aimed at by ELSA in long-term dimension. However, only well-proven products are accepted by commercial operators.
GEO commercial market could be addressed and the product may be offered to all major platform manufacturers, mainly EADS and Thales in Europe.
Alternative applications
ELSA developed technology could be applied for:
- High accuracy quasi-steady accelerometer
- Energy storage device for satellites
- Spherical stepper motor (actuator)
- Torque actuator in specific applications:
- Motion actuator in terrestrial / planetary exploration robots,
Exploitation
The partners of the ELSA consortium made different contributions to the exploitation of the project results, and they will have different benefits and sources of revenue:
The technology providers will use their contacts to decision makers in the space industry to promote and market the ELSA results. The research partners will make ELSA results known in the scientific community. They will benefit from ELSA by acquiring knowledge about user requirements and application fields of their methodology. We identify the role of each partner in this process
List of Websites:
http://elsa-project.eu/
