Figure 1. The Cluster operational orbits
The Cluster mission is a 'first' for the Agency in that it requires the delivery of four identical spacecraft for simultaneous launch on a single vehicle. This series production of spacecraft to meet a unique launch opportunity has never previously been attempted and has represented a major challenge for all involved: the Agency, Industry and the Experimenters.
Long before the start of the Cluster design phase (Phase-B) activities in October 1989, a launch date towards the end of 1995 had been fixed. This limited the time available for development, manufacture and testing of the four spacecraft to that which other satellite projects normally need to deliver only one flight model. Consequently, the logistics for Cluster hardware and software production and verification were crucial, leading to such requirements as modular designsavesystem level, allowing parallel pre-integration and late exchange of the PROMs in the Central Data Management Units (CDMUs)
Figure 1. The Cluster operational orbits
The mission orbits selected impose eclipse durations of up to 4 h and this has been a major design driver for the whole spacecraft system, and for the thermal-control, power and structural subsystems in particular. Another important design requirement originating from the scientific payload is the need for a high degree of spacecraft electromagnetic cleanliness, such that they do not disturb the local plasma fields or otherwise interfere with the sensor measurements. All of the particle experiments require unobstructed fields of view and chemical cleanliness to preserve the sensitivity of their micro-channel plate detectors. Very high magnetic cleanliness of the spacecraft is also required, to allow measurements to be made of the local DC and AC magnetic fields. In addition, these sensors (FGM and STAFF) require mounting in a position remote from the spacecraft's main body.
Globally speaking, each spacecraft has to accommodate 72 kg of payload, provide 47 W of power to the payload, and support the recording and transmission of the scientific data at rates ranging from 17 to 220 kbit/s.
All four Cluster spacecraft will be launched on a single Ariane-5 vehicle, the first qualification flight of which will deliver them into the Geostationary Transfer Orbit (GTO). The four spacecraft will be accommodated in two pairs within the two launcher payload compartments (Fig. 2). This requires that each of the lower spacecraft must have a dedicated separation system on top to carry the upper spacecraft of the pair.
Figure 2. The Cluster launch configuration aboard Ariane-5
Because the development of Cluster and of Ariane-5 have been proceeding in parallel, specific and sometimes severe requirements were imposed on Cluster to cover uncertainties in the launcher development programme. For example, conservative loads were specified for sine, random and acoustic vibration testing. Nevertheless, very late in the Cluster develop-ment programme - in fact afteracceptance of the Cluster hardware - unusually high predicted shock loads originating from launcher separations were identified by Ariane, necessitating additional, unforeseen verifica-tion activities at unit and system level for Cluster.
A further constraint resulting from the fact that it will be the first Ariane-5 launch is that launch must occur within a daily 2 h launch window during Kourou daylight time.
Since the four spacecraft will be injected into GTO rather than their mission operational orbits, each will have to perform a complex series of orbit transfer manoeuvres (Fig. 3), requiring large quantities of propellant. To make this transfer scenario manageable from the operational point of view, a strategy for handling the spacecraft as two pairs during the early mission phases has been adopted, leaving one pair twice as long as the other in GTO. The design of onboard equipment, including the solar arrays, has had to take the associated increased radiation damage into account. Operability requirements in general have played a major role during Cluster's development, to facilitate safe parallel operation of the four spacecraft by ESA's European Space Operations Centre (ESOC) in Darmstadt (D).
Figure 3. The Cluster transfer strategy from GTO
Another challenge in this context was presented by the normal ESA practice regarding the geographical distribution of work across industry in the ESA Member States. In Cluster's case, the Prime Contractor, Dornier (D), has had to coordinate the activities of 36 contractors for flight hardware and software alone.
Figure 4. Flight configuration of the Cluster spacecraft and a cutaway section
The spacecraft's cylindrical design is driven by the body-mounted solar array and also optimises the fields of view available to the experiments, which are accommodated around the rim of the main equipment platform on the upper side of the spacecraft. The height of the spacecraft body has been minimised to make optimal use of the fairing volume offered by the launch vehicle.
The compact spacecraft primary structure provides mass-efficient load paths to its mechanical interfaces. It consists of the central cylinder, the main equipment platform, a tank support structure, a platform internal to the central cylinder and a Reaction Control Subsystem (RCS) support ring (Fig. 5). The central cylinder is fabricated as a CFRP-skinned aluminium honeycomb sandwich, and the Main Equipment Platform (MEP) as an aluminium-skinned honeycomb panel reinforced by an outer aluminium ring. The MEP is supported by symmetrically arranged CFRP struts connected to the central cylinder. Overall, each spacecraft body is 2.9 m in diameter and 1.3 m high.
Figure 5. The Cluster Structural Model spacecraft during integration. The six propellant tanks, booms, main platform and central cylinder are visible, together with the harness and mass dummies of the various units
The overall design allows for parallel integration of all equipment with the Main Equipment Platform on one side, and the central cylinder with the RCS components on the other. This feature proved essential in maintaining the imposed schedule.
Six cylindrical titanium propellant tanks with hemispherical ends are each mounted to the central cylinder via four CFRP struts and a boss. The propellant carried in these tanks will constitute more than half of each spacecraft's launch mass of around 1180 kg.
Six curved solar-array panels together form the outer cylindrical shape of the spacecraft body and are attached to the Main Equipment Platform. This platform provides the mounting area for most of the spacecraft units, the payload units being accommodated on the upper surface and the subsystem units, in general, on the lower surface. The five batteries and their associated regulator units needed to provide energy to the spacecraft during the 4 h eclipses in mission orbit are mounted directly on the central cylinder.
At their lower ends, the solar-array panels support a ring accommodating many of the RCS components, including four radial 10 N thrusters. Four axial 10 N thrusters are mounted on studs on the upper and lower faces of the spacecraft. All thruster positions were carefully chosen to minimise the chances of contamination reaching the sensitive experiments.
Because the solar-array panels will experience extremely low temperatures during the eclipses, special care has had to be taken in designing their attachments to the structure and the thermal insulation of their inner faces, in order to minimise the onboard heating requirements in eclipse.
The inner equipment panel inside the central cylinder supports the single main engine, two high-pressure tanks and associated propellant-management hardware.
The RCS (Fig. 6) is configured as a conventional bi-propellant system, based on a single 400 N main engine and eight 10 N thrusters. It is arranged in two redundant branches (with the exception of the main engine), each of which is capable of performing a complex mission profile. Electrical cross-coupling permits the operation of either of the two branches from either redundant half of the Attitude Determination and Control Electronics (ADCE). The propellant is stored in six tanks pressurised by helium stored in two smaller spherical tanks. Pressure-regulation and propellant-delivery systems manage the pressurant and propellant conditioning and distribution functions. During launch, the pressurant, fuel, oxidiser and the manifold will be isolated from each other by pyrotechnically operated valves, to comply with launch-vehicle safety requirements. After each manoeuvre, the main engine and thrusters will also be isolated by additional latching valves, thereby increasing reliability by eliminating potential leak paths.
Figure 6. The Reaction Control Subsystem (RCS)
Two rigid, double-hinged radial booms on the upper surface of the Main Equipment Platform carry payload sensors. These booms are stowed for launch, as are the four payload wire booms and the two rigid, single-hinged antenna booms carrying the S-band antennas. The rigid booms consist of CFRP tubes with titanium-alloy end fittings and deployment mechanisms. The radial booms will be deployed mainl by the centrifugal force developed by the spinning spacecraft, while the antenna booms are driven by redundant springs.
The passive thermal control of the Cluster spacecraft is based on a low-emissivity concept, insulating the spacecraft from the exterior environment to the extent needed to survive the 4 h eclipses in mission orbit, whilst still allowing the internally generated heat to be rejected (Fig. 7). The thermal closeout is provided by three types of hardware: low-emissivity double foil shields on the upper and lower surfaces of the spacecraft; MLI on the top and bottom of the central cylinder, below the RCS ring and around the upper part of the satellite, enclosing the experiments; and thermal insulation of the inner sides of the solar-array panels and on the 400 N main engine. An Optical Surface Reflector (OSR) radiator is integrated into the top surface to allow for the high dissipation of the RF power amplifiers. An External Power Dumper (EPD) radiator located in the upper thermal shield within the central cylinder dissipates excess power generated by the solar arrays. Heaters are used to keep equipment within specified temperature ranges throughout all mission phases, including eclipses. Temperature control is achieved by a combination of thermostats with thermistor surveillance and thermistor-guided software control.
Figure 7. The Cluster thermal-control concept
The thermal design has been optimised for the almost constant solar-aspect-angle range (92 degrees less than SAA less than 96 degrees) that will apply throughout the nominal mission phase. During the orbit transfer manoeuvres, however, the spacecraft may experience a much wider SAA range (65 degrees less than SAA less than 115 degrees). The heat-rejection concept that has been selected therefore permits the satellite to dissipate heat through either the upper or the lower thermal shield. With these precautions, Cluster can safely withstand the complete range of solar aspect angles that will be encountered.
A heated-environment concept has been chosen for the lower spacecraft compartment, comprising RCS equipment, batteries and battery regulators. The complexity and duration of the assembly and integration activities were greatly reduced by this approach compared to a solution with insulated components, but it requires somewhat more heater power during eclipses.
All external surfaces, including the solar cells, blankets, double foils and radiator have been finished with an electrically conductive Indium Tin Oxide (ITO) coating to comply with the EMC requirements imposed by the experiments.
The central cylinder carries aluminium interface rings at both its upper and lower ends. The lower ring is compatible with the Ariane 1194 mm diameter adaptor and separation mechanism. The upper ring simulates the interface offered by the adaptor and is equipped with a separation mechanism. This allows two spacecraft to be stacked on top of one another, whilst themselves still remaining identical to the maximum extent possible.
Electrically, the spacecraft is configured in four major functional areas (Fig. 8): a power-supply subsystem including a pyrotechnics unit, an onboard data-handling subsystem, an attitude and orbit control and measurement sub-system, and a telecommunications subsystem. Dedicated, physically separated and carefully shielded harnesses interconnect the various subsystem and payload units. The payload includes a separate experiment interconnection harness.
Figure 8. The Cluster electrical system
Spacecraft power demands will be met by the body-mounted solar array and five non-magnetic silver/cadmium batteries. The batteries will power the spacecraft during eclipses and supplement the solar-array output during periods of peak power demand. Full payload operation will be supported throughout the entire mission orbit phase, except during eclipses. A block diagram of the Power Subsystem is shown in Figure 9.
Figure 9. The Cluster power subsystem
The solar arrays consist of 2 Ohm cm Back-Surface-Reflection (BSR) cells, arranged in self-compensating formations to minimise the generation of DC magnetic fields. A conductive coating on the cell cover glass minimises the build-up of differential charge potentials.
Spacecraft power is conditioned and distributed via a voltage-regulated bus and redundant current- limiting switches. Where required, permanent keep alive lines are provided to payload and subsystem units. Full protection against short-circuit or overload is provided by limiting the maximum current in any supply line. Excess solar-array output power is automatically routed to Internal Power Dumpers or shunted by commandable switching to an External Power Dumper. The power bus operates on a linear shunt regulation approach, rendering the main bus voltage extremely clean during payload measurement periods. This significantly reduces any potential electro-magnetic disturbances due to the power subsystem.
The Onboard Data Handling (OBDH) sub-system, which will perform the primary spacecraft control functions, is based on an ESA standard approach (Fig. 10). It consists of a Central Data Management Unit (CDMU), a Remote Terminal Unit (RTU), and two Solid-State Recorders (SSRs). The CDMU and RTU are internally redundant; each SSR provides memory for about 2.2 Gbit of data at beginning of life.
Figure 10. The Cluster onboard data-handling subsystem
The OBDH decodes and distributes commands, received by the telecommunication subsystem at a command bit rate of 2 kbit/s, and acquires and encodes telemetry from payload and subsystem units. This telemetry is delivered either to the telecommunications subsystem for real-time transmission to ground, and/or to the SSRs for later transmission. Dedicated high-data-rate interfaces are pro-vided to the Wide Band Data (WBD) experiment and the SSRs. Stored data will be played back at a much higher rate than real-time data, in order to reduce the duration of the downlink during the limited ground-station visibility periods. WBD telemetry will only be transmitted in real-time.
Telemetry-stream bit rates are fixed at about 2 kbit/s for housekeeping telemetry, 22 kbit/s for nominal science telemetry, 131 kbit/s for burst science-data/recorder playback, and 262 kbit/s for WBD transmission/recorder playback.
The OBDH will also provide timing and synchronisation signals to payload and subsystem units, as well as AOCMS data to the payload. It will perform a surveillance function using onboard software to provide the autonomy required because of the extended non-visibility periods.
Maintenance of the orbit and attitude of the spacecraft will be performed by the Attitude and Orbit Control and Measurement Subsystem (AOCMS; Fig. 11). Spacecraft attitude and spin data are provided by an internally redundant star mapper and an internally redundant X-beam Sun sensor. The reconstitution of attitude data, such as inertial attitude, spin rate and spin phase, will be performed on the ground. This information is essential for the interpretation of the payload science data. The necessary accuracies for these attitude data are comfortably met by the subsystem.
Figure 11. The Cluster attitude determination and control subsystem
Orbit and attitude maintenance will be performed by using control thrusters, both semi-radial and axial, together with the main engine, which will be used to perform the large orbital change manoeuvres required to reach the polar mission orbit from GTO.
Communications with the spacecraft will be established through the Telecommunications Subsystem (Fig. 12), which includes uplink and downlink capabilities to support the telecommand, telemetry and tracking functions. It will interface with the ESA Ground Segment and the NASA Deep-Space Network at S-band frequencies.
Figure 12. The Cluster telecommunications subsystem
The subsystem includes three low-gain antennas, a redundant set of transponders (including a NASA-supplied 10 W RF amplifier), an RF distribution unit, and associated RF harnesses. Two low-gain antennas are mounted on deployable booms attached to the upper and lower faces of the spacecraft. They will ensure full spherical coverage for uplinking and hemispherical coverage for downlinking. A third antenna mounted on the lower side of the spacecraft will be used until the in-orbit deployment of the lower antenna boom.
Whilst such characteristics also apply to other spacecraft programmes, the situation with Cluster differed considerably in that four flight models had to be produced within the time normally allowed for one. Although Cluster was of course not a series production in the full commercial sense, at system level it stands somewhere between the 'one-off' product of other scientific projects and the mass production in the telecommunications satellite sector, for example of satellites for mobile services.
The picture changes completely when looking at the unit and component levels: the hardware of the four flight spacecraft alone is comprised of a total of about 360 units, 16 rigid booms (without the 16 wire booms), 36 propellant tanks, 8 pressure tanks, 32 thrusters, 320 m of RCS pipework, almost 17 km of har-ness, 1440 connectors and more than 57 000 electrical contacts. Here, Cluster is indeed much closer to a series production product.
A mix of strategies originating from the two extremes of a 'one-off' and a series production were therefore employed in the Cluster programme. 'Standardised' or 'off-the-shelf' items are typically used in mass production because they represent readily available identical units at reasonable cost. Such items have therefore been used in many areas on Cluster; e.g. RCS equipment, silver-cadmium batteries, battery regulator units, sensors, booms, tape recorders and pressure tanks. Because such items are considered 'flight-proven' and usually have a long history of successful application in space, both development time and risk can be reduced by using them.
The designs of some 'existing items' like booms and battery regulator units had to be slightly changed to cope with Cluster's specific requirements, which required the initiation of a full new space-qualification process. In other cases such as the pressure tanks, waivers against the Cluster requirements were granted after careful evaluation of the item's acceptability.
Parallel work flows are a typical mass-production technique for reducing overall production time. For Cluster, the modular design at system level allowed the parallel integration of the RCS system with the central cylinder, and the various equipment items with the Main Equipment Platform. Mating of these two modules was performed after their pre-integration. Without this approach, the overall schedule could not have been met.
Parallel testing at unit and system level also became necessary to meet the overall schedule. This situation was not ideal in that the first system-level tests had to be performed in parallel with the production of the next spacecraft model's units. Compromises between requirements and actual performances and between unit/subsystem and system verification became necessary, tending sometimes to increase the risk. Further on, this situation resulted in extremely high workloads for longer periods than usual for those involved simultaneously in the unit, subsystem and system activities.
The system-level testing periods were extremely long, embracing static load testing of the Structural Model/Spacecraft Mass Dummy (SM/SMD) stack, electrical testing on the Engineering Model (EM), two sine/random/ acoustic vibration campaigns on the SM/SMD stack (before and after an SM failure), two sine/acoustic vibration campaigns for Flight Model stacks, and four thermal-balance/ thermal-vacuum tests. Most of these tests were performed at IABG in Munich (D) between March 1992 and March 1995, except for the period from mid- to end-1993 during which the EM tests took place at the Prime Contractor's site. The electrical system-level tests ran almost continuously on the various flight models.
While this situation presented logistic difficulties, the positive effect was that the various specialists worked with high efficiency because they could move almost immediately from one spacecraft to the next. A pronounced learning-curve effect due to repeated integration and testing activities also became visible at all levels, helping to achieve the prescribed overall schedule.
The flexibility during the production process that resulted from the production-line approach because more than one model per unit was available for much of the time, provided more possibilities for work arounds in the event of a failed unit. Under normal circumstances only one flight spare model is available (if at all).
The high motivation of all personnel involved in the activities was another prerequisite for success, allowing the timely solution of many unforeseen difficulties and problems. This included, for example, three-shift working at the Prime Contractor for an extended period during the early electrical-testing campaigns, to compensate for unit delivery delays of up to six months.
The timely availability of hardware and software was of the utmost importance to achieve the prescribed delivery date for the Cluster flight models. Problems with long-lead-time items and some specific items like CMOS devices, hybrids, specific heaters, double foils for the thermal top and bottom shields, etc. had to be resolved. Sometimes only limited quantities of particular parts or materials were available because of production stops in industry.
Traceability of the hardware and software has been another important element in Cluster's four-model programme. Each unit model has its own performance characteristics and its own calibration curves for the conversion of telemetry data into meaningful physical measurements, with slight nuances occurring from one model to another. It was therefore important to keep track of which model of a given unit was finally integrated to which spacecraft and in which position, not only for the ground testing, but even more so for later in-orbit operations. Databases containing this information are therefore being continuously updated until launch.
Identical functioning of the four Cluster spacecraft is extremely important, as noted earlier. This goal was achieved at unit level by applying common specification limits for all models of any given unit. Proof of specification compliance, and hence of identicality between models, was provided by the individual unit acceptance tests.
Identical functionality at system level could be checked via system-parameter measurements on each of the Cluster flight models. Only a very small scatter was found between the four spacecraft models in terms of mass, power, thermal, alignment, pointing, DC magnetic-field performance, RF link, etc., as evident from Table 1, Table 2 and Table 3.
Table 1. Launch masses of the four Cluster sapacecraft
Table 2. Achievable delta-V for actual spacecraft dry mass
Table 3. Cluster pointing performance
All four Cluster spacecraft flight models have been successfully built and verified as meeting ESA's stringent requirements. By taking maximum advantage of the production-line approach, and a number of specific design, production, integration and work-around strategies, and due to the dedication of all those involved, all four flight models have been completed in the time normally required for the production of a single flight-model spacecraft: 6 years from the start of Phase-B to launch, and just 4.5 years from the start of Phase-C/D to launch. As a consequence, the Cluster project has achieved a very low specific cost in terms of MAU per kg in orbit , making them probably the most efficiently developed ESA satellites to date. Moreover, all four are ready for launch at a date that was fixed seven years ago!
The author would like to thank H.K. Fiebrich, G. Gianfiglio, G. Jung and P. Witteveen for many fruitful discussions and ideas, which also helped in writing this article.
ESA Bulletin Nr. 84.