|
 |
|
Review of Gravity Probe B
4
Systems Engineering Assessment
SPACE VEHICLE
The present design of the GP-B space vehicle, which combines the science payload with a host spacecraft, has evolved over a long history dating back to the late 1960s. In addition to the normal mission objectives and launch-vehicle constraints imposed by NASA, requirements were also imposed for a set of fundamental measurements and constraints critical to the scientific goals. The vehicle requirements in turn have been allocated among the various subsystems and their hardware and software elements. The allocations were made by using a systems engineering procedure that includes feedback from the specific design criteria necessary to meet each requirement, combined with a comprehensive analysis of the contribution to the total expected measurement error from each candidate design.
The GP-B project had an unusually long period (more than 10 years) from early conceptual design through the preliminary design phase. In this interval the design team was able to develop new technologies, validate critical functional and hardware criteria, and assess their impact on the experiment. The extended development phase has allowed trade-offs among error sensitivities and design margins in order to balance risks over the whole program. The resulting development procedure for the spacecraft and its integrated payload has involved extensive prototyping of each selected element and subsystem, as well as demonstrations of most of the difficult integration processes.
At the time of the task group's review, the prototyping and integration work had demonstrated the validity and completeness of design criteria imposed to meet system requirements, as well as an ability to control the spacecraft hardware over a range of imposed environments. Final configurations of flight hardware have been established by using this foundation of experimental input to the systems engineering process. The GP-B requirements, design criteria, configurations, and interfaces now exist as a controlled database maintained at Stanford, Lockheed, and NASA, with elements as appropriate at selected subcontractors. The space vehicle subsystems are being developed to meet a set of hardware and software specifications derived from the allocated requirements by several "integrated product teams." Each team is composed of key experts selected and assigned from the personnel at Stanford, Lockheed, and major subcontractors. This approach helps to streamline the information flow, decision making, task direction, and execution and has recently come into favor at NASA. (For example, it is being implemented within the revised space station program.) the approach has been used very effectively by the GP-B project for several years.
Spacecraft Structure
The open-frame welded construction of the spacecraft permits maximum radiation from the Dewar shell to space. It also eliminates joint motion and can be machined to the precise interfaces required. The structures of the solar array panels are made of graphite epoxy and have a low coefficient of thermal expansion. This minimizes thermal shock at the day-night boundary, thereby eliminating a class of disturbing torques. Critical components of the release and deployment mechanisms for the solar array are flight-qualified and redundant. The important mechanism for trimming the spacecraft center of gravity is now in the incremental prototype phase and is expected to be finished by mid-1995. The design has adequate control authority to handle any plausible configuration and operational conditions.
Electrical Power
Peak power tracking is used to maximize the useful power from the solar arrays. A single nickel-cadmium battery unit (of two available) can support the mission. Most of the power subsystem hardware is already flight-proven, and only minor modifications are being made.
Communications
The communications subsystem is designed around flight-qualified hardware, including S-band links to the tracking and data relay satellite system (TDRSS) and redundant forward- and rear-facing antennas. Adequate data-rate link margins of 3 decibels have been incorporated.
Attitude and Translational Control
Proper operation of the attitude and translational control (ATC) subsystem is crucial to the scientific success of the mission. Primary pointing requirements are met with the proven fine-guidance system from the Hubble Space Telescope; its architecture and built-in protective measures have been well demonstrated under continuous operation in space. Backup or optional attitude control can be achieved without the gyros by using the helium thrusters described below. Other functions performed by the ATC subsystem include backup attitude and pointing using control gyros and magnetic torque rods, orbit injection and trim using GPS and/or star sensors as references, precise roll control, and position readout to 10-arc-sec accuracy.
Very-high-precision translation control is required to provide a zero-drag environment for the precision gyros and proof mass. The translation thrusters make use of the helium gas slowly boiling ot of the Dewar. The same system also maintains pressure by ejecting gas in a controlled, nearly isotropic manner. The desired thrust is produced by differential flow control through a set of low-expansion-ratio nozzles. These thrusters and their proportional control incorporate a new design, not yet proven in flight. One of the critical requirements to be met is adjustment of the sensitivity of individual thrusters to variations in inlet gas conditions. This sensitivity arises in part from the very low gas stagnation pressures, absolute temperatures, and Reynolds numbers in the nozzle. The design makes use of a nozzle-inlet pressure feedback to control a continuous flow into each thruster. The design criteria have been refined and validated in two development models, and a prototype engineering unit has been extensively tested.
Integrated Payload
The integrated payload consists of the Science Instrument Assembly (IA), the probe, and the Science Mission Dewar (MD). The SMD also forms a major structural element of the space vehicle itself. Component specifications, interfaces, and total configuration for the integrated payload were essentially complete at the time of the task group's review. Current activities are directed toward completing the verification testing of the component and subsystem hardware and addressing the cryogenic integration procedures.
The instrument package known as "Probe B" will be integrated into a ground-test Dewar early in 1995 (minus the telescope element in the SIA) and will undergo a series of design verification tests. In 1996 this probe will be upgraded with a flight-design telescope and reintegrated into the final SMD. The resulting integrated payload unit will then undergo a rigorous qualification program. Rotating it to a horizontal position will permit checkout of spin-up and caging of the science gyros in that orientation, for comparison with their vertical orientation characteristics. The flight unit will have its critical design review in the spring of 1995, and flight hardware is to be delivered in October 1996.
The SMD must provide a uniform very low magnetic field environment (10-7 gauss) for the probe and the SIA. It must maintain enough liquid helium capacity for the cryogenic needs of the SIA and still provide the required gas flow to the ATC thrusters over an operating period of up to 20 months. The task group notes, once again, that the available operating time for the experiment is one of the most important parameters determining the experiment's overall accuracy.
RISK ANALYSIS
One of the major objectives in this review is to appraise the risk that GP-B might not make an accurate measurement of the relativistic precession of a gyroscope in Earth orbit. The task group studied the objectives, design, analysis procedure, test data, and operational plans for the experiment. Using this information, and based on individual members' backgrounds in science and/or space missions, the task group arrived at varying opinions from which a consensus was formed. Summarized here is much of the information on which the group's risk assessments are based.
Overall Credibility
The scientific goal of GP-B requires putting gyros in Earth orbit with unmodeled spurious drifts no greater than 0.5 milliarc sec/yr. Before addressing the risks in achieving this spectacular performance, the task group lists some of the particulars that help to make the experiment credible:
1. Each of the four spinning gyros is a nearly perfect sphere of uniform density, operating in almost ideal free-fall conditions. Disturbances caused by atmospheric drag and other non-gravitational forces are eliminated exactly for one gyro. This is achieved by using small active thrusters to keep the case from contacting the spinning sphere, which is unsupported. With additional active control loops, the other three gyros, located close to the first, are individually given minute electrostatic supports to account for the small relative accelerations and gravity gradients. Because the support forces are tiny, the disturbing torques and gyro drifts should be tolerable.
2. To minimize any sensitive misalignments of axes, the cases of the four gyros and the reference-star telescope are made from single blocks of fused quartz. By thin-film cementing of the gyro and telescope blocks over their flat mating surfaces, the critical parts of the experiment are made into a single stable structure.
3. The quartz-block assembly and its readout electronics operate at liquid helium temperature, thus providing a number of essential properties: low mechanical creep, low thermal gradients, superconductive shielding of disturbing magnetic fields, ultrahigh vacuum to avoid disturbing torques on the gyros, and low-noise angular readouts of the reference-star telescope and gyros.
4. The spacecraft axis is nominally pointed at the reference star and given a controlled roll of about 0.25 revolutions per minute. Small misalignments of the individual gyro axes, the telescope axis, and the reference-star direction produce signal modulations at the roll frequency with amplitudes proportional to the misalignments. As long as the misalignments do not change significantly over the roll period, the signal can be processed to determine the misalignments and, in particular, the precise angle of each gyro axis from the reference star. Rapid changes in the quartz-block assembly would cause readout errors while such changes were happening, but such unlikely and occasional events could be readily identified and eliminated from the data.
5. The spacecraft roll helps in another way. Because many possible sources of spurious torques are tied to the case of the gyro, the direction of gyro drift correlates with roll phase and the net drift averages to zero over an integral number of roll cycles.
6. Aberrations caused by the Earth-orbital and annual motions of the spacecraft modulate the apparent direction of the reference star. The amplitudes, periods, phases, and directions of these aberrations are known very precisely. In the GP-B data they will have signatures similar to those of the relativistic precessions. because they are precisely known, they will not conceal the desired information; instead, the aberrations provide a built-in precise calibration of the gyro and telescope readouts that is continuously available throughout the mission.
7. If not measured independently, proper motion of the reference star during the experiment could limit the accuracy of the experiment. Consequently the proper motion will be determined by a new and very accurate technique. The selected reference star will be chosen to be bright enough for the GP-B telescope, detectable as a point radio source, and close in direction to a distant quasar. Changes in the star-to-quasar angular separation will be measured by VLBI, thus yielding the proper motion of the reference star with high accuracy.
Hardware Failure
The task group considers two possible kinds of failure of the GP-B experiment: a clear hardware malfunction leading to no credible measurement of gyro precession, and a failure to achieve the target accuracy of 0.5 milliarc sec/yr. Outright failure is a risk common to all space missions. However, much of the GP-B experiment's design and implementation has already been proven in flight. In particular, nearly all parts and functions except the science instrument package are identical to or derived from those of the Hubble Space Telescope (which was designed and built by the same Lockheed contractor team). Therefore the non-science part of GP-B should pose a smaller risk than did the more complex HST system when it was launched. The translation control for achieving local drag-free conditions has been successfully proven by the Navy's Transit navigation satellite. The control gyros that failed in the HST can be excluded from consideration becuase in GP-B they have been replaced by an entirely different design of proven reliability. The workhorse Delta launch vehicle and its operation are judged as having a low risk of failure for similar (if not stronger) reasons.
The most important concern, therefore, is the risk of failure in the GP-B science package and its supporting cryogenics. Included here are the four high-precision gyros, the reference-star telescope, the associated cryogenics and electronics, and the spacecraft translation and rotational controls that differ from equivalent HST functions. The functional reliability of the science payload depends in the first place on excellent engineering design and proven practices for the manufacture, test, and analysis of all subsystems. The task group has not identified any serious weakness in these areas; indeed, it is highly impressed with the thoroughness of attention to detail reflected in answers to its questions and the extensive documentation supplied. The functional reliability of GP-B also depends on multiple hardware and operational redundancies. The four gyros each have redundant suspension and readout electronics, as do the telescope readouts for each axis. In fact, functional redundancy throughout the spacecraft is such tat most single-point failures can be tolerated. The hardware configuration of redundant operational alternatives is fully controllable from the ground.
Dropouts of the gyro and telescope data can be tolerated over significant intervals without fatally compromising the experiment. Indeed, the telescope data are necessarily unavailable for half of each spacecraft orbit, due to occultation of the reference star by the Earth. Even with more serious and unintended dropouts, if the redundant support systems do not fail the gyros will continue to "remember" their precessions from the beginning of the yearlong experiment, and subsequent readouts can largely supersede the missing information.
Probability of Achieving the Desired Accuracy
Many factors contributing to the final experimental accuracy are testable on the ground at the component and subassembly level. These items, insensitive to the effects of weight and having been demonstrated stable, should operate reliably during the mission. Performance degradation, if it occurs, can be identified and either corrected or compensated for to the required level by any of several means.
However, such avenues can do nothing to avoid degraded accuracy caused by spurious torques on the gyros. Such torques could arise from many possible causes, and they might not be reduced sufficiently by roll averaging. Adequate control of disturbing torques is fundamental to the success of the experiment, and it cannot be demonstrated on the ground because relatively large electrostatic forces are then required to support the gyros. These supports cause correspondingly large spurious torques and consequent gyro drifts—drifts that would not exist in the free-fall conditions in orbit. Disturbing torques that might spoil the measurement in orbit are "lost in the noise" on Earth and cannot be observed or evaluated by their effects on gyro precession.
The GP-B team has made an extensive theoretical search and analysis of known phenomena that could be candidates for spoiling the experiment's accuracy. The considered list is a long one; moreover, the GP-B project has had many critical and comprehensive reviews over its long history. In these reviews no specific phenomena have been suggested that have not been proven negligible or acceptable in the overall error budget. Nor has anyone been able to fault these analyses. Needless to say, all reviewers are motivated as a matter of pride to identify new phenomena of possible concern. Nevertheless, the possibility of a new and fatal problem area cannot be ruled out by such arguments.
A commitment to launch GP-B must depend on the level of confidence remaining after allowing for concerns such as these. It is important to note that most of the tests the GP-B team would like to have performed on Earth, but could not, can be performed in orbit—before, during, and after the yearlong science experiment. An extensive plan for such measurements has been prepared, and the plan will be exercised in laboratory simulations using real hardware wherever possible. These simulations could confirm much of the preflight analysis of anticipated phenomena; they might also help to identify unanticipated sources of error and perhaps even point the way toward recovering lost experimental accuracy under some conditions.
Two powerful approaches are planned for the in-orbit tests. A series of measurements at low gyro spin frequency will be made to amplify the effects of disturbing torques. Other tests will involve explicit changes in various operating conditions, to confirm or expose their influence on the observed gyroscope precessions. Either or both techniques could reveal and calibrate a large class of anticipated and unanticipated effects that might otherwise remain hidden. Detection and measurement of a surprisingly important effect might suggest more favorable operating conditions or some other kind of accuracy-saving compensation.
The task group notes that the four gyros are made of amorphous quartz or crystalline silicon, in paired combinations with clockwise and counter-clockwise spins. Each gyro is therefore unique. This design feature was motivated by the possibility that an unexpected new effect might exhibit different signatures in one or more of the gyros. Obviously, such a result could provide further assistance with the identification and diagnosis of problems.
Sensitivity of Experimental
Errors to Key System Parameters
The task group asked a number of questions of the project management to help it assess quantitatively the risk of not achieving the design-goal accuracy of 0.5 milliarc sec/yr:
1. What are the sensitivities of the standard errors of the frame-dragging and geodetic precession measurements to key hardware design and operating parameters?
2. What are the margins of these key parameters relative to their design-allocated values?
3. What are the parameter values, either currently demonstrated or estimated, for likelihoods of 84 percent and 99.9 percent of being "better than or equal to"?
4. What is the margin in meeting the 0.5 milliarc sec/yr standard-error requirement, based on a a parameter set containing the most probable values and another set using the 84 percent likelihood values?
5. If the most critical parameters do not meet their likelihood profiles, will the experimental error degrade gracefully?
In responding to these questions the Stanford group identified 19 key hardware and operating parameters, 5 of which are especially critical to achieving the GP-B science objectives. Calculations were made to assess parameters from their currently estimated, most probable values. With the 14 noncritical parameters set at their conservative 3 values (one-sided 99.9 percent confidence limits), and the remaining 5 set at their most probable values, the geodetic and frame-dragging standard errors are estimated to be 0.20 and 0.18 milliarc sec/yr, respectively, for each of the four gyros individually. Uncertainty in proper motion of the guide star is common to all four measurements, but the total errors are dominated by effects that are uncorrelated among the gyroscopes. Taking this into account and assuming that there is adequate consistency among all four gyros, the team estimates a most probable 1 experimental error of about 0.11 milliarc sec/yr.
A more conservative approach uses the likelihood profiles for all 19 system parameters and yields an 84 percent probability of achieving standard errors for the geodetic and frame-dragging coefficients of 0.36 and 0.31 milliarc sec/yr, respectively, for each gyroscope individually. Again the largest contributions are expected to be uncorrelated, and so the total experimental error should be nearly a factor of 2 smaller.
Analysis shows that the standard errors degrade gracefully for all but 2 of the 19 parameters: gyro-readout nonlinearity, and root-mean-square pointing error on the guide star. However, sizable margins exist for these quantities (currently factors of 10 and 2, respectively) between their 3 values and the points at which they become a problem. The instrument team points out that these error analyses are based on current experimental data, without regard for expected improvements. As they move forward in their verification program, they expect many of the parameter values to be tightened up in the favorable direction.
Last update 5/10/00 at 3:35 pm
Site managed by Anne Simmons, Space Studies Board
|
|
The National Academies
Current Projects
Publications
Directories
Search
Site Map
Feedback