The SOHO Mission Halo Orbit Recovery from the

Attitude Control Anomalies of 1998

Craig E. RobertsComputer Sciences Corporation

Flight Dynamics FacilityGoddard Space Flight Center

Libration Point Orbits and Applications

10 - 14 June 2002

Parador d’Aiguablava

Girona, Spain

Contents

• Introduction to SOHO and its Halo Orbit• SOHO Stationkeeping Technique• June 1998 Anomaly and Recovery• December 1998 Anomaly and Recovery• Conclusion

Solar Heliospheric Observatory (SOHO) Spacecraft

SOHO Mission Overview• Launched 3 December 1995; Sun-Earth/Moon L1 Halo

Orbit Insertion (HOI) March 1996

• Joint ESA/NASA mission; second NASA mission dedicated to Sun-Earth L1 halo orbit

• Purpose: Sun and solar wind science

• 3-axis stabilized; closed loop attitude system with momentum wheels, gyros (formerly), Sun sensors, FHSTs

• Steering laws keep S/C X-axis and instrument boresights Sun-pointing; Z-axis nods to stay aligned with Sun’s spin axis over course of year

• Hydrazine blowdown propulsion for translational control, momentum management and backup attitude control

• Two fully redundant strings of 8 4.5 N thrusters per string

SOHO’s Halo: View from NEP

Lunar Orbit

To Sun

L1

SOHO6/10/02

ACEEarth's motion

Earth

SOHO spans:

1.33 (106) km in y

4.13 (105) km in x

Solar Rotating X-axis

Solar Ecliptic Rotating Frame XY Plane Projection

C.E. Roberts/CSCMay 2002

SOHO’s Class 2 Halo: Side View

SOHO L1 Halo Orbit and ACE L1 Lissajous Orbit

Lunar Orbit

To Sun

L1

SOHOCLASS 2

HALO

ACE CLASS 1

LISSAJOUS

4.5 deg SEZ Cone

Earth

Solar Ecliptic Rotating Frame XZ Plane Projection

AZ = 120,000 km

AX = 206,448 km

C.E. Roberts/CSCMay 2002

SOHO’s Halo: Looking Sunward from Earth

Ecliptic

Solar Disk

SOHO Halo Orbit -- Sunward View

SEZ: 4.5 deg radius

AZ = 120,000 km

AY = 666,672 km

Solar Ecliptic Rotating YZ -plane Perspective Projection View

L1

Maximum Elongation = 25.6 deg

North End Closest to Earth

10 June '02

C.E. Roberts/CSCMay 2002

SOHO Halo Orbit Stationkeeping Technique

• Single-axis control

• Thrust vectors aligned with S/C body X-axis (X-axis always Sun-pointed)

• Thrust/V then parallel or anti-parallel to S/C-Sun line

• Upshot: V basically along Earth-Sun line (GSE X-axis)

• Trajectory propagated to a candidate SK epoch; V applied and differentially corrected as trajectory propagated toward a target goal (RLP VX = 0) in subsequent RLP XZ-plane crossing; repeated until two halo revolutions are achieved

V toward the Sun increases orbital energy, preventing halo decay back toward Earth

V away from Sun decreases orbital energy, preventing escape into free heliocentric orbit

Stationkeeping Realities• LPO correction costs increase exponentially from epoch of

last impulse; doubling constant 16 days

• Burn performance deviations (errors) dominate orbit knowledge errors; so, are biggest contribution to the magnitude of the next SK burn

• For SOHO, we prefer to burn well before correction cost has grown to 1.0 m/sec

• SOHO SK technique nominal performance: 3 to 4 burns per year; 90 to 120 days apart; 2 to 3 m/sec/year

• However, SOHO attitude control realities intrude:

- Momentum dumps needed at intervals (3 to 4 per year); dump residual Vs can be 2 to 8 cm/sec along Sun line

- SK schedule tied to momentum dump schedule so that the dump’s residual Vs can be offset

Stationkeeping Realities, Part 2: ESRs

• Worse than momentum dumps, frequent ( 3 to 4 per year) onboard attitude anomalies called Emergency Sun Re-acquisitions (ESRs) occur randomly and require thrusters

V impact can be of order 1.0 m/sec before attitude stabilization restored (thrusters off/ wheels back on)

Vs from ESR events must be countered by special orbit recovery burns as soon as possible–usually within a day or two–after the ESR

Vs to recover from typical ESRs comparable in magnitude to normal SKs, i.e., < 1.0 m/sec

• Recovery burn Vs opposite in direction to ESR Vs

• Although identical in maneuver technique, we distinguish between normal SK and post-ESR orbit recovery burns

Post-ESR Recovery

• Meaningful post-ESR trajectory re-targeting requires that the ESR V first be modeled into the trajectory (can’t wait for OD)

• Unfortunately, in SOHO ESR mode, the telemetry lacks usual burn-related parameters like thruster counts; so ESRs cannot be reconstructed from propulsion data alone

• Fortunately, both the ESR timeframe and net V can be measured directly from DSN Doppler tracking data (actual ESR burn Doppler compared to no-burn predicted Doppler)

• ESR V then modeled into the trajectory as a finite burn event using trajectory design software (Swingby); model is adjusted until output radial delta-V matches Doppler

• Trajectory then propagated up to a candidate recovery burn epoch, and halo is re-targeted from that point

The June 25, 1998 Disaster• Problematic response to an ESR event by spacecraft

controllers led to the spacecraft rolling off into a tumble, with a resulting loss of communications and power

• Out of contact for several weeks, many were fearing the mission lost….though there were reasons for hope

• Attitude simulations suggested that SOHO likely settled into a slow spin about its major axis of momentum, but with the solar panels roughly edge on to the Sun (depriving SOHO of power)

• Predicted that attitude geometry would be such that by mid-summer the panels would begin getting some Sun

• Radar skin contact was made via Arecibo/Goldstone test on July 23, verifying predictions of position in halo

• DSN successfully made brief radio contact on August 3

Disaster’s Impact on Trajectory

• Doppler data covering the ESR prior to tumble was available; but once tumble began station lock was lost, depriving us of tracking and telemetry

• Though there was intermittent dropout prior to complete loss of contact, Doppler indicated that a total of only several cm/sec was imparted prior to the tumble, but resulting in a net delta-V of just 1.4 cm/sec Sunward

• Doppler and some later analysis suggested a tail-off of the thrusting and a possible thruster shutdown not long after tumble began (but this was highly uncertain)

• So, the best guesstimate was that a mere 1.4 cm/sec (net Sunward) was imparted to the orbit; this was modeled into the reconstructed “best estimated trajectory”, which was used to support subsequent search operations

Early Assessment

• Next 8 slides are a sampling of historical documents from the 1998 crisis; they are hand-annotated screen snaps from a brief study of possible post-accident trajectories

• They show the “Telemetry Dropout Case”, or TDC (the trajectory as known at the point of loss of communication), trajectory as well as a narrow range of dispersions from that case

• Wildly different outcomes from such a narrow range of dispersions underscore the extreme sensitivities of LPOs to small perturbations

• The plots were used to impress on the SOHO Project that there was a very wide range of possible outcomes, and that hopes of recovering SOHO depended on detecting it, and restoring it to health, sooner rather than later

Six Possible Escape Trajectories to Solar Orbit for +1 cm/sec Dispersions (+1 to +5 cm/sec) from Telemetry Dropout Case, or TDC

Post-Escape 11.6 Year Heliocentric Earth-Return Trajectoryfor the TDC Case (Solar Rotating Frame)

Escape Trajectories for other possible small dispersions:from TDC 1 cm/sec to TDC 6 cm/sec

First Dispersion Case (TDC – 7 cm/sec) with Fall-back to Earthwith Temporary Capture and Eventual Escape

TDC – 8 cm/sec Dispersion Case

TDC – 9 cm/sec Dispersion case

TDC – 10 cm/sec Dispersion Case: First Capture into High-Energy, Long-Duration Chaotic Orbit with Lunar Encounters

TDC – 11 cm/sec Dispersion Case:

SOHO Recovery #1

• Luckily and amazingly, the predicted trajectory turned out to be very close to truth, which led to the successful Arecibo contact

• Carefully, over 5 weeks, batteries were re-charged, the propulsion system was thawed, and communications were ramped up in a pre-planned manner

• OD operations resumed, though with spotty, poor-quality tracking data from intermittent contacts

• Though OD solutions were uncertain, they appeared to confirm that the June 25 anomaly could not have imparted more than 5 to 10 cm/sec, at most

• A planned 5-day, Sun-pointing attitude reacquisition procedure commenced on Sept. 16 and finished Sept. 22; however this involved roughly 5 days of thrusting, imparting 3 m/sec more in the Sunward direction

SOHO Recovery #1, Continued• Thrusting effects from the September attitude reacquisition

(done in ESR mode) were modeled into the trajectory, again relying on Doppler observations

• The first halo recovery V maneuver (6.2 m/sec) was performed on Sept. 25, correcting for both the original halo degradation and the excess energy imparted by the September ESR

• More or less normal operations resumed, and two more orbit recovery maneuvers in October and November fully restored the halo orbit

• SOHO systems came through it all in flying colors (science worked as well as, or better than, before), with the exception of two of the three gyros that were now dead

• Optimism was high going into December, yet the specter of just a single gyro remaining hung over the mission

December 21, 1998: SOHO Disaster #2

• As we were preparing for MM and SK maneuvers on Dec. 21, the last gyro suddenly died

• SOHO plunged into another ESR, with no way of returning to a normal mode of Sun-pointing stabilization

• In ESR mode, the two Sunward jets that control yaw, pulse as much as 5 times more frequently than the two canted, anti-Sunward jets responsible for pitch

• This behavior imparts a net V in the Sunward direction, increasing energy of the orbit, inducing it toward escape

• Hence, we had a virtual continuous, low-level thrust situation on our hands, with no known end in sight

• Initially, Sunward V amounted to as much as 0.65 m/sec imparted per day, using as much as 0.7 kg of fuel

SOHO Disaster #2 (continued)

• Given the sensitivity of halo orbits and exponential behavior of correction, SOHO was in a very grave situation

• While the SOHO team sought ways to re-configure SOHO to operate without attitude thrusting or gyros, my job was to come up with a scheme to keep SOHO home at L1

• We faced the prospect of never getting out of ESR mode; in that case, attitude thrusting would have exhausted the 186 kg of fuel then remaining in some 6 to 9 months (depending on assumed consumption rate)

• This estimate was made difficult by lack of thruster telemetry data

• Duty cycling had to be estimated via indirect means, and we could see from Doppler observations that thrusting varied

• Needed a way to model what was happening to the orbit

Orbit/Thrusting Modeling Method

• Monitor/record R/T Doppler and measure net radial (station line-of-sight) delta-V over a given time period

• Model ESR events in Swingby as reconstructed finite burn events using B-branch yaw and pitch thrusters

• The yaw:pitch pulsing ratio was 5:1, or adjusted as needed to track observed radial delta-V

• Thrust duty cycles were inferred that would yield the radial delta-V actually observed

• ESR duty cycles were low compared to planned burns, but over 24 hours cumulative V could approach 0.65 m/sec

• This approach was applied in the early post-anomaly period as a way to update the pre-anomaly orbit; the technique was then continued/extended over the course of the 40-day ESR to compute orbit updates

ESR Delta-V Day by Day (from Doppler): Dec. 21 to Jan. 6

ESR Cumulative Delta-V (from Doppler): Dec.21 to Jan. 6

“Low Thrust” Trajectory Evaluation

• Halo orbit was rapidly degrading; escape into solar orbit was inevitable if intervention was not forthcoming

• Computed single-shot correction maneuvers for various dates (assuming constant-level ESR thrusting), but they would prove too large for crippled S/C to perform

• Point of no return reached by February, if nothing done at all

• Constant ESR duty-cycle modeling showed fuel exhaustion by May at earliest, August at latest, depending on assumed duty-cycle level

• Continuous fuel drain was definitely a concern, but of lesser priority than prompt intervention and recovery on behalf of halo orbit

• S/C engineers estimated that they might devise a way to escape ESR mode by approximately February 1st, 1999

Nominal Halo Shown with an Indefinite ESR Escape Trajectory, and an Un-recovered One-day ESR Event Trajectory for Comparison

Indefinite ESR Escape Trajectory, and Guided Continuous Thrust Trajectory through fuel exhaustion and fallback with Earth Capture

SOHO Recovery #2 Maneuvers

• Could correct with one or two big burns; but it was deemed not advisable by S/C experts to do a V over 10 m/sec

• This meant corrections would need to begin in early January; waiting much longer would make it difficult to recover, and before too long, impossible to recover

• Conceived of a series of approximately four burns, all under 10 m/sec, spread over January to counter the ESR and reduce orbital energy

• But new, open-loop style commands needed to be devised and implemented; ultimately, regular orbit burn duty cycles were reduced from normal 75% to just 5%, making burns of given delta-V 15 times longer to execute than formerly

• Final cleanup burn(s) could wait until S/C back in normal mode (attitude thrusters off, control returned to wheels)

SOHO Recovery #2: Complicating FactorsHalo recovery maneuvers faced numerous complicating

factors, including but not limited to:

• No thruster counts available, making thrust reconstruction possible only via indirect methods/estimations

• Usable OD was not attainable during the entire ESR period due to presence of effective continuous thrust

• Loss of gyros meant onboard software no longer viable for closed-loop maneuver control; workaround methods improvised to implement delta-V maneuvers

• Perturbations from the ESR were unpredictably variable

• Continuous monitoring of the Doppler necessary to gauge accumulating ESR V and the spacecraft roll rate

• Correction maneuver schedules shifted frequently for a variety of reasons, making prediction/planning difficult

SOHO Recovery Maneuver History

OrbitManeuver

Event

Date(m/d/y)

Days sinceLast Orbit

Burn Event

JetsUsed

PlannedV

(m/sec)

AchievedV

(m/sec)

VError(%)

FuelUsed(kg)

SK-01 5/23/96 63 1,2 0.3067 0.3089 +0.714 0.3353SK-02 9/11/96 112 1,2 0.4541 0.4578 +0.808 0.4925SK-03 1/14/97 125 1,2 0.0432 0.0411 -4.861 0.0490SK-04 4/11/97 87 1,2 0.1887 0.1892 +0.235 0.2064SK-05 9/04/97 146 1,2 1.8876 1.8972 +0.506 2.0258SK-06 11/29/97 86 3,4 0.0396 0.0408 +2.84 0.0345SK-07 12/19/97 20 1,2 0.3984 0.3956 -0.703 0.4263SK-08 4/17/98 119 1,2 1.4375 1.4350 -0.179 1.5441RM-01 9/25/98 161 1,2 6.21 6.21 0.0 6.666RM-02 10/16/98 21 3,4 2.0 1.924 -3.78 1.687RM-03 11/13/98 28 3,4 2.285 2.293 +0.350 1.960RM-04 1/07/99 55 1,2 9.77 8.082 -17.28* 8.60RM-05 1/19/99 12 1,2 8.69 8.64 -0.576 9.20RM-06 1/26/99 7 1,2 4.06 3.95 -2.62 4.27RM-07 2/01/99 6 3,4 0.32 0.323 +0.77 0.267RM-08 3/05/99 32 1,2 0.125 0.122 -2.35 0.130SK-09 6/17/99 104 3,4 0.4652 0.4596 -1.21 0.3835

Notes: RM = Recovery Maneuver and C/I = Calibration Inconclusive * large V error due to onboard software problem connected to ESR mode

Conclusion: How Was It Done?

• Modeled the ESR event as a sequence of finite burns, updating the thrust rate piece-wise as the Doppler data accumulated; model extended daily over the entire 40-day event…providing us with the needed orbit updates

• Predicted trajectories were propagated in continuous low-thrust finite burn mode, assuming an average duty cycle

• A series of small (<10 m/sec) correction maneuvers were placed at intervals over a month, with the final burn of the series used as the independent halo re-targeting variable

• The series of correction burns countered the cumulative ESR damage and offset the halo energy just enough such that continued ESR thrusting added the energy back at just the right rate to maintain a halo-like orbit

• The correction series was “tuned” as we went, such that at the end of the ESR there was only 15 cm/sec error left