Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/institute/stuc/apr-06/HSTStatusApr06.pdf Дата изменения: Unknown Дата индексирования: Tue Feb 5 05:11:40 2013 Кодировка: Поисковые слова: m 43

Scheduling Experience in Two Gyro Mode
April 20, 2006 Rodger Doxsey

1


Scheduling Topics
Reminder of Two Gyro expectations Summary of Two Gyro scheduling performance
Achieved Orbits/week Two Gyro system improvements Cycle 14 Progress
Large and Treasury Programs Other C14 programs Targets of Opportunity Snapshots

Scheduling directions in the future

2


Expectations for Two Gyro Mode
Criteria for switching to Two Gyro Mode was to gain 2000 orbits relative to staying in Three Gyro Mode
Based on gyro lifetime models, this requires an effective rate of 67 successful orbits/week Expected 2% failure rate implied we need to schedule 68.5 orbits/week Three gyro scheduling rate was 80 orbits/week

To date, we are averaging 73.3 orbits/week
Failure rate has been 3.62% Effective rate has been 70.6 orbits/week

3


Scheduling performance - Orbits
Two Gyro Scheduling
100 90

80

70

60

50

40
241 248 255 262 269 276 283 290 297 304 311 318 325 332 339 346 353 360 002 008 016 023 030 037 044 051 058 065 072 079 086 093

Week

Through April 10, we have averaged 73.3 orbits per week, scheduled Peaks and valleys are due to SAA interactions
4

100


Scheduling Performance - Efficiency
Two Gyro

5


SAA interacts with Two Gyro mode

6


Two Gyro Acquisition success rate
Three Gyro Success rate in 2005: 99.02% Two Gyro Success rate: 96.38%
Bad guide star rate is the same (~1%) Additional failure rate (~2.6%) is due to errors in the FHST portions of acquisitions, prior to Guide Star Acq

FHSTs are much more critical in Two Gyro mode
Various causes for failures have been identified and are being addressed with FSW changes, Acquisition Logic changes, etc. Expect failure rate to come down as various failure scenarios are identified and fixed
7


Two Gyro Improvements
Timing changes improve schedulability Reduced Earth avoidance angle will improve schedulability Additional roll flexibility being investigated
Date Feb 2005 OOT At Time of TGM Entry Nov 1 Update Feb 27 Update April 3 Update Total Diff IRC 3.0 3.0 2.5 1.6 1.6
m

OBAD1 10.25 9.75 8.25 7.4 7.4
m

OBAD2 8.25 4.42 4.42 4.25 4.08
m

GOB 4.25 4.25 2.43 2.43 2.58
m

GSACQ 8.15 8.15 7.68 7.68 8.15 0.0
m

End-to-End Time 29.65 25.32 22.85 20.93 21.23 -8.42
m

Total FHST Time 27.78 23.45 19.63 17.71 17.53
m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

m

-1.4

m

-2.85

m

-4.17

m

-1.67

m

m

m

-10.25

m

8


Progress on Cycle 14
Execution of Cycle 14 program is proceeding well
Scheduling team has adjusted quickly to two gyro mode

Recent DDs Weaver - "Confirm the Discovery of Two New Satellites of Pluto" Mountain - "ACS Mosaic of M82" Choi - "Observations of the Active Centaur 60558 2000 EC98" de Pater - "Quit Winking: Jupiter Opens its Other Eye" Simon-Miller - "Rapid Response: Jupiter's White Oval Turns Red"

9


Cycle 14 Orbit Execution Rate
Cycle completion from start
100 90 80 70 60 50 40 30 20 10 0 Jul Aug Sept Oct Nov Dec Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Month of Cycle Cycle Cycle Cycle Cycle 11 12 13 14

10


Cycle 14 Treasury/Large Programs
Program Title PI This 10496 Dark Energy Studies Perlmutter 10632 Galaxies at z>6.5 in the UDF Stiavelli 10530 PEARS Malhotra 10775 Globular Cluster Survey Sarajedini 10504 Cosmic Reionization Ellis 10556 Neutral Gas Survey Turnshek 10551 Gamma-ray Bursts Kulkarni 10505 Formation Gallart Star 10592 Infrared Galaxies Evans 10610 Extrasolar Planets Benedict 10503 Formation Da Costa Star 10497 SN/Hubble Constant Ia Riess as of Apr 9: total: % week Observed Allocated %complete planne 1 132 219 60.3% 0 113 204 55.4% June 0 200 200 100.0% 2 58 134 43.3% April-Ju 0 31 110 28.2% May-Au 3 73 109 67.0% 0 2 100 2.0% 0 14 97 14.4% April, J 2 62 88 70.5% 0 69 72 95.8% 0 0 69 0.0% May/Ju 0 64 68 94.1% 8 818 1470 55.6% 56%
11


Targets of Opportunity
PI prop_id allocated completed scheduled unplanned to Left Perlmutter 10496 219 112 72 35 16% Smartt 10498 8 1 0 7 88% Grundy 10508 12 5 3 4 33% Bennett 10544 4 2 0 2 50% Kulkarni Berger Fox Tanvir 10551 10616 10624 10633 GRBs 100 24 42 22 188 2 6 18 2 28 9 0 0 0 9 89 18 24 20 151 89% 75% 57% 91% 80%

Backlog of unexecuted ToOs in Spring is common GRBs are SWIFT activated
12


Swift Pointings

13


Snapshot scheduling
Snapshots fill gaps that are left after GO programs are scheduled
Scheduled automatically, not manually Number and size distribution of gaps is determined by the success in scheduling GO programs

Improvements in scheduling GO programs have reduced gaps available for Snapshots

14


10

20

30

40

50

60

70

80

90

100

110

0 01190 01239 01288 01336 02021 02071 SM3B 02112 02161 02210 02259 02308 02357 03041 03090 03139 03188 03236 03286 03334 04019 04068 04116 04166 04215 04264 04313 04362 05045 05093 05143 05192 05241 05290 05339 06023 06072 Two Gyro Cycle 14 Cycle 13 prime Cycle 11 Cycle 10

Scheduling rates for Cycles 10-14

Weeks

Cycle 12

SNAPs

15


Recent Snapshot History
Snapshot opportunities have been decreasing over the last several cycles
Full orbit gaps have nearly completely disappeared Partial orbit gaps are decreasing
25.00 Cycle 10 (pre-SM3B) 20.00

In Three Gyro mode: GO Orbits + Snaps=95
Cycle 11 Cycle 12 Cycle 13 Plus

15.00

10.00 Cycle 14 (Two Gyro) 5.00

In Two Gyro mode: GO Orbits + Snaps=82 (so far)

0.00 72.00 74.00 76.00 78.00 80.00 82.00 84.00

16

Orbits/week Scheduled


Recent Snapshot History
Contribution of Snaps to Observing Efficiency has been decreasing
Reflects fewer Snaps scheduled Shift to shorter Snaps
18.00 16.00 14.00 12.00 10.00 8.00 6.00 4.00 2.00 0.00
01183 01239 01295 01351 02042 02105 02161 02217 02273 02329 03020 03076 03132 03188 03244 03300 03356 04047 04103 04159 04215 04271 04327 7 05073 05129 05185 05241 05297 3 06044 0501 0535

Time (2001 to present)

17


Recent Snapshot Durations
Snap Scheduling Statistics by Cycle
Cycle 12 - 03.181-04.180 250 Cycle 13 - 04.180-05.184 Cycle 14 (partial) - 05.184-06.100 Two-Gyro - 05.241-06.100

200

150

100

50

0 5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-45 45-50 50-55 55-60 >60 Snap Duration (minutes)

18


Scheduling and the SAA
SAA passages are responsible for gaps in the schedule
7-8 SAA passages per day (of 15 orbits/day) 1 - 30 minute duration SAAs are fixed in time by orbit Location relative to occultation pattern depends on target location

Tricks for increasing number of orbits scheduled
Find targets that "hide" the SAA
Not always possible

Use 3 SAA-impacted orbits for a nominal 2 orbit visit
Works best for short exposure programs

Schedule nominal orbits in CVZ "Craft" visits to be shorter than nominal visibility time

19


How many orbits might we schedule?
Physical orbit is the 96 minute period for one trip around the Earth
105 physical orbits/week

TAC orbit is the nominal visibility period for one uninterrupted physical orbit IF there were no SAA, we would schedule ~105 TAC orbits/week IF we could use no SAA orbits, we would schedule ~52 TAC orbits/week IF all the SAA time were put in one contiguous segment, we would schedule ~90 TAC orbits/week IF SAA passages never intruded on target visibilities, we would schedule ~105 TAC orbits/week
20


How do we do at utilizing orbits ?
112

Cycle 13
110

108

Cycle 11

106

104

102

Cycle 10

Cycle 12

100

98

96

Cycle 14
94

92 72.00 74.00 76.00 78.00 80.00 82.00 84.00

Orbits/Week Scheduled

21


Cycle 14 Snapshots
23 programs with 1843 targets were accepted for Cycle 14 Overall completion level is low
Project 23% overall completion by July 1
Cycle 14 Snap Completion level 340 of 1843 executed as of 4 April
100.00 90.00 80.00 70.00 60.00 50.00 40.00 30.00 20.00 10.00 0.00 10567 10512 10490 10612 10573 10488 10576 10514 10606 10536 10491 10548 10587 10534 10588 10525 10627 10626 10523 10629 10547 10634 10623

SNAP Program

22


Why is the completion level so low?
We solicited too many from the TAC (2000 vs. 1500)
At the time, we were concerned that the Long-Range plan might undersubscribe some weeks (seen in early Two Gyro scheduling tests) Such situations would leave many gaps to be filled

Two Gyro Snap scheduling rate is lower than Three Gyro (8/week vs. 12/week) when schedule is packed
No estimate was made of Two Gyro Snap scheduling rate (too much work) Two Gyro Snap scheduling rate was expected to be lower

Projected completion without these effects:
28% if we selected 1500 Snaps 34% if we scheduled current pool at Three Gyro rates 41% if we scheduled 1500 at Three Gyro rates Cycle 13 overall Snap completion rate was 40%
23


Distribution across Snap Programs
Some programs schedule more frequently than others
This has always been the case (e.g. STIS MAMAs) Shorter Snaps are more likely to schedule than long ones Restricted sky availability in Two Gyro mode may increase the dispersion among programs (especially with time)
Prop ID 10567 10512 10490 10612 10573 10488 10576 10514 10606 10536 10491 10548 10587 10534 10588 10525 10627 10626 10523 10629 10547 10634 10623 Cycle 14 Instrument ACS ACS ACS FGS ACS ACS ACS ACS ACS ACS ACS ACS ACS WFPC2/ACS ACS ACS ACS ACS ACS ACS ACS ACS ACS Executed 0 5 2 6 2 4 9 35 13 19 23 10 31 8 20 29 14 40 27 5 18 2 18 Next Cycle Requested Comp_Level_(%) Date_report 80 0.00 4-Apr-06 150 3.33 4-Apr-06 65 3.08 4-Apr-06 70 8.57 4-Apr-06 27 7.41 4-Apr-06 40 10.00 4-Apr-06 60 15.00 4-Apr-06 250 14.00 4-Apr-06 83 15.66 4-Apr-06 100 19.00 4-Apr-06 124 18.55 4-Apr-06 50 20.00 4-Apr-06 118 26.27 4-Apr-06 36 22.22 4-Apr-06 80 25.00 4-Apr-06 107 27.10 4-Apr-06 50 28.00 4-Apr-06 150 26.67 4-Apr-06 92 29.35 4-Apr-06 17 29.41 4-Apr-06 50 36.00 4-Apr-06 5 40.00 4-Apr-06 39 46.15 4-Apr-06 visibility 1-3 orbits 2200s 3300s 2000-2800s 1700-2500s 2000s 1500s 1900s 1500s 2500-2900s 2700s 1750s 1000-1500s 1400-1700s 1450s 900-3000s 1300-1800s 2350s 2500s 1900s 1200-1400s 2700s 1500s sky loc. NGC1407 J. Trojans scattered CYG OB2 Gplane scattered scattered Kuiper B scattered scattered scattered scattered scattered Uran/Nep scattered scattered scattered scattered scattered SMC scattered scattered scattered

24


How Do We Schedule Snapshots?
Step 1. Develop Snap list for scheduling each week
Randomize all legal Tack on randomized Tack on randomized Tack on randomized visits visits visits visits of of of of C14 C14 C13 C13 programs programs programs programs <50% >50% <50% >50% complete complete complete complete

Step 2. Take first 200 visits from this list
In practice, only take from the first list (C14 < 50% complete)

Step 3. Develop scheduling score for each visit of the 200
Uses same scoring/weighting algorithms as prime observations Factors included:
GO priority assignment Moving target Running out of time (last 8 weeks) Scheduling efficiency Others not entirely clear (code last fiddled ~10 years ago)
25


How Do We Schedule Snapshots?
Step 4. Sort 200 visits by score Step 5. Pass through entire list of 200, trying each visit in all remaining places in the schedule
If a visit can schedule, it is put in that place, which is no longer available for other Snaps

Step 6. Complete other steps of scheduling process (TDRSS, command loads, etc.) Step 7. Update database for scheduled visits

26


Potential Operational Improvements
Use more than 200 visits each week
Not likely to schedule more Snaps, but doesn't cost much

Revise scoring scheme
Develop weighting specific for Snap programs
Bias towards efficiency ? (try longest first) Bias towards flat completion across programs ?

Reduce 50% bias point in initial selection step
Small effect, drops some programs from list of 200 Could reduce Snap scheduling, since shortest Snaps will be biased against

Increase potential for C14 scheduling in C15
Put C14 < 50% ahead of C15>50%

27


Scheduling directions for the Future
Continued small tweaks to Two Gyro mode
FHST induced failures will be reduced to ~1% (my guess) Additional reductions in timing will provide small increases in sky availability (a few days for typical targets) Increase roll ranges will provide small increases in sky availability

Some continued improvements in GO program scheduling
SAA hiding Handling of Large/Treasury programs Dispersion of targets in selected C15 programs will have as large an impact on scheduling rate

Little overall change in snapshot situation
Improvements in Snap processing may add 1 per week Improvements in GO scheduling may reduce Snap opportunities Emphasizing very short exposures might open up some new ground

Some gains achieved in the next two years will apply after going back to 3 Gyro mode (SM4)
28