1. NGAO topical discussion: Observing Efficiency and Uptime Budgets
Telecon Meeting 12/7
D. Le Mignant and E. Johansson
7 Dec. 2006
Attendees; R. Campbell, J. Lyke, E. Gates, M. Perrin

2. Discussion Outline NG AO background information Our tasks
Proposal
NG AO SEMP and WBS…
Our tasks
Observing Efficiency Budget
Observing Uptime Budget
Current Observing Efficiency and Uptime: words from…
Keck
Lick?
Palomar?
Next Generation Definitions for
Observing efficiency
Observing Uptime
WBS Workscope:

3. WBS 3.1.1.11 & 12 Observing Efficiency:
The purpose of this performance budget is to determine what will be required to meet the Observing Efficiency requirement. Also, report on the lessons learned with current LGS AO systems (Keck, Gemini, ESO, etc). A list of all the items contributing to the loss of LGS AO-corrected integration time will be produced along with reasonable allocations of the observing efficiency budget amongst these items.
Observing Uptime Budget:
The purpose of this performance budget is to determine what will be required to meeting the observing uptime requirement. This budget is only intended to cover the NGAO facility and science instruments (and not the telescope or facility). A list of all the items contributing to downtime will be compiled along with a distribution of the uptime budget amongst these items.

4. 101 nights of Keck II LGS AO ops since Nov. 04 till Jul. 06
Our plans for Keck II LGS in the next couple of years are rather modest.

5. Keck Overall Efficiency: 101 nights
Bad weather impact:
~17% nights dome closed - winter weather
~21% nights impacted by marginal weather
Laser faults
Lost: 2 full and 5 1/2-nights
9 nights with ~ 1h lost
AO faults
Minor time lost yet present for 50% of nights
Laser Traffic
~ 2% impact
Overheads
A BIG chunck!

6. Keck Overall Efficiency: overheads
LGS AO checkout
30min/night
Telescope slew and pointing
Target ID and centering
LGS AO readiness
min/target
LGS AO optimization
2min per hour on target
Telescope/AO handshakes
30+ sec per dither
Scientific instrument setup and readout
Observing strategy    
Ref: 2006 SPIE papers and some Keck
internal discussion for K1 LGS AO

7. Keck Observing Efficiency: Lessons learned
Keck NGSAO observing efficiency for nights w/o weather or technical problems at best vary from 25% (snapshot surveys, Lp and Ms obs) to 60-80% for deep-exposure science programs.
LGSAO shows roughly the same values, except that it is more impacted by weather and technical problems
DLM’s conclusions: For a reliable system in good weather conditions, we are currently mostly limited by
Serial (vs parallel) algorithms (DCS /inst/AO) during observations
Under-designed telescope pointing and acquisition systems
Under-designed AO nodding/dithering hardware and software
Under-designed science instrument readout

9. Review of Keck II LGS AO uptime statistics
Major contributors to system downtime:
AO Faults
Laser Faults
Other Faults (instrument + telescope)
Space Command (?)
Spotters (?)
Laser traffic control (?)
Should all these categories be considered as downtime?

10. 1. Definitions for Observing Efficiency (what are we talking about?)
Currently: Science instrument open shutter time during dark time, including science data and calibrations (sky, telluric, photometric, PSF, astrometry, wavelength) / dark time
Does not take into account any metric science-data quality -> very difficult to understand how “observing efficient” an instrument is.
A future definition for NGAO?
Science instrument open shutter(s) time during dark time delivering science-quality data
Each data set is flagged with a science-quality idx
Good understanding of the “observing efficiency” for each type of science

11. 2. Definitions for Observing Uptime (what are we talking about?)
Currently: AO system and science instrument up and ready to go
Does not take into account any instrument performance metric (good calibrations, image quality, operation efficiency, etc)
A future definition for NGAO?
Science instrument uptime is when it is ready to collect data in support of the science program.
Uptime is the opposite of downtime!
Downtime includes any time (under adequate weather conditions) not spent on collecting data in support of the science operations.

12. Observing efficiency work plan
Lessons learned
Collect experience from other LGS AO systems (Palomar, Gemini, Lick, ESO) and a complex non-AO MOS instrument
Summarize, analyze and understand main factors
Provide spreadsheet to science and technical team to help build the efficiency budget
Look into big terms per science per sub-system
Circulate a first phase of requirements
Anyone welcome to work on this
Need observing experience with other AO/instrument
Need experience with high-level software
Need new ideas to break limitations of current observing paradigms
All need to work fast and efficiently (100 hours total!!)

13. Observing efficiency budget
Built for each science use case
Include all observing steps: target acquisition, ID & centering; dithering; science readout and reductions; dithering; command parsing and decision making process; calibrations; etc
Should assume a 100% core hardware/software reliability? Why separate Uptime and Obs. Efficiency?
Should look into other lost-time statistics (weather, technical, laser traffic)
Should look into benefits of:
Observing planning GUI and simulation tools
Calibration units and auxiliary systems/data during observing (seeing, photometry, air-glow monitoring?)
Other possible impact on science-quality data (cirrus, centering stability)
System monitoring and recovery to optimize system uptime?
etc

14. Roadmap to an uptime budget
Review Keck AO statistics
Currently have LGS AO stats
Looking for NGS AO stats
If not currently summarized, derive from Metrics database
Attempt to get uptime statistics from other institutions
Remove outliers
Failures with large downtimes that can be avoided in NGAO, e.g.:
K2 Laser timing board failure with no spares
K2 WFC crashes
K2 Laser failures may not apply to NGAO laser(s)
Use remaining statistics as a starting point for a budget
Iterate and fine tune

15. Conclusions
Not all current Keck failures/downtime will map directly to NGAO
Primary downtime drivers will be laser and AO system reliability
Need to break down AO and Laser faults into finer granularity to draw more and better conclusions