200 likes | 213 Vues
Learn about LCLS X-Ray FEL at SLAC, its components, operational goals, tuning strategies, and downtime statistics for researchers.
E N D
Availability Performance of LCLS X-Ray FEL at SLAC William Colocho for the LCLS team.
Outline • Quick LCLS Overview • Availability and downtime vs. time • Downtime per system and planned upgrades • Tuning and Availability • Operational Availability Goals • Questions ARW 2011/LCLS Availability Cape Town, April 2011
LCLS FEL at SLAC • Operated by Stanford University for the US Department of Energy • Study the molecular world in ways never before possible • 3 of 6 LCLS scientific instruments fully commissioned • By February 2011 1,162 unique scientists from 28 countries have submitted 427 proposals to use LCLS • Typically two experiments are scheduled for a five day user run. ARW 2011/LCLS Availability Cape Town, April 2011
A Tool for the Global Science Community “By 2013 six instruments will give researchers unprecedented tools for a broad range of research in material science, medicine, chemistry, energy science, physics, biology and environmental science” (LCLS fact sheet: http://www-group.slac.stanford.edu/com/docs/lcls_fact_sheets.pdf ) ARW 2011/LCLS Availability Cape Town, April 2011
LCLS Free Electron LASER • Each experiment selects a set of parameters to optimize their science goal. • Each configuration has different hardware and software requirements. For full list see: https://slacportal.slac.stanford.edu/sites/lclscore_public/Accelerator_Physics_Published_Documents/LCLS-parameters.pdf ARW 2011/LCLS Availability Cape Town, April 2011
LCLS Major Components • 197 Quadrupoles • 158 X, 159 Y Steering Correctors • 175 Beam Position Monitors • Klystrons/Modulators • 4 Injector Stations • 31 “Linac 2” Stations • 48 “Linac 3” Stations • 33 Undulators • 33x5 CAM Movers • ~1.8 Km of Vacuum • ~124 Vacuum Pumps • ~ 45 CAMAC crates • 2 Injector LASER systems • 33 Beam Shut Off Ion Chambers ARW 2011/LCLS Availability Cape Town, April 2011
LCLS Availability: 94.8% Percent of Scheduled Hours (%) COMM: Hutch Commissioning Programs. Plots and tables with the latest LCLS Availability information are automatically updated every shift during user runs. Run I: 92.8% Photon, 97.0% Hardware Run II: 92.5% Photon, 97.2% Hardware Runs I and II Photon Availability combined “Tuning” and “Configuration Change” as one category. ARW 2011/LCLS Availability Cape Town, April 2011
LCLS Availability Terms • Photon Availability: (Delivered + User Off + Configuration Change) / Scheduled On • Hardware Availability: (Delivered + User Off + Configuration Change + Tuning)/ Scheduled On • Scheduled On: Delivered + User Off + Configuration Change + Tuning + Down ARW 2011/LCLS Availability Cape Town, April 2011
Photon or Operational Availability Short term performance as important as long term. Users scheduled for machine time for five day intervals. ARW 2011/LCLS Availability Cape Town, April 2011
Trends Tuning, Downtime and Configuration Change time improved during the user run. Tuning and Configuration Change time reduction likely due to increased operational efficiencies. Downtime reduction not fully explained. Further analysis (Weibull analysis) needed. Notice correlation between tuning time and down time. (Down time leads to tuning time). ARW 2011/LCLS Availability Cape Town, April 2011
Downtime Statistics and future upgrades ARW 2011/LCLS Availability Cape Town, April 2011
MTBM and MDT Mean Time Between Maintenance Includes Corrective Maintenance and it does not include Preventive Maintenance. We accept that there will be preventive maintenance scheduled. This MTBM number should better reflect “Maintainability” when it most matters: During beam delivery to users. Only Selected Systems shown. See conference paper for full list. Data is 6 month snapshot (69.7 operating days). ARW 2011/LCLS Availability Cape Town, April 2011
Tuning Reduction Task Force • Members from Physics, Operations and Software groups. • Working on reducing tuning time during Machine Development and Repair Days recovery. • Expect this effort to also reduce tuning time during User runs. • Tuning reduced operational availability by ~2% ARW 2011/LCLS Availability Cape Town, April 2011
Tuning Task Force • Identify how tuning time is spent; Mine data from electronic logbook. Talk to Operators. • Measure execution time of software procedures. • Understand Machine configuration reproducibility. • Ongoing development of software to simplify and automate tuning procedures. ARW 2011/LCLS Availability Cape Town, April 2011
Tuning: Need Metrics Pie! • Injector Emittance, LI21 Emittance X, LI28 Emittance X, LTU Emittance X, Injector Matching, LI28 Matching, LTU Matching, Injector Phase Scans, Charge Change, Energy Change, Mirror reconfiguration, DL2 Dispersion with Quads, DL2 Dispersion with CORs, Undulator Re-pointing, Cathode Alignment, BSY Steering, LTU Steering, Laser Heater Alignment, Laser Heater Timing, C-Iris steering, MPS Trips at some settings, MPS Trips with Slotted Foil, Taper reconfiguration, Gas Detector Calibration, E-loss scan, Klystron Phasing, RF Amplitude Calibration, Zeroing e+ correctors, BBA, LI21 Emittance Y, LI28 Emittance Y, LTU Emittance Y, XCAV Set-points for FEL, L3 Steering… ARW 2011/LCLS Availability Cape Town, April 2011
LCLS Availability Goal is 95% • That is 95% Operational Availability to Users. • Before run III, goals did not take Tuning into account. • A tuning budget (~2%) will be included in the availability plan. This sets the hardware availability goal at 97%. ARW 2011/LCLS Availability Cape Town, April 2011
Realizable Goals • Goals based on past performance • Remove worst two weeks or events from historical data (Life of the project ~2.5 years). • System owners in charge and accountable for given system’s availability performance. ARW 2011/LCLS Availability Cape Town, April 2011
Conclusion • LCLS science program is ongoing. • Two year hardware availability ~ 97% • Photon availability of 95% realizable. • Multiple upgrade projects underway. • Further analysis needed to better understand long term time dependence of individual subsystem and component failure rate. ARW 2011/LCLS Availability Cape Town, April 2011
Thank You! • Questions. If we use failure rates as a metric for determining which systems (and therefore groups) will receive attention and/or resources; are we setting up a system where failure is rewarded? In a climate of decreasing resources and budget constrains: Is there a model that correlates Availability and Reliability with Budget Allowance? ARW 2011/LCLS Availability Cape Town, April 2011
ARW 2011/LCLS Availability Cape Town, April 2011