Do you believe in Operation Statistics Do you believe in Operation Statistics? Presented at the Accelerator Reliability Workshop 2011 Paul Scherrer Institut Andreas Lüdeke
Paul Scherrer Institute (PSI), Switzerland PSI: Swiss national research center 1300 employees Proton facility: 1 MW cw beam Spallation neutron source Proscan SC cyclotron Cancer therapy Swiss Light Source (SLS) e- storage ring 400mA, 2.4 GeV PSI West: Large Research Facilities Andreas Lüdeke, Paul Scherrer Institute at the ARW11 in Cape town, 11-Apr-2011 Slide 1
Swiss Light Source Operation Swiss Light Source is In operation since 2001 A user facility with 19 beam-lines in operation Providing ~ 5000 hours beam to users per year Users need to apply for beam time Just 30% of the proposals are accepted Accepted users need to wait about 6 month for beam time One important metric to judge machine quality Andreas Lüdeke, Paul Scherrer Institute at the ARW11 in Cape town, 11-Apr-2011 Slide 2
3rd Generation Synchrotron Light Sources: Availability Year Andreas Lüdeke, Paul Scherrer Institute at the ARW11 in Cape town, 11-Apr-2011 Slide 3 3
Availability: "Concours des Alpes" Year Andreas Lüdeke, Paul Scherrer Institute at the ARW11 in Cape town, 11-Apr-2011 Slide 4 4
Availability: "Concours des Alpes" Year Andreas Lüdeke, Paul Scherrer Institute at the ARW11 in Cape town, 11-Apr-2011 Slide 5 5
Metrics of the Swiss Light Source (SLS) Availability definition: “Delivered_Beam_Time” = “Beam_for_Users” during “User_Time” “User_Time" = “Scheduled_User_Time" + "User_Reserve_Time” Example for 2005: Uncompensated Availability Andreas Lüdeke, Paul Scherrer Institute at the ARW11 in Cape town, 11-Apr-2011 Slide 6
Elettra Metrics for Availability Thunderstorms are excluded from failure statistics at Elettra Example from 2005: Availability including outages from thunderstorms: Andreas Lüdeke, Paul Scherrer Institute at the ARW11 in Cape town, 11-Apr-2011 Slide 7 7
Availability: "Concours des Alpes” corrected Bare Availability [%] Year Andreas Lüdeke, Paul Scherrer Institute at the ARW11 in Cape town, 11-Apr-2011 Slide 8 8
Special Events: ESRF 2005 and the Crotch Absorber In March 2005 A water-to-vacuum leak Five days of user operation lost 120 hours downtime ESRF statistics 2005 Availability 97.6%, Total User Operation 5496 hours 131 hours downtime How can that be? All users of 5 days re-scheduled Not counted as downtime Andreas Lüdeke, Paul Scherrer Institute at the ARW11 in Cape town, 11-Apr-2011 Slide 9 9
What can we learn from this? Availability statistics not proportional to system reliability But large effort spend to compile downtime statistics Accurate failure data exists, but kept internal. Why? If that would be public, what could it be used for? Andreas Lüdeke, Paul Scherrer Institute at the ARW11 in Cape town, 11-Apr-2011 Slide 10 10
RF arcing at the Swiss Light Source Background SLS has 60 to 100 beam outages per year About 20 outages were caused by RF arcs Inquiry Asking around at the EPAC’08: “What is your experience with arc detectors?” Answer: arc detectors are often unreliable! Action We are replacing detectors with arc-coincidence detectors Result No arc interlock from any upgraded detector! Single diodes still show arcs, but do not cause interlocks Conclusion We had in 10 years 200 unnecessary beam losses! Andreas Lüdeke, Paul Scherrer Institute at the ARW11 in Cape town, 11-Apr-2011 Slide 11
What can we do? What we do now What we should do How to do that? We do your own mistakes Sometimes we share selected insights during workshops What we should do Learn from the failures of others Create a database of all outages of many facilities! How to do that? Define a simple document format Provide a web-platform to publish downtime event data I’ll start, will you join? Participate in the Friday morning discussion! . Andreas Lüdeke, Paul Scherrer Institute at the ARW11 in Cape town, 11-Apr-2011 Slide 12