Supplemental Material: Data on Tevatron Complex Reliability

1
Supplemental Material Data on Tevatron Complex
Reliability

• Elliott McCrory
• 25 February 2004

2
Presentation Outline

• Main talk is elsewhere
• The data here are supplemental to that talk

3
Downtime Logger, D18
Console Application Screen Shot
4
Web Page 4 Weeks of Downtime
Click Here
5
Focus on one subsystem BPS

• Entries like these are uses as the basis for the
  analysis

6
Tevatron Complex Subsystems

• BDIAG
• BMAG
• BMISC
• BPS
• BRF
• BVAC
• BWATR
• CMISC
• CNET
• CRLK
• CTIME
• EXPAR
• FDIAG
• FMISC
• FPS
• FWATR

• LDIAG
• LMAG
• LMISC
• LPS
• LRF
• LVAC
• LWATR
• MIDIAG
• MIMISC
• MIPS
• MIRF
• MISC
• MIVAC
• MIWATR
• MIMAG
• MIXFER

• POWER
• RADTRIP
• RRMISC
• SAFETY
• TCRYO
• TDIAG
• TMAG
• TMISC
• TPS
• TQPM
• TQUEN
• TRF
• TVAC
• TWATR
• WATER

• NTF
• PACC (L)
• PBCOOL
• PBDIAG
• PBMAG
• PBMISC
• PBPS
• PBRF
• PBTRGT
• PBVAC
• PBWATR
• PBXFER

7
Definition of Terms

• MTBF
• Mean Time Between Failure
• Simple arithmetic average of the time between
  failures
• ?
• Used to characterize the probability of an
  interval in the classic random queuing problem
• E.g., the arrival interval of cars at a toll
  booth
• ? 1/lttgt 1/s
• Actually, not s, but R.M.S.
• Calculated here as (2/(MTBF s(MTBF)))
• Average Downtime Fraction
• S (time down) / (specified interval)

8
MTBF For Each Major System
Running average over 30 days
Linac
Good
Fall 2003 Shutdown
Running average MTBF, hours
Bad
Running average over 120 days
Days since 1/1/2003
9
MTBF For Each Major System
Booster
Good
Running average MTBF, minutes
Bad
Days since 1/1/2003
10
MTBF For Each Major System
PBar
Good
Running average MTBF, minutes
A lot less downtime
Bad
Days since 1/1/2003
11
MTBF For Each Major System
Main Injector
Good
Running average MTBF, minutes
Bad
Days since 1/1/2003
12
MTBF For Each Major System
Tevatron
Good
Running average MTBF, minutes
Bad
Days since 1/1/2003
13
MTBF For Each Major System
Controls
Good
Running average MTBF, minutes
Bad
Days since 1/1/2003
14
Tevatron as 2 Machines

• The machines are called Low Beta and Not Low
  Beta
• Is Low Beta if energy is bigger than 900
• Otherwise, it is called Not Low Beta
• Only one machine exists at a time
• Time between failures
• Only count time in which the machine exists
• That is, count minutes at 980 to determine time
  between failures at 980
• Ditto for Not Low Beta
• Eliminate obvious bad points
• E.g., 5000 minutes of shutdown between failures
  at 150 for TeV this winter.

15
Tevatron as 2 machines, MTBF
MTBF is getting longer!
16
?(t) For Each Major System
Linac
Bad
Running average MTBF, hours
Running average over 30 days
Good
Running average over 120 days
Days since 1/1/2003
17
?(t) For Each Major System
Booster
Bad
Running average MTBF, hours
Good
Days since 1/1/2003
18
?(t) For Each Major System
Main Injector
Bad
Running average MTBF, hours
Good
Days since 1/1/2003
19
?(t) For Each Major System
PBar
Bad
Running average MTBF, hours
Good
Days since 1/1/2003
20
?(t) For Each Major System
Bad
Running average MTBF, hours
Good
Tevatron
Days since 1/1/2003
21
?(t) For Each Major System
Controls
Bad
Running average MTBF, hours
Good
Days since 1/1/2003
22
Average Downtime Fraction
Running average over 30 days
Running average over 120 days
Bad
Linac
Running average Downtime Duration, minutes
Good
Days since 1/1/2003
23
Average Downtime Fraction
Booster
Bad
Running average Downtime Duration, minutes
Good
Days since 1/1/2003
24
Average Downtime Fraction
Main Injector
Bad
Running average Downtime Duration, minutes
Good
Days since 1/1/2003
25
Average Downtime Fraction
PBar
Bad
Running average Downtime Duration, minutes
Good
Days since 1/1/2003
26
Average Downtime Fraction
Tevatron
Bad
Running average Downtime Duration, minutes
Good
Days since 1/1/2003
27
Average Downtime Fraction
Controls
Bad
Running average Downtime Duration, minutes
Good
Days since 1/1/2003
28
Radioactive Decay
? e ? t is the frequency distribution of the
classic random queuing problem
F(t) 2 -t/t
F(t) 2 -t/t e -ln(2) t/t e ?
t ? 0.0346
lttgt s 1/ ? 28.9
29
Controls Downtimes
Data Dump of All Controls Downtimes since
1-1-2003
30
? For All Systems
Probability of failure per hour 1 ? E.g.
Linac up for 5 hours (1-0.183)5 0.364
Log Scale
31
Quality of ? Fit Assumption
Q lttgt/s
32
Store Duration
Failures
Intentional 2003
dN / dT
Intentional 2004
Stores Duration, Hours