1. Tevatron run issues with higher luminosity
4th International Workshop on Heavy Quarkonia
Vaia Papadimitriou, Fermilab
BNL, June
BNL, June 2006
Vaia Papadimitriou

2. OUTLINE Tevatron performance and projections
CDF data sets and plans for higher luminosity
D0 data sets and plans for higher luminosity
Conclusion
BNL, June 2006
Vaia Papadimitriou

3. The Fermilab Accelerator Complex
√s  1.96 TeV
CDF D0 P P
MAIN INJECTOR: 150 GeV
RECYCLER / e-COOLING
BNL, June 2006
Vaia Papadimitriou

4. Tevatron Performance Tevatron (Run I 1992-96, ∫L dt = 110 pb-1 ):
p   pbar at s = 1.8 TeV, 3.5 ms between collisions
Tevatron (Run II 2002-Present, ∫L dt = ~1.53 fb-1 ):
p   pbar at s = 1.96 TeV, 396 ns between collisions
( original plan for 132 ns )
FY06
Best 1.72 x 1032 cm-1s-1
~ 1.53 fb-1 delivered per experiment in Run II
FY05
FY06
7.17 pb-1 delivered per experiment in one store, Feb. 12, 2006
FY04
FY05
FY03
FY03
FY04
FY02
FY02
BNL, June 2006
Vaia Papadimitriou

5. Collider Luminosity History (per detector)
Eng. Run I
.05 pb-1
Eng. Run II
9.2 pb-1
Run Ia ( )
32.2 pb-1
Run Ib ( )
154.7 pb-1
Run IIa ( )
1200 pb-1
Run IIb ( )
3,060 – 6,880 pb-1
Run IIa + IIb ( )
4,260 – 8,080 pb-1
Log Scale !
Projected
Projected
BNL, June 2006
Vaia Papadimitriou

6. Luminosity The major luminosity limitations are
The number of antiprotons (BNpbar)
The proton beam brightness (Np/ep)
Beam-Beam effects
The transverse antiproton emittance
Transverse beam optics at the interaction point (b*)
F<1
~30 cm
BNL, June 2006
Vaia Papadimitriou

7. Tevatron Performance
BNL, June 2006
Vaia Papadimitriou

8. Stacking Performance Stack size (1010) Zero stack stacking rate FY06
BNL, June 2006
Vaia Papadimitriou

9. Expected Integrated Luminosity
8.1 fb-1
Fermilab Tevatron
6.7 fb-1
DESIGN
30 mA/hr
5.3 fb-1
4.3 fb-1
BASE
15 mA/hr
BNL, June 2006
Vaia Papadimitriou

10. Accumulated Luminosity and Luminosity per fiscal year
30 mA/h
(fb-1)
25 mA/h
20 mA/h
15 mA/h
FY03
0.33
FY04
0.67
FY05
1.27
FY06
2.07
1.94
1.91
1.87
FY07
3.84
3.24
2.93
2.63
FY08
5.92
4.88
4.03
3.40
FY09
8.08
6.70
5.28
4.26
Fiscal Year
30 mA/h
(fb-1)
25 mA/h
20 mA/h
15 mA/h
FY03
0.33
FY04
0.34
FY05
0.61
0.60
FY06
0.80
0.67
0.64
FY07
1.77
1.30
1.01
0.76
FY08
2.08
1.64
1.10
0.77
FY09
2.17
1.82
1.25
0.87
BNL, June 2006
Vaia Papadimitriou

11. Expected Peak Luminosity
30 mA/hr
15 mA/hr
BNL, June 2006
Vaia Papadimitriou

12. Data sets
1.62 fb-1
CDF/D0 have about 10 million J/y’s each in 1 fb-1 of Run II data.
1.30 fb-1
1.44 fb-1
1.20 fb-1
BNL, June 2006
Vaia Papadimitriou

13. Trigger rates Trigger Level Maximum rate CDF Maximum rate D0 Level 1
30 kHz
1.6 (1.8) kHz
Level 2
0.7 (1.0) kHz
<10% dead time
0.95 (1.05) kHz
<5% (10-15%) dt
Level 3
150 Hz
300 – 400 Hz
BNL, June 2006
Vaia Papadimitriou

14. A Study of Store Lifetime
Collected data for all Tevatron stores of lasting longer than 24 hours
Used 116 stores
Fit first 24h of each store with:
Fit is typically good to better than ~ 5%
Model is easy to integrate/solve
Only two parameters (L0, t )
Phenomenological study of t vs. L0 to extrapolate to higher luminosities
Use results to predict integrated luminosities for low lum tables that “kick-in” only after instantaneous luminosity drops below threshold. t L ) ( + = 1
BNL, June 2006
Vaia Papadimitriou

15. Typical projected store evolution
Inst. Luminosity (E32)
Inst. Luminosity (E32)
66%
34%
64%
36%
Peak Lum = 3E32
Peak Lum = 2E32
hours
hours
< 1.5 E32
1.5 – 2.0 E32
2.0 – 2.5 E32
2.5 – 3.0 E32
BNL, June 2006
Vaia Papadimitriou

16. The D0 Detector Excellent muon and tracking coverage
Tracking up to |h|<3
Muons up to |h|<2
BNL, June 2006
Vaia Papadimitriou

17. J/y triggers
Central (|h|<1.6) muon pT requirements are 1.5 GeV/c and 3.0 GeV/c
Forward muons do not have tracking coverage and one cannot apply pT cuts at Level 1. (~1 GeV/c muons can penetrate the iron)
At higher trigger levels one requires either two forward muons with pT >2 GeV/c or one forward and one central muon with pTs greater than 1 and 3 GeV/c respectively.
Dielectron triggers as well in Run IIA but with roughly a factor of 500 smaller yield. Expect to collect more dielectron J/y’s in Run IIB ( ~ 5-10 times smaller yield than dimuon J/y’s)
No dynamic prescaling (DPS) used; change prescales every few hours
BNL, June 2006
Vaia Papadimitriou

18. The CDF Detector Excellent mass resolution Particle ID: dE/dx, TOF
Tracking triggers (Hadronic B’s):
L1: Tracks
L2: Secondary vertex
Central Muon Detectors: |h|<1.0
1.3<|h|<3.5
ToF counter for K/p separation
placed right before the solenoid
Central Outer Tracker: |h|<1.0
dE/dx for PID
3.5<|h|<5.1
Silicon: |z0|<45 cm, |h|<2.0
BNL, June 2006
Vaia Papadimitriou

19. J/y triggers Level 1 Level 2 Level 3 Prescale
DPS
CMU1.5/CMU df<1200, oppQ J/yCMUCMU L2 10:1:1
CMU1.5/CMX df<1200, oppQ J/yCMUCMX L2 10:1:1
CMU1.5/CMX df<1200, oppQ J/yCMUCMX L2 PS=2
CMU1.5/CMU df<1200, oppQ J/yCMUCMU L2 PS=2
CMU1.5/CMU auto J/yCMUCMU L2 PS=100
CMU1.5/CMX auto J/yCMUCMX L2 PS=100
CMUP auto J/yCMUCMU L2 50:10:1
CMUP auto J/yCMUCMX L2 50:10:1
CMUP CMUP J/yCMUCMU L2 10:1:1
CMUP CMUP J/yCMUCMX L2 10:1:1
DPS
calibration
- CDF psi mode systematics same as Belle’s best result! However result is statistically limited
- CDF hadronic modes have only slightly larger systematics, but improved statistical power
- D0 B+/B0 ratio at HFAG level – Interesting technique
- D0 Bs lifetime update is best measurement so far (improved CDF Run I measurement in semileptonics)
DPS
Single lepton
DPS
High pT
DPS
DPS
polarization
Dielectron triggers as well
BNL, June 2006
Vaia Papadimitriou

20. Trigger cross section - rate extrapolation
As the luminosity increases, higher average number of primary interactions per bunch crossing yield more complex events with higher occupancies and higher trigger rates which cause higher dead time fractions and lower efficiencies.
Current XFT
Upgraded XFT
One example: High Pt CMX Muon
In principle, a physics process
trigger cross section, s, is
constant .
In reality, a given trigger cross
section behaves as:
s = A/L + B + CL + DL2
Use existing data to
extrapolate
Confirmation of XFT tracks by stereo layers
is expected to yield a substantial reduction of fakes
BNL, June 2006
Vaia Papadimitriou

21. Trigger/DAQ Upgrades for higher luminosity
Goals
Increase bandwidth at all levels
Improve purity at L1
Status - Complete
COT TDC -- readout latency
COT Track Trigger (XFT)-- purity
Silicon Vertex Trigger (SVT)-- latency
L2/L3 trigger -- latency
Event builder -- latency
Data logger -- throughput
Add stereo layer info
Track trigger installation done, being commissioned
Data logger installation in progress
Proc power: 1THz  2.6THz
BNL, June 2006
Vaia Papadimitriou

22. Impact of L2 decision crate & SVT upgrades
on L1 bandwidth
After
Upgrade
Lumi~90E30
Before
Upgrade
Lumi~20-50E30 5%
Before: 5% deadtime
with L1A 18KHz
@ ~< 50E30
After: 5% deadtime
with L1A 25KHz
@ ~ 90E30
Dead time %
18KHz KHz
L1A rate (Hz)
BNL, June 2006
Vaia Papadimitriou

23. Trigger rate extrapolation – Jet 100 GeV
Primary vertex multiplicity vs inst. luminosity
Predicted cross section vs inst. luminosity
3rd order poly
3rd order poly
Trigger cross section vs
primary vertex multipl.
2nd order poly
BNL, June 2006
Vaia Papadimitriou

24. Trigger rate extrapolation – B hadronic two track trigger
Primary vertex multiplicity vs inst. luminosity
Predicted cross section vs inst. luminosity
3rd order poly
3rd order poly
Trigger cross section vs
primary vertex multipl.
2nd order poly
BNL, June 2006
Vaia Papadimitriou

25. J/y triggers for higher luminosity
Level 1
Level 2
Level 3
Prescale
DPS
CMU1.5/CMU df<1200, oppQ J/yCMUCMU L2 10:1:1
CMU1.5/CMX df<1200, oppQ J/yCMUCMX L2 10:1:1
CMU1.5/CMX df<1200, oppQ J/yCMUCMX L2 PS=25
CMU1.5/CMU df<1200, oppQ J/yCMUCMU L2 PS=2 5
CMU1.5/CMU J/yCMUCMU no pres.
CMU1.5/CMX J/yCMUCMX no pres.
CMUP auto J/yCMUCMU L2 50:10:1
CMUP auto J/yCMUCMX L2 50:10:1
CMUP CMUP J/yCMUCMU L2 10:1:1
CMUP CMUP J/yCMUCMX L2 10:1:1
DPS
1.75, 2.5 GeV/c
2< mT < 4 GeV
- CDF psi mode systematics same as Belle’s best result! However result is statistically limited
- CDF hadronic modes have only slightly larger systematics, but improved statistical power
- D0 B+/B0 ratio at HFAG level – Interesting technique
- D0 Bs lifetime update is best measurement so far (improved CDF Run I measurement in semileptonics)
DPS
DPS
DPS
DPS
BNL, June 2006
Vaia Papadimitriou

26. Preparation for doing physics at highest luminosity
Dedicated studies to understand evolution of Tracking, Lepton Identification, B-Jet Tagging, Missing Energy Resolution, Jet Corrections, etc.
Strategy:
Use Monte Carlo: over-lay additional minimum-bias events to simulate luminosity up to 3 E32
Use data: in bins of # of interactions/event; makes use of the bunch-to-bunch luminosity variations to gain a level arm to higher luminosity
Data vs MC comparison
Online
Trigger/DAQ
Offline
computing
detector
Analysis/meetings
PRL
~100s ns ~ µs to ~ms ~weeks ~ months
BNL, June 2006
Vaia Papadimitriou

27. Tracking: High Occupancy Physics
Avg
now
Peak (3 E32)
vs number of z vertices
At highest luminosities:
COT efficiency more significantly impacted
SVX efficiency minimally affected
Top, Higgs,…
on Average: 10% (relative) loss in B-tag efficiency
BNL, June 2006
Vaia Papadimitriou

28. 1% Tracking: Low Occupancy Physics
B, W, …
No significant effect on this type of CDF physics program
BNL, June 2006
Vaia Papadimitriou

29. Conclusions The Tevatron is running very well (1.53 fb-1 delivered)
Many new results
The Tevatron is expected to provide 4.3 – 8.1 fb-1 by October 2009
Typical peak luminosities of the order of x 1032 now and x 1032 expected
CDF and D0 have of the order of 107 J/y’s each in 1fb-1 of data
They expect to retain similar yields up to 2 x and 80-95% of the yield per fb-1 at higher peak luminosities
A lot of answers and surprises awaiting!!
BNL, June 2006
Vaia Papadimitriou

30. Backup
Backup Slides
BNL, June 2006
Vaia Papadimitriou

31. Tevatron Performance FY06 FY05 FY02 FY06 FY05 FY04 FY03 FY02
Design
FY05
Base
FY02
FY06
FY05
FY04
FY03
FY02
BNL, June 2006
Vaia Papadimitriou

32. Expected Weekly Luminosity
BNL, June 2006
Vaia Papadimitriou

33. Data Analysis and physics results turn around time
Data Analysis processing power:
8.2 THz - distributed among 10 Central Analysis Farms (CAFs)
5.8THz on-site (30% from non-FNAL funds), 2.4 THz off-site (for Monte Carlo)
Improvement - use a single entry point for job submission to offsite CAFs
expands CPU resources available for CDF and increases efficiency of their use (world-wide CDF-Grid of CPU clusters)
Physics results turn around time:
recent 1 fb-1 data to 1st physics result ~ 10 weeks
Online
Trigger/DAQ
Offline
computing
detector
Analysis/meetings
PRL
~100s ns ~ µs to ~ms ~weeks ~ months
BNL, June 2006
Vaia Papadimitriou

34. Antiproton Parameters
BNL, June 2006
Vaia Papadimitriou

35. Future Pbar Work Lithium Lens (0 – 15%) DRF1 Voltage (5%)
Lens Gradient from 760T/m to 1000 T/m
Slip Stacking (7%)
Currently at 7.5x1012 on average
Design 8.0x1012 on average
AP2 Line (5-30%)
Lens Steering
AP2 Steer to apertures
AP2 Lattice
Debuncher Aperture (13%)
Currently at 30-32um
Design to 35um
DRF1 Voltage (5%)
Currently running on old tubes at 4.0 MEV
Need to be a t 5.3 MeV
Accumulator & D/A Aperture (20%)
Currently at 2.4 sec
Design to 2.0 sec
Stacktail Efficiency
Can improve core 4-8 GHz bandwidth by a factor of 2
Timeline Effects
SY120 takes up 7% of the timeline
BNL, June 2006
Vaia Papadimitriou

36. Trigger cross section/rate extrapolation is based on existing data
One example: High Pt CMX Muon
Current XFT
Upgraded XFT
Main reason for the growth of
trigger cross section is the
increasing # of interactions
per bunch crossing
By counting the number of
vertices found offline,
one could estimate
the effective luminosity
Variation of bunch to
bunch luminosity due to
anti-proton intensity…
Those information is used for rate extrapolation and cross checks
Confirmation of XFT tracks by stereo layers
is expected to yield a substantial reduction of fakes
BNL, June 2006
Vaia Papadimitriou

37. At highest luminosities:
Tracking (SVX & COT): High Occupancy Physics
At highest luminosities:
SVX efficiency minimally affected
COT efficiency more significantly impacted
#hits on tracks
SVX
COT
Number of interactions per event
BNL, June 2006
Vaia Papadimitriou