Plans for lumi delivery in 2009-2010 LHC run

1. Plans for lumi delivery in 2009-2010 LHC run My apologies for dropping lumi units cm-2 s-1 What limitations can we expect ? How do we go from 1e28 to 1e32 ? How could the run look like ? After discussions with Ralph Assmann, Roger Bailey, Helmut Burkhardt, Werner Herr, Mike Lamont, Jorg Wenninger , and with the experiments.

2. 1st LHC Performance Workshop in Chamonix (feb 09) Message from the LHC experiments: • Typically, with 50-100 pb-1good data at 10-8 TeV many new limits set on hypothetical particles (some more stringent than Tevatron), or even discoveries possible! • With 200-300 pb-1 g.d. at 10-8 TeV  start competing with Tevatron for Higgs masses around 160 GeV • With 1 fb-1 g.d. at 10 TeV  find Higgs if around 160 GeV mass • The higher the energy, the faster it goes... • Note: below ~20-40 pb-1 g.d. at 10-8 TeV, or at any lower energy, one would probably start talking about an "engineering run" (can still be very useful, but perhaps not in terms of immediate physics results)

3. Outcome of First LHC Performance Workshop Held on 2-6 February 2009 in Chamonix “Many issues were tackled in Chamonix this week, and important recommendations made. Under a proposal submitted to CERN management, we will have physics data in late 2009, and there is a strong recommendation to run the LHC through the winter and on to autumn 2010 until we have substantial quantities of data for the experiments. With this change to the schedule, our goal for the LHC’s first running period is an integrated luminosity of more than 200 pb-1 operating at 5 TeV per beam, sufficient for the first new physics measurements to be made. This, I believe, is the best possible scenario for the LHC and for particle physics.” Rolf Heuer, CERN DG, 6 feb. 2009 Target: 300 pb-1 of good data at 10 TeV center of mass Ion run at the end

4. What do we need ? ∫Trun L(t) dt –––––––––– = 0.1 ... 0.2 L0 Trun Assume an overall “efficiency factor” Then, in order to reach few 100 pb-1 in a run of less than 10 months, we need a luminosity L0 of order ~ 1e32 The main parameters are (assume  is fixed to 5330) kb = number of bunches N = bunch population * = focusing at IP f kb N2 L0 = –––––––– F 4 N*

5. How do we get to 300 pb-1: input from LHC expts... ... on lumi strategy • ATLAS and CMS: highest possible lumi, even if average number of inelastic interactions of  = 7 (which we very probably never exceed in 2010) • ALICE: a couple of different running modes with typically very low lumi (1e28 to 1e30) and low pile-up per crossing ( ~ 0.1 or smaller), for example • kb=2-4, N~4e10, *=10m (largest possible bunch spacing) • kb=72, N~1e10, *=2m (bunch spacing > 100ns, "trains" spaced as much as possible) • LHCb: “low lumi” ~ few e32 => as much as ATLAS and CMS for this run, but keep pile-up at reasonable level (nominally  ~ 1, but may try working with higher pile-up) • TOTEM and LHCf: special short runs (typically of order few days)

6. Others, en-vrac,possibly incomplete ... some work ahead for the LPC • All experiments: • explore 25 ns spill-over effects, prepare the road • luminosity measurements (vdMeer, beam-gas, ...)cross-section normalisations • ALICE • collision spacing > 100ns whenever possible Pixel readout = 4 clock cycles • reversed dipole polaritysystematics • LHCb: • largest possible beam (at same luminosity) tranvserse vtx separation • polarity reversal systematics • fixed value of magnet (no ramp)save our spectrometer • inclined crossing (only possible at large *) tame the "bad" polarity • leveled luminosity (* adjustment during run) optimize integrated lumi • TOTEM/ALFA: • intermediate * , large * fwd physics program • LHCf • crossing angleincrease acceptance

7. How do we get to 300 pb-1 : input from LHC Main limitations may come from (as I understood it): • Collimation cleaning efficiency vs quench level => mostly limits total intensity I = kb N • see Kassandralph's talk at LMC #6 • Aperture limitations at the innner triplets • will limit * : *min ~ 1/ • gets worse with external crossing angle • Beam-beam effects => limit on N • head-on are always there (independent of * and ) • long-range appear with small bunch spacing (absent for > ~500ns spacing) • => see Werner Herr's talk at Chamonix: start with one bunch train per quadrant (full long range effects), then add trains. Collimator Year 1 settings limit Intermediate (start of physics run): I < 2e13 p Tight (towards end of run ?) I < 5e13 p Bunch spacing "easy" "challenging" and ext. angle * * > 500 ns , ext = 0 2m 1m > 500 ns , ext 0 3m 2m Nmax = ??? 5e10 "easy" 9e10 "challenging" ???

8. Some special complications... • Special complication for IP8: one polarity is worse than the other • see LPN 419. I assume, roughly: • ext = 0 and bunch spacing > 500ns: • like for IP1,5 (local bump is transparent to rest of machine) • ext  0 and bunch spacing = 50ns: • "good" polarity: can reach essentially same * as in IP1,5 • "bad" polarity: would have to step back by +1m • ALICE special bunches: • how to deal with BPMs ? • dynamic range for high intensity bunches may not allow to measure the small intensity bunches ? • switch to low scale a few BPMs ? • what spacing is possible for the special Alice bunches ? (150 ns ?) • how many can be fit in ?

9. Optimum number of bunches for lumi Simple-mindedly: • given a max intensity Imax and a max bunch charge Nmax, the number of bunches for maximizing luminosity is opt Imax kb = –––– Nmax For illustration: At the beginning, we are likely to have optimum at ~200-400 bunches (just beyond 156…)

10. Fill schemes: no crossing angle • Start with 2x2 • each IP gets 1 colliding pair and 1 non-colliding bunch/beam Then move on with equidistant schemes: see LPN 323+415 • start from a modified equidistant NxN scheme that gives N / 0 / N / ½N collisions in IP1/2/5/8 and shift symmetrically from IP1 a number of bunches to make them collide in IP2 (no change for IP1 and 5, but less collisions for IP8) • 43x43 • 43 / 2 / 43 / 20 colliding pairs in IP1 / 2 / 5 / 8 • 156x156 • 156 / 4 / 156 / 72 delayed by 75ns  that's about the maximum number of "equidistant" bunches (minimum spacing driven by length of common pipe section ~120m)

11. Fill schemes: with crossing angle Move on to crossing angle schemes • Many possibilities: 75 ns, 50 ns, 25 ns spacing (100 ns ? 150 ns ? ...) • NB1: the short spacing introduces long-range encounters in the IR => reduced reach in bunch charge ? • NB2: the crossing angle reduces the aperture in triplet => reduced reach in * ? => We most probably lose lumi with these schemes for the same intensity as in156x156!! Will require higher intensity to recover same lumi as in 156x156

12. Fill schemes: with crossing angle Why 50 ns instead of 25 ns or 75 ns ? • Allows to distribute lumi over 4 IPs in very flexible and optimal manner • Few colliding pairs in IP2 with “full” sequence at other three IPs • Hence, avoid large defocusing or lateral displacement at IP2 (ALICE) • More favorable in terms of beam-beam effects: • Only 3 IPs give the full head-on collisions (instead of 4 for 25 and 75 ns) • Required crossing angle almost independent of * • When reaching close to 50% nominal intensity and 100% nominal bunch charge (in 2011?), it should gives the highest possible luminosity (while waiting for phase 2 collimation system) For details, see LHC Project Note 415 and W. Herr in Chamonix 2009

13. Overview table (preliminary) 24 1.5 2

14. Physics run modes * expt magnet ON means at full nominal field ( as for 14 TeV) 900 GeV toroids, solenoids & spectr. dipoles ON* 10 TeV 2x2 5e10 10-11m toroids & solenoids ON*, spectr. dipoles OFF 2x2 5e10 10-11m 2x2 5e10 2m “few shifts” 43x43 5e10 2m “more few shifts” “few days” “more few days” Dominated by beam commisioning 156x156 5e10 2m Lumi goes over ~1e32 “few weeks” develop 50 ns, truncated introduces crossing angle short physics runs at 50 ns and go back to best luminosity (156x156) for mass lumi production until 50ns breaks even (then stay at 50 ns) try also 25 ns at the end 156x156 9e10 1m Dominated by physics running “few months” 50ns/432 9e10 3m

15. Proposal • Write in a document the target parameters for the 2009-2010 run (as discussed in Chamonix and right after) • Why ? • needed by many people for: radiation estimations, bkg studies, computing farm dimensioning, physics/trigger MC simulations • Replace (new version of) EDMS document 931921v2 • https://edms.cern.ch/document/931921/2 • Include summary table for physics conditions • But also add paragraphs to explain the strategy: • rough overview of requirements from experiments • what are the expected limitations ? • how do we plan to overcome them ? • how much beam commissioning time expected for the various steps ? • define "extreme" (but likely) case for background simulations

16. “Very draft schedule 2009-2010” • From first LHC Performance Workshop, Chamonix, feb. 2009 • Baseline • 1 month commissioning • 10 month proton physics • 1 month Lead Ions • Shutdown – end September 2010 • Built in slip potential see Mike Lamont in LMC_12c

17. Winter stop 2009-2010 • Unusual for CERN to run through winter • Assessed cost at LHC Performance Workshop 2009 in Chamonix, and decided to run through • Technical consequences and modalities being assessed • Main worry from accelerator sector is the manpower coverage (including burn-out risks) • Expts: no need for shutdown, more than happy with a 2-3 days technical stop per month • Proposal made at LMC (see Mike at LMC_12c): • Stop LHC with beam ~19th December 2009 • Earliest restart ~ 4th January 2010 • Could possibly use weekends either end =>12 days stop => Some expts questioned whether not better to have no stop at all • Difference in manpower investment and general impact for a no-stop scenario and a 2-week stop (standby) scenario is being assessed (see LMC 14)

18. Regular (Planned) Technical Stops (from Mike's talk) Survey across machine and expts needs was done => Programmed technical stop • 3 days including recovery and re-closure of ring • QPS plus power converters, controls, R2E etc. • Mon – Wed allowing weekday time for re-setup with beam • Followed by one day set-up with beam and systematic checks of machine protection system • Clearly if major breakdowns occur at other times – advantage will be taken. • Injector maintenance in parallel is an option • [Have not considered scheduling of MD…]

19. Very draft LHC schedule for 2009-2010 First draft produced (Mike), for discussion

20. TI8 test (6-7 June) It's the red beam, beam2, anticlockwise beam, the wrong beam, etc. LHCb's main purpose: • time and space align of silicon detectors (VELO, TT and IT) and • time align + tuning of Beam Conditions Monitor (BCM) + Beam Loss Scintillator (BSC) • good alignment is essential for tracking and triggering • remember: LHCb can't use cosmics for this... • requires special timing settings and special data analysis (not for free!), but proved to be very useful (2008 experience) => will give a good start in 2009 ! • considered important enough by the collaboration to put a special request requiring almost full operation of the experiment BSC TT BCM dwnstr IT BCM upstr

21. TI8 test (6-7 June) ...continued... LHCb wish list: • LHCb would like to take data on about 500 bunches of (optimally) 2e9 p/bch. This could be any time from saturday morning to monday morning, including nights. LHCb control room will be fully manned for data-taking from saturday morning to monday morning. • Need: • injection pre-pulse => ?? (Andy Butterworth ? JJ Gras ?) • LHC clock for beam2 => OK (Sophie Baron, RF) • LHCb will acquire data parasitically during machine TI8 studies and therefore would like to know what will be the maximum intensity (per bunch) that will be used: now assuming 1e10 p/bch. => yes, confirmed by Brennan • LHCb would appreciate if beam shots are given to them on the TED during any unforeseen break of beam studies of > ~1/2 hour (and where beam can still be delivered on the TED). • LHCb would appreciate if the bunch (or injected train) intensity could be published on DIP. • If multibunch operation is attempted, LHCb would be interested to take data as well, parasitically. Lowest possible intensity per bunch would again allow best usage for LHCb (the higher the intensity the least number of devices can be turned on: 2e9 p/bch is optimal, 1e10/bch is still useful).

22. TI2 test (11-12 July) ALICE: • no particular request. • will take data, but fully parasitically (means probably some minor requests on the fly, to be fulfilled if that does not disturb too much the test). LHCb: • depending on outcome of June 6-7 test (in TI8), would perhaps be happy to make use again of some beam in TI8 during the TI2 test, if not too disruptive for the test and for the work in adjacent sectors (s81, s78)