Beam losses through the cycle G. Papotti, A. Gorzawski, M. - PowerPoint PPT Presentation

LHC Beam Operation workshop - Evian 2012 Beam losses through the cycle G. Papotti, A. Gorzawski, M. Hostettler, R. Schmidt

outline motivation • – 2011 vs 2012, PM browser module beam losses per beam mode • – capture, ramp, flat top, squeeze, … • all 2012 fills until end of November – thanks to A. Gorzawski for a lot of data, analysis and plots some information on bunch-by-bunch differences • – appetizers on instabilities – bunch-by-bunch losses in stable beams and burn off disclaimer: • – give only overview of the year, no details on fill to fill differences – injection not treated 19 Dec 2012 giulia.papotti@cern.ch 2

at the SPS SPS page 1 and Larger • – e.g. LHC3, 72 bunches 19 Dec 2012 giulia.papotti@cern.ch 3

at the LHC (inject) • ramp • – 0.45 TeV to 0.5 TeV – 0.5 TeV to 4 TeV flat top • squeeze • adjust • stable beams • 19 Dec 2012 giulia.papotti@cern.ch 4

before stable beams zoom from injection to • end of adjust 19 Dec 2012 giulia.papotti@cern.ch 5

before stable beams zoom from injection to • end of adjust same for a fill in 2011 • 2011 ¡ 2012 ¡ 2011: used to have ~100% • transmission to stable beams 2012: <100% • … so look into the reasons • 19 Dec 2012 giulia.papotti@cern.ch 6

PM Beam Power Loss Module statistical analysis over many fills • 19 Dec 2012 giulia.papotti@cern.ch 7

PM Beam Power Loss Module 1 ¡ 2 ¡ statistical analysis over many fills • P = n 2 − n 1 peak power loss per mode • E 1 e 11 p @ 4 TeV t 2 − t 1 – based on sliding window (e.g. 5s, 20s, 80s) 19 Dec 2012 giulia.papotti@cern.ch 8

2011 vs 2012 statistics 2011 transmission: ~99.4% • – from end of injection to start of stable beams 2012 transmission: • – beam 1: ~96.2% – beam 2: ~95.3% 19 Dec 2012 giulia.papotti@cern.ch 9

Losses per beam mode

capture losses (.45->.5 TeV) MD3 ¡ energy ¡matching ¡ generally b1 worse than b2 • first energy matching improved the situation • – second not so much got generally worse ~after MD3 • – enhanced satellites? (Q20? batch-by-batch blow-up was often off) 19 Dec 2012 giulia.papotti@cern.ch 11

ramp (.5->4 TeV) TS3 ¡ non-negligible losses • transmission improved towards the end of the run • – when had the higher losses at capture? – Q20 and smaller emittances? 19 Dec 2012 giulia.papotti@cern.ch 12

ramp (.45 -> 4 TeV) max power loss during ramp • highest losses either at the start or at the end • – ramp function: 770s long beam 1, higher loss spikes • 19 Dec 2012 giulia.papotti@cern.ch 13

ramp (.45 -> 4 TeV) plots for longer window (now 80s, was 20s) • highest loss mostly at the end of the ramp function • very similar peak loss values for beam 1 and beam 2 • 19 Dec 2012 giulia.papotti@cern.ch 14

flat top few minutes long: manual tune trims and squeeze loading • negligible losses, generally b2 worse lifetime • – slightly worsened after octupole polarity change MO ¡polarity ¡change ¡ 19 Dec 2012 giulia.papotti@cern.ch 15

squeeze MO ¡polarity ¡change ¡ generally b2 worse than b1, • throughout the year both beams got worse after • MO polarity change • max power loss – very reproducible for beam 1, but at different times – at precise times for beam 2 930s (sq. function = 925s long) 820s ( β *~0.8-0.7m) 420s ( β *~3m) 19 Dec 2012 giulia.papotti@cern.ch 16

adjust split ¡coll ¡beam ¡process ¡ split collision beam process • made the difference – very reproducible losses and max power loss 19 Dec 2012 giulia.papotti@cern.ch 17

first 5 mins in stable beams b1 generally worse • – shift crews used to say: “b2 loses earlier in cycle, b1 loses at start of stable beams” split ¡coll ¡beam ¡process ¡ 19 Dec 2012 giulia.papotti@cern.ch 18

max power losses in 2011 generally factor 2-3 • smaller than 2012 beam 1 generally • higher losses no clear clustering • in time as in 2012 19 Dec 2012 giulia.papotti@cern.ch 19

Bunch-by-bunch observations

BBQ signal amplitude not ¡bunch-­‑by-­‑bunch! ¡ 1 ¡ ? ¡ 1 ¡ ? ¡ 1. ¡TS2 ¡ 2. ¡MO ¡polarity ¡change ¡ 3. ¡split ¡coll ¡beam ¡process ¡ ?. ¡Q’/MO ¡reducAon ¡incorporated ¡ in ¡collapse ¡funcAon ¡ 3 ¡ 2 ¡ 3 ¡ hard to correlate to max power loss • – preliminary analysis, more time required 19 Dec 2012 giulia.papotti@cern.ch 21

emittance from luminosity count blown-up bunches in stable 4 • beams Lumi Emittance [um] 3.5 – emittance from luminosity 3 • 1.65e11ppb, 2.5um -> 7.26e33Hz/cm 2 • 1.65e11ppb, 3um -> 6.25e33Hz/cm 2 2.5 2 MO ¡polarity ¡change ¡ 500 1000 1500 2000 2500 3000 3500 Bunch number 1000 800 Blown � up bunch count not easy to correlate to • 600 BBQ amplitudes – more analysis required 400 200 0 2500 2600 2700 2800 2900 3000 3100 3200 3300 3400 Fill number Q20 ¡& ¡split ¡coll ¡beam ¡process ¡ M. ¡Hoste6ler, ¡LBOC, ¡20.11.2012 ¡ 19 Dec 2012 giulia.papotti@cern.ch 22

b1 bunch length histogram splitting bunches with larger transverse • Bunch length histogram evolution, Fill 3287 1.5 emittance get lower bunch length Bunch Length [ns] 1.4 – build up during stable beams in beam 1 – no visible effect on beam 2 1.3 – for fills with selective transverse 1.2 emittance blow-up results in two distinct families • 1.1 0 1 2 3 4 5 6 7 Time in SB [h] Emittance vs. Length, 7h SB, Fill 3287 Emittance histogram evolution, Fill 3287 1.4 4.5 1.35 4 Emittance [um] Bunch length [ns] 3.5 1.3 3 1.25 2.5 1.2 2 1.15 0 1 2 3 4 5 6 7 Time in SB [h] 1.1 3 3.5 4 4.5 M. ¡Hoste6ler, ¡LBOC, ¡20.11.2012 ¡ Lumi Emittance [um] 19 Dec 2012 giulia.papotti@cern.ch 23

b1 loss structure in stable beams first ~30 bunches of each SPS batch • Beam 1 losses: Fill 3363, 10.0h SB 0.4 in beam 1 lose up to 10% less in stable beams (SB) 0.35 – very reproducible, also there in 2011 Rel. Loss [1] 0.3 – no correlation with number of long- 0.25 range interactions – not visible on beam 2 (or smaller?) 0.2 cause not clear yet • 0.15 500 1000 1500 2000 2500 3000 3500 Slot Number Bunch loss B1, Fill 3363 0.35 Beam 2 losses: Fill 3363, 10.0h SB 0.4 0.325 Intensity loss in SB [1] 0.3 0.35 0.275 Rel. Loss [1] 0.3 0.25 0.25 0.225 0.2 0.2 20 40 60 80 100 120 140 Offset in 144 bunch train 0.15 500 1000 1500 2000 2500 3000 3500 M. ¡Hoste6ler, ¡LBOC, ¡30.10.2012 ¡ Slot Number 19 Dec 2012 giulia.papotti@cern.ch 24

burn off collisions ¡in ¡ IP ¡8 ¡15 ¡158 ¡ 25 ¡ns ¡slot ¡ e.g. above plots for fill 3045 at 8 h of stable beams • residual loss has SPS batch structure • – after removing burn-off component from total losses ( σ proc = 101.8 mb) – particularly strong on b1 beam ¡2 ¡ beam ¡1 ¡ offset ¡in ¡SPS ¡batch ¡ offset ¡in ¡SPS ¡batch ¡ 19 Dec 2012 giulia.papotti@cern.ch 25

conclusions losses through cycle are non-negligible: transmission ~95% • – capture (degradation towards the end possibly related to enhanced satellites?) – ramp: ~1% towards the end – squeeze: peak losses at precise moments for beam 2 – adjust: split collision beam process improved the reproducibility – stable beams • observation of bunch length histogram splitting and correlation to transverse emittance blow up • 144-bunch loss pattern to be understood stronger on beam 1 – plenty more analysis possible • support fill-to-fill data analysis tool • – more useful than end-of-the-year overview! 19 Dec 2012 giulia.papotti@cern.ch 26

integrated lumi comparison 14 for full picture on • x 10 2 performance, need to Intensity at start of SB [p] fold in availability 1.8 intensity increase did not • 1.6 ‘clearly’ pay off in terms of integrated luminosity 1.4 1.2 2500 2600 2700 2800 2900 3000 3100 3200 3300 3400 7 x 10 7 � 1 ] Int. ATLAS Lumi, 8h SB [ub 6 0.35 Losses in 8h of SB [1] 0.3 5 0.25 0.2 4 0.15 0.1 3 2500 2600 2700 2800 2900 3000 3100 3200 3300 3400 2500 2600 2700 2800 2900 3000 3100 3200 3300 3400 19 Dec 2012 giulia.papotti@cern.ch 27