Summary notes of the ninth meeting of the LHC Commissioning Working Group

 

Wednesday June 28th, 14:30

CCC conference room 874/1-011

Persons present

Minutes of the Previous Meeting and Matters Arising

Stephane had sent some comments on the discussions following Gianluigi’s and Rogelio’s presentations, respectively:  (1) For the dipole end laminations, the longitudinal magnetic center of the dipoles shall be known within 1-2 cm (not 1mm) to exclude effects from a systematic offset of this magnitude. Presently we can warrant only +/-10 cm from the measurements (length of the short mole) by construction. Stephane did not know if more than 1-2 cm can be excluded. (2)  Concerning Rogelio’s presentation, the difference observed between the H and V beta-beat only partly arises from an asymmetry between beta_x at the SF and beta_y at the SD, but primarily is due to the fact that the SD sextupoles are twice more strongly excited (being half as efficient to correct Q'_nat because D_x at QD is only 1.1 m compared to 2.1 m at QF). In the model shown by Rogelio, the dominant source of beta-beat came from this source via feed-down effect in the MS for an assumed 2mm r.m.s CO in both planes at commissioning. Once this will be corrected, we should return to the more reasonable budget of 20-25% in both planes. Stephane’s comments were added to the minutes of the 8th LHCCWG meeting.

 

News on LHC Aperture Restrictions (Roger Bailey)

Roger reported news on the LHC aperture restrictions, which were discussed at the LTC on June 7th and at MARIC on June 21st. Around 400 of 500 installed dipole magnets have been checked with the reflectometer. A total of 19 problem cases have been identified to date, either from the reflectometer data or by visual inspection. In five cases, the problem could be cured, in one case a cure was attempted but failed. Typical objects found are metallic swarf from cutting the beam screens and plastic swarf from beam-screen wrapping. Small plastic pieces are not easily detected by the reflectometer. Further effort of removing obstacles from inside the installed magnets may involve an endoscope. Reopening of a few sectors may also be required. This effort is followed up in MARIC.

 

Paul mentioned a proposal that has been passed on to him for comment by Steve to intentionally introduce similar plastic pieces in the SPS/TT4 to explore their effect. Massimo asked whether the locations of the magnets with presumed obstacles are known. Paul replied that a database exists, possibly in the form of MTF entries. Roger commented that Paul Cruikshank has the mandate of overseeing the aperture curing effort and he would be the person to contact for detailed information on suspicious magnets.

 

News on LHC Schedule (Roger Bailey)

Roger informed the working group that a new LHC schedule was announced after recommendations and decisions by the LHC MAC, the SPC and the CERN Council. The cryomagnet cold tests are on schedule to finish in 2006. The magnet installation rate is now equal to or above 25 per week. Accordingly, the last magnet should be in the ring by March 2007. The interconnect rate achieved should allow closure of the vacuum in August 2007. The official schedule continues until the end of 2007. Collisions at 450 GeV in 2007 are now a project goal.

 

LHC Schedule to End 2007 (Roger Bailey)

Roger presented the detailed hardware and machine commissioning schedule for 2007. Sectors 5-6 and 6-7 will have a reduced period of hardware commissioning up to 7 TeV. Minimum hardware commissioning of sectors 1-2 and 2-3 proceeds in parallel with the cold check out. The 23-June CERN press release states that the high-energy physics run in 2008 will continue until a pre-defined amount of data has been accumulated. Massimiliano clarified that this refers to a few inverse femtobarns. Roger presented an extended schedule, including operations activities in parallel with hardware commissioning of the last two sectors. This is followed by two weeks for the full machine checkout., and 4 weeks of beam commissioning. The 450-GeV commissioning beam time required is estimated as 16 days, and the 4-week number assumes a 60% efficiency. Following beam commissioning is a two-week 450 GeV engineering run. Maximum intensity considered is 156 bunches with 1e11 protons per bunch. The peak luminosity can reach or exceed 1e30 cm-2s-1 and the corresponding event rate is of order 100 kHz. The number of Z or W bosons produced in the channels of interest is 10 or 100 per 24 hours of colliding beams. In 2008 the previously considered commissioning sequence could then take place.

 

Paul commented that this schedule does not mention ion operation. Massimiliano responded that the issue of ion running in 2008/2009 is yet to be discussed (by CRAG). Jan remarked that the present schedule may not leave sufficient time for the 3-month reliability run of beam-dump commissioning, which had originally been foreseen. In particular, he expressed his understanding that the planning for the straight sections has not been finalized. Paul replied that the reliability run could proceed at 450 GeV. Installation and bake out of sector 6 should be completed in April 2007, which should leave enough time for the beam-dump commissioning. Thijs asked whether the white space on the schedule could not be filled with other activities. Paul explained the philosophy of the schedule, namely that magnet transport and interconnect work go on in different sectors, in order to avoid interference.  The white space does actually contain other activities such as magnet transport.

 

The question was raised whether the LHC could be ramped in the 2007 run. Massimo declared that this is not possible since not all spool pieces will be available. Frank remarked that the LHCMAC has strongly recommended studies of snap back and persistent-current behavior at the start of the ramp. He suggested that sectors with missing spool pieces could be compensated by exciting those in other sectors more strongly. Massimiliano described the goal of the 450-GeV run from the viewpoint of the experiments, which is to obtain collisions as soon as possible and to have stable running periods.

 

Roger asked how long a 450 GeV run would be useful for the experiments. Massimiliano responded that this depends on the speed of the detector debugging, and he stressed that the 450 GeV in 2007 should be driven by machine commissioning, not by the potential use the experiments could make use of so many nb^-1 or pb^-1. Paul wanted to know whether the parasitic collisions – which arise if collisions are also delivered to LHCB - would be a problem for the other experiments. These parasitic collisions would occur with a longitudinal offset of 11.25 m from the main interaction point. Massimiliano answered that this type of operation had always been planned, and that this does not pose any problems to the detectors. Quoting Werner Herr, Massimo mentioned that 4 out of 43 bunches, or 24 bunches out of 156, would be offset for collisions in LHCB. Massimiliano will check these figures with the experiments. Roger listed two more questions: Assuming the 2007 run takes place, what will happen in 2008? And, if the 2007 run does not happen, do we still want to collide at 450 GeV in 2008? There was no clear answer yet.

 

A Look at 2008 (Roberto Saban)

Roberto presented two draft schedules for 2008, developed in collaboration with Paul, Luigi Serio, and Roger. He first explained that 2008 must contain a period of continued hardware commissioning, where all the tasks are performed that will be left out in 2007.   The two main items considered are the powering tests, which set the critical path, and the cryogenics systems.  In the first schedule, always one of two cryoplants will keep two sectors at 80 K, while the second compressor will undergo maintenance. Cryoplants in points 8 and 4 will experience the full maintenance, those in point 6 a somewhat reduced one, and those in point 2 for only half the regular period. This already highly aggressive schedule entails some risks of increased downtime and degraded operational efficiency.  The hardware commissioning would finish toward the end of April, followed by one month of cold checkout and two months of beam commissioning, so that the physics run would start in mid-July. An even more aggressive schedule is the second version, where many powering tests proceed in parallel, and there no longer is any cryogenics maintenance in half of the sectors. This schedule increases the two risks mentioned earlier and it includes as additional risk the availability of resources. With this schedule, the physics run could start at the end of May. This second schedule is not realistic and may be erased in the future.

 

Paul commented that the operations part of either schedule, namely 3-4 weeks of check out and 2 months of beam commissioning, is also very optimistic.

 

Collisions at 450 GeV (all)

The possible differences between the 450 GeV run in 2007 and the re-commissioning at 450 GeV in 2008 are the values of beta* in IPs 1 and 5, namely 6 m in 2007, compared with 17 m nominal (17 m is the design value since optics version 6.4 – resulting in 6% more luminosity than the previous 18 m) and a different magnetic state. Paul recommended that ramp studies in 2007 better be made for the 18-m beta* optics and not at beta*=6 m, and that in 2007 we should try to learn as much as possible for 2008. Replying to a question by Massimiliano, Jorg and Frank pointed out that a 17(18)-m optics at 450 GeV may be needed for the nominal LHC, because of the crossing angle required with the nominal number of bunches. There was some debate as to whether the 2007 run could use a 17(18)-m beta* rather than 6 m. The global aperture is the same for either value of beta*, but the local aperture at the triplets without crossing angle no longer has any margin at 6 m beta*. Massimo stated that running at 17(or 18)-m beta* indeed is a preferred option, if the experiments are not asking for highest luminosity.  Gianluigi commented that there may be other ways to easily gain, or lose, a factor three in luminosity, for example injection matching. A possible scenario is to first inject into the 17-m or 18-m optics, and only then into a squeezed optics. Paul cautioned that substantial parts of the optics set up may need to be redone. Another back-up option for higher luminosity is the flat-beam optics. It was finally agreed that we can try to start with an 18-m (17-m?) optics, if the experiments accept the implied lower luminosity, and then raise the luminosity by increasing the intensity, by which factors much larger than three can be obtained. However, this conclusion may already be superseded by the following information. 

 

After the meeting, Stephane commented that one should distinguish between the nominal optics, the pre-squeezed injection optics, and the "high lumi" injection optics (beta*=6 m, flat beams) which he generated a few weeks ago. He reminded us of the pre-squeezed optics with beta*=11m in IR1 and IR5 presented at the LTC on 31-03-2004. Applied in the present scenario, its advantages are twofold: First this optics will allow pushing the luminosity (~10^30 for 43 bunches and 4 10^10 protons per bunches), while leaving a comfortable aperture in the triplet w/o X-angle (~8 mm) and in the matching section. Then, and mainly, this optics will also allow a smooth transition to the nominal LHC commissioning ("just" by switching on the X-angle) and it will ease the later squeeze at 7 TeV (which indeed was its initial motivation). In view of this comment, it appears that the 11-m beta* optics should be the one to be used at the start of the 450 GeV run.

 

A brief discussion on beam finding at low luminosity in LHC followed. Hermann Schmickler is said to be worried that the initial luminosity of 1e28 cm-2s-1 may not be sufficient to use the luminosity signal for steering the beams into collision. Rhodri clarified that the luminometer measures bunch by bunch and sufficient resolution can be achieved by averaging in time. It was remarked that the low luminosity mentioned is still much larger than the design ion luminosity. Paul commented that Werner Herr is investigating possible alternatives. Frank added that bringing the beams into collision was never difficult at other 2-ring machines, such as HERA or the two B factories. He thought there should be no problem if the IR BPMs function reliably. Jorg commented that K modulation of the IR quadrupoles could be helpful for equalizing the orbits of the two beams.

 

Orbit Response Measurements and Analysis (Jorg Wenninger)

Jorg reviewed the measurement and analysis of orbit response data. This measurement consists in exciting orbit correctors and detecting the corresponding changes in beam position. It can be applied either on the first turn (trajectory) or for the closed orbit. The measurement does not give any direct information of beta function and phase advance. Rather it is used to fit many parameters in an optics model, including BPM and corrector calibrations, so as to obtain the best agreement of the simulated and measured response matrices. Nonlinear effects present at large amplitudes can also be analyzed. Also coupling between the horizontal and vertical plane can be included in a straightforward way. The result of the analysis are the BPMs and corrector calibrations, plus the quadrupole strengths. The sum of the squared differences between measured and predicted response is minimized in an iterative process. Constraints for LHC arise from the large number of BPMs and steering correctors, the available computer memory, precision, and CPU time. The program used derives from James Safranek’s LOCO code. Jorg uses MAD-X with the CERN-SPS optics model. The results are displayed in PAW. The program was applied to the SPS, TT10, TT40/TI8, and is ready to be used with CNGS. At TI8 Jorg obtained the results at 1 am in the morning, when nobody was watching. Running the program requires the presence of Jorg and there is no intent to change this for the LHC. Typically only a small number of correctors, e.g., 10-20, is excited. Th execution time can vary from a few minutes to hours or days, depending on the complexity of the problem and the number of quadrupole gradients to be fitted independently. Details on the SPS application can be found in CERN-AB-2004-009, on the TI8 test in AB-Note-2006-021. A gradient error of 20% was easily detected. Also a 1% vertical phase advance error was found over a few cells without any problem in TI8. At the LHC, the resolution should be much better as there are 4 times more BPMs per cell. That the analysis can also deal with strong coupling was illustrated with an example from TT10.

 

For the LHC Jorg expects that the orbit response analysis on the first turn can provide all kinds of polarity errors (BPMs, correctors, quadrupoles), other BPM errors, quadrupole errors down to a few % accuracy, the average phase advance per arc at the 1e-3 level, as well as information on the b3.  With the closed orbit, 4-20 correctors could be excited per plane, yielding information on the BPM quality and the  BPM calibration, the corrector calibration and polarity, and possibly the nonlinear optics. Jorg cautioned however that the optics fits are heavy and that the phase advance measurements presented by Rogelio in the previous meeting would be much “lighter”. He pointed out that there is some synergy between the two procedures.

 

In summary, arguments in favor of the LOCO analysis are (1) the method has been useful at the SPS, (2) the software chain is well tested, including automated data acquisition, (3) it can provide the calibration of BPMs and correctors, (4) it is powerful especially on the first turn, while with a stored beam we have an attractive alternative, and (5) it may provide information on the nonlinear optics.

 

Paul asked whether the technique can be used for matching the transfer line and the ring at the injection point. The answer is only partly, as the matching depends on the initial beam ellipse at the start of the line. However, one could use it to locate errors responsible for a mismatch. For past applications the transfer lines had BPM systems different from those of the SPS. For the LHC fortunately the BPM systems of the transfer lines and the ring will be the same. Frank asked whether constraints from symplecticity are taken into account in the fit. This is not the case, but may be difficult or impossible. Thijs asked about the role of BPM noise. Jorg replied that the noise is inferred from the reproducibility of the orbit without corrector changes, and used for weighting the fit. In the SPS the noise level is 15-100 micron, and the residual after the fit of order 100-150 micron is slightly above the noise floor. The residual is at least partly attributed  to the nonlinear BPM response.

 

Next Meeting

Wednesday July 12th, 14:30

CCC conference room 874/1-011

 

Provisional agenda

 

Minutes of previous meeting

Matters arising

450GeV optics – IR aperture and IR bumps (Yannis Papaphilippou)

Follow up of web documentation? (Verena et al)

AOB

 

 

 Reported by Frank