Joint NDC-LNO section meeting: OMC in beam test and LSTM Application for Automatic LHC Collimator Alignments
13/10/2021 Joined NDC-LNO on OMC in LHC beam test and automatic collimator alignment
News and round table :
Next week will be particularly busy : LHC beam test and HL-LHC collaboration meeting.
Concerning teleworking : the updated procedure was circulated by email; > 40%, TW from outside local area or TW combined with vacation requests require approval by section and group leader.
Injectors, Alex Huschauer
Currently checking and benchmarking SEM grids with Beam Gas monitors in dedicate PS beam time (require PS beams to be dumped after 50 turns to avoid SEM damage) and crab cavity test in SPS.
Booster intervention yesterday to fix a moving trim coil.
Small 1 mum emittances have been achieved at 1.6e11 intensity for BCMS beams in the PS.
Parallel SPS MDs are planned tomorrow to verify SPS beams and measure their emittance.
Some first LHC beam requests can be expected to check out transfer lines in the coming days.
Optics measurements in injectors, Ewen Maclean
Optics measurements including beta-beat are performed in the PS, more recently also at top energy using the AC dipole with extended acquisition time featuring good quality. For the Booster : an rf-mismatch was fixed at injection. Doing optics measurements in the booster is still a bit more difficult than in the PS. Acquision over longer time also to be tested at top energy in the PSB.
OMC activities during the beam test, Tobias Persson, slides
-----------------------------------------------------------
Tobias shows the schedule of the LHC beam tests. First beams in the LHC are scheduled next week for Thursday 21/10, but might actually start one or two days earlier if the preparations continue to go well as presently the case.
Optics measurements with AC dipole excitation and BPM recording (V,H, 2D) are planned from the first day with beams. The OMC-OP workshop in 2019 brought the two worlds closer together and some important outcome are about to materialize, like the definition and use of the target / correction trims to be able to extract machine settings. The extension to 3D (combined with RF-modulation) is also planned in tests. 3D measurements could enable in RUN3 to record dispersion and chromatic functions as well as transverse measurements without need for a separate ramp. New software and functionality has been implemented during the shutdown and can now be tested in real beam conditions from the CCC. The operational software is being improved to facilitate and trace changes of target values, like in particular changes of the (tune) working point.
It is planned to measure the beta-beat and local coupling at injection energy in the LHC. Sector by sector optics measurements using K-modulation and global Q'' and Q''' measurements are also planned. If things go well it is planned to also measure with a 60 degree optics as presented in the last LNO meeting. The beam stay clear at 60 degree gets reduced and the dispersion increased. Support with collimator experts will be needed for the 60 degree tests. The tests will be limited to pilot intensity (single bunch 1.e10).
Rogelio mentions that a second OMC-OP workshop could be organized end of 2022 to see the progress from first workshop in 2019.
Tobias wondered if some preliminary aperture measurements with AC dipole could be performed within collimation or OMC shifts. Stefano replied that it is not excluded but time is quite short (maybe actually just in between two adjacent OMC and collimation shifts something could be fit).
LSTM Application for Automatic LHC Collimator Alignments, Gabriella Azzopardi, slides
-------------------------------------------------------------------------------------
Gabriella presents her oral contribution for ICALEPCS’21 next week.
Machine learning has already been very successfully applied to align the numerous (now 123) collimators by moving the jaws and analyzing loss signals recorded with BLM, reducing the time required from initially over 10 hours to an hour at the end of RUN2. She explains how this could be further improved reducing the time needed potentially by 50% based on LSTM-RNN (Long Short Term Memory, Recurrent Neutral Network) techniques. The techniques are used to quickly classify loss spikes and identify actual loss spikes caused by collimator jaws moving into the beam.
Rogelio asks if similar techniques could also be used to automate aperture measurements: There is potential for gain but implementation is not evident, to be investigated.
Tobias asks what happens in case of faulty BLMs and if BLM verification could be automated using collimators. Grabrielle says faulty BLMs have not been an issue so far and that losses are also visible in the next downstream BLM. Stefano reminds that there are many collimators, but not as many as BLMs, such that a full verification of BLMs based on collimator movements appears difficult, particularly in the arcs.
Next meeting: after the LHC beam test, possibly in 3 weeks again as combined NDC-LNO