Beam-Beam and Luminosity Studies meeting

6/2-008 (CERN)



Show room on map
Guido Sterbini (CERN), Yannis Papaphilippou (CERN)
Videoconference Rooms
Beam-beam and luminosity meeting
Nikos Karastathis
Auto-join URL
Useful links
Phone numbers


G. Sterbini, M. Zampetakis, N. Triantafyllou, S. Kostoglou, K. Paraschou, A. Poyet, Y. Papaphilippou, G. Iadarola (remote), M. Carla', N. Karastathis,  F.F. Van der Veken, B. Salvanchua, J. Malewicz, E. Cruz Alaniz, A. Rossi, E. Metral, X. Buffat, K. Skoufaris, E. H. Maclean, N. Mounet. 

Julie presented the latest results of the tracking simulations to determine the beam-beam and more in general the losses of the LHC.

The goal of the study is to simulate 30 minutes of beam (or 20 million turns), and compare results to actual observed losses of about 0.1%. More in general the target is to address the simulation of beam losses on a time scale comparable with the fill's length.

The software used is the Sixtracklib (simple python support and GPU parallelization) coupled with GPU hardware (4 in Bologna and 12 in HTCondor).

The main ingredients for the approach are the normalization of the phase space, the factorization of the initial distribution and the Montecarlo integration.

To improve the convergence speed of the Montecarlo integration, a stratified sampling in the transverse space and a binning in time. This allows to reduce significantly the uncertainty on the computed losses and therefore to compare them with the measured ones.

Guido and Yannis praise the results of the works: it is very encouraging to see that the steady losses level could be reproduced. 

Guido  asked about the time to complete the study (10k particle and 20e6 turns) assuming ideal hardware availability.  Kostas answered that it takes 4.5 days.

Elias asked about the aperture model adopted. Julie answered that the tracking was performed using a DA approach with physical limit at 1 m.  Gianni commented that indeed it would be interesting to adopt a more realistic model of the machine's aperture. This could potentially address the discrepancy between the observations and simulations during the initial transient of the losses. Sofia added that in her studies the aperture model had a direct impact on the losses.

Xavier presented an update on the operational scenario taking into account of the constraints on coupling and non-linear optics corrections. The main requirements concern the coherent stability when putting the beam in collisions (transient effect considered as a static stability requirements, conservative approach) and the DA target of 6 sigma once they are in collision.

Studies showed that BB effects during the collapse of the separation increase the need of octupoles strength (collapsing one IP before the other is beneficial in this sense). 

Xavier presented the new estimate of the  stability during the ADJUST (beam separation collapse) obtained with a finer mesh.

Concerning the effect of the coupling, 

  • a residual ΔQmin~10-3 should be expected after optics correction
  • the effect of linear coupling is worse for the positive polarity of the octupoles as it affects mostly the indirect detuning term

These considerations bring to an educated guess of a tune split of 5e-3 during collision.

The current budget of the octupoles is partially absorbed by the lattice non linearities. An estimation of the needed octupoles current was conducted starting from the WISE model.

This resulted in a setup proposal that defines the $I_{oct}$ and $r_{ATS}$ parameters to ensure stability (for negative and positive polarity and for LS2 and Full upgrade).

  • The setup proposal feature a larger telescopic index and a reduced octupole current lower than the maximum (±570A), to allow for compensation of the potential detrimental effect of linear coupling and lattice NL
  •  Pessimistically, we require a sufficient dynamic aperture with the full current in the octupoles and the required telescopic index

The positive octupoles is not compatible with the stability for the non colliding bunches.

Starting from these configurations DA simulations studies were conducted (assuming the ($|I_{oct}|$=570 A).

The DA studies shows that it is not possible to fulfill together the requirements for DA and beam stability.

Similar conclusions can be drawn for positive polarity for the octupoles.

The DA and machine margin can be re-establish by reducing the bunch current with evident negative effect on the machine performance. 

Axel  asked if during the beam collapse the octupoles configuration could be optimized to gain stability margin. Xavier answered that the circuits are too slow for such rapid transition. Adriana and Yannis commented that the wire compensator circuit's could indeed cope with these fast transient.

Elias commented that it is important to convey a clear message on the coupling limit of 1e-3 after correction, since it has important implication.

Yannis suggested to also consider the chromaticity as a possible parameter to improve the situation. He asked if the damper configuration is the standard one. Xavier answered positively.

Adriana  asked why the CC impact negatively on the beam stability margin. Xavier explained that they increase the effect of the HO interaction by reducing the beams crossing angle.

As AOB, Guido informed about the next SWAN workshop. Riccardo de Maria is proposing a abstract for the next SWAN user's workshops. If you want to share your experience (mainly problems encountered) or you wants to ask a new features please contact Riccardo, Guido or Michi.

There are minutes attached to this event. Show them.
    • 16:00 16:30
      Simulation of Beam Losses at the Large Hadron Collider 30m
      Speaker: Julie Lise Aline Malewicz
    • 16:30 16:50
      Update on the operational scenarios for Run3 20m
      Speaker: Xavier Buffat (CERN)
    • 16:50 17:10
      BB and WP2 open actions 20m
      Speakers: Guido Sterbini (CERN), Yannis Papaphilippou (CERN)
    • 17:10 17:20
      AOB: SWAN workshop and proposed abstract 10m

      Proposal from Riccardo De Maria

      SWAN for machine studies

      R. De Maria, M. Hostettler, G. Sterbini

      SWAN can be used to extract machine measurements, query machine settings and beam dynamics simulations out of the box thanks to the pytimber, pjlsa, cpymad packages installed in the LCG stack. We present the status of those packages and our wishes for the SWAN platform.

      Limits to underline:

      • poor integration with lxbatch/htcondor, no sshfs, no  robust filesystem for home directories
      •  Little flexibility in resource usage & VM configuration, e.g. RAM, number of cores 

      • Automatic (non configurable) timeout after which the system kills the VM/Notebooks and all unsaved data in it.

      Should we refer to NXCALS and its integration on SWAN? Will there be a dedicated talk from CO/OP?