Minutes of the LIS Section meeting held on 16 April 2012
Present: G. Arduini, H. Bartosik, C. Carli, M. Fitterer, V. Forte, A. Garonna, C. Hernalsteens, A. Lachaize, B. Mikulec, A. Molodozhentsev, Y. Papaphilippou, V. Raginel, F. Schmidt, G. Sterbini, R. Wasef, C. Yu
General Information (Gianluigi)
Gianluigi will be absent from Thursday 19/4 afternoon until Friday 20/4 afternoon (included)
Gianluigi introduced Guido Sterbini who has joined the LIS section as staff member (officially from 1/6/2012). He will work on the PS and its upgrade.
CAS participant list approved at the BEMB this morning (from LIS V. Forte). Should go ahead with registration and the preparation of the travel.
IPAC12 papers to be uploaded on the BE server for group and departmental approval by 30/4
Deadline for the fellow requests (2 LAGUNA fellows for LIS) tomorrow in preparation for a first group discussion on Wednesday 18/4.
Requests for Technical and doctoral students have been submitted (1 for LIS/ICE for ELENA collective effects)
Group meeting will take place on 18/6 (presentations in the morning followed by BBQ in the Prevessin area)
Campaign for woodworm eradication started.
Gianluigi mentioned that the BTP pick-ups are not working at the BEMB. Follow-up by JJ Gras.
LHC status:
LHC operating with 1092 bunches and bunch population >1.3x1011 p at 4 TeV and 60 cm beta*. Luminosities close to 4x1033 cm-2s-1 have been achieved.
Plan to increase to 1380 bunches before the MD and Technical stop block. Issues with one injection kicker. Will be likely replaced during the next technical stop.
Space charge studies for PSB - preliminary status (Vincenzo) - slides
Main aim of the study so far is the individuation of the optimum simulation parameters to obtain convergence within reasonable execution times.
The evolution over 1000 turns has been simulated so far. This allows to cover more than a synchrotron period and it has been mostly limited by computation time so far. A significant improvement is expected, see the e-mail form F. Schmidt after the meeting and reproduced below:
"....
I did the math and I have a rough estimate of the gain in throughput of the 48 core option compared to lxbst (ENGPARA).
- The present system consists of 40 nodes 8 cores each. Special fast communication between the nodes is implemented.
- The 48 core nodes do not need fast communication between each other which should bring down the price. The administration should be easier as well. It goes without saying that only one job should be run per 48 core node. This option can easily be upgraded without setting up a new system.
- The benchmarking shows that a typical PTC-ORBIT job under ENPARA uses about 10 nodes with 1-8 cores. The speed of the 48 core machines is 1.5 slower than ENGPARA when 48 cores are used and a single user is on that machine.
- Given the fact that we have to share ENGPARA with others I figure that we can count on about 50% availability.
- There are considerable waiting periods of a day or even more on ENGPARA but I will not factor that in. I hope that in the new set-up each of the machine team will have 10 machines available avoiding long waiting periods.
=> 20 * 8 * 1.5 / 48 = 5, i.e. 5 0f the 48 nodes replace our present available ENGPARA system. Therefore the throughput will be 6 times larger for PTC-ORBIT than what we have now. For the three teams it means that they will have the double capacity of ENGPARA for themselves and no waiting period.
This analysis will be checked with Simone and Bernd Thursday in one week. We will also see about the costs for this system.
....."
The beam parameters to be used in the future simulations should be defined. Action: Christian.
A beam with moderate intensity and tunes below the half integer (4.27-H/4.43-V) has been considered for the convergence study. No significant emittance blow-up is expected for these parameters. The convergence as a function of the number of macroparticles and the number of grid points has been studied. The required CPU time depends mostly on the number of macroparticles and only slightly on the presence of space charge nodes and on the number of grid points to calculate space charge forces. This seems to indicate that teh CPU time taken is mostly dominated by PTC tracking.
So far boundaries have been considered for the calculation of space charge, in that case the mesh is uniform inside the metallic boundaries on the other hand the mesh adapts to the beam transverse distribution in the case of simulations with no boundary conditions (direct space charge). It is expected that direct space charge effect should dominate for for the PSB at injection and low energy.
From the discussion during the presentation it is evident that observables to be simulated and parameters and tools for the simulations should be defined as a function of the phenomena to be studied, in particular:
Evolution of the r.m.s. emittance vs 95% or 99% emittance. Evolution of the r.m.s. emittance might be important for understanding the core emittance blow-up (which is relevant for luminosity production and therefore for the LHC beams) in long term simulations while 95 or 99% emittance evolution might be important to understand halo formation and losses which are expected to occur mostly at low energy and are particularly relevant for
Direct space charge vs. space charge including boundaries. Direct space charge is dominating in the PSB at low energy.
ORBIT-PTC vs. ORBIT. ORBIT-PTC has several advantages for the lattice description and maintenance (introduction of non-linearities, double harmonics, time dependence, accurate tracking, etc.) but the first experience indicates that the required computing time is more important.
Space charge studies for PS - preliminary status (Raymond) - slides
A convergence study was performed in the PS for 1000 turns. The study showed convergence for approximately 1024 turns and 256-512 kMacro particles.
The tests performed by varying the number of macroparticles indicate that for the optimum parameters the CPU time is dominated by the tracking and only for mesh sizes larger than 1024 (e.g. 2048) a significant increase of the CPU time is observed. A 2nd order integrator as been used for the PTC tracking. In slide 9 the red tune footprint refers to a 90 ns bunch length.
Information from Simone (e-mail)
PS ventilation. Renovation under definition now. Documents for the call for tender ready. All 8 stations will be renovated in group of 4, taking in total 6 months.
To avoid R2E problems, the PLCs of the control system will be placed in the central building.
The system will have the same functionalities as today except the fresh air flushing before an access. Still under discussion is the smoke extraction in case of fire, and actually the procedure to follow in case of fire.
MSWG: during the last meeting we had a review of the beam instrumentation for the PSB/PS/SPS.
There are some concerns about: PU in BT/BTP and TT2, the cross calibration of the ring transformers between PSB and PS and the ones of TT2, the BWS in PSB and PS for some of the settings. For the SPS, the last pickups of TI41, before the target, are giving un-physical reading causing many interlocks. In total, for this about 10% of the total CNGS flux was lost in the last few days.
More issues will appear in the minutes. BI is following the different problems.
EAST area physics started on Monday. Only serious issue to be solved is that the ARCON system is giving spurious monitor faults.
The first Hiradmat Scientific Board & Technical Board Meeting will take place next monday.
Next meeting
Monday, 23rd April 2012 at 14:00 in room 6-R-012
Agenda:
General Information (Gianluigi)
Back to Minutes