Tuesday April 24th,
14:00
CCC conference room
874/1-011
Roger informed the team of a suggestion by Massimiliano to include representatives from the five experiments on the LHCCWG information mailing list. The team welcomed the representative Marzio Nessi from ATLAS at this meeting.
Next Roger reviewed the planning for the following LHCCWG meetings and LTC presentations. It was remarked that the topics “collisions” and “two-beam operation” were still missing in the LTC planning so far.
Laurette discussed the commissioning of the beam-loss monitor (BLM) system. Her talk was structured as follows: system overview, principle of simulations, strategy for BLM positioning and threshold settings, the available signals, hardware commissioning, and commissioning with beam. Laurette stressed that the BLMs are the only system protecting the LHC from fast losses occurring on a time scale between 0.4 and 10 ms. They also are the only system preventing magnet quenches. The BLM system is sensitive to ionizing particles over a huge dynamic range. Specifically, this system consists of 3600 ionization chambers and 310 SEM detectors, which measure the secondary shower outside the cryostat. A dynamic range of 1e8 is provided by the ionization chambers alone. The range can be extended to 1e13 by combining the ionization chambers with the SEM detectors. The dimensions of the various detectors are of order 10-60 cm. The signals are read out via optical links to surface cards (16 channels per surface card). In total there are 335 cards on the surface. Each channel (=monitor) provides data for 12 integration periods, taking into account 32 different energy levels. If the signal for one integration period is above threshold, for the respective beam energy, a beam dump is generated. Some channels can be masked with the safe-beam flag.
The expected loss pattern was obtained from simulations by the collimation team. The simulated losses reach a maximum before the center of a quadrupole. Simulations of the secondary shower were performed with GEANT3. GEANT4 was used to model the detector response. The longitudinal impact position was varied. Laurette presented typical simulation results for different impact locations. The maximum of the shower is reached about 1 m behind the primary impact point. The signal increases in magnet-free locations. A difference of about a factor 2 in signal is found between losses in quadrupoles and dipoles. Combining the loss map and the shower simulation for each loss location yields the expected “sum signal”. To catch the losses at the MB-MQ transition, at the middle MQ, and at the MQ-MB transition, three detectors per beam have been placed around each quadrupole. The chosen setup minimizes the uncertainty in the ratio of energy deposited in the coil and in the detector, and it allows discrimination between losses from the two beams.
Laurette further detailed the BLM strategy for quench protection. She reiterated that each cold quadrupole is surrounded by 6 BLMs, and she added hat this is also true for the quadrupoles in the LSS. Families of monitors (about 250) were defined for simplifying the problem and for reducing the number of degrees of freedom. The beam dump threshold is set to 30% of the computed quench level. In addition, also warm elements are equipped with BLMs, notably the collimators, warm magnets, kickers, septa, and masks. Here the beam dump threshold is set to 10% of the damage level. Simulations were executed by the FLUKA team for IR6 and IR7 and by MARS for IR3. Quench and damage threshold tables will be created for each family of BLM locations. They will be assembled into a MASTER table. The 7 TeV threshold is used as a seed for parametrization. The MASTER table as well as a MAPPING table (relating electronic channels and BLM locations) will be stored in LSA.
Commenting on calibration and model verification, Laurette recalled that simulations are needed for three different aspects: the secondary shower, the magnet quench and its dependencies, and the detector response. Ongoing experimental verification includes a measurement of the tails of the secondary shower with a BLM mounted on the HERA beam dump and a comparison of the measurement with GEANT4 simulations, a quench test campaign in SM18 for steady-state losses, and a check of the detector model with the CERN/H6 experiment (M. Stockner’s PhD thesis).
Laurette now turned to the proposed implementation. Thresholds can be changed via a dedicated GUI. The values loaded in the FPGAs are compared with those of the LSA database. The comparison of both is linked to the BIS. Of particular importance is a “switch” to enable the remote change of the threshold settings. Details are still under discussion. The flexibility of remote changes has to be balanced against reliability. The possibility of scaling the thresholds for an entire family is envisioned. The remote access will need to be validated by machine protection experts.
As for the signals available, 12 running sums for the various time scales fill individual buffers. Different systems receive and process BLM information, such as logging, post-mortem, XPOC and collimation. A fixed display is planned in the CCC, for which a normalization of the signals to the threshold values is being considered. A large amount of information needs to be represented, i.e. the readings from about 4000 monitors at 12 different integration times.
Other uses of BLMs include the mobile BLM system and BLMs for ions. For the mobile BLMs, 2 spare channels per card can be used, which provides 6 additional BLMs to place along the dipoles. A separate fixed display for non-active channels is foreseen. Concerning the BLMs for ions, a number of more specific loss locations need to be equipped with the same type of BLMs, e.g. at dipoles in the DS and the arcs of IR7 and IR3, and in certain cells of IR1, IR5 and IR2.
The hardware status is as follows. The ionization chambers are on time for the LHC start up. The SEM delivery is scheduled for the summer. The ionization chambers are installed in all of Sector 7-8 and halfway through Sector 4-5. Installation in 8-1 has also started. A prototype acquisition system is continually running at HERA. The fixed display is to be completed by CO. Extensive software tools for data analysis still need to be specified and implemented. The hardware commissioning procedure is documented in a manufacturing and test folder (MTF [C. Delamare et al , EPAC’02]). The machine protection sub-working group will review all pertinent functionalities. Connectivity verification is done with the help of a radioactive source in the tunnel. Laurette showed a flow chart of all BLM testing procedures, which were developed in the PhD thesis of G. Guaglio.
Stefano asked whether we require that all 4000 channels are correctly working all the time. Rudiger recalled that this question had been discussed in the past, including the associated difficulties. Bernd replied that the conclusions of the previous discussions are included in the BLM specification. One of these conclusions was that some arc monitors can be disabled. For critical monitors all channels are required, however. Laurette mentioned that we can always disable some of the channels. In particular, if we open the switch, we can disable the BLMs from the control room. Bernd commented that a promising statistics with high BLM availability exists from the SPS and the PS. It was remarked that the distinction of maskable and unmaskable BLMs is not completely defined.
Lastly, Laurette turned to the commissioning with beam, referring to the talk “Magnet Quenches with Beam” by A. Koschik in Chamonix’06. The main goals are the verification of the correlation between energy deposition in the coil (quench level) and BLM signal (threshold), the verification or establishment of “real-life’ quench levels, and the verification of simulated BLM signals and loss patterns. The idea of the experiment is simple: steer the beam into an aperture and cause a magnet quench. Both steady-state loss quench limits and fast-loss limits could be studied. This test could become part of the MPS commissioning. Laurette emphasized that parasitic quenches are not useful for a number of reasons, and that the clean conditions of an experiment are preferred. She specified the beam conditions and disagnostics needed for such beam test.
Laurette concluded that the BLM system is designed to be reliable and to fulfill the specified functionalities of machine protection and quench prevention. Uniform hardware and software for all monitors eases the maintenance. A controlled, defined test with beam is essential for an early verification of the BLM system. This beam test will yield the absolute quench limits and BLM threshold values, as well as validate the correlation of loss pattern, quench levels, and BLM signal. The remote access to the thresholds table still has to be approved by machine protection experts. The families to scale have to be defined by OP.
Jan asked for a confirmation of the maximum intensity of 1e11 protons. Laurette replied that this was a historical number proposed at the time of sector-test planning. Alternatively, 43 bunches with 1e10 protons, as foreseen at the moment, would also be a viable option.
A discussion ensured on the remote access to the thresholds with unclear conclusion. Helmut asked whether the BLM commissioning with beam will definitely be done. Jan responded that, yes, its priority had just been raised to the highest level 1. Paul remarked that we should not be afraid of quenches. He considered the planned threshold equal to 30% of the quench level as far too low. Rudiger cautioned that the difference between damage threshold and quench threshold depends on the beam energy and time scale. He warned not to approach the damage limit. Helmut commented that quenches entail the risk of damage and cost time. Concerning the scale factor, Laurette made the point that reliability is essential. Jan proposed including the damage levels as maximum values in the FPGAs. Laurette answered that she would need to check if there is enough place left on the FPGA to perform such additional comparison.
Roger asked who is looking at the fixed displays. Mike replied that LSA can provide this display easily. He already read and displayed 2000 channels in LSA. Oliver reminded of the HERA experience, where the causality of events was often difficult to understand. He recommended implementing a time trigger to record which events happen first. Roger commented that this is already provided by the post mortem system. Oliver proposed having some special BLMs displayed continually in a dedicated window, e.g. losses next to the primary collimators. Stefano remarked that a continuous display of BLM signals will be available within the collimator software for the dedicated BLMs that are attached to the collimators. These signals will be used for the beam based alignment of the collimator jaws with respect to the beam. Stefano asked whether the data saved in the logging include the longitudinal position. Laurette replied that the longitudinal position is available from the mapping table. Jean-Jacques commented that only the channel name comes from the front end system.
Gianluigi asked whether when disabling some arc BLMs the consistency between the database and the hardware is guaranteed. It was also asked why 12 different integration times were chosen. Bernd and Laurette responded that these 12 times are listed in the specification. The number was deduced from the dependence of quench & damage levels on the loss duration (plot shown).
Oliver asked for the BLM exit condition, e.g. do we test one BLM or how many? Helmut replied that studies for about 3 places and with a total of roughly 10 quenches are planned. Verena and Laurette pointed out that the strategy for warm elements is a bit different, requiring a “clean” injection. Another question posed was whether the quench tests are performed at one beam energy only, or also during the ramp, and at top energy. Helmut responded that we should also have a test at 7 TeV, or at least have the option to do such test, e.g., before dumping the beam.
Oliver highlighted that the person responsible for the BLMs should define the necessary tests. Jan suggested that it will be important to establish a coherent picture of the BLM readings all around the machine. The sub-working group will define the exact procedure for quench tests. Operational experience, some quench tests, and some failure scenarios together should ensure the emergence of the coherent picture.
=> ACTION: Define the necessary BLM beam tests (Helmut? Laurette?)
Ralph stressed that the understanding of quench levels is crucial. He proposed considering a different strategy for the thresholds, namely to start the commissioning with relaxed threshold settings. Then if quenches are experienced at higher intensity, the settings could be tightened, in the process yielding additional information. Jan asked about the relation to the damage threshold. Ralph remarked that the presently advocated strategy, with threshold settings at 30% of the computed quench levels, leads to premature beam dumps with zero information on the actual quench levels.
Thijs presented simulated particle energy spectra of protons, neutrons, and charged pions in the LHC underground areas, for various shielding materials, which illustrated that, in particular, the low-energy tail varies widely with the material. He emphasized that the dose alone is not sufficient to characterize the radiation, as it had been the case for LEP. He discussed the parametrization of a complex radiation field, taking into account some additional characteristics of the radiation, such as the rate of particles at energies above 20 MeV or above 100 keV, which are linked to single events and bulk damage. The various components of the RadMon system provide this additional qualification. Thijs showed photos of the monitor parts and assembly. The RadMon was calibrated in many different environments at five facilities. Almost all the LHC underground areas are equipped with a total of 309 RadMon monitors. The installation in some of the IRs is complete. Thijs described the location of the Radmon detectors for CMS and in the LHCb cavern, pointing out that the RadMon positions can be flexible for the start up.
A RadMon test was performed in the SPS primary target hall BA80 in 2004, where the monitor was able to resolve successive SPS supercycles. A further measurement showed data taken during the 2006 SPS start up. Another RadMon detector is deployed in the CDF collision hall at FNAL, where the monitor indicates that the low-beta quadrupoles act like a “line source” of radiation, and that the radiation could be responsible for a high-failure rate of a low-voltage power supply (LVPS) at the proton side of CDF. An interesting event was a failure of the electrostatic separator in the Tevatron. This accident was not accompanied by any measurable dose variation, neither by a change in the 1-MeV neutron flux. The only change in signal was seen in the particle fluence above 20 MeV in the Southwest region of the CDF detector
RadMon detectors will be used for the TI8 test in July 2007. Thijs presented the data storage and the real-time displays available in the control room. Then he commented on the possible use of the RadMons during a sector test with its anticipated beam intensities.
Thijs concluded that the RadMon system provides for mobile radiation measurements under SC magnets, in caverns and in the LSS. It features a 1 Hz data logging with real time data display and a spatio-temporal resolution complementary to that of the BLMs. Thijs’ proposal for LHC startup (TI8 test, sector test, eng. run) is selecting an adequate subset of BLMs and RadMons for each stage, and to display and store data of this subset for an immediate analysis in the CCC. For the nominal LHC he proposed creating a database of radiation measurements, where subsets of the radiation measurements are stored after each beam loss, allowing to perform statistical studies and to address a number of questions, like: How are beam losses and radiation levels related (spatio-temporal)? How is the LHC being operated? How is the equipment performing under radiation? How accurate are the simulations? How efficient is the shielding?
Rudiger asked for the read-out frequency of the RadMons. Thijs answered that they are read out at 100 Hz, but he mentioned a few subtleties, such as the need of multiple accesses to retrieve all data for one instant of time. Thijs also explained that the RadMon does not just give the integrated dose, but also the rate, flux, fluence, etc. The question was asked how the RadMon data will be interfaced to the logging and post-mortem systems.
=> ACTION: Interface RadMon to other systems (Thijs, Rudiger?)
This discussion had been meant to be web based, covering phase A.8 of the commissioning procedures for snapback and ramp. The presentation was cancelled due to computer problems. Phase A.8 will be discussed directly at the LTC of May 9.
Roger announced that the material of the LTC presentation is used for preparing EDMS documents on the commissioning procedures. These documents will provide a snapshot of where we are at the moment. The first document will come out soon.
Tuesday May 8th, 14:00
CCC conference room 874/1-011
Provisional agenda
Minutes of previous meeting
Matters arising
2 Beam operation
(Jan, Ralph)
MAD-x online model
(Frank S.)
Commissioning of accelerator system -
magnets (Walter)
AOB
Reported by Frank