LHC Report: a brief deceleration
The LHC has now transitioned from powering tests to the machine checkout phase. This phase involves the full-scale tests of all systems in preparation for beam. Early last Saturday morning, during the ramp-down, an earth fault developed in the main dipole circuit. Full evaluation of the situation is ongoing.
The various systems are put through their operational paces from the CCC. This includes important tests of the beam dump system and full-scale tests of the beam interlock system (BIS) and its many inputs from other systems around the ring. All magnetic circuits are driven through the ramp, squeeze, ramp-down, and pre-cycle along with the collimators and RF. Instrumentation, feedbacks, and the control system are also stress tested. Inevitably there is some final frantic debugging but, up to now, things seem to be in reasonable shape. The machine checkout is an important coming together of all LHC systems. During this final phase before beam, the operations team tests all of the LHC subsystems to make sure the entire machine is ready for beam.
The powering test phase has left all but two of the 1700 or so magnetic circuits fully qualified for 6.5 TeV. This is the result of a six-month long programme of rigorous tests of the circuits involving the quench protection system, power converters, energy extraction, UPS, interlocks, electrical quality assurance, and magnet behaviour. Sector 4-5 dipoles have proved a little stubborn but are now at the target value of 11,080 A (6.5 TeV + 100 A) after some 50 training quenches and sector 3-4 dipoles are also very nearly fully trained to the same value.
However, on 21 March, early morning, an earth fault developed in the main dipole circuit during the ramp-down following what was probably the last training quench of sector 3-4. All the protection systems functioned properly and there was no harm done. The fault developed at relatively low current and was intermittent in nature at this stage.
Three main options are being explored. The first would inject an energy-limited pulse of current and attempt to melt the metallic fragment that is causing the short circuit. The second option would involve pressurising the helium in the local cryogenic sector and then performing a fast pressure discharge to generate turbulent flow and so dislodge the object. Studies and preparation for both these options are ongoing and both could be attempted relatively quickly. In-situ measurements by system experts have located the fault to within 10 cm by injecting current locally and using the standard cold mass instrumentation, which includes voltage and current taps. The fault is located in the vertical tube that leads from the magnet enclosure to the diode box situated under the magnet (see below). The most probable scenario is that a small piece of metallic debris has inadvertently found its way into this tube and is making contact between the tube (earth) and one of the cables that leads to the diode. X-rays have been taken of the region. It’s a difficult location and although some debris can be seen, the results are inconclusive.
The third option involves a partial warm-up of the sector and opening the magnet interconnect concerned. This would allow direct access to the diode box. The warm-up, intervention, and subsequent cool-down would take around six weeks.
Full evaluation of the risks of each option is ongoing. It’s an interesting and frustrating problem; care is being taken that the only eventual cost is time.
| Diode box  Image credits: Arjan Verweij. Schematic of dipole circuit – diodes are the arrowheads under the magnet coils.When a magnet quench occurs, the current in the quenched magnet diverts to the diode in about 0.5 s, and the rest of the magnet chain in superconductive state slowly ramps down with time constants of the order of 100 s. Thus the diodes conduct a current pulse of up to 13 kA, which decays exponentially with a time constant of about 100 s. This can lead to a temperature rise inside the diode of up to 300 K. |
by Rossano Giachino & Markus Albert
