AROUND 8 p.m. on November 7, it became evident that the restart of the Large Hadron Collider (LHC) should occur very soon. On that day, for the first time since last years incident of September 19 (Frontline, January 30, 2009), protons were successfully sent around half of the ringed accelerators 27-kilometre circumference and stopped at one of the experiments located around the ring called the Compact Muon Solenoid (CMS). Low-energy protons from the LHC were dumped in a collimator of the massive CMS experiment. The CMS detectors recorded the tracks left by particles coming from the dumping point, called a splash event.
This was not the first instance of particles re-entering the reconstructed, refurbished and consolidated accelerator beam pipe after the 14-and-a-half-month hiatus. On October 23, the first beam of particles in the form of lead ions was injected into the accelerator, which travelled the entire length of a sector of the machine, Sector 1-2. The LHC ring is made up of eight sectors, each 3.3 km in length. These trial runs now signal the high probability of the restart, after many postponements, now actually happening in the second half of November, as stated recently by the CERN Director-General, Rolf Heuer.
The LHC is designed to accelerate counter-rotating beams of protons (which belongs to the family of particles called hadrons) with the maximum energy per beam of 7 trillion or Tera electronvolt (TeV). Housed in a continuous cryostat, a battery of superconducting electromagnets (1,232 dipoles and 392 quadrupoles), having a high magnetic field of 8.3 tesla and operating at ultracold temperature of 271.3 C (1.9 K), keep the particles circulating.
The recent initial trial runs, of course, were at a low energy of about 1.2 TeV (corresponding to a current of 2,000 amperes in the main dipole circuit). Even at the expected restart later this month, the LHC will initially run only at 3.5 TeV per beam, which is half the maximum design energy. Until July, the intended restart energy was pegged at 4-5 TeV per beam.
This decision, however, was changed and the restart energy brought down to 3.5 TeV in early August. We have selected 3.5 TeV to start with because it allows the LHC operators to gain experience of running the machine safely while opening up a new discovery region for the experiments, said Heuer on August 6 (see interview with Steve Myers, Director of Accelerators and Technology at CERN). The most powerful accelerator currently operating is the Tevatron at Fermilab in the United States, which makes protons with energy of 1 TeV per beam to collide head on.
In a statement on October 30, Heuer said that the first circulating beams would be realised by the second half of November, with particle collisions at injection energy following soon after. If things go well, he added, we may have the first high-energy collisions before the end-of-the-year break, which would be the best Christmas present I could wish for. The LHC is touted as a discovery machine. The road to these discoveries of new particles and new physics that lie at these energy scales was expected to be laid at the end of last year itself. But the unexpected mishap brought the gigantic machine costing over $8 billion to a standstill for over a year.
As would be recalled, the LHC was ingloriously shut down after its spectacular switch-on on September 10, 2008. The most likely cause is believed to be the melting of one of the 10,000-odd splices in the bus bar connecting the superconducting cables from two dipole segments in Sector 3-4 because of abnormally high resistance (220 nano-Ohm) over 500 times what it should have been due to bad soldering. This led to a thermal runaway (see Myers interview) resulting in an explosive arc that punctured the insulating vacuum and blew up the circulating liquid helium cooling system and other parts in the accelerator ring with the damage zone extending over a distance of 700 m of Sector 3-4.
There was extensive collateral damage as well. As the helium could not escape fast enough, there was a rapid build-up of pressure caused by the expanding helium into the insulating vacuum. A pressure wave travelled outwards on either side of D-zone and impacted other magnets in the neighbouring sectors as well.
The recommissioning of the LHC basically involves three parts: the various repairs in the damaged Sector 3-4; installation of systems to monitor the LHC closely and ensure that a similar incident does not recur; and, extra pressure relief valves (PRVs) for helium to escape in a safe and controlled manner should there be leaks inside the LHCs cryostat during the machines projected life of 15-20 years. It would seem that the tasks at hand have turned out to be much more involved than originally anticipated as the schedule for the restart date has got pushed every now and then. Some unforeseen problems have only ended up adding to the delay.
Soon after the release of the report on the analysis and diagnosis of the disastrous incident of Sector 3-4 in December 2008, a restart date in July had been set. But this soon got pushed following an LHC workshop in February in Chamonix, France. The new schedule aimed for first beams by the end of September with collisions in October. Apparently in Chamonix there was a consensus that the revised schedule was tight but realistic and achievable.
In mid-July there were helium leaks from the flexible hose in the liquid helium transport circuits into the insulation vacuum in Sector 8-1 and Sector 2-3 while they were being prepared for electrical tests on the copper stabilisers around 80 K.
In both cases the leak was at one end of the sector where the warm current from the power supply enters the cold atmosphere of the accelerator ring. This required a partial warming of the sectors and consequent delay. Apparently a similar leak had occurred in Sector 4-5 two years ago. Though the exact cause was not identified then, the hose material was changed. But this time around, the cause too has been identified (see Myers interview). The flexible tube has now been replaced in five places by a solid tube with an expansion loop to take care of the contraction during cooling with helium flow at half the speed.
Another delay occurred because of a short-circuit to ground in the dipole circuit of Sector 6-7 on August 20 requiring the cooling down that was under way to be stopped. Early in November, another minor helium leak occurred causing some disruption of ongoing activities briefly. The freakiest was, of course, the one that happened on November 3. A piece of baguette bread dropped by a bird caused a short circuit in an electrical outdoor installation that served Sector 7-8 and Sector 8-1 of the LHC. This knocked out the operation of the LHC cryogenic system. The incident, a CERN release said, was similar in effect to a standard power cut and normalcy was restored quickly.
Following the Sector 3-4 accident, a total of 53 superconducting magnets (39 dipoles and 14 quadrupoles) were removed from the tunnel and brought to the ground for inspection, cleaning and repair or reuse. Most of the removed magnets (30 dipoles and 7 quadrupoles) were replaced by spares. The remaining 16 were refurbished and recovered.
About 2.4 km of the vacuum beam pipes (78 per cent) of Sector 3-4 were spoiled, 19 per cent by soot and 59 per cent by multilayer insulation debris, requiring in situ cleaning. Now all the magnets have been installed. The last magnet was lowered into the tunnel on April 30.
According to Steve Myers, to be sure that those magnets left in the tunnel were indeed all right, they used all the electrical quality assurance techniques on them and did a very serious evaluation. We were worried about this just because it hasnt been visually damaged doesnt mean that there is nothing wrong, Myers said.
Further, they took out one of the magnets which they had decided to leave in but which could have been the worst of the ones that were going to be left behind, put it on the test bench and measured it completely. It was a magnet that was closest to the damaged area. It was the one that we were saying should we leave this one in or should we take it out? The decision was to leave it in. Then we said okay this is our most worrying one. Lets take it out and make some measurements on it. We did that and it was good, says Myers.
Besides Sector 3-4, Sectors 1-2, 6-7 and 5-6 also were warmed up for consolidation work. While the last one involved repair of the cryostat, the first two involved replacing of the dipoles because of anomalously high resistances (100 and 50 nano-Ohm respectively) in their internal superconducting joints. According to Myers, these did not suffer the fate of Sector 3-4, which was an external bus-bar joint, because of their inherently better mechanical stability. So there is a little bit of margin in that as well but there is also a factor of three in the resistance, he said on why these did not quench while Sector 3-4 did.
A magnet quench occurs when a superconductor becomes normally conducting. Interestingly, these magnets also had apparently been trained to 13,000 A (corresponding to full beam energy) but they had not quenched in spite of their high internal splice resistances. It shows you that 13,000 A is still not a problem if the resistance is 200 times the other good resistances, he remarked.
An important element in preventing collateral damage is the new pressure relief system. It has been designed in two phases. The first phase involves the installation of 100 mm PRVs in the existing vacuum ports in the whole ring. LHC engineers have calculated that in a Sector 3-4 like incident, the collateral damage (to the interconnects and superinsulation) would be minor with this first phase PRVs. The second phase involves additional relief valves on all the dipoles. According to Myers, a worst-case analysis has been done assuming that the worst thing that could happen would be twice as bad as last years incident. It is what is called the Maximum Conceivable Incident. If that were to happen, the first phase PRVs will not be enough. The second phase PRVs, which are 200 mm flanges (DM200), added on each dipole, will come into play.
Since four of the sectors were warm for other reasons following the accident, it was decided to put DM200s only on those four sectors and the others would be done a year later. Analysis has said we are safe for the entire machine for a September-like incident and we are safe for half the machine for the worst thing that anybody can think of, Myers said. So at the end of 2010, after we do the second half, we would be safe for the whole machine for the worst case, he added.
A key modification being made to the LHC is the so-called new Quench Protection System (nQPS), which essentially is addition of two new detectors to the existing QPS. When the protection system detects an increased potentially quench-causing resistance, the huge amount of energy stored in the magnet string is safely extracted and dumped into specially designed resistors. The first of the new detectors is designed to monitor the superconducting bus-bar resistance, including the joints between the segments, the part that caused the Sector 3-4 incident.
The new detector has a sensitivity of 0.3 mV and can measure the bus-bar resistance to about 1 nano-Ohm, which is 3,000 times better than the old system, and will provide an early warning for a potential quench. The second new detector protects against what are called symmetric quenches. When a magnet quenches, it can sometimes trigger a quench in other symmetrically located neighbouring magnets; in this case in the magnets in another aperture. It is caused by heat transfer between the magnets, but it is difficult to detect with the old QPS.
In the old QPS, voltage signals from two apertures were compared to detect a resistive build-up in either one. But if both quench simultaneously, such a comparison would give a null signal and the quench would go undetected. The nQPS monitors the voltage across four adjacent dipoles, allowing a symmetric quench to be detected and also provides a back-up detection method for detecting normal (asymmetric) quenches. It was originally planned to add this second component during the first winter shutdown period.
In Chamonix, however, it was decided to have it in place before the restart. Symmetric quenches were discovered only in June 2008 during the campaign of training quenches in Sector 5-6 (see Myers interview).
Another modification effected in the QPS is to halve the time constant for ramping down the current in the dipoles in case of a quench. Compared to 108 seconds it took during the Sector 3-4 incident, it will now be 50 seconds. This also guards against a symmetric quench because, according to Myers, the average time for the warm helium wave to arrive at the neighbouring magnet was 44 seconds. This is also one of the reasons for the decision to have the initial LHC run at 3.5 TeV with a lower dipole current, which makes this task a lot easier.
Yet another improvement in the QPS is accurate measurements remotely, which will also be automated. The new detector boards of nQPS are already in place in 436 crates around the inside of the tunnel. Interestingly, given the pressure of time for restart, instead of tendering its production to an external company, a production line was set up in a CERN workshop itself.
The central problem that led to the incident in Sector 3-4, namely poor soldering of bus-bar joints of both the superconducting cables and the copper stabilisers, has actually been bypassed in a sense by the complex techniques evolved to measure tiny resistances of the superconductor joints and the copper stabiliser joints (see Myers interview). The problem lies in the very nature of the soldering technique, which is actually a brazing technique, says a CERN scientist who does not wish to be named.
A kind of sandwich of the superconducting cable, the silver-tin solder and the surrounding copper stabiliser block is made and inductively heated so that the solder melts and propagates through. The solder is expected to flow in everywhere, but unfortunately gamma ray imaging technique has shown that it sometimes actually flows out leaving voids, or even complete absence of solder, in the splice.
This results in dangerous potentially quench-causing discontinuity in the contact between the superconducting cable and the copper stabiliser. Ideally, the scientist said, two different soldering materials, with different melting points, should have been used for the two joints. Right now, by soldering one, you actually end up melting the other.
But having actually re-soldered bad joints and obtained very good splices with the right resistances, Myers believes that the technique can be perfected. The other major unresolved problem is the loss of memory in the magnets that were trained up to 13,000 A. They now seem to quench at much lower currents. This is yet another reason for low start-up energy.
When we want to go to the full design, says Myers, two things are still outstanding: the training of the magnets and the copper stabilisers. For the training, I think we have to understand what happened and then we will try again to see if we can retrain them for all octants. For copper stabilisers, if we get this pulse current measurement technique working at 20 K, it should be okay. On the other hand, we need to gain some experience to see how these things will behave with time because there is no mechanical stability and all relies on this solder and I would like that.
COMMents
SHARE