Knocking on Heaven's Door (25 page)

After the budget history lowdown, Lyn’s talk turned to more happy news. He described the first
test string
of dipoles—a test of magnets combined together in a workable configuration—that took place in December 1998. The successful completion of this test demonstrated the viability and coordination of several of the ultimate LHC components and was a critical milestone in its development.

In 2000, when LEP, the electron-positron collider, had run its course, it was dismantled to pave the way for LHC installation. Yet even though the LHC was ultimately built in a preexisting tunnel and used some of the staff, facilities, and infrastructure that were already in place, a lot of man-hours and resources would be necessary before the transformation from LEP to the LHC could occur.

The five phases of the LHC’s development included civil engineering to build caverns and structures for experiments, the installation of general services so that everything could run, the insertion of a cryogenic line to keep the accelerator cold, putting in place all the machine elements including the dipoles and all the associated connections and cables, and ultimately the commissioning of all the hardware to make sure everything worked as anticipated.

The CERN planners started off with a careful schedule to coordinate these construction phases. But as everyone knows, “the best laid plans o’ mice an’ men gang aft agley.” Needless to say, this applied all too well.

Budget issues were a constant nuisance. I remember the frustration and concern of the particle physics community in 2001 as we waited to find out how quickly some serious budget problems at the time could be resolved to allow construction to proceed. CERN management dealt with the cost overruns, but at a price in terms of CERN breadth and infrastructure.

Even after these funding and budget problems were resolved, LHC development still wasn’t entirely smooth sailing. Lyn in his talk described how a series of unforeseen events periodically slowed down construction.

Certainly no one involved in excavating the cavern for the CMS (Compact Muon Solenoid) experiment could have foreseen digging into a fourth-century Gallo-Roman villa. The property boundaries were parallel to the farm field boundaries that exist to this day. Excavation was halted while archaeologists studied buried treasure, including some coins from Ostia, Lyon, and London (Ostium, Lugdunum, and Londinium at the time the villa was occupied). Apparently the Romans were better at establishing a common currency than modern Europe, where the euro still hasn’t displaced the British pound and the Swiss franc as a means of exchange—particularly annoying for British physicists arriving at CERN who don’t have the currency required to pay for a taxi.

Compared to CMS’s travails, the 2001 excavation of the ATLAS cavern proceeded relatively uneventfully. Digging the cavern involved removing 300,000 metric tons of rock. The only problem they faced is that once the material was removed, the cavern floor began to rise slightly—at the rate of about a millimeter each year. This might not sound like much, but the movement could in principle interfere with the precise alignment of the detector pieces. So the engineers needed to install sensitive metrology instruments. They are so effective that they not only detect ATLAS movements, but are sufficiently sensitive to have registered the 2004 tsunami and the Sumatra earthquake that triggered it, as well as others that came later.

The procedure for building the ATLAS experiment deep underground was rather impressive. The roof was cast on the surface and suspended by cables while the walls were built up from below until the vault could sit on them. In 2003, the completed excavation was inaugurated with a celebration, notable for the presence of an alpine horn echoing inside, which in Lyn’s description was a source of great amusement. Installation and assembly of the experimental apparatus subsequently followed with the components lowered one by one until ultimately the ATLAS experiment was assembled with this “ship in a bottle method” in the excavated cavern belowground.

CMS preparations, on the other hand, continued to face rough seas. It once again got into trouble during excavation since it turned out that the CMS site was infelicitously placed not only over a rare archaeological site but also over an underground river. With the heavy rains that year, the engineers and physicists discovered to their surprise that the 70-meter-long cylinder they inserted into the ground to transport materials down had sunk 30 centimeters. To deal with this unfortunate hindrance, the excavators created walls of ice along the cylinder walls to freeze the ground and stabilize the region. Supporting structures to stabilize the fragile rock around the cavern also had to be installed, including screws up to 40 meters in length. Not surprisingly, the CMS excavation took longer than foreseen.

The only saving grace was that because of CMS’s relatively compact size, experimenters and engineers had already been considering constructing and assembling it on the surface. Constructing and installing components is a lot easier aboveground, and everything is faster since there is more room to work in parallel. This aboveground construction had the added critical benefit that the cavern problems wouldn’t further delay construction.

However, as you might imagine, it was a rather daunting prospect to lower this enormous apparatus—which is something I had a chance to think about when I first visited CMS in 2007. Indeed, lowering the experiment was no easy task. The largest piece began its 100-meter descent into the CMS pit, carried by a special crane, at the dauntingly low speed of 10 meters per hour. Since there was only a 10-centimeter leeway between the experiment and walls of the shaft, this slow descent and a careful monitoring system were critical. Fifteen large pieces of detector were lowered between November 2006 and January 2008—a brazen piece of timing as the final piece was delivered pretty close to the scheduled LHC start-up date.

Following the CMS water trouble, the next crisis in the construction of the LHC machine itself struck in June 2004, when problems were discovered in the helium distribution line known as the QRL. The CERN engineers who investigated discovered the French firm that had taken on this construction project had replaced the material designated in the original design with what Lyn described as a “five-dollar spacer.” The replacement material cracked, allowing thermal contraction of the inner pipes. This faulty component wasn’t unique, and all the connections had to be checked.

By this time the cryogenic line had been partially installed and many other pieces had already been produced. To avoid blocking the supply chain and introducing further delays, the CERN engineers decided to repair what had already been produced while leaving industry to correct the problem before delivering the remaining parts. CERN’s factory operations and the need to move and reinstall large pieces of the machine cost the LHC a year delay. At least the delay was far less than the decade delay Lyn and others feared had lawyers been involved.

Without pipes and the cryogenic system, no one could install magnets. So 1,000 magnets sat around in the CERN parking lot. Even with the high-end BMWs and Mercedes that grace the lot at times, $1 billion worth of magnets probably exceeded the usual parking lot contents’ net worth. No one stole the valuable magnets, but a parking lot isn’t a great place to store technology, and further delays associated with restoring the magnets to their initial specification were inevitable.

In 2005, yet another near crisis occurred, having to do with the inner triplet constructed at Fermilab in the United States and in Japan. The inner triplet provides the final focusing of the proton beams before they collide. It combines three quadrupole magnets with cryogenic and power distribution—hence the name. This inner triplet failed during pressure tests. Although the failure was an embarrassment and an annoying delay, the engineers could fix it in the tunnel so the time cost wasn’t too severe in the end.

Overall, the year 2005 was more successful than its predecessor. The CMS cavern was inaugurated in February, though no horn graced the day. Another landmark event occurred in February—the lowering of the first cryodipole magnet. Magnet construction had been critical to the LHC enterprise. A close collaboration between CERN and commercial industry facilitated their timely and economical construction. Though designed at CERN, the magnets were produced at companies in France, Germany, and Italy. Initially, CERN engineers, physicists, and technicians placed an order for 30 dipoles in 2000, which they might then carefully examine to ensure quality and cost control before placing the final order for more than 1,000 magnets in 2002. CERN nonetheless maintained responsibility for procuring the main components and raw materials in order to maximize quality and uniformity and minimize cost. To do so, CERN moved 120,000 metric tons of material within Europe, employing an average of 10 big trucks a day for four years. And that was only one piece of the LHC effort.

After delivery, the magnets were all tested and carefully lowered through a vertical shaft into the tunnel near the Jura Mountains that overlook the CERN site. From there, a special vehicle transported them to their destination along the tunnel. Because these magnets are enormous and only a few centimeters separated the wall of the tunnel from the LHC installations, the vehicle was automatically guided by an optically detected line painted on the floor. The vehicle moved forward at a rate of only about a mile an hour in order to limit vibrations. That meant it took seven hours to get a dipole from the lowering point to the opposite end of the ring.

In 2006, after five years of construction, the last of the 1,232 dipole magnets was delivered. In 2007, the big news was the last lowering of a cryodipole and the first successful cooldown of a 3.3-km-long section to the design temperature of—271 degrees Celsius—which allowed the whole thing to be powered up for the first time, with several thousand amps circulating in the superconducting magnets in this section of the tunnel. As often happens at CERN, a champagne celebration marked the occasion.

In 2006, after five years of construction, the last of the 1,232 dipole magnets was delivered. In 2007, the big news was the last lowering of a cryodipole and the first successful cooldown of a 3.3-km-long section to the design temperature of—271 degrees Celsius—which allowed the whole thing to be powered up for the first time, with several thousand amps circulating in the superconducting magnets in this section of the tunnel. As often happens at CERN, a champagne celebration marked the occasion.

A continuous cryostat section was closed in November 2007 and everything was looking pretty good until yet another near disaster struck, this time involving the so-called plug-in modules, known as
PIMs.
In the United States, we didn’t necessarily follow all the reports about the LHC. But news spread about this one. A CERN colleague told me about the worry that not only had this piece failed, but it could be a ubiquitous problem all around the ring.

The problem is the almost 300-degree differential between a room-temperature LHC and a cool operating one. This difference has an enormous impact on the materials with which it is constructed. Metal parts shrink when cooled and expand when warmed. The dipoles themselves shrink by a few centimeters during the cooldown phase. This might not sound like much for a 15-meter object, but the coils must be accurately positioned to within a tenth of a millimeter to maintain the intense uniform magnetic field required to properly guide the proton beams.

To accommodate the change, dipoles are designed with special
fingers
that straighten out to ensure electrical continuity when the machine is cooled down and that slide back when warmed. However, due to faulty rivets, the fingers collapsed instead of recessing. Worse yet, every interconnection was subject to this failure, and it wasn’t clear which ones were problematic. The challenge was to identify and fix each faulty rivet—without introducing a huge delay.

In a tribute to the ingenuity of the CERN engineers, they found a simple method of exploiting the existing electrical pickup located every 53 meters along the beam that was initially installed so that the electronics would be triggered by the beam passage. The engineers installed an oscillator into an object about the size of a Ping-Pong ball, which they could send around the tunnel along the path a beam would take. Each sector was three kilometers long and the ball could blow through, triggering the electronics each time it passed a pickup. When the electronics didn’t record a passage, the ball had hit the fingers. The engineers could then go in and fix the problem without having to open every single interconnect along the beam. As one LHC physicist joked, the first LHC collisions were not between protons, but between a Ping-Pong ball and a collapsed finger.

After this last resolution, the LHC seemed to be on track. Once all the hardware was in place, its operation could begin. In 2008, many human fingers crossed when at long last the first test took place.

SEPTEMBER 2008: THE FIRST TESTS

The LHC forms proton beams and after a series of energy boosts injects them into the final circular accelerator. It then sends those beams around the tunnel so that they return to their precise initial position, allowing the protons to circulate many times before being periodically diverted to collide with great efficiency. Each of these steps needs to be tested in turn.

The first milestone was to check whether the beams would actually circulate around the ring. And they could. Amazingly, after its long history of trials and tribulations, in September 2008, CERN fired up its two proton beams with so few hitches that the results exceeded expectations. On that day, for the first time, two proton beams in succession traversed the enormous tunnel in opposite directions. This single step involved commissioning the injection elements, starting the controls and instruments, checking that the magnetic field would keep the protons in the ring, and making sure all the magnets worked to spec and could run stimultaneously. The first time this sequence of events was ready was the evening of September 9. Yet everything worked as well as or better than planned when the tests took place the next day.