GENEVA--The Large Hadron Collider is a marvel of both brute-force and sophisticated engineering.
To start, look at the mostly circular cavern, 27 kilometers in circumference, that houses the accelerator. It's got an average depth of 100 meters, but in fact it's actually horizontal: its plane is tilted 1.4 percent to keep it as shallow as possible to minimize the expense of digging vertical shafts while placing the cavern in a subterranean sandstone layer.
Tidal forces from the moon cause the Earth's crust to rise about 25cm, an effect that increases the LHC's circumference by 1mm. That may not sound significant, but it must be factored into calculations.
The cavern itself is recycled from an earlier accelerator, the LEP (Large Electron-Positron) accelerator, to cut costs.
"You get the biggest tunnel you can afford. It's the one thing you can't change after the fact," said said Tom LeCompte, physics coordinator for one of the major LHC experiments, ATLAS. Bigger accelerators are desirable because the more the path of charged particles curves, the more energy they lose through what's called synchrotron radiation.
There's something of a tension between those who operate the LHC and those who are running its experiments. In this earlier stage of its real work, a lot of time is devoted to working out the kinks and understanding the LHC.
"It's a machine in its very youth," said Mirko Pojer, a physicist and the engineer in charge of LHC operations. "Most of the things are not routine."
To get protons and lead ions up to full speed, the particles travel through some of CERN's history as a center of nuclear physics research. Two previously state-of-the-art accelerators, the Proton Synchrotron that started operation in 1959, and the Super Proton Synchrotron that started in 1976, are now mere LHC stepping stones on the way to higher energies.
The LHC has two beams that circulate in opposite directions on separate paths. At four locations, the two beams collide, producing the showers of particles researchers study. The nature of the particles is inferred from characteristics such as how much energy they release when they are absorbed at the edges of the detector; other detectors track the particles as they travel, letting physicists gauge their properties by how much their path is curved by an electric field, for example.
One of the standout features that lets the LHC reach such tremendous energy is its collection of magnets, supercooled to 1.9 Kelvin, or -456.25 degrees Fahrenheit. At this temperature, the magnets and the cables that connect them are superconducting, meaning that electrical current can travel within them without losing energy.
A total of 1,232 magnetic dipoles are responsible for steering the beams around the LHC's curve, and separate devices called a radiofrequency cavity accelerate the ions and group them into bunches.
These bunches pass through each other like two swarms of very tiny bees, and sometimes the protons actually collide. So far, the LHC operators haven't filled the accelerator to capacity. Eventually each beam will consist of 2,808 bunches, each with 100 billion protons, spaced about 7 meters apart, traveling around the ring 11,245 times per second, and producing 600 million collisions per second.
Maxed out today, the LHC produces collisions of two protons with a total energy of 7 tera-electron volts, or TeV, but after a coming shutdown period to upgrade the LHC, it's planned to reach a total energy of 14TeV.
That's actually kind of a feeble amount of energy by one view--a flying mosquito has about 1TeV worth of energy. What makes it impressive is that energy is confined to the very tiny volume of a proton collision. To attain it, the LHC consumes about the same amount of electrical power as all Geneva.
Pojer was off duty on September 19, 2008, when the LHC hit its roughest patch. The LHC had done well just days earlier when researchers fired it up with its first beam. Pojer had just sat down to a congratulatory glass of champagne when a colleague called: "Run here, immediately. It's urgent."
It turns out a faulty electrical connection between two magnets had heated up, causing an explosive helium leak. The helium used to cool the LHC's magnets is so cold it's what's called a superfluid, and heating it up so it turns into a gas causes big problems with gas pressure. Some of the massive magnet structures of the accelerator shifted by a half a meter by the forces involved.
"Everybody was depressed and astonished," Pojer said of the problem. "Slowly we recovered," though, and a happier day came on March 30, 2010, when he was the engineer in charge of ramping the LHC up to 3.5TeV energy level.
"It was amazing and exciting," Pojer said, and indeed there was jubilation among the crews and researchers. Now, when the engineers in charge running LHC aren't working on refinements such as a tighter beam that increases the chances of collisions, the LHC is used for its scientific mission.
The LHC's hardware is largely located underground, but on the surface are several control rooms, each festooned with large monitors, for overseeing the operations and the experiments. But a significant part of the LHC's effective operation actually takes place far from CERN.
That's because a worldwide grid of computers is used to process the data that actually comes from the experiments. CERN maintains a primary copy of the data produced, but several tier-one partners have copies of parts, and about 130 to 150 tier-two partners keep their own copies many scientists actually use in their research.
That's not to say CERN's computers aren't central. A data center with hundreds of machines churns night and day to process the data. "It never stops, not even at Christmas," said Ian Bird, worldwide LHC computing grid project leader, with a million computing jobs in the center in flight at any given moment.
That's because LHC experiments don't lack for data. "Experiments have a knob," a way to adjust how much data is captured. Collectively, the LHC experiments produce about 15 petabytes of raw data each year that must be stored, processed, and analyzed.
Data processing is needed to sift the interesting unusual events out from the sea of background noise, to find enough of them to prove they aren't flukes, and to offer precise measurements. There was a day when individual events at particle accelerators were observed in bubble chambers or recorded on film, but computers are an essential component in the LHC's scientific investigations.