Between 1998 and 2005 a total of almost 250,000 m3 of soil and rock was excavated from the CMS site at “Point 5” of the LHC, in Cessy. The contractors - Seli, an Italian company and Dragados, Spanish - had previously worked on the Madrid metro but here faced some unique surprises.

It started with an unexpected find: “We did trial pits around the site, because the archaeology of an area is always something we have to consider, and what we found was great – a Roman villa, complete with pots, tiles and coins,” explains John Osborne, project manager of CMS civil engineering. “And even better for us was that it wasn’t anywhere near any planned shafts or buildings.”

They were able to leave archaeologists to spend a year thoroughly investigating the site whilst the project got underway. But it wasn’t an easy task. “Funnily enough Point 5 was one of he worst places possible to build the cavern in terms of geology, though best from a physics point of view. And CMS was pretty much the only experiment that could cope with being built here,” remarks John.

The first challenge came in the form of underground water tables approximately ten to twenty metres below the surface. To dig down through them the team had to first freeze the ground around the shafts to act as a barrier to the water. By drilling holes around each shaft’s circumference and pumping down brine, cooled to -25oC, the water froze into a 3-metre wall of ice. But the water coming from Cessy was moving even faster than predicted, and combined with the channelling effect of water between the two shafts, pressure built up until it penetrated the walls.

Injecting the holes with an even colder substance: liquid nitrogen, at less than at -195oC, finally solved the problem. With its help engineers formed a wall of ice around the shafts that was solid-enough for teams to keep on digging.

However, the problems were not over yet. Once the shafts were dug out, work had to begin on the caverns, but because the ground materials were so soft, with no intervention any cavern excavated would collapse. “At Point 5 there is only 15 metres of solid rock. For the first 75 metres of digging down it’s just a type of glacial deposit called moraine - basically a mixture of sand and gravel,” explains John. “And the rock we did have was a kind of soft sandstone called Molasse. A large cavern built in this would just collapse.”

The solution was to build a large supporting structure underground that could hold up the caverns and withstand the mass of the soil above it. Engineers envisaged a concrete pillar as the divider between the two caverns that could do just this.

“Knowing we would have to build this structure anyway, we asked radiation teams how thick we’d need it to be to protect people from radiation in the cavern next-door to the experiment,” explains John. “They gave the figure at 7 metres. The width needed to ensure adequate support for the caverns was 7.2 metres, so this worked out very well, and in fact the second cavern can now be safely used when the machine is on.”

After these delays, engineers experimented with using explosives to clear away rock at a faster rate, but in the end it had little benefit for a big disruption. Instead the best trick seemed to be to excavate small sections at a time and immediately install “shotcrete”, a sprayed concrete that sets as soon as it hits the walls, and drilling steel anchors that reached 12 metres into surrounding rock: “If we didn’t do this, the whole thing would collapse whilst we were building it. We constantly monitored any movement with a host of instruments and adjusted the support as necessary.”

For the workers on the site, another exciting yet daunting nature of working on the excavation was descending into the CMS cavern in the lift. During the construction period, access underground was via a lift cage being lowered down the shaft on a system of ropes. “This was good as it meant you could always be lowered to exactly the right level, as opposed to fixed lifts,” explains John. “But being on a rope also meant that it had a tendency to sway as you went down. And 100 metres is a long way!”

As the environment 100 metres below the surface is also full of water, the final stages involved waterproofing, installing drainage systems and painting the cavern, as water could otherwise turn soft rock into mud. “Once everything was in place we could seal off the cavern with waterproofing and put in a permanent concrete wall up to 4 metres thick, reinforced with steel bars.”

Throughout the whole construction, consideration for the environment was at the forefront of people’s minds. In order to keep noise levels to a minimum, sound barriers were built all around the site. And instead of removing the many tonnes of earth and rock excavated from the site, causing noise and road disruption, it was deposited right by the buildings, covered with fresh soil and planted with vegetation, creating a now flourishing artificial hill.

Whilst this was going on, thanks to the “pre-packaged” design of CMS, split up into 15 slices, where each is assembled and as near to complete as possible on the surface, work on building the detector could carry on this whole time. And once the cavern was complete, in 2005, pieces of CMS could be lowered underground and installed.

But though the civil engineering stage of building CMS is long over, the underground nature of the experiment must always be taken into account. Though the detector weighs almost as much as the soil and rock it replaced, the caverns within the spongy earth are like bubbles in water, and they could potentially rise by as much as one millimetre (mm) per year. Geologists predict that CMS, along with the immediately surrounding sections of the LHC machine, will rise 15 mm over 10 years. This may seem small, but when you think that the detector’s tracker and muon systems, for instance, must align to within 0.15 mm, one hundredth of the distance, this small amount becomes significant. In fact the ability to adjust by this amount is built into the detector and movement will be monitored throughout the life of the experiment, so civil engineering continues to play a part in the running of CMS.