There cannot be many civil engineering clients who would calmly re-schedule more than $200M worth of complex works by four weeks, especially large-scale underground caverns. Fewer would bear the cost uncomplainingly if they did.
But the authorities at CERN, the international particle physics research centre just outside Geneva in Switzerland, decided to do just that in September. Shutdown of the Large Electron Positron collider was delayed, holding up some of works now under way to expand this huge high energy physics experimental facility.
The physicists had a good excuse – they might have cracked the Secret of the Universe – or perhaps one of them. A subatomic particle, predicted for decades but never found, may finally have shown up during a last overpowered test just before final shutdown. Probably. It was the Higg’s Boson, a key component to understanding matter.
The chance to confirm one of the most important of postwar scientific discoveries by more tests, was worth the cost of delay.
Despite this scientific interruption, major underground works to replace the accelerator at CERN with a much more powerful machine, are already well under way. Three big packages of tunnel and cavern construction works are going, creating vast experimental halls underground which exceed easily the size and scale of the crossover caverns on the Channel Tunnel.
Construction began three years ago, above ground for workshops and support facilities and underground on those shafts and tunnels which are either off the line of the 100m deep LEP main tunnel or above it. Work to be done directly in the line of the accelerator tunnel, is beginning after the shutdown finally happened last month.
In the event, the delay may even be absorbed in programme, says Tim Watson, deputy group leader of CERN’s civil engineering group. The dismantling of the complicated LEP within the tunnel can be re-sequenced to hold the programme.
“We should be able to take up most of this four weeks without unduly affecting the civil engineering project,” he says. The new accelerator is due to come onstream in 2005 and civil engineering work runs to 2003.
Even so, the re-scheduling did serve to underline the “secondary” nature of the civil work, which is dwarfed by the size of the overall project, some $1.5bn in all. Most of the cost goes on the complex magnets, supercooling equipment and the giant detectors that make up the LHC, and on the “edge of knowledge” design for the instruments.
Funding is from the 20 international participants in the centre, which include 12 European founder countries including the UK, France, Switzerland and Germany, and overseas participants including the US and India.
Construction scope
The work comprises several kilometres of new injection tunnels, detector chambers and underground experiment control caverns spanning just over 35m and several large access shafts.
The two main detectors for the LHC are four-storey high assemblies of steel, lead, copper and ceramics, magnets, screens and vacuum chambers. They demand big spaces to hold them in the line of the underground accelerator. Huge and powerful computers are needed almost alongside because the sub-nuclear events are so fast that the time delay transmitting data along longer wires becomes crucial.
The new excavations are all being added to the existing 3.8m diameter circular tunnel built in the 1980s and housing the 26.7km long underground LEP, which was commissioned in 1989.
The tunnels will connect the ring to the Super Proton Synchrotron, a smaller diameter accelerator which has its own higher level tunnel. The SPS is used for other types of experiment but is also to be the “injector”, to give particles their initial velocity before they enter the main LHC. The SPS is higher in the ground than the LHC and the tunnels slope down.
Civil engineering works, mainly underground, have been divided into three main packages, now four following subdivision of the third package. One of these for divers, mainly tunnelling, works at different points on the ring and two are at specific locations where the two biggest detectors, the ATLAS and the CMS will be situated, one in France and one in Switzerland.
The CERN ring lies 100m under the glacial terrain at the foot of the Jura mountains, straddling the border between Switzerland and France. The European Centre for Nuclear Research’s administration, accommodation and main laboratories lie in the Swiss sector but most of the tunnel is in France.
Ground comprises a varying thickness of glacial moraine, mostly around 20m thick, overlying so called molasse, a varying ground of marls and sandstones that varies considerably in hardness from 3MPa to 80MPa.
The marl contains smectite which produces a fairly quick swelling in the presence of moisture. Since the overlying moraine is not only saturated but has a fairly rapidly moving groundwater in it this can be a problem and watertightness is an issue for all the structures underground. CERN equally requires complete dryness for its equipment, which is highly sensitive. The design life for that is 50 years.
Package one
The first package is for works at point one where an existing detector chamber and shaft is being extended. Here there will be the biggest chamber, 325.1m across, 42m high and 56m long with another chamber at right angles. Full details follow in an article from consultant Knight Piésold.
Package two
The Compact Muon Solenoid, the second major detector on the LHC, weighs 15,000t but fits into a slightly smaller space than ATLAS. Even so, it needs a cavern 26.5m across, 35m high and 50m long. A second cavern alongside and parallel is 18m by 80m and 20m high.
Both are easier on design than Point One since they can be conventionally excavated before their waterproofed concrete lining goes in and big 2m thick floor slabs.
Even so, they are huge structures, forming together a system with a 50m span. It is too much for a single excavation, especially in the ground at this point which has more weak marls, says Ken Cole of UK consultant Gibb, part of the design team. “And we only have 20m of cover before the water bearing moraine. We have to be careful about creating any cracks which could lead to water flow paths” he says.
The lower cover is due to an awkward situation for the caverns, although they are set deeper here than on the Point One site. But the moraine is also much thicker, filling an old valley in the molasse.
The two caverns will, therefore, be separated by a huge concrete wall, or pillar, 7m thick, 30m high and 50m long. Excavation of the cavity for this mass concrete wall began in August. A Voest Alpine roadheader is taking out the ground, which needs fairly rapid support. A Tamrock drill puts in 8m long Swellex bolts and a Putzmeister robot machine applies a shotcrete layer. “We need that quickly because of the molasse’s sensitivity to water,” says Cole.
Some four benches will take the pillar to the bottom and then it will be mass concreted in 2m deep pours before the excavation begins on the main chambers. This is mostly unreinforced though there are a number of reinforced openings to make.
Work until now has focused on difficult and delayed construction of the two big shafts needed for both the excavation and as permanent parts of the facility used for lifting in and out the huge 2,000t elements that make up the detector. These massive pieces of high precision engineering are already being delivered to an above ground experimental hall, which is part of the contract works.
One shaft, dropping 74m to the top of the main hall is 20.4m in diameter and the other is 12m. But both have had to be formed in soaking ground and must, like nearly all the structures, be completely watertight.
“They have been a problem,” says Cole. A decision was made to use ground freezing for the shafts to get them through the water bearing moraine, because a diaphragm wall was considered problematical at the plus 50m depth required through the moraine. Accuracy of panels would be hard to achieve, particularly when there are large scattered boulders which could deflect the rigs.
But ground freezing was extremely difficult and subcontractor Rodio from Italy, found itself with a number of windows left in the freeze, after circulating its chilled brine through a ring of injection holes drilled at each shaft site.
“The water here is both stratified and moving at several metres a day,” says Cole “and that means it was bringing in heat all the time. As the ground partially froze restricting the flow path, the water moved even quicker.”
In the end a new pattern of holes was added to the first circle and grout injected to slow up the water. Liquid nitrogen was also used to superfreeze parts of the ground. In all, the shafts took seven instead of three months, says Cole. First of the shafts began in April 1999 and the bigger one in July.
Excavation of the frozen ground was done with hydraulic breakers mounted on excavators and a 10t capacity skip for mucking out. The contractor has mounted two large Paolo di Nicola portal cranes across the shafts which will also be used for the main excavations.
Shafts have a Paroi Marocaine wall of in situ cast concrete rings, as primary lining. The section in the molasse uses a 250mm layer of shotcrete with rockbolts, as required. Inside that goes a waterproofing membrane and protection before a slipformed secondary lining, successfully done for the small shaft in June and just finishing in October on the bigger shaft. The lining is between 300mm and 500mm thick.
“There is a widening of the shaft diameter at the bottom with ground anchors to create a thicker concrete for a shear key,” says Cole. This takes the load from the shaft lining rather than have it transfer running down to the caverns.
As well as excavation now beginning on the pillar, a small drilling gallery and instrumentation tunnel is now under construction, with a 3m by 3m horseshoe-shaped cross-section. This tunnel runs at the roof level of the main chambers and some $700,000 of instrumentation will be installed from it to monitor the main works, with drilling for grouting as and when necessary.
Package three
The third contract, though large, is a piecemeal collection of various tunnels’ access shafts and smaller chambers, as well as surface buildings.
The major work is two tunnels linking the LHC with a smaller ring accelerator, the SPS, which pre-accelerates particles. There is one tunnel for each direction.
One of these 3.5m diameter tunnels has been subdivided from the main contract because of funding. TI8, as it is designated, has been entirely funded by the Swiss and a requirement is that only Swiss contractors and suppliers are used.
Work began in September 1998, in areas away from the still running CERN facilities. Both tunnels are driven close to the rings and then await the shutdown before breaking in and forming connection chambers.
As with all the works, two constraints make it necessary to phase operations. First is radiation which, while not severe, is present during experiments on the ring and must be respected, and the other is the sensitivity of the equipment. Engineers work to centimetres or sometimes millimetres. Physicists adjust by fractions of a micron and can only cope with a small amount of ground heave. Though there is scope to adjust the accelerator levels by millimetres, it is not great and is sensitive to vibration.
Before the drive began on the first TI2 tunnel, a new shaft was created approximately halfway along its 2.1km length. As well as serving the drive this will also be used for delivery of the long magnets that make up the accelerator.
As a result the shaft has an 18m by 12m elliptical shape which made its design analysis quite difficult, says Mike Lepper, from the UK’s Brown & Root, part of the design and supervision team. “A circular one would have been easier,” agrees Fernando Polo from Intecsa of Spain, “but the area at the top was quite limited.”
To cope with additional stresses the diaphragm wall used for the top of the 30m deep shaft was 1m thick. The wall runs through the 27m layer of water bearing moraine and toes into the marl below by 3m.
Below that is a short 1.5m length of Paroi Marocaine, a concrete ring formed by step by step excavation and concreting, which ties together the wall panels. The remainder in the dry molasse rock layer is shotcreted and anchored.
“It was intended to put in a secondary concrete lining but we may not do that if the shaft proves stable,” says Lepper. It is being monitored at present. CERN will only require the shaft for some two or three years and is not insisting on a permanent lining unless it is needed for stability.
The tunnel drive, TI2, began from here using an AM50 roadheader and heading towards the SPS where it stopped about 60m away from the existing tunnel. It was brought back to drive the other way where is has run some 600m with about 800m to go.
Lepper describes the tunnelling as “based on a NATM system”. The molasse is quite variable and “you don’t know 10m ahead what you will find,” says Lepper. This heterogeneous rock varies in strength from marls with as little as 1MPa strength to sandstones of up to 80MPa. In the worst areas it is important to get support in fairly quickly and not leave more than a metre or so exposed.
Three categories of support are used, with varying levels of bolting, shotcrete from 50mm to 150mm and steel arches, if necessary. “Though we have only used those once,” says Lepper.
On the first part of the drive the support was 50% of each of the lower categories, on the second drive the ground has needed approximately half each of type two and type three support.
The tunnel has made good progress and is several weeks ahead, doing about 6m a day with a 56m maximum for one 35-hour week. In October it slowed up to allow test bores ahead above the roof. The drive was sitting under a paleo valley and there was the danger that moraine water might find a channel down through the reduced cover.
The other major drive has not gone so well. Because the shaft position here was near one end, 200m from the SPS ring, the Swiss contractor decided to use a TBM, a reconditioned Robbins machine. It could make one 2,200m push towards the LEP and breakthrough after an additional 60m could be finished with a roadheader, as could a shorter 200m length to the SPS.
But progress has been slow. “Though it is a faster machine I think perhaps a TBM has less flexibility in this kind of ground,” says Lepper. “Shotcrete could only go in 35m back from the face which meant a long pause although bolts could be installed close up to the face.”
A conveyor system was used for mucking out, the horizontal from Rowa and the vertical from Frei up the 45m deep shaft.
A rockfall early on lost a couple of weeks and then, perhaps through caution, progress was very slow after that. On its original schedule the permanent lining was due to start in February, but it is just getting under way. Re-scheduling means the contract is now three weeks behind.
Lining will be done with three sets of 10m collapsible steel shutters from Swiss firm Rauh, which should allow a rapid 30m a day.
Two more short 350m lengths of tunnel and some 8m wide chambers must be made elsewhere on the ring under the 3a contract; these are “beam dumps” filled with lead blocks to absorb the high energies of the beams when they are switched off. Work on these can only take place out of an existing chamber and is yet to start.
Timings for that get switched around, says Lepper. The designers have to be continually flexible to accommodate changes wanted because of advances in physics. For that reason, CERN has negotiated transfer of some work onto a target cost basis, although it remains within the FIDIC type of framework CERN’s civil engineers have the final say on engineering matters.
“We use an open book mechanism and share any gains between the client and the contractor.” adds Lepper.
He says, personally, he is very sold on the partnering approach to construction and thinks that the number of changes required here would lead to confrontation under the older system.
Related Files
CERN housing LHC
Cross section of the the CMS chambers
Three dimensional diagram
Plan view of the observation tunnel over the chambers at the CMS site
Ring tunnels at CERN
