When particle physicists talk about accuracy they are thinking microns.
Just outside Geneva at the CERN centre for nuclear research, equipment used in experiments is tuned to hair's breadth tolerances. Experiments themselves are conducted at a sub-atomic scale.
So CERN's scientific community was dismayed when tunnelling work being carried out next to the existing large electron positron collider (LEP) ring, part of a ú7.15bn upgrade now under way caused it to deflect.
Carefully aligned experiments were thrown out by millimetres.
'There is one holy law: Do not disturb experiments, ' summarises CERN liaison engineer Johanna Rammer.
The work is being carried out to construct a new large hadron collider (LHC) (see box). Engineers responsible for the minute but critical ground movement were ordered to predict and forewarn CERN about the effects of tunnelling work on surrounding ground and the LEP ring. The Swiss Fr345M ($205M) worth of civil engineering work has, in fact, been entirely geared around continuing experimentation.
Three construction packages were let in spring 1998 while the LEP ring was in full operation.
Lead consultants for package one are Electricite de France and Knight Piesold; Gibb, SGI Ingenierie and Geoconsult are on package two, while Brown & Root and Intecsa are designers for package three. All civil engineering must be complete and ready for hand over by the end of 2002.
Package one tunnelling works are most advanced. Excavation of the shafts, the main cavern vault and adjoining experimental hall began while the LEP tunnel was still in full operation - the ring was only shut down in December last year.
Here the accelerator tunnel will ultimately run through the main detector hall - the two structures are aligned. 'For the client, the danger was that we would excavate straight down to the level of the LEP tunnel, ' recalls Steve Heard, project manager with package one consulting joint venture.
'There is not enough time to excavate the cavern and then concrete the invert, walls and crown. We have to get the roof in now and then continue excavation to remain on schedule, ' Heard explains.
Just a few metres above the live LEP tunnel, the first bench of the detector hall was excavated - a space 35m wide, 56m long and 12m of a total 42m high. Ground across the CERN site is soft molasse - a combination of sedimentary and tertiary materials including sandstones, laminated marls, clays, gravels and sand.
Strength of materials ranges from 2-3MPA to 50MPA, while layer thickness varies between a few centimetres and several metres. The molasse is prone to swelling, heave and rapid decay in contact with air and water.
'The client could accept excavation of the crown and the heave that would produce. If CERN knew what to expect it could adjust equipment, but it could not deal with disruption, ' says Heard. EDF/KP has been predicting ground movement to the nearest millimetre.
CERN has also called for early delivery of the experimental hall, adjoining the detector cavern. The 23m wide, 65m long and 18m high space has been fully excavated. Final concrete lining is now taking place.
A 200mm thick fibre reinforced shotcrete primary support layer was applied within 12 hours of excavation to prevent deterioration and resist squeeze.
Movement is closely monitored.
Final reinforced concrete linings are slipformed over a geotextile drainage layer and a 2mm thick, heat welded, polythene waterproof membrane, using bespoke shuttering supported by a rail-mounted falsework 'horse'. To resist long term swelling pressures, the linings are between 500mm and 1.3m thick, with as much as 10% of wall mass made up of rebar.
Heard says: 'From an ideal design point of view, we would have excavated both chambers and let the ground move. The hard structure attracts forces like nobody's business - displacement and convergence forces mean earth wants to fill the hole. The structure has to be strong enough to resist inward forces from the surrounding ground.' As a result EDF/KP has 'hugely increased' the tonnage of reinforcing adjacent to the main cavern.
Meanwhile, the main detector hall invert, a flat slab, will be a massive 5m thick, balancing out swelling pressure, Heard adds.
The most significant challenge facing package one is construction of the main detector hall roof. Falsework and formwork are now in place and the vault is due to be cast in seven pours.
However, once complete, 30m of additional excavation is required to bring the cavern down to its final depth. Unusual temporary works must be carried out to support the huge vault slab until secondary, structural linings have been cast in the invert and walls.
Part of the vault's load will be taken by the two shaft linings, locked into surrounding ground by 5m deep, key-profile, C100 concrete shear collars. A total of 38, 1,800kN, 13 strand Freyssinet anchors will be installed to carry the remainder.
Contractor CCC, a joint venture between Swiss firm Zschokke Locher, Baresel of Germany and Austrian tunnelling specialist Asdag Tunnelbau, has excavated two 12m long suspension galleries off each of the main hall shafts, just above a lens of hard, supportive sandstone. Locked into the vault roof with specially designed concrete 'plugs', the cables will be tensioned first to 25% before being taken to full load. Stressing of anchors is due this autumn.
On package two, a similarly striking strategy is being executed to tackle swelling and deformation.
Here, the two halls are larger than on package one and lie one alongside the other, separated by just 7m of rock. Analysis of the marl's load bearing capacity showed this pillar was inadequate, posing a danger of collapse during excavation of the chambers.
Consulting joint venture Gibb/ SGI Ingenierie/Geoconsult and contractor Dragos-Seli joint venture is now preparing for work on the main caverns by excavating the pillar itself. The 30m tall, 54m long cavity will be back filled with C40 reinforced concrete, providing stable roof support for the structures either side.
'There will be heavy reinforcing in the top of the pillar, ' says Gibb resident engineer Angus Spiers. 'The weight of ground on to the pillar will impose high shear loads, like a wedge trying to split it apart.' Concreting will take three to four months.
Excavation in the pillar cavity, as on all bar one of the other CERN tunnelling projects, is by hydraulic breaker and road header. The contractor has completed four benches of 4m to 5m so far with two to go. Walls are supported using 25mm diameter 6m to 12m long rock bolts at 1.5m centres, and 200mm of fibre reinforced shotcrete. Naturally occurring hydrocarbons have caused problems in getting shotcrete to adhere to excavated walls, notes Spiers. Mesh provides a key.
Cavern excavation will start in the second half of this year, says Geoconsult principal engineer Roland Mayr. As on package one, massive floor slabs will resist squeeze in the cavern inverts.
Dense bolting will limit swelling in the crown.