For Charles Bassford, Miller contracts manager for the 8.7km long tunnel on Yorkshire Water's £200M Humbercare project, 16 November 1999 is a memorable date. Events on this day are etched into his memory - as the day that the tunnel collapsed (NCE 18 November 1999). Bassford was called early in the morning to hear that water and silt was spurting through a joint in the precast concrete lining of the tunnel near to shaft T3 - some 180m behind the Lovat earth pressure balance tunnel boring machine. He says: 'I couldn't understand what they were telling me - how could a failure happen there, two weeks after the TBM had passed?' The leak soon turned into an inrush, which caused a 100m section of the £70M tunnel to collapse.
Leaks were first noted by a fitter walking along the tunnel on the previous day, who saw water squirting through the right hand knee of the first joint in the lining east of shaft T3. A few hours later as he walked back, he noticed it had got worse.
By 12.30am, the joint was leaking all around the tunnel periphery. A loco driver got soaked to the skin as he passed through on his way in to the TBM, and reported this to the pit boss. Two hours later the tunnel was flooded to a depth of 2.5m, and a number of lining segments were becoming displaced and beginning to crack. The decision was then made to abandon the TBM and evacuate the tunnel.
By 6am, a settlement trough up to 2.4m deep and 60m long had appeared in a car park on the surface above. To stabilise the flooding and prevent further collapse, Miller decided to put the tunnel under two atmospheres of compressed air. Work began immediately to backfill two shafts to provide an airtight seal to the tunnel.
So did the investigation in to what had gone wrong.
Observations from the bottom of T3 suggest that the section of collapsed tunnel have sunk by up to 2m, and in some areas the lining may still be intact. The Miller team has established that fine windblown sand from below the tunnel invert to the west of the shaft washed into the tunnel - causing it to sink.
The epicentre of the collapse is at the point where the leak was first noticed, some 4m to the east of the shaft. From this point the collapse extends 70m to the east and 30m to the west.
Re-excavation of the tunnel is expected to provide vital clues to the reason for the failure. It occurred in a 400m wide alluvial channel where ground conditions varied slightly from those encountered in other sections of the project.
To get below the marina and piles near the centre of Hull the tunnel was driven at a depth of 20m through glacial clays and gravels. In the area of the collapse this material was underlain by the layer of fine windblown sand and overlain by a 1m layer of fibrous peat at the crown of the tunnel (see diagram).
Tunnellers noticed the peat material in the TBM spoil when passing through the alluvial channel. Investigations during the re-excavation will concentrate on the quantity and compressibility of the peat material present, although initial testing has revealed that the material had relatively high shear strength of 150kPa.
Ground conditions in the alluvial channel are complicated by the tunnel drive being effectively sandwiched between two separate groundwater systems. The sands and gravels underlying the tunnel are more directly connected to the tidal effects of the river Humber - which has a 6m tidal range - than the alluvium above.
This differential effect may have caused an uplift force on the tunnel, causing cumulative displacements of the lining over the three weeks after construction. Vibrations from the construction railway inside the tunnel may have exacerbated the problem. These movements could have eventually caused the lining to breach.
The situation is further complicated by the breach occurring at the first flexible lining joint to the east of the shaft, suggesting possible interaction between the rigid 12.3m diameter shaft - which was unaffected by the collapse - and the 'flexible' tunnel.
Investigators will also concentrate on examining the concrete lining segments as they are extracted from the collapsed tunnel. Each ring of the lining consisted of six trapezoidal concrete segments, which were erected on to the tailskin of the TBM. After pulling the tailskin forward to expose the rings to the ground, the 65mm annular void behind the rings was grouted to 0.5bar above hydrostatic pressure, to provide intimate contact with the ground and a watertight seal.
Each segment was also fitted with a gasket, which allowed up to 20mm of segment movement.
Investigations will centre on the possibility that the ring that initially leaked may have been adjacent to a void in the peat that was not fully grouted, although rings uncovered so far have all showed no problems.
The Miller team is waiting to hear the results of physical modelling at Cambridge University using a geotechnical centrifuge.
The tests are attempting to reproduce the collapse by modelling the interaction between a rigid shaft and flexible tunnel, with a layer of compressible material above the tunnel.
Since the fateful events of 16 November the compressed air has come off to be replaced by water, and subsequently ground freezing. The first stage in the recovery operation was to reexcavate the backfilled T3 shaft that - with the water table at 2m - was originally constructed underwater. Vertical freezing tubes were also drilled down in to the flooded tunnel to produce an ice wall, to plug the flooded section. Liquid nitrogen was used to provide a more intense freeze than the conventional brine.
Re-excavation of the shaft is complete and work has begun re-excavating the collapsed tunnel to the west of the shaft in an attempt to reach the TBM, nicknamed 'Maureen', which has now been trapped for nine months.
Horizontal ground freezing tubes, 20m long, have been installed from the base of the shaft, and a 1.5m thick ring of frozen ground has been created around the original tunnel drive.
Miners are now digging into the frozen ground by hand, and applying a 350mm thick sprayed concrete lining. Their work is complicated by having to dig through debris from the collapsed tunnel, including lining segments and a locomotive.
Tunnel re-excavation will be conducted in 20m long sections, and two layers of sprayed concrete lining, with a total thickness of 600mm, will be applied before the freezing comes off.
Heated water and aggregate, along with an insulating layer of polyethylene foam will be used to ensure that the shotcrete gains strength normally, despite the ice.
Ground freezing and re-excavation of the collapsed tunnel by hand and roadheader is expected to extend into the spring, and will include recovery of the trapped TBM.
Bassford is confident that Maureen will be still serviceable, although it may need an electrical refit. During the nine months since the collapse the remaining tunnelling has been completed by a replacement TBM, nicknamed Gloria, and Maureen will only need to drive 50m to a new reception shaft to the west of the marina.
The combination of physical investigations and laboratory modelling will provide enough clues to piece together the mechanism for the collapse, Bassford believes. But his team is still bewildered by the disaster.
'It's unprecedented, ' he says.
'Some of my tunnellers have over 20 years of experience and never heard of anything like it.'
This sentiment is shared by Steve Tindall, Yorkshire Water project manager for the scheme, although he concedes that in some ways they have been lucky.
The surface settlement, was completely contained in the carpark, at Blanket Row in Hull, and has not had any effect on the surrounding buildings. 'It couldn't have happened in a better place, ' says Tindall.
Why read this:
Major tunnel collapse
Ground freezing using liquid nitrogen