Storebaelt - the final chapters May 1995

When the DKr3.1 billion contract for the 8km long twin-tube bored tunnel under the Eastern Channel of the Storebaelt was awarded by Storebaelt A/S to the MT Group in November 1988, tunnel excavation was programmed to be completed between January 1990 and September 1991. To meet the programme, MT Group mobilised four 8.75m diameter EPB TBMs from James Howden of Scotland in association with its recent acquisition of TBM manufacturer Wirth of Germany. These would work from the four portals - two from the Halsskov landfall on Zealand and two from the island of Sprogø in the middle of the Storebaelt. Each was programmed to excavate about 4km, advancing towards a mid-tunnel junction on the Sprogø side of the nadir, or deepest part of the channel (Fig 1).

View of Sprogø and the East Bridge under construction
(Inset) Rail bridge and tunnel in red, motorway bridges in yellow

However, the catalogue of disasters and problems that befell this extremely challenging project are well known.

First, the TBMs arrived up to 10 months behind schedule and were late being commissioned. Problems continued when impurities were found in the hydraulic systems of the new TBMs and went on to include the disastrous inundation and flooding of the two tunnel headings from Sprogø Island in October 1991. Failure of the main bearing seals of the large screw conveyors of the EPBMs, when less than 200m into each of the drives, then caused five to six months delay on each machine. Add to this the nine months required to rebuild the cutterheads of the Halsskov TBMs after suffering incredible wear caused by the highly abrasive glacial till material and the inability to enter the cutterhead chamber to execute timely cutter changes due to the very high compressed air pressures required to accomplish safe man-entry.

Other mechanical problems required time consuming and expensive support methods to allow safe repair. These included the vertical nitrogen ground freezing required to repair a jammed screw conveyor discharge door on the TBM 1 Selandia from Halsskov in March 1992 and the horizontal ground freezing needed to allow safe man-entry to replace worn cutters on the Hlsskov TBM 2 Dania later in 1994. All of these events conspired against progress of the project, leading to tremendous strain in the contractual relationship between the contractor and the client and breakdown of communication between the contractor and the machine supplier.

The fire

Despite extremely desperate times, these problems were overcome and culminated in October 1994 with the breakthrough of the southern tube. But celebrations at the breakthrough were subdued, for not 25m away in the parallel northern tube, tunnelling to complete the last 58m of that tunnel was in a critical situation.

In June 1994, when only 1% of tunnel excavation was left to complete and when few could believe that anything more could plague this ill-starred project, a devastating fire broke out on the Dania TBM 2 heading from Halsskov. Fortunately, no-one was killed or seriously injured in the fire, which is believed to have been caused by the ignition of oil vapour escaping from a broken hydraulic hose on the TBM. The fire blazed for several hours, causing considerable damage to the TBM and, more seriously, nearly destroying the top segments of the leading six rings of the tunnel’s 400mm thick precast concrete segmental lining. In some places only 160mm of the crown segments remained as the barrier between the free air tunnel environment and the enormous ground and water pressures of up to nearly 5 bar acting on the outside of the tunnel. Had the fire burnt a few hours more the last few millimetres of the lining would have also spalled away and inundation would have followed inevitably. Such an inundation would have flooded the entire Halsskov tunnel system, flowing back along the main tunnel bore and through the completed hand excavated cross passages which link the two main tunnel bores.

Prince Joachim emerges from the second TBMs air lock after being the first person to walk through the completed link in the southern tunnel of the twin tube link

Retrieving this situation, when the end of all tunnel excavation was so tantalisingly close, has been a difficult test, not only technically but also for the morale of all involved on the project. To mitigate the immediate risk of flooding and to protect workers during the repair phase, a carpet of plastic material was laid over a 25m x 25m area of the seabed directly above the damaged tunnel. Next a substantial bulkhead immediately behind the TBM back-up trailing gantries was erected as quickly as possible to isolate the damaged area from the rest of the tunnel system. The bulkhead was a concrete plug of about 5m thick which was equipped with a man and materials airlock. The bulkhead was about 200m from the tunnel face, about 2km from the Halsskov portal and just west of cross passage number seven which was being completed when the fire occurred and has since been finished.

While the situation in the tunnel was being stabilised, the engineering teams from COWl-Mott, the tunnel design and construction supervision JV; MT Group, the construction contractor; and Storebaelt A/S, the client, finalised a reparation plan. During most of this period, the damaged area was under constant observation by CCTV. Air pressure was applied once the bulkhead was in place and was gradually increased to about 0.75 bar pressure.

The reparation plan entailed reconstructing the damaged area of the tunnel lining using surplus stock of cast iron rings left over from the Channel Tunnel project. The Channel Tunnel rings have been built inside the existing rings, reducing the internal diameter at this point from 7.7m to 7.2m over a distance of about 36m or 22 Storebaelt rings. This however does not adversely affect the operation of the trains. Crews of up to 12 persons at a time worked in the compressed air zone ahead of the concrete bulkhead. In case of emergency, it was calculated that workers would need six minutes to evacuate the heading and retreat into the safety of the bulkhead airlock.

Reconstruction of the damaged area began in mid-February 1995 and was completed by early March. At that time, the Fionia TBM from Sprogø, which faced the Dania TBM from Halsskov at a distance of 58m, started boring again to complete the last 35 rings of the northern tunnel. Fionia was stopped in July 1994 to prevent any possible adverse influence on the extremely vulnerable section of fire-damaged lining in the Dania heading.

On 7 April 1995, Fionia placed the last segment in the bored tunnel programme. This comes almost five years after the first segment was placed in August 1990 but is not the end of excavation. Now begins the final phase to excavate by hand the last 0.3m still existing between the two TBMs.

Frozen junctions

Originally, all four TBMs were programmed to meet within the better quality, underlying marl on the Sprogø side of the nadir where a junction could take place at normal atmospheric pressure. However, the highly abrasive and technically demanding nature of the overlying glacial till, particularly on the Halsskov side, resulted in much slower progress than anticipated. In the event, it was the two Sprogø machines which completed the passage under the nadir and progressed toward a junction on the Halsskov side of the channel.

The last segment in the northern tunnel tube was erected by the final TBM on 7 April 1995

Very slow progress by the Halsskov machines then forced a junction for both tunnels in the till, rather than the better quality marl. This in turn demanded the use of extensive ground freezing above the TBMs to create a safe environment for hand excavation of the junction.

It took four months to create this umbrella of frozen ground above the first two TBMs, Jutlandia and Selandia, to junction in the southern tunnel. The two TBMs were stopped in May 1994 and hand excavation of the final 0.3m began in September 1994. The same process was then repeated for the junction between the Fionia and Dania TBMs in the northern tunnel. After stopping these two TBMs on 7 April 1995, the final breakthrough was expected to be accomplished toward the end of the Summer in August/September 1995.

In the meantime, the components of the two southern tunnel TBMs, inside the sacrificial skins, have been cut up and removed and a cast iron lining has been erected, providing a completely finished tunnel bore in which track laying and other mechanical and electrical installation work now continues. Excavation of the last two of the 29 cross-passages between the bored tunnels is also under way.

According to a status report distributed by the Storebaelt Public Relations Department, a team which has kept the media fully informed of all events during the bedevilled development of the project, the rail component of the Storebaelt link, comprising the road/rail bridge across the Western Channel from Knudshoved on Funen to Sprogø, and the twin tube rail tunnel under the Eastern Channel from Sprogø to Halsskov on Zealand is expected to be open for traffic by the end of 1996 or early 1997. The road facility, with traffic passing over the Eastern Channel on one of the world's longest single span and most impressive high-level bridges, is expected to open a year later in 1998.

The current budget for the entire Storebaelt link, the two bridges and the tunnel, is DK 21.6 billion in January 1988 prices (approximately US$3.5 billion). This includes all costs relating to the delays incurred on the tunnel project. A settlement with the contractor in May 1994 resolved all claim situations, resulting primarily from the flood at Sprogø in 1991, increasing the original DK 3.1 billion contract to DK 4.9 billion. This however did not affect the overall cost of the project as there were sufficient reserves in the original budget. Since this settlement was binding to the end of the contract, the final cost of the tunnel for Storebaelt A/S remains unchanged despite the damaging fire in the Dania TBM heading in June 1994 and its consequences. MT Group can only look to its insurers for damages caused by the fire. For Storebaelt, the fire impacts significantly on its financial situation causing yet a further extension of the debt servicing period on borrowed money because of the extra delays to the opening of services. Danish State Railways, as the sole user of the rail tunnel facility, will actually shoulder this increased cost.

The project is financed by loans raised both in Denmark and overseas at competitive interest rates. The all-in borrowing cost on the Storebaelt debt of DK 27.6 billion was 7.2%, including exchange rate adjustments. Such interest rates are made possible by the Danish Government guarantees which underwrite all loans raised for the project. These loans will be paid back from revenue accrued by the collection of a toll charged for using the road link and the fees paid by the Danish State Railways to Storebaelt A/S for use of the rail link. Debt for the road link is expected to be repaid by 2012, and on the rail link by 2026.

Fig 1: Plan and geological section of the Eastern Channel railway tunnel crossing with location and date of major events

Thankfully work on the two bridges, despite taxing logistical and technological demands in their own right, has progressed exceptionally well. Both bridges have progressed in close accordance with schedules and are within their original budget prices. Whereas the road and rail is carried on the Western Channel bridge, the rail had to be separated from the road bridge over the Eastern Channel because the very long 1.624km free-standing span of the bridge could not carry the weight of both.

Storebaelt challenges

While almost every conceivable calamity that could befall a subsea tunnelling job has befallen Storebaelt - including inundation, the greatest risk of all - it must be also remembered that this tunnel was recognised from the start as being one of the most challenging of this century and one that would require the most advanced tunnelling technology available. Despite the succession of disasters, and in some cases directly because of them, the job has undoubtedly advanced the leading edge of tunnelling technology and the support systems used in association.

Geology and TBM selection

The design of the EPBMs specified for use in the complex geology beneath the Storebaelt was of the most demanding to date. These had to work in two distinct layers of material - 40% in the upper glacial till which contains boulders up to 3m in size, 40% in the underlying chalk marl which is highly fissured, and 20% in mixed face of both. The water content of the material is influenced by the full hydrostatic pressure of the sea above with the tunnel alignment being under about 1 to 5 bar pressure in the till and between 5 and 8 bar in the marl. The sand, silts and gravels of the till also have a high quartzite content and are extremely abrasive. Irregular pockets or lenses of sand, under the full hydrostatic head of the sea above, also added to the difficulties of working in the till. The marl is a more favourable tunnelling medium generally but the heavily fissured areas, particularly under the nadir, exposed the TBMs to the maximum 80m hydrostatic head of the Storebaelt.

To cope with these conditions,

• the EPBMs had to be capable of working in both the open and closed modes;
• the cutterheads had to incorporate both disc cutters and drag pick teeth;
• the main bearing seal was designed to resist water pressures of up to 15 bar;
• the bulkhead and other closed mode components had to withstand water and soil pressures of up to 8 bar with an adequate safety margin; and
• the tail sealing system was designed to withstand not less than 12 bar pressures.

Fig 2: Simplified section of the four 8.75m diameter EPBMs used on the subsea rail project

The four TBMs manufactured by James Howden and Wirth complied with all of these specifications. At the heart of the EPB systems was a double screw conveyor of 1.2m in diameter x 20m long with each piece weighing about 80 tonne. The tail seals of each machine comprised four rows of grease fed wire brush seals with an emergency rubber lip seal fitted forward of these.

To the front screw was added a boulder trap. The TBMs had no stone crusher. Instead, a grill of spacer bars on the face of each TBM cutterhead prevented boulders of larger than 600mm from entering the excavation chamber. Such boulders could then feed up the first part of the first screw conveyor, which is a ribbon type, but could not progress into the second part of the screw which is a central shaft auger type. They were therefore expected to fall into the stone trap. Once in the trap, a door would isolate the trap from the conveyor and allow the stones to be removed at atmospheric pressure.

Project Moses

The TBMs were built according to specification but the high abrasivity and technically-demanding nature of the overlying glacial till in particular was seriously underestimated. Rapid wear of the cutterhead tools demanded much more frequent entry into the excavation chamber on all four TBMs for inspections and tool replacement than originally anticipated. However, man-entry was not always possible when required because of the very high pressures required to execute safe entry. The very high water and soil pressures were also placing high demands on the sealing systems of the TBMs. Early in the tunnel boring programme, it became evident that the pressures acting on the TBM systems had to be reduced in some way. This led to the implementation of Project Moses, the largest sea bed dewatering scheme ever carried out in connection with a sub-aqueous tunnelling project.

Project Moses cost about DK 180 million and involved installing 43 dewatering wells of between 35m and 115m deep into the marl under the sea bed. These wells were fitted with dewatering pumps which were operated from six barges anchored in the sea channel, each barge supporting six or eight pumps. At its peak Project Moses comprised 37 wells working simultaneously, extracting 3,100m³ of water/hr from the ground of the sea bed. Over the area of influence, the system achieved the objective of reducing the ground water pressure at tunnel axis to less than 3 bar, thus reducing the demands on the pressure-withstanding components of the TBMs and allowing man-entry into the excavation chambers in areas where this would otherwise have been impossible, and under more reasonable compressed air pressures.

Unfortunately, Project Moses could not be installed in the busy traffic lanes of the nadir, and would have been ineffective in this zone in any case. However, the marl in this area proved less fissured than anticipated and the area was passed without serious incident.

Cross passages

Excavation by hand of the 29 sub-sea cross passages between the two main tunnels could be considered the most hazardous operation on the project due to the possible risk of inundation from the sea, particularly in the till sections. A team of highly experienced hand excavation tunnellers from Ireland was employed on these 17m long x 4.5m diameter connections which are lined with rings of bolted and gasketed SGI segments. In the till, excavation started with a 1.8m diameter SGI lined pilot tunnel which was subsequently broken out to full 4.5m diameter in a sequence of small controlled steps.

While Project Moses assisted cross passage excavation substantially, further local ground treatment measures were installed to ensure safety. A lateral system of vacuum dewatering was installed around the cross passage area from the main tunnel bores. Grout injection from the main tunnels also consolidated any ground loosened by the passage of the TBMs and reduced water flow through bands of sands and gravels. For cross passages in the nadir, outside the influence of Project Moses, ground dewatering was used for this excavation. Work was carried out under the protection of an emergency door which could be closed rapidly in case of threatened inundation. All but two of the cross passages were complete by mid-April 1995 and excavation, at the time of writing, had taken place without major incident.

Precast concrete segmental lining

The 1.65m wide rings of the bolted and gasketed precast concrete segmental lining comprise six segments and key. It is a tapered ring system, with minimally-sized bolt pockets for curved rather than straight bolts. The 400mm thick segments are reinforced with fully welded reinforcement cages giving a modest steel proportion of 80kg/m3 of concrete.

The Eastern Channel rail link will provide a top quality, highly durable low maintenance structure though its 100 year design life

The lining is a one-pass system and must meet a specified 100 year durability in a potentially corrosive environment. To meet this, the reinforcement cages are protected with a 200 to 400 micron fusion-bonded epoxy coating. This was applied by preheating the cages to 200-240°C and dipping them into a pit filled with epoxy powder fluidized by injected air.

Being a one-pass system, the quality of the ring build had to be within very tight tolerances. Following a long learning curve, this has been attained, but a final pass through the tunnels will be required to carry out back grouting and other water-tightening measures once the dewatering system of Project Moses has been switched off and the natural hydrostatic pressures return to normal.

Lessons to be learned

After all the trials and tribulations, the tunnelling industry must ask: What can be learned from the Storebaelt tunnel project experience?'If nothing else, the Storebaelt has confirmed yet again the inherent risk associated with tunnelling, be it a short length micro tunnel or a major subsea tunnelling project, and has emphasised the essential need to appreciate and manage these risks from the start.

It also revealed how unforgiving tunnelling is as a civil engineering discipline. A mistake made at the beginning of a tunnelling job must be carried through to the bitter end. There are few opportunities for a u-turn other than complete abandonment of the job, or substantial modification of the system once installed, which is often extremely difficult and expensive and requires enormous effort and time.

But other questions arise. For example:

• Could soil investigation have been more accurate and, if so, what different or better methods could have been used?
• How can auxiliary methods be used efficiently and to best effect to assist the chosen tunnelling system?
• Why exactly did the flood and the fire occur and how could they have been prevented?
• Why did the concrete segments spall so badly in the event of the fire?

The root causes of the troubles endured by the Storebaelt project may never be fully understood but blame certainly cannot be placed at the door of any one party.

Discussions about reasons why Storebaelt proved to be such a troubled project would have to include:

• Cross examination of all parties, and investigating the run up to various contract awards; The relationships between the client and its consulting engineer and contractor;
• The relationship between the parties of the contracting joint venture and its management of the contract;
• The relationship between the contractor, the client and the consulting engineer with the TBM manufacturer; and
• A review of the specialist tunnel workers noting in particular their appreciation of the job and their previous tunnelling experience.

The Storebaelt experience however, must not pass into history without the tunnelling industry taking stock of the messages it carries and without taking steps to redeem the reputation of bored tunnelling as a feasible option to other civil engineering alternatives for providing major transportation infrastructure.

• Client: Storebaelt A/S, a semi-autonomous state owned organisation.
• Consulting Engineer: COWI-Mott, a joint venture between COWiconsult of Denmark and Mott MacDonald of the UK.
• Contractor: MT Group, led by Monberg & Thorsen A/S of Denmark, with Campenon Bernard and SOGEA of France, Dyckerhoff & Widmann of Germany and Kiewit Construction of the USA.