From the start, the fourth tube for the highway tunnel under the River Elbe in Hamburg was going to be special. First, it would be bored, unlike the three existing tunnels, which were built as concrete immersed tubes in the 1960s. Disrupting shipping in the port of Hamburg to construct another immersed tube was politically and financially unacceptable.
Secondly, the new two-lane tunnel, with a third crawl/emergency lane, was going to be large. The minimum required an internal diameter of 12.4m to accommodate large freight lorries and autobahn traffic speeds. Add to this the 700mm thickness of the colossal precast concrete segmental lining and the 230mm annulus, and the minimum outer diameter had to be about 14m. “In the 1960s, the technology was not available to complete a bored tunnel under the river, where ground water pressures at the tunnel alignment are up to 5 bar,” explained Gert Wittneben, Site Manager for the Fourth Elbe Tunnel construction project. “We have this ability now and the bored option was the only option for building the new tube.”
Next, the geology is complex. The Hamburg area of Germany has endured five or six glacial advances from Scandinavia, and the erosion of the melting and freezing periods has created mixed geology, from hard clay to boulders of up to 6m in dimension.
Finally, the new tube had to be adjacent to and share the same portal structures as the immersed tube to link with the existing highway lanes on either side. With a required maximum roadway gradient of 3.7%, these specifications meant less than a tunnel diameter of cover beneath the riverbed all the way across the river.
In 1994, the contract to design, build and finance the Fourth Elbe Tunnel was advertised internationally and four proposals were received; three from German companies and one from a French-led group. Of the three German groups, two combined forces to become the Arge 4 Röhre Elbetunnel, comprising Billfinqer-Berger; Dywidag; Heitkamp; Hochtief; Philipp Holzmann; Wayss & Freytag; and Züblin. This was the successful group.
In negotiating 100% funding for the project, five support banks pay the Arge according to progress certificates signed by the client. Once the new tube is open to traffic in 2003, the Government will pay 15 annual payments to the Arge for the construction cost plus interest. The Arge will then repay the lending banks.
The Arge’s fixed price bid to design and construct the 4,403m two-lane highway, including the 2,561m bored tunnel under the river, is $421.1 million. "The final cost, at the end of the construction, is expected to be about $630.5m depending on interest rates," said Rolf Berger Project Manager of the Arge.
In preparing its proposal for the bored tunnel section, the Arge worked with TBM manufacturer Herrenknecht to procure a machine capable of meeting the difficult geological and logistical challenges. For the first 500m or so of the total 2,561m drive, the tunnel passes through reclaimed land in the harbour, which comprises sand, gravel and ungraded fill material incorporating all sorts of debris.
For the 1,000m drive under the river, the glacial material was said to be often a conglomerate of very abrasive clay and gravel with big stones frequently embedded. “The material, even the sand, is all very hard, having been compacted by the weight of the glaciers,” said Wittneben. “It is not very permeable but is very dense and there is little air loss from the tunnelling operation. But it is highly variable and we are most often in very mixed faces.”
For the last 1,000m on the north landfall of the river, soil conditions remained similar to those under the river, but the tunnel passes 12-35m under old buildings and homes. Preventing settlement in this area would be a top priority. In addition, there are two emergency cross passages of up to 70m long between the new tunnel and the existing immersed tube to be completed. These are to be excavated by a 4.4m diameter pipejacking machine with the operator controlling the excavator from behind a pressure bulkhead.
The enormous Herrenknecht TBM, designed specifically for the project, incorporates several technical innovations, of which the disc changing facility in atmospheric conditions is the most advanced.
After gaining access into the pressure-protected space of the cutterhead spokes, the discs are replaced as complete pressurised units. First, a gate valve closes off the disc unit from the surrounding ground and water pressure. A small ram is then engaged laterally to hold the unit in place while the bolts are disconnected. As the ram is slowly retracted, the material in the disc housing spills out and a small crane attachment lifts the worn disc unit out and lowers the replacement one in.
There are 30 double-ring, mono-block disc units on the cutterhead, of which eight are quipped with hammerheads. These Herrenknecht-designed hammerheads do not turn but are made of layers of hard carbide, embedded with hard carbide teeth. These have proved effective in sticky abrasive material with the occasional boulder. If there are too many boulders, these hammer discs are worn after about 10m, but they last well in sticky abrasive material.
After completing 2,000m of 14.2m diameter tunnel, 300 discs and 50 hammerheads had been changed. This was much higher than originally expected, but these changes were performed in free air. Only the scraper tools at the edges of the cutterhead slots and the tools on the small centre cutterhead had to be changed under compressed air.
The central 3m diameter cutterhead in the middle of the huge cutting wheel can operate either in tandem or completely independently of the main wheel. It has its own slurry circulation system and has a rotating speed of 2.2 rev/min. The huge 14.2m diameter outer wheel rotates at 1.4 rev/min. The central cutterhead prevents material balling up in the middle of the huge slow turning wheel and so reduces torque and thrust pressure requirements. There are 64 thrust rams on the machine, capable of providing a total thrust of 18,000 tonnes, of which some 14,000 tonnes is used for normal operation.
At the deepest point, the tunnel crown lays 11m below the riverbed, including a 1.5m blanket of copper stone. At the edges of the river, this cover reduces to 9m, including the blanket. With a maximum river depth of 15m at sea level and 17m at high tide, the saturated glacial silts and the water in the charged sand lenses are subject to the full hydrostatic pressure. Together with a margin for safety, plus pressure to support the ground itself, the working pressure of the support slurry is up to 5 bar. The TBM is tested to 5.5 bar, and compressed air work for maintenance and repair has been up to 4.5 bar. Control of the support slurry pressures across the 160m2 face of the TBM is completely automatic.
The TBM’s slurry system is a water based system drawing water directly from the river. A bentonite slurry was needed to start through the fill material but there was sufficient natural clay in the glacial deposits under the river. The separation plant, supplied by Schauenburg and Bird Humboldt of Germany, is working well and is handling some 2,500m3 of slurry/hr.
To circulate the slurry, two high performance 950kW variable-speed Warman slurry pumps from Australia are used on the discharge pipeline - one on the TBM and another at the low point of the tunnel to maintain the outbound velocity - and there are two Habergamm pumps from Germany in the separation plant feeding cleaned slurry back to the TBM.
The pressurised slurry is working well to support the face of mixed materials and reacting swiftly to balance the variable pressures across the 160m2 face. The slurry support system has also controlled settlement over the wide settlement trough to a maximum 10mm. Settlement under a residential area on the north landfall of the tunnel and under a cover of only 12m is minimal.
When TunnelTalk visited the site in early October 1999, the TBM was 2,031m into the total 2,561m drive. It had completed the under river crossing and was under the north bank landfall with only 530m to complete before breakthrough into the reception open cut ramp.
“In preparing the tunnel drive, we were concerned mostly about the high pressures, the very low cover; and the difficult ground conditions,” said Wittneben. “The size of the TBM was not a particular worry - it was getting under the river that commanded our attention.”
In choosing the machine, both Herrenknecht and the Arge team agreed on the slurry rather than the EPB option of the two soft ground pressurised TBM systems. Wittneben explained that both systems were appropriate but that their team was more experienced with the slurry system. “It provides good face support, responds more rapidly to changes in working pressures, and provides better control of settlement. The cutterhead torque is lower with a slurry system, which is important for such a large diameter wheel, and a slurry system allows easier entry into the plenum under compressed air,” he said.
This last distinction was to become of major significance. They were not expecting to need man entry into the plenum except on rare occasions and the Herrenknecht team designed a system whereby the principal cutters on the cutterhead could be changed under atmospheric conditions from within the spokes of the cutterhead. But the scrapers and discs on the central TBM cutterhead and the scraper teeth on the slots of the main wheel would have to be changed under compressed air. The potential for a blowout was a major risk. With no opportunity to increase the cover by taking the alignment deeper, it was necessary to improve the condition of the overburden as much as possible.
First, the sand layers of the riverbed for a 20m wide corridor across the river were compacted using the vibroflotation technique. Five vibrators on the rig penetrate the loose material to harder soil resistance at about 9-10m deep and compact the soils with vibration. The ungraded and saturated deposits of the fill on the south landfall side of the river bank were treated in the same way.
Next, a 1.5m thick blanket of heavy copper stone rock armour was placed on the 20m wide tunnel corridor of the riverbed. This will be removed once the tunnel is complete. If this operation had not been undertaken, they would have needed to grout an area of the riverbed to undertake man-entry to the plenum, and this was unacceptable.
Man entry into the plenum has been much more frequent than had been hoped. After a regular exchange of all the tools at 400m, the first extensive period of man entry was just after heading out under the river when all the tools on the cutterhead had to be replaced. On entering the plenum, the teams noted considerable wear and damage caused by large pieces of scrap iron and metal in the landfill at the river edge. It took seven weeks to repair the damage under compressed air pressures of 3-4 bar.
The teams also noticed the start of wear on the back side of the rotating cutterhead. The glacial material was known to be highly abrasive and the front of the cutterhead had been fitted with extra protection. However, in the plenum, the material was caking to the dividing wall and the shield can and grinding the cutting wheel with every turn. Under compressed air, protection scraper type tools were welded to the backside of the cutting wheel.
“At the deepest point of the river crossing, we could feel in the TBM operation that this wear was much more rapid and more serious than we had anticipated,” explained Wittneben. “We had to go in under 4-4.5 bar compressed air to exchange all tools on the centre cutterhead and undertake repairs. Under such pressures, we could work in the air for about 80 minutes maximum and then had to spend two hours in oxygen-assisted decompression. We spent many days on those repairs.”
Advance resumed but further inspection by ultrasonic measurement indicated that continued abrasion to the back part of the cutterhead had completely removed the repair welding and had ground the original steel structure down from 80mm thick to just 15mm! A six-week downtime was needed at this point to build in a complete second skin on to the most worn sections.
Compressed air work was undertaken only during the week when full support services were available. In general, four men entered the plenum together under compressed air. There are two manlocks and an emergency lock in the bulkhead of the main cutterhead wheel and another for access to the centre wheel.
But man-entry at any time is a serious activity undertaken only by experienced crews. Under the Elbe, the danger factor was increased significantly, first by the enormous 14.2m diameter of the tunnel, which gives a tunnel face area 160m2, and secondly, by the complex nature of the material.
In the mixed glacial conditions, the low cover is not the main problem. The main problem is the presence of both clay and large sand lenses in the face. “If the clay completely surrounds the lens,” said Wittneben, “the water in the sand has nowhere to go when pushed by the compressed air. Although the face may look stable, the water is still in the sand and reduces significantly its potential compressed air stand-up time. The sand starts to collapse without notice”
Such collapses occurred several times during man entry sessions, causing the compressed air teams to stop the compressed air work at once. The face regained stability after the chamber was refilled with slurry but the compressed air works had to be postponed. Only once did a collapse develop to the river, causing the overburden silt and the hard stones of the protective blanket to fall into the void. Each turn of the cutterhead wheel drew more of the copper stones into the plenum until the head was locked solid.
Restarting the cutterhead wheel was only possible after an umbrella of cement grout was formed above the face through the drilling and grout injection valves in the TBM’s shield skin. It then took two days of intensive slurry pumping to get first the centre wheel and then the main wheel turning to restart the advance.
During the planning stage, it was expected that a programme average of 10m/day for the big machine would be reasonable. “As it turns out, our best progress of 14m in a 24h day was achieved at the start of the job during the learning curve and in the land fill material, where a 2m stroke was taking about 60 minutes to bore,” said Wittneben. “Since moving into the glacial soils, progress has slowed dramatically. Torque is up, penetration is down and we are taking up to 5hr to complete a 2m stroke. Average progress is down to about 7m/day.”
Following the excavation stroke, it takes about 1hr 10min to build the ring. Each ring of the massive 2m wide x 700mm thick segmental lining of the tunnel comprises eight main segments and a key. The lining is reinforced with 100kg/m3 of steel and the eight main segments weigh 20 tonne each. A lining thickness of about 450mm thick would have been suitable for a tunnel this size, said Wittneben, “but we needed the extra weight here to counteract buoyancy under such shallow cover.”
Each segment has a ‘panelled’ sealing gasket, where gaskets across the 700mm thickness of the segment and between an inner and outer gasket will allow any leaks to be grouted locally. Bolts in the lining are provided to assist ring building only and these are removed after a day or two as the trailing back-up advances and the bolt pockets are filled.
The segments were manufactured by a JV of Dywidag, Hochtief and Strabag in a field casting yard using moulds supplied by CBE of France. Eight grout pumps feed annular grout through a tailskin grouting system to grout up the lining. Each pump operates independently to ensure an even fill of the annulus as the TBM advances.
In anticipation of deformation of the huge ring of lining, a support ring was included in the TBM design. This runs around the full circle of each ring of segments as it is built and must be installed into the new ring before the back-up can run forward with the next stroke. The support ring, has, however, been found to be unnecessary. Continuous survey of the tunnel lining shows deformation of the huge ring is negligible.
Only once was there significant deformation of the lining and this was over a 15m length under the river, where a pocket of unexpected swelling clay overstressed and cracked the segmental lining. As the TBM passed through this zone, it was being forced up on level and, once through, the invert segments heaved by as much as 100mm in the centre. All movement has now stopped and the cracked lining of the zone will be repaired with grout injection. The heave in the invert lining will not cause a problem as it will be covered by the road deck backfill.
When TunnelTalk visited in early October, there was no strict end date for the TBM. They had hoped originally to complete the tunnel by the end of October 1999, but with downtime and the slow penetration rates in the till, the internal end date for the TBM is now set for spring 2000.
According to contract, the new 4th tube was due to be opened in May 2003. The client, however, needs to undertake urgent repairs in the three sections of the existing 6-lane immersed tube and so needs the fourth 2-lane tube a year earlier in order to keep at least six lanes of traffic open. This gives a revised opening date for the fourth tube of May 2002.
When the TBM completes its job, it could be on its way to Russia to excavate a twin-tube road tunnel in a highway ring road around Moscow. On breakthrough in Hamburg, it will return to Herrenknecht under a negotiated buy-back arrangement and will be refurbished and fitted with a new cutterhead for the Moscow project. The Russians are also negotiating to buy the segment casting equipment.
Providing the Moscow project retains political support through the local political elections in Moscow in December 1999, the TBM could arrive in Moscow for April 2001 and start work by the following August. Negotiations to secure the experienced teams to operate and maintain the huge TBM are under way.