Anacostia challenges America's first EPBM Mar 1987

No one expected it to be easy. Providing a link across the Anacostia River for the new Green Line of the 165km Washington DC Metro could prove to be the most difficult undertaking for the system so far (Fig 1).

Fig 1. Route of the WMATA Green Line under the Anacostia River

There are already four river crossings in the system. To the west, the Orange/Blue Lines cross the Potomac River in a twin tube tunnel completed by drill+blast through competent dry rock in the mid-1970s. To the east, the same two lines cross the Anacostia River on a low level bridge. On the Yellow Line, a steel/concrete immersed tube crosses the Washington Canal between L'Enfant Plaza and the Pentagon station, and crosses over the Potomac by bridge.

The master plan takes the Green Line across the Anacostia about 2 miles (3.2km) upstream of its confluence with the Potomac where the Anacostia is some 1,500ft (457m) wide and between 30ft and 40ft (9m to 12m) deep (Fig 2). A bridge was readily rejected by WMATA (the Washington Metropolitan Area Transit Authority), the client, and its design engineer for this section, the Parsons Brinckerhoff/Sverdrup & Parcel JV. Mid-sized ships of the US Navy use this stretch of the river to approach naval yards in the area and to provide adequate clearance, the ramps of a bridge would be too long and steep to be incorporated in the subway system.

The decision between an immersed tube and a bored tunnel was less easy and after lengthy discussions, the immersed tube was eliminated on economic grounds. The only facility capable of fabricating the steel tubes locally had closed down and transportation costs from factories further afield were prohibitive. Also, the muck dredged from the riverbed would require special handling and disposal at extra cost.

In August 1984, WMATA invited tenders for two 2,500ft (770m) long x 18ft 10in (5.8m) o.d. tunnels between two 75ft wide x 62ft long (23m x 19m) construction shafts either side of the Anacostia. The launch shaft on the south bank of the river was 67ft (20.5m) deep and supported by steel soldier piles and timber lagging. The reception shaft on the north bank was 105ft (32m) deep and supported entirely with slurry diaphragm walls.

Fig 2. Four river crossings on the Metro system to date

The Engineer's estimate for this work was US$38 million. At US$25.55 million, Harrison Western, as leader of a JV with Franki Denys (FDI), the US subsidiary of the Belgian companies Franki and Denys, returned the lowest of ten competing bids and was given notice to proceed in January 1985.

Geological conditions

Not being particularly fast flowing along this reach, the river bed is a deep layer of saturated sands and alluvial deposits with pockets of dense, cohesive clay containing occasional large boulders (Fig 3). Of the 2,500ft long tunnel, 1,500ft (492m) would be under water with a maximum 90ft (30m) between the surface of the water and the crown of the tunnel. This equals a maximum hydrostatic head of 40psi (2.7 bar) at the tunnel invert.

Based on data from several boreholes and by incorporating two 1,400ft (427m) radius horizontal curves and two 400ft long vertical curves, the tunnel was placed, within gradient tolerances, in the most favourable geology. This was particularly important as there were fears of channels in the riverbed that would provide direct links to the river.

Compressed air, EPBM or pressurised face shields were the tunnelling options specified by the contract and, although still relatively new technology, there were several large diameter soft ground tunnelling techniques from which Harrison Western could choose.

Fig 3. Geological section of the river crossing for which compressed air, EPB or pressurised face shields were specified

In Lyon, France for example, a 6.5m diameter Hydroshield operating on the Wayss & Freytag air pressure controlled bentonite slurry principle with an extruded concrete primary lining was making good progress under the Rhône and Sâone Rivers after early setbacks.

Lovat soft ground TBMs with their hydraulically controlled flood gates working in conjunction with compressed air had proven themselves on many subaqueous jobs such as under the River Dodder in Ireland and the sewer siphon under harbour channels in Hamburg. They had also proven themselves through wet ground conditions in Washington DC itself.

The Herrenknecht Mixshield, also based on the Wayss & Freytag bentonite slurry principle, was beginning to make a name for itself after recording its debut on the 6.5km long Hera nuclear experimental accelerator ring in Hamburg.

Then there were the many different large diameter soft ground tunnelling machines being designed and used in Japan. Few of these machines have been sold outside Japan, but with the huge domestic market showing signs of easing, Japanese manufacturers are beginning to concentrate on the export market for survival.

Hitachi Zosen's EPBM ready for shipment

Most of the alternatives were examined by Harrison Western-FDI and several soft ground tunnelling projects were visited, and it was in Japan that the JV made its choice. Price, delivery time and horsepower were said to be the criteria upon which the US$1.7 million order was placed with Hitachi Zosen.

Meanwhile, construction of the two deep working shafts were well underway. Franki was well experienced in the slurry wall technique of shaft construction. This is one of the principal reasons for its partnership in the JV. The slurry walls are 3ft thick and must withstand ground water pressures of 40psi (2.7 bar), the water table being within 1m of the surface. There are also several heavy duty steel bracing beams across the shaft at various depths. Deep well dewatering as a method of controlling ground water was forbidden by the client for fear of causing settlement damage particularly to a 120ft (36.5m) high chimney stack on a power plant only 180ft (55m) from the reception shaft on the north bank.

The EPBM was assembled in six weeks and set off on its first drive on 16 December 1985.

First EPBM for the USA

Of the 650 tunnelling machines built so far by Hitachi, 70 are of the bentonite slurry type and some 175 are EPB shields. The largest EPBM in the world is a 7.5m Hitachi machine, which worked on the Sapporo Metro in Japan in the early 1980s. In 1986 only six shields had been exported by Hitachi and this is the second of only two Hitachi fullface cutter head shields working outside Japan. The first is in Singapore excavating metro tunnels.

For Harrison Western-FDI, the Japanese were presented with the challenge of making one of the highest capacity machines it has ever built. Harrison Western wanted a machine with a maximum 3,000 US ton thrust capacity to be delivered through 24 independently operated thrust jacks. It wanted a 394 tonne metre cutterhead torque and a 900kW (1,200hp) cutterhead drive to provide a maximum 4.1 rev/min and push a 4ft (1.22m) stroke in 20min.

It was said that, had the machine been ordered for the Japanese domestic market, an advance of 1m in 40min would have been accepted with a 1.5rev/min cutting speed. Thrust jacks would also have been designed to work in sets instead of independently.

"After a successful launch and installation of backup equipment at the 200ft mark, the machine did well achieving up to 60ft (20m) in a 24/day," said Armand Fournier, Assistant Manager. However, despite best efforts, the first tube took about two months more than scheduled to complete. Inexperience of Harrison Western's operators and engineers with this tunnelling technology, and extremely difficult and unexpected ground conditions all contributed to the delay.

Muck is fed through the bulkhead onto the primary conveyor by the central ribbon screw, which is capable of handling stones up to 18in (457mm) maximum. Larger stones would have to be excavated manually under compressed air conditions, a small air lock in the bulkhead allowing man entry to the 36m³ excavation chamber. This lock is on site but so far its installation has not been necessary.

From the primary conveyor, muck is transported by continuous overhead conveyor to the shaft where it is dumped into one of four 8yd³ (6.12m³) muck skips arranged on a carousel. Full skips are hoisted out of the shaft for discharge.

Fig 4. The Hitachi Zosen EPBM was designed to cope with saturated sands and gravels under 2.7 bar water pressure at the invert

Of the two lining options, Harrison Western chose precast concrete segments over steel segments on economic grounds. Steel segments would have added more than $1 million to the cost. The six 10in (25.4cm) thick reinforced segments and key that make each 16ft 8in (5.5m) i.d. ring were designed by WMATA and its section designer to carry the full overburden. They are of a stronger compressive strength than the normal 6,000psi in order to withstand the stresses induced by the relatively tight curves in line and level. A dense rubber gasket provides a watertight seal in the joints. The segments are manufactured by Charcon Tunnels of the US established by the British company Charcon, and the gaskets are supplied by D S Brown.

It was build of the segments that caused the first problem. The manually operated segment pick-up pin of the erector supplied by Hitachi proved inadequate and had to be replaced with an air operated catching device. It was taking up to 70 minutes to erect a ring. This was improved to 50 min/ring.

Backfill grouting when driving through the so called P2 sands which are charged with water, also proved problematical and hazardous to the safety of the tunnellers. No annular space exists in these sands. Therefore, grouting did not encapsulate the tunnel lining and when removing the grout packer, the miners were subjected to high pressure water and sand blow-ins. Pressure needed to inject grout to contract specifications in these conditions was 45psi. With WMATA's permission, the primary and secondary grouting specification was relaxed to suit the prevailing conditions.

Water leakage from behind the lining

Loss of pressure in the tailseal was a major cause for concern. The seal, designed to withstand 3 bar pressure, was established by two sets of wire brush seals with the 245mm long x 38mm high space between the segment and tail skin filled with grease. There is a shallow groove at each joint on the outer surface of the lining and it is believed that as the tailseal passes across these, backfill grout enters the tailseal area. Recurring electrical failures wore on every one's patience. These regularly interrupted progress because the electrical equipment of the machine was drip-proof but not waterproof.

That sinking feeling

Keeping the tunnel within the 1.5in (38mm) line and level tolerances in the soft, water-logged soil was also extremely difficult. The 250 ton machine, the shield of which is made of 40mm thick SS41kg/mm² steel plate, was occasionally sinking up to 1in at the leading edge when standing still during ring build. In addition, as the ring left the tailskin and became subjected to the full load, it deformed by more than the +0.5in (13mm) i.d. tolerances. The horizontal diameter should not, according to the specifications, increase by more than +1in (25.5mm). "This could not however be controlled, and the horizontal diameter tolerance was increased to a maximum of + 2in (50.8mm)," explained Richard Horton, WMATA's Resident Engineer.

Discrepancies in line and level will be compensated for and corrected during the casting of the in-situ concrete track bed invert and safety walkway, which are part of the contract.

Morale on the job site was lowest when, about 300ft from the reception shaft, the machine ran into dense, sticky clay. Not only did this clog the screw conveyor and excavation gate, it also became compacted against the sides of the excavation chamber. A combination of compressed air and water was used to help force the clay out of the screw. The sand plug, which prevents water escaping through the excavation gate during normal EPBM operation, was also removed which helped. The trough in which the screw conveyor lies had been removed earlier.

A six week delay then occurred when built up clay in the excavation chamber fell, damaging the screw. Fortunately, being in tight clays, the face was boarded up and repairs took place in free air.

The muck skip carousel at the shaft bottom

When the machine reached about midway under the river, the articulation seal failed, a setback that reduced still further the undermined confidence of the tunnelling crews and operators. What the job needed was a well experienced tunnelling engineer who could restore confidence and push the troubled drive to its end. The man asked to take on the job was Helmut Kobler, a retired Harrison Western Manager and a tunnelling veteran that TunnelTalk first met on the equally difficult adit for the 28km long Yacambu irrigation tunnel in Venezuela.

Kobler agreed to come out of retirement in California and, together with Assistant Manager Armand Fournier and Egon Franzisz, a long time Harrison Western tunnel superintendent, succeeded in getting the best out of the operators and tunnel crews to complete the drive. With no articulation, the machine had to be steered the rest of the way by means of the thrust jacks. Harrison Western’s foresight in specifying independent operation of these jacks was now being appreciated.

However, the job was far from over. Hole-through was also expected to be difficult, but not as difficult as it proved. It was known that the tunnel face at break out would comprise about 16ft of clay in the lower part with about 2ft of P2 wet running sands under about 30psi pressure in the top.

Knowing too that grouting in these sands proved ineffective, it was decided to apply liquid nitrogen ground freezing to control this ground water during breakthrough, deep well dewatering having been forbidden. Freezing was applied about 40 rings from the shaft wall.

All seemed to be going well and on 16 September 1986, the EPBM bored its way through the lm thick slurry wall. However, while on site TunnelTalk witnessed a minor in-rush of wet sand from the annulus between the shield and the slurry wall which proved a bad omen. Two days later, just as the shield was to make its final shove, more than 70yd³ (53m³) of material burst through. Freezing had also proved inadequate.

Once the situation was recovered the client accepted a claim of unforeseen geological conditions and it was by rescinding its original ban on dewatering that the shield finally emerged. Three dewatering wells were sunk to lower the water table by about 20ft. This did not result in any serious settlement and the client will allow the same dewatering procedure for break through of the second drive.

This same shaft will be used by Mergentime as the reception shaft for its twin running tunnel contract from Navy Yard station. For breakthrough of its Lovat EPB TBM dewatering by up to 20ft for about two or three weeks will be permitted but the wells will have to be installed and the pumps tested well in advance and not established as an emergency measure as in Harrison Western's case.

Prepare for drive number two

While on the surface after the first drive, the machine was completely overhauled. The cutter teeth were changed, a new screw conveyor was fitted and the wire brush tail seal and failed articulation seal were replaced. On December 16, 1986, the machine started the second drive and true to the prediction by Eric Smith, the main operator for Harrison Western, and Hitachi Engineer Hiroshi Tomiki, it has taken half the time to complete the first 200ft (61m) as it did to do the same distance on the first drive. "We were much more experienced by the time we set out on the second drive, having learnt the lessons the hard way on the first drive and we will finish the second tube in half the time, you'll see," said a confident Smith when TunnelTalk was on site.

So confident indeed is everyone about the second drive, that Helmut Kobler has retired again to his vineyard in northern California and Hitachi's technicians have returned to Japan.

But work does not stop after completing the second tube. At present, area manager Dan Harrison is controlling the open cut excavation of the double cross over box on the Green Line contract F3a which Harrison Western won as lowest of ten bids late in 1985. The two 2,200ft (670m) long tunnels involved will be excavated with the same EPBM.

Transfer of segments from the flat car to the erector pickup area

Conditions are expected to be fairly similar on the next contract. Ground water pressures of 15psi are expected but there will not be the threat of the river overhead this time. Stones and boulders are expected to be a greater problem and the compressed air lock will be permanently installed.

With the advantage of hindsight, and as is always the case on troubled projects, the wisdom of earlier decisions is questioned not only by those who were directly involved, but by all interested tunnellers. Was the use of the first ever EPBM in the US the correct choice? "I believe it was," Jay Carlson, Vice President for Harrison Western told TunnelTalk. "I do not believe we could have completed the project economically any other way."

Most agree that to rely on compressed air would have proven much too dangerous although this is believed to be the first tunnel ever driven through wet ground under 40psi pressure without compressed air.

Helmut Kobler is of the opinion that a slurry shield of the Hydroshield type would have been more suitable. However, the Japanese technicians and others doubt the ability of the bentonite slurry to transport the sticky clays encountered and fear it would have been difficult to separate these clays from the bentonite.

Was the decision to buy the first ever Hitachi EPBM used outside Japan the correct one? Some would say not, but it is unlikely to be the last Japanese EPBM used in the US and Harrison Western and FDI can claim to have had the courage and foresight to introduce what is undeniably a sophisticated and successful soft ground tunnelling technique.

Above all, was it right for WMATA to specify a bored tunnel. "We agonised over this decision for a long time and it was only at the eleventh hour, just before inviting tenders, that we eliminated the immersed tube option," said Vernon Garrett, the Design Engineer for WMATA responsible for this contract.

No doubt the metro tunnel under the Anacostia will be a topic of discussion for some time yet and the experience called upon by other transit authorities around the world contemplating similar subaqueous crossings. But it is through projects like this that tunnelling moves forward and no-one can take it away from the Harrison Western-FDI JV. It has pioneered a new tunnelling technique in the US. It paid the price for taking the chance but succeeded.