The three stations along the central underground section of the line, Rokin, Vijzelgracht and Ceintuurbaan, have all been a major challenge to construct, each within a huge underground box of diaphragm walls and concrete floors. They are on the critical path for the work and delays here have been the main factor in stretching out the project schedule.
Firstly they must be built up to 32m below ground, which is no easy matter in the waterlogged soft soils of the city and with the high water pressures that implies. It affects both the design and the construction work.
Secondly all the excavation and construction must be done in some of the busiest streets of the historic centre, close alongside ancient buildings and their pile foundations and with traffic, trams, tourists and residents passing above and alongside. Noise and working time limits are tight, space constrained, and equipment restricted.
The three stations all have their own characteristics. First is the Rokin, the most central and close the main Dam square and the palace. It sits halfway along the Rokin which is a historic street, once the course of the river Amstel and then infilled when the Dam was built some centuries ago. It is now on one of the rings in old Amsterdam’s wheel-and-spoke street and canal layout.
Its wide dimensions allows for the easiest of the three station structures since the station has relative space to spread out. That means it is wider than the others at 33m and can be kept shallower, going only 27m down. It is 250m long.
On the edge of the old centre but still alongside historic houses, perhaps unfortunately, is Vijzelgracht. There is less room here but it can still be 20m wide, this time dropping to 30m deep. It is here that problems were encountered with the diaphragm wall joints, leading to an inflow and building settlement.
Further south Ceintuurbaan is located in a narrow street in a newer part of the town, built just post-WW2 and near to a well-known street market which is one of the city’s tourist attractions.
Space limits between the buildings mean the finished station here is 11m wide, just under 13m during excavation. To fit the line in, the tracks have to be stacked one over the other instead of side by side. The box is therefore deeper at 31m, if not much more than Vijzelgracht. Full excavation depth reached 33m.
A confident strut
Like all the stations, the buildings around are on relatively shallow piled foundations through the soft peaty ground into the first sand layer, though rather then the timber piles found on the older buildings, the postwar foundations are mostly steel. "But they are often not that sound and close to a safety factor of one," says Henk de Pater, the city client’s project manager for the station. The design for each station is similar, a perimeter diaphragm wall of 1.5m thickness and dropping as deep as 47m for Ceintuurbaan, reaching the impermeable clay which sits above the deepest ‘third sand layer’. It has to resist high water pressures in particularly the various sand layers encountered, which means it reaches a maximum of 3.5 bar.
Because of these high pressures, project main designer Witteveen+Bos added a jet grout ‘strut’ at the base of the diaphragm walls on each station to supplement the normal steel struts and concrete floors installed higher up, which usually prevent deformation of the walls.
The designer believed that the position of walls close to buildings just 3m away, and the depth of excavation well below the bottom of the house piles, which are just 13m long, made the addition important.
The jet grout columns are in contact but are specifically meant not to overlap completely such that they form a floor adds Witteveen+Bos design engineer Richard de Nijs. "The water is supposed to come through," he adds.
Installation was done using an observational method with modifications to the installed pattern made as measurements were taken, and core samples retrieved during the subcontractor’s work. This took around five months at Ceintuurbaan and similar times at the other stations. The aim was to achieve a stiffness between fairly tight limits 1000MPa and 2200MPa, which was done.
Sinking the walls
Ground strutting work came at the end of the diaphragm wall installation at the stations, which was mostly carried out by Belgian firm Franki working for the German main contractor Max Bogl which did all three stations. Work began in 2003.
Installing the diaphragm walls took longer than supposed initially, particularly because of the working constraints at the site which allowed only a limited number of the big diaphragm rigs to work at any one time. This was especially true at Ceintuurbaan where an initial site occupation layout, taking up most of the street, had to be abandoned because of traffic and pedestrian problems. A method of working one side of the street and then the other, leaving access to local shops, was found.
"Here at Ceintuurbaan it took us three and a half years or so to do the walls," says de Pater. "It had been estimated by the city that it could be under two years" he says "but it was realistic to believe it would be longer. It is very difficult working in this very wet and soft ground to achieve verticality."
One problem is local vibration in the soft peat layers at the top he says. "You can feel when heavy machines or trucks pass by and there is some movement between the wall panels."
Walls were built conventionally with a small concrete guidance wall set up at ground level and using a grab system.
Once in place the station top could be formed. For the very wide Rokin station this was done with precast units laid out and then stitched with in situ concrete. The beams rest on internal columns halfway across. For the other stations a top slab was cast. At each station excavation then was carried out by top down construction, using three apertures in the slab for mucking out.
Spoil was lifted out by crane in 8m3 skips which load in threes onto trucks to run out through the narrow city streets. There are various options for the disposal. Top and potentially contaminated layers go to disposal sites, but lower material can be used for harbour infill and in the case of the dense clean sand layers, makes an excellent raw material for building work.
"Mucking out has been quite a constraint on the contractors," says Pelle de Wit, Amsterdam City assistant project manager at the Vijzelgracht station. "Inside the station particularly, logistics are quite difficult as there is considerable strutting which has to be worked around for excavation and then concreting work later."
Most stations are strutted at 6m intervals with largescale temporary steel tubing. Further down a permanent concrete slab is made which becomes the eventual station roof for the train platforms that are at the bottom of the station box.
"The space above, over four floors, will be used for car parking" says de Wit. At Rokin this will comprise four floors of concrete underground parking space, the floors of which will provide permanent strutting. At Vijzelgracht the space above will "probably have parking using an underground robotic system to store the cars in steel racks," he says.
Under pressure
The initial concrete slab roughly 19m deep at the bottom has a secondary purpose, namely for containing compressed air. The depth of the stations and the permeability of the sand layers meant that uplift from ground water pressure needed to be contained by working under air once a depth of 25m is reached. Up to 1.9 bar overpressure was used in the case of Ceintuurbaan and just under one atmosphere at Vijzelgracht. At this lower pressure the health and safety issues are much reduced and shift times can be close to normal.
"For design the air pressure loads are the main loading factor to consider in the specification of the slabs" says project engineer Richard de Nijs, at consultant Witteveen+Bos.
It was originally supposed that some compressed air work would be needed at the less deep Rokin station too but calculations by the consultant showed that friction from the diaphragm walls could play
a significant part in restraining uplift. As its depth was less, the work could be done without the air. It also helped that the station did not have an ‘intermediate’ sand layer with water under pressure.
This extra sand layer is found at about 47m deep in some locations in the city sitting in between the main second and third layers. It was present at both the other two stations and for each a dewatering system was used to reduce the pressures from it with wells drilled down to around 45m below ground, just above the end of the diaphragm walls. An incoming water flow of about 5m3/hour had to be pumped at Ceintuurbaan, and nearly twice that at Vijzelgracht station.
Dewatering did not completely eliminate the need for compressed air working but it did reduce the incoming water pressures from below the station and allow for some reduction in the compressed air overpressures.
For a while it was thought the pumping and friction mechanisms would allow Vijzelgracht to eliminate compressed air working altogether. But after some dramatic incidents in 2008, when diaphragm wall leakage led to settlement of some historic buildings, this was not possible. The City wanted the most conservative way of working to be adopted and air overpressure was used for the deeper parts of the excavation to add a margin of certainty.
Pumping was also made more secure by installing vacuum pump technology on the well heads and making a duplicate set of wells, some forty in all, as a redundant backup system.
The settlements came early on in the excavation work when it had reached about 13m deep, roughly the level of the first sand layer and the end of the timber piles on the historic buildings.
Uncontrolled settlement
Two major events occurred which stopped the station work for over a year, and raised questions of cancelling the entire scheme. First was in June caused by leakage between two panels of the diaphragm wall, most likely because of a misplaced or damaged water stop though it has not been fully determined yet. A rubber stop seal is added during panel excavation which is held in place by a steel frame as the next panel is grabbed out and then removed before concreting. Taking it out can be tricky. Whatever the cause there was some water inflow.
"That was not good but the real problem was that the inflow brought ground material in with it," says de Nijs, "with a 15m head of water driving the erosion."
In a first small incident there was only 2mm of settlement but two days later a second leak caused a large volume of water entry, along with sand. The result was very rapid settlement propagated to the surface impacting on buildings close to the station works.
That was serious enough, causing subsidence at the surface of about 160mm underneath some 18th century houses. Fortunately the buildings though damaged were still usable says de Nijs.
"Work stopped and everyone sat down to consider the incident" says his colleague Sujeet Bhageboe. Over three months the joint could be repaired using resin injection and the diaphragm walls were inspected for further problems but although some ‘sweating’ was observed, it was nothing out of the ordinary.
It was agreed that work could continue but with a specified sequence for spoil removal and only one joint at a time. So in September excavation continued. Within two days there was another inflow, which was even larger causing a drop of 4m in the water table at the surface, and a total 230mm of settlement.
A row of historic Weavers Houses dating back to 1780 suffered major cracking and movements lasting four hours forcing evacuation of the occupants, mostly local businesses. Some 15km of timber shoring was used to prop up the structures in subsequent days and the city ordered a stop of works.
A year long hiatus followed while the cause was examined and investigated and the city decided on what to do.
The immediate difficulty was clear; there was a hole in the diaphragm joint several centimetres wide caused more likely by a bentonite inclusion. How that happened is again subject to varying opinions which may yet be contested legally.
But rather than attribute blame at the time, and potentially bankrupting various parties, the city decided to rebuild relationships, establishing new pain-gain partnering contracts and a new set of procedures for monitoring and organisation of the work. Disputes may yet be fought out but after completion.
A key element was how to monitor the diaphragm walls joints and how to remedy flaws. The solution eventually agreed was to use ground freezing to seal any possible leaks ahead of excavation.
The freezing would be internal rather than outside. External freezing would allow ground pressure to help seal the joint as an ice body built up. "But there would be a risk of ice buildup disrupting the ground structure" says de Nijs.
On the inside wall therefore, twin freezing pipes were set up by each joint, to form an ice body about 2m thick. Excavation could then proceed very carefully joint by joint, with the possibility of replacing spoil if any leaks were found.
"The ice body itself was carefully milled away using a roadheader attachment on an excavator " explains Felix Paleari, the city’s project leader for Vijzelgracht. Since this also milled away the ice pipes during each 2m deep bench, they were reattached across the gap with flexible hoses.
Once exposed the joints were assessed and sealed with a low temperature setting grout mix. Steel plates were then fixed across the joint for sealing in one of five categories, of increasing heaviness. "There is a later concrete lining to go in which covers this and adds to the wall strength" says Paleari. With these methods the remaining excavations have been completed, the last at Vijzelgracht finishing in December last year. After installing the 2m thick base slab for the station the attention is now on building up the internal concrete structures and eventually fitting out for the stations, continuing until 2014.
Once the TBMs have passed through the stations to complete the main tunnels drives, track and signalling will begin too.
