Report on the British Geotechnical Association informal discussion Design and construction of Ashford cut and cover tunnels, held on 21 January 2004 at the ICE, London, by Paul Tester, Cementation Foundations Skanska.
The cut and cover tunnels of Contract 430 form the centrepiece of the 15km stretch of Section 1 of the Channel Tunnel Rail Link (CTRL) through the centre of Ashford (Figure 1).
Starting with an overview, Dave Twine from Rail Link Engineering (RLE) explained that the tunnels carry four tracks that run in parallel as the CTRL line approaches Ashford Station.
Crossing above the tunnel are the Maidstone railway line, which was built in two, 52 hour possessions and a main road also which was rerouted a number of times during construction.
Geotechnical design Ground conditions at the tunnel site were superficial deposits over Hythe Beds, Atherfield Clay and Weald Clay, with groundwater 3m below ground level. The behaviour of the Atherfield and Weald clays is very similar to weak rock (Figure 2).
Pump tests showed silt partings in the Weald Clay led to high horizontal permeabilites (10 -7 m/s) yet low vertical permeabilities (10 -9m/s). This led to almost hydrostatic wall loads (Figure 3).
Tender design was based on effective stress parameters and used Eurocode 7 safety factors, with the most conservative of serviceability and ultimate limit state calculations. During the design process a number of value engineering exercises were carried out. However, once the contractor was appointed, some of these were reversed, highlighting the importance of the early contractor involvement for value engineering. Figure 4 shows the design programme.
Construction Adam Chodorowski of RLE covered the move from design to construction. Co-operation proved key on site as commitment to true partnering from the top downwards was realised on site with:
l Co-location of the field teams in the site offices l True open book working with one set of files l Common access to the electronic database and document control.
About 3,000, 0.9 to 1.5m diameter piles, 20m to 25m deep, were installed for the retaining walls. Foundation piles were 30m to 45m deep. Piles were bored under bentonite to avoid degradation of the pile bore and a significant number included plunge columns for top down construction.
During excavation, a number of cage imprint features were found on the retaining wall, just below the temporary casing toe level. Piles were cored, slots were cut in the piles and ultrasonic testing was carried out to check concrete quality and cover to reinforcement. These proved acceptable, although there was a layer 5mm to 10mm thick of poor quality concrete on the edge of the piles.This was grit blasted off where necessary.
Remote cameras used during pile concreting found the concrete flow had led to poor quality concrete on the pile edge (Figure 5). To prevent this, concrete workability was kept to the high end of the specified tolerance, bar spacings were kept above a minimum of 100mm and good cover was maintained using heavy duty wrap-around 'ladder spacers' rather than the usual plastic spacer wheels.
Tunnel construction involved a number of different techniques including bottom up, top down, top down retained cut, cast insitu with open cut, cantilever and propped excavations, external relief digs, and perhaps the most challenging - dewatering.
Dewatering was needed to reduce pressure on the wall and also to keep stability within the excavation. It was not possible to dewater the Atherfield Clay effectively as the strata was recharged by water from the Hythe beds.
Ejectors were installed in the Weald Clay to reduce pore pressures (although the volume of water extracted was not high - 0.15 litres/s/150m).
These were installed 5m to 10m outside of the excavation at 12m to 15m spacing and ensured a water head reduction of over 20m in the Weald Clay. Sand drains within the excavation ensured stability.
The observational method was introduced during construction, jointly agreed by Skanska and RLE, including the construction sequence, design predictions and trigger levels. Some £1M worth of instrumentation was installed at critical design sections, including inclinometers, convergence points, prop strain gauges and piezometers.
Instrumentation was designed to have a degree of redundancy while avoiding information overload.
Raw and processed data was available to all the site teams on the shared database.
Value engineering Howard Roscoe of Cementation Foundations Skanska explained more about the value engineering contribution which the contractor brought to the project. By changing as much construction as possible from top down to bottom up, access and ventilation issues were reduced allowing more efficient construction.
Wall movements during construction were typically up to 35mm, significantly increased by a surcharge of soils above the capping beam level. It should be noted that a 5mm movement was typically recorded at the toe of the retaining wall.
Figure 6 shows the relationship between retained height and wall movements.
General maximum movements were 0.4% retained height, even with significant surcharges on the active side of the wall.
Typical movements of 0.2mm per day were recorded during excavation.
Monitoring and research The instrumentation and observational method approaches were used to great benefit on site. Initially, the excavation and concreting of the base slab were to be carried out from two access points but restrictions meant only one was available.
Monitoring of the retaining walls meant the whole stretch could be excavated before the base slab was concreted, retreating from the end of excavation. This avoided concreting and excavation being carried out simultaneously which would have been restrictive, slow and costly.
Strain gauges on the tubular steel temporary props provided additional information (Figure 7). Gauges were corrected for temperature effects as thermal effects in concrete during curing can confuse results. Results are shown in Figure 8, showing that using typical calculated prop stiffness values is overly conservative - prop loads do not generally reach those predicted and using 50% prop stiffness gives more sensible results. It was also found that the upper of two props attracts a higher proportion of the load.
Originally it was proposed to use 'sway correction' for movements of the tunnel walls due to an imbalance of forces on the active sides of the two retaining walls and the props in between.
However, instrumentation showed prediction by this method does not work well, as the elastic prop shortening is of the same order as the 'sway correction' Finite element analysis predictions produced were much closer to measured wall movements.
Conclusions It was proposed that a design standard or guidance for design of cut and cover tunnels is needed. This should be method dependent and so the construction team should be involved as early as possible.
Also, when designing for the Weald Clay, its anisotropic permeability properties must be considered.
The partnering environment was vital, leading to the resolution of problems encountered such as those with the pile concrete and reinforcement cages. Monitoring played a key role in the project, with ease of access to up to date data in an easily usable format.This led to excellent research information on time dependent soil and concrete movements and prop loads which allowed the development of predicting wall movements, prop loads and tunnel 'sway' Questions and answers Chris Clayton from the University of Southampton questioned the apparent contradiction between clays not behaving as clays, the high permeability of the Weald Clay and the small volumes of water abstracted from the ejectors.
Twine said descriptions suggested a clay-type material but further investigation showed pile design parameters were different to those that would be expected for a typical clay. The anisotropic nature of the Weald Clay meant the volumes of water abstracted were small, coming from the thin silt partings. However, this did provide pore pressure relief on the active side of the wall.
BGA chairman Tony Bracegirdle asked about the swelling properties of the clay. Twine said extensometers gave results in close correlation with effective stress analysis. Typical recorded movements were 10mm to 20mm.
Adam Chodorowski added that the Atherfield Clay was typically fissured in matchbox sized pieces which led to some weeping and a wet clay face, a degrading of the cut face and occasional slumping.
William Parry of the University of Southampton asked about wall movements which were not related to consolidation of the clay.
Howard Roscoe said the issue should be the subject of more research, particularly with reference to when concrete changes from short term to long term crack conditions. This is faster than five years but how much faster is the question.
Duncan Nicholson of Ove Arup & Partners asked the panel to comment on the dewatering and the pore water pressure profile outside the wall. Roscoe said design was based on effective stress parameters after trials. These looked at how to dewater the Atherfield Clay with ejectors.
The assumed pore pressures were approximately hydrostatic and being allowed to use these parameters avoided the risk of delay. If design had been based on pore pressures and for any reason piezometer readings were not available, the excavation process would have been held up.
Tony O'Brien of Mott MacDonald asked why a hard soft secant wall approach was not adopted to prevent the lateral movement of water. Twine said the design was progressing as the problem of lateral water movement evolved.
A number of ideas were considered including filling the gaps using jet grouting or changing to diaphragm walls or gravity relief structures.
However, in terms of the overall contract programme the solution constructed was the best. Roscoe added that to do primary and secondary piles would have meant the programme would have overrun.