Foundations for the £350M Bluewater development, initially challenging because of the size of the project, also presented some interesting technical issues.
One was avoiding overdriving in the tougher than expected ground in the 1km long, 0.85km wide and 60m deep former chalk quarry near Dartford in Kent. Chalk had been quarried by bucket grab excavators in large 'cells', separated by bunds up to 15m high and 10m wide.
Continual dewatering to ease excavation lowered the regional groundwater table (the chalk is a class A aquifer). Over 40 to 50 years the excavated cells were backfilled with Thanet Sand during quarrying.
The 102-week piling contract was carried out by Kvaerner Cementation Foundations in joint venture with Stent. Prior to piling dynamic compaction was carried out. Any remaining areas were backfilled with engineered fill up to 20m thick, with dewatering carried out where necessary.
In total over 16,000 piles were formed comprising 7298, 270mm square and 8173, 350mm square precast driven load bearing piles, along with 1002, 450mm diameter CFA piles required to resist uplift from the regional groundwater recovery.
The original scheme was to install 270mm square driven precast concrete segmental piles through the Thanet Sand fill, socketing between 3m and 4m into the chalk.
However, value engineering led to a combined scheme comprising 270mm and 350mm square piles. This allowed pile production to be carried out by three factories available to the joint venture, enabling it to supply piles at a rate to meet the tight programme. The use of larger piles allowed for a reduction in the number of piles as well as savings in time and money.
Design called for maximum loads of 1000kN and 1750kN for the 270mm and 350mm piles. These loads were influenced by the amount of negative skin friction generated by the effective depth of fill, determined by probe piles.
Precast piles were driven using five rigs although preboring was required in some locations to penetrate the engineered fill and the chalk bunds. This allowed piles to be driven to similar depths, thus minimising the risk of differential settlement between pile caps.
One of the biggest problems was that of piles breaking during driving. This was thought to be due to the high quality and strength of the chalk and the uneven profile of the quarry floor. As the square pile toes hit the steep chalk rockhead profile, they had a tendency to be deflected sideways until they met softer chalk or a more horizontal profile. This set up bending movements and stresses leading to failure of individual pile segments.
Different pile toe designs were suggested to reduce breakages, but trials on site revealed that the strength of the chalk meant that these had little effect.
Maintained load and dynamic pile testing confirmed that the high risk of pile breakages was due to the chalk's high quality and strength. However, it also showed that the pile design was conservative, with ultimate bearing capacity governed by the structural strength of the pile rather than geotechnical properties of the chalk.
A relaxed driving set of 30mm was based on test pile performance. Back analyses of the test pile showed that the original 10mm set was conservative.
Pile breakages often led to problems in planning rig movements and programming because installing adjacent piles would restrict access to replacement pile positions.
A quality control system was put in place to ensure no broken piles and replacement piles were missed. A record, updated hourly, was kept to monitor any broken piles, piles failing to reach prebore depth and any piles out of vertical tolerance.
CFA piles were also installed in the deep basements to resist uplift generated by the recovery of the regional water table. They were considered a more economic solution than using large driven piles where the tension capacity of the precast pile was governed by the capacity of the pile joints.
Unfortunately, the CFA piling also turned out to be problematic in places where surface water had softened the near-surface chalk. When covered by the Thanet Sand fill, soft zones formed at depth causing 30% to 60% overbreak in some of the piles. This also caused problems in rig planning because it was feared that installation of adjacent piles could cause slumping in these piles.
At the peak of construction, five precast rigs, two prebore rigs and one CFA rig were working to produce 450 piles a week. Due to the various hold-ups on site, planning had to be flexible. This was achieved by using GPS setting out technology. Pile positions were downloaded directly from electronic drawings stored on the site computer network. However, traditional methods of setting out were used where interference and signal loss occurred.
Paul Wiltcher is a graduate engineer at Kvaerner Cementation Foundations.