A new high pressure gas pipeline is being construct- ed between Goxhill, on the south bank of the Humber Estuary in North Lincolnshire, and Paull, on the north bank in the East Riding of Yorkshire. The pipe- line route includes a new 5 km bored tunnel under the Humber Estuary through which the pipeline pass- es. A launch pit at Goxhill and a reception shaft at Paull were required for the TBM drive. This paper covers the design and performance of the temporary works groundwater control arrangements used to facilitate excavation of the launch pit and associated ramp structure. The site layout is shown in Figure 1.

The site was previously agricultural land with a ground level of approximately 2 mOD. The launch pit was to be primarily excavated in superficial glacial soils, which act to confine groundwater in the underlying chalk, although at its deepest point the excavation extended into the upper portion of the chalk. Groundwater levels monitored in the chalk prior to the works were typically around 1 mAOD. Preliminary studies suggested the likely presence of a thin horizon (up to 5 m) of potentially highly permeable chalk „bearings‟ (clast dominated frost shattered chalk), which are a feature of Northern Province Chalk Group, immediately below the superficial de- posits. The launch pit was 80 m long by 7 m wide, narrowing to 5 m, with an associated ramp structure 48 m long by 5 m wide. The maximum dig depth was 11.2 m, with a formation level between -3.8 mOD and -9.2 mOD. The launch pit had side support com- prising secant piles extending to 19 m AOD and sheet piles extending to 17.6 mAOD. Following excavation a base slab was cast to accept full hydrostatic loading, effectively isolating the launch pit from the groundwater regime and allowing cessation of pumping. A cross section through the launch pit is shown in Figure 2. 

Figure 1. Site layout showing site investigation borehole locations and pumping test layout.

A groundwater control solution comprising a partial cut-off with internal pumping wells was developed to reduce upward pressure from the chalk and facilitate the excavation of the launch pit.

The site was subject to a number of environmental constraints that informed the design of the groundwater control system. The chalk in North Lincolnshire was classified as “fully committed” by the Environment Agency under their Catchment Abstraction Management Strategy meaning pumping to a surface discharge point was not viable. Additionally, the proximity to the Humber Estuary, which presented a risk of inland migration of the saline interface, and the presence of a nearby private supply borehole necessitated the use of external re-injection wells to minimise the drawdown in water levels outside the launch pit.

Figure 2. Cross-section through structure showing general soil profile and proposed cut-off.

A further consideration was the driveability of sheet piles in the relatively dense Northern Province Chalk Group. In order to achieve a suitable barrier to flow, the piles needed to be driven without causing disturbance at the pile toe, which would create preferential pathways. Vibratory driving of the sheet piles was deemed viable following review of the available ground investigation information. This proved to be the case, with only a small fraction of the piles refusing just short of the design depth and no indication of damage to the chalk below the toe of the pile leading to the development of preferential flow paths. 

A pumping test programme was carried out to un- derstand the variability of hydraulic conductivity with depth. The pumping test results were used to de- termine the required target depth for the internal pumping wells, and the required drive depth for the piles to key into the competent chalk underlying the chalk bearings.

Seepage flows into a cofferdam are largely controlled by the permeability at the toe of the retaining wall which in this case is the Grade A5 Burnham Chalk Formation, see Figure 2 and 3. There was also a requirement to recharge the abstracted groundwater which it was known from experience would lead to significant flow recirculation unless it could be shown that the cofferdam provided an effective cut- off for any superficial higher permeability chalk. It was clearly necessary to design the pumping test to investigate the variation in permeability with depth. In order to achieve this, a total of three wells were in- stalled with response zones at increasing depths as shown in Table 2 and Figure 3. 

The shallow well PW1 targeted the superficial Grade Dc chalk bearings. PW2 targeted the horizon at about the design level of the retaining wall. PW3 targeted the high grade chalk below as a contingency measure in case the results from PW1 and PW2 indicated that a deeper cut-off was necessary. The response to the pumping was monitored in an array of existing and new piezometers, at varying distances from the pumped wells from 5 m up to 600 m. 

The test programme involved step tests and 48 hour constant rate discharge tests on each well followed by a 72 hour abstraction recharge test. A log time plot showing the drawdown data from the constant rate discharge test is shown in Figure 4.

Model Calibration

The values derived from a conventional single vertical well pumping test, as given in Table 3, are predominantly horizontal permeability values. Inflow to a cofferdam with retaining walls which penetrate to a significant depth below dig level, as in this case, is primarily controlled by the vertical permeability of the stratum in the vicinity of the toe of the cut-off. The Grade Dc chalk should be fully cut-off by the retaining wall. The inflow to the cofferdam during the construction works will then largely be controlled by the vertical permeability of the stratum at the retain- ing wall design toe level which in this case is the Grade A5 chalk targeted by PW2. 

Calibration was carried out by setting the abstraction flow from PW2 at 4.5 l/s with recharge to PW1 at the same rate as in the test. Commencing with the lower permeability value derived from the PW2 distance drawdown data for A5 chalk, the permeability and k_v/k_h ratio were then modified to obtain a close match to the actual groundwater level data recorded during the test. Good correlation was obtained for the differing response of the shallow and deep piezometers installed at SP4. The final calibrated numerical model used the permeability values and ratios given in Table 4.

Figure 5:  Numerical model cross-section.

An array of piezometers was used to monitor water levels. This included internal piezometers to check that the target drawdown could be achieved and external piezometers to monitor the effects of the re- charge. In addition the existing array of far field piezometers allowed an assessment of any impact on neighbouring abstraction licence holders. These were also used for sampling to assess any impact on the water quality, particularly saline intrusion from the estuary. 

Figure 7:  Site data from initial system operation.

The data gathered during the first week of operation are shown in Figure 7. The initial abstraction flow rate was approximately 12 l/s. The initial commissioning is represented by the period of relatively unstable flow as the dewatering system was trimmed and optimised to ensure the drawdown was achieved, and that the system was operating within consent constraints. A switch-off test was undertaken on 19 October which confirmed the expected rapid recovery in the event of an interruption to pumping. This reaffirmed the need for redundancy and automatic standby generator power supplies. Once this commissioning period was complete and the required drawdown levels had been established, the system was operating at an abstraction flow rate of just under 8 l/s which provides a comfortable margin relative to the base case model results. At all times during the project all of the abstracted water was recharged to the aquifer, with only a minimal rise in external water levels, typically 0.2 m – 0.3 m within the vicinity of the recharge wells; no discharges to surface waters was necessary.