An observational approach was adopted to determine the efficacy of passive wells and the need for active pumped wells for pressure relief at the London Clay/Lambeth Group interface during construction of the deep shafts for the foundations to the piers for the Millennium footbridge across the River Thames. This paper presents and compares initial predictions and hydrological modelling with the piezometric data obtained during and after construction.
The foundations for the Millennium footbridge were designed in detail by Ove Arup & Partners after submission ofan alternative proposal by the contractor, Sir Robert McAlpine/Monberg and Thorsen joint venture. Initial design was for a piled foundation and the site investigation data available at the time of tender had been obtained with the assumption that such foundations would be constructed.
The alternative proposal for 8m diameter shafts sunk to a depth of approximately 12m below River Thames riverbed level were subsequently developed by the engineer using 3D finite element modelling analysis (Ground Engineering, January 2000) to a final diameter of 6m and a depth of about 18m below bed level. The stiffness of the foundations was critical for the superstructure design and for its capacity to resist high impact forces to cater for a safety case involving collision from a large vessel travelling at speed downstream.
The shafts were constructed from within two sheet piled cofferdams in the River Thames (Figure 1). Two shafts were required for each pier foundation. Initial site investigation data was inconclusive in the measurement of piezometric pressures and further work was undertaken in an attempt to measure the piezometric profile over the critical depths towards the base of the shafts and beyond into the Lambeth Group. This was done using a piezocone and interpretation of this data suggested a hydrostatic profile with a water level at -1. 5m OD.
As the design was developing during the preferred bidder stage of the contract award, the contractor had allowed for the possibility of shafts being constructed to a level very close to the London Clay/Lambeth Group interface. At this depth there would have been a high risk of base failure in the shaft due to the expected pressures contained within more permeable layers of the Lambeth Group. Proposals were sought from specialist groundwater contractors for a scheme that would eliminate all risk of failure or excessive heave in both the base and sides of the shaft, which was to be constructed by traditional underpinning methods using precast segmental rings.
Three proposals ofdiffering concept and methodology were received and their technical merits were assessed by an independent consultant. The conclusion reached by McAlpine's engineer was that all proposals had merit and that in order to eliminate risk and obtain the most cost-effective solution, an observational approach should be adopted. The tenderers were therefore asked to resubmit prices, based upon a minimum ofwork involving passive wells and the installation of piezometers, as well as additional rates to cover the design ofan active (pumped) system using the data that would become available. A schedule of rates to accommodate these additional wells also had to be included. A contractor was appointed and the work put in hand.
It should be noted that the responsibility for design of this pressure reliefwork was part of McAlpine's temporary works to enable construction of the deep reinforced concrete infilled shafts for bridge piers in a safe and cost-effective manner.
The ground conditions for the area ofthe pier foundation are typically (Figure 2): 0m - 3m Thames Terrace Gravels 3m - 23m London Clay 23m+ Lambeth Group with Thanet Sands and Chalk lying underneath.
The London Clay is an overconsolidated stiff to very stiff, closely to very closely fissured clay. The Lambeth Group in the area consisted of interdigitated layers ofsilty sands, very stiffsilty clays, silts and clays. Groundwater was encountered in the sand layers of the Lambeth Group during drilling. It was reported to rise from about -28m OD to -13. 8m OD within 20 minutes following initial water strike.
Historic data suggested a piezometric profile influenced by rising groundwater conditions in the London Basin (Simpson et al,1989). Interconnection between sandy layers of the Lambeth Group and an upper aquifer within the Thames Gravels could not be discounted with knowledge ofscour hollows in the vicinity.
Considerable doubt therefore existed in relation to the piezometric conditions within both the London Clay and Lambeth Group formations. The contractor sought the opinion of specialist contractors and consultants in the assessment of the original site investigation and from relevant published data. It was concluded that further investigation work was required.
The key areas ofconcern were: a) the continuity of permeable layers within the Lambeth Group and the basement beds of the London Clay b) the precise piezometric pressures at critical depths c) the insitu strength ofthe soils at varying effective stress conditions beneath the shaft base.
Within the time available to the contractor to finalise temporary works design it was considered that the necessary data could best be obtained by insitu testing using cone penetrometer equipment with a piezocone tip. The construction programme allowed windows ofopportunity at nights and weekends for a jack-up platform with both shell and auger and CPT equipment on board to enable cone testing to the required depth.
The data obtained assisted the contractor in the assessment of the true piezometric pressures and the continuity and value ofpermeability in the silty/sandy zones of the basement beds of the London Clay and the critical upper levels of the underlying Lambeth Group. The piezometric data available to the contractor (Figure 3) was obtained from the pre-tender ground investigation report (Ove Arup & Partners,1998) and literature records (Simpson et al,1989).
Having sought and obtained various opinions upon the most cost-effective and safest method of eliminating the risk of base failure during shaft construction, the contractor, with benefit of additional data from the piezocone investigation, concluded that an observational approach to shaft construction was: a) essential, as doubts persisted b) cost-effective, as a fail-safe option would impose high costs for the project and also significantly delay the construction works c) technically the best solution for an indeterminate array ofgeotechnical parameters.
The River Thames at this site is tidal, with variations in height from MLWS of -2. 67m OD to MHWS of+3. 88m OD. This variable 'live load'was thought to induce load/pressure changes in the soils below the riverbed. However the precise effect was unknown.
Cofferdams were installed at the site of the pier to provide access enclosures for the shaft excavation and construction works (Figure 1). These cofferdams were about 9m wide by 19m long (Figure 4) and installed to a toe depth of-15m OD. Excavation within these cofferdams was made to a formation level of-9. 25m OD before shaft excavation. The excavated diameter of the two shafts in the cofferdam was 6. 27m, with a depth ofabout 14m from the cofferdam base (Figure 3).
As shown in Figure 3, the piezometric level ofgroundwater in the sand layers of the Lambeth Beds was much higher than the formation level ofexcavations. With the unit weight of the London Clay measured at about 20 kN/m 3(Ove Arup & Partners,1998), this level of pore water pressure would clearly produce an uplift force on the soils between the excavation base and the top of the water-bearing layers. This uplift force would be greater than the weight of the soils and the stability of the excavation therefore needed to be considered with respect to both hydrostatic forces and the shear strength ofthe plug ofsoil beneath the shaft.
Full mobilisation of the measured shear strength of the London Clay would provide an adequate factor of safety. However, there is considerable diversity ofopinion with respect to the value (%) of shear strength that should be adopted, taking into account the fissured nature of the clay and the effective stress conditions during excavation. A depressurisation system, either using a passive or an active approach, had to be designed and installed to provide sufficient pressure relief in the ground before excavation could proceed.
In order to define the risk and derive solutions to mitigate this risk, independent assessments in regard to the uplift pressure were carried out by both the main contractor Sir Robert McAlpine and groundwater dewatering specialist Oxford Geotechnica International (UK) (OGI).
Both assessments concluded that the best way to manage this risk was to adopt an observational approach to the problem. First, interpretation of the site investigation data indicated that there were great uncertainties as to the exact equilibrium water pressure in the sand layers and the depth and distribution ofthese granular soils in the Lambeth Group.
Second, model simulations as part of the risk assessment suggested that the requirements for the pressure relief system were very sensitive to the existing level of the water pressure in the sand layers of the Lambeth Group and their distribution. Thus it was considered to be impossible or impractical to design a system prior to construction to guarantee sufficient pressure reliefwithout incorporating significant potential redundancy in the system. Such redundancy would not only increase the cost of the project but also cause significant delay to the construction programme.
Therefore, the actual requirements of the pressure relief system had to be decided on site during installation once the piezometer data and drilling results were obtained.
To ensure the installation of a sufficient pressure relief system for the excavation works, a risk management plan was developed to provide controls and procedures for the management of the temporary works and the subsequent excavation works. This plan was developed to reflect both the uncertainties in the ground conditions and the tight time schedule for the construction works. The central purposes of the risk management plan were:
to provide a good standard of construction quality assurance for the installation of the pressure relief system to define and clarify the decision-making process with regard to the installation of the pressure relief system and the subsequent excavation works to minimise any unpredictable delays due to the possibility of machinery breakdown, difficult ground conditions, insufficient pressure relief, etc to anticipate any such problems and formulate contingency plans.
The objectives for the installation of the depressurisation system were:
to provide sufficient pore pressure reduction for safe excavation to protect the integrity of the ground formations below and around the foundations to minimise the hindrance to the shaft construction works.
The key components ofthis plan included:
1 Installation of pore pressure monitoring facilities to provide real time data for the decision making process during the installation of the pressure relief system and during the excavation works. These facilities would allow the establishment of background conditions and permit monitoring ofpressure changes as excavation progressed.
2 Establishing the critical pressure levels at which appropriate actions (such as pumping) were required to ensure that the uplift pressure was reduced to a safe level for stability of the excavation.
These actions were clearly defined and set up in accordance with the critical 'alert'and 'action' pressure levels (defined later) for management ofthe excavation works.
3 Installation ofpressure reliefwells to lower the head ofgroundwater in the sand layers underneath the shaft excavations. Modelling studies based on the data from the piezometers were used to determine the requirements of the pressure reliefsystem.
4 Model predictions and simulations ofpiezometer responses to the shaft excavation before, during and after well installation. Model simulations were used to test scenarios that might have been encountered on site and to examine the depressurisation requirements under different conditions. Similarly, model predictions of the measured pressure conditions were used to assist in interpretation of site data and to define and anticipate the alert and action levels (discussed later).
5 Real time measurement and analysis ofpiezometer data that would enable the project manager to make informed decisions in controlling the excavation operation on site.
Following this plan, a preliminary pressure relief system was designed based on the site investigation data. This system incorporated:
four vibrating-wire piezometers installed from the base of each cofferdam to monitor the pore pressure four deep wells located inside each ofthe shafts to facilitate pressure reliefduring excavation.
Well screens and casings were installed to the base of each of the pressure reliefwells and extended up to the level of the shaft base plug at -20m OD. Well casings consist of 11m of 75mm diameter 0. 5mm slotted PVC screens and 6m of the same diameter plain casings. This configuration was designed to facilitate pumping operations should they be required, while minimising possible obstruction to the excavation works. After the installation ofwell screens and casings, the well was backfilled with 1mm-2mm graded clean filtration sand up to the bottom ofthe cofferdam (ie the top of the shafts). In this way, excavation of the shaft below the natural piezometric head ofthe water-bearing strata would activate passive pressure relief.
The ability to convert passive wells into active pumping wells would further enhance the capacity of this pressure relief system to cope with any unexpected higher water head during the excavation which may be caused by dynamic loads such as tidal changes.
A quantitative understanding of depressurisation in terms of hydrogeological principles was required to predict, control, and possibly optimise the pressure relief system. An analytical model was used in this project to carry out the simulations. This model was developed by OGI based on CIRIA Report 113 (1986) for temporary groundwater control, based on standard well hydraulics theory for porous medium, with modification taking into account interference between individual wells (Power,1992). This predictive model was implemented in an Excel spreadsheet which allows manual control of input parameters as new measurement data becomes available during the drilling and installation.
Two specific issues were examined at this stage for the proposed design of the pressure relief system described above. First, modelling simulations were applied to examine the system capacity of the preliminary design for pressure relief, ie the maximum ambient pore water pressure in the sand layers of the Lambeth Group that the system could cope with in order to provide sufficient pressure reduction. This was calculated for systems that were both passive and active. The results provided the critical threshold values to assist the site supervising engineer to make a decision as to whether additional pressure reliefwells would be required for the conditions encountered on site.
The model was also used to predict the piezometer responses to the excavation process. These were used to identify the critical excavation levels at which active pumping would be required for the assumed ambient pressure levels.
Rotary auger and cable percussion drilling were used to install piezometers and pressure relief wells (Figures 5 and 6). Drilling started from the base of the cofferdam at -9. 25m OD, about 4m below the riverbed outside the cofferdam, and terminated at depths of about -30m OD for the piezometers and about -37m OD for the pressure reliefwells.
Vibrating wire piezometers were chosen to monitor the pore pressure changes within the Lambeth Group. These transducers were robust and had an instantaneous response to the pressure changes within the measured layers, essential for real time risk assessments for the management of the excavation. The transducers were installed to a depth ofabout -29. 5m OD with a 3m sand cell surrounding the piezometer sensor. The piezometer cables were connected to a data logger (Datataker) (Figure 7) on the access bridge. The data logger was set up to record the pore pressures as measured by the piezometers. Recorded data was stored and subsequently downloaded, or directly displayed on a notebook computer as excavation proceeded. Four piezometers were installed in each cofferdam at the locations shown in Figure 4.
Four 150mm diameter pressure reliefwells were installed in each of the shafts to a depth ofabout 28m from the base of the cofferdam. The wells were installed with varying lengths of 75mm diameter well casings and screens depending on the precise depths at which the sand layers in the Lambeth Group were encountered. Each of the wells consisted of6m long upper plain casing with the lower remainder being slotted. The top of the wells was set at -20m OD, about 11m below the base of the cofferdam. Filter sands were installed in the annular space between the well screen and the borehole wall and in the open hole above the top of the well casings. In this way the installed wells would facilitate passive pressure relief for excavation above -20m OD, while leaving options for pumping below -20m OD if necessary. The 6m long plain casings were designed to protrude through the proposed concrete base plug and allow upward extension of the well casing to facilitate the first concrete pour within the shaft without losing the pressure relieffunction ofthe wells.
Following the installation ofall pressure reliefwells, revised analytical modelling was carried out to predict the piezometer responses to the excavation, based on the data collected during the installation and the actual positions ofthe wells. Although it was recognised that the actual pressure was changing with the tide, for the purpose ofmanaging the risk of uplift only the maximum pressure at the high tide was needed for the predictive studies. The results of this model prediction (Figure 8) clearly show the pressure reductions at the piezometer locations in response to the excavation.
For the range ofambient piezometric levels measured (Figure 9), it was considered unlikely that pumping would be required. This model was also used to assist in the interpretation and understanding of the actual pressure responses measured on site during the excavation works. Any unexpected changes in pore pressure during the excavation works could be identified promptly in time to initiate appropriate actions. Also shown in Figure 8are the calculated safe excavation levels with and without passive pressure reliefwells. Alert and action levels are defined below.
The vibrating wire piezometer measurements were logged at 30-minute intervals throughout the construction of the foundations. This data was converted into readable engineering units during logging and stored in the Geologger 515 system (Figure 7).
The data could be downloaded for inspection at any time using a portable computer when required. It was also possible to obtain measurements in real time by issuing Datataker commands directly from the computer without disrupting the running ofthe logging program.
This system provided the real time pore pressure data for the observational approach to risk management. It also provided the opportunity to monitor the trend of uplift pressure changes during excavation, allowing prediction ofpossible adverse effects ofpore pressure on the safety of the excavation.
In order to simplify the decision-making process for the site managers without the need to understand the hydrology involved, two critical levels were set up by engineers to provide risk control for the management of excavation works. The 'alert' level was defined as the piezometric level at which the weight of the soil below the excavation formation was equal to the uplift force generated by the pore water pressure in the sand layers of the Lambeth Group. This level was set at - 18. 2m OD (ie equivalent to a piezometer reading ofabout 110kPa) based on the borehole logs and piezometric data.
The 'action'level was defined as the piezometric level at which the uplift force on the soil below the excavation formation was equal to the weight of the soil plus some of the shear resistance of the London Clay. This level was set at -15. 5m OD (ie equivalent to a piezometer reading of about 140kPa). For safety reasons, the shear resistance ofLondon Clay used only took account of20% of the measured shear strength of the London Clay across a single shear plane. If the measured piezometric level exceeded the action level, pumping of the installed wells would be activated, or alternatively, excavation could switch to the adjacent shaft to activate additional pressure relief wells. The latter method was used during construction of the shafts in the north cofferdam (Figure 10).
The alert and action levels adopted were considered conservative and therefore safe. The risks associated with the base failure were onerous as softening of the soils at and around the base of the shaft would occur with ingress ofwater. Reassessment of the soil properties and the consequent redesign for the pier foundations would cause significant contract delays and additional work in deepening the shafts or provision ofalternative piled underpinning to the shaft.
The results ofthe piezometer measurements for the south cofferdam are shown as P1, P2, P3, and P4 in Figure 9and for the north cofferdam as P5, P6, P7 and P8 in Figure 10. Tidal data shown at the top of these figures was obtained from the Port of London Authority and were measured at Westminster Bridge (upstream of the site) and Tower Bridge (downstream) over the period of concern, respectively. Comparison of the piezometer data with the tidal data suggests that the pore pressure in the Lambeth Group was closely related to the tidal heights. It was apparent that the pore pressure changes reflected an instantaneous elastic response occurring with the tidal variations.
Evidence ofpressure reliefduring excavation is clearly shown in these figures as indicated by the pressure reduction in the piezometer measurements.
The London Millennium bridge project presented an excellent case study for the management of risk relating to ground uplift during construction. It demonstrated that a risk management approach was a viable and cost-effective means to mitigate pore water pressure problems during construction works, particularly when there were uncertainties with regard to the ground conditions.
The project also demonstrated that proper application of recent technologies by a competent engineering contractor could save both time and money for construction projects. An essential ingredient to the success of this project was the close co-operation between the contractor's and specialist subcontractor's design and management teams and the integration of these teams within the construction management organisation.
It is suggested that further research is required to provide adequate guidance for engineers in choosing an appropriate level of plug shear strength, contributing to an overall factor of safety against base failure in shaft construction in overconsolidated fissured clay formations.
The measurements obtained validated the predictive model ofpore pressure responses to the excavation works. The spreadsheet presentation of the model has proven to be a useful tool in this project for the management of the risk of ground uplift. Because of its simplicity in comparison with a sophisticated numerical model, it allows real time, on-site assessment of pore water pressure conditions while the drilling and installation are in progress. The model proved sufficiently accurate in prediction of the magnitude and behaviour of the pore water pressure reduction during excavation (comparing Figure 8with measured data in Figure 9). Further future improvements may include the capability of modelling the effects of real time tidal variation on the pore water pressure at depth.
Ground Engineering (2000). Poetry in motion, Volume 33, Number 1, January 2000.
Ove Arup & Partners (1998). Geotechnical interpretative report, the London Millennium bridge, November 1998.
Powers JP (1992). Construction dewatering - new methods and applications. John Wiley & Sons.
Somerville SH (1986). Control of groundwater for temporary works. CIRIA REPORT 113.
Simpson B, Blower T, Craig RN and Wilkinson WB (1989). The engineering implications of rising groundwater levels in the deep aquifer beneath London. CIRIA Special Report 69.
The authors wish to express their appreciation to the London Millennium bridge project team for allowing the publication of this paper, to Sir Robert McAlpine and OGI for providing the monitoring information. To J Bartley and JP Welch for their assistance in drilling and installation. Special thanks also to Stephen Thomas and Steve Hoare for their contributions to the formulation and design of these temporary works and providing invaluable support in many areas of the project.
Soil Instruments provided the DATATAKER logging system. Newcastle Mining and JB Site Investigations provided drilling rigs and crews. Site installation of the temporary work was supervised by Fajin Yuan and managed by Ken Foulds (formerly of Sir Robert McAlpine) in association with Yuan and Steve Hoare of OGI. Special thanks also to A Brown of Sir Robert McAlpine for his hard work in downloading and processing of data while construction works continued.
At conceptual stage the input of Jim Pontin of Stuart Well Services is acknowledged. David Norbury of CL Associates provided valuable advice and Soil Mechanics are thanked for their prompt services in site investigation work.