Thames Tideway Tunnel: Blackfriars Bridge Foreshore: innovative in-river construction Open Access

London relies on a Victorian combined sewer system which, although in excellent condition, was too small to serve its nine million inhabitants. When it rained, tens of millions of tonnes of untreated combined sewage were being discharged into the River Thames every year.

The Thames Tideway Tunnel project addresses this by intercepting or managing flows from 34 combined sewer overflows (CSOs) located in the heart of the capital and safely transferring the flows, by way of new drop shafts and tunnels, to Beckton Sewage Treatment Works. The new 25 km tunnel system of predominately 7.3 m internal diameter, starts in Acton, West London and follows the approximate alignment of the Thames before it connects to the existing Lee Tunnel in Stratford, East London. It is the UK’s single greatest intervention to tackle combined sewage pollution.

Weaving its way through a congested city, the safe and successful construction of London’s deepest tunnel system required 21 worksites, each with unique design and construction solutions to address challenging constraints both above and below ground. The safe operation of multiple 24/7 tunnelling sites and interception at multiple CSO locations in a densely populated metropolis, especially during a global pandemic, was never going to be easy. Sensitive stakeholders required the project teams to think differently and challenge usual solutions to provide innovative mitigation measures, minimising the impacts of the work.

This paper describes the unique challenges associated with the interception of the Fleet Main CSO and the Northern Low Level No1 (nLL1) sewer at the Blackfriars Bridge Foreshore site (Figure 1) and focuses on the bespoke solution implemented for safe and efficient construction of the hydraulic interception culvert.

All the CSO interception sites on the project presented unique challenges in highly constrained locations and the Blackfriars Bridge Foreshore site was no exception. The foreshore-based construction site was surrounded by the following key stakeholders and critical infrastructure.

• It was adjacent to the upstream side of the listed 1869 Blackfriars Road Bridge.

• Sir Joseph Bazalgette’s original Fleet Main CSO discharged to the river beneath the northern arch of the bridge.

• On the north of the site boundary, was the Bazalgette Grade II listed wall which retains the Victoria Embankment, behind which lies the 2.54 m dia. brick nLL1 sewer and an upper 2.74 m brick service subway. Critical services within the subway included low-pressure and medium-pressure cast-iron gas mains, each 36 inch (914 mm) in dia., electricity and strategic fibre optic communication links.

• The shallow depth of the London Underground Waterloo and City lines (located approximately 8 m below foreshore level) had to be delicately worked around during drop shaft construction and CSO interception works.

• The Blackfriars Millenium Pier had to be relocated to the downstream side of the Blackfriars road and rail bridges as part of the initial site enabling works.

An aerial view of the site area before works commenced is shown in Figure 2.

Interception of flows from the Fleet Main CSO avoids approximately 521 000 m³ of untreated combined sewage being discharged into the tidal River Thames in a typical year (Thames Water, 2013). The location is one of the larger CSOs intercepted on the Thames Tideway Tunnel project. During design development, wider hydraulic modelling of the London catchment identified that, by diverting flows from the nLL1 sewer at the Chelsea Embankment Foreshore, the Victoria Embankment Foreshore and Blackfriars Bridge Foreshore sites would free sufficient capacity in the network to manage flows without having to physically intercept them at ten other CSOs, avoiding a further 2 138 000 m³ being discharged into the tidal Thames. The design flow rate for the Fleet Main CSO interception was 46 m³/s and the design flow rate for the nLL1 sewer was 15 m³/s.

Intercepted flows from the original outfall flap valves located beneath the northern arch of Blackfriars Road Bridge arch are turned 90° to run parallel to the river wall. They are then passed through a series of new connection culverts and hydraulic chambers before transferring safely from the upper and relatively shallow level of the existing sewer to the deeper level of the main tunnel, which is approximately 53 m deep. Transfer of CSO flows is by way of a tangential vortex intake above a 3.1 m dia. stainless steel lined vortex tube housed within a wider 22 m dia. drop shaft that is situated on the alignment of the 7.3 m dia. main tunnel.

A separate interception arrangement diverts flows using a weir structure from the nLL1 sewer located within the Victorian river embankment by a series of separate hydraulic chambers into a secondary vortex intake and 2.3 m dia. vortex tube structure located within the drop shaft. Figure 3 shows a cross-section of the original river wall at the location of the nLL1 sewer interception and a photo of the river wall during demolition prior to construction of the interception structure. Both vortex drop tubes discharge into separate stilling basin structures located on the perimeter of the shaft to enable vertical deaeration of flows, before being combined with other flows in the main tunnel (see Fricker et al. (2025) in this issue).

The Fleet Main CSO interception works included a relocated CSO outfall to allow flows to discharge to the river should the new Thames Tideway Tunnel system be unavailable on rare occasions, the system becomes full or is not available due to planned maintenance. A double set of new river flap valves provides security to the tunnel and the new hydraulic infrastructure by providing protection from tidal inflow on the relocated outfall.

Figure 4 shows a schematic diagram of the worksite arrangement, showing the locations of the nLL1 sewer overflow weir chamber and the Fleet Main CSO interception chamber in relation to existing structures and the new CSO drop shaft and main tunnel.

A significant challenge presented to the design and site teams, which needed an innovative and buildable solution, was how to construct the Fleet Main CSO interception culvert. Construction of the interception culvert in its final location within a more traditional dry cofferdam was investigated but assessed as not being feasible for several reasons, including the following.

• Damage assessments showed building a twin-wall cofferdam for the whole worksite would cause an unacceptable amount of movement to the gas mains located within the service subway.

• The eastern end of the culvert needed to connect to the old CSO outfall located beneath the restricted overhead clearance of the northern arch of Blackfriars Road Bridge and in unacceptable proximity to the adjacent foundations of the northern support pier of the bridge.

• It needed to be situated over the shallow London Underground Waterloo and City line tunnels, which are at a depth of 8 m beneath the foreshore approximately 60 m to the west of the old CSO outfall, thereby preventing installation of sheet piles over and in close proximity to this asset.

• The river wall toe of the existing listed river wall embankment further restricted the available width to construct the cofferdam and fulfil hydraulic size requirements for the culvert between the river wall and the bridge pier.

Further design work for the interception arrangements and associated cofferdam configuration also impacted the construction programme of the drop shaft and tunnelling across the wider project. The shaft was on the critical path for the project as it was to be utilised as an intervention location for one of the tunnel boring machines (TBMs) constructing the main tunnel before it transited beneath London’s iconic Tower Bridge. This meant that a delay to the cofferdam completion and subsequent shaft construction would delay TBM launch and have significant wider programme implications.

Although beyond the scope of this paper, the drop shaft solution developed by the team was to split the overall cofferdam with a smaller twin-wall cofferdam serving the shaft footprint area and to only sink the shaft part way to a level with sufficient cover to the crown level of the intended main tunnel drive beneath. The construction methodology of the drop shaft’s primary walls was changed from the intended full-depth diaphragm walls to a combination of upper secant walls and lower sprayed concrete lining (SCL).

Some of the deepest jet-grouted columns in the UK (up to 52 m) ensured watertightness during SCL construction and formed a cutoff ring to allow a successful TBM head intervention in what would have been permeable ground conditions and groundwater pressures of up to 5 bar. Once the tunnel drive east to Chambers Wharf had been completed, the tunnel was filled with foamed concrete and SCL was used to complete the remaining lower section of the shaft around the completed tunnel.

The chosen solution for the interception culvert, taking all these constraints into account, was to construct a concrete box culvert (approximately 96 m long and weighing 3700 t) in a cofferdam located upstream of its final position and then float this as a single ‘precast’ structure into its permanent position beneath the bridge arch (Figure 5).

Although unconventional in nature, building the culvert as a large concrete vessel and floating it into position provided an innovative solution to the site constraints, but it also raised a whole new set of challenges for the team.

The long box interception culvert would be part constructed as a precast shell within a temporary single-skin, tubular pile cofferdam located to the west of the Waterloo and City line London Underground tunnels and connected to the drop shaft’s cofferdam. Once the culvert box was constructed, this cofferdam would then act as a dry dock, enabling ‘launching’ of the culvert to float it into its permanent position onto a ground bearing foundation on a prepared rock and concrete mattress on the riverbed. After completion of the floating operation, the cofferdam would be resealed and used for the construction of the Fleet Main CSO and nLL1 sewer valve chambers, finally having the shared wall with the shaft cofferdam removed and completing the remaining structures.

The design, the moving and subsequent placement of the culvert were undertaken by several parties, each working in very close collaboration. The permanent works designer (Aecom) of the main works contractor for the central section (joint venture of Ferrovial Construction and Laing O’ Rourke (Flo)) designed the culvert for the permanent load cases. Temporary loading conditions for the part-constructed culvert during construction in the cofferdam and subsequent float-out and immersion operations were undertaken by Flo's temporary works designer, Robert West.

The culvert geometry design was set by several criteria, including external site constraints and hydraulic flow requirements. The long, narrow structure has a rectangular cross-section along its western half, transitioning to a trapezoidal cross-section in its eastern half to align with the profile of the existing river wall. The narrow profile of the culvert would also minimise permanent river encroachment while assisting the hydrodynamics during the culvert flotation.

The permanent geometry of the culvert was a closed, sealed structure which needed to provide sufficient weight to be stable on the riverbed and prevent the structure becoming buoyant. The restricted overhead clearance of the bridge, however, required the culvert to have a temporary layout without upper wall sections and roof slab to enable suitable space between the structure and the bridge arch when floated into position at high water. This temporary U-shaped geometry also made the culvert lighter and more stable for the float-out operation. A temporary steel parapet structure was installed above the culvert walls to provide freeboard for a dry working area up to flood defence level above high spring tide.

Different wall thicknesses (700 mm for the inner wall and 750 mm for the outer wall) were developed to facilitate the transition to the permanent works and ensure stability under flotation. Cast-in bolts and additional reinforcement for temporary bulkheads, props, CSO outfall gates, a lifting gantry and a parapet structure were also incorporated to coordinate the temporary and permanent designs.

The concrete class for the structural elements of the culvert was C40/50. The concrete cover was determined to be 60 mm with an additional 25 mm applied to the internal elements for protection against hydrogen sulfide corrosion and abrasion for the 120-year design life. Reinforcement was typically B32 bars at 150 mm centres throughout the base, walls and – once in permanent position – the culvert’s roof.

The durability crack width limit for serviceability limit state calculations was 0.3 mm for flexure crack control (in accordance with clause 7.3.4 of BS EN 1992-1-1:2004 (BSI, 2004)) and early thermal crack control, with through-crack limits interpolated between 0.05 mm to 0.2 mm based on the hydraulic gradient ratio for watertightness according to clause 7.3.1 of BS EN 1992-3:2006 (BSI, 2006a).

All the elements were designed for early thermal cracking to satisfy watertightness class 1 in accordance with standard procedures (Bamforth, 2007; BSI, 2006a). Re-injectable grout hoses were installed at construction joints and used to pressure test and measure the watertightness of completed works before and after the float out. These hoses were then used for resin injection of any joints where the pressure test was shown to be below expectations for water seepage.

The culvert was designed to resist ship impact loads where there is no protection from the bridge pier, based on probabilistic impact force modelling for a 1 in 10 000 years return period. Various vessels were modelled to determine impact forces, and the refinements made included estimating a reduction in impact velocity when propeller failure is the failure mechanism. This was in accordance with clauses C4.3 (5) (inland vessels) and C.4.4 (2) (sea-going vessels) of BS EN 1991-1-7:2006 (BSI, 2006b). A dynamic amplification factor of 1.7 was used in the final design with loading of 13.6 MN (frontal) and 5.4 MN (lateral).

Buoyancy calculations for the structure in the temporary floating case and the final position were developed between the permanent works and temporary works designers. The base slab was designed to be 900 mm thick for the permanent case but, to achieve floating buoyancy, part of this structure was left out until after placement in its final position. A 350 mm thick slab was maintained for stability during the float out and placement, with temporary reinforcement added for the temporary condition (Figure 6).

Detailed calculations for buoyancy and an assessment of the marine stability of the structure were also reassessed during construction to incorporate as-built geometry. A detailed three-dimensional (3D) model of all permanent and temporary works was used to assess weights and the centre of gravity. Density measurements from 121 samples of the as-built structure were used to confirm concrete mass.

Due to the geometry of the culvert, there was eccentricity between the centre of gravity of the structure and the centre of buoyancy, so the culvert would tend to rotate with one end being deeper than the other until water pressures at the base balanced the system. To overcome this imbalance, ballasting was used to counter the rotation forces. A combination of concrete blocks and polystyrene buoys were used to limit rotation roll and pitch to below 0.7°. Buoyancy testing trials were undertaken before the float-out operation.

The permanent design was carried out with 3D finite-element (FE) modelling using Lusas Bridge Plus, with various elements of the structure analysed using a 3D thick-shell model. The geotechnical capacity of the foundation was checked with a 3D Geotechnical FE Plaxis model to provide realistic soil stiffness input for the 3D Lusas model. Geotechnical parameters were defined in a geotechnical interpretive report produced by the designer based on extensive geotechnical investigations carried out by the client prior to contract award.

To minimise the risk and enable a fuller understanding of the float-out procedure, physical hydraulic modelling was undertaken for the floating operations. This informed the design and strategy for the float-out operation in order to evaluate, validate and optimise the proposed installation manoeuvre. Successful modelling required

• reproducing the environmental conditions of the River Thames – the tidal cycle (currents and river water level), wake waves and wind loads

• building a reduced-scale mock-up, including all the elements involved during the deployment operation

• reproducing a reliable culvert positioning control system that replicated the real dynamics of the manoeuvre.

A 1:25 scaled simulation basin was built, taking into account the unique site conditions within the area. The physical model was built at a specialised metal workshop at the testing facilities of IHCantabria, in Cantabria, Spain. The basin represented a half-width section of the River Thames and reflected the dimensions of the manoeuvre and the space available at the hydraulic modelling facility (Figure 7).

The proposed float-out operation was designed to be executed with three towing/pulling lines (two stern and one bow) connected to three independent winches placed inside the culvert. An additional line at the bow was installed as a contingency measure, although the use of this line was not envisaged during the float-out operation.

Three different line manoeuvre control strategies were implemented

• pre-defined mooring line loads as derived from an OrcaFlex simulation, although winch control did not consider the culvert path affected by environmental loads

• manual operation of the winches considering the theoretical culvert path

• autonomous operation, with the loads in line three automatically modified to follow the theoretical culvert path.

The culvert installation manoeuvre was evaluated under different environmental load cases. To cover all possible scenarios during deployment of the culvert, the load cases employed for the simulation combined the effects of tidal level, river currents, wind loads and wake waves (generated by a police boat). The physical modelling was divided into three blocks

• constant level tests

• variable level tests: simulation of the River Thames tidal cycle

• set-down test: simulation of the culvert set down.

Monitoring of the culvert movement was recorded using the track motion system Qualysis, while the culvert accelerations were recorded by an IMU (accelerometer). Winch lines were equipped with axial load cells to track evolution of the mooring loads. Ultrasonic wave gauges and acoustic Doppler velocimeters (Vectrino) were installed in the basin to monitor the environmental loads.

The geometry of the mock-up was obtained by applying the Froude scaling laws of similitude with the test basin divided into three principal sections

• effective testing area – the central part of the basin, 20 m long, 4 m wide and variable water depth (0.1–0.5 m)

• auxiliary tank – placed parallel to the testing area to communicate through a set of pipes and pumps (two pumps in parallel with a maximum flow of 140 l/s) to modify the water depth of the testing area and simulate the tidal cycle

• instrumentation working area – basin installed just behind the river wall to house all the measurement devices.

Wind loads were simulated using a set of 20 portable fans that were placed in front of the testing area. Generation of wake waves was performed using a remote controlled boat and recorded by an array of free surface sensors. The winches were simulated using a set of stepper motors that allowed adjustment of the length of the lines, simulating the effect of the real winches.

Mooring lines were simulated using a nylon fishing line, which in model scale compared closely to the weight of the full-scale mooring lines. At laboratory scale, it was important to accurately reproduce the axial stiffness of the lines (bending and torsion stiffness were considered negligible), so a calibrated spring was introduced at the end of the line.

The main conclusions obtained from the physical testing that assisted the float-out planning were as follows.

• The theoretical path from the cofferdam to the bridge pier was proven and validated.

• The manoeuvre was carried out without evidencing the need for large pulling loads.

• From tests with currents (constant or variable water depth), increasing the current velocity induced separation between the culvert and the river wall. To avoid this phenomenon, it was necessary to increase the mooring lines’ pre-tension.

• Wind tests (wind direction perpendicular to the river) showed the culvert’s centre of gravity was slightly shifted from the theoretical one. The final culvert position was rotated around −0.5° (yaw). To avoid the final rotation, auxiliary mooring lines should be added during the set-down operation.

• For the set-down operation, a fourth mooring line would improve the performance and the safety of the operation.

• A set of accidental scenarios was reproduced to assess potential risks to the manoeuvre, replicating potential real emergencies happening during the float-out operation. The system showed robustness since the culvert was safely moored during all the different accident scenarios.

Preparatory work to facilitate the culvert float-out operation included dredging, partial demolition of the existing river wall toe and the installation of a gravel bed to create a suitable foundation to receive the culvert.

The river wall to the north of the floated culvert was constructed during the 1960s as the embankment underpass was constructed. The existing river wall varied in geometry and a section of river wall toe (40 m long, 2.3 m wide and up to 1.8 m deep) clashed with the location of the new culvert and had to be removed.

The existing riverbed was made up of mud, gravel and rocks (both natural and debris) that was removed during dredging to create a flat and level platform. On top of the dredged riverbed, a layered gravel bed 400 mm thick was placed, with D₅₀ = 25 mm for the initial layer and D₅₀ = 10 mm for the top layer. Near to the old Fleet Main CSO outfall, an overlying 100 mm precast concrete mattress was placed to prevent scouring in case of a CSO discharge before the float-out operation was completed.

To validate the design assumptions and the physical modelling results, and to ensure the culvert was watertight and behaved as expected when buoyant, a number of flotation trials were undertaken.

Following naval stability principles, key success criteria for the flotation trials were to confirm the as-built culvert’s stability and tilt, and ensure the culvert avoided excessive pitch and roll. To achieve this, the cofferdam was flooded, and a series of trials was undertaken before the cofferdam wall was removed for the final floating manoeuvre.

Water management to control filling and draining of the culvert and cofferdam was achieved through a pre-installed system of valves and 152 mm (6 inch) intake pumps, located along the length of the cofferdam and culvert walls (Figure 8). Environmental impacts were an important consideration and mitigations such as including eel screens into the valve design and facilitating the removal of potential contaminants to the river environment before flooding were implemented.

The first phase of testing involved pumping water into selected compartments of the culvert to provide ballast. This formed part of the temporary works assessment for the method of sequential compartmental filling and testing of the culvert’s temporary steel bulkheads. Where minor leakage was identified, waterproofing was used to reseal in advance of subsequent testing.

The second phase flooded the culvert to a higher level, to test the sealing of the temporary structural steel parapets. This was carried out by flooding individual culvert compartments to test the steel bulkheads sequentially (Figure 9). In a reverse process, water acted externally to test compartments where the rising tide reached a level within 100 mm of the top of the concrete. Water tests were typically held for 12 h.

The final series of tests focused on the structure’s flotation. Pre-positioned draft plates were fixed externally in pairs at starboard and portside along the length of the culvert, as well as at the stern and bow. Once maximum mean differences in pairs of readings were deemed within an acceptable limit (about 50 mm), stability and tilt values had achieved the design intent. In addition, a trim of 0.1–0.2° was required to avoid local disturbance of the gravel and avoid any ‘aquaplaning’ effects during set down of the culvert. Concrete kentledge blocks provided the primary means of adjustable ballast, and several configurations were tested before the optimum results were obtained.

When all the flotation tests had been successfully completed and all relevant parties were comfortable to proceed further, the cofferdam was opened to the river to allow the final float out of the culvert. A section of the tubular steel piles forming the cofferdam was removed by wire saw cutting, thereby flooding the cofferdam and enabling the passage of the culvert into open water (Figure 10).

Although the actual distance the culvert would travel was relatively small (approximately 100 m between the temporary dry dock cofferdam and the permanent position), the level of preparation and associated checks undertaken to ensure a safe passage was considerable.

On the Bank Holiday weekend of August 2020 – amid the worldwide Covid-19 pandemic – the tides and weather forecasts were favourable, and the decision made to proceed with the move. Over the long weekend, the culvert was moved in a two-stage operation.

The first spring high tide on 30 August enabled a sufficient tidal range to winch the culvert the first 20 m of its journey through the cofferdam and into open water. Over the period of the high tide, a series of three winches attached between the culvert and bespoke reaction frames located on the existing river wall carefully eased the culvert on its maiden journey. Visual and electronic monitoring ensured the lateral, tilt and rotational movement of the culvert was maintained within anticipated and agreed limits. As the tide fell, the culvert was temporarily seated on a pre-prepared granular riverbed platform that acted as a suitable support for the culvert at low tide. A photograph of the first stage of the flotation is shown in Figure 11, with two blue winches attached to the end of the culvert visible in the foreground.

The low-tide period enabled all parties to pause and review how the initial flotation had gone and raise any concerns at a debrief session. Once all parties were comfortable to proceed, the decision was made to conduct the second stage of the move on the subsequent high tide.

Following the successful first stage, over a period of less than an hour in the darkness of the early hours of Bank Holiday Monday, the culvert was then traversed the remaining 80 m of its journey to its permanent position in front of the old CSO outfall beneath the bridge arch. On the falling tide, the culvert was guided using a positioning frame and guide wheels before it became supported by the prepared riverbed. Once the team had established that the culvert was in the correct and specified location, the culvert was ballasted by pumping in river water.

With the culvert successfully located in its permanent position, the teams were able to progress on two specific areas.

The culvert was permanently restrained to prevent future movement from unintentional buoyancy. This included the culvert being ballasted with an additional 300 mm layer of concrete internally across its length as the water was pumped out. Once dry internally and isolated from river ingress, the subsequent steps of vertically extending the culvert side walls and completion of culvert structural elements, including additional internal division walls and roof slab, were then able to commence.

After completion of the main structure, the seal around the old CSO outfall was completed during dry weather windows (Figure 12). The old river flaps valves were then removed. The temporary openings and temporary flap valves on the end of the culvert were sealed after the main tunnel was commissioned to receive flows.

Meanwhile, another site team re-established a dry area within the cofferdam by installing steel bulkheads connected to pre-fitted channels on the sides of the floated culvert across the opening in the cofferdam wall. These bulkheads were then sealed to the existing cofferdam structure, allowing for the final drain down. Once secure and dry, this enabled construction of the remaining section of the Fleet Main CSO and nLL1 sewer interception structures. Figure 13 shows the interception culvert in its final location adjacent and beneath Blackfriars Road Bridge (right), the cofferdam for the remaining interception structures (middle) and the drop shaft in its cofferdam (left).

The constraints of the Blackfriars Bridge Foreshore site presented the team with a unique set of challenges that required an innovative and collaborative approach to problem solving.

Open and honest communication between all parties was fundamental to ensuring the safe and timely flotation of the 3700 t concrete interception culvert. Works carried out by the Blackfriars Bridge Foreshore team was supported by the wider project’s ‘do it safely or not at all’ ethos for everyone, irrespective of training, experience or role. This enabled all those involved with the culvert flotation regardless of their position, to have the confidence to raise concerns or challenge the accepted.

Now operational, the new super sewer complements the original Victorian sewer network designed by Sir Joseph Bazalgette and provides protection to the central London sections of the River Thames against combined sewage discharges for current and future generations of Londoners.

The river-based site has also enabled the project to leave a new permanent riverside public space as part of its long-term legacy (see Donnelly et al. (2025) in this issue). The space, formally named Bazalgette Embankment, was the location for the formal operational opening of the wider Thames Tideway Tunnel project by His Majesty King Charles III on 7 May 2025. An aerial photo of the site as it neared completion is shown in Figure 14.

Although hidden from view, the new 120-year design life combined sewer system will provide a vital function beneath Londoners’ feet to ensure a cleaner river in the twenty first-century and beyond.

The authors would like to acknowledge Allen Summerskill, Programme Completion and Closeout Manager, Jacobs, Tideway, London, UK for his valuable contribution to this paper.