On 10 April 2006 the pipeline supplying the Queen Mother Reservoir failed spectacularly. Local residents reported seeing plumes of water reaching 7m in the air just outside Datchet, Berkshire. Fortunately no one was injured, particularly since it occurred just to the side of a busy road.

The failure occurred in a 2.54m i.d wedge-block tunnel, 27m deep in London Clay. This inlet tunnel, built in 1967-9, has a nominal capacity of 900 megalitres/day, at pressures up to 5 bar, and serves the second largest raw water storage reservoir on the Thames Water network. With the risk of drought and water shortages looming it was critical that the tunnel be brought back into service quickly, ready for the autumn refill period. The deadline imposed by Ofwat, the UK water regulator, was 29 September, leaving just 26 weeks to carry out the repairs.

Mobilisation

Thames Water (TW) invited Mott MacDonald to assess the situation and to propose suitable options for carrying out the repair. A team of design and construction engineers from Thames Water Engineering, Mott MacDonald, Barhale Construction and Amec Construction was established in site offices adjacent to the collapse site at the Datchet Raw Water Pumping Station.

Site constraints

The collapse location was adjacent to the B376, Horton Road. Due to the size of the crater this road was closed, and a local diversion was installed around the site. In the road was a range of sensitive services: 30” high-pressure gas main, eight fibre optic cables, 33kV and 11kV electric supply cables, traditional copper telephone lines, and a 4” water main. The water main was damaged in the incident and a temporary diversion was installed.

The Queen Mother Reservoir embankment was only 60m away from the collapse. After initial assessment it was decided that this was not close enough to put the reservoir at risk. However, several of the options for the repair would have required deep excavations within the zone of influence of the embankment. This put an additional requirement on the site team to gain approval from the statutory inspector of the reservoir on the proposals for the repair.

The water level in the crater stabilised 1m below ground level (the local ground water level). The depth of this crater was found to be approximately 3m. The area was then backfilled to allow the ground investigation equipment better access.

Initial investigation

Information regarding the tunnel’s condition was limited. Aside from the 15m wide crater, there was a large amount of gravel and clay scattered around the site. Additionally the adjacent road had settled noticeably and significant cracking had occurred.

A submersible remotely operated vehicle (ROV) was used to investigate the water-filled tunnel. At this point it was not possible to pump out the tunnel as there was a direct connection between it and the upper aquifer. The ROV was used from the shafts at either end of the tunnel, and a video feed allowed the team to assess its condition. The fine material suspended in the water made this assessment difficult; however no sections of damaged tunnel were discovered. Near the position of the failure the tunnel was filled with what appeared to be gravel. Based on the length of the cable paid out when the ROV was at the gravel, the tunnel appeared to be blocked for some 40m (though this turned out to be inaccurate and no more than 7m). This evidence forced the team to consider options for dealing with anything up to 40m of collapsed tunnel, and the resulting disturbed ground above. Proposals discussed ranged from a 75m x 20m top down excavation, to the abandonment of the tunnel and installation of a new pipeline over the reservoir embankment.

Ground investigation

A major ground investigation was undertaken in and around the collapse. The primary purpose of this was to establish the extent of the disturbed ground, and how much tunnel was damaged. The testing consisted of rotary and cable percussion boreholes, cone penetration tests, resistivity imaging, microgravity tests, ground probing radar, and seismic cross-hole testing. Results from the geophysical methods were disappointing, giving no clear evidence of the extent of the disturbed ground.

The most informative ground investigation was done using the CPT rig. Once the crater had been backfilled the rig was used to test two lines of points, at 1.5m centres, at right-angles across the centre of the crater. All but one of these tests indicated typical stratigraphy and undisturbed London Clay. The non-standard result, and the result of a follow up test carried out 1m away, showed disturbed ground/gravel to the full extent of the test (~20m). A third line of tests was carried out along the centre of the road to show whether the surface settlement was due to a deep level disturbance, or to water undermining the road whilst escaping to the nearby brook. All these tests showed typical strength profiles in the clay. The CPT results all indicated that the failure was limited to a small zone within the crater. This gave the team a high level of confidence that the disturbed ground could be contained with the proposed 9m recovery shaft.

Option selection

All options investigated involved some kind of cut-off between the water-bearing Terrace Gravels, and the tunnel. This was necessary to prevent the upper aquifer from draining down, through the collapsed ground, into the tunnel. Since the River Thames was only 0.4km to the west of the site, the recharge rate of the gravels would have been high, preventing tunnel pump-out.

Installing a secondary lining in the existing tunnel (except in any sections bypassed) was a key element of the project. A risk assessment carried out by Thames Water recommended that the secondary lining would have to extend from the pumping station to at least the inner side of the reservoir embankment. After this point the factor of safety was significantly higher due to the water above the tunnel.

For any of the options considered it would be necessary to provide isolation between the tunnel and the reservoir. The design of the inlet tower allowed planking plates to be installed on each of the inlet jets. These, in addition to the wedge-gate control valves, provided the “double-isolation” required.

An option selection workshop was held to determine the best way forward. In this the eight most promising options were scored against finish date, programme certainty, flexibility & contingency, construction health & safety, construction environmental impact, effect on local residents, integrity of reservoir, functionality of product, cost, and understanding the cause of the failure. The two highest scoring options were the “High Level Connection” and “Shafts within a cut-off”. Other options were rejected mainly due to longer programmes, higher risk of delay or higher risk of injury to personnel.

The High Level Connection option was to abandon the gravel filled section of tunnel, and then install a sub-surface pipeline to join the existing tunnel via a new shaft on the reservoir side of the collapse (see figure 1). The cut-off would be effected by using slurry-walls or similar.

The second option, to install shafts within a cut-off would have involved sinking two shafts down onto the tunnel, one over the collapse, one closer to the reservoir. These shafts would then give access to the tunnel so that the gravel could be mined out. This option was reliant on the tunnel being in sound condition on either side of the collapse (rather than damaged and possibly unstable throughout the entire gravel filled section). Mining out a damaged tunnel would have been possible, but slow and with a greater risk of delays.

Since these options both involved sinking a 9m i.d shaft and constructing slurry walls, the decision of which option to take forward was postponed to await further geotechnical information. That meant that precast shaft rings and steel pipe for the secondary tunnel lining were procured immediately. The steel pipe needed for a sub-surface bypass was also ordered, even though they were only required for one option. The unacceptability of delay to this project meant that this sort of contingency was deemed prudent.

Programme

The 29 September deadline was onerous. Of the options considered at the workshop only two were initially programmed to finish by the required date, just 19 weeks later. In order to assess which option to choose, a detailed programme was created for each.

As the project continued the emphasis on programme remained. An updated programme was produced for each weekly progress meeting. This gave everyone the opportunity to see what was on the critical path, and to discuss what could be done to ensure the estimated finish date was still on or before the deadline. This allowed the designers to keep abreast of the requirements of the construction team.

Design

The design team consisted of two engineers working on site for the duration of the project, one site based draftsman, and several other head-office based Mott MacDonald engineers and geologists, all under the overall direction of a Thames Water project manager. The Lead Design Engineer (LDE) coordinated the design for Thames Water, and liaised with the Mott MacDonald project manager to ensure that the necessary resources were available. The author was working on site as a design engineer, and was responsible for carrying out much of the design work and ensuring that drawings were ready for issue when required by the contractors. Temporary works design (for all the works except the shafts) was done by the two contractors.

As the ground investigation progressed, a decision was made to push forward with Option 3A, the shaft over the crater, without the cut off wall (figure 1). The slurry wall equipment was kept on site so if cut-off was not achieved, there would be no real delay in changing to an alternative approach. One of the major drivers for this was the desire to avoid diverting the services in the road, in particular the 30” gas main and the 33kV electrical supply. These were on the reservoir side of the road, away from the collapse, but would have to be disrupted at least temporarily to install a sub-surface bypass.

The key factor that allowed the change in strategy was the findings from the CPT tests. Of the 40+ tests carried out only two indicated any significant disturbance to the clay. This gave the team sufficient certainty to attempt to repair the tunnel on-line.

The challenge for the designers was the limited time available. The construction programme consisted almost entirely of critical path items. So any delay in the production of drawings would lead directly to days lost on the programme. This urgency meant that efficiency and elegance in design was traded for robustness and speed.

The main elements of the design were the shafts down to the tunnel, the secondary tunnel lining, and the temporary overpumping equipment. The inclusion of a second shaft was to mitigate the risk of delay when sinking the shaft over the collapse. This was deemed likely since the forensic investigation would require time to assess the condition of the tunnel and the surrounding clay within this shaft.

The temporary overpumping equipment was a response to the drought in south-east England in summer 2006. This consisted of three 450mm o.d pipes which were laid on the ground surface over the reservoir embankment. This allowed 130Ml/day to be pumped into the reservoir whenever the river level was sufficiently high. Since several other tunnels were working at reduced pressure, this extra abstraction capacity was important to avoid water shortages.

The shaft design was carried out by the author. Both shafts were constructed using precast concrete segments and were 9m i.d. This size was chosen to ensure that the entire collapse was within the shaft, and to allow 6m lengths of steel pipe to be lowered down the shaft and jacked into the tunnel. Both shafts were intended to be temporary, i.e. would not subsequently be used as operational/ inspection shafts by TW.

However, since it would not be possible to secondary line the entire length of the tunnel within the time available, one of these shafts would remain available for a future project. The shaft over the crater (Shaft 16) was chosen for this since it was within the TW pumping station site. Although this work was planned to go ahead in the subsequent year, given the uncertainty of this timing it was deemed necessary to design this shaft for a permanent loading case.

The design of the tunnel portals required close working between the author and the contractors’ temporary works designers. In order to withstand the ground stresses at this depth in the permanent case, a large jamb frame would be required. More sophisticated options were ruled out due to the urgency of the project. The design of these elements had to take consideration of the contractors’ intended method of work.

At this stage the quality of the clay around the tunnel was unknown, so the contractors’ method included provision for timbering of the exposed face. Also, the temporary works design included corner bracings at the top of the jamb frame. These were not necessary in the permanent case, but it was desirable for the contractor to be able to install the sill after the upper section of the frame, so these members would be loaded temporarily. The design of the permanent works considered the effect these additional struts had on the frame, and web stiffeners were added to the primary members.

One of the critical issues when designing the jamb frames was determining acceptable tolerances between the tunnel and the frame. The alignment of the tunnel was known from record drawings only (one of which mis-located Shaft 13 by 2m), so the position of the shaft could have been significantly away from this line. Although it would have been possible to position the jamb frames at any point around the shaft circumference, if the tunnel had been too far off-line it would not have been possible to install the lining pipe jacking frame and allow for the 6m long pipes. The vertical position was also critical, since the jamb frames were to be positioned directly below the last complete shaft lining ring. Fortunately the tunnel was where predicted, with the shaft aligned to within 200mm of the tunnel centreline, so the jamb frames were installed largely as planned.

The base slab for Shaft 16 was designed to withstand the anticipated long-term uplift due to the unloading of the ground beneath and consequent bending moments generated. These forces were considerable due to the depth and diameter of the shaft. A 1m under-ream was used to prevent heave of the slab. This was critical since the pipeline within this shaft would be supported on a frame anchored to this slab, so adverse movements could potentially result in damage to the tunnel. The contractor requested that reinforcement in the base slab be T25 or smaller to reduce the risks from manual handling. This constraint lead to a base slab depth of 1900mm. In order to avoid excavation beneath the suspended jamb frames, the under-ream was only taken around 2/3 of the shaft circumference. Careful detailing of the rebar in these locations was necessary to ensure a full moment connection was formed.

In many instances the author had to design from first principals, useful spreadsheets or software being inappropriate or unavailable. This required careful interpretation of relevant British Standards, and forced the author to justify departures from precedence to the checking engineer at Mott MacDonald. Despite this being emergency works, Thames Water QA procedures were used throughout.

The tunnel lining pipes were 20mm thick, 6m long and of a diameter to give a 105mm annulus to the existing tunnel. This thickness was chosen to reduce the risk of damage due to the jacking forces during installation. 350m of pipe was to be installed in the tunnel. Depending on the curvature of the existing alignment, the welding was to be either done in the shafts, or in the tunnel. Once the team gained access to the tunnel a survey was carried out. This indicated that the tunnel was sufficiently straight to jack in a rigid pipe on a series of fixed rollers. The welding of this pipework was critical to the project and various alternative joint details were brought forward. The eventual method used was to fillet weld an external collar to each of the pipes above ground, then to lower the pipe into the shaft, jack it together with the preceding one, then to fillet weld the other side of the collar.

Design changes

As the construction progressed several temporary design changes were made. A key one was the arrangement at the base of Shaft 17. The design had called for the wedge-block tunnel lining to be removed, jamb frames to be installed, and a 300mm deep base slab to be cast. The use of identical jamb frames to those in Shaft 16 was overly conservative for a temporary shaft, but was justified because it simplified fabrication, and would allow good practice learnt in Shaft 16 to be used speeding up the installation. Once the contractor reached that stage of the work however, it was realised that several days could be saved if the lower half of the existing tunnel were retained. The author investigated whether this would be safe for the three-month design life required before the shaft was back-filled. Calculations showed that the expected hoop loads could be transferred without a jamb frame, providing the gap between the shaft rings and the existing tunnel was completely filled. Monitoring was recommended to give early warning of any instability. This change saved three days on the programme.

Forensic investigation

A full forensic investigation is being undertaken by Thames Water. The author has not been directly involved in this, so this paper does not attempt to pre-empt the outcome of that work. The intent of this summary is to give the reader a picture of the collapsed tunnel as it was found (figure 2).

The shaft sunk over the collapse site enabled a detailed investigation of the collapse chimney to take place. This shaft was initially sunk as a wet caisson, but once cut-off was achieved, it was possible to dewater and inspect the ground. The collapse was found to be largely circular and of 2.5m-3m diameter. Within this chimney was Terrace Gravels with occasional blocks of London Clay. The shape of the chimney was logged daily and the condition of the surrounding ground assessed. As the shaft excavation depth approached tunnel level, the exposed face of the clay was found to be discoloured in places. This rust-like colouring was increasingly frequent until the last ~1m above the tunnel crown. Below this level it was not present.

The shape of the chimney became elongated along the tunnel axis as the excavation progressed. At tunnel crown level the profile was almost rectangular and encompassed five damaged wedge-block lining rings. Four key segments and five left tops were missing, with the fifth key still in place, but having dropped significantly. One of the missing keys had been uncovered 10m above the tunnel during the shaft excavation, whilst the remaining eight were uncovered either directly above the tunnel crown or in the tunnel. As anticipated, gravel filled the tunnel completely at the opening, but was found to taper off from either side.

The tunnel was returned to service on time, on 29 September 2006.

The project team went on to win the Capital Project Management Award and the Team of the Year Award at the Utility Industry Achievement Awards 2006.

Conclusions

The decision to adopt the “shaft over crater” option was shown to be a good one, with the reliance on CPT results clearly being vindicated. The other options available would have increased the amount of construction work to do, increasing the risk of delay. Additionally the sinking of a shaft over the crater provided a great deal of useful information regarding the collapse.

Although the individual elements of the design were not technically difficult, the pace of the project made this work very challenging. To carry out and thoroughly check designs, with two demanding contractors and a Thames Water project manager in the next office, required careful time management and composure under pressure. This situation proved to be an important learning experience for the author.

The site based design team contributed strongly the success of this project. The regular discussion between the designers and the contractors’ engineers ensured that designs were ready when required and appropriate for the conditions on site. Design changes could be considered and adopted as required. The close working relationships made within the team ensured that everyone was committed to the project, and that the four firms involved cooperated fully throughout.


The crater (facing Horton Road) Jamb frame in Shaft 16 Shaft 16 base slab underream Installation of secondary lining pipe Gravel filled tunnel at collapse Harwich Plan of works Fig 1