Crossrail’s C305 Contract is in the east of London with its tunnels, forming the G-drive, running from the Limmo access shaft to the Victoria Dock portal where they terminate. Canning Town is located just north of the River Thames and the structure of concern was the flyover that carries the Docklands Light Railway (DLR). The primary objective was to keep the DLR trains running while the G-drive TBMs passed beneath the area. It was important to complete the G-drive tunnels on time as track laying operations were to start from the Victoria Dock Portal, passing through these tunnels, working westwards into central London: any delay would have held up the whole Crossrail project.
The flyover (viaduct) structure is supported by piled piers along its length and is formed in three composite sections: a 50m long central steel deck which is bolted into longer concrete end-spans either side of it. The flyover is slightly sigmoidal in plan, with the steel deck spanning the point of inflection.
The first westbound (WB) G-drive ran only just beneath the southern end of the flyover while the eastbound (EB) tunnel passed directly beneath a large proportion of the piers and so was much more challenging. Numerous bearings are incorporated into the flyover structure along with a fixed point in the central section and a movement joint at the southern end (denoted SA0).
As a consequence of its shape in plan, the bridge bows laterally as it warms up, an important aspect to appreciate in advance of the tunnelling works. Tunnelling was mostly in London Clay (and just into the Harwich Formation below it), a good tunnelling medium, although the nature of the London Clay in East London can sometimes be unpredictable and not as expected from the ground investigation boreholes.
Prior to starting the tunnel drives, various options were available to safeguard the flyover structure with varying degrees of risk and inconvenience: (i) close the DLR during tunnelling operations, providing an alternative bus service; (ii) proceed without any protection measures (e.g. jacking of the bridge piers and deck); (iii) control tunnelling-induced settlements using jacks; (iv) extend the length of engineering hours to tunnel only during this period to allow movements to be assessed before the DLR trains started running at the beginning of each day; and (v) provide some alternative alignment so the works would have no impact on the DLR. Option (i) would have been the safest and preferred Crossrail option but would not be popular with DLR or their customers. Realigning the tunnels, option (v), was not practicable. If sufficiently low volume losses could have been guaranteed then option (ii), to proceed without jacking could have been feasible but this was not the case. Extending engineering hours, option (iv), would have been inconvenient for both Crossrail and DLR.
The option chosen was to tunnel after installing a controlled jacking system to be run in conjunction with a comprehensive monitoring scheme, option (iii). The choice was strengthened by the fact that it was possible to drive the WB tunnel first, which was expected to have a much lower risk associated, thus enabling instrumentation feedback and control systems to be tested.
Pre-construction analyses
Detailed appraisals and analyses were made to assess potential movements (at a range of volume losses) to assess acceptable levels of hogging, sagging, curvature and twisting in conjunction with the viaduct bearings (for both final and transient conditions). Each step had to be agreed between Crossrail and DLR engineers as even at a 1 per cent volume loss, predicted crack widths would exceed limiting values.
Even with the jacking control systems in place there were still concerns about exceeding crack widths beyond 0.2 or 0.3mm which could lead to problems with long-term concrete durability.
After much discussion it was decided that rather than working to absolute tolerances it would be better to introduce a ‘reference line’ concept that would allow the flyover to tilt but without impacting on the structure.
Twisting and bending of the structure were of greater concern and the instrumentation set-up was designed with this in mind: relative movements were assessed rather than absolute levels. It was also considered better to focus on slope measurements, rather trying to assess other variables such as bending and differentials between slope measurements, as confidence in overall control would reduce when many of the piers are moving together both vertically and horizontally.
Therefore having good differential settlement rules was a practicable compromise.
Apart from settlement and twist another consideration was the load on the bearings, especially the lateral ones. In the case of the latter, without horizontal jacks the loads would increase by an amount that would be difficult to quantify. The horizontal jacks were therefore considered necessary, especially at the southern movement joint SA0.
Without this capability there was a risk of step-dislocation of the rails, which was essential to avoid.
Settlement predictions for a volume loss of 1 per cent indicated that the WB tunnel would cause settlements reaching a pre-defined green trigger level while the resulting trough from both the WB and EB tunnels combined, would cause settlements more than 30 per cent in excess of those assigned as the red trigger level.
The concept of considering relative rather than absolute settlements was tested in advance of the tunnelling by the ground movements induced by dewatering operations. A total of about 6mm was measured but as this occurred over a widespread area all the piers were equally affected and so the relative movements and their consequent impact on the flyover structure were negligible, despite the fact that the red trigger level was 5mm.
An assurance process using the Railways and Other Guidance Systems (ROGS) regulations was also undertaken, involving a number of steps to assess the level of risk the works posed. This suggested that there were significant risks that had to be dealt with, necessitating the Crossrail engineers to set up a system with their DLR counterparts to obtain sequential approvals for the various stages of works.
A methodology was implemented to model the response of the piled pier foundations based on the location of piles within the settlement trough. A depth of two-thirds the pile depth was chosen as a representative point: for the WB tunnel drive the pile groups of the northern piers were outside the zone of influence while some of the southern pier pile groups were just inside. Much greater interaction was to be expected for the EB tunnel drive, which passed beneath a significant number of the piers. This provided further evidence of the benefit of driving the WB tunnel first. As the viaduct is curved there were also concerns about tunnelling-induced horizontal as well as vertical displacements and so both were monitored using water levels and tilt meters respectively (judged to have sufficiently high resolution and accuracy) and equally horizontal and vertical jacks were planned. After many discussions between all parties and iterations, a trigger system was decided upon based on the available capacity of the jacks.
Initial surveys of the flyover deck and piers showed that the central two piers had settled over the four-year period of its operation and so the spans either side of the central span already has some curvature (differential slopes). Detailed predictions were made of settlements during intermediate stages of the transit to assess critical locations of the TBM.
This indicated that on completion of the transit the existing slope would reduce at the southern end, i.e., improving the situation, but that some jacking might be necessary at some of the intermediate stages. In view of this the philosophy adopted was to observe closely during the transit, knowing that slopes might improve and as well as worsen, and then to take action (after discussion) using the jacks. Therefore a passive approach was adopted rather than the original concept of controlling absolute levels to within tight tolerances.
The conclusion from all the analyses performed in advance of the works was to implement progressively the monitoring system to understand the background response of the structure (especially during the base-line period), adopt the strategy to construct the WB tunnel first and to allow some parts of the flyover to sag (improving conditions of existing slopes).
Monitoring systems and Mitigation Measures
Following detailed discussions and appraisals of the conditions on site, many modifications were made to the original desired scope of monitoring and control to provide systems that were achievable.
Very high tolerances of 0.3 mm were required from the monitoring systems, which precluded the use of traditional optical instruments such as automated total stations. Precise digital levels, with an accuracy of 0.3 mm, were therefore implemented with them daisy-chained from the north to the south of the flyover with a precise level set up midway that could sight to a deep datum, thus allowing absolute displacements to be determined along the whole structure.
Tilt sensors were installed on all the piers and so by basic geometry the absolute level of all the bearing locations could be accurately determined. Displacement transducers were also installed to monitor the distance between the top of the piers and the deck, which would allow the movements of the jacks to be monitored during the corrective jacking operations. Once the positions of the key parts of the structure were known the virtual reference line was set up, to which all relative movements were to be controlled within a set tolerance.
As the monitoring system was so key to the successful protection of the flyover structure, assurances had to be provided that it would be as reliable and fail safe as practicably possible. The cabling to all the instruments was assessed like an electrical circuit so that if any part was cut the instruments would not be affected.
Therefore back-up power supplies were provided along with redundancy in the control systems so that none of the instruments would lose power or data.
In designing the hydraulic systems for jacking careful appraisal of the flyover structure was made, assessing factors like its articulation and the forces transmitted through the bearings. Bespoke pumps were then designed to provide the necessary capacity and control, in both horizontal and vertical senses and during both individual bearing replacement and tunnelling activities. Prior to installing the jacks and after testing all the individual components, a full-scale trial modelling one of the piers was set up and the hydraulic system tested by imposing displacements and twists similar to those anticipated during tunnelling. Even factors such as realistic hose lengths of 150m were implemented (as this can affect response time). This exercise gave invaluable experience and provided confidence in the planned system.
Once installed, the vertical jacks essentially became temporary bearings (the original bearings were removed), thus supporting the structure, and they had to sustain this role for an 18-month period – an unusual requirement. In designing the jacks, in addition to the vertical forces they had to carry, consideration also had to be given to the fact that they should be able to operate when the deck is translating laterally or rotating. Locking collars were provided as a safety precaution, so that only a fixed displacement (of about 1mm) would take place if there were a malfunction of the hydraulic system. The horizontal jacks were essentially floating and were to hold the deck in position but to allow the piers to rotate.
Forewarned
As the Crossrail works were anticipated at the time the flyover was designed and constructed, anchors were cast into it to support the jacks. However, these were found to not be usable due to their location and form and so structural steelwork frames and clamping devices had to be designed and installed to apply the hydraulic systems.
Planning the installation of the various hydraulic and monitoring systems was very complex due to the small and very congested nature of the site, the need for DLR possessions for access, and additionally the time frame for the installation works was tight. As some parts of the site were inaccessible most of the time, a temporary bridge was constructed to access a central island area so that work could run 24 hours a day.
This was also useful for maintenance and calibration of the instrumentation. At other locations, some of the works were so close to the existing tracks that the extent of the kinematic envelope of the trains had to be considered.
Lifting and installing the jacks were major operations, as they each weighed 0.5t. In some cases comprehensive steelwork systems had to be constructed just to install the jacks. The entire operation had to be very carefully planned and precautions and back-up strategies were implemented at all stages to minimise risk to the flyover and DLR operations.
In total there were more than a thousand sensors with vast quantities of raw and processed data that had to be managed. As well as the trigger levels linked to the structural tolerances, additional functionality and integrity alert systems were set up to ensure that the instrumentation was working properly. Templates were developed to allow the vast data sets to be presented in a meaningful way so that the various parties could assess them quickly and make decisions and take necessary actions.
Observed responses
The jacks were installed 18 months in advance of the tunnelling and the early setting up of the monitoring allowed background, and in particular thermal responses, to be identified.
A linear thermal response of the flyover to temperature change was observed with expansion and contraction with respective warming and cooling. Reassuringly the fixed point was very close to the point where zero change occurred.
With the flyover supported on the jacks, the instrumentation was sensitive enough to register when the DLR trains passed over.
As the layout of the piers was not uniform, the forces on the various jacks acted in multiple directions. It was very important to understand this behaviour in advance of the tunnelling when it would be necessary to control the jacks.
Installation of the fixed bearing at this point was very difficult and once in position the temporary structural steelwork and jacks were subjected to enormous forces from the structure.
The flyover at this location did move longitudinally, initially it was thought that the capacity of the jacks would not be able to match the magnitude of the forces, but with the jacking loads applied in incremental steps, it was possible to induce small displacements and to control the movements.
In the case of the movement joint SA0, horizontal and vertical jacks were installed to take over its function during the tunnelling works – without horizontal control there could have been problems with the railway and the vertical control helped maintain lateral tilt of the transition slab within tolerance. Initially there was much noise with data spikes extending into the red trigger level zone.
It was important to have better conditioned data prior to the first tunnel drive and this was achieved by installing voltage regulators and modifying the method of processing the data. There was a clear change evident in the graph, in time for the tunnel drive, allowing careful control of the jacks.
During the tunnelling works, the monitoring data were continuously assessed and reasons for irregularities checked immediately. Frequent meetings were held to ensure that the data were interrogated properly and tolerances maintained.
Excellent control of tunnelling volume losses were achieved for the WB tunnel (<0.2 per cent), which allowed the whole system to be checked without operating it to its full capacity. Larger movements occurred during the EB drive (volume loss still only about 0.4 per cent) but the viaduct was successfully controlled by carefully staged control of the jacks using feedback from the monitoring systems.
Primary conclusions
The importance of team work and interaction with all parties involved was emphasised. The relationships developed among individuals from all parties resulted in what could have been major issues being resolved smoothly, even under pressure. The main lessons from this project can be summarised as follows.
- It was essential to set up a workable approvals structure with all the various parties on board and an interface control plan in place.
- The primary mitigation measure was to control face loss (this was assisted greatly by adopting EPBMs).
- Assessment of overall risk meant identifying the unmanageable risks and controlling those with which could be dealt. Installing the comprehensive monitoring and jacking systems allowed much better control of risk.
- Having more instruments means more data (the need to condition and interpret them) and potentially more false alarms.
- Simple checks are essential (and development of check sheets, flow charts, etc.).
- Experience was gained in steps, gradually making progress, increasing knowledge (e.g., base line readings, dewatering operations and the experience from the WB drive).
- Adopt a passive approach where possible (i.e., not reacting too hastily without carefully thinking through consequences, assessing whether conditions might settle down, and pre-emptive jacking in anticipation of tunnelling-induced ground movements).
- Importance of teamwork, enabling issues to be resolved quickly and smoothly, even under pressure
Questions from the floor
Gary Brierley, Doctor Mole Inc., USA: How did you distribute the predicted settlement troughs across the deep piled foundations?
Answer: It was assumed that the solid pier would behave monolithically [en masse] and the piles be tied together by it. The settlement at two-thirds of the pile group depth were taken on the left- and right-hand sides and the assumption made that the pier would move as a linear element in conformance with that. It seemed a reasonable approximation given all the other factors to be taken into account.
Barry New, Geotechnical Consulting Group. I have two questions: (i) How much did the jacking system cost; (ii) what would have happened given the ground loss performance, if you hadn’t done anything at all?
Answer: The scheme cost several million GBP. The structure came to the limit of the amber threshold on the basis of the 0.2 and 0.4 per cent volume loss. In hindsight it might have been just about acceptable but decisions had to be made six months before the TBMs went through and Crossrail had to give DLR a guarantee that their trains would be able to continue running safely. The 1 per cent volume loss was the worst credible at that time and this was the value that we worked to. Things could have been much worse if the tunnelling had not been in London Clay and it would not have been very risky to have an assumed a 0.3 per cent volume loss at that time.
Neville Harrison, retired: The values set as trigger levels were less than 10 mm – these seem low in view of the fact that weather and temperature can have a strong influence on a structure’s movements. He recalled an example where Grey’s Monument in Newcastle was being monitored and was found to move by about 25 mm due to sun.
Answer: This is definitely true, for example during the Jubilee Line Extension works the Big Ben clock tower at Westminster was found to move diurnally as a result of tidal and solar effects. At Canning Town the flyover could expand and contract by up to about 35mm longitudinally. The curved alignment of the flyover meant that the longitudinal movements also caused it to bow by about 5mm, which was exacerbated by the sun shining on alternate sides of it during the course of the day. Changes in the ground water table were evident from tidal effects but these did not seem to affect the bridge. A further intriguing response was caused by the fact that as the steel central section of the flyover reacted more quickly than the concrete side spans a spiralling pattern was observed. These movements were all greater than those anticipated from the TBM, it was important to understand them in advance so as to be able to isolate them.
Vinno Balakumarasingham, Waterman Group. Two questions: (i) with so many variables was your engineering judgement challenged at times? (ii) Is there any counter-intuitive that you will carry forward to future similar projects?
Answer: At the daily review meetings the data from the instrumentation invariably looked erroneous and were checked and found to be correct – sometimes the behaviour of the structure was counter-intuitive – this was particularly so during the base-line monitoring and highlights the importance of this period to help understand the background responses so that during the transit – when occasionally there were unclear responses – at least the non-tunnelling associated movements were reasonably well understood. A particular point worth mentioning, re counter-intuitive responses, concerns the CPN fixed point where longitudinal jacks were located. These were not capable of shifting the 7,000t deck but in conjunction with the longitudinal thermal movements it was possible to control the deck movements northwards: it was surprising in this case that the theory worked out in practice.
Roger Bridge, Balfour Beatty (BTS chairman): The flyover was designed with mitigation measures to take into account the tunnelling that was due to take place some years later (i.e. the designers knew that the flyover would be influenced by future Crossrail construction). How well did that work? Are there lessons that can be taken forward with this regard?
Answer: It would have been good if the structure were more tolerant to displacements. The fact that the deck was continuous from end to end made safeguarding it difficult. If there had been individual spans with greater articulation it would have been a simpler structure to work with but then it would not have met the objectives of the original structure – the only way of achieving it would have been to have a continuous S-shaped span.
Rapporteur: Jamie Standing, Imperial College London
