The hong kong-zhuhaimacao bridge runs for a total length of 55km, with 29km over the water, passing through important shipping channels. Because of the reservation for local development, a requirement for limiting the blockage of the Pear River Estuary and the limitations put in place by Hong Kong International Airport, a solution combining a bridge and tunnel (with the changeover on artificial islands) was chosen.
The HKZMBT project is the controlling and crucial part of the link (Figure 1). Construction works comprise two 100,000m2 man-made islands, with support structures built on the islands, a 6.7km-long subsea tunnel with all associated fit out works, and the tunnel element prefabrication factory. The full highway is laid over six lanes in total, with a vehicle design speed of 100km/h. Two bores, each with a single-span gallery, no divides, measure 11.4m high by 37.95m wide. The immersed tunnel length is exactly 5.664km consisting of 33 elements and one closure joint; the typical element length is 180m and weight 76,000t.
Regarding challenges, the construction site is 30km away from the contractor’s facilities on the main island, and located in China White Dolphin conservation region. Over 4,000 vessels pass by the area daily. The foundations of tunnel and islands rests in a thick but soft soil stratum (Figure 2). Typhoons in summer and strong convections in autumn and winter are frequent.
The project is a design-build scheme and work began in January 2011, while hand-over was in February, fulfilling the schedule. The construction of the tunnel element production factory took 14 months, tunnel element production 58 months, and the installation of tunnel elements, including closure joints, 48 months.
RAPID FORMATION OF TWO ISLANDS
It is known from the above that the total schedule is seven years. To provide a condition for the connection of the first tunnel element the man-made islands and the cut and cover section cast in-situ on the island shall be completed in the first place. However, the soft soil stratum under the man-made islands is up to a thickness of around 30m and it was estimated that with this soil condition island construction will take three years, which left time inadequate for tunnel construction.
The soft soil foundation can be a double-edged sword for island construction. With the traditional method of creating artificial islands, arbitrarily speaking, a soft soil foundation is either improved or removed before being replaced with sand and stone above. That is to say, the softness of the soil is regarded as an adverse factor that should be modified.
Taking a different approach for this project, the authors observed that soft soil can be easily penetrated. Thus, a 22m-diameter, 50m-high steel cylinder with wall thickness of 16mm can be inserted into the soft soil to around 30m deep. In this way, two islands can be formed by inserting 120 steel cylinders. Meanwhile auxiliary cells used to connect the adjacent cylinders are inserted into the impermeable layer of the soft soil. The connected cylinders and auxiliary cells can form a temporary, impermeable wall so that dewatering inside the area becomes possible. The critical job is the joint work, using multiple hydraulic hammers for cylinder driving and also the insert-ability of the auxiliary cells.
Applying this method, the length of time taken to form an island is shortened to seven months and the reduced marine operation also greatly reduced the impact on the dolphins, and minimised extra traffic in the already busy area.
Why is this method such an improvement? It can be explained in three points:
¦ The construction of the temporary island wall is fast. The site operation time is reduced by means of industrialisation and large scale basis (i.e., the bigger each individual piece, the less the marine operation time). The dominant factor of the construction time is production and transportation of the cylinders. To ensure rapidity, steel structure factories can be compared in terms of processing capacity and transport ability.
¦ The foundation work is fast. The impermeable wall allows for sand filling and vertical drains inside the wall on the island to achieve surcharge precompression by a large overconsolidated ratio (Figure 3). Thus the foundation consolidation/ improvement takes around 100 days. Compared with traditional island construction methods, a great amount of marine foundation work is unnecessary.
¦ After-foundation ground improvement can directly meet the requirements for constructing buildings, and thus the need for separate pile foundation for land structures is eliminated. Additionally, an 18m deep trench is needed for cut and cover work, the cylinder can be used as a fence structure for this. The construction procedure of the deeplyinserted cylinders is shown in Figure 4b. Its characteristics can be seen when comparing it to a traditionally constructed island in Figure 4a.
¦ The soft soil is conserved to a large extent and used for two functions: one is it’s almost impermeable, and the other is that it is penetrable. The latter secures the stability of the steel cylinders exposed to marine conditions.
¦ The island wall of steel cylinders and auxiliary cells not only isolates the seawater outside the island, but also allows for parallel work on the permanent island revetment outside, and the land works inside. Therefore, the island construction speed no longer depends on the speed of permanent revetment works, but of the production and installation of the steel cylinders and auxiliary cells. It may be worth mentioning that despite the development and execution of the new method, the cost of the island did not outstrip the budget. This is largely because the marine work is significantly reduced.
Since the implementation of this methodology for the HKZMBT, other projects in China have adopted it. A good example is the Shenzhen-Zhongshan link, currently under construction, which features steel cylinders with a diameter up to 28m that are driven by 12 hydraulic hammers.
THE TUNNEL’S FOUNDATION
Under the original design, the support piles start from both ends of the tunnel and as you travel along the alignment, become settlement-reducing piles and then gradually it switches to the natural foundations in the mid part of tunnel.
Preparations were made for this methodology. However, during testing of the design parameters for the load performance of the pile cap and gravel bed, the settlement was not convergent with the increase of loading level. Hence the original scheme was abandoned and the tunnel foundation scheme redesigned. The final foundation scheme is double-bedded foundation layer, underwater surcharge, and ground improvement (mostly by sand compaction piles).
A double-bedded foundation layer means additional placement of cobble mass below the gravel bed to reduce the uncertainty of foundation settlements.
For the HKZMBT, the tunnel is excavated to a maximum depth of 30m, so the soil stress generated by all the loads after completion of the tunnel will not exceed the historical stress. The main settlements will not occur at a deep layer of ground, but instead at or close to the surface, where settlement was a matter of great uncertainty due to two factors: the disturbance of underwater excavation, and sedimentation. The interlayer formed by the disturbed soil or sediment is very weak compared to the intact soil. To overcome this, the authors proposed the installation of an additional 2m-thick layer of cobbles below the gravel bed, which was to be compacted by vibration using hydraulic hammer. The compaction eliminates the disturbance of excavation; the diameter of the cobbles is 300 to 500mm enabling the multi-beam to easily tell the sediment from the cobble layer, i.e., cobble layer provides a reference surface for sediment cleaning, during which the cobbles will also not lose stability due to unit weight.
In short, the cobble layer reduces the uncertainty of subsea foundation works and provides a hard base for the placement of the gravel bed layer above and of the landing of the tunnel elements.
Sand compaction piles (SCP) are applied to improve the soft soil layer and to increase the bearing capacity of ground. Depending on the load level and soft layer thickness, three replacement ratios of 70 per cent, 55 per cent and 42 per cent of SCP were used. Also, the majority section of SCP serves as a vertical drainage passageway for surcharge and precompression. Underwater surcharge and pre-compression aims to achieve a uniform ground stiffness along the alignment and to eliminate the initial settlement of SCP. The transition of stiffness starts from middle part of the tunnel (deeply-excavated trench) to the shallow part (thick, soft soil below) and then to the cast in situ part of the tunnel structure in the island. The settlement of surface and deeper layers is monitored during surcharge and the ground consolidation is estimated by the monitored date and 90 per cent of consolidation is the criteria for unloading. Gravel and cobbles are applied for surcharge, and the width of the surcharge cross section is the same as the width of backfill protection on the tunnel. The widest part is around 120m which is equivalent to the width of apron protection of the tunnel for ship collisions near the man-made islands.
After the implementation, settlement monitoring data demonstrated that the settlement along the majority is controlled in the range between 50 and 60mm (an exception is at the location of immersion joint between elements E32 and E31 where settlement is much larger possibly due to sedimentation interference). Figure 5 shows settlementtime curves of four tunnel elements. It can be seen that the curve rapidly converges once the tunnel element was loaded. What is surprising is that the monitored settlement was far less than the predicted value of 150 to 200mm and monitored settlement, considered longitudinally, shows better uniformity than that of the predictions. The results for the other 30 tunnel elements (monitoring points at longitudinal interval of 22.5m during construction period) also show the same tendency. Therefore, it is highly likely that the main settlement of immersed tunnels only occurs at the surface of foundation and is not particularly related to geology. Hence, good control of surface settlement, or a foundation scheme favourable for quality control of underwater foundation works such as the double embedded foundation layer in this project, is vital to the foundations of an immersed tunnel.
Considering the above, that the quality of foundation works is critical to control the settlement of an immersed tunnel; a series of special-purpose equipment was developed/modified to assure the quality in the HKZMBT project. A grab dredger was modified to have the plane dredging function so as to reduce disturbance and increase accuracy, which is controlled within a tolerance of ±500mm for the tunnel trench bottom. A dredger cleans the sediment at the bottom of tunnel foundation trench before gravel bed placement. A stone compacting and levelling vessel was developed to place the cobble layer as discussed and this vessel places, vibrates, and levels the cobble mass under the water by two falling pipes and one hydraulic hammer. In addition, a gravel levelling platform is made for placing the gravel bed, and was equipped with a suction head to directly remove the local sediments on the gravel bed almost without disturbing the gravel bed placed in position.
THE SEMI-RIGID TUNNEL ELEMENT
In this project it is highly possible that the vertical shear keys of the segmented joints of the tunnel could be damaged due to differential settlement. The tunnel is placed in a pre-dredged trench. Once the tunnel is in operation, a 21m-thick sediment cover will be formed on tunnel roof, and part of the sediment will be re-dredged in future for a deep navigation channel. In addition, the geology is soft soil layers with varied thickness ranging from 0 to 30m along the tunnel length.
The original vision for the immersed tunnel element structure was segmented, with vertical concrete shear keys arranged at the joints with a spacing of 22.5m. Due to the thick coverage on top of the tunnel, relying on the shear keys at segmented joints would result in insufficient bearing capacity. In order to solve this problem, two solutions aiming at reducing the load on top of the tunnel were proposed. One solution was constant dredging to remove the sediment during the operation stage throughout the full 120-year lifetime of the tunnel. The other solution was to place light material on top of the tunnel in advance, this would have a material unit weight close to that of seawater.
Rather than changing the external surroundings of the structure, change of the structure itself is preferable. The authors proposed the concept of a semi-rigid element structure to solve the problem without the need for load reduction. This solution can reduce the necessary marine works compared to the previous two solutions, and the main change for the tunnel element structure is keeping the temporary prestressing tendon in place permanently, instead of releasing the temporary pre-stressing tendons after immersion of tunnel element, which is normal practice for a segmented tunnel.
The semi-rigid element is a type of segmented structure in which the frictional force at the vertical face of segmented joint is mobilised to resist (part of) the shear force so as to strengthen the shear capacity of the segmented joint. Adequate frictional force is ensured by arranging a rational amount of prestressing tendon longitudinally so as to gain adequate compression at the joints. Further, a certain amount of rotation is allowed between segments to enable the element structure to adapt to foundation settlement through longitudinal deformation. In short, semi-rigid elements are a type of immersed tunnel in which tension and frictional force are utilised and longitudinal flexibility of the structure is kept to improve both robustness and integrity of the tunnel structure.
The high robustness of semi-rigid elements can be illustrated by comparing their structural behavior when resisting the uneven settlement with that of monolithic element and segmented element (Figure 6). For a colourful metaphor, if the monolithic element is a a muscular person and the segmented element a flexible person, then semi-rigid element is a person with both musculature and flexibility.
Since the semi-rigid element retains prestressing tension compared with segmented element the joint is not vulnerable to opening hence the integrity is improved. The advantages gained are:
¦ The watertightness of a segment joint is better, particularly with an injectable waterstop
¦ Less likelihood of reflection cracking on the road surface.
MEMORY BEARING
The semi-rigid element has increased the shear resistant of the segment joint as discussed in the previous section. However, the vertical locking between elements (i.e., the immersion joint) is a weak point too. For the HKZMBT the vertical locking is by steel shear keys mounted after tunnel element immersion.
Although the force on vertical shear keys can be reduced by delaying the locking time to allow some relative vertical displacement at the immersion joint, still 21m of sediment will accumulate and will increase the load on tunnel roof for years, and thus the steel shear keys and the concrete walls adjacent to the connections are subject to damage due to excessive loading. To protect the shear keys and avoid cracking of concrete, especially on the external walls without any external membrane (where the cracking may lead to corrosion), the author proposed setting ‘memory bearings’ between vertical steel shear keys as shown in Figure 7.
The memory bearing functions to protect the structure near immersion joints from shear damage by the compression of itself to release the differential settlement between elements and continually sustaining a constant force during its compression. This force is set to be no greater than the capacity of the shear key or of the adjacent structure. On the whole, the memory bearing protects the structure by rationally diverting the force direction and distributing the force. The memory bearing ‘remembers’ the design capacity of the concerned structure and when the memorised value is about to be exceeded by the load, the bearing will divert the excess part of the load to the foundation right down the structure rather than to the foundation of the adjacent structure via shear keys. By this, the memory bearing can free the structure to play its role/utility without damaging itself. The character of force-compression of the memory bearing is calculated by tearing and fracture testing of special purpose materials.
Any uncertainty of geology is compensated for by strengthening the structure. But for an immersed tunnel the structural bearing capacity is often constrained by the dimensions of its walls, so the safety levels of immersion joint can be insufficient. Memory bearings compensate for this uncertainty and even provides redundancy
DEPLOYABLE ELEMENT: A NEW METHOD TO BUILD THE CLOSURE JOINT
The closure joint is located between elements E29 and E30, where the depth of the lower extreme of the element is -27.9m and is exposed to waves and currents. Before its construction, five years ago, the authors made inspections and found five construction methods for the closure joint — the cofferdam, the panel (most broadly used), the V-block developed by Kawasaki, the terminal block developed for the Osaka port tunnel and the key-element method also used in Japan. It was found, however, that none of them could reliably connect to the adjacent tunnel elements and stop the water.
A deployable element was developed as the closure joint for HKZMBT project. During installation, the closure joint stretches itself until making contact with the adjacent tunnel elements. The closure joint is equipped with two retractable joints, which can expand out or retract longitudinally, in relation to movements of the main structure. To provide space for works on the permanent connection from tunnel’s internal side, the joints are arranged at the outer edge of the main structure.
Further, a waterstop system is provided at the ends of the joint to ensure dry conditions in the connection chamber for the works of permanent connection. The works procedures, in brief, are: 1) The structure is made and outfitted in a steel processing factory or element prefabrication yard, or both, prior to marine operation; 2) A floating crane lifts the closure joint up in one piece, lowers it and places it on a gravel bed in the closure gap; 3) The retractable joints on both ends stretch out and make contact with adjacent elements; a watertight chamber are formed at the connections at both ends; 4) Water is removed from the chambers; 5) Construction of permanent connections from the intrados of the tunnel by welding steel plates and grouting.
Three benefits have pushed the realisation of this work.
1) Short marine operation time. The time taken by installation (for the first time, not counting the re-connection) is less than one day, for the HKZMBT project. Comparatively, the panel method needs approximately eight months of marine operations. Hence the construction risk is greatly reduced with the reduced duration of marine works, particularly with respect to the diving work;
2) Positive connections and watertightness. The connection and compression of the Gina gasket is completed by the operating jacks. 3) Re-connection is possible so the tunnel alignment at the closure joint location can be better managed. After the first time installation, a gothrough survey reported that the plane offset of the closure joint part and the adjacent tunnel elements was about 100mm. Despite this acceptable accuracy, the teams re-flooded the chamber, withdrew the retractable joints, slightly lifted up the closure joint and made the connection for the second time. The survey showed that the offset was reduced to 20mm. However, there are three drawbacks in applying this method:
1) A large floating crane is required.A fully revolving 12,000t floating crane was used in the HKZMBT project and it may not be used in another project for the limited water depth;
2) The local disparity in ground stiffness is high. For this project the longitudinal length of the closure joint is approximately 9.6m at the bottom slab and about 12m at the roof. Despite the bulkhead, the buoyancy of the closure joint is still far less than its dead weight, that is to say, its dead weight cannot be neutralised by its buoyancy while the submerged tunnel element can. Therefore the average pressure of the closure joint imposed to the ground is approximately 30 times of that of the adjacent tunnel elements, causing a foundation stiffness difference. Hence the problem had to be dealt with by over-loading the neighbouring elements before the installation of the closure joint and by pressurised grouting below the connecting part after the installation.
3) In total, 54 jacks are deployed, but they are sacrificial and cannot be re-used.
For the first two of the three drawbacks, the authors suggest improving matters by making the closure joint’s longitudinal length the same as the tunnel element’s length or long enough to make it meet the two conditions. When submerged, its own buoyancy can neutralise its dead weight; the other the closure joint can be installed by using the immersion rigs shared with the tunnel elements. In other words, no gap is reserved for the closure joint and instead the retractable joints are equipped at both ends of the last installed element and treated as a closure joint, integrating the closure joint and tunnel element.Several possible tunnel longitudinal layouts are illustrated in Figure 8.
For the last drawback, two conceptual configurations that could allow for recycling the jacks are illustrated in Figure 9. Jacks are installed at corresponding locations of corbels supporting the bulkheads or arranged at the inner side of the tunnel structure. The bracket conveys the force of jacks to the frame to achieve the extension. In addition, the quantity of jacks can be reduced along with the increased stiffness of the frame structure in front of it. The dismantling of jacks can be performed after connection, water discharging, or during the fixing of permanent connections inside.
TUNNEL ELEMENT FACTORY
The selection of a prefabrication yard started in 2008, and six locations in areas close to the tunnel site were compared; the abandoned quarry yard on Guishan island which is the closest to the tunnel site was selected. The factory method initiated in Øresund project2 was adopted to produce the 33 tunnel elements for the immersed tunnel. The factory is divided into three zones: the production zone, the management zone and the living zone.
To match the speed of tunnel element installation, which is one every month, two elements were produced every two months on two production lines.
The critical path for the production line is rebar assembly, formwork and concrete casting, of which rebar assembly is the dominant procedure. Thus, rebar assembly was further separated by three parts in the production line namely the assembly of the bottom slab, of the walls and of the roof, to achieve a flow operation.
Production and concrete casting is done alongside the launch of the segments and finally the entire element, typically massing 76,000t, is launched forward as a block over the 200m to the shallow dock. Four sliding tracks are arranged below the four walls of the tunnel element structure and a multiplepoints launch is undertaken. The jacking speed reaches up to 0.13m/min and transverse accuracy is controlled to within 5mm. It is worth noting that the launching of five curved elements with plane curvature radius 5,500m was achieved for the first time ever, with this operation.
Fitting out is undertaken in the shallow dock. The deep dock can accommodate four elements to prevent delays in element installation. A sliding gate divides the shallow dock and the production line; it is a steel structure that weighs about 750t and is approximately 200m wide. The gate has a regular triangle cross-section that allows the use of water pressure for stability (Figure 10). The sliding operation takes 30 to 40 hours. The floating dock gate in the deep dock is gravity structure with a weight of 13,000t and width of 60m.
ELEMENT INSTALLATION
The natural water depth is insufficient for the transportation of the tunnel elements, so they can only be towed within the dredged channel. If an element was out of control the project would suffer an enormous loss and the stranded element could become an obstacle to the busy marine traffic. To reduce the risk of transportation over 10 MSA boats escorted the towing works. Alongside these, 12 tugboats (occasionally 13) were employed for towing the tunnel elements. Among them four tugboats are directly connected to the element with two at the fore to provide forward momentum and two aft; the remaining eight tugboats sailed alongside to control the position of the element by pushing the immersion rigs riding above the element whenever large cross currents are encountered. A navigation piloting system was developed for the towing works and the position of tunnel element and all tugboats were monitored in real time with the data displayed on the screen in the towing commanding room on the immersion rig as well as on all of the tugboats. This way, the lead captain in the command room of the immersion rig can give clear orders to all other captains on tugboats. Despite the above, it was a big challenge for the main captain (despite his rich towing experience) to coordinate the 12 tugboats simultaneous and keep a 76,000t element on track (i.e., within the safe region of the dredged channel). Hence, prior to formal towing works, exercises were done by towing barge of the similar mass for purpose of skilled cooperation meanwhile calibrating the newly developed navigation piloting system.
The tunnel was to be placed in a deeply excavated trench as mentioned previously, which has caused two problems1. It was observed that the current velocity at tunnel trench bottom was sometimes greater than that of the surface. This leads to a transverse installation offset of approximately 100mm (greater than expected) when installing element E10. After this happens the forecasting and alerting system was updated to add a particular window for the period of element connection with the current velocity at the trench bottom monitored and forecast; the installation schedule was adjusted accordingly. 2The sedimentation rate is rapid and sediment can cover the gravel bed within one day. A special case was the installation of tunnel element E15. For the first two attempts, the element was on its way to the immersion location only to find it half buried with sediment, so the element has to be turned around and sent back to the prefabrication yard, to wait for another appropriate time, which could be up to one or two months. After E15, a sedimentation forecasting system was developed to aid subsequent installations and the special purpose sedimentation cleaning head mentioned previously was also developed.
Finally 33 elements were placed during 35 installations attempts with no major incidents.
MISCELLANEOUS
The temporary steel bulkheads for element installation were non-monolithic, but water sealing was achieved by welding thinner steel plate; each bulkhead was re-used five to six times and no leakage was found. For element immersion, not only wave, current and weather conditions were monitored but also the displacement and speed of the element had to be calculated in real time by analysing the data from an accelerometer installed inside each element. For positioning of an element during immersion, wireless sonar was employed at the primary end of the element (i.e., connection end or Gina side) and a survey tower observed the secondary side. The positioning accuracy was within 100mm and 50mm. The go-through survey accuracy is within 50mm. With the accuracies from the tunnel alignment control strategy, installation of 28 elements (following the installation of the first five tunnel elements) was achieved without re-aligning works (i.e., no longer using jacks to correct the position of a secondary end of the newly installed element). After typhoon “Haiyan” in 2013, the project’s hazard standards were re-evaluated ito take into account the local tendencies for hazardous weather, and the hazard-prevention capabilities of the site were increased comprehensively. During construction, the project successfully withstood a frontal assault from the super strong typhoon Hato.
During the construction period the contractor carried out over 140 tests including the sediment accommodating mechanism; capabilities of gravel bed; the frictional coefficient of concrete element and gravel bed; element fast towing. Zero leakage into the tunnel has been achieved from installation of E1 in May 2013 up to present (February). In the authors’ opinion, the result is attributable to the innovative approach to tunnel foundation work, which limits the settlement of the semi-rigid element and limits the opening of segment joint, in addition to the project standard management, concrete crack control system, and the strict quality control of underwater foundation works.
