The city of Durban lies on the east coast of South Africa and is a thriving business and tourist destination. The harbour, with an area of highland called the ‘Bluff’ to the south and the business district to the north, is the busiest port in Southern Africa, operating 24 hours a day, 365 days a year. The harbour entrance has a navigable channel of 120m and can accommodate vessels with a maximum draught of about 12m.
The current harbour entrance was established in the late 1800s and is quite difficult for larger ships to navigate in heavy weather. The harbour is also unable to accommodate larger new-generation ships. Consequently, the Transnet National Ports Authority intends to widen the harbour entrance by relocating the breakwater and deepen the channels within the port.
At present a number of utility services are carried across the harbour entrance in an existing 350m long, 3.6m diameter immersed tube. This tunnel was built in the 1950s and is located at a shallow depth. Before the widening and deepening of the harbour channel can take place, these services need to be relocated into a deeper bored tunnel. The existing immersed tube will then be removed.
A feasibility study for the project was undertaken in 2003/2004, with the work for the replacement tunnel commissioned by eThekwini Municipality Water & Sanitation Department, and several options for carrying the services across the harbour entrance were evaluated. A shallow bored tunnel was eventually selected as the preferred option.
The lead consultant for the project is Goba (Pty) of South Africa, with the UK’s Mott MacDonald providing specialist advice on the tunnelling. The tunnel contract was awarded to Durban Harbour Tunnel Contractors (a JV between Hochtief of Germany and South Africa’s Concor).
Alignment
The proposed profile of the new harbour entrance largely governs the vertical alignment of the tunnel. However, commercial and residential development constraints on the north side and topographic constraints (the Bluff) to the south meant that deep shafts would be needed for TBM launch and reception. To minimise the depth of the shafts, the position of the tunnel under the shipping channel was as shallow as considered prudent. Studies were carried out to determine the maximum grade that could practically be negotiated by a TBM (and any associated increased cost), which could then be compared with the cost of tunnelling at lesser gradients but with increased shaft depths.
This led to the adoption of unusually steep gradients of 20% on the incline and decline of the tunnel for about 100m at each side with a 300m section at 0.5% grade across the main part of the harbour entrance. The cover for the bored tunnel below the existing seabed was chosen as a minimum of 9m (two tunnel diameters). After the dredging of the deepened harbour channel (following tunnel construction) the minimum cover chosen was 5m. These depths also allowed for any over dredging.
Geology
Eleven boreholes were drilled as part of the site investigations, two of which were within the existing shipping channel. The geology is largely Holocene and Pleistocene marine and lagoonal sediments. This is mostly classified as dense fine cohesionless sand with a clay/silt content in the order of 10%.
However, lagoonal sediments comprising mostly silt and clay were identified, which could occupy up to 50% of the tunnel face. Approximately 420m out of the 515m total tunnel drive would be through these sediments. The strata changes to aeolianite reef sandstone on the south side of the harbour forming the harder geology of the ‘Bluff’ and the tunnel drive would pass through this geology for the remainder of the drive. The sandstone on the south side was uniformly grained but heavily laminated with a strength varying between 0.5-50MPa. A full head of water pressure of around 3.5 bar was also expected.
Tunnel lining
Though a 4.5m id tunnel was initially specified, it was known early on in the project that a suitable Herrenknecht Mixshield slurry TBM with a 4.4m id lining had been used on the recently completed Kai Tak Scheme, in Hong Kong. Fortunately, all four tenderers for the Harbour Tunnel included the Kai Tak TBM in their bids and as it turned out, both the TBM and lining moulds from the project were procured.
The 4.4m id lining was a tapered ring, nominally 1200mm long, fitted with primary Phoenix EPDM and secondary hydrophilic gaskets. The manufacture of the lining segments took place on-site close to the North Shaft. With the limited production requirement for the short length of tunnel, and the generally favourable weather, the use of an indoor production facility was not warranted. However, portable screens were provided for the moulds to give protection against sun, wind and rain. The moulds were retrofitted with form vibrators to assist with compaction of the concrete.
Extensive trials were carried out on site to determine the optimum consistency/slump for the specified 50MPa mix, and to obtain the best sequence of casting, opening the moulds and floating. Steel reinforcing cages amounting to 465kg per ring were pre-assembled in the segment manufacture area. Cover to the steel was 50mm.
Shaft design
Both the North and South Shafts were constructed using diaphragm walls. The North Shaft was designed as a twin cell structure, with a central diaphragm wall. This wall had an opening allowing for shield and gantry installation. Pit One was designed to accommodate TBM support during tunnelling, while Pit Two allowed early commencement of the cut and cover works.
The South Shaft required a different configuration, incorporating a single cell reception shaft and rectangular inclined shafts for cut and cover works.
On completion of the diaphragm walls, the shafts were excavated using a combination of dry and underwater excavation using a mechanical grab or a dredge pump in the softer material. Prior to casting the base plug, the formation was cleaned by divers using suction airlift equipment. The 3.5m thick un-reinforced base plugs were then cast underwater using a high slump pump mix, incorporating an anti-washout admixture.
A problem was encountered on the North Shaft shortly after dewatering, with a leak coming from under the base plug of Pit Two through a joint on the intermediate cross panel. The leak caused sand washout and a sinkhole to form outside the shaft adjacent to the tower crane. The shaft was immediately flooded for stability. Extensive underwater grouting was carried out using a combination of cementitious and polyurethane grout before dewatering again.
Tunnel construction
It was a requirement of the project that only a slurry TBM was considered for excavation. This was due to the ground conditions expected along the tunnel drive and the high water pressure of up to 3.5 bar. The Herrenknecht Mixshield excavated material, mixed with bentonite slurry, pumped through a pipeline to the surface separation plant.
Due to the gradients, a trackless transport system was developed for the segments and grout mix. This Multi-Service Vehicle was specially designed and fabricated by Techni Métal Systèmes (TMS) in France and tested under simulated conditions for full load operations on a 20% Slope.
In the initial stages of the down-grade drive, a combination of the soft material being excavated and the gradient of the drive necessitated low thrust pressures to prevent the TBM “running away”. Diving of the cutterhead was also experienced.
The consequence was some difficulty in steering since it was hard to develop the differential thrust needed to keep the front of the shield up. An average advance of about 12m per day (or ten rings) was originally planned for the tunnelling operation, working on a 6-day week, 24 hrs a day.
However, problems such as the leak in the North Shaft resulted in a delayed start and led to the decision to change to a 7-day week in order to recover the schedule.
The initial rate of advance was slower than originally planned, which persisted through most of the down-grade drive. The rate of advance started to recover when the TBM moved into the sub horizontal drive, but was then affected by encounters with significant quantities of clay, requiring interventions to clear blockages and problems caused at the separation plant. However, once the clay areas had been negotiated, the rate of advance for the remainder was essentially as originally planned (figure 4).
Of the total 2519 hours of working time, total downtime due to TBM, backup and separation plant was 1370 hours, giving an impressive tunnel system availability of 46%.
On the decline, control of annulus grouting proved critical. The TBM was not fitted with a skirt at the rear of the shield to prevent backflow; and on the down-grade drive, the grout showed an increased tendency to flow forward over the shield.
Problems were experienced with very high thrust pressures being required on the middle part of the down-grade drive, which resulted in a loss of directional control with the inability to develop a differential thrust.
This was thought to be due to the presence of annulus grout, which had accumulated and hardened on the outside of the shield. Attempts were made to dislodge this, including fitting form vibrators to the inside of the shield, which generally proved unsuccessful. It was also thought that the accumulation of annulus grout on the shield might have blocked the articulation gap, again hampering the ability to steer effectively. This was partly alleviated by continuous manoeuvring of the shield to remove grout from the articulation joint.
Eventually, an intervention was carried out and the gauge cut was increased by 20mm to reduce the friction on the shield. This allowed the advance to proceed using normal thrust pressures and with improved directional control.
The main effect of this problem with directional control was a deviation from the designed alignment of some 450mm horizontally and vertically. While this was out of specification, the decision was made not to force the TBM immediately back onto the designed alignment, as this could have resulted in damage to the segmental lining. Instead, a revised alignment was adopted.
The circumferential and radial joints of the segmental lining were fitted with dowels and bolts respectively. Problems were initially noticed during the installation of the first rings in the tunnel, with severe spalling around the dowel sockets.
On investigation, it was found that the dowels had a nominal diameter greater than that of the inserts. This resulted in stressing of the concrete around the socket, which resulted in failure around the dowel.
Site trials were conducted to test the effect of reaming the sockets on the pull-out force of the installed dowel. It was determined that limited reaming would ensure adequate lock-in force and this solution was implemented successfully.
The 600m3/h slurry treatment plant was provided by Piggot Shaft Drilling (PSD) from the UK and consisted of a primary shaker, two de-sanding units and two 15m3/h centrifuges. The clay encountered on the tunnel alignment caused some problems with the plant, with the capacity of the hydrocyclones on the de-sanding units being exceeded, resulting in overtopping.
The amount of clay being dispersed into the slurry also resulted in high mud weights, which required extensive cleaning through the separation process.
Summary
The Durban Harbour Tunnel was the first use of a slurry TBM in South Africa and demonstrated how the difficulties of tunnelling on a steep gradient of 20% were able to be overcome. The successful completion of the tunnel means the proposed widening of the harbour entrance can now proceed as planned.
The project has recently won the Concrete Society of South Africa’s Fulton Award for Excellence in the use of concrete and the South African Association of Consulting Engineers (SAACE) award for projects over R5 million (US$690,000).
Questions & Answers
Michael Cole (Retired) observed that the 46% tunnel system availability was indeed impressive and asked how much time in the shift was spent maintaining the machine.
Andrew Officer responded that two 12 hour shifts were worked with no specific periods allowed for maintenance or pipe changes.
Gordon Torp-Peterson (London Underground) asked what measures had been incorporated into the ring design to protect the tunnel from the high external water pressures.
Andrew Hindmarch responded that the ring was fitted with a hybrid EPDM gasket and hydrophilic gasket. A dense concrete mix with a large 50mm cover to reinforcement was also specified to ensure long-term serviceability of the tunnel. He also noted that leakages into the tunnel were below the specified amounts.
Martin Knights (Jacobs) asked who owned the tunnel and whether there were any specific problems with the insurance of the asset.
Andrew Officer replied that the Durban Municipality owned the tunnel, but the services themselves were owned by multiple authorities. Extensive work had been undertaken to ensure robust fire and vapour detection systems were included in the design, although no fire suppression system was included. The authors were not aware of any problems relating to the insurance of the scheme.
Donald Lamont (HSE) asked what rules had been followed in relation to the compressed air working during head interventions. He also asked what the incident rate of decompression sickness had been on the project.
The speakers replied that there had been no reported cases of decompression sickness during the eleven interventions required.
Neville Harrison (Mott MacDonald) congratulated the speakers on communicating the difficulties encountered, particularly with the dowels, and wished that the industry in general communicated lesson learned on projects to avoid future repeats.
The speakers said the contractor had done very well, but subsequently found out that exactly the same problems with the dowels had been suffered on the previous project.
Rapporteur: Colin M Eddie
Fig 1 – Durban harbour’s entrance channel The Herrenknecht Mixshield is lowered into the North Shaft Fig 2 – Longitudinal section and geology On site segment yard Fig 3 – Design of the North Shaft Fig 4 – Time/chainage plot The TBMs tailskin arriving on site The Mayor’s wife visits the project