Shani Wallis, Editor
Collapse of a heading during excavation of the Lane Cove highway tunnel in Sydney has added to the heat of toll road tunnelling investigation in Australia in recent times. TunnelTalk examined the source of such intense scrutiny.
Southwest corner of the apartment block damaged by the November 2005 tunnel collapse
- Collapse of an exit ramp excavation on the Lane Cove highway tunnel in November 2005 tarnished an otherwise flawless undertaking and dealt another blow to tunnelling in Australia. From the start, Lane Cove was intended to raise the standard for well organized, well managed, and well financed tunnel construction in Australia and in Sydney in particular. A project to put the industry back on a solid footing following delayed completion of the Northside Stormwater Storage Tunnel(1); the government bailed-out Airport Rail Link(2) (both built ahead of Sydney’s 2000 Summer Olympic Games), and more recently in the face of Sydney’s troubled and now bankrupt CrossCity public-private-partnership (PPP) toll road tunnel debacle(3).
- Lane Cove is also a 33-year PPP concession awarded by the NSW Roads and Traffic Authority (RTA) to the Lane Cove Tunnel Company (LCTC) in December 2003 to finance-design-build, own, operate and maintain the new highway facility. Concession partners are Theiss, Transfield, John Holland and the financial institution ABM AMRO.
Fig 1. Location of Lane Cove within the Sydney motorway network
- The new 7.1km motorway with its 3.1km of twin tube tunnel completes a missing link in Sydney’s orbital highway, connecting the M2 Freeway at North Ryde in the north-west to the Gore Hill Freeway to the east and rerouting much of the central city’s commuter traffic off the congested Epping Road into the highway tunnel below. It also includes a highway connection between the tunnel and the north-south Pacific Highway.
- Design and construction of the project was awarded as a $Aust1.1 billion lump sum contract to the Theiss/John Holland JV. Parsons Brinckerhoff (Australia) was engaged as the lead design engineer with Coffey Geosciences and Pells Sullivan Meynink (PSM) providing specialist geotechnical services. URS Australia was appointed the Independent (design) Verifier.
- The tunnel lies largely in Hawkesbury Sandstone to the east and in weaker Ashfield Shale to the west. The twin 54m2 two-lane highway tubes run on average 30-40m deep and have a span of 9m, expanding to 13.5m in the breakdown bays, 75m2 for the three-lane sections, and 12.5m for the Pacific Highway interchange with spans of 17.5m in its breakdown bays.
Fig 2. Schematic of the twin tube Lane Cove tunnel and its associated underground works
- Excavation progressed from the two portals and a mid-point access adit using seven roadheaders (five Voest Alpine AM105s and two Mitsui SLB 300s). Systematic rockbolting (installed using automated Atlas Copco Boltec rockbolting rigs) and steel fibre-reinforced (sfr) wet-mix shotcrete (applied using two BASF UGC Potenza robots and two Jaycon Viper shotcrete machines) provided the core of the support regime throughout. Steel sets were included for zones of weaker rock with pre-excavation grouting to control ground water where necessary. Final lining is shotcrete and to finish, the ceiling is painted black and the walls are clad with reflective panels.
Fig 3. (Left) Elevation of the MC5B/MCAA intersection in relation to the main tunnel tubes beneath and (right - from the WorkCover report) from the opposite direction.
The property-damaging collapse in November 2005 occurred on one of the last phases of excavation to be completed and at the intersection of the MC5B ventilation adit and the rising MCAA exit ramp to Pacific Highway. Cover at this point and at about 120m from the ramp’s portal, is some 17m.
The incident occurred on the night shift, starting at about 1.30am, with small blocks of rock and shotcrete falling from the crown. Workers evacuated the heading as the fall progressed and reaching the surface within 20 minutes. By 2.30am a 6m-diameter hole had appeared, extending under the corner of a three storey apartment block and under the retaining wall of an at-grade exit ramp from Pacific Highway, known at this point as Longueville Road. The apartments and those of the neighbouring block were evacuated moving some 50 resident groups. By noon the following day the cavity had been filled with about 2,750m3 of concrete and backfill grout, burying with it the abandoned roadheader and loader.
Fig 4. Plan of the collapse zone after backfilling (WorkCover report)
- At the time, the MC5B ventilation adit was fully excavated to the end wall on the far side of the transition zone with the exit ramp (MCAA) and the ramp tunnel was being excavated in both directions from the ventilation adit. The up heading towards the portal had progressed about 22m and excavation of the down heading towards the main eastbound traffic tunnel had just started. It was in the north-west corner of the transition zone, at its junction with the down heading of the ramp, that the collapse began.
There were no injuries or fatalities to the workers or the general public.
Following the event, two reports were generated. The first, an independent incident report commissioned by the Theiss/John Holland JV from rock mechanics specialist E. T. (Ted) Brown, consultant with Golder Associates and Emeritus Professor at the University of Queensland, Australia, and before that at Imperial College London. The second is the investigation report by WorkCover, the New South Wales occupational health and safety commission(4). According to the JV’s Project Director, Brendan Donohue, who met with TunnelTalk in April 2006, the JV co-operated fully with both investigations releasing all necessary documents and making available project managers and workers for interview and questioning.
Both reports cite the low cover, the jointed nature of the area’s weak shale, and the presence of an almost vertical, low-strength dolerite dyke cutting across the wide 17m-20m+ diagonal span of the intersection, as being principal contributing collapse factors. Both ruled that water was not an influence although water from a resulting burst water main exacerbated expansion of the subsequent collapse crater.
- Both examined in detail the design, and described the excavation sequence being progressed at the time. Although not stated specifically, it is believed the ventilation adit and the 9m wide x 7m high exit ramp headings were being excavated on full-face 1.5m rounds with support comprising 4m long mechanically anchored, fully grouted rockbolts on 1m x 1m spacings plus layers of sfr shotcrete to a minimum 125mm thickness. Support in the crown of the intersection zone was supplements with additional 5m long bolts and extra layers of shotcrete. This augmented support was applied in two stages. First, an initial 100mm of shotcrete followed by 4m long bolts on 1m x 1m spacings. Second, another 100mm layer of sfr shotcrete with a pattern of 5m long bolts on 1.5m x 1.5m centres with a final layer of shotcrete to cover the bolt heads for a total thickness of 200mm. Design required that bolts be installed and grouted to within 500-600mm of the face.
- In his 40 page report, submitted in January 2005, Brown found no fault with the design in principle, reporting that “the design methodology … was in accord with the best practice in Sydney and elsewhere” and that the JV had “in place, appropriate and best practice processes for the safe and productive execution of the underground construction works” that “were implemented in a professional and effective manner”.
- The report concludes that the collapse was caused by a “complex combination of factors that were not present together at any other location in the underground works” stating that the support regime of rockbolts and layers of sfr shotcrete was “clearly adequate for the normal circumstance for which it was developed and used” and that “pre-production tests of rockbolting in the Ashfield Shale and proof tests on production installations in the shale in MC5B had been carried out with satisfactory results”.
- It reported however that: “In retrospect, it could be concluded that the level of support in the western side of the [MC5B/MCAA] transition was inadequate to ensure the excavation’s stability, given the large effective span, the low rock cover, the presence of a persistent vertical discontinuity (the dyke) transecting the excavation, and the inferred poor mechanical properties of the overlying rock mass.”
Fig 5. Mapping of the geological conditions and a vertical dyke transecting the excavations
Without listing them as direct contributing factors, Brown mentions as well the following as possible short comings.
• Apparent failure to apply recommendations of a design guide for tunnels in Sydney that suggests 6m bolts should be used in excavation spans of 17m, and of at least 7m long for spans of 21m, but adding that such charts are used “only as a guide or as one of several approaches used in reaching design decisions”.
• The absence of modelling for the wide span junction remarking that this might have been warranted during the design phase.
• The dyke featured at the collapse zone had not been identified in pre-excavation geotechnical investigations but it had been intersected and mapped during excavation of the main eastbound traffic tunnel below (MCA1) and in the adjacent MC5B ventilation adit, which, Brown suggests, would have signalled extra attention needed for excavation of the intersection and might have initiated review of the design and excavation sequences of the junction.
• The provisions of the Geotechnical Mapping and Ground Support Determination Work Method Statement may not have been followed fully in that there is “no evidence that the most complete longitudinal geological section….to aid projection of conditions ahead of the face” had been produced for MCAA or MC5B.
• “Preparation and maintenance of an up-to-date plan of excavation progress and geological features encountered may also have been helpful” in predicting and preparing for conditions ahead.
Among other possible mitigating factors Brown lists:
• The proximity of the pre-existing Longueville Road retaining wall and ground anchors to the crown of the MC5B/MCAA junction excavation stating that these “may have contributed by influencing the loads applied to the rock immediately above the excavation, or by weakening that rock mass, or both”.
• The crimped plastic corrosion protection sheathing of the rockbolts that appear to have interfered with the ability of bolts to “develop sufficient dilational shear strength at the interface of the bolt and the weak and relatively low stiffness weathered shale” and that
• “mechanical anchors may not have achieved their initial anchorage”.
Roadheader tunnelling in grey shale at the west end of the contract
The WorkCover report released in March 2006 is more direct.
After recognising the “complex and difficult environment in which to achieve suitably stable tunnel roof conditions”, it found that an “initial project design allowance to install steel sets or lattice girders in the excavations in poor shale ground conditions was superseded in the MC5B and MC5B/MCAA intersection with a less conservative rockbolt and shotcrete support option”. It does not say when this change was initiated but reports that it was “provided by the project designers in response to a request from [the JV] for a possible alternative to steel sets”.
The report states that “failure of the ground support system has two possible explanations – either the design was inadequate and not capable of supporting conditions that prevail[ed]” or, “the support system was not installed in full accordance with the design requirements and technical specifications”.
It explains that investigations to date indicated that:
• the suitability of the rock bolting strategy adopted to stabilise the geological conditions of the intersection “was marginal”;
• drill hole diameters, bolt end-anchorage, and full grout encapsulation specifications were not universally achieved; and that
• details of the “excavation advance and the timely establishment of a suitably thick and cured shotcrete lining to the crown and walls, as integral facts to the lead up to the incident, required further review”.
Neither report included data regarding time lapses between applications of support elements or between excavation rounds.
Since the downfall, the JV has progressed the project toward the planned opening in March 2007. The exit ramp tunnel was redesigned to bypass the collapsed zone, which was isolated behind a wall of about 17m deep bored piles. The residents of the 31 apartments in the two evacuated bocks (21 in 11-13 and 10 in 15-19 Longueville Road) were allowed to return by early January 2006. As an early compensation scheme, the JV offered to buy the apartments of those who chose not to return. The JV now owns all but two apartments in both blocks. The damaged building has been repaired to necessary safety standards and most apartments are currently occupied.
Excavating the invert in the grey shale
- On the legal front, WorkCover has continued its investigations and retains the right to bring charges against the JV and its designer for the incident. Just exactly what these charges might be is open to speculation but there are hints in the WorkCover report that these might be similar to those brought against the contractor and designer of the Heathrow Express railway tunnels project in London following collapse of its Central Terminal Area NATM station excavation in October 1994. Paraphrased, these charges accused the contractor and its design-build-contract engineer of failing in their duty to ensure the safety of their employees during construction, and the safety of those not in their employ (i.e. the general public) by exposing them to danger. A charge of failing in their duties towards technical supervision of the works was also brought against the engineer.(5)
- While the Heathrow Express contractor changed its plea to guilty just before start of the criminal trial in January 1999, the designer defended itself through a 26-day court heading and was found guilty by the lay jury of 12. The case for the defence was based on claims of unforeseen ground conditions and a range of mitigating contractual, procedural, organisational and practical circumstances. Many at the time suggested that the exact wording of the charges made them difficult, if not impossible, to defend. In his final instruction, the Judge reminded the jury of the charges, asking them to consider; did the accused endanger the lives of the workers and those of the general public and did they do every thing ‘reasonably practicable’ to avoid such risks? If ‘yes’ to the first and the second parts – not guilty. If ‘yes’ to the first and ‘no’ to the second – guilty.
Rockbolting with the Atlas Copco automated Boltec rig
- International comment and coverage of the Lane Cove incident in Sydney has referred to the tunnelling method also as NATM. According to Donohue however, the Lane Cove project has never been described as an NATM operation, by either the designer or the construction JV or the PPP consortium. The term is not used to describe the design or excavation process in either the Brown or the WorkCover reports nor in any of the Australian media’s coverage of the project or the incident itself.
- In response to the Lane Cove collapse; and similar to the Heathrow collapse response; WorkCover published an up-date of its Code of Practice for Tunnels Under Construction(6). The UK publication(7) spoke specifically of tunnels constructed in soft ground urban conditions using NATM and renamed the method SCL or ‘sprayed concrete lined’ tunnels. The WorkCover Code supersedes the 1991 Code of the same name and follows a format that guides all involved, through project risk assessment and identification of legal obligations. It does not include the terms NATM or SCL. As a concluding remark and as one commentator suggested, referring to NATM in the context of Lane Cove is “a total red herring” and regardless of the name or the method, “it is all about good engineering”.
Split heading excavation in Hawkesbury Sandstone
Tunnel junction in Hawkesbury Sandstone
- When TunnelTalk visited the Lane Cove project in 2004, Project Director at the time Russell Cuttler, confirmed that the JV had adopted the undertakings of the 2002 British Tunnelling Society (BTS) and Association of British Insurers (ABI) Joint Code of Practice for procuring project insurance(8). He also stated that there are some 2,500 properties and many public infrastructure assets in the 100m wide tunnel corridor (determined as twice the tunnel depth) through the Lane Cover city suburb. In addition, more than 250 Minister for Planning’s Conditions of Approval (controlling conditions from noise and dust levels to traffic management and environmental protection issues) had to be met at all times(9). Private property in Sydney is owned to the centre of the earth and the State government arranged all necessary project easements and procurement of the subsurface title to under-pass the properties. All 2,500 properties within the tunnel corridor were surveyed prior to start of construction and four community liaison groups were established to manage incoming calls, distribute project information, and conduct community meetings once a month.
- Insurance aspects in relation to the collapse were not disclosed by the JV at the meeting with TunnelTalk in April 2006 except to say that negotiations were on-going. There are however some points in the incident reports that could be construed as not complying with the UK BTS/ABI Joint Code of Practice (JCoP). The absence of an up-to-date excavation and geological record of progress for projection into the transition zone, for example, may be considered as failing to comply with JCoP requirements.
- As investigations of the Lane Cove incident continue, WorkCover is also investigating a fatal accident recorded on Sydney’s Cross City Tunnel. During excavation of its ventilation tunnel, a slab of rock fell from the crown killing one of the workers. The accident is currently before a Coroner’s Court of Enquiry with an initial hearing scheduled for late 2007.
- In addition to these construction incident investigations, financial collapse of the Cross City operating company, less than 16 months after opening to traffic, raises questions also about the viability of Lane Cove’s PPP economic model and of the new North-South Bypass Tunnel toll highway concession recently awarded in Brisbane. Independent studies report that these projects also rely on overly optimistic traffic/toll income forecasts and flawed financial concepts that leave them vulnerable and exposed to potential insolvency(10).
1 Northside Storage Tunnel Project; T&TI May 2000, p37 and Nov 2002, p28
2 Sydney Airport Rail Link; T&TI Sept 2000, p16
3 Cross City Tunnel, T&TI 2004, Feb p20 and Nov p32
4 WorkCover Lane Cover collapse incident report – http://www.workcover.nsw.gov.au/Pages/SearchResults.aspx and
5 Heathrow failures highlight NATM misunderstandings; Shani Wallis; TUNNEL Issue 3, 1999 – pdf download
6 Australia Code of Practice for Tunnels Under Construction; pdf download
7 HSE report on NATM safety; T&TI March 1997, p49
8 Risk Management of Tunnel Works in the UK; BTS/ABI Joint Code of Practice; http://www.britishtunnelling.org.uk/publications.php and pdf download
9 Lane Cove 2004 site visit report; T&TI, Nov 2004, p25
10 The fatal flaw in financing of private road infrastructure in Australia; John L Goldberg; Executive Summary pdf download; ATRF 2006 Conference Paper pdf download