FAILURE OF under-designed rock bolts triggered the collapse of the Lane Cove motorway tunnel in Sydney, according to an independent report published last month.
The collapse occurred during construction on 2 November last year at the junction of the Pacific Highway exit ramp tunnel and the Marden Street ventilation tunnel in north Sydney (GE January 2006). At this point the spans were larger than those for a single bore section.
The tunnel roof caved in an area where ground conditions where known to be poor, the report said.
Rock bolts at least 2m longer than the 5m bolts used should have been incorporated to support the tunnel roof, said report author Ted Brown, senior consultant at Golder Associates Pty. After the collapse, a 6m diameter, 25m deep hole opened up from street level down to the tunnel, undermining a block of flats.
Construction of the tunnels was by NATM with road headers used to excavate rock. Up to 2.5m was excavated at a time to form a 7m high, 8.1m wide bore. A 100mm thick shotcrete layer was then applied before 4m long rock bolts were installed on a 1m by 1m grid.
A secondary support system of 100mm of shotcrete and 5m long rock bolts installed on a 1.5m by 1.5m grid was then applied before a final shotcrete layer to cover the bolt heads.
Ground conditions in the collapse area were weathered, highly jointed and faulted Ashfield Shale, with a near vertical and weathered dolerite dyke passing through it at the junction of the two tunnels. Many of the joints were clay filled and sometimes slickensided.
The joints, faults, dyke and excavation boundaries acted together to isolate blocks that were free to fall or slide if not adequately supported, the report said. This led to progressive collapse.
Brown's investigation also found that the dyke, which was encountered throughout the course of tunnelling, had not been not picked up by the site investigation and was not included in design for the junction.
'Under these circumstances, the junction is regarded as an excavation meriting special design attention, particularly in terms of the provision of support. But in the design stage, no special analysis of the junction was carried out.' Brown said the effective span of the tunnel at the junction was underestimated. Although calculating the effective span was a matter of opinion, he believed it to be 21m. Design and build contractor Thiess John Holland (TJH) calculated it at 17m.
The report continued: 'Spans of 17m to 22m are very large effective spans for a material as weak as the weathered Ashfield Shale and could well be outside the limits of precedent practice in this material.' Brown added that rock bolts at least 7m long should have been installed at the tunnel intersection.
But he praised TJH for its best practice approach. 'Some of the documents setting out these processes are models of their kind, ' the report said. 'Up to the time of the incident, designs and processes had been executed in a highly professional and productive manner.' TJH has continued tunnelling the rest of the twin 3.6km motorway tunnels using the same method.
A company spokesperson said the tunnel would be diverted from the collapse site.
Timing of a tunnel collapse
2 November, 1.38 am: Small blocks of rock and shotcrete fall from the crown of a recently excavated tunnel section. Fractured rock against the dyke surface falls into the tunnel, destroying the structure's arching action. Shotcrete begins to peel away as the void caused by falling rock enlarges.
Rock, including rock bolts, fails along the vertical plane of a fault and the dyke and horizontally along joints.
1.40am: Workers evacuate as the void propagates upwards and towards a retaining wall under Epping Road and the foundation of the block of flats above within 20 minutes.
2.30am: A hole up to 6m in diameter opens up above the tunnel.
6.50am: Longueville Road closed to traffic.
8.25am until noon on 3 November: concrete pumped in to stabilise the hole.