John Harrison looks at Sydney's Lane Cove collapse and asks questions about the state of current practice.
Work seemed to have been progressing well at Lane Cove and ventilation tunnel driving had completed. At its end, where the intersection with the highway tunnel would be formed, excavation of the up-dip branch was well under way, and excavation by roadheader of the down dip branch was about to begin.
Then, at about a quarter to two in the morning, blocks of rock began falling from the crown and shotcrete started to peel from the roof. Within 10 hours the intersection had partially collapsed and had propagated to the surface, forming a substantial void and damaging property.
So, in November 2005, the Lane Cove project in Sydney went from being just another tunnelling project to one with a degree of infamy.
How did this happen- According to an investigation into the failure, the project was exemplary in being constructed in accordance with the best of industry practice.
Except, that is, for the facts that the intersection had a somewhat larger span than had previously been attempted in this particular rock, and that a weathered dyke cut through the intersection. Arguably, these features are the nub of the problem - not with Lane Cove, but with rock engineering in general.
They highlight the need to undertake rigorous engineering design, to continually and carefully monitor ground conditions and to be ready and able to revise the design to deal with changed conditions.
These three issues deserve detailed consideration. First, how do we move beyond precedent practice with our designs- Is it as a result of rigorous engineering research, by the construction of - possibly sacrificial - trial excavations, or by 'guess and try'?
Presumably most engineers would plump for the first of these. But given that most rock engineering design is fundamentally rooted in empiricism (eg empirical rock mass classification schemes, empirical strength criteria), the first job must surely be to replace this with something scientifically derived.
Lane Cove, like most projects that require substantial rock engineering design, used sophisticated numerical modelling to help in the design process, but how much empiricism is embodied in this practice- Where does the input data, like rock mass strength and stiffness and mechanical behaviour of discontinuities, come from?
Is the industry fully aware of the shortcomings posed by all of this underlying empiricism and is it happy with it- If not, is it prepared to instigate and support the fundamental research necessary to replace it- Or do ever decreasing margins rule this out?
The failure report notes the site geologist diligently recorded ground conditions. But can we guarantee this is always the case- Do we always have careful and continual monitoring- Do we genuinely have suf'ciently experienced geotechnical engineers to do this- Do we give them the time and tools to do so?
How often do we hear of monitoring being relegated to those few minutes when the miners are taking a break, and records being just a few hastily drawn field notebook sketches- There is high street technology available to allow production of fully 3D records of the rock mass, but is it routinely applied?
And finally, what about on-the-fly changes to the design- Are site staff sufficiently well versed in design to know when a change is required?
Are designers given the opportunity to properly revise designs- Does the form of contract explicitly allow time and cost for this?
Presumably, the insurers of the Lane Cove project will pay the collapse cost, which is likely to be substantial. How many trained geotechnical engineers and how much fundamental research could this have funded- And for how long will the insurance industry permit us to continue using current procedures?
John Harrison is senior lecturer in rock mechanic at Imperial College London