Late last year in Norway, in the quiet of Christmas night there were few motorists out driving between Sande and Holmestrand, to the south west of Oslo, which was fortuitous. Along that stretch of the E18, the dual carriageway nips in and out of twin bore tunnels (figure 1), among the longest being the 1.76km Hanekleiv tunnel. Just before midnight, the peace was shattered when, about 600m from the exit, part of the roof came crashing down.
Fortunately, as the dust settled, it became clear that no one had been hurt. Above, in the exposed syenite, a fissure was revealed up to 2.5m wide, slicing diagonally across the tunnel axis. The hole wasn’t too deep, mainly 3m-4m, but the damaged stretched over some 25m (T&TI, January 2007, p7).
Like in many other sections of the 7m high by 10.5m wide excavated tube, the crown had been supported by rockbolts and steel-fibre shotcrete. The rockbolts were mainly epoxy end-anchored and mostly 3m long on the project, as is common in Norwegian road tunnels. In a few sections they were 4m long. Many now lay mangled on the road surface amidst 200m3 of syenite mixed with clay, and also the shotcrete and PE foam from the water/frost proof inner tunnel lining that blocked the two, 3.5m wide lanes and shoulders.
Statens Vegvesen, the national public roads authority (NPRA), moved quickly to investigate the collapse as the Department of Transport (SD) appointed its own expert group to report by mid-February, which was around when the tunnel was expected to be reopened, according to early estimates. But not only does the southbound tube remain closed, it is not due to open before mid-year. Neither is the northbound tube, which was also closed for precautionary checking. To add to this, some neighbouring tunnels have also been closed for checks.
The checks of the tunnels began as the external expert group was beginning its investigation of the collapse site in the southbound tube at Hanekleiv. As part of their research, the group requested the original design and construction paperwork from the client. While there was no detailed description of the ground exposed at the location of the collapse from the time of the excavation, the records noted that the rock in the vicinity was generally more fractured and that there was some ‘thick, swelling clay’, or smectite. On site, following the collapse, the investigators did not find clay in the fissure; it was amongst the 600 tonnes of debris on the tunnel floor (figure 2).
Collapse precedents
Tunnel collapse in Norway is rare but the investigators realised that the fall at Hanekleiv would not have been the first time a layer of smectite might prove culpable in a tunnel crown collapse.
Norway has more than 1500km of road and rail tunnels, and prior to Hanekleiv there had been only two collapses – during construction of the Ellinsoy road tunnel, and at Kvineshei rail tunnel.
At Ellinsoy, in 1987 as road tunnel excavation was progressing from gabbro into a zone of crushed gneiss the fracturing got so bad, the ground weakening so much that the face collapsed. The ground ravelled upwards to reveal a near vertical fissure lined with clays, following which was a band of extremely fractured gneiss and then, not quite so bad, a mass of highly fractured gneiss, all of which proved to be far worse ground than the tunnellers had come through.
In the early 1960s, the Kvineshei rail tunnel was in operation in the south of the country, not too far from Kristiansand, when a stretch of the tunnel crown collapsed. Above the tunnel was a void of about 500m3 within a fault zone, again steeply inclined upwards (figure 3). While the geology was different, the fault structure has some similarity to the fissure at Hanekleiv and there was smectite in the weak zone before the fall. As the Hanekleiv investigators were to report, the rock may not have been similar but the mechanism was the same – swelling clay.
Near vertical faults have also given problems in other types of tunnels. The country has some 3500km of water tunnels in hydropower projects. Almost 30 years ago at the Hemsil I plant a cave-in blocked a tunnel that was filled with water, ready for generation to commence. The 11m2 rock tunnel had 50mm of shotcrete lining. It was learned that a fault was to blame, traversing the line of the tunnel at 45°. The 2.5m wide fault had collapsed and left behind a void 10m deep above the tube.
The key information drawn by the Hanekleiv investigating group, therefore, was not that there was insufficient pre-investigation but that it was known that fault zones had proved problematic in the past, especially with smectite in crushed rock. The phenomena had preceded the excavation at Hanekleiv – they were not unknown, just new. The investigators felt they’d shown that such risks, or potential collapse mechanisms, could be reasonably anticipated in general from the broad geological conditions, though the exact location of particular potential difficulties cannot always be pinpointed in advance.
At Hanekleiv, therefore, at the site where the particular problem did emerge, the investigators turned their focus on the swelling clay in the fissure.
Weak ground at Hanekleiv
The tunnels at the north end of the Hanekleiv twin bore pass through sandstone, which underlies a bed of slate and conglomerate tens of metres thick, and the formations are overlaid by basalt. The geology on the southern half is completely different, consisting only of younger, intrusive syenite. The edge of the mass of syenite is a near vertical boundary and has its own weak zones. There are parallel lines of weakness farther into the mass, including one about 200m away to the south at the fracture zone that saw the tunnel collapse (figure 4).
Smectites are formed by hydrothermal processes acting on discontinuities and other weak zones in bedrock, such as the syenite mass at Hanekleiv. The materials, though, vary widely in their properties and hence their relative risk. When samples of smectite from the rubble at Hanekleiv were tested in the laboratory, it was established that the particular swelling clay was not high risk, meriting only a ‘medium’ classification. The swelling pressure was barely 0.16MPa and only 14% of the sample had particles of less than 20um/microns. Clays classified as ‘highly active’ can exert pressures of more than 2MPa.
But the trigger mechanism for the crown collapse was still ‘very likely’ to be the smectite, according to the head of the expert investigation group, Bjorn Nilsen of the Norwegian University of Science & Technology (NTNU). Such relatively low expansive pressure, however, means such a material would take a long time to exert any appreciable influence or detrimental effect. Hence, it was concluded that this was why, so long after construction – 10 years later -the weakening effect of the swelling clay had progressed sufficiently to undermine the integrity of the local fractured zone.
But the swelling clay was not enough to cause a collapse. What was found to have compounded the problem of the zone of weak rock and smectite was inadequate tunnel support at the collapse site.
Given the previous fissure and clay-related problems for tunnel stability, there was ‘obvious’ precedent and so particular caution was needed in determining the adequacy of tunnel support options in areas of geological weakness, even over relatively short stretches. Therefore, when assessing what had happened at Hanekleiv, it was important for the expert group of investigators to go further into the records and learn about the choices made, and rationale, during the construction of Hanekleiv tunnel.
The twin tubes were built following some years of planning and design by NPRA, which had traditionally been the project developer, had design input, and also would build and operate roads. In the early, pre-investigation work the client was helped by Norconsult, though the consultant did not become involved in the execution phase, in the mid-1990s. To deliver the project, NPRA used its own resources.
However, at the time there were changes being introduced to the traditional, integrated structure of NPRA. In the re-organisation the construction arm was separated and ring-fenced from the planning, ownership and operating functions but it remained within the organisation, to be prepared for eventual spin-out. In 1996, the Hanekleiv road project was the first major scheme to be executed under this new arrangement, which, in effect, saw two new organisations within the one group undertaking the job.
Excavating Hanekleiv
Records from the time, and site inspection, also showed that rockbolts and shotcrete were not the only type of tunnel support employed on the project. Beyond more extensive use of the technique plus spiling at the face in more difficult ground, where the geology was particularly poor – such as at points earlier in the excavation, and also at other parts of the twin bore – the tunnellers had used in-situ concrete lining. At the location of the later tunnel crown collapse, it was decided not to employ concrete as a more substantial lining (figure 5).
“The wrong support was used,” says Nilsen. By selecting only rockbolts and shotcrete for excavation support at the collapse location, the ‘wrong judgment’ was made, he told T&TI.
However, there are no clear and transparent reasons in the project records that explain why engineers decided not to use in-situ concrete lining at the collapse location, and nor is there a detailed description of the observed, local geological features when the ground was opened at the time. Yet, as Nilsen further observes, selecting a support method for a particular geological situation is ‘not a simple decision to take’ in all cases.
Beyond the official records, the investigators also contacted individuals involved in building the Hanekleiv tunnel. What became more apparent in the course of the discussions was that there had been difficulties with clarity in decision-making during the period. It is understood to have been a direct consequence of the organisational changes as the two, newly-hatched sister units were getting on their feet within NPRA.
The result on site was the use of rockbolts and shotcrete to pin and hold a weak zone in place. While the method employed clearly worked for a time, it was not a durable solution in light of the gradual pressure from within the fractured rock mass and exerted on the rockbolts by the smectite (figure 6).
Nilsen says it is ‘very obvious’ what the problem was, technically. But organisationally, the investigators’ remit in preparing their report did not involve further probing into decision-making responsibility for construction choices made within NPRA at the time.
Around two years after the tunnel was excavated, NPRA called in the Norwegian Geotechnical Institute (NGI) to check the northbound bore for possible strengthening requirements. The contractor installing the inner tunnel lining, for frost isolation and waterproofing, had expressed concerns to the client. As a consequence, extra rockbolts and also some concrete lining were installed in parts of the tube. The E18 in the area opened in late 2001, and its structures included 7km of tunnel and 2km of bridges, and had apparently functioned well.
Now, NPRA faces consequential challenges of the Hanekleiv findings, not only of repairing the damaged southbound tube but also of deciding if, how and when to assess many of its other tunnels. The adjacent, northbound tube is already having extra full in-situ concrete lining built in three locations. How much, if any, extra support might be needed in the southbound bore will have to await the end of investigations.
T&TI was also told that beyond Hanekleiv, the public roads authority is preparing for the possibility of a wide programme of checking tunnels, beginning in the vicinity in southern Norway, and then the need elsewhere will be judged. A spokeswoman for NPRA said it was too early to detail the survey plans.
How the aspects of the Hanekleiv tunnel collapse might, or should, be adopted in tunnel checks remains to be judged. NPRA has its own, internal review group at work, reviewing the external report as well as procedures. What might be reasonably concluded at present from the investigators’ findings, however, is the need to closely heed the lessons of history, most especially where tunnellers have faced choices over support near zones of fissured rock, and especially where smectite is present.
Where such decisions have been required amidst a changing organisational culture, the additional lesson of Hanekleiv is that there may be benefits from attentive hindsight as well as promoting foresight. Precedence in the tunnel collapses in Norway has, unfortunately, a new dimension. Yet, that this awareness has been garnered can also be seen as fortunate, though, for the benefit and safety of the country’s tunnel assets. For tunnellers elsewhere, it can also raise awareness or be a reminder about organisational risk.
Location of Hanekleiv twin tunnels on E18 near Sande, south west of Oslo. Nearby tunnels have also been closed Figure 1 – Location of Hanekleiv twin tunnels The Hanekleiv tunnel suffered a crown collapse on Christmas Day, 2006
Hanekleiv crown collapse Plan and sections showing voids created at site of crown fall in southbound tube Fig 2 – Plan and sections Investigators reviewed previous tunnel collapses in Norway due to inadequate support in zones of geological weakness. The crown at Kvineshei rail tunnel collapsed in the early 1960s due to the same mechanism as at Hanekleiv – swelling clay Figure 3 – Kvineshei rail tunnel collapse Longitudinal section showing geological formations Figure 4 – Longitudinal section of Hanekleiv Indicative arrangement of excavation support along with inner water/frost lining and its positioning bolts Figure 5 – Indicative arrangement of excavation support Investigators believe swelling clay in weak rock plus inadequate excavation support were to blame for the collapse Swelling clay in weak rock plus inadequate excavation support were to blame Position and length of rockbolts at site of largest opening in tunnel crown prior to the collapse Figure 6 – Position and length of rockbolts
