Northside Storage Tunnel, Sydney: investigation, design and construction R. A. Gee Connell Wagner Pty Ltd C. J. Parker Coffey Geosciences Pty Ltd R. J. Cuttler Transfield Pty Ltd
ABSTRACT: The Northside Storage Tunnel Project comprises 22km of tunnel driven with 4 TBMs of 3.8m to 6.6m diameter and 250,000m3 of other tunnels excavated by roadheaders. The A$460million project, realised under an Alliance between Sydney Water, Transfield, Connell Wagner and Montgomery Watson, commenced in February 1998; portal construction started that July and all tunnelling was completed 24 months later. The paper describes the geotechnical investigations and aspects of the design and construction of the tunnels. Reference is made to: innovative geotechnical investigation tools; the investigations for the crossing of Middle Harbour; forward probing and grouting work during the drive under Middle Harbour; the impact of high in-situ stresses; the design of the ground support; high-speed TBM tunnelling and the selection of fibreglass bolts; observations on the level of support employed; tunnel spoil handling and disposal; construction rates and the benefits of an alliance approach to such works. 1 INTRODUCTION The Northside Storage Tunnel is a hard rock TBM driven tunnel system 20km long and varying in diameter from 3.8m to 6.6m. It was designed to capture and detain the major storm overflows from four points along the existing Northern Suburbs Ocean Outfall Sewer (NSOOS) to reduce the raw sewage entering and polluting Sydney Harbour. The overflow points are situated at Lane Cove, Scotts Creek,
The underground works also included the excavation of 3km of declines and associated tunnels, seven major caverns and ten shafts using of a range of excavation equipment and techniques (ie. roadheaders, tunnel boring machines, raiseborers, rock hammering and splitting, and d rill-and-blast). A total of 750,000m3 of ground was excavated. To realise the A$460million project, Sydney Water formed an alliance with engineering consultants
Scotts Creek Middle Harbour Crossing
Lane Cove West North Head Site
The North Shore
Tunks Park Site
Figure 1 Alignment of Storage Tunnel System Tunks Park and Quakers Hat Bay. The connection point for the Storage Tunnel to the existing system is at the Sewage Treatment Plant at North Head, Manly (Figure 1).
Connell Wagner and Montgomery Watson and construction contractor Transfield to undertake the design and construction of the facility based on the EIS concept design (Henderson & Cuttler, 1999). The site investigations, interpretation of the data and provision of geotechnical advice to the Alliance,
throughout the project, were undertaken by Coffey Geosciences. The project commenced in early 1998 and, as scheduled, the tunnel system was ready to receive overflows prior to the Olympic Games in Sydney in September 2000. The tunnelling works had taken exactly 24 months from groundbreaking to final breakthrough. 2 GROUND INVESTIGATIONS AND IMPACT OF FEATURES ON CONSTRUCTION 2.1 General Tunnel Geology The geotechnical investigations carried out during 1998 comprised the usual activities of cored boreholes (45 no.), surface mapping, seismic, magnetic and gravity surveys and laboratory and in-situ testing. In addition, cross-hole and surface-to-borehole seismic tomographic imaging and downhole camera imaging (RaaX) were employed in selected locations. Further works were undertaken during the project, as the need for more information in certain locations became apparent. This was particularly the case at features identified as having higher than normal risk, such as Manly, Middle Harbour and the Lane Cove River. The geotechnical investigations undertaken for the project identified that the tunnel would be constructed within sub-horizontally bedded rocks occupying the eastern portion of the Sydney Basin. The majority of the tunnel length (over 90%) would be excavated through the Triassic Hawkesbury Sandstone, while the eastern 1.5km of the tunnel beneath North Head would be excavated in the underlying Newport Formation rocks (Narrabeen Group: Triassic-Permian). The Hawkesbury Sandstone formation is up to 280m thick. The Hawkesbury Sandstone rocks comprise high strength, quartzose sandstones and lesser shales, siltstones and laminites. The fresh sandstone typically has an unconfined compressive strength of about 40Mpa. Three distinct facies types, formed by differing deposition processes, are present in the Hawkesbury Sandstone. The three facies are massive sandstone, cross -bedded or sheet sandstone and shale/siltstone facies. Each of the facies, typically 0.5m to 3m thick, has different engineering properties which influenced tunnel construction (in particular the laminite interbed units). The percentage of each of the facies types in the project investigations was as follows: • massive sandstone facies (22% of core) • cross-bedded or sheet sandstone facies (70% of core)
• shale/Siltstone facies (8% of core) The occurrence of the shale/siltstone facies is more prominent in the basal third of the Hawkesbury Sandstone with the tunnel alignment intersecting two major units of about 10m and 19m thickness between Middle Harbour and Manly. The underlying Newport Formation consists of subhorizontally bedded sandstones, fine to medium grained with some siltstone laminae, and siltstone/laminite lithology. The investigation intersected approximately 60% siltstone/laminite beds and 40% sandstone beds in the Newport Formation. The stratigraphy along the tunnel route has a regional dip down to the southwest of about 1° At North Head, on the eastern margin of the basin, a dip of 2.5° down to west across the 1.5km wide headland. Rock structure assessed to be encountered in the tunnel horizon within the Hawkesbury Sandstone and Newport Formation included the following discontinuities and features: • tight, sub-horizontal bedding planes of similar strength to the rock mass • bedding plane contact seams • incipient cross bedding at angles of 15° to 30° within individual sandstone beds • sub-horizontal very fine sandstone laminae within the siltstone lenses and interbeds • subvertical jointing or joint swarms • low angled thrust faults and bedding shears • higher angled displacement faults and shear zones • basalt dykes 2.2 Deeply Incised Palaeochannels At the EIS stage of the project, it had been decided that tunnelling would be much simplified and high rates of progress made possible if the drives avoided mixed faces. To achieve this an alignment was set which p assed below the deeply incised palaeochannels at Manly and Middle Harbour. This led to a deep tunnel aligned in rock throughout the tunnel length (Figure 2). Most of the tunnel route was therefore to be excavated in high strength rock with some 50m to 150m of rock cover. The paleaochannels, formed during periods of fluctuating sea level over the last 150,000 years, are up to 70m deep and in-filled with marine, alluvial and back swamp deposits. More sophisticated investigation methods were warranted to identify the geotechnical risks associated with tunnelling in rock beneath these features. Previous rock tunnelling experience beneath palaeochannels in Sydney indicated mixed success in relation to poor ground and high groundwater inflows. (Cable tunnel from Greenwich to Balmain, 1913-1924; Parramatta River Water Supply Tunnel
Figure 2 â€“ Longitudinal section of main tunnel from Rose Hill to Rydalmere, 1926-1928; Lane Cove siphon tunnel, 1928-1930.) An assessment of the depth and location of the palaeochannel floors at each location was necessary to refine the tunnel vertical and horizontal alignment. Initial concerns were expressed that the incised channels were controlled by pre-existing sub-vertical structural features such as faults or dykes, with the features extending to the tunnel horizon. Targeted inclined boreholes coupled with down -hole camera imaging of defects (RaaX imaging), in -situ stress measurements and surface-to-borehole seismic tomographic imaging were used to characterise the rock beneath the palaeochannels (Parker et al, 1999). The investigation data identified that the rock beneath the palaeochannels contained primarily low angled thrust fault and bedding plane shear features. Such features were assumed to be related to horizontal stress concentrations in the floor of the channels. 2.3 Manly At Manly, the palaeochannel floor was identified at a depth of 68m beneath the ground surface of the Central Business District. The invert level selected for the 6.6m TBM drive was 95m below sea level, which obtained 23m of rock cover. Bedrock beneath the channel floor comprised interbeds of sandstone, shale and siltstone from the Hawkesbury and Newport Formations. Some thrust fault shears and subvertical joint swarms were assessed to be within the proposed tunnel horizon over a relative short tunnel length of less than 100m. Permeability testing in boreholes either side of the palaeochannel floor suggested relatively tight rock but with higher permeability immediately beneath the channel floor. Horizontal rock stress measure-
ments also showed results consistent with high stress concentrations beneath incised channels, as noted by other workers in Sydney. When tunnelling progressed beneath the Manly palaeochannel, high strength and relatively tight rock was encountered. Though forward probing was undertaken to prove rock quality and provide information on likely groundwater inflows, the tunnelling progressed relatively quickly. Stress concentrations in the channel floor and around the tunnel opening were evident in the form of roof spalling at the tunnel crown. Low angled thrust faults of less than 200mm thick were mapped in the tunnel over a length of 60m and correlated to features identified in the investigation borehole logs. These features, together with several swarms of north-south trending sub-vertical joints yielded small groundwater inflows of less than 2L/s over the 60m tunnel length. No forward grouting was deemed necessary as the TBM progressed through this area. However, some months after the TBM had passed through the area, a small roof fall occurred, which temporarily halted tunnelling. The fall appeared to have been caused by groundwater penetrating the interface between the shale just above the crown and the overlying sandstone. Increased water inflows occurred as a result and the installation of steel sets throughout the area was carried out. As in other areas where steel sets had been installed, these were regarded as temporary and the permanent support, installed later, was a specifically designed steel fibre reinforced shotcrete ring.
2.4 Middle Harbour The tunnel alignment across Middle Harbour is from Parriwi Head at The Spit to Clontarf Beach on the northern foreshore. Middle Harbour is a drowned river valley which is tidal and flows eastward to connect to Sydney Harbour some 2.5km downstream. At the crossing point, Middle Harbour is 250m wide though the skewed alignment of the tunnel resulted in 500m of tunnel length being beneath open water. The maximum water depth is about 20m. The geotechnical investigations identified the base of the palaeochannel to be at a depth of about 67m, located beneath the shoreline at Clontarf Beach. The palaeochannel was assessed to be about 70m wide with a deeper, 30m wide section. Detailed investigations into the palaeochannel floor beneath Clontarf Beach involved vertical cored boreholes and three inclined cored boreholes each about 100m to 150m in length. RaaX downhole camera imaging was performed in the inclined boreholes to provide defect orientation and width characterisation. In-situ permeability testing, an extended period pump test and groundwater chemistry testing were carried out. Surface-to-borehole and boreholeto-borehole seismic tomographic imaging was performed to confirm bedrock levels and attempt to assess low velocity bedrock zones between boreholes (Parker et al, 1999). In -situ stress measurements were conducted in one of the boreholes. Stratigraphy intersected by the boreholes drilled beneath the palaeochannel consisted of 37m of sandstone overlying an unusually thick (19m) siltstone/laminite interbed unit within the Hawkesbury Sandstone. Close examination of bedding plane contacts in the core enabled correlation of specific beds. Groupings of several similar beds were made to develop a geotechnical model with nine sub-horizontal geotechnical units. The 37m thick sandstone stratigraphy also contained numerous zones of “no core”. These zones, up to 1.6m thick, were assessed to occur parallel to bedding in the more massive, coarser grained beds or as low angled shear zones and thrust faults. RaaX camera imaging in the inclined boreholes indicated the “no core” zones consist of fragmented sandstone or very low strength sandstone. Very high permeability was recorded within the fractured sandstone rock and good hydraulic connection was measured between boreholes up to 150m apart. Horizontal stress measurements suggested a lower than expected prevailing principal stress whereby the stress may have been partly relieved. Similarly, the extensive structure in the rock mass, a small anticlinal fold beneath the channel floor and the high
permeability, was assessed to be a consequence of valley floor stress relief (valley bulging). This was the key crossing in terms of fixing the tunnel vertical alignment for the whole project. An alignment was selected which placed the invert at 91m below sea level and obtained some 20m of rock cover. Significant tunnelling risk was identified for the crossing in the form of potentially high groundwater inflows, consequential dewatering induced surface settlements and areas requiring heavier tunnel support. Groundwater inflows of well in excess of 100L/s were predicted for a “wished -in-place” tunnel unless forward grouting was undertaken to reduce the permeability around the tunnel. Tunnelling progressed slowly beneath Middle Harbour with the Alliance adopting a strategy of forward probing and grouting from the TBM. Forward probing and grouting was carried out over a tunnel length of about 210m which equated to the area beneath the palaeochannel assessed to be have been influenced by valley bulging. Up to 54 grout holes were drilled prior to each 4 to 6m advance of the TBM. Groundwater inflow in the tunnel after the grouting was estimated to have been reduced from a potential 200L/s based on back-calculated permeabilities to 18.5L/s for this section of tunnel, i.e. a grouting “efficiency” of some 90% (Ashton et al, 2001). Subsequent chemical grouting of discrete inflows reduced the total inflow in this section to the order of 10L/s. Steel sets (as initial tunnel support) were required over a tunnel length of about 120m. This zone represented the area beneath the palaeochannel floor where significant low angled defects had been identified in the rock cores and where the seismic tomographic imaging had detected a low velocity zone beneath the southern flank of the palaeochannel. Tunnel mapping beneath the palaeochannel revealed primarily high strength sandstone with narrow defects contributing to much of the groundwater inflow. However, a 45m length of tunnel, containing more intense low angle thrust faulting with displacements and shears, was mapped on the southern flank of the palaeochannel in the area of the seismic anomaly. Alternatively, the more intense shearing may also represent a bend in th e palaeochannel floor. 2.5 Lane Cove River The last of the major palaeochannels to be crossed was at the Lane Cove River within the final 150m drive for the 3.8m Wirth TBM. The site investigation here comprised three boreholes, one of which
was an inclined hole drilled from a berm out in the river, aimed at intercepting the base of the palaeochannel. In the case of this crossing, a precedent had been set by a similar sized tunnel driven by drill and blast for the NSOOS siphon crossing of the river in 1928. The records for that tunnel were very sparse, the only information being available on conditions encountered coming from some record photographs and the annual reports of the Water Board directors. It appeared that the tunnel had encountered significant saline water inflows, but not more than could be handled by the pumps. The new tunnel was to be at a similar horizon but some 30m downstream, possibly placing it in significantly different conditions. Being TBM driven, this tunnel would not have been able to tolerate excessive water inflows without the ability of the machine to muck being affected. The floor of the palaeochannel was located at 30m below sea level; the alignment of the tunnel was lowered to place the invert at 43m below sea level giving some 10m of cover. Ground conditions were assessed to be in good but fractured sandstone. A low velocity zone was identified, which could have represented a zone of significant water inflow. The alignment was subsequently probed by a 140m long horizontal hole drilled from the reception shaft on the west bank of the river due to the difficulty of carrying out such an operation from within a relatively small TBM driven tunnel. The probe did not encounter any significant features. In the event, whilst tunnelling ben eath the channel, the rock was found to be of good quality but significant water inflows (>12L/s) were encountered through a bedding -parallel shear zone just beyond the reach of the probe hole. Grouting was largely ineffective but the inflow diminished with time and the TBM was able to complete the drive. The inflow eventually reduced to a level of 5 to 7L/s and was, in terms of the tunnel design, considered to be beneficial for flushing the tunnel invert. 2.6 High In-situ Stresses Stress testing carried out during the investigations suggested that the ratio of horizontal to vertical stress is generally in the order of 2-5, with the horizontal components often reaching magnitudes of 1316MPa in an approximately north-south direction. Failures in the tunnel crown and floor and movement along weaker seams, such as sandy laminae in shales, were expected to result from these high locked -in horizontal stresses in regions where the tunnel orientation was unfavourable (ie. east-west). Qualitative estimations of in-situ rock stresses could be made during tunnelling operations by observing
the modes of tunnel failure, while taking into consideration factors such as rock type, rock strength, defect nature and abundance, and local topography. Failure in the crown or invert was not, of itself, su fficient grounds to confirm high in-situ horizontal stresses. In a low strength stratum, such as a shale breccia, or in areas containing abundant defects, such as joints or tension fractures, moderate or major spalling can occur with relatively low in-situ stress conditions. In contrast, spalling in massive or medium to high strength sandstone or shale, with few significant defects, was taken to indicate a high in situ stress field. Such high-stress areas are sometimes associated with anticlinal warps resulting from the buckling of stressed strata. In many areas, high regional in -situ stresses have been relieved by horizontal shearing prior to tunnel excavation. This is particularly the case beneath the more deeply incised palaeochannels and associated valley walls, reflecting the process known as valley bulging. It is also noted that high in-situ rock stresses can also be relieved, in part, by reactivation of normal faults in a reverse sense. Overbreak was common throughout the roadheader excavation of the declines and chambers for both North Head and Tunks Park. It was generally associated with the intersection of bedding planes and joint zones in the crown of the tunnel. However, high stress -induced spalling in the roadheader excavations was rare. During the excavation process the approximate depth of spalling and overbreak throughout each TBM tunnel was observed and recorded. Graphs of the combined spalling and overbreak depth were plotted against chainage along each TBM drive to make observations on the likely stress regimes and to attempt to predict where significant spalling might occur. Spalling caused significant problems in the main east west drives from Tunks Park and in the initial section of the drive to Scotts Creek before the orientation of the latter became approximately north south. Breakage in the invert caused problems with the trackwork for the supply trains and in the crown necessitated more bolting than would have been required from structural considerations alone, to support precautionary mesh. The damaged invert was also a contributory factor to the decision to concrete the tunnel invert throughout to minimise ponding of the effluent.
3 TUNNEL DESIGN 3.1 The Design Process The challenge in the support design of the underground works was twofold: to maximise the benefits bestowed by the generally benign tunnelling env ironment in the Sydney Basin and to develop designs which assisted the constructors to maximise production rates. Design development commenced with determination of the arrangement of the tunnel components needed to satisfy the operational requirements of the project, the hydraulic designers and the constructors. This was a very time consuming phase which was kept live well into the detail design and construction phases to suit construction constraints and priorities which varied as the project progressed. The arrangement of the components was then developed in the light of construction methods and s equences. An early constraint on the use of explosives on the project led to extensive use of roadheaders in the access and pump station works. In order to maximise rates of advance, high power machines were employed (Voest Alpine AM105s and Mitsui S200s and S300s). The size of these machines and their limit ed manoeuvrability was a significant determinant of the detailed geometry of the openings, junctions and intersections. The staged construction of large chambers and junctions which required heading and bench necessitated careful consideration of the sequencing, access and approach directions. As would be expected, TBMs were selected to drive the main storage tunnels. The approximate size of each drive was set out in the concept design to yield a tunnel storage capacity of between 470 and 510 megalitres. The actual final tunnel diameters were determined by the TBMs available on the world market at that time. In addition, an analysis was carried out to determine which machine would be appropriate for each drive given each machineâ€™s capabilities, the project program and physical constraints and obstacles. The TMBs presented little difficulty in terms of geometry although selection of the precise alignment was a protracted affair through determining the safest route across Middle Harbour from North Head to the North Shore. Support for the TBM drives was tailored for each machine based on its stroke length and gripper pad arrangement. 3.2 Design Methods The tunnel design team itself carried out the rock mass classification from the cores yielded by the site investigation campaign in order to gain a better understanding of the tunnelling media. Two classifica-
tion systems were used, the Q-system (Barton et al, 1974) and the RMR system (Bieniawski, 1989). Co rrelations were carried out between the results from the two systems to ensure that the determinations were consistent. It was decided at the outset that support design would be based on a number of methods depending upon the circumstances in each case. The emphasis would be on empirical methods with assistance given by numerical analysis using commercial software to identify and address areas where particularly severe loading conditions were likely to be encountered. The rock mass classifications were used to provide initial ranges of support requirements from the chart prepared by Grimstad and Barton (1993). The range was determined for each location based on best, average and most conservative interpretation of the available information (bearing in mind that borehole spacing could be many hundreds of metres). 3.3 Design Tuning However, Connell Wagner, as the designers of the rock tunnel section of the New Southern Railway in Sydney (now the Airport Link Railway), also constructed in the Hawkesbury Sandstone, had access to the records of the initial support installed on that project. It was therefore possible, through back analysis, to assess the relationship between the support regimes suggested by empirical methods, based on rock classification systems, and the support actually installed. Experience has shown that these empirical methods, which were developed from rock with randomly oriented joints, tend to result in conservative support prediction. Their direct use therefore fails to benefit from the features of the Sydney basin rocks that are favourable for tunnelling, namely sub -horizontal bedding and sub-vertical, low persistence jointing. The analysis was carried out for the Q-system. It was also observed that the Q values derived from the cores were heavily dependent on parameters which are difficult to determine from cores such as the Jn value (indicative of the number of joints) and the Jr value (joint roughness). Wherever there is uncertainty about a design parameter, the designer will tend to err on the side of caution often leading to heavier and more uneconomical support than might be necessary. It was therefore apparent that a better understanding of the geological characteristics of the rock mass was necessary before undertaking the Q rating. (Asche & Quigley, 1999). The output from the study enabled a range of more economical support types to be derived for each section of the work from which the appropriate support for the ground
could be more readily selected, as the rock condition was exposed. However, in the event, more support than was indicated was often installed by the TBM crews for practical and other reasons. Provided such additional support did not impact on TBM progress it was not disputed. 3.4 Use of Shotcrete An early decision was made, in conjunction with the constructor, to avoid the use of shotcrete as a component of the initial support to simplify the construction process. Omitting the shotcrete removed the need for a third process at each advance after excavation and installation of initial support, and eased the burden on the ventilation. This applied to all the underground works. The decision required that the support predicted by the Grimstad/Barton chart be modified. The omitted shotcrete had to be replaced by additional bolts in the form of a tightened up bolting pattern. The revised pattern was determined using a method developed from the work of Bischoff and Smart (1975). The numerical analyses, using commercially available software, were based on both the continuum and discontinuum models. The continuum models were used primarily for intersections, tunnels in close proximity and complex chambers. The discontinuum models are more appropriate for the analysis of excavations in the blocky Sydney Basin rocks. The leading two-dimensional discontinuum modelling package UDEC (Itasca Consulting Group Inc., 1996), a distinct element program, was used. Normally, the actual joint locations and orientations are not known with respect to the tunnel and statistical parameters about the joint sets are used. UDEC is then run as a series of “dice-throws” to see what happens. The output remains statistical but it gives a good indication of what can be expected in reality. However, being a two-dimensional model, threedimensional effects are not recognised. Again, by using data from the Airport Link Railway it was possible to recreate the observed behaviour in that tunnel by modifying the jointing representation. The modified jointing representation was then applied to this project to assist in determining the required support for the TBM drives (Figure 3 - UDEC modelling of TBM tunnel support) and the optimum geometry of the crown of the 15m span Raw Sewage Pumping Station chamber. 4 HIGH SPEED TUNNELLING The NSTP was recognised from the outset as a very tightly programmed project requiring high speed tunnelling, particularly of the storage tunnels. It was
Figure 3 – UDEC modelling of TBM drive support also recognised that, unusually for a sewage facility, the storage tunnels would be accessible for maintenance for substantial periods of time. The latter criterion meant that an economical roof support in these tunnels could be developed bearing in mind that a higher level of maintenance of the support system would be possible. The former criterion meant that a support system compatible with fast TBM drivage would be needed. A one-shot chemically anchored and encapsulated system met this requirement. The chemically anchored rockbolt is also the preferred system in the ground through which these tunnels were driven due to the prevalence of siltstone and shale bands which often cause mechanical end-anchors to fail. The drawback of chemical anchorage and encapsulation is that full encapsulation is not certain. This inferred that a normal steel bolt would be particularly prone to corrosion at “holidays” in the protection afforded by the encapsulation but also at cracks developing in the resin as the bolts come under load as the ground relaxes. These cracks would occur at just the locations where the bolts are most vulnerable: where they are most heavily stressed. A variety of solutions to the corrosion issue were investigated; these included stainless steel bolts, epoxy coating, metal coating, galvanising, composite stainless and high tensile steel bolts and even the use of impressed current corrosion protection. The conclusion was reached that a fibreglass bolt with similar capacity to a steel bolt would provide the required support and be less prone to corrosion since the load carrying elements, the glass fibres, were effectively inert. GRP bolts of 30 tonne capacity were readily available in the market being much used in the coal mining industry where cuttable support is required in the mining process. A 50 year working life was determined to be a reasonable expectation.
It was possible, with UDEC, to model the properties of the rock reinforcement and it was determined that there was no significant difference between the support offered by either steel or fibreglass bolts. It appeared that the flexibility of the fibreglass bolt could sometimes overcome its inherently lower shear strength. This determination enabled the decision to be made to use 30 tonne fibreglass bolts in the TBM drives. 5 TUNNEL SPOIL HANDLING AND DISPOSAL Removal of spoil from the tunnelling operations was a major issue on the project. Because of the location of the major tunnelling worksites in highly developed residential areas, an alternative to spoil removal by road transport had to be found. The final arrangement entailed spoil removal from the tunnel boring machines by continuous conveyors and removal from the underground works by a co mbination of inclined, vertical and horizontal conveyors to barge loading points on the harbourside. Eventually there were over 25km of conveyors employed on the project. Barges transported the spoil some 18 km across Sydney Harbour to a commercial railhead in White Bay. From there spoil was transported, predominantly by rail, to points on the western outskirts of Sydney where it was used for industrial development earthworks. The barge loading point for the works at North Head was located at the end of a 1.4km long conveyor tunnel at Little Manly Point Reserve in Spring Cove, North Harbour. Spring Cove is the habitat for Sydney Harbour’s only colony of Little Penguins. Problems to be overcome at this site were the prevention of disturbance to the pengu in colony and the avoidance of damage to sea grasses in the dedicated marine park area. After completion of the spoil conveyor tunnel, the wharf and noise shielded barge loading facilities were constructed at Little Manly Point and the spoil conveyor from North Head was installed in the tunnel. A similar spoil handling facility was installed at the other main construction site at Tunks Park, Cammeray, within an environmentally sensitive residential area and adjacent to parkland and a public recreational reserve. Stringent noise and space constraints had to be overcome to allow the project to proceed. Two underground caverns were constructed for storage of TBM spoil overnight and to allow tunnelling to continue 24 hours per day. Offices and amenities buildings were constructed on piles over the bay and all surface workshops and barge loading facilities were totally enclosed within buildings lined with acoustic suppression cladding
designed to minimise noise transmission and disturbance to the local community. 6 CONSTRUCTION SUMMARY During the two -year tunnel construction period, 21km of TBM driven tunnel, 250,000m3 of declines, caverns and associated tunnels and 1050m of vertical shafts were constructed. Two underground pumping stations, one located 100m below sea level and each of over 16MW installed pump capacity, constructed at the North Head STP also required the placement of over 7000 m3 of concrete. A total of 1,860,000 tonnes (750,000 bcm) of spoil was removed from the tunnelling works and transported by barge and train to a remote location for beneficial re-use. This involved over 2300 barge movements and over 1000 train loads. The four tunnel boring machines achieved a combined best advance in one week of 1093 m. Individually, the peaks of performance for the largest machine (6.6 m diameter Wirth TBM) and the smallest machine (3.8 m diameter Wirth TBM) were as follows: Best shift (12 hr): Best day (24 hr): Best week (7 day):
6.6 m φ 47 m 70 m 328 m
3.8 m φ 57 m 80 m 377 m
7 THE ALLIANCE ADVANTAGE The integrated approach of an alliance contractual arrangement allowed the Northside Storage Tunnel to be delivered as a true, ‘fast-track’ project. At a capital cost of approximately A$460 M, the project was conceived, planned and delivered in a total period of less than four years. A significant feature of the alliancing agreement was the ease with which mutually acceptable designs could be developed. The close working relationship between Sydney Water, the designers and the constructor enabled the design to respond rapidly to changing construction circumstances and for decisions on design to be agreed and implemented with the minimum of delay. Amongst many examples, the advantage of the alliance contractual arrangement was clearly evident when extremely difficult ground conditions were encountered in sections of the tunnel crossing under Middle Harbour. Identified as a high-risk undertaking, this area required a forward probing and grouting regime to be implemented from the TBM to reduce the ground water inflows from a potential maximum of 200L/s down to an actual 10L/s. The
alliance arrangement allowed a total focus to be maintained by all parties on overcoming the construction problems. Under a conventional contract serious disputes could have arisen over the allocation of responsibility for such ‘latent’ conditions, leading to inevitable delays and associated cost increases. 8 REFERENCES - Asche H R and A Quigley. Tunnelling Design on the Northside Storage Tunnel Project. Tenth Australian Tunnelling Conference, Melbourne, Australia, March 1999 - Ashton, G B, R A Gee, P N Groves and S T Warren, Northside Storage Tunnel Project, Sydney: Tunnelling and Groundwater Control Beneath the Middle Harbour. Rapid Excavation and Tunneling Conference, San Diego, USA, June 2001 - Barton N R and E Grimstad. The Q-system Following Twenty Years of Application in NMT Support Selection. Geomechanik. 1994 - Barton, N R. Investigation, Design and Support of Major Road Tunnels in Jointed Rock using NTM Principles. Proceedings of the IX Australian Tunnelling Conference, Sydney, Australia, August 1996 - Bieniawski Z T. Engineering Rock Mass Classifications. Wiley Interscience. New York. 1989 - Bischoff J A and J D Smart. A Method of Computing a Rock Reinforcement System Which is Structurally Equivalent to an Internal Support System. 16th Symposium on Rock Mechanics. Pub. Design Methods in Rock Mechanics. 1977 - Grimstad E and N J Barton. Updating the Qsystem for NTM. Int’l Symposium on Sprayed Concrete. Fagernes, Norway. 1993 - Henderson, A D and R J Cuttler. Northside Storage Tunnel Project. Tenth Australian Tunnelling Conference, Melbourne, Australia, March 1999 - Parker C, B Whiteley and B Gee, Innovative Investigation Techniques used for Geotechnical Modelling of the Northside Storage Tunnel Project, Sydney. Tenth Australian Tunnelling Conference, Melbourne, Australia, March 1999