ABSTRACT

The City Rail Link project connects the existing Britomart rail terminus in Auckland’s CBD with the Mt Eden Station on the North Auckland Line. The c.3km-long route traverses a wide range of geology and ground conditions. Most of the tunnelling will be within Waitemata Group rocks (East Coast Bays Formation, ECBF) that underlie much of the wider Auckland region, while shallower structures at the north and south of the project will also encounter younger sediments (Tauranga Group); volcanic materials (Auckland Volcanic Field, AVF); made ground (including reclamation). Desk studies and early site investigations established a broad model of ground and groundwater conditions, while later investigations focused on specific design requirements, material characterisation and/or ground risks (and opportunities). Targets included a 10m-thick mega-bed (in ECBF) at Karangahape Rd Station; anomalous ECBF weathering profiles; complex AVF materials and geometry at the south end of the project; a pre-Holocene paleo-channel under Britomart; post-volcanic spring/pond deposits in the valley between two volcanoes in the south; and undocumented structures in the built environment. Various works packages were procured using a variety of contract styles. However, procurement documentation has generally included a geotechnical baseline report (GBR), supported by ground data in geotechnical data reports. An observational approach was specified for parts of the works, particularly where opportunities for ground investigation during the design phase were limited. Innovation during procurement includes the use of 3D ground models in two of the contracts. A static 3D model was included (for information) for the Mt Eden enabling works area, to illustrate the complex relationships between proposed structures and varied ground units. In a world-first, for Seequent’s Leapfrog/Central the main works tender included a cloud-based 3D geology model and geotechnical database for each tenderer, readily allowing development of their own geology model and facilitating alignment on geotechnical risk during the Alliance GBR process.

Introduction

The City Rail Link (CRL) project comprises twin 3.5 km-long rail tunnels linking the CBD’s Britomart Station with Mt Eden Station on the North Auckland Line, with two new stations along the route (Aotea and Karangahape). The project was started by Auckland Transport, and is now managed by City Rail Link Ltd., owned jointly by Auckland Council and central government. The project commenced in 2010 and is presently (May 2020) under construction via six contracts (for enabling works, early works and main works).

The project lies mainly in the weak rocks of the Waitemata Group (East Coast Bays Formation: a turbidite sequence of alternating sandstones and siltstones). In places the alignment cuts into Pleistocene alluvial deposits, and through complex deposits (basalt lava, scoria and ash) of the three volcanic centres that lie adjacent to the route. Holocene marine sediments and significant areas of reclamation are present at the Britomart end of the alignment.

This paper outlines the site investigations, that ran alongside the development of the design (Concept, Reference, Detailed), and informed the Resource Consenting process. The ground model evolved with the design, and the phased approach to investigations allowed areas of ground uncertainty or ground hazard to be identified and investigated in a prioritised way. We also describe the development of our use of 3D geology models for procurement. Geology modelling was introduced initially as an internal process to help designers understand the locally complex ground conditions, but as software advances allowed increasingly sophisticated documentation of both the models and the supporting geotechnical database the models took an increasingly prominent role in procurement.

Ground risk assessment and documentation have followed a traditional model with desk studies, multiple phases of ground investigations, with associated factual reporting; interpretation and reporting (Geotechnical Interpretative Report, Geotechnical Engineering Report); leading to a series of Geotechnical Baseline Reports and supporting Data Reports for procurement. Hydrogeology (Pattle, Delamore Partners) and contaminated land investigations (Golders Associates) were designed and executed together but reported separately (but with shared interpretative sections for the Hydrogeological Assessment Reports). Resource consenting was supported by a range of assessments of environmental effects (for groundwater, settlement, vibration, contamination, and other topics).

Site INVESTIGATIONS

The ground investigation database was established with extensive research of desk study sources ultimately finding more than 1300 historic holes in the project corridor. A series of five project-specific investigation phases were carried out between 2010 and 2018:

• Preliminary (Stage 1, 2010) investigations (by AECOM, Parsons Brinkerhoff and Beca, APB&B) focussed on the Newton area (originally intended to be a deep station) and environmental sampling;
• The main investigation (Stage 2, 2012-2013) was extensive, establishing the main elements of the ground model and installed the bulk of the ground water monitoring;
• Supplementary investigations were carried out in 2015 (Stage 3) and 2016 (Stage 4), investigating a series of potential ground risks and/or design changes including a pumping test at the ECBF ‘mega-bed’ at Karangahape Station, the depth of foundations of a existing retaining wall at the Central Motorway Junction, investigations of depth of basalt infilling paleo-valleys and detailed environmental investigations of specific sites;
• Pre-construction investigations were finalized with tender-stage investigations in 2017-18 (Stage 5), requested by the parties expressing interest in bidding for the main works (construction contract C3).

The site investigation works included drilling, CPT probes, trial pits, downhole geophysics and geophysical probing, a wide range of in situ and laboratory testing and groundwater monitoring.

Figure 1: City Rail Link Alignment with 5 stages of ground investigations (Stage 1 yellow; Stage 2 blue; Stage 3 red; Stage 4 green, Stage 5 purple), with historic investigations. Letters show locations discussed in text.

Figure 2: City Rail Link geological sections. Top (green) section on main tunnel alignment (Britomart to Mt Eden). Blue section along North Auckland Line (see also Figure 1). Supporting drillholes not shown on sections (but locations shown on

Figure 1 shows the project area with five stages of ground investigations, together with some of the locations discussed in the next section.

GROUND CONDITIONS AND RISKS

We do not attempt to fully describe the ground conditions of the alignment here, rather present examples of the ground conditions, spanning both the stratigraphic column and the length of the alignment, that illustrate the ground conditions, our investigations and potential risks (and opportunities). The locations of the sites specifically mentioned are labelled on Figure 1 (plan) and Figure 2 (long sections) and are described below in their stratigraphic context.

Waitemata Group (East Coast Bays Formation)

The main engineering geology units for the ECBF are defined by weathering state. We log weathering grade according to the standard six-fold classification, but we layer code (and assign engineering geology unit) on three pairs of grades: ER representing residual and completely weathered material; EW for highly and moderately weathered; and EU for slightly and un-weathered rock. We have developed logging guidelines for each grade and find the resulting units returning greatly reduced spreads of test results and notice a lack of correlation when using externally sourced logs. Weathering profiles are varied, mainly depending on the site’s age and preservation, but we most commonly see relatively sharp weathering fronts (where ER is thicker than EW) spanning 5-10m in total. The deepest weathering profile (over 20m) encountered in the CRL alignment is at location “F”, where both ER and EW are thick and likely due to the age and preservation of this hilltop site.

Relatively unweathered ECBF (EU) is predominantly a very weak (UCS < 5MPa), sedimentary rock, where the main variance in geotechnical properties derives from the relative proportions of the alternating sandy siltstone and silty fine sandstone layers (turbidite). We assign these normally cemented strata to the EG Unit “EU2” and have much less common end members “uncemented sandstone” (EUs1, as this applies exclusively to sandstone) and well-cemented sandstone (or conglomeratic sandstone) EUs3 (and EUg3) respectively.

We apply “well-cemented” where EU exhibits or could support open joints (as secondary permeability is a potential concern while tunnelling in ECBF), and this applies to strata with UCS strength of 8-10 MPa or more (our “weak+”). We prefer to avoid the widely-used term “Parnell Grit” (except in an informal way) as any stratigraphic merit of this ECBF Member has long since been disproved, and the term is associated with simplistic sedimentogical models.

The CRL alignment is unusual in that it includes a significant body of well-cemented ECBF, interpreted as a single “mega-bed” (10-12m thick, “D” on the long section) and corelated for about 800m along the alignment (below Karangahape Station). The feature is well documented, as we carried out a pumping test at the unit and installed numerous piezometers for the purpose. The feature is so distinctive that it is readily mapped and fault offsets of 5-10m were identified. The unit is stronger than normal (UCS 15-25 MPa) and though it does contain open joints, the (low-volume) pumping test demonstrated that the storativity is low (joints are tight), but transmissivity is high (and anisotropic, with drawdown axis matching interpreted faulting). With the groundwater considered relatively benign, the unit offers construction opportunities rather than risks. The geometry of the mega-bed suggests that it is the source of natural springs that used to flow into the upper part of Myers Park, and therefore likely the source of the Waihorotiu Stream once a feature of the Queen Street valley.

We use the term “uncemented sand” (EUs1) to describe drillers’ “running sands” that are difficult to recover during drilling of boreholes, though routinely test as very dense (SPT N=50+). Because of recovery problems the unit is rarely sampled, and our testing approach has focussed on samples that are likely at the upper end of the strength range. Petrographic studies on previous projects (P. Black, unpublished report) suggested these samples had never been cemented and largely lacked a clayey matrix (typical of EU2, for example). The units rarely exceed 1m in thickness and are usually randomly distributed within an alignment. However, in CRL, there is a cluster of records of the uncemented sand close to tunnel levels at the mined cavern section of the Y-junctions (where the main alignment splits to join the existing NAL). This is a potential ground risk in this method of tunnelling, and therefore needs to be mitigated in the design of supports.

Tauranga Group and Auckland Volcanic Field

Auckland’s Tauranga Group covers a range of diverse sedimentary deposits that spans most of the Quaternary. Much of it is inferred to be alluvial (erosion products of the ECBF). Organic layers and peat deposits are present locally, and the unit incorporates Taupo Volcanic Zone-derived rhyolitic ashes (primary and redeposited), and commonly grades into basaltic volcaniclastic materials in the proximity to Auckland’s many volcanoes. Differentiation between ECBF residual soils and ECBF-derived fine alluvium can be genuinely difficult especially in borehole core. Some practitioners still differentiate simply by consistency, but this has proved to be an unreliable approach and has confused the record of how the different materials behave – for example during dewatering (potential settlement is a common concern in resource consent applications).

In the CRL alignment, we identify at least four Tauranga Group sediment types: two at Britomart (A); and at Albert Street Gullied Terrain (B) and post-AVF swamp deposits (J, on the NAL section).

At Britomart (A), twin paleochannels cut into the ECBF substrate, below the reclaimed land of the CBD. The easternmost channel (documented from older boreholes investigating the original Britomart site) represents alluvial deposition during a mid-late Pleistocene glacial period and comprises stiff over-consolidated fine soils infilling a proto-Waihorotiu Stream. Cutting these deposits, and into the underlying ECBF weathering profile, is a younger channel (our pre-Holocene channel) inferred to have formed after the lava flows of the (c.100ka) Albert Park volcano infilled the stream valley upstream from the CRL alignment. This younger channel was later infilled as sea levels rose in the Holocene, with marine sand deposits near the original shoreline, and increasingly fine sediment further into the bay. The pre-Holocene channel with its infilling of soft young marine mud has been a source of geotechnical challenges for the design team (Michael Chung pers. comm.). The Figure 2 interpretation was based on a few historic holes, and construction observation, resulting in a revised interpretation (and re-design) as discussed again in section 5.1, below.

At Location “J” on the NAL long section, an entirely different Tauranga sediment is encountered. About 30,000 years ago a valley was formed where the tuff and ash from the older Domain volcanoes (to the northeast) met the basalts from Mt Eden (from the south). Springs formed at this point, likely fed by artesian groundwater that collected in the coarse basal ash (locally tuff) that mantles the slopes to the north. The site was still a pond during early settlement (but associated deposits are extremely soft lacustrine silt and organic fine soils, that have caused earlier construction difficulties (see discussion of Contract 6, below). The C6 project (replacement stormwater line) also needed to tunnel under a basalt flow (about 100m north of location “K”). Historic drilling and some project holes had identified the broad paleo-topography, and the ReMi passive seismic technique was employed (over a network of profiles to further define the geometry of the basal contact of the flow, and to locate the deepest part). This area was then drilled to find the actual level of this area. This geophysical investigation is described further in Pancha et al, 2019.

Geomorphology – Built environment

Although it is well understood that the modern urban landscape has been modified over time, it has sometimes been surprising to discover the extent of these historic works. Major infrastructure developments may have historic plans and designs, such as the detailed plans and designs for the successive reclamations of Commercial Bay (now Auckland’s CBD). Histories of other areas can sometimes be pieced together from historical photographs or even paintings, for example Point Stanley (location B, south side of the Customs/Albert St intersection) was once a prominent headland standing 12 – 15m above the wave cut platform below. Initial reclamation in the 1850s and 1860s left Customs and Albert streets vertically separated, but they were later (early 1900s) joined by cutting back the headland and forming an apron of fill with the debris.

Pre-European ECBF ridges and side-slopes were also more prominent and incised. But as tracks, became paths, then roads, which were later lined with buildings there was a progressive series of cuts and fills to establish building sites. These were typically formed by cutting into weathered ECBF slopes and infilling the gullies. More than 100 years later, the resulting profiles (ECBF fill over in situ ECBF), can be difficult to distinguish from a natural profile. Examples are the Karangahape Road ridge (location D) and the side slopes leading from Albert St. to Queen St.

Another recurring theme for the project, is that while we can often find ground investigation data for a construction project, usually with accompanying design drawings, it is much less common to find as-built documentation, confirming that construction proceeded as designed, or if modifications were made during the construction process (perhaps accommodating unexpected ground conditions).

A case study for this is the existing piled retaining wall for the Upper Queen Street overbridge at the Central Motorway Junction (location E) where the western-most three piles are located above one of the tunnels. Without any as-built records it was not possible to confirm that the piles did not clash with the proposed tunnel. The investigation into the actual depth of the piles was extensive – including engineering analysis, geodesy and surveying, liaison with the Auckland Motorway Alliance, and ultimately very precise drilling (from a platform perched over SH1) of three boreholes as close as permitted to the three piles, enabling a range of downhole geophysical methods to be deployed. Ultimately, measurements of magnetic field and susceptibility (from the rebar cages used in the pile construction) together with off-site simulations allowed analysis of pile termination depth (Matt Watson, Scantec). Consistency of the three test results, together with the ground conditions (matching stated pile termination criteria) provided a high degree of confidence that the piles terminated above the level of the proposed tunnel. .

3D GROUND MODELLING

From the start the CRL project has been one in which digital innovation was encouraged. In this context the geotechnical team was involved in early work to build a comprehensive 3D model of major aspects of the project, including the adoption and integration of BIM, and 3D GIS approaches (similar to approaches later formalised in ISO19650). 3D ground models were initially used on an internal basis during design because they had become our preferred method of handling (including Quality Assurance) the large datasets available for the project, and particularly the use of historic data, much of which required re-interpretation to achieve consistency with our project-specific ground investigation data. The development of 3D geology also helped collaboration with our colleagues at PDP, who were developing 3D groundwater models for parts of the project area. The ability to export a geology model directly to 3D groundwater modelling software saves time and ensures consistency between the two parts of the team (A. Perwick PDP, pers comm). However, above all technical advantages, the use of 3D ground models improves the communication of geological interpretation with designers, clients and other stakeholders. Our preferences are to build 3D ground models where the geology is complex, or the ground risk is significant, and where supporting data is already digital, or can be converted into a digital database at reasonable cost.

Ground Risk in Procurement

During the CRL design and consenting, the software for developing and managing 3D ground models became increasingly sophisticated. As it turned out, the timing of procurement of the various works packages charts a progression of our use of 3D models (see Table 1, below), as we progressively introduced the technology into procurement.

Aspects of selected projects are further discussed below, illustrating the ways in which ground risk is addressed conventionally, as well as providing background to the introduction of ground models for this purpose.

Table 1 documents the chronological development of the use of 3D models on the CRL project, commencing with internal use by the designers in 2014 on Contracts 1 and 2. For contract 5 and 6, the 3D model was used to develop the 2D long sections and was provided to tenderers within a non-editable viewer. Feedback at that time from tenderers was that the viewer was very useful on communicating the geological conditions and risks. For the main tunnels and stations of Contract 3, Leapfrog Central was used to host the client tender model as well as the models that were developed and enhanced by each of the tenderers. One of the main advantages of this approach was that tenderers did not need to expend time and effort on processing geotechnical data but could instead focus their efforts on revised interpretation and development of their risk mitigation approaches. The software (Seequent Central), through it’s model comparison tools, also facilitated a transparent evaluation of the geotechnical risk profile inherent in interpretations developed by each tenderer (as the client requested full access to the tenderers developed 3D models).

Table 1: Approach to ground risk in CRL procurement.

*After initial contract award, Contract 5 was discontinued and re-assigned to successful Contract 3 tenderer.

Contract 1: Observational Approach at Britomart

At Britomart Station, where part of the historic CPO building including its façade were underpinned as part of tunnel constructions works, geological interpretation was challenging due to limited opportunity to carry out investigations prior to construction and therefore the interpretation was based on some historic GI and limited project-specific GI (summarised in section 3.2). The interpretation shows the site is underlain by a paleo-channel and later infilled by soft Holocene marine mud. Although the presence of the channel was known, there were uncertainties about its extents: depth, width, and axial position. For this reason, construction stages relied on an observational approach (e.g. pre-drills for diaphragm wall panels) to define actual site conditions, and final depths for panels and associated piles. Construction phase drilling proved that the channel was slightly east of the design-stage interpretation, and the panel and pile depths therefore required revision.

This work was carried out conventionally (i.e. not applying 3D modelling). The model illustrated in Figure 3, was developed later, as part of a lessons learnt exercise to show what would have been possible with 3D. The erosional extent of a paleochannel can be challenging to represent in a limited number of 2D sections and much better communication would likely have been achieved using the 3D model. Also, our present software has limited capabilities of illustrating the reliability or uncertainty of 3D geological features, and we documented uncertainty (and allocated risk) as baselines in the Geotechnical Baseline Report (both in terms of geometry and geotechnical properties).

Figure 3: Pre-Holocene Channel from below, represented by thicker marine deposits (blue), with channel lag deposits in purple. Red structures are prosed founding structures from CPO building and Queen Street. Yellow and orange sticks are the representations of boreholes intersecting the underlying strata (Tauranga Group and ECBF).

Contract 6: Static 3D ground model (for information only)

The geology of the southern portal area of the CRL (where the main alignment at G, splits and joins the NAL at J and H, figure 1) is a complex interaction between the deposits of two volcanoes, and the sediments that deposited between them (Section 3.2). The area has a history of adverse ground conditions and associated engineering difficulties. When the existing stormwater was being constructed the first tunnelling attempt unexpectedly encountered Mt Eden basalt (with a TBM not designed for hard rock), and the second attempt (on a modified alignment) encountered unanticipated very soft ground (but managed to negotiate it, location “I”).

The area is complex both geologically and in terms of engineering, and the large suite of geological long and cross sections we had developed was frankly quite bewildering. Our designers used a combination of both our simplified 3D model and the detailed sections to get to grips with the ground conditions.

For procurement in this area we wanted to emphasise that we had developed a robust ground model and understood the limitations of earlier models, and for this reason we issued the 3D ground model (showing all the supporting GI data) for information only, in conjunction with the GBR sections. The model was available in a free 3D viewing format, and included the Revit model of the reference design, allowing ready understanding of how the proposed design related to the ground conditions.

The construction of the C6 tunnel has been successfully completed. Tunnelling proceeded (with a machine capable of handling basalt) under the basalt. Shafts were designed to handle very soft ground and basalt. The soft ground shaft was completed successfully, while the basalt shaft encountered abrupt transitions in lithologies that had not been anticipated in the model, though variation was largely within that anticipated in the GBR. Data from the (contractor’s) shaft excavations and tunnels has been incorporated into the (client’s) 3D geology model to inform contractual communication relating to differing ground conditions. The ability to show construction stage ground observations in a view with tender stage (3D) ground interpretation is a powerful means of communication when addressing unexpected ground conditions.

Contract 3 Main Works: shared database and modelling environment

The final main procurement package comprises the main works for the project, including station construction and tunnelling. The geology is not as complex as at Britomart or the south portal, but the scope of works is considerably larger.

For this procurement we adopted a more ambitious 3D ground model environment comprising a cloud-based 3D geology model and geotechnical database (internal to Leapfrog) for each tenderer. This environment (Leapfrog Central) allowed each tenderer access to the database of GI (with our interpretation) and a partial lineage of model development. Moreover, the environment allowed download of data and model to their own software for further analysis and model development and we required upload of their developed models as part of tender assessment. This process contrasts markedly from common approaches to ground risk, where factual data is lodged in a data room, commonly in pdf format, and tenderers teams must wrangle large volumes of data before they can start providing geotechnical advice.

During the procurement phase, CRLL placed high value on design and construction mitigation measures proposed to address ground risks, as is appropriate for a major underground project. To achieve alignment on the level of geotechnical risks, an Alliance Geotechnical Baseline Report (AGBR) process was developed, which involved multiple alignment workshops and submissions that were all facilitated through the shared development of the cloud-based 3D model. This interactive 3-D model facilitated the required interaction with tenderers on ground risk.

The AGBR process and 3D model development process contributed to a successful tender process and resulted in both tenderers having a similar ground risk profile. This meant the tender process was not skewed by one tenderer having a significantly higher project risk profile. Neither tenderer included unmanageable ground risks.

6. Conclusion

Auckland’s varied geology always poses problems for linear infrastructure projects, but particularly for tunnels. This paper gives examples of a range of geological processes, active across Auckland’s geological history – and extending to the as-built environment, that were investigated for the CRL project, using a traditional approach of phased site investigations, integrated with developing design. We also outline how we developed a 3d geological modelling capability for the project, to facilitate interpretation and improve communication of that interpretation (and its limitations). Geology models and the associated database of site investigation data were used both to support (for information only) a traditional set of ground risk documentation; and as an integral part of an innovative procurement environment (supported by geological model management software).

Many projects in New Zealand are on accelerated programmes, which places demands on both client-side delivery of ground information and tenderers ability to process the information to make sense of ground risk. We support the use of digital data transfer standards (eg AGS) or other formats to transfer the raw data, but consider digital data supported by 3D geological models to be the optimum method for rapid assessment (and alignment) of ground risk, pricing of mitigation, and setting reasonable baselines for risk allocation.

REFERENCES

Ireland, T., Newns, B, Chau, S.F. Hawkins, S. 2017. Tunneling Challenges on the Auckland City Rail Link, New Zealand.

Kirk P A & Ireland T.J. 2019, ‘A whole-of-life’ approach to project 3D ground-models: City Rail Link, New Zealand. Hong Kong Geotechnical Division Annual Conference, 2019.

Pancha A., Kirk P.A. Sabiston R.A., Apperley R.A. 2019. ‘Listening to the Earth’: An unconventional approach to mapping basalt flows. Australian Geomechanics Society, Victoria Symposium 2019

Rose G, Kirk P.A., Gibbons, C., Lander A. 2018. Three dimensional geological models in ground engineering: when to use, how to build and review, Benefits and potential pitfalls. Australian Geomechanics Journal Volume 53: No. 3.