The Second Avenue Subway (SAS) project, the first major subway extension in over 60 years in New York City, is being planned for the east side of Manhattan Island. The entire project will run in tunnel from the north of Manhattan at 125th street, along the eastside of central park to the south tip of the island at Hanover Square Station, the Financial District. Currently one subway tunnel, servicing three major subway lines the “Lexington Line” runs north-south on the east side of central park. This is now at full capacity and any disruption causes major delays to the entire subway network, due to the lack of a redundant system along this stretch (figure 1).
The planned SAS project consists of around 13.7km of bored tunnels, sixteen subway stations (6 mined stations caverns with spans of up to 26m, 10 cut and cover stations with depth of up to 16m), 6 cross over caverns, and 8 shafts for ventilation and emergency egress. The capital construction cost is some US$16bn (2007 prices). To make a project like this feasible and to acquire the funds and labour, the project must be broken up into phases with each designed to deliver maximum rider-ship for the investment and to suit the construction market. For the SAS project, the phases in Table 1 have been developed based on passenger estimates, construction packages sizes and importance to develop a redundant system along the east side of Manhattan.
Project setting
The geological setting of New York City poses many challenges to subsurface projects because of the undulating rock profile, the presence of shears and faults and deep-seated alteration and deterioration of the rock. The subsurface investigation for the SAS was a comprehensive phased program including conventional drilling and extensive borehole, surface and cross-hole geophysics. The investigation included the collection of existing information such as old maps from colonial to present for Manhattan, construction records, geomorphology, geology, land-use and more than 600 historic borings. The field investigation included more than 350 borings to sample soil and rock, laboratory and in situ testing and mapping exposed rock in existing tunnels and excavations.
The rocks in the project area are of the Precambrian Manhattan Schist Formation, which are essentially strong and abrasive crystalline quartz, mica and feldspar with abundant garnet, amphibole rich zones and pervasive alteration. Intrusions of varying size occur within the schist typically along and occasionally across the foliation. The rock mass is characterised by three principal joint sets with sub-sets and the dominant joint set is parallel to the foliation in the schist. Faults and shears occur throughout the project area with major fault zones in the vicinity of 64th to 69th Street, 86th Street and 92nd Street. The area is a region of high tectonic stresses and local measurements for the SAS project and other projects indicate a stress ratio of between 4 and 10. Groundwater generally follows the interface between soil and rock or stands approximately 5m below ground surface. The rock mass permeability is generally very low with locally high permeability associated with fracture zones, faulting and alteration. Glaciation has caused erosion followed by deposition of poorly sorted granular and cohesive soils and the most recent deposits include organic silts and made ground.
The intact rock has demonstrated consistent properties from the laboratory testing. At an early stage in the investigation it was decided that the key to understanding the engineering behaviour of the ground was the structural geology (table 2).
Because schistose rocks are extremely difficult to correlate between boreholes, an acoustic televiewer was used. This was a significant success in defining the location and orientation of faults and shear zones in the rock mass and it was the basis for establishing discrete geological domains for the purposes of tunnel support and contract provisions. The ATV images were correlated with the core samples to provide engineering properties of the structures, orientation and true thickness. This was an important part of the design and the risk mitigation strategy.
Deep excavations in rock are common in Manhattan and groundwater is not a significant consideration. However, there is a risk of third party impacts such as mobilisation of groundwater plumes due to sustained groundwater flows for the large tunnels and caverns. Therefore a program of constant head injection tests were carried out to identify the transmissive features and assign them to a structural group so that specific measures could be developed to mitigate the risk from inflows. The prevalence of tall buildings and proximate transportation and utility tunnels is a testament to the good quality of the Manhattan rocks. However, because of the faults and shear zones the ground has been broken up into geological domains – those controlled by the structural features and those that are not. This is the basis for the modelling, analysis, classification and design and ultimately the construction contracts.
Cut and cover vs shallow caverns
The choice of construction method in the past was dependent almost purely on cost. Generally cut and cover won. In addition the techniques and ability to mine were not available, but new improved methods mean that underground construction is now feasible in increasingly poor ground.
In today’s tunnelling industry other factors come into play; utilities, environmental impacts, surface disruption to traffic, dust, noise, vibration, etc.
For the SAS project the application of cut and cover stations was investigated by the project design team during the Preliminary Engineering (PE) phase. From the start the following constrains were given:
• At 96th Street Station the new subway line has to tie into existing cut and cover tunnels constructed in the 60’s, and the tunnels utilised for storage
• To the south, the project has to connect to the existing 63rd street station box, where the newly constructed line will connect to an operating existing line
• Desire for shallow stations to make swift access time for passengers
• Reduce the surface disruption to a minimum on a heavy traffic area, both for vehicles and pedestrians
Multiple schemes have been developed showing pros and cons for cut and cover vs. mined station caverns. Based on detailed studies in terms of utilities, disruption, ground conditions, construction costs it was decided that a mixture of cut and cover boxes and mined stations will be the best solution to make the project feasible, in terms of program and cost. One of the most challenging issues in a subway construction is the investigation, exploration and due diligence of existing ground conditions, infrastructure and utilities. During the PE the design team encountered major issues in terms of utility assessments, which required multiple trial trenches to expose and investigate what is actually in the ground. The New York City underground can only be described as a maze of sewer lines, power lines, water pipes, steam lines, etc. Existing or as built drawings are available for all the utilities but are almost never updated, therefore cut and cover/open cut boxes add a lot of risk to the client in terms of utilities.
Due to the strong and massive underlying rock in the areas of the 72nd and 86th Street Station, and a rapid drop of the Manhattan Schist to loose to medium dense fill overlaid by soft organics over medium dense glacial deposits, in the area of 96th Street as well as the need to connect to existing shallow cut and cover tunnels in this area, the following concept was chosen for Phase 1 (figure 2):
• 3962m of twin rock TBM tunnels and 610m of D&B, 6m diameter tunnels from 96th Station to 63rd Street Station
• 96th Street Station: 540m of cut and cover station, up to 18m wide and 15m deep, with 1m thick slurry walls, and a 1.8m thick reinforced invert slab. The structure is designed for full water pressures
• 86th Street Station: 295m long D&B cavern in Manhattan Schist with cover to span ratios (C/S) as low a 0.26. The Cross sections width is ~21m, with a cavern height between 14.7-19.1m. The cavity is designed as drained structure, with a flat invert as shown in Figure 3
• 72nd Street Station and crossover caverns: 408m long drained D&B cavern in Manhattan Schist with cover to span ratios as low a 0.22. The cross sections width is ~29.5m, with a height of 15.1m.
Having established the benefits and feasibility of cavern construction for the project, the designers were faced with challenges that included large excavation spans with relatively low rock cover and variable geotechnical conditions, all located within a complex urban environment.
Geometrical issues
In the Upper East Side of Manhattan, property acquisition is a major consideration. Only certain properties have been identified by the client as feasible to acquire to use as entrances. Many of these were designated as such during the 1960’s design work, when SAS was first on the radar. They determined the possible entrances to the stations, and therefore defined their general horizontal alignment, together with the tie in at the existing cut and cover tunnels at 96th Street station and the existing 63rd Street Station to the project’s south.
Drained vs. undrained
In order to comply with the client’s specification of completely dry tunnels, the circular TBM tunnels were designed as undrained with 360° waterproofing. Discussions evolved about whether the same 360° waterproofing should be applied in station caverns, and thus be designed for full hydrostatic pressure, or, whether stations should be designed with a drained invert and loaded by partial hydrostatic pressures. The proposed undrained design of station caverns with spans of up to 32m would require arched concrete inverts, resulting in a larger volume of excavation and consequently higher construction costs. A drained design, on the other hand, may lead to increased operation and maintenance costs, including pumping and groundwater treatment, and the associated risks of potential contaminant transport and handling of unexpected groundwater inflow. A detailed study was performed to evaluate the pro and cons of drained vs undrained. This concluded that:
• The best way to provide the client with a dry station was to manage groundwater, rather than trying to block it completely
• The groundwater inflow anticipated would be manageable and it would be economic to treat and pump groundwater and maintain its management systems
• There is a long history of drained tunnels and deep drained inverts in New York, giving the approach precedent
• The groundwater flow was unlikely to introduce contamination, however space would be provided for in the design to include a comprehensive treatment works in the station, the actual treatment installed would be based on the quality of groundwater encountered during construction. This would be adjusted over the working life of the cavern to respond to changes in groundwater quality
A drained solution was therefore adopted.
Design concept for the caverns
Design for a shallow cavern with wide spans, as proposed for the SAS station caverns (Cover to Span Ration C/S 0.22-0.25), is an engineering challenge that hasn’t been performed often around the world. Therefore no design procedures, and limited guidance, have been published to date. To overcome this challenge a detailed methodology, based on empirical, analytical and numerical methods, has been developed to guide designers through the design process.
An Initial Ground Support Requirement (IGSR) chart was developed by the SAS design team to select support for the tunnels and caverns and to ensure that a consistent design approach is applied for all the mined structures along the project alignment. With these IGSR charts the tunnel engineer, as well as the cost estimating and quantity surveyors and the contractor on site, can get an initial understanding of the type of support that will be applied to stabilise the cavern excavation. The functions of the design chart are as follows:
• Establish that the cavern is constructible
• Stipulate minimum support requirement to potential contractors and to assist them in bid preparation
• Define the design ranges of five support types by cavern design rules that are selected to estimate ranges of similar ground behaviour
• Estimate quantities of materials associated with the five support types
• Provide guidance for excavation sequence design
• Determine design rock loads for final lining design
For the SAS five ground support classes were developed to cope with the anticipated ranges of ground conditions and boundary conditions in the vicinity of a cross section (overburden, tunnel intersections, building loading, etc.). The ground behaviour rules, which recognised impacts of both the rock quality and the excavation geometry, arranged the five support types in a hierarchy from bolt-reinforced rock to structural shotcrete arch in recognition of a boundary between “good” and “bad” ground behaviours. The ISGR chart is based on the following input parameters (Lagger et.al. 2007); Norwegian Geotechnical Institute’s Rock Tunnelling Quality Index – Q; Geomechanics Classification, Rock Mass Rating – RMR; Rock Cover to Span Ratio C/S; Rock Mass Class after Cording; Rock Load; Peak Friction Angle; Proximity to Adjacent Structures.
To update the developed ground support from PE (IGSR) to Final Engineering (FE), a sophisticated, well structured design approach has been put forward as routine. It provides the tunnel and cavern engineer from the initial ground support, developed based on empirical models, through 2D and 3D numerical calculation, structural geology (kinematics), to beam spring models, to the final initial ground support.
The design guideline is based on eight steps to ensure a robust cavern design solution.
1: Employ the IGSR chart to create design slices along the cavern
The first step in the design procedure is to divide the cavern into design slices that serve as a basis for modelling the cavern based upon the cavern design rules as presented in the IGSR chart: The cavern will be divided into zones of similar rock support classes based upon the cavern design rules. Every design slice should include the geological information of a minimum of two boreholes. Each cavern penetration (entrances, auxiliary space, etc.) is a separate design slice, including 3m on either side of the penetration as shown in Figure 4.
2: Evaluate the intact rock and rock mass, rock joint and soil properties per design slice
For each design slice the available in-situ data for the intact rock, rock mass, rock joint and soil properties are interpreted and best estimate and lower bound values are given. For the intact rock mass the Hoek-Brown Strength Criterion Edition 2002 is used to determine the appropriate strength properties for the intact rock mass. Using the Rocklab 2005 Software package, by Rocscience, the input parameters c’ rock and ø’ rock, as well as the rockmass modulus Em rock, are calculated from lab data for each design slice. For the rock joint model, a Mohr-Coulomb joint model is adopted, using the existing shear box data where available to calibrate the joint behaviour. Where geometrically possible, the correct geometry of the joint is modelled, e.g. instead of a plane linear joint the waviness of the joint may be incorporated.
3: Evaluate structural properties of materials
The initial ground support elements for the cavern consists of resin grouted, tensioned rock bolts, Swellex rock bolts, and un-reinforced and fibre reinforced shotcrete. Lattice girders are not included in the lining’s structural capacity calculations. Shotcrete arches reinforced with steel sets, were used and are incorporated with increased stiffness properties. For rock, an elasto-plastic material model of the Mohr-Coulomb type is assumed, where the soil is modelled as a stiff elastic media above the rock.
4: Development of geometry, boundary conditions and adjacent structure loads
The mesh size for UDEC analysis is set to 12*D (D=Cavern width) width and the lowest point of the excavation is 2*D away from the boundaries of the mesh to avoid boundary effects and to achieve reasonable computation times. Other boundary conditions include the following; building loads are applied on top of the soil according to the estimated loads from a building survey; horizontal stress ratio K0 is applied according to the Geotechnical Interpretive Report; no water pressures are applied because the caverns are drained.
5: Develop a 2D UDEC numerical model
The Universal Distinct Element Code (UDEC) 2D is used to evaluate the global stability of the main cavern sections due to the excavation, i.e. to evaluate the response of the discontinuous media (such as a jointed rock mass) subjected to static loading. For each different IGSR class, different excavation sequences have been developed during FE. For Type I to IV, either side drift excavations or centre cuts can be chosen by the contractor. For Type V ground (adverse ground conditions) only side drifts will be allowed. Using the UDEC code these excavation sequences are modelled to ensure the stability of the rock mass and adjacent structures and to limit displacements to the specified levels.
6: Block calculation via UNWEDGE
One of the major concerns in rock tunnels and caverns is the kinematic failure of discrete blocks in the excavation area, i.e. rock blocks become loose and fall out of the rock matrix, destabilising the rock mass. Therefore the local stability of the rock mass between the rock support necessary for the global stability is checked using UNWEDGE. Based on engineering geology input data (dip and dip direction) and excavation direction, the maximum geometrically formable blocks between rockbolts are calculated. The rock wedges are scaled in terms of excavation area, no larger than the bolt spacing of the surrounding rock support. Using gravity loading and hydrostatic water pressure, the factor of safety (FSmin=1.5) is then calculated for the local stability of the rock blocks between the rock bolts.
7: Shotcrete stress evaluation
After the global and local stability of the rock mass is achieved, the internal forces of the shotcrete shell are checked using a beam spring model.
8: Evaluate the stability of cavern penetrations via a 3DEC – 3D model
During PE the global stability of the cavern section was checked using empirical models and plain strain discontinuum 2D UDEC models. The global stability in areas where the cavern cross sections were penetrated by entrances, ventilation structures, and auxiliary buildings have been calculated using a reduced Q-value. Barton 1974 and 2003, recommend that due to changes in the in-situ stress state (transfer of membrane stresses) around a tunnel junction, the calculated Q-values should be divided by a factor of 3 to account for this stress transfer. Using this method, initial ground support in the junction and penetration areas have been designed using Q-values only during PE to evaluate an indicative initial ground support for cost estimate reasons.
During FE a more sophisticated analysis was chosen to determine a more accurate ground support system in the areas of the penetration. The main cavern cross section is designed using the seven previous steps with a 2D UDEC model. This model incorporates various excavation sequences, overburden and building loads, etc. to evaluate the global stability of the main cavern section.
For the areas of penetrations and junctions a three dimensional numerical model 3-DEC has been applied to evaluate the effect of the stress change in the initial ground support of the main cavern station and to determine a detailed updated ground support in the penetration areas.
The 3-DEC calculation is based on the same discontiuum algorithm as the 2D-UDEC code, but in all three dimensions. It is possible with computer technology today to model the entire station in three dimensional space, including detailed modelling of joint systems, geometries and excavation sequences, and ground support.
Conclusions
The approach to cavern design for the large shallow caverns of the SAS has developed during the design phase into a robust procedure consisting of the best-available design tools. This procedure has been employed to verify the designed ground support system, incorporating cost savings opportunities into the IGSR chart where possible. Numerical modelling via 3-DEC has been added during FE to verify ground support systems around penetrations. In this manner, the high risk arising from unknown conditions that could have severe impacts have been met with a comprehensive analysis.
Several advantages are offered by this approach. The design can be implemented step by step throughout the design stages of the project. Designed support can be modified and support quantities easily recalculated throughout the project. Multiple approaches are employed to cover the bases where geology may differ from what is anticipated. Finally, it provides for some consistency with and between large-scale projects.
Typical New York utility situation Project map with phases Construction methods along Phase 1 Fig 2 Typical cross section cavern stations, minimum rock cover Fig 3 Initial ground requirements, assessed for each ‘design slice’ Fig 4
