JFK International Airport

The Port Authority of New York and New Jersey has planned an automated light rail transit (LRT) service to link John F. Kennedy (JFK) International and La Guardia Airports with each other and the New York City Transit subway to midtown Manhattan. At JFK, the project would include a cut-and-cover tunnel and a pump station. Construction excavation would be as much as 10 m (35 feet) below the groundwater table.

Tunnel Construction

The cut-and-cover tunnel and associated retained earth sections, totaling about 610 m (2,000 feet), would include both a twin LRT alignment and a barrier-separated, restricted service road that breaks away from the LRT at approximately the tunnel’s midpoint. The tunnel would pass underneath two taxiways at a maximum depth of about 10 m (35 feet). The retained earth sections would connect to at-grade sections running to the Central Terminal Area and along the Van Wyck Expressway (VWE).

Tunnel construction can proceed by either the bottom-up or top-down method. Both types have advantages and disadvantages, but a critical restriction on both for this project is construction duration. Only one taxiway can be taken out of service at a time, and only in a narrow time slot at a given time of the off-season travel period.

• Bottom up. Either a permanent slurry wall or a temporary soil-cement mix wall with a separate reinforced concrete box is permitted. The latter has scheduling advantages over the former.
• Top down: Construction would proceed with a slurry wall as both temporary and permanent support walls.

This article focuses on the geotechnical design considerations of the project, if it were to be built.

Geotechnical and Site Considerations

The relevant geotechnical conditions in sequence below the ground surface are as follows:

• Loose to medium dense hydraulic fill with thickness varying between 3 m to 6 m (10 feet to 20 feet)
• An extensive but not continuous layer of soft to firm, slightly over consolidated, clay/peat deposit from 0.6 m to 3 m (2 feet to 10 feet) thick
• A medium-dense to dense glacial sand layer with thickness varying between 21 m to 30 m (70 feet to 100 feet)
• An impervious silty clay stratum at about a 38-m (125-foot) depth.

Where present, the low-permeability clay/peat layer creates a perched water table in the fill. The clay/peat layer is absent in scattered areas inside and outside the tunnel perimeter, thus permitting hydraulic connection between the perched water table and the glacial aquifer. The perched water table in the fill is generally 1.5 m to 2.4 m (5 feet to 8 feet) below the existing grade. The potentiometric head in the glacial aquifer is about 0.3 m to 1.2 m (1 foot to 4 feet) lower.

The permeabilities of the fill, clay/peat and glacial sand were estimated to be about 1x10-2, 1x10-5, and 5x10-2 cm/sec, respectively. The fill and glacial sand are relatively homogeneous and isotropic, while the clay/peat vertical permeability was estimated to be about one to two orders of magnitude smaller than its horizontal permeability.

The site is constrained by many utilities and is within 30 m (100 feet) of the VWE, which passes underneath the taxiways. Three large groundwater contamination plumes known to be adjacent to the site consist mostly of hydrocarbons (including gasoline and jet fuel) and glycol.

Geotechnical Design Considerations

Regardless of whether the construction proceeds as top-down or bottom-up, construction using a conventional external dewatering system was eliminated and the need to control groundwater drawdowns to a range of a few feet was instituted. Cutoff walls, either as slurry walls or soil-cement mix walls are required. Any cutoff wall would need to extend relatively deep, in the range of 38 m to 41 m (125 feet to 135 feet), to reach an impermeable stratum and therefore be effective.

We determined that it would be more cost effective and faster to minimize the wall depth by using a horizontal
cutoff spanning between support walls. A horizontal cutoff would be located below the excavation invert and would consist of either soil-cement mixing, jet grouting or chemical grouting. Thickness of this cutoff varies from about 1.5 m to 3.7 m (5 feet to 12 feet) depending on excavation invert depth. Despite having both vertical and horizontal cutoffs, some drawdown is still expected due to a small but measurable permeability of the cutoffs.

Groundwater Flow Analyses

Groundwater analyses indicated that the most important factor for drawdown control is the permeability and thickness of the horizontal cutoff. Various thicknesses and material characteristics of the horizontal cutoffs were assessed for drawdowns by using two-dimensional groundwater flow analyses in vertical cross sections based on the finite element method. The thicknesses were later verified by three-dimensional finite difference based groundwater flow analyses (Figure 1). Case histories of similar applications were also reviewed for drawdown performance and plausibility of our approach.

Figure 1: Cross-sections through fill and peat/clay layers in N-S direction (left) and E-W direction (right). Shown also is a trace of water table in fill along model boundaries (right).

Figure 2: Contours of drawdown in feet in upper glacial aquifer in the vicinity of cut & cover tunnel.

Figure 3: Capture zone for construction period (15-month).

Both vertical and areal distributions of groundwater drawdowns are very important. We had to look at settlements in specific areas as a result of groundwater drawdowns in compressible soils. Also, the nearby contamination plumes needed to be assessed as to whether construction would influence contamination movement and dispersion. Furthermore, during construction, drawdowns would be controlled by an observation program that includes a network of multilevel piezometers.

Maximum allowable drawdowns are assigned to each piezometer of the network and enforced by the specifications. A three-dimensional groundwater model was developed for the project site to simulate general groundwater flow, to
estimate distributions of drawdowns, and to examine contaminant transport. The U.S. Geological Survey’s Groundwater Simulation Model, MODFLOW, was used to simulate groundwater flow in three dimensions in steady-state modes.

The MODFLOW model was calibrated for an average annual distribution of hydraulic head in the fill layer with a uniform areal recharge of 0.15 m (6 inches) per year. The vertical cutoff wall and the horizontal invert cutoff were both modeled as horizontal and vertical flow barriers, respectively, with specified thicknesses and permeability.
Figure 2 shows the distribution of drawdowns in the glacial aquifer with the cutoffs in-place and the excavation advanced to maximum limits both areally and vertically for the entire length of the tunnel. The drawdowns predicted by both the two-dimensional and three-dimensional analyses were within 20 percent of each other, considered good agreement.

Transport Analyses

PATH3D, a MODFLOW companion model, was used to find contaminant migration pathways and related travel times, and to evaluate the capture zones for the excavation. The travel times estimated for contaminants are generally much longer than the estimated construction period (15 months) for all contamination locations, except the contamination near Terminal 3. The perimeter of the capture zone for the construction period includes or passes near the contamination centers near Terminal 3 (Figure 3 on page 22), indicating that the contaminant could enter into the excavation during construction.

Also important is that this contamination plume would be moved offsite, making future cleanup more difficult. Hence, by utilizing our 3-dimensional groundwater model, we looked at a series of options for maintaining the plume in place. Our model indicated that a low volume ground-water recharge system is a practical solution. Our client also requested that we have a system that will not create drawdowns near another nearby contamination site, which is undergoing limited cleanup operations.

An observational approach to managing the drawdowns and adjacent contamination plumes is contained in the project specifications. Each of the multilevel piezometers in the monitoring network has specific drawdown limitations. The contractor will be required to develop a recharge system to prevent drawdown in the areas closest to the contamination plumes.

Conclusion

The interaction of cutoff walls with the soil and groundwater conditions can be difficult to determine. The groundwater models we used for this complex site provided a valuable insight into the interaction in addition to serving as a decision and design tool aid.