The Northgate Link Extension project is the third phase of the Sound Transit master plan to extend light rail throughout the greater Seattle metropolitan area. When completed, the Northgate Link Extension will extend the existing light rail system north by about 7 km (4.3 miles) from the University of Washington to the Northgate neighborhood. A general overview of the project alignment is provided as Fig. 1. The project was funded by a combination of federal grants and allocations from local taxes. The general contractor on the project was JCM Northlink LLC (JCM), a joint venture between Jay Dee Contractors, Frank Coluccio Construction Company and Michels Corp. SoilFreeze, Inc. was retained as a subcontractor to provide ground freezing design and construction services.
The project involved the construction of two parallel 5.5-m, 25-cm (18-ft, 10-in.) (inside) diameter tunnels using earth pressure balance (EPB) tunnel boring machines (TBM) and precast segmental concrete tunnel liners. The two tunnels maintained roughly the same elevation along the alignment and were spaced roughly 12 m (40 ft) apart (centerline to centerline). A total of 23 cross-passages connecting the two tunnels were included at approximate 244-m (800-ft) intervals. Cross passage excavations were elliptical in shape with dimensions of approximately 5.6 m (18.5 ft) in height and 5.3 m (17 ft) in width. The cross-passages were 5.8 to 6.1 m (19 to 20 ft) long as measured between the exterior of the tunnels at springline elevation. Cross passages were intentionally located directly beneath public right-of-ways thus avoiding buildings. Many of the cross passages were located beneath narrow roads in residential neighborhoods with mature vegetation, existing utility structures and piping and limited parking areas. Excavation and support for each cross passage was completed using the sequential excavation method (SEM).
The subsurface soil and ground water conditions at the cross-passage locations were described in the Northgate Link Geotechnical Baseline Report (GBR) prepared by Jacobs Associates and the Northgate Link Geotechnical Data Report (GDR) prepared by Shannon and Wilson, Inc. Soils along the project alignment were glacially overridden and characterized as very dense or hard. Subsurface soils consisted of cohesionless sand and gravel (CSG), cohesionless silt and fine sand (CSF), cohesive clay and silt (CCS) and till and till-like deposits (TLD).
The contract documents identified three categories of ground support systems for the cross-passage excavations based on the soil conditions identified at each location. Category 3 ground support systems required ground improvement before excavation could begin. Category 3 cross passages were typically the most difficult soil conditions located near a transition between very permeable sandy CSG and the less permeable CCS or CSF. The boundary between the two dissimilar soils would have rendered dewatering efforts ineffective, as sufficient ground water drawdown could not be achieved. Initially five cross passages were identified as Category 3, but freezing was used to stabilize the soils for an additional five after soil explorations identified difficult soil conditions at those locations.
The five cross passages initially requiring Category 3 ground support and frozen from the ground surface are as follows:
- Cross Passage No. 21 (CP 21): The center line of the excavation was approximately 34 m (113 ft) below the ground surface (bgs).
- Cross Passage No. 29 (CP 29): The center line of the excavation was approximately 39 m (125 ft) bgs.
- Cross Passage No. 30 (CP 30): The center line of the excavation was approximately 37.5 m (123 ft) bgs.
- Cross Passage No. 31 (CP 31): The center line of the excavation was approximately 35 m (114.5 ft) bgs. In addition, CP31 was located at the low point of the tunnel alignment and included a sump pit which extended about 5 m (16 ft) below the centerline of the cross passage.
- Cross Passage No. 32 (CP 32): The center line of the excavation was approximately 23 m (76 ft) bgs.
Ground freezing for Category 3 cross passages
Ground improvements for Category 3 ground support systems were originally specified to be jet grouting or ground freezing. JCM was concerned with community and environmental impacts of the jet grouting in the densely populated urban/residential environment. In addition, freezing presented a more flexible schedule than jet grouting as the work could be done before or after the TBM had passed the location. As a result, JCM opted to pursue ground freezing as the preferred method to improve soils conditions.
Ground freezing involves circulating calcium chloride brine, chilled to -15 °F (-26 °C) or colder, through an array of steel pipes installed within the subsurface. Heat is extracted from soils immediately around each chilled steel pipe, freezing ground water within the soil matrix. As soils freeze radially outward from individual chilled pipes, the frozen soils from adjacent pipes eventually overlap to form an impermeable barrier with increased strength and stiffness. Ground freezing only needs to change the temperature of in-situ soils in contrast to more intrusive methods such as injection, permeation or replacement. Advantages of ground freezing include: the ability to pinpoint the location of the freeze pipes and thus the extent of the ground freezing; the ability to verify the extent of ground freezing through ground temperature monitoring; and minimal disturbance and spoils management at the ground.
At the request of JCM, SoilFreeze provided an approach that installed the majority of freeze pipes from the ground surface so that larger freeze equipment would be located out of the tunnels. Smaller freeze systems installed within the tunnels would supplement the primary freeze system elements from the ground surface. JCM felt that freezing from the ground surface would achieve two principal goals: 1) limit the work and equipment that needed to be located within the tunnels and 2) remove the bulk of the freeze system installation off the critical path of the project by installing the freeze pipes independent of the completion of the tunnels.
Design and installation of the ground freeze system
The choice to freeze from the ground surface contained technical and outreach challenges that were successfully addressed and mitigated through careful, innovative design and progressive, forward-thinking during the construction process. The primary challenge of freezing from the ground surface was to maximize the benefit of freezing for successful excavation of the cross passages while minimizing impacts to existing near surface infrastructure and the newly installed tunnels. Specific challenges can be categorized in these two areas: 1) within the freeze around the tunnels and 2) at the ground surface.
Specific challenges within the freeze zone included:
- Development of a frozen soil zone within time frames required by the construction schedule.
- Coordinating freeze pipe installations to minimize the number of pipes intersecting the path of the cross-passage excavations.
- Freezing the areas in the “shadow” of the tunnel, beneath the tunnel haunches where freeze pipes from the surface cannot access.
- Ensuring that frozen soils sufficiently adjoin to the extrados of the warmer tunnel liner segments.
- Ensuring that sufficient freeze continues during excavation and subsequent construction of the cross passage.
- Methods to manage and/or mitigate pressures that develop on the tunnel liner segments.
The specific challenges at the ground surface included:
- Optimize the freeze system to minimize electrical demands due to power supply restrictions.
- Coordination with utility entities to protect, maintain, and not interrupt, existing infrastructure crossing through work zones.
- Mitigating potential impacts to nearby residences resulting from equipment noise, limited vehicular access, and damage to landscaping.
- Methods to manage and/or mitigate ground movement associated with freezing activities.
Design of the frozen soil shoring system
In general, the design of a frozen soil shoring system requires two types of analysis. The first is a thermal analysis where the freeze pipe spacing, extent of frozen soil with time and frozen soil temperatures within the frozen soil shoring are determined. This information is then used to estimate frozen soil geometry and strengths to be implemented in a constitutive geotechnical analysis. The constitutive geotechnical analysis assesses deformations, stresses and strength-based factors of safety that could be anticipated within a given frozen soil shoring system.
Highly specialized two-dimensional (2D) engineering software was used for the thermal analysis. The thermal software utilizes finite element methodology to accurately evaluate thermal variations in the ground. Specifically, the thermal software calculates transient (timedependent) frost growth around chilled freeze pipes within a 2D plane.
Three-dimensional (3D) constitutive geotechnical analyses were calculated using finite element software. This computer software is capable of analyzing stress and strain behavior for complex 3D soil geometries under static or dynamic loading conditions. Soil-structure interaction between unfrozen soils, frozen soils, ground water and structural elements can be effectively analyzed.
Iterative thermal analyses were used to conclude that a grid of 30 freeze pipes was sufficient to freeze the soils between the two tunnels within a six- to eight- week time frame. This time frame was governed primarily by the cohesive soils that were present at each location. Cohesive soils typically have a higher moisture content, and therefore, a higher latent heat. Freeze pipes were arranged in a grid pattern consisting of six columns of five pipes as shown in Fig. 2. The profile of the approach is provided as Fig. 3. The spacing between each column of freeze pipes was approximately 1.4 m (4.5 ft) while the spacing between rows of freeze pipes was about 1.2 m (4 ft).
The outermost columns of freeze pipes, in the plan, were positioned outside the excavation limits of the cross passages. This left approximately 20 pipes that would be encountered during excavation for each cross passage. A grid pattern with fewer pipes would not adequately freeze the soils above and below the cross passage within the project schedule.
The circular shape of each tunnel restricted access from the ground surface to soils below the springline elevation of the tunnels directly adjacent to the extrados of either tunnel liner, thus forming a freeze shadow. Short haunch freeze pipes were installed through the tunnel liners below the springline of both tunnels to freeze just beyond the limits of the excavation. Although a shadow did not exist above the tunnels, it was decided that haunch freeze pipes would also extend above the springline of the tunnels instead of drilling additional freeze pipes from the ground surface.
Typical locations of haunch freeze pipes are shown in Fig. 4. The number and location of the haunch freeze pipes varied at each location based primarily on the potential for ground water gradients following the tunnel alignments. Installation of the haunch freeze pipes did require a chiller to be installed in each tunnel, however the size of the chiller was significantly smaller than one that would be required to freeze the entire freeze zone. The in-tunnel chillers were placed on elevated platforms installed along the interior the tunnel liners, near springline elevation. The in-tunnel chillers did not impede traffic or work within either tunnel.
A final design hurdle involving freeze pipe positioning was to develop a system that would ensure that frozen soils would remain frozen directly adjacent to the tunnel liner extrados. This interface represented the weakest point of the frozen soil shoring. Ventilation systems actively circulating air through the tunnels warmed the interior face of the segmental concrete tunnel liners. This air circulation acted to create a warm internal boundary that could generate enough of a thermal gradient through the concrete liner to thaw a thin layer of soil and induce a seepage path into the excavation. To combat the warmth imparted by the ventilation system, steel pipes circulating chilled brine were affixed to the interior face of the tunnel liners surrounding the proposed cross passage portals, as shown in Figs. 4 and 5. Chilled brine was supplied to these pipes using the same chiller systems that were installed for the haunch freeze pipes. Additionally, insulation blankets were affixed to the interior surfaces of the tunnel liner to limit warm air exposure.
The system of vertical freeze pipes installed from the ground surface in combination with the in-tunnel freeze pipe systems, were analyzed using both the thermal and constitutive geotechnical models previously discussed. Thermal models predicted a zone of ground improvement significantly smaller than the specified ground modification extents presented in the project drawings. Despite the frozen soil zone being smaller than specified, constitutive models indicated that maximum frozen soil deformations during excavation would be less than 1.3 cm (0.5 in.) at the crown of the cross passage, with little to no deformation at the invert of the cross passage. In addition, the calculated strength- based factor of safety for the frozen soil shoring generally exceeded 5.0 at each cross-passage location. Reviewers eventually allowed for a reduction of the modified soil zone extents for three reasons: 1) the location of each freeze pipe would be known; 2) the radius of frozen soil around each pipe is relatively uniform and predictable using thermal modeling and 3) the extent of the frozen soil could be verified and monitored using thermocouple sensors placed at key locations.
SoilFreeze employed zone freeze pipes as the primary freeze elements that would be installed from the ground surface. Zone freeze pipes are a patented system designed to isolate freezing within a targeted zone at depth. Zone freeze pipes work by circulating chilled brine through a 10-cm (4-in.) steel vessel with high heat transfer characteristics that forms a freezing length at depth. The portion of the zone freeze pipe extending from the top of the steel vessel to ground surface is comprised of high-density polyethylene (HDPE) brine delivery lines confined within a gasketed, watertight, polyvinyl chloride (PVC) sleeve. The HDPE lines are buffered from the PVC sleeve and the surrounding soils using air to form an insulated length with intrinsically low heat transfer characteristics. The insulated length of the zone freeze pipe served to minimize freezing of soils extending above the targeted freeze zone to the ground surface, thus reducing the required chiller capacity and maximizing system efficiency. This resulted in lower energy demand and also minimized the freeze system footprint within the narrow above-ground streets. It was estimated that the zone freeze pipe technology reduced the required chilling capacity and power requirements by at least half. Assembly of the insulated PVC portion of a zone freeze pipe installation is shown in Fig. 6.
Installation of the freeze system
The installation process for each freeze system posed its own unique set of challenges. When the freeze pipe grid was superimposed on existing conditions at each site, conflicts were apparent, requiring adjustments. The following is a summary of the conflicts at the ground surface for each cross-passage location.
- CP 21 — The location of the cross passage was partially located beneath a low traffic, two-lane roadway at the southeast corner of the University of Washington campus. The entire eastern half of the cross passage was located beneath a vegetated slope that could not be disturbed, requiring that all work had to be completed within an area that was less than half the size of the designed freeze pipe grid at depth. In addition, a 20-cm (8-in.) ductile iron water line and manhole structure were located within the work area.
- CP 29 — This cross passage was located beneath a narrow neighborhood side street. There were no buried utilities at this site, however overhead power was present on the north side of the road and local service lines crossed over the road and above the freeze pipe installations.
- CP 30 — This cross passage was located beneath a narrow side street. Large heritage trees were located on the north side of the road, with additional trees just to the south. The heritage trees had protected root zones that extended out into the work zone. Utilities consisted of overhead power on the south side of the road, two buried fiber ducts, a 5-cm (2-in.) gas line, a 31-cm (12-in.) cast iron water line and a 51-cm (20-in.) cast iron water line. Neither of the water lines were located within the work area.
- CP 31 — This cross passage was located below the northbound lane of a residential avenue. Accordingly, the work area was located within the northbound lane and extended slightly beyond the east curb line. The buried utilities extending through the work area consisted of a 104-cm (42- in.) diameter steel riveted water line, a 20-cm (8- in.) cast iron water line and a 5-cm (2-in.) gas line. A fire hydrant was removed prior to mobilizing to the site.
- CP 32 — The shallowest cross passage was located directly below a small residential road located near the intersection with a busier thoroughfare. Overhead power lines were present to the south of the work area and buried utilities crossing the work area consisted of storm drain laterals, an 20-cm (8-in.) concrete sewer line, a 10- cm (4-in.) gas line and a small 5-cm (2-in.) water line lateral.
Where conflicts were identified, the location of the freeze pipes at ground surface were adjusted. In a few cases the adjustments were minor and the freeze pipes could still be installed vertically. However, in most cases the adjusted freeze pipe locations were significant enough that the pipes needed to be installed at an angle (or batter) to maintain the required spacing at depth. Freeze pipes that were installed at angles so that the freeze vessels passed through the original design grid point in plan at the mid-level elevation of the cross passage excavation. Due to the depth of the cross passages, installation angles were minor, typically ranging between 2 and 10 degrees from vertical. The battered freeze pipes typically resulted in slightly closer freeze pipe spacings at the top of the frozen soil column and a greater spacing at the bottom of the frozen soil column at depth. Thermal models were adjusted to include spatial variations of the freeze pipes with depth at each cross passage. Ultimately it was determined that the extent of the frozen ground was not significantly impacted by these spatial variations.
After layout of the freeze pipe grid was completed and conflicts were identified at each site, JCM returned to the site and installed 31-cm (12-in.) diameter PVC sleeves to mark freeze pipe installs that were located directly adjacent to utilities. In general, this process was safely completed with a vactor truck. The PVC sleeves extended to depths just below the utilities and served as a guide during the installation of the freeze pipes.
The second advantage provided by the PVC sleeves installed by JCM was to provide redundant insulation around the freeze pipes to reduce freezing around utilities. Freezing does not adversely affect most utilities including gas, fiber-optic, sewer, power and running water. However, during negotiations with the various utility agencies, there was an expressed concern about the impact of the freezing temperatures. Therefore, the additional insulation provided around the utilities was an added benefit of the PVC sleeves.
Freeze pipe installations from the ground surface were completed by Cascade Drilling. A steel casing was advanced using sonic drilling techniques to design depths. For each installation, a steel freeze vessel with HDPE brine supply/return lines attached was incrementally lowered into the cased hole while 6-m (20-ft) sections of PVC sleeve were stacked and joints waterproofed. After the freeze pipe was lowered to depth, the steel casing was retracted and the annular space around the freeze pipe was backfilled with grout to the ground surface.
The work zones required to install the freeze pipes blocked the narrow roadways at CP 21, 29, 30 and 32, however a single lane was maintained past the drilling zone at CP 31. Drilling was completed during hours dictated by the city and site cleanliness was maintained throughout the installation process. No complaints were received from the general public during the installations. After freeze installations were completed, roads remained closed at CP 29, 30, and 32. Pipes installed at CP 21 were recessed below the street level and covered with a steel sheet which allowed the road to remain open during freeze operations.
Tracking the locations of each freeze pipe at depth was critical, therefore the steel drill casing was surveyed prior to the installation of the freeze pipe. A downhole gyroscopic survey tool was used to determine the spatial locations of each installation between the tunnels for varying depths. If the space between two adjacent freeze pipes was excessive as determined by an updated thermal model utilizing as-built data, an additional freeze pipe was installed in the gap to maintain the necessary freeze coverage. A total of 14 additional freeze pipes were added for all five cross passages due to drilling inaccuracies. The downhole survey tool was also implemented mid-drilling process for pipes located immediately adjacent to the tunnels to verify that the hole alignment had not drifted towards the tunnel. All drill holes verified in this manner were found to be on target and were installed to depth.
Within the tunnels, JCM was responsible for installation of the haunch freeze pipes to better control schedule and space constraint impacts within the tunnels. JCM installed the short haunch pipes by pre-drilling through a pneumatic packer, removing the drill tooling, then driving the steel freeze pipes to depth. The location and alignment of the short haunch freeze pipes were surveyed by JCM using conventional optical surveying methods.
Freezedown and excavation
Freezedown. Start of freezedown and freezedown durations varied for each cross-passage location. Chilling units servicing the freeze pipes at ground surface were initially powered by generators. At some cross passage locations, chillers at the surface were eventually switched over to the local power grid. Chiller units at the ground surface had footprints of 2.4 m and 5 m (8 ft by 16 ft). One chiller unit was sufficient to freeze each cross passage, except at CP 31 where additional freezing capacity was needed to support the excavation for the deeper sump pit.
Chillers and generators operated full-time during freezedown, excavation and final lining construction phases of each cross passage. Accordingly, mitigation of mechanical noise for each freeze system was a requirement by the city. This was accomplished by using specially designed low-noise fan blades on each chiller unit, working with the power company to get a power drop at each site thereby eliminating the need for generators, and building a sound dampening structure around the generators and chillers. The sound dampening structures were constructed by JCM and consisted of plywood walls internally lined with sound dampening blankets and insulation. Each sound dampening structure was large enough to provide ample air circulation to the chillers, while remaining small enough to fit within the limited footprint at each site. These measures successfully dampened the mechanical noise to less than 63 decibels, the operational threshold allowed by the city.
In general, the frozen soil shoring was formed within the time frame calculated during the design process. Brine and ground temperatures were closely monitored during the freezedown process, using thermocouples installed at various depths around each site. Recorded ground temperature data was used to calibrate as-built thermal models. Measured brine temperatures were used as input to the thermal models and the soil and freeze pipe parameters were slightly adjusted so that the calculated temperatures closely matched the observed temperature trends. The calibrated thermal models were instrumental in estimating spatial extents of the ground freezing at varying depths, as well as extrapolating future development of the frozen soils.
Ground movement. One of the challenges on this project was monitoring and controlling the ground deformation associated with freezing the ground for extended periods of time. The expansion that occurs when pore water undergoes the phase change to ice does not typically cause significant volumetric expansion of a soil unit. However, when freezing is maintained for a long period of time, cryogenic suction will draw water toward the freezing front and can form lenses of pure ice. If conditions permit, such ice lenses can result in segregation heave and excessive deformations at the ground surface.
The zone freeze pipes were never designed to completely prevent freezing of soils extending above the freeze zone at depth. As a result, near surface soils eventually froze during prolonged freezing operations. As the near surface soils froze, some minor heaving at the ground surface was captured by optical monitoring points and extensometers. Observed ground movements were minor and slightly varied at each site. The observed movement was highly dependent on surface water runoff, available ground water, site soils, and instrumentation accuracies. Ground deformation was limited and controlled using a number of techniques which included:
- Managing surface water to limit surface runoff from entering the work area and collecting.
- Adjusting/raising the temperature of the brine. This was a delicate balance between being cold enough to maintain freezing at depth while simultaneously warming near surface soils to limit the development of ice lenses.
- Use of heat trace tape and blowing warm air into the upper portion of the zone freeze pipes.
Where there were no utilities located within the freeze grid, ground movement was not controlled and any deformation was repaired as part of the site restoration. Where utilities were present, particularly the 107-cm (42-in.), riveted-steel water main at CP 31, movement was successfully maintained below thresholds defined in the project specifications. No damage to any utility was observed.
Deformations were also observed along the tunnel liners at depth. Some deformation of the tunnel was acceptable and remediation actions were limited to varying supplied brine temperatures. After taking such measures, observed deformations stopped and rebounded slightly after excavations began.
Excavation. Prior to excavating each cross passage, short probe holes were drilled through the tunnel liner to verify that ground water had been completely cut off, and the tunnel liner segments could be removed without ground water intrusion. Excavation was completed using SEM construction methods and a remotely operated roadheader. Ten cm (4 in.) of an initial shotcrete lining was placed over frozen soils exposed during SEM work. Conservatively, two of the 10 cm (4 in.) of initial shotcrete lining were intended to be sacrificial and provided an insulating barrier prior to application of the final shotcrete lining. This initial layer, paired with welded wire fabric, facilitated successful placement and strength gain of the dry shotcrete mix.
As the excavation progressed, rows of freeze pipes were deactivated and purged of brine ahead of the excavation work. This allowed the freeze pipes in the path of the excavation work to be cut and removed without releasing the calcium chloride brine into the excavation, which could thaw frozen soils, potentially impacting the integrity of the frozen soil shoring. Once the freeze vessels were cut and deactivated, soils along the base of the cross-passage excavations would no longer be actively frozen for the remainder of the project. This was anticipated during the design phase, and the outer rows of freeze pipes and the haunch pipes were strategically positioned to remain active and compensate for the additional heat introduced during the excavation and subsequent construction of the cross-passage.
During the excavation, the majority of the pipes installed from the surface were cut, recapped, pressure tested and then buried in the shotcrete. Two of these cut freeze pipes were brought back online immediately over the portals at each tunnel to help maintain the freeze along the crown of the excavation, directly adjacent to the concrete tunnel liners. In addition, several of the interior freeze pipes were replumbed with heavy duty rubber hoses that connected freeze vessel segments at the crown of the cross passages with freeze vessel segments extending beneath the excavation. The hoses were buried into the walls of the excavation and covered with shotcrete. This innovative approach was sufficient to keep the soil mass below the excavation actively frozen for the duration of the project.
After the excavation was completed, the freeze system remained in operation for approximately eight weeks during which time the waterproofing system was installed, reinforcing steel was placed and the permanent final concrete lining was cast in-place. During this time, the frozen soil system remained operational to maintain the robust and water-tight shoring.
Serious project challenges were successfully overcome with freezing from the ground surface providing safe, stable, and water-tight shoring for five Category 3 crosspassage excavations along tunnel alignments for the Sound Transit Northgate Link Extension project. Each of the above-ground freeze systems was successfully installed and operated within crowded and difficult urban environments with minimal impacts to the population. Zone freeze pipes were successfully employed to limit freezing impacts to near surface infrastructure, optimize chilling capacities and focus freeze efforts at depth to develop a robust frozen soil shoring system. Downhole survey techniques were established that can be implemented on future projects requiring drilling of freeze elements in close proximity to sensitive structures at significant depths. Methodologies to mitigate propagation of frozen soils and limit frozen-soil-induced heave deformations around existing infrastructure were successfully employed.