Sequential Excavation Method with ground freezing for DC Water’s First Street Tunnel
DC Water’s First Street Tunnel (FST) is part of the $2.6 billion Clean Rivers Project designed to reduce the occurrence of combined sewer overflows into local waterways. Due to flood events in the Bloomingdale and LeDroit Park neighborhoods, the FST Project was accelerated by DC Water to mitigate future flooding by boosting storage capacity and thus relieving the undersized combined sewers.
The FST Project was finalized in 2016 as a collaborative effort between DC Water, its consulting team consisting of Greeley & Hansen and McMillen Jacobs Associates, and the design-build team of Skanska, Jay Dee Contractors (SKJD) with WSP Parsons Brinckerhoff as the designer.
Scope of project. The FST Project (Fig. 1) included four primary shaft sites, a large 6 m (20 ft) internal diameter, 823- m (2,700-ft) long bored tunnel and three adit connections from the off-line shafts to the large-diameter tunnel. The adits which vary in size and length, are the subject of this article. The Channing Street site (Fig. 2) was established as the main site for an Earth pressure balance tunnel boring machine (TBM) and included a 49-m (160-ft) deep shaft with slurry wall support of excavation and 20-m (65-ft) internal diameter final cast-in-place permanent concrete liner. The construction sites at Adams Street, V Street and Thomas Street each consisted of off-line shaft structures tied to near-surface sewer diversion chambers and ventilation facilities. These three off-line shafts were connected to the large bored TBM tunnel via adit tunnels. Each of the three adits had a portion or the entire length excavated by sequential excavation method (SEM) in frozen ground.
The Adams Street adit (Fig. 3) was constructed in two phases. The first phase is the 23-m (75-ft) long and 5-m (16-ft) diameter section immediately outside the drop shaft which also serves as the deaeration chamber and was excavated by SEM. The second phase which is 91 m (300 ft) long and 3 m (10 ft) in diameter was excavated by Micro TBM (MTBM) launched from within the permanently lined deaeration chamber. MTBM excavation commenced toward a reception chamber that was excavated by SEM from the previously bored FST TBM tunnel. This SEM excavation had a 5 m (17 ft) diameter and a length of 4.5 m (15 ft).
The V Street adit was the largest SEM excavation of the FST Project and included a 6 m (20 ft) diameter, 21 m (70 ft) long SEM tunnel. This SEM tunnel was excavated with a 24 m (80 ft) radius curve from the 33 ft (110 ft) deep drop shaft toward the previously bored TBM tunnel. The reference design by DC Water initially included a straight alignment that was revised by the design-build team to a constant curve to avoid passing beneath a residential property and to allow the utilization of vertical ground freezing for both the SEM excavation and the deep Near Surface Structure excavation above.
The Pumping Station adit at Thomas Street had an excavation diameter of 4 m (13 ft). The 15-m (50-ft) long SEM tunnel was excavated from a 27 m (90 ft) deep shaft that will function as the temporary Pumping Station until the adjacent Northeast Boundary Tunnel connects the FST to the Clean Rivers Project.
Ground conditions. Ground conditions along the FST alignment consisted of an upper layer of recent fill, followed by Quaternary Alluvium, Cretaceous Potomac Group above Bedrock.
The Potomac Group was previously overlain by several hundred feet of soil deposits that were later eroded away and fine-grained cohesive soils are hard and overconsolidated. The coarse-grained cohesion-less soils are dense to very dense. The upper portion of the Potomac Group consists of Patapsco/Arundel Formation with transitional layers of fine-grained soils with high plasticity (G1) and lower plasticity (G2). SEM excavation was fully located in the lower portion of the Potomac Group, and consisted of the Patuxent Formation which included predominantly non plastic silty or clayed sand (G3A) or non plastic silty or clayed gravel (G3B). These G3 subgroups are distinguished from each other solely on the basis of whether the sand or the gravel fraction has the higher percentage in the particle size analysis. The G3 soil group within the Patuxent Formation is transitional interlayered with G1 and G2 clay, fine to coarse sand with traces of gravel and fines (G4) and fine to coarse gravel with traces of sand and fines (G5).
A shallow ground water aquifer exists, predominantly in the fill and Quaternary Alluvium. This unconfined alluvial aquifer is generally perched on the confining clay units of the upper Potomac Group of the Patapsco/Arundel Formation. Ground water within the confined water bearing layers and lenses of the lower Potomac Group of the Patuxent Formation exhibit artesian conditions.
All three adits have been excavated predominantly through G3A soils with various amounts and thicknesses of G1 and G2 layers in the face. When unsupported in the excavation, these G3A soils will exhibit fast-raveling to flowing behavior. The G1/G2 clay is prone to slide or fall as discrete blocks or wedges along fissured and slickensided fractures and therefore ground freezing was utilized to stabilize these soils for all three SEM excavations.
Design and constructions
SEM designs. Due to desired low impact on the urban and historic residential neighborhood and to mitigate risks of SEM excavation in unstable ground under high ground water pressure, ground freezing to temporarily improve the ground was utilized for all three adit excavations.
In a first step, the properties of the freeze body around the excavation were established based on empirical correlation of similar soils with a target temperature of –10 °C (14 °F). With these pre-established frozen ground properties, the required limits of frozen ground around the SEM excavation were developed with the help of numerical modeling. Initial support for the adits included shotcrete with lattice girder and welded wire mesh. The final liner for the V-Street and Adams Street adits included cast-inplace reinforced concrete and the Pumping Station adit was furnished with a hobas pipe.
In a second step, in order to verify the frozen ground properties derived by empirical correlation, frozen soil testing was performed. Representative split spoon samples from the ground investigation program were selected and recompacted to simulate undisturbed field conditions. Samples were saturated and tested in the laboratory under design target temperatures. A total of three unconfined compression tests and three pulse velocity tests on frozen soil test specimens were performed in accordance with ASTM standards. The test results confirmed previous empirical assumptions with a Young’s Modulus of 2,280 ksf to 4,100 ksf, thus finalizing the SEM design.
A freeze pipe layout with vertical and slightly inclined freeze pipes drilled from the surface was generated to provide a required freeze boundary with a target temperature of –10 °C (14 °F). However, since the SEM excavation was cutting through vertical freeze pipes, the numerical model assumed a reduction of frozen ground properties (thawing) while the shotcrete liner was gaining strength at the same time. Information on shotcrete strength gain curves, sprayed on frozen ground, was utilized from previous projects and a trial test was performed prior to and during construction to support and verify the required shotcrete strength.
Brine, as the freeze medium, was circulated via a utility trench to all three sites from the central freeze plant located within the Channing Street site. The brine had an initial temperature of –29 °C (-27 °F) going out and typically returning around –27 °C (-16.6 °F), with a warming of about 2 °C (35 °F). Freeze formation of the three adits, in order to generate the required freeze boundary with predefined properties, took between 40 and 60 days and was confirmed via temperature-monitoring pipes with thermocouples at various depth intervals that were read automatically.
Convergence monitoring in the SEM excavation was installed to verify proposed excavation length and initial support measures to allow possible adjustments during excavation.
V Street adit construction. The V Street adit (VS-A), which was the largest of the headings to be excavated, was the first adit to be constructed. The drop shaft size limited the equipment selection due to size constraints. The physical constraints required that the adit be excavated with a heading and bench excavation (Fig. 4). The top heading cross section was excavated at a 4.2 m (14 ft) height for the full length of the adit. Once the shotcrete initial liner was installed, the bench was excavated to the full 6 m (20 ft) cross section.
Vertical freeze pipes were installed to accelerate the overall schedule of adit construction, as well as facilitating ground support for the North Capital Street Diversion Chamber, directly above the adit that was being excavated concurrently with adit construction. The vertical freeze pipes were activated as the final liner of the drop shaft was being constructed. This saved more than six weeks of schedule as the freeze was able to be developed during the lining process. Excavation of the adit began as soon as the cast-in-place liner of the 7 m (23 ft) diameter V Street Drop Shaft was complete. The SEM was developed for the first 2.2 m (7.5 ft) using a Brokk 400 electric demolition robot with an Atlas Copco SB552 hydraulic hammer. The first two lattice girder arch sections were installed and shotcreted to full lining thickness. The excavation duration in the top heading dictated that the freeze pipes be re-established both on the crown and in the bench excavation to maintain the freeze temperatures (Fig. 5). A separate glycol freeze unit was set up onsite and a header was installed down the shaft into the adit. Each invert freeze pipe was hooked into this header once it was cut out of the top heading cross section (Fig. 6).
Once the initial heading was developed with the Brokk 400, an Antraquip AQM 150 roadheader was selected for excavation of the remainder of the adit. A skidsteer was used to tram the muck from the conveyor to the shaft. Excavation progressed by excavating 1.2 m (4 ft) of total cross section, installing the lattice girder, wire mesh and shotcrete. Both dry and wet shotcrete were tested and approved for use in the adits, but due to low sprayed shotcrete quantities and advantages in timely supply, dry shotcrete was selected for the initial support. The small size of the site and concurrent excavations on the site dictated that the shotcrete setup had to be set up and torn down each shotcrete cycle. Quickcrete 5,000 psi shotcrete was the supplied mix with 3 percent dry accelerator premixed into the super sacks. A silo and Putzmeister GM 060 were used for the shotcrete plant.
Construction of the adit’s top heading excavation used a Meyco Oruga shotcrete robot that was modified with a dry shotcrete nozzle. Due to the surface area and height of the heading, spraying with the robot was more safe and efficient as the nozzleman did not have to stand under the shotcrete as it was being sprayed. A temporary concrete invert was placed with the skidsteer to protect the bench frozen ground. This was completed before the shotcrete was applied. The adit was excavated to within 5 ft of the First Street Tunnel (FST). The tunnel was not constructed by the time the excavation of the adit was complete. A shotcrete bulkhead with wire mesh was installed until the FST tie in was made. The bench was excavated in 2.4 m (8 ft) advances with a Brokk 400 diesel demolition robot. As the re-established invert freeze pipes were encountered, each pipe was removed and backfilled. At each 2.4 m (8 ft) advance, the lattice girders from the top heading were completed through the invert. Wire mesh and shotcrete were applied to complete the initial support (Fig. 7).
Adams Street adit construction. The Adams Street adit (AS-A) was excavated concurrently with the bench excavation for the V Street adit and the Pumping Station adit SEM excavation. The Brokk 400 electric demolition robot was used to develop the first 2.2 m (7.5 ft) of heading similar to the V Street adit. The cross section of the Adams Street adit allowed for full-face SEM excavation. The Antraquip AQM 150 Roadheader was used to excavate the deaeration section of the adit. Mucking was done by a 2 cy muckbox on light gage rail. Excavation was advanced in 1.2 m (4 ft) sets. Each set would allow for installation of lattice girder, wire mesh and shotcrete. Due to limited space, a shotcrete robot could not be utilized and shotcrete was applied by hand in this excavation. Once the adit was excavated to the end station of the deaeration chamber a shotcrete bulkhead with wire mesh was installed to allow for construction of the cast-in-place lining (Fig. 8).
Pumping Station adit construction. The Pumping Station adit (PS-A) was excavated in full face with a Brokk 400 electric demolition robot with hydraulic breaker. Muck was removed by skidsteer from the heading to the shaft. The 3.6 m (12 ft) excavation was excavated in 1.2 m (4 ft) advances. Each advance allowed for lattice girder, mesh and shotcrete installation. Shotcrete was applied by hand for this heading (Fig. 9). Similar to the other headings the adit excavation was completed before the FST tunnel was mined. A shotcrete bulkhead with wire mesh was installed 1.5 m (5 ft) from the FST. The adit was then lined with a 2.4-m (96-in.) Hobas pipe and grouted.
Adit connections to First Street Tunnel. As the TBM excavated the FST, each location of the adit tie-in was blanketed with a surface freeze loop on the concrete segments and a series of invert freeze pipes were drilled and installed through the precast segmental liner from within the FST to allow for continuance of freeze and water cutoff at the invert of the adit tie in. This was work done on nonproduction shifts during mining of FST. Additionally, the aluminum freeze pipes that were cut while the TBM advanced were sleeved to the surface and reestablished to maintain the freeze block at the tie-in location at the crown and springline of the FST. Once the TBM was decommissioned and buried in place at the tunnel termination, a steel propping ring assembly was installed at each adit tie in location. The segments were then saw cut and removed in the cross section of the excavation. The SEM excavation at the Pumping Station adit (Fig. 10) and V Street adit were tied into the adits that were constructed from the drop shafts. Each were 1.2 m (4 ft) advances with lattice girders, wire mesh and shotcrete. The Adams Street adit connection was excavated by the same method but was terminated 4.5 m (15 ft) from the FST springline to provide enough space to receive the MTBM within the reception chamber (Fig. 11). A shotcrete bulkhead was constructed for the MTBM reception chamber. Once the cast-in-place tie-ins were constructed, the invert and surface freeze pipes were removed from service.
Lessons learned
Overall, the freezing process and SEM excavation worked out very well, with stable ground for both shaft and adit excavations (Fig. 12). Maintenance of freeze was achieved with resleeving cut freeze pipes (either by TBM or SEM excavation) from the top with a smaller diameter pipe and reestablishment of brine flow. A couple of freeze pipes were punctured by excavation equipment, but shut off valves enabled system isolation which enabled freezing to be maintained in other sites while repairs were made. Overall, aluminum pipes were easily mined through by the TBM. Due to concerns of warming in the invert zone once the TBM passed through, additional freeze pipes were successfully installed by drilling through the precast segmental lining to maintain the invert freeze which was not able to be reached by the resleeving from the top.
There was a learning curve in shotcrete operations. The pumping station shaft SOE was selected as a field trial location to spray the shotcrete on exposed frozen ground. The shotcrete was then cored at the required time after application in order to determine and verify actual shotcrete strength grain on frozen ground under field conditions. Nozzleman testing and certification was done in the frozen shaft, enabling the nozzleman to practice shooting on frozen ground. Utilization of premixed dry shotcrete that already included 3 percent accelerator was very beneficial to eliminate an additional step during shotcrete application and to avoid any over dosage. During placement, the heat of hydration of shotcrete coupled with frozen ground caused thick fog to develop and limited visibility of the shotcrete nozzleman, thus causing high rebound. Increased ventilation and blow-in exhaust proved to be sufficient in clearing the fog. The design for the initial support of shotcrete allowed for 50 mm (2 in.) of sacrificial shotcrete that was not required for the full design loads. This sacrificial layer insulated the remainder of the shotcrete layer from contact with the frozen ground and maintained the heat of hydration for the initial 152 mm (6 in.) pass of shotcrete.
The sizes of the drop shafts were the main factor in the selection of equipment for SEM operations. Due to the multiple heading sizes, choosing a single piece of equipment that could satisfy the size constraints of the shaft as well as heading size took collaboration with the equipment suppliers. Antraquip Corp. was selected to provide the roadheader (Fig. 13) for excavation of the headings. A modified Antraquip AQM 150 was utilized which used a AQM 150 turret, boom and transmission assembly mounted to an AQM 100 body which allowed for installation in the smaller diameter shafts and allowed full reach of the machine for the SEM headings. Other modifications were made to accommodate the unusual conditions caused by the freeze in the roadheader’s cooling system, electrical and hydraulic system.
Due to scheduling constraints, all three headings were excavated concurrently at a certain phase of the schedule. The Brokk 400 demolition robot was used due to its nimbleness and small footprint from within the shafts. The unit was able to reach the full height of the headings and muck during the initial development of the portal in each heading. The Brokk 400 was used to completely excavate the Pumping Station adit which had the least effect on the schedule. Due to removal for mucking this was the most inefficient of the mucking cycles. The V Street adit bench excavation utilized a Brokk 400 diesel demolition robot which sat on the bench and chipped material into the bucket of a skidsteer at the toe of the bench. In general, the headings cycled roughly 2 to 2.5 cycles/week which included mobilization and demobilization of the shotcrete plant for each cycle.
The adit tie-ins were excavated with the Brokk 400 Diesel demolition robot due to the necessity to move within each adit location in the tunnel. The machine was able to tram itself to the subsequent adit tie-in once excavation was complete. Both the roadheader and Brokk 400 proved to be good and suitable equipment in the abrasive sand and gravel frozen ground. Roadheader teeth were checked after every round, and worn teeth were replaced. Selected equipment was suitable for the sizes of SEM excavation and confined space.
Convergence monitoring in the adit excavations showed very little to no deformation in the excavation and stayed well below the normal predicted convergences of 10.6 mm (0.4 in.) in the design. This verified that the design parameters for the frozen ground were prudently selected. Excavation at the face and within the unsupported section prior to shotcrete application was stable at all times and allowed application of shotcrete initial support the following day.
All adits were constructed at locations with existing structures and utilities in close proximity. These structures had to be monitored for potential movements due to both ground deformations due to the adit excavation and the freeze-related heave and/or settlement. Because of the difficulty in accurately estimating freeze-related heave and/or settlements it is important to have a robust and proactive instrumentation and monitoring program. There is need for close coordination between the freeze designers and the instrumentation team to allow for early mobilization of mitigation actions when excessive freezerelated movements are observed. There were several successful mitigations to counter impacts of heave on adjacent utilities that included installation of heat trace rows to counter the growth of the ground freeze body. The success of these types of measures is dependent on proactive monitoring. Furthermore, the selection and locations of ground monitoring equipment should consider the frozen mass. For example, an inclinometer will not be responsive to excavation related ground deformation if it is installed within a mass that eventually freezes. Our recommendation for adit instrumentation is to have beyond SEM monitoring a plan that addresses the effects of ground freezing and installation of heat pipes at critical utilities such that freeze mitigation measures can be implemented immediately as necessary.
Summary
Ground freezing proved to be a good solution for ground improvement of SEM excavation for the FST project. This low-impact and less intrusive method of ground improvement enabled SKJD to adhere to the strict contract working hours of 7 am to 7 pm. The stable ground created by ground freezing enabled efficient use of the limited working hours. A significant benefit that ground freezing provides during the SEM mining cycle is the flexibility in stand-up time. For example, during the week work days, partially excavated ground could be left by covering with blankets at the end of the day, without the need for temporary support, such as a flash coat of shotcrete. This is unlike other ground improvement techniques, where temporary support needed to be installed before leaving for the day or in some instances, the requirement of round-theclock construction. Re-establishment of freeze pipes was considered a more critical SEM construction factor than partially completing the initial support and ground freezing allowed for this flexibility in the sequence of construction.
