The City of Atlanta’s Water Supply Program Tunnel project includes a 7,315-m (24,000-ft) long, 4-m (13-ft) diameter, hard rock tunnel. A complex aspect of the project involves connecting five, blind bore shafts to the tunnel in a location close to current drinking water reservoirs. During the supplemental geotechnical investigation following the initial contract award, results from additional borehole geophysics were reviewed. Unfavorably oriented fracture sets forced a change in the original pre-excavation grouting program designed for that site. Real-time grout monitoring and geophysical data were compared to provide assurance that the program, as designed, correlated with the in situ ground conditions. Following grouting, results were modeled, and all grouting data were reviewed to determine if grouting was complete.
The current water supply system operated by the city of Atlanta’s Department of Watershed Management (DWM) consists of four, aged, raw-water pipelines, one of which dates back to 1893. Based on previous assessments completed by the DWM, the entire water supply system is at, or will soon reach, its recommended useful life. As such, the city acquired the Bellwood Quarry in 2006 with the intention to create a water storage facility with a volume of approximately 9.1 billion L (2.4 billion gal) to serve approximately 1.2 million people.
The project location is shown in Fig. 1, which is generally in the northwest part of downtown Atlanta, GA. The overall project has been divided into two phases. The Phase 1 project connects the quarry and the Hemphill Water Treatment Plant (HWTP), and the Phase 1 Extension project connects the HWTP to the Chattahoochee Water Treatment Plant (CWTP) and the Chattahoochee River. A 7,315- m (24,000-ft) tunnel with a finished diameter of 3 m (10 ft) connects all three elements. The HWTP location is also where the city’s two most proximate drinking water reservoirs are located.
Procurement method. The construction-manager-at-risk (CMAR) model was used as the overall project contracting method. Specific to this procurement method involves producing a pricing set of design drawings, specifications, and data and baseline reports, which represent a partial design (typically 60 to 70 percent is used), to allow the CMAR to start pricing the work as design progresses toward final design. As the design evolves and changes are made, assumptions in bid pricing from various subs to the CMAR are reflected in various bid stages.
As changes to the evolving design are made and submitted to the CMAR and owner, specific details are delineated and given to all parties describing changes in the design. This allows for various subcontractors bidding on certain, niche parcels of work (as released by the CMAR) to either adjust their prices based on the revised design or keep their submitted prices based on their perceived risks, design detail revisions, and overall effect of design package revisions.
Hemphill site. The Hemphill site is the point on the project where the conveyance and storage parts of the project needs meet. During the initial phase of the subsurface investigation, a pump station shaft was intended to provide transmission of raw water to and from the tunnel to the HWTP as well as a construction shaft to provide access during construction. During this period, it was communicated from the city that any disturbance from the excavation processes to the existing unlined reservoir was an unacceptable consequence. A pre-excavation grouting program was designed as a risk mitigation tool to reduce the potential for any communication between the reservoir during either of the excavation shafts.
Schedule and blasting restrictions steered the design away from traditional shaft sinking techniques toward using blind boring methods. Blind bores allowed shaft excavation at the Hemphill site to be decoupled from tunnel excavation, subsequently removing it from the critical path of the project. A result of this change in connecting the surface components and the tunnel was that the location of the pumps now needed to be a lot closer to the tunnel. The size and breadth of the pre-excavation grouting program was reduced in accordance with the change in shaft size selection, construction methodology and arrangement.
Geologic conditions. The Atlanta Water Supply Program is located within the Piedmont physiographic province. Many underground components for the project are within a single geologic unit, the Clairmont Mélange. Characteristic of the mélange are interbedded biotitequartz- feldspar schists and gneisses, with minor granitic lenses. Foliation is very well developed and highly contorted wrapping around the granitic lenses while often displaying a sheared texture. Strike and dip of foliation commonly varies by 35° and the mélange is locally described to as “consistently inconsistent.”
During the initial subsurface investigation, boring RWB-15 was drilled based on accessibility while along the tunnel alignment at the Hemphill site. Two other borings, RWB-25 and RWB-26, were drilled based on the original proposed pump shaft and construction shaft locations. Boring RWB-15 showed extremely poor rockmass conditions, as the hole was reamed six times due to stability issues. Packer testing was not performed nor were down hole borehole geophysics due to concerns of lost tooling. While the other borings, RWB-25 and RWB- 26, showed some signs of similar geologic conditions (increased weathering along fracture planes, decreased RQD within discrete intervals, and increased permeability values), nothing observed nor tested was as pervasive or severe as RWB- 15, which was proximate to the tunnel, while the others were more distal.
As the design evolved from traditionally excavated shafts (drilling and blasting) to blind bores, it was determined that the rockmass close to RWB-15 (as suggested poor ground conditions) warranted additional borings. Additional drilling occurred proximate to RWB-15 and along the tunnel alignment close to the blind bore shaft locations. Borehole stability proved to not be an issue with the additional borings and packer testing and downhole geophysics were performed. Rock cores collected exhibited characteristics of the lineament hit by RWB-15, but fracturing was less penetrative while packer testing results indicated similar to slightly less permeabilities as shown in RWB-25 and RWB-26.
Design and grouting considerations
There were two underlying premises behind the pre-excavation grouting program. The first was the decree from the owner that under no circumstances shall the existing city water reservoirs be affected by blind drilling processes. Since the #2 reservoir was constructed prior to the 1930s and less than 30 m (100 ft) away from the shafts, it was unlined and, while all information from the subsurface investigation suggested that there was not a connection to the local water table, a risk mitigation measure was required. Second, as common with blind bore shaft sinking techniques, the area around the shaft is traditionally grouted to lower the potential for catastrophic fluid loss. During excavation, the shaft is filled with water to maintain hydrostatic balance and provide for a stable excavation as ground support is not installed. As the large diameter reaming process proceeds to the target elevation, water is maintained in the drilled shaft to keep the excavation open.
During supplemental drilling, the pricing set of documents needed for soliciting bids for the work by the CMAR were then issued and the tunnel contractor, Atkinson-Technique JV, solicited bids for the pre-excavation grouting work package. The initial preexcavation grouting layout for the blind bore shafts were used. Geophysical results from RWB-25 and RWB-26 suggested moderately open foliation joints (common within this unit) and that inclined grout holes oriented toward the tunnel boring machine (TBM) tunnel would be sufficient to intersect open, variable (relatively flat lying) foliation joints.
Once the data analysis was complete, it was determined that a change to pre-excavation grouting program was needed. Geophysical information from supplemental borings HDB-2 and HDB-3 indicated two open, high-angle joint sets with apertures ranging from two to four inches, in conjunction with open foliation joints along the “inconsistent” foliation. As designed, the existing preexcavation grouting program had a high likelihood of missing the newly identified joint sets.
Grout holes were to be drilled at a bearing of 260° at 10° from vertical. This orientation provided the highest probability of hitting both newly identified high-angle joint sets (Set 2 and Set 3) while also targeting the known foliation joint set, as shown in Fig. 3.
Primary grout holes were spaced at 5-m (16-ft) centers with secondary holes split spaced in between them. This pattern results in 2.4-m (8-ft) spacing between primary and secondary grout holes. Grouting started with primary grout holes and, once complete, drilling and grouting of secondary grout holes occurred. Each row had targeted grout elevations from which stages below would be grouted under pressure. This creates a block of treated ground that surrounded the future blind bore drilled shafts. One row near the tunnel alignment contained vertical holes and was drilled and grouted last to help seal off the grouted block, as shown in Figs 4 and 5.
Grout details. Pre-excavation grouting work was paid for as unit rates per bid quantity estimates. Estimated drill footages for both overburden and rock, estimated grout pump times, cement bag estimates and grout stages were provided. Grout mixes provided were by volume starting at 2:1 and progressing to a maximum 0.5:1 water to cement ratio. Refusal criteria was 0.25 gpm or less for five minutes at the full grouting pressures, which were 0.8 psi per foot from the point of injection. For the program, Type III cement was required. Bentonite was not used in any of the grout mixes.
The steps for thickening the grout mix were straightforward per grouting industry standards. Pressure and flow rate were tracked to determine if a mix change was warranted. If pressure increased while flow rate decreased, then the grout mix stayed the same, as the grouting system was functioning properly. When the pressure reached the target injection pressure and the flow rate was below the refusal injection rate, refusal on the stage was called and the packer assembly was moved. If flow rate was constant and pressure did not increase, the grout mix would be stepped down and thickened and injection continued at the specified grout mix until changes in pressure or flow rate were observed. Typically, mix changes were made after 380-757 L (100-200 gal) of grout were injected or if it was immediately apparent that the interval was open and a thicker grout would be needed.
Hayward Baker mobilized to the Hemphill site in April 2016 with drilling scheduled to commence in May. Overburden casing was advanced 1.5 m (5 ft) into rock to ensure the pvc casing was properly socketed into competent rock. Hayward Baker proceeded to install all surface casing for primary grout holes. Grouting was staggered between primary rows as to lower the potential for compromising grout holes that had not yet been grouted through communication from an active grout hole. Production drilling began in July 2016. Grouting of primary grout hole rows started in July 2016 and ended in August 2016. Drilling production secondary grout holes started in August 2016 and grouting all secondary holes and Row 1 (which was designated to be the last row drilled and grouted) concluded in October 2016. In total, 84 grout holes were drilled between primary and secondary portions of the program.
Hayward Baker utilized its proprietary grout monitoring equipment during all grouting. Cement was stored in bulk and then batched and mixed onsite. Grout was delivered to grout carts, small highly mobile tracked equipment that allowed for quick grout injection at various specified grout stages.
The main challenge associated with the program was associated with the ground. It was known that the ground was highly fractured around RWB-15, but conditions around HDB-2 and HDB-3 were better, and they were off the main lineament. With packers traveling along the hole, small pieces of rock would bridge the grout hole. This would prevent the packer from any further traverse towards the targeted depth. This was overcome through either re-drilling the hole and removing the obstruction or by grouting the interval and then re-drilling the interval. Grout hole drilling was the critical path item of the project. Drilling hours were extended to allow for second shift drilling to reduce the potential for grouting downtime, i.e. the grout carts have nowhere to go. During production grouting, this only occurred once.
Is grouting complete?
Toward the completion of secondary grouting, the natural question asked was whether grouting was considered complete. Schedule impacts concerning other subcontractors were at risk, as there were two other subcontractors who were scheduled to complete work prior to North American Drilling arriving onsite to begin blind bore shaft drilling. At this point, direct evidence of grouting efforts was not available. The decision was based on numerical grouting results and analysis. The following breaks down what grouting factors were analyzed.
Grout takes. The overall grout take for the program was about 200,600 L (53,000 gal). Primary grout holes took approximately 132,500 L (35,000 gal) of grout and the secondary grout holes took about 68,200 L (18,000 gal) of grout.
The secondary holes took about half of the grout that the primaries took. This reduction in grouting quantities follows the trend of what one would want to see — a progressive reduction in grouting quantities as you traditionally progress past primary grout holes through subsequent grouting.
Grout curtain. The grout curtain, as identified by two rows of grout holes around the perimeter of the drilled grout holes, took 68 percent of the total grout injected. This curtain is comprised of both secondary and primary grout holes. This was judged as reasonable, as the holes are on the edge of the area and not within the middle. Grout injected along the edge is not confined and will travel outside the treatment area as far as injection pressure, fracture aperture and cement particle size will allow.
Location of future blind bores. The cross-section of the treated area from around the future blind bores was looked at in relation to grout injected. Grout takes on 3 m (10 ft) of either side of the blind bores along the inclined grout holes, including the area within the blind bore (~9 m or 30 ft total) were assessed. Grouting data from grout stages that fit within this window around the blind bores were analyzed to see grout takes in the immediate vicinity of the bores, not the area overall. Of the total grout injected over the treatment area, the area immediate to the blind bores took only 29 percent of the injected volume. Of the grout injected proximate to the blind bores, 68 percent was 2:1 by volume, 17 percent was 1.5:1 by volume, 8 percent was 1:1 by volume and 7 percent was 0.75:1 by volume. The percentages of mixes used over the entire treated area are within 1-2 percentage points of the values just listed for the 9 m (30 ft) area around the blind bore shafts.
Packer testing data. The packer testing data from HDB-2 and HDB-3 reflect most of the permeabilities in the range of 1 x 10-5 cm/sec to 1 x 10-7 cm/sec, with the remaining zones around 1 x 10-4 cm/sec. Typically, permeabilities lower than 1 x 10-5 cm/sec are considered not groutable. It was judged that the data reflects this as just less than 70 percent of all the grout injected was 2:1 by volume. The thicker mixes pumped reflects the open fractures observed from the borehole geophysical results. Typically, when 0.75:1 by volume was injected, refusal occurred quickly, as one would expect.
Grout model. All grouting data were compiled into Civil3D, and modeled, as presented in Fig. 6. The first graphic is only primaries, the second graphic only secondaries and the third graphic is the composite. The blind bore shafts are shaded all the way down. What was observed is that the models corresponded to the grouting data. Following just primaries, large grout takes intersecting with the blind bore shafts was not observed. All the larger grout takes are within the treatment zone, but distal to the blind bore shafts. Following just the secondaries, there are few large to moderate grout takes, but as with the primaries, these are within the treatment zone, but distal to the blind bores. The composite log is judged to demonstrate good overall coverage of the grouting program.
Correlation. In some instances on the grout logs, zones were observed where takes were slightly larger (~54 m or ~180 ft, ~76 m or ~250 ft, and ~103 m or ~340 ft below ground surface). These were interpreted as the foliation joints identified in the field investigation. When looking at individual hole grout takes, there were sometime depths that would correspond between holes, but never really across more than two holes or so. Also, if a zone at 122 m (400 ft) on row four took 5,700 L (1,500 gal), the holes around it (both primary and secondary) were reviewed to see if there were any corresponding elevated values (even if not as large). Similarities in grout take volumes was not readily observed within the area proximate to the blind bores. This condition is also not apparent from the model.
iGrout logs. Lastly, when looking at Hayward Baker’s iGrout logs, a large portion of the time is spent achieving refusal. At thinner mixes, this is when pressure filtration takes over and the water is squeezed out of the grout and the finer fractures are filled. All the logs from the area proximate to the blind bore shafts reflect proper refusal without pre-mature thickening of the grout mix or poor injection trends (the trends reflected are what they should be — high initial rate of grout injection with flow rates slowly dropping while pressure remains constant until the refusal criteria is met).
Is grouting complete? The rationale for initiating a tertiary grouting program was not observed. One could inject more grout into the ground (you can always inject more), but the program would be at a point of diminishing returns. Also, based on the overall grout takes, and the grout takes around the blind bores, there was not data available that suggested that there could have been increased take around the blind bores. It was judged that the data set reflects a grouting program implemented as designed. Grout takes decreased between primary and secondaries, there were a few large takes (all outside the blind bore area), but overall the takes were not large, and the amount of thinner mix used reflects the permeabilities of the rockmass from the borings. The data suggest good encapsulation around the treatment area with grout penetration and many of the grout holes communicated to one another. It was judged that impacts to the blind bore construction schedule and the additional cost for a tertiary grouting program are greater than a small benefit from additional grouting that may be gained.
Hayward Baker demobilized offsite in October 2016. North American Drilling began reaming blind bore shafts number 1 and number 4 in February 2017. Large diameter reaming to the final diameter began in May 2017 and was completed in July 2017. During the large diameter reaming process, drill cuttings are recirculated to the ground surface. Once on the surface, cuttings are transported to sedimentation ponds via the drill return water where the large and small particulates settle out of suspension and water is then pulled from the ponds for further use.
During the large diameter reaming process, drill cuttings were inspected. Numerous, irregular shaped pieces of grout were picked from the drill cuttings pile. In addition, fluid levels in the shafts and at the reservoir were monitored during drilling. Both the reservoir water level and the water level in the blind bore shafts were constant during shaft reaming. The presence of grout pieces in the blind bore drill cuttings and stable water levels in both shafts and the reservoir further demonstrated that grout penetration was sufficient during the injection process.
An evolving design, coupled with interpretation of supplemental geotechnical information, required a change in the design pre-excavation grouting program as originally submitted to the CMAR for bidding purposes. Grouting results were quantitatively analyzed as well as modeled. Grout logs indicated proper grout injection and grout-rockmass interaction. Prior to blind boring operations, the grouting program was considered complete. During shaft reaming, grout chips and pieces were recovered from the drill cuttings pile and observed stable water levels in both the shafts and the reservoir substantiated grout penetration into the rockmass and provided assurance that the grouting program was implemented as designed.
Currently, the tunnel has been excavated by the five blind bore shafts and four of the five blind bores have been excavated. During mining, some of the grout holes were observed in the crown of the tunnel, but grout was not observed along fractures. Following four to six weeks after excavation, calcium was observed along fractures within the tunnel. Calcium leaching from the grout injected from the surface has been observed in the tunnel by the location of the blind bore adits, providing further substantiation to the effectiveness of the program.