The six-foot tunnel was nearly one mile long and accessible from just two 80-foot deep vertical shafts, one of which is in the river. Closure of the tunnel presented a unique remediation challenge because of the limited access, considerable volume of the tunnel and because the tunnel was full of potentially contaminated water. A plan was developed and implemented that closed the tunnel by filling it with cementitious grout while simultaneously removing and treating the displaced water. The grout used to fill the tunnel had to meet demanding requirements for both regulatory acceptance and workability.

The project work plan had to take into account a number of unique complicating factors:

• The tunnel was level, making it difficult to displace water upward with a heavier grout.

• Access for material placement was really only practical from the land end of the tunnel.

• Access by divers into the flooded tunnel was limited to about 200 meters (600 ft) from each end.

• The tunnel could not be dewatered due to the nearly unlimited volume of water from both the tunnel and infiltration from the river that would need to be treated.

• Once work would begin to fill the tunnel, no further personnel access would be permitted, requiring a remote operation.

• Redundant systems would be required to account for multiple variations and breakdowns that might occur. It would be difficult to ever restart the work in the event of a disruption.

• The work plan had to account for the fact that the first stage of grouting might not be totally effective in sealing the tunnel up to the roof, so a secondary grouting system would need to be devised.

• Because of the dimensions of the project and the problems of grout setting, the system would be designed to operate continuously once work started until completion.

The key to success on the project was the selection of a grout with parameters that would fit the situation as well as the design of a placement system that could reliably place a large amount of grout over a period of a few days. Grout for an application like this had never been designed and it was necessary to go back to the laboratory to search out the ideal combination. Since the grout had to be mixed and pumped from shore and based on the placement work plan, the initial grout would have to pass through almost 5,000 feet of pipe to the point of placement and would have to flow back through the tunnel, displacing water, for a distance of at least 1,000 feet over a period of 30 hours or more before it would set too much to pump. The volumes were considerable. It would take approximately 6,000 cubic yards of grout to fill the tunnel, so more than 1,300 cubic yards would have to be placed before it would start to set. Based on these requirements, as well as regulatory requirements for the completed grout fill, parameters for the grout design were set as follows:

• Unconfined strength at 28 days in the range of 100 to 200 kPa (15-30 psi).

• Heavier and more viscous than water so that water would be displaced out of the tunnel as the grout was placed.

• The grout should be immiscible in water, so that it would form a face displacing the water, rather than a semi-mixed zone of water and grout.

• The grout should have an extended set time (24 hours or more) to allow significant volumes of grout to be placed from a single point.

• The mixed grout had to have low viscosity, preferably less than 60 seconds Marsh Funnel to allow it to be placed through small diameter pipes over long distances without significant head losses.

• The permeability of the hardened grout had to be no higher than 1 x 10-6 centimeters per second.

Model test of grout placement in a tunnel full of water.

Laboratory Testing

The grout mixtures were first subjected to a series of tests including: viscosity, density, set time, bleed, shrinkage, unconfined compressive strength and permeability. With respect to viscosity, all of the grout mixtures were workable or could be made workable using additives. The set times of the CB and BFSB mixtures were acceptable, but some of the CF grouts set too quickly for the placement conditions and BFSB grouts that did not include some portland cement did not set at all. The most significant finding was the variability in the bleed and shrinkage of some of the grouts. While the CB and BFSB had minimal shrinkage, the CF grouts performed poorly. No additive provided significant improvement in the bleed, so the CF mixtures were deleted from the program. While the strength and permeability of the CB and BFSB grouts were both acceptable, the BFSB grout had better properties.

Three kinds of tests were performed to check the compatibility of the grout mixes still under consideration with site leachate, specifically DNAPL (dense non-aqueous phase liquid) and APL (aqueous phase liquid). In the first test, the fluid grout is poured into pans full of leachate and of water for comparison. The grout is tested and a comparison is made between the times for the grout to set in leachate compared to times to set in water. The results showed no effect due to the leachate.

The hardened grout was subjected to an immersion test designed to predict the long-term performance of cement products exposed to chemicals. The test is performed by soaking cured grout specimens in sealed jars filled with leachate and tap water (for comparison) for up to 45 days. No effect due to the leachate was observed.

Pipe puller and shaft at the shore end of the project.

Model Testing Program

The first test was a simple tremie test with the grout placed through a tube into a container full of water. The grout should not mix with the water and the bleed of the grout should still be acceptable as it sets underwater. The selected grout passed this test with no problem. Even when it was placed in a manner so that it dropped through the water, the grout bulb remained intact until it rejoined the grout at the bottom of the container with essentially no mixing with the water.

The second bench-scale test was devised to model the horizontal displacement of the water in the tunnel as the grout is placed. The setup was a half-pipe full of water with a grout tube inserted at one end. A long slope of grout pushing the water forward was expected. There actually was a surprisingly steep face of grout that formed. Again, there was essentially no mixing of the leading edge of the grout with the water.

Placing the steel "stool" in the river shaft in preparation for the tremie seal.

Field Implementation

The first step was to prepare the tunnels by inserting the grout pipes. The difficulty here was that the tunnel had to remain flooded and was only accessible to divers for a short distance. The shore shaft was a six-foot diameter riser pipe 82 feet deep. At the bottom there was an immediate transition into the horizontal tunnel that generally was about 41?feet wide and six feet high, lined with concrete. The only other access was on a small concrete platform out in the river one mile away, where there was an eight-foot-diameter riser shaft.

Fabricated rollers were installed at the base of the access shafts to allow pipes and cables to be pulled through without damage. This was accomplished using divers. Next a small remote-operated vehicle went through the tunnel carrying a small diameter cable. This cable was used to pull through a larger cable that was about 10,000 feet long. This cable was wound onto cable drive pipe pullers at both ends of the tunnel so that it could travel back and forth through the tunnel, dragging in the many components of the grouting system. Considerable attention was paid to the details of cable connections to avoid the possibility of snagging on a previously placed component.

The first part of the grout placement system to be installed was the secondary grout line. Since the most likely place for voids to form would be above the grout as it settled, this line had to be at the top of the tunnel. It consisted of a 3-inch diameter HDPE pipe with rubber sleeve valves placed over holes drilled in the pipe on approximate 50 foot centers. The pipe was dragged through then evacuated by blowing the water out of the line with compressed air to a sump pump at the opposite end, forcing the line to float to the top of the tunnel.

The cable was then rewound on the pipe puller at the shore and attached to the first and longest primary grout pipe that was then dragged through using the river-end pipe puller. This pipe reached all the way to the base of the river shaft and succeeding pipes were each about 1,000 feet shorter than the last. A total of five primary grout pipes would be placed, ranging in diameter from 6 inches to 4 inches. To speed placement of the pipes, they were prefabricated into sections of approximately 500 feet. Each pipe had a conical tip designed to keep out sediment as the pipe was dragged into the tunnel, yet would open when grout pressure was applied. An emergency sleeve valve was mounted a short distance back from the tip to provide an outlet in the event the tip was blocked.

Once the pipes were all placed, it was time to block the river shaft so that grout would be forced back toward the shore once pumping started. This was accomplished by placing a steel stool in the shaft, sealing around the edges and pouring tremie concrete on top of the structure to seal the shaft. After this was done, the only grout pipe left accessible from both ends was the secondary grout pipe.

The tunnel now was ready for the grouting operation to begin. To simplify the mixing process, all the bentonite slurry needed was premixed and stored in large ponds on the site. Cement and blast furnace slag cement were delivered to the site pre-blended, so only one dry component would have to be mixed, allowing for easier quality control and a faster placement rate.

The one-component mixing system was crucial to the project design since it would allow for a continuous - as opposed to a batch - mixing system. The mix plant consisted of a 6-cubic-yard-capacity colloidal enclosed mixer fed by a variable flow liquid system and a variable speed dry material feed system. The level in the mixer was controlled by a sonar device; feed rates of the materials were adjusted based on real-time continuous density measurements provided by a highly accurate coriolis density meter. Slurry was pumped in a rate sufficient to keep the level in the mixer constant as the material was pumped out and the rate of feed for the cement blend was controlled based on density.

This highly productive mix plant was designed to produce grout at a rate of more than 1,000 cubic yards in a 24-hour period. After some test runs where grout was mixed and pumped into a pit on site, the work was finally ready to begin.

A few moments pushed the system to its limits. The grout barely could be pumped the long distance out to the end of the first pipe. Once it began to consistently flow, water started to rise in the shore shaft. Work continued on a 24-hour-a-day basis until completion. Problems arose during the first run when a violent nighttime thunderstorm forced a cessation of operations and the pipe was lost before the grout reached the next pipe end. Work continued from the second pipe but it was known that there might be a gap at this location. Grouting through the succeeding pipes went according to plan. At one location, there was a breakout of grout into the riverbed through an old shaft that was supposed to have been sealed. The grout came up the shore shaft on schedule.

Once the primary grout had been allowed to set, the secondary grout program began. The secondary grout pipe was pressurized, allowing the seals to pop open in any locations where there might be a weakness and theoretically no grout. After a period of time, the secondary line was flushed out by pumping water through it and the secondary grout allowed to set before starting another phase of secondary grouting. By the end of this process, grout was coming out of the ground at various locations, so it appeared that the tunnel was tightly sealed.

Grout mixing plant.

Field Verification

An additional quality control requirement of this project was confirmatory drilling after placement and curing of the grout to visually confirm the effectiveness of the fill and to check for the presence of any voids above the grout. Because of problems that arose during the grouting, there were two locations in which grout was introduced into the tunnel from a tremie pipe that was in front of the grout face. This created the potential for a gap caused by hydraulically confined water. Three borings were advanced into the tunnel -- one from land where the tunnel passed under an island and two from a barge in the river.

Locating the tunnel required great care, as none of the original construction plans were available. The drilling coordinates were calculated from known locations of the tunnel shafts. At each location, a well casing was carefully set vertically to the top of bedrock and a full-hole outer tube core barrel system was used to ensure a vertical hole through the bedrock down to the 82-foot depth of the tunnel. To check for the presence of a void in the grout at the top of the tunnel, caution was used at the appropriate depth and the drilling operation was videotaped to capture the drop of the drill stem if it were to occur.

The boring on land intercepted the tunnel on the second attempt and confirmed the presence of one of the suspected gaps (caused by the stoppage of grouting during the lightning storm). When the tunnel was penetrated, air erupted from the borehole and the drill bit dropped six feet to the tunnel bottom. A relief hole was bored into the gap some distance away so that additional grouting could fill the gap.

For the work in the 6-mph current of the river, the drilling rig was placed on a barge equipped with a 2-foot-diameter hole for the drill bits. The hole was positioned directly over the calculated drilling coordinates using global positioning. When the barge was close, one spud of the barge was sunk into the riverbed while a tug was used to rotate the barge to achieve the final location. Both of the borings in the river confirmed a complete fill with competent grout.

Pumping unit with density meter to the right and flow meter to the left.

Conclusions

Blast furnace slag cements need a percentage of portland cement to perform at all in this application. Without it, they do not set. With a proper mix ratio, the grout will have a low viscosity, low bleed, low shrinkage and will form a grout of low strength.

This project clearly tested the limits of the current knowledge of grout mix design as well as the technology of grout mixing and pumping. The project parameters and the design requirements made the job one of the most challenging imaginable. The combination of a far-sighted owner and a competent contractor to do the design, testing and construction supervision made the project a success.