The eglinton crosstown lrt is a new public transportation project in Toronto, Ontario. It is being constructed by Metrolinx, a provincial Crown corporation. The project is a 19km corridor for rapid transit using light rail vehicles and includes a 10km underground section. The tunnels west of Yonge Street consist of approximately 6.2km of two TBM-bored tunnels, eight cross passages, three shafts and multiple station headwalls. Contract ECLC1-15 was awarded in October 2012 to Crosstown Transit Contractors (CTC), a joint venture of Obayashi, Kenny, Kenaidan and Technicore.

As of June 1, 2016, CTC has completed the excavation of the two TBM-bored tunnels, five hand-mined cross passages, three shafts, and all station headwalls on the western portion of the contract.

This paper describes the ground improvement methods applied to the first five cross passages and shares the authors’ lessons learned from the experience.

CONTRACT CONDITION

Contract ECLC1-15 documents included a Geotechnical Baseline Report (GBR) that defined the subsurface conditions. The GBR identified plastic till and interstadial sand to gravel, which were anticipated along the tunnel alignment. The conditions included a regional confined aquifer at the location of the cross passages with a maximum head of 6m above the crown of the cross passage (Figure 1). Thus, pre-excavation dewatering was necessary to maintain the stability of each cross passage throughout construction. The GBR also provided the anticipated range of steady state groundwater pumping rate for each cross passage to be from 110 to 440 m3/day. Similar to other contracts, the design of the dewatering system was the contractor’s responsibility, and to determine the most effective way to lower the groundwater level to below the invert of the cross passage excavation.

DEVELOPMENT OF THE INITIAL PLAN

After studying the GBR and the construction conditions, CTC developed an initial dewatering plan with the dewatering contractor – Insitu Contractors Inc. This called for the installation of deep wells from ground surface. Because there were many utility and easement restrictions, the number of wells that could be installed from the surface was limited. The initial plan was simulated with 3D-Finite Element Modelling (3D FEM) by Terraprobe Inc., and the effectiveness of the initial plan was evaluated. The results of 3D FEM were that two deep wells were not sufficient to lower the groundwater level completely. This was due to limited aquifer depth and the low hydraulic conductivity of the soil. The analysis showed that, in order to lower the groundwater table to an acceptable level, 10 wellpoints needed to be located between the deep wells inside the tunnels.

The initial plan called for two deep wells to be installed from the surface on either side of the cross passages. Wellpoints would then be installed from inside the tunnels to further lower the groundwater table.

A major concern was ground loss that could incur during wellpoint installation. Since many of the wellpoints needed to be installed under hydraulic pressure, there was a risk of loss of ground during the installation. In order to minimise ground loss, driven-type wellpoints were initially installed instead of the conventional jetted wellpoints.

PERFORMANCE OF THE INITIAL PLAN

The wells were installed using a sonic drill rig (Sonic, SDC500), which provides a good quality well since it leaves no borehole smear. Fortunately, a small sonic drill rig was available to fit in the tight traffic constraints on Eglinton Avenue. The drill installed a temporary casing of 8 inches (200 mm) in diameter using a mixture of air and water as a flushing medium. After drilling the hole, a 5-inch (127mm) PVC well screen was installed with filter sand packed in the annulus between the screen and the temporary casing. The upper portion of the annular space was sealed with bentonite to prevent surface water infiltration. The well screen and filter pack was developed by air lifting until the discharged water was substantially free of sediment. Pumps were then installed in the wells.

Once pumping began, the groundwater level reacted quickly, but stabilised after about 24 hours with little more drawdown being achieved (Figure 2). The drawdown was limited by the proximity of the interface between the clay till and the sand layer above the till.

According to the plan, five driven type wellpoints were installed on either side of the bored tunnels. The driven type wellpoints (1.0m long, 50mm diameter) were made from stainless steel and were installed by pushing them into a hole pre-drilled with a jackleg drill. These wellpoints produced a limited amount of water (1 L/min) but not nearly enough to significantly lower the groundwater level. In some, the water inflow from the wellpoints stopped after only a couple hours. The ground had a difficult time to give up the water due to its low hydraulic conductivity.

FIELD TRIAL

Several field trials were performed in an attempt to overcome the situation. Different sizes of well screen were tried but there was no ideal screen size that was more effective at producing water. If the screen opening size was too large, sand particles flowed in and eventually clogged the wellpoint. If the opening size was too small, no water came in from through the wellpoint screen. It became apparent that the installation of filter sand around the well screen was an important element to prevent the plugging of the well screens in low conductivity ground. The filter sand reduced the groundwater velocity around the well screen and prevented the small soil particles from moving with the water to plug the screen.

In order to place the filter sand, a concrete core drill was used initially. The soil inside the core barrel was flushed out with a blow pipe using air and water as a flushing medium and filter sand was placed around the wellpoint. However, the filter sand came out when the core barrel was retracted and this was not successful.

Based on the field trial, it was necessary to find a way to place the filter sand under hydraulic pressure. Consideration was given to using a hollow stem auger drill and other methods, but there was difficulty in obtaining sub-contractors to provide such a service. In the meantime, more deep wells were added from surface to further lower the groundwater level as much as possible.

ADDITIONAL DEEP WELLS

Additional deep wells were not easily installed because new permits were needed and there were only limited locations where additional deep wells could be fit in between the utilities, the existing deep wells, and the bored tunnels. Some deep wells were installed at an incline in order to avoid these obstacles (Figure 3). A track-mounted sonic drill rig (Sonic, SDC550) was utilised for a one-way inclination and a tieback drill rig (Klemm, KR806) was utilised for two-way inclination to give more freedom to the well alignment. If the inclination was steeper than 15 degrees, it was difficult to place filter sand evenly around the well screen. Even with a centraliser attached to the well screen, the screen tended to pull out when the temporary drill casing was retracted.

On some wells, the PVC screens were replaced with stainless screens, which offered more opening area with the same slot size. It was hoped that, with more screen exposed to the formation, the well yields would improve, prevent clogging of the well screen and, hence, a greater effect would be realised. These modifications were intended to increase the deep wells’ effectiveness; however, the results were marginal.

Because deep well submersible pumps are not designed to cavitate, they must be cycled to prevent failure. Therefore, submersible pumps in some of the less productive wells were replaced with an eductor pumping system where the lower permeability of the ground prevented the water from freely flowing to the deep well. This system used venturis to pump water to the surface. These can create negative pressure, which effectively increased the yield of the wells and created negative pore pressure around the well. From the authors’ observations, the eductor system did not make much of difference in the flow, but the eductor system provided continuous water flow and this minimised the movement of soil particles into the wells, which consequently reduced maintenance headaches.

WELLPOINT INSTALLATION

In order to place filter sand around the wellpoints, the conventional water jetting method was applied to the installation of wellpoints in the tunnel. An aluminium pipe (3m long, 100mm diameter) was connected to a fire hose in the tunnel and jetted into the ground with erosion of the soil caused by the water flow. After submersing the pipe in the ground, the fire hose was disconnected and a PVC pipe (3m long, 50mm diameter) with a PVC well screen at the tip was immediately installed with sand poured into the remaining annulus. This installation method did not require specialised equipment and is very handy to work with in the tunnel. Crews could install an average of six wellpoints in a 10-hour shift.

Since the aluminium pipe had to be jetted through the tunnel precast segment, a core hole was drilled at each wellpoint location. Because adding core holes could change the structural integrity of the tunnel precast segment, the designer was involved and determined that it was allowable to install a maximum of two core holes (150mm diameter) per segment of a tunnel ring. In addition, these two holes must be split horizontally with 250mm offset vertically and horizontally.

There was concern that the method could create a lot of ground loss during the water jetting process. After some trials, it was recognised that the ground loss was minimal since the core hole in the tunnel segment was limited to the open area. Of course, it was important to immediately fill the annulus with well sand after jetting in the ground before the ground heaved. Approximately 10 bags (approx. 0.18 m3) of well sand was dumped into the annulus around the well screen. Ground loss during installation was almost completely compensated with the filter sand.

The layout of the wellpoints was also important to ensure their efficiency. The 3D-Finite Element Model suggested that 10 wellpoints should be located between the two deep wells at the cross passages; thus, five wellpoints on each side of the bored tunnel were recommended. The wellpoints were installed as per the designed layout. It lowered the groundwater but once again, the drawdown diminished after 24 hours. The number of wellpoints was increased between the two deep wells but with the same result.

In the end, the wellpoint system could keep the groundwater lower than the base of the cross passage excavation base by spreading the wellpoints more widely away from the cross passages and installing 14 wellpoints at either side (Figure 4). It was considered that the wide-spreading layout minimised the head difference and prevented recharge from the side and back of the tunnel. This maintained the groundwater level to be sufficiently low for a long period.

WELLPOINT INSTALLATION UNDER HYDRAULIC PRESSURE

Some of the cross passages proved to be especially difficult. Despite the addition of more deep wells from surface, the groundwater level was lowered only to the spring line of the tunnel. Consequently, many of the wellpoints had to be installed under hydraulic pressure. Two rows of wellpoints were installed to lower the groundwater level step by step (Figure 5). Since the tunnel is a round shape, these wellpoints could only be installed as high as the spring line of the tunnel. Since the upper row of the wellpoints was installed at a nearly vertical face, the wellpoints had to be inclined. This was a difficult operation where a man-lift was utilised to position personnel above the tunnel spring line.

Sometimes the ground water level was found to be higher than the spring line. As it was only possible to install the wellpoints as high as the spring line level, a few additional measures were included to prevent ground inflow during the installation and drawdown period. Following the removal of the core and annulus grout, there was a latency period before the ground began to move. This allowed enough time to install a wellpoint by water-jetting and to apply a bulkhead over the remaining opening in the segment. The process included placing bentonite chips at the top of the core hole and fastening a plywood bulkhead over the opening. This created enough of a seal to prevent ground inflow. Six wellpoints in each tunnel were installed at the spring line level, which had the effect of lowering the water level by about 1m to 2m.

With the water table now lowered below the spring line of the tunnel it was then possible to install wellpoints at a lower elevation in the tunnel. The wellpoints installed at knee level were installed with two separate wellpoints in the same core hole. One wellpoint was installed at a shallow level (1.5m deep), while the other was installed deeper (3.0m deep). The shallow wellpoints were able to quickly depressurise around the core hole area to minimise inflow. Eventually, the deeper wellpoints lowered the water enough to be effective on their own and the shorter wellpoint were turned off to prevent airlocks in the pumping system.

CROSS PASSAGE MINING

The cross passages were excavated after the dewatering system lowered the groundwater to 1m below the invert base. Two piezometers installed from surface were located outside the cross passages but they did not exactly present the groundwater table at the excavation area because the cone of depression was small due to low conductivity ground.

In order to evaluate the groundwater table properly, some drive-type piezometers were installed at the centre of the cross passage in the bored tunnels. The data was monitored carefully as the wells were installed and helped make the decision whether we need to add more wells. Figure 6 shows the drawdown history by adding deep wells and wellpoints.

In addition to dewatering, other pre-excavation supports such as splining and chemical grouting were carried out as secondary measures.

The purpose of splining was to support the roof and the purpose of chemical grouting was to stabilise the face. They prevent crumbling of the ground after dewatering. Both preexcavation supports supplemented the dewatering system.

However, dewatering left residual moisture and capillary tension between the soil pore which stabilised the low permeable ground and provided longer stand-up time.

Where the ground contained more than 25 per cent of soil finer than the #200 sieve (silt), the chemical groutability was poor. In such ground, dewatering provided greater stability without chemical grouting. The ground was easily shovelled and the crews did not need to cut the ground with clay spades. The crews appreciated the less physical work during mining.

DEWATERING MAINTENANCE

Proper maintenance of the dewatering system was another issue which required attention. The dewatering systems had to run continuously to maintain the groundwater level below the invert at all times. If the system failed or for some other reason was unable to maintain this level, the consequences could be catastrophic. Fortunately, the deep well and wellpoint systems ran as independent systems so that the consequences of interruption of the operation of one system was minimised, and there was time to get the malfunctioning system repaired, if need be.

For the deep wells, a standby generator with an auto-transfer switch was set up beside the primary generator. If the primary generator failed, the standby generator automatically started to power the deep well system.

A cellular-based alarm system was installed so that if the auto-transfer system was activated, the alarm system immediately sent a text message to personnel who could verify the system was functioning properly. This system was limited only to a power failure, so that if a pump itself malfunctioned, the alarm system would not send out a warning. This made daily monitoring of the wells very important.

The vacuum wellpoint pumps (model BBA-PT90) required 600 Volt and 7.5kW power.

The in-tunnel power supply line for tunnel lights was utilised to power the vacuum wellpoint pumps. Power was fed from Toronto Hydro main lines. However, an emergency generator was hooked up for tunnel safety and it could also power the vacuum wellpoint pumps. The vacuum wellpoint pumps were almost maintenance-free although crankshaft oil had to be replaced due to the high levels of dust in tunnel.

The most important maintenance item was to regularly adjust the wellpoints to maintain adequate vacuum in the system. As the groundwater was lowered, the wellpoints would begin to draw more air which decreased the vacuum level. In order to maintain suitable vacuum (approximately 6m of water column), the ball valves connecting each wellpoint to a common “header pipe” had to be adjusted. Clear swing hoses connecting the wellpoints to the main header pipe allowed for visual observation of the flow of water from the wellpoints. Although damage to the swing hoses and header pipe was a concern, the supervisors and the crews well understood that the dewatering systems were critical and so damage to the dewatering systems was limited during the construction.

WELL DECOMMISSIONING

All deep wells and wellpoints were decommissioned after the cross passage construction was completed. The decommissioning of the wellpoints and deep wells took place in reverse order to how they were installed. This involved first placing cement grout into the wellpoints while the deep wells were still operational. As wellpoints were decommissioned, the hydraulic pressure behind the wellpoints built up.

Thus, wellpoints were carefully decommissioned step by step to prevent the excess hydraulic pressure.

For the wellpoints installed under hydraulic pressure, polyurethane grout was injected behind the tunnel segment prior to the decommission (Figure 7). It created a chemical grout bulb to withstand the hydraulic pressure and then the core holes of the wellpoints were patched with the hydraulic cement. Finally, the wellpoints were grouted with cement.

Where dampness presented around the core holes due to the excess hydraulic pressure, Xypex Crystalline Technology was applied to seal it and this met with good success. Thereafter, the deep wells were decommissioned from surface by grouting with a bentonite/grout mixture pumped with a Tremie pipe.

LESSONS LEARNED

Planning phase

¦ A dewatering design provides the guideline to start the work but it is not everything.

¦ The team anticipated that actual ground conditions were likely to be more complicated, and boundary conditions would be complex.

¦ The other factors such as the construction method or surface restrictions can dictate the design capability.

¦ The plan needs to be adjusted by observing the field situation carefully.

Installation phase (deep well)

¦ Deep wells are the primary method to depressurise the ground prior to in-tunnel wellpoints.

¦ There is little improvement in well yield using stainless steel wire-wrapped well screen when compared to a slotted PVC screen in low permeability soil.

¦ Eductor wells do not increase well yield and pumping rate, but do reduce the well maintenance.

¦ An incline deep well can be installed up to 14 degrees from the vertical.

¦ There are many restrictions to installing deep wells from the surface and it takes longer to install wells compared with in-tunnel wellpoints.

Installation phase (wellpoint)

¦ Filter sand placed around the well screen is an important element to install functional wellpoints.

¦ The installation of wellpoints by the water jetting method does not create much ground loss when filter sand is used to fill the hole.

¦ The water jetting method is convenient in the confined space of a tunnel. ¦ Wide wellpoint spacing can be beneficial to create a wide drawdown curve.

¦ Sequential installation of shallow and deeper wellpoints are effective in lowering the groundwater level step by step.

¦ Wellpoints can safely be installed under certain excess hydraulic pressure and under certain geological conditions with extra caution.

¦ There is much less restriction to installing wellpoints in tunnel compared to surface deep wells.

Mining phase

¦ Driven piezometers are good tools to monitor the groundwater level around a tunnel.

¦ Supplemental pre-excavation measures may need to be installed, while considering the ground condition after dewatering.

¦ Dewatering of the silty soil stabilised the ground and increased the stand-up time.

Maintenance phase

¦ Maintenance is just as important as the installation itself.

¦ Deep wells and wellpoint dewatering provided two independent systems and provided redundancy.

¦ Manual daily monitoring is most important to keep deep wells and wellpoints running efficiently.

¦ Every person on site needs to understand the significance of the dewatering system so that no damages are caused