Like other technologies, those in underground construction have been developing at a rapid pace over the last 30 years. Each new development helps to solve old problems, but also introduces new risks. It takes time to learn how best to use the new technology. Procedures, organisations and systems that were adequate previously may need to be significantly changed.

The limitations of the new technology, and any unintended side effects, have to be established and considered in the planning and implementation of the work. There is, in effect, a learning curve that the industry needs to go through as each new technology is introduced. The more rapidly the technology changes, the more learning curves have to be negotiated; with attendant risks while the technology becomes established.

The first phases of the Singapore Mass Rapid Transit (MRT) system, starting in 1983, comprised 83km (figure 1) including 22km of twin tube tunnel divided almost equally between bored and cut and cover methods. The majority of the tunnelling was carried out using open face shields and compressed air. Earth Pressure Balance (EPB) shields were used on just one section of the East-West line. For the North East Line though, starting in 1997, over 90% of the tunnelling was by EPB shields. Although ground control was generally good, with low settlements, there were some problems of large localised ground loss, particularly when tunnelling in a mixed face of rock and soil resulting from weathering of the Bukit Timah granite.

Slurry application in Singapore

The use of slurry machines, where the face pressure can be maintained irrespective of the rate of advance of the shield, seemed to offer a means of eliminating the face control problems experienced in weathered rock with EPB machines. Much of the MRT Circle Line (figure 1) had to be constructed through variably weathered rock, often directly below buildings, so face control and avoidance of ground loss was a critical issue.

Risk Assessments for EPB or slurry shield tunnelling in Singapore have typically considered the risk of face instability to be negligible, provided a pressurised face machine and an experienced tunnel crew were provided. This has not proven to be the case in practice, and there have been a number of incidents of localised loss of ground. One factor has been the variation in ground conditions, which change frequently and rapidly in Singapore. In one instance, in the first phases of MRT construction, the tunnels from Raffles Place station to Bugis station, a length of less than 1.5km, encountered five distinct changes of strata. In this case, three different tunnel drive segments allowed the equipment to be suitably varied to cope with the changing conditions (figure 2).

With the use of (more expensive) pressured face machines, longer tunnel drives have been planned. However, longer drives lead to greater variation in ground conditions. The machines have not always performed equally efficiently throughout the different conditions encountered.

There is therefore a need to review the geotechnical and topographic data available, recognise its reliability, or otherwise, assess the machinery obtainable and the work force available and evaluate the risks that may arise. Finally, but as important, is the need to ensure that contingency plans are prepared for all of these eventualities, to allow speedy implementation if and when required.

To illustrate the difficulty of identifying all of the risks introduced by a new technology, a number of the modifications that proved necessary on one slurry TBM drive, are given below. These are examples only, a total of 16 modifications have been required in all.

Some slurry experience

On one section of the construction of the Circle Line the opportunity arose for twin drives of 2.6km each. Over this length it was necessary to select TBMs that would operate satisfactorily in the recent deposits of the Kallang Formation and in the weathered rocks of the sedimentary Jurong Formation and the igneous Bukit Timah granite. Tunnelling conditions in these strata are discussed in Shirlaw (2002). Generally, the tunnels were expected to lie in the Jurong formation, overlain in parts by the Kallang deposits, in the first 60% of the drive. The Bukit Timah Granite would then predominate for the remainder of the drives. To cope safely with these varied conditions, the use of slurry type TBMs was specified by the client, and the Contractor supplied two Herrenknecht slurry type TBMs. It was of course recognised that these TBMs would not work consistently in all of the conditions anticipated. However, it was felt that the use of slurry TBMs would reduce the risk of sinkholes and other disruptions previously suffered by the EPBMs, particularly in the Bukit Timah granite.

Cutting chamber pressure spikes

Almost from the start of tunnelling it became evident that some beds of the moderately and highly weathered rock of the Jurong Formation were breaking down, with excavation, into a sticky clay. There was a tendency for the clay, derived from the breakdown of weathered mudstone of the Jurong Formation, to clog up the suction entry gate area in the invert of the cutting chamber. Even with the addition of a dispersant such as Condat TFA 6 the blockages could not be entirely eliminated.

In the Herrenknecht slurry TBM, the face pressure is controlled by an air bubble in the plenum chamber, with the pressure governed by the Samson equipment. The equipment is sufficiently sensitive to control pressure fluctuations, generally to within +/-0.2 bars (20kPa). However when the clay created a blockage at the suction entry gate, there was no connection between the excavation and plenum chambers, so the Samson device was no longer controlling the pressure in the excavation chamber. The feed line delivered fresh slurry at 4 to 5 bars and at a rate of 1000m3/hour.

With the route to the suction line blocked the pressure in the excavation chamber inevitably increased very rapidly. Spikes in the pressure in the excavation chamber occurred suddenly, before the TBM operator had time to react by putting the slurry circuit into bypass mode.

For large diameter (>8m diameter) slurry TBMs it is common for the TBM supplier to provide twin pressure balancing pipes. These pipes provide an alternative means for the slurry to pass from the excavation to the plenum chamber, bypassing the blockage and ensuring that the pressure in the two chambers is equal. In smaller diameter machines the pressure balancing pipes are not normally installed, due to space constraints. This was the case for the TBMs delivered for this project. The tendency of the weathered rock of the Jurong Formation, when broken down by mechanical excavation, to clog was not recognised as a risk in the early hazard analyses.

The more severe of the sudden spikes in face pressure led to loss of slurry to the surface. Where this occurred under or near a road the resulting conditions were a risk to road users. To control this risk, two balancing pipes were installed in the first TBM, by retrofitting the machine underground (figure 3). The congestion this caused in the head was obvious and seriously interfered with maintenance. Two pipes were installed, to minimise the possibility of the pipes both becoming blocked. Despite the provision of a water flush to these balancing pipes, blockages did occasionally occur. However, overall, the pressure balancing pipes greatly reduced the number of pressure spikes.

For the second TBM, a second line of defence was instituted in the form of a 50mm pipe from the excavation chamber to a relief valve which would, when it opened under a predetermined excess pressure level, bleed slurry into the invert of the tunnel. This would both relieve the pressure and give a quick visible warning to the TBM operator. With this in place it became possible, to install only one balancing pipe in the second TBM, and preserve access to the other equipment in the plenum chamber, most notably the rock crusher.

With these modifications in place the tunnelling continued more consistently and with significantly less disruption than previously. However the provision of these measures in tunnels under construction was obviously a much greater challenge than had they been incorporated during the original manufacture.


Even with the provision of balancing pipes to avoid the pressure spikes when a blockage occurred, the problem of clearing the blockage remained. The traditional method of running the slurry circuit in bypass, and attempting to remove the obstruction by suction only was not always effective. Turning the head without advancing the tunnel, to agitate the slurry, carried with it the risk of over excavation and sinkholes.

To improve the agitation of the slurry and break-up blockages, it was decided to extend the agitator paddles into the suction zone, and double their number from two to four. Two of the slurry supply nozzles were also extended down towards the invert to provide jets of slurry into the area where blockages were of greatest concern.

These jets of fresh slurry agitated and diluted the material in the suction gate area, and helped maintain the excavated material in suspension, aiding the flow into the suction pipe.

Operating parameters

At least as important as the hardware is the need to have all the operating parameters accurately defined and detailed procedures prepared so that all employees are aware of their responsibilities and the actions to be taken in case of unforeseen circumstances.

In compressed air tunnelling, with open face shields, the face pressure was typically adjusted on the instructions of the shift engineer or the leading miner, based on observations of the visible water in the face. This observational approach was not effective where there were sudden changes in ground conditions, requiring sudden changes in the compressed air pressure. A number of spectacular losses of ground, often inundating the tunnel, have occurred due to the failure to adjust air pressures in advance of such interfaces.

Initially, it was common for contractors using pressurised shield machines to use a similarly devolved approach to assessing face pressures, with the shield operator making the decision as to how to adjust the pressure. This approach has not worked in Singapore’s rapidly varying ground conditions. Shirlaw[1] records a number of sinkholes that occurred due to failing to plan face pressures to allow for known interfaces.

As recorded in Shirlaw and Boone[2], there were 16 incidents of very large settlement or sinkholes during the EPB tunnelling for the construction of the North East Line, and at least 21 incidents during the construction of the deep sewer system in Singapore. In order to help reduce the number of such incidents, it has been found that it is necessary to:

a) Calculate target face pressure values for every few rings of the tunnel, before starting tunnelling

b) Have enough borehole and piezometric information along the tunnel alignment to have a reasonable basis for the ground and ground water conditions used in the analysis

c) Define levels of authority for changing the face pressure from that calculated, based on observations in the tunnel; in particular the shield operator should not be authorised to drop below the target pressure without confirmation from the tunnel manager

d) Carry out rigorous checking of the net material excavated, either continuously or every few hundred millimetres of advance, to guard against over excavation

Implementing these measures will help to minimise the risk of a local loss of ground.

Where groundwater levels are close to the ground surface, as is the case in Singapore, there is a very small margin between the face pressure required to support the face and the pressure that will expel slurry or foam (for EPBs) to the surface up an open path. It has been found that such open paths are reasonably frequent in urban areas.

The most likely routes are the boreholes and instrumentation installed for the project. However old boreholes, instrumentation, wells and areas where temporary works has been extracted can also be present. There have been cases in Singapore of the loss of up to 100m³ of slurry to the surface, and in urban areas this can cause almost as much of a problem as a major loss of ground. Open paths should be grouted in advance of tunnelling, where they have been identified.

However, in urban areas it is unlikely that all such paths can be identified prior to tunnelling. It is therefore necessary to:

• Carry out the face pressure calculations as accurately as possible, and without unnecessary conservatism

• Minimise the pressure fluctuations during tunnelling

The measures discussed above are essential to reduce the risk of a major loss of ground or loss of slurry/foam to the surface. Even with the implementation of such measures, in the highly heterogeneous soils of Singapore there remains a residual risk of loss of ground or loss of slurry/foam, and contingency measures for such losses still need to be planned for.

It has been found advantageous, in particularly sensitive areas, for someone to be posted on the surface, above the tunnel, to identify any signs of a loss of ground or slurry/foam. Simple measures, such as warning signs for a loss of ground or bunds, for loss of slurry, can then be implemented immediately.

Regular look ahead

The need for calculated target pressures for pressurised TBM tunnelling has been identified. It is important that such calculations are not just a paper exercise; the results of the calculations need to be communicated to all of the key personnel in the tunnel, including the shield operators and shift engineers. In addition to the target face pressures they need to be aware of anticipated changes in the geology, and what is above the tunnel. A convenient way to communicate this information is to produce a weekly ‘look-ahead’ sheet. On this sheet are summarised, for every few rings:

• The anticipated ground conditions

• The target face pressure

• Buildings, roads, railway tracks and utilities above the tunnel

• Any geological hazards, such as faults

• Other hazards, such as sudden changes in ground level due to retaining walls

• Borehole and deep subsurface instrumentation locations

The ‘look ahead’ is prepared weekly, with about twice the expected week’s production covered on a single sheet of paper. The sheet is intended to be a simple means of communication of key aspects of the planned tunnelling to those actually carrying out the work.


The Code of Practice for tunnelling rightly requires risk assessments to be carried out prior to tunnelling. However, it has been found in Singapore that it is common for contractors to list, simply, an appropriate machine from a major manufacturer and an experienced tunnelling team as adequate risk control measures for pressurised tunnelling in an urban environment through highly varied ground conditions.

A number of actual problems, and implemented solutions for slurry shield (primarily) and EPB tunnelling in Singapore have been outlined above. The purpose of giving these examples is to show that trouble free tunnelling is unlikely to happen even with an experienced crew and a specially selected new machine. The problems, and solutions, are also documented as an aid in the identification of risks and solutions on other tunnelling projects.

Preferably the selection of the TBM should be removed from the competitive tendering process. If not, suppliers will always be pressurised to supply a basic machine whereas it is essential that the experienced manufacturer should be encouraged to incorporate the lessons learnt from its earlier machines.

The Contractor must be encouraged to carry out a risk assessment such that the wide experience available in the industry is utilised to make the exercise as comprehensive as possible.

Since no risk assessment can ever claim to have identified every possible hazard then comprehensive contingency plans must be in place and practiced so they can be rapidly implemented if the need arises. T&T

Slurry escape to the surface Slurry escape to the surface Balancing pipes in the TBM – for flushing provision Balancing pipes – 1 Balancing pipes in the TBM – for flushing provision Balancing pipes – 2 A sinkhole above one of Singapore’s deep sewer tunnels A sinkhole above one of Singapore’s deep sewer tunnels Figure 2. Geology between Raffles Place and Bugis Station Fig 2 – Geology between Raffles Place and Bugis Station Figure 3. The balancing pipes Fig 3 – The balancing pipes Figure 1. The 2007 MRT system superimposed on Singapore’s complex geology Fig 1 – The 2007 MRT system superimposed on Singapore’s complex geology