The Land Transport Authority (LTA) of Singapore has been procuring the expansion of their underground rail network at a prodigious rate over the last twenty years. EPBM tunnelling has been extensively used in Singapore for the construction of the North East Line, Changi Airport Line (East West Line extension), Circle Line and on the ongoing Downtown Lines One and Three.

Tunnelling using slurry shields was first introduced in Singapore (except for microtunnelling) for the construction of parts of the Circle Line and Downtown Line Two to cope with particularly challenging mixed ground conditions of the Bukit Timah Granite formation. Slurry shields will also be used on some sections of the future Thomson Line and on the recently awarded Singapore Power cable tunnels.

Downtown Line Two Contract C916 for the construction of Beauty World station and the twin 1km bored tunnels between Beauty World and King Albert Park stations was awarded to McConnell Dowell SEA. The tunnel alignment is located entirely within the Bukit Timah Granite formation. The first TBM installation began in June 2011 and both tunnels were completed in May 2012. Two Mixshield type slurry TBMs were selected, where a pressurised air bubble acts on the slurry behind a submerged wall and the air pressure is transferred to the slurry improving the face pressure regulation.

Description of Bukit Timah Granite

The Bukit Timah Granite is an early to middle Triassic igneous strata found in central and northern areas of Singapore. It comprises a number of acidic rocks, predominantly granite, micro-granite and granodiorite, it has a very wide range of strength, from less than 1MPa to in excess of 300MPa, with the fresh granite causing major wear as it falls into the ‘extremely abrasive’ range. This strength range is a consequence of the severe tropical weathering, creating six weathering grades, from soil to fresh rock (named GVI through to GI), which can occur within a very short distance of each other. Consequently the problem of supporting a mixed face exists, particularly in the intermediate weathering grades where intact rock is encountered in the invert and corestones within a loose sandy or clayey matrix in the crown. This situation is further exacerbated by the high permeabilities in the intermediate weathering grades, increasing the instability by forcing the ground to ‘flow’ into the TBM [2].

Function of slurry in TBM tunnelling

Slurry has several functions, including:

  • To transmit pressure and support the face
  • To form a cake and prevent slurry loss into the ground
  • To support spoil in a suspension in the fluid for effective pumped transport
  • To lubricate the spoil transport system and reduce abrasion
  • To encapsulate soil particles and prevent hydration of minerals or dispersion of fines

Tunnelling slurry used on C916 was conventional water-based bentonite slurry, using various additives as necessary to modify its rheological properties to achieve these functions.

Rheology can be defined as the science of flow and deformation of matter. It studies the reactions of all solid or liquid substances to deforming forces and classifies them using precise terms.

It thus serves to accurately define properties that we refer to with vague terms such as ‘thick’, ‘thin’, ‘sticky’, ‘viscous’ [4] .

The importance of the five listed functions above varies in differing ground conditions, and some can become primary properties or can be considered of secondary importance. The desired properties (known as key performance indicators, or KPIs) for the anticipated ground condition, were determined in advance of tunnelling, based initially on the foreseen geology, but then refined further during the tunnelling works.

Tunnelling in rock generally requires slurry with properties related mostly to its secondary functions (lubrication, transport of muck).

Tunnelling in soil would require slurry with properties related to both primary (face support) and secondary functions. Tunnelling in mixed ground conditions requires slurry with specific properties, which are not simply the one we would apply to the worst soil conditions encountered at the face.

As the slurry is constantly circulated through the TBM and back to a separation or slurry treatment plant (STP) at the surface, the properties continually change as the slurry is exposed to the ground and groundwater at the face and as it is processed at the STP.

Therefore it is essential to test these properties frequently, at least one time before each ring advances in order to verify that they still fall within the desired KPI range.

Slurry treatment plant

McConnell Dowell used two STPs, one for each TBM, with some components shared between both plants.

Each STP was designed to handle slurry flows of up to 1,400m3/hr, the total equipment included:

  • Two primary declined deck shaker units, each with wedge wire screens,
  • Six desanders, each with a ‘660’ primary hydrocyclone,
  • 4.8m long dewatering shaker deck, 18 no. 5in long bodied desanding and desilting hydrocyclones
  • Dilute electrolyte mixing system, 125m3 capacity dilute electrolyte storage tank
  • Dilute flocculant mixing system, 125m3 capacity dilute flocculant storage tank
  • One S6-1-G, two S5-1-G and one S4-1-G (KHD) centrifuge
  • Two bentonite mixing systems, each with a Venturi jet mixer, 20m3 capacity bentonite mixing tank with twin vertical shaft agitators, two 125m3 fresh bentonite storage tanks, four 250m3 cleaned slurry storage tanks with large diameter inter-connecting pipework.

Key slurry properties

Testing of these properties is fully described in various standards [3].

The tests produce a numerical value, but in most cases the importance is to track the trend of change between subsequent tests and the observed behavior on the tunnelling process rather than focus on the actual value and its units.

Marsh funnel viscosity

The Marsh funnel viscosity (MFV) reports the number of seconds required for a quantity of fluid to flow through the Marsh funnel. MFV also gives an indication of other properties.

Density

Measurement of density indicates the fines contents of the slurry being recycled to the TBM after separation.

In fact the density depends on the slurry components, on physical contamination by fines from the ground, which are not separated and returned back to the TBM and on regeneration by fresh bentonite.

pH

A change to the pH of the slurry affects its properties due to changes in ionic attractions. Beyond a pH range from 7.5 to 9.5, there is likely to be poor slurry performance.

Fluid Loss

Is determined with a filter press by applying a gas pressure onto a slurry sample in a pressure vessel that has a permeable filter paper base. Over a fixed time period, a volume of filtrate is collected and this is measured. It is a measure of the capability of the slurry to form an impermeable cake against permeable formations, which is vital to allow the face pressure to exert a supporting force on the face. A high value indicates a risk of the drill fluid migrations through micro fractures and porous formations.

Filter cake

From the previous test the thickness and quality of the filter cake can be observed. Ideally a filter cake should be thin and flexible as this indicates that the cake has a low permeability.

A thick cake indicates that filtrate has continued to pass through the cake encouraging deposition of further material to build the cake thickness.

In the TBM excavation chamber, cake is continually scraped off the face and destroyed with each sweep of a cutting tool, but must reform it immediately after the tool has passed.

Sand content

The sand content of the slurry is an indicator of the performance of the separation plant. Any sand remaining in the slurry directly affects the permeability of the cake and its stability on the ground interface.

Plastic viscosity (PV)

PV is determined using a rheometer – a device that spins a vane (referred to sometimes as a weighted bob or bobbin) in a cup containing slurry, it measures torque transfer through the slurry onto the cup, it indicates viscosity of the fluid under dynamic fiow conditions.

This is a particular property that is best considered on the basis of tracking changes in its value in subsequent tests.

Yield point (YP)

YP and PV are similarly determined with rheometer, it is an interpolated value that indicates the stress or pressure required to start fluid moving, i.e., a resistance to initial fiow. Therefore yield is an important characteristic in supporting the face.
Again this property is best considered on the basis of tracking changes in its value in subsequent tests.

0-10 minutes gel strength

Again this is determined with the rheometer. It compares the shear strength of the slurry when freshly stirred, and then 10 minutes afterward.

Key performance indicators

Key performance indicators for slurry performance were predetermined prior to tunnelling. These were the defined ranges of the above properties in the various ground conditions to be encountered.

Initially the geological conditions were predicted at 20m intervals and one of the KPI ranges allocated to each range. These predictions and allocations were based on, among other things, the contract site investigation, additional post contract award site investigation, conditions encountered during construction on this and neighbouring construction site, anecdotal accounts of local geological properties, circumstances encountered elsewhere in Singapore and globally and their effect on tunnelling and STP performance.

The range of properties may seem wide, but these were general ranges.

These were the initially selected KPIs, but significant refinements were carried out during tunnel excavation. With the key focus on maintaining fluid loss and YP/gel strength, the other KPIs then tend to fall into place.

Management of KPIs during tunnelling

The STP technician was required to carry out the full suite of tests each advance, record the results and compare with the KPIs (updated to meet encountered geology and foreseen geology) If the KPIs were met then advance could continue, otherwise the STP operator would be required to improve the properties to meet the KPIs.

In certain circumstances, e.g. in borderline cases where the result was within the accuracy of the test or where the unattained property was a secondary or tertiary property, or where actual conditions (perhaps observed in the muck pit) could allow a different KPI to be applied, this decision was in the hands of the senior engineer on duty, who could call on the mud engineer or tunnel manager for support and guidance.

If a slurry property outside the planned KPI range was accepted for use, this was documented, with reasons.

Typical challenges of tunnelling in mixed ground

Unstable face and ground loss during advance/ uncertainty about actual ground conditions being encountered

It became apparent that transitions between the weathering grades could be more abrupt than originally predicted. To combat this, the use of the GG class was discontinued, and an increased minimum yield strength limit was applied in the remaining KPIs.

Unstable face during compressed air intervention

During cutter head intervention (for cutter changes or other reasons), the slurry level in the excavation chamber is drawn down and simultaneously replaced with compressed air at the same pressure. If there is no filter cake then air loss into the ground and face instability can occur.

Therefore particular care is required to ensure that the properties of slurry are suitable, and it is necessary to be pessimistic when predicting the ground that will be encountered during the intervention (and also along the TBM annulus between the face and last grouted ring). At C916 an intervention mix was designed and this was delivered into the excavation chamber before the intervention, however the delivery of this mix into the excavation chamber and retaining it there can be a challenge.

Of equal importance is the quality of slurry injected into the excavation chamber after completing the intervention, especially considering that slurry may have been stationary in the TBM and tunnel pipework for several days.

Solids contaminations: handling of fines

When encountered soils have higher than foreseen fine particles (less than 63 m) these may exceed the STP limits on the separation of clay particles and silt particles. Therefore slurry accumulates high presence of fines, the density rises and this causes the performance of hydrocyclones on the STP to deteriorate, exacerbating the problem. Therefore it is necessary to take action such as:

  • Stop advance while processing continues
  • Dilute the slurry with water (not a good solution as it reduces the other rheological properties)
  • Dilute the slurry by adding fresh bentonite slurry.

Solids contaminations: sand

The sand content of slurry depends upon the performance of the separation plant. Any sand remaining in the slurry directly affects the permeability of the filter cake and its stability on the tunnel/ground interface, and therefore the sand content should be kept as low as possible.

Solids contaminations: Cement

During TBM operation slurry may become contaminated by the backfill grout which can ow into the slurry, cement present along the drive in areas which soil has been treated for ground improvement or even from the leading TBM (if the tunnel alignments have relatively small separation in permeable or fractured ground) Contamination by cement is recognisable by the changes in the following slurry properties:

  • Increase in pH
  • Increase in filtrate loss to very high values (> 50cc)
  • Increase in filter cake thickness to several millimeters (5mm or more)

Recovery can be done with the addition of additives in the slurry in order to recover its properties to an acceptable level, although if too severe it may be necessary to dispose of/replace it with fresh one. As a result of the fine soil and cement contamination encountered during tunnelling on C916, maintaining a low fluid loss (in the range of 20-30) was one of the greatest challenges and required constant attention by the supervision team and the use of additives and bentonite.

Conclusion

McConnell Dowell commenced installation of the first TBM on C916 in May 2011 and both TBMs completed tunnelling in May 2012. The best day’s production for one TBM was 18.9m, and its best month’s production was 259m. Reliability and processing capacity of the STPs was generally satisfactory.