The Land Transport Authority (LTA) of Singapore is responsible for the implementation of the Mass Rapid Transit (MRT) System on the island. The rail network is currently being expanded with the Circle Line (CCL) Project, which when complete will add 29 underground stations and 34km to the existing rail transit network.

Circle Line Contract C855 comprises four stations and 5.2km of tunnels, mostly constructed by cut & cover or bored tunnel. However, for a 70m tunnel section under a road called Ayer Rajah Avenue (AYA), connecting a cut & cover box and a station, there are major utilities along the road corridor above the tunnel route that cannot not be diverted. It was decided to construct these AYA Tunnels using the Sprayed Concrete Lining (SCL) method.

The twin tunnels are part of the Contract 855, awarded to JV contractor WSA, formed by Singaporean company Woh Hup, Austria’s Alpine Mayreder, and Chinese company STEC. The consultant appointed by the JV for the design of the SCL temporary works is Amberg & TTI, part of the Swiss-based Amberg Group.

The 70m long tunnels were divided into nine construction stages (figure 1). The widths of the tunnels are 11.6m for inner tunnel module 1, 10m width for inner tunnel module 2 and about 9m for the remaining modules.

The selection of SCL was a direct result of the need to minimise ground movements and the risk of settlement damage to the major utilities above the tunnels.

One peculiarity of the AYA tunnels is the narrow pillar of soil between the two tunnels ranging from 1m to 4m accompanied by a shallow overburden in relation to the tunnel diameter (a cover of approximately 1 diameter for inner tunnel module 1).

Geological conditions

The AYA tunnel is situated in Tengah Facies of the Jurong Formation, belonging to Upper Triassic to Early Jurassic (100M to 200M years ago) consisting of muddy marine sandstone with occasional grit beds and conglomerate.

A fill layer between 0.7m and 7m thick overlies the Jurong formation. The presence of this backfill was confirmed in most of the boreholes.

The sedimentary rocks of the Jurong Formation, consisting of shale, siltstone, sandstone and limestone of various weathering grades are encountered beneath the fill layer. The rocks are also found to be frequently interbedded.

Shotcrete lining

Generally, tunnel supports consist of a shotcrete shell 300mm thick with a double layer of T8-100 mesh, with lattice girders every 1m. A pipe arch consisting of a series of 114mm diameter pipes was installed prior to the excavation of the top heading. Horizontal H beams were designed to be installed across the footing of Stage 1A to provide a lateral restraint to excavation.

For the first module in the inner tunnel, the excavation face is 92m2 with a maximum horizontal span of some 21m. Based on the designer’s experience in similar ground conditions, the tunnel face was divided into four stages, 1A, 1B, 2A and 2B (figure 2) using the side wall drift method to provide better control over ground deformation. In excavation stage 1A, “elephant’s feet” are provided at the invert on each side of the excavation to reduce the tunnel settlement. H-beams were proposed spanning the width of the 1A excavation to provide lateral restraint against the active lateral pressure.

In the design, finite element analysis was performed to simulate the sequential construction of the tunnels and the soil-structure interaction between the ground and the SCL lining. The envelope of the forces acting on the tunnels in all construction stages was used to compute the required thickness of the shotcrete lining wall. The constitutive model for the various soil types is based on the elastic – perfectly plastic Mohr-Coulomb failure criteria.

In the computation, a relaxation factor of 10% was considered in assessing the ground deformation during the excavation stage. This 10% factor has accounted for the presence of the pipe arch roof, which provides a pre-confining effect to stabilise the soil at the crown prior to excavation, as well as the type of soil material that was expected to be excavated. The shotcrete lining and elephant footings were simulated as beam elements. The horizontal beam spanning between the 1A and 1B face was modelled as a node-to-node anchor. Based on the geometry of the section especially the joints at the crown and invert (in 1A and 1B), high bending moments were expected due to the drastic change in curvature. Due to the high bending moment, plastification and cracking of the shotcrete lining would likely occur. To simulate the behaviour of this plastification, a limiting moment capacity was modelled at the joint (a partial plastic hinge) to allow for a re-distribution of the moment from the joint to the shotcrete lining wall. This was critical in carrying out the design adopting the side drift approach. The final closure of the ring, by removing the side drift and excavating 2A/2B in a staggered fashion, resulted in a more stable, rounded geometry and the forces in the lining were mostly hoop stress with a small magnitude of bending moments and shear forces.

Finite Element Model for the TBI

For the typical module in the inner (module 3 & 4) and outer tunnels (module 5 to 9), the excavation face is about 55m2 to 70m2 spanning a maximum width of about 10m. As the geometry is relatively smaller both in size and width, the excavation face is broken into just three stages, top heading, temporary invert followed by bench and invert excavation. The temporary invert is necessary due to the presence of residual soil with low SPT N-values (figure 3).

Similar to the side drift approach, the shotcrete lining was simulated in the finite element program as a beam element which includes the elephant footings on both sides of the excavation. A relaxation factor of 10% was again considered in the analysis during the excavation to assess deformation.

As the profile of the excavation generally consists of gentle curvatures, except near the footings of the temporary invert, the bending moments and shear forces were quite small. The predominant force was the hoop force acting within the tunnel lining. The footings were modelled as a series of beam elements of varying thickness to form the transition from the normal section to the enlarged section near the invert. The beam with the larger stiffness naturally attracted higher forces and the lining thickness for the transition was designed accordingly.

In the design for the face stability, a circular tunnel with equivalent diameter and a sliding wedge in front of the tunnel face were assumed. Vertical loading adopting Terzaghi’s silo theory was applied on the top of the sliding wedge. The driving force was derived from the overburden soil weight and weight of the wedge. Resisting forces are contributed by frictional and by shear developed both at the inclined slip surface and side areas. GRP anchors (12m) were used at the face to increase resisting forces.


Presently the Inner Bound tunnel has been completed and permanent lining works are ongoing, while 13m of the Outer Bound has been completed.

The construction of the Inner Bound tunnel started on December 2006, but the construction was almost immediately interrupted by a ban on sand importation from Indonesia to Singapore. After a few heading excavations of 1A, the shotcrete strength was found to be low and did not achieve the required 12hr strength.

Investigations were carried out that found the sand quality to be the primary cause. The quality of available sand from other sources after the Indonesian ban was irregular and mainly too coarse, which resulted in a lower strength shotcrete. During this period, extensive analysis and trials were carried out with the available sand to achieve the required shotcrete strength but was still found to be inconsistent. The situation was only cleared up at the beginning of March 2007 when alternative sources of quality sand became available in Singapore.

The first Tunnel Section to be excavated (A1) was the largest of the entire tunnel. Due to the vicinity of the Outer Bound tunnel, a typical conical shape for the pipe arch umbrella would have further reduced the soil pillar width. For this reason the Section A1 was designed with a constant geometry, taking advantages of the dimension reduction in the next section to have the clearance for executing the next umbrella.

The first challenge for construction was the execution of the pipe arch umbrella that, for a 21m section A1, had to be a minimum of 27m long with a minimum overlap of 6m to the next pipe arch umbrella. In order to verify the alignment of the pipe arch, a survey was executed after each pipe installation, measuring each pipe deflection. Based on this execution/monitoring process, continuous adjustments were done to the drilling methodology in order to achieve the desired result. Grouting was executed with double packer.

The results on the pipe roof alignment were better than expected (considering that the execution was undertaken with a normal two boom jumbo), as summarised in Table 1.

The spacing of 300mm was therefore achieved for most of the pipes. This was critical as it was designed as support for the overburden over the excavation length.

Another challenge has been the construction of the joint between the temporary drift and the final section profile (between 1A/2A and between 1B/2B).

The joint is heavily reinforced due to the high bending moments created during the side drift stage, and due to the need to install lapping bars to assure continuation of the reinforcement within the final profile, creating difficulties for a good application of shotcrete and also the potential for shadows.

In order to assure high quality work, a mock up of the joint was used for training the shotcrete applicators and a second one was used to verify the shotcreting quality.

The construction using the side drift created some difficulties in term of machinery accessibility, especially in the section 1B where mining had to been done below the horizontal H beams.

For the excavation of the first drift 1B, accessibility was further restricted by the pre-constructed permanent base slab for the cut & cover tunnel, causing a restriction in the headroom between the slab level and the installed horizontal H beams. The problem was resolved by replacing the H beam with an “A” frame to increase the headroom for machinery access by around 1m.

During the excavation of the inner tunnel module 2, accessibility for excavation for 1B was also found to be difficult due to the horizontal H-beam constraints. An alternative sequence has been adopted – excavation of the side drift (1A), followed by enlargement (2A), followed by a close excavation of the bench and invert to achieve a faster ring closure. The problem of head room was eliminated as the H-beam could be progressively removed and excavation can be done at the bench and invert.

During the 1st 3m of excavation of the bench and invert, the bench and invert faces experienced localised face instability below the elephant’s foot at the side drift and an ongoing settlement trend was observed.

Micropiles at the partially completed dome and along the temporary side drift were installed to transfer the load below the excavation level and reduce instability problems of bench/invert faces. This had immediate benefit in reducing ongoing settlement trends.

For the initial 10m of the inner tunnel, as the siltstone level was higher than the tunnel invert, the design of the tunnel profile was modified to socket the sprayed shotcrete lining into the siltstone. In simulating this profile, the socketing end was modelled as a pin joint with a small footing, which provides the bearing capacity.

For the design to work well on site, adequate quality control in ensuring that the rock layer wads sound and not heavily jointed was implemented. This ensured a good base for the loading to be transferred directly to the confined rock layer. There was little tunnel convergence monitored for this modified approach.

One safety benefit of this approach was that it removed the need for workers to excavate out the solid rock in the tunnel invert to achieve the original full ring closure. This would have resulted in significant vibration of the fresh shotcrete sealing layer above, causing a risk of falling shotcrete.

In the 3rd and 4th modules the top heading bench invert method was adopted with benefits on workability and productivity. The slowest cycle time for a 1m round has been 36hrs for the drift 1B during section A1 (within a mixed face of rock and soil) whilst the fastest has been 8hr for a top heading round in the Section A4.

Generally during excavation, problems of face instability have occurred mainly due to the unfavourable orientation and dips of narrow spaced strata and discontinuities, and the occurrence of distinct slip planes.

Soil nailing of the face was included in the design for the drifts 2A, for the top heading excavation and at the end of each tunnel section where a full face is formed. During construction soil nailing of the face has been extensively used also at all stages (1A, 1B, bench invert). The face nailing also helped as a confinement action in front of the face in reducing pre-excavation deformation. As a result, both in-tunnel deformation as well as surface settlement has been within the predicted design values, although a few settlement markers on surface exceeded the designed values that required a verification/assessment process by the designer (see Table 2).

Pillar stability

As mentioned, the pillar of soil between the twlo tunnels ranged between 1m and 4m. The stability of the pillar was a concern when excavating the second tunnel (outer tunnel).

LTA routinely appointed a board of international renowned consultants to review construction risk. The Board of Advisors advised that the permanent lining in the inner tunnel must be cast to increase the tunnel stiffness to improve the stability of this pillar before outer tunnel excavation could start.

The outer tunnel where the pillar is narrowest is now completed. The measures implemented comprise the casting of the side walls and base slab in the inner tunnel, the stitching of the two tunnels with tie rods, continuous face nailing ahead of the face and controlled excavation to minimise over excavation at the pillar side.


Both the side drift and top heading, bench, and invert (TBI) methods have shown to be successful in controlling the ground movements and avoiding excessive settlement, which could damage the existing utilities or public road.

he side drift method had been adopted for the first two larger modules of the inner tunnels, where a full top heading face would have been too large. The side-drift method presented some problems with workability, machinery access, space constraints, but with this system the inner tunnel module 1 (11.6m width tunnel with a cover of approximately 1 diameter) were successfully excavated with minimal deformations.

The TBI method used in the remaining portion of the inner bound tunnel has provided more working space and hence faster cycle time.

With the side drift method most of the deformation occurred during the temporary drifts excavations (1A/1B) and stabilised only once the full circular section was completed. Therefore the deformations on 1A/1B are sort of time related which might increase until the final stage of ring closure.

The top heading/bench/invert method resulted in generally lower deformation not only due to the reduced tunnel cross section compared to that of the side drift method, but also the formation of the temporary invert which immediately acted as a ring closure.

Micropiles have been shown to be an effective and easy method to increase the bearings of the ground below the elephant’s feet and limit settlement.

Face nailing has also shown to be effective in order to maintain the stability of the face and to reduce the correlated hazards to an acceptable level.

No damage to the utilities has occurred as settlement of the utilities was within the designed Slope values.

The pillar stability was ensured via making the first tunnel stiffer through casting the base slab and side walls, continuous face nailing of the outer tunnel, pillar stitching between the two tunnels and controlled excavation of the pillar side to avoid over-excavation.

Finite element model of 1A/1B excavation (side drift approach) Finite element model of 1A/1B excavation (side drift approach) Finite element model of the top heading excavation (TBI approach) Finite element model of the top heading excavation (TBI approach) Improved accessibility of machinery made possible by the A-frame installation Improved accessibility – 1 Improved accessibility of machinery made possible by the A-frame installation Improved accessibility – 2 Mock up of the junction of the side drift Mock up of the junction of the side drift Figure 1. Construction modules for the 70m long twin tube AYA Tunnels Fig 1 – Construction modules for the 70m long twin tube AYA Tunnels Figure 2. Partitioned face for initial excavation through module 1 Fig 2 – Partitioned face for initial excavation through module 1 Figure 3. Sequence of the top heading, bench and invert approach Fig 3 – Sequence of the top heading, bench and invert approach Table 1 Table 1 Table 2 Table 2