RETC2007.

bk Page 976 Thursday, April 19, 2007 1:48 PM

MONITORING SUCCESSFUL NATM IN SINGAPORE

K. Zeidler
Gall Zeidler Consultants, LLC

T. Schwind
Gall Zeidler Consultants, LLC

ABSTRACT
The Fort Canning Tunnel, Singapore, is an approximately 50 feet (15m) wide
vehicular tunnel under shallow ground cover that has been constructed according to
the principles of the NATM. Its shallow location in soft ground combined with a high
groundwater elevation in close vicinity of historically important features called for spe-
cial design considerations combined with a rigorous monitoring scheme. Surface, sub-
surface and in-tunnel instrumentation was installed to monitor the performance of the
ground and tunnel support during construction. Comparison of actual reading data
with the results from the computer modeling carried out during the design phase is
provided in this paper.

INTRODUCTION
The Fort Canning Tunnel forms a part of the project “Contract PE101A—Design
and Construction of Fort Canning Tunnel and Realignment of Stamford Road,” spon-
sored by the Land Transport Authority (LTA) of Singapore. At Fort Canning Tunnel
(FCT) the New Austrian Tunneling Method (NATM) was used for the first time for a
large span (14.7m–48ft) tunnel in Singapore. The contract has been tendered in a
Design–Build framework and the team with the members Sato Kogyo (S) Ltd. (Con-
tractor), TY Lin Ltd. (Engineer) and Gall Zeidler Consultants (NATM designer) has
been awarded the contract. Fort Canning Tunnel is a 180m (590ft) long, three-lane
highway tunnel, and has a cross section area of 135m2 (1440ft2). It was constructed in
residual soils under an overburden between 3m and 9m (10ft and 30ft). The tunnel
approaches at the north and the south portal of Fort Canning Tunnel are constructed
using cut-and-cover techniques.

LOCAL SETTINGS

Surrounding Structures
Whereas at the northern end of the tunnel no buildings are located in close vicinity
to the tunnel structure, a retaining wall for the newly constructed Singapore History
Museum is positioned in immediate proximity of the tunnel next to the south portal.
The alignment leads underneath the historic Fort Canning Park with its preserved
trees and a historic cemetery. Fort Canning Rise, a public road crosses the tunnel
alignment at a vertical clearance of approximately 5 m between the tunnel roof and the
road surface.
In vicinity of the south portal, a settlement sensitive Church building is located.

976
RETC2007.bk Page 977 Thursday, April 19, 2007 1:48 PM

MONITORING SUCCESSFUL NATM IN SINGAPORE 977

Figure 1. Original schematic geological profile (from initial site investigation report by Kiso-Jiban
Consultants Co. Ltd., 2003)

Figure 2. Updated schematic geological profile (Kiso-Jiban Consultants Co. Ltd., 2004)

Geology
The Fort Canning Tunnel is constructed in the residual soils of the Fort Canning
Boulder Beds. The Fort Canning Boulder Bed is a colluvial deposit of Pleistocene age
that underlies parts of the central business and commercial district of Singapore. It
consists of sandstone boulders in a matrix of hard sandy silt or sandy clay with silt. The
matrix is of deep-red, red and white or mottled red, yellow and white color. For classifi-
cation purposes the residual soil layer was subdivided based on the SPT N-value; RS I
(N<15), RS II (15<N<30), RS III (30<N<50) and RS IV (N>50). The residual soils of
Fort Canning Boulder Bed are overlain by a man-made fill layer of varying thickness
(1m to 5m) and are underlain by the bedrock of the Jurong Formation, a sedimentary
rock (Sandstone).
Ground water level at the Fort Canning Hill is established at approximately 1m
below surface level, i.e., 2m to 8m above tunnel crown level.
The deposits were found in non-continuous layers and lenses with a varying
amount of boulders leading to a inhomogeneous excavation medium with a wide vari-
ety of ground properties. Observations in the tunnel face revealed that the ratio
between matrix vs. boulders varied between 70% boulders vs. 30% matrix and 30%
boulders vs. 70% matrix (Figures 1 and 2).

DESIGN AND CONSTRUCTION

General
The project was tendered based on a Design–Build contract format. The NATM
mined construction approach was the Contractor’s proposed alternative. The design
RETC2007.bk Page 978 Thursday, April 19, 2007 1:48 PM

978 2007 RETC PROCEEDINGS

Table 1. Soil parameters

Soil Parameters
γ E φ' C' Cu
Layer (kN/m3) (MPa) (deg) (MPa) (MPa) n'
Fill Original 19/ 10/ 30/ 0.005 0.040/ 0.35/
1st Variation 19/ 10/ 30/ /0.001/ N/A/ 0.35/
2nd Variation 19 10 30 0.001 N/A 0.2
Residual Soil (RS I) Original —/ —/ —/ —/ —/ —/
1st Variation 20/ 15/ 30/ 0.010/ 0.045/ 0.35/
2nd Variation 20 15 30 0.010 0.045 0.2
Residual Soil (RS II) Original 20/ 60/ 30/ 0.015/ 0.180/ 0.35/
1st Variation 20/ 30/ 30/ 0.010/ 0.090/ 0.35/
2nd Variation 20 30 30 0.010 0.090 0.2
Residual Soil (RS III) Original —/ —/ —/ —/ —/ —/
1st Variation 20/ 60/ 30/ 0.020/ 0.180/ 0.35/
2nd Variation 20 60 30 0.020 0.180 0.2
Residual Soil (RS IV) Original 20/ 100/ 30/ 0.035/ 0.300/ 0.35/
1st Variation 20/ 100/ 30/ 0.035/ 0.300/ 0.35/
2nd Variation 20 100 30 0.035 0.300 0.2
Boulder Bed Original 22/ 300/ 30/ 0.040/ 0.500/ 0.35/
1st Variation —/ —/ —/ —/ —/ —/
2nd Variation — — — — — —
Jurong Rock Formation Original 24/ 600/ 35/ 0.040/ 1.000 0.3/
1st Variation —/ —/ —/ —/ /—/ —/
2nd Variation — — — — — —

for both the tunnel initial support (temporary works) and the permanent support (per-
manent works) had to be provided by the Contractor for approval by the LTA and the
Building & Construction Authority (BCA).
The tragic incident at the Nicoll Highway in early 2004 triggered a series of addi-
tional ground investigations and the request for additional numerical analyses for the
design of the Fort Canning Tunnel. These analyses included comprehensive ground
parameter studies and sensitivity analyses to ensure that a robust design was devel-
oped leading to a save and successful tunnel construction.

Design Parameters
Investigation Program. The soil investigation program included approximately 40
borings along the alignment to retrieve cores for laboratory testing and to carry out
borehole tests such as Standard Penetration Tests and groundwater observations.

Variation of Design Parameters

During the design verification phase the geological model along the alignment
was modified (Figure 2) and a series of parameter studies was carried out. Further-
more, a weak residual soil layer (RS I) was assumed to be present above the crown of
the tunnel and a “worst credible” geological profile was introduced. The corresponding
FE model was generated. Table 1 summarizes three sets of parameters used during
that phase. These were used for the design analyses at different design stages.
RETC2007.bk Page 979 Thursday, April 19, 2007 1:48 PM

MONITORING SUCCESSFUL NATM IN SINGAPORE 979

Figure 3. FEM analysis model (CH 240) at the location near the new support wall for the Singapore
Natural History Museum

Design Analyses Results

The finite element program Phase2, V5.04 by Rocscience, Inc. was employed for
the analyses of loading conditions, ground response to the tunnel construction and to
assess the lining forces. Triangular solid material elements were used to model the
ground, and beam elements were used for the representation of the linings. The soil
was modeled as an elastic-plastic material using the Mohr-Coulomb failure criterion;
the linings were simulated as ideally elastic-plastic materials.
All finite element models for the primary lining design utilize a multi-staged model-
ing approach. An initial stage describing the in-situ stress state of the soil prior to tun-
neling formed the start point. The modeling stages were established to asses the
individual construction stages. Top heading excavation was modeled by softening the
soil within the excavation limits of the heading followed by the excavation of the soil
elements within the excavation area. This approach simulates the excavation and
resulting relaxation of the soil surrounding the tunnel prior to installation of the shot-
crete support.
The increase of compressive strength and stiffness of the shotcrete lining after
installation was modeled by gradually increasing the stiffness of the lining elements
during the following construction stages. By this method, the interaction between
ground deformations of the surrounding soil and the initial shotcrete lining is simulated
as the excavation progresses.
The modeling sequence of the excavation and support sequence is concluded
with bench and invert excavation and installation of the shotcrete lining in the invert.
The shotcrete lining was checked for its integrity during all the intermediate con-
struction stages. The structural capacity of the shotcrete lining was determined in
accordance British Standard BS 8110 and Singapore Standard SS CP65.
In addition to the two-dimensional finite element analyses, a three-dimensional
finite element analysis was performed to assess the effects of the AGF pre-support
umbrella and to confirm the performance of the two dimensional finite element models.
All structural design, however, was based on the results of the two-dimensional finite
element analyses. Figures 3 and 4 show typical finite element models.
RETC2007.bk Page 980 Thursday, April 19, 2007 1:48 PM

980 2007 RETC PROCEEDINGS

Figure 4. FEM analysis model (CH 340) underneath the Fort Canning Rise

Table 2. Summary of FE results

Surface Settlements Tunnel Lining Deformation
CH 240 max. 104 mm max. 104 mm max. 18 mm
(roof settlement) (settlement @ springline)
CH 340 max. 39 mm max. 57 mm max. 21 mm
(roof settlement) (settlement @ springline)
Worst credible max. 114 mm max. 138 mm max. 48 mm
(roof settlement) (settlement @ springline)

Utilizing the last iteration of geotechnical parameters (2nd Variation in Table 1) in a

series of FE analyses, the deformations at ground level and within the tunnel structure
are determined in Table 2.

Pre-Support
The Fort Canning Tunnel is constructed under very limited overburden ranging
from 3m (9ft) to 9m (19ft); i.e., 20% and 60% of the tunnel width. Due to this shallow
overburden an AGF pipe arch was used as a continuous pre-support system over the
full length of the tunnel (Figure 5).
The AGF (“All Ground Fastened”) pipe arch consisted of a single row of steel
pipes drilled at 400mm spacing in the crown of the tunnel top heading. The outer diam-
eter of the steel pipes was 114mm (4.5"), pipe wall thickness 6mm (1⁄4"); overall pipe
length is 12.5 m (41ft), installed in four sections (3.5m–3m–3m–3m). The steel pipes
were installed at a 7% outwards angle. A 3.5m overlap between succeeding pipe
umbrellas was provided. 9m long tunnel sections were excavated followed by the
installation of the next AGF umbrella in sequence.
A polyurethane two-component grout was injected through the AGF pipes via
grouting ports at 0.25m (3⁄4 ft) spacing along the length of each pipe. The grouting pro-
cess was both volume and pressure limited for each individual pipe.

Excavation and Support Sequence

The Fort Canning Tunnel was constructed using an excavation and support
sequence comprising of top heading excavation (with temporary invert) and combined
RETC2007.bk Page 981 Thursday, April 19, 2007 1:48 PM

MONITORING SUCCESSFUL NATM IN SINGAPORE 981

Figure 5. Pre-support umbrella installation

Figure 6. Excavation and support sequence

bench/invert excavation. The top heading extended over the entire width of the tunnel
and its height was 6.0m that increased to 6.5m (20ft to 21.5ft) at the AGF installation
location. Advance length in the top heading was 1m (3.3 ft). During the trial period, the
temporary ring closure of the top heading (temporary invert) was installed in 2m incre-
ments at maximum 4m distance behind the excavated tunnel face. The temporary
invert installation was increased to 3 m blocks at max. 6 m distance to the face follow-
ing the trial section. In order to enhance the stability of the tunneling face a face stabi-
lization wedge of 2 m depth at its base is left in place.
The combined bench and invert excavation with two 2m increments in bench fol-
lowed by the two 2m increments in the invert was carried out in two steps immediately
following each other. A minimum distance of 20m between top heading and bench/
invert excavation faces was maintained (Figure 6).

Tunnel Support
Initial Support. The primary lining consists of minimum 300mm thick shotcrete
with a design compressive strength of 40 N/mm2. The primary lining is reinforced by
one three bar lattice girder installed in every excavation round at 1m spacing and two
layers (inside and outside) welded steel mesh of 10mm diameter bars at 200mm
RETC2007.bk Page 982 Thursday, April 19, 2007 1:48 PM

982 2007 RETC PROCEEDINGS

Figure 7. Initial shotcrete tunnel support in top heading

spacing. The temporary invert had a design shotcrete thickness of minimum 150mm
and was reinforced by one layer of welded steel mesh of 10mm diameter bars at
200mm spacing (Figure 7).
Tunnel Final Lining. The Land Transport Authority (LTA) as the Client of Fort Can-
ning Tunnel requested in the project design criteria that no ground water may be
drained by the tunnel structure. A waterproofing system that extends around the full
tunnel circumference and that consists of a geotextile layer and a welded PVC water-
proofing membrane was utilized to achieve this requirement. The waterproofing sys-
tem is segmented by a series of circumferential and longitudinal water barriers.

OBSERVATIONS DURING CONSTRUCTION

General
Excavation for 180m (590ft) long tunnel started in March 2005 and was completed
in March 2006. Excavation was carried out using excavators and breakers.
In general, the ground conditions and ground behavior was found in agreement
with the expectations.

Ground Behavior
The deposits were found in non-continuous layers and lenses with a varying
amount of boulders leading to inhomogeneous ground conditions with a wide variety of
ground properties. Depending on the density of the soil matrix, number and size of
boulders in the excavation face and water saturation the ground varied from very hard
to very soft (Figure 8).
Most of the sandstone boulders displayed a more-or-less thin weathered mantle.
Inside the weathered zone the boulders were very hard generating significant resistance
to excavation and breaking. While blasting of the larger boulders was not an option,
excavator mounted hoe rams were used to remove the boulders from the excavation
face and to break them up. Boulder removal lead sometimes to loosening of the sur-
rounding soil matrix and soil layers due to the vibration energy exerted by the breakers.
While the clay typically yields low permeability, sand and silt admixtures, layers
and lenses generated water paths that lead to groundwater discharge through the tun-
nel face. Despite the limited overburden thickness, the hill foot location in combination
RETC2007.bk Page 983 Thursday, April 19, 2007 1:48 PM

MONITORING SUCCESSFUL NATM IN SINGAPORE 983

Figure 8. Boulders in the bench face

with frequent tropical rainfall generated sufficient water recharge to cause wet tunnel-
ing conditions. Pump sumps had to be provided to avoid softening of the ground in the
bench and invert area.
The inherent stand-up time provided by the ground was considered too limited for
a safe tunnel support installation. That led to the decision to employ a pre-support sys-
tem over the entire length of the tunnel early on in the design phase. The lateral extent
of the pre-support arch proved sufficient to safely install the shotcrete tunnel support
after each excavation round. The grouting material penetrated the various soil deposits
including the very fine silt materials introducing cohesion for improved ground strength.
The slightly domed tunnel face of the top heading was sealed after each excavation
round. A flash layer of shotcrete was sufficient to stabilize the face and to prevent desic-
cation of the soil between the excavation operations. Face instabilities have not been
observed. This may have been caused—in parts—by the continuous pre-support but
also by the face stabilization wedge. During longer excavation interruptions, such as for
the AGF umbrella installation, full shotcrete face support was installed (Figure 9).

Surface Settlements
Surface settlements were monitored using surface monitoring points either
anchored into paved surfaces or into the soil. Readings were carried out using precise
survey instruments. Generally, settlement readings were carried out on a daily basis. If
undue readings were recorded, the reading frequency was increased to twice or sev-
eral times a day (Figures 10 and 11).

Tunnel Deformation
Tunnel deformation monitoring was carried out with total stations. The daily read-
ings yielded 3-D movement data that were evaluated immediately after taking the
readings. In tunnel sections were the full final ring closure had been established and
the tunnel lining deflections had ceased, the reading frequency was reduced to once
per week (Figures 12 and 13).
RETC2007.bk Page 984 Thursday, April 19, 2007 1:48 PM

984 2007 RETC PROCEEDINGS

Figure 9. Top heading face shortly after excavation with exposed pre-support pipes

Figure 10. Surface settlement CH 245 (overburden 5.3m)

Figure 11. Surface settlement CH 345 (overburden 5.1m)

RETC2007.bk Page 985 Thursday, April 19, 2007 1:48 PM

MONITORING SUCCESSFUL NATM IN SINGAPORE 985

Figure 12. In-tunnel deformations CH 242 (overburden 4.8m)

Figure 13. In-tunnel deformations CH 350 (overburden 4.5m)

COMPARISON OF ACTUAL READING DATA TO DESIGN DATA—

SURFACE SETTLEMENTS AND TUNNEL DEFORMATIONS
Comparison of the analytical predictions with deformation measurements during
construction yields partially inconclusive results. Using the ground properties stated
above (2nd Variation), surface settlements for CH 240 are considerably over-predicted
(104 mm predicted vs. 36 mm measured). Similarly, tunnel lining deformations were
predicted too high (104 mm roof settlement and 21 mm settlement at the spring line)
compared to the measured data (25 mm roof settlement and 13 mm settlement at the
spring line). Back analyses of the FE model at CH 240 show that the introduction of an
improved material layer in the tunnel roof modeling the AGF umbrella and an increase
of the stiffness of the soils by 50% and the shear strength by a factor of 10 yield results
more comparable to the measured deformations (24 mm roof settlement and 15 mm
settlement at the spring line). The effect of the AGF pre-support in the model was
ignored during the later design stages, to ensure a more conservative design
approach (Figures 14 and 15).
For the back analysis, the strength and stiffness parameters of the improved soil
have been assessed by multiplying the strength of each component (soil, steel) by the
cross sectional area of each component and then dividing the sum of these products
by the total area of improved soil under consideration (see Equation 1).
RETC2007.bk Page 986 Thursday, April 19, 2007 1:48 PM

986 2007 RETC PROCEEDINGS

Figure 14. Surface and tunnel settlement along the alignment

Figure 15. Tunnel deformations CH 240—back analysis with AGF umbrella

( P P ⋅ AP ) + ( P S ⋅ AS )
PAGF = -------------------------------------------------------
- (1)
( AP + AS )
When using the same soil properties (2nd Variation from Table 1) at CH 350 and
ignoring the beneficial effects of the AGF umbrella (as it was done for all final design
calculations), the numerical analyses significantly under-predicted the deformations
measured in the field. Whereas, following numerical analyses, surface settlements of
39 mm should have been expected for the prevalent soil conditions, surface settle-
ments of up to 148 mm were observed. Tunnel lining deformations at this location were
also considerably underestimated. Tunnel settlements of 57 mm in the roof and 21 mm
at spring line level are predicted; 93 mm and 71 mm respectively are measured.
A closer match to the actually measured data at this location forms the model used
as the worst credible situation at Fort Canning Tunnel. In the latter analysis, which
formed the design case for the tunnel lining, surface settlements of 114 mm and vertical
tunnel deformations of 138 mm (crown) and 48 mm (spring line) are determined.
RETC2007.bk Page 987 Thursday, April 19, 2007 1:48 PM

MONITORING SUCCESSFUL NATM IN SINGAPORE 987

Under normally consolidated soil conditions it may be expected that surface set-
tlements are somewhat smaller, or approximately equal for very shallow tunnels than
settlements measured in the tunnel crown. The surface settlement values measured at
Fort Canning tunnel are locally significantly higher than the tunnel lining deformations
(up to 50% more surface settlement). This may be attributed to insufficiently com-
pacted fill above the tunnel alignment that has not been identified during the soil inves-
tigation and a short-term consolidation effect triggered by the tunneling operation
below. However, measurement data are not yielding completely conclusive indications
for this assumption.

CONCLUSION
Categorizing inhomogeneous residual soils—as prevalent at Fort Canning Hill—
according to their SPT-N-values, as well as assessing elasticity and Mohr-Coulomb
shear parameters in order to predict the ground response to a tunneling operation con-
stitutes a challenging task. Numerical analyses utilizing a wide range of soil parame-
ters are required to reliably predict upper and lower bound of the expected surface and
tunnel lining deformations. In the Finite Element analyses performed for Fort Canning
Tunnel, the upper bound settlement values were predicted acceptably well while lower
bound values were not closely matched by analytical results using the stated elasto-
plastic soil parameters. By using SPT N-values to categorize soil strata, an exact pre-
diction of surface and tunnel deformations at locations where detailed geological strat-
ification data were available (i.e., borehole location) however was not achieved.

ACKNOWLEDGMENTS
The authors wish to thank Mr. Osamu Hasegawa, Mr. Hitoshi Suzuki and the engi-
neering staff of Sato Kogyo (S) Ltd. and Mr. Tan See Chee of TY Lin Ltd., as well as the
engineering staff of LTA for their continued support during the design and the construc-
tion of Fort Canning Tunnel.

REFERENCES
Kiso-Jiban Consultants Co. Ltd., 2003: Fort Canning Tunnel, unpublished
Geotechnical Interpretative Report.
Kiso-Jiban Consultants Co. Ltd., 2004: Fort Canning Tunnel, unpublished
Geotechnical Interpretative Report on additional soil investigation borings.
Wallis, S., (2006). Singapore tests NATM at Fort Canning, Tunnels & Tunneling
International. May 2006.
T. Schwind, K. Zeidler, V. Gall, NATM for Singapore, North American Tunneling 2006,
ed. Levent Ozdemir, Taylor & Francis, 2006.