-
1 INTRODUCTION Tanjong Pagar Tower and Capitol Development are
two major mixed use building developments which started formal
design in early 2011. Capitol is set for completion in 2014/2015
with Tanjong Pagar following just behind in 2016, artistic
impressions of each development are shown in Figures 1 and 2. Both
projects offered significant design and construction challenges for
the project teams. Both have significant basements to be
constructed within close proximity to sensitive existing and high
profile structures including MRT stations, historic and iconic
buildings. Arup were the geotechnical design-ers for both projects.
PLAXIS 3D v2011 was used by Arup on each project to confirm
suspected behaviour due to 3D ef-fects of the excavation that could
not be well modeled by other 2D software or easily through
manipu-lation of output from 3D software. This paper will detail
lessons learnt in using the Plaxis 3D finite element software to
verify ground movements estimations to lend for leaner design
decisions, gain confidence from authorities and push construction
programme.
Tale of 2 deep Singapore basements and how 3D modeling led to a
leaner design and keener construction programme. J. Tan, J. Austin,
J. Pang, L. Dan Arup Pte Ltd, Singapore S.S. Soh (QP Geo) Arup Pte
Ltd, Singapore
3D analysis is being increasingly used on major projects to
refine traditional analyses to facilitateleaner design and keener
construction programmes. This paper presents two recent projects,
still un-der construction at the time of writing where 3D analyses
has been used in just this way. Capitol De-velopment in downtown
Singapore comprises a 6 level (25m) deep top down basement
constructedwith contiguous bored pile walls with superstructure up
to 12 storeys high and the conservation ofCapitol Theatre, Capitol
Building and Stamford House. Two underground connections are also
made to nearby MRT City Hall MRT and the new Stamford Hotel. The
second, Tanjong Pagar mixed use development will stand as
Singapores tallest tower at 290m high once complete (due in 2016)
and comprises an 18m (3 level) deep basement immediately adjacent
to raft founded Tanjong Pagar MRT station. Both developments have
in common a proximity to high profile neighbours, sensitive
toground movements, making it important to understand with accuracy
the anticipated ground move-ments that would be caused by the
developments. Consequently top down construction methodology was
selected for both sites, which also allowed faster progress of the
above ground structures. This pa-per outlines the challenges
experienced during the ground investigation works on both projects,
the subsequent design and its verification through the use of 3D
finite element modeling, with some com-parison with initial
construction/field monitoring.
Underground Singapore 2014
-
Figure 1: Capitol Development Artists impression Figure 2:
Tanjong Pagar Artists impres-sion
2 TANJONG PAGAR MIXED USE DEVELOPMENT 2.1 Overview of the
project Tanjong Pagar Mixed Use Development is located at the
corner of Peck Seah Street and Choon Guan Street, adjacent to
Tanjong Pagar MRT station, Singapore. Figure 3 shows the site
location. The de-velopment comprises the construction of a 290m
tall mixed-use residential and office tower with an 18m deep three
level basement. Direct links will also be made to connect with the
existing under-ground MRT station. A full top-down construction
method was adopted for the basement excavation works.
Figure 3: Tanjong Pagar Site location Figure 4: Tanjong Pagar
Site Geology (DSTA
2009) The development will tower over the other buildings in the
vicinity with its 64 storeys, making it the tallest building in
Singapore. Figure 5 shows the detailed site plan showing the
surrounding buildings and boundaries. The main tower is founded on
a 4m thick raft supported on 1.8m diameter bored piles, extending
about 40m below the base of the raft.
-
Figure 5: Tanjong Pagar Site location and surrounding buildings
2.2 Ground Conditions The project area is underlain by the Jurong
Formation. The geological map indicates that the site is lo-cated
where the Queenstown Facies and the Rimau Facies meet, with the
Queenstown Facies at the western end and the Rimau Facies on the
eastern end of the site. An extract of the DSTA map relevant to the
Tanjong Pagar site is given in Figure 4. The approximate site
location has been highlighted for clarity.
The geotechnical investigation was carried out for the
development in mid-2011. The drilling of 20 boreholes to depths up
to 60m was carried out using rotary drilled (wash bored) methods.
In-situ soil testing included Standard Penetration Testing (SPT),
falling head permeability testing and pressure-meter testing.
Undisturbed samples were obtained using thin walled tubes and a
mazier sampler. Some of the undisturbed samples recovered generally
appeared disturbed and due to the interbedding of Ju-rong Formation
S-IV and S-V. Groundwater monitoring using standpipes and
Casagrande-type pie-zometers positioned in the Residual Jurong
soils stratum and in the completely weathered Jurong For-mation
rock stratum. Laboratory testing was conventional and included
moisture content, Atterberg limits, particle size distribution,
specific gravity, unconfined compressive strength, point load index
and chemical testing. Generally the stratigraphy of the site
appears relatively consistent across the footprint of the proposed
structure with some localised deepening of more weathered Jurong
Formation. Figure 6 gives an ex-ample cross section of the
stratigraphy. In general, the soil profile starts with around 3m
Fill, under-lain by residual soils of the Jurong Formation and
completely weathered Jurong Formation. The Ju-rong Formation soils
to 97mRL are typically residual soil to completely weathered rock
with an SPT N value of between 50 and 100. Below 97mRL, SPTs
Increased to indicate completely weathered rock has an SPT N value
consistently above 100 (extrapolated).
-
Figure 6: Example geological cross section
2.3 Key risks to the project The site is located immediately
adjacent to the existing underground Tanjong Pagar MRT station that
is supported by raft foundations. One of the biggest challenges of
this project was to control the movement of the MRT structures
keeping within the stringent criteria set by the LTA. There are two
other buildings founded on shallow foundations located less than
20m away from the proposed devel-opment Maxwell Chambers and shop
house at 76 Peck Seah Street. These adjacent sensitive struc-tures
necessitate the needs to understand likelihood movement.
2.4 Why 3D modeling was employed Initially 2D Plaxis analysis
was used to size the wall elements and for early predictions of
ground movement behind the new basement structures. Due to the
complex soil-structure and structure-existing structure
interaction, a full 3D finite element model was employed to study
the effect of exca-vation and long- term settlement of the
structure itself. To better control the retaining wall deflections,
sequential excavations of zones was proposed by the Contractor
excavation works were divided into four phases :
- Office/residential (Phase 1) - Hotel (Phase 2) - MRT link
(Phase 3 and 4)
This phasing was expected to significantly impact on predicted
movements around a non-regular basement and simpler analyses
software was not deemed appropriate for estimation of the resulting
wall deflections and movement of soils surrounding the new
basement. For this reason 3D Plaxis was used to better capture the
complexity of the 3D behavior 3D modeling was employed in a number
of different ways for the project:
- To investigate proposed excavation sequence - To investigate
soil structure interactions (for both the new structure and
existing MRT box)
3D modeling was employed part way through the project when
addressing anticipated movement at the MRT structures.
-
2.5 Approach to the 3D model Geometry and construction sequence
A top-down construction method has been proposed for the basement
excavation works. The typical construction sequence begins with the
installation of Contiguous Bored Pile (CBP) and Secant Pile Wall
(SPW). The basement columns and foundation piles (steel plunge in
columns/king posts) will then be constructed before any excavation
works commence. For each side of the excavation, the floor slab
(L01) will be constructed first, with at least one construction
hole left open to allow for the re-moval of the excavated soil.
Further excavation down the opening will take place only when the
L01 slab has gained the sufficient strength. The openings will be
left open within each newly formed base-ment floor slab and the
procedure will repeated for subsequent basement levels. The finite
element mesh The soil is modeled using 10- node tetrahedral
elements with the wall and slab modeled as a 6 node triangular
plate element. Ground conditions Section 2.2 of this paper
discusses the ground conditions. In the 3D model, all layers are
modeled as drained. The ground water table is at 2m below ground
level. Constitutive modelling A Hardening Soil model was adopted
for all soil layers. The Hardening Soil model is an elastoplastic
type of hyperbolic model, formulated in the framework of shear
hardening plasticity. Specifically the model involves compression
hardening to simulate irreversible compaction of soil under primary
com-pression.
3 CAPITOL MIXED USE DEVELOPMENT 3.1 Overview of the project
Capitol mixed use development is located in the centre of Downtown
Singapore. The development comprises a 6 level (25m) deep top down
basement constructed with contiguous bored pile walls with
superstructure up to 12 storeys high and the conservation of
Capitol Theatre, Capitol Building and Stamford House. Two
underground connections are also made with a 110 m long B2 level
connection to City Hall MRT to be excavated beneath the busy North
Bridge Road and a three level deep base-ment linkway connecting the
main basement to the new Stamford House hotel. Once complete, the
12 storey Capitol Development will boast one of the deepest
basements of its size in Singapore. Figure 7 shows the buildings
surrounding the site and highlights the different elements.
Figure 7 : Capitol Surrounding buildings Figure 8: Model of
basement layout incl. MRT
tunnel, 3 and 6 level basement
-
3.2 Ground Conditions The DSTA geology map (Figure 9) indicates
that the Capitol Site sits over the Kallang Formation. Ground
Investigation was carried out in 2 stages early 2011 by AGA with
supplementary holes drilled by Kiso Jiban later in the year.
Together the Ground Investigations comprised 27 boreholes spread
across the site up to depths of 80mbgl. It was discovered that the
Fort Canning Boulder Bed extends under the site, and a number of
boulders were identified at approximately 70mbgl. The established
stratigraphy is summarized in schematic form in Figure 11.
Figure 9 : Capitol extract from geology map (DSTA, 2009)
Figure 10: Updated FCBB locations (Shirlaw 1990)
Figure 11: Schematic of ground model
3.3 Key risks to the project There were a number of key
geotechnical challenges on the project. The shallower ground
conditions made it difficult to ascertain the geology
(distinguishing between Jurong Formation and FCBB), deep-er GI was
required to establish this. Additionally there was a large number
of existing H piles which had supported the old car park build-ing
which obstructed the new works and required removal. To add to
this, the basement layout was complex, irregular in plan shape and
of varying levels. The main basement was 6 levels deep with a large
area it connected to 2 other structures including the MRT and
Stamford House via a 2 level deep passageway and 3 level deep L
shaped basement respec-tively. The Capitol site is surrounded by
sensitive buildings such as St Andrew Church, Capitol Theatre,
Cap-itol Building and Stamford hotel. Furthermore, it is located
close to the City Hall MRT station and its two running tunnels. Top
down construction method was adopted for capitol project to better
control ground movement and allow for both superstructure and
basement construction at the same time. 3.4 Why 3D modeling was
employed 3D modeling was employed part way through the project when
addressing anticipated movement at the MRT structures and the model
was extended to include the 3 level linkway. An acceptable and
accurate prediction of excavation-induced movement and associated
damage as-sessment for the adjacent sensitive buildings was
essential for the feasibility of this project. Initially 2D Plaxis
analysis was used to size the wall elements and for early
predictions of ground movement behind the new basement structures.
Oasys XDisp was used to superimpose these results to estimate the
3D effects. Unfortunately, the superposition did not work well
possibly due to the geometry of the excavation (complex plan shape
and varying excavation levels). It was felt the 2D results were
over predicting actual movement that would be experienced by the
neighbouring structures, 2D finite element analysis was considered
inadequate to predict movement at the structure location.
Therefore, to more realistically predict the ground behaviour under
applied loading, 3D finite element analysis was proposed. Later in
the programme, once piling was progressing on site, the Contractor
wanted to adjust their con-struction sequence to release pressure
on the construction programme, the Main Contractor was opti-
-
mising construction sequence to shorten the Main basement
construction period through progressing excavation in the floor
below before the casting of the slab for the floor above was
complete. For basement construction, it was proposed to cast A1
area slabs by two phases. First phase is to cast half slab, second
phase is to excavate under this first half slab while the second
half slab is being cast. 3D finite element analysis is essential to
capture this two phases slab casting and excavation process to
validate the optimized construction sequence. 3.5 Approach to the
3D model 3.5.1 Constitutive modeling To model the soil behaviour,
the elastic perfectly plastic Mohr-Coulomb model is adopted for all
soil types. Different element models were adopted to simulate
structural members. Retaining walls, perma-nent slabs and temporary
struts as well as MRT tunnels were modelled as linear elastic plate
element, while kingpost in A2 area was modelled adopting beam
element. In addition, embedded pile element was used to simulate
all foundation piles. Structural members and its corresponding
material element are illustrated in Figure 1 below. Mixed ground
conditions are sometimes found in the Central Business District in
Singapore. Materi-als which are similar to the residual soil of
Jurong Formation are encountered above the Fort Canning Boulder
Bed, occasionally. In some locations the material so described may
form part Fort Canning Clay, but without the characteristic
quartzite boulders. In other locations, it may form part of the Old
Alluvium. (quoted from CP4 2.2.1.3.3). Given the variability of
these Mixed ground conditions, the Mohr Coulomb soil model was
adopted as it is generally considered more conservative for
estimation of ground movement and therefore would better
accommodate local variation in soil properties without adversely
impacting on design.
Figure 12 Model geometry, structural members and corresponding
material properties
General design water level is at 103mRL which is 2m below ground
level. For the soil clusters below excavation area, the water
pressure is generated by implying a water head according to the
excavation level. Due to the complexity of the 3D modelling, no
water flow calculation is performed in this analy-sis, and the
water pressure within clusters is generated by its phreatic
level.
The proposed excavation is surrounded by existing buildings with
various types of foundations. In or-der to capture the effects of
the existing building loads on the proposed excavation, surcharges
are ap-plied around the excavation to simulate the loading effects
from above mentioned existing structures. 3.5.3 Finite element mesh
and calculation stages The ground profile adopted in the model is
interpolated among 5 selected critical boreholes across the site to
capture the worst ground condition. The model was extended to a
vertical dimension of 1.5 times length of the retaining wall to
eliminate any boundary effect interference. As illustrated in
Figure 2, boundary fixities were located along every vertical face
and the bottom boundary. Ground surface is free to move in all
directions. The numerical model meshing is generated in the mesh
mode automati-cally by Plaxis 3D program with 10-node tetrahedral
elements for basic soil element, 3-node line ele-ments for beams,
6-node plate elements to simulate the behavior of plates, and
12-node interface ele-ments to simulate the soil-structure
interaction behavior. Embedded pile consists of beam elements with
special interface elements providing interaction between beam and
the surrounding soil. The cal-
-
culation stages were built based on the scheduled construction
program and sequence. Plastic calcula-tion was performed for every
stage following the initial equilibrium calculation. 4 FIELD
COMPARISON 4.1 Monitoring Instrumentation Located immediately
adjacent to the existing underground MRT station and buildings,
tight control over the construction works are required. Therefore,
a comprehensive monitoring scheme was imple-mented to ensure that
the construction induced movements stayed within the allowable
limits. Automatic Tunnel Monitoring System (ATMS) which consist of
automated total stations and prisms to check the lateral and
vertical movement of existing tunnel structure. Monitoring prisms
are installed within monitoring zone at 3m interval within zone of
influence and 5m interval before and after influ-ence zone. MEMS
Tilt Beam Sensors were installed between tracks to measure the
possible twist dur-ing the construction period. Vibration meter
were installed on the tunnel wall to catch possible vibra-tion
caused by the piling or other construction activities. Building
movements and tilt were measured by building settlement marker and
tiltmeter. Performance of the ERSS walls were measured by cluster
of instruments which consists of inclinometer, water standpipe and
3 tips piezometer. 4.2 Tanjong Pagar Mixed Use Development
Figure 13 Tanjong Pagar Layout of monitoring and instrumentation
4.2.1 Inclinometer readings Currently phase 1 excavation has been
completed. Figure 14 below shows the comparison of corrected field
measurements with predicted movements of the phase 1 ERSS at each
stage of excavation works. The inclinometer readings were plotted
with the adjustment at top coordinate. Inclinometer readings could
be misleading if the toes of the inclinometers were used as
reference points for plotting wall de-flections. Back analysis was
carried when the excavation reaching B2 level, soil stiffness was
in-creased by 1.5 times to have a closer match between analysis and
actual instrumentation readings. Due to the nearby critical MRT
structure, more onerous loading condition was assumed in the
analysis, hence actual wall movements are generally less than
predicted values.
-
Figure 14 Staged inclinometer results against predictions
Figure 15 Trend plot for cluster 7 instruments
70
75
80
85
90
95
100
105
110
10 0 10 20 30 40
Elevation(
mRL)
WallDeflection(mm)
IW4(WallichStreet)
Predicted_EXC_B1
Predicted_EXC_B2
Predicted_EXC_B3
Actual_EXC_B1
Actual_EXC_B2
Actual_EXC_B370
75
80
85
90
95
100
105
110
10 0 10 20 30 40
Elevation(
mRL)
WallDeflection(mm)
IW7(Maxwell)
Predicted_EXC_B1
Predicted_EXC_B2
Predicted_EXC_B3
Actual_EXC_B1
Actual_EXC_B2
Actual_EXC_B370
75
80
85
90
95
100
105
110
10 0 10 20 30 40
Elevation(
mRL)
WallDeflection(mm)
IW8(MRTSide)
Predicted_EXC_B1
Predicted_EXC_B2
Predicted_EXC_B3
Actual_EXC_B1
Actual_EXC_B2
Actual_EXC_B3
-
4.2.2 Trend plot The plot shown in Figure 15 shows the water
standpipe, piezometer, ground settlement marker, build-ing
settlement marker readings and excavation depth with time for
cluster 7. There is a large reduction in piezometric readings
during the excavation stage. The large reduction in the pore
pressure can be attributed to the Jurong formation and its
behaviour during construction. The stress relief during excavation
causes the fissures and fractures to open and significantly
increases the permeability from its preconstruction state, a
phenomena noted elsewhere in other excavations in Ju-rong
Formation, where estimated permeabilitys of 1.15x10-6m/s was
observed (quoted from Ong 2006 ). The depth of the SBP/CBP walls
does not reduce the large draw-down as the reduction is primarily
due to the stress changes in the ground. Within the excavation, the
unloading caused the reduction in the major principal stress in the
soil elements and this resulted in large reduction of water
pressure be-low the formation level. Behind the retaining wall, the
horizontal stress is reduced to the ka condition from the original
ko condition, i.e. a reduction in 3(confining stress). In soft and
compressible subsurface ground condition, the localised piezometric
drawdown can lead to consolidation settlements and subsequent
ground settlement. However, in the vicinity of Tanjong Pa-gar site
along the Wallich Road, the subsurface ground conditions are
relatively stiff residual soil un-derlain by weathered weak rocks.
Ground settlement is rather minimal due to the reduction in pore
wa-ter pressure. 4.2.3 Long Term Settlement The performance of the
pile raft will be monitored by flat cells, piezometers and strain
gauges. Flat cells and strain gauges will measure the load
distribute to the raft and pile when superstructure load coming
down, piezometer will record the pore pressure build up. While
writing this paper, the long term foundation monitoring just
started. Figure 16 show the predicted movements of pile-raft using
Plaxis 3D. Table 3 shows the comparison between finite element
analyses with empirical calculation method.
Figure 16 Raft settlement results
-
Table 3 Comparison between finite element analysis and empirical
method
4.3 Capitol Mixed Use Development
Figure 17 Capitol Layout of monitoring and instrumentation
4.3.1 Inclinometer readings The predicted lateral deflection
from the 3D Plaxis analysis for IW-21 and IW-9 is shown in red for
the final stage of excavation. The work suspension level indicated
is obtained from 2D analysis. The measured inclinometer readings
are lower than the predicted deflection. For IW-9, the adjacent
building has 2 basements and is founded on bored piles. In the 3D
analysis, on-erous loading condition was assumed in each of the 6
level excavation stage and the deflection at each stage was
accumulated until the final excavation stage. The actual wall
movements are generally less than predicted values as the onerous
loading conditions may not occurred. From the measured inclinometer
readings, it is likely that the stiffness of the mixed soil within
the ex-cavation above the FCBB is much higher than the values
adopted in the FEM analysis.
-
Figure 18 Inclinometer results against predictions 5 LESSONS
LEARNT Parametric studies with 2D Plaxis was carried out to find
out the sensitivities of the parameters before 3D finite element
analysis was employed. Observation method is important when the
excavation is on-going, especially when the contractor changed the
excavation sequence. Back analysis was performed as part of the
observation method as changing soil stiffness to reflect the
measured movements.
Due to the limitation of Plaxis 3D (version 2011 at the time
analysis was carried), the connection be-tween plates to plates
could be modelled as fully fixed connection, which in reality most
cases are pinned connection especially at the slab-wall connection.
Therefore, the analysis results for retaining wall reaction are not
reliable for design purposes. For thickness of 1.5m base slab, it
is more realistic to be modelled as volume element instead of plate
element to avoid unrealistic sudden peak in the wall reaction.
However, for simplicity purpose on Capitol project, it has been
modelled as plate element.
6 CONCLUSIONS 3D Plaxis offered an effective tool for capturing
3D effects on high profile and geometrically complex basement
excavations. The purpose of estimating ground movement induced by
excavation has been achieved well with satis-fying results at
especially MRT tunnel locations and sensitive building locations,
which led to more accurate damage assessment of these existing
structures. Furthermore, this 3D analysis provided effec-tive
justification for contractor proposed construction sequence which
could not be captured if it is 2D analysis. The predicted retaining
wall movement during excavation stages is in reasonable order
within allowable movement limits.
-
References Christian, J.T. and Carrier III, W.D. (1978). Janbu,
Bjerrum and Kjaernslis chart reinterpreted, Canadian Ge-
otechnical Journal, 15, 123, 436. DSTA (2009). Geology of
Singapore 2nd edition. Defence Science and Technology Agency,
Singapore. H.G. Poulos & E.H. Davis (1980). Pile foundation
analysis and design. J.C.W.Ong and N.Osborne (2006). Piezometric
changes during deep excavation, International Confer-
ence on Deep Excavation, 28-30 June 2006, Singapore. JN Shirlaw,
KB Poh, RN Hwang (1990). Properties and origins of Singapore
Boulder Bed, Tenth
Southeast Asian Geotechnical Conference, 16-20 April, 1990,
Taipei. Tomlinson, M. J. (Michael John) (2008). Pile design and
construction practice. Wen, D. & Lin, K.Q. (2002). The effect
of deep excavation on pore water pressure changes in the Old
Alluvium and under-drainage of marine clay in Singapore.
Kastner, Emeriault, Dies, Gullloux (ed.), Geotechnical aspects of
underground construction in soft ground; Proc. intern. symp.,
Toulouse, France, 23-25 October 2002.