Beruflich Dokumente
Kultur Dokumente
Lim Kok Kim, Teh Hee Seang and Madan Bikram Khadka
T.Y. Lin International Pte Ltd, Singapore
ABSTRACT
Located along the southern coastline of Singapore, The Reflections at Keppel Bay is a premier
waterfront development overlooking the spectacular open blue sea at the front and the lush greenery of
Mt Faber at the back. The site is formerly a shipyard surrounded by a deep natural water harbour
dotted with jetties, boat ramps and large dry docks built along its shoreline. The jetties and boat ramps
have since been demolished and a vertical seawall put in place to enable the land boundaries to be
pushed out and optimized via land reclamation. The pre-existing dry docks have mostly been retained
to create water channels to enable the sea to extend right into the heart of the development.
The initial authority planning restrictions had limited the height of buildings in Reflections to 28
storeys apparently due to vista view controls from Mt Faber. It was subsequently relaxed following the
developers and architects appeal. The architect had shown that the towers would be located mainly
away from the central vista view and emphasized that the iconic composition and its curves would
greatly enhance the aesthetics of the area where sea channels outside is also the southern gateway into
Singapore.
The eventual design yielded 6 curved sky towers and 11 villa style apartments housing 1,129 new
homes. The sky towers are the main feature of the design. The taller three are 41 storeys high whilst
the shorter three are 24 storeys. They are arranged in pairs but with each rotated strategically for
architectural reasons and also to ensure that views from the apartment units are maximized.
Conceptualized by world renowned architect Daniel Libeskind, the alternating towers with double
curvature symbolizes an ascending symphony of chords. Each pair of the towers are connected by
skybridges at 3 levels and capped off with a steel tower crown each. The main structure is
conceptualized and designed using reinforced concrete.
This paper shall present mainly the challenges in the design and construction of the curved sky towers
in concrete. The skybridges and tower crowns are in steel. It shall also discuss how the towers are
designed to incorporate the many variations in apartment unit layouts contained within the curved
body form, each staggered differently at every floor on top of one another. Special considerations
arising from the continuous change in centre of gravity of the structure during construction causing
lateral movements are presented. Construction and instrumentation monitoring methods used are also
discussed. The figures and charts presented shall mainly be those of Tower Type 1B/3B (see Figure
2b) for consistency and clarity.
1
IStructE Conference on Structural Marvels Reflections at Keppel Bay
KEY WORDS
curved towers; vertical curvature; curved concrete frame; inclined columns; construction engineering;
creep; skybridges
INTRODUCTION
In 1995, T.Y. Lin International Pte Ltd was appointed as the C&S Consultant to engineer the
redevelopment of Keppel Shipyard where the Reflections is now sited. The proposal is to convert the
then pre-existing shipyard into an integrated premium waterfront residential development with a world
class marina with commercial and hotel amenities. The entire site is about 32 hectares and the
proposed plan is to have 2500 to 3000 residential apartment units developed in phases. Phase 1 started
with regularization and improvement of the land through land reclamation. This commenced in year
2000 with the enhancement of infrastructure by constructing seawalls to regularize the land and link
bridges along the shoreline to create a continuous promenade around the bay. Phase 2 is the
development of the 1st residential parcel around the pre-existing docks ie. the Caribbean. Phase 3 is the
construction of a cable-stayed bridge and the marina on Keppel Island.
The Reflections is Phase 4 of the project and occupies the largest land parcel in the whole
development. The site area is 83,591m2 and the permissible GFA is 193,400m2. The construction
contract was awarded in January 2008 to Woh Hup Pte Ltd under a lump sum, semi design and build
arrangement. The contract is for the full scope of works including foundation piling and the contract
period is 48 months. Figures 1a & 1b show images of the site before and after re-development (ie.
status as is today).
2
IStructE Conference on Structural Marvels Reflections at Keppel Bay
4
IStructE Conference on Structural Marvels Reflections at Keppel Bay
DESIGN
General Description
The actual form of the Reflections sky tower consists of a double vertical curvature that is about 500m
radius at the front & back, and a gently curved side face that is tapered about two degrees inwards
from bottom to top. On the taller tower, the entire floor plate shifts out by 3.9m at Level 19 and then
shifts back by 4.3m at roof level. See Figure 3.
Isometric Form
Front Elevation
Side Elevation
5
IStructE Conference on Structural Marvels Reflections at Keppel Bay
7
IStructE Conference on Structural Marvels Reflections at Keppel Bay
Figure 9 shows the natural deformation of the tower subjected to its centre of gravity shifts at every
floor, without alignment correction. The total deformation is made up of 2 components, one from
immediate elastic storey deformation and the other from creep of concrete which is time-dependent.
The immediate elastic component (plus a fraction of short term creep) will automatically be
cancelled out during construction as the vertical alignment is re-set at every floor to the designed
profile. However, the long term creep that happens after re-alignment is carried out will accumulate
over time. This is the time-dependent component of deformation that needs to be ascertained by
construction engineering and corrected by pre-camber.
characteristic dead load. The building models were sent for a wind tunnel test [2]. A summary of the
derived range of design forces for the towers are as follows:
(i)
(ii)
Base Shear:
Load Source
Wind (from Wind Tunnel Test)
Notional
(iii)
Material:
Concrete (28-day Cube Strength)
a. General: Grade 35 (All horizontal elements)
b. Vertical structure (lower levels): Grade 60
c. Vertical structure (higher levels): Grade 40
Structural Steel
a. Skybridge: Grade S355
b. Tower Crown: Grade S275
Design Performance
Based on the loads and criteria prescribed, the building is expected to perform as follows:
(i) Lateral Drift
Lateral drift analyses based on Normal Analysis and Construction Sequence Analysis carried
out are compared in Table 1.
a. Normal Analysis is based on load applied on a fully assembled model
b. Construction Sequence Analysis is based load applied on a progressively assembled model
according to construction sequence
(ii) Peak acceleration response
The 10-year peak acceleration response derived from wind tunnel analysis, based on 1% critical
damping is as follows:
a. Transverse, x-direction peak = 6.6 to 10 milli-g
b. Combined peak
= 10 to 12 milli-g
The recommended 10-year peak acceleration response limit is 19 milli-g [1].
10
IStructE Conference on Structural Marvels Reflections at Keppel Bay
TABLE 1
LATERAL DRIFT ANALYSIS
Load
Normal Analysis
Gravity (eccentric)
Elastic, SW
SDL
LL
Creep, ST
LT
Wind, max
Total
Target Limit (H/500)
Note:
105*
40
15
120
125
300
165*
20*
10*
195*
310
30 years after
completion
20
15
10
90
125
260
CONSTRUCTION
Construction of the curved towers presented several challenges. Firstly, the geometry of the form
creates a continuously changing floor plate and changing inclination of columns at every floor. The
temporary stresses, stability of the inclined structure and continuous movements arising from the time
dependent properties of concrete are complex. The centre of gravity at every floor shifts as the tower is
constructed upwards. The load history is dynamic and is dictated by the progressive load build-up on
every floor. The load timing is important too as the deformation of the structure is affected by concrete
age. The temporary deformations will be cumulative. The boundary conditions of the temporary
structure are often not so definite.
It was anticipated during the design stage that the construction of the curved structure would be carried
out as stand-alone without the deployment of any lateral strutting or props. Because of this, any
lateral movement in the structure and stresses developed during the temporary stages will be locked-in
and irreversible. It will not be possible or practical to make any alignment adjustments or corrections
after it is built.
The biggest concern arising from the above knowledge is whether the builder is able to construct the
lift shaft to satisfy the stringent verticality tolerances of present day high-rise elevator systems. Most
elevator systems can tolerate transient deformations but not many of them are designed to cope with
permanent deformations. In the case at hand, permanent lateral deformation is inevitable.
As such, construction engineering was prescribed and specified under the scope of the builders work.
The design brief in the specifications called for construction analyses to be carried out as soon as the
construction sequence and schedule are known and approved. The chief purpose is to determine the
required construction pre-camber profile and demonstrate/verify that the achieved target profiles will
be okay for the lifts. The builder is required to coordinate this with his appropriate lift sub-contractor
to establish the acceptable tolerances and work towards achieving the required verticality accordingly
11
IStructE Conference on Structural Marvels Reflections at Keppel Bay
by adjusting his pre-camber. The analysis is also necessary to check whether there will be any
temporary overstress situations in the partially and progressively constructed structure up from the
foundation. For example, it is envisaged that the most critical condition for the concrete shear walls is
when structure is in its temporary intermediate stage. This may happen when the structure reaches the
apex of its curved profile somewhere near mid height. The walls or foundation piles may be in tension
at that point when the vertical compression load component is at its lowest.
Pre-cambering of the tower structure was implemented by the builder after the construction
engineering analyses. The pre-camber profile was derived based on the amount of calculated creep
deformation to be cancelled out at the appropriate target time. The ideal time should not be too distant
in the future yet long enough to allow the bulk of the creep to be dissipated.
As part of specified requirement, the builder had implemented an appropriate laser-guided
instrumentation system to monitor the actual lateral movements of the structure. The monitoring
prisms were placed mainly on the external wall of the lift shaft core as this was the most critical
element. Prisms were also placed at each corner of alternate tower floors to monitor possible
distortion. The instrumentation plan and system used by the builder is shown in Figure 10.
Method
(i)
(ii)
(iii)
(iv)
(v)
Builder carried out the construction engineering analyses using SAP 2000 non-linear
version.
Initially, the natural total horizontal displacement of the structure was determined to assess
its relative magnitudes at Completion of Construction and 5 years, 30 years after. See
Figure 9.
The immediate elastic component of displacement plus a fraction of short term creep
(automatically corrected during construction by repeated realignment) is then removed to
determine the component of time dependent displacement (long term creep) that is
locked-in and irreversible. This is the deformation that has to be corrected by pre-camber.
See Figure 11.
The amount of pre-camber to set depends on the builders desired final target profile.
Figure 12 shows the projected displacement curves at various intermediate and key stages
of the structure. To attain the appropriate end target profile, the construction pre-camber
can be set to reverse any of the projected displacement profiles eg. to achieve an end
vertical profile at Completion of Structure, the pre-camber curve is simply the reverse of
its projected displacement curve.
In consideration of the lifts and expected creep expiry, the final Pre-camber profile
adopted is that of 5 years after completion of structure. See Figure 13.
13
IStructE Conference on Structural Marvels Reflections at Keppel Bay
Assumptions
(i)
Boundary conditions:
a. Base of tower is Fixed
b. Floor diaphragm restraints at Basement Level 1 and 1st storey released. Pour strip
provided.
(ii)
Load History:
a. Floor cycle = 10 days
b. Superimposed dead load = 50% average upon structure completion
(iii)
Concrete properties:
a. Concrete time-dependent properties modeled according
recommendations in SAP2000, with the following parameters:
i. Cement Type Coefficient, s = 0.25 (normal cement)
ii. Relative Humidity, RH = 80%
iii. Notional size, h = 0.2 (200mm)
iv. Shrinkage coefficient, sc = 5 (normal cement)
v. Shrinkage start age = 0
to
CEB-FIP
1990
C35
35
C40
40
C50
50
C60
60
33
34
36
38.5
Results
Realistically, some of the above assumptions may not be completely representative or realized
precisely at site. For example, the pour strips specified at Basement B1 and 1st storey which had to be
closed off early. It was initially specified to eliminate uncertainties in the degree of restraint that can
be achieved. There was also concern that the high restraint force may harm the floor structures. As
such, initial analyses model excluded these restraints. Subsequently, the pour strips were closed off
due to construction access and practicality issues.
The age of concrete and timing of load also cannot be precisely captured. Such deviations are to be
expected and can be resolved by calibration against actual survey readings.
The actual survey results were conscientiously checked against the analyses model predictions
throughout the construction. No calibration was necessary. The comparison can be said to be rather
good overall. See Figure 14.
15
IStructE Conference on Structural Marvels Reflections at Keppel Bay
Figure 14: Actual (Surveyed Profile) vs. Predicted (Target Profile) curves
16
IStructE Conference on Structural Marvels Reflections at Keppel Bay
CONCLUSION
The selection of reinforced concrete as the preferred scheme for the curved tower structure of
Reflections has proven to be an appropriate and a good choice. Earlier reservations in its selection
with regards to its variability, performance and constructability are solvable. Shortcomings anticipated
can be resolved through proper attention during design, simple detailing and stringent quality control.
Early anticipation of potential problems and rigorous, accurate engineering analyses are crucial in
minimizing problems in the construction of complex structures. In concrete structures especially, the
loading history, time-dependent properties and construction quality control of concrete are critical
factors influencing the accuracy of prediction models. Established parameters and research models are
available in technical literature and certain engineering software to simulate the behaviour of concrete
structures at different ages.
Construction engineering analysis is a must for building projects where the stress/strain build up is
highly dependent on the sequence of construction and its temporary condition at intermediate stages. It
is shown on this project that the stress and strain patterns along the structure can be significantly
different. It was found using sensitivity studies that a shorter floor cycle time would have increased the
short term lateral creep deformation of the structure by as much as 25%!, and the permanent stresses in
the vertical structure are greater at lower levels when construction sequence is taken into account.
Results of the actual survey readings of the tower lateral movements monitored throughout
construction compared well with that of the predicted movements by construction engineering
theoretical analyses. Allowance should be considered for deviations due to construction variations and
tolerance. Monitoring should check on accuracy, rate of deformation and direction of movement to
detect undesirable signs. Accuracy of assumptions and parameters that are difficult to ascertain in
theoretical analyses can be adjusted/corrected by intuitive calibration during the process of
construction by comparing with actual readings.
The speed of construction using cast-in-situ concrete on this project does not appear to be a hindrance
or disadvantage to the project construction schedule. The tower structures are all completed ahead of
scheduled time. As a matter of fact, the flexibility and adaptability of cast in situ concrete has its
advantages in the construction of buildings with extensive floor profile variations and little repetition.
In the case of this project, none of the tower floors are similar. Awareness of potential problems,
sufficient rigorous engineering studies, simplicity and clarity of details are probably the more critical
factors influencing the constructability of structures.
REFERENCES
[1] Council on Tall Buildings and Urban Habitat (1995), Structural Systems for Tall Buildings,
McGraw-Hill, Inc.
[2] Vipac Engineering and Scientists Ltd, Australia (2007), Wind Tunnel Test report on Structural
Wind Load Study of Keppel Bay Plot 1 Towers Singapore, Revision 1.
[3] Woh Hup Pte Ltd / Meinhardt (Singapore) Pte Ltd (June 2008), Construction sequence analysis
report on Tower blocks, Reflections @ Keppel Bay.
17
IStructE Conference on Structural Marvels Reflections at Keppel Bay