DEPOT SITES

8 8 BROMS - PROFFESOR, SCHOOL OF CIVIL AND STRUCTURAL ENGINEERING, NT/

J N SH/RLAW- GEOTECHNICAL ENGINEER, MR1C

SUMMARY INTRODUCTION
All three MRT depot sites have required extensive Much of Singapore Island has already been
reclaimation and soil improvement. At each of the three developed to meet the housing and industrial needs of
depots the particular subsoil conditions and maintenance the country. Because of the scarcity of land in Singapore
requirements have led to different techniques being most of the Mass Rapid Transit (MRl) system has been
used. For the main depot, at Bishan, trials were carried located in tunnels below ground or is elevated supported
out on the use of vertical drains, preloading and on piers. Therefore the system does not require much
dynamic replacement and mixing, before it was finally land, except for depots needed to se!Vice, stable and
decided to use a piled concrete deck. At Ulu Pandan operate the trains. However, the soil conditions in the
depot the relatively shallow poor ground, and availability areas which were available for the depots near the MRT
of fill from cuttings, made excavation and replacement lines were very poor. This has made previous
economical. At Changi depot the time available for development unattractive.
construction allowed for use of preloading with vertical The locations of the three MRT depots, Bishan
drains to preconsolidate the underlying marine clays. (Phase 1), Ulu Pandan (Phase 1A) and Changi (Phase
The ground conditions and the particular considerations IIA) are shown in Fig. 1. Bishan is the main depot of
involved in selecting the appropriate methods are the MRT system, while Ulu Pandan and Changi are
discussed. The use of high strength geotextile for the auxilliary depots and smaller in area. The construction
successful reclaimation of old tailings ponds at Changi of these three depots and the different soil improvement
is also described. schemes that have been used to improve !he soil
conditions are described in this paper.

IIII61SHAN

J
0
0
llil CHANGI
0
~~~

0=
,.,o''•'"•'''•'''~

FIG. 1 LOCATION OF MRT DEPOTS

71
t. BISHAN DEPOT Fig. 2). Below the Kallang formation is weathered
granite (G4) or weathered cemented sand, the old
1.1 General alluvium (0).
The Bishan depot is located just south of the new The simplest, and normally the cheapest method of
town of Ang Mo Kio. Within the 30 hectare cjepot are raising an area to grade is by filling. However, the soft
area offices, maintenance and repair workshops, stabling estuarine and marine clays at Bishan were highly
s1d1ngs and a 2km long test track. The original elevation compressible. Under an imposed load of 36kpa due to
of the site was about + 102.5mPD, apart from a hill to 2.2m of earthfill and 54kpa due to ballast, tracks and
the northwest of the depot. The lower parts were prone trains, it was calculated that a total settlement of up to
to flooding, particularly when heavy rains overloaded the 2.5m could occur over a 30 year period. Such large
Kallang River immediately to the south of the site. It was settlements would be totally unacceptable. It was
therefore decided to raise the area about 2.2m to therefore decided to investigate different ground
+104.7mPD. improvement methods to reduce the settlements. The
A layer of very soft, organic (up to 60%) estaurine following methods were investigated:
clay 5 to 7m thick covered much of the site. Other i) Preloading with vertical drains. Vertical drains
deposits of the Kallang formation, marine and fluvial accelerate the consolidation of the soft clay below the
clays and fluvial sands, underlaid the estuarine, giving a fill. The consolidation rate can also be increased by
total depth of soft soil varying between 7 and 25m (see increasing the surcharge load.

c
KEY
D 0 - 5m

m 5 -10m

ITJ 10- 20m

• Over 20m
0 100m 200m

scale
FIG. 2 BISHAN DEPOT
THICKNESS OF KALLANG FORMATION &
LOCATION OF EMBANKMENTS

KEY
• Piled concrete deck

~ Main Depot Buildings

FIG. 3 BISHAN DEPOT

LOCATION OF CONCRETE DECK & MAIN BUILDINGS
72
ii) Dynamic Replacement and Mixing (DRM) without higher than DRM. A pile supported concrete deck had
vertical drains. A heavy weight is used in this the following advantages:-
method, which is dropped repeatedly at the same i) Simple, readily available technology which had
location from a height of 10m to 20m. The hole after already been used and proven in Singapore. The end
each blow is filled with sand, and gradually a sand results could almost be guaranteed, and delays
column is formed in the soft soil as the number of avoided by mobilising additional resources.
blows increases. The sand is also mixed with the soil. ii) Maintenance costs are low.
After this treatment the area is preloaded. The sand iii) Stabilization of the slopes along the Kallang River was
columns will increase the consolidation rate during avoided.
the preloading.
iii) Dynamic Replacement and Mixing (DRM) with vertical The Bishan Depot is a key facility both with respect
drains. Vertical drains are used to accelerate the to the construction of the MRT and its operation. The
consolidation of the soil below the sand columns. depot is currently used as a base for the track-laying
and for the installation of electrical and mechanical
1.2 Test Embankments equipmment for the MRT. After the MRT becomes
Four trial embankments were constructed. Three were operational, regular track maintenance would be
used to investigate the different soil improvement extremely disruptive due to the concentration of facilities
methods as outlined above, the fourth embankment at the Bishan Depot. Since only a year was available for
served as reference. the construction of the depot between the award of the
The size, cost and method of construction of the four contract and the start of the tracklaying, it was decided
embankments are shown in Table 1. to choose the safer option of a piled concrete deck for

TABLE 1 -COSTS OF TEST EMBANKMENTS AT THE BISHAN DEPOT

Embankment Area Cost• Height Ground

m• S$/m• m Improvement Method

A 1600 61 3.6 Pre loading and drains

2.0m and 1.5m spacing
B 1800 124 3.0 DRM and drains at 2.2m
202** spacing
c 1600 36 3.5 Pre loading only (no drains)
XI 1500 119 3.9 DRM only (no drains)
197**

• Excluding costs of instrumentation, site clearance, offices and insurance.

• • Extrapolated coast for DRM with heavy tamping.

It was found that stability was a problem for all four the most critical part of the depot, a 120,000m• area.
trial embankments, due to the very low shear strength The eastern part of the site would simply be filled since
of the underlying very soft estaurine clay. Slip failures only the test track would be affected by the resulting
occurred at Embankments A and B while C and XI large settlements. The maintenance of the test track
showed signs of incipient failure. would not affect the operation of the depot but would
The four embankments were monitored for allow the crews to be trained in reballasting the tracks.
approximately six months after the filling. The
settlements of Embankments A and C were large, 0.5 to 1.3 Construction
1.0m. No particular acceleration of the settlements of the The 14,500 piles in the contract were driven and the
two embankments by the drains could be observed. 120,000m• concrete deck was constructed successfully
The settlements of embankments B and XI were on time. The first piles were driven in November 1984.
relatively small. The settlements at embankment XI, The tracklaying started one year later in November
which varied between 150 and 400mm, occured much 1985. No significant settlements have been recorded of
faster than for the other embankments. the deck since its installation. The cost of the piles and
The trial embankments indicated that preloading of the deck was about $250 per m•. This is somewhat
alone, or preloading combined with vertical drains, higher than the unit cost for the DRM, stablized )rial
would not be sufficient to reduce the long term embankment. However, it is probable that the cost of
settlements within the available time to an acceptable the trial embankment did' not fully reflect the overall cost
level. The best results were obtained with high energy of a complete project (see 3.3 below).
DRM, and the remaining long term settlement after
three months of preloading would be acceptable. 2. UB..U PANDA.N DEPOT
However the cost for this method was very high, about 2.1 General
$200 per m2 , and the slope stability problems noted The Ulu Pandan depot, including a 600m long spur
earlier were not fully resolved. line together with a 450m long by 140m wide siding
It was therefore decided to investigate an alternative area, is located at the western extension of the MRT.
using a concrete deck supported on precast concrete or The main function of this depot is to provide the
steel H-piles driven into the underlying weathered Bukit stabling for the trains. There are also some other minor
Timah Granite (G4) or the Old Alluvium (0). The cost of facilities in the depot.
this alternative was estimated to be similar to, or slightly The depot is located in an old stream valley. Up to
73
6m of estaurine and fluvial materials have been The tailings in the ponds were extremely soft, and
deposited in the centre of the valley up to El. the depth varied between 1m and Sm. Cone penetration
+ 106mPD. Along the sides of the valley there are tests carried out from a pontoon showed no measurable
outcrops of weathered sedimentary rock, or the rock cone resistance as shown in Fig. 6. Vane tests indicated
occurs just below the surface. Uncontrolled filling had a sheer strength of about 1kPa but only altere the vane
already taken place over much of the site so that the had been pushed down 2m into the tailings. The water
existing ground level was generally between + 110mPD content was very high, well in excess of its liquid limit
and +115m PD. (68).
The spur line to the depot cuts through a series of It was planned to build a workshop and covered
small hills, with elevations of up to +129m PD. To sidings in the western half of the depot and open
provide a stable platform level at about + 110mPD both sidings in the eastern half. Since the original ground
cutting mainly for the spur line and filling were required. level varied from about + 100mPD to + 103mPD it was
Calculations of the long term settlements and of the necessary to provide a stable platform at + 103.05mPD
stability of slopes around the sidings area showed that, to prevent flooding. The design load from ballast and
as was the case at the Bishan Depot, the major trains, which was similar to that at the Bishan depot,
problems were connected with the soft estaurine clay. was equivalent to an additional 3m of fill. Calculations
Even relatively thin deposits of this material could cause indicated for this loading a total settlement which varied
large settlements and unstable slopes. Various methods from 0.2m after 30 years at the centre of the site, where
to treat the clay were considered. there was no soft marin"e clay, to nearly 2.0m at the
Because the clay was localised, it could be removed eastern end where the soft marine clay was thick.
and replaced easily. Since suitable fill was available Although the calculated settlements were almost as
from cuttings within the site this method was very large as those at the Bishan depot, Changi presented a
economical. · slightly different problem. At the Bishan depot the
The earthworks started in November 1985, and were settlement was predicted to occur largely in the shallow
largely completed by August 1986. estuarine clay, while at the Changi depot the settlement
The following volumes of material were moved during was mainly associated with the soft marine clay, which
the construction:- was up to 14m thick. There was more time available at

TABLE 2 VOLUME OF REWORKED SOIL AND ROCK AT THE ULU PANDAN DEPOT

Excavation and removal of unsuitable material 11,30oma

Excavation of fill material 196,00oma

Placement of earth and rockfill material 235,00oma

11>
'
The gross area of the site was 105,600m 2 including Changi, since tracklaying in the western half was not
cut slopes. The usable area was about 70P00m2. scheduled until 18 months after the award of the
Total cost of the contract was approximately S$4M. The earthworks contract, while 30 months were available at
cost ($571m 2) for the trafficed area was very low in the eastern end. The maintenance requirements at
comparison with the concrete deck required for the Changi were also less stringent than at the Bishan
Bishan Depot. depot. With spare sidings. available, it would be possible
to maintain the track without disrupting the operation of
the depot. However, the cost for the maintenance had to
3 CHANGI DEPOT be considered in the evaluation of the methods.
A variety of ground improvement methods were
3.1 General considered in order to reduce the settlements, including
The Changi depot is bounded on one side by the a piled deck similar to that of the Bishan depot. After
Sungei Ketapong drainage canal and on the other side much discussion tt was decided to use preloading and
by the Bedok sewage treatment plant as shown in Fig. 4. vertical drains in order to reduce the time required tor
The total area of the depot is about 100,000m 2. There the consolidation. The main reasons for this decision
was a recently cleared kampong area in the western of were:-
the site wtth vegetable gardens, fish ponds and a (a) The successful use of vertical drains at the Changi
Chinese temple when the site was taken over by MAT airport in similar ground conditions (2).
In the eastern part there was an abandoned sand (b) The area needed first, the western part, had
quarry with an old wooden sand extraction plant. There relatively favourable ground conditions and small
were also some large ponds, abandoned sand pits, with predicted settlements. A relatively long time was
a total area of 40,000m 2 (Fig. 5). They had partially available for the eastern part where the ground
been backfilled with washings from the sand extraction. conditions were less favourable.
Investigations showed that the depth of the soft (c) If the scheme was not fully successful, and
Kallang formation was extremely variable within the site. settlements would exceed the predicted 0.1m during
The depth to the underlying Old Alluvium (0) varied the first three years of operation, adjustment of the
from 5m to 19m as shown in Fig. 4. The composition of tracks would not disrupt the operation of the depot.
the Kallang formation was variable with extensive fluvial (d) The settlements caused by secondary consolidation
deposits as well as marine clay. In the eastern part of of the marine clay, with an organic content of
the depot the marine clay was relatively deep and generally 3% to 5%, would be small compared with
uniform while in the western part there was more fluvial those of the highly organic estaurine clay at the
clay and sand. Bishan depot.
74
 Sungei Ketapang drainage canal

KEY .··,.

D 0 - 5m
Bedok sewage treatment
~ 5-10 m plant

0. 10- 15m

~ 15-20 m

D 20 + m FIG. 4 CHANGI DEPOT

THICKNESS OF THE KALLANG FORMATION
ABOVE THE OLD ALLUVIUM (0)

0 300m

scale

FIG. 5 CHANG! DEPOT - LOCATION OF PONDS

Cone resistance, ~c, MPa

0 5 10
0 '
~
.j'CP10

E >
.r::;"
--~CP11 GEOFABRIC DETAILS
a(!)
5 I Material: Woven polypropylene
0 Tensile strength: 80 kN/m
Mass: 500 gm/m.sq.
Elongation: 19%

10
~-- - --- CBR puncture: 14.4 kN

FIG. 6 - RESULTS FROM CONE PENETRATION TESTS, CP 10 & CP 11

75
3.2 Pond reclaimation strength woven geotextile. The different construction
Before a suitable preloading scheme could be stages of this method are shown in Fig. 7. After placing
selected, it was necessary to find a suitable and the geofabric over the pond, the fabric is restrained by
economical method to stabilize the pond tailings. berms placed along the two sides of the pond (Stage
Because of the extremely low shear strength of the II). Narrow berms are then constructed across the pond
tailings ( < 1kPa) the maximum height of fill that could (Stage Ill) which had to be built at the same time and at
be placed without causing mud-waves or slip failures the same rate to prevent heaving, since the stability and
was only 0.3 to 0.6m. Various methods were considered:- the bearing capacity depend ·on the restraint from
adjacent berms. The fabric is gradually stretched as the
(a) Excavation and replacement. The main soft clay is displaced below the berms. The increase of
disadvantages with this method was the transport the bearing capacity provided by the fabric depends on
and the disposal of the tailings. the stress-strain properties of the fabric, the width and
(b) Excavation, treatment and replacement. This method the height of the berms as well as on the shear
was not investigated in detail, but the treatment with strength of the underlying clay. Once the berms have
lime was considered to be a possibility. been constructed across the pond the berms can be
(c) In situ stabilization. It was preferable to use the widened (Stage IV). Once the first layer of the fill has
same method as for the rest of the site i.e. vertical been placed the height can be increased following the
drains and preloading. same sequence (Stage V). It is important that the
Before preloading and vertical drains, Method (c), bearing capacity of the underlying soft clay should not
could be adopted, it was necessary to develop a be exceeded during the construction of the berms.
procedure so that the drains required for the With the method outlined above it was considered
consolidation could be installed and the fill required for possible to construct a stable platform over the ponds
the preloading could be constructed without exceeding from which the vertical drains could be installed in order
the bearing capacity of the tailings. to increase the consolidation rate. After the installation of
A method of improving the bearing capacity on very the drains the fill could be brought up to the final
soft soil has been developed in Japan as in the so- height.
called "Roped Sheet Method" described by Aboshi (1) The scheme outlined above was specified in some
and Wateri (3). Other methods using meshes (5) and detail in the contract documents. Tenderers were also
grids (4) have also been applied. At the Changi depot specifically asked to price alternative methods if these
the bearing capacity was increased using a high were preferred.

utockpile

STAGE I - PLACEMENT OF FABRIC STAGE Ill - CONSTRUCTION OF BERMS

STAGE IV - WIDEN BERMS

7~--rr;z;;-,..;.,<::>7:.:~ .· .. ··
I II ...·.
very soft clay

STAGE II - STABILIZING BERMS STAGE V - LAYERED FILLING

FIG. 7 POND RECLAIMATION USING GEOFABRIC

76
3.3 Construction $100/m2 at Changi and $250/m2 at the Bishan depot,
None of the contractors tendering for the earthworks where a concrete deck was required. The difference in
at Changi depot proposed an alternate scheme for the cost between Ulu Pandan and Changi was largely
stabilizing the ponds. The successful tender price was due to differences in soil conditions. The deep soft clay
just under S$10 million (S$99.7/m 2). This unit price is at Changi was stabilized by preloading and vertical
significantly higher than that (S$61/m2) for the test fill at drains, whereas the localised soft areas at Ulu Pandan
the Bishan depot using preloading and vertical drains. could be excavated and replacecj with readily available
However, the price in the tender tor the Changi depot fill.
included a 6m high till compared with a 3.6m fill at the Some of the higher cost at the Bishan depot was
Bishan depot trial. A breakdown of the price showed due to a layer with highly organic soil which increased
that the tender also included several other major items the cost considerably. In this particular case the
which were not included in the test embankment at the overriding need of a guaranteed settlement tree platform
Bishan depot as indicated by the following table:- coupled with a very tight schedule limited the options.

TABLE 3 ADDITIONAL COSTS AT THE CHANGI DEPOT

Item Cost, S$/m2

Offices, stores etc. 2.8

Instrumentation, monitoring, site

investigation 7.1

Insurance, fencing, miscellaneous 1.3

Reinforced earthwall along Sungei

Ketapong 14.0

Geotextile for the ponds 2.0

Temporary ditches, drainage and turfing 5.8

33.0

When these items are excluded it can be seen that References

the cost per ms of till was significantly less at the (1) Aboshi H 1985 Soil Improvement Techniques in
Changi depot than for the trial embankment at the Japan. Pro c. Seminar on soil Improvement &
Bishan depot, as could be expected because of the Construction Techniques in soft Ground, Singapore 10 -
larger volume of work. 11 January 1984, pp3 - 16.
The proposed method of reclaimation of the ponds (2) Choa V, Vijiaratnam A, Karunaratne G P. Ramasamy
using geofabric proved to be very effective in practice, S D and Lee S L (1979) Consolidation of Changi
without any significant problems. The surcharging was Marine Clay of Singapore using flexible drains. Proc.
completed by April 1986 at the western end of the site, 7th European Cont. Soil Mech. and Found. Engrg.,
and by December 1986 at the eastern end. The Brighton, Volume 3, pp29 - 36.
consolidation of the soft soils is proceeding in (3) Watari Y, 1984 .Reclaimation with Clayey Soils and
accordance with the predictions. The settlements at the Method of Earth Spreading on the Surface. Proc.
ponds at the eastern end have been large. The Seminar on Soil Improvement and Construction
surcharge will be removed by the middle of 1987 at the Techniques in Soft Ground. Singapore 10 - 11
western end and by November 1987 at the eastern end. January 1984, pp103 - 119.
The only significant problem at the Changi depot has (4) Watari Y, Higuchi Y and Aboshi H 1986 Field Studies
been the stability of the slopes along the Sungei of the Behaviour of Geogrids and Very Soft Ground.
Ketapong drainage canal. Reconstruction of the canal by Proc. 3rd tnt. Cont. on Geotextites, Vienna, Austria,
the relevant authority was carried out at the same time Vol. 1, pp187- 191.
as the construction of till required for the preloading (5) Yamauchi H and Kitamori I 1985 Improvement of Soft
next to the canal. By limiting the height of the fill and Ground Bearing Capacity using Synthetic Meshes.
by careful monitoring of settlements and of the lateral Journal Textiles and Geomembrances, Elsevier, Vol. 2,
displacements the problem with the stability of the No. 1, pp3 - 22.
slopes could be overcome.

4. CONCWSIONS
The cost of providing a raised platform at the three
depot sites varied from $571m2 at Ulu Pandan, to

77