Underground Singapore 2011

ii

CONTENT

Session 1 ................................................................................................................................ 1

Results of a rock fissure grouting trial, Bukit Panjang Station ................................................. 3

N. Shirlaw, W. Kay, M. Creutz

Three-Dimensional terrestrial fluid analysis for assessment .................................................. 14

of water loss due to tunneling

Y. Goto, K. Orihara, T. Sadamura, Y. Huang, K. Miura

Effectiveness of large spaced PVD surcharging for inland .................................................... 24

Kallang formation and residual soil for a MRT depot

K.F. Wong, H.T. Lim, S.C. Tan, E.T. Hong, W. Maw

State of the art piling technique ............................................................................................ 31

G.U. Ulrich, P.P. Platzek

Session 2 .............................................................................................................................. 43

Numerical study on pile capacity, pile movement, soil settlement ......................................... 45

and dragload with settlement analysis

S.A. Tan, S.S. Chuah

Application of the hardening soil model in deep excavation analysis .................................... 59

P.L. Teo, K. S. Wong

Influence of various modelling assumptions in numerical analysis ........................................ 73

for deep braced excavation

K.K. Loh, C.C.M. Kho

Back analysis of a braced excavation with DCM ground improvement ................................. 85

G.J. Li, K.S. Wong, P.B. Ng

Prediction of ground settlement due to adjacent deep excavation works ................................ 94

W.M. Cham, K.H. Goh

Session 3 ............................................................................................................................ 105

Experience of hard rock tunnelling in Spain – ..................................................................... 107

Case studies of Pajares, Guadarrama & Barcelona Line 9

M. Merrie, T. Camus

Volume loss caused by tunneling of Circle Line projects .................................................... 114

Y.H. Zhang, W.M. Cham, J. Kumarasamy

iii

Prediction of pile responses due to tunneling-induced soil movement ................................. 122

W.M. Cham

Design and construction of NATM/SCL tunnel under ......................................................... 139

an operating MRT tunnel for DTL1 Contract 905

O. Sigl, J.J. Lin

A design concept for a deep underground and high-rise ...................................................... 152

building complex in Tokyo, Japan

M. Higashino, N. Ito, Y. Kobayakawa

Session 4 ............................................................................................................................ 165

A pressuremeter’s perspective on soil stiffness ................................................................... 167

K.H. Goh, W.M. Cham, D. Wen

Understanding the engineering behavior of Bukit Timah Granite ........................................ 177

during deep excavation and the benefits of early design review

N. H. Osborne, A. Yang, D. Macphie, S. H. Ra, K.M. Soh

Past experience on subsurface ground condition of Bukit Timah ......................................... 189

Granite formation in Circle Line Contract 854

S.A. McChesney, M. Maw

Geotechnical & engineering challenges for Downtown Line ............................................... 196

Stage 2 C917 & 918 projects

D.C.C. Ng, R. Prasad, B.W. Tew, C.W. Neo, K.F. Pong, R. Supargo

Application of multiple-deck-charge blasting with .............................................................. 208

electronic detonator at DTL2 Contract 915 in Singapore

C.O. Shin, T.Y. Ko, S.C. Lee, M.S. Cho, J.W. Yoon, H.S. Lee, S. Hoblyn, E.M. Aw

Session 5 ............................................................................................................................ 215

Innovative excavation design for Marina Bay Sands ........................................................... 217

J.W. Pappin, W.K. Leong, P. Iskandar

Safety by design – Clearance of a buried sea wall at Marina Bay, Singapore ...................... 226

P.J. Clark, I. Askew, O. Thoren

Design and construction of DTL Promenade Station and bore tunnels ................................ 233

for Contract 902

Y.Y. Loh, T.G. Ng, S.B. Tay, Y.K. Lee

Partial top-down method & high capacity ground anchors for ............................................. 244

optimization of temporary works design

S.Y.H. Low, A.P.C. Yong, D.C.C. Ng

iv

Session 6 ............................................................................................................................ 255

General observations on wall and ground deformations during ........................................... 257

the multi-propped excavations in CCL5

K.H. Goh, W.M. Cham

Field measurements of strut loads in LTA Contract C907 ................................................... 267

A.S. Jadhav

Settlement assessment – The key components in identifying building damage .................... 277

J. McCallum, N. Osborne, B. Vontivillu

Observed apparent pressure diagrams from actual strut monitoring of ................................ 289

excavations in Circle Line project

W.M. Cham, K.H. Goh

Settlement of sewer pipes in consolidating soft clays .......................................................... 298

S.S. Agus, N. Mace

1. INTRODUCTION

Bukit Panjang Station is the northern station of Downtown line 2. The temporary works for the station

and tunnels consist primarily of strutted secant pile walls. Diaphragm walls are used in short sections

under existing viaducts. The secant piles and diaphragm walls are socketed 1 to 2m into granite bed-

rock; in local areas where the rockhead is significantly above final excavation level, use is made of

nailed rock slopes. A grout curtain has been installed under selected sections of the temporary secant

and diaphragm walls, in granite bedrock, to extend the cut-off.

The main contractor for the project is Lum Chang Building Contractors Pte Ltd., and LSW Consulting

Engineers have carried out the detailed design of the temporary works. The required extent of the grout

curtain is part of the temporary works design. The installation of the grout curtain was carried out by

WAK Consultants (Pte) Limited, to method statements prepared by Golder Associates (Singapore) Pte

Ltd. The grouting was based on modern Scandinavian rock grouting techniques, but these had to be

adapted to the particular climatic and geological conditions in Singapore. In order to develop suitable

grout mixes and to demonstrate the effectiveness of the proposed methods, a series of grout mix tests

were carried out, followed by a full scale field trial.

Results of a rock fissure grouting trial, Bukit Panjang Station

N. Shirlaw Golder Associates (Singapore) Pte Ltd

W. Kay WAK Consultants Pte Ltd

M. Creutz

Golder Associates AB, Sweden

ABSTRACT: Rock fissure grouting was required as part of the excavation retaining support system for

Bukit Panjang Station on Contract 912 of the DownTown Line. The grouting was required to control

seepage under a secant bored pile wall, and was carried out through ducts installed in the piles. The first

10 piles grouted were used as a trial section to demonstrate the effectiveness of the methods proposed.

Before commencing grouting, grout mixes were tested for suitability. Two grout mixes were developed,

one using Ordinary Portland Cement and the second using micro-fine cement. It was found that the mi-

cro-fine grout was very sensitive to temperature. The high ambient temperatures in Singapore caused

some problems in developing a grout with the required characteristics. The grouting was carried out us-

ing the split spacing technique. The first phase of the grouting was to grout at the toes of the piles. This

was followed by rock fissure grouting, in six successive steps. At each step, water testing (lugeon test-

ing) was carried out. The testing showed how the permeability of the rock was being reduced succes-

sively as the grouting progressed. The trial was also used to confirm the quality assurance testing re-

quirements for the grouts.

Underground Singapore 2011

3

Figure 1. The Downtown Line. Bukit Panjang Station is within the red circle, and is the northern station of the

line.

2. GROUND AND GROUNDWATER CONDITIONS

Based on the Geological Map of Singapore (DSTA 2009) the site is in an area of the Bukit Timah Gra-

nite. Granite, gneissose granite, and minor intrusions of basalt, dolerite, dacite porphyry and rhyolite

were identified in the site investigation. At the site, 3 to 12m of fill and superficial deposits overlie the

soil grades of the weathered rock (grades IV, V and VI). The soil grades of the weathered rock (sapro-

lite) extend to 7 to 51m below ground surface. The rockhead profile is highly variable along the line of

the station and tunnel walls.

All of the grouting was carried out in the granite rock, as the secant piles and diaphragm walls extended

into bedrock. The Geotechnical Interpretative Baseline Report for the project (Maunsell 2008) gives

baseline permeability values of 2 x 10-6

m/s for the 10m of rock immediately below rockhead, and 1 x

10-7m/s for rock that is greater than 10m below rockhead. The fracture index in the zone 10m below

rockhead was variable, but commonly 5 or greater. The average RQD values for the rock up to 10 m

below rockhead varied between 20% and 56 %. The monitored groundwater level prior to the start of

construction was 0.4 to 2.6m below ground level.

3. SPECIFIED EXTENT OF THE GROUT CURTAIN

Because of the highly variable depth to rockhead, and the varying sensitivity of the adjacent buildings

and structures to groundwater lowering, the grout curtain is not continuous around the site. The grout

curtain is specified for three zones (zones 1, 2 and 5) of the site. Sections of wall comprising approxi-

mately 600 secondary secant piles and 14 diaphragm wall panels require a grouted cut-off in the rock.

The secant pile wall consists of 800mm diameter primary piles and 1200mm secondary piles. The sec-

ondary piles are at 1500mm centres.


4

Figure 2. Plan of Bukit Panjang Station, showing the outline of the secant pile wall

The majority of the grouting, and the grouting trial, have been carried out in Zone 2 of Bukit Panjang

Station, shown in Figures 2 and 3.

Figure 3, secant piling in progress for the east wall of

Bukit Panjang Station

4. DRILLING AND GROUTING METHODS PROPOSED

The specified minimum width of the grout curtain is 1m, with a minimum depth of 10m below rockhead

or final exaction level, whichever is less. It was considered impractical to control rock fissure grouting

to limit the width to 1m. The grouting was planned on the basis of achieving a radius of grouting of 3m

in the smallest fissure that could be penetrated by the grout. This gave a theoretical minimum curtain

width of 6m. For simplicity of control on site, the depth of the curtain was set at 10m below the toe of

the secant pile wall, which resulted in a slightly greater grouted depth than the minimum required.

It was decided to use conventional rock fissure grouting, in open holes, rather than the tubes-a-

manchette (TaMs) commonly used for grouting in Singapore. For rock fissure grouting, TaMs, and the

sleeve grout installed around the TaM, close off many of the fissures intercepted by the drill hole, so

that they may not be grouted. Provided that the drilled hole in the rock is stable, grouting in an open hole


5

is more effective, as the intercepted fissures are accessible to the grout. Single packers were used so that

the grouting was carried out in stages (Figure 4).

Figure 4. Inflatable single rock packer, just after being re-

trieved from the drill hole

In order to minimise the amount of drilling and to avoid, as far as possible, the need for overburden

drilling, it was decided to install two steel ducts in each secondary secant pile, fixed to the rebar cage

(Figures 5 and 6). Drilling for the grouting took place through these ducts. This allowed all of the drill-

ing to be carried out using percussive methods. Initially, top hammers were used, but this was later

changed to down-the-hole hammers, to improve progress and reduce noise levels. With two ducts in each

secondary pile designated for grouting, the average spacing of the grout holes was 750mm along the line

of the wall. It was decided that toe grouting of the piles and diaphragm walls would be carried out prior

to the fissure grouting, to seal the disturbed zone at the base of the wall.

Figure 5. Ducts for grouting in pile cage

Figure 6. Cage with ducts being installed

Based on modern rock grouting practice, it was decided to use only stable grouts, with control of bleed a

major criterion for grout mix selection. Two types of grout were proposed. An Ordinary Portland Ce-

ment (OPC) based grout was proposed for injection at the toes of the piles and diaphragm walls, and for

sealing of the larger fissures in the rock. The d95 for OPC grout, about 100 microns, limits the size of

fissure that this grout can be injected into. The practical limit is a minimum aperture of about 0.3mm.


6

Due to this constraint, the practical limit for OPC grouting is rock permeability of about 2 to 5 lugeon

units (2 to 5 x 10-7

m/s). In order to grout finer fissures, and so achieve a lower permeability, a microfine

cement (MFC) grout with a d95 of 15microns was proposed for the majority of the rock fissure grouting.

Although there had been some prior experience of MFC rock grouting in Singapore prior to this project,

this was the first grouting project of this type undertaken by the specialist subcontractor, WAK Con-

sultants, and the first time that modern Swedish control and testing methods had been applied in Singa-

pore. It was therefore decided that a carefully planned series of tests and a full trial were needed to es-

tablish the appropriate grout mixes to be used, and to confirm the effectiveness of the injection

sequences and procedures proposed.

From the outset, it was decided that the grouting would be carried out using a number of key principles,

listed in Table 1. These principles are based on modern Swedish rock grouting practice. Many of these

principles are also found in modern US dam grouting practice, as outlined by Bruce (2011). However,

compared with exceptionally high standards required for permanent grouting for dams, the requirements

for the temporary curtain at Bukit Panjang were less onerous. As a result, not all of the principles out-

lined by Bruce were incorporated into the work at Bukit Panjang.

Table 1. Key principles for the Bukit Panjang grouting

Principles

1. Grouting in open holes drilled into the rock

2. Use of stable grout mixes

3. Grouting using a split spacing sequence

4. Preliminary grouting trial, as required in EN-12715

5. Use of typical grout testing equipment used in Sweden, for the testing of the grout test mixes, and for routine

quality control during production grouting

6. Use of the maximum allowable grouting pressure, provided that adverse effects, such as heave, loss of grout

or hydrofracturing could be avoided.

5. TESTS AND TRIALS

Climatic and other conditions for grouting in Singapore were significantly different to those in Sweden.

The differences included:

High temperature and humidity all year

The temperature of the rock mass is approximately 25° C compared to temperatures between 5 – 10°

C in Sweden

Deep tropical weathering of the rock, with weathering extending below rockhead

Tap water temperature of about 20 – 25° C

A generally less educated and experienced workforce, particularly in respect of grouting: most of the

construction labour in Singapore is brought in from third world countries on limited term engage-

ments

The high air and water temperatures in Singapore significantly reduce the time that a particular grout is

workable, compared to Sweden.

In order to develop suitable mixes and confirm the proposed procedures, a sequence of grout mix testing

and a field trial was planned. The sequence consisted of:

1. Initial mix tests in a laboratory in Sweden (Vatenfall Research AB)

2. Mix tests in a site laboratory in Singapore

3. Final mix test on site, using the site mixers

4. A field trial consisting of grouting below ten secondary secant piles


7

5.1 Preliminary Mix Tests

Preliminary mix tests were carried out on site in Sweden and Singapore, using a variety of plasticizers

and retarders. Although tests were carried out for both OPC and MFC based grouts, the majority of the

testing was for the MFC grout, which proved to be particularly sensitive to the effects of the high tem-

perature environment in Singapore.

For the preliminary mixes a variety of tests were used to assess the properties of the grout, including:

Mud balance to confirm the water cement ratio

Marsh-cone to measure the flow properties

Bleed test in measuring cylinder (Figure 7)

Filter pump testing, to check the size of the rock fissures that could be penetrated by the grout, based

on evaluating bcrit and bmin (Figures 8 and 9)

Fall-cone apparatus to measure the increase in strength with time

Rheometer test to evaluate the rheological properties, i.e. yield strength and viscosity, which are basis

for the calculation of the penetration length

The cup test, to check the time to initial set

Most of the test equipment, including filter pumps, fall-cone and stick test (for rheology) were imported

from Sweden.

Figure 7. Bleed and cup tests at Bukit Panjang Station

Based on the preliminary tests it was decided not to use a retarder in either of the mixes, as there was no

retarding effect on the grout once the grout temperature reached 30OC. In order to provide adequate time

to pump the MFC grout, and maintain penetration capability, it was found that a slightly increased wa-

ter:cement ratio, compared with typical grouts in Sweden, together with control of the initial temperature

of the grout were necessary.


8

The final MFC grout mix had a water cement ratio of 1.2, compared with the typical 1:1 ratio used in

Sweden. The increased water:cement ratio increased the bleeding for the grout, however the latest grout-

ing research has shown that an increased water cement ratio can be used in a MFC cement grout, as

long as the bleeding can be kept at an acceptable limit and the strength requirements are not critical. In

this case, the measured bleed with the 1.2:1 mix was 5 – 7 %, which was considered acceptable. The

mixing water was cooled to 10° C. The cement power, which was stored in the open, on site, had an ini-

tial temperature of more than 30° C. Combining the chilled water and the cement power resulted in an

initial temperature of about 20° C. The resulting MFC mix had a pot life of approximately 30 – 40.

The final microfine cement mix consisted of a 1.2:1 water to Microcem 650, with 1% Muraplast120S

additive. When mixed in the high speed colloidal site mixer and tested with a filter pump, bcrit was 95

µm and bmin 41 µm. An OPC grout mix was developed for toe grouting and for the injection of large fis-

sures. For the grouting trial the OPC grout mix was 1:1 water : cement, with 1% bentonite used to con-

trol bleeding. As discussed below, a thicker OPC mix was used after the grouting trial.

Figure 8. Filter pump testing on the Bukit Panjang site Figure 9. The filter used in the filter pump

5.2.1 Outline of grouting trial The grouting trial was carried out to assess the effectiveness of the mixes, injection sequences and limit-

ing pressures proposed. The trial was a ‘working’ trial: the grouting for the trial formed part of the fin-

ished grout curtain. Grouting was carried out under 10 secondary secant piles. Each pile included two

110mm preformed ducts (Figure 5), so that drilling commenced just above the toes of the piles. The drill

holes were split into A1, A2, B3, B4, C5 and C6 holes, to define the grouting sequence, as shown in

Figure 10.

Figure 10. Plan view of the grout trial


9

Before carrying out rock fissure grouting, toe grouting was carried out to seal the disturbed area at the

base of each pile. Toe grouting was carried out only in the A and B holes. The holes were drilled to 1m

below the base of the pile, and grouted using the selected OPC grout. Toe grouting on a hole continued

until reaching either of the following criteria: a pressure of 10 bar, held for 10 minutes, or a maximum

volume of 200 l.

After completing the toe ground, the holes were extended 10m below the pile toes, and stage grouting of the rock was carried. The grouting was on a split spacing basis, in the sequence A1, A2, B3,

B4, C5, C6. The cumulative grout hole spacing was: 6m (A1), 3m (A2), 1.5m (B3 and 4) and 0.75m

(C5and 6). For the trial, the A holes were grouted with OPC cement grout, while the B and C holes were

grouted with microfine cement grout. Each hole was grouted in three stages using an inflatable packer.

The three stages were:

Stage 1: Packer is placed 5 m below bottom of the secant pile

Stage 2: Packer is placed 1 m below bottom of the secant pile

Stage 3: Packer is placed just above the bottom of the secant pile

If there was evidence that the unsupported drill hole in the rock was unstable, the upward stage grouting

sequence listed above was reversed, so that a downward stage sequence was followed. Each stage was

terminated when either a maximum grouting pressure (and hold time) was reached first, as shown in

Table 2.

Table 2. Termination criteria for the grouting

OPC

Stage 1

OPC

Stage 2

OPC

Stage 3

MFC

Stage 1

MFC

Stage 2

MFC

Stage 3

Grouting pressure [bar] 25 15 10 25 15 10

Hold time [min] 2 3 4 3 5 7

Max.volume [litres] 400 320 80 400 320 80

If the maximum pressure was reached, grouting would continue for the ‘hold time’, holding the maxi-

mum pressure. This final step was used to gently squeeze grout into the finer fissures accessible to the

grout. The hold time given in Table 1 was calculated based on achieving 3m penetration from the point

of injection. For the microfine grout the calculation was based on a rock fissure with a width of 100 µm.

The ‘hold time’ could be cut short if it was observed that the piston of the pump was not moving over 2

minutes or more.

To assess the results of the grouting trial, water injection (lugeon) tests were carried out at the start of

each phase of the grouting. The last two C6 holes were used as a final test, after all of the other holes

had been grouted.

5.2.2 Quality Assurance during the trial grouting

A critical factor in the effectiveness of the grouting is the ability of the grout to penetrate the fissures in

the rock. The preliminary grout mix tests were carried out to develop grout mixes that would do this. In

order to ensure that the grouts used continued to be effective, a programme of continued quality assur-

ance testing was set up. The routine tests are summarised in Tables 3 and 4. The tests were carried out

each morning and afternoon that there was active grouting, with additional testing when there was a new

delivery of cement.

Table 3: Testing for OPC grout

Test Limiting value Comment

Bleed <10%

Marsh Cone < 55 seconds


10

Table 4: Grout testing, MFC grout

Test Limiting value Comment

Bleed <10%

Marsh Cone <36 seconds

Filter pump, 77μ >200l To target Bcrit of <100μ

Pot life >30 minutes

The grout quality assurance programme was continued throughout the trial and then through

the production grouting that followed.

5.2.3 Results of the grouting trial The grouting trial was carried out between 25

th August and 22

nd September 2010. Figure 10 shows mix-

ing of MFC in progress for the trial.

Figure 10. Mixing in progress for the grouting

The water test results at each phase of the grouting, expressed in terms of lugeon units, are shown in

Figure 11, and the average volume of grout per hole used in each phase is summarised in Table 2.

Lugeon Values based on progress of Grouting works

0.000

5.000

10.000

15.000

20.000

Luge

on

Val

ue

s (l

itre

s/m

/min

/MP

a)

Top (ST3) 19.027 6.656 10.640 11.040 7.860 10.743 4.982

Middle (ST2) 0.947 1.498 1.253 0.630 0.052 2.255 0.040

Bottom (ST1) 0.692 0.226 1.109 0.520 0.005 0.000 0.000

A (before OPC

Grouting)

A (before MFC

Grouting)

B3 (before MFC

grouting)

B4 (before MFC

Grouting)

C5 (before MFC

grouting)

C6 (before MFC

grouting)

C6 (before MFC

grouting) -52,

Figure 11. Results of water (lugeon) tests in successive phases of the grouting trial


11

Table 5. Average volume of grout used per hole, in litres

Average take MFC Average OPC Total

A1 holes 197.70 275.43 473.13

A2 264.20 287.35 551.55

B3 575.27 109.20 684.47

B4 378.95 94.90 473.85

C5 72.74 72.74

C6 128.43 128.43

C6 (last) 8.15 8.15

5.2.4 Assessment of the grouting trial As can be seen from Figure 11 and Table 5, in stages 1 and 2, i.e. the lower 9m of the 10m grout hole,

there was a satisfactory pattern of reducing permeability and grout take as the grouting progressed. The

final results, with a very low residual lugeon value in the rock, showed that the rock had been effectively

grouted.

Although there was the same general pattern of reducing grout take and permeability in the upper (stage

3) grouting, the residual permeability in this stage was much higher than in the lower two stages. There

was clear evidence that a significant amount of soil and drilling fluid was trapped at the toes of the piles,

based on observations during grouting. There was frequent connection between adjacent ducts, even

though the undrilled duct was supposed to be buried in the concrete of the pile. On some holes, (pile)

drilling fluid and sludge trapped at the toes of the piles were driven, by the grouting, up the connecting

hole.

Following the trial, major changes were made to the toe grouting. These changes included:

Thickening the toe grout to a 0.5 : 1 w : c ratio, to increase viscosity and improve the ability of the

grout to flush out sludge from the base of the pile

The maximum grout volume for the toe grouting was increased to 1,000 litres

The maximum volume for the Stage 3 grouting was increased, with an option for thickening the MFC

grout

The drilling for the toe grouting was reduced to 0.5m

Excluding the toe grouting, all stage grouting in the rock was by means of microfine grout. This

change was made to simplify procedures and improve site control.

With these changes, the average volume of grout injected during toe and Stage 3 grouting increased very

significantly, with many holes taking 0.5 to 1m3 of grout in the limited zone just at the toe of the pile.

Despite these massive grout takes, there was generally limited benefit in terms of reducing the measured

permeability at the toes of the piles. This was considered to be due to the presence of the remaining un-

consolidated sediment trapped at the toes of the piles; such sediments cannot be grouted as effectively as

rock fissures. However, drill hole stability improved markedly once the changed toe grouting procedures

were implemented.

During the grouting trial it was found that the two grouting mixes developed in the initial testing pro-

vided an adequate pot life for practical grouting. Although there were incidents of grout setting in the

lines, this was because of a labour force that was entirely new to grouting. Regular testing of the grout

mixes, using the filter pump, also identified some initial problems with equipment cleaning, leading to

flakes of set grout getting into the fresh batch, significantly reducing the penetration capability of the

grout. Regular testing allowed this to be identified quickly, and improved washing practices imple-

mented. During later grouting the filter pump tests also showed where a consignment of grout had been


12

stored on site for too long, despite the storage environment being controlled. Microfine grout from a po-

tential alternative supplier was also tested, and found to have poor penetration properties in comparison

with that used.

6. CONCLUSIONS

The grouting test mixes and grouting trial at Bukit Panjang Station were carried out to assess what

changes needed to be made to typical rock grouting practice, as applied in Sweden, in the very different

climatic conditions in Singapore. The different climatic conditions had a very significant effect on the

grouts, and major adjustments to the planned grout mix and mixing process had to be made to achieve

practicable grouts.

Routine testing of the grouts during the trial and subsequently during injection was necessary to verify

that they had, and continued to have, the required properties. Due to the initial inexperience of the work-

force, changes to simple working procedures, such as the standard and methods for cleaning of the mix-

ing equipment had to be made, in order to ensure grout quality. The regular Quality Assurance testing of

the grout was a key factor in identifying the need for changes to procedures.

REFERENCES

Bruce D.A. (2011). Rock grouting for Dams and the Need to fight Regressive Thinking. Geotechnical News,

vol 29, Number 2, June 2011, pp 36 to 43

Defence, Science and Technology Agency (2009). Geology of Singapore, 2nd Edition

AECOM (2008). Geotechnical interpretative Report for Contract 912, Rev C.


13

1 INTRODUCTION The road tunnel is proposed to be constructed in the Chugoku Mountains in the central part of Hiroshi-ma Prefecture, Japan. There are numerous water sources being used for irrigation of paddy field around the tunnel. The water sources are streams, reservoirs and wells located above the proposed tunnel level. Since the tunnel is designed to be non-watertight NATM tunnel, it is expected that the tunnel construc-tion could cause water loss in use for irrigation. To study the influence by the tunnel construction to the irrigation, a hydrological simulation using 3D terrestrial fluid flow programme was conducted. This pa-per presents the approach and results of the 3D simulation. 2 PROPOSED TUNNEL AND HYDROGEOLOGICAL CONDITION 2.1 Proposed Tunnel and Water Sources for Irrigation The tunnel length is 2.1km and diameter is 12.4m. The tunnel is planned to run under 300 to 500m high hills and the maximum overburden is 180m. Several streams are crossing the tunnel route and wells and reservoirs are located near the tunnel. The water from these sources is extensively used for irrigation in a period from April to September every year. Figure 1 shows the tunnel alignment and water sources. The amount of water usage was investigated for each source by direct measurement and/or estimation based on the area of paddy field. 2.2 Hydrogeology The geology around the tunnel was revealed by the geological investigation using exploratory drilling, electrical resistivity and seismic refraction survey. The basic geology is Mesozoic Rhyolite and Grano-Porphyry. The rocks are generally strong but intensively weathered into residual soils near the ground surface. Recent river and talus deposits cover the gentle slopes at the foot of hills, where the tunnel en-trances/exits are located. The geological investigation and aero-photographic analysis revealed 18 faults and most of faults are located across the tunnel as shown in Figure 2.

Three-Dimensional terrestrial fluid analysis for assessment of water loss due to tunneling

Y. Goto Hiroshima Prefecture, Japan

K. Orihara

Kiso-Jiban Consultants Co’Ltd, Singapore

T. Sadamura, Y. Huang, K. Miura Kiso-Jiban Consultants Co’Ltd, Japan

ABSTRACT: The proposed 2.1km long road tunnel is located in a hilly area in Japan where the surface water and groundwater are used extensively for irrigation. The non-watertight tunnel is designed to run mainly through hard competent rhyolitic rocks with low permeability. However in the geological investi-gation, high permeable faults were found crossing the tunnel route. Significant amounts of water ingress into tunnel and water loss are expected. A 3D terrestrial fluid flow programme with coupling of surface water and groundwater was used to estimate the water loss due to tunneling. The rainfall, evaporation, surface water, infiltration and groundwater are simulated in the 3D hydrogeological model. The water loss in the individual streams, wells and reservoirs are estimated as well as the amount of water ingress along the tunnel. The results of analysis indicated the necessity of countermeasures to supplement water loss for irrigation.


14

Figure 1. Tunnel Route and Water Sources for Irrigation

⑧to ⑯ denote faults

D, CM and CH are rock mass classification.

Figure 2. Electrical Resistivity, Seismic Tomography and Geological Profiles along Northern Half of Tunnel

Well No.H20-4

Well No.H20-4

Seismic Tomography

(km/sec)

Electrical Resistivity

Tomography (Ωm)

Interpreted Geological Profile

Tunnel

Tunnel

Rhyolite CH

CM

CH CH CH CH

CM CM

CM

D

D

Grano-Porphyry

New talus deposit


15

Table 1 summarizes geological units and coefficients of permeability. The permeability of fault varies in a rage from 10

-6 to 10

-8 m/sec as confirmed by the limited numbers of in-situ tests. The faults are gener-

ally associated with highly fractured rocks and form relatively high permeable aquifers. Table 1. Geological Units and Coefficients of Permeability. _________________________________________________________________________

Geological Unit Soil/Rock Type Coef. of Permeability (m/sec) _________________________________________________________________________

Surface soil Sandy soil 5x10-6 River deposit Gravelly soil 1x10-5 New talus deposit Gravelly sand 1x10-5 Old talus deposit Clayey soil 1x10-7 Rhyolite/Grano-Porphyry *Residual soil Clayey soil 1x10-8

*Weathered/Fresh Hard rock 1x10 -9 Fault Fractured rock 1x10-6 to 5x10 -8 ________________________________________________________________________

3 COMPUTER SIMULATION 3.1 Programme”GETFLOWS” The 3D programme “GETFLOWS” was used for simulation. The programme was developed by Tosaka et al. (1996) to simulate fluid flows with coupling of surface water and groundwater. The rainfall, eva-poration, surface water in streams, ponds, water wells, reservoirs, groundwater, saturated or unsatu-rated seepage flows, etc., are simulated. 3.2 Models The 3D model consists of a total of 440,000 elements created according to geometry and hydrogeologi-cal structures as shown in Figure 3. The model covers approximately an area of 3.5km by 3.5km. The tunnel is located at the center of the model. The boundaries are set along the rivers/valleys at the lower levels than the tunnel level in order for boundary condition not to affect results. The vertical elements form the soil and rock layers with various coefficients of permeability. The vertical elements include the first layer of atmosphere and the second layer of surface water running with the Grauckler-Manning coefficients.

Figure 3. 3D Model Mesh


16

Figure 4 shows a model section along the tunnel. The faults are modeled as closely as possible to their extent revealed by the geological investigation.

① to ⑯ denote faults

Figure 4. Model Section along Tunnel

3.3 Simulate Current Conditions (Calibration) Prior to the tunnel construction simulation, the 3D model is calibrated by comparing the calculated val-ues to the observed data. The stream flows, water levels in wells and reservoirs were observed for a year from December, 2008 to November, 2009. The observed flows and fluctuation are simulated in the model using actual rainfall during the observation period. An initial run and a sensitivity analysis indi-cated that the most influential parameters to the result are a coefficient of permeability of surface soil, the Glauckler-Manning coefficients, and local geometry. These parameters and input data were adjusted until the square of correlation coefficient of 0.9 or greater is obtained between the observation and cal-culation. Table 2 presents typical values used in the parametric study. To find out the final parameters in Case 2, the parameters adjustment was repeated several cases in addition to those shown in Table 2. Table 2. Parametric Study. _________________________________________________________________________

Parameters Initial Case 1 Case 2 (Final) _________________________________________________________________________

Coefficient of Permeability 5x10-6 5x10-6 1x10-6 of Surface Soil (m/sec) Glauckler-Manning Coefficient

*Forest 0.8 1.0 1.0 *Agriculture land 0.4 0.5 0.5 *Other area 0.2 0.5 0.5

________________________________________________________________________

The evaporation is also one of important factor in the simulation and the evaporation rate is obtained by the following Harmon’s equation;

Ev = 0.14D02Pt,

where Ev = amount of daily average evaporation (mm/day); D0 = sunlight rate (12hrs/day); and Pt = sa-turated absolute humidity related to temperature (g/m

3).

Figure 5 compares the observed and calculated water flow in the stream No.B-2. The calculated flows agree reasonably well with the observed flow except at the large flows in June, July and August, 2009. The discrepancy at the large flows may be because the large flow is influenced by a localized heavy rainfall on the catchment area of stream. The rainfall on the catchment area of each stream may not be the same amount of rainfall in the meteorological station. In the simulation, the rain is assumed to fall


17

uniformly over the entire ground surface of the model. Therefore, the square of correlation coefficient of 0.9 is obtained only after neglecting the large flows. This is acceptable because the small volume is more important in this study for water loss. There is no significant difference between Cases 1 and 2 with different coefficients of permeability of surface soil. Figure 6 shows groundwater level in a well No.H20-4, which was constructed in the exploratory bore-hole near the stream No.C-2. The 90m deep borehole No.H20-4 was sunk to the tunnel invert level and the well strainer was provided at the tunnel level. The calculated groundwater fluctuation is fairly rea-sonable as a square of correlation coefficient is 0.98.

0

500

1000

1500

2000

2500

3000

3500

4000

08/11 08/12 09/01 09/02 09/03 09/04 09/05 09/06 09/07 09/08 09/09 09/10

Month

ly

Ave

rage

Flo

w (

liter/

min

)

Observation

Case1

Case2

Figure 5. Observed and Calculated Flows in the Stream No.B-2.

-5.0

-4.0

-3.0

-2.0

-1.0

0.0

1.0

08/11 08/12 09/01 09/02 09/03 09/04 09/05 09/06 09/07 09/08 09/09 09/10

Month

ly A

vera

ge L

eve

l (m

)

Observation

Case1

Case2

Figure 6. Observed and Calculated Groundwater Levels in Well No. H20-4.

4 SIMULATE TUNNEL CONSTRUCTION 4.1 Rainfall and Simulation Cases The simulation cases are selected based on the rainfall record in the last ten years from 2000 to 2009 collected at the nearest meteorological station. The cases include 10 years average rainfall, poor and rich rainfall as shown in Table 3. The year 2005 is selected for the study of water loss after focusing on the smallest rainfall in the irrigation period of April to September although the annual rainfall of 1382mm is


18

not the smallest in the last 10 years. For the study of water ingress into tunnel during excavation, we se-lect the rich rainfall year of 2006 with the maximum annual rainfall of 2038mm in the last 10 years. Figure 7 shows 10 years average daily rainfall and evaporation rate in each month used in the simula-tion.

Table 3. Tunnel Construction Simulation Cases _____________________________________________________________________________________

Case Rainfall Tunneling Purpose _____________________________________________________________________________________

3 10 years average before compare with Case 4 4 10 years average after assess water loss in general year

5 poor in 2005 before compare with Case 6 6 poor in 2005 after assess water loss in poor year 7 rich in 2006 during assess water ingress into tunnel _____________________________________________________________________________________

Figure 7. 10 years Average Daily Rainfall and Evaporation Rate in Each Month

4.2 Calculated Water Loss The water flows in the irrigation period from April to September are summarized in Table 4 before and after tunneling under the 10 year average rainfall. The water usage for irrigation is also shown in the ta-ble. The estimated water loss is small and the stream can supply enough amounts of water more than the usage. The rate of reduction varies from 1 to 25% and the maximum reduction occurs in the stream No.C-2 which has the largest flow. Remarkably, the major faults are located along the stream No.C-2. Table 4. Water Loss in Streams due to Tunneling under 10 years Average Rainfall (Cases 3 and 4) _____________________________________________________________________________

Stream Water Usage Before Tunneling After Tunneling Shortage No. for Irrigation Min. Flow Max. Loss Min. Flow occurs

_____________________________________________________________________________

A 51 72 4 (4%) 69 no B-2 270 618 67 (10%) 562 no

C-2 7 653 204 (26%) 486 no E 38 107 24 (17%) 89 no F 25 34 4 (9%) 30 no I 0.06 4 0.1 (1%) 4 no Q 31 72 2 (2%) 70 no _____________________________________________________________________________

Unit in liter/min

Table 5 presents the results of poor rainfall year of 2005. Even before the tunneling, the shortage of wa-ter is to take place in all the streams except at streams No.C-2 and I. It should be noted that the water of


19

stream No.C-2 is once kept in several reservoirs shown in Figure 1. The small water usage of C-2 in the table is based on the water usage directly from the stream, while main water usage relies on the reser-voirs in this region. The amount of water usage from the reservoirs is unknown as the water flow and usage among the reservoirs were complicated.

Table 5. Water Loss in Streams due to Tunneling under Poor Rainfall in 2005 (Cases 5 and 6) _____________________________________________________________________________

Stream Water Usage Before Tunneling After Tunneling Shortage No. for Irrigation Min. Flow Max. Loss Min. Flow occurs

_____________________________________________________________________________

A 51 13 8 (15%) 11 yes B-2 270 154 123 (24%) 116 yes

C-2 7 130 349 (81%) 25 no E 38 24 39 (36%) 15 yes F 25 1 9 (67%) 1 yes I 0.06 0.6 0.3 (24%) 0.6 no Q 31 5 4 (13%) 4 yes _____________________________________________________________________________

Unit in liter/min

Table 6 shows groundwater drawdown in wells after tunneling under 10 years average rainfall. The change is as small as 0.12m in water level and practically not significant for the usage. Table 6. Groundwater Drawdown in Wells due to Tunneling under 10 years Average Rainfall (Cases 3 and 4) ________________________________________

Well No. Water Drawdown in m ________________________________________

3 0.01 to 0.05 21 0.03 to 0.12

________________________________________

4.3 Calculated Water Ingress into Tunnel The tunnel excavation is simulated in accordance with the construction schedule in Figure 8.

Figure 8. Tunnel Excavation in Two Drives

The tunnel excavation is assumed to commence from two entrances in January. The commencement month was selected to avoid the tunnel passing under major streams in the irrigation period of April to September based on the tunnel advance speed so that the water loss are expected to be reduced as much as possible. Figure 9 shows the calculated unit water ingress during excavation in transient condition under the rich rainfall year of 2006 with an annual rainfall of 2038mm (Case 7). The water ingress increases drastical-ly when the tunnel passes through high permeable faults as expected. The maximum unit water ingress of 4.5 liter/min/m is registered under the stream No.C-2. Figure 10 shows the water ingress after con-


20

struction of non-watertight tunnel in steady state condition. As expected, the unit water ingress is re-duced by about 30% as compared to transient condition.

Figure 9. Water Ingress into Tunnel during Excavation in Transient Condition

Figure 10. Water Ingress into Non-watertight Tunnel after Construction in Steady State Condition

A total amount of water ingress after construction is estimated to be 340 liter/min as shown in Figure 11. Conventionally in Japan, the total water ingress in steady state is calculated based on the hydrological method proposed by Takahashi (1965). The so-called Takahashi’s method is based that the water in-gress in steady state relates to the stream flow within an influence zone of tunnel in the dry season. The total water ingress by the Takahashi’s method is calculated to be 218 liter/min based on the observed water flow in May, 2009, which is about 65% of the amounts calculated by 3D analysis. The rainfall in the dry season of May, 2009 was extremely low, only 30% of the last 10 years average rainfall in the same month. Therefore, the total water ingress of 218 liter/min by Takahashi’s method is probably un-derestimated. For the comparison, the water ingress is calculated under the poor rainfall year of 2005 with an annual rainfall of 1382mm. The calculated total water ingress is 335 liter/min, which is about the same as the value obtained in rich rainfall year of 2006.


21

Figure 11. Accumulative Water Ingress into Non-watertight Tunnel after Commencement of Excavation

5 COUNTERMEASURE The water loss exceeds the demand in water usage for irrigation in the poor rainfall year in 2005. Partial watertight segments are proposed at fault zones to cut off the water ingress and to reduce water loss. The partial watertight segments are simulated in 3D analysis as presented in Table 7. Watertight seg-ments cover major fault zones under the streams No. B-2 and C-2 in Case 8, while the watertight seg-ments are extended to a width of catchment area of each stream in Case 9. Table 8 shows effectiveness of watertight segments with various section lengths. To recover the water flow to 90% of the amount be-fore tunneling, watertight segments are required for about 350m for each stream. Table 7. Analysis Cases of Watertight Segments ___________________________________________________________

Case Rainfall Watertight Section (m) Stream No. B-2 Stream No. C-2 ___________________________________________________________

8 poor in 2005 94 67 9 poor in 2005 358 347 ___________________________________________________________

Table 8. Effectiveness of Watertight Segments ___________________________________________________________________

Stream No. Case Watertight Section (m) min. Flow (liter/min) ___________________________________________________________________

B-2 5 before tunneling 154 (100%) 6 0 116 (75%)

8 94 125 (81%) 9 358 150 (97%)

C-2 5 before tunneling 130 (100%) 6 0 25 (19%)

8 67 58 (45%) 9 347 114 (88%) ____________________________________________________________________

The following countermeasures are considered to overcome the water loss problems including the water-tight segment. New reservoir Water tank at tunnel entrance to utilize seepage water from tunnel Partial water tight segments The countermeasure will be selected after the consideration of cost, users’ convenience and compensa-tion.


22

6 CONCLUSION The 3D simulation is useful to assess water problems in the tunnel design. The water loss due to tunne-ling can be assessed in the individual water sources. The water ingress through the high permeable aqui-fer of fault is well demonstrated. The calibration can increase the credibility of the 3D analysis. Howev-er, the calibration has some limitations in the case that the ground water flow is changed from horizontal to vertical direction through localized aquifers of faults due to tunneling. The calibration method may require some improvement in such a case. The detail tunnel design is now in progress and construction works will start soon. The observation of stream, wells, and reservoirs will continue. The water ingress will be recorded in the progress of tunnel excavation. These observed data will enhance the accuracy of 3D simulation to be carried out during excavation. It will result in better estimation of foregoing water ingress and water loss. 7 ACKNOWLEDGEMENT Authors are grateful to Mr. Tameto and Mr. Sunakawa of Kiso-Jiban Consultants Co’Ltd, Chugoku Branch for their assistance in conducting this study and to Mr. T. S. Chua of Kiso-Jiban Singapore Pte Ltd for his review of this paper. REFERENCES Takahashi, H. 1965, Characteristics and Problems of Tunnel Seepage Water in Japanese. Journal of Japan So-

ciety of Engineering Geology: Volume 6 No.1, 25-52 Tosaka, H. Kojima,K. Miki, A. Senno, T. 1996. Development of 3D Terrestrial Fluid Flow Simulation with

Coupling of Surface Water and Ground Water in Japanese. Journal of Japanese Association of Groundwa-ter Hydrology, Volume 38 No.4, 253-267


23

ABSTRACT

The use of prefabricated vertical drain (PVD) is an established method for soil improvement in soil with very weak engineering properties. This paper primarily discusses the effectiveness of soil surcharging with and without PVD in three cases of soil condition within the Kallang Formation and Bukit Timah (BT) Formation: the first being surcharging with PVD in inland Kallang Formation. The second case considers surcharging with PVD in BT Formation only and the third case being surcharging on top of residual soil without the use of PVD. This paper will also focus on theoretical calculation compared against site performance and finally on the effectiveness of large spaced PVD.

1. INTRODUCTION

The location of the Depot is along Old Woodlands Road, opposite Stagmont Ring Road. Presence of terraced slopes suggests presence of farming previously, with Old Woodlands Road being the base of the valley. An abandoned quarry is located behind the proposed Depot, with parts of the site being used by the Singapore Armed Forces. The rest of the proposed Depot was used as a cemetery previously. Figure 1 shows the process of placing excess surcharge for consolidation of the ground.

Figure 1- Location of the site with surcharging works in progress

Existing borelogs around the site indicated pockets of Kallang Formation, including Fluvial Sands & Clays and Estuarine Clay. Ground improvement is suggested to consolidate these compressible soils,

Effectiveness of large spaced PVD surcharging for inland

Kallang formation and residual soil for a MRT depot

K.F. Wong, H.T. Lim, S.C. Tan CPG Consultants Pte Ltd, Singapore

E.T. Hong Land Transport Authority (LTA), Singapore

W. Maw GS-HLS (JV)


24

considering that differential settlement caused by train load will be detrimental to the Mass Rapid Transit (MRT) train operation in the long term. The most economical method of ground improvement therefore will be excess temporary surcharge with PVD to accelerate consolidation settlements due to the MRT train load. This eliminates further settlement when the MRT train starts operation on the allocated site.

Advanced studies evaluated that existing soft soils in the Kallang Formation would easily compress under increased loading due to primary consolidation, and the time period for the soil to fully consolidate will be long due to its low permeabilities. Similarly, the soft to firm residuals soils have low sand content, with relatively high compression indices. These soft layers have low SPT values of 7 blow counts Additional loading onto the soft residuals soils would result in large settlements, over a long period of time. It was recommended to carry out ground treatment measures to expedite the consolidation due to additional loads, using vertical drains and surcharging where large thickness of the residual soil are encountered.

2. GEOLOGY OF THE SITE

Soil Investigation (SI) works indicates the uppermost layer to be Fill in general, with Fluvial Sands, Fluvial Clay and Peaty Clay underneath. The bottom most layer are residual soil of BT Formation, with rock fragments spread across the site. Fluvial Clay and Residual Soil of BT Formation are predominantly silt with varying amount of sand. The thickest Kallang Formation is located at the proposed Stabling Yard area (borehole DT/2229), containing up to 10m of Fluvial Sand and Fluvial Clay.

Figure 2 Location of the thickest Kallang Formation


25

Figure 3 Geological Profile of ground at location with thickest Kallang Formation

3. METHODOLOGY

Preliminary settlement analysis was based on borehole DT 2229 which has the thickest Kallang Formation, approximately 10 metres. Analyses were then refined based on surrounding clusters, to give an estimated settlement. Clusters are defined as grid squares of 40 m wide. An excess surcharge of 2m was placed on site which is in excess of the train load, to accelerate consolidation. PVD was to be placed in a 2.7m triangular spacing with the time period set to be 9 months, with a layer of sand blanket of 300mm thick under the temporary surcharge. The presence of the sand layer facilitates the dissipation of the excess pressure in the Kallang Formation under the static weight of the temporary surcharge. The spacing of the PVD is relatively large.

Figure 4 Cross section of ground showing ground improvement using excess surcharge with PVD

The PVDs must be embedded into firmer residual soils of BT Formation for the consolidation (compressible Kallang Formation and residual soils of SPT N values less than 10) to be effective. The length of the PVD is taken to be either 3m below the toe of the Kallang Formation or 3m below residual soil of SPT N value of 10 blow counts. However, the termination depths of the PVDs can be shorter if the gauge in the mandrel rig hits its maximum pressure of 250 kN. The coefficient of horizontal consolidation (Ch) is assumed to be 1.5 times of vertical consolidation (Cv). The rate of consolidation is important in surcharging works; tapering of settlement implies that excess surcharge can be removed. The rate of consolidation will also assess the behavior of soil


26

(whether there is residual settlement) upon removal of the surcharge. Ground improvement along DSTA Road is to be done without the use of PVD for the purpose of determining the effectiveness of PVD in Kallang Formation and residual soils of BT Formation. 4. FIELD RESULTS

Field methods of both Asoka’s method (1976) and Hyperbolic method (Tan, 1995) were used to assess the effectiveness of soil surcharging. Readings were taken on a daily basis for the first 3 months, after which, were reduced to 3 times per week. Data were plotted on a 10-day scale, to predict the final settlement. Soil surcharging is considered effective if the 90% of the targeted consolidation (due to train load) was achieved. Surcharge is to be removed when the residual settlement is less than 10% of the total settlement. Field data indicates settlement to be significantly lesser than preliminary analyses. The following sections discuss the differences between theoretical and field settlements; and the effectiveness of PVD in residual soil of BT Formation. 5. BACK-ANALYSIS

The rate of settlement is consistent with the suggested parameters. In the case of the Kallang Formation, the coefficient of vertical consolidation (Cv) is approximately 2m

2/yr and the coefficient of

horizontal consolidation (Ch) is approximately 3m2/yr. Solutions from Barron’s theory (R.F. Craig,

2001) and field methods from Victor Choa (VM) (M.W Bo & V Choa, 2004) were also plotted against field data to determine the rate of consolidation. 5.1. Effectiveness of PVD (Kallang Formation)

Figure 5- Settlement time graph at Area 2 (Kallang Formation)

-50

-40

-30

-20

-10

0

0 50 100 150 200 250 300

Sett

lem

en

t (m

m)

Time (days)

Settlement (mm) vs Time (days) graph

Area 2 SM 1

Design Settlement with PVD-Barron Field Settlement with PVD

Design Settlement with PVD- VM


27

Figure 6 Settlement Time Graph at Area 3 (Kallang Formation)

The figures in Figure 5 and Figure 6 shows the comparison of the field settlement compared to the predicated settlement. The settlement time graphs between the theoretical solutions (graphical and formulae methods) are consistent with each other. Matching theoretical settlements with field settlements results in compression ratio at Kallang Formation to have Cc/(1+e0) values less than 0.2. This implies that the actual ground is not as compressible as suggested in the baseline report. Piezometer readings in this ground improvement area did not capture a rise is pore water pressure upon placement of the excess surcharge. This could be due to the presence of fluvial sand in the ground, resulting in quick dissipation of the pore pressure.

5.2. Effectiveness of PVD (residual soil of BT Formation)

Figure 7 Settlement time graph at Area 2 (residual soil of Bukit Timah Formation)

Figure 7 shows the location at Area 2 where there are 2 types of soil improvement: Surcharging with PVD and without PVD. Surcharging without PVD is represented by SM 45, SM 46 and SM 48, while SM 4 and SM 5 are at locations with excess surcharge and PVD. PVD is extremely effective in accelerating consolidation settlements in the initial period, as shown by the steep gradient in Figure 7. Settlement readings in the PVD region taper off approximately after 140

-80.0

-60.0

-40.0

-20.0

0.0

0 50 100 150 200 250 300

Sett

lem

en

t (m

m)

Time (days)

Settlement (mm) vs Time (days) graph

Area 3 SM 16

Field Settlement Design Settlement- VM 1

Design Settlement- Barron

-12.00

-10.00

-8.00

-6.00

-4.00

-2.00

0.00

0 50 100 150 200 250 300

Sett

lem

ent (

mm

)

Time (days)

Settlement (mm) vs Time (days)

SM4 SM5 SM45 SM46 SM48


28

days. Readings at regions without PVD were settling gradually over a time period of 200 days. Consolidation settlement tapers off approximately after 180 days. 6. PLATE LOADING TESTS

Plate loading tests were also proposed to determine the bearing capacity of the ground after the completion of surcharging.

Figure 8 Plate Loading Test at Stabling Yard

The Figure 8 above shows the pressure induced onto the soil at 300 kPa. The maximum settlement is 4.08 mm. The modulus of sub-grade reaction ks is given by:

ks= q/δ (1)

where q is the pressure induced onto the soil and δ is the displacement. The modulus of sub-grade reaction from the above test indicates that ks is approximately 73529 kN/m

3.

The modulus of both the soil and sub-grade reaction is correlated by:

(2)

where Es is the modulus of soil and B is the width of the plate.

From the above test, Es is approximately 20000 kN/m2. The increment in the soil’s Young modulus

can be attributed to ground improvement.

7. DISCUSSION

The coefficient of consolidation (Cv , Ch) for residual soils and Kallang Formation are consistent with the design geotechnical parameters in the baseline report. This is shown in the trend of consolidation settlements plotted in Figures 5, 6 and 7. The rate of consolidation for Kallang Formation tapers after 180 days in both the design and field settlement data, as shown in Figure 5 and Figure 6. This supports the design parameter of Cv and Ch for Kallang Formation being 2m

2/yr and 3m

2/yr.

The compression ratio for Kallang Formation and residual soil of BT Formation are not as compressible as the value of 0.2, given in the baseline report. The value of 0.2 given in the baseline report is on the conservative side. Applications of PVD have been used in many projects. One of the more critical factors for consideration is the time period allowed for surcharge. From Figure 6, it can be observed that with the use of PVD, the rate of consolidation of the Kallang Formation is increased. It is recommended to


29

consider a combination of excess surcharge and prefabricated vertical drains when improving these grounds. For residual soil of BT formation, large spaced PVDs are effective in accelerating consolidation settlements. However, the difference in the rate of consolidation is not significant after 180 days. Thus, one may argue about the necessity of vertical drains for a time period given 9 months to achieve 90% consolidation of the design load. For a time period of less than 180 days, ground improvement with PVD and excess surcharge should be considered. 8. CONCLUSION Data from settlement markers placed at Area 2 (DSTA road) suggests that for a surcharge period of 9 months, ground improvement using both excess surcharge with PVD may not be necessary after 180 days to eliminate consolidation settlement caused by train load. Excess surcharge and PVD are necessary if the surcharge period is limited to 180 days. The trend in consolidation suggests that the coefficient of consolidation (Cv and Ch) are close to the

design parameter of 2m2/yr and 3m

2/yr. The increment in the soil’s Young modulus can be attributed

to ground improvement.

REFERENCES.

M.W Bo & V Choa, 2004. Reclamation and Ground Improvement. Pg 231- 239. R.F. Craig. Soil Mechanics. Sixth Edition 2001, Pg. 248 - 297


30

1 INTRODUCTION

The first double rotary drilling method was developed around 30 years ago for the installation of deep excavations close to existing structures in inner cities. The main advantage was to minimize the dis-tance between the existing building and the required retaining wall. The "Front of Wall" (FOW) me-thod used two separate rotary heads for the auger and the casing, which are rotating opposite direc-tions simultaneously. Initially the system was limited to relatively small diameters. The development of larger and more powerful drilling rigs has allowed the double rotary system to install cased piles with a diameter of up to 1200 mm for various applications, namely King Post, contiguous and secant pile walls and discreet foundation piles.

State of the art piling technique

G.U. Ulrich & P.P. Platzek BAUER Maschinen, Germany

ABSTRACT: In the changing construction environment clients are searching for new technology to complete their projects in shorter times. Manufacturers are therefore developing solutions to speed up established construction methods. The significant progress in recent years with the development of increasingly powerful drilling rigs is providing piling companies increase options for the execution of specialist foundation projects. The double rotary drilling methods are primarily used for the construction of pile walls as support for deep excavations. Typically these retaining walls are designed and installed as cased piles using the es-tablished Kelly drilling mode. However, more recently it is more popular to undertake these projects with the Cased Continuous Flight Auger (CCFA) system. This paper provides an overview of drilling with double rotary methods, in particular the monitoring during installation.


31

2 CONSTRUCTION PROCEDURE

Figure 1. Working steps for CCFA Piling

3 DOUBLE ROTARY DRIVE SYSTEMS

The fundamental principle of double rotary drive systems is a continuous flight auger in combination with an outer casing, that are drilled simultaneously, but in the opposite direction, into the ground. These characteristics resulted to the common term Cased Continuous Flight Auger Pile (CCFA). The system allows relative movement between the casing and the auger of +/- 300 mm. Depending on the soil conditions the auger can drill in advance or within the casing. Normally the casing needs to be used in advance or at the same depth to stabilize the surrounding soil. But in case of drilling through obstacles or hard, stable layers the auger is in advance of the casing. The pile spoil is transported upwards by the auger flights surrounded by the casing and exits through openings at the top of the casing. The pile is formed by pumping concrete through the hollow stem of the auger during withdrawal, whilst monitoring and the concrete pressure and volume. After cleaning the pile head a reinforcement cage is positioned centrally over the pile and is pushed or vibrated into the fresh pile, all as in the established flight auger technique. Generally the double rotary drive systems can be classified into the following main types:

Figure 2. Overview in CCFA methods


32

The advantages of CFA piles and cased bore piles are combined in the double rotary system. The in-stallation rates of CFA piling and the high quality standards of traditional cased piles in terms of bore-hole stability and verticality are achieved.

4 FOW - SYSTEM

A major issue in the construction of retaining walls adjacent to existing buildings is the loss of space. Standard piling rigs require typically 1000 mm from an adjacent wall to the center of the piled wall due to attachments at the rotary drive. The FOW system allows the installation of piled walls in con-fined areas much closer to existing structures.

Figure 3. drilling rig with attached FOW equipment

The rotary drive is optimized so there are no components outside the drilling diameter. This results in a distance of around 400 mm between the building and the center of the piled wall.

5 "CASED CFA" SYSTEM WITH TWO SEPARATE ROTARY DRIVES

In the mid-90s more powerful drilling rigs with two discreet rotary drives were introduced which were capable of installing for the execution of CCFA piles of diameters up to 1200 mm.


33

Figure 4. BAUER BG 42 with two rotary drives

The system worked very efficiently, achieving high installation rates. The main disadvantage was the costly installation of the second rotary drive with additional hydraulics and the difficulty of converting back to conventional drilling modes in terms of time and cost.

6 "CASED CFA" SYSTEM WITH TORQUE MULTIPLIER BTM

In view of this further developments of the drilling method were investigated. The result was the in-troduction of a torque multiplier in combination with a spoil discharge chute system. The Bauer Torque Multiplier (BTM) is a mechanical unit attached to the cardanic joint below the rota-ry drive. The ratio is fixed at 1:2, so that the BTM doubles the incoming torque, halves the speed and inverts the direction of rotation of the casing. The resultant has to be limited to remain within the structural capacity of the mast of the drilling rig. The incoming torque from the rotary drive needs to be limited to suit the specific application and the bearing capacity of the mast.

Figure 5. BAUER BG 36 with torque multiplier BTM 400 and spoil discharge chute system


34

As with other double rotary system the spoil exits the top of the casings after a certain depth. To man-age and reduce this risk the spoil discharge chute system was developed. When the spoil reaches the top of the casing it enters an oversized collector. It is then sweeped round to an opening and then falls through a series of telescoping chutes to the ground in a controlled manner. During the drilling process the chutes are withdrawn to maintain a safe drop height for the spoil. To achieve the maximum drilling depth, the spoil discharge chute system needs to be temporarily moved. This is achieved by the use of a manipulator arm as shown in figure 5. An additional improvement has been the introduction of air to the drilling process, via an air valve lo-cated on the concrete swivel. While drilling down air is pumped through the hollow stem to the tip of the auger. The air has several benefits - it prevents water or soil ingress from the base. Additionally it has a drying effect on the spoil and assists the transportation of the spoil up the augers. The conversion to this drilling mode is also relatively straightforward with only additional equipment being added without major alterations to the structure, hydraulics or electrics of a standard BG rig. Another straightforward application of the BTM system is for single column soil mixing, where lower torque and increased rotation is required. In this case the BTM would be attached to the cardanic joint upside down. Therefore the BTM would double the incoming speed and halve the torque.

7 QUALITY CONTROL SYSTEM OF CCFA METHOD FOR SECANT PILE WALLS

The CCFA method is mainly used for the installation of secant pile walls. The rate of installation re-duces many of the sequence issues experienced with traditional methods. The quality standards for pile walls are necessarily very high as the costs of repairs would be significant and time consuming. It is therefore essential to employ an accurate quality surveillance system

Figure 6. typical guide wall

As primary requirements a stable working platform with less than 3% inclination and a guide wall (see figure 6) are highly recommended. The guide wall ensures the correct starting location of every pile and facilitates the set up. The verti-cality of the casing is controlled manually with a spirit level, however, the verticality of the mast is both measured and recorded by the electronic sensors. Verticality corrections are easily made in the x & y directions either manually or automatically ensuring that the masts and tools are kept vertical at all times. For quality control all drilling data and data relative to the verticality of the mast are availa-ble in real time.


35

t

D

d

D

s

2

1

8 DRILLING ASSISTANT CCFA

Many factors contribute to the efficient drilling and subsequent concreting of CCFA piles. To assist and allow the operator to monitor all the various parameters the drill and extraction assistant programs have been developed. During the drilling phase the optimal performance is affected by the applied tor-que and crowd forces and rotation speed. These will need to be varied for each particular soil condi-tions encountered. In addition to these rig related factors the geometry of the casing and augers need also have an influence. All these factors are used in the drill assistant program to achieve the desired optimal performance, which is measured as the penetration per revolution. Control of this measure is widely accepted as a major factor in avoiding problem in CFA and CCFA piling. In particular there are the following benefits:

avoids over-flighting

avoids corkscrew of auger

optimizes the filling grade of the auger

assists of operator

The exact details of the used drilling tools must be entered into the system:

Figure 7. input data sheet

Different ground conditions require consideration of the filling of the auger during drilling. The pene-tration rate per revolution is calculated using the following parameters: d = diameter of the mandrel D1 = outside diameter of casing D2 = outside diameter of auger Ψ = filling grade of auger s = auger pitch t = penetration depth

Figure 8. geometrical consideration of the penetration rate


36

9 EXTRACTION ASSISTANT FOR CCFA

An efficient concreting process is achieved by monitoring a combination of the concrete volume and pressure and the extraction rate.

10 MEASUREMENT OF CONCRETE PRESSURE AND QUANTITY

There are two basic methods of measuring the volume of concrete supplied by a concrete pump. The first is the direct method where the volume is measured by a flow meter placed in the concrete pipe-work. It is a reliable method mainly used for grout with fine aggregates and a small diameter of the concrete line. Another method is to measure each pass of the concrete pump. This is done with a pressure gauge in the concrete pipe at the drilling rig. A typical graph of the concrete pressure within the pipe is:

Figure 9. recorded concrete pressure for counting strokes during the run-up of a concrete pump

The single strokes can be distinguished clearly by a characteristic rise and drop of the pressure. It is also a common approach to count the stroke directly at the concrete pump. This is done by the electrical system of the pump itself or with a distance sensor at the hydraulic cylinders of the S-valve. Then the signal is transferred to the drilling rig by radio signal. The amount of concrete that is pumped per stroke needs to be measured on site and entered into the software The advantage of this system is its simplicity. Any considerable mistake can be encountered easily on site. Each stroke that is not counted or counted too much is shown as an abrupt discontinuity of the concrete flow. Therefore it is important to keep the speed of the concrete pump constant. Now the extraction assistant for the concreting process has sufficient information to guide the machine for the casting of the pile. The retraction speed is calculated by volume formulas. A significant over consumption of concrete depending on ground conditions is adjusted in the computer program. The working screen of the machine software shows the operator all important information of the produc-tion process:

Figure 10. working screen of the extraction assistant


37

An additional pressure gauge is measuring the concrete pressure at the concrete swivel head to double check the concrete flow and giving a warning signal in case of overpressure.

11 QUALITY AND CONSUMPTION OF CONCRETE

The concrete needs to fulfill the particular strength specifications for secant pile walls, but the consis-tency of the concrete must fulfill special requirements, too. The high flowability ensures a smooth pumping process and facilitates a successful installation of the cage. The concrete must not emit water when it gets in contact with soil. The use of pulverized fuel ash or furnace slag in the concrete mix re-duces the tendency for segregation and bleeding. To avoid any interruptions of the concreting, it should be ensured that the amount of concrete to com-plete the pile is on site before any pile is started. The over consumption of concrete depends on the soil conditions and the appropriate drilling process. For the CCFA piles it should be within a range of 5 - 15% of the theoretical volume. The over consumption is between the CFA pile and the standard cased pile. The consumption of concrete should be inspected accurately, especially if it changes abruptly. Any unexpected changes are an indication for a failure in the drilling process or unexpected cavities in the ground.

12 DRILLING TOOLS AND VERTICALITY

The drilling tools consist of continuous flight auger over the full length. Especially for secant pile walls a double flight start auger, that is equipped with round shaft chisels, is used. One (FOW) or two walled (CCFA) casings are used. The starter should be fitted with changeable blocks as the wear of the cutting head is relatively high due to the cutting of the female piles. Worn teeth or blocks must be re-placed in time to ensure the proper cutting of the female piles. Thereby the retraction forces are mi-nimzed, too.

Figure 11. double wall casing with exchangeable blocks, double flight auger and casing guide


38

The verticality of the pile is controlled and influenced by the outer casing. Attention must be paid to the accurate alignment of the casing. The relative stiffness of the combined drill string is significantly higher than that of a CFA auger. The counter rotation of the casing and auger also has a beneficial sta-bilizing effect.

13 SOIL CONDITIONS

Double rotary drive systems have been used successfully in many different soils. In granular soils the system is limited where boulders are encountered. Boulders tend to cause blockages on the auger flights. The spoil gets compressed on the boring head, which stops the drilling process. The CCFA method has been used successfully in rock with a very weak to moderately strong unconfined com-pressive strength up to 20 MPa. It is also possible to drill through thin layers of strong rock. The CCFA system performs very well in stiff to very stiff clay - e.g. Blue London Clay. The stability of the surrounding soil is ensured as the borehole is supported by the casing. During the entire drilling process no soil decompression occurs and additional air pressure prevents any soil intrusion from be-low. Ground water is pressed back in the ground and the spoil reaches the ground surface, moist in many soil conditions.

14 PILE QUALITY

The high degree of verticality and pile quality is shown in Figure 12, showing a project where CFA and CCFA were employed adjacent to each other:

Figure 12. secant pile wall, dia 30 in (750 mm), Paddington, UK

During the execution of the secant pile wall by CFA method, the operator noticed a deflection in the verticality of the piles while drilling. Especially when drilling a male pile, the auger drifted strongly out of direction. Therefore the piling method was changed to CCFA. The result was an imposing im-provement of pile quality in the same soil conditions, what could be seen after the excavation of the pit. With the CCFA method a 1 in 200 verticality tolerance could be achieved, whereas the male CFA piles show high deviations.

cased CFA standard CFA


39

15 PROJECT EXAMPLES

Many piling projects using CCFA piles for secant pile walls have recently been completed with great success in the last years in England, Germany, Italy and Canada. The normal diameter for the piles is 750 mm or 880 mm. The length of the piles is limited to approx. 18 m to 21 m by the length of the mast, size of the drilling rig and the required pile diameter.

Figure 13. BG 36 with BTM 400, 18,00 m pile length, dia 880 mm, city centre Bath, UK

On a jobsite for a secant pile wall in Bath a peak performance of up to 14 piles per day was far ahead of the scheduled rate of completing 4 piles per day. An average production of 8 piles per day during the whole project could be realized. In Naples, Italy, 12 excavation pits as secant pile walls have been built during the new development of the subway. It is needed for the mechanical equipment of the air ventilation system of the tunnel. The working times per pile are split up in the following spreadsheet:

Table 1. working times per pile

reinforced pile

setting up on pile location 2 min

drilling 8 min

pouring concrete 9 min

installation of cage 10 min

cleaning of auger 1 min

Total: 30 min


40

Although space on site was very limited, the works could be executed without problems. The spoil was nearly earth moist, so that there were no special facilities for the treatment of water on the surface needed. The spoil fell down through the soil discharge chute. The required length to keep the chute short above the ground surface is adjusted by a small winch. Therefore it was possible to bypass the traffic directly along the adjacent street and pavements.

16 CONCLUSIONS

The major application for the CCFA system is the construction of secant pile walls. Against the tradi-tional systems with fully cased bored piles in Kelly mode or standard CFA piling, the CCFA system has advantages in costs and time against the fully cased system and in quality against the standard CFA system:

short production time

high verticality of piles

pile is protected from surrounding soil influence (e.g. ground water)

calculable concrete over consumption

constant quality due to drilling and concreting assistant

clean jobsite due to absence of water

simple modifications on the drilling rig

penetration through hard layers possible

less noise disturbance than cased piles in Kelly mode

Especially in the beginning of projects the team needs to get some experience to adjust the method to the soil conditions. The drilling rig must have sufficient torque, pull down and retraction force to keep the drilling process steady at all times, even in difficult soil conditions. A perfect concrete supply must be ensured in the work preparation to ensure a constant operating sequence. Good communication and coordination among the crew on site is important and changes of the staff should be avoided. These improvements and advances contribute to a better quality product. However, these will only be suc-cessful if used properly. Therefore the importance of training, understanding and briefing to all per-sonnel involved is vital to its success.


41

ABSTRACT: This paper examines pile capacity, pile movement, soil settlement in a settling ground by numerical study. A selected 2-pile group and 9-pile group cases were used to examine pile-soil

interaction under dead load and ground settlement through finite element analyses. Plaxis 3D

Foundation program is used to analyze two types of cases, one involving a uniform soil and the second a layered soft soil with hard strata below respectively. Unipile program is used to determine

the neutral plane for pile group and its pile group settlement using the concept of an equivalent

footing seated at the neutral plane. Unipile program’s settlement results at equivalent footing are used for comparison with 3D FEM results. It has been shown that the pile would experience the largest

axial load at the neutral plane, due to dragload from settling soil around the pile. The conservatism of

CP4 factor of safety approach is discussed for pile capacities calculation due to negative skin friction.

The neutral plane can be obtained from the finite element analysis on the interaction results of pile movement; and soil settlement and from the Unipile program, but the two programs may not give a

close estimation of neutral plane as Unipile do not model the actual pile-soil interaction in its analysis.

1. INTRODUCTION

There have been huge numbers of past projects involving deep piled foundation in Singapore for many decades in the private and public sectors. However, the understanding of pile-soil interaction have not

been well developed until the use of powerful numerical tools such as finite element analyses to model

realistically the soil-pile interaction of pile behavior, pile capacity and movement under various

loadings with known soil parameters, geological formations and piles configurations. This paper presents a purely numerical study. There has been great interest to understand how the pile behaves

under a settling ground which creates another set of interaction to cause dragload on the pile. Fellenius

(Fellenius, B.H., 2009) presented a much more detailed study on soil-pile interaction problem with dragload under a settling ground.

This paper examines pile capacity, pile movement, soil settlement with settlement analysis for the cases below using Plaxis 3D Foundation program version 2.2 . It is to be noted that embedded piles

model is used in analyzing the numerical results of pile forces, pile movement, etc. Rinter at the

interface of 1.0 is assumed for this study. Volume pile is adopted instead of embedded pile in using

Plaxis 3D Foundation to model the piles described in this paper.

2-PILE GROUP: 400x400 mm RC PILES ANALYSES UNDER SETTLING GROUND WITH SURCHARGE

This section tries to simulate a simple case of soil-pile interaction by using 400x400 mm precast reinforced concrete (RC) piles installed through a 30 m thick soft clay layer and 3 m socket

Numerical study on pile capacity, pile movement, soil settlement and dragload with settlement analysis

S. A. Tan National University of Singapore

S.S. Chuah Former Graduate Student, National University of Singapore


45

penetration into a completely weathered siltstone SV layer with the following soil parameters; long term drained condition is considered for this case to account for all complete consolidation under

surface surcharge load (refer to Figure 1 to 3):

a) Soft clay (Mohr-Column model; Drained): = 16 kN/m3; Cu = 20 kPa; φu = 0;

E’ = 5000 kPa (assuming that Cu is unchanged by the soil consolidation)

b) Completely weathered siltstone SV layer (Mohr-Coulomb model; Drained): = 20 kN/m

3; C’ = 50 kPa, φ’ = 45

o; E’ = 150,000 kPa

Figure 1. 3D FEM model with 30

m soft clay overlaying 10 m completely weathered siltstone SV

layer

Figure 2. Total 33 m long RC piles

in the model (3 m into hard strata SV layer)

Figure 3. 3D FEM model with

column load on top of pilecap and surcharge load above settling

ground with various magnitude of

soil settlement over depth

To simulate the soil-pile interaction alone and to avoid the contribution of soil bearing resistance below pile-cap, a 0.5 m gap was created and all the loads imposed would be fully supported by piles alone. The pile is spaced far apart at 3 m centre to centre distance to reduce the interaction between the piles. In normal cases, the piles should be spaced at least 2.5 times pile diameter. 2.1 Discussion on load-transfer and resistance curves The piles have been subjected to various dead loads under separate analyses to examine the resistance curves due to pile and soil interaction through shaft friction. For a working load (WL) of 1200 kN for each 400 x 400 mm RC pile when the allowable working stress of 7.5 N/mm

2 is adopted, each pile is

subject to high loading to examine the pile resistance with respect to pile toe resistance and toe movement for this type of founding soil layer. It is noted that toe resistance is very much dependent on toe movement for its resistance to be mobilized (refer to Figure 4). This is assuming that the pile is able to withstand the load without material failure at ultimate limit state. From the analyses, the pile would be expected to move at 27mm at pile head and 8 mm at pile toe at 2.5xWL of 3000 kN. The pile will not fail even up to 5000 kN (about 4.2xWL) with the 58 mm and 26 mm pile head and pile toe movement respectively.

2.2 Review and determination of Neutral Plane with pile movement, soil settlement and shaft

friction curves

Another case would be to examine the effect of soft clay settling effects on the piles with the simulation of surcharge to cause the ground to settle (refer to Figures 5 to 7). It is noted that the ground settlement presented using Mohr-Coulomb model here for soft clay, as a function of elastic modulus, is much lower than what the one dimensional consolidation settlement would estimate. The ground settlement values are not realistic since the model cannot produce volumetric strain. However


46

it would not affect the results in the analysis since the negative skin friction would quickly become mobilised with small ground settlement. With the effect of surcharge ranging from 20 kPa to 80 kPa, the pile would experience dragload or negative skin friction (NSF) caused by the surrounding settling ground. Assuming the pile is subject to the maximum of WL of 1200 kN for a start, the pile would experience the largest axial force at the neutral plane (N.P.) with the dragload increasing with depth till the pile experience the upward positive shaft resistance below the neutral plane. At the neutral plane, the pile moves and the soil settle together by the same magnitude. Neutral planes, N.P.1, 2 and 3 in Figure 6 are established based on this phenomenon. This study is illustrated in Figures 6 and 7 where neutral planes were lowered with the increase of surcharge effects to cause larger amount of dragload on the pile as shown in Figure 5.

Figure 4. Pile resistance curves under various loadings with pile toe movement and its toe resistance

Observation can be made from Figures 5 and 7 that as the surcharge increases from 60 kPa to 80 kPa, there is very little increase in the dragload as the skin friction is almost fully mobilised at 60 kPa ground surcharge. This can be seen from Figure 7 when the shaft friction increases at the top first where it is first mobilised under surcharge of 20 kPa and the shaft friction becomes constant at surcharge 60 and 80 kPa before it changes from negative skin friction to positive shaft resistance at the neutral plane N.P. 2 and N.P. 3 locations respectively. Figures 8 to 10 depict the Plaxis 3D Foundation pile soil interaction settlement plots under various surcharge conditions where neutral plane can be observed where pile and soil settles by the same amount at the neutral plane.

-35

-30

-25

-20

-15

-10

-5

0

0 1000 2000 3000 4000 5000 6000

Pil

e dep

th (

m)

Load (kN)

Axial load vs pile depth under

1.0WL, 1200 kN

Axial load vs. pile depth under

2.5xWL, 3000 kN


4.2xWL, 5000 kN

-30

-20

-10

0

0 1000 2000 3000

Pile toe resistance (kN)

Pile toe resistance vs.

pile toe movement

Qb = 1280 kN

Toe

move-

ment

(mm)


47

Figure 5. Load transfer curves due to NSF effects of settling ground to the piles under various surcharge

conditions and its neutral plane found

Figure 6. Respective pile movement and soil settlement are shown for various surcharge cases. Neutral plane

has been lowered down with the increase of dragload with larger soil settlement.

-35

-30

-25

-20

-15

-10

-5

0

0 2000 4000 6000

Pil

e dep

th (

m)

Load (kN)


1xWL 1200kN without NSF


1xWL with NSF due to 20 kPa

surcharge



surcharge



surcharge


4.2xWL without surcharge

-45

-40

-35

-30

-25

-20

-15

-10

-5

0

-0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35

Pil

e d

epth

(m

)

Settlement (m)

pile movement under 1xWL with 20 kPa

surcharge

soil settlement under 20 kPa surcharge


surcharge



surcharge


N.P. 1

N.P. 2 N.P. 3

N.P. 1

N.P.2 N.P. 3

NSF 1

NSF 2 NSF 3

Qsp

1200 kN


48

Figure 7. Shaft friction under various surcharge conditions which affect the neutral plane location

Figure 8. Soil settlement of 73 mm

under 20 kPa surcharge

Figure 9. Soil settlement of 243

mm under 60 kPa surcharge

Figure 10. Soil settlement of 331

mm under 80 kPa surcharge

2.3 Discussion on CP4 negative skin friction computation which diminishes geotechnical capacity of

piles

From Figure 5, we have noted that at the start of the working load of 1200 kN for each 400x400 mm RC pile, there is an increase of negative skin friction due to surcharge conditions. Under 20 kPa, the negative skin friction amount is NSF 1 as indicated in Figure 5. Then, there is an amount of NSF 2 and NSF 3 respectively for surcharge of 60 kPa and 80 kPa respectively. From Figure 7, we can conclude that the maximum load the pile can experience due to dragload is 1200 kN + NSF 3 (1422 kN) = 2622 kN assuming the negative skin friction is fully mobilised. NSF 3 value in the analysis appears to be

-35

-30

-25

-20

-15

-10

-5

0

-100 0 100 200 300 P

ile

dep

th (

m)

Shaft friction (kPa)

Shaft friction vs. pile depth

uinder surcharge 20 kPa


under surcharge 60 kPa


under 80 kPa surcharge N.P. 1

N.P. 2 N.P. 3

N.P. 1

N.P. 3 N.P. 2


49

higher than the computation of max. NSF with α method for this case (assumed α = 1.0, max. NSF = 30x0.4x4x1.0xCu = 960 kN) . This maximum load the pile can experience should be checked against the structural capacity of pile with appropriate structural factor of safety required. The maximum concrete stress obtained is 16.4 N/mm

2 which is about 0.4 fcu of grade 40 precast RC pile. For this

case, it turned out to be acceptable for design consideration and there is no need for any reduction of pile capacity due to the potential presence of negative skin friction. Hence, it is important to understand the load transfer curve of pile subject to dragload on what the pile would be experiencing. It is worth noting that the use of CP4 allowable geotechnical capacity of pile subject to negative skin friction in the long term (QaL) = [Qb+Qsp] / FoS - *Qsn, where Qb = ultimate toe resistance, Qsp = ultimate positive shaft resistance below neutral plane, Qsn = summation of negative skin friction mobilised in soils above the neutral plane, FoS is geotechnical factor of safety (2.0 to 2.5), = degree of mobilization (0.67 to 1). CP4 formula is used here for computation as an illustration. Qb from Figure 4 is 1280 kN, Qsp and Qsn from Figure 5 are 795 kN and 1422 kN (NSF 3 value) respectively. Qb from Figure 4, can be considered to be ultimate toe resistance when the pile head is subject to 2.5xWL (3000 kN) without surcharge on the ground. For FoS = 2.5, = 1.0 for end bearing pile; also it appears from Figure 7 that degree of mobilization is close to 1.0, QaL = -592 kN! (i.e. no geotechnical capacity!). For the estimation of Qb from CP4’s recommendation for driven pile, for Kb = 6, qb = Kb (40N) kPa = 12000 kPa for N = 50 typical value from a range of N = 50 to 70 for driven piles, Qb = 12000x0.4x0.4 = 1920 kN, which is close to the numerical analysis result of 1827 kN; fs = KsN kPa = 2x50 = 100 kPa; Qsp = 100x0.4x4x3 = 480 kN; NSF = 960 kN using α method; For FoS = 2.5, = 1.0 for end bearing pile, QaL = 0 kN (i.e. no geotechnical capacity) Hence, from CP4 geotechnical capacity equation, the dragload would have diminished the pile geotechnical capacity. This approach will make the pile becomes unnecessarily longer in order to gain more geotechnical capacity than what is required when the dragload is high– but in reality the dragload does not actually diminish capacity. Dragload plus dead load is a matter for the pile structural strength of pile material in a settling ground. The geotechnical capacity is determined from a head-down static loading test and it represents the maximum soil resistance available for a pile embedded in a particular soil profile. NSF does not reduce the geotechnical capacity of the soil resistances. From the above illustration of load transfer curve of pile subject to dragload, the 400x400 RC pile with 3 m penetration into the hard strata has adequate structural capacity to resist the maximum axial load (dead load plus dragload) at the neutral plane. There is no inadequacy in this approach and the pile is safe. 2.4 Discussion on CP4 for an appropriate allowable concrete stress consideration

The use of 0.25fcu may be too conservative for preformed piles with full reinforcement such as precast r.c. piles, prestressed spun piles with well defined geometry and concrete quality. The Canadian Foundation Engineering Manual allows for 0.67fcu, but their pile design is based on LRFD (Load and Resistance Factor Design method) for which load is applied with appropriate factor with the use of higher concrete stress. In the BS codes, the allowable stress in concrete piles is given as 0.4fcu. (refer to summary shown in Table 1). In view of this, the authors wish to recommend the use of 0.4fcu for allowable concrete stress which is still within the elastic behaviour of concrete and not 0.25 fcu which is too conservative for all preformed piles. This will have substantial costs saving in piling works without compromise on piling safety when higher capacity of preformed piles can be utilized. As for the bored cast-in-place piles, 0.25fcu is recommended as allowable compressive stress capped at 7.5MPa in view of concrete quality which is subject to workmanship and site casting of piles.


50

Table 1: Summary of BS code requirements on permissible stress for concrete and steel. This applies to BS 8004 for Code of Practice for Foundations

The proposed 0.4fcu is supported by the fact that the piles installed at a greater depth is subject to high confining pressure from the surrounding soils for which no buckling of piles are expected. As an example, Figure 10a shows a stress-strain curve of a grade 50 concrete. For a 0.4fcu at 20 N/mm

2, the

concrete strain is 0.9 milli-strain at which concrete is still in the elastic zone. Fellenius (Fellenius, B.H., 2009) proposed that the strain resulting from the dead load and dragload be not more than 1 milli-strain. Hence, the pile is still very safe at 0.4fcu from the elastic strain point of view with respect to pile failure consideration at much higher strain. It is important to include cube test result on stress-strain curve in the design consideration as it will show directly what stress we are at 1 milli-strain subject to the maximum of 0.4fcu for the pile design.

Figure 10a. For an illustration of 0.4fcu stress and its corresponding strain in a grade 50 concrete stress-strain

curve


51

According to Canadian Foundation Engineering Manual’s recommendation, at or below the pilecap, the structural strength of the embedded pile is determined as a short column subjected to the permanent load plus the transient live load, but dragload is to be excluded. It is the authors’ view that under such circumstances, 0.25fcu would be reasonable for a check without the addition of dragload when the piles may be subject to high bending in addition to axial loading at a shallow depth of a long pile. 3 9-PILE GROUP OF 1 METER DIAMETER BORED PILES ANALYSES

3.1 3D FEM for 9-pile group of 1 meter diameter bored piles in a layered soft soil with hard strata

below

This section tries to simulate a 9-pile group of soil-pile interaction by using 1 m diameter bored pile each installed through a 30 m thick very soft clay layer and 5 m penetration into a completely weathered siltstone SV layer with the following soil parameters; long term drained condition is considered for this case to account for all complete consolidation under surcharge load (refer to Figure 11 to 12):

a) Soft clay (Mohr-Column model; Drained): = 16 kN/m3; Cu = 20 kPa; φu = 0;

E’ = 5000 kPa (assuming Cu is unchanged by ground surcharge load)

b) Completely weathered siltstone SV layer (Mohr-Coulomb model; Drained): = 20 kN/m

3; C’ = 50 kPa, φ’ = 45

o; E’ = 150,000 kPa

Similar exercise is carried out to impose surcharge loading on top of a soft clay layer to examine the load distribution among the piles. A total of 45000 kN is applied on the centre column with 80 kPa surcharge on the soil. Figure 14. shows the centre pile would experience slightly more load than the edge pile or corner pile at the pile top. Dragload is increasing with depth and the axial load is the largest at the neutral plane.

Figure 11. 3D FEM model with 30

m thick very soft clay overlaying

20 m completely weathered

siltstone SV layer

Figure 12. Total 35 m

long 1 m diameter bored

piles in the model (5 m

into hard strata SV layer)

Corner pile = CnrP;

Edge pile = EP

Centre pile = CtrP

Figure 13. Piles under review

CnrP

EP

CtrP


52

Figure 15 depicts the shaft friction distribution along pile depth for corner pile, one without surcharge and one with surcharge. The neutral plane is about -30m of pile depth. Figures 16 and 17 are 3D FEM model and diagonal cross section to show the soil settlement under surcharge load of 80 kPa. Figures 18 and 19 show the settlement of pile group to be around 34 mm maximum at the centre of the pile group for “without surcharge” case. This is to be compared with Unipile results for pile group settlement as described in Section 3.2

Figure 14. Load distribution among the corner pile, edge pile and centre pile

Figure 15. Shaft friction for corner pile under 80 kPa surcharge and without surcharge case

-40

-35

-30

-25

-20

-15

-10

-5

0

0 2000 4000 6000 8000

Pil

e dep

th (

m)

Load (kN)

Corner pile load vs.

pile depth under 80

kPa surcharge

Edge pile load vs.

pile depth under 80

kPa surcharge

Centre pile load vs.

pile depth under 80

kPa surcharge

-40

-35

-30

-25

-20

-15

-10

-5

0

-200 0 200 400 600 800

Pil

e dep

th (

m)


Shaft friction

under 80 kPa

surcharge

Shaft friction vs.

pile depth without

surcharge case


53

Figure 16 3D FEM model with column load on top of

large pilecap without surcharge to compare with

Unipile results

Figure 17 Soil settlement and pile movement plot

without surcharge case by cutting a diagonal cross

section

Figure 18. Pile group settlement as depicted by the

large pilecap settlement of 34 mm maxium at the

centre (without surcharge case) for comparison with

Unipile results

Figure 19. Pile settlements vary from 25 mm to

34 mm (without surcharge case)

3.2 Unipile analysis for 9-pile group of 1 meter diameter bored piles in a layered soft soil with hard

strata below

Unipile program version 4 is used to model the settlement of pile group using equivalent footing concept. Since the bored piles are embedded by 5 m in a completely weathered siltstone, the layer where bored piles are socketed is a reinforced soil with piles; Hence, the soil properties in the reinforced soil is computed for the combined equivalent Young’s modulus based on the area ratio of piles in the soil area. E combined = [EpAp + EsAs]/[Ap + As] where Ep and Es are Young’s modulus of pile and soil respectively, Ap and As are the area of piles and soil respectively within the pile group. From the pile capacity calculation of Unipile V.4 program, the neutral plane is found to be at -32.2.m which is quite close to 3D FEM estimation at -30 m. The Equivalent Footing is placed at -32.2 m and the compressibility of the reinforced soil from -32.2 m to pile group toe at -35 m is that of pile and soil combined. Unipile assumed a 2(V):1(H) distribution of stress below the “Equivalent Footing” level. The settlement of piled foundation is then caused by the compression of the soil due to the increase of effective stress below neutral plane. Note that the Rt, toe resistance is calculated based on user’s input


54

of limiting toe resistance stress, 10,000 kPa for this case. From Figure 20, it is to be noted that Qd = Dead load applied at individual pile top from the pile group, Rs is the fully mobilised pile shaft friction (the curve due to Rs can be both positive and negative and mirror image to each other), Ru is the total resistance. Pile group settlement at “Equivalent Footing” level is about -25 mm. The elastic compression of pile due to Qd = 5000 kN, pile length of 35m and area of 1m diameter bored pile with Ep = 30E+06 kN/m

2 is 7mm. Hence the total pile group settlement is -32 mm which is very close to

the 3D FEM estimation of -34 mm.

Figure 20 Unipile load transfer and resistance curves (LHS) and pile group settlement curves (RHS)

3.3 3D FEM for 9-pile group of 1 meter diameter bored piles in a uniform sand layer

This section tries to simulate a 9-pile group of soil-pile interaction by using 1 m diameter bored pile each installed through a uniform sand layer. Total pile penetration is 35 m in a 50 m thick sand layer with E’ = 10,000 kPa at the top with increasing E’ such that E’ = 70,000 at -50m. (refer to Figure 21):

Uniform sand layer: (Mohr-Column model; Drained): = 20 kN/m3; C’ = 0 kPa;

Φ’ = 35o; E’ = 10,000 kPa varying with E’incr at 1,200 kN/m

2 per m depth

Figure 21. 3D FEM model with 35 m long bored piles in 50 m thick uniform sand and surcharge load on top of sand layer

-40

-35

-30

-25

-20

-15

-10

-5

0

0 5000 10000 15000

Pil

e dep

th (

m)

Load (kN)

Load

transfer

curve in

Unipile,

Qd+Rs

Capacity

curve in

Unipile,

Ru-Rs

-60

-50

-40

-30

-20

-10

0

0 10 20 30

Dep

th (

m)

Settlement (mm)

Pile group settlement at

Equivalent Footing

Pile group

settlement at

Equivalent

Footing


55

Similar to the exercise carried out previously for layered soft clay above the hard strata case, a total of 45000 kN is applied on the centre column supported by 9-pile group of 1 m diameter bored pile each with equal pile penetration depth of 35 m, with 80 kPa surcharge on top of the sand layer. Corner pile is observed for its neutral plane at -15 m for this case. (refer to Figures 22 and 23). For comparison with Unipile results without surcharge case, see Figures 24 to 27 where the movement of pile group is about -52 mm.

Figure 24 3D FEM model with column load on top of

large pilecap without surcharge to compare with

Unipile results for uniform sand case

Figure 25 Soil settlement and pile movement plot

without surcharge case by cutting a diagonal cross

section for uniform sand case

Figure 22. Load transfer curve for corner pile with and

without 80 kPa surcharge at the top of the ground

Figure 23. Shaft friction along corner pile

depth due to 80 kPa surcharge

-40

-35

-30

-25

-20

-15

-10

-5

0 0 2000 4000 6000 8000

Pil

e dep

th (

m)

Load (kN)

Axial load

vs. pile

depth

under 80

kPa surcharge Axial load

vs. pile

depth

without

surcharge case

-40

-35

-30

-25

-20

-15

-10

-5

0 -50 0 50 100 150

Pil

e dep

th (

m)


Shaft

friction vs.

pile depth

under 80

kPa surcharge

N.P. at

- 15m


56

Figure 26. Pile group settlement as depicted by the

large pilecap settlement of -52 mm maxium at the

centre (without surcharge case) for comparison with Unipile results

Figure 27 Pile settlements vary from -44 mm to -52

mm (without surcharge case)

3.4 Unipile analysis for 9-pile group of 1 meter diameter bored piles in a uniform sand layer

Since the bored piles are embedded throughout a thick layer of uniform sand, the layer where bored piles are below the neutral plane is a reinforced sand with piles; Hence, the soil properties in the reinforced sand is again computed for the combined equivalent Young’s modulus based on the area ratio of piles in the sand area such as E combined = [EpAp + EsAs]/[Ap + As] where Ep and Es are Young’s modulus of pile and soil respectively, Ap and As are the area of piles and soil respectively within the pile group. From the pile capacity calculation of Unipile V.4 program, the neutral plane is found to be at -25.3m whereas there is no neutral plane found in 3D FEM in non settling sand case without surcharge. The “Equivalent Footing” is placed at -25.3 m and the compressibility of the reinforced sand from -25.3 m to pile toe level at -35 m is that of pile and sand combined. Unipile assumed a 2(V):1(H) distribution of stress below “Equivalent Footing” level. The settlement of piled foundation is then caused by the compression of the sand due to the increase of effective stress below neutral plane. Note that the Rt, toe resistance is calculated basing on user’s input of limiting toe resistance stress, which is 10,000 kPa for this case. From Figure 28, pile group settlement at “Equivalent Footing” level is about -29 mm. The elastic compression of pile due to Qd = 5000 kN, pile length of 35m and area of 1m diameter bored pile with Ep = 30E+06 kN/m

2 is 7mm. Hence the total pile group settlement is -36 mm which is

smaller than the 3D FEM estimation of -44 to -52 mm. It is noted that for pile group in uniform sand layer, Unipile result on neutral plane is not correct though the pile group settlement appears to be reasonably realistic and close to the 3D FEM result.

Figure 28 Unipile load transfer and resistance curves (LHS) and pile group settlement curves (RHS) for sand

layer case without surcharge

-40

-35

-30

-25

-20

-15

-10

-5

0

0 10000 20000

Pil

e dep

th (m

)

Load (kN)

Load

transfer

curve in

Unipile,

Qd+Rs Capacity

curve in

Unipile

Ru-Rs -60

-50

-40

-30

-20

-10

0

0 20 40

Dep

th

(m)

Settlement (mm)

Pile group

settlement at

Equivalent Footing


57

4 CONCLUSIONS

The following conclusions can be made from the above study:

a) Plaxis 3D Foundation is able to analyze full 3D soil-pile interaction problems for different soil types with realistic estimates of pile group settlement and of neutral plane determination for a settling ground condition.

b) Unipile program will not always provide the correct estimate of neutral plane for sand layer

though the pile group settlement in uniform sand layer appears to be reasonably realistic. For pile group in soft clay layer and settling ground where piles are socketed into hard stratum below soft clay, Unipile results appear to be acceptable for pile group settlement and neutral plane determination.

c) Toe resistance is very much dependent on toe movement for its resistance to be mobilized.

Hence, it is always a good practice to measure the toe movement in static pile load tests to estimate the pile end bearing behavior under various loading conditions.

d) CP4 does not provide a reasonable geotechnical capacity calculation for piles subject to

negative skin friction as the dragload is taken to reduce its capacity. Maximum axial load is experienced by a pile at the neutral plane where the total dead load and dragload are checked against the structural pile capacity with the appropriate material safety factor. Dragload does not cause reduction of the geotechnical capacity of piles. However, it does cause settlement which is required to be taken into design consideration.

e) It is recommended to use 0.4fcu for allowable concrete stress for all preformed piles

with full reinforcement instead of 0.25fcu. The strain at which 0.4fcu should be checked such that it does not exceed 1 milli-strain. As for the bored cast-in-place piles, 0.25fcu is recommended as allowable compressive stress capped at 7.5MPa.

REFERENCES

Fellenius B.H. 2009. Basics of foundation design, Electronic Edition, Calgary, Alberta, Canada

Plaxis 3D Foundation version 2.2 Manual, 2009 Unipile User Manual version 3 & 4, 2002

Potts D.M. and Zdravkovic L. 2001. Finite element analysis in geotechnical engineering application, 1st ed.,

Thomas Telford Ltd., London, UK.

Singapore Standard CP4: 2003 Code of Practice for Foundations, Spring Singapore, Singapore.

Canadian Foundation Engineering Manual 4th Edition, 2006, Canadian Geotechnical Society, Canada


58

1 INTRODUCTION The Mohr-Coulomb model (MC) is commonly used in practice despite of its many shortcomings.

(Wong 2011). The Hardening Soil model (HS) is an advanced soil model that can better simulate real

soil behaviour. It is a big improvement over the MC model in several aspects.

This paper presents an overview of the HS model and highlights its strengths over the MC model in

excavation analysis. Back-analysis of deep excavation case records in Singapore are presented to compare the predicted wall and ground movements between both soil models and against measured

data. The paper also discusses shortcomings of the HS model.

2 THE HARDENING SOIL MODEL

2.1 An overview

The HS model is an elastic-plastic soil model based on the classical plasticity theory (Brinkgreve,

2004). For every stress increment, there is a corresponding incremental elastic and plastic strain if the

soil is undergoing primary loading or only the elastic strain if it is unloading-reloading. The main attraction of this model is its ability to simulate the nonlinear, inelastic and stress dependent behaviour

of soil.

The model adopts the Mohr-Coulomb failure criteria.

(1)

1 This paper is adapted from a paper submitted for the Hulme Prize Award of the Tunnelling and Underground

Construction Society (Singapore).

Application of the hardening soil model in deep excavation analysis1

P.L. Teo & K. S. Wong

WKS Geotechnical Consultants, Singapore

ABSTRACT: The Mohr-Coulomb model (MC) is commonly used in deep excavation

analysis for its simplicity. However, it has shortcomings that may produce unrealistic soil

behavior. The Hardening Soil model (HS) is an advanced soil model that is able to generate

more realistic soil response in terms of non-linearity, stress dependency and inelasticity. This

paper highlights some of the shortcomings of the MC model and presents a simplified

approach to determine the HS parameters for drained and undrained analysis of deep

excavations. Three case studies were back-analyzed to validate the application of the HS

model for practical excavation analysis. The HS model suffers the same problems as the MC

model in using effective stress parameters c and to determine the undrained shear strength.


59

It has two yield surfaces as shown in Figure 1(a). The first one deals with yielding due to shear stress. The second one handles the expansion of the cap due to changes in mean effective stress p’. Figure

1(b) shows the yield surfaces in three-dimensional principle stress space.

(a) (b) Figure 1. (a) Illustration of double yield surface of HS model; and (b) Yield surfaces of HS model in 3-D.

The hardening law for shear is given in Equation 2. For a given shear stress increment, the plastic

shear strain increment can be computed from this equation and the corresponding plastic volumetric strain increment can be computed from the flow rule in Equation 3. The hardening law for the cap is

given in Equation 4. For a given increment in mean stress, the plastic volumetric strain increment can

be computed using this equation with no shear strain generated.

(2)

(3)

(4)

where Kc is related to the compression index Cc and Ks is related to the recompression index Cr .

2.2 Parameters of the Hardening Soil Model

The required soil parameters are summarised in Table 1. Most of the soil parameters can be

determined from consolidated drained triaxial compression test (CD) and consolidation test. In the absence of consolidation test, the parameter Eoed

ref can be set equal to E50

ref. If the unloading-reloading

cycle is not carried out in the CD test, the parameter Eurref

can be set equal to 3E50ref

.

Table 1. Parameters for the HS model.

Parameter Description Type of Test

c Effective apparent cohesion CD or CU

Effective peak friction angle CD or CU

E50ref Effective secant modulus (50% stress level) at confining pressure of 100 kPa CD

Eoedref

Effective 1-D compression modulus at a vertical stress of 100 kPa. Typically

set equal to E50ref in the absence of test data.

Oedometer

Eurref

Effective unloading-reloading modulus at a confining pressure of 100 kPa.

Typically set equal to 3E50ref in the absence of test data.

CD

m Modulus exponent controlling the stress-dependency of the modulus with

values typically varying between 0 and 1. CD

ur Unloading-reloading Poisson’s ratio. Typically set equal to 0.2. CD

Ψ

Angle of dilation. Typical set equal to zero for undrained analysis and

( - 30o) for drained analysis. CD

Ko,nc Coefficient of earth pressure at-rest. Typically set equal to (1-sin ). -

E50 & Eoed

combined hardening

E50

shear hardening

Eur

elastic

Eoed

cap hardening

q

p’p p’

1

3

2


60

Once these parameters are known, the modulus of a soil under any stress condition can be computed using the following equations in Table 2.

Table 2. Equations for stiffness parameters.

*pref = reference pressure of 100 kPa.

The parameter E50

ref can be determined from the CD test. Assuming that the test is carried out on three

samples under different confining pressures, the secant modulus E50 can be determined from each

sample. By plotting ln (E50) versus ln

on a natural scale and fitting a straight line

through the data, the y-intercept gives the value of ln(E50ref

) and the slope is the parameter m. An illustration is shown in Figure 2(a). The parameter Eoed

ref can be determined from consolidation test.

From the stress-strain curve, the tangent modulus at 100 kPa is Eoedref

as illustrated in Figure 2(b).

(a) (b)

Figure 2. Illustration on how (a) E50ref and m; and (b) Eoed

ref are determined.

2.3 Proposed correlations for analysis of deep excavations

In the absence of CD test or consolidation test data, the following approach is proposed to determine

the parameters.

2.3.1 Sand

Based on the data from Wong and Duncan (1974), the parameters E50ref

and m were determined for

several sands. Results of the E50ref

and m are plotted against the relative density Dr in Figures 3 and 4, respectively. The proposed HS parameters for sand are summarised in Table 3.

Figure 3. Plot of E50ref against relative density. Figure 4. Plot of m against relative density.

ln

ln E50

1‘

1

10

100

1000

0 20 40 60 80 100Relative Density (%)

Sacramento River Sand

Port Allen Lock Sand

Fine Silica Sand

Monterey No. 0 Sand

E50ref (MPa) = 6 e (0.025 Dr)

E5

0ref(M

Pa

)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 20 40 60 80 100

m

Relative Density (%)

Port Allen Lock Sand

Fine Silica Sand

Monterey No. 0 Sand

m = 0.45 + 0.003 Dr


61

2.3.2 Clay

For analysis involving undrained behaviour in clay, it is proposed to adopt the parameters shown in

Table 3. By setting m = 0, the stress dependency of the soil modulus is eliminated which is valid for

undrained behaviour of clay. The HS model still captures the nonlinear and inelastic response. Table 3: Proposed relationships to determine HS parameters for sand and clay.

Parameter Proposed Value / Correlation

Sand Clay

c' (kPa) 0 cu

' (o) 28 + 0.15 Dr (%) 0

E50ref (MPa) 6 e0.025 Dr (%) 0.8 Eu (Eu based on Eu / cu ratio)

Eoedref (MPa) E50

ref E50ref

Eurref (MPa) 3 E50

ref 3 E50ref

m 0.45 + 0.003 Dr (%) 0

ur 0.2 0.2

(o) ' – 30o

Ko Ko = (1 - sin ' ) Ko = (1 - sin ' )

3 LIMITATIONS OF THE MC MODEL AND IMPROVEMENTS BY THE HS MODEL

A comparison between the stress-strain behaviour of real soil, MC and HS models is illustrated in Figure 5. The MC soil is elastic before failure and only switches to plastic upon reaching failure. In

contrast, real soil response is nonlinear even before failure. The HS model captures this nonlinear

behaviour and also uses different modulus for primary loading and unloading-reloading to capture the

inelastic response whereas the MC model uses the same modulus. In the CD tests on sand or clay, the modulus is stress-dependent. This stress-dependency is captured by the HS but not by the MC model.

Figure 5. Stress-strain behaviour of real soil, MC soil and HS.

3.1 Importance of modelling non-linear behaviour before failure

A case study to illustrate the importance of the nonlinearity attribute is shown in Figure 6(a). A

surcharge of 13 kPa was applied on the steel plates behind the shallow excavation. In the zone nearer to the excavation, the soil was sheared to a higher stress level but not to failure. Since there were no

failures, the MC soil behaved elastically and produced a near symmetrical ground settlement response

as shown in Figure 6(b). The HS model, on the other hand, captured the reduction in stiffness in this

e

Elastic

Plastic

Mohr-Coulomb Soil

e

Elastic

Plastic

UU Test on Clay

cu > 0

u = 0

CD Test on Clay or Sand

c' ≥ 0

' > 0

e

Elastic-plastic

Real Soil

e

Inelastic

e

Elastic-plastic

Hardening Soil

e

Inelastic


62

zone and yielded a non-symmetrical response as shown in Figure 6(c). The nonlinearity attribute can be important in many practical problems.

(a) (b) (c)

Figure 6. (a) Case study to illustrate the importance of nonlinearity attribute; (b) response computed by MC

model; and (c) response computed by HS model.

3.2 MC model cannot model stress-dependent stiffness

Under drained condition, the stiffness of real soil is stress-dependent; while the stiffness of a MC soil

is stress independent as illustrated in Figure 5.

Figure 7 illustrates the settlement of a rigid footing on sand with modulus increasing with depth.

Figure 7(a) shows that the MC model generated large ground surface settlement because the sand modulus near the surface was small and remained constant. Consequently, Figure 7b shows a zone of

large settlement immediately below the footing. The stress-dependent stiffness is not captured in the

MC model but is captured in the HS model. When the applied load and stresses below the footing increased, the HS model responded with an increase in soil stiffness and hence resulted in smaller

settlement.

In excavations, Figure 8 illustrates that due to stress-dependency, the stiffness of soil inside the excavation decreases as the excavation depth increases. The modulus of a soil element is initially

20 MPa and decreases to 3 MPa at the end of excavation. This change in modulus is captured by HS

model but not by the MC model.

(a) (b) (c)

Figure 7. Rigid footing in sand (a) load-settlement curves; and (b) stress contours of vertical displacement.

MC Model HS ModelLoad on Footing

Footi

ng S

ettl

em

ent

HS Model

MC Model

Zone of large

settlement

0

35

0 50 100 1500

35

0 50 100 150

Fill

Soft Marine Clay

e

e


63

Figure 8. Stress-dependent behaviour of soil under drained condition.

3.3 MC model cannot properly model unloading-reloading behaviour

Figure 9 presents a case study of a hypothetical excavation in sand. In the MC analysis, the sand was

divided into layers to simulate the increase in initial modulus with depth. During excavation, the

modulus remained unchanged. The modulus of a HS soil however, changed continuously according to the change in stress and the stress-path involved. While the MC model produced larger wall

deflection, toe movement, bottom heave and ground settlement, the HS model produced a more

reasonable response.

(a) (b)

(c) (d)

Figure 9. Case study of excavation in Sand (a) finite element model; (b) wall deflection; (c) ground settlement;

and (d) ground heave.

3.4 MC model cannot generate the correct 1-D compression behaviour

The non-linear stiffness of real soil is evident from consolidation test. The stiffness increases with

stress as the soil becomes more compact. The MC model produces a linear response as shown in Figure 10. The HS model is able to capture the non-linear behaviour because of the cap yield surface.

0

10

20

30

40

50

60

70

80

90

100

110

-5 0 5 10 15 20 25

He

ave

(m

m)

Distance (m)

HS - excav to 1.5m

HS - excav to 5m

HS - excav to 7.5

HS - excav to 10.5

MC - excav to 1.5m

MC - excav to 5m

MC - excav to 7.5m

MC - excav to 10.5

-15

-10

-5

0

5

10

15

20

25

0 20 40 60 80 100 120

He

ave

/ S

ettl

em

en

t (m

m)

Distance (m)

HS - excav to 1.5m

HS - excav to 5m

HS - excav to 7.5m

HS - excav to 10.5m

MC - excav to 1.5m

MC - excav to 5m

MC - excav to 7.5m

MC - excav to 10.5m

e

Mohr-Coulomb Model

Hardening Soil Model

-30

-25

-20

-15

-10

-5

0

0 10 20 30 40 50 60 70

Dep

th (

m)

Deflection (mm)

HS - excav to 1.5m

HS - excav to 5m

HS - excav to 7.5m

HS - excav to 10.5m

MC - excav to 1.5m

MC - excav to 5m

MC - excav to 7.5m

MC - excav to 10.5m


64

Figure 10. Stress-strain curves for one-dimensional consolidation

3.5 MC model may produce an incorrect response under certain stress paths

There are certain stress paths where an elastic soil will produce an incorrect response. Figure 11

illustrates one of them where a soil sample is subjected to a simultaneous reduction in 3 and ( 1-

3). An elastic MC soil sample rebounded upwards whereas the real soil actually moved downwards. The elastic-plastic HS model is able to simulate the correct response.

3.6 MC model under-estimates the horizontal stress in certain stress paths.

Consolidation tests were simulated to compare the horizontal stresses generated by the MC and HS

models and the empirical relationship Ko,OC = Ko,nc OCR0.5

. Figure 12 shows that the MC underestimated the horizontal stress during loading and unloading. The HS gave a better agreement

with the empirical relationship.

Figure 11. Behaviour in a drained triaxial test subjected to

reduction in 3 and ( 1- 3).

Figure 12. Comparison of horizontal stresses

computed using empirical relationship, the MC

and the HS model.

3.7 MC model results may be sensitive to Poisson’s Ratio in a drained analysis

For some problems, results can become very sensitive to the Poisson’s Ratio when using the MC model. Figure 13 shows an example of excavation in sand. By varying the Poisson’s Ratio from 0.2 to

0.4, the wall deflection doubled and strut forces increased by up to 1.5 times. The uncertainty and

sensitivity of the Poisson’s Ratio is not an issue in the HS model. The Poisson’s Ratio only affects the elastic strain but not the plastic strain. The unloading-reloading Poisson’s Ratio value of real soil is

typically between 0.15 and 0.25. A value of 0.2 is commonly used.

p

Ver

tica

l Str

ain

Vertical Stress (kPa)

Mohr-Coulomb Model depending


depending on E’

0

100

200

300

400

500

0 200 400 600 800

Ho

rizo

nta

l Str

ess

(kP

a)

Vertical Stress (kPa)

MC HS Emiprical

1.5

2.0

2.5

-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1e1 (%)

1- 3

(kg/cm2)

-0.2

-0.1

0

-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

ev(%)

0.5

1.0

1.5

2.0

2.5

0 0.3 0.6 0.9 1.2 1.5

3 (kg/cm3)

Measured

HS

MC

1- 3

(kg/cm2)

Mohr-Coulomb Soil

Hardening Soil


65

Figure 13. Stress-path and Strain behaviour under drained triaxial test.

4 CASE STUDIES Back-analysis of three case studies of deep excavations in Singapore was carried out using the finite

element (FE) program Plaxis version 9.02. They were the Rocher Complex, the Lavender MRT

Station and the Hougang-Buangkok cut-and-cover tunnel. Results of analysis using the MC and the

HS model were compared with the measured data. The HS parameters were determined according to Section 2. The total stress approach was used in the MC analyses.

4.1 The Rochor Complex

The Rochor Complex excavation was 95 m wide and 6.3 m deep. The excavation was retained using sheetpiles and three levels of preloaded struts. The ground water level was at 1.5 m below ground

level. A cross-section of the excavation and the soil profile are shown in Figure 14. (Lim et al 2003;

Halim and Wong 2005).

Figure 14. Cross-section of excavation at the Rochor Complex (Halim & Wong, 2005; Lim et al, 2003).

In the FE model, the Upper Marine Clay and Lower Marine Clay layers were divided into sub-layers with increasing undrained shear strength and stiffness. The HS stiffness parameters E50

ref, Eoed

ref and

Eurref

were determined for each Eu /cu ratio.

Results shown in Figure 15 indicate that the Eu /cu ratio of 250 for the HS model and 300 for the MC model produced a reasonable match with the measured deflection. The MC model produced larger

ground settlement and bottom heave. It is interesting that the MC model also produced more plastic

points (Figure 16), which indicated more soil reaching failure.

SAND = 20 kN/m3

’ = 30o

FIRM CLAY = 17kNm3 Cu = 100 kPa PI = 20%

LOWER MARINE CLAY = 16kN/m3 Cu = 33.1 to 36.1 kPa PI = 40%

UPPER MARINE CLAY = 16 kN/m3

Cu = 15 to 30 kPaPI = 45/%

10 20 30 400

0.0

5.05

18.5

21.25

24

Dep

th, z

(m

)

SAND = 20 kN/m3

’ = 30o

VERY STIFF SILTY CLAYCu = 200 kPaPI = 20%

PI = 40%

Dep

th, z

(m

)

FSP IIIA Sheetpile

0.6m Preload=28kN/m

3.8m Preload=175.1kN/m

6.3m

1.5m Preload=104.3kN/m

Width of excavation = 95m

24m

1.5m

Undrained Shear Strength, Cu (kPa)

0

5

10

15

20

25

30

-20 0 20 40 60 80

Dep

th (

m)

Wall Deflection (mm)

=0.2 =0.4

Mmax ,kNm/m 298 477

Strut 1, kN/m 77 114

Strut 2, kN/m 226 335

Strut 3, kN/m 163 178

c’=5 kPa

’=35o

E’=8000 kPa

H=9 m

=0.2

=0.4


66

(a) (b) (d)

Figure 15. Comparison of results from analysis using the MC and HS models at the final excavation of Rochor

Complex (a) & (b) wall deflection; (c) ground settlement; and (d) bottom heave.

(a) (b)

Figure 16. Comparison of plastic points at the Rochor Complex final excavation for (a) HS model; and (b) MC

model.

4.2 The Lavender Station

The excavation at Lavender Station was 15.7 m deep and 23 m wide. It was supported using 1 m thick diaphragm wall at 28 m penetration depth and 6 levels of preloaded struts. The ground water level is at

1.5 m below the ground level (Halim and Wong 2005).

The analysis wall deflection was compared with measured deflection profile 3, which applied to the

excavation section and soil profile presented in Figure 17 (Halim and Wong 2005). The other two

measured deflections were reported for another excavation section and soil profile (Lim et al. 2003) and are presented to show the similarity in the deflection profile.

Figure 17. Soil profile and cross section of excavation at the Lavender Station.

1000 mm Diaphragm Wall

11.5 m

Fill

Very Dense Clayey Silt (N > 100)

Dense Silty Coarse Sand (N = 83)

Medium Dense Silty Course Sand (N = 27)

Lower Marine Clay

Upper Marine Clay

-1.5 m

-40 m

-26.6 m

-22.6 m

-17.5 m

-13 m

-3.6 m

0 m

-40 m

-26.6 m

-22.6 m

-15.66 m

-13.21 m

-11.31 m

-9.4 m

-6.53 m

-3.6 m

-0.5 m

Half excavation width

Preload = 190kN/m

Preload = 390kN/m

Preload = 327kN/m

Preload = 260kN/m

Preload = 233kN/m

Preload = 220kN/m

= 18 kN/m3

= 16 kN/m3

= 16 kN/m3

= 20 kN/m3 (N = 27)

= 20 kN/m3

= 20 kN/m3

-26

-24

-22

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

0 50 100 150 200

Dep

th (

m)

Deflection (mm)


Measured final excav

HS model - Eu/Cu =

200HS model - Eu/Cu =

250HS model - Eu/Cu =

300-26

-24

-22

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

0 50 100 150 200

Dep

th (

m)

Deflection (mm)

Mohr-Coulomb Model


MC model. Total

stress. Eu/Cu = 250MC model. Total

stress. Eu/Cu = 300MC model. Total

stress. Eu/Cu = 350

-120

-100

-80

-60

-40

-20

0

0 10 20 30 40 50 60

Set

tlem

ent

(mm

)

Distance from wall (m)

HS model. Eu/Cu = 250

MC model. Eu/Cu = 300

0

20

40

60

80

100

0 10 20 30 40 50

Hea

ve

(mm

)




(c)

Plastic points zonePlastic points

zone


67

Table 4: Parameters used in the Lavender Station excavation case study.

Figure 18(a) shows that the HS gave a close prediction of the wall deflection profile using Eu /cu ratio

of 250 and 300. Similar agreement was also achieved using the MC model at Eu /cu ratio of 300 and

400. Compared with the MC model, the HS model predicted larger settlement but lesser ground heave.

The MC analysis produced more plastic points, i.e. more extensive yielding of the soil mass.

(a) (b) (d) Figure 18. Comparison of results from analysis using the MC and HS models at the final excavation of Lavender

Station: (a) wall deflection; (b) ground settlement; and (c) bottom heave.

4.3 The Hougang-Buangkok Cut-and-Cover Tunnel

The Hougang-Buangkok Cut-and-Cover Tunnel was a 16m-wide, 13.3m-deep excavation retained using sheetpile wall and four levels of struts. The ground water level was 1.5 m below ground level (Li

2001). Strut preloading was not modelled in the back-analysis.

Soil Type

Unit

Weight (kN/m3)

Both Models HS Model MC Model

cu

(kPa) '

(deg)

c'

(kPa)

E50ref = Eoed

ref

(MPa)

Eurref

(MPa) m

E (MPa)

Eu (MPa)

Fill 18 80

- 0.8 Eu

Eu = 250 cu; 300 cu

3 E50ref

0 - 300 cu; 400 cu Upper Marine Clay 16

25

Lower Marine Clay 35

Medium dense silty sand

20

- 38 1 37 0.67 40

- Dense silty sand 42 1 73 0.75 95

Very dense clayey silt 500 - 0.8 Eu

Eu = 250 cu; 300 cu 0 - 300 cu; 400 cu

-30

-28

-26

-24

-22

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

-20 0 20 40 60 80 100

Dep

th (

m)

Deflection (mm)

HS Model

Measured final

excav 1

Measured final

excav 2

Measured final

excav 3

Hs model. Eu/Cu =

250

HS model. Eu/Cu

= 300

-30

-28

-26

-24

-22

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

-20 0 20 40 60 80 100

Dep

th (

m)

Deflection (mm)

MC Model

Measured final

excav 1

Measured final

excav 2

Measured final

excav 3

MC model,

Eu/Cu = 300

MC model,

Eu/Cu = 400 -30

-20

-10

0

10

0 20 40 60 80 100

Set

tlem

ent

(mm

)




0

10

20

30

40

50

0 2 4 6 8 10 12 14 16 18 20

Hea

ve

(mm

)




(c)


68

Figure 19. Soil profile and cross-section of excavation at the Hougang-Buangkok Cut-and-Cover Tunnel

Table 5: Parameters used in the Hougang-Buangkok Cut-and-Cover excavation case study.

Soil Type

Unit

Weight

(kN/m3)

Both Models HS Model MC Model

cu

(kPa) '

(deg)

c'

(kPa)

E50ref = Eoed

ref

(MPa)

Eurref

(MPa) m E (MPa) Eu (MPa)

Fill 19 - 33 0.1 14.4

3 E50ref

0.555 15.6 -

E 13 15 -

0.8 cu

Eu = 250 cu;

300 cu

0 - 300 cu;

350 cu; 400 cu

F1 19 - 31 0.1 9.9 0.510 10.7 -

OA1 20 - 31.5 0.1 11.2 0.525 12.1

OA2 20 135

-

0.8 cu

Eu = 250 cu;

300 cu

0 -

300 cu;

350 cu;

400 cu

OA3 20 400

OA4 20 600

OA5 20 900

Figure 20(a) shows that at the final excavation level, both the HS and the MC models gave reasonable

agreement with the measured deflection. The larger predicted than measured deflection at the lower

wall portion could be due to underestimation of the soil stiffness. For the HS model, Eu/cu ratios of both 250 and 300 yielded good agreement with the measurement. Using the MC model with Eu /cu of

400 appeared to work well. As for ground settlement and bottom heave, the HS model seemed to

generate more reasonable results.

GL 104.5 m

13.3 m

9.45 m

6.4 m

1 m 2W24, 610 X229X101 kg/m

Sheetpile Wall, LX32

3.95 m 2W24, 610X229X125 kg/m

2W24, 610X324X155 kg/m

2W24, 610 X229X101 kg/m

N = 27Very Stiff Clayey Silt

Very Dense Silty Sand

OA3 =20 kN/m

N = 80

N = 5

N = 4

N = 0~1

N = 4

Organic Silt

Loose Silty Sand

OA2

11.5 m

Loose Clayey Sand

OA1

8 m

6.5 mF1

Loose Silty Sand

E

3 m

Fill

=20 kN/m

=20 kN/m3

=19 kN/m3

=13 kN/m3

=19 kN/m3

4.6 m

7.0 m

11.5 m

16 m

3

3

2 m

Very Stiff Clayey SandOA5 =20 kN/m 3

18 m

Half Excavation Width = 8 m

Very Dense Silty Sand

OA4

N = 120

=20 kN/m

21 m

3

N = 150


69

(a) (d)

Figure 20. Comparison of results from analysis using the MC and HS models at the final excavation of

Hougang-Buangkok Cut-and-Cover Tunnel (a) deflection at final excavation level; (b) ground settlement; and (c)

bottom heave.

Figure 21 shows the wall deflection at the different stages of excavation. The differences in wall

deflection generated by both models were not significant and comparisons with measured deflection for both models were reasonable. Figure 20(c) shows the comparison of strut force. The HS model

gave slightly better agreement with the measured strut forces.

Figure 21. Comparison of wall deflections from analyses using MC and HS models (a) & (b)

deflection; and (c) strut force.

-22

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

-5 0 5 10 15 20 25 30 35 40 45 50 55 60

Dep

th (

m)

Deflection (mm)Final Excavation






MC model. Eu/Cu = 400 -32

-28

-24

-20

-16

-12

-8

-4

0

4

0 20 40 60 80 100

Set

tlem

ent

(mm

)


Measured






6

8

10

12

14

16

18

20

22

0 1 2 3 4 5 6 7 8 9 10

Hea

ve

(mm

)


HS model. Eu/Cu = 250HS model. Eu/Cu = 300MC model. Eu/Cu = 300MC model. Eu/Cu = 350MC model. Eu/Cu = 400

(c)

-22

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

-5 0 5 10 15 20 25 30 35 40 45 50 55 60

De

pth

(m

)

Deflection (mm)Hardening Soil Model. Eu/cu = 250

Measured 1stexcavMeasured 2ndexcavMeasured 3rdexcavMeasured 4thexcavMeasured finalexcavAnalysis 1stexcavAnalysis 2ndexcavAnalysis 3rdexcavAnalysis 4thexcavAnalysis finalexcav -22

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

-5 0 5 10 15 20 25 30 35 40 45 50 55 60

De

pth

(m

)

Deflection (mm)Mohr-Coulomb Model. Eu/cu = 400

Measured 1stexcavMeasured 2ndexcavMeasured 3rdexcavMeasured 4thexcavMeasured finalexcavAnalysis 1st excav

Analysis 2ndexcavAnalysis 3rd excav

Analysis 4th excav

Analysis finalexcav

-12-10

-8-6-4-20

0 50 100 150

Dep

th (

m)

Strut Force (kPa)

Measured Strut Force HS model. Eu/Cu = 250 MC model. Eu/Cu = 300

(a) (b)

(c)


70

5 SHORTCOMINGS OF THE HARDENING SOIL

The HS model is a great improvement over the MC model. However, it has its own set of short

comings. Some of them are listed below.

a. Over-estimation of undrained shear strength, cu of soft clay

b. Under-estimation of pore pressure of soft clay

c. Over-estimation of cu/p ratio for normally consolidated clay

d. Under-estimation of cu/p ratio for over-consolidated clay e. Over-estimation of undrained shear strength cu under simple shear and triaxial extension condition f. For stress paths below yield surface, the HS soil becomes elastic and has all the shortcomings

associated with an elastic soil.

Figure 22 shows the stress paths of a CU test. The MC model greatly over-predictes the cu. The HS model fares a little better. The over-prediction by the HS model is due to the location of the elliptical

cap adopted by the model as shown in Figure 23. The Modified Cam Clay also uses an elliptical yield

surface but predicts a much lower cu than the HS model.

This problem could be overcome by using a lower friction angle or by setting c = cu and =0. This method works well for excavation analysis in most cases. However, it would not be appropriate for

problems involving consolidation analysis and changes in strength with time.

Figure 24(a) shows that the HS model over-predicts the cu/p ratio for normally consolidated clay and under-predicts the ratio for over-consolidated clay. This issue may not be crucial for excavation

analysis. However, for problems with complex stress paths involving unloading-reloading and changes

in effective stress, it may be prudent to scrutinise the results carefully.

The HS model also over-predicts cu tested under simple shear or triaxial extension conditions as compared to the cu from real soils as shown in Figure 24(b).

It is interesting to note that from tests on real soil and simulations on HS soil, the moduli obtained from the simple shear, triaxial extension and triaxial compression tests are different. Test results also

show that the strength and modulus obtained under plane strain condition are also different from those

obtained from triaxial test. Therefore, it should be reminded that the parameters obtained from the conventional UU, CU, CD and consolidation tests can only be used as a crude approximation.

Figure 22. Stress paths of a CU test. Figure 23. Effect of yield surface on cu.

p or p’ (kPa)

q (

kPa)

Real Soil

Hardening Soil

Mohr-Coulomb Soil

AC BDqD

u at qD = UAB

u at qD = UAC

u at qD = UAD

Total Stress

Effective Stress


71

(a) (b)

Figure 24. HS prediction of (a) cu/p ratio and (b) cu under different test conditions.

6 CONCLUSIONS AND RECOMMENDATIONS

The HS model overcomes some of the shortcomings of the MC model. The three case studies show that the HS model can produce reasonable wall deflection and ground movement that compared well

with measured data. Highlights from the back-analyses are summarised below.

a. For the HS model, the application of Eu/cu ratio of 250 to the clayey soils seemed to work well for these three case studies.

b. To achieve a similar agreement with the measured deflection using the MC model, it would be

necessary to vary the Eu/cu ratio, as in the case of the back-analysis, to 300 or 400. c. The HS model produced a more realistic ground settlement profile. The consideration of nonlinear

and inelastic stiffness in the HS model gave a better prediction of the settlement near the

excavation, as shown in the Hougang-Buangkok cut-and-cover tunnel case study. d. The HS model produced a smaller toe movement than the MC model.

e. The HS model predicted smaller bottom heave than the MC model.

f. The HS model generated lesser plastic points because the model is able to simulate the softer soil

behaviour as the soil approaches failure. The MC model generated more plastic points which gave a false impression on the extent of soil yielding.

REFERENCES Brinkgreve, R.B.J et al. 2004. Plaxis Version 8, Material Model Manual. Kulwawy, F. H. & Mayne, P. W. 1990. Manual on estimating soil properties for foundation design. Report EL-

6800, Electric Power Research Institute, Palp Alto, CA.

Lim, K. W. et al. 2003. Comparison of results of excavation analysis using Wallap, Sage Crisp and Excav97.

Proceedings of Conference on Underground Singapore 2003, Singapore, 83-94.

Halim, D. & Wong K.S. 2005. Evaluation of Modified Cam Clay Parameters for Deep Excavation Analysis,

Proceedings of Conference on Underground Singapore 2005, Singapore, 188-200.

Li, W. 2001. Braced Excavation in Old Alluvium in Singapore, PhD Thesis, School of Civil and Environmental

Engineering, Nanyang Technological University, Singapore.

Wong, K.S. & Duncan, J.M. 1974. Hyperbolic stress-strain parameters for nonlinear finite element analyses of

stresses and movements in soil masses. Geotechnical Engineering Report, Department of Civil Engineering,

University of California, Berkeley.

Wong, K.S. 2011. Things you should know about the Mohr-Coulomb Model. Seminar on Infrastructure in Soft Ground – Challenges and Solutions, Singapore.

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

c u/p

'

Drammen Plastic clay

Vaterland Clay

Studenterlunden

Drammer Lean Clay

HS Model φ' = 25°

ExtensionSimple Shear

0

0.1

0.2

0.3

0.4

0.5

0.6

0 0.5 1 1.5 2 2.5 3 3.5

c u/p

OCR

f' = 20

f' = 25

f' = 30

0.22 OCR^0.8

’ = 20o

’ = 25o

’ = 30o

0.22 OCR0.8

Compression


72

1 INTRODUCTION Deep braced excavation is a common method employed in the construction of major underground struc-tures in densely built-up areas. In the design of earth retaining support systems in such geotechnical en-gineering circumstances, it is standard practice to use 2D finite element (FE) software to derive maxi-mum wall bending moments and strut forces. While the advancement of geotechnical software has provided powerful tools in modelling soil-structure in deep braced excavation, the results they produce are still subject to modelling assumptions made by their users. Thus, for a particular geotechnical prob-lem, a wide range in the results can be obtained by different users of the same software. In this paper, two popular geotechnical 2D FE software, PLAXIS version 8.6 and SAGE CRISP ver-sion 5.3, are used to simulate a common deep braced excavation using steel soldier-pile wall with sheet-pile lagging as the earth retaining and support system. The standalone PLAXIS is a 2-stage analysis software based on the establishment of a pore pressure distribution using a separate steady seepage analysis, and then using that pore pressure distribution for the solution of the stresses. On the other hand, SAGE CRISP is a fully coupled loading-consolidation (Biot) analysis software. The problem is simplified so that uncertainties that are common to the real circumstance, such as actual groundwater regime during each stage of construction, and variability of soil permeability with depth and with respect to soil anisotropy, are not dealt with. Instead the objective is to see the range of maxi-mum wall bending moments and strut forces attainable by the practicing geotechnical engineers by only varying common assumptions made regarding soil-wall interface and wall permeability, and the capabil-ity of the software to model such problem. The output obtained from the two mentioned software will be compared and discussed. Additionally, certain pitfalls that designers should be wary of when simulating the drained behaviour of a deep braced excavation in SAGE CRISP are highlighted.

Influence of various modelling assumptions in numerical

analysis for deep braced excavation

K.K. Loh & C.C.M. Kho CPG Consultants Pte. Ltd., Singapore

ABSTRACT: The advancement in geotechnical software has made modelling soil-structure interaction a common practice in understanding practical geotechnical engineering problems, especially in the design of braced earth retaining structures for deep excavation. However, with no standard guidelines or rec-ommendations on how to model such geotechnical problems in practice, such analyses are subject to various assumptions made by the consulting agencies. The influence of assumptions such as the inter-face elements between structure and soil, wall permeability and input parameters, are studied on an earth retaining wall system comprising steel soldier-pile wall with sheetpile lagging using two popular software, PLAXIS version 8.6 and SAGE CRISP version 5.3. The results are presented and discussed. Pitfalls that Sage Crisp users may not be aware of are also presented.


73

2 SPECIFICATION OF FINITE ELEMENT MODEL The deep braced excavation modelled is about 30m wide and 21m deep and comprises 5 levels of steel struts. The retaining wall comprises steel soldier-piles placed at 1.60m spacing with sheet-piles as lag-ging that terminate at the final excavation level. The ground comprises Fill material overlying residual soil of Jurong Formation in horizontal layers. The benchmark problem is therefore simulated as a plane strain problem by considering half of the geometry using a mesh with a width of 125m and a depth of 50m. A cross section of the problem is depicted in Figure 1. A surcharge of 20kPa is assumed to extend 20m behind the wall. The boundaries are horizontally restrained at the lateral boundaries and fixed in both directions at the bottom boundary. Initial stresses have been taken as σ‟v = γ‟.h and σ‟h = Ko.γ‟.h where h is the depth below ground surface. The vertical and horizontal boundaries are assumed closed for consolidation while the water table is as-sumed „perched‟ at 1.5m below the ground surface i.e. not allowing drawdown to occur. Only a drained analysis is performed where the assumption regarding steady ground water drainage will be performed at every construction/excavation stage of the analysis. This is unrealistic in a real problem but greatly simplifies the problem for this exercise. Unit weight of the wall is assumed the same as the soil replaced. Material parameters of the underlying subsoil and the structural support system are listed in Tables 1 and 2 respectively. Computational steps are listed in Table 3. 5 models with various assumptions are carried out with the 2 mentioned software and they are summa-rised in Table 4. In Table 4, the strength of the interface between the retaining wall and the adjacent soil is represented by R, a factor of the strength of the adjacent soil. The constitutive model to be adopted is the elastic-perfectly plastic model with the Mohr Coloumb criteria.

Figure 1. Geometry of FE model


74

Table 1. Material parameters of the subsoils _______________________________________________________________________________________________________________

Soil layer E‟(kPa) ν‟ Ko c‟ (kPa) Φ‟ (°) Ψ (°) γbulk (kN/m3) kx = ky (m/s) _______________________________________________________________________________________________________________

Fill 8696 0.3 0.5 0 28 0 19 10-7 SVI N=5 8695 0.3 0.8 5 28 0 20 10-7 SVI N=17 29565 0.3 0.8 2 33 0 20 10-7 SVI N=19 33044 0.3 0.8 10 28 0 20 10-7 SVI N=31 53913 0.3 0.8 15 28 0 20 10-7 SVI N=42 73043 0.3 0.8 15 28 0 20 10-7 SVI N=30 52174 0.3 0.8 15 28 0 20 10-7 SVI N=40 69565 0.3 0.8 15 28 0 20 10-7 _______________________________________________________________________________________________________________

Table 2. Structural elements _____________________________________________________________________________________________

Description Element E (kPa) Spacing (m) Preload (kN) _____________________________________________________________________________________________

Wall 610x324x155 kg/m 2.05x10-8 1.6 - Strut No.1 2x610x178x82 kg/m 2.05x10-8 8.0 75 Strut No.2 2x610x324x155 kg/m 2.05x10-8 8.0 150

Strut No.3 2x610x324x174 kg/m 2.05x10-8 8.0 200 Strut No.4 2x610x324x195 kg/m 2.05x10-8 8.0 200 Strut No.5 2x610x324x174 kg/m 2.05x10-8 8.0 200 _____________________________________________________________________________________________

Table 3. Computational steps to be performed ____________________________________________________

Step Description ____________________________________________________

1 Surcharge 2 Install wall 3 Excavate to 1.0m below Strut No.1

4 Preload & install Strut No.1 5 Excavate to 1.0m below Strut No.2 6 Preload & install Strut No.2 7 Excavate to 1.0m below Strut No.3 8 Preload & install Strut No.3 9 Excavate to 1.0m below Strut No.4 10 Preload & install Strut No. 4 11 Excavate to 1.0m below Strut No.5 12 Preload & install Strut No.5 13 Excavation to formation ____________________________________________________

Table 4. FE models ___________________________________________________________________________________________________

Model Type Modelling assumptions ___________________________________________________________________________________________________

A Drained Wall impermeable only from top to base of sheet-pile lagging; R = 1.0 B Drained Wall impermeable only from top to base of sheet-pile lagging; R = 0.5 C Drained Wall impermeable from top to toe; R = 1.0

D Drained Wall impermeable from top to toe, R = 0.5 E Drained Similar to Model A but R = 0.5 on active side of wall only ___________________________________________________________________________________________________

Models C, D and E are unrealistic scenarios but are included to demonstrate the effects it has on the wall bending moments and the strut forces induced. Models A and B on the other hand are considered as respectively the best and worst credible scenario of the real problem, i.e. real problem will more likely to be in between Models A and B. In addition, a specific permeability value of 5E-8 m/s will be assigned to the wall below the formation for Models A and B assuming that the steel soldier-piles are embedded in pre-bored 800mm diameter concrete. This assumption accounted for the views that may be adopted by some engineers that the soldier-pile wall cannot be infinitely permeable although this may also be viewed by others (the authors included) as academic considering that the design soil permeability is in order of 10

-7 m/s.


75

3 MODELLING IN PLAXIS 3.1 Modelling aspects Modelling a deep braced excavation problem such as the one specified is a straightforward exercise us-ing PLAXIS version 8.6. This is so because of the features readily available in the software to perform the modelling specified. In PLAXIS, the retaining wall is normally modelled as plate elements where stiffness properties are „smeared‟ 2D equivalents of the real elements. Preloading and installing struts and introducing interface elements to model soil-wall interactions are simple tasks in the software. Prohibiting water flow or allowing infinite permeability though the wall can be performed by respective-ly activating or deactivating the interface elements in „water pressure mode‟. Assigning a specific per-meability value to the wall in the software, however, is not directly available, but could be done by as-signing thin columns of adjacent soil with the specific permeability behind the wall and with the interface deactivated in „water pressure mode‟. The tension cut-off option is selected for the soil layers and the default value of zero tensile strength is used. Triangular 15-noded elements of „very fine‟ mesh are used.

3.2 Comments on selected results

Only the output from the penultimate and final excavation stages is shown as it was found to generate the maximum stresses and movements for the analyses performed. Figures 2a and 2b depict the profile of the wall bending moment and horizontal deflection against depth at the penultimate and final excava-tion steps respectively. Pertinent results are summarised in Tables 5 and 6. Table 5. Selected results (PLAXIS)

Penultimate excavation step Final excavation step _________________________________________________________ __________________________________________

Model Wall BM (kNm/m) Lateral displ. (mm) Wall BM (kNm/m) Lateral displ. (mm) _________________________________________________________ __________________________________________

A 707 105 571 120 B 837 136 735 160 C 773 103 841 142 D 912 134 1047 200 E 723 108 601 125 _________________________________________________________ __________________________________________

By assuming wall-soil slippage to be R = 0.5, the bending moment increases 28.7% from 571kNm/m to 735kNm/m for Model A, and 24.5% from 841kNm/m to 1047kNm/m for Model C. Perhaps the most interesting result comes from Model E, which show that passive interface on the wall has far more influ-ence on the wall bending moment than the active interface for the problem considered. By assuming R=0.5 only on the active side of wall, the increase in wall bending moment is only 2.3% and 5.3% for Model A at the penultimate and final excavation steps respectively. If R = 0.5 is on both sides of the wall (Model B), the wall bending moment increase with respect to Model A is 18.4 % and 28.7% re-spectively in the penultimate and final excavation steps. When the same is applied to Model C, the in-crease in bending moment is 18.0% and 24.5% respectively in the penultimate and final excavation steps. The strut forces produced by Models A, B and D for the penultimate excavation step and the final steps are shown in Tables 5 and 6 respectively. The highlight will be the 72% increase in derived strut force for Strut No. 5 at the final excavation step, from 832kN/m in Model A to 1432kN/m in Model D.


76

Table 6. Strut forces (kN/m) (PLAXIS)

Excavation Step No. 5 ________________________________________________ ________________________________

Strut Model A Model B Model D Increase from Model A to D (%) ________________________________________________ ________________________________

No.1 16 37 37

No.2 371 429 420 13.2

No.3 631 652 639 1.3

No.4 949 1026 1091 15.0

No.5 - - - - ________________________________________________ ________________________________

Final Excavation Step ________________________________________________ ________________________________

Strut Model A Model B Model D Increase from Model A to D (%) ________________________________________________ ________________________________

No.1 13 51 63

No.2 382 452 471 23.3

No.3 627 649 645 2.9

No.4 875 932 914 4.5

No.5 832 907 1432 72.1 ________________________________________________ ________________________________

Figure 2a. Wall bending moment and horizontal movement at penultimate excavation step

Excavation step No.5

65.0

70.0

75.0

80.0

85.0

90.0

95.0

100.0

105.0

-1000 -500 0 500

kNm/m

mR

L

Model A Model B Model C

Model D Model E

Excavation step No.5

65.0

70.0

75.0

80.0

85.0

90.0

95.0

100.0

105.0

-0.150 -0.100 -0.050 0.000

m

mR

L


Model D Model E


77

Figure 2b. Wall bending moment and horizontal movement at final excavation step

4 MODELLING IN SAGE CRISP 4.1 Modelling aspects

Modelling the benchmark problem is a far more tedious process using SAGE CRISP version 5.3. This is due to a variety of reasons:

Computation steps listed in Table 3 cannot be sequenced directly as the software does not have ready features that provide specific functions like the simulation of preloading and installing of struts. This particular construction step in SAGE CRISP will therefore require 2 discrete steps to simulate the effect. Each discrete step in SAGE CRISP is called an Increment block. Similar-ly, each excavation will comprise 2 increment blocks: one for removal of soil material and the next for long-term equalisation of pore pressures. This is in line with the software‟s recommen-dation that an increment block should be used whenever loading (including removal of elements) or boundary conditions are applied, modified or when allowing consolidation to take place. Consequently, assigning time steps to each computational step in SAGE CRISP then becomes important when simulating drained conditions as described in the benchmark problem. As such, the first increment block for excavation will have to be kept very short so that tensile stresses and calculation errors due to undefined pore water boundaries cannot develop. To simulate the benchmark problem, the computational steps necessary for SAGE CRISP are summarised in Table 7.

2 Mohr Coulomb soil models are available in SAGE CRISP. One is termed “Original Mohr Coulomb Elastic Perfectly Plastic” while the other is “New Mohr Coulomb Elastic Perfectly Plastic”. The difference between the 2 is the plastic deformation or flow: one is associated with the yield surface of the material whereas the other is not. The Original model has associated plasticity flow rule and does not require the user to input the dilation angle (i.e. friction angle equals dilation angle), whereas the New model does and has non-associated plasticity flow. The New model in SAGE CRISP is notoriously difficult to use because of numerical difficulties or convergence issues, especially when the degree of non-associativity is high, e.g. Φ‟ = 33° and Ψ‟ = 0°. A solution may not be possible when non-associativity is high. Although more appro-priate for this exercise, it still deters the practising geotechnical engineer from using this partic-ular soil model in SAGE CRISP. This difficulty does not apply to the Original model, which however is known to predict unrealistically large volumetric change during shearing for fric-tional material described by the Mohr Coulomb criteria. Thus, although not totally keeping

Final excavation step

65.0

70.0

75.0

80.0

85.0

90.0

95.0

100.0

105.0

-1050 -550 -50 450

kNm/m

mR

L


Model D Model E

Final excavation step

65.0

70.0

75.0

80.0

85.0

90.0

95.0

100.0

105.0

-0.200 -0.150 -0.100 -0.050 0.000

m

mR

L


Model D Model E


78

within the specification of the benchmark problem, the Original model is used for this paper, but this will be factored in when assessing the output produced.

The wall can be modelled either with beam elements (3-noded beam) or linear strain quadrila-teral (LSQ) elements (isotropic elastic). The LSQ (consolidation) elements can be assigned spe-cific permeability. Using beam elements, on the other hand, does not provide barrier to water flow but this can be easily resolved in principle by providing a thin column of soil element with lower permeability of several orders on the active side.

The software provides an in-built interface element which is a flat 8-noded element with 2 „dummy‟ nodes midway along each of the narrow edges. It can be used to simulate the interface between soil and structure which begins to slip after a limiting stress condition has been reached. The slip model requires these inputs: c‟, Φ‟, normal, shear and residual stiffness and thickness. However it cannot be used for Models A and B as water flow through the wall cannot be simulated. To get around this, an alternative modelling method is required and this is com-monly done by placing thin columns of soil (LSQ-consolidation) elements with interface proper-ties around the wall to simulate slippage.

The decision was made to use the LSQ (consolation) elements to model the wall for this bench-marking exercise as it is most applicable to all the FE models specified in Table 4. To simulate wall-soil slippage, thin (0.1m) columns of LSQ (consolidation) elements with the required interface properties are used in Models B, D and E. The wall parameters are shown in Table 8 while interface parameters are shown in Table 9. Finally, structured finite mesh of 8-noded (20 d.o.f) linear strain equilateral elements is used.

Table 7. A sample of computational steps in SAGE CRISP ______________________________________________________________________________________________________

Increment Block Description Time step for block (second) ______________________________________________________________________________________________________

0 In Situ 1E15

1 Set pore water fixities 1E15

2 Surcharge 1E15

3 Excavate 1.0m below Strut No.1 1

4 Allow long term equalisation of pore pressures 1E15

5 Apply preload at Strut No.1 1E15

6 Release preload & install Strut No.1 1E15


8 Allow long term equalisation of pore pressures 1E15

9 Apply preload at Strut No. 2 1E15

10 Release preload & install Strut No.2 1E15


… …Continue to final excavation step… _______________________________________________________________________________________________________

Table 8. Wall (0.61m wide LSQ-consol elements) ______________________________________________________________________________________________________________

Description Soil model E (kPa) ν‟ γw (kN/m3) γbulk (kN/m3) kx = ky (m/s) ______________________________________________________________________________________________________________

Wall (impermeable) Isotropic Elastic 8751610 0.3 10 19-20 1E-12

Wall (permeable) Isotropic Elastic 8751610 0.3 10 20 5E-8

______________________________________________________________________________________________________________

Table 9. Material properties of soil elements to model slippage between wall and soil _________________________________________________________________________________________________________

Slip layer E‟(kPa) ν‟ c‟ (kPa) Φ‟ (°) γbulk (kN/m3) kx = ky (m/s) _________________________________________________________________________________________________________

Fill (slip) 2424 0.3 0.1 14.89 19 10-7

SVI N=5 (slip) 2424 0.3 2.5 14.89 20 10-7

SVI N=17(slip) 8245 0.3 1.0 17.99 20 10-7

SVI N=19 (slip) 9216 0.3 5.0 14.89 20 10-7

SVI N=31 (slip) 15040 0.3 7.5 14.89 20 10-7

SVI N=42 (slip) 20370 0.3 7.5 14.89 20 10-7

SVI N=30 (slip) 14550 0.3 7.5 14.89 20 10-7 _________________________________________________________________________________________________________


79

4.2 Comments on selected results SAGE CRISP results corresponding to those produced by PLAXIS are shown in Table 10. Table 10. Selected results (CRISP)

Penultimate excavation step Final excavation step _________________________________________________________ __________________________________________

Model Wall BM (kNm/m) Lateral displ. (mm) Wall BM (kNm/m) Lateral displ. (mm) _________________________________________________________ __________________________________________

A 761 (707) 109 (105) 655 (571) 131 (120) B 929 (837) 151 (136) 887 (735) 188 (160) C 812 (773) 107 (103) 878 (841) 157 (142) D 993 (912) 150 (134) 1160 (1047) 231 (200) E 773 (723) 115 (108) 671 (601) 141 (125) _________________________________________________________ __________________________________________

Brackets indicate corresponding PLAXIS results

Table 11. Strut forces (kN/m) (CRISP)

Excavation Step No. 5

_________________________________________________________ __ ________________________________

Strut Model A Model B Model D Increase from Model A to D (%) ___________________________________________________________ ________________________________

No.1 19 (16) 18 (37) 12 (37) No.2 302 (371) 379 (429) 375 (420) 24.2

No.3 621 (631) 674 (652) 649 (639) 4.5 No.4 996 (949) 1103 (1026) 1170 (1091) 17.5 No.5 - - - - __________________________________________________________ ______________________________

Final Excavation Step __________________________________________________________ ________________________________

Strut Model A Model B Model D Increase from Model A to D (%) _________________________________________________________ __ ________________________________

No.1 15 (13) 31 (51) 18 (63) No.2 315 (382) 407 (452) 407 (471) 29.2

No.3 611 (627) 666 (649) 641 (645) 4.9 No.4 916 (875) 976 (932) 935 (914) 2.1 No.5 937 (832) 1050 (907) 1521 (1432) 62.3 _________________________________________________________ __ ________________________________

Brackets indicate corresponding PLAXIS results

The results by SAGE CRISP show similar pattern to PLAXIS with respect to maximum values pro-duced in the penultimate and final excavation stages. For Model E, the difference in only placing slip-page on the active side of wall compare to no slippage is an increase of only 1.6% and 2.4% in wall bending moment in the penultimate and final excavation steps respectively. This is marginally lesser than the corresponding result in PLAXIS: 2.3% and 5.3%. If R = 0.5 is on both sides of the wall (Model B), the wall bending moment increase with respect to Model A is 22.1 % and 35.4% respectively in the penultimate and final excavation steps. When the same is applied to Model C, the increase in bending moment is 22.3% and 32.1% respectively in the penultimate and final excavation steps. These increases are larger than the corresponding increases in PLAXIS. The strut forces produced by the various FE models are shown in Table 11. 5 COMPARISON OF RESULTS (PLAXIS AND CRISP) 5.1 Comments on PLAXIS vs CRISP results It can be seen from Table 10 that bending moments and wall displacements produced by SAGE CRISP are significantly larger than those produced by PLAXIS, especially for FE models (B, D and E) that in-troduced wall-soil slippage. Compared to PLAXIS, at the final excavation step for Models B, D and E, SAGE CRISP produced 10.8-20.7% higher bending moment and 12.8-17.5% larger wall displacement. For FE models without wall-soil slippage, SAGE CRISP predicted 14.7% and 4.4% higher bending moment and 9.2% and 10.6% larger wall displacement respectively for Models A and C. The higher prediction is due to the associated flow rule adopted by SAGE CRISP‟s Original Mohr Coulomb model,


80

but the difference with the PLAXIS result could be even larger as coarser finite element mesh were used to perform the analysis in SAGE CRISP.

Figure 3a. Wall bending moment and horizontal movement at final excavation step (Model A)

Figure 3b. Wall bending moment and horizontal movement at final excavation step (Model D)

At the final excavation step, Strut No. 3 force predicted by the 2 software is comparable. For the lower 2 struts, No. 4 and 5, SAGE CRISP predicted higher loads than PLAXIS and lesser for the upper 2 struts, No. 1 and 2. This strut load distribution at the final excavation step is related to the larger wall displacement profile or bending moment (occurring at or close to the excavation level) predicted by SAGE CRISP.

Model D: Final excavation step-Wall bending

moment

65.0

70.0

75.0

80.0

85.0

90.0

95.0

100.0

105.0

-1200 -700 -200 300

kNm/m

mR

L

PLAXIS CRISP CRISP-In-built slip element

Model A: Final excavation step- Wall bending

moment

65.0

70.0

75.0

80.0

85.0

90.0

95.0

100.0

105.0

-800 -300 200

kNm/m

mR

L

PLAXIS CRISP CRISP-Beam element (wall)

Model A: Final excavation step-

Horizontal displ.

65.0

70.0

75.0

80.0

85.0

90.0

95.0

100.0

105.0

-0.150 -0.100 -0.050 0.000

m

mR

LPLAXISCRISPCRISP-Beam element (wall)

Model D: Final excavation step-

Horizontal displ.

65.0

70.0

75.0

80.0

85.0

90.0

95.0

100.0

105.0

-0.300 -0.200 -0.100 0.000

m

mR

L

PLAXIS

CRISP

CRISP-In-built slip element


81

Figure 4a. Vertical movement behind wall and excavation floor at final excavation step (Model D). Figure 4b

Pore water pressure

5.2 Graphical comparison of PLAXIS and CRISP Results of Models A (penultimate excavation step) and D (final excavation step) produced by PLAXIS and SAGE CRISP are compared. The comparisons comprise wall bending moment profile, wall hori-zontal displacement, total pore pressure (active and passive), total pore pressure contour, flow net dia-gram, vertical movement of ground surface behind wall and excavation floor and these are shown in Figures 3 to 6 . Results of Model A modelled with Sage Crisp‟s beam element for wall is shown in Fig-ure 3a and 4 (vertical displacement on excavation floor). Finally, Model D, modelled with Sage Crisp‟s in-built slip element is also shown in Figure 3b.

Figure 5. Pore water contour and flow net (Model A: PLAXIS and CRISP)

Model A: Final excavation step- Vert ground movement behind

wall

-0.150-0.100

-0.0500.000

0.050

15 25 35 45 55 65Model A:PLAXIS Model A:CRISP

Model D:CRISP Model D:PLAXIS

Final excavation step- Vert movement of excavation floor

-0.050

0.150

0.350

0.550

0.750

0 2 4 6 8 10 12 14 16Model A:PLAXIS Model A:CRISP


Model A:Crisp (Beam element)

Wall pore pressure diagram

65.0

70.0

75.0

80.0

85.0

90.0

95.0

100.0

105.0

-300-200-1000100200300

kPa

mR

L

Model A:PLAXIS Model A:CRISP


Fig. 4b Fig. 4a


82

Figure 6. Pore water contour and flow net (Model D: PLAXIS and CRISP)

6 CONCLUSION

In this paper, simulation of the benchmark problem by PLAXIS and SAGE CRISP are performed. Al-though behavior of the earth retaining wall system predicted by the 2 software compares well, signifi-cant differences in results are obtained. This is due largely to the differences in the constitutive soil model and meshing adopted in the 2 software. The higher values predicted by SAGE CRISP can be at-tributed to the usage of the Mohr Coulomb elastic-perfectly plastic soil model with associated flow rule, and by simulating wall-soil slippage with thin columns of LSQ elements with lower soil properties (half adjacent soil strength). This promotes plastic shearing in interface soil and thus larger movement. As de-scribed in this paper, simulation of the problem by SAGE CRISP is a far more tedious process and care has to be exercised when performing drained analysis reflecting that of the benchmark problem by allo-cating long time step (1E15 seconds) to each computational step. Drainage or pore water boundary has to be correctly specified at each increment block. Therefore when 2 increment blocks are used to simu-late excavation: one for removal of material and the next for steady state ground water, then the time step for the first block should be set at 1 second (a very short time) to prevent calculation issues due to the undefined pore water boundary.

Additionally, irrespective of the 2 software used, assumption of wall-soil slippage and wall permeability had been shown to greatly impact the wall bending moment and strut force predicted. The wall and its support system can be designed based on maximum bending values ranging from 707kNm/m to 837kNm/m (predicted from Models A and B with PLAXIS) and will be well within the capacity of the original soldier pile wall . However, if the range of maximum bending values from Models C and D (841-1047kNm/m) had been used for design, the original wall system would have to be strengthened substantially or may even be changed totally.


83

REFERENCE Brinkgreve, R.B.J. (ed.) 2002. PLAXIS Finite Element Code for Soil and Rock Analyses 2D version 8. Rotter-

dam: A.A. Balkema. Ong, D.E.L., Yang, D.Q., and Phang, S.K. 2006. Comparison of finite element modelling of a deep excavation

using SAGE CRISP and PLAXIS. Int. Conf. on Deep Excavations Singapore, 28-30 June 2006. Singapore. Rahim, A. 1998. The significance of Non-associated Plasticity- Part 1. CRSIP NEWS (6): 2-3. Sage Engineer-

ing Limited. Tan, T.S., Setiaji, R.R. and Hight, D.W. 2005. Numerical analyses using commercial software – A black box?

Proceedings of Underground Singapore 2005, Singapore: 250-258. Woods, R. & Rahim, A. (ed.) 2007. CRISP2D Technical Reference Manual for use with CRISP version 5. The

CRISP Consortium Ltd.


84

1 INTRODUCTION

A new underground expressway project is being built in the southern part of Singapore at a reclaimed

site. The excavation is 14 m deep by 20 m wide. The retaining wall system consists of tubular steel pipe

piles with three levels of struts. In order to control the wall deflection and improve the base stability,

ground improvement was implemented using the Deep Cement Mixing (DCM) method.

Deformation analysis using the finite element method has become a routine exercise since the start of

mass rapid transit system in Singapore from the early eighties. Even after thirty years of experience,

there are still many issues and challenges facing the local engineers. One of the issues is the modulus of

reclamation sand fill. It is common to adopt a value of 10 MPa for the entire stratum. Another issue is

the mass modulus of the marine clay improved by the DCM method (Wong et al., 2006 and Yogarajah,

2009). The third issue is the consideration of unbalanced surcharge in the design of retaining wall. The

fourth issue is the use of advanced hardening soil model in the analysis. The fifth issue is the modeling

of the OA soil as a drained or undrained material. This paper addresses these five issues.

2 GROUND PROFILE

The project is located in the southern part of Singapore at a reclaimed site. The soil profile consists of

14 m of reclamation fill overlying 30 m of soft marine clay and fluvial sand and clay layers locally

known as the Kallang Formation. Below this formation is a medium to hard Old Alluvium soil with va-

rying degree of weathering. The CPT test data of the Old Alluvium are shown in Figure 1.

Back analysis of a braced excavation with DCM ground

improvement

G.J. Li Land Transport Authority, Singapore

K.S. Wong WKS Geotechnical Consultants, Singapore

P.B. Ng

Kiso Jiban Singapore Pte Ltd

ABSTRACT: This paper back-analyzed a 14 m deep by 20 m wide excavation for an underground ex-

pressway project in Singapore. The soil profile consists of a thick reclamation fill underlain by a deep

deposit of soft marine clay and fluvial sand and clay layers locally known as Kallang Formation. The

retaining wall system consists of tubular steel pipe pile wall with three levels of struts. In order to con-

trol the wall deflection, ground improvement was implemented using the Deep Cement Mixing (DCM)

method. The 5 m thick DCM layer was located just below the formation level. Instruments were in-

stalled to monitor the performance of the earth retaining system during the excavation. Results from this

study show that (i) it is important to adopt a modulus that increases with depth for the thick sand fill in-

stead of using a constant modulus for the entire layer; (ii) the back-analyzed DCM stiffness is much

higher than the design value; (iii) the high DCM stiffness resulted in high bending moment in the wall;

(iv) unbalanced surcharge should be modeled in the analysis where appropriate; (v) the Hardening Soil

model produced better results than the Mohr-Coulomb model and (vi) the drainage condition (i.e.

drained or undrained) of the OA soil has negligible effect on the results in this study.


85

The reclamation material was placed about 30 years ago consisting of excavated old alluvium soils from

Bedok area in the eastern part of Singapore. The moisture content of samples collected from reclamation

fill ranges from 10% to 60% indicating mixed sandy and clayey materials. The unit weight varied from

16 to 22 kN/m3. The fill material is typically classified as silty sand (SM) and clayey sand (SC) under

the British Soil Classification System. The fines content varying from 5% to 80% is plotted against ele-

vation in Figure 2a. The Atterberg Limits from samples with high fines content are plotted in Figure 2b.

Some samples are located close to A-line and can be classified as intermediate to highly plastic inorgan-

ic clays.

The Kallang Formation comprises of very soft upper marine clay, lower marine clay, loose fluvial sand,

peaty estuarine deposits and occasionally a stiff clay layer. The Old Alluvium is an alluvial deposit that

has been variably cemented and weathered. The OA deposits are non-uniform and can be classified as

clayey sand, silty sand, silty clay or clayey silt for engineering purposes. The SPT N values for OA

layer ranges from 3 to more than 100 depending on the cementation, weathering grade and composition

of the soil. The OA soils in Singapore are classified under 5 different weathering grades: A to E, with A

indicating unweathered and E being completely weathered as a residual soil.

Figure 1 CPT field test results at the site.

Figure 2 Fill material fine content and plasticity


86

3 GROUND IMPROVEMENT WITH DEEP CEMENT MIXING (DCM) LAYER

To control the wall deflection, ground improvement was implemented using the DCM method. The

DCM layer is 5 m thick and is located just below the formation level.

The DCM method is a process of improving the ground by cutting a column of soils and mixing it me-

chanically using a rotating mixing tool with cement or lime. The DCM column can be formed either by

dry mixing or wet mixing method. The wet mixing method was adopted in this project. In the wet mixing

method, slurry (mixed cement and water) is discharged into the ground and mixed with existing soil by

rotary wings to form an improved column.

Before the mass production of the DCM on site, laboratory trial and field trial tests were carried out to

determine the operation parameters of binder type (cement type), grout mix ratio, binder consumption

per cubic meter of soil, mixing duration, and penetration speed etc. The details of the DCM operation

parameters and DCM column configuration were reported in the paper by Ng et al. (2011).

Core samples were collected from the DCM layer to determine the total core recovery (TCR), uncon-

fined compression strength and Young’s modulus. The test results show that the TCR ranged from 93%

to 99%, UCS strength ranged from 880 to 3226 kPa and Young’s modulus ranged from 198 to 492

MPa.

4 THE EARTH RETAINING STRUCTURE SYSTEM

The 14 m deep excavation was supported with 1.4 m diameter 16 mm thick steel tubular pipe pile wall

at 2.8 m c/c to a depth of about 55 m with the toe penetrated into the hard OA layer. Sheet piles were

used as lagging between the pipe piles. There are three layers of struts which are located at 6.5 m spac-

ing horizontally and 4.0 m vertically. The 5 m thick DCM layer just below the formation level acts as a

strut to minimise the wall deflection. The DCM layer is anchored by bored piles which resist tension due

to unloading and base stability during excavation stage and compression due to structural loading dur-

ing permanent stage. The bored pile size is 1.5 m diameter at 7.8 m c/c with penetration depth about 20

m into OA layer. The bored pile and DCM as well as tubular pipe pile wall and struts are the integral

part of the earth retaining system.

5 RETAINING WALL DEFLECTION, DESIGN VS MONITORED AND BACK ANALYZED

The configuration of the excavation and FEM mesh used for back-analysis is shown in Figure 3. The

predicted wall deflection at design stage and back-analyzed wall deflection profiles are shown in Figure

4 together with monitored results when excavation reaches the formation level at 14 m depth. The pre-

dicted wall deflection is unsymmetrical since the right side ground belongs to main tunnel and therefore

was improved with DSM from 14.0m depth to 21.0m depth. It should be noted that the back-analyzed

wall deflections cannot fully match the monitored deflection due to the model limitation and other imper-

fections.

The excavation was analyzed using Plaxis FEM code and the modeling parameters are shown in Table 1

for both design stage and back-analysis stage. It can be seen that the back-analyzed DCM stiffness is

about 1000 MPa which is more than 5 times larger than the initial design value of 116 MPa. The back

analyzed secant modulus (E50, ref

) of the reclamation fill based on the Hardening Soil model is only 1.0

MPa which is much less than the constant value of 10.0 MPa used in the initial design with the Mohr-

Coulomb model.


87

Figure 3. FEM mesh used in Plaxis back analysis

Figure 4. Back-analyzed, predicted and monitored wall deflections

Table 1A Ground layer parameters used during design stage (all layers MC)

Layer

Type

Unit

Weight

(kN/m3)

Poisson’s

ratio

Eref

(kN/m2)

cref

(kN/m2)

Einct

(kN/m3)

cinct

(kN/m3)

yref

UMC MC-UD 16 0.3 10 0.1 0 228 1.1 103.0

LMC MC-UD 16 0.3 10 0.1 0 250 1.2 103.0

F2 MC-UD 16.5 0.2 5200 25 0 312 1.5 93.0

OA(D) MC-UD 20 0.2 6.6E4 125 0 0 0 0

OA(B) MC-UD 20 0.2 1.25E5 250 0 0 0 0

OA (A) MC-UD 20 0.2 1.6E5 250 0 0 0 0

DCM MC-UD 16 0.35 1.16E5 400 0 0 0 0

Fill MC-D 20 0.3 10000 0 30 0 0 0

50

60

70

80

90

100

-60 -40 -20 0 20 40 60Deflection (mm)

Left Wall Deflection (Fill HS

drained)

Back analysed

Predicted during design

IW204 Monitored

50

60

70

80

90

100

-40 -20 0 20 40 60Deflection (mm)

Right Wall Deflection (Fill HS

drained)

Back analysed

Predicted during design

IW203 Monitored


88

Table 1B Back-analyzed ground parameters in Plaxis input

Layer

Type

Unit

Weight

(kN/m3)

Poisson’s

ratio (ur)

E50,ref

(kN/m2)

Eoed,ref

(kN/m2)

Eur,ref

(kN/m2)

cref

(kN/m2)

Power

m

Fill HS-D 20 0.15 1000 1300 3000 0.1 25 0.5

Layer

Type

Unit

Weight

(kN/m3)

Poisson’s

ratio

Eref

(kN/m2)

cref

(kN/m2)

Einct

(kN/m3)

cinct

(kN/m3)

yref

UMC MC-UD 16 0.3 10 0.1 0 228 1.1 103.0

LMC MC-UD 16 0.3 10 0.1 0 250 1.2 103.0

F2 MC-UD 16.5 0.2 5200 25 0 312 1.5 93.0

OA(D) MC-UD 20 0.2 1.0E5 125 0 0 0 0

OA(B) MC-UD 20 0.2 3.0E5 250 0 0 0 0

OA (A) MC-UD 20 0.2 4.0E5 250 0 0 0 0

DCM MC-UD 16 0.35 1.0E6 400 0 0 0 0

Table 1C Soil parameters for all layers with Hardening soil model

Layer

Type

Unit

Weight

(kN/m3)

Poisson’s

ratio (ur)

E50,ref

(kN/m2)

Eoed,ref

(kN/m2)

Eur,ref

(kN/m2)

cref

(kN/m2)

Power

m

Fill HS-D 20 0.15 1000 1300 3000 0.1 25 0.5

UMC HS-UD 16 0.2 3740 3740 11220 18.7 0 0

LMC HS-UD 16 0.2 6960 6960 20880 34.8 0 0

F2 HS-UD 16.5 0.2 14600 14600 43800 73.0 0 0

OA(D) HS-UD 20 0.2 48000 48000 1.4E+5 125 0 0

OA(B) HS-UD 20 0.2 1.4E+5 1.4E+5 4.3E+5 250 0 0

OA (A) HS-UD 20 0.2 2.3E+5 2.3E+5 6.9E+5 250 0 0

DCM HS-UD 16 0.2 1E+6 1E+6 3E+6 400 0 1.0

6 MODELING OF RECLAMATION FILL MATERIAL

Results of the CPT and laboratory tests indicate that the reclamation fill material is a soft to firm sil-

ty/clayey soil inter-bedded with irregular sandy layers. During the design stage, the fill material was

modeled as drained type using Mohr-Coulomb model in Plaxis. However, it can be seen in Figure 5 that

the monitored wall deflection differs greatly from the predicted both in shape and magnitude. In this

study, the fill material was modeled as drained material using the Mohr-Coulomb (MC) and Hardening

Soil (HS) models. In the MC model, four constant moduli (2, 4, 6 and 10 MPa) were studied. In the HS

model, four different E50,ref (1, 2, 6 and 10 MPa) were used. The results are shown in Figure 5. The MC

model yielded almost the same deflection profile regardless of the modulus. For the HS model, E50,ref = 1

MPa yielded the best agreement with measured value.

7 EFFECT OF DCM LAYER STIFFNESS ON RETAINING WALL

The DCM layer was modeled as undrained type using Mohr-Coulomb model. The modulus of the DCM

layer Eref was varied from 150 to 2500 MPa trying to match the wall deflection profile. The wall deflec-

tion and the accompanying bending moment profiles are shown in Figure 6 for the different DCM stiff-

ness.

It can be seen that the back-analyzed DCM stiffness is about 1000 MPa, which is more than 5 times the

initial design value of 116 MPa. This modulus has greatly reduced the wall deflection at formation level


89

from 39 to 11 mm. However, the bending moment has also increased from 1760 to 2100 kNm/m. De-

pending on the problem and wall properties, the increase in bending moment can be even higher. There-

fore, it is recommended that the design of retaining wall with ground improvement layer should consider

both the lower and upper bound of the DCM stiffness. This will be especially critical for the design of

rebar in concrete walls. The wall may be under designed which will be difficult to amend later. When

comparing with the laboratory test results of samples collected from the DCM layer, it seems that the

current practice of measuring stiffness from UCS test may seriously under-estimate the stiffness of the

DCM mass properties. More studies are needed in the estimation of the lower and upper bound design

stiffness for the DCM layer.

Figure 5 Wall deflections due to different modeling and stiffness of fill

Figure 6 Effect of DCM stiffness on wall deflection and bending moment

8 EFFECT OF UNBALANCED SURCHARGE

It is a normal practice that balanced surcharge is placed outside both sides of retaining wall in FEM

modeling. However, the monitored inclinometer deflections indicated sway type movement of the retain-

ing wall. This is partly caused by the very soft fill material, and may also be affected by unbalanced

surcharge since one side is construction access road and another side is not used as access road. The

50

60

70

80

90

100

-20 0 20 40 60

Deflection (mm)

Fill Stiffness Effect (HS Drained)

Fill Eref=6000 kN/m2


IW204 Monitored



50

60

70

80

90

100

-20 0 20 40 60

Deflection (mm)

Fill Stiffness Effect (HS Drained)



IW203 Monitored



50

60

70

80

90

100

-20 0 20 40 60

Deflection (mm)

Fill Stiffness Effect (MC

Drained)

Fill MC Eref=6000 kN/m2


IW204 Monitored



50

60

70

80

90

100

-20 0 20 40 60

Deflection (mm)

Fill Stiffness Effect(MC

Drained)



IW203 Monitored


MC Eref=2000 kN/m2

50

60

70

80

90

100

-20 0 20 40 60

Deflection (mm)

Left Wall Deflection Due to DCM

Stiffness Effect

Left Wall E=2.5E6 kN/m2


IW204 Monitored

Left Wall E=5E5 kN/m2


50

60

70

80

90

100

-50 0 50 100

Deflection (mm)

Right Wall Deflection Due to DCM

Stiffness Effect

RightWall E=2.5E6 kN/m2


IW203 Monitored

RightWall E=5E5 kN/m2


50

60

70

80

90

100

-3000 -2000 -1000 0 1000

BM (kNm/m run)

Left Wall BM Due to DCM Stiffness Effect





50

60

70

80

90

100

-1500 -1000 -500 0 500 1000

BM (kNm/m run)

Right Wall BM Due to DCM Stiffness Effect






90

surcharge due to heavy lorries and lifting cranes may be the main reason for the sway type movement.

In order to produce similar wall deflection, unbalanced surcharge is placed in the Plaxis model. The sur-

charge on the left side was varied from 20 to 60 kPa, and right side surcharge is zero. The effect of dif-

ferent surcharge on the wall deflection is shown in Figure 7 which indicates that unbalanced surcharge

of 50 kPa matches the monitored wall deflection well. The soil parameters used in the analyses to gener-

ate Figure 7 are from Table 1B.

Figure 7 Effect of unbalanced surcharge on the retaining wall deflection

9 COMPARISON OF MC AND HS SOIL MODEL

The MC model is commonly used in practice in Singapore although advanced models have been in exis-

tence and available for many years. In this study, four analyses were conducted. The first one uses the

MC model for all soils with parameters shown in Table 1A. The second one uses the HS model for the

fill and MC for all other soils with parameters shown in Table 1B. The third one uses the HS model for

all soils with parameters shown in Table 1C. The fourth one is similar to the third one except the DCM

modulus was reduced from 1000 to 500 MPa. The back-analyzed wall deflection profiles are shown in

Figure 8. It can be seen that the all layers using HS model with DCM E50, ref = 1000MPa produces bet-

ter agreement with the field measured than those with the MC model.

10 COMPARISON OF UNDRAINED AND DRAINED MODELLING FOR OA LAYER

The underlying old alluvium material in this project shows variable sand and clay contents resulting in

highly variable permeability. Some limited permeability tests show that OA permeability varies from

1.0E-4 m/sec to 1.0 E-8 m/sec depending on the and silt and clay content. It is a common practice to run

both drained and undrained analysis for this material. The OA parameters for the undrained case are

given in Table 1B and the drained case in Table 2. Parameters for all other soils are given in Table 1B.

Results of both analyses are shown in Figure 9. The differences between both cases are negligible main-

ly because the OA soils are located far below the final excavation level.

50

60

70

80

90

100

-20 0 20 40 60

Left Wall Deflection Due to

Unbalanced Surcharge Effect

Surcharge = 50 kPa

Surcharge = 60 kPa

IW204 Monitored

Surcharge = 35 kPa

Surcharge = 20 kPa

50

60

70

80

90

100

-20 0 20 40 60

Deflection (mm)

Right Wall Deflection Due to

Unbalanced Surcharge Effect

Surcharge = 50 kPaSurcharge = 60 kPaIW203 MonitoredSurcharge = 35 kPaSurcharge = 20 kPa


91

Figure 8 Comparison of MC and HS material model effects

Table 2 Drained parameters for OA soils

Layer

Type

Unit

Weight

(kN/m3)

Poisson’s

ratio

Eref

(kN/m2)

cref

(kN/m2)

Einct

(kN/m3)

cinct

(kN/m3)

yref

OA(D) MC-D 20 0.2 1.0E5 5 28 0 0 0

OA(B) MC-D 20 0.2 3.0E5 10 30 0 0 0

OA (A) MC-D 20 0.2 4.0E5 10 32 0 0 0

Figure 9 Comparison of drained and undrained modeling of OA soils

50

60

70

80

90

100

-20 0 20 40 60

Left Wall Deflection duen to

OA Modelling Effect

OA undrained

OA drained

IW204 Monitored

50

60

70

80

90

100

-20 0 20 40 60

Deflection (mm)

Right Wall Deflection Due to

OA Modelling Effect

OA undrained

OA drained

IW203 Monitored

50

60

70

80

90

100

-20 0 20 40 60

Left Wall Deflection

HS and MC Comparison

Only Fill HS, Table 1B

Table 1C

IW204 Monitored

All layer MC, Table 1A

Table 1C except DCM Eref=500 Mpa

50

60

70

80

90

100

-20 0 20 40 60

Deflection (mm)

Right Wall Deflection

HS and MC Comparison

Only Fill HS, Table 1B

Table 1C

IW203 Monitored

All layer MC, Table 1A

Table 1C except DCM Eref=500 Mpa


92

11 CONCLUSIONS

The following conclusions can be drawn from the study.

1) Deep cement mixing (DCM) is a successful ground improvement method in deep excavation of soft

soil.

2) Reclamation fill material deserves better attention in terms of material model, material type and stiff-

ness in Plaxis analysis. Constant stiffness in the Mohr-Coulomb model may not be able to give satisfac-

tory results if the sand deposit is thick. The HS model in this study for fill matches well the deflection.

3) The back-analyzed DCM layer is much stiffer than the designed value. Both the lower and upper

bound values should be considered in the design.

4) The unbalanced surcharge should be considered in future retaining wall design where appropriate.

This is especially important when the ground layer is sloping and access way is only on one side.

5) The HS model produced better results than the MC model in deep excavation analysis.

6) For this project, both drained and undrained modeling of the OA soils produce similar results.

Acknowledgements

The authors wish to thank the Land Transport Authority for allowing them to publish the data, and the

design consultant, main contractor and instrumentation contractor for their support during the prepara-

tion of the paper.

References Plaxis V8.0 User Manual by Plaxis BV, 2008.

Wong K.S., Halim, D. and Goh, A.T.C. (2006) “Modelling JGP Slab in Deep Excavation Analysis”, Proceed-

ings of International Conference on Deep Excavations, Singapore, June 2006.

Yogarajah, I. (2009). Concerns of Grouting in Soils, GeoSS Seminar, 30 June 2009

Ng P.B., Li G.J., Wong K.S. (2011). Application of DCM in the construction of an underground expressway in

Singapore, International Symposium on Advances in Ground Technology and Geo-Information, Singa-

pore 2011.


93

1 INTRODUCTION

Most of the deep excavations in Singapore are carried out in densely urbanized areas and at close prox-imity to buildings and infrastructures. The excavation effects on these adjacent structures can be detri-mental as problems such as settlement of the ground could arise. Excessive ground movements are likely to cause damage to adjacent buildings, roads and utility lines, which would result in delay of construc-tion work and cost overrun. The situation is more acute if the site is located near critical structures such as MRT structures, old buildings, or historical structures. These types of structures are very sensitive to ground movements, and the tolerance on the allowable ground movements on those structures is strin-gent. For example, LTA limits the total allowable movement on its structure or track to be less than 15mm in any direction as specified in the Code of Practice for Railway Protection.

The estimation of ground settlement profile around an excavation thus becomes an essential task of building assessment. Two-dimensional (2D) FEM is commonly used to predict ground movements caused by excavations, however, realistic settlement prediction is often difficult to achieve without using accurate representation of small-strain nonlinearity in a finite-element analyses. For example, previous analysis on deep excavation problems often produced significant amount of toe movement which does not replicate the field measurement of wall deflection. Unless adjustment is made to the stiffness of soil around the toe, the real toe behaviour is impossible to match. Whittle et al. (1993) has also shown that the accuracy of ground settlement predictions by FEM can be significantly improved if the soil beha-viour at small strain levels can be properly modeled. However, it is generally difficult to model correctly

Prediction of ground settlement due to adjacent deep excavation works

W.M. Cham & K.H. Goh Land Transport Authority, Singapore

ABSTRACT: The prediction of ground settlement is critical for excavation works adjacent to buildings. In the past, it was acceptable to adopt a factor of safety approach in designing a deep excavation against collapse, which indirectly offers acceptable ground and building settlement during construction. Today serviceability predictions of ground and building deformations are becoming more important as owners now require more assurance that existing structures and services will not be damaged. Over the last twenty years, numerical analysis has been increasingly used to predict ground settlement; however they tend to make unreliable estimates because the user often fits straight lines to the non-linear stress-stain curve of soil. Those programs that use sophistical soil model can make more accurate predictions but only if the user can provide reliable values in their input parameters. The lack of practicable means of accurately predicting ground and building settlement caused by deep excavation sometimes require expensive design options, which are shown to be unnecessary when instrumentation monitoring indicate a much smaller ground and building settlement. Therefore, further studies would be required to develop a better understanding of the problem and contribute to an economic design. In this paper a simplified semi-empirical model is proposed for predicting maximum ground settlement and ground settlement profile due to excavations. This can be done by investigating the actual ground settlement behaviour using the instrumentation and monitoring data that was collected during the exca-vation works of Circle Line Projects.


94

the soil behaviour at small strain levels as the measurement of soil parameters at small strains requires a specialized triaxial apparatus with high resolution instruments, which is generally not readily available. Over the last 20 years, methods for estimating ground settlement caused by deep excavations have re-ceived considerable attention. Numerical modeling is used for predicting the performance of excavation support systems, but this often comes at the errors or misleading results for ground settlement. The con-struction of large scale infrastructure projects such as Circle Line and North-East Line has opened up enormous scope for underground works in Singapore, which offers a great opportunity to study the ground response due to adjacent excavation. In this paper, a simplified semi-empirical method in esti-mating the magnitude and profile of the ground settlement caused by excavation are described.

2 GROUND MOVEMENTS AROUND EXCAVATION

Two main factors are known to induce ground movement around an excavation. Firstly there is stress relief due to the removal of overburden. The soil that is being removed initially serves as a support to the boundary of the excavation. When it is removed, the soil around the excavation starts moving in-wards due to the loss of this support, and ground movement is then generated. Secondly, if the excava-tion is carried out below the groundwater table and if the soil is permeable or the elapsed time is long enough to allow the water table to drop, additional ground movement may occur due to the increase in effective stress in the soil generated by the fall of the groundwater table. Hulme et al. (1989) and Yong et al. (1989) have also reported of the time-dependent change in ground movement, which is associated with the dissipation of negative pore water pressure. Excess pressure begins to dissipate immediately and typically continues well beyond the completion of the excavation, resulting in progressive move-ments even when no excavation is being carried out. This paper focuses on the ground movements aris-ing from immediate stress relief due to overburden removal during excavations, rather than the influence of consolidation arising from drop in piezometric pressures. There are two general settlement troughs during excavation: (a) maximum settlement occurs at some distance from the retaining wall (Figure 1a), and (b) maximum settlement occurs very close to the wall (Figure 1b). For case (a), the top level of struts is effective in restraining the cantilever mode of initial deflection of the wall. As excavation proceeds, the wall at the prop level can only deform a little while the rest of the wall continues to deform, thereby inducing further ground settlement at some distance from the wall. As for case (b), the wall deflects in a cantilever mode with larger settlement occurring near the face of the wall. In many excavations, the first stage of excavation is often in this mode. (a) (b)

Figure 1 Different types of settlement profile (After Ou et al., 1993)

Ou et al. (1993) reported that the extent of ground settlement behind the wall increased with excavation depth. He observed from a number of case studies that although the location of the maximum wall def-lection varies with excavation depth, the location of maximum ground surface settlement does not seem to vary as much with excavation depth.


95

Ou et al. (1993) also highlighted the complexities of factors affecting the ground movement. For exam-ple, the use of inadequate excavation support system and improper dewatering will affect ground settle-ment and retaining wall deflection. Others include soil and groundwater conditions, excavation geome-try, sequence, duration, surcharge conditions, existence of adjacent buildings, retaining wall construction method, penetration depth and stiffness, spacing and stiffness of lateral supports and pre-loading.

3 DESIGN APPROACH

Empirical relations and charts have been proposed by various researchers for the evaluation of ground movements in deep excavations (Peck.1969; Clough and O’Rourke 1990; Ou et al. 1993; Hsieh and Ou 1998). Peck (1969) proposed a settlement chart (Figure 2) to estimate the magnitude of ground settlement. This chart is divided into three zones and estimates the settlement according to the in-situ soil conditions, and was based on excavations in Chicago and Oslo clays supported by soldier piles or sheetpiles with struts and tieback. With the greater use of stiffer wall recently such as diaphragm wall, it would be more ap-propriate to use charts that are developed for stiffer walls.

Figure 2 Settlement chart by Peck

Clough and O’Rourke (1990) summarized methods to estimate maximum ground settlement deforma-tions associated with construction of excavation and provided another series of charts illustrating enve-lopes of ground settlement for different soil types (soft clay, stiff clay, and sand). With knowledge of the maximum ground settlement, the dimensionless diagrams in Figure 3 may be used to estimate the ground settlement profile. Based on several case studies, Ou et al. (1993) proposed an empirical formula to estimate ground set-tlement at distance away from excavation. He reported that the maximum settlement is equal to around 0.5 to 0.7 times the maximum wall deflection.


96

Figure 3 Design charts for estimating the profile of ground settlement adjacent to excavation by Clough and

O’Rourke (1990)

Hsieh and Ou (1998) proposed a procedure for estimating the excavation-induced surface settlement profiles. These profiles are divided into the primary influence zone and the secondary influence zone, and can be determined with the prerequisite that vm is already known. They further suggested that vm could be estimated from the relationship between hm and vm which is expressed as vm = Rhm (where R=deformation ratio). Based on excavation case-history data, the deformation ratio generally falls in the range of 0.5 to 1.0 for soft to medium clays. However, there is a high scatter in the deformation ratio de-termined from field observations and thus, improved procedures for estimating the deformation ratio and ground settlement are needed.

4 WALL DEFLECTION AND GROUND SETTLEMENT MONITORING FROM CIRCLE LINE PROJECTS

4.1 Geological conditions and type of retaining walls

The Circle Line (CCL) is a fully underground orbital line, linking the strategic Mass Rapid Transit lines leading into the city. It is 33 kilometres long, and comprises of 29 underground stations which are all constructed using the cut-and-cover method. For this study, the wall deflection and corresponding ground settlement monitoring at several braced excavation sites in Stages 3 and 4 of the CCL are ana-lysed and reported. Figure 4 shows the geological formations through various stages of the Singapore Circle Line. Specifically, CCL3 alignment was into the Bukit Timah Granite & Old Alluvium forma-tions, whilst CCL4 alignment was into the Bukit Timah Granite and Jurong formation. At several of the stations, the Kallang Formation was also encountered, but at different extents depending on local condi-tions.


97

Figure 4. Geology through various stages of the Singapore Circle Line

Table 1. Construction details at various stations in CCL3 and CCL4.

Station Max. excava-

tion depth Ground conditions Type of Retaining walls

Farrer 28m Fill, GVI, GV D-Wall

Botanic Garden 28m Fill, KF (10m), GVI, GV 0.8m - 1.0m thk D-wall

Bukit Brown 18m Fill, GVI, GV, GIII/GII 1.0m dia CBP wall

Thomson 24m Fill, GVI, GV, GIII/GII 1.0 - 1.2 dia. CBP wall

Holland Village

20m

Jurong formation SIII,

SIV, SV 1m thick D-wall

C&C near AYE 20m Jurong formation SIV, SV

SVI CBP wall

One North 23m Jurong formation SIV, SV

SVI 1.2m CBP wall

NUH 35m Jurong formation SIV/SV CBP wall

Table 1 summarises the station locations from CCL3 and CCL4 that were studied and where inclinome-ters and settlement markers monitoring are reported in this paper. Eight excavation case histories of braced excavations in soft to stiff clays are collected and used for developing the intended model. The stations were constructed using cut and cover method of construction with multiple layers of struts. The depths of excavation, geological conditions, as well as the types of retaining walls are also listed for each station excavation.

4.2 Monitoring of wall deflection and ground settlement

An intensive programme of wall deflection and ground settlement monitoring was implemented during the excavation works. More than 250 settlement markers and 40 inclinometers readings were extracted and these are presented in Figure 5 and 6.

CCL1

CCL21

CCL321

CCL421

CCL5421


98

Figure 5 Measured maximum lateral wall movement vs Depth of excavation

Figure 6 Measured maximum ground settlement vs depth of excavation

Figures 5 and 6 are measured maximum wall and soil movements plotted against He for braced excava-tion. Both figures have disclosed a general pattern of wall movement and adjacent ground settlement, which can be useful as design tools to estimate maximum wall and soil movements. In Figure 5, it can be seen that the data scatter between hm/He of 0.1% and 0.5%, which is similar to those presented by Ou et al. (1993). In Figure 6, the data scatter between vm/He of 0.05% and 0.4%.

Observations of movements of soil adjacent to excavation provide the best way to define ground settle-ment likely to occur in the field. However, it is well known that field measurements can involve signifi-cant movement components caused by activities such as dewatering and construction loading. Therefore, it is logical to treat these components separately and to focus on the soil movements caused by excava-tions. In this study, each set of field measurements involving inclinometers and corresponding settlement markers was screened to preclude movements not primarily related to the excavation works.


99

5 ANALYSIS OF MONITORING DATA

The objective of this study is to develop a semi-empirical model which consists of four components: (a) predicting max wall deflection, (b) estimate deformation ratio, (c) estimate max ground settlement and (d) generate ground settlement profile caused by a braced excavation in soft to stiff clays.

5.1 Predicting maximum wall deflection

Generally, the wall deflection caused by excavation can be adequately predicted by FEM, but the ground movement is usually far from the field observation. In most cases, the ground settlement near the retaining wall predicted by FEM is significantly smaller than that observed in the field. On the contrary, the prediction of surface settlement by FEM far away from the retaining wall is larger than the observa-tions. Similar observations have been reported by Ou et al., where the shear strain of soil was less than 0.01% and hence high stiffness in the regions far from the wall. There are numerous published reports that indicate a close match between measured wall deflection and results of 2D FEM analysis. As such, in this semi-empirical model, the maximum wall deflection is rec-ommended to be obtained from FEM.

5.2 Predicting the deformation ratio

With the generated data described previously, attempts are made to develop an empirical equation that relates deformation ratio (R) to hm and vm, which can be expressed as: R = vmhm (1)

Figure 7 Deformation ratio

Figure 7 shows the vm/He plotted against hm/He. It is observed that the deformation generally falls in the region of 0.2 to 1.0. The accuracy of the estimated ground settlement depends on the accuracy of the estimated deformation ratio. Given the high scatter of the deformation ratio determined from field obser-vations, further steps to refine the model are needed. As discussed by Kung et al (2007), the deformation ratio R may be affected by many factors such as the geometry of excavation, the wall system (wall thickness, wall length, and strut stiffness) and the ground conditions (soil properties). To ascertain the effect of these factors on the deformation ratio, a series of parametric studies was conducted using finite element modelling to examine the possible effect of the above factors. It was found that the geometry of excavation, the excavation width, wall thickness, wall length and strut stiffness have little effect on the deformation ratio R. However the results showed that


100

the deformation ratio R is significantly affected by the soil parameters of shear strength and Young’s modulus. With this in mind, an attempt was made in this study to correlate the deformation ratio to the soil para-meters using SPT-N values. Although SPT-N value may be an approximate measure of soil strength and stiffness, it has the benefit of being routinely tested in soil investigations and this allows it to be used as a proxy of ground conditions. Figure 8 shows the plot between deformation ratio (R) and the corresponding average SPT-N values for the various CCL sites. The average SPT-N values were calcu-lated for the excavated soil above the final excavation level (excluding rocks). Notwithstanding the scat-ter in data plots, there is a general trend showing a reduction in deformation ratio with increasing SPT-N value of soil. It should be noted that conclusive result is not achieved for very soft clay where SPT-N value is zero. Nevertheless, this trend line can be used to provide reasonable guidance on estimating the deformation ratio R.

Figure 8 Deformation ratio vs average SPT-N values

With the estimated deformation ratio, the maximum ground settlement can be calculated from the pre-dicted maximum wall deflection using the relationship in Equation (1).

5.3 Predicting the surface settlement profile

Based on the findings presented by Clough and O’Rourke (1990) and Hsieh and Ou (1998), a concave-type settlement profile is generally observed for braced excavations. Figure 9 shows the normalized ground settlement profile observed for the eight case studies, where the ground surface settlement was normalised by dividing by the maximum surface settlements and the distance behind the excavation was normalized by dividing by the distance behind the excavation coinciding with the maximum settlement. An envelope can be drawn to all the points in the case studies, and a simple curve can be fitted to this envelope so that the maximum settlement profile due to excavation-induced movements can be described by an empirical equation. The proposed settlement profile, denoted herein may be expressed as:

(2)


101

Where x = distance from the wall (m); v = ground settlement at the distance x (m); vm=maximum ground settlement (m) and xvm= distance from the wall to point of maximum ground settlement (m)

Figure 9 Normalized ground settlement profile

To estimate the location behind the excavation where maximum surface settlement is located, a chart (Figure 10) is also developed from the CCL data points to establish the relationship between the dis-tance from the wall to the point of maximum ground settlement (Xvmax) and depth of excavation (He). As seen in Figure 10, the maximum settlements in the CCL case studies occur between 0.5 times to 1.1 times of the excavation depth. This compares well with the assumption by Hsieh and Ou (1998) where the maximum settlement was assumed to occur at 0.5He and Clough and O’Rouke (1990)’s assumption of maximum settlement occurring at 0.75He. A median trend line of 0.8 as indicated on the chart is rec-ommended.

Figure 10 Relationship between the distance from the wall to the point of maximum ground settlement (Xvmax)

and depth of excavation (He)

The proposed semi-empirical model may be used to estimate the deformation ratio, the maximum ground settlement, and the ground settlement profile. This model is applicable to braced excavations in soft and stiff clays, and a step-by-step procedure to implement this model is presented below:


102

a) Determine maximum wall deflection from FEM; b) Estimate the deformation ratio; c) Calculate the maximum ground settlement; and d) Estimate the ground settlement profile from equation 2

It should be noted that this procedure would be more suitable for wall deflections showing a deep in-ward wall movement shape typical of braced excavations, rather than for wall deflections that are show-ing cantilever shapes predominantly. Furthermore, it should also be noted that settlement arising from consolidation effects should be considered separately from this procedure.

6 CONCLUSION

Accurate prediction of ground settlement adjacent to an excavation is often difficult to achieve without using accurate representation of small-strain nonlinearity in a finite-element analyses. In this paper a simplified semi-empirical model is proposed for predicting deformation ratio, maximum ground settle-ment and ground settlement profile due to excavations. The model is developed based on the results of large number of actual data in CCL projects. However it should be noted that the proposed semi-empirical model assumes normal workmanship and no basal failure in the braced excavation. Possible uncertainties caused by the soil variability at the ex-cavation site and construction-related issues such as dewatering activity and over excavation are not ex-plicitly addressed in this paper. The effects of consolidation settlement are also not considered for the proposed semi-empirical model. Engineering judgment is required and must be carefully exercised to ad-just the model as necessary, to account for these factors.

It is hoped that the proposed semi-empirical model based on the measured database recorded from the CCL excavations can become useful references and design check for engineers undertaking future build-ing assessment in Singapore where accuracy of ground settlement profile is of great importance.

REFERENCES

Clough, G. W., and O’Rourke, T. D. (1990). “Construction induced movements of in-situ walls.” Proc., Design

and Performance of Earth Retaining Structure, Geotechnical Special Publication No. 25, ASCE, New York,

439-470.

Gordon T.C. Kung, C. Hsien Juang, Evan C.L. Hsiao and Youssef M. A. Hashash (2007). “Simplified model

for wall deflection and ground-surface settlement caused by braced excavation in clays” Journal of Geotechnical

and Geoenvironmental Engineering, 731- 745

Hsieh, P.G ., and Ou,C.Y.(1998). “Shape of ground surface settlement profiles caused by excavation.” Can.

Geotech.J., 35(6), 1004-1017.

Hulme, T.W., Potter, L.A.C. and Shirlaw, J.N. 1989. Singapore MRT System: Construction. Proc. Institution of

the Civil Engineers, Vol.86, pp. 709–770.

Mana,A.L., and Clough,G.W.(1981). “prediction of movements for braced cuts in clay.” J.Geotech.Engrg.Div.,

107(6), 759-777.

Ou,C.Y., Hsieh,P.G., and Chiou,D.C.(1993). “Characteristics of ground surface settlement during excavation.”

Can. Geotech.J.,30(5), 758-767.

Peck, R. B. (1969). “Deep excavation and tunneling in soft ground” Proc 7th Int. Conf on Soil Mechanics and

Foundation Engineering, 225-290 Whittle, A. J., Hashash, Y. M. A., and Whitman, R.V. (1993). “Analysis of deep excavation in Boston.” J. Geo-tech. ngrg., 119(1), 69-90. Yong, K.Y., Lee, F.H., Parnploy, U., Lee, S.L. 1989. Elasto-plastic consolidation analysis for strutted excava-tion in clay. Computers and Geotechnics, Vol.8, pp. 311-328


103

1 INTRODUCTION

Work done in the Spanish rock massifs over the past years clearly illustrate the various methods that can be applied in tunnel boring, depending on the geology. We will discuss hereafter the example of three tunnelling works that were executed with three different machine types: Guadarrama, Pajares and Barcelona. Location of Pajares, Guadarrama and Barcelona Projects

Experience of hard rock tunnelling in Spain – Case studies of Pajares, Guadarrama & Barcelona Line 9

M. Merrie Technical Manager SE Asia, NFM Technologies, Singapore

T. Camus R&D Manager, NFM Technologies, France

ABSTRACT:

NFM Technologies have delivered TBM’s for the construction of several tunnels in Spain, which ex-

hibit particularly difficult geological conditions.

In this paper it shall be demonstrated how technological developments enabled Rock TBM’s to deal

with squeezing ground by use of displaceable main drives and over-cutting, to deal with carboniferous

shales where methane was an ever present concern through to the handling of complex geology in an ur-

ban environment where the innovative use of a Dual Mode TBM saved the contract the price of a new

TBM.

Guadarrama

Barcelona Pajares


107

2 GUADARRAMA

This site is located along the high-speed rail link connecting Madrid to Segovia. The tunnel goes through the Sierra de Guadarrama mountain range located North of Madrid. It is a 28 km long twin-tube tunnel of 8 m in diameter that are bored in a very hard granitic massif. Four TBMs were used, which were launched 2 by 2 on both sides of the mountain. Although the rock is generally only slightly fractured, both the launching phase and the fault zones could be a problem for gripper type machines. This led the Spanish contractors to opt for double-shield TBMs, with segmental lining laid along the complete drive length. Two contracts for one such machine were awarded to the Wirth-NFM group, both companies working together in close collaboration in order to design, build and commission the TBMs.

Double Shield Rock TBM

The main characteristics of the two TBMs are as follows: Table 1 - TBM Data – Guadarrama Double Shield

Type of Tunnelling Double Shield TBM

Excavation Diameter Ø 9.46 m, extendable up to 9,53

Length of TBM + Back Up 271 m with California switch Number of 17” disc cutters 66 No Total power 5600 kVA Drive unit power 4 000 kW Main / Auxiliary Thrust 10 450 kN / 108 000 kN Nominal / Unlock Torque 20 750 kN.m / 27 000 kN.m Variable speed 0 to 5 rpm Shield weight 1 000 Tonnes Total weight 1 750 Tonnes

The main challenge on this project consisted in organising the logistics for the disc cutters supplies and repairs. With an average progress rate of 600 m per month (and peaks at 900 m/month) through a very hard and abrasive rock the disc cutter consumption was indeed very high. More than 9,000 disc cutter replacements were needed over the boring of the complete drives, which represents a daily supply of 6 repaired disc cutters in order to keep the machines running. Manufacturing the corresponding spare parts, delivering new disc cutters, repairing and replacing worn parts, and installing refurbished disc cutters on the cutter head required a perfect organisation involving all partners so as to keep up with the machine progress.


108

Figure 1 – Production Rates - Guadarrama

3 PAJARES

This project is part of the high-speed railway line between León and Asturias in the Northwest part of Spain. This 25 km long twin-tube tunnel of 8,5 m inside diameter goes through the Cordillera Cantábrica mountain range, a carboniferous shale area. The tunnelling works involved 4 TBMs, 2 on each side of the mountain; a fifth TBM was used to bore an inclined shaft for safety and ventilation. Geology in this massif is complex, with sandstone and shale (~80 MPa) that can be highly fractured in some areas, dolomitic lime (~60 MPa), veins of carbonaceous shale, numerous fault and contact zones, folded structures, through which the tunnel was bored. The many problems raised by this difficult geological context are:

Risks of convergence – up to 100mm amplitude Wide range of UCS from high (sandstone) to low (carbonaceous shale) Water inflow (~800m overburden) High risks of methane High abrasiveness (can reach cerchar ~5,75)

Several reasons oriented the choice of the machine type to single shield: shorter shield length improving the machine capacity to escape squeezing in the converging ground, possibility of coping with water in-flows by sealing the shield, thrust ensured independently from the ground characteristics. The Wirth-NFM group delivered two such machines for the project. The main features characteristics of these machines are related to the ground characteristics. The cutter head can be translated both longitudinally by 600 mm and radially by 100 mm. This allows a 200 mm diameter extension in high convergence areas, avoiding trapping the shield. Diameter change from 9.9 m to 10.1 m was achieved by gauge cutters and adding two disc cutters on the cutter head periphery. Fur-thermore, fighting the squeezing conditions involved implementing a very high thrust capacity.


109

Cutterhead of Single Shield TBM

Ventilation was also designed with care, in order to ensure the methane/air mix would remain below the explosion point: this was achieved with powerful blowers (52 m

3/s) and housing part of the electrical

circuits in an ATEX enclosure. As it was planned to dismantle the machines inside the tunnel, they were designed with a sacrificial skin allowing this operation without the need for a cavern Disassembly of Shield with Sacrificial Skin

The main characteristics of the two TBMs are as follows: Table 2 – Single Shield TBM Data Type of Tunnelling Single Shield TBM

Excavation Diameter Ø9.9 m extendable up to 10,1 Length of TBM + Back Up 290 m with California switch Number of 17” disc cutters 69 (+2) Total power 8 000 kVA Drive unit power 5 000 kW Nominal / Unlocking Thrust 130 000 kN / 180 000 kN Nominal / Unlock Torque 25 000 kN.m / 30 000 kN.m Variable speed 0 to 5 rpm Shield weight 1 150 Tonnes Total weight 1 850 Tonnes


110

TÚNELES DE PAJARES

AVANCES ACUMULADOS A ORIGEN (17/01/2010)

0

500

1.000

1.500

2.000

2.500

3.000

3.500

4.000

4.500

5.000

5.500

6.000

6.500

7.000

7.500

8.000

8.500

9.000

9.500

10.000

10.500

11.000

11.500

12.000

12.500

13.000

13.500

14.000

14.500

15.000

0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1.000 1.050 1.100 1.150 1.200

DIAS DE CALENDARIO

ME

TR

OS

EX

CA

VA

DO

S

LOTE 1 - TÚNEL OESTE - HK

LOTE 1 - TÚNEL ESTE - NFM

LOTE 2 - GALERÍA + TÚNEL ESTE - HK


LOTE 4 - TÚNEL OESTE - MHI-DF


LOTE 5 - TÚNEL OESTE - HK

MONTAJE ANILLO Nº 1

LOTE 1 - HK 19/07/2005

LOTE 1 - NFM 13/09/2005

LOTE 2 - HK 14/09/2005

LOTE 3 - NFM 23/03/2006

LOTE 4 - MHI-DF 15/07/2006

LOTE 5 - NFM 09/10/2008

LOTE 5 - HK 29/05/2009

ANILLOS A ORIGEN

LOTE 1 - HK 9.982 anillos

LOTE 1 - NFM 6.571 anillos


-Galería: 3.634 anillos

- Túnel este: 2.755 anillos


LOTE 4 -MHI-DF 6.314 anillos



METROS A ORIGEN

LOTE 1 - HK 14.973,00 m

LOTE 1 - NFM 9.856,50 m

LOTE 2 - HK 9.583,50 m

-Galería: 5.451,00 m

- Túnel este: 4.132,50 m

LOTE 3 - NFM 10.236,00 m

LOTE 4 -MHI-DF 9.471,00 m

LOTE 5 - NFM 5.862,00 m

LOTE 5 - HK 4.759,50 m

INICIO DE EXCAVACIÓN

LOTE 1 - HK 22/07/2005

LOTE 1 - NFM 20/09/2005

LOTE 2 - HK 23/09/2005

LOTE 3 - NFM 8/04/2006

LOTE 4 MHI-DF 24/08/2006

LOTE 5 - NFM 20/11/2008

LOTE 5 - HK 16/07/2009

21

GALERÍA TÚNEL ESTE

4

3

5

6

Ver NOTA al pie

NOTA: Se incluye gráfica tomando como origen el inicio

del tramo correspondiente al Lote 2 (día 904)

7

8

LEYENDA

1.- DÍA 616 (21-05-07)

LOTE 1-TÚNEL ESTE : INICIO TRABAJOS DESMONTAJE TBM

2.- DÍA 630 (04-06-07)

LOTE 1-TÚNEL ESTE: FIN EXCAVACIÓN (anillo 6.571)

3.- DÍA 629 (04-06-07)

LOTE 2-GALERÍA DE ACCESO : FIN EXCAVACIÓN (anillo 3.634)

4.- DÍA 731 (19-07-07)

LOTE 1-TÚNEL OESTE : FINAL DEL TRAMO CORRESPONDIENTE AL LOTE 1

(anillo 7.326)

5.- DÍA 803 (25-11-07)


6,- DÍA 904 (08-01-08)

LOTE 1-TÚNEL OESTE: COMIENZO EXCAVACIÓN TRAMO LOTE 2

7,- DÍA 1.064 (16-06-08)

LOTE 1-TÚNEL OESTE: FIN EXCAVACIÓN TRAMO LOTE 2 (anillo 9.577)

8,- DÍA 1.074 (26-06-08)

LOTE 1-TÚNEL OESTE: COMIENZO EXCAVACIÓN TRAMO LOTE 4

9,- DÍA 1.116 (07-08-08)

LOTE 1-TÚNEL OESTE: FIN EXCAVACIÓN TRAMO LOTE 4 (anillo 9.982)

10.- DÍA 892 (30-08-08)


9

10c

The outstanding events concerning the tunnelling works concern the handling of expected geological ac-cidents. The team had to cope with important water inflows, roof collapse due to low cohesion rock, dif-ficulties in the directional driving at low thrust imposed by the low resistance of the excavation front, trapped machine as a result of maintenance stops, heavy wear of disc cutters in the sandstone areas. However, as the design had taken all these risks into account, the advance rate was very good with an average of 500 m per month and highs at 900 m/month. Figure 2 – Production Rates – Parjares

4 BARCELONA

The tunnel discussed here is part of line 9 of the Barcelona metro. This line, which is under construc-tion, is along one of its sections (26 km out of the 48 km total) built as a single tube tunnel of 10.9 m inner diameter.

Mixed Rock-EPB cutterhead for Barcelona TBM


111

The machine that NFM delivered for this job is in charge of boring a total of 12 km in several sections. It goes across a widely varied geology, from very hard rock (granitoid rock with porphyry inclusions, up to 180 MPa), folded and alternating shale layers, fragmented and weathered rock, clayey marls and completely weathered granite. The choice was made to use a Dual Mode machine that can operate in either EPB or open mode, with a belt conveyor muck transport system. In order to handle the geology changes along the drive, it was de-cided to design and manufacture several cutter heads, each one of them specifically adapted to the given corresponding geological context. Three cutter heads have therefore been provided. The first one follows a pure hard rock design, but with openings in the face and drag bits for soft ground, that allows it to be used in EPB mode. The second one is of the soft ground type, with arms and plates, bearing drag bits for both directions of rotation, and on which disc cutters or tilting scrapers can be mounted indifferently. The third cutter head is a genuine hard rock design, with a fully closed face, perfectly adapted to operate in open mode through the grani-toid rock section. The main characteristics of the Barcelona machine are as follows: Table 3 – Dual Mode TBM Data

Type of Tunnelling Dual mode machine

Shield type Single shield , passive articulation Excavation Diameter Ø11.95 m Length of TBM + Back Up 126 m Number of 17” disc cutters 71 (cutter heads 1 and 2)

81 (cutter head 3) Total power 8150 kVA Drive unit power 4725 kW Nominal / Max Thrust 90 000 kN / 110 000 kN Nominal / Unlock Torque 28930 kN.m / 37 000 kN.m Shield weight 1500 Tonnes Total weight 2200 Tonnes

The main challenge was to organise the design, fabrication and delivery of every cutter head according to the planning of operations, and to perform the exchange operation in sometimes difficult conditions. As an example, the following describes the original cutter head exchange operation executed at Maragall metro station where access directly above the cutter head was not possible.

The site configuration is as follows: a shaft is bored 15 m to the side of the tunnel, and a gallery (6.5 m in width and 15 m in height) is excavated using traditional methods so as to communicate with the main tunnel. This gallery is further equipped with two tracks on which a self-propelled truck can move. Upon arrival of the TBM, its cutter head emerges in the communication gallery. The current cutter is disconnected from the drive unit and positioned onto the truck with the help of hydraulic jacks. The truck can then be moved below the shaft opening, from where a crane hoists the cutter head up to the surface. Installing the new cutter head follows the same principle in reverse sequence. The complete exchange operation was executed in 3 weeks.


112

Cutter head exchange operation

Figure 3 – Production Rates - Barcelona

5 CONCLUSION

The Guadarrama project is quite standard, both for its geology and the design of the TBMs used. Works were difficult but the problems met were expected and the design ideas known for 20 years. Still many improvements were incorporated into these machines, resulting in significant progress in terms of per-formance and safety. Overall the Guadarrama TBMs are technically sophisticated and powerful. Both other projects are more unusual by the way the end users and the contractors tackled the associated difficulties. Both tunnelling jobs could have been executed with more standard machines, with a clear risk of geological accidents delaying the works. In both cases the contractors chose the approach exhib-iting the lowest risk level, planning for sometimes complex systems for the crossing of the geological ac-cidents. This approach requires great skills and experience in working in mixed faces, from both the TBM manufacturer and the civil engineering company. It also implies a close cooperation work between sup-plier and client, in co-designing the machine based on experience and know-how from the TBM operator and TBM manufacturer. The results of both above projects show the merits of this approach. Despite the precautions taken the machines were stopped because of the geological accidents, but the consequences of which on the TBMs and environment remained without gravity thanks to the specific design features.


113

1 INTRODUCTION

The Circle Line (CCL) is a fully underground orbital line, linking all existing Mass Rapid Transit (MRT) lines leading to the city. It comprises 29 underground stations of 33 kilometers long. This paper studies the settlement induced by CCL3 to 5 tunneling works. Circle Line Stage 3 to 5 which run through a long corridor around the city, starting from Bartly station and ending at HarbourFront station.

The Circle Line was built in stages. Most of the tunneling works were completed successfully by 17 Aug 2009. The circle line was opened in stages, the section from Bartley to Marymount was opened on 28 May 2009, the second section from Dhoby Ghaut to Bartley, was opened for passenger service on 17 April 2010. The final section from Marymount to Harbour Front is to be opened on 8 October 2011. Tunnelling along CCL alignment has been challenging due to the different geology encountered with properties and characteristics that vary very much along the route. The alignment of Circle Line was in-dicated on the geology of Singapore as shown in Fig 1. The construction of Circle Line tunnels encoun-tered all four major formations in Singapore as follows:

CCL3 tunnels were constructed in the residual soil and completely weathered granite of Bukit Timah Granite Formation (GV and GVI) and the Old Alluvium.

CCL4 tunnels were constructed predominately in the Bukit Timah Granite Formation in various

weathering grade.

CCL5 tunnels were constructed in residual soils and completely weathered sedimentary rocks of Jurong formation (SV and SVI).

The CCL 3 to 5 twin bored tunnels have an internal diameter of 5.8m. The lining is precast reinforced concrete, comprising of 5 segments plus one key. The cover of the soil above the tunnel crown ranges

ABSTRACT: Tunnel construction in urban areas is becoming popular for underground Metro Lines, thus resulting in many tunnels being constructed close to existing buildings and infrastructures. The ground settlement due to tunneling is largely dependent on the volume loss induced by the tunnel exca-vation. Little information has been published on actual volume loss encountered during tunnel construc-tion in various soil types. The construction of Singapore MRT has offered a great opportunity to study the ground response due to tunneling works. This study presents and discusses the volume loss caused by tunneling in different soil conditions and different TBM machines. Back analysis on trough width pa-rameter from the available data is also carried out. The main findings of the study are useful as a refer-ence for future projects.

Volume loss caused by tunneling of Circle Line projects

Y.H. Zhang Land Transport Authority, Singapore

W.M. Cham, J. Kumarasamy Land Transport Authority, Singapore


114

from 13m to 20m. The lateral distance between the centre-line of the twin tunnels varies from 9.2m to 16m. Generally, both the tunnels were located at the same vertical alignment. The two common types of closed-face tunneling shields are the slurry shield and the earth pressure bal-ance shield (EPB). Both were used in the tunneling works in CCL 3 to 5 projects. Detail information is summarized in Table 1 & Figure 2.

Fig.1 Geology of Singapore Circle Line

Fig.2 Circle Line 3, 4 & 5 tunneling drive


115

Table 1. Tunnelling information of Circle Line 3, 4&5

Circle Line Stage 5

854 856

Type TBM EPBM Slurry Slurry Slurry EPBM EPBM

CCL Drive LRC-SERSER-

BLYBSH-LRC MRM-BSH

BKB- TSN, TSN- MRM

BKB- BTN, BTN-FRR

ONH-BNV-

HLV-FRR

ONH-KRG-

HPV

LBP-PPJ-HPV LBP-

TLB-HBF

Max Drive Length (m)700 1200 1513 1237 3170 5800 3000 1100

GeologyGranite &

OA

Weatherred

Granite Granite

ONH-HLV:

Jurong HLV- Jurong Jurong

Outside Dia.(mm)6680 6720 6720 6630 6630 6600

TBM Length/ Back-Up (m)9.9/64.1 9.9/74.1 10.1/110 9.6/95 8.7/74 7.9/70

Circle Line Stage 4Contract

OA & Granite

6630

Circle Line Stage 3

7.6/70.4

855

EPBM

852 853

2 GROUND SURFACE SETTLEMENT MONITORING

Most of the monitoring data was extracted from LTA Geotechnical Database. Fig 3 shows the settle-ment trend plotted against time for settlement markers located within a typical array. It should be noted that negative displacements indicate settlement and positive displacements indicate ground heave. During the passage of TBMs, it can be observed that the settlements are mostly immediate settlements that happened within a short period. The short-term settlements are usually found to be almost complete when the TBMs are at a distance about approximately 15m beyond the monitoring array, i.e. about 2.5 times the tunnel diameter. .

Typical Ground Settlement Monitoring Data

-7

-6

-5

-4

-3

-2

-1

0

1

20/02/06 06/03/06 20/03/06 03/04/06 17/04/06 01/05/06 15/05/06 29/05/06

Date(DD/MM/YY)

Se

ttle

me

nt(

mm

)

Figure 3 Surface settlement trends plotted by the readings from database

The amount of settlement induced by TMB passages of only one tunnel bound was obtained from read-ing the surface settlement trend. Other factors such as site conditions, TBM operating parameters, con-struction timing and ground condition are also considered in the interpretation of the data. Cases such as unfavorable soil condition, cutter-head replacement, TBM launching and receiving, long term settle-ments due to consolidation were not considered in the study. Sufficient data were analyzed in order to ensure that the calculated volume loss is representative of the soil conditions and tunneling methods.

3 BACK ANALYSIS OF TROUGH WIDTH PARAMETER & VOLUME LOSS

3.1 Basis

Bored tunneling work will generally produce a settlement trough which is related to a Gaussian distribu-tion curve. To quantify the amount of volume loss due to tunneling, an empirical approach by Peck (1969) and New and O’Reilly (1991) is adopted. In the approach, the tunneling-induced transverse sur-face settlement trough is assumed to be given by a normal Gaussian distribution curve as in Equation 1


116

Where S is the surface settlement (mm) at a horizontal distance y (m), away from the tunnel centerline. Smax is the maximum surface settlement (mm) above the tunnel centerline and i is the horizontal dis-tance from the tunnel centerline to the point of inflection of the settlement trough (m). The point of inflection is determined from an empirical relationship given by the linear expression shown in Equation 2.

i = K zo

Where Zo is the depth of the tunnel axis (m) and k is the trough width parameter which depends on the soil type. By integrating Equation 1, the volume of the settlement trough per unit length, vs. is obtained by Equa-tion 3.

The above equation is valid based on the assumption that volume loss is equal to the volume of settle-ment trough. This is true as excavation takes place relatively quickly and therefore, it is unlikely to have any volume change of the clay. The loss, VL, can be written as in Equation 4.

Where D is the diameter of the tunnel (m). The equations described above are applicable to ‘Greenfield’ ground condition. The volume loss is ob-tained by fitting the Gaussian curve to each of the measured settlement troughs.

3.2 Study of trough width parameter

The settlement and the maximum settlement Smax for each monitoring array can be used to determine the settlement trough width parameter K based on the approach described by Mair et al (1993). According to Equation 2, parameter 0/ zi is needed to obtain trough width parameter and this could be obtained by plotting max)/(log SSe versus

2

0 )/( zx , where x is horizontal distance from the tunnel axis. This correlation is based on assuming that the shapes of the settlement profiles are characterized by a Gaussian distribution, shown in Fig 5. By rearranging Equation 1 and substituting K into it, the relationship between

max)/( SSLoge and 2

0 )/( zx is established as equation 5

2

0

22

max )/(/5.0)/( zxKSSLoge Equation 5 Once the max)/(log SSe -

2

0 )/( zx is plotted for each monitoring arrays, a best fit trend line by li-near regression can be developed. Trough width parameter K can be calculated from the slope of the best fit linear line as Equation 6

Equation 6

Equation 1

Equation 2

Equation 3

Equation 4

])(x/zversus)(S/Slogof[slope0.5K2

0maxe/

max2 iSVs

%1004

2

D

vV s

L

2

2

max2

expi

ySSV


117

Determination Trough Width Parameter K

(Residual Soil / Bukit Timah Granite)

y = -1.9337x

k=0.51

-2.25

-2

-1.75

-1.5

-1.25

-1

-0.75

-0.5

-0.25

0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

square(x/Zo)

Lo

g e

(S

/Sm

ax

)

Bukit Timah Residial Soil

Best Fit Line for GVI/GV ( from 10 arrays)

Determination Trough Width Parameter K

(SVI/ SV)

y = -1.6529x

k=0.55

-2.25

-2

-1.75

-1.5

-1.25

-1

-0.75

-0.5

-0.25

0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

square(x/Zo)

Lo

g e

(S

/Sm

ax

)

Jurong Residial Soil

Linear (Jurong Residial Soil)

Determination Trough Width Parameter K (OA)

y = -2.4173x

k=0.46

-2.5

-2.25

-2

-1.75

-1.5

-1.25

-1

-0.75

-0.5

-0.25

0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

square (x/Zo)

Lo

g e

(S

/Sm

ax)

OA

Best Fit Line for OA (from 9

arrays)

It is found that the trough width parameter is independent with volume loss. Before carrying out the study of volume loss, numbers of suitable sections are selected to work out the trough width parameter for residual soil / completely weathered granite (GVI/GV), residual soil / completed weathered sedimen-tary rock (SVI/SV) and the Old Alluviums. Couples of representative arrays have been selected for the study. Considering the influence zone range that normally one time of the tunnel depth, the selected settlement markers are within this distance. The results show (Fig 3) K values of 0.51, 0.55 and 0.46 for the residual soil and completely weathered gra-nite of Bukit Timah Granite Formation, the residual soils and completely weathered sedimentary rocks of Jurong Formation and the Old Alluviums respectively. It is to be noted that K of 0.45 is recommend-ed in LTA Civil Design Criteria (CDC) for the three soil types. Assuming a same amount of volume loss, the slightly lower K value would result in narrower settlement profile and larger maximum settle-ment; whereas higher K value results in wider settlement profiles, but smaller maximum settlement. Since the trough width parameter is critical for settlement prediction and building damage assessment. More reliable data are needed to verify the sensitivity and accuracy.

(a)Bukit Timah Granite

(b) Jurong Formation

(c) OA

Fig 4 Determination of trough width parameter from settlement data, (a)Bukit Timah Granite, (b)Jurong For-mation, (c)OA


118

CCL5 Ch80+950 Settlemnt Trough (Monitored vs Calculated)

-15

-10

-5

0

-30 -20 -10 0 10 20 30

Distance from Tunnel Center (m)

Sett

lem

en

t (m

m)

VL=0.35% LG1150 -1152

0

0.5

1

1.5

2

Numbers of points

Vo

lum

e L

oss(%

)

Mixed ground (Soil/Rock) + Soil like material

3.3 Study of volume loss

Calculation of volume loss is based on the equations indicated in section 3.1 above. The general se-quence is shown in Figure 5 for a typical cross section. The trough width parameter of 0.5 and 0.6 is adopted in the volume loss study in Bukit Timah Granite and Jurong Formation respectively, while OA remains 0.45 as CDC recommended.

1. Select surface monitoring arrays along the

alignment

2. Identify the immediate settlement due to tunneling

3. Plot the settlement trough from monitoring

data

4. Input some basic info such as existing ground level, top or rail level, distance of each monitoring points to tunnel centre line, select appropriate K( according to soil profile and produce the Gaussian curve

5. By adjusting the volume loss input to

match the measure settlement trough, the volume loss is obtained.

Figure 5. An indicative cross section settlement trough extracted from spreadsheet

240 monitoring arrays along the chainage of CCL Stage 3, 4 and 5 tunnels were selected for volume

loss calculation. The Figure 5 to figure 9 summarize the volume loss in different ground conditions with

either EPB or Slurry Machine. The volume loss ranges from maximum of 1.5% to a minimum of 0.2%.

Generally, the magnitude of volume loss is kept below 1%. The results are consistent with observation

from current construction practice using EPB shields in North-East Line.

0

0.5

1

1.5

2

Numbers of points

Vo

lum

e L

oss(%

)

Slurry in Soil like material (GVI/ GV)

0

0.5

1

1.5

2

Numbers of points

Vo

lum

e L

oss(%

)

Slurry in Soil like material (GVI/ GV)

0

0.5

1

1.5

2

Numbers of points

Vo

lum

e L

os

s(%

)

Slurry in Rock material (GIII/ GII)

0

0.5

1

1.5

2

Numbers of points

Vo

lum

e L

oss(%

)

Slurry in Mixed ground (Soil/Rock)

Fig 5 Volume loss in Bukit Timah Granite Formation by Slurry Machine

Fig 6 Volume loss in Jurong Formation by Slurry


119

0

0.5

1

1.5

2

Numbers of points

Vo

lum

e L

os

s(%

)EPB in Soil like material (GVI/ GV)

0

0.5

1

1.5

2

Numbers of points

Vo

lum

e L

oss(%

)

EPB in Mixed Ground (Soil/Rock)

0

0.5

1

1.5

2

Numbers of points

Vo

lum

e L

oss(%

)

EPB in Soil like material (SVI/ SV)

0

0.5

1

1.5

2

Numbers of points

Vo

lum

e L

oss (

%)

EPB in Old Alluvium

Ground

TypeGV/GVI GIII/GII

G(Soil/

Rock)

S(Soil/

Rock)GV/GVI

G(Soil/

Rock)SV/SVI SIII/SII

S(Soil/

Rock)OA

Tunnelling

MethodSlurry Slurry Slurry Slurry EPB EPB EPB EPB EPB EPB

K 0.5 0.5 0.5 0.6 0.5 0.5 0.55 0.55 0.55 0.45

V 0.7 1.5 1.5 1 0.8 1.5 0.9 0.5 1.5 0.9

Fig 7 Volume loss in Bukit Timah Granite Formation by EPB

Fig 8 Volume loss in Jurong Formation by Slurry Machine

Fig 9 Volume loss in OA by EPB

4 SUMMARY

This paper has presented and discusses the volume loss caused by tunneling in different soil conditions

and by different TBM types, the settlement data is from CCL Stage 3, 4 and 5 monitoring database. On-

ly immediate settlement is considered as the settlement induced by one bound tunneling passage, any

subsequent settlement or long term settlement is not taken into account. Large, or localized settlement is

not included, such like sink holes or other similar incidents. Back analysis of trough width parameter

from limited data has also been carried out. The main findings of the study are summarized in Table 3,

which are useful as a reference for future project. However it should be noted that the results were li-

mited to specific tunnel configurations and TBM performance. Nevertheless, these results provide valu-

able knowledge on ground responses to tunneling.

Table 3. Summary & Recommended Volume Loss and Trough width parameters

0

0.5

1

1.5

2

Numbers of points

Vo

lum

e L

oss(%

)EPB in Rock material (SIII/SII)

0

0.5

1

1.5

2

Numbers of points

Vo

lum

e L

oss(%

)

EPB in Mixed Ground (Soil/Rock)


120

Tunnelling in mixed ground condition is critical as it has resulted in larger settlement as com-

pared to other soil types. However with well controlled TBM operations, volume loss in such

ground conditions could be achieved within 1.5%.

Tunnelling in Old Alluvium has achieved a volume loss ranging from 0.1% to 0.9%. However

the majority of the volume loss is found within 0.8%.

Trough width parameter affects the magnitude and width of the settlement trough; it is indepen-

dent of volume loss. More accurate and complete monitoring data could be obtained from future

projects to refine the K value.

REFERENCE

Bowers K.H. & Moss. N.A., 2006, Settlement due to tunneling on the CTRL London Tunnels, Geotechnical

aspects of underground construction in soft ground – Bakker et al (eds), London, Pg 203 – 208

Eisenstein.Z., Geotechnical challenges in soft ground tunneling – examples from significant projects, Proc.

XIII ECSMGE, Vanicek et al. (eds), Prague. Pg 779 – 782

Lim .P.C. & Sigl O., 2000, An estimate of subsurface settlement trough width parameter for G4 material,

Tunnels and underground structures, Zhao, Shirlaw & Krishanan (eds), Balkerna, Rotterdam, Pg 507 -0513

Mair.R.J. 2006, TC28-tunnelling: reflections on advances over 10 years, Geotechnical aspects of under-

ground construction in soft ground – Bakker et al (eds), London, Pg 3 – 10

Osborne N.H., Williams O.I. & Lim W.B., 2000, Ground control of tunneling beneath an operating railway

tunnel in Singapore, Tunnels and underground structures, Zhao, Shirlaw & Krishanan (eds), Balkerna, Rotter-

dam, Pg 515 – 520

Taylor.R.N. 1995. Tunnelling in soft ground in the UK, Underground Construction in Soft Ground, Fujita &

Kusakabe(eds), Balkema, Rotterdam. Pg123 - 126

Wongsaroj J., Borghi F.X., .Soga K & Mair R.J. et al, 2006, Effect of TBM driving parameters on ground

surface movements: Channel tunnel rail link contract 220, Geotechnical aspects of underground construction

in soft ground – Bakker et al (eds), London, Pg 335 - 340


121

1 INTRODUCTION

Tunnel construction is becoming popular for underground Metro Lines which are usually built in densely urbanized areas. The influence of such works near to structures supported on piled foundations

has increasingly become a source of major concern. Tunnelling effects can be detrimental to pile

foundations as problems such as settlement of the ground and passive loading on the piles could arise.

The interaction between piled foundation and tunnel excavation is a complicated soil-structure-

interaction problem and it is not well understood at present. Little information is available for design

practice regarding the interaction between tunnels and piles. Land Transport Authority, Singapore, for example, specify that bored piles should not be constructed closer than 6m to existing tunnels.

However, no limits are specified for the reverse problem.

Conventional design procedures for the assessment of the risk of tunnelling-induced damage on

structures focused mainly on the assessment of the ’greenfield’ surface settlement trough above the

tunnel and its effect on the differential settlement to adjacent buildings. Although this may be

appropriate for buildings supported on shallow foundations, this is unfortunately, in most cases, ignoring the piled foundation due to difficulty and time consuming to perform accurate pile analysis.

However, it should be noted that apart from the building assessment, the additional loading caused by

sub-surface soil movement could lead to excessive pile settlement or pile structural capacity being exceeded. Under-designed pile foundations will be reflected on the superstructure such as cracks on

beam, column or wall and ultimately collapse if the damage is substantial. As a result, expensive

protective or mitigation works are usually proposed and this leads to high construction cost. Therefore,

Prediction of pile responses due to tunneling-induced soil

movement

W.M. Cham Land Transport Authority

ABSTRACT: In land scarce Singapore, the development of a comprehensive and efficient underground public transport system is the key to a sustainable transport system. This has resulted in

many tunnels being built in densely urbanized areas and at close proximity to buildings and

infrastructures. The tunnelling effects on these adjacent structures can be detrimental as problems such as settlement of the ground and passive loading on the piles could arise. The construction of large

scale infrastructure projects such as Circle Line and North-East Line has opened up enormous scope

for tunnelling and underground works in Singapore, which offers a great opportunity to study the

response of piles to tunnelling. In this paper, a two-stage approach using simplified numerical method in analyzing the performance of piles subjected to tunneling induced soil movement are described. The

accuracy of this approach is verified by comparison with reported two case histories. Subsequently,

this simplify numerical approach was used for parametric studies where it is shown that the influence of tunneling on pile response depends on a number of factors such as tunnel sizes, volume loss, soil

strength, pile diameter, lateral distance, and ratio of pile length to tunnel depth. Simple design charts

are also presented for estimating pile responses and may be used in practice to assess the behavior of

existing piles affected by tunneling operations.


122

further studies would be required to develop a better understanding of the problem and contribute to an

economic design

This paper intends to improve current knowledge for the prediction of pile response to tunnelling-

induced ground movements using proposed two-stage simplifies numerical approach. The field results from instrumented piles presented here will also provide a useful source for the validation of the

numerical approach. The construction of the Circle Line (CCL) and North-East Line projects in

Singapore offered the opportunity to obtain valuable data on the effects of tunnelling on piled

foundations, particularly Circle Line Stage 3 (CCL3) C852 and North East Line C704.

Field measurements on building settlements as a result of close proximity tunnelling are also presented

and analyzed. More than 300 building settlement markers were installed for buildings that are located within the influence of tunnelling.

2 TWO-STAGE NUMERICAL APPROACH ON PILE ASSESSMENT

An accurate assessment of existing piles affected by tunneling induced soil movements would require

the use of sophisticated three dimensional (3D) finite element modeling, which is complicated and

time consuming to perform. In this paper, a two-stage numerical approach is developed to analyze the lateral and axial responses of piles caused by tunneling. In the first stage, Greenfield soil movements

are estimated using an analytical method based on the work of Logananthan and poulos (1998). In the

second stage, these estimated soil movements are imposed on the pile in simplified boundary element analyses to compute the pile responses. A brief description of these two stages is given in the

following section.

2.1 Estimation of soil movement

The Gaussian curve shows a close match to tunneling induced soil movements at the ground surface.

However for pile foundation, it is expected that subsurface soil deformations will play an important role in influencing the pile response and Gaussian method does not give practical subsurface soil

deformations. Therefore a more representative tunneling induced subsurface movement estimation

method is required.

The estimation of the subsurface ground movements caused by tunneling is a complex soil-structure

interaction problem. This often has to be studied using a 3D finite element program and a good

prediction of soil movements will require a refined constitutive soil model such as nonlinear soil behaviour at small strains. However, in view of the complexities of the program and uncertainties of

the soil parameters, it is considered more practical to adopt simple analytical solutions which

considered an upper limit in prediction of ground movements.

Several analytical methods were developed in the past, such as those given by Verruijt and Booker

(1996) and Loganathan and Poulos (1998), which may be useful, particularly when there is inadequate detailed site information to carry out a proper complex finite element method. In this study, subsurface

settlement and the lateral deformations induced by tunneling are calculated based on close-form

analytical solutions presented by Loganathan and Poulos (1998) as follow:

(1)

(2)

(3)


123

where Uz=0= ground surface settlement; z = depth below ground surface;

Uz = subsurface settlement; R = tunnel radius; Ux = lateral soil movement; h = depth of tunnel horizontal axis level;

n = Poisson’s ratio; x = lateral distance from tunnel centerline

e = volume of the settlement trough

2.2 Axial response of pile

The analysis of axial pile response uses simplified boundary element analysis of the type described by

Phoon K K (2005). The numerical finite element method and its parameters are explained as follow:

In the analysis, the pile is modeled as an elastic element and the surrounding soil as soil spring

expressed in term of stiffness. The pile is divided into a series of finite number of cylindrical elements

of equally-spaced nodes. A schematic representation of the problem is shown in Figure 1. The vertical

pile movement of each element depends on the pile-soil interaction stresses, pile compressibility, stiffness of the pile and soil and also on any greenfield subsurface movements that are imposed on the

pile. The load deformation behaviour of the pile can be expressed as:

[Kp] {wp} = {P}

Where [Kp] = pile stiffness matrix

{wp}= nodal displacement vector {P} = nodal load vector

Figure 1. Discretization of problem, Phoon KK (2005)

To simulate real pile response more closely, allowance has been made for slip at the pile-soil interface,

i.e., the pile-soil interaction stresses cannot exceed the limiting pile-soil skin friction, fs. The -method is adopted as follow:

Ws

Wp

Ws

Pnsf P

Ws-Wp

Limiting fs

d


124

fs = ’v

where ’v = effective overburden pressure

= 0.25

For constant EA, the pile element stiffness matrix [Kp] is given by:

And for constant k

’s, the soil element stiffness matrix [Ks] is given by:

where z= depth below ground;

Ni = element node; k’s = soil stiffness;

d = equally-spaced nodes

Generally, by adopting

[Kp] {wp} = [Ks] {ws-wp}

Where [Kp] and [Ks] are pile and soil stiffness matrix and {ws} and {wp} are soil and pile movements.

Rearranging the equation, we can obtain:

[Kp+Ks] {wp} = [Ks] {ws}

By applying known {ws}, displacement of the pile {wp} at any depths can be calculated.

Once {wp} is obtain, pile axial force can be solved by

[Kp]{wp}={P} or

[Ks] {ws-wp}={P} or

Pz = EA/d (wp1-wp2)

The main soil parameters required for this analysis are the soil stiffness (k’s) and the soil base stiffness

(kb). They can be approximated based on soil shear modulus as follow:


125

Pile shaft Pile base

Displacement can be

approximated as:

Method based on Randolph and Wroth

(1978)

where G=soil shear modulus, rm=limit

radius, ro=pile radius

For a homogeneous soil in which the

stiffness increases with depth, Randolph And Wroth proposed the following

expression for rm.

Where L=pile length, vs=poisson ratio,

=ratio of GL2/GL

Method based on Fleming

(1992)

Soil stiffness

Soil parameters

Where E = soil modulus vs = poisson’s ratio

2.3 Lateral response of pile

The analysis of lateral pile response uses finite element approach of the type described by Chow and

Yong (1996). In the analysis, the pile is modeled as beam elements and the soil is idealized using the

modulus of subgrade reaction. A schematic representation of the problem is shown in Figure 2.

Figure 2 Idealisation of pile and soil

The input parameters for the analysis requires the pile bending stiffness, the distribution of lateral soil

stiffness, limiting lateral soil pressure acting on the pile with depth and the Greenfield horizontal soil

movements. The load deformation behaviour of the pile can be expressed as: p = kh(y-yo}

Where p = soil pressure acting on the pile

kh = modulus of subgrade reaction of the soil y = pile lateral deflection

yo = lateral soil movement

The element matrix equation for the pile-soil system subjected to lateral soil movements is given by

w

Pk s

bbb

v

Grk

1

4

bb

sbb

Gr

vPw

4

)1(

)1(2 sv

EG

Lvr sm )1(5.2

o

m

ss

r

r

Gk

ln

2'

y

dzzyNKyKyK ohsp )(}{}]{[}]{[


126

Where [Kp]= pile element matrix

[Ks]= soil element matrix {y} = vector of pile deformation

yo(z) = lateral soil movements

Kh = soil stiffness per unit length of the pile

The element stiffness matrix for pile and soil is given by:

Where Ep = young modulus of the pile

Ip = second moment area of pile section

{N} = vector of shape function L = length of pile

For a given set of lateral soil movements, the deformation of the pile can be determined by solving the

global stiffness matrix, and the bending moments and shear forces in the pile obtained from the resulting pile deformations.

The main soil parameters required for the analysis are lateral soil stiffness Kh and the limiting soil pressure py. They can be approximated as follow:

Lateral soil stiffness Limiting soil pressure

Estimated based on soil Young’s

modulus as follow:

For clay, Es is usually correlated to the

undrained shear strength cu as follow:

where 1 typically lies between 150 and 300. Cu can be estimated as 5N (SPT

values)

For sand, Es can be correlated to the

SPT blow counts as follow:

where 2 typically lies in the range between 1 and 2.

Based on Broms (1964)

For clay, py is related to the

undrained shear strength cu as

follow:

where Np can be assumed to vary

linearly from 3 at ground surface to

a limiting value of 9 at a depth of

about 3.5 pile diameter.

For sand, py can be determined as

follow:

where Kp is Rankine passive earth

pressure coefficient and ’v is effective overburden pressure.

3 CASE STUDY

Case histories on full-scale instrumented piles subjected to tunnelling effects are very scarce in the literature. The construction of the Circle Line stage 3 C852 and North-East Line C704 in Singapore

offered the opportunity to obtain valuable data in which pile settlement and pile forces induced by

nearby tunnelling were recorded. The two case histories have been analyzed to investigate the validity of the method presented in this paper.

3.1 Case study 1- North East Line C704

The first case history involved full-scale trial during construction of the new North East Line Mass

dzdz

Nd

dz

NdIEK T

L

ppp }}{{][2

2

2

2

0

dzNNKK TL

hs }}{{][0

upy cNp

sh EK

us cE 1

NEs 2

vpy Kp '3


127

Rapid Transit in Singapore where instrumented piles were used to monitor the effects of twin tunnel

construction on adjacent groups of bored cast in-situ piles as reported by Pang et al. (2005). The project involved the construction of twin bored tunnels adjacent to a 1.9km long vehicle viaduct. The

viaduct was constructed with 39 piers, which were supported by pile groups of 4 to 6 bored piles of

1.2m or 1.8m diameter. The 6.2m diameter twin bored tunnels were constructed at a depth of about 20m using EPB machines and were constructed close to the piles with clear distance ranges from 1.6m

to 4.4m. The tunnels were driven through residual soil of Bukit Timah Granite.

To monitor the tunnelling effects on the piles, twelve piles were instrumented with vibrating wire strain gauges and settlement markers. The piles were installed to depths ranging from 28 to 69m.b.g.l.

and generally varied in length due to the degree of weathering of soil. All piles were base grouted to

avoid soft toe problem. Sets of four strain gauges were installed at different levels along the piles corresponding to the tunnel axis level and above and below the tunnel crown and invert.

Figure 3 Cross section of tunnels adjacent to piled-foundation and plan view of instrumented pile

relative to tunnels (after Pang et al., 2005)

The measured and predicted results were presented for pier 20 as shown in Figure 4. The ground loss

was estimated at 0.5%. In general the results showed an increase in the induced axial load along the pile after the advancement of twin tunnels. The measured maximum axial forces were as high as 66 %

of the structural capacity and recorded near to the centre line of the tunnel. The results also showed

that the piles underwent transverse bending after the passage of the tunnels. The highest bending

moments were measured at the centre line of the tunnels and they were estimated to be considerably small.

The agreement between the predicted and the measured induced forces is good, although the predicted maximum induced forces are slightly over-predicted by 10%.


128

Figure 4 Measured and predicted induced forces for case study 1

3.2 Case study 2 - Circle Line stage 3 C852

The study described herein considered the tunnelling between Serangoon and Bartley station. One of the main challenges faced in Contract 852 was the construction of the twin tunnels directly beneath a

plot with five piled terrace houses.

The five terrace houses are situated within the tunnelling corridor as indicated in Figure 5. As the piles are adversely affected by the tunnelling works, it was decided to acquire temporarily the five houses

and rebuild them with foundation straddled over the tunnels. In view of the uncertainty, some of the

new piles were instrumented and several ground instruments were installed near to the piles to monitor the effects of tunnelling.

Figure 5 Layout of the 5 houses relative to tunnels’ alignment and instrmented piles

East-

bound West-

bound


129

The overlying geology comprises of backfill, Kallang Formation, residual soil of Bukit Timah Granite

and bedrock of varied depth due to different degrees of weathering. The tunnels were mainly driven through residual soil using Earth Pressure Balance Machines (EPBMs).

The foundations were bored piles with diameter 600mm, 800mm and 1000mm, ranging from 27 to 34 m long. In this study, 5 piles were installed with strain gauges. Figure 5 shows the layout of the five

instrumented piles relative to the alignment of the twin tunnels. The levels at which the strain gauges

were installed below ground were 2.5m, 5.5m, 19.3m, 21.8m, 25.3m and 28.3m. They were mostly

located between tunnel crown and tunnel invert level.

The results of the five working piles instrumented with strain gauges were evaluated to check the

response of piles during tunnelling. The study concluded that the maximum induced axial forces can be as high as 48% and 72% of the pile structural capacity after the advancement of single and twin

tunnels. Maximum induced bending moments are measured near to tunnel level and their magnitude is

small compared to their ultimate capacity. In addition, induced forces are found to be negligible for horizontal offset beyond 3 times the tunnel diameter.

Figure 6 shows the measured and predicted induced axial forces and bending moments. It is observed

that the shapes of the profiles are identical, however the predictions over-estimate the induced axial force and the bending moment by about 5% to 10%. In conclusion, although the proposed approach is

relatively simple, it is capable to perform a reasonable good analysis of pile response caused by

tunnelling.

Figure 6 Measured and predicted induced forces for case study 2

4 PARAMETRIC STUDIES OF PILE RESPONSES DUE TO TUNNELLING

The valuable field monitoring data of the case studies described in the paper was only limited to a

specific range of tunnel-pile configurations, tunnelling parameters and soil properties. Further

understanding of the tunnel-pile interaction problem outside the range is therefore required. To bridge


130

the gap of knowledge, a set of parametric studies is implemented using the two-stage numerical

analysis.

Details of the parametric studies and the influence of the parameters is assessed as follow: pile-tunnel

distance (Xpile

), pile length to tunnel depth ratio (Lp/Z

tun), volume loss (V

L), soil Young’s modulus (Es)

and pile diameter (Dpile

). This will provide an insight into the significance of each parameter to the

analysis of pile responses due to tunnelling. In the study, VL

of 1% was assigned and the soil was

modeled as homogeneous soil with Young Modulus, Esoil of 100MPa. Figure 7 shows the geometry

and pile position investigated. Xpile

/Dtun

of 0.5, 1.0 and 2.0 and Lp/Z

tun of 0.75, 1.0 and 2.0 were

investigated. Pile diameter was 1.0m and concrete stiffness of 28GPa was assumed. These positions were arranged in a practical way to cover the zones of influence on piles.

Figure 7 Tunnel-pile configuration

4.1 Pile settlement

Figure 8 shows the settlement of piles at various Lp/Z

tun and constant X

pile/D

tun of 1.0. Greenfield

subsurface soil settlement at the same location is also plotted in the same figure for comparison. Generally, the pile settlements were significantly influenced by the location of the pile tip relative to

the tunnel horizontal axis. For pile base located above the tunnel axis level, it was observed that the

pile settled in an almost rigid body movement. This implies very little slip occurred between soil and pile. It is also noted that the pile head settlement was almost equal to the greenfield subsurface

settlement. However, the pile settlement reduced when the pile length was increased (i.e. Lp/H

tun of 1.0

and 2.0). In this case, the pile head settlement became lesser than the greenfield subsurface settlement.

This is consistent with the field observation where pile settles differently at different zone of influence.

However, it should be noted that the piles in the analyses here were without working load and hence, lesser pile settlement.

The magnitude of pile head settlement was observed to be varying significantly with Xpile

, VL

and Dpile

as shown in Figure 8. Several important features were identified. Firstly, the pile settlement decreased with lateral distance. This implied that the pile settlements were more significant for pile located near

to the tunnel. Secondly, small diameter piles generally settled more than large diameter piles. On the

other hand, there was little effect of soil Young modulus on pile settlement. The change of pile

settlement was negligible for a reduction of soil young modulus, Esoil from 100000kPa to 50000kPa.

Ztunnel

Lpile/Ztunnel = 0.75, 1.0, 2.0

Xpile/Dtunnel = 0.5,1.0, 2.0

Xpile

Dpile

Dtunnel


131

Figure 8 Pile settlement

4.2 Pile axial force

Figure 9 shows the variation of maximum axial force with Lp/Z

tun, V

L, Dpile and Esoil. Generally, the

axial force increases with increase of Lp/H

tun and V

L. However the rate of increase for increasing

volume loss is not significant for Lp/Z

tun less than 1. Besides, the induced axial force also increases

with soil Young modulus and pile diameter. This is because, in a stiffer soil, the pile is subjected to a higher shear stress mobilization. This is also true for a bigger diameter pile, where the shaft area

subjected to shear stress mobilization is bigger and therefore, caused higher axial force.

Figure 9 Pile axial force


132

4.3 Pile lateral deflection

Figure 10a shows the pile lateral deflection at various Lp/Z

tun and constant x/Z

tun of 1.0. The greenfield

lateral soil movement at the same position is also plotted for comparison purpose. It can be observed

that the piles deflected similarly to the shape of soil movement. This was due to the low bending

stiffness of the piles compared to the surrounding soil. However, for a pile which is installed at a

deeper soil where the pile base is located outside the zone of influence (i.e. Lp/H

tun>1.0), the

restraining effect took place. The soil outside the zone of influence is subjected to a negligible soil movement and it provides a restraint to the pile deflection. Therefore, a slightly smaller pile deflection

is expected, comparing to the soil movement.

Figure 10b shows the pile lateral deflection at various tunnel-pile distance. As observed, the pile

lateral deflection is highly dependent on the distance between tunnel and pile. At a closer distance to

tunnel (i.e. x/Dtun

=1.0), the lateral deflection is higher and the maximum deflection point occurs at the

tunnel centerline. The lateral deflection became smaller as the distance increases and the maximum lateral deflection shifted to the pile head where the lateral soil movement is largest.

(a) (b) Figure 10 Pile lateral deflection

4.4 Pile bending moment

Figure 11a shows the bending moment profile of piles at various Lp/Z

tun and constant x/D

tun of 1.0.

Generally, the tunnelling-induced bending moment increased with Lp/Z

tun. For pile base located

outside the zone of influence, the bending moment was greater comparing to the pile base located

within the zone of influence. This was caused by the restraining effect of pile length extended outside

the zone of influence. Similarly, the bending moment increased with decreasing lateral distance between the pile and the tunnel and the maximum values were observed to locate near to the tunnel

centerline.

(a) (b) Figure 11 Pile bending moment


133

Further analyses show that the maximum bending moment increased with increasing Lp/Ztun as shown

in Figure 12a and it seen to reach a constant value when Lp/Z

tun was greater than 1.5 and for x/D

tun

greater than 2.0. It is also observed in Figure 12b that maximum bending moment increased with increasing pile diameter. However, there was little effect of change in soil Young modulus on pile

maximum bending moment.

(a) (b) Figure 12 Pile bending moment (effects on pile length and pile diameter)

5 DESIGN CHARTS

From the two-stage numerical approach, a set of design charts are developed for use to estimate the

maximum pile responses due to tunnelling such as the induced bending moment, axial forces, pile

settlement and lateral deflection. Figure 13 shows the maximum values of various pile responses varying with the distance away from the vertical axis of the tunnel. These values will form part of the

basic design charts for the three simple cases involving the following assumptions:

a. The soil is a deep homogeneous soil layer with Young’s modules Esoil=100MPa b. The pile diameter Dpile=1.0m

c. Three different pile lengths and depth of tunnels are considered. The first case is a short pile of

Lp=15m and Ztun=15m; the second case is a medium pile of length Lp=20m and Ztun=20m and the third case is a long pile of length Lp=30m and Ztun=30m

d. The tunnel diameter Dtun=6.6m

e. The volume loss is assumed at 1% and 3% for practical purpose

From the design charts, it can be seen that the induced maximum axial force (Fp), maximum bending

moments (Mp), maximum lateral deflection (δH) and pile head settlement (δV) all decrease with

increasing lateral distance (xpile). Other observations include: a. bending moment increases significantly when xpile <10m

b. maximum axial force is larger for longer pile

c. lateral deflections are almost independent of pile length d. vertical pile head settlements are larger for shorter pile length at xpile <10m


134

Figure 13 Maximum pile responses for short, medium and long pile

Correction factors for soil Young’s modulus, pile diameter and pile length to tunnel depth ratio were

derived from the parametric study and are presented in chart form as shown in Figure 14, 15 and 16.

The basic response for specific case is first read followed by applying correction factors for particular case. The various maximum pile responses may be approximated as follow:

Axial response:

Lateral response:

F

ZL

F

D

F

Ep tunppilesxCxCCF /max

V

tunp

V

pile

V

s ZLDEv xCxCC /max

M

ZL

M

D

M

Ep tunppilesxCxCCM /max

H

tunp

H

pile

H

s ZLDEH xCxCC /max


135

Figure 14 Correction factor for Dpile

Figure 15 Correction factor for Esoil


136

Figure 16 Correction factor for Lp/Ztun

5.1 Limitations

The following limitations are to be noted when using the two-stage numerical approach and the design charts:

1. The lateral and axial pile responses are computed separately in this study. In reality, both

responses occur together and ignoring the combined loading could under-estimate the bending moment in the pile.

2. The design charts compute only the additional pile response, assuming that the pile is stress free

with no working load on pile head before tunnelling. 3. Pile groups are beneficial in reducing the effects of pile responses. Therefore the evaluation of the

effects of tunnel construction on a single pile considered in this study will be conservative.

4. Pile head condition has a significant effect on bending moment in pile. In this study, the pile head

is freed from translation and restrained from rotation. 5. The use of design charts inevitably resulted in some discrepancy in the values calculated as

compared to values computed directly from the two-stage numerical analysis. In general, the

discrepancy is found to be no more than 15%

6 CONCLUSIONS

A simplified two-stage numerical approach is proposed for the prediction of pile responses due to

tunnelling-induced soil movement. Two published case histories on full-scale instrumented piles are

studied in which the measured lateral and axial pile responses are compared with the predicted values.

Reasonable good agreement is found and the prediction is found not greater than 10% over-estimation. A parametric study shows that the pile responses are influenced by a number of factors, including

tunnel-pile configurations, ground loss ratio, soil strength and stiffness and pile diameter. Simple

design charts are also presented for estimating the maximum pile responses and may be applied to


137

practical problems. The study intends to add to the knowledge and contribute towards a better

understanding of the effects of tunnelling-induced soil movements on the behavior of piles.

REFERENCES

Loganathan, N. and Poulos, H. G. 1998. Analytical prediction for tunneling-induced ground movements in clays.

Journal of Geotechnical and Geoenviromental Engineering, Vol. 124, No. 9, pp. 846-856.

Chen, L.T., Poulos, H. G. and Loganathan, N. 1999. Pile responses caused by tunneling. Journal of Geotechnical

and Geoenvironmental Engineering, ASCE, Vol. 125(3), pp.207-215.

Chen, et. al. 2000. Approximate design charts for piles adjacent to tunneling operations. Proceeding of GEOENG

2000, Melbourne.

Pang, C. H. 2006. The effects of tunnel construction on nearby piled foundation. PhD thesis, National University

of Singapore.

Cham, W.M. 2007. The response of piles to tunnelling. Degree of Master of Science (M.Sc) and Diploma of Imperial College (DIC), Imperial College, London.

Chen, L. T. Poulos, H.G. and Loganathan, N. 1999. Pile responses caused by tunneling. Journal of Geotechnical

and Geoenvironmental Engineering, ASCE, Vol. 125(3), pp.207-215.

Chow, Y.K. and Yong, K.Y. 1996. Analysis of piles subject to lateral soil movements. Journal of The Institution

of Engineers, Singapore Vol.36, No.2, pp. 43-49.

Loganathan, N. and Poulos, H. G. 1998. Analytical prediction for tunneling-induced ground movements in clays.

Journal of Geotechnical and Geoenviromental Engineering, Vol. 124, No. 9, pp. 846-856.

Pang, C.H. 2006. The effects of tunnel construction on nearby piled foundation. PhD thesis, National University

of Singapore.


138

ABSTRACT: Following rapid development, underground space in Singapore is fast becoming more and more congested. Tunnels crossing existing infrastructure, such as underground stations and tunnels, have become a feature of practically every new MRT line in Singapore. This paper gives an example of stacked mined connection tunnels crossing under an existing C&C tunnel with live MRT tracks, using sprayed concrete linings and sequential excavation. The upper connection tunnel had to cut through the toe of the existing D-wall panels of the Circle Line (CCL) cut & cover tunnels, with live running MRT track. It was also necessary to remove the toes of two bored piles supporting the exiting underground linkway of CCL Promenade station, which was operational and open to the public at the time of tunnel construction. The paper presents design considerations for the SCL tunnel lining as well as the impact assessment on existing adjacent structure. Some aspects of the construction of SCL tunnels, the actual performance of cut & cover tunnel as well as a summary of monitoring results are also presented

1. INTRODUCTION

Contract C905 of the Singapore MRT Downtown Line 1 was awarded to Shimizu Corporation on August 2007.It comprises design and construction of tunnels under Marina bays connecting Promenade station to Marina Bay Sands. Layout plan and major structures of C905 are shown in Figure 1.

Figure 1 - Site Layout plan and Major Structure of C905

Design and construction of NATM/SCL tunnel under an operating MRT tunnel for DTL1 Contract 905

O. Sigl & J.J. Lin Geoconsult Asia Singapore


139

2. DESIGN CONSIDERATION

2.1. Site layout

The majority of connection tunnels had to be constructed beneath the base slab of the existing cut and cover tunnels at the South East corner of Circle Line (CCL) Promenade station. In addition, 2m to 3m of the bottom part of the existing diaphragm wall together with the toe of two bored piles supporting the underground linkway of CCL Promenade station, had to be removed. Therefore, two NATM/SCL connection tunnels are proposed to fulfil this purpose. These two tunnels are stacked vertically and have similar circle shape and internal size. Figure 2, Figure 3 and Figure 4 show the layout plan and section of connection tunnels respectively.

In this paper, we will focus on the upper connection tunnel.

Figure 2 –Layout Plan of Connection Tunnels to existing CCL Promenade Station

2.2. Ground Condition

Subsurface condition along the connection tunnels was well established from adjacent boreholes. It revealed that the soil types from ground to tunnel invert comprise mainly the old alluvium, which was overlain by Fill and Kallang Formation. The soil to be encountered within tunnel face was described as hard Sandy Clay and hard Sandy Silt with SPT values more than 100.The depth to upper tunnel axis was approximately 28.7 m below ground level.


140

Figure 3– Section of Connection Tunnel

2.3. Existing Cut & Cover Tunnel and Linkway

Existing cut & cover tunnel was a rigid stacked box structure, which consisted of 3 layers of concrete slabs and a 1.2m thick diaphragm wall as retaining wall and foundation.

The Underground linkway was a rigid single box structure. Roof and base slab were all connected to the existing Promenade station diaphragm wall by using cast-in reinforcement couplers. The base slab was designed to sit on the roof slab of the cut & cover tunnel. Along its alignment the linkway was sitting on a number of 800mm diameter bored piles, providing foundation along the linkway’s outer edge.

Figure 4– Section through Existing Structure of Circle Line C&C Tunnel


141

2.4. Necessity of Underpinning Works

The upper connection tunnel will pass underneath the cut & cover boxes at an acute angle affected the C&C diaphragm wall over a length of approximately 10m. These diaphragm wall panels were originally designed as earth retaining structure as well as deep foundation elements. Removing part of the diaphragm wall will cause loss of their original design capacity.

In the design of the SCL tunnels, the lost capacity was converted to a force and applied as a force to the tunnel lining. The same concept was also applied to the two numbers of bored piles, PI-9 & PI-10. In other words, the diaphragm walls and piles were designed to partly sit on the SCL tunnel lining.

However, it is crucial to make sure that the existing cut & cover and linkway structure are safe during tunnel excavation stage, especially just after the diaphragm wall or bored piles are cut off. This related to a stage, where the SCL lining has not been applied yet or the freshly applied shotcrete has not obtained the required strength yet.

Therefore, the first step is to investigate what is the actual load taken by the diaphragm walls and piles as well as their remaining structural capacities, in order to determine the requirements for the intended underpinning works.

Based on As-built drawings, detailed loading computation for the affected diaphragm wall panels was carried out, which included diaphragm wall panels Y39, Y40 & Y40a, Y41, Y41a & Y41b, Y42 and Y43. In addition, the geotechnical and structural capacity was calculated, based on shorter length.

It was found that, except for panels Y39, Y40 and Y40a, the capacity of remaining D-wall was still capable to resist the design loads. However, for panels Y39, Y40 and Y40a the difference between the design loading and remaining D-wall capacity was found to be only marginal. Therefore, instead of underpinning, it was proposed to remove about 1m of top soil above tunnel, meanwhile ensuring that no ground surcharge can occur in this area during the construction period. As shown in Table 1, by implementing this simple measure, all the design loading of the existing D-walls are less than their remaining structural capacity. The same assessment was carried out for bored piles PI-9 and PI-10. Again, following this same approach, no underpinning works for existing linkway was required.

Table 1: Comparison of D-wall/pile Design Load and Capacity - 1m of Top Soil Removed

Item Number

Design Loading Design Loading (1m soil removed) Remaining

Capacity

DL

(KN/m)

LL

(KN/m) DL+LL

DL

(KN/m)

LL

(KN/m) DL+LL KN/m (KN)

D-wall (1.2m)

Y39, Y40 1535 155 1690 1462 69 1531 1533

Y40a 1477 155 1632 1458 69 1527 1533

Y41 1060 79 1139 1035 79 1114 1200

Y41a, Y41b 782 79 861 782 79 861 1200

Y42 1039 111 1150 1014 83 1097 1200

Y43 1037 111 1148 1012 83 1095 1200

Bored

pile (Dia

0.8m)

PI-9 2950 559 3509 2510 70 2580 2943

PI-10 3284 631 3915 2778 69 2847 2943

2.5. General Arrangement of Lining System of Connection Tunnel

Since tunnel design life of all permanent structures was specified to be 120 years, besides using Sprayed Concrete Lining to provide ground support, it was proposed to install standard precast


142

segmental tunnel lining as secondary lining in order to fulfil the durability requirement. The precast segmental lining segments are the same as used in the bored tunnel section (275mm thick with internal diameter of 5.8m). Therefore, the excavation diameter of the connection tunnel was finally determined at 7.65m. The design of the precast segment lining will not be presented here.

Figure 5–Removal of 1m of Top Soil on Site

3. DESIGN OF SPRAYED CONCRETE LINING (SCL)

3.1. General

The following loads and partial safety/material factors were considered in the analysis, acting simultaneously in Ultimate Limit State considerations with reference to SS CP65.

Table 2: Load Factors

Loads Partial Safety Factor

Effective soil pressure 1.4

Groundwater pressure 1.2

Ground surface surcharge of 20 kN/m2 1.6

Loading from D-wall or Piling (applicable only for the upper connection tunnel)

1.5

Both upper as well as lower bound of the groundwater level were taken into account in deriving hydrostatic loads, whichever was more critical for the structure or the loading condition being considered.

Upper bound: 0m below ground

Lower bound: 5m below ground


143

3.2. Design Methodology

The analysis for the upper connection tunnel SCL lining was carried out using plain frame analysis by computer program Staad.pro. The lining was modelled as beam elements and the ground bedding support to the lining was modelled using radial uni-directional (axial) compression–only springs, along the beam elements.

Total of four analysis sections along the upper connection tunnel were selected in order to carry out plane frame analyses, which considered different load locations at which the diaphragm wall would transfer to the shotcrete lining:

Section 1 – Loading from diaphragm wall apply to lining at 45 degree direction

Section 2 – Loading from diaphragm wall apply at crown of tunnel

Section 3 - Loading from diaphragm wall apply at crown of tunnel plus pile loading of PI-9

Section 4 - Loading from diaphragm wall apply at 45 degree direction plus pile loading of PI-10

For each analysis section, two types of geometrical section considered in the analysis:

1) Full circle lining – long term case after construction works completed

2) Half circle lining – short term case during construction, in order to simulate the top heading excavation.

Figure 6–Analysis Sections along the Connection Tunnel


144

Table 3: Distribution of D-wall/pile Design Load

Analysis Section S1 Analysis Section S2

Analysis Section S3 Analysis Section S4

3.3. Lining Structural Design

The distribution of loads which are described in Section 3.2 above, were calculated and applied as primary load cases in the analysis. The relevant primary load cases (LC 1 to 10), as summarised in

Table 4 below, were then combined with the respective partial load safety factors to form the Ultimate Limit State and Service Limit State design. Reinforced concrete design for the SCL lining was carried out in accordance with CP65. Respective M/N-interaction diagrams were used to check sections subjected to axial compression and bending moments. Shear Force (V) considering Axial Force (N) was checked as well.

In addition, since the diaphragm wall of the cut & cover tunnel and piles PI-9 & PI-10 will eventually rest on the SCL lining under long term condition, the design also checked compliance to the serviceability requirement of permanent works. Therefore the crack width was checked to be less than 0.3mm.

The design requirements for the shotcrete lining for upper connection tunnel are summarised in Table 5 below.


145

Table 4: Summary of Primary Load Cases & Factor for Load Combination

Primary Load

Case

(LC)

Loads

Primary Load Combination - ULS

Primary Load Combination -SLS

Upper Bound

GWT

Lower Bound

GWT

Upper Bound

GWT

Lower Bound

GWT

1 Lining self-weight 1.4 1.4 1.0 1.0

2 Groundwater pressure (Upper Bound) 1.2 – 1.0 –

3 Effective soil vertical pressure (Upper Bound) 1.4 – 1.0 –

4 Effective soil horizontal pressure (Upper Bound) 1.4 – 1.0 –

5 Groundwater pressure (Lower Bound) – 1.2 – 1.0

6 Effective soil vertical pressure (Lower Bound) – 1.4 – 1.0

7 Effective soil horizontal pressure (Lower Bound) – 1.4 – 1.0

8 Vertical surcharge pressure 1.6 1.6 1.0 1.0

9 Horizontal surcharge pressure 1.6 1.6 1.0 1.0

10 Loading from D-wall or piling 1.5 1.5 1.0 1.0

Table 5: Design Requirements for Upper Connection Tunnel

Location Stage Total Lining

Thickness[mm]

Reinforcement

Each Face

Shotcrete

Grade

Upper (outside of cut & cover tunnel)

Top Heading

Bench

Invert

550 T8 – 150 / 150 wire mesh C30

Upper (Rest)

Top Heading

Bench

Invert

550 T8 – 150 / 150 wire

mesh+T25-150 vertical C30

4. IMPACT ASSESSMENT

As shown in Figure 7 below, a 3D model using program code Plaxis 3D Foundation program was

developed in order to study the impact of the construction of connection tunnel to the adjacent

structures. Three grid line length (about 50m long) of existing station and cut and cover tunnel

together with shaft B were simulated in the 3 dimensional analysis model. Embedded piles are

introduced to simulate the existing underground linkway bored piles. The length of D-wall and bored

piles which was cut during the tunnel excavation was deactivated in the process. However, due to the

limitation of the program, the connection tunnels were simulated by soil cluster with equivalent

excavation area and applying a volume metric strain to the soil cluster in order to simulate the volume

loss caused by the excavation of the lower and upper connection tunnels.


146

Plan View Perspective

Figure 7–Plaxis 3D Model

Displacements of existing structure from analysis are presented in Table 6 below. The values are all

less than the allowable values as per the LTA performance criteria.

Table 6: Summary of Predicated Structure Displacement

Structure

Maximum Displacement due to

Excavation of connection NATM Tunnel LTA Performance Criteria

Vertical

Settlement

(mm)

Lateral displacement (mm)

Vertical

Settlement

(mm)

Lateral

displacement

(mm)

C825 Station D Wall 8.34 8.36 10 10

C825 Station Slab 8.34 2.6 10 10

cut & cover D wall (Y39, Y40 & Y40a) 7.15 5.30 10 10

cut & cover D wall (Y42, Y41, Y41a&

Y41b) 8.28 8.24 10 10

Cut & Cover Tunnel Slab 8.28 1.5 10 10

Linkway wall 8.5 1 10 10

Linkway Slab 8.27 1 10 10


147

Figure 8–Upper Tunnel Layout Plan

Figure 9–Upper Tunnel Layout Plan

5. CONSTRUCTION OF CONNECTION TUNNEL

Prior to commencement of excavation, a detailed pre-construction study of exact location of the

diaphragm wall at each round of the excavation was conducted and drawn up. This approach enabled

the construction team to have a clear understanding of the situation that they will face during mining

works. It was also beneficial for teams of QP(D), LTA and QP(S) when supervising the works. Figure

8 presents the layout plan and Figure 9 shows a few selected sections of this study.

Excavation of upper connection tunnel was started in early June 2010. Circle Line Stages 1 and 2 were

opened to the public on 17 April 2010. Thus excavation of the connection tunnels had to be carried out

underneath the live MRT running tracks.

The excavation sequence was sub-divided into stages. The staging was kept on constant review and

revised slightly if required. During construction, face mapping for each round of top head, bench and

invert was conducted. Actual ground conditions and actual locations of encountered diaphragm wall


148

panels and bored piles were found to accurately correspond with the expectations established in pre-

construction studies.

Figure 10 present a few selected photos of the excavation face with exposed D-wall panels and bored

piles. Construction of the upper tunnel was successfully completed in early September 2010.

Round 6 Top Heading Round 8 Top Heading

Round 9 Top Heading Round 21 Top Heading

Figure 10–Photos of Excavated Face at Various Stages

6. INSTRUMENTATION AND MONITORING

In order to verify the validity of design assumptions, a comprehensive instrumentation and monitoring system was implemented. In addition to ground surface monitoring arrays, optical 3D prisms were installed within the cut & cover structure to monitor the performance of the life running track bed.

Monitoring data was accessible online through a dedicated web site. If a particular reading exceeded the alert level during excavation, the system automatically sent out an SMS to the key persons of the project team identifying the instrument location, the alert level and actual current reading value.

D Wall Panel

Y39

D Wall

Panel

Y40

D Wall Panel Y40

D Wall

Panel

Y41

Existing Bored

Pile PI-9


149

Plan of instrumentation at cut and cover tunnel Typical section of cut and

cover tunnel

instrumentation

Figure 11–Instrumentation at Cut and Cover Tunnel

7. PERFORMANCE OF EXISTING STRUCTURE

Monitoring result showed that the displacement of cut and cover tunnel remained relatively small, with maximum vertical settlement and horizontal displacements in the range of about 2 mm. Figure 12 and Figure 13 present 3D real time monitoring results. It can be concluded that the construction of connection tunnel did not cause any damage to the existing CCL structures or caused concerns with respect to train operations.

-12

-10

-8

-6

-4

-2

0

2

4

6

8

10

12

1-O

ct-

09

21

-Oc

t-0

9

10

-No

v-0

9

30

-No

v-0

9

20

-De

c-0

9

9-J

an

-10

29

-Ja

n-1

0

18

-Fe

b-1

0

10

-Ma

r-1

0

30

-Ma

r-1

0

19

-Ap

r-1

0

9-M

ay

-10

29

-Ma

y-1

0

18

-Ju

n-1

0

8-J

ul-

10

28

-Ju

l-1

0

17

-Au

g-1

0

6-S

ep

-10

26

-Se

p-1

0

dX

(mm

)

Change in dX at C825 Upper Tunnel_above DTL(SCL)

XYZ8214 XYZ8215 XYZ8217 XYZ8218

Sta

rt of D

TL-L

(SC

L)

W.S.L

W.S.L

A.L

A.L

70% A.L

70% A.L

Sta

rt of D

TL-U

(SC

L)

Figure 12–Vertical settlement of upper cut and cover tunnel


150

-12

-10

-8

-6

-4

-2

0

2

4

6

8

10

12

1-O

ct-0

9

21

-Oct

-09

10

-No

v-0

9

30

-No

v-0

9

20

-De

c-0

9

9-J

an-1

0

29

-Jan

-10

18

-Fe

b-1

0

10

-Mar

-10

30

-Mar

-10

19

-Ap

r-1

0

9-M

ay-1

0

29

-May

-10

18

-Ju

n-1

0

8-J

ul-

10

28

-Ju

l-1

0

17

-Au

g-1

0

6-S

ep

-10

26

-Se

p-1

0

dZ(

mm

)

Change in dZ at C825 Upper Tunnel_above DTL(SCL)

XYZ8214 XYZ8215 XYZ8217 XYZ8218

Sta

rt of D

TL-L

(SC

L)

W.S.L

W.S.L

A.L

A.L

70% A.L

70% A.L

Sta

rt of D

TL-U

(SC

L)

Figure 13–Horizontal displacement of upper cut and cover tunnel

8. CONCLUSIONS

Construction of the Downtown Line Stage 1 C905 connection tunnel using sequential excavation method and sprayed concrete linings was successfully completed without causing any damage to the existing CCL structures. Construction impact assessment using 3D modeling was useful tool to predict settlements of the existing structures.

In addition to the proper design approaches and analysis methods, detail pre-construction studies proved to be very important and useful, such that the relevant project teams including contractor, QP(D), QP(S) as well as the LTA could prepare adequate and reliable plans based on the expected conditions.

The organizational set-up allowed the construction team to closely review all construction information, like monitoring results, encountered face conditions, shotcrete testing results. Based on this set-up, the team had enough information available at all time, in order to react quickly and implement adjustments to construction elements such as round length, application of shotcrete sealing layer, use of forepoling, probe drilling, etc.


151

1 INTRODUCTION An economy leading large city like Tokyo has made its prosperity by concentrating people in one place. This concept has encouraged the construction of high rise buildings in the city. However the area which is appropriate to construct high rise building seems to have left a small space already and the city sys-tem above ground has become quite complex now. Thus current large city has excessive number of population and the community is quite crowded. This problem is not only seen in Tokyo, but also found in many large cities around the world. The underground construction technology has developed dramatically recently and deep underground is now an attracting new frontier. In facing above issue, Japanese government has been encouraging re-searches and constructions to utilize deep underground frontier extensively (Ohtsuka et al. November 1986, Takatsuji et al. September 1987, Kazama et al. July 1996). Several civil construction projects in-cluding tunnels and infrastructures has been conducted and built already. The utilization of deep under-ground by civil constructions is gradually increasing. However, few approaches from building construc-tion have been conducted yet. The authors have been participating in research project, “New usage of underground space in the fu-ture”, led by Urban underground space center of Japan, which is sponsored by Mori Building Co. Ltd. This report summarizes the proposal submitted by working group 3 of this research project. Major advantages in using deep underground are summarized as follows: High seismic performance can be achieved. Lateral support by soil in underground reduces earth-

quake load demand in structure. Structure above ground has to withstand not only the lateral ground movement but also the amplification in the structure itself. There were extensive amount of damages found in structures above ground, whereas very few were found in underground constructions after Kobe earthquake in 1995.

A design concept for a deep underground and high-rise

building complex in Tokyo, Japan

M. Higashino & N. Ito Takenaka Corporation, Japan

Y. Kobayakawa Working Group 3 of the Research Project “New usage of underground space in the future”, Japan

ABSTRACT: The deep underground frontier is now very attractive to improve complex city system

above ground in a large city like Tokyo. Many infrastructures at deep underground level have been con-

structed by now, but few developments have been made in building constructions. Major issues to solve

to construct deep underground structures are soil and water pressure, seismic load, environmental condi-

tions, and construction cost. The authors propose a new configuration of super high rise complex build-

ing utilizing 100m deep underground. This proposal employs circular retaining wall, sunshine court be-

tween ground building and super high rise building, and also energy dissipating system between these

two buildings. This paper summarizes the technical merits of proposal and discusses the issues to over-

come. The subjects in this paper is the result from research project “New usage of underground space in

the future”, led by Urban underground space center of Japan. The report was submitted by Working

Group 3 of this research project.


152

Heat load is less and the running cost will be minimal. As the sides of the building are not exposed to sunlight and air, the heat load, hot and cool, is less compared to the building above ground.

Greening is facilitated by leaving ground surface open. To enjoy above advantages, following issues need to be solved: Low habitability without natural ventilation and sunlight. High construction cost. Usually the cost amounts 1.5 to 3.5 times that of constructing buildings

above ground. Evacuation route is complex. Giving solutions to these issues and to maximize the advantages, the authors propose the new concept of high rise complex building utilizing deep underground. This paper summarizes the outline of the concept and explains the strong points. 2 OUTLINE OF THE PROPOSED CONSTRUCTION 2.1 Assumed location of the site and its soil characteristics Since the theme is to facilitate city functionality by building construction utilizing deep underground, one of the down town area in Tokyo, Kasumigaseki, was chosen as the site for study. The area of the site is 140m by 140m, and the depth of basement is 100m below ground level. The outline is shown in Figure 1. Soil at the site is mainly composed of sand, gravel, silt, and loam. They are all diluvia under top rec-laimed layer of few meters. The stiffness of the diluvia layers 10m below ground level is high enough that N value by standard penetration test is over 50, which is high enough to hold piles to support super high rise buildings. However, the average water level of Tokyo Bay, Tokyo peil, is found at 4.9m below ground level, and effect from the water needs to be taken into consideration in engineering the retaining walls. 2.2 Architectural features Three major architectural features are summarized as follows: The building complex is largely divided in two sections, private zone and public zone. Private zone is

further divided in two parts, office which is in super high rise building, and retailing stores which is in underground building which incorporates retaining wall. This underground building also includes parking lots. The public zone is in the most deep underground construction. This section structurally supports super high rise building and circular underground building. Total floor area of private zone sums up to 381,000 m

2, which is 1900% in floor space index, whereas the total area of public zone

is 75,000 m2, which is 380% in floor space index.

The “Sunshine court” is provided to divide circular underground building and super high rise building. This hollow section allows natural sunlight to reach the bottom of the court. By finishing the exterior wall facing sunshine court of underground section of two buildings with glass curtain wall, the sun-light is designed to reach as deep as 56m from ground level. This depth is adopted also from the limit of natural ventilation inside the court.

The roof level of circular underground building is set equal to ground level and left open as Public Park which incorporates greens. The area of this park extends as large as 60% of the area of the site.

The outline of the usage of the building is summarized in Table 1. 2.3 Configuration of structural system The structural system is summarized in following four aspects: Outer wall of underground building is designed as circular diaphragm wall which functions as self

supporting retaining wall. This construction enables to avoid shoring, strutting, supporting works thus the construction period is shortened and costs for temporary works are reduced. The circular di-aphragm wall is permanently resistant against lateral static pressure and seismic loads, thus the inner


153

structural system of underground building can be constructed only by columns and flat slabs. The adoption of flat slab allows freer space for usage. Figure 1 Outline of proposed building

Table 1 Usage of building and floor area

Location Usage Floor area (m2) Notes

Private zone

Super high rise build-

ing

Above underground

park

Office 250,000 Daily habitant zone.

Also can be used as hotel,

residence or retail shops.

Underground building

B1F to B12F

Retail shops

Parking lots

90,000 Environment is controlled

same as floors above ground.

B13F Underground

public park

11,000 Environment is controlled

same as floors above ground.

B14F and B15F Administration

Machinery

30,000

Subtotal floor area 381,000 Floor space index 1900%

Public zone

B16F Administration

Machinery

15,000

B17F to B21F Emergency evac-

uation,

Storage, and

Interconnection of

underground in-

frastructure

61,000 Water tank, reservoir, sewage

processing, transformer

room, DHC, refuse disposal,

Bus terminal, subway station

Subtotal floor area 75,000 Floor space index 380%

Total floor area 456,000

平面図

20 20 60 20 20 140

20

20

60

20

20

140

地下階棟

サンシャインコート

地表面

超高層棟

Sunshine court

▽GL

B15F

B1FLE

10F

20F

30F

40F

50F

20ｍ 20ｍ

57F

ダンパー

B21F

耐圧版

地下階棟

サンシャインコート

公共地下広場

管理・機械室

公共施設ゾーン

超高層棟

断面図

民間ゾーン(

建築ゾーン)

公共施設ゾーン

100

m

27.

5m

56m

2m

Super high

rise build-

ing

Dampers Underground

building

Super high

rise building

Underground

building

Horizontal plan Cross section

Mat slab

Public park

Administration

Machine

Sunshine court

Public zone

Private zone

Public

zone

Dampers

(m)


154

Pre-stress is applied to flat slab with compression by taking reaction from diaphragm wall since the diaphragm wall is the retaining wall to support external soil and water pressure. When open cut ex-cavation is adopted, oil jack is used to apply pre-stress into flat slab from inside surface of the wall. This pre-stress enables to expand the span of columns to support the flat slab that it further increases the open space for usage. If top-down construction method is adopted, flat slabs are constructed as excavation progresses. Compression from soil and water let the diaphragm wall to incline inside and this deformation induces compression inside the flat slabs. Pre-stressing will be automatically done in this case and jacking system may not be necessary. Method using open cut construction is studied in the following section.

Adoption of pumping water method to prevent floating is considered. Although there still are numbers of discussion if the floating force of underground structure can be reduced by water pumping, this method is worth considering that the cost of structure to transfer the weight of the building is quite large when the pumping method is not employed. Since the soil at the bottom of underground build-ing is mudstone, the coefficient of permeability is less than 10

-6cm/s. Also the welling of the water at

nearby site was measured as almost 50m3/day. By considering these facts, floating force can be tech-

nically controlled to almost zero by penetrating the bottom of the diaphragm wall into the mudstone layer and pumping up the water outside.

The employment of structural control system is considered to reduce earthquake response. As de-scribed in architectural features, the structural system is divided by Sunshine court into underground building and super high rise building. These two structures are connected with dampers at ground level. Also, since super high rise building is standing from GL-100m level, the bottom of the building is considered as bedrock and there is no amplification of earthquake by surface ground characteris-tics. By these two configurations, the earthquake response of super high rise building is reduced dramatically. Earthquake response of underground building is very small since the outside wall of the building has the contact with surrounding soil.

3 EXTERNAL FORCES External forces to consider are earth pressure, water pressure subjected to diaphragm wall of under-ground building, earthquake load and wind load. The wind load to this super high rise building is same as usual super high rise building that it is not discussed in this paper. 3.1 Lateral static pressure Based on Foundation Design Guideline of 2001 (Architectural Institute of Japan, Japan Association of Wall Foundation 2006) total static earth pressure and water pressure subjected to diaphragm wall is evaluated with following equation.

zzz ww PPzKP 00 (1)

Where P0 is total pressure at depth z, K0 is the coefficient of earth pressure at rest, is the density of soil, z is depth, and Pw(z) is water pressure at depth z. The coefficient of earth pressure at rest for sandy soil is evaluated with Jaki’s theory (J. A. Jaki ). The pressure is evaluated as shown in Figure 2. When referring to design and construction guideline of retaining wall (Advanced construction technolo-gy center 1994), and assuming the soil is basically composed of clay below GL-50m, the pressure is calculated by following equation.

qzKP az 0 (2)

Where, Ka is coefficient of active earth pressure and q is vertical pressure at ground surface. While pressure at GL-100m is calculated as 1160kN/m

2 using equation (1), 915kN/m

2 is obtained using equa-

tion (2). Pressure using equation (1) is used in following study.


155

Table 2 Dimension of underground building

Member Depth of wall Thickness (mm)

Diaphragm

wall

0 to -31.5m 1000

-31.5 to -58.5m 2000

-58.5 to -100.0m 2400 (Current

study case)

Flat slab 300mm

Column B1 to B13F =1000

B13F to B 21F =1500

Mat slab 5000

Figure 2 Profile of the pressure (a) At engineering bedrock (b) At ground surface Figure 3 Earthquake spectra: Pseudo velocity

Figure 4 Stress inside diaphragm wall Figure 5 FEM analysis model

P

P

P

P

Psinθθ

Rｄθ

σTσT

RT

ｄθ θP

P

P

P

P

Psinθθ

Rｄθ

σTσT

RT

ｄθ θP

0

10

20

30

40

50

60

70

80

90

100

0 200 400 600 800 1000 1200

圧力（ｋN/m2)

深度(m)

土圧

水圧側圧

Pressure (kN/m2)

Depth (m)

Soil pressure

Water pressure

Total pressure

0

50

100

150

200

0.1 1.0 10.0

Period

Kobe

Hachinohe

Kushiro

Kobe

Hachinohe

Kushiro

0

50

100

150

200

0.1 1.0 10.0

Period(s)

200

100

0 0.1 1.0 10.0

Period

(m/s) 200

100

0 0.1 1.0 10.0

Period

(m/s)

(s)

Foundation: Mat slab

Column

Diaphragm wall

Pre-stressed flat slab


156

3.2 Earthquake load Since the building is constructed at GL-100m level and the bottom of super high rise building is at GL-56m level, the earthquake ground motion is considered at engineering bedrock which does not have am-plification in surface soil. The spectrum of ground motion stipulated in Building Standard Law at engi-neering bedrock is shown in Figure 3. To artificially compose earthquake ground motion, phase spectra from Hachinohe 1968 NS, Kobe 1995 NS, and Kushiro were employed. To compare the earthquake re-sponse of proposed super high rise building with ordinal super high rise building which is built on ground surface, artificial earthquake ground motions, which have amplification of surface soil, were composed for the ground surface also. The spectra for these earthquakes are also shown in Figure 3. 4 STRUCTURAL SYSTEMS AND THEIR PERFORMANCES 4.1 Circular diaphragm wall To understand the mechanics in the diaphragm wall, the concept of the relation between external and in-ternal forces are illustrated in Figure 4. In contrast to the square diaphragm, moment is not induced in circular diaphragm either in circumferential or radial direction but only the compression force. The compression stress is evaluated as follows:

T

PR (3)

Where is compression stress, R is the radius of circular wall, P is external pressure and T is the thick-ness of wall. The pressure at 100m depth is 1.1N/mm

2. Also assuming R=70m, T=2.4m the compres-

sion stress is calculated as 34.0N/mm2. This value is well below the allowable stress when the high

strength concrete is employed. Fc=80N/mm2 is assumed in this structural design considering the stress in

re-bars. FEM, Finite Element Method, analyses were conducted to evaluate internal force and the displacement precisely. External force shown in Figure 2 for vertical distribution and uniform distribution in the cir-cumference are assumed. Also the dimensions shown in Table2 were adopted. Analyzing code MSC/NSATRAN was used. Structural model is shown in Figure 5. Among several cases of analyses conducted, the case considering the construction process is shown as follows. Open cut construction me-thod is employed and maximum stress and deformation as excavation inside the diaphragm wall pro-gressed to the depth 100m. The slab at the bottom of the structure is not considered in the analysis that construction stage without bottom slab is the condition induces largest stress and displacement. Maximum radial displacement: 50.8mm ( at GL-100m ) Maximum circumferential stress: 0.92N/mm

2 ( at GL-0m)

Minimum circumferential stress: -24.5N/mm2 ( at GL-100m)

To study the effects from eccentric distribution of soil and water pressure, a case study by increasing 10% of external force on half of diaphragm wall, from 0 degree point to 180 degree point in the circum-ference, was carried out. The essences of the results of analysis are as follows: Maximum radial displacement: 41.1mm to 65.3mm ( at GL-100m ) Maximum circumferential stress: 2.2N/mm

2 ( at GL-0m )

Minimum circumferential stress: -36.6N/mm2 ( at GL-100m )

Displacements at the bottom of diaphragm wall are translated as 1/1970 deformation angle in average for the uniform external load and 1/1530 for eccentric loading. The deformations in both cases are small enough that cracks induced by displacement will be minimal. The compression stress evaluated in the analysis is smaller than the estimation by approximate calculation. Small level of tension is observed at GL-0m, but this is well within the level to be carried by re-bars. These results support the adoption of the concrete strength of Fc=80N/mm

2 in construction.


157

Figure 6 Details of jacking point Figure 7 Stress model of circular flat slab Figure 8 Stress distribution inside circular flat slab 4.2 Pre-stressed flat slab The dimensions of the flat slab system of the underground building are 140m in diameter, 20m in width and 0.3m thick. These dimensions are the same for all slabs in underground building. The slab carries 5kN/m

2 live load together with the weight of the slab itself.

The slab system is pre-stressed with hydraulic jacks at 36 jacking points on the rim of flat slab. Each jack is designed to apply 1960kN toward the center of the circular slab by having the reaction from out-er diaphragm wall. The detail of the jacking point is illustrated in Figure 6. These jacks are removed af-ter applying the load and filling the slit between flat slab and diaphragm wall with grout. The stress state induced by pre-stress inside the slab is expressed theoretically as follows considering the model shown in Figure 7(Timoshenko & Goodier 1951):

2

2

22

2

1r

a

ab

bpr (4)

ｒ

アンカー鉄筋

主筋

地下連続壁

波形鋼板

ジャッキ

スラブ

後施工スラブ受け機械式継手

溶接

ジャッキ

スラブ地下連続壁

後施工スラブ受け

アンカー鉄筋

主筋

地下連続壁

波形鋼板

ジャッキ

スラブ

後施工スラブ受け機械式継手

溶接

ジャッキ

スラブ地下連続壁

後施工スラブ受け

Anchor rod Welding

Corrugated steel plate

Jack

Flat slab

Slab support Mechanical

joint Main re-bar

Diaphragm wall

B13F～B20F

左下：孔なし(a=0)の場合 1F～B12F

右下：孔半径 a=50ｍの場合

P

P

P

P

P

P

P

P

-3.08p

-4.08p

σθ

σr σr

σθ

-1.0

p

-1.0p

50m

70m

0

-1.0

p

Left bottom: Without center hole

(a=0)

Right top: With center hole

(a=50m)


158

2

2

22

2

1r

a

ab

bp (5)

Where r is radial stress, is circumferential stress, p is external pressure, a is outer diameter, b is in-ner diameter of circular slab and r is the distance from the center of the circle to the point in considera-tion. In case of current dimension of the building, a=50m and b=70m, the stress state is calculated as shown in Figure 8. Detailed stress state and deformation state are studied by FEM analyses. The results for full loading condition of dead load, live load, and pre-stress are summarized as follows: Circumferential stress: -1.99 to -1.24N/mm

2

Radial stress: -4.04 to 0.37N/mm2

Vertical deformation: -9.9mm at the midst of the column and rim Horizontal deformation: -2.5mm at rim The stress inside flat slab is small compared to the concrete strength of Fc=80N/mm

2. Tension is ob-

served as 0.37N/mm2 but the level is further smaller. With these results, the reinforcement required in

the slab is minimal but enough to make the span of columns to 12m. If top-down construction method is employed, compression stress inside slab becomes higher without using jacking system. Slab system may be able to carry earthquake load as well as static loads in this case. Study of stress state for top-down construction case needs to be conducted in future study. 4.3 Measures to deal with floating force in underground building Total floating force by underground water at the bottom of underground building is 13,600MN since the water level is GL-10m. Total weight of the building is calculated as is shown in Table 3. Vertical load by the total weight of the building exceeds the floating force at the bottom thus the floating force will be cancelled by the weight of the building if the weight is equally distributed at the bottom with well engi-neered shear transmission system. However the water pressure overwhelms if the weight of diaphragm wall and the friction cannot be taken into account. So the method to deal with remaining floating force need to be considered. Conventional method to deal with floating force is to place earth anchors at the bottom to counteract. To cancel the floating force by the weight of the building, the shear transmission wall system of 18.5m high and 4.0m thick in the basement structure is considered herein as shown in Figure 9. However this wall system blocks horizontal transportation inside the basement and the cost amounts very large. Table 3 Vertical load by the weight of each section of the building (MN)

Mat slab at bottom 1,900

Underground building: B1 to B15 2,100

B16 to B21 1,400

Super high rise building 2,400 Diaphragm wall 2,700

RC inner wall 900 Friction from surrounding soil 5,200

Subtotal 8,700 Subtotal 7,900

Total 16,600

Table 4 Estimated seepage water depending on permeability and evaluation method

Coefficient of permeability (cm/s) 1.0 x 10-4 1.0 x 10-5 1.0 x 10-6

Artesian well method (m3/day) 451 45 5

Gravity well method (m3/day) 4,385 873 216


159

Figure 9 Wall system to deal with floating force. Figure 10 The idea to draw out water.

Instead of dealing with structural system to transmit vertical loads, the method to draw the water out

from under the bottom of basement to reduce the floating force is studied. The evaluation of the volume

of seepage water needs the evaluation of the coefficient of permeability. Since the data to evaluate per-

meability is not enough, three cases of the coefficients were used in the study (Miyake et al. September

2000, Tomizawa et al). There are two different methods, gravity well method and artesian well method,

which corresponds to shallow well and deep well respectively, to estimate seepage water and evaluated

volumes for these two methods are compared in Table 4. There are large differences among the results

by two methods, but observation more coincides with the results by artesian well method. Extensive dis-

cussions still need to be carried out if the measure to draw out the water is applicable in practice since

the drawn water could be called as “waste” depending on the volume and might not be permitted to be

flowed out. One idea of drawing water method is illustrated in Figure 10. 4.4 Structural control system to reduce earthquake load

Time history analyses subjected to earthquake excitation are carried out for the super high rise building

model and for the conventional super high rise building model built on ground surface as the comparison.

Analytical model for these two buildings are shown in Figure 11. Building is modeled in 57 lumped

masses for section above ground and 13 masses for the section of underground level. Hydraulic dampers

are modeled as dashpots which connect 13th mass with ground surface. Nonlinear characteristics are

considered for the stiffness of the structures. Earthquake excitation at engineering bedrock is used for

the proposed building and earthquake at ground surface is used for conventional building built on

ground surface.

▽GL

B15F

B1FLE

67.5

m22.5

m5

5

B21F

耐圧版壁（4層分：ｈ＝18.5ｍ）

2015601520

140

2015601520

140

20

15

60

15

20

140

20

15

60

15

20

140

壁（ｔ＝2,000mm）壁（ｔ＝4,000mm）

Wall (t=2,000mm) Wall (t=4,000mm)

Mat slab Wall: 4 floors, h=18.5m

Reservoir

Connecting pipe

Reservoir

Connecting pipe Mat slab Water tank Machine room

Underground building

Super high rise building

Sunshine court

Lowered water level by continuous pumping


160

Figure 11 Dynamic analysis models. Figure 12 Maximum response profiles. Table 5 Comparison of initial cost, running cost and emission of CO2.

Initial cost

(million Yen)

Running cost

(million Yen/year)

CO2 emission

(ton/year)

Conventional building (A) 50,900 4,600 61,700

Proposed building (B) 64,800 3,500 5,000

(B-A) Increase rate (B/A-1) 13,900 +27% -1,100 -28% -11,000 -18%

The profiles of maximum responses for the acceleration and story drift angle are shown in Figure 12.

Artificial earthquake using Hachinohe phase spectrum is used in these analyses. Acceleration response

of proposed building is smaller compared to the response of conventional building. This result is the ef-

fect of longer natural period in proposed building and the smaller excitation level for earthquake at engi-

neering bedrock. Also the story drift angle of conventional building is reduced by lower excitation level

and energy dissipation of dampers. For both acceleration and story drift angle, the effect of higher

damping over 5% capacity is not as effective compared to result of 5% damping which is clarified by

parametric study. Thus 5% is enough. Also maximum force of one damper is 80,000kN and maximum

stroke is 33cm for each direction. 5 EVALUATION OF COST AND THE EMISSION OF CARBON DIOXIDE

Since the outer wall of underground building is not exposed to the air and also does not get sunlight, the

heat load is minimal. The advantages of Lithium battery, LED light bulb, solar batteries, biomass, geo-

thermal heat etc are fully taken into consideration in facility planning. The estimated initial cost, running

cost and emission of carbon dioxide are summarized in Table 5. The annual emission of carbon dioxide

is estimated as 50,730 t. This amount is 18% less compared to that of conventional building with same

total floor area. Initial construction cost is estimated to increase 27%, i.e. 14 billion yen, compared to

conventional building because of deep underground construction and the employment of state of the art

technologies in facility planning. However, the state of the art facility contributes to reduce running cost

by 28%, i.e. 1.1 billion yen per year. This saving in running cost will balance the increase in initial cost

in 12.7 years.

57質点

地表波

G.L.

基盤波

57質点

13質点

G.L.

57 nodes 57 nodes

13 nodes Earthquake at ground surface

Earthquake at bed rock

Proposed model Conventional model

Damper

-15

0

15

30

45

60

0 200 400(gal)

(F)

-15

0

15

30

45

60

0 0.004 0.008 0.012(rad.)

(F)

Maximum acceleration Maximum story drift

Proposed

Conventional Proposed

Conventional

(cm/s2)


161

6 STUDY OF EVACUATION Assuming 58,370 people are working in whole building, evacuation time is evaluated by simulation ana-lyses as follows: Maximum floor evacuation time from underground building: 12.8 minute Maximum entire building evacuation time from underground building: 56.4 minute With these results, all people working in the complex is evaluated that they can evacuate within 1 hour in case of emergency event. Following data are of a 47 floor conventional super high rise building. We can see that evacuation time of proposed underground building is not too large compared to current su-per high rise buildings. Maximum floor evacuation time from conventional super high rise building: 18.5 minute Maximum entire building evacuation time from conventional super high rise building: 40.0 minute 7 CONCLUSIONS The results of the study on deep underground use by super high rise complex building are summarized as follows: The employment of circular diaphragm wall as the retaining wall of underground building facilitated

the engineering of the mechanics in the structure to deal with high soil and water pressure. This con-struction enables to avoid shoring, strutting, supporting works thus the construction period can be shortened and costs for temporary works can be reduced. The idea has been utilized in civil engineer-ing for more than 20 years but this proposal is the first to utilize the idea in building construction.

Since the diaphragm wall is the retaining wall to support external soil load and water pressure, it is proved to be used as reaction force point of oil jack to apply pre-stress into flat slab from inside the wall. This pre-stress enables to expand the span of columns to support the flat slab that it increases the open space for usage. Also, since the shape of the slab is as simple as pure circular plate with circular hole at the center, the stress state is easily predicted by the theory of elasticity by Timoshen-ko.

Since the soil at the bottom of underground building is mudstone, the coefficient of permeability is less than 10

-6cm/s. Also the welling of the water at nearby site was measured as almost 50m

3/day.

Considering these facts, floating force can be technically controlled to almost zero by penetrating the bottom of the diaphragm wall into the mudstone layer and pumping up the water outside. This will be the first to employ the technique in building construction.

The “Sunshine court” is provided to divide circular underground building and super high rise building. This hollow section allows natural sunlight to reach the bottom of the court and allows natural venti-lation inside the court. By connecting the underground building and super high rise building with dampers, the earthquake response of super high rise building will be 30% less compared to construct building above ground.

The building utilizes underground construction thus reduces the running cost. This will balance the increase in construction cost in almost 13 year. Less energy consumption will reduce the emission of carbon dioxide dramatically.

ACKNOWLEDGEMENT Subjects shown herein are the results from the research project “New usage of underground space in the future”,

led by Urban underground space center of Japan, which is sponsored by Mori Building Co. Ltd. This report

summarizes the proposal submitted by working group 3. The authors would like to express their thanks to Dr.

Shigeru Ito, Dr. Toshikazu Takeda, Mr. Shinya Kobayashi, Mr. Shyoichi Kobayashi, Dr. Takashi Shiokawa,

Mr. Tetsuya Matsumoto, Mr. Masahiro Aburakawa, and Mr. Masayuki Muto.


162

REFERENCES Ohtsuka, Masahiro et al. November 1986. Evaluation of external force on deep underground cylindrical retain-

ing wall, Proceedings of 41st Annual meeting of Japan Society of Civil Engineers. Takatsuji, Tetsu et al. September 1987. Construction planning of No. 5 and 6 TEPCO Futtsu thermal power

station, Electric power civil engineering No. 210. Kazama, Ryo et al. July 1996. Design and construction of large scale self standing retaining wall ( No. 1

through 8), 31st Meeting of Japan Geotechnical Society (Kitami). Architectural Institute of Japan, Recommendations for design of building foundations. Japan Association of Diaphragm Wall Foundation. January 2006. World’s No.1 Diaphragm Wall Foundation.

Foundation engineering and equipment. J. A. Jaki. Pressures in soil, Proceedings of 2nd ICSMFE Vol. 1 Advanced construction technology center 1994, Guideline of design and construction of deep underground re-

taining wall. Timoshenko, S. P. and Goodier, J. N. 1951. Theory of Elasticity. McGraw-Hill. Miyake, Noriharu, et al. September 2000. Permeability of large scale variable thickness cylindrical diaphragm

wall and the behavior of underground water during and after the construction. The proceedings of AIJ an-nual meeting(Tohoku).

Tomizawa, Shuji et al. The characteristics of mudstone which consists the foundation soil of trans Tokyo Port bridge. Soil and Foundation. No. 35-3: 350.


163

1 INTRODUCTION

The stiffness of soil is an important parameter that affects the deformations of ground and adjacent structures arising from construction activities such as tunnelling and deep excavations. Whilst constitu-tive models and analytical methods have been derived to predict soil deformation from its stiffness, en-gineers face a difficult task of identifying soil stiffness from routine site investigations. A tool to deter-mine the soil modulus in-situ is the pressuremeter. The pressuremeter consists of a long cylindrical device, placed in a borehole and radially expanded into the surrounding ground. It is designed to apply a uniform radial pressure to the sides of the borehole through a flexible membrane, and the pressure is increased to create an expansion of the borehole cav-ity. During the pressuremeter test, measurements of the applied pressure and the corresponding expan-sion of the cavity are taken so that they may be interpreted into ground properties. The cavity expansion is measured either by measuring the volume of fluid change in the cylindrical device or by measuring the change in cavity radius using displacement transducers. Furthermore, the type of pressuremeter is also differentiated in terms of how the pressuremeter is installed. Figure 1 (as extracted from BRE, 2003) shows the three main types of pressuremeter testing that are available – (a) the pre-bored pressuremeter which is installed in pre-formed boreholes, (b) the self-boring pressuremeter which is able to form its own hole with minimal disturbance on the ground, and (c) the full displacement pressuremeters which is inserted into the ground without soil removal and the ground is displaced by the passage of the pres-suremeter. Mair and Wood (1987) and Briaud (1992) discussed the different types of pressuremeter test methods and interpretation. The pressuremeter tests are usually conducted to procedures described in ASTM D4719 (2000) and BS 5930 (1999).

A pressuremeter’s perspective on soil stiffness

K.H. Goh, W.M. Cham & D. Wen Land Transport Authority, Singapore

ABSTRACT: The stiffness of soil is an important parameter that affects the deformations of ground and adjacent structures arising from construction activities such as tunnelling and deep excavations. Whilst constitutive models and analytical methods have been derived to predict soil deformation from its stiffness, engineers face a difficult task of identifying soil stiffness from routine site investigations. This paper discusses the use of pressuremeter testing in site investigations to estimate the soil modulus for design. The paper begins by summing up the use of pre-bored pressuremeters in in-situ testing. In particular, the interpretation of pressuremeter curve to obtain the reload modulus will be discussed. Instead of de-scribing the pressuremeter test results using a single modulus value, it is proposed to interpret the pres-suremeter modulus in relation to the corresponding strain levels for each test. By correlating the pres-suremeter modulus to the strain levels, it is possible to identify guidance for elastic modulus in geotechnical design and analysis. This is illustrated using actual pressuremeter tests completed in Bukit Timah Granite soils and other soils in Singapore. All the pressuremeter tests results presented in this paper are OYO type pressuremeters, where the cavity expansions are monitored directly using dis-placement transducers.


167

In this paper, the role of pre-bored pressuremeters in in-situ testing is first discussed. This will be fol-lowed by a review on the pressuremeter test results obtained from typical soil investigation reports, spe-cifically for those which measure the cavity expansion directly using displacement transducers (LVDTs). The second part of the paper proposes an alternative method to interpret the pressuremeter modulus, and the advantage of this method over conventional method is illustrated by studying pres-suremeter curves of recently completed tests in various geological conditions around Singapore. All the tests reported in this paper were conducted using the OYO Elastmeter-2 apparatus, which is a pre-bored type pressuremeter and which measures the radial expansion of the cavity directly using displacement transducers. The final part of the paper discusses the implications of pressuremeter modulus on geo-technical design. Based on data from 300+ pressuremeter tests, relationships between pressuremeter modulus and strains can be derived for various soils in Singapore.

Figure 1.Three main types of pressuremeter testing (after BRE, 2003).

2 PRE-BORED PRESSUREMETERS FOR IN-SITU TESTING

2.1 Characteristics of the pre-bored pressuremeter curve

It is usual to present the pressuremeter testing data in terms of an applied pressure against cavity strain curve. Figure 2 shows an idealized pressuremeter curve obtained using a pre-bored pressuremeter, from which important phases of pressuremeter testing can be identified. The initial part of the curve OAB usually shows a high increase in radial strain before the pressure starts to increase. This is a result of the relaxation of soil during the pre-boring process, where the borehole is left open before the pressure-meter device is installed. Once this has been overcome, the pressuremeter starts to load the soil along the primary compression curve (BCE). However, due to the significant disturbance introduced during the borehole installation stage, the soil response is usually not representative of its in-situ behaviour even after the initial inflation stage – this limitation can be mitigated by minimising soil disturbance in a self-boring pressuremeter. Thus, for pre-bored pressuremeters, estimates on earth pressures at-rest, undrained shear strength and modulus within the initial loading curve have to be treated with careful consideration.

Typically, it is common practice to carry out at least one unload-reload cycle (section CDC in Figure 2)

within the pressuremeter test from which the unloading stiffness may be obtained. This is increasingly

recommended for two reasons. Firstly, the unloading curves are less sensitive to imperfections in the

pressuremeter installation process, and hence the results can be representative of in-situ soil behaviour.

Secondly, the unloading curves could give important information about the elastic behaviour of soil,

which geotechnical analysis to predict ground deformations is dependent upon. Once the unload-reload

cycle has been overcome, the pressuremeter curve continues its path along the primary loading line until

the end of the test with a final unloading cycle (section CEF in Figure 2).


168

Figure 2. Typical pressure-strain curve for pre-bored pressuremeter.

2.2 Pressuremeter modulus

The slope of the pressuremeter curve gives an indication on the shear modulus of soil during the pres-suremeter test. For a pressuremeter where cavity expansion is measured directly using LVDTs, the pressuremeter modulus (Ep) can be calculated as follows:-

Ep = (1+)*p / (R/Ro) (1)

where is the Poisson’s ratio, p is the increase in the applied pressure, R is the increase in the cavity radius and Ro is the initial radius of the cavity. The radial strain R /R of the pressuremeter would be equal to half the shear strain in the cavity wall as explained in Mair and Wood (1987). Furthermore, it should be noted that the pressuremeter test measures shear modulus from the slope of the pressuremeter curve, and that the pressuremeter modulus is actually deduced depending on the drainage conditions to which that value of Ep would be used in design. A Poisson’s ratio of 0.3 is usually assumed when re-porting the pressuremeter modulus, and this would make it similar to an elastic modulus under drained condition. Typically, the pressuremeter modulus would be calculated by assuming a best-fit-line for the initial loading curve (line BC in Figure 2) and a best-fit-line for the unload-reload curve (line DC in Figure 2). These would be provided in the soil investigation reports and usually denoted as initial modulus and unloading modulus respectively. The pressuremeter modulus may be correlated to SPT-N values to derive relationships for design. SPT-N value is a useful parameter for correlations not only because it gives a rough indication on soil stiff-ness, but also because it is routinely tested for the entire length of all the boreholes in soil investigation works. Figure 3 plots the initial and unloading pressuremeter moduli against the nearest SPT-N value within the same borehole. These are obtained from a total of 115 pressuremeter tests that are conducted in the residual and completely weathered soils of Bukit Timah Granite (G-V and G-VI) in 3 different lo-cations in Singapore – Gambas Avenue, Ang Mo Kio Avenue 6 and Thomson Road. Specifically, all the pressuremeter tests reported in this paper were conducted using OYO Elastmeter-2.

As expected, there is a significant difference between the stiffness of the initial (or primary) loading

curve with the stiffness of the unload-reload curve. As seen in Figure 3, the initial modulus are generally

lower than the correlation Ep = 1.5N (in MPa) whereas the unloading modulus are much higher. This is

due partly to the primary loading behaviour of soils being less stiff than the unload-reload behaviour of

soils, but also influenced by the relaxation in the borehole before the pressuremeter test. Furthermore,

there is a significant scatter in the plot of unloading modulus versus SPT-N, where the unloading

modulus can range between 1.5N to 15N with a mean correlation of Ep=5.2N based on linear regres-

sion. With such a large scatter in the pressuremeter test results, it can be quite a challenge to identify a

suitable soil modulus for design and analysis.


169

Ep = 5.2N (MPa)

0.00E+00

2.00E+05

4.00E+05

6.00E+05

8.00E+05

0 20 40 60 80 100

SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)

Initial modulus (SI report)

Unloading modulus (SI report)

Linear (Unloading modulus (SI

report))

E = 1.5N (MPa)

E = 15N (MPa)

Figure 3. Pressuremeter modulus by assuming linear initial loading curve and linear unload-reload curve as re-ported in the soil investigation reports.

2.3 Detailed interpretation on the unload-reload portion of pressuremeter curve

Instead of assuming a linear function to work out a single unloading stiffness, an alternative method of

interpreting the unload-reload portion of the pressuremeter would be to examine the elastic secant

modulus in relation to the corresponding strains. Figure 4 shows the secant modulus of a particular

pressuremeter curve (Ep) plotted against its corresponding radial strains (R/Ro). There is a decrease in

the pressuremeter modulus, even within the range where a best-fit line is used to estimate the unload-

reload modulus in soil investigation reports. For example, as the radial strain is increased from 0.2% to

1% within the reloading curve, the elastic modulus dropped to about a quarter of its value from

8.5x104kPa to 2.3x10

4kPa. This rapid degradation of pressuremeter stiffness with radial strains reflects

the strain-dependent behaviour of soil within the elastic region.

Figure 4. Strain dependent behaviour of pressuremeter modulus

p’

R

0.E+00

1.E+05

2.E+05

3.E+05

4.E+05

5.E+05

0.010% 0.100% 1.000% 10.000%

Radial strain

Ela

stic m

odulu

s E

p (

kP

a)

R / Ro

Plot of elastic modulus vs radial strain

(from pressuremeter curve)


170

3 INTEPRETATION OF PRESSUREMETER MODULUS WITH STRAIN

3.1 Pressuremeter modulus for soils of Bukit Timah Granite Formation (G-V and G-VI)

With this new interpretation of pressuremeter modulus in mind, the reload pressuremeter modulus is plotted against the nearest SPT-N value, from the same 115 pressuremeter tests in the residual and completely weathered soils of Bukit Timah Granite, but this time with elastic modulus corresponding to three radial strains (0.1%, 0.5%, 1%) instead of just a single value reported in the soil investigations. Compared to the best-fit approach to obtain the unloading modulus (Figure 5a), there is a substantial reduction in the scatter of the results – particularly for pressuremeter modulus at 0.5% and 1% strains (Figures 5b, 5c, 5d). Relating the elastic modulus to the corresponding strains gives a rational and con-fident means to estimate the stiffness of soils for design purposes rather than estimating the unloading modulus from Figure 3 previously. It is observed that the pressuremeter modulus is highest when the radial strains are lowest. Using linear regression to fit in a correlation by least square method, the pressuremeter modulus corresponding to 0.1% strain is about 7.8 times the SPT-N value, and this decreases to 3 times the SPT-N value at 0.5% strain and 1.8 times the SPT-N value at 1% strain.

(d) Elastic modulus @ 1% strain

Ep = 1.8N (MPa)

0.E+00

2.E+05

4.E+05

6.E+05

8.E+05

0 20 40 60 80 100

SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)

(c) Elastic modulus @ 0.5% strain

Ep = 3.0N (MPa)

0.E+00

2.E+05

4.E+05

6.E+05

8.E+05

0 20 40 60 80 100

SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)

(a) Unloading modulus from SI report

Ep = 5.2N (MPa)

0.E+00

2.E+05

4.E+05

6.E+05

8.E+05

0 20 40 60 80 100

SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)

(b) Elastic modulus @ 0.1% strain

Ep = 7.8N (MPa)

0.E+00

2.E+05

4.E+05

6.E+05

8.E+05

0 20 40 60 80 100

SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)

Figure 5. Pressuremeter modulus of reload curves at various radial strains (Bukit Timah soils G-V & G-VI).

3.2 Pressuremeter modulus for soils of Jurong Formation (S-V and S-VI)

40 OYO-type pressuremeters were conducted in the residual and completely weathered soils of Jurong Formation (S-V and S-VI). These tests were done in Tuas West area of Singapore. In a similar manner, the elastic modulus of the reload pressuremeter curves are calculated at various cavity strains and these are plotted in Figure 6, together with the unloading modulus reported in SI reports using a best-fit line to the unload-reload curve. Similar observations may be made about the pressuremeter response of the Ju-rong Formation soils as with the Bukit Timah soils – namely that there is a significant reduction in scat-ter of the unload-reload modulus particularly at the 0.5% and 1% strain levels, and also that the elastic


171

modulus reduces with increasing strains measured by the pressuremeter. Using linear regression to fit in a correlation by least square method, the pressuremeter modulus corresponding to 0.1% strain is about 8.4 times the SPT-N value, and this decreases to 2.8 times the SPT-N value at 0.5% strain and 1.6 times the SPT-N value at 1% strain.


Ep = 8.4N (MPa)

0.E+00

2.E+05

4.E+05

6.E+05

8.E+05

1.E+06

0 20 40 60 80 100SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)


Ep = 2.8N (MPa)

0.E+00

1.E+05

2.E+05

3.E+05

4.E+05

5.E+05

0 20 40 60 80 100SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)


Ep = 1.6N (MPa)

0.E+00

1.E+05

2.E+05

3.E+05

4.E+05

5.E+05

0 20 40 60 80 100SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)


Ep = 8.6N (MPa)

0.E+00

2.E+05

4.E+05

6.E+05

8.E+05

1.E+06

0 20 40 60 80 100SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)

Figure 6. Reload pressuremeter modulus for Jurong Formation soils (S-V & S-VI).

3.3 Pressuremeter modulus for rocks of Jurong Formation (S-III and S-IV)

Pressuremeter tests (OYO type, 9 nos.) were also conducted in the moderately weathered and highly weathered rocks of the Jurong Formation at River Valley and Clarke Quay area. Figure 7 plots the elas-tic modulus with pressuremeter test depth for various corresponding strains, and compared to the unloading modulus reported in the SI report. There is a reduction in the scatter of the elastic modulus when the strain levels are considered during its evaluation from the pressuremeter curve. Notwithstand-ing the small sample size, the average elastic modulus (Ep) was found to reduce from 1.73x10

6MPa at

0.1% strain to 0.59x106MPa at 1% strain.

3.4 Pressuremeter modulus for Fort Canning Boulder Bed (FCBB)

OYO pressuremeter tests (24 nos.) were also conducted within the Fort Canning Boulder Bed in the vi-cinity of Bencoolen Street. The FCBB typically consists of quartzite boulders – UCS in the range of 100-200MPa – embedded in a hard sandy clayey silt or sandy silty clay matrix with undrained shear strength of at least 150kPa (Shirlaw et al, 2003). As most of the pressuremeter tests were conducted at depths between 20m to 50m, the FCBB was very hard and SPT refusal conditions were met even before the test depth (i.e. SPT-N > 100). Figure 8 plots the elastic modulus with pressuremeter test depth for various corresponding strains, and compared to the unloading modulus reported in the SI report. The average elastic modulus (Ep) was found to reduce from 1.03x10

6MPa at 0.1% strain to 4.45x10

5MPa at

0.5% strain and 3.27x105MPa at 1% strain.


172


0.E+00

2.E+06

4.E+06

6.E+06

8.E+06

0 10 20 30 40 50

Test depth of soil (m)

Ela

sti

c m

od

ulu

s (

kP

a)

Average Ep = 3.22x106

kPa


0.E+00

2.E+06

4.E+06

6.E+06

8.E+06

0 10 20 30 40 50


Ela

sti

c m

od

ulu

s (

kP

a)


kPa


0.E+00

1.E+06

2.E+06

3.E+06

4.E+06

0 10 20 30 40 50


Ela

sti

c m

od

ulu

s (

kP

a)


kPa


0.E+00

1.E+06

2.E+06

3.E+06

4.E+06

0 10 20 30 40 50


Ela

sti

c m

od

ulu

s (

kP

a)


kPa

Figure 7. Reload pressuremeter modulus for Jurong Formation rocks (S-III & S-IV).


0.E+00

2.E+06

4.E+06

6.E+06

8.E+06

0 10 20 30 40 50


Ela

sti

c m

od

ulu

s (

kP

a)


kPa


0.E+00

2.E+06

4.E+06

6.E+06

8.E+06

0 10 20 30 40 50


Ela

sti

c m

od

ulu

s (

kP

a)


kPa


0.E+00

1.E+06

2.E+06

3.E+06

4.E+06

0 10 20 30 40 50


Ela

sti

c m

od

ulu

s (

kP

a)


kPa


0.E+00

1.E+06

2.E+06

3.E+06

4.E+06

0 10 20 30 40 50


Ela

sti

c m

od

ulu

s (

kP

a)


kPa

Figure 8. Reload pressuremeter modulus for Fort Canning Boulder Bed (SPT > 100).


173

3.5 Pressuremeter modulus for Old Alluvium

Finally, 136 OYO-type pressuremeters were conducted in the various weathering grades of the Old Al-luvium in eastern part of Singapore, specifically at Jalan Besar/Kallang Bahru area, Tampines area and Upper Changi/Expo area. Figure 9 shows the elastic modulus of the reload pressuremeter curves calcu-lated at various cavity strains and plotted against the SPT-N values. Using linear regression to fit in a correlation by least square method, the pressuremeter modulus corresponding to 0.1% strain is about 9.6 times the SPT-N value, and this decreases to 3.7 times the SPT-N value at 0.5% strain and 2.5 times the SPT-N value at 1% strain. Compared to the unloading modulus reported in SI reports using a best-fit line to the unload-reload curve, the scatter in the elastic modulus is reduced when the modulus is correlated to the corresponding strains during testing, especially for those corresponding to 0.5% and 1% strains.


Ep = 2.5N (MPa)

0.E+00

2.E+05

4.E+05

6.E+05

8.E+05

1.E+06

0 20 40 60 80 100

SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)


Ep = 3.7N (MPa)

0.E+00

2.E+05

4.E+05

6.E+05

8.E+05

1.E+06

0 20 40 60 80 100

SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)


Ep = 11N (MPa)

0.0E+00

5.0E+05

1.0E+06

1.5E+06

2.0E+06

2.5E+06

3.0E+06

0 20 40 60 80 100

SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)


Ep = 9.6N (MPa)

0.0E+00

5.0E+05

1.0E+06

1.5E+06

2.0E+06

2.5E+06

3.0E+06

0 20 40 60 80 100

SPT-N values

Ela

sti

c m

od

ulu

s (

kP

a)

Figure 9. Reload pressuremeter modulus for Old Alluvium.

4 IMPLICATIONS ON GEOTECHNICAL DESIGN AND ANALYSIS

Conventional geotechnical analysis usually assumes a bilinear constitutive model for soil that is linear elastic and perfectly plastic, such as the popular “Mohr-Coulomb model” in Plaxis finite element soft-ware. Whilst it is a well-known fact that soil behaviour is non-linear even at small strains within the elastic region (Clayton, 2011), the immense experience developed using bilinear models in geotechnical analysis makes the linear elastic approach relevant as an approximation in the design process. Under serviceability conditions and for deformation prediction, an important decision to be made is the selec-tion of an appropriate modulus for the elastic model. The pressuremeter presents an opportunity to investigate the stress-strain response of the ground in-situ. Although the installation process may have disturbed the primary loading behaviour of the soil, the unload-reload response is less sensitive to such disturbances and the results would be representative of in-situ soil behaviour. Moreover, in a construction where ground is being removed or unloaded (such as excavations, tunnelling, retaining structures, etc.), the unload-reload behaviour of the ground may be more important than its loading behaviour. For example, Simpson et al (1979) observed that the stiff-ness modulus derived from unload-reload cycles in self-boring pressuremeter tests are generally in rea-sonable agreement with back-analysed modulus of field case studies of deep excavations in London Clay. There may be merits to describe the stiffness parameters of such geotechnical analysis using the


174

unload-reload curve rather than the primary loading curve of pressuremeter tests. Nevertheless, it is noted that the stress paths involved can be rather complicated, and could vary between stiff soils and soft soils and from one location to another. More back-analysis of field case studies would be needed to verify this observation. However, Mair (1993) highlighted that the ranges of soil strains encountered in typical geotechnical works (such as retaining walls, foundations and tunnels) occur over the range where there is the greatest variation of soil stiffness with strain. See Figure 10. For a linear elastic model, it would be quite errone-ous (and/or onerous) if the analysis is based on a stiffness parameter that is not corresponding to the an-ticipated strains in the geotechnical works. Thus, it would be useful to select a stiffness parameter corresponding to the anticipated soil strains for the geotechnical analysis, rather than the conventional approach of selecting a “one-size-fit-all” stiffness for all geotechnical analysis. For example, if the desired outcome of a geotechnical analysis is to limit the shear strains in the ground to less than 0.5%, then the elastic stiffness used for the geotechnical model should correspond to modulus at 0.5% shear strains. Or if the results of the analysis shows that the strain level in the soil had reached 1%, then the stiffness used for the soil model should have been capped at the stiffness corresponding to 1% strain.

Figure 10. Typical strain ranges for geotechnical works in relation to stiffness-strain behaviour of soil

(after Mair, 1993) Table 1 summarises the average pressuremeter moduli for different types of soils that are reported in the earlier sections of this paper. The “initial” modulus that is usually reported in the soil investigation re-ports is not meaningful for pre-bored pressuremeters due to significant disturbance and relaxation dur-ing the installation phase. The “unloading” modulus reported in the soil investigation reports would have been useful, except that there is no guidance on how these are measured other than a best fit line to the unload-reload curve and often results in a large scatter of elastic modulus for soils with similar proper-ties. Interpreting the elastic modulus with respect to the corresponding radial strains (say 0.1%, 0.5%, or 1%) provides a rational approach to estimate the elastic modulus of soil. This method also reduces the scatter in the pressuremeter modulus for soils of similar stiffness, and gives the designer greater con-fidence to use the correlations derived from pressuremeter tests.

One limitation of using pressuremeters to estimate the small strain stiffness of soil is the accuracy in

measuring radial displacements. For an OYO-type pressuremeter, the accuracy of displacement detec-

tion using the LVDT is 0.001cm. For an initial radius of say 3.8cm, the error in estimating the strains

would be in the range of 0.05% strain. As a result, there is a high scatter in the pressuremeter modulus

for the low strain (0.1%) cases. However at larger strains (say 0.5% and 1%), the scatter in the pres-

suremeter modulus is less attributed to the error in measuring the radial strain, and would be more at-

tributed to non-homogeneous sampling and on the use of SPT-N correlations as an indirect measure of

soil stiffness. Consequently, the error in estimating modulus would be the lowest at larger strains.


175

Table 1. Pressuremeter moduli for different soils.

Value reported in SI report Elastic modulus interpreted from reload

portion of pressuremeter curve

Initial modulus

Unloading modulus

Reload modulus at 1% radial

strain

Reload modulus at 0.5% radial

strain

Reload modulus at 0.1% radial

strain

Pressuremeter moduli correlated to SPT-N (in MPa)

Residual soils of Bukit Timah granite (GV, GVI)

0.9N 5.2N 1.8N 3.0N 7.8N

Old Alluvium (OA) – various weathering grades

1.6N 11N 2.5N 3.7N 9.6N

Jurong Formation soils (SV, SVI)

1.3N 8.6N 1.6N 2.8N 8.4N

Average pressuremeter moduli (in kPa)

Jurong Formation rocks (SIII, SIV)

4.82x105 32.2x10

5 5.93 x10

5 10.4 x10

5 17.3 x10

5

Fort Canning Boulder Bed (SPT>100)

2.34 x105 20.1x10

5 3.27 x10

5 4.45 x10

5 10.3 x10

5

5 CONCLUSION

In this paper, the use of pressuremeter testing to estimate soil modulus was discussed. Using data from actual pressuremeter tests conducted in Singapore, it was observed that the conventional approach of calculating the elastic modulus from a best-fit line on the unload-reload curve results in a wide scatter of pressuremeter modulus. This makes it difficult to estimate the design elastic modulus reliably. An alternative approach that relates the pressuremeter modulus to its corresponding radial strains (which is equal to half the shear strains at cavity wall) was investigated. This method addresses the variation of soil stiffness with strain and results in a reduced scatter in the pressuremeter modulus. Se-lection of the elastic modulus for design and analysis should depend on the strains induced in the soils. Using this new approach, the results from more than 300 pressuremeter tests (OYO types) were studied. This study provided some reference on pressuremeter modulus for various types of soils in Singapore, and presents a new perspective on the selection of soil stiffness for geotechnical design and analysis.

REFERENCES

ASTM D4719. 2000. Standard Test Method for Prebored Pressuremeter in Soils, American Society for Testing and Materials.

Briaud, J.L. 1992. The Pressuremeter. Rotterdam: Balkema. BRE. 2003. A simple guide to in-situ ground testing Part 5: Pressuremeter testing. Building Research Estab-

lishment BS 5930. 1999. Code of Practice for Site Investigations, British Standards Institution. Clayton, C.R.I. 2011. Stiffness at small strain: research and practice, Geotechnique, Vol. 61(1), pp.5–37. Mair, R. J. 1993. Developments in geotechnical engineering research: applications to tunnels and deep excava-

tions. Unwin Memorial Lecture 1992. Proc. Instn Civ. Engrs, Civ. Engng 3, No. 1, pp.27–41. Mair, R.J. and Wood, D.M. 1987. Pressuremeter testing: methods and interpretation, CIRIA Ground Engineer-

ing report on in-situ testing, London: CIRIA. Shirlaw, J.N., Broome, P.B., Chandrasegaran, S., Daley, J., Orihara, K. Raju, G.V.R., Tang, S.K., Wong, I.H.,

Wong, K.S. and Kyi Yu. 2003. The Fort Canning Boulder Bed. Proc. conf. Underground Singapore 2003, pp.388–407.


176

1 INTRODUCTION

Contract C913, Downtown Line 2 involves two cut and cover stations (Hillview Station and Cashew Station) and bored tunnels. The North Launch Shaft (NLS) is adjacent to the proposed Cashew

Station, as presented in Figure 1, a 25m by 25m in plan extending to a depth of 20mbgl. The earth

retaining support system comprises 1.0m thick reinforced concrete diaphragm walls with 4 layers of diagonal struts. The reduced level at the top of the wall is about RL114.0 m and the toe of the wall

ranges from RL81.0m to RL92.3m. Upper Bukit Timah bounds the site on the west and there is a 8m

wide canal on the east side. The piled ESPA condominium is 140m to the south east side of the shaft.

There are no other important structures within 100m.

ESPA Condominium

Upper Bukit Timah Road

Cashe

w R

oad

Cashew

Station

North

Launch

Shaft

Canal

Canal

NESPA Condominium

Upper Bukit Timah Road

Cashe

w R

oad

Cashew

Station

North

Launch

Shaft

Canal

Canal

NN

Figure 1: North Launch Shaft plan layout and photograph of excavation to final excavation level.

Understanding the engineering behavior of Bukit Timah

Granite during deep excavation and the benefits of early

design review

N. H. Osborne, A. Yang, D. Macphie Mott Macdonald, Singapore

S. H. Ra

GS Engineering and Construction, Korea

K.M. Soh Land Transport Authority, Singapore

ABSTRACT: Geotechnical parameters are derived from traditional site investigation and laboratory

testing, combined with previous experience; however once excavation commences in such variable ground geotechnical prediction rarely matches the outcome observed. The reasons for this frequently

lie in the choice of the parameters together with the variability of the ground. It is therefore essential

that early within any construction project the performance of the design is assessed, any significant

deviation highlighted, the mechanisms for this difference fully investigated and understood. This then insurers safe construction or enables potential refinement of the design, providing early feedback to

make adjustment and allows the project to move forward in a controlled manner and without undue

complication. This process is carried out for a deep excavation in the Bukit Timah Granite. The observations from the

instrumentation are fedback into the design model with various sensitivity options run to refine the

design to more closely reflect the actual performance on site. This allows the actual behavior of the

Granite to be understood far more clearly than at the onset of the project.


177

2 GROUND CONDITIONS AND EARTH RETAINNING SYSTEM

The general site stratigraphy comprises sillty sand Fill overlying occasionally present and thin

interdigitating F1 (sand)/F2 (clay) of the Kallang Formation on variably weathered Bukit Timah Granite Formation of up to 9m thick residual silty clay to slightly sandy silt (GVI); 7m thick

completely weathered sandy silt (GV), 4m thick highly weathered rock (GIV) underlain by moderately

weathered rock (GIII) or better grade. Moderately conservative ground parameters were prescribed

within the contract document, they are summarised in Table 1 and Figure 2.

Table 1: Base line ERSS geotechnical design parameters

Strata

Unit weight Permeability

Drained

Young's

Modulus

Poisson's

ratio

Effective

cohesion

Friction

angle

Lateral earth

pressure

coefficient

γunsat γsat kx, ky, E’ref ν c' φ K0

kN/m3

kN/m3 m/s

MN/m2

kN/m2 º

Fill 16 18.5 1.E-06 6.96 0.3 0 30 0.5

F1 17.2 19.2 1E-06 10.44 0.3 0 30 0.7

G - VI 16 18.5 3.E-06 3.48 0.3 3 30 0.8

G - V 17 19 3.E-06 34.8 0.3 3 32 0.8

G - IV 22 23 2.E-06 300 0.25 50 40 0.8

G - III 24 24 2.E-06 3000 0.25 300 45 0.8

G -III (10m

deeper) 24 24 2.E-07 3000 0.25 300 45 0.8

80

85

90

95

100

105

110

115

0 25 50 75 100

SPT-N

RL (

m)

1.5 2.0 2.5

bulk unit weight (kN/m3)

0 25 50 75 100

Atterberg Limits (%)

Moisture Content

Plastic limit

Liquid Limit

111.5

S1 - 113.5

GL

86.5

107.0

S2 - 108.0

103.0

S3 - 104.0

94.0

S4 - 99.0

FEL

0 25 50 75 100

Fraction (%)

clay

clay+silt

clay+silt+sandfissure

grouting

98.0

diaphragm

wall

111.0

98.0

FILL

107.0

102.0

GV (a)

GV (b)

95.0

GV (c)

GIV

GIII

91.5

87.5

GVI (a)

GVI (b)

114.0

Figure 2: ERSS system and stratigraphy

The Earth Retaining Support System (ERSS) comprised 1 metre thick diaphragm walls with 4 layers of diagonal double steel I-beam struts. To ensure toe stability, panel bites of typical width of 2.85m or


178

5.7m were installed a minimum of 1 metre into GIII bedrock. As a groundwater control measure, fissure grouting was undertaken for either 10 metres below the GIII rockhead or 10 metres below

formation level, whichever was shallowest. The construction sequence comprised conventional cut

and cover technique as follows

Install Diaphragm Walls

Undertake fissure grouting

Excavate to 1st level & install strut 1, preload 100kN/m

Excavate to 2nd

level & install strut 2, preload 150kN/m

Excavate to 3rd

level, &install 3rd

strut, preload 200kN/m

Excavate to 4th level, & install 4

th strut, preload 200kN/m

Excavate to final excavation level

The structural properties for 2D and 3D analysis are presented for the struts and the wall in Table 2 and 3 respectively.

Table 2: Strut properties (* for 2D analysis) for 2 & 3D analysis

Table 3: Wall properties (*for 2D analysis) for 2 & 3D analysis

*EA

(kN/m)

*EI

(kNm2/m)

γt

(kN/m3)

E1

(kN/m2)

E2

(kN/m2)

G12= G13

(kN/m2)

G23

(106kN/m2) V12

Schematic

of plate plane and

axis

directions

2.4×107 1.63×106 24.0 19.6×106 0.98×106 8.167×106 0.208×106 0.2

3 PERFORMANCE OF ERSS

The performance of the shaft was monitored using wall inclinometers installed at the centre of each

wall; strain gauges on struts; and vibrating wire piezometers and water standpipes as presented on the instrumentation plan (Figure 3).

3.1 Pore Water Pressures Performance

Two to three level piezometers were installed in GVI, GV, GIV, GII strata outside the shaft. Two piezometers were installed inside the shaft within the fractured GIV/GIII rock. Initial groundwater

levels at different strata are comparable suggesting constant pressure head distribution with phreatic

level of 112.5m (i.e.1.5mbgl). During excavation a passive sump system was utilised when necessary to maintain groundwater level at excavation level. Figures 4 & 5 compare pressure readings from

Strut Type *EAθ=45

(kN)

*Spacing

(m)

γt

(kN/m3)

A

(cm2)

I2=I3

(10-3

m4)

I23

(10-3

m4)

S1 2/HR 400 × 400 × 172

kg/m 4.05×106 3.4 78.5 395 1.18 0

S2 2/HY 650 × 300 ×

158 kg/m 4.12×106 3.4 78.5 402 2.83 0

S3 2/HY 700 × 350 ×

206 kg/m 5.36×106 3.4 78.5 523 4.35 0

S4 2/HY 700 × 350 ×

206 kg/m 5.36×106 3.4 78.5 523 4.35 0


179

piezometers within (GWV2203 & GWV2204) and outside (GWV2205 & GWV2206) the shaft over time.

Recorded groundwater pressures remained relatively constant (1.5m bgl) during excavation to the 1st

and 2nd

struts generating excess hydraulic head of 5m relative to excavation level. During the 3rd

excavation stage an abrupt drop in pressures of 5m occurred within the shaft with 3m excess head

remaining relative to excavation level; whilst outside the shaft the drop was 3m.. This sudden response

can be accounted for by removal of a low permeability capping layer of upper GVI or F2 triggering

seepage flow (Figure 6). Also to a lesser degree decreasing overburden effective stress around preferential flow paths such as around king posts as indicated by observed high seepage flow, between

the walls and the soil, and along relict subvertical fissures may have contributed to this observation.

During the 4th excavation stage, groundwater pressures dropped by 6m to become hydrostatic within

the shaft and continued to remain hydrostatic with excavation to FEL. Outside the excavation

groundwater pressures showed significant drops from resulting in a 10m reduction by the final

excavation level. Such notable pore water pressure reduction infers fissure grouting was relatively

ineffective, or targeted at the wrong strata., not the most permeable.

Figure 7 shows the groundwater pressure drop with distance from the shaft, with respect to depth

below ground. The groundwater influence extent was about 200m away from the excavation with higher pressure reductions occurring close to the excavation (Figure 4). The discontinuity network

within the bedrock is envisaged to play an important role in pressure loss propagation over distance.

The low storage capacity of the rock means dewatering can occur relatively quickly, whilst the rock permeability can also vary over the course of the excavation due to stress release, flushing fines from

the apertures, changing degree of saturation, pore water pressure and indeed the permeability.

Generally the pore water pressure distribution with depth remained hydrostatic with slight

underdrainage approaching rockhead. Such behaviour is consistent with laboratory data indicating higher sand and silt content and lower clay content at depth.

GWV2205

GWV2204

GWV2203

GWV2202

GWV2206

GWV2201

I2201

I2202

I2202

I2204

GWS2201

GWS2203

GWS2204

GWS2202

30m

GWV2205

GWV2204

GWV2203

GWV2202

GWV2206

GWV2201

I2201

I2202

I2202

I2204

GWS2201

GWS2203

GWS2204

GWS2202

30m

75

80

85

90

95

100

105

110

115

120

-350 -300 -250 -200 -150 -100 -50 0 50 100 150 200 250 300

Pore Water Pressure (kPa)

Ele

vati

on

(m

RL

)

1st strut

2nd strut

3rd strut

4th strut

FEL

strut levels

hydrostatic

ground leve;

Diaphragm wall

hydo gwl exc to 3rd

hydo gwl exc to 4th

hydro gwl exc to fel

Series4

Initial groundwater

levelhydro gwl exc to

4th struthydro gwl exc to

3rd struthydro gwl exc to

4th struthydro gwl exc to fel

Hydro gwl exc to

2nd pas

Figure 3- Instrumentation plan Figure 4- Groundwater Pore Pressures for different

excavation stages


180

90

95

100

105

110

115

18/11/2010 08/12/2010 28/12/2010 17/01/2011 06/02/2011 26/02/2011 18/03/2011 07/04/2011 27/04/2011

Date

Pie

zom

etr

ic h

ea

d (

RL

m)

excavation periods

excavation level

GWV-2203

GWV-2204

GWV-2205-1

GWV-2205-2

GWV-2206-1

GWV-2206-2

GWV-2206-3

Exc to S1 Exc to S2 Exc to S3 Exc to S4 Exc to FEL Remove S4

Figure 5 Piezometric head change with time.

Figure 6 Mechanism to create seepage flow.

85

90

95

100

105

110

115

0 50 100 150 200 250

Distance from d-wall (m)

Elev

ation

(RLm

)

>100

90-100

<90

* tip level of

piezometer

Figure 7: Groundwater drawdown curve.

(1) excavation

(2) -ve

pore water

pressure

develop (3) immediate

water in flow

Permeable

Impermeable

(4) Upward

flow impeded

by

impermeable

layer

(1) excavation

(2)

seepage

Permeable

(3) pumping

(4)

reduced

pwp

beyond

excavation


181

3.2 Wall Deflections

Wall deflections were monitored against the original ERSS design predictions. Figure 8 shows actual

wall deflections as recorded by wall inclinometers for selected construction stages and the original design predictions which appear conservative. During the first excavation stage the wall deformed in

cantilever. With successive excavation and strut installation, maximum movement tended to increase

and shift downwards to form a bell shape curve. Higher deflections, particularly during stage 1

excavation, were recorded at the north wall (I2001) and largely attributed to shaft excavation commencing at this zone (Figure 1). The lowest deflection was indicated at the east wall (I2004),

likely owing to nearby canal diversion excavation, as supported by an adjacent soil inclinometer

recording 20mm of movement away from the excavation towards the canal.

Figure 8: Comparison of development of wall deflections and ERSS Prediction

3.2 Strut loads

Consistent with wall deflections, strut loads were significantly lower than the original ERSS design predictions but indicated similar trends (Figure 9). After reaching a peak following excavation to the

1st stage, loads for the 1

st strut level decreased for the remainder of excavation as deeper strut levels

were added. Strut loads for the 2nd

level remained relatively constant after excavation to 3rd

level, whilst loads recorded at 3

rd and 4

th strut levels continued increasing to final excavation level.

75

80

85

90

95

100

105

110

115

120

0 10 20 30 40 50 60 70 80


75

80

85

90

95

100

105

110

115

120

0 10 20 30 40 50 60 70 80


Red

uced

Level (R

Lm

)

75

80

85

90

95

100

105

110

115

120

0 10 20 30 40 50 60 70 80


75

80

85

90

95

100

105

110

115

120

0 10 20 30 40 50 60 70 80


75

80

85

90

95

100

105

110

115

120

0 10 20 30 40 50 60 70 80


Predicted

Actual

EXC to S1 EXC to S2 EXC to S3 EXC to S4 EXC to FEL

75

80

85

90

95

100

105

110

115

120

0 10 20 30 40 50 60 70 80


I2001-North I2002-West I2003-South I2004-East

North (design) West (design) South (design) East (design)

75

80

85

90

95

100

105

110

115

120

0 10 20 30 40 50 60 70 80


75

80

85

90

95

100

105

110

115

120

0 10 20 30 40 50 60 70 80


Red

uced

Level (R

Lm

)

75

80

85

90

95

100

105

110

115

120

0 10 20 30 40 50 60 70 80


75

80

85

90

95

100

105

110

115

120

0 10 20 30 40 50 60 70 80


75

80

85

90

95

100

105

110

115

120

0 10 20 30 40 50 60 70 80


Predicted

Actual

EXC to S1 EXC to S2 EXC to S3 EXC to S4 EXC to FEL

75

80

85

90

95

100

105

110

115

120

0 10 20 30 40 50 60 70 80


I2001-North I2002-West I2003-South I2004-East

North (design) West (design) South (design) East (design)


182

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

08/12/2010 28/12/2010 17/01/2011 06/02/2011 26/02/2011 18/03/2011 07/04/2011

Date

Lo

ad

pe

r d

ou

ble

str

ut

(kN

)SG 2201 -1

SG 2202-1

SG 2203-1

SG 2204-1

SG 2201 -2

SG 2202-2

SG 2203-2

SG 2204-2

SG 2201 -3

SG 2202 -3

SG 2203 -3

SG 2204 -3

SG 2201 -4

SG 2202 -4

SG 2203 -4

SG2204-4

1st level strut

2nd level strut

3rd level strut

4th level strut

excavation periods

Figure 9: Development of load as inferred by strain gauges against predictions with time.

4.ORIGINAL ERSS DESIGN AND DESIGN REASSESSMENT

4.1 Base line ERSS design

A PLAXIS 2D plane strain model showing the ERSS design stratigraphy at the north wall is presented in Figure 10, utilizing design parameters as outlined in table 1. The wall is modelled as an elastic plate

(beam element), the struts as single node anchors and the soil mesh comprises 15 node triangular

elements. Other design aspects include a 20kPa ground surface surcharge extending 20m from the

excavation and groundwater table at the ground surface level. The constitutive soil model was linear elastic perfectly plastic Mohr Coloumb model with either all materials modelled as drained or

undrained (except made ground as drained), with the former proving more onerous and governing the

design. Drained analysis was run with groundwater steady state seepage analysis.

4.2 Design Reassessment

Recorded wall deflections and strut forces, as discussed earlier, confirm the ERSS baseline design to

be over conservative. A variety of variables including stratigraphy, ground parameters, groundwater

pore pressures, surcharge and strut preloading will act in combination to affect the shaft performance.

Figure 10: Plaxis model (excavation to FEL) baseline ERSS design (north wall

stratigraphy)

F1

GVI

GV

GIV

GIII (upper)

GIII (lower)

Fill

76.5

101

112.5

109.5

88.5

86.5

115


183

Each variable has been re-evaluated with the sensitivity assessed using PLAXIS whereby wall deflections at final excavation level are compared relative to the original design at the north wall. A

discussion for each re-evaluated variables is provided below, with results summarised in Figure 12 and

13. Figure 11 presents the wall deflection profile for each sensitivity study and that for all combined re-evaluated variables compared with inclinometer profiles. Figure 12 presents a bar chart summary of

the percentage change of maximum wall deflections for assessed variables.

4.3 Stratigraphy

Site stratigraphy is particularly complex laterally. Diaphragm wall panels were contractually required

to socket 1m minimum into GIII rockhead level to ensure a fixed hinge point with no movement at the toe, as confirmed by modelling predictions and thus represent the most critical strata level. The ERSS

baseline design involved adopting stratigraphy based on 4 boreholes. Subsequent additional probe

holes (at 12m spacing) along the proposed wall for assisting panel installation were undertaken. Utilising this information and that obtained during diaphragm wall construction, a better defined

rockhead level was subsequently ascertained, with the GIII rockhead profile and GIV thickness

confirmed as highly variable and generally higher than assumed in the design (Figure 11).

The average GIII rockhead level was re-evaluated as RL87.5m, approximately 1m higher than the

design for the north wall. The overlying transitional layer, GIV, was considered to be of average

elevation RL87.5m, with an average thickness of 4m compared to the north wall design of 2m. The re-evaluated strata levels are relatively similar to those of the north wall explaining the relatively small

reduction of maximum predicted deflections (25%). In contrast larger differences of predicted wall

deflection are apparent for different sides of the excavation at FEL within the original ERSS baseline design (Figure 11) whereby deeper GIII rockhead levels (east->north, ->south and west) are associated

with larger deflections (Figure 8). Applying this principle to individual panels, significant variations of

deflection may be expected along a wall with varying GIII rockhead and hence panel depth. However

differences are anticipated to be counteracted by shearing resistance acting between the panels; load transfer through arching of the retained soil and the capping beam on top of the panels, thus helping to

equilibrate movement of one panel relative to the adjacent panel. Therefore predicting movement and

designing reinforcement for the deeper panel could result in a slight overestimate and underestimate for the shallower panels, thus indicating a moderately conservative estimate of rockhead elevation

along a section width to be suitable for deflection predictions opposed to selecting the lowest rockhead

value, although subject to confidence of the derived rockhead profile.

Figure 11: diaphragm toe elevation levels; GIII and GIV rockhead elevations


184

The strata levels overlying the rock have been re -evaluated primarily based on SPT-N values as presented in Figure 2. It should also be noted granular (F1) and cohesive (F2) Kallang Formation have

been incorporated into Fill material and GVI material respectively owing to their discontinuous nature

and similar properties.

4.4 Ground parameters

Given the performance of the ERSS it was clear that the differences in stratigraphy were not the only contributory factor in the under prediction of the wall deflections. The additional site investigation and

laboratory testing was revisited and re-evaluated to derive more probably soil and rock parameters,

table 4. As much of the excavation occurred within the GVI and GV this included attempts to further subdivide these strata to more accurately predict their behaviour.

The degree of weathering tends to decrease with depth and heavily influences the material type and engineering behaviour of the Bukit Timah Formation. The grain size distribution data for GVI and GV

soil, as presented in Figure 2, indicates decreasing clay and increasing sand content with depth

consistent with results of Sharma et al, 1985. Moisture content, liquid limit and plasticity index tend

to decrease with depth, further supporting an increasingly drained response with depth whilst limited portions of the upper profile behave in a more undrained fashion, as corroborated by the pore water

pressure response discussed earlier. The upper GVI strata (SPT-N <8) is therefore modelled as

undrained. All other strata have been modelled as drained.

Drained strength parameters (c’ and phi) have been re-evaluated based on laboratory strength testing

and a new set proposed with higher strengths are displayed in Table 4. Derived effective cohesion of GVI and GV appears to increase with depth from 5 to 20kN/m

2, likely owing to increasing

cementation as the material becomes less weathered and larger proportions of the original rock

structure are preserved, but counterintuitive in the respect of decreasing clay component encountered

at depth. Poh et al (1985) previously cited a direct linear relationship between increasing effective cohesion and increasing clay content for fine grained weathered soil of the Bukit Timah Granite

Formation (i.e. 1kN/m2 increase per unit increase in clay content, with zero cohesion for no clay

present); whilst also acknowledging it is common for samples with low clay content to have significantly higher effective cohesion values due to iron oxide cementation. Such data also indicate

the original effective cohesion values quoted by Poh et al (1985) are higher than the baseline ERSS

values of 3kN/m2.

Table 4: Derived Geotechnical Parameters

Strata SPT-N γsat

(kN/m3)

Mc

(%)

Ll

(%)

Lp

(%)

, c’

(kN/m2)

φ

(°)

Su

(kN/m2)

FILL 4-10 19.0 29 56 27 2 29 GVI (a) <8 17.5 50 64 34 40

GVI (b) 8-20 18.5 40 58 31 5 32

GV (a) 20-30 18.5 35 51 29 5 33

GV (b) 30-50 19.0 26 51 30 6 34

GV (c) 50-100 19.0 23 43 27 10 35

GV-IV >100 19.0 15 - - 20 35

GIV - 23.0 - - - 50 40

GIII - 24.0 - - - 300 45


185

The effective friction angle, φ’, re-evaluated on lab strength data exhibit an increase of 32 to 35 degrees from GVI to the base of more granular

GV soil. Such values are higher than ERSS

baseline values of 30° for GVI and 32° for GV, whilst enabling a less abrupt strength transition

into the GIV strata.

The soil stiffness correlation used for the ERSS base line design (Young’s modulus, EU =2 × SPT-

N) is equivalent to reduced stiffness at large

strains. Because of the linear elastic component of the adopted Mohr Colomb model high small

strain stiffnesses are unaccounted for (Atkinson,

2000). CIRIA 143, (1993) suggests a relationship of E =6.3 to 10.4 × SPT-N at small strains. As the

deflections are limited, only small strain stiffness

are expected to be mobilised particularly further

from to the excavation. A global correlation of Eu =4 × SPT-N has been assumed.

The effect of higher strength and stiffness profile reduces maximum predicted deflections by about

50% from the ERSS design, more accurately

mirroring those observed.

4.5 Groundwater Pore Pressures

As presented in Figure 3, recorded groundwater pressures are lower than those generated in the

design where phreatic level was modelled at

ground surface level and steady state seepage condition simulated. Through modelling the

recorded groundwater pressures with successively

lower phreatic levels and hydrostatic distribution with depth, there is a significant reduction of

maximum deflections by 45%.

This indicates the potential benefits of pore pressure reduction for ERSS performance, if

settlement effects on utilities and buildings can be

managed accordingly.

Figure 12: Wall deflections at final excavation level

for different sensitivities compared with inclinometer

readings

-20-1001020304050607080

stratigraphy

ground parameters + stratigraphy

surcharge

groundwater pore pressure

preload

combined variable change

Figure 13: % change of maximum deflections relative to ERSS baseline design model.

FEL


186

4.6 Surcharge

At the north wall a surcharge of 20kPa was assumed 20m from the shaft. However 5kPa is understood

to be a more realistic value, given the absence of structures or even shallow foundations, producing a 5% reduction of predicted maximum wall deflection; however this is noticeably higher for deflections

around to the top of the wall.

4.7 Strut Preload

High strut preloads were assumed in numerical analyses to reduce wall deflections during the design

process (Table 2). However it was apparent during the groundwater pore pressure sensitivity study that unrealistic wall deflection into the retained soil at ground surface level is predicted largely due to

higher preloads in Plaxis analysis than were maintained in reality. CIRIA 157 suggests preloading

carries little practical significance, whereby a significant proportion of the preload is lost due to temperature effects. About 50% reduction of the initial preloading over a few days is discernible from

the strain gauges installed on the NLS struts (Figure 8) and therefore adopted for re-evaluated preloads

applied in the sensitivity study. Converse to all other parameters considered, re-evaluated preloading

values lead to an 8% increase of maximum predicted wall deflections.

4.8 Combined Variable Changes – 2D analysis

All re-evaluated variables as discussed above (Table 4) and structural properties within Table 2 and 3

were applied in a 2D PLAXIS analysis. The combined re-evaluated variables predict wall deflections

comparable to wall inclinometer records at the final excavation level as presented in Figure 12.

4.9 Combined Variable Changes – 3D analyses

To assess lateral arching of the retained and underlying soil, 3D PLAXIS analyses were undertaken

9Zdravkovic et al), (Figure 14). Similar geometry parameters were adopted as the re-evaluated 2D

model, with the exception of the wall properties (Table 3).

The predicted wall deflections along the diaphragm wall are presented in Figures 15 & 16. The

maximum wall deflection at the centre panel is similar to 2D analysis which suggests the 3D effect is

insignificant near the centre. This may be due to the fact that the base rock strata and d-wall are already very stiff, consequently limiting the benefits of the 3D analysis and the benefit of arching

effect from soil (GV & GVI) is not obvious. The 3D effect is significantly more apparent within 6m

from corner, where less than 70% of deflection is predicted to occur.

Figure 14: 3D model of North Launch Shaft


187

80

85

90

95

100

105

110

115

0 10 20 30 40 50

Deflection (mm)

Depth

(m

)

Exc to S1Exc to S2Exc to S3Exc to S4Exc to FEL

0

2

4

6

8

10

12

01020304050

Deflection (mm)

Dis

tance f

rom

cente

r (m

)

Figure 15: Wall deflections vs depth Figure 16: Wall deflections along the

D-wall.

5 CONCLUSION

It is well known that the Bukit Timah Granite is a very variable strata making any design challenging

and therefore important that the design must err towards the robust side. By reviewing the early

performance of the design a better understanding of the ground and its’ behaviour can be gained. Five major factors within the design were adjusted and their influence on the design quantified. It is evident

that the variable stratigraphy, the ground parameters and groundwater pressures all have a significant

impact on the design output and eventual ERSS system used. As a further check 3D analysis was undertaken to understand what potential benefits it would bring, in this case this was limited. It is

however clear that once better understanding of the actual site geology can be verified there is

potential scope to review future design and amend design approaches to reduce construction time and

costs.

References

CIRIA 157 Temporary Propping for Deep Excavations, 1999. Page 23

CIRIA 143 SPT methods and Use, 1993, page 81

Atkinson, J.H (2000) Non linear Soil Stiffness in Routine Design. Geotechnique. 50, No. 5, p 487 to 508.

Poh, K.B et al (1985), Residual Granite Soil of Singapore, Eighth Southeast Asian Geotechnical Conference.

Sharma. J.S et al (1985),Geological and geotechnical features of Singapore: an overview.Tunnelling and

Underground Space Technology. Volume 14, No. 4, 1999.

Zdravkovic, L, Potts D.M & St John, H.D (2005). Modelling of a 3D excavation in finite element analysis.

Geotechnique 55, No. 7, p 497-513


188

1. INTRODUCTION The scope of Contract 854 (C854) of the Circle Line Project consisted of the construction of 6.2km long twin bored tunnels, plus three underground stations namely Caldecott, Botanic Gardens and Farrer Road Stations and also Bukit Brown RTS facility building (Fig. 1). The underground geological profile and conditions made the tunneling on this contract profoundly challenging for the Tunnel Boring Ma-chines (TBMs) and the Engineers carrying out the tunneling works. Extensive ground investigation had been carried out along the tunnel alignment prior to the commencement of tunnel and station construc-tion. A total of 320 site investigation (SI) boreholes were carried out, 155 pre tender boreholes by the Client, plus an additional 66 numbers of boreholes for stations, and 99 additional boreholes for the tunneling, were carried out by the Contractor. The extensive SI was deemed adequate to confirm the subsurface geological profiles and the nature and characteristics of the soils and rocks to be encountered along the C854 alignment.

Figure 1: Tunnel Alignment of C854 showing probable boundary of Central Singapore Granite and localized Dyke Rock intrusions

Past experience on subsurface ground condition of Bukit Timah Granite formation in Circle Line Contract 854 S. A. McChesney Land Transport Authority, Singapore

M. Maw SsangYong E & C Co. Ltd, Singapore

ABSTRACT: The Bukit Timah Granite Formation was the predominant Geological Formation in the construction of the tunnels and stations on Contract 854 of the Circle Line Project. This paper presents the different characteristics of the subsurface geological profile of the Bukit Timah Granite encountered along the 6.2km length of the C854 tunnel alignment, including localized igneous intrusive dyke (dike) rocks and the variable thickness of residual soils with different weathering grades. The Bukit Timah Granite in C854 can be categorised into two distinct types, firstly hornblende-biotite GRANITE (or hornblende GRANITE), and secondly, feldspar rich GRANITE, which is similar to Central Singapore Granite. A detailed borehole investigation provided the predicted underground condition generally before construction, however the actual ground encountered provides updated information that is a valuable record of the geology along the alignment of C854. The difficulties posed on the TBM tunneling from the different rock types and weathering grades are also valuable lessons learnt that can be used by fu-ture tunnel projects in Singapore.

Predominant GRANITE (Similarity of

Central Singapore Granite)

Caldecott

Station

Marymount

Station

Bukit Brown

RTS

Predominant Hornblende GRANITE

associated with Dyke Intrusion

Botanic Gardens

Station

Farrer

Road Station

North

PIE

Towards

Changi


189

2. GEOLOGY OF SINGAPORE

There are four prominent geological formations in Singapore. Firstly, the recently deposited Kallang Formation which can be found across the entire Singapore island but predominantly in the South-Eastern region. The Old Alluvium which is found mostly in the Eastern area with some localised depo-sits in the North-West of the island. The Jurong Formation appears across the Western side of the island extending to the South. The Bukit Timah Granite Formation covers one quarter of Singapore from the North to the central area of the Island. The Bukit Timah Granite mainly consists of acid igneous rock formed during the early to mid-Triassic Period, 225 million years ago in the geological time scale. A number of hybrid rocks are associated in this formation.

Figure 2: Geology of Singapore

3. SUBSURFACE GEOLOGY OF CONTRACT 854

The general ground conditions along the majority of the C854 tunnel alignment comprised of a relatively small layer of Fill underlain by the Bukit Timah Granite Formation Residual soils. However at certain locations, Kallang Formation soils, consisting of organic soft clays and sand layers were present above the Bukit Timah Granite Formation. Approximately 70% of the tunnel alignment lay within the Bukit Timah Granite Formation residual soils (GVI) and completely weathered rock (GV). The characteristic of SPT-N value for the completely weathered rock and residual soil is generally increasing with depth due to the decrease in weathering. The remaining 30% of the tunnel alignment encountered Bukit Timah Granite rock with weathering grades ranging from highly to moderately weathered rock through to fresh rock as designated GIV, GIII, GII and GI respectively. The Bukit Timah Granite rock can have uncon-fined compressive strengths ranging from 10MN/m

2 to 280MN/m

2. Residual soil and rock is mostly

found to be in irregular contact with either incline or horizontal position. The C854 tunnel alignment encountered sections of full face rock and also full face residual soils, how-ever it also encountered significant section of mixed face soil / rock conditions during the tunnel drives. Intrusions were also frequently found in the granite host rock as dyke (dike) consisting of quartz DIORITE or sometimes as ANDECITE, DECITE rock. The dyke rocks, especially quart DIORITE, proved to be extremely strong rock with UCS test strength of 360MPa. The encountered soil/ rock conditions in tunnel face along the 6.2km tunnel route are recorded statisti-cally as 52% full face residual soil, 34% mixed face, and 14% full face rock which are based on four tunnel boring machines’ records (Fig. 3). Records showed that the full face soil condition is more signif-icant than mixed face and full face rock between Bukit Brown and Caldecott station (BKB-TSN). How-ever, mixed face condition was slightly more than full face soil between Caldecott station and Mary-


190

mount station (TSN-MRM). Similar full face soil condition was encountered between Bukit Brown and Botanic Gardens station (BKB-BTN), and Botanic Gardens Station to Farrer Road station (BTN-FRR) but encountered mixed face condition was recorded almost same percentage as full face soil in this stretch.

Full Face Rock Mixed Face Full Face Soil

Figure 3: Statistical percentage of soil/ rock face in Contract 854 Tunnel Route Granite (GI-IV), Residual Granite Soil (GV), Residual Granite Soil (GVI), Kallang Formation F1, Kallang Formation E Figure 4: One of the soil/ rock profile in Contract 854, 6.2km Tunnel Route

4. BUKIT TIMAH RESIDUAL GRANITE SOIL

The residual soil of the Bukit Timah Granite Formation generally classifies as GV and GVI based on weathering grade. The completely weathered soil with original rock texture preserved is considered as GV while the completely weathered soil with no rock relic texture in soil is considered as GVI. The re-sidual soil is mainly clayey, sandy silt and silty sand with plastic limit ranging from 20% to 40%. The permeability of residual soil generally recorded as 10

-6m/s to 10

-8m/s. Identifying the interface of resi-

dual soil GV and GVI is subjective and difficult to determine precisely as the weathering grade turns gradually. The GV is more favorable than GVI soil in consideration of face stability during cutter tools

TBM1- BKB-TSN (Outer)

18%

11%

71%

TBM1- TSN-MRM (Outer)

8%

32%

60%

TBM2- BKB-TSN (Inner)

63%

10%

27%

TBM2- TSN-MRM (Inner)

26%25%

49%

TBM3- BKB-BTN (Inner)

8%

60%

32%

TBM3- BTN-FRR (Inner)

14%59%

17%

TBM4- BKB-BTN (Outer)

16%

61%

23%

TBM4- BTN-FRR (Outer)

32%

10%

58%

Tunnel Route

Botanic Gardens

Station

Farrer Road

Station


191

change and construction of the Cross Passages with SPT-N values typically found to be in the region of 60 to 100 for the GV material. However, the interface of residual soil and granite rock is distinctive. The stability of soil face during cutterhead interventions or Cross Passage construction is very impor-tant. For example, the GV soil face (Fig. 5) was proven stable enough for Cross Passage construction with sufficient stand up time to allow safe excavation without any ground treatment. The GV condition is also providing a stable tunnel face for cutterhead intervention. GV soil – During Cross Passage Construction GV soil – During Tunnel Cutterhead Intervention

Figure 5: Bukit Timah Granite Residual Soil - GV

5. BUKIT TIMAH GRANITE IN CONTRACT 854

Several published papers in recent years discussed the appearance of Bukit Timah Granite. As men-tioned in the earlier section headed Geology of Singapore, it is found predominantly in the North to the central areas of the Singapore Island. The granite is divided informally into two groups, firstly the Cen-tral Singapore Granite, which lies from Siglap to Punggol, and secondly the Pulau Ubin Granite in the Southern part of Singapore. However these two types of Granite are not formally distinguished on the geological maps of Singapore. Nevertheless, the Bukit Timah Granite found in C854 can be recognized as part of Central Singapore Granite and also divided into two groups based on rock forming minerals. The first type (Fig. 6), consisted of light grey to whitish grey medium, coarse grained, with the presence of pale pinkish orthoclase feldspar and quartz. This was found from Marymount Station to Bukit Brown RTS. No secondary dyke intrusions were associated with the granite in this area which may differ from the Pualu Ubin Granite which does have hybrid intrusions. The UCS was recorded from 10MPa to 180MPa. Figure 6: Bukit Timah Granite (Central Singapore Granite)


192

Second group (Fig. 7), the granite is hornblende mineral rich namely Hornblende GRANITE was dis-covered from western edge of the Bukit Brown RTS towards Farrer Road station. Hornblende granite is whitish grey with dark greenish hornblende minerals associated with plagioclase feldspar, which has light milky colour intergrowth quartz. The UCS test results of hornblende granite range from minimum 80MPa to maximum 280MPa. Figure 7: Bukit Timah Granite – Hornblende GRANITE

Dyke intrusions were not present in eastern side of contract 854 from Bukti Brown to Marymount, how-ever frequent intrusions were encountered along the tunnel route towards west side between Bukit Brown RTS through botanic gardens station to Farrer Road.

Dyke rock consists of quartz DIORITE which belongs to intermediate igneous rock (geologically), fine grained, extremely hard (UCS – 360MPa), very deep grey or dark greenish grey and associated with more than 10% of quartz and rock forming minerals rich of hornblende. Dyke intruded nearly vertical into the host rock, hornblende granite, from the fracture zone. Dyke intrusion is not only vertical and al-so different direction into the granite body was discovered. Figure 8: Bukit Timah Granite – Hornblende GRANITE and Typical Dyke Rock Intrusion

Petrography of dyke rock contributed to a better understanding of major and minor mineralogy, texture, structure and genetic interpretation. Using XRD (X-Ray Diffraction) scan method and petrography ex-amination revealed the source of origin, high temperature/ high pressure during massive intrusion.


193

Quartz DIORITE – sample 1 Quartz DIORITE – sample 2 Figure 9: Dyke Rock -Quart DIORITE

Figure 10: Quart DIORITE – XRD Scan Results

Table 1. Composition of Dyke Rock – Quartz DIORITE

Mineralogy Observed Alteration Size (mm) Morphology Comments min max av

Primary Plagioclase 33% 60% 0.4 1 0.8 Mostly subhedral Altered K-Feldspar 15% 66% 0.2 3 1 Mostly subhedral Altered Hornblende 12% 63% 0.2 2 1 Mostly subhedral Altered (Amphibole) to euhedral Biotite 11% 61% 0.2 2 0.3 Euhedral Foliation masses Quartz 11% 0.1 4 1 Anhedral Strained, embayed boundaries Magnetite 1% 0.05 0.15 0.08 Euhedral Minute grains

Secondary Actinolite/ 6% 0.02 0.5 0.1 Euhedral Replaces hornblende, Tremolite foliation masses (Amphibole) Chlorite 5% 0.02 0.5 0.1 Euhedral Replace hornblende, Biotite, foliation Epidote 2% 0.03 0.3 0.08 Mostly subhedral Replace hornblende


194

6. DISCUSSION

The geology along the alignment of Contract 854 is entirely within the Bukit Timah Granite Formation and can be considered to belong to the Central Singapore Granite type. The full face massive granite, the soil / rock mixed face ground conditions and also the extremely hard dyke rock within the tunnel alignment posed a real challenge to the TBMs and the Engineers carrying out the tunneling works, as recorded in the paper by McChesney S.A, Gasson P.A, Nair R.S, entitled Slurry TBM Tunneling Risk Control & Lessons Learnt on CCL4 Stage 4 – Contract C854 (International Conference on Deep Exca-vations 2008). However, sufficient soil investigation provided the relevant information to the engineers for better understanding and experienced on underground construction. This paper also discussed the types of granite which can be categorised as GRANITE and Hornblende GRANITE based on actual field data and the records of the many SI boreholes carried out along the alignment. This information was used on C854 to aid the tunneling works and help distinguish the locations of the different types / grades of the Bukit Timah Granite Formation rocks. This information can also be used to update the geological correlation to Singapore Geology for future underground construction projects in the same area.

REFERENCES

CPG Consultants Pte Ltd, 2005, Circle Line Contract 854, Settlement Analysis and Risk Assessment on Struc-tures Report for Bored Tunneling, Volume 1, DOC/854/DTW/CA/1091B,

CPG Consultants Pte Ltd, 2005, Circle Line Contract 854, Geotechnical Interpretative Report

Defence Science and Technology Agency, 2009, Geology of Singapore, 2nd Edition

J.S. Sharmar, J.Chu, J.Zhao, 1999, Geological and Geotechnical Feature of Singapore: an Overview, Tunneling and Underground Space Technology, Vol 14, No.4, pp. 419- 431 McChesney S.A, Gasson P.A, Nair R.S, 2008, International Conference on Deep Excavations, Slurry TBM Tunneling Risk Control & Lessons Learnt on CCL4 Stage 4 – Contract C854 Read H.H, Watson. J, 1970, Introduction to Geology, 2nd Edition, Volume 1 S. Yugami, M. Maw, Y.K. Liew, M.Y. Wijesinghe, C.M. Kho, R. Nair, 2009, Underground Singapore, Back Analysis of Tunneling Settlement Parameters for Circle Line Contract 854 Taisei Corporation, MinServ (Mineral Services) Australia, January 2007, Geological Report


195

1 INTRODUCTION

Downtown Line Phase 2 Project (DTL2P), as shown in Figure 1, which is currently under construction

in Singapore, presents a number of major geotechnical and engineering challenges. It comprises the con-

struction of twelve stations and associated tunnels with a total length of 16.6km.

Figure 1 Plan showing the alignment of DTL2 and the location of C917 & C918

Geotechnical & engineering challenges for Downtown Line

Stage 2 C917 & 918 projects

D.C.C. Ng, R. Prasad, B.W. Tew, C.W. Neo, K.F. Pong, R. Supargo Meinhardt Infrastructure Pte Ltd (Member of Meinhardt Group)

ABSTRACT: Downtown Line Phase 2 Project (DTL2P), currently under construction, presents a num-

ber of major geotechnical and engineering challenges. It comprises the construction of twelve stations

and associated tunnels with a total length of 16.6km. Contracts 917 & 918, which are part of DTL2P,

comprise of the construction of 3 stations, cut & cover tunnels and twin bored tunnels. The C917 & 918

alignment is approximately 4km long. Based on the interpretation of the ground conditions, the rock

head level and the thickness of the in-situ materials of Residual Soil of Bukit Timah Granite Formation

are anticipated to be rather undulating and variable along the cut & cover construction of the stations

and tunnels. The engineering properties of the Bukit Timah Granite Formation and the high variability

of the rock head level pose great challenges to the design and construction of the excavation support

system (ERSS) for the project. The stations and cut & cover tunnels are to be constructed along the

busy Bukit Timah Road & Dunearn Road and the critical Bukit Timah Canal. This paper describes in

details two of the geotechnical and engineering challenges in designing the ERSS for the cut & cover

tunnels and mined tunnel undercrossing the Bukit Timah Canal.


196

Across the project, there are numerous engineering challenges driven principally by the complex ground

conditions and site constraints. Contracts 917 & 918 (C917 & C918), which are part of DTL2P, com-

prise of the construction of 3 stations, cut & cover tunnels and twin bored tunnels, which is approx-

imately 4km long.

Deep excavation project in a similar scale in Bukit Timah Granite have been previously done and pub-

lished for Circle Line Stage 3 Contract 853 - Construction and Completion of Marymount Station in-

cluding Tunnels (Loo et al, 2006). In this paper, two major challenges in the design and construction of

C917 and C918 are described in details. Firstly, it is the geotechnical and engineering challenge in de-

signing a safe and practical earth retaining and stabilizing system (ERSS) given the conditions of the

completely weathered Bukit Timah Granite Formation. Secondly, it is challenge of designing and con-

structing an adit undercrossing the critical Bukit Timah Canala and the existing ever busy Dunearn

Road.

2 EARTH RETAINING SYSTEM (ERSS) FOR C918 CUT AND COVER TUNNELS

2.1 Ground Conditions

Based on an interpretation of the inferred ground conditions, the rock head level and the thickness of the

in-situ materials of Residual Soil, which is the Completely Weathered materials and Highly Weathered

materials of Bukit Timah Granite Formation, are anticipated to be rather undulating and variable along

the cut and cover construction of the stations and tunnels. Areas where the rock head declined sharply

have been identified as much as possible along the alignments. The highly variable or sharply declining

rock head may be related to the effects of geological features such as faults and other tectonic activities.

There are areas where the rock head drops sharply which suggest that geological features such as faults

may be present in the area and likely to traverse the proposed cut and cover construction for the stations.

Figure 2 shows the geological profile of the cut and cover tunnel area.

Figure 2 Geological profile for C918 cut and cover tunnel area

The engineering properties of the Bukit Timah Granite Formation and the high variability of the rock

head level pose great challenges to the design and construction of the excavation support system for the

project. Loo et al 2006 has given the properties of the Bukit Timah Granite Formation for the Circle

Line Stage 3 Project. Table 1 shows the design parameters adopted in this project for design.


197

Table 1 Design parameters Unit Fill F1 F2 E (Peat) Insitu Weathered Soil

G-VI G-V G-IV G-III G-II

Bulk unit

Weight

Yb

kN/m3

18.7 19.0 18.5 15 18.5 19.5 23 23 24

Effective

Angle of

Friction

Degree G:30

C:27

G:29 25 22 30 31 40 45 50

Effective

Cohesion

c’ kPa G:0

C:0

0 0 0 3 4 50 300 500

Undrained

Shear

Strength

Cu kPa 20 - 21 d≤5m:

Cu=5kPa

d>5m:

Cu=0.22Po’

≤20kPa

d≤5m:

Cu=20kPa

d>5m:

Cu=4.8

d≤10m:

Cu=30kPa

d>10m:

Cu=(20d-

110)/3

≤230kPa

- - -

At Rest Earth

Pressure Co-

efficient

Ko 0.5 0.7 1.0 1.0 0.8 0.8 0.8 0.8 0.8

Young’s

Modulus

Eu MPa 8 Eu=1.5N E

u/C

u=300 E

u/C

u=200 2N

(max 200)

2N

(max 200) 300 3000 6000

E’ MPa E=0.87E

u

E=0.87E

u

E=0.87Eu E=0.87E

u E=0.87E

u E=0.87E

u

Permeability K m/s G:1x10

-6

C:1x10-8

2.5x10-6

1x10-7

1x10-9

1x10-6

2x10-6

2x10-6

At the first

10m below

rock head:

5x10-6

From 10m

below the

rockhead:

1x10-7

At the first

10m below

rock head:

2x10-6

From 10m

below the

rockhead:

1x10-7

2.2 ERSS Design for Cut and Cover Tunnels

The layout of the ERSS for the cut and cover tunnel in C918 is shown in Figure 3. The ERSS design

consists of diaphragm wall, 7 layers of struts and a layer of jet grouting (JGP) slab beneath the forma-

tion level, as shown in Figure 4. The ground treatment (JGP) layer acts as an improved stiffen soil to

provide better passive support for controlling the deflections of the pile below strut level S7. The ho-

mogeneous ground improvement zone corresponds to approximately 6 800 jet grout columns, each 3 m

long, to be formed approximately 23 m below the construction platform level. Based on the progress,

there is a risk that this might delay the construction program. Hence, sensitivity analyses were carried

out to determine whether it was possible to reduce or completely eliminate the ground improvement

layer. It was later concluded by the sensitivity analyses that the ground improvement layer was indeed

critical for ensuring the deflection of the wall to be within the allowable limit.

Figure 3 Layout of C918 cut and cover tunnel


198

Figure 4 Typical Cross Section of cut and cover tunnel ERSS

With the design parameters stated in Table 1, analyses were carried out for the staged excavation using

bottom up construction method. Figure 5 shows the Plaxis model used for the analysis for the construc-

tion of the cut and cover tunnel by bottom up method with the use of a layer of ground improvement be-

low the formation level. Figure 6 shows the wall deflection and bending moment diagram from the ana-

lyses output. Although the GVI and GV soil are considered as relatively good material, the wall

deflection of the staged excavation exceeded the allowable limit stipulated in the BCA Advisory Notes,

which is 0.5% of excavation depth for this case, if no ground improvement was used. Therefore, a layer

of ground improvement below the formation level was introduced to limit the wall deflection to be within

the BCA limit.

Figure 5 Plaxis model used for the analysis for the construction of the cut and cover tunnel by bottom up me-

thod with the use of a layer of ground improvement below the formation level


199

Figure 6 Wall deflection and bending moment diagram from the analyses output

2.3 Forming of JGP in Completely Weathered Bukit Timah Granite Formation

There have been a number of literatures reporting the successful modeling of JGP and application of

JGP to control the wall deflection for deep excavation (Lim et al. 2003, Wen 2005 and Wong et al

2006). The quality, strength and stiffness characteristics of JGP form has also been widely discussed in

literatures (Osborne et al 2009, Kamruzzaman et al 2005, O’Carroll et al 2003 and Shirlaw et al 2003).

While the literatures show that it is common to use JGP in deep excavation projects, there has not been

much case histories reporting on the installation of JGP in residual soil made up of completely wea-

thered Bukit Timah Granite with SPT ranging from 15 to 40 (GVI and GV material of Bukit Timah

Granote Formation). Check with the JGP contractor has also reviewed that it is not commonly done and

the operating parameters for forming JGP in such ground have not been widely established. Therefore

there was a concern on whether the JGP can possibly be formed in such ground condition. This was a

crucial factor as the ERSS design hinged on the use of JGP below the formation level.

Trials were carried out to determine the suitability of the selected operating parameters to form JGP in

the GVI and GV soil. The trial panel which consisted of 8 nos of 1200mm diameter jet grout columns

was installed. The depth of the improvement zone is from RL80 to RL83, which consisted of stiff to

hard sandy silt with SPT values ranged from 37 to 42. Four nos of boreholes were selected to obtain

cored samples through full depth of the grouted block. Of the four boreholes, one was located in the

centre of the pile, one at the point two third of the pile radius from the centre, and the remaining two at

the overlapping areas of the piles. Four additional locations were selected at the overlapping areas of

the piles to perform SPT.

From the coring results, all cored samples have achieved at least 85% Total Core Recovery (TCR), ex-

cept one which has a TCR of 48%, which is probably due to disturbance during sampling process. SPT

N-values ranged from 13 to 100, with an average of 66. Unconfined compressive strength (UCS) re-

sults show values ranged from 342kPa to 1049 kPa, with an average of 637 kPa. The E-modulus re-

sults obtained ranged from 99 MPa to 289 MPa, with an average of 179 MPa. These results of UCS

and E-modulus have exceeded the minimum values of 600kPa and 90MPa respectively.

The JGP Results from the trial shows that it is feasible to install JGP within the completely weathered

Bukit Timah Granite. Figure 7 shows the core samples from the trial. The core samples show that the

JGP can be formed based on the tested operating parameters. At the time of writing this paper, the ac-

tual work for the JGP and the corresponding quality control tests are in progress. The results will be

presented during the conference.


200

Figure 7 Core samples from the JGP trial

3 DESIGN AND CONSTRUCTION OF ADIT UNDER BUKIT TIMAH ROAD FOR C917

In C917, to facilitate the construction of the ventilation shaft which will be connecting to the proposed

King Albert Park (KAP) Station at the Bukit Timah road side, the construction method using temporary

piped box tunnel with mining method has been proposed. The temporary piped box tunnels will under-

crosss the existing Dunearn Road and Bukit Timah Canal. The temporary steel pipes box tunnel con-

sists of the temporary horizontal steel pipes and steel support closed frames, using the universal steel

column members.


Based on an interpretation of the inferred ground conditions, the rock head level and the thickness of the

in-situ materials of Residual Soil. Completely Weathered materials and Highly Weathered materials of

Bukit Timah Granite Formation are anticipated to be rather undulating and variable along the cut and

cover construction of the stations and tunnels. Figure 8 shows the geological profile of the adit area.

Figure 8 Geological profile of the adit area


201

3.2 Mined Tunnel Design for Adit Construction

The layout of the adit is shown in Figure 9. There were various constraints to the construction of Adit

structure by open cut method. There is the existing KTM Rail Line next to the site which made tempo-

rary diversion of traffic and canal impossible. The utilities locations could not be determined accurately

and therefore careful excavation need to be carried out to locate these utilities and support them during

open cut excavation. The open cut method will induce excessive deformation on the retaining wall for

the construction of station main box structure. As a result, mined tunnel was considered to be the most

appropriate construction method to minimize the risk and disturbance to the surrounding. The cross

sections for the design of mined tunnel are shown in Figure 10. The mined tunnel consists of a 3-cells

tunnel supported by jacked-in pipe piles and will be excavated sequentially.

Figure 9 Layout of adit structure for the vent shaft to be constructed by mined tunnel

The construction sequences of the excavation of temporary piped box tunnel will be as follows (Cell 1

being the left most tunnel cell and Cell 3 being the rightmost tunnel cell):

• As the pipe piles will be terminated into 1.2m thick of improved soil by Wet Soil Mixing

(WSM) Method, the ground improvement by WSM will first be completed.

• The temporary piped box tunnel will be constructed by ramming the interlocking steel pipes

from the temporary launch shaft at the existing Dunearn Road side, to go underneath the Du-

nearn Road, Bukit Timah Canal and reach the outer side of diaphragm walls of KAP station

• The temporary piped box tunnel is divided into 3 cells. Excavation in Cells 1 & 3 will com-

mence first. Prior to the excavation inside each cell, high tensile steel dowel bars will be in-

stalled in the exposed tunnel face and serve to stabilize the ground ahead of the excavation and

ensure the stability of the face. The high tensile steel dowel bars (Deformed Bar Type II) will be

overlapped for a minimum of 2 round lengths or at least 4.6m length.

• Drainage holes will be drilled into the exposed face to drain off the ground ahead within that cell

to ensure the excavation in dry condition.

• Subsequent excavation will proceed in round lengths of maximum 2.3m length.

• Upon the completion of each round length excavation, the temporary steel support closed frame

will be installed immediately near the excavation face.

• The above construction steps will be repeated until the excavation has reached KAP station.

• The temporary steel support closed frames will be removed alternately, leaving the remaining

support frames spanning at 4.6m spacing. After which, the casting of ventilation shaft reinforced

concrete structure in the pipe-jacked tunnel will continue in between the alternate bays of 4.6m

length.

Proposed adit structure vent shaft of KAP Station to be constructed

underneath the existing Dunearn Road and Bukit Timah Canal

Exsiting Bukit Timah Canal

Existing Dunearn Road


202

• After the concrete of the ventilation shaft structure has gained the sufficient design strength as re-

quired, the remaining reinforced concrete work of the ventilation shaft structure will be con-

structed upon removing the remaining temporary steel support frames.

• Subsequently the abovementioned steps will be repeated for the last cell, Cell 2.

Figure 10(a) Typical longitudinal section of the mined tunnel

Figure 10(b) Typical cross sections of mined tunnel

3.3 Geotechnical Analyses for the Excavation Sequence for Mined Tunnel Design

The excavation sequence of the mined tunnel was modeled with Plaxis. Figure 11 shows the Plaxis

model used for the design. The slope cut will be made prior to advance the excavation of the ground

ahead. The slope stability analyses using un-drained and drained parameters have been performed using

PLAXIS 2D FE analysis – -c reduction method for 3 critical sections along the longitudinal direction

to determine the factor of safety against face stability of the tunnel excavation i.e. At the beginning,

middle and at the end of the tunnel excavation mostly within the Estuarine Clay (E) layer. The high ten-

sile steel dowel (Deformed Bar Type II) bars will be installed to stabilize the ground ahead of the exca-

vation. It has been modeled by increasing the cohesion value of the original ground ahead of the excava-

tion due to the presence of those dowel bars which is assumed to be of minimum 3 round lengths. The

dowel bars will be overlapped by at least 2 round lengths. The minimum factors of safety of 1.2 for

drained and 1.5 for un-drained condition have been met in accordance to the design criteria.

38m

7m 12m

Cell 1

Cell 2

Cell 3


203

Figure 11(a) -c reduction analysis at the stage upon the first round, at the beginning end of pipe section

Figure 11(b) -c reduction analysis at the stage upon excavation of the 8th round near to the middle of pipe section

Figure 11(c) -c reduction analysis at the stage upon excavation of the 13th round beneath the E layer at the end

of the excavation


204

3.4 Settlement Prediction

Figure 12 show the predicted settlement contour at area around the mined tunnel induced by the con-

struction of the mined tunnel. The maximum predicted settlement on the adjacent ground is 65mm due

to the construction of the mined tunnel. The nearest structure to the mined tunnel is Dunearn Road and

Bukit Timah Canal. These structures will be monitored closely during the construction of the mined

tunnel.

Figure 12 Predicted settlement contour at area around the mined tunnel

3.5 Instrumentation and Monitoring

Comprehensive instrumentation and monitoring scheme has been proposed for monitoring the perfor-

mance of the design for the mined tunnel. Figure 13 shows the instrumentation plan and section for

monitoring the performance of the mined tunnel construction. Two arrays of ground settlement markers,

one horizontal real time in-place inclinometer, strain gauges and convergence bolts will be installed prior

and during the excavation of the mined tunnels to monitor the performance of design. Figure 14 shows

the monitoring plan for monitoring the effect on the adjacent structures and ground response by settle-

ment markers, inclinometers and piezometers.

(a) Monitoring Plan for Mined Tunnel (b) Monitoring Section for Mined Tunnel

Figure 13 Instrumentation plan for monitoring the performance of the mined tunnel construction

Main Box Structure for KAP Station

Mined Tunnel

Convergence Bolt

Electrical Strain Gauges

Horizontal Inclinometers


205

Figure 14 Monitoring plan for monitoring the effect on the adjacent structures and ground response by settle-

ment markers, inclinometers and piezometers

However, at the time of writing this paper, the monitoring results is not yet available as the preparation

works are still underway. If available, the monitoring results will be presented during the conference to

be compared against the prediction for evaluation of the design and construction method.

4 CONCLUSIONS

In conclusion, the above two design case studies have demonstrated that the tremendous effort put in

during the design stage has great benefit to the project. Various risks were identified during design stage

and measures were implemented in the design to mitigate the risks. This has greatly minimized the

problems arise during the construction stage and therefore able to ensure that the project can be con-

structed smoothly within the planned schedule.

This paper has shown that JGP can be installed in residual soil if appropriate parameters were selected

and verified by carefully planned and executed trial. The will benefit the future excavation projects in

similar ground conditions where there is need to limit the wall deflection in order to minimize impact to

the surrounding structures. The design of the mined tunnel to construct the adit structure undercrossing

the existing Dunearn Road and Bukit Timah Canal has demonstrated that with proper planning and de-

sign, the construction of the underground structure can be executed with minimum impact to the sur-

rounding and disruption to the daily life and activities of the stake holders.

At the time of writing this paper, the preparation works for the construction of the mined tunnel is still

underway and therefore no monitoring data was available yet to be presented in this paper. The moni-

toring results will be presented during the conference

ACKNOWLEDGEMENT

The authors would like to thank Mr James Stabler (Design Manager of Alpine) for providing valuable

input to the preparation of this paper. The authors are grateful for LTA permission to publish these

case studies in this conference.


206

REFERENCES

BS8002.1994, British Standard Code of Practice for Earth Retaining Wall, British Standards Institution.

P.C.Loo, W.L. Chow and T.O.Nay. 2006. Grouting in Weathered Granite – A risk Management Approach for

Deep Excavation in Bukit Timah Granite Formation. Proceedings International Conference on Deep Excava-

tions. Singapore 2006.

S.Khazaei, H.Shimada, K.Matsui. 2004, Analysis and Prediction of Thrust in Using Slurry Pipe Jacking Me-

thod. Proceedings World Tunnel Congress. Singapore 2004.

M.Kurokawa, Takashi Kuhara, A. Shima, M. Iizumi. 2004, Construction of Shaft Tunnel by Slurry Pipe Jack-

ing Through Sedimentary Ground Under High Water Pressure. Proceedings World Tunnel Congress. Singapore

2004.

N.Osborne, C.C.Ng. 2009. Jet Grouting (JGP) for Deep Excavation – The Importance of Quality Control. Pro-

ceedings International Symposium on Ground Improvement Technologies and Case Histories. Singapore 2009.

Kamruzzaman A.H.M., Lee F.H., Chew S.H. and Ong T.S. 2005. Strength and Stiffness Characteristics of Ce-

ment Treated Singapore Marine Clay. Proceedings Underground Singapore 2005.

Lim P.C., and Tan T.S. 2003. A floating type braced excavation in Marine Clay. Proceedings Underground

Singapore 2003.

O’ Carroll J., Loganathan N., Flanagan R., Ratty D. 2003. A Correlation between Energy Input and Quality for

Jet Grouting in Marine Clay. Proceedings Underground Singapore 2003.

Shirlaw J.N., Wen D., Kheng H.Y. and Osborne N.H. 2003. Controlling Heave during Grouting in Marine Clay.

Proceedings RTS Conference Singapore.

Wen D. 2005. Use of Jet Grouting in Deep Excavations . Chapter 13 of Ground Improvement – Case Histories.

Wong K.S., and Goh A.T.C. 2006. Modelling JGP Slab in Deep Excavation Analysis. Procedings International

Conference on Deep Excavations, 2006.


207

1 INTRODUCTION Downtown Line Stage 2 (DTL2) is a 16.6 km-long, underground subway system in Singapore, consist-ing of one depot and 12 stations including 3 interchange stations, and is scheduled to be completed by 2015. Contract 915 (C915) comprises twin-bored Tunnel Boring Machine (TBM) tunnels between the Beauty World and Hillview stations, 9 cross passages and cut and cover structures. Figure 1 shows the location of the DTL2 C915 construction site.

Figure 1. Location of the DTL2 Contract 915 construction site.

Application of multiple-deck-charge blasting with electronic detonator at DTL2 Contract 915 in Singapore

C.O. Shin, T.Y. Ko, S.C. Lee SK E&C, Geotask Team, Korea

M.S. Cho, J.W. Yoon SK E&C, DTL2 Contract 915, Singapore

H.S. Lee SK E&C, Singapore Branch, Singapore

S. Hoblyn, E.M. Aw Land Transport Authority, DTL2 Contract 915, Singapore

ABSTRACT: A TBM launching shaft in DTL2 Contract 915 site is located in a typical hard Bukit Timah granite formation and lots of blasting work is required for shaft sinking. The original blast de-sign used the electric detonator and ANFO blasts consisting of 30 holes per one blast with 1.5 m depth of drilling hole because the rock head was expected at deeper locations. However, significant delay of work and poor progress were expected due to a very shallow rock head profile during construction. To overcome such constraints, an efficient new blasting method which can improve productivity and sa-tisfy vibration limit was required. The revised blast design, using triple-deck blasts with electronic de-tonators and cartridge emulsion explosives, gives better construction performance and can reduce con-struction time. Such a new blasting technique can be effectively used for similar underground projects in the future where the volume of rock blasting is significant.



208

Because the tunneling routes are located in typical mixed ground conditions, three slurry TBMs with very high capacity are used to overcome mixed ground problems. Cut and cover structures at the C915 consist of 2 TBM launching shafts and a cut and cover box. Since about 20% of the entire excavation volume for cut and cover structures are rock excavation, lots of rock blasting work is required. Be-cause construction-related blasting activities are not common in Singapore, rock blasting in urban area is regarded as a special and difficult construction activity. Most blasting work use conventional elec-tric detonators and Ammonium Nitrate Fuel Oil (ANFO) as blasting agents, therefore blast productivi-ty and vibration control are unfavorable. This paper presents a blasting method of multiple-deck blasts with electronic detonators and cartridge emulsion explosives to overcome construction constraints and show how the new method can improve blast productivity in urban environment. 2 CUT AND COVER EARTH RETAINING OR STABILIZING STRUCTURE (ERSS) Figure 2 shows the plan view of the cut and cover ERSS in C915 site. The dimensions of TBM launch-ing shaft B are 31.2 m (width) 25.7 m (length) 26.2 m (height).

Figure 2. Plan view of TBM launching shaft B at the C915 site

a) Estimated rock cover at tender stage b) Real rock cover after additional site investigation

Figure 3. Rock cover thickness estimated at tender stage and after additional site investigation.


209

Geological formation at the site composed of residual soil and Bukit Timah granite. As shown in Fig-ure 3, launching shaft B has thick Bukit Timah granite layer with an uni-axial compressive strength of 200 MPa. The excavation rock volume of the launching shaft B is estimated at about 10,504 cubic meters. Dur-ing the tender stage, the maximum rock thickness from the bottom was estimated to be about 5.6 m. However, the maximum rock thickness was found to be 22.2 m after additional site investigation dur-ing the construction. Cut and cover Earth Retaining or Stabilizing Structure (ERSS) consist of Secant Bored Pile (SBP) wall, king post, diagonal strut, horizontal strut, and metro deck. Only two blasts per day is allowed and al-lowable peak particle velocity is 30 cm/sec for the ERSS and 1.5 cm/sec for residential buildings. An original blast design involved the electric detonator and ANFO explosive consisting of 30 holes per one blast with 1.5 m depth of drilling hole. Based on careful analysis with the data from adjacent construction sites, delayed work progress was expected due to low blast productivity. Shaft sinking activity is one of the critical paths to fulfill planned construction schedule because of highly mixed ground condition for tunneling. Thus, an efficient blast method which can improve blast productivity and also satisfy vibration limit is required to overcome schedule constraint. The following factors are considered in selecting the revised blast design.

- Delay periods and charge weight for one blast per day - Maximum hole depth per one blast - The number of decks for blasting efficiency - Selection of detonators for precision blast - Selection of explosives for vibration control - Cover system to reduce flyrock - Drilling rigs to improve efficiency

After considering the above factors, a revised blast design was proposed using triple-deck blasts with electronic detonators and cartridge emulsion explosives based on contemporary state-of-the-art of blasting technology. 3 ELECTRONIC DETONATOR SYSTEM The most widely used explosives in construction site are dynamite, emulsion and ANFO. These explo-sives need detonator (blasting caps) to blast. There are three types of detonators according to firing methods : electric detonator, non-electric detonator and electronic detonator (Figure 4).

Figure 4. Three types of detonators (After Miller & Martin, 2007)

Recently developed electronic detonators, having itself electronic chip and control device, can deliver a variety of benefits in accuracy and precision. The delay of an electronic detonator is controlled by a


210

quartz clock within a central processing unit on the electronic chip. This allows an electronic detonator to fire as accurately as 0.01% of its nominated delay time (Miller & Martin, 2007). The introduction of electronic detonation revolutionizes rock blasting technique, making it possible to overcome the prob-lems with ground vibration and increase the number of blast holes per one blast (Persson et al., 1993). Table 1 summarizes the characteristics of an electronic detonator. Table 1. Characteristics of electronic detonators

Detonator shell Copper

System operating temperature -20 ~ +50 degrees centigrade

Maximum delay 20,000ms

Maximum number of detonators per one blast 450

Maximum surface wire length 2.5 km

4 MULTIPLE-DECK BLASTS WITH ELECTRONIC DETONATORS

4.1 Problems

The construction of cut and cover ERSS involves stages of excavation, mucking out, and support (strutting). Excavation in rock was planned using electric detonator and ANFO as a blasting agent, but the following problems were found.

- Limitation of the number of electric detonators (maximum 30 nos per blast) to satisfy vibra-

tion restriction - Large ground vibration due to ANFO characteristics - Less efficient work sequence in limited area

Thus, the original blast design causes delayed work progress and it is hard to keep the construction schedule. It was concluded that rock blasting works would affect the delay of construction and subse-quent tunneling works significantly. In order to improve rock blasting efficiency, multiple-deck charge blasting method was suggested for increasing the number of holes and volume of shot rock per one blast. Also electronic detonator was chosen to increase efficiency of deck blasting by accurate control of the delaying. As drilling length is increased with deck blasting, more powerful explosives are required to overcome rock confinement condition. Therefore, cartridge emulsion explosives were selected. Cartridge emul-sion type explosive has higher strength and efficiency than ANFO and has the additional advantage of reducing ground vibration (Olofsson, 1988). To minimize flyrock, soil covering with thickness of 2~3 m was suggested at initial stage. However, blasting mats consisting of Tatami (rice straw) mat and rubber tire mat were used to reduce time in covering works during the blasting cycle. Table 2 shows the advantages and disadvantages of each blasting design. Table 2. Advantages and disadvantages of original and revised blast design

Blasting design Original blasting design Revised blasting design

Advantages Conventionally used method in Singapore Increase in the number of blast holes per

one blast

Reduction of ground vibration

Increase in blasting efficiency

Disadvantage Limits in the number of blast holes per one

blast (30 holes per one blast)

First time use in Singapore

More costly (expensive)

4.2 Revised blast design

The revised blasting design is composed of triple-deck blasts with electronic detonators and cartridge emulsion explosives, and it is suggested for improving construction productivity and maintaining sta-bility of retaining walls and support structures. The drilling length per round was determined to be 4.5


211

m considering the height between struts and working spaces. The number of decks was selected to be 3 based on test blasting results. Figure 5 shows the principle of triple deck blasting method. Three ver-tical decks are blasted sequentially in one blast round by very accurate control of delay with electronic detonator. The number of blast holes was determined to be 233 for considering time in charging and installation of blasting mast. The cut and cover section were divided into 3 zones as shown in Figure 6 and each zone was blasted alternately. Table 3 summarizes the comparison of the original and revised blast de-sign. Figure 7 shows work sequences for triple-deck blasting preparation.

Figure 5. Principle of triple decking blasting.

Figure 6. Layout of blasting area and schematic diagram of blast hole.

Table 3. Comparison of the original and revised blast design

Blasting design Original method Revised method

Excavation Area 24 m 32 m 24 m 32 m

Depth 4.5 m 4.5 m

Drilling depth per round 1.5 m / step 4.5 m / step

Burden and spacing 1.1 m 1.1 m

Required sector 24 3

Total number of drilling holes 2097 (699 ea 3 steps) 699

Total number of detonators 2097 (699 ea 3 steps) 2097 (699 ea 3 decks)

Total drilling length 3145.5 m 3145.5 m

Max. nos of blast holes per round 30 233

Max. nos of detonators per round 30 699 (233ea 3 decks)

SBP wall Line

Drilling

3 Deck

Blasting


212

a) Drilling b) Charging

c) Electronic detonator with connecting wire d) Scanning

e) Covering f) Input delay time

Figure 7. Work sequence for the revised blasting method on the shaft B in C915.

4.3 Comparison of cycle time

The revised blast method is able to reduce 21 days in excavation time for depth of 4.5 m at the TBM launching shaft B. If it is assumed that both blasting designs have the same mucking out time, the re-vised method can save about 105 days in the whole excavation of TBM launching shaft B with 5 strut levels. This construction site is subjected to restricted working and vibration regulations such as two blasts per one day only and strict vibration limits control. Such restrictions can make big difference in con-struction time when electronic detonators are used. Also, soil covering method with thickness of 2~3 m was changed to blasting mats consisted of Tatami (rice straw) mat and rubber tire mat to reduce significant time in covering works. Table 4 shows comparison of cycle time in each blast design to excavate rocks with depth of 4.5 m be-tween two methods.


213

Table 4. Theoretical comparison of cycle time for a strut layer

Blasting design Original method Revised method

Total number of drilling holes 2,097 (699ea 3 steps) ; 1.5m 699 ea ; 4.5m

Total blasting sector 72 (24 sectors 3 steps) 3

Drilling length 1.5 m 4.5 m

1 Blast

Max. nos of blast holes 30 233

Drilling time 0.25 days 1.5 days

Charging time 0.125 days 1.0 days

Blasting time (including

covering and ventilation

time)

0.125 days 0.5 days

Sum 0.5 days 3.0 days

Total

Blasting 36 days (0.5 days 72 sectors) 9 days(3 days 3 sectors)

Mucking - 6 days

Strutting 6 days 6 days

Cycle time per strut 42 days 21 days

5 CONCLUSION This paper presents a case study involving the application of s revised blasting method to overcome low productivity and schedule constraint during the shaft sinking at DTL2 C915.

The revised blast design applied to the work can be summarized as follows;

1) In order to improve construction productivity under restrictive vibration regulations (two blasts per one day and strict vibration limits), triple-deck blasts with electronic detonators are proposed. The revised blast design can reduce 21 days in excavation time for a depth of 4.5 m at the TBM launching shaft B and save about 105 days in the whole excavation of TBM launching shaft B with 5 strut levels.

2) The drilling length is determined to 4.5 m based on the height between struts and working spaces.

The number of decks is selected to be 3 through test blasting results. 3) As drilling length is increased with deck blasting, more powerful explosives are required to

overcome rock confinement. So, cartridge emulsion explosives are selected instead of ANFO. 4) In order to prevent flyrock, soil covering with thickness of 2~3 m was initially considered. How-

ever, this was later changed to blasting mats consisting of Tatami (rice straw) and rubber tire mat to reduce time in covering works.

A lot of blasting works are anticipated for future MRTs, tunnel and cavern projects in Singapore. New and efficient blasting methodology should be applied to enhance the productivity of the blasting works. The method applied to C915 site can provide a good start pointing for such efforts.

REFERENCES

Miller, D. & Martin, D. 2007. A review of the benefits being delivered using electronic delay detonators in the

quarry industry. Proc. Quarrying 2007 Conference, Hobart, Tasmania, 24-27 October 2007.

Olofsson, S. 1988. Applied explosives technology for construction and mining. Arla, Sweden: Applex.

Persson, P., Holmberg, R., & Lee, J. 1993. Rock blasting and explosives engineering. New York: CRC Press.


214

ABSTRACT: Singapore’s newest integrated resort, Marina Bay Sands was completed on record time

and has garnered numerous engineering awards. The development sits on sand infill, which in turn rests on deep soft marine clay deposits. With an average excavation depth of around 20 metres, the

15.5 hectare waterfront development involved some of the largest marine clay excavation in

Singapore. About 2.8 million cubic metres of fill and marine clay were excavated from the site

equating to about 800 trucks a day for two years. To add to the challenge, a 35metre deep ‘cut and cover’ tunnel next to the Singapore’s longest bridge, the Benjamin Sheares Bridge, was engineered.

To overcome the challenges of the bulk excavation and minimize shoring in difficult soil

environments, innovative excavation solutions and scheme were developed to enable an accelerated construction timetable involving densely packed site works with complex staging and interface issues.

This paper provides an overview of the geotechnical innovations and schemes adopted for the award

winning project.

1 INTRODUCTION

Marina Bay Sands Integrated Resort (MBSIR) is a 15.5 hectare tourist resort development which includes three 55 storey hotels with connecting skypark, a casino, a state of the art convention centre

(MICE), an art science museum, theatres, retail outlets and floating pavilions. As shown on Figure 1,

the west and northern perimeter of the site is bounded by the sea, while the eastern side is interfacing with existing East Coast Parkway highway (ECP) and the Benjamin Sheares Bridge (BSB).

Figure 1. Overall Development at MBSIR.

Innovative excavation design for Marina Bay Sands

J.W. Pappin & W.K. Leong

Arup, Hong Kong

P. Iskandar

Arup, Singapore


217

The development is sitting on reclaimed land, comprising sand infill overlying deep soft clay marine deposits, above an underlying very stiff-to-hard Old Alluvium (OA) layer. The soft marine clay,

coupled with the proximity of the East Coast Parkway highway and the Benjamin Sheares Bridge,

posed significant challenges to the design of the excavation works (Figure 2).

MBSIR involved some of Singapore's largest excavations, with overall some 2.8Mm3 of fill and

marine clay being taken from the site, equating to about 800 trucks a day for two years. In addition, a

35m deep cut-and-cover tunnel had to be engineered for the construction of Singapore Mass Rapid Transit (SMRT) next to the Benjamin Sheares Bridge, which links the island’s east and west coasts

and had to remain operational throughout construction. With more than 40% of the concrete

construction occurring between 18m and 35m underground, the required timetable was only made possible by innovative approach to the excavation in the first year.

Some of the innovations include constructing two 120m-diameter and a 100m-diameter circular

cofferdams, a twin-cell 75m-diameter cofferdam, a 65m-radius semi-circular cofferdam, a T-shaped diaphragm wall and modification to the Benjamin Sheares Bridge to ensure the bridge could safely

tolerate the horizontal movement imposed upon it to facilitate excavation.

2 SITE GEOLOGY

2.1 Reclamation History

Over the past decades, the Marina Bay area has undergone several reclamation phases with the latest

being completed in mid 1990's. As shown in Figure 2, the majority of the development is sitting on Phase VIII (1990's) reclamation zone, while the eastern side is located within the Phase VB

reclamation zone, completed in late 1970's.


Ground level across the site is generally flat at about +103 to +103.5mRL, with the recorded groundwater table being at approximately +100.5mRL. The subsoil conditions typically comprise 12

to 15m thick of reclamation Fill overlying 5 to 35m of Kallang Formation soils, underlain by the stiff

to hard Old Alluvium. The Kallang Formation consists of predominantly soft marine clay, with some

interbedded firm clay and medium dense sand of fluvial origin (Figure 2).

In the main podium area which covers the MICE, Casino, Retail, Theatre and Museum, a thick marine

clay deposit (up to 35m) is generally encountered on the southern end and it gets thinner towards the northern end (Geological Section 1 & 2). On the eastern side, where the Hotel, DCS and DTE are

located, the soft marine deposit is approximately 10m thick, except at the northern and southern end

where a deep OA valleys are encountered (Geological Section 3).

3 CIRCULAR DIAPHRAGM WALLS FOR MINIMAL STRUTTING

Deep excavations were required across the site with general excavation depths ranging from 18m to

35m underground. To overcome the challenges of the bulk excavation and minimise shoring in the

difficult soil environment, the excavation design included four large reinforced concrete cofferdams (see Figure 3) comprising:

• two circular, 120m diameter, cofferdams in the Expo and Convention Centre (MICE) area,

• one circular, 103m diameter cofferdam and a 75m diameter twin-celled and peanut-shaped cofferdam in the Hotel area, and

• one semi-circular, 65m radius, cofferdam in the Museum area.


218

Figure 2. Site location plan and geological profile.

Each circular cofferdam was a dry enclosure, within which excavation and subsequent construction could be carried out without the need for conventional temporary support. The hoop compression

forces within each cofferdam provide an open underground space which allowed a prop-free

environment for the substructure works to be carried out. Apart from the cost savings in steel, elimination of struts reduced congestion and therefore allowed the work to be accelerated (Figure 4).

The only constraint was that excavation within a cofferdam must be deeper than the excavation

outside of it.

The 120m diameter cofferdams were among the largest ever deployed in Singapore generally, and

notable for their depth – down to 18m below ground. They allowed work to progress across the site

simultaneously. The design of the MICE cofferdams in conjunction with a steel truss system to the perimeter diaphragm walls at the MICE area allowed independent excavation between the MICE area

and the Casino and Theatre areas to the north. The single-layer steel truss / strut system (see Figure 5)

enabled the 11m deep excavation to be completed outside the two cofferdams in the MICE area.

Legend for Geological sections

Fill

Marine Clay

Old Alluvium


219

Excavation Depth Across the Site

Minimal Strutting

Areal View of the Site Showing the Cofferdams

Figure 3. Excavation plan and location of circular cofferdams.

Due to the vicinity of the East Coast Parkway, the use of the peanut-shaped diaphragm wall, without

any crosswall above excavation level, enabled unhindered bulk excavation of the breakwater mole that had been buried during previous reclamation (see left hand side of Figure 6). Parts of the diaphragm

walls of the two Hotel cofferdams doubled as permanent hotel basement walls and load bearing

elements for the Hotel towers. The remaining parts of these walls, and both the 120m diameter

cofferdam diaphragm walls in the MICE area (shown as red in the plan on the right hand side of Figure 6), had to be removed down to the excavation level by "wire cutting" them into liftable blocks

before removal.

Cofferdams :

120m dia. at MICE

103m dia. at Hotel

75m dia. Twin Cell at Hotel

Semi Circular at Museum


220

Excavation Within Circular Cofferdam Excavation Within Semi Circular Cofferdam

Figure 4. Areal view of excavation inside the cofferdam.

Figure 5. Isometric view of analysis model of MICE truss support at +100mRL.

SAP2000 Model for Hotel Twin Cell Cofferdam Removal of Diaphragm Wall to Excavation Level

Figure 6. Overview of twin cell (peanut) cofferdam and the cutting of diaphragm wall above excavation level.


221

5 TOP DOWN CONSTRUCTION AT CASINO AREA

As the layer of soft marine clay is generally thinner in the northern part of the site, a top-down

excavation method with minimum temporary supports was used to facilitate the 4 level basement construction in the Casino area in conjunction with simultaneous strutted excavation in the adjacent

cut-and-cover SMRT tunnel (DTE). After various considerations, it was decided that the practical

way forward was to design the B2 slab to act as a continuous support between the two retaining walls

on the western and eastern sides, which then allowed excavation to B4 and construction of sub and superstructure above B2 to proceed concurrently. Considerable time savings was achieved by allowing

these two activities to go on simultaneously. The construction sequence for this area is shown in

Figure 7.

Stage 1 :

Excavation to B2 level and casting of B2 slab with temporary prop

Stage 2 :

Partial Excavation to B4 and Completion of B2 Slab at Retail

Stage 3 :

Excavation to B4 level and construction of

remaining structures above B2

Stage 4 :

Structure Completion

Figure 7. Construction sequence at Casino area.


222

6 CONTINUOUSLY REINFORCED DIAPHRAGM WALL FOR DCS BOX

For energy efficiency, the Singapore Government required MBS to incorporate a District Cooling

System (DCS), its plant housed in a deep reinforced box north of the peanut-shaped cofferdam. Shear walls constructed with the DCS box enabled unhindered bulk excavation across the Theatre area to the

west. Within the DCS box, a "top-down" excavation method within minimum temporary strutting was

used. The DCS box also doubled as a retaining structure for the deepest excavation in the adjacent

cut-and-cover tunnel where a deep valley of soft marine clay is present. As the Theatre structures are isolated from the remainder of the development, the DCS box has to permanently support the lateral

loads coming from the ground to the east of the DTE tunnel. The large shear forces required to be

transferred into the underlying Old Alluvium require continuously reinforced diaphragm walls. To achieve this support, three east-west shear walls were constructed from diaphragm walls at the

locations indicated in Figure 8. Each shear wall is 1.5m thick x 50m long (approximately) and

comprises a series of male (6.4m) and female panels (3.0m). To ensure continuity, the shorter female

panels are cast with steel end plates on both ends, leaving about 1.5m of reinforcement bars unconcreted at each end for future lapping with the subsequent male panel reinforcement. The general

design principle used to ensure the lateral load transfer from the shear walls into the underlying Old

Alluvium is illustrated in the upper part of Figure 9. Movement predictions across the DCS were made using 2 dimensional Plaxis analyses where the shear wall effect was modelled as a strut

supporting the floor plates of the DCS (see lower part of Figure 9).

Shear Walls Location Plan Construction of Shear Wall

Arrangement of DCS Shear Wall Panels

Figure 8. Shear wall construction at DCS area.


223

Schematic View of Load Path from the Shear Wall into the Old Alluvium

Plaxis Section Across DCS and DTE

Figure 9. Overview of the shear wall design.


224

7 MANAGING THE IMPACT OF EXCAVATION WORKS ON THE BENJAMIN SHEARES BRIDGE

Excavation within the deeper end of the cut-and-cover SMRT tunnel was carried out adjacent to the operational Benjamin Sheares Bridge (BSB) and therefore movement control became critical. To limit

the impact of excavation works on BSB, a stiff temporary strutted T-shape diaphragm wall was

installed and as explained earlier, the DCS box was also designed to resist lateral earth pressure

coming from the east of the MRT tunnel. Despite using these stiff retaining systems, calculations showed that the predicted movement, while not affecting ride comfort, would overstress the shear

connections between piers and deck. Rather than attempting to stiffen the earth retaining system

further, which could lead to an uneconomical and impractical design, the existing fixed shear pins between the deck and the southernmost pier were replaced by fewer, but adjustable, pins (Figure 10).

Periodic adjustment of these pins enabled the last section of the bridge deck to articulate in plan and

render the whole bridge tolerant of the ground movement inevitably caused by the deep excavation for

the MBS and SMRT tunnel. On-going monitoring throughout the excavation enabled comparisons with design predictions and timely adjustment of the shear pins.

Overview of Benjamin Sheares Bridge

Adjustable Shear Pin

Original Connection at Pier 22 (With Dowel Bars) Modified Connection With Adjustable Pins

Figure 10. Overview of adjustable pin connection for BSB.

8 OVERVIEW

The basement structure was completed in 2009. Innovative approaches to the excavation design in

these difficult site and time constraints set a benchmark for future large-scale excavations both within

Singapore and elsewhere.


225

INTRODUCTION 1.1 Industry safety performance Safety standards in the construction industry are far from satisfactory. Government published safety re-cords suggest over the last decade that the number of construction accidents has generally reached a pla-teau in SE Asian countries. Data from Singapore is shown in Figure 1. Falls from height and being trapped or struck by moving objects are consistently the most common causes of accidents, accounting for about 75% of industry injuries.

24 24 25 31 32

0.35

0.30.29

0.27

0.22

0

5

10

15

20

25

30

35

2006 2007 2008 2009 2010

Nu

mb

er

of

Fata

liti

es

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

AF

R (p

er

100,0

00 M

an

ho

urs

)

Construction Fatalities AFR (per 100,000 manhours)

Figure 1. Singapore construction industry safety performance

Safety by design – Clearance of a buried sea wall at Marina Bay, Singapore

P.J. Clark & I. Askew Lambeth Associates, Hong Kong

O. Thoren Gammon Pte, Singapore

ABSTRACT: The Urban Redevelopment Authority of Singapore presented contractors with a unique problem - the removal of a 70m wide by 1350m long tract of breakwater, constructed of granite boul-ders and buried to a depth of approximately 20m in soft, marine clay beneath the Marina Bay reclama-tion. An innovative earthwork support method was developed that made use of a deep embedded retaining wall utilising steel pipe casings. Only two layers of struts were installed. The lower half of the excava-tion was carried out under water, making use of the weight and hydrostatic pressure of the water to pro-vide stability. A systematic observational design approach was introduced to validate geotechnical as-sumptions and define when it was necessary to flood the cofferdam. This project showcased the vital role of design in improving safety performance. To reduce the risk of struts being damaged by plant, working space was maximized with the use of mega-trusses and concrete walings. Protective deflector panels were provided to distribute possible impact loads during gab re-moval of the boulders and the mega-trusses were designed with added robustness to cater for accidental collisions. Other examples of safety by design include maximizing off-site prefabrication and continu-ously improving designs by incorporating site feedback.


226

1.2 Layers of protection People make mistakes so it is necessary to have several layers of defence to ensure that these mistakes do not translate into accidents. Robust safety management systems have several layers of protection with an accident only occurring when all of these layers are breached. The first layer of protection is de-sign. Reliance on front line workers to protect themselves is the last layer (Figure 2).

Figure 2. Layers of protection

Worldwide statistics suggest that 60% of major and fatal construction accidents could have been pre-vented by better application of the design process. It is suggested that a similar ratio will apply to con-struction projects involving underground excavations.

2 PROJECT OVERVIEW 2.1 The challenge Removal of Underground Obstructions (the Mole) at Marina Bay South – Phases I and II was awarded to Gammon by the Urban Redevelopment Authority in late 2006. The final phase of the project will be completed in 2012. Works comprise the removal of a 1350m long; 70m wide seawall buried 20m below ground. The seawall was formed of large granite boulders, some weighing up to 8 ton. It was to be re-moved to clear the land of major obstructions for future development. No temporary works were to be left in-place. Soil conditions at the site comprise 10m reclamation sand fill overlying up to 30m of soft to very soft marine clay interbedded intermittently with layers of soft alluvial deposits. The site is lo-cated in close proximity to a major road so limiting ground movements was a key consideration. The huge span of the excavation, very soft soil, requirement to remove all temporary works and onerous construction regulations meant it was necessary to look beyond conventional approaches. 2.2 The solution Steel tubes,1.5m and 1.8m in diameter, were drilled 6m into firm/hard soils about 40-50m below ground, to form cofferdam walls. The steel tubes were installed using „hard rock‟ piling techniques, employing machines and specialist operators brought-in from Hong Kong. Gaps between the casings were sealed using sheet piles and grouting over the depth of the permeable fill soils. Huge mega-truss supports were installed progressively in pre-fabricated elements as supports to the cofferdam walls and to provide work access. Excavation was undertaken partly in the dry. The cofferdam was then flooded using sea-water to allow excavation of the deeper sections underwater. (Figure 3)


227

Figure 3. Visualisation of construction method (end of dry stage)

Underwater excavation allowed the omission of a jet grout mat that is a common feature of similar ex-cavations and furthermore resulted in at least a 60% reduction in strutting compared with the industry norm for a dry braced cofferdam of similar depth. Minimising the strutting translated to safety benefits in the form of increased working space, fewer heavy lifts and reduced site storage and transportation of support elements. The working procedure was simple, with plant largely operating from ground level, and prefabrication and re-use was maximised. (Figure 4)

Figure 4. Installation of steel tubes to form cofferdam walls

Staged excavation of the seawall was broadly divided into 2 stages, namely the “dry excavation” and the “wet excavation” corresponding to the flooded condition. Strutting installation was carried-out in the dry. To achieve a higher production rate, dry excavation continued until projected wall movements indicated a need to fill the cofferdam with water to maintain movements within predefined limits. Inclinometers were installed along the sides of the cofferdam and monitoring data was reviewed on a regular basis to allow informed decisions on when to flood. The back-analysis of results indicated that the maximum de-flection of cofferdam wall was predominantly controlled by the depth of dry excavation and there was only a negligible increase after flooding. Adopting an observation method, with predefined actions linked to field observations, was a rational means to remove inherent conservatism in design parameters with-out compromising robustness. It also allowed the performance of the cofferdam to be accurately evalu-ated before flooding such that any adjustments in the strut configuration could have been made in dry conditions. Once the cofferdam was flooded it would have been highly problematic to make any adapta-tions. (Figure 5)


228

Removal of rock boulders in a flooded cofferdam posed accessibility challenges. Boulders near the cof-ferdam walls could be removed using grabs supported from cranes sitting adjacent to the wall. However, boulders in the middle of the 70m wide cofferdam could not be reached without access platforms. By bracing the top struts in pairs horizontally, 6m wide access platforms were provided at strut locations. The platforms were designed to support a 120 ton crawler crane and two 30 ton lorries. Each access platform was restrained vertically by bracing to the bottom struts to form a box truss structure (which became known as a mega-truss). With the truss action in place, the mega-trusses performed as bridges running across the cofferdam. They were only supported mid-span by a pair of steel piles. The operating design condition took account of the mid-span piled support; however a check was also carried out to verify that the mega-truss would remain stable even if one of the piles was dislodged.

Utilising the long-span struts as working platforms was not only efficient in terms of material usage, it also minimised the number of vertical supports required within the excavation, essential for ease of re-moval of the large boulders. (Figure 6)

2.3 Sustainability Given that the cofferdam was very long and of a uniform width, and that the steel pipes had to be ex-tracted from the ground, both the cofferdam wall and mega-trusses were designed to be re-used in dif-ferent phases of the project. In excess of 6000 ton of steel tubes and mega-trusses have been re-used up to four times on the project. Of note, one re-use equates to 11700 ton of embedded carbon. Two sizes of steel casing were used on the project (1.5m and 1.8m). The smaller diameter was slotted into the larger

Figure 6. Crane placing deck on mega-truss which serves as excavation

support and as a working platform

Figure 5. Flooded strutted cofferdam


229

to reduce the shipping volume and thereby save carbon cost in transportation. Damaged tubes were re-cycled as steel road plates.

3 CONSTRUCTION SAFETY 3.1 Safety by design Maximising working space inside deep excavations is a key objective to promote safer construction. Us-ing fewer widely spaced struts is favoured to create open working areas and to reduce the number of heavy lifts. Prefabricating struts and temporary working decks off-site is a further positive step in creat-ing a safe construction site. (Figure 7)

Locations for spoil removal should be identified at an early stage so that potential risks of damage to earthwork supports from accidental plant impacts can be designed-out. For example on this project, at mucking-out locations, fender panels were introduced to protect struts and those structural members most likely to be damaged during underwater excavation were deliberately oversized to build-in robust-ness. (Figure 8)

Access and egress provisions for workers are rarely shown on design drawings. This is often left to "trade practice". Accidents due to falls from height and plant impacts could be mitigated if access and egress risks are designed-out with details of access walkways and stairs clearly identified on construc-

Figure 7. Maximising working space and prefabrication

Figure 8. Fender panels for the protection of earthwork supports


230

tion drawings. Barriers, walkway details, plant routes and limiting surcharges were identified on the de-sign drawings for this project. The wider adoption of observational design approaches should be promoted for deep excavations. Ob-servational approaches, when correctly applied, provide a rational means to remove often inherent con-servatism in design parameters without compromising robustness. A further benefit is that observational approaches tend to advance understanding of soil-structure interaction by making designers monitor the actual performance of their designs against predictions. Consideration of the removal of earthwork supports requires an equal emphasis to the installation. On this project temporary concrete walings were used and they required demolition. The designer working closely with the site team introduced slots, void formers and reinforcement breaks to aid removal. Simi-larly, lifting eyes and cut locations for earthwork support installation and removal should be pre-planned and shown on design drawings. 3.2 Communicating design information Temporary works for deep excavations are often shown as "wished-in-place" on design drawings. Gen-erally there is no information provided on how earthwork supports such as struts and walings or tempo-rary access decks are to be installed, used or dismantled. This is usually left to "trade practice" and de-scribed in generic method statements that are generally of low value for risk management. A desired position would be for every step in the construction cycle to be clearly identified on design drawings, in-cluding access provisions for site workers and material delivery. If for operational reasons it is neces-sary to depart from the drawings, changes will be agreed with the designer to allow construction to pro-gress safely.

It is not always possible to mitigate all risks through design. Residual risks should be identified on de-sign drawings as "Designers Notes" and highlighted, through the use of colour or a text box, so that they stand-out above general notes that tend to be fairly generic. On this project the designer highlighted in “Designers Notes” that temporary cut slopes inside the cofferdam required particular care to avoid over-steepening and prevent slips. When presenting information on design drawings it is favoured to use more images and fewer words. Plans, sections and developed elevations are typically shown on deep excavation drawings. Adding iso-metric views can be effective to show spatial arrangements and connection details for earthwork support systems. Including images of people and plant on design drawings is also a favoured practice since it can trigger designers and planners to re-evaluate working space provided. Animations showing the steps in seawall removal, prepared by the designer, proved useful to improve communication of the design in-tent on this project. Designers must encourage feedback on buildability from those engaged in the construction of their de-signs. Ideally dialogue should start during the early stages of design development and continue through the entire construction cycle. Traditional procurement approaches, where designs are produced with lit-tle or no contractor engagement, competitively tendered and awarded shortly before site works com-mence do not drive safer construction. Designs produced by designers integrated within construction teams tend to be more efficient and easier to build safely as was the case for this project. During the construction of deep excavations, design changes are often necessary due to unexpected ground conditions, utilities and because of differences between as-built conditions and construction drawings. Changes must be referred back to the designers to ensure that appropriate checks are con-ducted so that the work can proceed safely. Internet based systems, such as Gammon's GEMS, used on this project, allow the electronic issue and endorsement of all design changes as part of broader platform for temporary works control. Systems that rely upon paper "change requests" tend to be less effective because they often suffer communication lags and changes are difficult to track. 3.3 Harnessing technology There have been major advances in instrumentation for ground and structure monitoring, particularly during the last decade, with the development of ever more powerful microelectronics and communication


231

networks. Innovation in instrumentation will continue to accelerate with more accurate devices available at reducing cost. Automatic strut load and tilt sensors, connected to an internet based monitoring plat-form were used on this project. Statutory frameworks must be sufficiently flexible to allow advances in instrumentation and modelling to be harnessed by industry. For example, advances in instrumentation should promote the wider appli-cation of observational design approaches for deep excavations. This is not happening. Most statutory frameworks are tailored towards having only one approved design outcome rather than a framework of design decisions based on field observations. Similarly, it is becoming common to have web-based ac-cess to instrumentation databases yet there are still onerous contractual and statutory requirements to have daily or weekly hard copy reports circulated. 4 CONCLUSIONS Improving safety on construction sites is a major challenge for our industry. For all projects, including those involving deep excavations, designers have a leading role to make projects easier to build safely. A degree of habitual reluctance on the part of designers to bear a higher level of responsibility for the safety of those who build their designs will need to be overcome. Raising the emphasis in education and training of temporary works design, since they are almost always safety critical, would be a positive step to change this mindset. An industry culture that encourages designers and contractors to invest in the latest technologies should be promoted. In the context of deep excavations, instrumentation and modelling tools are developing at ever increasing rates. Harnessing such technology to deliver safer and more efficient construction must be a shared objective of all those involved in the industry. REFERENCES 2006-2010 Safety Statistics, Ministry of Manpower of Singapore Government, Occupational Safety and Health Division website


232

1 INTRODUCTION Contract 902 under DTL 1 Line comprises the design, construction and completion of the works for

the non civil defence Promenade Station, modification to the CCL Promenade Station and connecting

tunnels between contract boundaries C905 and C903. PB Consultants Pte Ltd in association with GeoEng Consultants (S) Pte Ltd and AEP Consultants/SAA Architect were appointed by Shanghai

Tunnel Engineering Co. Ltd to carry out the design for the Contract 902.

The proposed DTL Promenade Station is being constructed immediately adjacent to the existing CCL Promenade Station to allow passenger movement between the existing CCL and new DTL Promenade

Station. The DTL Promenade Station will be a stacked station with rail levels at RL73m and RL64m at

the outer track and inner track, respectively. The depth of the existing CCL Promenade Station is ap-

proximately 24m from the ground level to the soffit of the base slab while the depth of the new DTL Promenade Station is approximately 43m below the existing ground level. The DTL Promenade Sta-

tion is constructed using diaphragm wall acting as both temporary and permanent earth retaining wall.

At some portion, for a stretch of about 85m, the new diaphragm walls were constructed beneath the existing underground walkways of the CCL Promenade Station. The underground CCL walkway (un-

Design and construction of DTL Promenade Station and bore tunnels for Contract 902 Y.Y. Loh Parsons Brinckerhoff, Singapore

T.G. Ng GeoEng Consultants (S) Pte Ltd, Singapore

S.B. Tay Land Transport Authority, Singapore

Y.K. Lee Shanghai Tunnel Engineering Co. Ltd. ABSTRACT: The C902 project is a Design and Build contract consist of a 42m deep underground in-terchange stacked platform station and twin bored tunnels (600m long) constructed towards the pro-posed station from a 32m diameter circular launching shaft located adjacent to Rochor Flyover. The DTL Promenade Interchange Station is abutting the existing CCL Promenade MRT station on the south and at close proximity to existing high rise buildings on the north. The earth retaining system comprises of 1.0m to 1.5m thick diaphragm walls. Major part of the new diaphragm wall abutting the existing CCL Station is located below the existing underpass leading to the CCL Station. The under-pass has to be underpinned before the foundation piles are cut to allow the low-headroom diaphragm wall machine to operate beneath the underpass. Top-down construction supplemented by multi-levels steel strutting support system is adopted to control the wall and ground movements in compliance with the stringent movement criteria imposed on the existing CCL MRT Station. During the course of ac-tual construction, the measured wall deflection and movement in the CCL Station were found to be much less as compared to the design prediction. The design of the ERSS was subsequently reviewed and optimized for area where the excavation has yet to be completed. The twin bore tunnels passed through the Rochor Flyover at close proximity to the existing piles supporting the south abutment. A cement soil mix continuous piles wall was formed to shield the piles from the effect of the bore tunne-ling. During the course of the tunneling, the bore tunnels also encounter ground anchor strands left-in from Suntec City basement construction and foundation piles supporting the CCL5 M&E room’s. This paper presents the challenges encountered in the construction of the new DTL Promenade Station and bore tunnels from the launch shaft to the station.



233

derpass) is underpinned with barrette piles during construction stage and permanent stumps were con-

structed to transfer the CCL walkway load onto the permanent roof slab of the proposed DTL Station.

A launch shaft was constructed next to the Rochor Road Flyover. The twin tunnels were launched from the launching shaft toward the Promenade Stacked Platform Station which serves as the receiving shaft for the TBM drive. The inner tunnel, which has a lower rail level when entering the DTL station as compared to the outer tunnel, was launched first followed by the outer tunnel. Figure 1 shows the C902 project site plan.

Figure 1: Contract 902 Project Layout Plan

2 GROUND CONDITION AND GEOLOGY PROFILE

Figure 2 shows the borehole layout plan and interpreted soil profile of C902. The ground levels are relatively flat along the route at between +101.5mRL and +103.5mRL except for the localised road

abutment for Rochor Flyover.

From the History Land Use Survey Report, the C902 project is located on 2 separate parcels of re-

claimed land. The land to the north of Nicoll Highway where the launch shaft is located was reclaimed

in the 1950’s. The land between Nicoll Highway to the contract boundary between C902 and C905

was reclaimed between 1971 and 1975.

As such, the stratification of subsoil for C902 is typical that of reclaimed land along Downtown Ma-

rina and East Coast regions of Singapore where reclamation Fill overlying Kallang Formation which in

turn overlying Old Alluvium Formation could be encountered in all boreholes. Exception was noted in the boreholes within the DTL Promenade Station where the reclamation Fill was found directly overly-

ing the Old Alluvium Formation without the presence of Kallang Formation.

3 PROPOSED EARTH RETAINING SYSTEM FOR THE DTL PROMENADE STATION Figure 3 and Figure 4 show the plan and sections of the existing CCL and new DTL Promenade Inter-change Station. In view of the close proximity of the deep excavation to existing MRT station, top-

down construction method was adopted for the DTL Promenade Station to limit the wall deflection

and ground movement associated with the excavation works. The earth retaining wall comprises of

diaphragm walls with thickness range between 1.0m to 1.5m. The typical toe level of the diaphragm wall is RL 55.0m or at least 6m below the final excavation level. The diaphragm walls serve as tempo-

rary as well as permanent earth retaining system for the DTL Promenade Station.


234

Fig

ure

2:

Soil

Pro

file

fro

m L

aun

chin

g S

haf

t to

Pro

men

ade

Inte

rch

ang

e S

tati

on


235

During the excavation phase, the permanent slabs at roof, B2, B3 DTL Transfer Level, B4 Mezz level,

B5 Upper Under Platform and B6 Lower Under Platform levels were constructed floor by floor in top-down construction to act as lateral support against the diaphragm walls. Additional temporary steel

struts were installed when the span between slabs is large or the slabs are not continuous across the

width of the excavation.

Cross walls span between the diaphragm walls below the final formation level were constructed to fur-

ther control the wall and ground movements between GL-13 and GL15 because of the presence of Ma-

rine Clay and deeper excavation at the TBM receiving shaft.

Figure 3: Plan of CCL and DTL Promenade Interchange Station

Figure 4: Longitudinal and Cross Section of Interchange Station

One of the main challenge in constructing the DTL Promenade Station is the need to construct the diaphragm wall abutting the existing Promenade Station diaphragm wall beneath the existing CCL underpass. The latter is a reinforced concrete box structure supported on 800mm diameter bored piles spaced at 6m c/c. These piles have to be removed in order to make space for the low-headroom diaphragm wall machine to maneuver and work below the underpass. Prior to cutting the existing bored piles, new barrette piles were installed by the side of the underpass. The barrette piles were then “stiched” to the sidewall of the underpass structure through shear connection. Once the underpass was underpinned by the barrette piles, excavation proceeds to below the DTL roof slab in stages with


236

installation of two levels of steel strut. A temporary slab is cast at RL88.0 as working platform for the low-headroom diaphragm wall machine. The vertical clear span required between the working platform and to the soffit of the underpass’s slab is 6m. The bored piles supporting the CCL underpass were then cut to allow the load to be transferred to the barrette piles. Upon completion of the low-headroom diaphragm wall, the roof slab is cast encasing the barrette piles. In the subsequent excavation below the roof slab, the barrette pile segments beneath the roof slab were cut and removed as the excavation progress. The load of the CCL walkway is now transferred to the new diaphragm walls through the segment of barrette piles above the roof slab. Fig-ure 5 and Figure 6 show the plan and cross section of the underpinning described above.

Figure 5: Part plan of DTL B2 level showing the existing bored piles and new barrette piles.

Figure 6: Cross Section of Existing Underpass Being Underpinned during Construction Stage and Roof Slab De-

signed to Undertakes the Underpass Loads at Completion Stage


237

4 COMPARISON BETWEEN PREDICTION AND MEASURED DISPLACEMENT Figure 7 shows the diaphragm wall deflection profiles measured by in-wall inclinometer on the east wall (IW2062) and west wall (IW2061) across GL-11 when the excavation level reached RL67m. The maximum wall deflection measured was 12mm and 17mm on the east wall and west wall, respectively. Figure 8 shows the prediction of wall deflections for the design section across GL-11. The maximum diaphragm wall deflection predicted at the stage of excavation to RL67m was 27.5mm and 28.1mm for east wall and west wall, respectively. It is noteworthy that the actual wall deflection measured was much less as compared to the design prediction, especially at the level below RL80 where OA(A) was found. The over-prediction on wall deflection is attributed to the limit of 150MPa set for Young’s Modulus (E’) of OA(A) as specified in the GIBR. To better simulate the actual soil behaviour, the analysis at GL-11 was re-run keeping all the design sequence and soil parameters the same as the original design but E’ of OA(A) was increased from 150MPa to 250MPa . The predicted deflections of east and west walls with increased E’ are shown in Figure 9. The maximum wall deflection predicted at excavation to RL67m reduced to 19mm and 21mm for east wall and west wall, respectively. It can be seen from Figure 9 that even with E’ of OA(A) increased from 150MPa to250MPa, the wall deflection predicted is still larger as compared to the actual measurements. The wall deflection predicted below RL80, which is within the OA(A), is still grossly over-predicted. With the above back-analysis in mind, the construction sequence between GL-1.5 and GL-9.5 was reviewed. The two layers struts below B6, originally designed at RL68 and RL64.5 was revised to one layer strut at RL67, and the works have been sucessufully completed.

Figure 7: Diaphragm wall deflections measured by in-wall inclinometer across GL-11.

40

44

48

52

56

60

64

68

72

76

80

84

88

92

96

100

104

-6-4-202468101214

East Wall Measured Deflection (mm)

Ele

vation (

mR

L)

5/3/2010 (RL99)

30/4/2010 (RL96)

10/6/2010 (RL92)

19/6/2010 (RL88)

15/8/2010 (RL84.5)

14/9/2010 (RL81.5)

20/10/2010 (RL78)

27/11/2010 (RL74.5)

10/1/2011 (RL71)

11/2/2011 (B6 Slab)

44

48

52

56

60

64

68

72

76

80

84

88

92

96

100

104

-4 -2 0 2 4 6 8 10 12 14 16 18

West Wall Measured Deflection (mm)

Ele

vation (

mR

L)

5/3/2010 (RL99)

30/4/2010 (RL96)

10/6/2010 (RL92)

19/6/2010 (RL88)

15/8/2010 (RL84.5)

14/9/2010 (RL81.5)

20/10/2010 (RL78)

27/11/2010 (RL74.5)

10/1/2011 (RL71)

11/2/2011 (B6 Slab)

IW2061 IW2062


238

Figure 8: Diaphragm wall deflections predicted by FEM across GL-11 based on OA(A) –E’=150MPa

Figure 9: Diaphragm wall deflections predicted by FEM across GL-11 based on OA(A) –E’=250 MPa

54

58

62

66

70

74

78

82

86

90

94

98

102

0 5 10 15 20 25 30

West Wall Predicted Deflection (mm)

Ele

vatio

n (

mR

L)

|Max|=28.61

54

58

62

66

70

74

78

82

86

90

94

98

102

-30 -20 -10 0 10

East Wall Predicted Deflection (mm)

Ele

vatio

n (

mR

L)

|Max|=27.75

58.0062.0066.0070.0074.0078.0082.0086.0090.0094.0098.00102.00

Exca2 to RL99 Exca3 to RL96 Exca4 to RL92 Exca5 to RL88 Exca6 to RL84.5

Exca7 to RL81.5 Exca8 to RL78 Exca9 to RL74.5 Exca10 to RL71 Exca11 to RL67

54

58

62

66

70

74

78

82

86

90

94

98

102

0 5 10 15 20 25

West Wall Predicted Deflection (mm)

Ele

vatio

n (

mR

L)

|Max|=20.77

54

58

62

66

70

74

78

82

86

90

94

98

102

-20 -15 -10 -5 0 5

East Wall Predicted Deflection (mm)

Ele

vatio

n (

mR

L)

|Max|=19.1

58.0062.0066.0070.0074.0078.0082.0086.0090.0094.0098.00102.00

Exca2 to RL99 Exca3 to RL96 Exca4 to RL92 Exca5 to RL88 Exca6 to RL84.5

Exca7 to RL81.5 Exca8 to RL78 Exca9 to RL74.5 Exca10 to RL71 Exca11 to RL67


239

5 LAUNCHING SHAFT The launch shaft is located at the north side of Nicoll Highway next to the Rochor Road Flyover’s

north embankment. The twin tunnels were launched from the launching shaft toward the Promenade Station which serves as the receiving shaft for the TBM drive. The inner tunnel, which has a lower rail

level when entering the DTL station, was launched first followed by the outer tunnel (upper tunnel).

This reduce the risk and impact to the first constructed tunnel therefore the construction at the receiv-ing shaft at the station was complete in advance to receive the lower track Tunnel Bored Machines.

The launching shaft depth is 28m from ground level, the F1 and marine clay thickness is 35m. In or-

der to comply with LTA(DBC) requirements of maximum 15mm horizontal and vertical movements

to existing Rochor Flyover structures, the shaft design was changed from rectangle shaft to the circular shaft Three Dimension analysis of the shaft and Rochor Flyover foundation were carried out to com-

ply with the Authority requirements. The prediction of diaphragm wall deflection was 11mm and ac-

tual deflection of 8mm of the Rochor Flyover was recorded during the construction.

The proposed 32m diameter circular launch shaft was constructed with 1.2m thick diaphragm wall

(minimum DW toe level 53.5 or minimum 3m key into OA soil). The connection to C903 cut & cover

tunnel with NATM tunnel with circular pipes roofing were designed so that the hoop stress of the cir-cular shaft is maintained (Refer to Figure 10). Prior to construction of the NATM tunnel, cement soil

mix piles were constructed to strengthen and improve the marine clay layer to prevent soil from col-

lapsed during NATM excavation. This circular shaft with ring beams do not required any struts or

walers for earth retaining temporary supports (refer to Figure 11), therefore all constructions activities were able to be carried out in columns free space within the circular shaft. These have expedited con-

struction activities and cost saving of approximate 15% compare to the rectangular shaft construction.

Figure 10: NATM tunnels connecting to C903 cut & cover tunnel were designed to maintained the hoop stress of

the circular shaft


240

Figure 11: The 32m diameter circular shaft with ring beams without any struts or walers and have provided col-

umns free space for construction activities

6 BORE TUNNEL FROM ROCHOR FLYOVER TO CENTENNIAL TOWER With the Tunnel Bore Machine (TBM) ready to launch from the circular shaft, there are three con-straints and risk that ahead of TBM boring path. a) Close proximity 3.2m distance between TBM and Rochor Flyover South abutment pile. b) 16 rows of decoupled ground anchor wires leave by Suntec City basement wall construction. c) Existing CCL Promenade station M&E room’s bored piles ob-structing path of TBM. a) Close proximity of 3.2m distance between TBM and Rochor Flyover South abutment pile The TBM will be boring at F1 layer at about 3.2m from existing Rochor Flyover south abutment pile. There is risk that sand will move into TBM during tunneling operation and the piles will lost the skin friction and hence cause the bridge to tilt. In order to mitigate this problem, the sacrificial cement soil mix continuous piles wall was designed and installed to prevent soil loss in Rochor Flyover piles dur-ing tunnel boring operation. Refer to Figure 12 below. Figure 12: Separation wall between TBM and existing Rochor Flyover Pile at South Abutment


241

b) 16 rows of decoupled ground anchor leave behind by Suntec City basement wall construction After the construction of Suntec City basement, the ground anchors have been decoupled and decom-mission. However, these left in ground anchor still in the path of TBM enveloped. Since the ground anchor wires were 16mm diameter and TBM torque and cutter head is able to bore through these ob-structions, therefore observation and inspection was carried out at Suntec basement wall during TBM bore through the ground anchors and no cracks or sign of water seepage at wall after the TBM boring through these obstructions. Refer to the ground anchor wires on Figure 13. Figure 13: Ground anchors wires collected in TBM conveyor belt after bore through debonded ground anchors.

c) Existing CCL Promenade station M&E room’s bored piles obstructing path of TBM. Five number of CCL5 M&E room’s bored piles are in the way of TBM boring path. In order to avoid damaged to the existing structures, LTA has called separate contract to construct compensation piles outside the envelope of TBM path and truncate these five piles 300mm away from the base slab, refer to Figure 14. As built drawings were check and conformed that the reinforcement cage of the 600mm diameter bore piles are on the top half of the piles only, therefore it is safe for the TBM to bore through these truncated bore piles. The lower TBM had successfully bore though these obstructions and reach the receiving shaft at PMN Station on 15 February 2011. Then the upper TBM successfully constructed the upper tunnel and reach the receiving shaft at PMN Station on 26 July 2011.

Figure 14: Compensation piles constructed to Existing M&E room’s decoupled bored piles for TBM to bore

through the decoupled piles


242

7 CONCLUSION The noteworthy and critical work elements are as below:-

- Compliance of Code of Practice for Railway Protection (CPRP) for the 42m deep station excava-

tion alongside the running MRT station founded at about 24m below ground.

- Underpinning of existing MRT station underpass and low headroom diaphragm wall construction underneath existing station underpass

- Close proximity of 32m diameter launching shaft to Rochor Flyover and mitigation measure to

comply with 15mm horizontal and vertical movement requirements

- Bore tunnels overcoming underground obstructions. All these construction challenges have been successfully overcome and resolved within the contract period. With more and deeper infrastructures being built in the land scare in Singapore, there will be more construction projects like C902 where the construction of MRT project in congested Central Business District in future. C902 project demonstrate that with all the constraints and difficulties be-ing identified and cost in during tender together with careful planning, design and construction, all these difficulties and constraints can be overcome successfully within budget and on time. This will not only benefit to LTA, STEC, PB and all the parties involved in the project but to all the public commuters where the operator will be able to open the line on time to serve the public on time.


243

1 INTRODUCTION The proposed Downtown Line Stage 3 (DTL3) will be an underground Mass Rapid Transit (MRT) System extending from Downtown Line Stage 1 (DTL1) Chinatown Station and run through MacPher-son, Bedok Reservoir, Tampines and ending at the East West Line Expo Station. Contract 937B, de-sign and construction of Tai Seng Facility Building (TSFB), is part of DTL3. The proposed TSFB will provide maintenance, operation and staff facilities for DTL3.

Figure 1 Layout plan of DTL3 alignment and location of C937B

Partial top-down method & high capacity ground anchors

for optimization of temporary works design

S.Y.H. Low, A.P.C. Yong, D.C.C. Ng Meinhardt Infrastructure Pte Ltd (Member of Meinhardt Group)

ABSTRACT: Contract 937B, design and construction of Tai Seng Facility Building (TSFB), is part of Down-

town Line Stage 3 (DTL3). The proposed TSFB will provide maintenance, operation and staff facilities for

DTL3. It is a 52m wide and 295m long underground building which consists of 2 level basements and entrance

structure at ground level. The schedule for completion of this project is very tight and therefore it is necessary

to optimise the design of earth retaining and support system (ERSS). Contiguous Bored Pile Wall (CBP) re-

taining system supported by temporary ground anchorage and a top-down roof slab is proposed as the ERSS for

the excavation and construction of the underground TSFB. Partial-top down construction method is adopted

where the roof slab with large access opening will be cast first. Excavation will proceed and ground anchors

will be installed as excavation progress downward. The use of partial top-down construction method and incor-

porating the roof slab as part of the support system to the excavation help to reduce one layer of temporary sup-

port system. Optimisation in design of ground anchors by using high capacity ground anchor was also adopted

to minimise the total number of ground anchors to be installed. This paper presents the design approach for the

partial top-down method and the challenges in designing with high capacity ground anchors.

C937B TSFB


244

TSFB is a 52m wide and 295m long underground building which consists of 2 level basements and en-trance structure at ground level. The location plan of DTL3 and C937B is given in Figure 1. This de-sign and build contract will be constructed by cut and cover construction with a maximum excavation depth of about 20m. One of the major challenges in the project is the extremely tight schedule of work. Hence it is crucial to optimize the design in order to ensure that work can be completed within schedule.

2 GROUND CONDITIONS

The ground conditions encountered at the site are generally in line with those anticipated from available

published regional geological information.

The Geological survey by PWD (1976) subdivides this formation into five members namely, Marine,

Alluvial, Transitional, Littoral and Reef. The different members of the Kallang Formation encountered

are Estuarine, Fluvial Sand, Fluvial Clay and Marine Clay. The Old Alluvium is an alluvial deposit

that has been variably cemented, often to the extent that it has the strength of a very weak or weak rock.

The upper zone of the OA has typically been affected by weathering and has typically penetrated as a

discernible front from the surface. All five classes of weathering classification of the OA are encoun-

tered at this site.

Figure 2 Typical soil profiles at the site

A typical soil profile comprising mainly of Fill and Old Alluvium is encountered along the route as on

Figure 2.SI has been carried out for both sides of the walls to assess the ground variability across the

site. The interpretative subsurface profiles have four soil layers, namely Fill, Kallang Formation, Old

Alluvium (OA) (N<50) and OA (N>50). The soil layer boundaries were interpolated in between two

boreholes to develop the subsurface profile. The subsurface profiles indicate that Fill and OA layers

consistently exist along the alignment but in varying thickness. The surface of the Old Alluvium (OA)

is encountered between +96mRL and +109mRL which is above formation of the TSFB. The material

below the formation level is expected to be Class A, B or C OA. Pockets of Kallang Formation are

shown to be present in at least two locations. The upper few meters of OA is weathered with SPT-N

value of less than 50. The lower OA material has SPT-N of more than 50. The engineering properties

of OA material were comprehensively described by Wong et al (2001), Chiam et al (2003) and Chu et al

(2003). Table 1 shows the summary table of the design parameters for the soils.

From the soil investigation report, the variability of the OA material is apparent in the particle size dis-

tribution curves for the various OA of different degree of weathering and depth. The clayey OA has a


245

very wide range of particle size distribution (PSD) curves the data reveals thatthe material ranges be-

tween 40% and 80% fines. While the sandy OA has a tighter band of PSD curves, showing the material

to be silty or clayey sand with a clay/silt content of about 20%.

Table 1 shows a summary table of the geotechnical design parameters for the various soil materials en-

countered for this site that are used in the design of the earth retaining and support system.

Table 1 Summary table of soil parameters

3 ERSS DESIGN AND PROPOSED CONSTRUCTION METHOD The schedule for completion of this project is very tight and therefore it is necessary to optimize the de-sign of earth retaining and support system (ERSS). Contiguous Bored Pile Wall (CBP) retaining sys-tem supported by temporary ground anchorage and a top-down roof slab is proposed as the ERSS for the excavation and construction of the underground TSFB. Partial-top down construction method is adopted where the roof slab with large access opening will be cast first. Excavation will proceed and ground anchors will be installed as excavation progress downward. The use of partial top-down con-struction method and incorporating the roof slab as part of the support system to the excavation help to reduce one layer of temporary support system. Optimization in design of ground anchors by using high capacity ground anchor was also adopted to minimize the total number of ground anchors to be installed.

Figure 3 Layouts of ERSS for C937B

Extent of Area using Partial Top Down Construction Method

Existing viaduct piers of Bartley Flyover


246

Figure 3 shows the layout of the ERSS. Partial-top down construction method will be provided between grids 3 to 16, where the roof slab with large access opening will be cast first. Figure 4 shows a typical section of the ERSS with partial top-down construction. Excavation will proceed and ground anchors will be installed as excavation progress downward. The roof slab will be connected to the CBP wall and supported by bored pile as internal supports at temporary stage. Once the excavation reaches the forma-tion level, base slab will be cast, walls, column and B1 structure will be cast from bottom up. The de-sign of ERSS has utilized high capacity ground anchors of up to 1400kN force per ground anchor in or-der to optimize the ERSS design with larger vertical spacing for the ground anchors. This has reduced the construction time significantly.

Figure 4 Typical section of ERSS using partial top-down construction method with high capacity ground anchors

3 PLAXIS ANALYSES

Computer software will be used for the design of the earth retaining structure and structural elements for the underground station. Finite element analysis to account for any staged construction effects on the retaining structure will be assessed by Geotechnical software PLAXIS. In this project, the behaviour of the temporary earth retaining system was analyzed using finite element software PLAXIS Version 9.02.

Figure 5 Typical models of the Plaxis analyses


247

The interaction between the excavation and temporary retaining system is to be carried out using

PLAXIS version 9.02. PLAXIS is a geotechnical finite element program intended for two-dimensional

deformation and stability analysis. PLAXIS uses fully automated calculation procedures based on ro-

bust numerical methods. The program is able to model soil-structure interaction for excavations in a

realistic simulation that closely follows the actual construction sequence. It allows the application of

surcharge loads, point loads and also models changes in ground water pressures.

Figure 6 Result of analyses for wall bending moment

The soil behavior was modeled using Mohr- Coulomb Model, which is an elastic-perfectly plastic con-stitutive model. Tension cut-off option was activated to cut off all possible developed tension points in the soil elements. The CBP wall was modeled as beam elements with the corresponding axial and bend-ing stiffness. Figure 5 show the typical model of the PLAXIS analyses. The result of analyses for wall bending moment and deflection are shown in Figures 6 and 7 respectively. In this project, the wall def-lection for south wall is controlled by a limit of 50mm from LTA contractual requirement to limit movement of viaduct structure as shown in Figure 3. However the allowable deflection at north wall is greater, which is 140mm or 0.7% of total excavation depth as required by BCA.


248

Figure 7 Result of analyses for wall deflection

5. HIGH CAPACITY TEMPORARY REMOVABLE GROUND ANCHORS

High capacity temporary removable ground anchors of up to 1400kN force per anchor were used to op-

timize the vertical spacing of the ground anchors for the ERSS design. Case history of using high capac-

ity ground anchors has reported by Barley et al, 1999. The proposed 1400kN ground anchor will used

300mm grout hole with 16 numbers of 12.9mm diameter strands There are a number of methods that

has been used in Singapore namely Dummy Strand, Explosive, Water Jet Cutting, Coupling, U-turn

(Chua et al, 1994 and Chua et al, 1997) and SBMA (Chua, 2003). Figure 8 shows the typical details of

the ground anchor. This ground anchor system is a removable ground anchor tie back system using the

U-turn system. Figure 9 shows the typical strand and anchor block for the ground anchor U-turn system.

The ground anchor design and testing was according to BS 8081 and CIRIA Report 580 (Gaba et al,

2003). Each anchor was tested to prevent over-stressing of the strands by measuring the load increments

and measurement of strand displacements adhering to the proposed load cycle as per BS 8081. All ten-

dons were stressed at one end of the anchor head to test load of 125% and locked off at 110% of the

preload value, which was about 70% of the design force of the ground anchors. The rationale for adopt-

ing about 70% of the design forces can be found in paper by Chua et al (2006). After completion of construction and backfilling has reached approximately 0.5m below the ground anchor level, the strands were de-stressed and removed. For extraction of the strands, the tendon is in-itially de-stressed with the help of a mono-jack or first cut one end of the strand behind the bracket and then pulled in the other end. Two trials were carried out for the removable ground anchors. One 100T and one 140T ground anchor was tested up to 150% of the working load and locked off at 125% for a period of 24 hours. The allowable elongation for the trial ground anchor test is 90% to 110% of the theoretical elongation.

Wall deflection

limit to 50mm

Wall deflection

limit to 140mm


249

Figure 8 Typical details of the removable ground anchor

Figure 9 Photo of typical strand and U-turn anchor block

The ground anchor using U-turn system has different lengths. The load transfer mechanism of the ap-

plied load is therefore more complicated. Steps have to be taken to ensure that applied load is distributed

evenly to each holding piece. This is achieved by taking care of the initial differential elongation of each

pair of strands. Each pair of strands has to be stressed sequentially, from the longest to the shortest, to

overcome the differential elongation between them so that they can be stressed together and acts as a


250

group. The shortest pair elongation is the baseline against which remaining pair must be altered to suit.

The theoretical elongation of each pair is calculated as follows:

Theoretical elongation, e = (P x L) / (A x E)

Where P = applied load on each pair of strands, kN

L = total length of strand, m

A= area of strands, m2

E = Young’s modulus, kN/m2

The applied load, P is the ultimate load divided by the number of holding pieces. After tabulating the

elongation, the values are subtracted from each other to find the different in elongation. This differential

elongation is used to calculate the load required by each pair to achieve balanced or equal load at the

maximum test load of 125% WL. The elongation during the test is compared with the theoretical values

specified by the code. If these criteria are not met, the affected ground anchor will be downgraded or re-

placed. Pre-load affects the behavior of ground anchor during subsequent excavation work. For this

project, the preloading of ground anchor is actually 75% for south wall and 60% for north wall.

The following general expression has been used to provide design estimates for the ultimate geotechnical

capacity of individual ground anchor:

TU = DLCa where D is the diameter of the fixed anchor length

L is the length of the fixed anchor

Ca is the unit value of the shaft adhesion, which is taken to be 2N

The magnitude of the grout/tendon interface ultimate bond stress is assumed to be uniform over the ten-

don bond length and is equal to 2.0 N/mm2 for clean strand for minimum grout compressive strength of

30N/mm2 prior to stressing and clear spacing between tendons is not less than 5mm.

6. CASE HISTORIES ON SIMILAR DESIGN AND CONSTRUCTION METHOD A case study was carried out for the ERSS design and construction for the world’s largest underground train depot, Kim Chuan Depot based on the presentation material for Geotechnical Society of Singapore by Dr Ng Tiong Guan on 19

th May 2010. This site is selected for case study as the location is very close

to C937B site and adopting same supporting system, ground anchor. The ground condition is very simi-lar to C937B site which consist of Fill, Kallang Formation and Old Alluvium. Figure 10 shows the ele-vation view of the excavation with ground anchor at Kim Chuan Depot site. The design of ERSS for Kim Chuan Depot consists of soldier pile with sheet pile lagging and supported by 5 layers of removable ground anchor for an excavation depth of 19m. The finite element method (FEM) computer program Sage Crisp was used to analysis the wall and ground anchor performance. Figure shows the analysis results for a typical section with 5 layers of ground anchor.


251

Figure 10 Elevation of Excavation works at Kim Chuan Depot

Figure 11 Predicted deflection and unfactored force in soldier piles

Figure 12 depicts the monitored wall deflection for the excavation up to formation level. The observed

maximum wall deflection was about 16mm which is smaller than the predicted value of 33mm in the

analysis.


252

Figure 12 Inclinometer monitoring results

7 CONCLUSIONS In conclusions, this case study has shown that the optimized design of the ERSS with partial top-down method and high capacity removable ground anchors has been successful in meeting the design objective for the ERSS and ensuring safety in the deep excavation for this project. The schedule for completion of this project is very tight and therefore it is necessary to optimize the design of ERSS to ensure that the project can be completely safely and on schedule. This case study has demonstrated that by using partial-top down construction method with the permanent roof slab as part of the support system to the excavation, it help to reduce one layer of temporary support system and yet make the ERSS a more ro-bust design. The total number of layers of ground anchors can be minimized by using high capacity ground anchor. These optimization of ERSS design has helped in reducing the time of excavation and therefore ensure that the project can be completed within the schedule. At the time of writing this paper, the excavation of the C937B site has only progress up to first layer of ground anchor. Hence there is insufficient monitoring data to be presented in comparison with the design prediction. The monitoring results of the site against the prediction will be presented during the confe-rence. However, from the case history of the ERSS for Kim Chuan Depot with similar supported system og ground anchor, the ERSS has performed better than expected. Therefore, the author has confidence that the performance of the high capacity, 1400kN ground anchor at the partial top down construction me-thod would perform according to design. ACKNOWLEDGEMENT The authors would like to thank Mr Young Pong Chin (Senior Project Manager of LTA) and Mr Ed-mund Ting (Deputy Project Manager from LTA) for their leadership and for motivating safety culture in this project. The authors are grateful for LTA permission to publish this case study in this confe-rence.


253

REFERENCES

BS8081.1989, British Standard Code of Practice for Ground Anchorages, British Standards Institution.

Gaba, A.R., Simpson, B., Powrie, W., & Beadman, P.R. 2003. Embedded Retaining Walls – Guidance for Eco-

nomic, Construction Industry Research and Information Association (CIRIA) Report No. C580, London, UK.

Wong K.S., W.Li, J.N. Shirlaw, Ong J.C.W., Wen D. and Hsu J.C.W. 2001. Old Alluvium: Engineering Prop-

erties and Braced Excavation Performance. Proceedings Underground Singapore 2001

Chiam S.L., Wong K.S., Tan T.S., Ni Q., Khoo K.S. and Chu.J. 2003, The Old Alluvium. Proceedings Under-

ground Singapore 2003

Chu J., Goh P.P., Pek S.C. and Wong.I.H, 2003. Engineering Properties of the Old Alluvium Soil. Proceedings

Underground Singapore 2003 Chua T.S., Marican S.S., Kok T.W., Andrew. 2006. Critical Levels for Monitoring of Ground Anchor System for Deep Excavation. Proceedings International Conference on Deep Excavation 2006 Singapore

Barley, A.D, Payne, W.D. & McBarron, P.L. 1999. Rows of High Capacity Removable Anchors Support Deep

Soil Mix Cofferdam, X11th European Conference on Soil Mechanics and Geotechnical Engineering, Amster-

dam.

Chua, T.S. & Lai, H.P. 1994. Micropiling And Excavation Works for The Republic Plaza Project – Singapore’s

Next Tallest Building, 3rd International Conference on Deep Foundation Practice incorporating Piletalk, Sin-

gapore.

Chua, T.S. & Prasanthee, R. 1997. Performance of Temporary Removable U-Turn Ground Anchors in Singa-

pore, Proceedings of 3YGEC, Singapore.

Chua, T.S. 2003. Recent Development of Ground Anchor Technology in Singapore, 12th Asian Regional Con-

ference on Soil mechanics and Geotechnical Engineering, Singapore.

Ng, T.G. 2010. ERSS for the World’s Largest Underground Train Depot – Kim Chuan Depot, Presentation

Material for Geotechnical Seminar jointly organized by Geotechnical Society of Singapore (GEOSS) and Cen-

tre for Soft Ground Engineering, Singapore.

PWD Geology of Singapore (1976)


254

1 INTRODUCTION

The final phase of the Singapore Circle Line, CCL Stage 5, was constructed between 2006 and 2010. Apart from the bored tunnel sections, cut-and-cover tunnels were constructed using multi-propped exca-vations at five station boxes and two cross-over boxes. The dominant geology is Jurong Formation, with local spots of Kallang Formation in some locations. An intensive instrumentation and monitoring pro-gramme was implemented during the excavation works at each of the sites. This paper reports and ana-lyses the maximum surface settlements and maximum wall deflections at West Coast Station, Pasir Pan-jang Station, Pasir Panjang Cripple Siding, Labrador Park Station, and Telok Blangah Station, as well as some study on the surface settlement profile and the horizontal displacement behaviour of the ground during these excavations.

2 MAXIMUM WALL DEFLECTIONS AND SURFACE SETTLEMENTS

2.1 West Coast (WCT) Station The geological condition at WCT station (now renamed as Haw Par Villa station) comprised mainly the Jurong Formation soils with a 3-5m thick overlying layer of Fill. The excavation at WCT station ranged from about 15m-22m deep. The Earth Retaining or Stabilising Structure (ERSS) for WCT station was made up of contiguous bored pile walls (1m) whose toes are about 29m-35m deep and propped with five levels of struts. Figure 1 shows the geological profile at WCT station and the layout of its ERSS during construction.

General observations on wall and ground deformations during the multi-propped excavations in CCL5

K.H. Goh & W.M. Cham Land Transport Authority, Singapore

ABSTRACT: The final phase of the Singapore Circle Line, CCL Stage 5, was constructed between 2006 and 2010. Apart from the bored tunnel sections, cut-and-cover tunnels were constructed using multi-propped excavations at five station boxes and two cross-over boxes. The dominant geology is Ju-rong Formation, with local spots of Kallang Formation in some locations. An intensive instrumentation and monitoring programme was implemented during the excavation works at each of the sites. This pa-per reports and analyses the surface settlement and wall deflection monitoring results at several of the cut-and-cover locations, namely West Coast Station, Pasir Panjang Station, Pasir Panjang Cripple Sid-ing, Labrador Park Station, and Telok Blangah Station. By relating the deformation behaviour to local ground and excavation support conditions, some general observations about wall and ground deformations induced by multi-propped excavations can be made. Such as the influence of ground and excavation support stiffness on wall deflection, and the influence of consolidation on ground settlements. The database of ground and wall deformations can thus be com-piled into various charts so that these can become useful references for engineers undertaking future ex-cavation designs in Singapore.


257

Figure 2 plots the maximum wall deflection and the maximum ground settlement during the construction of WCT station. The maximum wall deflection (w) ranged from 0.11% to 0.63% times of excavation depth (H) at the end of backfilling, whilst the maximum surface settlement (Sv) ranged from 0.06% to 0.49% times of excavation depth. Furthermore, it is observed that the maximum ground settlement is slightly lower than the maximum wall deflection. Specifically, the ratio of maximum settlement to max-imum wall deflection ranged mostly between 0.5 to 1.

Figure 1.Layout of WCT station.

Max wall deflection vs excavation depth

0

50

100

150

200

0 5 10 15 20 25

Depth of excavation, H (m)

Ma

x w

all d

efl

ec

tio

n,

w (

mm

) Excavate to FEL

Strut removal &

backfill completed

w / H = 1%

w / H = 0.5%

w / H = 0.25%

Max settlement vs excavation depth

0

50

100

150

200

0 5 10 15 20 25


Ma

x s

ett

lem

en

t, S

v (

mm

)

Excavate to FEL

Strut removal &

backfill completed

Sv / H = 1%

Sv / H = 0.5%

Sv / H = 0.25%

Normalised Settlement vs Deflection

0.0%

0.2%

0.4%

0.6%

0.8%

1.0%

0.0% 0.2% 0.4% 0.6% 0.8% 1.0%

Normalised max wall deflection w / H

No

rma

l. m

ax

se

ttle

me

nt

Sv

/ H Excavate to FEL

Strut removal &

backfill completed

Sv / w = 1

Sv / w = 0.5

Figure 2. Maximum wall deflection and ground settlements at WCT station.

2.2 Pasir Panjang (PPJ) Station The soil condition at PPJ station consisted of 3m of fill material from the ground surface, followed by a 17m thick Kallang Formation on the eastern half of the station. This was underlain by the various grades of Jurong Formation, consisting of soils (SV-SVI) to highly and moderately weathered rocks (SIII-SIV). The ERSS for PPJ station was described by Chua et al (2008). Due to the proximity to the existing piers of the Pasir Panjang semi-expressway, a rigid earth retaining system was designed for the construction of PPJ station. This comprised of 1m thick diaphragm wall with six levels of struts, and supplemented with 1m thick cross-walls that are 10m deep and perpendicular to the main diaphragm wall at the pier locations. The toes of the diaphragm walls ranged from 25.5m to 30m deep. Figure 3 shows the geological conditions and ERSS scheme at PPJ station. The maximum excavation depth was

Jurong Formation (soils)

Jurong Formation (rocks)

1m contiguous bored pile walls


258

about 20m. Figure 4 plots the maximum wall deflection and the maximum ground settlement during the construction of PPJ station. Due to the very rigid ERSS, the wall deflections are low and the maximum wall deflec-tions are less than 0.2% times of the excavation depth. However, due to the thick compressible layer from the Kallang Formation, there is some consolidation during the excavation and the maximum sur-face settlements went up to 0.39% of the excavation depth. The effect of this consolidation is to cause settlement of the ground to go beyond two times that of the wall deflection where there is thick Kallang Formation – this is in contrast to the excavation response at WCT station in the absence of compressible clays. Another observation of Figure 4 is that whilst the maximum wall deflection remained nearly the same between the final excavation stage and the end of backfilling stage, the maximum surface settle-ment increased substantially. This is because the wall is rigid enough so that there is not much addition-al deflection during backfill stage, but the consolidation was still ongoing and this result in increased surface settlement during backfilling stage.

Figure 3.Geological conditions & layout of ERSS at PPJ station.


0

50

100

150

200

0 5 10 15 20 25


Ma

x w

all d

efl

ec

tio

n,

w (

mm

) Excavate to FEL

Strut removal &

backfill completed

w / H = 1%

w / H = 0.5%

w / H = 0.25%


0

50

100

150

200

0 5 10 15 20 25


Ma

x s

ett

lem

en

t, S

v (

mm

)

Excavate to FEL

Strut removal &

backfill completed

Sv / H = 1%

Sv / H = 0.5%

Sv / H = 0.25%


0.0%

0.1%

0.2%

0.3%

0.4%

0.5%

0.0% 0.1% 0.2% 0.3% 0.4% 0.5%


No

rma

l. m

ax

se

ttle

me

nt

Sv

/ H

Excavate to FEL

Strut removal &

backfill completed

Sv / w = 1

Sv / w = 0.5

Sv / w = 2

Figure 4. Maximum wall deflection and ground settlements at PPJ station.

Kallang Formation Jurong Formation (soils)



259

2.3 Pasir Panjang Cripple Siding (PCS) The soil condition at PCS comprised of an overlying Kallang Formation of between 9m to 13m thick and with varying thickness of marine and estuarine clays, above the soils and rocks of Jurong Forma-tion. Figure 5 shows the geological conditions and the layout of the ERSS for the cut-and-cover tunnel section at PCS. The excavation at PCS was about 17m to 20m deep, and was supported using 0.8m and 1m thick diaphragm walls with six levels of struts. The lengths of the diaphragm walls ranged from 21m to 29m deep.

Figure 6 plots the maximum wall deflection and the maximum ground settlement during the construction

of PCS. The normalized maximum wall deflections (w/H) ranged from 0.02% to 0.21% at the final ex-

cavation level, and this increased to between 0.11% and 0.35% after completing the strut removal and

backfill stages. The normalized maximum ground settlement (Sv/H) was even higher and ranged be-

tween 0.27% and 1.09% at the final excavation level, and increased to between 0.52% and 1.95% after

the strut removal and backfilling stage. Due to the consistently thick compressible Kallang Formation

along the PCS excavation, there is a significant component of consolidation-induced settlement. The

maximum ground settlement is more than two times the maximum wall deflection monitored along the

PCS excavation (i.e. Sv/w > 2).

Figure 5. Geological conditions & layout of ERSS at PCS.


0

50

100

150

200

0 5 10 15 20 25


Ma

x w

all d

efl

ec

tio

n,

w (

mm

) Excavate to FEL

Strut removal &

backfill completed

w / H = 1%

w / H = 0.5%

w / H = 0.25%


0

50

100

150

200

250

300

350

400

0 5 10 15 20 25


Ma

x s

ett

lem

en

t, S

v (

mm

)

Excavate to FEL

Strut removal &

backfill completed

Sv / H = 1%

Sv / H = 0.5%

Sv / H = 0.25%


0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

0.0% 0.2% 0.4% 0.6% 0.8% 1.0%


No

rma

l. m

ax

se

ttle

me

nt

Sv

/ H Excavate to FEL

Strut removal &

backfill completed

Sv / w = 0.5

Sv / w = 1

Sv / w = 2

Figure 6. Maximum wall deflection and ground settlements at PCS.

2.4 Labrador Park (LBD) Station The maximum excavation depth at LBD station is about 21m. Figure 7 shows the geological conditions and the layout of the ERSS for the cut-and-cover tunnel section at LBD station. The soil condition at LBD station comprised predominantly of Jurong Formation, except at the west end of the station where there is a localized valley of Kallang Formation (up to 10m thick) above the Jurong Formation. Due to its proximity to the existing buildings and the piers of the Pasir Panjang semi-expressway, a rigid ERSS consisting of 0.8m and 1m thick diaphragm walls with toes ranging from 20m to 26.5m deep was used to support the excavation. The excavation was constructed using 6 levels of struts. Figure 8 plots the maximum wall deflection and the maximum ground settlement during the construction of LBD station. There was not much difference in the wall deflections between the final excavation

0.8m-1.0m thick d-walls



260

stage and the end of backfilling, and the maximum normalized wall deflection (w/H) was 0.32%. The maximum normalized surface settlements (Sv/H) were greater than 0.5% at the western end of the sta-tion coinciding with the valley of Kallang Formation, but elsewhere the maximum normalized surface settlements were less than 0.34%. This is consistent with earlier observations about the maximum sur-face settlements being substantially higher than maximum wall deflections in the presence of compressi-ble clays during excavations.

Figure 7. Geological conditions & layout of ERSS at LBD station.


0

50

100

150

200

0 5 10 15 20 25


Ma

x w

all d

efl

ec

tio

n,

w (

mm

) Excavate to FEL

Strut removal &

backfill completed

w / H = 1%

w / H = 0.5%

w / H = 0.25%


0

50

100

150

200

0 5 10 15 20 25


Ma

x s

ett

lem

en

t, S

v (

mm

)

Sv / H = 1%

Sv / H = 0.5%

Sv / H = 0.25%


0.0%

0.4%

0.8%

1.2%

1.6%

0.0% 0.2% 0.4% 0.6% 0.8%


No

rma

l. m

ax

se

ttle

me

nt

Sv

/ H

Sv / w = 1

Sv / w = 0.5

Sv / w = 2

Figure 8. Maximum wall deflection and ground settlements at LBD station.

2.5 Telok Blangah (TLB) Station Figure 9 shows the geological conditions and the layout of the ERSS for the cut-and-cover tunnel sec-tion at TLB station. The soil stratigraphy comprises of a thin layer of Kallang Formation (not more than 5m) overlying the Jurong Formation. Due to the excavation being shallower and further away from ex-isting buildings, the ERSS at TLB station was less rigid and comprised mainly of soldier pile wall at 1.6m spacing with sheetpile lagging above and shotcrete lagging below 12m, and 1m contiguous bored pile walls at both ends of the station. The excavation depth was around 16m to 18m and this was sup-ported using four levels of struts. The details of the design and construction at TLB station are de-scribed in Ng et al (2010). Figure 10 plots the maximum wall deflection and the maximum ground settlement during the construc-tion of TLB station. The normalized maximum wall deflections (w/H) at the final excavation level were less than 0.19%, and the maximum w/H increased slightly to 0.40% after the strut removal and backfill stages. Considering that the soldier pile wall would be fairly flexible, the wall deflections are



0.8m-1m thick d-walls


261

not high. This may be attributed to the stiff ground at this location. The normalized maximum ground settlements (Sv/H) were less than 0.45% at the final excavation stage, increasing slightly to 0.68% after the strut removal and backfilling stage. The ratio of settlements to wall deflections (Sv/w) ranged from 1 to 2. There is some influence of consolidation on ground settlements although the Kallang Formation is considerably thinner at this location – this would be due to the higher wall permeability in the ERSS compared to other locations.

Figure 9. Geological conditions & layout of ERSS at TLB station.


0

50

100

150

200

0 5 10 15 20 25


Ma

x w

all d

efl

ec

tio

n,

w (

mm

) Excavate to FEL

Strut removal &

backfill completed

w / H = 1%

w / H = 0.5%

w / H = 0.25%


0

50

100

150

200

0 5 10 15 20 25


Ma

x s

ett

lem

en

t, S

v (

mm

)

Excavate to FEL

Strut removal &

backfill completed

Sv / H = 1%

Sv / H = 0.5%

Sv / H = 0.25%


0.0%

0.2%

0.4%

0.6%

0.8%

1.0%

0.0% 0.2% 0.4% 0.6% 0.8% 1.0%


No

rma

l. m

ax

se

ttle

me

nt

Sv

/ H Excavate to FEL

Strut removal &

backfill completed

Sv / w = 0.5

Sv / w = 1

Sv / w = 2

Figure 10. Maximum wall deflection and ground settlements at TLB station.

2.6 Discussion Generally, the monitored wall deflections at all the five locations were less than 0.5% times of the exca-vation depth – this is consistent with the study by Clough & O’Rourke (1990) on excavation case histo-ries where they showed that the maximum wall deflection for excavations in stiff clays were less than 0.5% times of the excavation depth. Table 1 compiles the average maximum wall deflection and maxi-mum ground settlement at the various sites. It is interesting to note that the wall deflections for the excavations at WCT were higher than those at TLB, even though a more rigid contiguous bored pile wall was used compared to the soldier-pile wall. This was similar to the observation made by Clough & O’Rourke for excavations in stiff clays, where they found no significant influence between trends of maximum wall movements for various types of walls and suggested that soil stiffness could have a stronger influence on wall behaviour than wall stiff-ness. Nevertheless as the excavation at PPJ station has illustrated, it is possible to install a very rigid

Soldier pile walls

CBP walls

CBP walls

Soldier pile walls

Jurong Formation (soils)

Kallang Formation



262

ERSS by using cross-walls to restrict the wall deflection to less than 0.2%. This method would be more effective than increasing the thickness of the main diaphragm wall. Another observation is the influence of consolidation on surface settlements. In the absence of compress-ible clays, such as the excavation at WCT station, the ground settlement is generally lower than the wall deflection. This had been investigated by Mana & Clough (1981) using field case histories and finite element studies, where they proposed to estimate the maximum ground settlement (Sv) from the wall deflection (w) using the relationship Sv / w = 0.6 – 1.0. The ratio of settlement to deflection (Sv / w) observed at the locations where Kallang Formation was absent, was generally found to be within the bounds proposed by Mana & Clough. However where there is compressible clay, the effect of consoli-dation-induced settlements would be quite significant. As shown in the earlier figures, the maximum set-tlement in such cases was easily greater than twice the maximum wall deflection. Table 1. Mean wall deflections and ground settlements at end of backfilling stage.

WCT station PPJ station PCS LBD station TLB station

ERSS Contiguous bored pile

wall

Diaphragm wall

Diaphragm wall

Diaphragm wall

Soldier-pile wall

Geological conditions Jurong For-

mation 17m Kallang above Jurong

13m Kallang above Jurong



Normalized wall def-lection (w / H)

0.38% 0.09% 0.21% 0.12% 0.18%

Normalised ground settlement (Sv / H)

0.28% 0.21% 1.10% 0.29% 0.35%

Settlement to deflec-tion ratio (Sv / w)

0.75 2.6 5.7 1.9 2.1

3 SURFACE SETTLEMENT PROFILES AND HORIZONTAL GROUND DISPLACEMENTS

3.1 Profiles of excavation-induced settlements Any study on excavation-induced settlement profiles would have to take into account whether consolida-tion effects are important or not. From the ground conditions and maximum settlement trends, the ef-fects of consolidation were not evident at WCT station but were most pervasive at PCS tunnel excava-tion. The surface settlement profiles at these two locations would be presented. Figure 11 plots the final surface settlement normalized by the maximum settlement (Sv / Sv,max), against the distance from excavation normalized by the excavation depth (i.e. distance D / excavation depth H), at WCT station. The classical settlement envelopes by Clough & O’Rourke (1990) for soft to medium clays and for stiff to very hard clays – which were based on case history data – are also plotted for comparison. When consolidation effects are absent, Clough & O’Rourke’s envelope gave an upper bound to the response of surface settlement during excavation. Figure 12 plots the normalized final surface settlement trough at the end of PCS backfilling. It was ob-served that consolidation effects had caused surface settlement that is above the settlement envelope by Clough and O’Rourke for soft to medium clays, and also increased the extent of the settlement trough to more than two times the excavation depth. The effect of consolidation is well observed in case histories (Hulme et al. 1989, Nicholson 1987, etc.) but no attempt has been made to quantify this so far. The rea-son is because consolidation effects are dependent on several factors, such as thickness and extent of compressible clays from excavation, degree and duration of dewatering, wall permeability, etc. More re-search would need to be carried out to investigate the few mechanisms that have been proposed to ex-plain consolidation, including underdrainage-induced consolidation (Wen & Lin, 2002) and wall lea-kage-induced consolidation (Halim & Wong, 2006). Nevertheless, the settlement profiles in Figure 12 does illustrate that due considerations should be given to consolidation effects, particularly if these are to be used for assessing impact to adjacent structures and buildings.


263

Surface settlement profile at WCT station

0.00

0.20

0.40

0.60

0.80

1.00

0 0.5 1 1.5 2 2.5 3 3.5 4

Normalised distance from excavation D / HN

orm

alis

ed s

ettl

emen

t S

v /

Sv,

max

Array C1

Array B1

Array C5

Array C3

Array C6

Clough & O'Rourke (1990) Soft clay

Clough & O'Rourke (1990) Stiff clay Figure 11. Normalised settlement profile at WCT station.

Surface settlement profile at PCS

0.00

0.20

0.40

0.60

0.80

1.00

0 0.5 1 1.5 2 2.5 3 3.5 4

Normalised distance from excavation D / H

No

rmal

ised

set

tlem

ent

Sv

/ Sv,

max

Array B3

Array B2

Array C3

Array C1

Clough & O'Rourke (1990)

Figure 12. Normalised settlement profile at PCS.

3.2 Horizontal displacements induced during excavations Another effect of excavation which is not well-understood is the horizontal ground displacements in-duced during an excavation. In contrast to wall deflection and surface settlements which are well-reported in case studies, horizontal ground displacements were less frequently monitored during excava-tions. Nevertheless, as engineers become more aware of the influence of horizontal displacement on ad-jacent structures, the importance of monitoring horizontal displacements has caught up. For example, at various excavation sites in CCL5, horizontal displacements were monitored using soil inclinometers. Specifically, the horizontal displacements at WCT station and TLB station will be presented in this sec-tion, as the soil inclinometers at these locations were generally further away from any pre-existing pile foundations and hence more reflective of excavation-induced ground displacements. Figure 13 shows the horizontal displacements of the ground behind the CBP wall at WCT station for several sections, plotted with the corresponding wall deflections. The wall deflections are highest near to the final excavation levels, resulting in deflection profiles that have a deep bulge at the centre of the wall length. This causes a deep inward movement of the soil nearest to the wall, and as it gets further away, the maximum lateral movement of the soil moves up to the surface. The horizontal displacement of the soil was still being observed, even at a distance of 20m~27m away (which is about 1.5 times the exca-vation depth). It is also noted that there is a rapid decrease in the magnitude of the horizontal displace-ments, with the maximum value reducing to less than 50% of the wall deflection at a distance of 10m~12m away from the wall (about 0.5 times the excavation depth).


264

North wall (Array C3)

70

75

80

85

90

95

100

105

-0.02 0 0.02 0.04 0.06 0.08 0.1

Horizontal displacement (m)R

ed

uced

level (m

)

Deflection of wall

Soil displacement at

x = 10m behind wall


x = 27m behind wall

South wall (Array C1)70

75

80

85

90

95

100

105

-0.02 0 0.02 0.04 0.06 0.08 0.1

Horizontal displacement (m)

Red

uced

level (m

)

Deflection of wall


x = 10m behind wall


x = 20m behind wall

South wall (Array B1)70

75

80

85

90

95

100

105

-0.02 0 0.02 0.04 0.06 0.08 0.1


Red

uced

level (m

) Deflection of wall


x = 12m behind wall


x = 22m behind wall

Figure 13. Horizontal displacements at WCT station.

South wall (Array C1)

75

80

85

90

95

100

105

-0.01 0 0.01 0.02 0.03 0.04 0.05


Red

uced

level (m

)

Deflection of wall


x = 13m behind wall


x = 25m behind wall

North wall (Array A2)

75

80

85

90

95

100

105

-0.02 0 0.02 0.04 0.06 0.08


Red

uced

level (m

)

Deflection of wall


x = 6m behind wall


x = 12m behind wall

Figure 14. Horizontal displacements at TLB station.

Figure 15. Two types of horizontal displacement profiles.

(b) Cantilever-shaped wall deflection (a) Deep-inward movements of wall


265

Figure 14 shows the horizontal displacements of the ground behind the soldier pile wall at TLB station, where the final wall deflection profiles were cantilever-shaped. In this case, the lateral displacements of the soil further behind the wall were also following the same cantilever-shaped, but with a reduced mag-nitude. At about 12m~13m away from the wall (approximately 0.75 times the excavation depth), the soil displacements had reduced to less than 50% of the soil displacements at the wall. Figure 15 illu-strates the differences of horizontal displacement profiles arising from the deep-inward wall movement type seen at WCT station and the cantilever-shaped wall profile seen at TLB station. More studies are needed to understand the horizontal displacements of the ground (in terms of its extent, magnitude, etc.) behind an excavation.

4 CONCLUSION

The instrumentation monitoring data from several cut-and-cover locations of CCL5 were analysed and reported in this paper. The first part of the paper presents the maximum wall deflection and maximum surface settlement at all the sites, and discusses the monitored results in relation to the studies by Mana & Clough (1981) and Clough & O’Rourke (1990). A key deviation of the CCL5 study from these clas-sical studies is the influence of consolidation on surface settlements. It was found that excavation-induced settlements were easily greater than two times the maximum wall deflection under the influence of consolidation. Except at WCT station, the maximum wall deflections monitored were less than 0.5% of the excavation depth. The tight control on wall deflection arises from the proximity of the excavation to the existing viaduct piers of the Pasir Panjang semi-expressway as well as the existing buildings be-side the deep excavations. The surface settlement profiles and the horizontal displacement profiles caused by the excavations were also studied. When consolidation effects are absent, the settlement envelope by Clough & O’Rourke gives an upper bound to the settlement profile. When consolidation effects are present, the settlement profile could exceed the maximum envelope but the effects would depend on the extent and degree of excavation-induced consolidation. Horizontal displacements of the ground were also observed up to a distance of 1.5 times the excavation depth from the excavation at WCT and TLB stations. Furthermore, two types of horizontal displacement profiles were observed and this depends on whether the wall def-lection profile is of cantilever-shaped or of a deep-inward movement type. More studies would be needed to improve our understanding of the consolidation mechanism and the horizontal displacement behaviour of the ground.

REFERENCES

Chua, T.S., Lew, M., Rama, V. & Phua, H.L. 2008. Robust earth retaining wall system to control ground movement and minimize impact on adjacent structures. Proc. International Conference on Deep Excava-tions 2008, 10-12 Nov 2008, Singapore.

Clough, G.W. & O’Rourke, T.D. 1990. Construction induced movements of in-situ walls. ASCE Geotechnical special publication No.25 – Design and Performance of Earth Retaining structures, pp.439–470.

Halim, D. & Wong, K.S. 2006. Post-Excavation Settlement in Deep Excavation, Proc. Of International Confe-rence on Deep Excavations, 28-30 June 2006, Singapore.

Hulme, T.W., Potter, L.A.C. and Shirlaw, J.N. 1989. Singapore MRT System: Construction. Proc. Inst Civil Engineers, Vol.86, pp.709–770.

Mana, A.I. & Clough, A.M. 1981. Prediction of movements for braced cuts in clay. ASCE Journal of Geotech-nical Engineering Division, Vol.107(6), pp.759–777.

Nicholson, D.P. 1987. The design and performance of the retaining walls at Newton Station. Proc. conf. Singa-pore Mass Rapid Transit Conference, Singapore, 6-9 April 1987, pp.147–154.

Ng, C.Y., Veeresh, C., & Rama, V. 2010. Design and construction of Circle Line Telok Blangah station. Proc. of World Urban Transit Conference, 20-22 Oct 2010, Singapore.

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, Proc. 3rd intern. symp. on Geotechnical Aspects of Under-ground Construction in Soft Ground, Toulouse, 23-25 October 2002, pp.447–452.


266

INTRODUCTION LTA Contract C907 consists of excavation for cut and cover tunnels and excavation for CST 3A (Common Services Tunnel). Earth retaining system for CST 3A consists of 1.1m diameter Contiguous Bored Pile (CBP) wall with 5 levels of strutting with horizontal spacing of 5m. Excavation for under-ground CST 3A tunnel is 12m wide and 16m deep. Earth retaining system for cut and cover tunnels consists of 1m thick diaphragm wall with 4 levels of strutting with horizontal spacing of 6m. Excavation for cut and cover tunnels is 14m wide and 15m deep in section B7. Strain Gauges (SG) and Load Cells (LC) are installed on struts to monitor strut loads.

1 MEASUREMENTS OF STRUT LOAD IN CST 3A EXACAVATION

1.1 Instrumentation in CST 3A In CST 3A, loads in struts were monitored using vibrating wire strain gauges and load cells. Strut CP1 has 5 levels (S1 to S5) as shown in strut layout and photograph on next pages. Strut CP1, from level 1 to level 5, was instrumented. Similarly, strut CP5, from level 1 to level 5, was instru-mented. Strut CD3, from level 1 to level 3, was instrumented. Similarly, strut CP13 was also monitored. Further details of instrumented struts are shown in Table 1. All strut loads were monitored real time. Contiguous Bored Pile (CBP) wall movements were monitored by in wall inclinometers but results of these inclinometers are not presented here.

Field measurements of strut loads in LTA Contract C907

A.S. Jadhav GeoEng Consultants, Singapore

ABSTRACT: This paper describes measurements of strut loads, including temperature loads, in LTA (Land Transport Authority, Singapore) Contract C907. Strain Gauges & Load Cells are used to moni-tor loads in installed struts. Observed data indicates that strut forces are maximum during excavation of bottom most level. Upon reaching formation level, lower level struts showed substantial increase in load. Struts are removed after casting of base slabs and roof slabs and effect of strut removal on other remaining struts is also studied through data available from instrumented struts. For CST3A and Cut and Cover excavation in Contract C907, measured strut loads were smaller than those predicted in FE analysis.


267

1.2 Strut layout and sectional details of CST 3A Excavation

Figure 1. Strut layout of CST 3A

Table 1. Details of instrumented struts in CST 3A


268

Figure 2. Photograph of struts in CST3A

1.3 Variation of CP1 – S4 strut load with excavation Strut load in strut CP1-S4 (level 4) breached Alert Level of 1847 kN in September 09. Load-time history of this strut was studied and when related to excavation activities, it was concluded that excavation to formation level (RL87.5) might have caused strut to pick up additional load. Load-time history plot of Strut CP1-S4 is shown in Figure 3 below.

Figure 3. Variation of strut load with excavation (Strut CP1-S4)

1.4 Comparison of predicted strut load and measured strut load (Strut CP1 - S1 to S5)

(For excavation till formation level)

Comparison of predicted strut load and measured strut load, for strut CP1, is shown in Figure 4 below. Comparison shows that for struts S2 and S3 strut loads are over predicted. For struts S5 there is a match between predicted load and measured load. For strut S4, predicted load decreased gradually for excavation below strut S5 and till formation level, while measured load for S4 showed increment. Lower level struts (S4 & S5) carried about 50% of total measured load in all 5 level struts while measured strut loads in upper level struts remained approximately constant. Generally, for excavation until formation level, measured strut loads were less than predicted loads.


269

Figure 4. Comparison of predicted strut load and measured strut load for strut CP 1 - S1 to S5

1.5 Effect of temperature on strut load Following plot shows fluctuation of strut load due to temperature on a particular day. Monitored strut is CP5 at Level 1 (topmost strut). Strut load starts to increase in the morning and reaches peak in the afternoon. The plot is compared with temperature measurements of that strut on that same day. Minimum temperature is 24°C while average maximum is 36°C at 1200hrs.

One more example of fluctuation of strut load due to temperature is presented below. Contract C907 has diverted drain where sheet piles are used to support the excavation. Monitored strut is DP9 at Level


270

1 (topmost strut). Variation of strut load is compared with temperature measurements and both plots are shown below

From all instrumented struts it was observed that average increment in strut load due to temperature was about 400kN. Strut forces due to temperature change are given by equation 1.

αΔtAE (1)

Where α = Thermal coefficient of expansion for steel; Δt = Change in strut temperature (°C); A = Cross sectional area of strut; E = Young’s Modulus of steel


271

In C907, theoretical strut load due to temperature variation (of 10°C) is about 600kN for single struts and about 1000kN for double (compound) struts. Where as strut load increment due to temperature is found to be about 400kN, which is less than computed theoretical increase in strut load due to tempera-ture. 1.6 Effect of strut removal on remaining struts in the system (load distribution to upper level struts) In CST 3A excavation, struts were removed from Level 5 to Level 1 as per sequence mentioned in TERS analysis. (Generally, strut removal was after casting of base slab & after casting of roof slab etc). It was observed that after removal of particular level strut, upper level struts picked up load. Load distribution to upper level struts, after removal of bottom level struts, is presented with monitored re-sults below. The struts studied for this effect are CP1 and CP5.


272

Table 2. Effect of strut removal on strutting system in CST 3A

Above data, for instrumented struts, indicates that upon removal of particular level strut some of the strut load was distributed to strut immediately above. Generally, it was observed that about 40% of load carried by the strut to be removed was redistributed to upper level struts.

2 MEASUREMENTS OF STRUT LOAD IN CUT AND COVER EXCAVATION (SECTION B7)

2.1 Instrumentation in Cut and Cover Excavation (Section B7) In section B7 of Cut and Cover tunnel, loads in struts were monitored using vibrating wire strain gauges and load cells. Strut EP2 has 5 levels (S1 to S5) as shown in strut layout and longitudinal section on next page. Strut EP2, from level 1 to level 4, was instrumented. Similarly, strut EP7 and EP14 (EP14 is not shown in section) were also instrumented from level 1 to level 4. All strut loads were monitored real time. Diaphragm wall movements were monitored by in wall inclinometers but results of these inclinometers are not presented here.


273

2.2 Strut layout and sectional details of Cut and Cover Excavation (Section B7)

Figure 5. Strut layout and sectional details of section B7 of cut and cover tunnel


274

Figure 6. Partial longitudinal sectional view of section B7 of cut and cover tunnel

2.3 Comparison of predicted strut load and measured strut load (Strut EP7- S1 to S4)

(For excavation till formation level)

Figure 7. Comparison of predicted strut load and measured strut load (Strut EP 7 - S1 to S4)

Comparison of predicted strut load and measured strut load, for strut EP7, is shown in Figure 7 above. Comparison shows that for struts S2, S3 and S4 strut loads are over predicted. Generally, for excavation until formation level, measured strut loads were less than predicted loads.


275

2.4 Effect of strut removal on remaining struts in the system (load distribution to upper level struts)

Table 3. Effect of strut removal on strutting system in section B7 of cut and cover tunnel

Above data, for instrumented struts, indicates that some of the strut load was distributed to strut imme-diately above. Generally, it was observed that about 50% of load carried by the strut to be removed was redistributed to upper level struts. It is obvious that some of the balance load might have been distri-buted to base slab and roof slab.

3 CONCLUSION

Observed increase in strut load due to temperature was about half that of theoretical increase in load due to temperature. Observation of data from instrumented struts indicated that strut forces increased during final level excavation to formation level. Upon reaching formation level, lower level struts showed substantial increase in loads. Generally, it was observed that about 40% of load carried by the strut to be removed was redistributed to upper level struts. For CST3A and Cut and Cover excavation measured strut loads were smaller than those predicted in FE analysis.


276

1 INTRODUCTION With the increasing amount of construction within the urban environment, the public’s awareness of construction projects and the advent of viral media any building damage can rapidly become headline news providing negative public relations to the project and potential stop work scenarios. Therefore un-derstanding and controlling building damage plays a key role in the success of any major urban project. Assessing damage, recognizing potential areas of concern and putting measures in place to control movements and therefore controlling damage to acceptable limits is crucial. There are a number of high profile cases where building damage has occurred with significant cost and programme impact to the project (Shirlaw et al (2003), Osborne et al (2005)). An all encompassing and logical approach is there-fore required to identify all potential sources of movement, correctly model their manifestation and pre-dict the structural response to the movements, with such an approach building damage can be unders-tood and controlled so that it does not adversely impact the construction project. 1.1 Excavation induced settlement Excavations cause the loss of support to the surrounding ground. Retaining systems are used to support the excavation walls but these will undergo movement which is transmitted to the ground and leads to settlement. The degree and extent of the ground movement depends on the dimensions of the excavation, the retaining system used, the geological profile and groundwater control. Settlement is usually pre-dicted using empirical methods as numerical modeling tends to under predict near field movements and over predict far field movements unless complex constitutive models are used. Settlement predictions are fundamental in the analysis of building damage and in recent years it has been found that the usual models used are too conservative, leading to uneconomic and inefficient practices both during the design and construction phases of projects involving underground works. 1.2 Spandrel curves The first published curves were produced by Peck (1969) and were based on field data collected from many different excavation projects (Figure 1). He generalized the data into a set of generic curves de-pendent on the ground conditions and support type. The curves did not have a mathematical formula to

Settlement assessment – The key components in identifying building damage

J. McCallum, N. Osborne & B. Vontivillu Mott MacDonald Ltd, Singapore

ABSTRACT: Ground settlement curves perpendicular to excavations given in literature typically follow a “spandrel” shape. This curve can be mimicked with the hogging limb of an inverted Gaussian curve. Modern numerical analysis and field data suggest that strutting and frictional effects create a “concave” profile. Both curves are presented in this paper and resolved into a single equation, followed by a short discussion of the impact of settlement on piled structures. By correctly identifying buildings likely to suffer damage appropriate design solutions can be used to ensure damage does not occur, contributing to a successful construction project.


277

describe them but they followed the classical spandrel curve shape with a maximum settlement imme-diately adjacent to the wall and decreasing with distances forming an asymptote with zero. There was no prediction for the horizontal movement of the ground.

Figure 1. Spandrel curves suggested by Peck Slight variations on these curves have been produced by various authors. Bowles (1996) replicates the curves with a parabola; Bowers & New (1994) corroborate this approach with field data from the Heathrow trial shaft. The differences between the curves are marginal but they can affect the building damage assessment due to the difference in curvature, particularly in the near field region. Tunnelling induce settlement is generally recognized to conform to the inverted Gaussian curve. Alterna-tives such as Loganathan & Poulos (1998) (which includes depth calculations) and the ribbon sink mod-el of Bowers & New (1994) make slight modifications to the shape of the curve but for the majority of purposes the Gaussian curve is sufficiently accurate. By manipulating the equation for settlement and forcing the point of inflection of the curve to coincide with the boundary of the excavation it is possible to mimic the curves of Peck with the hogging limb of the Gaussian curve (Figure 2):

22

12

1

2

1

max

W

y

v eyS (1)

where S is the vertical displacement, y is the distance from the wall, vmax is the settlement at the wall (usually calibrated to the maximum wall deflection and expressed as a ratio to the excavation depth, Z) and W is the trough width of the settlement curve (often related to the depth of excavation, Z). This derivation was first used by Mott MacDonald on the Jubilee Line Extension project in London as part of the advanced works for the Parliamentary Bid.


278

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

y/Z

s/s

max

Figure 2. Spandrel curve based on the hogging limb of the inverted Gaussian curve (from equation 1) Spandrel curves are typically applicable to retaining systems that have the greatest deflection near the ground surface and respond in a cantilever mode. The spandrel shape is most applicable in modern con-struction to diaphragm wall installation where the excavation is supported by bentonite slurry. Recent data taken from an MRT project in Singapore is plotted against the normalized curve in the Figure 3 and shows a reasonable match after adjusting the influence zone to be equivalent to the installation depth. For this particular project the dvmax/Z ratio was set at 0.065% based on the average of historical records for diaphragm wall installation in Singapore and this represents a good prediction for moderate levels of workmanship, though would need to be increased in the even that a panel is left open for an ex-tended period of time. The spandrel curve is most appropriate to cantilever type retaining structures and diaphragm wall installations for the prediction of settlement.

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.5 1.0 1.5

y/Z

s/s

max

Figure 3. Field data for settlement associated with diaphragm wall installation.


279

1.3 Concave curves With improved monitoring and the use of numerical modeling it became apparent that frequently the area immediately adjacent to an excavation did not suffer the most settlement. The maximum ground movement was some distance from the excavation. The first suggestions of a concave curve were postulated by Clough & O’Rourke (1990) who presented a bi-linear model of settlement that was flat immediately adjacent to the wall and then decreased with distance from the wall (Figure 4). This approach produces no curvature which was noted by Skempton & MacDonald (1956) as being one of the primary causes of building damage. Using the currently ac-cepted staged approach to building damage would split any building straddling the elbow of the curve into a sagging and hogging zone with no apparent deflection ratio as both “curves” are linear. With no deflection ratio there can be no bending or shear strain, therefore only horizontal strains would contri-bute to building damage.

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

y/Z

s/s

max

Figure 4. Clough & O’Rourke (1990) concave curve

Hsieh & Ou (1998) proposed a more complex 3 part linear model which mimics the concave curves seen in the results of numerical models (Figure 5). There model is specifically for retaining walls that experience more lateral movement at depth than at ground surface. This is more representative of mod-ern top down construction where a stiff first strut is often used to prevent excessive lateral movement near the top of the wall. The Hsieh & Ou linear model defines the settlement at the wall as 50% of the maximum. The maximum occurs at a distance equivalent to half the excavation depth from the wall. At twice the excavation depth from the wall the settlement is reduced to 10% of the maximum, and by four times the excavation depth settlements are negligible.


280

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

y/Z

s/s

max

Figure 5. Concave linear profile suggested by Hsieh & Ou (1998). A normal distribution curve can be fitted to these points and has the following form:

2

2 2

5.0

max

Zy

Z

v eyS

(2)

where S is the vertical displacement, y is the distance from the wall, vmax is the maximum settlement and Z is the depth of excavation. is the standard deviation for the Gaussian curve used, with a fixed value of 0.699 the maximum settlement and the two further points can be achieved but the condition at the wall is higher than Hsieh & Ou suggest. While this gives a more onerous condition on settlement the gradient of the curve is shallower and hence the bending strains will be less. In order to force the curve to fit the points given by Hsieh & Ou a variable has to be defined and is done so using the cumulative normal distribution, though for ease of calculation this can be replaced with the error function as fol-lows:

42465.02

27435.0

Zy

erf (3)

where y is the distance from the wall and Z is the depth of excavation. This function varies the standard deviation from 0.425 at the wall through to 0.699 at distance. All the conditions of the 3 part linear curve given by Hsieh and Ou are now satisfied (Figure 6).


281

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

y/Z

s/s

max

Figure 6. Concave curve based on a modified inverted Gaussian curve. 2 VERTICAL MOVEMENTS The two models for vertical movements can be combined into a single equation by introducing a term for the ratio of position of maximum wall movement to excavation depth (R). For a spandrel curve the maximum wall movement will occur at the ground surface and therefore R = 0, for a concave curve the maximum wall movement will occur at some depth and R = 1 if this depth coincides with the formation level of the excavation. This is convenient as most retaining wall analysis will give the maximum hori-zontal movement (which can be used to predict the maximum vertical movement of the ground) and the position of that movement on the wall. The single equation takes the form:

2

max

Z

yDCB

A

v eeyS (4)

where S is the vertical displacement, y is the distance from the wall, vmax is the maximum settlement and Z is the depth of excavation. In addition:

RA 15.0 (5)

211

5.0

RB

(6)

RC 5.11 (7)

RRD 12

(8)

where R is the ratio of position of maximum wall movement to excavation depth, is the standard devi-ation used in the normal distribution curve (varying with distance from the wall) and is the ratio of trough width to excavation depth. The last equation locks the extent to that described by Hsieh & Ou when R is equal to 1. Reducing it to 2/ instead will allow the extent to be varied but the position of the maximum settlement will now not conform with the parameters given by Hsieh and Ou. Allowing some


282

variance in the position of the maximum settlement will be beneficial as more field data is gathered that leads to a clearer definition of parameters relevant to the expected site conditions. Field data from various projects around Singapore are compared to the normalized concave curve in the follow graphs (Figures 7, 8 and 9). The excavations are in various geological formations, all of them use strutted temporary support and have relatively low horizontal deflections of the retaining wall. The first graph is for a shaft mainly founded in Old Alluvium, the second in Bukit Timah Granite and the last in reclaimed land which is still consolidating and has a very wide trough width due to the loose na-ture of the soil. Obtaining accurate and reliable field data is always difficult, especially on an active construction site – however the general shapes give a reasonable approximation to the settlement curves presented here. As most cut and cover structures tend to use a stiff propping system particularly at the top of the support wall it would be appropriate to use the concave curve for the prediction of settlement for these situations.

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

y/Z

s/s

max

Figure 7. Field data for settlement associated with excavation in Old Alluvium.

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

y/Z

s/s

max

Figure 8. Field data for settlement associated with excavation in Bukit Timah Granite.


283

0.0

0.2

0.4

0.6

0.8

1.0

0.0 1.0 2.0 3.0 4.0 5.0 6.0

y/Z

s/s

max

Figure 9. Field data for settlement associated with excavation in reclaimed land. 3 HORIZONTAL MOVEMENTS Horizontal strains tend to govern the equations for total bending and shear strains used in the standard staged approach to building damage. While recent work by Potts & Addenbrooke (1997) and Mair & Goh (2008) have demonstrated that horizontal strains can be greatly reduced when consideration of the structural stiffness of a building foundation is taken into account, for the first stage of the assessment a valid model for horizontal movements needs to be adopted. Horizontal ground movements are usually monitored with inclinometers, but often the upper portion of these instruments can be damaged due to site traffic, therefore attaining accurate readings is very diffi-cult. The formulation of the spandrel curve presented here can be further adjusted to give a horizontal deflec-tion curve by considering the vector of movement toward the imaginary tunnel centre that has been used to generate the vertical curve. This gives the maximum horizontal movement at the top of the wall (which seems logical for a cantilever retaining system) and reduces with distance (Figure 10):

ySW

yyH .21

(9)

where H is the horizontal ground movement, y is the distance from the wall, W is trough width of the settlement curve and is the ratio of maximum horizontal to vertical movement of the ground.


284

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

y/Z

h/h

max

Figure 10. Horizontal movement related to a spandrel settlement curve. Hsieh & Ou (1998) provided no equivalent curve for their vertical model. If a similar approach of vec-tor movement is considered for the adopted curve presented here, then the centre of the imaginary tunnel is positioned on the retained earth side of the wall. This would mean that between the wall and the loca-tion of maximum vertical settlement (over the tunnel centerline at y = 0.5Z) the horizontal movements would be away from the wall. This is completely counter intuitive and does not support what little field data there is available. A more realistic approach would be to direct the vector to the location of the maximum wall movement. For the concave curve this is at some depth but for the sake of argument the formation level of the exca-vation will be presented here. Making use of the ratio R (position of maximum wall movement to excavation depth) again the equa-tions for spandrel and concave equivalent horizontal movements can be combined into a single equation of the form:

ySZ

yDFEyH

(10)

where H is the horizontal ground movement, y is the distance from the wall, Z is the depth of excavation and S is the vertical ground movement (Figure 11). In addition:

RRE 1 (11)

RF 1 (12)

RRD 12

(13)

where is the ratio of maximum horizontal to vertical ground movement, R is the ratio of the position of maximum wall movement to excavation depth and is the ratio of trough width to excavation depth.


285

0.0

0.2

0.4

0.6

0.8

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

y/Z

h/h

max

Figure 11. Horizontal movement related to a concave settlement curve.

4 GROUND MOVEMENTS AND PILED STRUCTURES The classical limiting-strain approach to building damage assessments, Mair et al (1996), is based on masonry structures. Allowance is often made for the depth of the foundation but there is no way to ac-count for piled foundation directly within the theory. One method around this is to de-couple the vertical and horizontal movements. The vertical movements induced at the toe of the piles are calculated using Greenfield displacement at depth while the horizontal movements at the pile cap are used. Often this leads to the situation where no vertical movement is pre-dicted as either the pile toe is below the excavation depth or are sufficient distance from the excavation that they do not fall within the influence zone of the excavation (comparable to the active wedge). Hori-zontal movements are greatly reduced once the stiffness of the pile cap and ground beams is taken into account, and so the building damage assessment returns a negligible damage category. However, while the building may not suffer damage queries are often raised about the impact on the piles themselves and whether or not they will be compromised either structurally or geotechnically by the ground movements and in particular negative skin friction effects. 4.1 Geotechnical capacity To fail a pile geotechnically the pile must move more than the surrounding ground. Piles are used to transmit loads to a deeper more competent geology as founding the structure on a shallow foundation would leave to bearing capacity failure and excessive settlement. If the pile settles excessively then it is not performing the task it was designed for and has failed. The geotechnical capacity of the pile is generated through skin friction along the shaft of the pile and end bearing at the toe. Skin friction is generated when the pile is pushed through a surrounding soil that is moving less than the pile. Should the surround soil start to move the skin friction will be reduced and if the soil moves more than the pile itself then negative skin friction can be developed leading to addition drag forces on the pile. Therefore negative skin friction by definition can not cause geotechnical failure of a pile.


286

4.2 Structural capacity While the geotechnical capacity of a pile is not of great concern, the structural capacity can be. Addi-tional loads in the vertical axis due to negative skin friction will need to be checked for and compared to the design capacity of the pile. In addition, horizontal load may also be induced in piles. Most buildings transfer their loads vertically to the ground and therefore any horizontal capacity the piles may have is merely a by-product of the vertical design. With this in mind it is important to assess the impact on the pile of horizontal movements which could bend and deflect due to horizontal ground movements behind the retaining wall. The empirical methods outlined above do not provide a means to predict horizontal ground movements with depth for excava-tions. There are two alternatives currently available. The excavation and building can be model in a nu-merical package to account for soil structure interaction and the horizontal capacity checked. Alterna-tively, the work of Poulos can be utilized using the normalized charts produced to predict pile loads due to horizontal ground movements. In addition, recent studies have been conducted locally on this pheno-menon by Leung et al (2000) and Ong et al (2003) and provide greater insight into the effects of ground movements on piles. 5 BUILDINGS WITH MIXED FOUNDATIONS For buildings with mixed foundations the standard limiting strain approach of analysis for building damage is not appropriate. This applies to buildings that have different foundation systems and also to series of buildings that share party walls and have different foundation systems. Both scenarios are common in Singapore. The response to ground movement will be different for each part of the structure and so the limiting strain approach no longer applies. Detailed assessment is often required but as a first stage the limits suggested by Skempton & MacDo-nald can be used assuming that the different parts of the foundation respond in the worst manner, the deep foundation experiencing no settlement and the shallow foundation following green field deforma-tion. The differential settlement between the components can be compared to the limits with 1:150 in-ducing possible structural damage. Often buildings with mixed foundations are identified too late within a contract, only as the damage ma-terializes, as details of the foundations and any alterations or partial upgrade are frequently not recorded on any as built drawings as such work was carried out before such records were produced. However, given the potential impact of mixed foundations to a project more attention should be given to identify-ing mixed foundations. During the precondition survey stage potential mixed foundation can be identi-fied through inspection of the state of the building and its surroundings. Trial pits can be used to con-firm any suspicion. Once identified different design solutions can be utilised to estimate the likely damage; these can range from complex structural models to simple comparison of differential settle-ments. Appropriate design solutions can also be employed to prevent the differential settlements form occurring. 6 CONCLUSION In this paper a method of combining the spandrel and concave curves into a single function has been demonstrated in line with the currently available theory for settlement perpendicular to a shaft or sup-ported excavation. It should be noted that while the concave curve might represent a more accurate pre-diction of the ground movements immediately adjacent to the wall, frequently this zone is of little conse-quence due to location of the works and proximity to structures of importance. Building damage assessments can be more conservative using the concave profile but this is dominated by the horizontal movements which need to be carefully considered in any assessment and account made for the soil-structure interaction as noted by Potts & Addenbrooke. It should be highlighted that the parameters used in the formulae still need additional confirmation through the collection of field data and that while the curves appear to represent a reasonable fit to the data presented they are still empirical in nature. The curves also make no allowance for consolidation settlements, though a combination of the spandrel and concave curves with appropriate parameters could take account of installation, excavation and consolidation effects. The importance of accurately


287

predicting horizontal movements is very important in assessing building damage and overly conservative predictions will lead to unnecessary effort in protecting structures that do not require it. Therefore the application of the horizontal movements associated with the concave curve presented here must be treated with caution given that field data is relatively sparse and the curvature is relatively large. In addition some basic guidelines for the assessment of piled structures and the impact on them caused by excavation induced settlement have been put forward in a hope that a more unified approach can be adopted. By correctly identifying building movements and their associated impact on structures adjacent to exca-vations in the urban environment, a significant concern to the project can be eradicated. Frequently it is not the overall magnitude of movement but its manifestation to the building that is important, this needs to be understood so that the buildings likely to suffer damage can be identified, focused upon and design solutions put in place to ensure damage is controlled. REFERENCES Bowles J E 1996 Foundation Design and Analysis, 5th Edition, McGraw-Hill Bowers K M & New B M 1994 Ground Movement Model Validation at Heathrow Express Trial Tunnel, Proc.

Tunnelling 1994, IMM London Clough G W & O’Rourke T D 1990 Construction Induced Movements of In situ Walls, Proc. Design and Per-

formance of Earth Retaining Structures, 439-470, ASCE Special Conference, Ithaca, New York Hsieh P G & Ou C Y 1998 Shape of Ground Surface Settlement Profiles Caused by Excavation, Can. Geotech J

35, 1004-1017 Leung C F, Chow Y K & Shen R F 2000 Behavior of pile subject to excavation-induced soil movement. Journal

of Geotechnical and Geoenvironmental Engineering ASCE Vol 126, No 11, pp 947-954 Loganathan N & Poulos H G 1998 Analytical Prediction of Tunnelling Induced Ground Movements in Clay, J

Geotech. Geoenv Engng ASCE 124 No 9 846-856 Mair R J & Goh K M 2008 Response of Building under Excavation Induced Ground Movements, International

Conference of Deep Excavations 2008, Singapore Mair R J, Taylor R N and Burland J B 1996 Prediction of Ground Movements and Assessment of Risk of Build-

ing Damage, Geotechnical Aspects of Underground Construction in Soft Ground, 712-718, Balkema, Rot-terdam

Ong D E L, Leung C F & Chow Y K 2003 Time-dependent pile behavior due to excavation-induced soil movement in clay. Proc 12th Pan-American Conference on Soil Mechanics and Geotechnical Engineering, Massachusetts Institute of Technology, Boston, USA

Osborne N H, Ong J & Chang K B 2005 Minimizing Construction Impact on a Settlement Sensitive Building, Proc. Underground Singapore 2005, Singapore

Peck R B 1969 Deep Excavations and Tunnelling in Soft Ground, 7th ICSMFE, State-of-the-Art Volume Potts D M & Addenbrooke T I 1997 A Structure’s Influence on Tunnelling Induced Ground Movements, Proc.

Inst Civ Engrs Geotech. Engng Vol 125 (2) 109-205 Shirlaw J N, Wen D, Algeo R A & Patterson-Kane K J 2003 The Effect of Excavation on some Buildings on

Mixed Foundations in Singapore. Proc. Underground Singapore 2003, 111 – 122, Singapore Skempton A W & MacDonald D H 1956 The Allowable Settlement of Buildings, Proc. Inst of Civ Engs Part

III 5, 727-784


288

1 INTRODUCTION

Terzaghi and Peck (1967) developed an empirical method based on strut load envelopes to estimate the loads in the struts of a multi-braced excavation. This is based on constructing apparent earth pressure diagrams based on theoretical and practical studies. It was called apparent because the earth pressure was not meant to represent actual earth pressure or its distribution with depth, but for the sole purpose of evaluating the strut loads from the empirical envelopes. By using measured strut loads from a number of subway projects in New York, Oslo, Toronto, etc., and back-calculating the apparent pressure by dis-tributing the loads over some assumed contributing area of the soil mass, Terzaghi and Peck proposed apparent pressure diagrams (APDs) for various general types of soil. This is shown in Figure 1. To distinguish between soft and stiff clays, Peck (1969) proposed to calculate a stability number N as N = (*h + q)/Cu, where is the unit weight of soil, Cu is the undrained shear strength, H is the depth of excavation and q is the surcharge load at top of excavation. When N is less than 4, the soil is considered stiff clay and the apparent pressure would range from 0.2H to 0.4H depending on how stiff is the clay (see Figure 1c). When N is greater than 4, the soil is considered soft to medium stiff clays and the ap-parent pressure is calculated (*H – 4*m*Cu) as shown in Figure 1(b), where m is an empirical coeffi-cient used to account for the low stability of excavations underlain by soft clays. The design strut loads of braced excavations may then be calculated using a tributary method applied onto the relevant APDs.

Observed apparent pressure diagrams from actual strut monitoring of excavations in Circle Line project

W.M. Cham & K.H. Goh Land Transport Authority, Singapore

ABSTRACT: The apparent pressure diagrams proposed by Terzaghi and Peck (1967) have long been used in practice. However, these apparent pressure diagrams were developed based on field measure-ments involving relatively flexible walls. The popularity of the more rigid diaphragm walls in recent years substantiates the need to further investigate the applicability of the apparent pressure diagrams. The construction of the Circle Line Stage 3, 4 and 5 projects in Singapore offered the opportunity to ob-tain valuable instrumentation data on the effects of excavation. This paper reports, analyses and dis-cusses the strut forces at most of the station and cut-and-cover locations. The study looks into refining apparent pressure diagram for multi-propped excavation based on local soil conditions involving flexi-ble and rigid walls. The effects on the strut forces from factors such as wall stiffness, excavation width and depth to hard stratum are also studied. The results are compiled into various graphs so that these can become useful check for engineers undertaking future excavation designs in Singapore.


289

Figure 1. Apparent pressure diagrams for computing strut loads in braced cuts (after Peck, 1969)

Since then, the dimensionless APD envelopes as proposed by Peck (1969) have been well used in design practice. However, there were several limitations of Peck’s method. In particular, it was noted that the original APDs were developed based on field measurements involving relatively flexible walls. Although the original 0.4H envelope remains a good fit for case histories of flexible walls in stiff clays, the fur-ther case histories by Twine and Roscoe (1999) suggested that Peck’s tentative recommendations for stiff walls in stiff clays are not conservative. Furthermore, Chang and Wong (1996) had also conducted a parametric study using finite element analysis to investigate the apparent pressure diagram for braced excavations in soft clay with diaphragm wall. It was also concluded that the original APDs by Terzaghi and Peck underestimates the strut loads for soft clay excavations supported by stiff walls. With these in mind, it becomes necessary to review the suitability of Peck’s original APDs in estimating strut loads during excavations. This is further accentuated by the increasing use of more rigid diaph-ragm walls in today’s construction and which the earlier studies had suggested that the original APDs are inadequate. The construction of the Circle Line projects in Singapore offered the opportunity to ob-tain valuable instrumentation data on the strut forces due to braced excavation. The study looks into re-fining apparent pressure diagram for multi-propped excavation based on local soil conditions involving flexible and rigid walls. The effects on the strut forces from factors such as wall stiffness, excavation width and depth to hard stratum are also studied. The objective of the proposed revised apparent pres-sure diagrams is not to replicate the exact pressure distribution but to provide a reasonable basis for de-sign and check.

2 STRUT MONITORING FROM CIRCLE LINE PROJECTS

2.1 Geological conditions and type of retaining walls

The Circle Line (CCL) is a fully underground orbital line, linking the strategic Mass Rapid Transit lines leading into the city. It is 33 kilometres long, and comprises of 29 underground stations which are all constructed using the cut-and-cover method. For this study, the strut monitoring at several braced exca-vation sites in Stages 3, 4 and 5 of the CCL, starting from Bartley station to HarbourFront station, are analysed and reported. Figure 2 shows the geological formations through various stages of the Singa-pore Circle Line. Specifically, CCL3 alignment was into the Bukit Timah Granite & Old Alluvium for-mations, CCL4 alignment was into the Bukit Timah Granite and Jurong formation, and CCL5 alignment was into predominantly the Jurong Formation. At several of the stations, the Kallang Formation was al-so encountered, but at different extents depending on local conditions.


290

Figure 1. Geology through various stages of the Singapore Circle Line

Table 1. Construction details at various stations in CCL3, CCL4 and CCL5.

Station Max. excava-

tion depth (m)

Max. excava-tion width

(m) Ground conditions Type of Retaining walls

Bartley 21 35 Fill, KF, OA, GVI, GV,

GIII/GII Contiguous Bored Pile

(CBP) walls

Serangoon 22 35 Fill, GVI, GV Diaphragm Walls (D-wall)

Marymount 20 32 Fill, KF (5-10m), GVI,

GV, GIII/GII Soldier pile walls

(1.2m secant walls locally)

Holland Village 12

(entrance) 12

Jurong formation SIII, SIV, SV

1m CBP wall

Holland Village 20

(station) 26

Jurong formation SIII, SIV, SV

1m thick D-wall

Buona Vista 21 30 F1 and F2(5m) + Jurong

Formation SIII/SIV 1m thick D-wall

One North C&C 23 20 Jurong formation SIV, SV

SVI 1.2m CBP wall

NUH 35 26 Jurong formation SIV/SV CBP

Telok Blangah 20 33 5-10m Kallang formation + Jurong formation SIV,

SV

Soldier pile walls (1m CPB wall at both

ends)

Alexandra (now renamed as Labrador Park)

18 25 Jurong Formation SIII/SIV 1m thick D-wall

Pasir Panjang 18 21 15m Kallang formation + Jurong formation SIV, SV

SVI

1m thick D-wall (with cross walls)

Table 1 summarises all the station locations from CCL3, CCL4 and CCL5 that were studied and whose strut monitoring are reported in this paper. The stations were constructed using cut and cover method of construction with multiple layers of struts. The size and depth of excavation, geological conditions, as well as the types of retaining walls are also listed for each station excavation. It can be noted that the re-taining walls can be differentiated into two categories – firstly is a stiff retaining wall system comprising of a diaphragm wall (sometimes with the addition of cross-walls), and secondly is a more flexible retain-ing wall system using soldier piles and contiguous bored piles.

CCL1

CCL21

CCL321

CCL421

CCL5421


291

2.2 Monitoring of strut loads

An intensive programme of strut monitoring was implemented during the excavation works. In line with the APD method outlined by Peck (1969), the maximum value of the strut load was extracted for the en-tire monitoring period for each strut at the various excavation sites. More than 1500 strain gauges and load cell readings were extracted and these are presented in Figure 2.

Figure 2. Maximum strut loads monitored for each strut at the various station locations

3 ANALYSIS OF STRUT MONITORING DATA

3.1 Analysis of strut forces using apparent pressure method

For each strut location, the maximum strut loads were back-analysed into the definition of Peck’s ap-parent pressure. This is done by dividing the strut loads (F) into the tributary area derived from inter-strut spacings (A), and then further normalising the average pressures by dividing this with H. The normalized apparent pressure (F / A / H) is then plotted against the location of the strut with respect to the final excavation depth (i.e. z / H). The back-analysed apparent pressures from actual strut monitor-ing are then compared against the original APDs that would have been estimated using Peck’s recom-mendation. The results are plotted in Figure 3 and Figure 4. In order to compare against Peck’s APDs, it is necessary to estimate the stability number for the exca-vation conditions. Table 2 calculates the stability number corresponding to each cut-and-cover excava-tion. The undrained shear strength was estimated based on the average value corresponding to the strut-ted excavation depth. At some locations, the excavation was done in rock and without struts at the lower levels. In such cases, the maximum excavation depth and the average undrained shear strength were es-timated based on the soils above the unsupported excavations in rock. When the stability number is more than 4, the ground would comprise of soft to medium stiff clays and the APD in Figure 1b would be used to estimate the strut loads. As seen in Table 2, the excavation at Pasir Panjang is an example of such an excavation, whose high stability number reflects the thick Kal-lang Formation encountered during excavation. When the stability number is less than 4, the ground comprises of stiffer clays and the APD in Figure 1c would be used to estimate the strut loads. It is also to be noted that the lower the stability number, the stiffer would be the ground. For example for the ex-cavations at Telok Blangah and at Marymount, although the lower half of the excavation is done in rock


292

and without props, the upper half of the excavation – which is strutted – is done in medium stiff soils. This is compared to the low stability numbers for excavations, say in Holland Village Alexandra, Buona Vista for example, where the excavations were done in stiffer clays in the Jurong Formation. Table 2. Calculation of stability number at various station locations.

Station

Ground condi-

tions at strut

location

Max. excava-

tion depth, H

(m)

Unit weight of

soil,

(kN/m3)

Ave undrained

shear strength,

Cu (kN/m2)

Stability

number,

Nc

Bartley Kallang/

GIV/GV/GVI 21 20 175 2.4

Serangoon Kallang/

GV/GVI 22 20 135 3.3

Marymount Kallang /

GV/GVI 20 20 122 3.3

Holland Village

(entrance) SIV/SV 12 22 189 1.4

Holland Village

(station box) SIV/SV 20 22 294 1.5

Buona Vista

SIV/SV 21 22 252 1.8

One North

C&C

SIV/SV/SVI 23 22 230 2.2

NUH

SIV/SV 35 22 260 3.0

Telok Blangah Kallang

/SIV/SV 20 22 131 3.4

Alexandra

SIV/SV 18 22 315 1.3

Pasir Panjang Kallang Forma-

tion 18 16 24 12.0

The original APDs for various types of ground conditions can then be plotted against the apparent pres-sure maximum strut loads monitored. Figures 3a and 3b show the measured normalized apparent pres-sure plotted against the normalized strut depth for the excavations that are supported in more flexible re-taining walls (soldier piles, CBPs), for excavations in medium stiff clays (3<N<4) and for excavations in very stiff clays (N<3) respectively. The back-analysed apparent pressures are observed to fall within the bounds of Peck’s APDs. From the results of the CCL excavations, the original APDs recommended by Peck (1969) may still be suitable as a design check against the maximum strut loads estimated for excavations supported using flexible retaining walls. Figures 4a and 4b show the measured normalized apparent pressure plotted against the normalized strut depth for the excavations that are supported using more rigid diaphragm walls, for excavations in soft clays (N>4) and for excavations in stiff clays (N<4) respectively. It is observed that the back-analysed apparent pressures are higher than the original APDs by Peck. This is not surprising since the original APDs were based on flexible retaining walls. A stiffer retaining wall would restrain ground deforma-tions but would attract more loads into its supports. Therefore, the tentative recommendations by Peck for excavations supported by stiff walls might not be suitable. This has already been recognised in earli-er works, such as by Twine and Roscoe (1999) and Chang and Wong (1996). Furthermore, by drawing upper bound limits based on the CCL data, it is possible to obtain the apparent pressures that could be used to estimate the maximum strut loads for excavations supported using stiff walls.


293

(a) (b)

Figure 3. Normalized apparent pressure for Soldier pile/CBP wall in (a) medium stiff clay and (b) stiff clay

(a) (b) Figure 4. Normalized apparent pressure for diaphragm walls in (a) soft clay and (b) stiff clay

3.2 Further analysis of apparent pressures

Using the apparent pressures back-analysed from the CCL data, further analysis is undertaken to study the effects on the apparent pressures due to factors such as wall stiffness, thickness to hard stratum and excavation width. 3.2.1 Effect of wall stiffness

Three retaining walls with different flexural stiffness (EI) ranging from 1m and 1.2m diameter conti-guous bored pile walls and 1m thick diaphragm walls were investigated. The back-analysed apparent pressures are plotted at various depths (normalized as a depth ratio, z / H) for walls of different stiff-nesses. These are as shown in Figure 5. For all the depths, there is a clear trend of increasing apparent pressures with wall stiffness. Indirectly, this indicates that the maximum strut force during an excava-tion is dependent on wall stiffness. This is similarly observed by Chang and Wong (1996), who went on


294

to relate this phenomenon to the arching effect induced by wall displacement. Less stiff walls allow for larger displacements of the wall and the soil, thereby inducing stronger arching effects as the soil de-forms and shears. This would reduce the forces on the wall and on the struts as the wall stiffness is re-duced.

Figure 5. Effect of wall stiffness on apparent pressures

3.2.2 Effect of thickness to hard stratum

The restraining effect of a hard stratum in reducing the maximum strut loads may also be investigated from the back-analysed apparent pressures in the CCL projects. For each strut level, the thickness to the hard stratum is calculated in terms of its distance to the final formation level (D) plus the depth of the soil layer above the hard stratum (T). Figure 6 plots the back-analysed apparent pressures at selected struts in various stations against the normalized thickness to hard stratum. As the thickness to hard stra-tum increases, the apparent pressure increases due to the reducing restraint provided by the hard stra-tum. When the normalized thickness to hard stratum (i.e. (D+T) / H ) is less than 1, the restraining ef-fect from the presence of hard stratum at shallow depth could potentially reduce the apparent pressure, and this would reduce the maximum strut force during excavation.

Figure 6. Effect of hard stratum on apparent pressures


295

3.2.3 Effect of excavation width

Figure 7 investigates if there is any influence from the width of the excavation (B) on the apparent pres-sures. From the plots which were done for both CBP walls and diaphragm walls at various station loca-tions, there is no evidence to suggest that the excavation width may have an influence on apparent pres-sures. Hence, it is believed that the maximum strut load may not be influenced by the width of the excavation.

Figure 7. Effect of excavation width on apparent pressures

3.3 Proposed modification to the original apparent pressure diagrams

It is noted in Figure 3 and Figure 4 that the envelope of the apparent pressure diagrams based on the CCL data are similar in shape to the original apparent pressure diagrams. To account for the influence of stiffer retaining walls in increasing strut loads, it is proposed to include a modification factor into the original APDs. This modification factor (Mf) should be based on empirical experience and data of com-pleted case histories. Based on the CCL data, the modification factor is as proposed in Figure 8.

Figure 8. Proposed apparent pressure diagrams with modification factors

4 CONCLUSION

The evolution of actual earth pressure distribution behind a retaining wall during an excavation can be rather complicated. The shape of the earth pressure distribution and its magnitude depends on factors such as the relative stiffness of the wall system in comparison to the strength of the ground, sequencing and time of construction, drainage conditions, etc. Furthermore, measurements of earth pressures on full-scale walls, model tests and computer simulations have also shown that earth pressures are typically

Flexible walls

Stiff walls

Mf = 1

Mf = 1.2

Mf = 1

Mf = 1.8

Soft to medium

clay

Stiff to hard

clay


296

more concentrated near wall supports depending on soil type and wall type. These complications led Potts (1992) to comment that all simplified methods of analysis (such as stress field and limit analysis using upper and lower bounds of plasticity and equilibrium methods such as the Coulomb wedge analy-sis) would be flawed, and their application in the design of struts would be inappropriate unless it has been verified by field observations. The empirical approach of using apparent pressure diagrams by Tergazhi and Peck attempts to estimate the maximum strut load for excavations in different soil conditions based on actual field data. However, there are several limitations from the apparent pressure diagrams as these were developed from limited field data and primarily from excavations supported by flexible walls. Although Twine and Roscoe (1999) developed a distributed prop load method substantiated with more field data to improve on the limitations of the apparent pressure method, the original Peck’s method is still popular today as a sim-plified design check for the predictions from finite element analysis. In this paper, the strut monitoring data from the completed excavations in Circle Line was used to vali-date the use of the apparent pressure method. It was found that the back-analysed CCL data fit in with the original apparent pressure diagrams for flexible retaining walls in medium stiff to stiff clays, but the original apparent pressure diagrams were under-estimating the maximum strut loads for excavations supported using rigid diaphragm walls. This can be overcome by including a modification factor (based on empirical data) to the original apparent pressure diagrams. A trend of increasing apparent pressure for higher wall stiffness and greater strut distance to the hard stratum can also be interpreted from the CCL data. It is hoped that the compiled database of maximum strut forces recorded from the CCL ex-cavations can become useful references for engineers undertaking future design of strut system for braced excavation in Singapore.

REFERENCES

Chang, J.D. and Wong, K.S. 1996. Apparent pressure diagram for braced excavations in soft clay with diaph-

ragm wall, Proc. International of Symposium on Geotechnical Aspects of Underground Construction in Soft

Ground, London, 15-17 April 1996, pp.87-92.

Peck, R.B. 1969. Deep excavations and tunnelling in soft ground. Proceedings of the 7th Int. Conf. on Soil Me-

chanics and Foundation Engineering, Mexico City, state-of-the-art volume, pp.225-290.

Potts, D.M. 1992. The analysis of earth retaining structures. Proc. Conf. Retaining Structures, ICE, London,

pp.167-186.

Terzaghi, K. and Peck, R.B. 1967. Soil Mechanics in Engineering Practice, 2nd Ed., John Wiley & Sons.

Twine, D. and Roscoe, H. 1999. Temporary propping of deep excavations – guidance on design, CIRIA Publi-

cation C517, London.


297

1 INTRODUCTION A cut-and-cover road tunnel is to be constructed in the Marina East area in Singapore, where thick lay-ers of Kallang formation soils can be found. Alongside this development, a 2.5m diameter trunk link sewer pipe of about 800m long will also be constructed at a depth of about 15m below ground, i.e. in the Upper Marine Clay (UMC) layer. A typical cross section showing the cut-and-cover tunnel and the sewer pipe is given in Figure 1. The area had been reclaimed in the past (from 1979 to 1985) and the soft soils in this area are still undergoing consolidation settlement. Hence, the sewer pipe will settle along with the soft soils except at manhole locations, where no settlement is expected as the manholes will be supported on piles. The requirements stem from the fact that connections to the manholes are made by sewers, which undercross the road tunnel. No settlement is expected for these connections due to the presence of ground improvement for the cut-and-cover tunnel excavation. Thus the piling for the manholes is required to prevent large differential settlements. Since the thickness of the Kallang forma-tion soft soils varies along the sewer alignment, non-uniform sewer pipe settlement is also expected in the long term. Since the sewer pipe is constructed by jointed reinforced concrete segments, the non-uniform long term settlement will affect its serviceability as the joints can normally only tolerate a joint rotation of 1 in 120.

This paper elaborates a methodology, which has been used to compute settlement of sewer pipe con-structed in the Marina East area. Cone penetration tests (CPT) using piezocones were carried out, which can provide some indications of the degree of consolidation of the soft soils in this area. Back-analyses were carried out to derive parameters required for the consolidation analysis using large strain soft soil model finite element method. Results that have been obtained in terms of soil consolidation settlement, sewer pipe long-term settlement and its comparison with Mohr-Coulomb and classical one-dimensional Terzaghi consolidation analysis are also presented and discussed.

Settlement of sewer pipes in consolidating soft clays

S.S. Agus, N. Mace Mott MacDonald Singapore Pte Ltd, Singapore

ABSTRACT: A section of 2.5m diameter trunk link sewer pipe of about 800m long will be constructed in conjunction with and alongside a cut-and-cover tunnel road tunnel in consolidating soft soils at a depth of about 15m below ground. In this paper, a method to calculate the sewer settlement considering soil-structure interaction is presented. Back-analyses of cone penetration (CPT) data were carried out to determine the coefficient of permeability of the consolidating soils. Consolidation settlement of the soft soils was further computed using a large strain soft soil model. A separate soil-structure interaction analysis was then performed in order to determine the longitudinal response of the sewer. In order to re-duce the magnitude of differential settlement, a soil zone around the sewer will be improved by means of prefabricated vertical drains (PVDs) installed locally to accelerate the consolidation process of the soft soils such that the remaining settlement after the sewer is constructed can be limited. Discussion on the results is given to provide an understanding on the effects of localized PVD ground improvement and soil-structure interaction between the sewer pipe and the consolidating soft clays.


298

8m (typical)

Trunk link sewer Cut-and-cover road tunnel

PVD-improved zone

~60m (typical)

Figure 1 Typical cross section

2 GEOLOGICAL CONDITIONS

The site is located in a generally “green field” area at Marina East. The proposed route continues from the one end near the Marina Bay Channel and terminates in the Marina East area. Geological profile for the project has been determined through extensive soil investigation works and is shown in Figure 2. The soils consist of an about 10m up to 15m thick reclamation fill layer underlain by Kallang formation lay-ers with varying thickness from about 20 to more than 40m. The Kallang formation soils in this area in-clude Upper Marine Clay (UMC), Lower Marine Clay (LMC) sandwiching Fluvial Clay (F2) and/or Fluvial Sand (F1) and Estuarine (E) layers. The reclamation history in this area has been described in the earlier section. As described, the Kallang formation soils are generally still undergoing consolidation and thus excess pore-water pressures exist. Old Alluvium (OA) is encountered beneath the Kallang for-mation soils across the entire site between approximately 30m and 55m below the existing ground level.

Figure 2 Geological profiles along sewer alignment

3 METHOD OF ANALYSIS For predicting the long-term settlement of the soils and thus the trunk link sewer, two-dimensional finite element (FE) method using Plaxis was adopted. Since the soil settlements in the area of concern are mainly attributed to consolidation settlements, accuracy and reasonability of the consolidation parame-ters used in the computation play an important role. The most important consolidation parameter for predicting the remaining long-term settlements is the coefficient of consolidation (cv), which is strongly related to the coefficient of permeability (k) of the soils. The values can be back-calculated from the in-situ test data that have been obtained. Once the consolidation parameters are determined, the current de-gree of consolidation can then be evaluated so that the remaining settlement magnitude can be estimated.

Manhole A Manhole B Manhole C

Distance from Manhole A (m)


299

3.1 Back-analyses of Cone Penetration Test (CPT) Data

Since the consolidation process is basically a densification process, for soft clays the shear strength of the consolidating layers increases with the degree of consolidation. The shear strength profile therefore gives an indication of the degree of consolidation of the soils. The CPT data that have been obtained from tests along the trunk sewer alignment can be related to the field vane shear test (FVST) data and the laboratory shear strength data to provide the shear strength profiles of the soils along the sewer alignment. The analysis commenced with calibration of CPT cu-p’ relationship and the cone bearing factor (Nkt) for the soils encountered, where p’ is the effective overburden pressure. Subsequently, the cu profile for each CPT location was back-analysed to obtain k values of the Kallang formation soils. In the FE ana-lyses, the following factors were considered:

1. Old reclamation took place over 6 years from 1979 to 1985 (t = 0 day started in 1979). 2. Shear strength data were obtained in 2008 (i.e. 29 years after 1979). 3. The maximum k value for marine clays as reported by Buttling et al (1987) (i.e. 2x10

-9m/s)

was adopted as an initial value of k (i.e. ko). Buttling et al (1987) reported k value of Singa-pore marine clays between 10

-10m/s and 2x10

-9m/s.

4. The E layer was assumed to have a similar k value to the LMC. 5. The F2 layer was assumed to be a consolidating layer as well with an initial value of k of 10

-

9m/s.

6. Large strain analyses were adopted. The FE geometry (or the FE mesh) was updated from time to time in the FE analyses to account for the large strains that occur. Wong and Choa (1987) indicated that ignoring the large settlement effect (or the updated mesh) under-predicts the rate of settlement.

7. The k value was assumed to vary with void ratio (e) following: log (k/ko) = e/ck, with ck taken to be in the order of the compression index (Cc) (Plaxis, 2008).

8. The anisotropy factor for the k value was assumed to be equal to 2 or kh/kv = 2. This value was also reported by Buttling et al (1987) for Singapore marine clays.

3.2 Results of Back-analyses

At the 29th year (i.e. at the time of site investigation), the shear strength profiles were checked. It was

found that the FE analyses predict the shear strength profiles quite closely with the above factors taken into consideration. Figure 3 shows the predicted cu profiles for one of the CPT points. The measured cu profile from the CPT data was obtained from the relationship cu = (qt – v)/Nkt. The Nkt value adopted was based on the formula proposed by Bo et al (2002). The value of Nkt was consi-dered to follow the relationship Nkt = 23.8 – (0.263 Ip) where Ip is the plasticity index. Table 1 summa-rises the Nkt values for different soft soils in the area. The adopted values are seen to give a good fit to the measured cu profile. The theoretical line for the UMC and LMC shown in the above figure was calculated as cu/p’ = 0.22 (p’ is the effective overburden pressure). This is within the range given by Buttling et al. (1987). For the other soft clays, cu/p’ = 0.11 + 0.0037 Ip (Skempton, 1957) was adopted. For the E and F2 layers, the value of cu/p’ can be taken to be equal to 0.25 and 0.295, respectively based on the range of Ip of these soils (Table 1). The initial k values (ko values) obtained from the back-analyses are shown in Table 2. The k values re-duce as the soils consolidate following the relationship given in the earlier section. The ko values were adjusted in the calculation until the cu profiles gave a good match. Table 2 summarises ko values ob-tained from the back-analyses compared with the measured and original estimated values. Table 2 reveals that the k values obtained from the back-analyses generally agree well with the esti-mated values with less than one order of magnitude variation, which indicates the accuracy of the back-analyses carried out. Buttling et al. (1987) also reported k values ranging from 10

-10 to 2x10

-9m/s for

Singapore Marine Clays.


300

50

60

70

80

90

100

110

0 50 100 150 200

Cu (kPa)

Ele

vati

on

(m

RL

)Computed from CPT

Laboratory data

Field vane shear data

Theoretical at t=infinity

Simulated at t=29years

UMC

F2

LMC

E

Reclamation Fill

Figure 3 Comparison of undrained shear strength (cu) profiles for one of CPT points

Table 1 Nkt and cu/p’ ranges as a function of Ip for soft soils in the area

Layer Ip range Nkt range cu/p’

Upper Marine Clay (UMC) 38 – 55 9.3 – 13.8 0.22 Lower Marine Clay (LMC) 28 – 59 8.3 – 16.4 0.22

Estuarine (E) 15 – 52 8.0 – 19.3 0.25 – 0.295 Fluvial Clay (F2) 17 – 60 10.1 – 19.9 0.25 – 0.295

Table 2 ko values obtained from the back-analyses

Layer ko value used (m/s)

k at the time of SI works (t=29 years) (m/s)

Original estimated k (m/s)

Upper Marine Clay (UMC) 2x10-9 1.1x10-9 10-9 Lower Marine Clay (LMC) 2x10-9 1.7x10-9 10-9

Estuarine (E) 2x10-9 1.6x10-9 10-9 Fluvial Clay (F2) 2x10-9 7.8x10-10 10-9

4 ESTIMATE OF REMAINING CONSOLIDATION SETTLEMENT

4.1 Finite Element Modeling

Once the coefficient of permeability values had been back-calculated, the analyses were subsequently continued to consider construction stages for the trunk link sewer construction. It was envisaged that ground improvement by means of prefabricated vertical drains (PVD) would be required to accelerate the consolidation process such that the remaining consolidation settlement affecting the sewer pipe is minimized. It was also foreseen that only localized areas would require PVD treatment near to the man-holes where differential settlement would be high as an economical solution to the settlement problem.

The following factors were considered in the analyses:


301

1. Conditions at the project commencement (i.e. in 2009), i.e. 30 years after the old reclamation works commencement in 1979.

2. New reclamation was planned to be carried out (with PVD) along parts of the alignment, which was in this case was assumed to be done in 1 year duration. The PVDs for the new reclamation were assumed to be installed in January 2010.

3. The PVD for sewer construction was assumed to be installed in June 2010. 4. The cut-and-cover tunnel basic structure was assumed to be completed in June 2012, after

which the trunk link sewer is to be constructed. 5. The excavation works for the cut-and-cover tunnel construction were not modeled.

Several sections were modelled for a good representation of geological profile variation on site. The fol-lowing typical section (Figure 4) was adopted with the toe level of the PVDs limited by the equipment used for the installation (maximum depth = 50m from the ground level). The width of PVD improved zone ranges from 8 to 20m for the area where the new reclamation works are carried out and is typically 8m for the area without new reclamation works. It is expected that the consolidation settlement is accelerated in the PVD improved zone. In the area without PVDs, the consolidation settlement will proceed slowly (delayed consolidation) as the permea-bility of the soft clays is low.

Figure 4 Typical cross section analysed for the area with new reclamation works

4.2 Geotechnical Parameters Adopted

The geotechnical parameters for the analysis are given in Table 3 with the coefficients of permeability, which were taken from the values obtained from the back-analyses (see Table 2). The soils were mod-eled using the soft soil model in Plaxis.

The reclamation fill, Old Alluvium (OA), and F1 soils were modeled as drained Mohr-Coulomb soils. The armor rock was modeled as a Mohr-Coulomb soil with = 22kN/m

3; c’ = 3kPa; ’ = 35

o; E’ =

10,000kPa; and = 0.25. The cut-and-cover tunnel (solid element) was modeled as an elastic non-porous material with a reduced density of 17kN/m

3 and a reduced stiffness of 25% that of concrete to

allow for ageing. The interface layer was conservatively considered to be a soft soil similar to UMC with a lower effective friction angle of 15

o. The interface layer was used to represent the soil distur-

bance due to the construction of the cut-and-cover tunnel temporary wall and excavation works.


302

Table 3 Soft soil parameters adopted in the analyses

Soil type

Effective stress parameters Compression parameters

Bulk den-sity,

(kN/m3)

Effective cohesion, c’ (kPa)

Effective friction

angle, ’ (

o)

Initial void ra-tio (eo)

Compres-sion in-dex,Cc

Recom-pression index, Cr

(Cc/2.3(1+

eo))

(Cr/2.3(1+

eo))

UMC 16 0 22 2.0 0.90 0.1500 0.130 0.043

LMC 17 0 22 1.4 0.60 0.100 0.109 0.036

E 17 0 17 1.6 0.70 0.1167 0.117 0.039

F2 20 10 23 0.5 0.25 0.0417 0.072 0.024

4.3 PVD-Treated Soil

The PVD-treated zones were modeled by means of zones with equivalent k values. The method as de-scribed in Chai et al (2001) was adopted. The equivalent vertical coefficient of permeability of the PVD-improved soil (kve) is formulated as:

v

v

h

e

ve kk

k

Dk

2

25.21

(1)

Where kv and kh is the coefficient of permeability of the soil in the vertical and horizontal direction, re-spectively; ℓ is the drainage length; and De is the diameter of a unit cell. De is equal to 1.13S and 1.05S for the square (or rectangular) and triangular pattern, respectively (S = PVD spacing). The value of can be expressed as:

w

h

s

h

q

ks

k

k

s

n

3

2

4

3lnln

2

(2)

Where n = De/dw (dw = diameter of drain = 2(a+b)/); a and b is the dimensions of the PVD; s = ds/dw (ds = diameter of smear zone); ks is the smear zone coefficient of permeability; and qw is the discharge capacity of PVD.

The ds value can be taken as 5dm (FHWA, 1986) with dm is the equivalent diameter of mandrel calcu-lated from the largest mandrel cross sectional area (dm = (4wm lm/)

0.5); wm and lm are the mandrel di-

mensions (mandrel area is normally in between 60cm2 and 70cm

2). The qw value can be taken up to

100m3/year. The ratio of kh/ks ranges from 1 to 5.

In the computation, the PVD parameters shown in Table 4 were adopted.

Table 4 PVD parameters adopted

Parameters Value

PVD width (a) 100mm PVD thickness (b) 3.3mm Spacing 1 to 2m (triangular pattern) PVD discharge capacity (qw) 100m3/year Mandel width (wm) 444mm Mandrel thickness (lm) 14.6mm Ratio kh/ks 2.5

In the model, only the hydraulic properties (or the k values) of the PVD-improved soils are modified us-ing kve. The strength and deformation properties of the soils follow those of the original soils. Iterative procedures were adopted to determine the most optimum PVD scheme (spacing, pattern and depth), which also satisfies the requirement of a maximum differential settlement of less than 1 in 120.


303

4.4 Results of modeling

Figure 5 shows the typical time-settlement curve of the soil at the sewer position for the cross sections taken through one of the CPT points. The plots indicate that for this particular section 450mm consoli-dation settlement has been accelerated approximately 11.5years earlier that it should occur when no PVDs are used. However, the total consolidation settlement remains almost similar to that without PVDs. The ratio of settlement, which occurs during the PVD treatment period to the total settlement, is related to the width of the PVD-treated zone and the treatment period. Hence, the proportion of the re-maining settlement can be reduced by widening the PVD-treated zone and/or lengthening the PVD treatment period.

0

500

1000

1500

2000

2500

3000

1000 10000 100000

Elapsed time (days)

Sett

lem

en

t at

sew

er

(mm

)

No PVD

With PVD

Start of PVD treatment

End of PVD (~2 years)

120yrs dsg life

Construction of sewer

(June 2012)


(June 2012)

End of 120years

design life

(2132)Settlement prior to PVD

installation

Settlement with PVDs

installed (~450mm)

Settlement after sewer construction

(remaining settlement) (~620mm)

new reclamation

(with PVDs)

0

200

400

600

800

1000

1200

1400

10000 11000 12000 13000 14000

Elapsed time (days)

Se

ttle

me

nt

at

se

we

r (m

m)


(June 2012)


(June 2012)

Figure 5 Typical time-settlement curves at the sewer location

The excess pore-water pressure versus time plot (Figure 6) indicates that during the PVD treatment pe-riod the excess pore-water pressure at the sewer location drops from approximately 100kPa to about 10kPa suggesting that the consolidation process under that particular overburden pressure has almost ceased. However, the excess pore-water pressure rebounds to approximately 45kPa once the PVD treatment is stopped before it reduces again to zero at the end of the 120-year sewer design life. On con-trary the soils continue to settle regardless the increase or decrease in the excess pore-water pressure. Further assessment of the modelling results indicates that when the PVD treatment is terminated volu-metric strain of the soil element at the sewer location reduces signifying a volumetric expansion due to the excess pore-water pressure outside the PVD-treated zone being higher than that inside the treatment area. This volumetric expansion is predominantly caused by lateral strain reduction while settlement still takes place.


304

0

20

40

60

80

100

120

140

160

1000 10000 100000

Elapsed time (days)

Excess p

ore

-wate

r p

ressu

re (

kP

a)

No PVD

With PVD


End of PVD (~2 years)

120yrs dsg life

0

20

40

60

80

100

120

10000 11000 12000 13000 14000

Elapsed time (days)

Excess p

ore

-wate

r p

ressu

re (

kP

a)


(June 2012)


(June 2012)


(June 2012)


(June 2012)

End of 120years

design life

(2132)

Figure 6 Typical time-excess pore-water pressure curves at the sewer location

The soil settlements at the sewer level at several distances from the sewer position have been extracted from the analysis results and are plotted as time-settlement curves in Figure 7. The plot provides infor-mation on the zone of influence of PVD treatment effects. The curves plotted in Figure 7 have been gen-erated for one section without new reclamation works. Hence, the time-settlement curves for the points, where the PVD treatment effects are not prevalent follow the common S-curve in semi-logarithmic scale. The figure appears to show that the PVD treatment effects diminish at a distance of 50m away from the sewer.

0

500

1000

1500

2000

2500

3000

1000 10000 100000

Elapsed time (days)

So

il s

ett

lem

en

t (m

m)

At sewer location

4m away

10m away

20m away

30m away

40m away

50m away

60m away

80m away

100m away

PVD effects

Affected by

PVDs

Unaffected

by PVDs

Figure 7 Time-settlement curves extracted at various points at sewer level several distances away from sewer


305

It is also revealed from the figure that the points located close to the cut-and-cover tunnel (i.e. up to 10m away from the tunnel) exhibit lesser total consolidation settlements compared with those located away from the cut-and-cover tunnel. This fact indicates the influence of tunnel, which reduces the consolida-tion settlements since it acts as a stiff reinforcing element.

5 COMPARISON WITH MOHR-COULOMB MODEL AND TERZAGHI ONE-DIMENSIONAL CONSOLIDATION ANALYSIS

Analyses were carried out in attempt to compare the FE analyses performed using the soft soil model for the consolidating clays with that when the Mohr-Coulomb model is adopted. A further comparison was also made with the analysis using the classical Terzaghi’s one-dimensional consolidation formula.

In these analyses, a section from the area without new reclamation works was considered. Since the PVD ground improvement used was a localized improvement, the one-dimensional Terzaghi’s consoli-dation formula is unable to compute the settlement correctly due to two-dimensional nature of the prob-lem. Thus, the comparison was only made for the time-settlement curve of the soil without PVD treat-ment. Table 5 provides a summary of comparisons of various modelling approaches. Comparison was also made for small and large strain models.

Table 5 Various modeling carried out for comparison

Model Analysis type Soil model for conso-lidating clays

Permeability of con-solidating clays

Large/small strain model +

Model 1 * 2-dimensional FE analysis

Soft soil model Varies with void ra-tio

Large strain model

Model 2 2-dimensional FE analysis

Soft soil model Varies with void ra-tio

Small strain model

Model 3 2-dimensional FE analysis

Mohr-Coulomb mod-el

Constant Small strain model

Model 4 1-dimensional analysis

Terzaghi’s theory Constant Small strain model

Note: * the model adopted in this project; + in large strain model, mesh and pore-water pressure is updated as calculation proceeds

Figure 8 shows the time-settlement curves obtained from the computation using the models given in Ta-ble 5. Lines of the 50-year, 100-year and 120-year design lives are also plotted. The time-settlement curve obtained form the one-dimensional Terzaghi’s consolidation analysis (further refered to as Ter-zaghi’s model) is denoted as “Manual 1D” in the plot.

-3000

-2500

-2000

-1500

-1000

-500

0

1000 10000 100000

Elapsed time (days)

Se

ttle

me

nt

at

se

we

r le

ve

l (m

m)

Model 1

Model 2

Model 3

Manual 1D

50yrs

100yrs

120yrs

Figure 8 Time-settlement curves obtained from various modeling


306

The figure reveals that the modeling using the Mohr-Coulomb soil model for the consolidating soils (Model 3) greatly underestimates the total consolidation settlement of the soil at the sewer location as no volumetric plastic strain is produced. The Mohr-Coulomb model may still be used to estimate soft soil consolidation settlement provided adjusted soil modulus of elasticity is used in the analysis. The other three models yield a comparable total consolidation settlement with a variation of less than 10%. In general, Terzaghi’s model (Manual 1D) estimates total settlement almost similar to the soft soil model (i.e. Model 1 and Model 2) but with a lesser rate of settlement. The model without the large-strain effect consideration (Model 2) underestimates the consolidation settlement by about 20% (maximum) compared to that of Model 1 at elapsed times before the end of 120-year design life. However, the total consolidation settlement at the end of 120-year design life agrees well with that of Model 1. The above discussion suggests that Terzaghi’s model can be adopted for computing the total consolida-tion settlement of layered soft soils with sufficient accuracy (within 10% variation). It is also true for the FE consolidation analysis using soft soil model, but without large-strain considerations (in this case represented by Model 2). However, the latter is not recommended for computation of short-term consol-idation settlement of soft soils as it gives a significant underprediction.

6 COMPUTATION OF SEWER SETTLEMENT The trunk link sewer settlement and differential settlement can be computed from the input soil settle-ment obtained at each CPT point. Settlement of a flexible sewer pipe will generally follow the settlement profile of the surrounding soil. Due to the manhole local effects, the sewer settlement at the manhole lo-cations can be assumed to be zero as the manholes will be founded on piles. The sewer pipe also has a certain bending stiffness in the longitudinal direction. These two factors provide some restraints to the sewer pipe to settle with the surrounding soil.

In terms of the bending stiffness, another factor, which has to be taken into account in the computation of sewer differential settlement, is the fact that the sewer is not continuous and monolithic, but will be constructed by jointing segmental reinforced concrete (RC) pipe segments of a certain length. The pres-ence of joints reduces the longitudinal bending stiffness of the sewer pipe. Chen and Wen (2003) pre-sented a method to account for joints in the calculation of axial and bending stiffness of shield tunnel due to uneven ground settlements. Although the method was developed for bored tunnels where tunnel lining segments are bolt connected longitudinally, a similar concept can also be used in the case of circu-lar RC sewer pipe. Computation of sewer pipe settlement from the soil settlement was carried out using a soil-structure in-teraction software, LPile. This software is designed for the analysis of the lateral response of piles using p-y curves. However, the software does allow for the imposition of a lateral displacement field onto a pile to determine the effects by making appropriate allowances for the manhole locations, the sewer is treated as a horizontal pile. A reduced axial stiffness (EAred) and bending stiffness (EIred) of 75% and 50% respective segment stiffness was adopted to allow for the presence of joints. Figure 9 shows the comparison of long-term soil and sewer settlements (i.e. at the end of 120-year design life) obtained from the analyses. The long-term sewer settlements for 50-year and 100-year design lives are also pro-vided for comparison. The sewer settlement was computed based on the PVD treatment scheme obtained through optimization. PVD is mainly required near the manhole where maximum differential settlement is likely to occur.

Figure 9 shows that the settlement of the sewer pipe at its mid span between two manholes generally fol-lows that of the soil. This indicates that the flexibility of the sewer pipe significantly controls the beha-vior of the sewer pipe. However, the reverse is true near the manhole locations, where the sewer pipe settlement is restrained by the piled manholes. Figure 10 indicates that the maximum differential settle-ment occurs near the manholes, where transition from sagging to hogging sewer settlement profiles oc-curs. The figure also reveals that the PVD ground improvement scheme adopted is effective in limiting the sewer pipe differential settlement to below the maximum allowable value of 1 in 120.


307

0

200

400

600

800

1000

1200

0 100 200 300 400 500 600 700 800 900

Distance from contract

start edge (m)

Sett

lem

en

t at

sew

er

level (m

m)

soil (50years)

soil (100years)

soil (120years)

sewer (50years)

sewer (100years)

sewer (120years)


1st

CP

T p

oin

t

2n

d C

PT

po

int

3rd

CP

T p

oin

t

4th

CP

T p

oin

t

5th

CP

T p

oin

t

6th

CP

T p

oin

t

7th

CP

T

ad

dit

ion

al secti

on

ad

dit

ion

al secti

on

ad

dit

ion

al secti

on

ad

dit

ion

al secti

on

ad

dit

ion

al secti

on

Figure 9 Soil and sewer settlements obtained from the analyses

134

172123 122

10

100

1000

10000

0 100 200 300 400 500 600 700 800 900

Distance from contract

start edge (m)

Sew

er

dif

fere

nti

al sett

lem

en

t (1

in

ve)

sewer (50years)

sewer (100years)

sewer (120years)

allow. diff sett


1 in 120

1st

CP

T

2n

d C

PT

3rd

CP

T p

oin

t

4th

CP

T p

oin

t

5th

CP

T p

oin

t

6th

CP

T p

oin

t

7th

CP

T p

oin

t

Figure 10 Differential settlement of the sewer

7 SENSITIVITY ANALYSES

Sensitivity analyses were carried out to investigate the effects of sewer pipe span length (manhole to manhole) and bending stiffness. The ratio of sewer pipe reduced to full bending stiffness (EIred/EIfull) was varied for various sewer span lengths. The results obtained are shown in Figure 11. It is seen from the figure that the effect of sewer reduced bending stiffness becomes obvious for the sewer pipe span of less than 200m. This suggests that sewer behaves more like a flexible material as its span length in-creases. In this project, the maximum span length of the sewer pipe is about 450m and thus the sewer pipe behaves as a flexible material and its settlement at the mid span follows closely that of the soil.

0

2

4

6

8

10

12

14

0 50 100 150 200 250 300 350

Sewer pipe span length (m)

So

il s

ett

lem

en

t / sew

er

sett

lem

en

t

1.00

0.75

0.50

0.25

0.10

Series6

EIred / EIfull :

Soil settlement = sewer settlement

Figure 11 Effect of sewer pipe bending stiffness

50 years 100 years

120 years

PVD PVD PVD

Input soil movement (120 years)

50 years

100 years 120 years

PVD PVD PVD

Distance from

Manhole A (m)

Distance from

Manhole A (m)


308

7 CONCLUSIONS

An analysis of sewer settlement in consolidating soft clays has been presented. The following conclu-sions can be drawn:

1. Back-analyses of CPT data using finite element method with consolidating soils modeled using soft soil model provide reasonable match with the field and laboratory shear strength data. The coefficient of permeability of the consolidating soils from the back-analyses represents the true value based on the analyses.

2. The use of Mohr-Coulomb model for analyzing consolidation of soft clays greatly underpre-dicts the consolidation settlement and thus is not recommended. The Mohr-Coulomb model may still be used to estimate soft soil consolidation settlement provided correctly adjusted modulus of elasticity of the soils is used in the analysis. Terzaghi’s model estimates almost similar magnitude of total consolidation settlement compared to the soft soil model but with lesser rate of settlement. It is recommended to use the soft soil model for modeling consolidat-ing clay layers with the large-strain consideration.

3. The effect of reduced sewer pipe bending stiffness becomes significant for sewer pipes with a span length of less than 200m. The use of 50% reduction in the sewer pipe bending stiffness is reasonable.

4. The PVD ground improvement scheme adopted is effective in limiting the maximum differen-tial settlement to below the allowable value of 1 in 120.

5. Since this is only a prediction of sewer settlement, it is important to instrument the soils after the PVDs are installed to get a more realistic picture of the behavior of the consolidating soils. Back-analyses will be required to re-estimate the long-term behavior of the sewer pipe.

8 REFERENCES

Bo, M.W., Tint, S.M. & Choa, V. 2002. Correlation of physical properties to strength and compression parame-ters of Singapore marine clay at Changi. In Proceedings of the 9th Cong. Engineering Geology for Develop-ing Countries, Durban, South Africa.

Buttling, S., Shirlaw, J.N., and James, J. 1987. The shear strength of Singapore marine clays. In Proceedings of the 5th International Geotechnical Seminar on Case Histories of Soft Clay, Singapore, 1987: 251-260.

Chai, J.C., Shen, S.L., Miura, N., and Bergado, T. 2001. Simple method of modelling PVD-improved subsoil. Journal of Geotechnical and Geoenvironmental Engineering, Vol. 127, No. 11: 965-972.

Chen, B. and Wen, Z. 2003. Elasto-plastic analysis for the effect of longitudinal uneven settlement on shield tunnel. (Re)Claiming the Underground Space, Saveur (ed.), Swets & Zeitlinger, Lisse: 969-973.

FHWA. 1986. Prefabricated vertical drains. Vollume 1: Engineering Guidelines. Federal Highway Administra-tion (FHWA), US Department of Transportation. Report No. FHWA/RD-86/168, August 1986.

Plaxis (2008) Plaxis 2D-Version 9.0 Reference Manual. Plaxis BV, The Netherlands. Skempton, A.W. 1957. Discussion on the planning and design of the new Hongkong airport. Proceedings of In-

stitution of Civil Engineers, Vol. 7: 305-307. Wong, K.S. and Choa, V. 1987. Settlement analysis of soft clay by finite difference method. In Proceedings of

the 5th International Geotechnical Seminar, Nanyang Technological Institute, Singapore, 1987: 283-289.


309

Underground Singapore 2011

Documents

Transcript of Underground Singapore 2011